Enabling or constraining? Unraveling the influence of organizational slack on innovation

Enabling or constraining? Unraveling the influence of organizational slack on innovation Abstract We employ theories of organizational search and agency costs to propose a contingency perspective that reconciles mutually contradictory prior findings on the relationship between organizational slack and innovation. First, we argue that influences of organizational slack depend on whether we consider exploitative innovation or exploratory innovation. Further, absorbed slack and unabsorbed slack differ in their forms of relationship with innovation. The ways in which a certain type of innovation is enabled by organizational slack are conditioned by distinct modes of organizational search associated with alternative types of innovation, as well as by the extent to which effective shareholder monitoring is disturbed by different types of organizational slack. An empirical analysis of 37 Japanese pharmaceutical firms’ new product developments over a 20-year period supports our argument. One of the most critical backbones of the organizations literature is that managers’ rationality is bounded (Simon, 1955). Given that managers are not able to compare all alternatives by precisely forecasting their consequences, it is important to maintain certain amount of flexibility to accommodate future adjustments. A typical source of organizational flexibility is organizational slack (Cyert and March, 1963; Bourgeois, 1981), or a class of organizational resource whose usage is left unspecified. However, the empirical results of works examining whether organizational slack benefits organizations are mixed. Particularly, findings from prior work concerning influences of organizational slack on innovation initiatives (or more generally, on risk-taking) are inconclusive at best. Some argue for a positive relationship (Meyer, 1982; Singh, 1986; Greve, 2007), while others argue for a negative relationship (Bromiley, 1991; Wiseman and Bromiley, 1996; Palmer and Wiseman, 1999; Latham and Braun, 2009). Further, some scholars argue for a curvilinear relationship in an inverted U-shape (Nohria and Gulati, 1996; Geiger and Cashen, 2002) as well as in a U-shape (Wiseman and Catanach, 1997), while nonfinding is also reported (Greve, 2003). Upon the examination of differential effects of organizational slack on exploratory innovation and exploitative innovation, Greve (2007) calls for more work with the larger sample to obtain more precise inferences. Some authors also examine the relationship between organizational slack and an important precursor of innovation initiatives, or risk-taking, by theorizing shifts in the locus of attention of decision makers (March and Shapira, 1992; Audia and Greve, 2006; Chen and Miller, 2007; Iyer and Miller, 2008; Lungeanu et al., 2016). Although they advanced the literature by uncovering the mechanism in which organizational slack influences risk-taking, the authors generally assume positive influences of organizational slack, thereby downplaying the possibility that organizational slack may disturb or discourage risk-taking initiatives under certain conditions. Furthermore, as central interests of the authors lie in the mechanism of performance feedback, organizational slack is theorized to moderate the mechanism, rather than to directly influence risk-taking. In this manuscript, we propose a contingency perspective to reconcile these inconclusive research findings—we argue that organizational slack differentially influences exploitative innovation and exploratory innovation (March, 1991). Extending prior work on alternative search modes enables us to theorize that organizational slack exerts negative influences on exploitative innovation, while positive influences on exploratory innovation. By building upon the theory of agency costs (Jensen and Meckling, 1976; Fama, 1980), we also argue that absorbed slack and unabsorbed slack (Singh, 1986) differ in their forms of relationship with innovation, as the former is more difficult to monitor than the latter. Specifically, we hypothesize that a negative association is observed between absorbed slack and exploitative innovation in a linear form, while the association between unabsorbed slack and exploitative innovation is curvilinear, in that the slope reveals diminishing negative effects of unabsorbed slack. As for exploratory innovation, we hypothesize linear positive influences of unabsorbed slack, whereas diminishing positive effects of absorbed slack. Our empirical analysis of the unique data on new pharmaceutical development by the Japanese firms supports our argument. In short, we attempt to advance our understanding of organizational slack by uncovering boundary conditions on whether we observe a positive, a negative, or a curvilinear relationship. We develop our arguments by showing that the agency theory complements the behavioral theory of the firm (Cyert and March, 1963), BTOF hereafter, in a previously unreported manner. Our findings imply that we could also bridge theories of slack and managerial attention (Ocasio, 1997; Joseph and Ocasio, 2012; Li et al., 2013) by indicating that organizational slack regulates attention of decision makers. 1. Theory and hypotheses 1.1 Types of innovation and organizational search Our first contingency refers to the differences between alternative types of innovation influenced by organizational slack. We particularly consider the possibility that organizational slack’s influences may not be identical between exploitative innovation and exploratory innovation (March, 1991; Voss et al., 2008). We define exploitative innovation as innovation initiatives targeted at improving or modifying existing knowledge utilized for current business, while innovation initiatives targeted at identifying or generating new knowledge that is beyond the scope of current business are defined as exploratory innovation (March, 1991; Sørensen and Stuart, 2000; Rosenkopf and Nerkar, 2001; Benner and Tushman, 2002; Katila and Ahuja, 2002; Lee et al., 2003; Piao, 2010). Accordingly, we refer to innovation as initiatives or resource commitments targeted to innovate new products, processes, or businesses, rather than as outcomes achieved by those initiatives. We particularly focus on the different search modes associated with exploitative and exploratory innovation. Exploitative innovation is enabled by problemistic search, or “search that is stimulated by a problem (usually a rather specific one) and is directed toward finding a solution to that problem” (Cyert and March, 1963: 121). Although one may be tempted to associate problemistic search with high risk-taking choices in that departures from the status quo entail some risk, problemistic search leads organizations to select less, rather than more, risky alternatives, e.g. cost cuts, for more reliable and predictable resolution of the problem (Deephouse and Wiseman, 2000; Iyer and Miller, 2008; Bromiley and Washburn, 2011; Kacperczyk et al., 2015). Accordingly, problemistic search “emphasizes relatively immediate refinements in the existing technology, greater efficiency, and discoveries in the near neighborhood of the present activities” (Levinthal and March, 1981: 309). Put differently, problemistic search is the first-order search (Argyris, 1976; Watzlawick et al., 1974), in that it is a reaction in terms of “existing rules” (Levinthal and March, 1981: 309). Therefore, problemistic search is an important precursor of exploitative innovation because problemistic search is characterized with intentional efforts to improve current performance by addressing deficiencies of current knowledge (Levinthal and March, 1981; Stuart and Podolny, 1996; Levinthal, 1997; Knudsen and Levinthal, 2007), or by searching “in the neighborhood of the current alternative” (Cyert and March, 1963: 121). Organizations may also adopt slack search (Levinthal and March, 1981; March, 1981; Nohria and Gulati, 1996; Greve, 2003) in a hope to identify new ideas and knowledge that may be useful for future new business opportunities, rather than for addressing current business requirements. Unlike problemistic search, the locus of slack search is nonlocal (or distant). Put differently, slack search is not tightly linked to current organizational goals because it reflects “the irrelevant wanderings of loosely controlled subunits” (March, 1981: 214). Slack search is also characterized as “second-order” (Watzlawick et al., 1974; Argyris, 1976) search because it enables “changes in performance targets, technological opportunities, search behavior, and knowledge about opportunities” (Levinthal and March, 1981: 310). Although organizations may discover “something of considerable value” (March, 1981: 214) through slack search, the likelihood of such discovery is not high. In other words, slack search entails higher risk compared to problemistic search that is characterized with certain and predictable outcomes. Accordingly, slack search is an important precursor for exploratory innovation that is enabled by new knowledge beyond the scope of current business. Those differences in underlying search modes inform our discussion on differential influences of organizational slack on alternative types of innovation as discussed below. 1.2 Types of slack and agency theory Second, we consider alternative types of organizational slack as a potential contingency because the type of organizational slack may determine its contribution to performance (Lecuona and Reitzig, 2014). One challenge to examine absorbed slack and unabsorbed slack separately is that there is no a priori theory about their differential effects (Wiseman and Catanach, 1997; Singh, 1986). In this manuscript, we argue that the agency theory represents a promising theoretical perspective with which to evaluate how absorbed slack and unabsorbed slack differ in terms of their influences on managerial risk-taking, and thus on subsequent innovation. Accordingly, we argue that absorbed slack and unabsorbed slack differ in terms of the degree that managers versus shareholders have conflicts regarding risk-taking. Absorbed slack is organizational slack that is distributed to particular usages, or “absorbed into the system design as excess costs” (Bourgeois and Singh, 1983: 43). Examples of absorbed slack include excess inventory, excess machine capacity, and indirect staff. It also is denoted as recoverable slack (Bourgeois and Singh, 1983), or low-discretion slack (Sharfman et al., 1988). On the other hand, unabsorbed slack is an alternative type of organizational slack that is excess, liquid, and uncommitted resources in an organization. Unabsorbed slack is also more readily redeployable because it is not assigned to any particular usages (Bourgeois and Singh, 1983; Singh, 1986). The best examples of unabsorbed slack are cash and marketable securities. Scholars also use available slack (Bourgeois and Singh, 1983), or high-discretion slack (Sharfman et al., 1988) to denote unabsorbed slack. The differences between alternative organizational slacks pertain to the degree of conflict of interests between shareholders and managers. Compared to shareholders, managers are less willing to pursue risk-taking initiatives because managers’ well-being is closely tied to the fate of their organizations (Fama, 1980; Hill and Snell, 1989; Baysinger et al., 1991; Francis and Smith, 1995), while shareholders can reduce their exposure to financial risks by diversifying their investments (Fama, 1980). Consequently, managers often choose to benefit themselves by avoiding risk-taking at the cost of shareholders unless appropriate monitoring tools are in place (Jensen and Meckling, 1976; Fama, 1980; Walkling and Long, 1984; Jensen, 1986; Malatesta and Walkling, 1988). We argue that such conflicts of interest between managers and shareholders are particularly relevant to the usage of absorbed slack because monitoring the usage of absorbed slack is more difficult than monitoring the usage of unabsorbed slack. Put differently, although the expected usage may be more narrowly specified for absorbed slack than for unabsorbed slack, that does not mean monitoring the usage of absorbed slack is easier. First, it is very difficult for external stakeholders to identify excess costs, or absorbed slack (Bourgeois and Singh, 1983; Jensen and Meckling, 1976; Love and Nohria, 2005). In fact, it is also difficult (or at least very costly) for managers to precisely identify excess portions of total costs. Furthermore, it is very difficult to monitor whether general and administrative expenses, one of the typical examples of absorbed slack, are properly used because firms do not disclose how much is spent on what. On the other hand, it is relatively easy to monitor the usage of unabsorbed slack because unabsorbed slack can be clearly identified as liquid resources in excess of current business requirements by capturing current assets that exceed current liabilities. For example, investment in marketable securities, a typical example of unabsorbed slack, is fully disclosed for shareholders’ examination of its appropriateness. Consequently, quantifying gains from absorbed slack is very difficult, whereas returns from unabsorbed slack can be quantified relatively easily. These characteristics of absorbed slack pose substantial challenges for shareholders who try to monitor its appropriate usages. In contrast, monitoring the amount or appropriate usages of unabsorbed slack is relatively straightforward. Therefore, we argue that managers may avoid risk-taking to the extent that monitoring by shareholders is difficult, or that more absorbed slack is available. On the other hand, unabsorbed slack does not pose such challenges for shareholders’ monitoring, and therefore shareholders can pressure managers to be more risk-taking in their decisions to innovate. 1.3 Organizational slack and exploitative innovation Building upon the above, we first argue that organizational slack is negatively associated with exploitative innovation (as show in the left-hand side of Figure 1). This is because increases in organizational slack are associated with being less responsive to competitive changes and feeling less urgency to solve current performance problems (Litschert and Bonham, 1978; Yasai-Ardekani, 1986). It is due to the fact that organizations with more organizational slack are less likely to make changes in their technical core because organizational slack buffers organizations from competitive requirements by absorbing environmental variation (Thompson, 1967). Figure 1. View largeDownload slide The relationship between organizational slack and innovation. Figure 1. View largeDownload slide The relationship between organizational slack and innovation. More specifically, managers can use organizational slack to “pay the price” of “a relatively loose fit between” (Litschert and Bonham, 1978: 216) their choices (in terms of organizational design, strategic decisions, and operational initiatives) and those “dictated by contextual variables” (ibid.). For example, with more absorbed slack, organizations can reduce their bankruptcy risk (Reuer and Leiblein, 2000) because they can avoid profit decreases by simply cutting excess costs, or by decreasing absorbed slack (Cyert and March, 1963). Organizational slack is also associated with lower responsiveness to competitive requirements because organizations may accumulate organizational slack to avoid excessive upward adjustment of organizational aspiration (ibid.). More formally, organizations can keep their attainment discrepancy (Levinthal and March, 1981; Lant, 1992), or the discrepancy between a performance target and achieved performance, at a minimum by adjusting realized performance either by decreasing or by increasing organizational slack. Put differently, organizations with more organizational slack are associated with having smaller and less frequent attainment discrepancies. Consequently, an organization’s search for solutions to a performance problem, or a “problemistic search,” is expected to be less intensive when more organizational slack is available (Cyert and March, 1963: 80). Accordingly, organizations are less motivated to improve current performance by innovations that closely address deficiencies of current knowledge, or exploitative innovation, to the extent that organizational slack increases. We further argue that organizational slack’s influences to discourage exploitative innovation should reveal different forms depending on the types of organizational slack, or between absorbed slack and unabsorbed slack (Singh, 1986). Given that monitoring the usage of absorbed slack poses greater challenges to shareholders (as discussed above), we expect that the negative association between organizational slack and exploitative innovation is particularly salient in the case of absorbed slack. Put differently, organizations with more absorbed slack, often characterized with internally oriented resource allocation patterns (Cheng and Kesner, 1997), are less subject to supervision by shareholders who aim to ensure that managers pursue sufficient risk-taking initiatives, thereby showing lower tolerance for risk-taking associated with innovation (Deephouse and Wiseman, 2000; Steensma and Corley, 2001). Our first hypothesis is stated as follows. H1. The association between absorbed slack and exploitative innovation is negative. On the other hand, we expect that the negative association between organizational slack and exploitative innovation grows less explicit as more unabsorbed slack is available. Put differently, the form of relationship between unabsorbed slack and exploitative innovation is nonlinear, in that the negative influences of unabsorbed slack on exploitative innovation weakens (or the slope grows less steep) as unabsorbed slack increases. Like absorbed slack, unabsorbed slack may also allow organizations to be less responsive to current environmental requirements, thereby discouraging exploitative innovation. However, because shareholders can more easily monitor the usage of unabsorbed slack (as discussed above), it would be increasingly difficult for organizations to forego opportunities (or ignore requirements) of exploitative innovation, as available unabsorbed slack grows more substantial and salient. Accordingly, we expect to observe diminishing negative effects of unabsorbed slack on exploitative innovation because the marginal decrease in exploitative innovation diminishes as unabsorbed slack’s buffering effects are offset by effective shareholders monitoring to the extent that more unabsorbed slack is available. Therefore, we propose our second hypothesis as follows. H2. The association between unabsorbed slack and exploitative innovation is curvilinear, in that the former exerts diminishing negative effects on the latter. 1.4 Organizational slack and exploratory innovation On the other hand, we argue that organizational slack is positively associated with exploratory innovation (as show in the right-hand side of Figure 1). Organizations insulated from current competitive requirements decrease their efforts in exploitative innovation, but they may instead pursue exploratory innovation. As is widely acknowledged, current competitive pressure discourages organizations’ efforts in exploratory innovation (Cooper and Smith, 1992; Christensen and Bower, 1996) because returns from exploratory innovation are uncertain and remote, if any returns are gained at all (March, 1991). Accordingly, organizational slack may enable exploratory innovation by buffering organizations from current competitive environments. In short, the effects of organizational slack are asymmetrical between exploitative and exploratory innovation. More formally, organizational slack allows organizations to satisfice during searches by lowering the threshold for acceptability (Bourgeois, 1981: 36) so that “projects that would not necessarily be approved in a tight budget” are accepted (Cyert and March, 1963: 279). Such resource munificence may be associated with less rigorous evaluation of alternatives, which encourages distant search (Knudsen and Levinthal, 2007). Consequently, “slack provides a source of funds for innovations that would not be approved in the face of scarcity” (Cyert and March, 1963: 279). Put differently, managers’ decisions and actions become more exploratory as more organizational slack is available (March, 1994, 2006; March, 2007). The underlying assumption is that organizational decisions are often outcomes of political concessions among competing managerial coalitions (Cyert and March, 1963). Without uncommitted excess resources, or organizational slack, initiatives with uncertain future consequences are closely scrutinized by competing managerial coalitions before approval is given, if any is given at all. Conversely, organizations with more organizational slack may more likely to accept experimental initiatives that would not be justified based on short-term profit forecasts but that look promising in terms of long-term profit potential. Such an experiment or a “irresponsible” search (Levinthal and March, 1981: 309) that is motivated and enabled by organizational slack is termed “slack search” (Levinthal and March, 1981; Nohria and Gulati, 1996; Greve, 2003); these searches are distinct from “problemistic search” in that the motivation for a slack search is not associated with the need to address a particular performance problem. Instead, slack searches enable organizations to “foster future growth through the development of new and different products or processes” (Souder and Shaver, 2010: 1318) that are not constrained by current competitive requirements. Accordingly, we argue that innovation targeted at identifying or generating new knowledge that is beyond the scope of current business, or exploratory innovation, increases to the extent that more organizational slack is available for slack search. We further argue the positive relationship between organizational slack and exploratory innovation may be particularly salient in the case of unabsorbed slack because shareholders are able to effectively monitor the usage of unabsorbed slack, and then force managers to pursue exploratory innovation vigorously. Accordingly, organizations with more unabsorbed slack are characterized with externally oriented resource allocation patterns (Cheng and Kesner, 1997) that may enable them to search for nonlocal and novel knowledge (Smith et al., 1991; Mishina et al., 2004; Carnabuci and Operti, 2013), or to explore new technologies and markets (Danneels, 2008). H3: The association between unabsorbed slack and exploratory innovation is positive. On the other hand, we argue for diminishing positive effects of absorbed slack on exploratory innovation. Put differently, the form of relationship between absorbed slack and exploratory innovation is nonlinear, in that the positive influences of absorbed slack on exploratory innovation weakens (or the slope grows less steep) as absorbed slack increases. We expect that absorbed slack also allows organization to pursue exploratory innovation by relaxing managerial coalitions’ selective screening. However, marginal increase in exploratory innovation diminishes because managers grow increasingly cautious toward additional risk-taking. Furthermore, shareholders find it increasingly difficult to make sure that their managers pursue sufficient risk-taking initiatives because increases in absorbed slack disturbs shareholders to effectively monitor the usage of absorbed slack (Jensen, 1986; Kim et al., 2008). Consequently, we argue that absorbed slack’s enabling effects on exploratory innovation diminish as available absorbed slack increases. H4: The association between absorbed slack and exploratory innovation is curvilinear, in that the former exerts diminishing positive effects on the latter. 2. Methods 2.1 Sample We tested the hypotheses with data from the Japanese pharmaceutical industry. We particularly leveraged data on their new pharmaceutical products development to operationalize our sample firms’ degree of exploitative, as well as exploratory innovation because new product development is often used as a measure of firms’ innovation (Greve, 2003; Voss et al., 2008; Natividad, 2013). The data on the Japanese pharmaceutical firms’ new products development are appropriate for our study for following two reasons. First, upon the approval of all new ethical drugs, independent specialists determine whether each new pharmaceutical contains an NCE (new chemical entity). This distinction enables us to precisely operationalize exploratory innovation and exploitative innovation because an NCE-based pharmaceutical product represents exploration of new knowledge in the context of new pharmaceutical development, while a non-NCE-based pharmaceutical product captures exploitation of existing knowledge (Bierly and Chakrabarti, 1996; Cardinal, 2001; Dunlap-Hinkler et al., 2010). An NCE is a completely new chemical entity whose medical effects were unknown before. Therefore, finding an NCE requires a search beyond known libraries of active ingredients, while pharmaceutical firms reuse NCEs already approved for medical use to develop non-NCE-based products. One good example of an exploratory pharmaceutical product is Eli Lilly’s Prozac which is based on an NCE, called fluoxetine. Non-NCE version of the same chemical entity is Sarafem, which is an example of an exploitative pharmaceutical product. Fluoxetine was successfully developed as an anti-depressant (Prozac) before Eli Lilly redeveloped it for a different indication of premenstrual dysphoric disorder (Sarafem) upon Prozac’s patent expiration. It is important to note that we are concerned about knowledge underlying the NCE. In particular, we aim to capture the knowledge’s degree of continuity or similarity with the current knowledge base of the organization. Accordingly, the degree of innovativeness or radicalness of benefits enabled by the focal knowledge is irrelevant to our operationalization of exploitative innovation and exploratory innovation. Second, rich data on sample firms’ new product development activities are available. Pharmaceutical firms are required to report on their clinical trial activities to the regulatory agency, which then discloses the information to the public. Leveraging these disclosed data, we are able to measure sample firms’ degree of exploitative, as well as exploratory innovation objectively. A professional medical magazine, called New Current, has been publishing exhaustive lists of pharmaceuticals under development (or pipelines) on a quarterly basis since 1990. The list shows each pharmaceutical firm’s detailed pipeline information, including the name of pipelines, targeted therapeutic indications, stages of clinical trials, and whether each pipeline contains an NCE. Our database consists of 37 Japanese pharmaceutical firms who gained new pharmaceutical approvals during January 2001 to December 2010 in the Japanese market. Combined revenue of these 37 firms represents 58.0% of the total Japanese health-care market as of 2010. We constructed a panel database on these 37 firms over 20 years (from 1991 to 2010). After removing observations due to missing values in at least one variable of interest, we end up with a final data set of 597 firm-years. 2.2 Variables and analysis To test our hypotheses, we constructed a measure of exploitative innovation and exploratory innovation and tested their associations with sample firms’ degree of organizational slack. Natividad (2013) points out the possibility that managers endogenously adjust the amount of organizational slack in response to prior organizational performance. Accordingly, we needed to address the omitted variable problem and the reverse causality problem. Specifically, we employed an instrumental variables estimation (Bascle, 2008; Wooldridge, 2010) with panel data. Furthermore, because panel data include multiple observations per sample firm, observations for the same firm are likely to be correlated. As such, our models applied an instrumental variable method to a data set with possible autocorrelation and heteroskedasticity. Therefore, we chose to employ a continuous-updating estimator (Hansen et al., 1996), which provides us HAC (heteroskedasticity and autocorrelation consistent) estimation. We used a Stata 13 command “xtivreg2” (Schaffer, 2010) with “cue” and “robust” options. Below, we describe variables employed in our models. We used annual data to construct all variables. We also lagged most right-hand-side variables 1 year, so that we can mitigate the possibility of reverse causation. Our dependent variable to test H1 and H2, or exploitative innovation, is operationalized by counting sample firms’ non-NCE-based pipelines at the end of each fiscal year. Because larger firms are likely to be associated with more pipelines, we divided the measure by annual research and development (R&D) expenses. We measured our dependent variable to test H3 and H4, exploratory innovation, in a similar way by counting NCE-based pipelines. Our independent variables are absorbed slack for H1 and H4, unabsorbed slack for H2 and H3, respectively. Following prior works (Bourgeois, 1981; Bourgeois and Singh, 1983; Singh, 1986; Bromiley, 1991; Greve, 2003; Bromiley and Washburn, 2011), we measured absorbed slack by dividing sample firms’ annual selling, general, and administrative expenses with total revenue. As for unabsorbed slack, we divided sample firms’ current assets by current liabilities at the end of each fiscal year. As Natividad (2013) points out, our independent variables (i.e. organizational slack) may be endogenous variables. Accordingly, a set of instrumental variables was employed to obtain fitted values of the independent variables, which were used to estimate our dependent variables (Bascle, 2008; Wooldridge, 2010). We lagged all instrumental variables 1 year to the independent variables, and therefore 2 years to the dependent variables, so that we could minimize the possibility that our instrumental variables influence the dependent variables. As for absorbed slack, we employed R&D intensity and total assets (trillion yen) as instruments. Because absorbed slack is “absorbed into the system design as excess costs” (Bourgeois and Singh, 1983: 43), it is highly conceivable that absorbed slack would be used to buffer absorbing processes or subunits from unpredictable variations. In other words, absorbed slack is not committed to specific purposes, but some processes or subunits define its default usages by absorbing it. Therefore, we argue that organizations are motivated to increase absorbed slack to the extent that they recognize their current processes or subunits as unpredictable. One typical source of unpredictability for pharmaceutical firms’ current operation is R&D. Accordingly, we employed R&D intensity, or annual R&D expenditure divided by total revenue, as a measure of the degree to which sample firms are motivated to increase absorbed slack as a buffer against negative surprises. Furthermore, stabilizing processes or subunits calls for substantial buffer resources to the extent that the focal processes or subunits have deployed larger resources. Therefore, we expect a positive association between total assets and absorbed slack because firms with larger total assets may need more absorbed slack as their buffer. Instruments for unabsorbed slack include environmental dynamics and revenue growth. Unabsorbed slack differs from absorbed slack in that unabsorbed slack is not absorbed into “the system design,” indicating that no organizational decisions are made as to which processes or subunits can claim a privilege to use this portion of uncommitted resources. Put differently, organizations reserve unabsorbed slack for usages beyond current operations. Accordingly, we argue that organizations increase unabsorbed slack to the extent that they anticipate major changes in their competitive environments due to a high degree of environmental dynamics. Our measure of environmental dynamics reflects the volatility of competitive environments. Following Wang and Li (2008), we regressed the Japanese health-care industry sales over a moving window of 30 years preceding the focal year on the year variable, and then used the standard error of the regression coefficient related to a year variable to produce an index of environmental dynamics. Conversely, managers may feel that major changes in their competitive environments are unlikely to the extent that their current business performance is favorable. We operationalized the degree of favorable business performance by revenue growth, or annual growth rate of sample firms’ total revenue, which we expect decreases firms’ unabsorbed slack in a subsequent year. Weak identification tests by Cragg-Donald Wald F statistic (Cragg and Donald, 1993) revealed that we can reject the null hypothesis that our instruments are weak, or only marginally relevant for both sets of instruments. Tests of overidentifying restrictions by Hansen J statistic (Hansen, 1982) indicated that the null hypothesis that all instruments are valid is not rejected. Furthermore, n times the R2 from the first stage of two-stage least squares are much larger than the number of instruments, indicating that two-stage least squares are less biased than ordinary least squares for our models (Murray, 2006). We also employed several time-varying control variables that corresponded to each time period in which sample firms were at risk of innovating. First, organizational size is our sample firms’ number of employees (in thousands). R&D is also employed as a measure of the degree of sample firms’ innovative capacity operationalized by their annual R&D expenditure (billion yen). We also included sample firms’ age to control for effects of sample firms’ senescence. A dummy variable that indicates whether sample firms experienced mergers and acquisitions in a preceding year (M&As) controls for influences of drastic changes in their pipelines. We also employed a measure of sample firms’ attainment discrepancy because it influences their degree of risk-taking. We operationalized it by the discrepancy in ROA (return on assets) between the focal firm and the industry average. Because we measure sample firms’ degree of innovation by counting pipelines, it also is important to control for the degree of development resources intensity. Therefore, we also included average size of pipelines by dividing annual R&D expenditure by total counts of pipelines at the end of each fiscal year. It also may be important to consider influences of government regulations. The Japanese government periodically reviews official reimbursement prices of pharmaceutical products downward. The revisions strongly influence pharmaceutical firms’ new product development strategy. Accordingly, we included a variable that shows the industry average degree of reimbursement price revisions. Managers may be more anxious to raise their assets’ productivity to the extent that their assets increase (Abernathy, 1978). Therefore, we included a measure of annual expansion of total assets, or sample firm’s asset growth, to control for assets size effects on managers’ risk avoidance. Furthermore, because sample firms’ potential slack (Bromiley, 1991; O'brien, 2003) may influence their degree of innovation, each firm’s debt–equity ratio is also employed. Finally, lagged exploratory innovation is also included for models with exploitative innovation as a dependent variable (H1 and H2) because new ideas and knowledge gained through exploratory innovation may enable subsequent exploitative innovation. 3. Results Table 1 shows descriptive statistics and a correlation matrix for all variables employed in our models. Overall, the independent and control variables show considerable variability, and most correlations among the variables range from small to moderate. We also checked the variance inflation factors for all variables, and none of them exceeds 10.0, which is the rule of thumb threshold of potential multicollinearity (Cohen et al., 2003). Table 1. Descriptive statistics and correlations Variables  Mean  Standard deviation  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  1. Exploitative innovation  0.53  0.75                                  2. Exploratory innovation  1.01  1.10  0.64*                                3. R&D intensity  0.12  0.24  0.11*  −0.01                              4. Total assets  0.50  0.61  −0.32*  −0.39*  −0.14*                            5. Environmental dynamics  0.19  0.06  0.00  0.13*  0.06  0.00                          6. Revenue growth  0.03  0.12  −0.09*  −0.03  −0.26*  0.07  0.06                        7. Organizational size  7.46  8.90  −0.30*  −0.39*  −0.18*  0.84*  −0.02  0.05                      8. R&D  26.86  35.85  −0.29*  −0.37*  0.00  0.62*  0.04  0.12*  0.39*                    9. Age  68.79  18.88  −0.08*  −0.09*  −0.31*  −0.05  −0.06  0.08*  −0.01  0.23*                  10. M&As  0.01  0.09  0.05  0.02  0.00  −0.02  −0.03  −0.03  −0.02  0.01  0.06                11. Attainment discrepancy  0.00  0.06  0.11*  0.14*  0.48*  −0.12*  0.01  −0.30*  0.07  −0.27*  −0.22*  −0.02              12. Size of pipelines  1.50  2.20  −0.28*  −0.35*  −0.09*  0.51*  0.03  0.03  0.58*  0.33*  0.01  −0.04  0.00            13. Price revisions  0.04  0.04  0.02  0.10*  −0.05  −0.04  0.01  0.03  −0.04  −0.08  −0.05  −0.05  −0.02  0.01          14. Asset growth  1.04  0.16  −0.01  0.00  0.03  0.06  0.03  0.48*  0.00  0.06  −0.08  −0.01  −0.06  0.01  0.03        15. Potential slack  0.37  0.56  −0.08*  0.03  −0.11*  0.07  0.08  −0.03  0.21*  −0.08*  −0.09*  −0.04  0.26*  0.10*  0.09*  −0.06      16. Absorbed slack  0.31  0.10  0.27*  0.32*  0.33*  −0.46*  0.05  −0.14*  −0.54*  −0.14*  −0.13*  −0.03  0.06  −0.33*  0.02  −0.05  −0.15*    17. Unabsorbed slack  2.93  1.63  0.14*  0.05  0.47*  −0.18*  0.01  −0.11*  −0.36*  0.01  −0.09*  −0.01  −0.08  −0.17*  −0.05  0.05  −0.39*  0.37*  Variables  Mean  Standard deviation  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  1. Exploitative innovation  0.53  0.75                                  2. Exploratory innovation  1.01  1.10  0.64*                                3. R&D intensity  0.12  0.24  0.11*  −0.01                              4. Total assets  0.50  0.61  −0.32*  −0.39*  −0.14*                            5. Environmental dynamics  0.19  0.06  0.00  0.13*  0.06  0.00                          6. Revenue growth  0.03  0.12  −0.09*  −0.03  −0.26*  0.07  0.06                        7. Organizational size  7.46  8.90  −0.30*  −0.39*  −0.18*  0.84*  −0.02  0.05                      8. R&D  26.86  35.85  −0.29*  −0.37*  0.00  0.62*  0.04  0.12*  0.39*                    9. Age  68.79  18.88  −0.08*  −0.09*  −0.31*  −0.05  −0.06  0.08*  −0.01  0.23*                  10. M&As  0.01  0.09  0.05  0.02  0.00  −0.02  −0.03  −0.03  −0.02  0.01  0.06                11. Attainment discrepancy  0.00  0.06  0.11*  0.14*  0.48*  −0.12*  0.01  −0.30*  0.07  −0.27*  −0.22*  −0.02              12. Size of pipelines  1.50  2.20  −0.28*  −0.35*  −0.09*  0.51*  0.03  0.03  0.58*  0.33*  0.01  −0.04  0.00            13. Price revisions  0.04  0.04  0.02  0.10*  −0.05  −0.04  0.01  0.03  −0.04  −0.08  −0.05  −0.05  −0.02  0.01          14. Asset growth  1.04  0.16  −0.01  0.00  0.03  0.06  0.03  0.48*  0.00  0.06  −0.08  −0.01  −0.06  0.01  0.03        15. Potential slack  0.37  0.56  −0.08*  0.03  −0.11*  0.07  0.08  −0.03  0.21*  −0.08*  −0.09*  −0.04  0.26*  0.10*  0.09*  −0.06      16. Absorbed slack  0.31  0.10  0.27*  0.32*  0.33*  −0.46*  0.05  −0.14*  −0.54*  −0.14*  −0.13*  −0.03  0.06  −0.33*  0.02  −0.05  −0.15*    17. Unabsorbed slack  2.93  1.63  0.14*  0.05  0.47*  −0.18*  0.01  −0.11*  −0.36*  0.01  −0.09*  −0.01  −0.08  −0.17*  −0.05  0.05  −0.39*  0.37*  *P <0.05. Table 1. Descriptive statistics and correlations Variables  Mean  Standard deviation  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  1. Exploitative innovation  0.53  0.75                                  2. Exploratory innovation  1.01  1.10  0.64*                                3. R&D intensity  0.12  0.24  0.11*  −0.01                              4. Total assets  0.50  0.61  −0.32*  −0.39*  −0.14*                            5. Environmental dynamics  0.19  0.06  0.00  0.13*  0.06  0.00                          6. Revenue growth  0.03  0.12  −0.09*  −0.03  −0.26*  0.07  0.06                        7. Organizational size  7.46  8.90  −0.30*  −0.39*  −0.18*  0.84*  −0.02  0.05                      8. R&D  26.86  35.85  −0.29*  −0.37*  0.00  0.62*  0.04  0.12*  0.39*                    9. Age  68.79  18.88  −0.08*  −0.09*  −0.31*  −0.05  −0.06  0.08*  −0.01  0.23*                  10. M&As  0.01  0.09  0.05  0.02  0.00  −0.02  −0.03  −0.03  −0.02  0.01  0.06                11. Attainment discrepancy  0.00  0.06  0.11*  0.14*  0.48*  −0.12*  0.01  −0.30*  0.07  −0.27*  −0.22*  −0.02              12. Size of pipelines  1.50  2.20  −0.28*  −0.35*  −0.09*  0.51*  0.03  0.03  0.58*  0.33*  0.01  −0.04  0.00            13. Price revisions  0.04  0.04  0.02  0.10*  −0.05  −0.04  0.01  0.03  −0.04  −0.08  −0.05  −0.05  −0.02  0.01          14. Asset growth  1.04  0.16  −0.01  0.00  0.03  0.06  0.03  0.48*  0.00  0.06  −0.08  −0.01  −0.06  0.01  0.03        15. Potential slack  0.37  0.56  −0.08*  0.03  −0.11*  0.07  0.08  −0.03  0.21*  −0.08*  −0.09*  −0.04  0.26*  0.10*  0.09*  −0.06      16. Absorbed slack  0.31  0.10  0.27*  0.32*  0.33*  −0.46*  0.05  −0.14*  −0.54*  −0.14*  −0.13*  −0.03  0.06  −0.33*  0.02  −0.05  −0.15*    17. Unabsorbed slack  2.93  1.63  0.14*  0.05  0.47*  −0.18*  0.01  −0.11*  −0.36*  0.01  −0.09*  −0.01  −0.08  −0.17*  −0.05  0.05  −0.39*  0.37*  Variables  Mean  Standard deviation  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  1. Exploitative innovation  0.53  0.75                                  2. Exploratory innovation  1.01  1.10  0.64*                                3. R&D intensity  0.12  0.24  0.11*  −0.01                              4. Total assets  0.50  0.61  −0.32*  −0.39*  −0.14*                            5. Environmental dynamics  0.19  0.06  0.00  0.13*  0.06  0.00                          6. Revenue growth  0.03  0.12  −0.09*  −0.03  −0.26*  0.07  0.06                        7. Organizational size  7.46  8.90  −0.30*  −0.39*  −0.18*  0.84*  −0.02  0.05                      8. R&D  26.86  35.85  −0.29*  −0.37*  0.00  0.62*  0.04  0.12*  0.39*                    9. Age  68.79  18.88  −0.08*  −0.09*  −0.31*  −0.05  −0.06  0.08*  −0.01  0.23*                  10. M&As  0.01  0.09  0.05  0.02  0.00  −0.02  −0.03  −0.03  −0.02  0.01  0.06                11. Attainment discrepancy  0.00  0.06  0.11*  0.14*  0.48*  −0.12*  0.01  −0.30*  0.07  −0.27*  −0.22*  −0.02              12. Size of pipelines  1.50  2.20  −0.28*  −0.35*  −0.09*  0.51*  0.03  0.03  0.58*  0.33*  0.01  −0.04  0.00            13. Price revisions  0.04  0.04  0.02  0.10*  −0.05  −0.04  0.01  0.03  −0.04  −0.08  −0.05  −0.05  −0.02  0.01          14. Asset growth  1.04  0.16  −0.01  0.00  0.03  0.06  0.03  0.48*  0.00  0.06  −0.08  −0.01  −0.06  0.01  0.03        15. Potential slack  0.37  0.56  −0.08*  0.03  −0.11*  0.07  0.08  −0.03  0.21*  −0.08*  −0.09*  −0.04  0.26*  0.10*  0.09*  −0.06      16. Absorbed slack  0.31  0.10  0.27*  0.32*  0.33*  −0.46*  0.05  −0.14*  −0.54*  −0.14*  −0.13*  −0.03  0.06  −0.33*  0.02  −0.05  −0.15*    17. Unabsorbed slack  2.93  1.63  0.14*  0.05  0.47*  −0.18*  0.01  −0.11*  −0.36*  0.01  −0.09*  −0.01  −0.08  −0.17*  −0.05  0.05  −0.39*  0.37*  *P <0.05. Table 2 reports the results of our tests of hypotheses on absorbed slack’s influences (H1 and H4). Models 1 and 4 show the first-stage models, where we regress instrumental variables against our independent variable, i.e. absorbed slack. Exogenous variables for exploitative innovation and for exploratory innovation are also included in Models 1 and 4, respectively. All instrumental variables show strong association with our measure of absorbed slack. Models 2 and 5 are models with control variables. We then used fitted values of absorbed slack to estimate their associations with exploitative innovation (Model 3, H1), and with exploratory innovation (Models 6 and 7, H4). Table 3 follows the similar format with alternative independent variable of unabsorbed slack. We examine the association between unabsorbed slack and exploitative innovation (H2) by Models 10 and 11. Model 14 tests the association between unabsorbed slack and exploratory innovation (H3). Table 2. Results of the panel continuous-updating estimator (CUE) instrumental variable regression analysis for the effects of absorbed slack on innovationa Dependent variables  Absorbed slack   Exploitative innovation   Absorbed slack   Exploratory innovation   Model 1  Model 2  Model 3  Model 4  Model 5  Model 6  Model 7  R&D intensity  0.36***  (0.07)          0.36***  (0.07)              Total assets  0.06**  (0.02)          0.06***  (0.02)              Organizational size  0.01  (0.01)  0.10†  (0.05)  0.12**  (0.04)  0.01  (0.01)  0.07  (0.11)  −0.18  (0.17)  −0.13  (0.09)  R&D  0.00  (0.01)  0.00  (0.03)  0.03  (0.02)  0.00  (0.01)  0.13*  (0.06)  0.95*  (0.38)  −0.01  (0.03)  Age  −0.05†  (0.03)  −0.03  (0.20)  −0.09  (0.14)  −0.07**  (0.02)  −1.48**  (0.46)  −2.04***  (0.37)  −1.00***  (0.18)  M&As  0.00  (0.02)  −0.18†  (0.09)  −0.17  (0.13)  0.01  (0.01)  0.04  (0.19)  0.15  (0.19)  0.00  (0.19)  Attainment discrepancy  0.25†  (0.14)  −1.05*  (0.51)  −0.63  (0.44)  0.26†  (0.14)  1.62  (1.32)  1.67  (1.73)  0.13  (1.01)  Size of pipelines  0.00  (0.00)  0.00  (0.00)  0.00  (0.00)  0.00  (0.00)  −0.02  (0.02)  −0.03†  (0.02)  −0.02  (0.01)  Price revisions  0.04  (0.02)  0.26  (0.31)  0.32  (0.48)  0.04†  (0.02)  0.74  (0.45)  0.64  (0.64)  0.56  (0.60)  Asset growth  0.04  (0.04)  −0.16  (0.12)  −0.11  (0.11)  0.04  (0.04)  −0.07  (0.15)  −0.27  (0.23)  0.06  (0.17)  Potential slack  0.01  (0.01)  −0.03  (0.04)  −0.01  (0.05)  0.01  (0.01)  −0.04  (0.08)  −0.07  (0.07)  −0.13  (0.12)  Exploratory innovation  0.01†  (0.00)  0.34***  (0.07)  0.36***  (0.10)                  Absorbed slack          −2.02*  (1.02)          0.67  (1.01)  8.36***  (1.72)  Absorbed slack, squared                          −9.02***  (2.42)  Constant  −0.12*  (0.05)  0.33*  (0.13)      −0.12*  (0.05)  1.02***  (0.16)          N firm-years  597    597    597    597    597    597    597    N Firms  37    37    37    37    37    37    37    RMSE  0.04    0.42    0.44    0.04    0.67    0.83    0.71    Log likelihood  1054    −329    −344    1050    −604    −719    −625    F  25.6    27.4    3.68    25.7    6.08    8.37    8.21    Dependent variables  Absorbed slack   Exploitative innovation   Absorbed slack   Exploratory innovation   Model 1  Model 2  Model 3  Model 4  Model 5  Model 6  Model 7  R&D intensity  0.36***  (0.07)          0.36***  (0.07)              Total assets  0.06**  (0.02)          0.06***  (0.02)              Organizational size  0.01  (0.01)  0.10†  (0.05)  0.12**  (0.04)  0.01  (0.01)  0.07  (0.11)  −0.18  (0.17)  −0.13  (0.09)  R&D  0.00  (0.01)  0.00  (0.03)  0.03  (0.02)  0.00  (0.01)  0.13*  (0.06)  0.95*  (0.38)  −0.01  (0.03)  Age  −0.05†  (0.03)  −0.03  (0.20)  −0.09  (0.14)  −0.07**  (0.02)  −1.48**  (0.46)  −2.04***  (0.37)  −1.00***  (0.18)  M&As  0.00  (0.02)  −0.18†  (0.09)  −0.17  (0.13)  0.01  (0.01)  0.04  (0.19)  0.15  (0.19)  0.00  (0.19)  Attainment discrepancy  0.25†  (0.14)  −1.05*  (0.51)  −0.63  (0.44)  0.26†  (0.14)  1.62  (1.32)  1.67  (1.73)  0.13  (1.01)  Size of pipelines  0.00  (0.00)  0.00  (0.00)  0.00  (0.00)  0.00  (0.00)  −0.02  (0.02)  −0.03†  (0.02)  −0.02  (0.01)  Price revisions  0.04  (0.02)  0.26  (0.31)  0.32  (0.48)  0.04†  (0.02)  0.74  (0.45)  0.64  (0.64)  0.56  (0.60)  Asset growth  0.04  (0.04)  −0.16  (0.12)  −0.11  (0.11)  0.04  (0.04)  −0.07  (0.15)  −0.27  (0.23)  0.06  (0.17)  Potential slack  0.01  (0.01)  −0.03  (0.04)  −0.01  (0.05)  0.01  (0.01)  −0.04  (0.08)  −0.07  (0.07)  −0.13  (0.12)  Exploratory innovation  0.01†  (0.00)  0.34***  (0.07)  0.36***  (0.10)                  Absorbed slack          −2.02*  (1.02)          0.67  (1.01)  8.36***  (1.72)  Absorbed slack, squared                          −9.02***  (2.42)  Constant  −0.12*  (0.05)  0.33*  (0.13)      −0.12*  (0.05)  1.02***  (0.16)          N firm-years  597    597    597    597    597    597    597    N Firms  37    37    37    37    37    37    37    RMSE  0.04    0.42    0.44    0.04    0.67    0.83    0.71    Log likelihood  1054    −329    −344    1050    −604    −719    −625    F  25.6    27.4    3.68    25.7    6.08    8.37    8.21    a Robust standard errors adjusted for clustering by firm are in parentheses. Two-tailed tests for all effects. † P < 0.1, *P < 0.05, **P < 0.01, ***P < 0.001. Table 2. Results of the panel continuous-updating estimator (CUE) instrumental variable regression analysis for the effects of absorbed slack on innovationa Dependent variables  Absorbed slack   Exploitative innovation   Absorbed slack   Exploratory innovation   Model 1  Model 2  Model 3  Model 4  Model 5  Model 6  Model 7  R&D intensity  0.36***  (0.07)          0.36***  (0.07)              Total assets  0.06**  (0.02)          0.06***  (0.02)              Organizational size  0.01  (0.01)  0.10†  (0.05)  0.12**  (0.04)  0.01  (0.01)  0.07  (0.11)  −0.18  (0.17)  −0.13  (0.09)  R&D  0.00  (0.01)  0.00  (0.03)  0.03  (0.02)  0.00  (0.01)  0.13*  (0.06)  0.95*  (0.38)  −0.01  (0.03)  Age  −0.05†  (0.03)  −0.03  (0.20)  −0.09  (0.14)  −0.07**  (0.02)  −1.48**  (0.46)  −2.04***  (0.37)  −1.00***  (0.18)  M&As  0.00  (0.02)  −0.18†  (0.09)  −0.17  (0.13)  0.01  (0.01)  0.04  (0.19)  0.15  (0.19)  0.00  (0.19)  Attainment discrepancy  0.25†  (0.14)  −1.05*  (0.51)  −0.63  (0.44)  0.26†  (0.14)  1.62  (1.32)  1.67  (1.73)  0.13  (1.01)  Size of pipelines  0.00  (0.00)  0.00  (0.00)  0.00  (0.00)  0.00  (0.00)  −0.02  (0.02)  −0.03†  (0.02)  −0.02  (0.01)  Price revisions  0.04  (0.02)  0.26  (0.31)  0.32  (0.48)  0.04†  (0.02)  0.74  (0.45)  0.64  (0.64)  0.56  (0.60)  Asset growth  0.04  (0.04)  −0.16  (0.12)  −0.11  (0.11)  0.04  (0.04)  −0.07  (0.15)  −0.27  (0.23)  0.06  (0.17)  Potential slack  0.01  (0.01)  −0.03  (0.04)  −0.01  (0.05)  0.01  (0.01)  −0.04  (0.08)  −0.07  (0.07)  −0.13  (0.12)  Exploratory innovation  0.01†  (0.00)  0.34***  (0.07)  0.36***  (0.10)                  Absorbed slack          −2.02*  (1.02)          0.67  (1.01)  8.36***  (1.72)  Absorbed slack, squared                          −9.02***  (2.42)  Constant  −0.12*  (0.05)  0.33*  (0.13)      −0.12*  (0.05)  1.02***  (0.16)          N firm-years  597    597    597    597    597    597    597    N Firms  37    37    37    37    37    37    37    RMSE  0.04    0.42    0.44    0.04    0.67    0.83    0.71    Log likelihood  1054    −329    −344    1050    −604    −719    −625    F  25.6    27.4    3.68    25.7    6.08    8.37    8.21    Dependent variables  Absorbed slack   Exploitative innovation   Absorbed slack   Exploratory innovation   Model 1  Model 2  Model 3  Model 4  Model 5  Model 6  Model 7  R&D intensity  0.36***  (0.07)          0.36***  (0.07)              Total assets  0.06**  (0.02)          0.06***  (0.02)              Organizational size  0.01  (0.01)  0.10†  (0.05)  0.12**  (0.04)  0.01  (0.01)  0.07  (0.11)  −0.18  (0.17)  −0.13  (0.09)  R&D  0.00  (0.01)  0.00  (0.03)  0.03  (0.02)  0.00  (0.01)  0.13*  (0.06)  0.95*  (0.38)  −0.01  (0.03)  Age  −0.05†  (0.03)  −0.03  (0.20)  −0.09  (0.14)  −0.07**  (0.02)  −1.48**  (0.46)  −2.04***  (0.37)  −1.00***  (0.18)  M&As  0.00  (0.02)  −0.18†  (0.09)  −0.17  (0.13)  0.01  (0.01)  0.04  (0.19)  0.15  (0.19)  0.00  (0.19)  Attainment discrepancy  0.25†  (0.14)  −1.05*  (0.51)  −0.63  (0.44)  0.26†  (0.14)  1.62  (1.32)  1.67  (1.73)  0.13  (1.01)  Size of pipelines  0.00  (0.00)  0.00  (0.00)  0.00  (0.00)  0.00  (0.00)  −0.02  (0.02)  −0.03†  (0.02)  −0.02  (0.01)  Price revisions  0.04  (0.02)  0.26  (0.31)  0.32  (0.48)  0.04†  (0.02)  0.74  (0.45)  0.64  (0.64)  0.56  (0.60)  Asset growth  0.04  (0.04)  −0.16  (0.12)  −0.11  (0.11)  0.04  (0.04)  −0.07  (0.15)  −0.27  (0.23)  0.06  (0.17)  Potential slack  0.01  (0.01)  −0.03  (0.04)  −0.01  (0.05)  0.01  (0.01)  −0.04  (0.08)  −0.07  (0.07)  −0.13  (0.12)  Exploratory innovation  0.01†  (0.00)  0.34***  (0.07)  0.36***  (0.10)                  Absorbed slack          −2.02*  (1.02)          0.67  (1.01)  8.36***  (1.72)  Absorbed slack, squared                          −9.02***  (2.42)  Constant  −0.12*  (0.05)  0.33*  (0.13)      −0.12*  (0.05)  1.02***  (0.16)          N firm-years  597    597    597    597    597    597    597    N Firms  37    37    37    37    37    37    37    RMSE  0.04    0.42    0.44    0.04    0.67    0.83    0.71    Log likelihood  1054    −329    −344    1050    −604    −719    −625    F  25.6    27.4    3.68    25.7    6.08    8.37    8.21    a Robust standard errors adjusted for clustering by firm are in parentheses. Two-tailed tests for all effects. † P < 0.1, *P < 0.05, **P < 0.01, ***P < 0.001. Table 3. Results of the panel continuous-updating estimator (CUE) instrumental variable regression analysis for the effects of unabsorbed slack on innovationa Dependent variables  Unabsorbed slack   Exploitative innovation   Unabsorbed slack   Exploratory innovation   Model 8  Model 9  Model 10  Model 11  Model 12  Model 13  Model 14  Environmental dynamics  2.14†  (1.10)              2.24†  (1.13)          Revenue growth  −0.74*  (0.29)              −0.75*  (0.29)          Organizational size  −0.51***  (0.13)  0.10†  (0.05)  −0.15  (0.15)  −0.56  (0.36)  −0.51***  (0.13)  0.07  (0.11)  0.28*  (0.12)  R&D  −0.06  (0.13)  0.00  (0.03)  −0.01  (0.04)  −0.05  (0.11)  −0.05  (0.13)  0.13*  (0.06)  0.14**  (0.05)  Age  1.88***  (0.50)  −0.03  (0.20)  0.71  (0.48)  1.68†  (0.88)  1.78***  (0.46)  −1.48**  (0.46)  −2.14***  (0.35)  M&As  −0.25  (0.31)  −0.18†  (0.09)  −0.26  (0.20)  −0.26  (0.45)  −0.19  (0.30)  0.04  (0.19)  0.13  (0.15)  Attainment discrepancy  −0.37  (1.86)  −1.05*  (0.51)  −1.13  (0.78)  −4.18†  (2.15)  −0.30  (1.90)  1.62  (1.32)  1.58  (1.04)  Size of pipelines  0.00  (0.01)  0.00  (0.00)  0.01  (0.01)  0.01  (0.01)  0.00  (0.01)  −0.02  (0.02)  −0.03*  (0.01)  Price revisions  0.82†  (0.46)  0.26  (0.31)  0.54  (0.60)  1.15  (0.96)  0.87†  (0.46)  0.74  (0.45)  0.50  (0.67)  Asset growth  −0.02  (0.34)  −0.16  (0.12)  0.03  (0.21)  −0.34  (0.41)  −0.01  (0.34)  −0.07  (0.15)  0.04  (0.19)  Potential slack  −0.02  (0.08)  −0.03  (0.04)  −0.05  (0.04)  −0.09  (0.09)  −0.02  (0.08)  −0.04  (0.08)  −0.04  (0.05)  Exploratory innovation  0.07  (0.06)  0.34***  (0.07)  0.38***  (0.10)  0.35***  (0.10)              Unabsorbed slack          −0.45†  (0.27)  −1.42*  (0.70)          0.42**  (0.16)  Unabsorbed slack, squared              0.17*  (0.08)              Constant  −0.44  (0.47)  0.33*  (0.13)          −0.39  (0.45)  1.02***  (0.16)      N firm-years  597    597    597    597    597    597    597    N Firms  37    37    37    37    37    37    37    RMSE  0.73    0.42    0.57    0.96    0.73    0.67    0.74    Log likelihood  −652    −329    −492    −801    −653    −604    −648    F  7.48    27.4    2.30    1.94    7.41    6.08    10.1    Dependent variables  Unabsorbed slack   Exploitative innovation   Unabsorbed slack   Exploratory innovation   Model 8  Model 9  Model 10  Model 11  Model 12  Model 13  Model 14  Environmental dynamics  2.14†  (1.10)              2.24†  (1.13)          Revenue growth  −0.74*  (0.29)              −0.75*  (0.29)          Organizational size  −0.51***  (0.13)  0.10†  (0.05)  −0.15  (0.15)  −0.56  (0.36)  −0.51***  (0.13)  0.07  (0.11)  0.28*  (0.12)  R&D  −0.06  (0.13)  0.00  (0.03)  −0.01  (0.04)  −0.05  (0.11)  −0.05  (0.13)  0.13*  (0.06)  0.14**  (0.05)  Age  1.88***  (0.50)  −0.03  (0.20)  0.71  (0.48)  1.68†  (0.88)  1.78***  (0.46)  −1.48**  (0.46)  −2.14***  (0.35)  M&As  −0.25  (0.31)  −0.18†  (0.09)  −0.26  (0.20)  −0.26  (0.45)  −0.19  (0.30)  0.04  (0.19)  0.13  (0.15)  Attainment discrepancy  −0.37  (1.86)  −1.05*  (0.51)  −1.13  (0.78)  −4.18†  (2.15)  −0.30  (1.90)  1.62  (1.32)  1.58  (1.04)  Size of pipelines  0.00  (0.01)  0.00  (0.00)  0.01  (0.01)  0.01  (0.01)  0.00  (0.01)  −0.02  (0.02)  −0.03*  (0.01)  Price revisions  0.82†  (0.46)  0.26  (0.31)  0.54  (0.60)  1.15  (0.96)  0.87†  (0.46)  0.74  (0.45)  0.50  (0.67)  Asset growth  −0.02  (0.34)  −0.16  (0.12)  0.03  (0.21)  −0.34  (0.41)  −0.01  (0.34)  −0.07  (0.15)  0.04  (0.19)  Potential slack  −0.02  (0.08)  −0.03  (0.04)  −0.05  (0.04)  −0.09  (0.09)  −0.02  (0.08)  −0.04  (0.08)  −0.04  (0.05)  Exploratory innovation  0.07  (0.06)  0.34***  (0.07)  0.38***  (0.10)  0.35***  (0.10)              Unabsorbed slack          −0.45†  (0.27)  −1.42*  (0.70)          0.42**  (0.16)  Unabsorbed slack, squared              0.17*  (0.08)              Constant  −0.44  (0.47)  0.33*  (0.13)          −0.39  (0.45)  1.02***  (0.16)      N firm-years  597    597    597    597    597    597    597    N Firms  37    37    37    37    37    37    37    RMSE  0.73    0.42    0.57    0.96    0.73    0.67    0.74    Log likelihood  −652    −329    −492    −801    −653    −604    −648    F  7.48    27.4    2.30    1.94    7.41    6.08    10.1    a Robust standard errors adjusted for clustering by firm are in parentheses. Two-tailed tests for all effects. † P < 0.1, *P < 0.05, **P < 0.01, ***P < 0.001. Table 3. Results of the panel continuous-updating estimator (CUE) instrumental variable regression analysis for the effects of unabsorbed slack on innovationa Dependent variables  Unabsorbed slack   Exploitative innovation   Unabsorbed slack   Exploratory innovation   Model 8  Model 9  Model 10  Model 11  Model 12  Model 13  Model 14  Environmental dynamics  2.14†  (1.10)              2.24†  (1.13)          Revenue growth  −0.74*  (0.29)              −0.75*  (0.29)          Organizational size  −0.51***  (0.13)  0.10†  (0.05)  −0.15  (0.15)  −0.56  (0.36)  −0.51***  (0.13)  0.07  (0.11)  0.28*  (0.12)  R&D  −0.06  (0.13)  0.00  (0.03)  −0.01  (0.04)  −0.05  (0.11)  −0.05  (0.13)  0.13*  (0.06)  0.14**  (0.05)  Age  1.88***  (0.50)  −0.03  (0.20)  0.71  (0.48)  1.68†  (0.88)  1.78***  (0.46)  −1.48**  (0.46)  −2.14***  (0.35)  M&As  −0.25  (0.31)  −0.18†  (0.09)  −0.26  (0.20)  −0.26  (0.45)  −0.19  (0.30)  0.04  (0.19)  0.13  (0.15)  Attainment discrepancy  −0.37  (1.86)  −1.05*  (0.51)  −1.13  (0.78)  −4.18†  (2.15)  −0.30  (1.90)  1.62  (1.32)  1.58  (1.04)  Size of pipelines  0.00  (0.01)  0.00  (0.00)  0.01  (0.01)  0.01  (0.01)  0.00  (0.01)  −0.02  (0.02)  −0.03*  (0.01)  Price revisions  0.82†  (0.46)  0.26  (0.31)  0.54  (0.60)  1.15  (0.96)  0.87†  (0.46)  0.74  (0.45)  0.50  (0.67)  Asset growth  −0.02  (0.34)  −0.16  (0.12)  0.03  (0.21)  −0.34  (0.41)  −0.01  (0.34)  −0.07  (0.15)  0.04  (0.19)  Potential slack  −0.02  (0.08)  −0.03  (0.04)  −0.05  (0.04)  −0.09  (0.09)  −0.02  (0.08)  −0.04  (0.08)  −0.04  (0.05)  Exploratory innovation  0.07  (0.06)  0.34***  (0.07)  0.38***  (0.10)  0.35***  (0.10)              Unabsorbed slack          −0.45†  (0.27)  −1.42*  (0.70)          0.42**  (0.16)  Unabsorbed slack, squared              0.17*  (0.08)              Constant  −0.44  (0.47)  0.33*  (0.13)          −0.39  (0.45)  1.02***  (0.16)      N firm-years  597    597    597    597    597    597    597    N Firms  37    37    37    37    37    37    37    RMSE  0.73    0.42    0.57    0.96    0.73    0.67    0.74    Log likelihood  −652    −329    −492    −801    −653    −604    −648    F  7.48    27.4    2.30    1.94    7.41    6.08    10.1    Dependent variables  Unabsorbed slack   Exploitative innovation   Unabsorbed slack   Exploratory innovation   Model 8  Model 9  Model 10  Model 11  Model 12  Model 13  Model 14  Environmental dynamics  2.14†  (1.10)              2.24†  (1.13)          Revenue growth  −0.74*  (0.29)              −0.75*  (0.29)          Organizational size  −0.51***  (0.13)  0.10†  (0.05)  −0.15  (0.15)  −0.56  (0.36)  −0.51***  (0.13)  0.07  (0.11)  0.28*  (0.12)  R&D  −0.06  (0.13)  0.00  (0.03)  −0.01  (0.04)  −0.05  (0.11)  −0.05  (0.13)  0.13*  (0.06)  0.14**  (0.05)  Age  1.88***  (0.50)  −0.03  (0.20)  0.71  (0.48)  1.68†  (0.88)  1.78***  (0.46)  −1.48**  (0.46)  −2.14***  (0.35)  M&As  −0.25  (0.31)  −0.18†  (0.09)  −0.26  (0.20)  −0.26  (0.45)  −0.19  (0.30)  0.04  (0.19)  0.13  (0.15)  Attainment discrepancy  −0.37  (1.86)  −1.05*  (0.51)  −1.13  (0.78)  −4.18†  (2.15)  −0.30  (1.90)  1.62  (1.32)  1.58  (1.04)  Size of pipelines  0.00  (0.01)  0.00  (0.00)  0.01  (0.01)  0.01  (0.01)  0.00  (0.01)  −0.02  (0.02)  −0.03*  (0.01)  Price revisions  0.82†  (0.46)  0.26  (0.31)  0.54  (0.60)  1.15  (0.96)  0.87†  (0.46)  0.74  (0.45)  0.50  (0.67)  Asset growth  −0.02  (0.34)  −0.16  (0.12)  0.03  (0.21)  −0.34  (0.41)  −0.01  (0.34)  −0.07  (0.15)  0.04  (0.19)  Potential slack  −0.02  (0.08)  −0.03  (0.04)  −0.05  (0.04)  −0.09  (0.09)  −0.02  (0.08)  −0.04  (0.08)  −0.04  (0.05)  Exploratory innovation  0.07  (0.06)  0.34***  (0.07)  0.38***  (0.10)  0.35***  (0.10)              Unabsorbed slack          −0.45†  (0.27)  −1.42*  (0.70)          0.42**  (0.16)  Unabsorbed slack, squared              0.17*  (0.08)              Constant  −0.44  (0.47)  0.33*  (0.13)          −0.39  (0.45)  1.02***  (0.16)      N firm-years  597    597    597    597    597    597    597    N Firms  37    37    37    37    37    37    37    RMSE  0.73    0.42    0.57    0.96    0.73    0.67    0.74    Log likelihood  −652    −329    −492    −801    −653    −604    −648    F  7.48    27.4    2.30    1.94    7.41    6.08    10.1    a Robust standard errors adjusted for clustering by firm are in parentheses. Two-tailed tests for all effects. † P < 0.1, *P < 0.05, **P < 0.01, ***P < 0.001. As Model 3 shows, the coefficient for absorbed slack is negative and significant (P < 0.05), indicating a negative linear association between absorbed slack and exploitative innovation (H1). We also find a support for diminishing positive effects of absorbed slack on exploratory innovation (H4) in Model 7, that shows a positive and significant (P < 0.001) coefficient for the linear term of absorbed slack and negative and significant (P < 0.001) squared term. A Wald test indicates that adding a squared term creates a statistically significant improvement in the fit of the model (P < 0.001), lending a further support for the hypothesized curvilinear relationship. In addition, our hypothesis for diminishing negative effects of unabsorbed slack on exploitative innovation (H2) is also supported by a negative and significant (P < 0.05) as well as positive and significant (P < 0.05) squared term of unabsorbed slack in Model 11. Improvement in the fit of the model with the squared term is shown as statistically significant (P < 0.05) as well. Finally, Model 14 shows a positive and significant (P < 0.01) coefficient for unabsorbed slack, lending a support for our hypothesis on a positive association between unabsorbed slack and exploratory innovation (H3). Figure 2 shows predictive margins of exploitative innovation for ±two standard deviations of organizational slack (while other covariates are held constant at their means). Because xtivreg2 routines of Stata (Schaffer, 2010) does not estimate or report a constant with the fixed-effects model, reporting raw values of predictive margins is quite misleading. Accordingly, we report standardized value of predictive margins by dividing deviations from the predicted value at the mean of organizational slack with standard deviation of exploitative innovation. Figure 3 shows standardized predictive margins of exploratory innovation following the same format. Figure 2 graphically supports H1 and H2 by showing a linear negative relationship between absorbed slack and exploitative innovation (H1), as well as a diminishing negative relationship between unabsorbed slack and exploitative innovation (H2). Figure 3 provides graphical supports for H3 and H4, in that it shows a linear positive relationship between unabsorbed slack and exploratory innovation (H3), as well as a diminishing positive relationship between absorbed slack and exploratory innovation (H4). Figure 2. View largeDownload slide Influences of organizational slack on exploitative innovation. Figure 2. View largeDownload slide Influences of organizational slack on exploitative innovation. Figure 3. View largeDownload slide Influences of organizational slack on exploratory innovation. Figure 3. View largeDownload slide Influences of organizational slack on exploratory innovation. Observed patterns are, although not explicitly hypothesized, consistent with our theoretical arguments in that the organizational slack which is more flexibly redeployed as a readily available buffer, or unabsorbed slack, exerts more salient influences on exploitative innovation than less flexibly redeployable one, i.e. absorbed slack, does. On the other hand, we hypothesized that organizational slack relaxes discipline on resource allocation to exploratory innovation because slack, be it absorbed or unabsorbed, is an indication that each dominant coalition has satisfied its claim for the share of resources as “total necessary payments” (Cyert and March, 1963: 36) to members of respective coalition. Put differently, we have no reasons to expect major differences in the salience of influences on exploratory innovation between absorbed and unabsorbed slack. Our empirical results shown in Figure 3 are consistent with our hypothesized rationale that the sense of adequacy associated with organizational slack regulates the magnitude of influences on exploratory innovation. 4. Robustness tests We checked robustness of our findings by examining two alternative model specifications. First, we added lagged dependent variables to control for any residual unobserved heterogeneity across firms leading to systematic differences in innovation performance, although the fixed-effect estimator may be inconsistent when lagged regressors are introduced in instrumental variables estimation (Cameron and Trivedi, 2009). The results show that all hypotheses are statistically supported, corroborating our original findings. We further examined our results’ robustness by employing annual increases of non-NCE-based pipelines and NCE-based pipelines (rather than raw counts, which we included as controls) as alternative measures of exploitative as well as exploratory innovation, respectively. Firm size differences are taken into account by dividing the measures by annual R&D expenses. Results shown in Appendix Tables A1 and A2 are fully consistent with our original findings, except for Model 25 (Table A2) that examines H2. Because the model reveals a weak identification largely due to the addition of a squared term of unabsorbed slack, we employ an alternative approach to examine the hypothesized curvilinear relationship by including natural logarithm of unabsorbed slack instead of its linear and squared term. The Model 26 (Table A2) shows a statistically significant negative coefficient for this alternative independent variable, indicating that the relationship between unabsorbed slack and exploitative innovation is shown in a diminishing negative slope. Overall, the results show that our findings reported above are robust. 5. Discussion In this concluding section, we discuss our contributions. First, we reconciled mutually contradictory prior findings on organizational slack by proposing a contingency perspective on the relationship between organizational slack and innovation. The form of associations between organizational slack and innovation differs across different types of innovation, as well as across different types of organizational slack. Types of innovation and organizational slack are boundary conditions on whether we observe a positive, a negative, or a curvilinear relationship. Accordingly, our argument reconciles mutually contradictory perspectives on organizational slack provided by BTOF (Cyert and March, 1963) and the agency theory (Jensen and Meckling, 1976; Fama, 1980) in a new way. In prior works, scholars who argue for an inverted U-shaped relationship employ the two theories by grafting them together in that they apply BTOF and the agency theory separately to the inverted U-shape’s left-hand side and right-hand side, respectively (Nohria and Gulati, 1996; Geiger and Cashen, 2002). Specifically, as organizational slack increases, innovation increases because latent goal conflicts between managerial coalitions are resolved (Cyert and March, 1963). However, once organizational slack exceeds a certain optimal level, managerial discipline wanes. Managers then act in self-serving manners by selecting “pet R&D projects” at the cost of “value-added innovations to firms” (Nohria and Gulati, 1996: 1248). Several opportunities to refine the prior works’ theoretical explanation motivated our hypotheses development. First, when they argue that organizational slack increases innovation by resolving latent goal conflicts, they ignore an alternative argument that managers with more organizational slack may reduce diligent effort to pursue immediate innovation opportunities, as they feel they are buffered from the competitive pressure (Thompson, 1967). Because these two arguments are mutually contradictory, it is very difficult to reconcile them unless we consider alternative types of innovation influenced by organizational slack differentially. Second, the authors underplay organizational slack’s disturbing influences on shareholder monitoring, thereby obscuring differences between absorbed slack and unabsorbed slack in influences on the extent to which managers are forced to pursue risk-taking initiatives. We build on the argument for a complementarity between BTOF and the agency theory, and extend it by showing that we need to combine (rather than graft) a theory of search (Cyert and March, 1963; Levinthal and March, 1981; Greve, 2003) and a theory of shareholder monitoring (Jensen and Meckling, 1976; Fama, 1980) when we try to explain differential influences of organizational slack on innovation. We employ the theory of search to argue that organizational slack influences exploitative innovation and exploratory innovation differentially. Predominant influences of absorbed as well as unabsorbed slack are negative for exploitative innovation because organizational slack discourages problemistic search. On the other hand, organizational slack enables exploratory innovation by increasing slack search. Put differently, to the extent that absorbed, as well as unabsorbed slack is available, organizations shift their efforts from problemistic search to slack search, or from exploitative innovation to exploratory innovation. However, alternative slack’s forms of relationship with innovation are not exactly identical. Our argument theorizes about the differences between absorbed slack and unabsorbed slack by adjusting their predominant influences (derived from the search theory) with the differential degree of difficulties associated with monitoring by shareholders. Put differently, given the type of innovation considered, whether the relationship is linear or curvilinear depends on the type of organizational slack examined. We argue that absorbed slack is associated with more substantial monitoring challenges for shareholders than unabsorbed slack is. Monitoring appropriate usages of organizational slack is challenging for external stakeholders to the extent that constraints placed on the usage of organizational slack is idiosyncratic to organizational contexts. Without understanding such idiosyncratic contexts, it is very difficult to judge whether focal slack is appropriately used. Therefore, monitoring the usages of organizational slack is more challenging for shareholders to the extent that the focal slack is absorbed to particular expenses with idiosyncratic usages. On the contrary, unabsorbed slack is not constrained by such idiosyncrasy, thereby enabling more effective shareholder monitoring, as it grows more substantial and salient. Because those influences of organizational slack on shareholder monitoring increase as organizational slack increases, predominant influences of organizational slack on search diminish as more organizational slack is available. Specifically, absorbed slack disturbs efforts of shareholders to monitor their managers’ extent of risk-taking, thereby attenuating its positive relationship with exploratory innovation (H4). On the other hand, unabsorbed slack weakens its negative relationship with exploitative innovation by enabling shareholders to monitor more effectively (H2). In short, the type of slack examined is the second contingency that governs the form of the relationship between organizational slack and innovation. The theory of shareholder monitoring enables us to more appropriately explain influences of organizational slack by distinguishing the two types of organizational slack. As such, it is interesting to pursue implications of our findings on some important research fields related to studies of innovation. First, our findings inform the research on search (Cyert and March, 1963; Rosenkopf and Nerkar, 2001; Katila and Ahuja, 2002; Laursen and Salter, 2006; Lopez-Vega et al., 2016). We build our argument on the premise that search is an important precursor of innovation (Laursen, 2012), and accordingly an important contingency which characterizes differential types of innovation. The prior work on search focuses on the intensity of search (or resultant innovation), while efforts to examine search (or innovation) by acknowledging the existence of several distinct types of search are relatively limited (exceptions include Laursen (2012) or Lopez-Vega et al. (2016)). We employed distinct types of search as contingencies that explain differential relationships between organizational slack and innovation. Empirical supports to our argument that types of search regulate influences of organizational slack on resultant innovation attest to the importance of the distinction between different types of search. Furthermore, a perspective that dichotomizes problemistic search (Cyert and March, 1963) and slack search (Levinthal and March, 1981; Greve, 2003) is informed of the possibility that these two types of search are pursued as substitutes in that organizational slack encourages the latter while discourages the former as slack allows organizational decision makers to shift their locus of attention from attainment discrepancy to distant profit potential. Stated differently, our underlying rationale to relate organizational slack to differential types of search is premised on the attention-based theorizing (Ocasio, 1997) because search presupposes certain focus of attention by organizational decision makers. Accordingly, it also is important to note our findings’ implications for a theory of managerial attention (Ocasio, 1997; Joseph and Ocasio, 2012; Li et al., 2013). We argue that organizational slack provides buffers against environmental changes that influence exploitative innovation and exploratory innovation oppositely. Specifically, buffered organizations are less motivated to address current performance problems by way of exploitative innovation. In contrast, the same buffer encourages organizations to invest in exploratory innovation for possible future performance gains. Put differently, organizational slack influences managers’ temporal orientation (Souder and Shaver, 2010; Bromiley and Washburn, 2011), thereby allowing managers to shift their attention from “current viability” toward “future viability” (Levinthal and March, 1993). Accordingly, our findings indicate that the relationship between organizational slack and innovation initiatives is mediated by the distribution of managers’ attention across short-term concerns and long-term concerns. The prior work (March and Shapira, 1992; Audia and Greve, 2006; Chen and Miller, 2007; Iyer and Miller, 2008; Lungeanu et al., 2016) relates organizational slack to shifts in the focus of attention by showing that slack influences managers’ choice from their aspiration level, survival point, and organizational slack. Our findings extend the prior work by acknowledging more general roles of organizational slack as a regulator of attention (Ocasio, 1997). Specifically, we conjecture that organizational slack regulates the shift in the locus of attention of decision makers across alternative reference points, as well as across alternative timeframes they could employ. Put differently, organizational slack may regulate attention of decision makers beyond the context of performance feedback. As such, our findings call attention of scholars on the potential roles of organizational slack as a general regulator of attention, thereby informing the literature of attention about an influential antecedent of attention. Irrespective of the contributions, there are several limitations. First, we were not able to examine the way in which measures to reduce agency costs moderate the relationship between organizational slack and innovation within the scope of current research. Given that alternative types of organizational slack are associated with different degrees of agency costs, several measures to reduce agency costs—including managerial stock holding (Jensen and Meckling, 1976; Walkling and Long, 1984; Eisenhardt, 1989) and stock concentration (Hill and Snell, 1989; Baysinger et al., 1991; Francis and Smith, 1995)—may attenuate differences observed across different types of organizational slack. For example, we argue that absorbed slack and unabsorbed slack reveal different forms of the relationship with innovation because stakeholders experience more difficulty to monitor absorbed slack than unabsorbed slack. Accordingly, the extensive disclosure of usages of absorbed slack may render such differences in the form of the relationships less salient. One example of such disclosure may include explicitly disclosing the usage of general and administrative expenses because the expenses are generally characterized as highly discretionary. More specifically, it is expected that marginal negative influences of absorbed slack on exploitative innovation, which we hypothesized as constant, gradually diminish to the extent that firms disclose their usages of absorbed slack more extensively. As for the influences of absorbed slack on exploratory innovation, we expect to observe hypothesized diminishing positive influences less explicitly to the extent that firms disclose more extensively. Put differently, negative influences on exploitative innovation are attenuated, while positive influences on exploratory innovation are reinforced. Accordingly, search, and resultant innovation, is expected grow more intensive as more information on the usage of absorbed slack is disclosed. Empirical examination of such influences of increased disclosure on the differences in effects between alternative organizational slack, not only replicates our findings but also opens up one of promising lines of inquiry on the relationship between organizational slack and innovation. Another straightforward avenue to extend our argument would be to apply our findings to the relationship between organizational slack and organizational performance. Furthermore, examining long-term influences of organizational slack on subsequent innovation may be worthwhile. It also is important to note that our results may suffer from the omitted variables bias, as alternative slack variables are separately employed. We had to examine them separately because we felt it more theoretically grounded to model them as endogenous variables with distinct sets of instruments. Although it is difficult to completely deny concerns on omitted variables bias, we employed several measures to alleviate the concerns. First, we employed instrumental variable method, which is recommended when one suspects the existence of omitted variables bias (Bascle, 2008). Second, we also employed fixed-effect models, which effectively remove the influences of firm-specific effects including time invariant portion of organizational slack. Finally, usual caveats of external validity apply as well, i.e. we encourage researchers to replicate our findings in other empirical contexts. For example, in a context where performance-related payment is more widely adopted than our empirical contexts, i.e. Japan, managers are more willing to pursue risk-taking initiatives, thereby rendering differences in effects of absorbed and unabsorbed slack less explicit. Furthermore, as our argument is built on the assumption that innovation is closely associated with organizational search, our findings may not be replicated in a context where innovation is institutionalized. It also may be conceivable that our argument is not applied to firms held by governments or business partners, as these shareholders may not necessarily pursue maximizing financial returns from their investment. It is very difficult to identify organizations with no organizational slack. Furthermore, organizational slack influences various aspects of organizational phenomena, including coalition building (Cyert and March, 1963), political behaviors (Bourgeois and Singh, 1983), risk-taking (Singh, 1986; Bromiley, 1991), competitive responses (Thompson, 1967; Smith et al., 1991), and profitability (Tan and Peng, 2003; George, 2005; Wang et al., 2014). However, our understanding of organizational slack is still limited, fragmented, and confused. We hope our arguments stimulate further discussion on this interesting construct. Acknowledgements The author is grateful for the valuable comments and suggestions of Editor Fredrik Tell, Paul Nightingale, and anonymous reviewers. Errors and omissions remain the author’s own. References Abernathy W. J. ( 1978), The Productivity Dilemma: Roadblock to Innovation in the Automobile Industry . Johns Hopkins University Press: Baltimore, MD. Argyris C. ( 1976), ‘ Single-loop and double-loop models in research on decision making,’ Administrative Science Quarterly , 21( 3), 363– 375. Google Scholar CrossRef Search ADS   Audia P. G., Greve H. R. ( 2006), ‘ Less likely to fail: low performance, firm size, and factory expansion in the shipbuilding industry,’ Management Science , 52( 1), 83– 94. Google Scholar CrossRef Search ADS   Bascle G. ( 2008), ‘ Controlling for endogeneity with instrumental variables in strategic management research,’ Strategic Organization , 6( 3), 285– 327. Google Scholar CrossRef Search ADS   Baysinger B. D., Kosnik R. D., Turk T. A. ( 1991), ‘ Effects of board and ownership structure on corporate R&D strategy,’ Academy of Management Journal , 34( 1), 205– 214. Benner M. J., Tushman M. ( 2002), ‘ Process management and technological innovation: A longitudinal study of the photography and paint industries,’ Administrative Science Quarterly , 47( 4), 676– 706. Google Scholar CrossRef Search ADS   Bierly P., Chakrabarti A. ( 1996), ‘ Generic knowledge strategies in the U.S. pharmaceutical industry,’ Strategic Management Journal , 17( S2), 123– 135. Google Scholar CrossRef Search ADS   Bourgeois L. J.III. ( 1981), ‘ On the measurement of organizational slack,’ Academy of Management Review , 6( 1), 29– 39. Google Scholar CrossRef Search ADS   Bourgeois L. J.III, Singh J. V. ( 1983), ‘ Organizational slack and political behavior among top management teams,’ Academy of Management Proceedings , 1983( 1), 43– 47. Google Scholar CrossRef Search ADS   Bromiley P. ( 1991), ‘ Testing a causal model of corporate risk taking and performance,’ Academy of Management Journal , 34( 1), 37– 59. Bromiley P., Washburn M. ( 2011), ‘ Cost reduction vs innovative search in R&D,’ Journal of Strategy and Management , 4( 3), 196– 214. Google Scholar CrossRef Search ADS   Cameron A. C., Trivedi P. K. ( 2009), Microeconometrics Using Stata , revised ed. Stata Press: College Station. Cardinal L. B. ( 2001), ‘ Technological innovation in the pharmaceutical industry: the use of organizational control in managing research and development,’ Organization Science , 12( 1), 19– 36. Google Scholar CrossRef Search ADS   Carnabuci G., Operti E. ( 2013), ‘ Where do firms' recombinant capabilities come from? Intraorganizational networks, knowledge, and firms' ability to innovate through technological recombination,’ Strategic Management Journal , 34( 13), 1591– 1613. Google Scholar CrossRef Search ADS   Chen W.-R., Miller K. D. ( 2007), ‘ Situational and institutional determinants of firms' R&D search intensity,’ Strategic Management Journal , 28( 4), 369– 381. Google Scholar CrossRef Search ADS   Cheng J. L. C., Kesner I. F. ( 1997), ‘ Organizational slack and response to environmental shifts: The impact of resource allocation patterns,’ Journal of Management , 23( 1), 1– 18. Google Scholar CrossRef Search ADS   Christensen C. M., Bower J. L. ( 1996), ‘ Customer power, strategic investment, and the failure of leading firms,’ Strategic Management Journal , 17( 3), 197– 218. Google Scholar CrossRef Search ADS   Cohen J., Cohen P., West S. G., Aiken L. S. ( 2003), Applied Multiple Regression: Correlation Analysis for the Behavioral Sciences . Lawrence Erlbaum Associates: Mahwah, NJ. Cooper A. C., Smith C. G. ( 1992), ‘ How established firms respond to threatening technologies,’ Academy of Management Executive , 6( 2), 55– 70. Cragg J. G., Donald S. G. ( 1993), ‘ Testing identfiability and specification in instrumental variables models,’ Econometric Theory , 9( 02), 222– 240. Google Scholar CrossRef Search ADS   Cyert R., March J. G. ( 1963), A Behavioral Theory of the Firm . Prentice Hall: Englewood Cliffs, NJ. Danneels E. ( 2008), ‘ Organizational antecedents of second-order competences,’ Strategic Management Journal , 29( 5), 519– 543. Google Scholar CrossRef Search ADS   Deephouse D. L., Wiseman R. M. ( 2000), ‘ Comparing alternative explanations for accounting risk-return relations,’ Journal of Economic Behavior and Organization , 42( 4), 463– 482. Google Scholar CrossRef Search ADS   Dunlap-Hinkler D., Kotabe M., Mudambi R. ( 2010), ‘ A story of breakthrough versus incremental innovation: Corporate entrepreneurship in the global pharmaceutical industry,’ Strategic Entrepreneurship Journal , 4( 2), 106– 127. Google Scholar CrossRef Search ADS   Eisenhardt K. M. ( 1989), ‘ Agency theory: An assessment and review,’ The Academy of Management Review , 14( 1), 57– 74. Google Scholar CrossRef Search ADS   Fama E. F. ( 1980), ‘ Agency problems and the theory of the firm,’ Journal of Political Economy , 88( 2), 288– 307. Google Scholar CrossRef Search ADS   Francis J., Smith A. ( 1995), ‘ Agency costs and innovation some empirical evidence,’ Journal of Accounting and Economics , 19( 2–3), 383– 409. Google Scholar CrossRef Search ADS   Geiger S. W., Cashen L. H. ( 2002), ‘ A multidimensional examination of slack and its impact on innovation,’ Journal of Managerial Issues , 14( 1), 68– 84. George G. ( 2005), ‘ Slack resources and the performance of privately held firms,’ Academy of Management Journal , 48( 4), 661– 676. Google Scholar CrossRef Search ADS   Greve H. R. ( 2003), ‘ A behavioral theory of R&D expenditures and innovations: Evidence from shipbuilding,’ Academy of Management Journal , 46( 6), 685– 702. Greve H. R. ( 2007), ‘ Exploration and exploitation in product innovation,’ Industrial and Corporate Change , 16( 5), 945– 975. Google Scholar CrossRef Search ADS   Hansen L. P. ( 1982), ‘ Large sample properties of generalized method of moments estimators,’ Econometrica , 50( 4), 1029– 1054. Google Scholar CrossRef Search ADS   Hansen L. P., Heaton J., Yaron A. ( 1996), ‘ Finite-sample properties of some alternative GMM estimators,’ Journal of Business and Economic Statistics , 14( 3), 262– 280. Hill C. W. L., Snell S. A. ( 1989), ‘ Effects of ownership structure and control on corporate productivity,’ Academy of Management Journal , 32( 1), 25– 46. Iyer D. N., Miller K. D. ( 2008), ‘ Performance Feedback, Slack, and the Timing of Acquisitions,’ The Academy of Management Journal , 51( 4), 808– 822. Jensen M. C. ( 1986), ‘ Agency costs of free cash flow, corporate finance, and takeovers,’ The American Economic Review , 76( 2), 323– 329. Jensen M. C., Meckling W. H. ( 1976), ‘ Theory of the firm: Managerial behavior, agency costs and ownership structure,’ Journal of Financial Economics , 3( 4), 305– 360. Google Scholar CrossRef Search ADS   Joseph J., Ocasio W. C. ( 2012), ‘ Architecture, attention, and adaptation in the multibusiness firm: General Electric from 1951 to 2001,’ Strategic Management Journal , 33( 6), 633– 660. Google Scholar CrossRef Search ADS   Kacperczyk A., Beckman C. M., Moliterno T. P. ( 2015), ‘ Disentangling risk and change: Internal and external social comparison,’ In the Mutual Fund Industry', Administrative Science Quarterly , 60( 2), 228– 262. Google Scholar CrossRef Search ADS   Katila R., Ahuja G. ( 2002), ‘ Something old, something new: a longitudinal study of search behavior and new product introduction,’ Academy of Management Journal , 45( 6), 1183– 1195. Kim H., Kim H., Lee P. M. ( 2008), ‘ Ownership structure and the relationship between financial slack and R&D investments: evidence from Korean firms,’ Organization Science , 19( 3), 404– 418. Google Scholar CrossRef Search ADS   Knudsen T., Levinthal D. A. ( 2007), ‘ Two faces of search: alternative generation and alternative evaluation,’ Organization Science , 18( 1), 39– 54. Google Scholar CrossRef Search ADS   Lant T. K. ( 1992), ‘ Aspiration level adaptation: an empirical exploration,’ Management Science , 38( 5), 623– 644. Google Scholar CrossRef Search ADS   Latham S. F., Braun M. ( 2009), ‘ Managerial risk, innovation, and organizational decline,’ Journal of Management , 35( 2), 258– 281. Google Scholar CrossRef Search ADS   Laursen K. ( 2012), ‘ Keep searching and you’ll find: what do we know about variety creation through firms’ search activities for innovation?,’ Industrial and Corporate Change , 21( 5), 1181– 1220. Google Scholar CrossRef Search ADS   Laursen K., Salter A. ( 2006), ‘ Open for innovation: the role of openness in explaining innovation performance among U.K. manufacturing firms,’ Strategic Management Journal , 27( 2), 131– 150. Google Scholar CrossRef Search ADS   Lecuona J. R., Reitzig M. ( 2014), ‘ Knowledge worth having in ‘excess’: the value of tacit and firm-specific human resource slack,’ Strategic Management Journal , 35( 7), 954– 973. Google Scholar CrossRef Search ADS   Lee J., Lee J., Lee H. ( 2003), ‘ Exploration and exploitation in the presence of network externalities,’ Management Science , 49( 4), 553– 570. Google Scholar CrossRef Search ADS   Levinthal D. A. ( 1997), ‘ Adaptation on rugged landscapes,’ Management Science , 43( 7), 934– 950. Google Scholar CrossRef Search ADS   Levinthal D. A., March J. G. ( 1981), ‘ A model of adaptive organizational search,’ Journal of Economic Behavior and Organization , 2( 4), 307– 333. Google Scholar CrossRef Search ADS   Levinthal D. A., March J. G. ( 1993), ‘ The myopia of learning,’ Strategic Management Journal , 14( S2), 95– 112. Google Scholar CrossRef Search ADS   Li Q., Maggitti P. G., Smith K. G., Tesluk P. E., Katila R. ( 2013), ‘ Top management attention to innovation: The role of search selection and intensity in new product introductions,’ Academy of Management Journal , 56( 3), 893– 916. Google Scholar CrossRef Search ADS   Litschert R. J., Bonham T. W. ( 1978), ‘ A conceptual model of strategy formation,’ Academy of Management Review , 3( 2), 211– 219. Google Scholar CrossRef Search ADS   Lopez-Vega H., Tell F., Vanhaverbeke W. ( 2016), ‘ Where and how to search? Search paths in open innovation,’ Research Policy , 45( 1), 125– 136. Google Scholar CrossRef Search ADS   Love G. E., Nohria N. ( 2005), ‘ Reducing slack: the performance consequences of downsizing by large industrial firms, 1977–93,’ Strategic Management Journal , 26( 12), 1087– 1108. Google Scholar CrossRef Search ADS   Lungeanu R., Stern I., Zajac E. J. ( 2016), ‘ When do firms change technology-sourcing vehicles? The role of poor innovative performance and financial slack,’ Strategic Management Journal , 37( 5), 855– 869. Google Scholar CrossRef Search ADS   Malatesta P. H., Walkling R. A. ( 1988), ‘ Poison pill securities: Stockholder wealth, profitability, and ownership structure,’ Journal of Financial Economics , 20( 0), 347– 376. Google Scholar CrossRef Search ADS   March J. G. ( 1981), ‘Decisions in organizations and theories of choice’, in Van de Ven A. H., Joyce W. F. (eds), Perspectives on Organization Design and Behavior . Wiley: New York, NY, pp. 205– 248. March J. G. ( 1991), ‘ Exploration and exploitation in organizational learning,’ Organization Science , 2( 1), 71– 87. Google Scholar CrossRef Search ADS   March J. G. ( 1994), A Primer on Decision Making: How Decisions Happen . Free Press: New York, NY. March J. G. ( 2006), ‘ Rationality, Foolishness, and Adaptive Intelligence,’ Strategic Management Journal , 27( 3), 201– 214. Google Scholar CrossRef Search ADS   March J. G. ( 2007), ‘ Perspective—Scholarship, Scholarly Institutions, and Scholarly Communities,’ Organization Science , 18( 3), 537– 542. Google Scholar CrossRef Search ADS   March J. G., Shapira Z. ( 1992), ‘ Variable risk preferences and the focus of attention,’ Psychological Review , 99( 1), 172– 183. Google Scholar CrossRef Search ADS   Meyer A. D. ( 1982), ‘ Adapting to environmental jolts,’ Administrative Science Quarterly , 27( 4), 515– 537. Google Scholar CrossRef Search ADS   Mishina Y., Pollock T. G., Porac J. F. ( 2004), ‘ Are more resources always better for growth? Resource stickiness in market and product expansion,’ Strategic Management Journal , 25( 12), 1179– 1197. Google Scholar CrossRef Search ADS   Murray M. P. ( 2006), ‘ Avoiding invalid instruments and coping with weak instruments,’ Journal of Economic Perspectives , 20( 4), 111– 132. Google Scholar CrossRef Search ADS   Natividad G. ( 2013), ‘ Financial slack, strategy, and competition in movie distribution,’ Organization Science , 24( 3), 846– 864. Google Scholar CrossRef Search ADS   Nohria N., Gulati R. ( 1996), ‘ Is slack good or bad for innovation?,’ Academy of Management Journal , 39( 5), 1245– 1264. Google Scholar CrossRef Search ADS   O'brien J. P. ( 2003), ‘ The capital structure implications of pursuing a strategy of innovation,’ Strategic Management Journal , 24( 5), 415– 431. Google Scholar CrossRef Search ADS   Ocasio W. C. ( 1997), ‘ Towards an attention-based view of the firm,’ Strategic Management Journal , 18( 1), 187– 206. Google Scholar CrossRef Search ADS   Palmer T. B., Wiseman R. M. ( 1999), ‘ Decoupling risk taking from income stream uncertainty: a holistic model of risk,’ Strategic Management Journal , 20( 11), 1037– 1062. Google Scholar CrossRef Search ADS   Piao M. ( 2010), ‘ Thriving in the new: implication of exploration on organizational longevity,’ Journal of Management , 36( 6), 1529– 1554. Google Scholar CrossRef Search ADS   Reuer J. J., Leiblein M. J. ( 2000), ‘ Downside risk implications of multinationality and international joint ventures,’ Academy of Management Journal , 43( 2), 203– 214. Rosenkopf L., Nerkar A. ( 2001), ‘ Beyond local search: boundary-spanning, exploration, and impact in the optical disk industry,’ Strategic Management Journal , 22( 4), 287– 306. Google Scholar CrossRef Search ADS   Sørensen J. B., Stuart T. E. ( 2000), ‘ Aging, obsolescence, and organizational innovation,’ Administrative Science Quarterly , 45( 1), 81– 112. Google Scholar CrossRef Search ADS   Schaffer M. E. ( 2010), ‘xtivreg2: Stata module to perform extended IV/2SLS, GMM and AC/HAC, LIML and k-class regression for panel data models,’ http://ideas.repec.org/c/boc/bocode/s456501.html. Sharfman M. P., Wolf G., Chase R. B., Tansik D. A. ( 1988), ‘ Antecedents of organizational slack,’ Academy of Management Review , 13( 4), 601– 614. Google Scholar CrossRef Search ADS   Simon H. A. ( 1955), ‘ A behavioral model of rational choice,’ The Quarterly Journal of Economics , 69( 1), 99– 118. Google Scholar CrossRef Search ADS   Singh J. V. ( 1986), ‘ Performance, slack, and risk taking in organizational decision making,’ Academy of Management Journal , 29( 3), 562– 585. Smith K. G., Grimm C. M., Gannon M. J., Chen M.-J. ( 1991), ‘ Organizational information processing, competitive responses, and performance in the U.S. domestic airline industry,’ Academy of Management Journal , 34( 1), 60– 85. Souder D., Shaver J. M. ( 2010), ‘ Constraints and incentives for making long horizon corporate investments,’ Strategic Management Journal , 31( 12), 1316– 1336. Google Scholar CrossRef Search ADS   Steensma H. K., Corley K. G. ( 2001), ‘ Organizational context as a moderator of theories on firm boundaries for technology sourcing,’ Academy of Management Journal , 44( 2), 271– 291. Stuart T. E., Podolny J. M. ( 1996), ‘ Local search and the evolution of technological capabilities,’ Strategic Management Journal , 17( S1), 21– 38. Google Scholar CrossRef Search ADS   Tan J., Peng M. W. ( 2003), ‘ Organizational slack and firm performance during economic transitions: Two studies from an emerging economy,’ Strategic Management Journal , 24( 13), 1249– 1263. Google Scholar CrossRef Search ADS   Thompson J. D. ( 1967), Organizations in Action: Social Science Bases of Administrative Theory . McGraw-Hill Inc.: New York, NY. Voss G. B., Sirdeshmukh D., Voss Z. G. ( 2008), ‘ The effects of slack resources and environmental threat on product exploration and exploitation,’ Academy of Management Journal , 51( 1), 147– 164. Google Scholar CrossRef Search ADS   Walkling R. A., Long M. S. ( 1984), ‘ Agency theory, managerial welfare, and takeover bid resistance,’ The RAND Journal of Economics , 15( 1), 54– 68. Google Scholar CrossRef Search ADS   Wang D., Sun D., Yu X., Zhang Y. ( 2014), ‘ The impact of CEO duality and ownership on the relationship between organisational slack and firm performance in China,’ Systems Research and Behavioral Science , 31( 1), 94– 101. Google Scholar CrossRef Search ADS   Wang H., Li J. ( 2008), ‘ Untangling the effects of overexploration and overexploitation on organizational performance: The moderating role of environmental dynamism,’ Journal of Management , 34( 5), 925– 951. Google Scholar CrossRef Search ADS   Watzlawick P., Weakland J. H., Fisch R. ( 1974), Change: Principles of Problem Formation and Problem Resolution . W.W. Norton: New York, NY. Wiseman R. M., Bromiley P. ( 1996), ‘ Toward a model of risk in declining organizations: an empirical examination of risk, performance and decline,’ Organization Science , 7( 5), 524– 543. Google Scholar CrossRef Search ADS   Wiseman R. M., Catanach C. ( 1997), ‘ A longitudinal disaggregation of operational risk under changing regulations: evidence from the savings and loan industry,’ Academy of Management Journal , 40( 4), 799– 830. Wooldridge J. M. ( 2010), Econometric Analysis of Cross Section and Panel Data , 2nd edn. MIT Press: Cambridge, MA. Yasai-Ardekani M. ( 1986), ‘ Structural adaptations to environments,’ Academy of Management Review , 11( 1), 9– 21. Google Scholar CrossRef Search ADS   Appendix Table A1. Results of the panel continuous-updating estimator (CUE) instrumental variable regression analysis for the effects of absorbed slack on innovation increasesa Dependent variables  Absorbed slack   Exploitative innovation   Absorbed slack   Exploratory innovation   Model 15  Model 16  Model 17  Model 18  Model 19  Model 20  Model 21  R&D intensity  0.33**  (0.10)          0.33**  (0.10)              Total assets  0.05*  (0.02)          0.05*  (0.02)              Organizational size  0.02  (0.04)  0.01  (0.06)  0.01  (0.06)  0.02  (0.03)  −0.03  (0.06)  −0.03  (0.05)  −0.12  (0.08)  R&D  0.00  (0.00)  0.06†  (0.04)  0.17*  (0.07)  0.00  (0.01)  0.11**  (0.03)  0.09***  (0.02)  0.06†  (0.03)  Age  −0.06*  (0.03)  −0.05  (0.06)  −0.12  (0.09)  −0.07*  (0.03)  −0.07  (0.13)  0.02  (0.09)  0.12  (0.12)  M&As  0.01  (0.02)  −0.13  (0.13)  −0.09  (0.16)  0.02  (0.02)  0.01  (0.13)  −0.02  (0.15)  −0.01  (0.15)  Attainment discrepancy  0.18  (0.19)  −1.39*  (0.55)  −1.12**  (0.43)  0.18  (0.19)  −0.05  (0.48)  −0.47  (0.45)  −0.68  (0.61)  Size of pipelines  0.00  (0.00)  0.01  (0.01)  0.00  (0.01)  0.00  (0.00)  −0.02  (0.01)  −0.02  (0.01)  −0.03  (0.02)  Price revisions  0.01  (0.03)  −0.20  (0.26)  −0.20  (0.40)  0.01  (0.03)  0.09  (0.42)  −0.10  (0.47)  0.04  (0.49)  Asset growth  0.01  (0.05)  0.23*  (0.09)  0.14†  (0.08)  0.02  (0.05)  0.03  (0.07)  0.02  (0.08)  0.07  (0.12)  Potential slack  0.05  (0.04)  0.07  (0.09)  0.14†  (0.08)  0.05  (0.04)  0.02  (0.08)  0.12  (0.07)  −0.15  (0.16)  Non-NCE pipelines  0.00  (0.00)  −0.02***  (0.01)  −0.03***  (0.01)                  NCE pipelines              0.00  (0.00)  −0.02***  (0.01)  −0.02***  (0.00)  −0.02***  (0.00)  Exploratory innovation  0.00  (0.00)  0.27***  (0.06)  0.28*  (0.12)                  Absorbed slack          −1.28*  (0.52)          −0.15  (0.37)  3.92*  (1.78)  Absorbed slack, squared                          −5.28*  (2.30)  Constant  −0.31  (0.27)  −0.21  (0.53)      −0.29  (0.26)  0.49  (0.45)          N firm-years  581    581    581    581    581    581    581    N Firms  37    37    37    37    37    37    37    RMSE  0.04    0.39    0.40    0.04    0.39    0.40    0.43    Log likelihood  1000    −270    −276    1000    −277    −279    −312    F  34.5     4.46     5.52     37.2     3.19     3.73     3.11     Dependent variables  Absorbed slack   Exploitative innovation   Absorbed slack   Exploratory innovation   Model 15  Model 16  Model 17  Model 18  Model 19  Model 20  Model 21  R&D intensity  0.33**  (0.10)          0.33**  (0.10)              Total assets  0.05*  (0.02)          0.05*  (0.02)              Organizational size  0.02  (0.04)  0.01  (0.06)  0.01  (0.06)  0.02  (0.03)  −0.03  (0.06)  −0.03  (0.05)  −0.12  (0.08)  R&D  0.00  (0.00)  0.06†  (0.04)  0.17*  (0.07)  0.00  (0.01)  0.11**  (0.03)  0.09***  (0.02)  0.06†  (0.03)  Age  −0.06*  (0.03)  −0.05  (0.06)  −0.12  (0.09)  −0.07*  (0.03)  −0.07  (0.13)  0.02  (0.09)  0.12  (0.12)  M&As  0.01  (0.02)  −0.13  (0.13)  −0.09  (0.16)  0.02  (0.02)  0.01  (0.13)  −0.02  (0.15)  −0.01  (0.15)  Attainment discrepancy  0.18  (0.19)  −1.39*  (0.55)  −1.12**  (0.43)  0.18  (0.19)  −0.05  (0.48)  −0.47  (0.45)  −0.68  (0.61)  Size of pipelines  0.00  (0.00)  0.01  (0.01)  0.00  (0.01)  0.00  (0.00)  −0.02  (0.01)  −0.02  (0.01)  −0.03  (0.02)  Price revisions  0.01  (0.03)  −0.20  (0.26)  −0.20  (0.40)  0.01  (0.03)  0.09  (0.42)  −0.10  (0.47)  0.04  (0.49)  Asset growth  0.01  (0.05)  0.23*  (0.09)  0.14†  (0.08)  0.02  (0.05)  0.03  (0.07)  0.02  (0.08)  0.07  (0.12)  Potential slack  0.05  (0.04)  0.07  (0.09)  0.14†  (0.08)  0.05  (0.04)  0.02  (0.08)  0.12  (0.07)  −0.15  (0.16)  Non-NCE pipelines  0.00  (0.00)  −0.02***  (0.01)  −0.03***  (0.01)                  NCE pipelines              0.00  (0.00)  −0.02***  (0.01)  −0.02***  (0.00)  −0.02***  (0.00)  Exploratory innovation  0.00  (0.00)  0.27***  (0.06)  0.28*  (0.12)                  Absorbed slack          −1.28*  (0.52)          −0.15  (0.37)  3.92*  (1.78)  Absorbed slack, squared                          −5.28*  (2.30)  Constant  −0.31  (0.27)  −0.21  (0.53)      −0.29  (0.26)  0.49  (0.45)          N firm-years  581    581    581    581    581    581    581    N Firms  37    37    37    37    37    37    37    RMSE  0.04    0.39    0.40    0.04    0.39    0.40    0.43    Log likelihood  1000    −270    −276    1000    −277    −279    −312    F  34.5     4.46     5.52     37.2     3.19     3.73     3.11     a Robust standard errors adjusted for clustering by firm are in parentheses. Two-tailed tests for all effects. † p <0.1; *p <0.05; **p <0.01; ***p <0.001. Table A1. Results of the panel continuous-updating estimator (CUE) instrumental variable regression analysis for the effects of absorbed slack on innovation increasesa Dependent variables  Absorbed slack   Exploitative innovation   Absorbed slack   Exploratory innovation   Model 15  Model 16  Model 17  Model 18  Model 19  Model 20  Model 21  R&D intensity  0.33**  (0.10)          0.33**  (0.10)              Total assets  0.05*  (0.02)          0.05*  (0.02)              Organizational size  0.02  (0.04)  0.01  (0.06)  0.01  (0.06)  0.02  (0.03)  −0.03  (0.06)  −0.03  (0.05)  −0.12  (0.08)  R&D  0.00  (0.00)  0.06†  (0.04)  0.17*  (0.07)  0.00  (0.01)  0.11**  (0.03)  0.09***  (0.02)  0.06†  (0.03)  Age  −0.06*  (0.03)  −0.05  (0.06)  −0.12  (0.09)  −0.07*  (0.03)  −0.07  (0.13)  0.02  (0.09)  0.12  (0.12)  M&As  0.01  (0.02)  −0.13  (0.13)  −0.09  (0.16)  0.02  (0.02)  0.01  (0.13)  −0.02  (0.15)  −0.01  (0.15)  Attainment discrepancy  0.18  (0.19)  −1.39*  (0.55)  −1.12**  (0.43)  0.18  (0.19)  −0.05  (0.48)  −0.47  (0.45)  −0.68  (0.61)  Size of pipelines  0.00  (0.00)  0.01  (0.01)  0.00  (0.01)  0.00  (0.00)  −0.02  (0.01)  −0.02  (0.01)  −0.03  (0.02)  Price revisions  0.01  (0.03)  −0.20  (0.26)  −0.20  (0.40)  0.01  (0.03)  0.09  (0.42)  −0.10  (0.47)  0.04  (0.49)  Asset growth  0.01  (0.05)  0.23*  (0.09)  0.14†  (0.08)  0.02  (0.05)  0.03  (0.07)  0.02  (0.08)  0.07  (0.12)  Potential slack  0.05  (0.04)  0.07  (0.09)  0.14†  (0.08)  0.05  (0.04)  0.02  (0.08)  0.12  (0.07)  −0.15  (0.16)  Non-NCE pipelines  0.00  (0.00)  −0.02***  (0.01)  −0.03***  (0.01)                  NCE pipelines              0.00  (0.00)  −0.02***  (0.01)  −0.02***  (0.00)  −0.02***  (0.00)  Exploratory innovation  0.00  (0.00)  0.27***  (0.06)  0.28*  (0.12)                  Absorbed slack          −1.28*  (0.52)          −0.15  (0.37)  3.92*  (1.78)  Absorbed slack, squared                          −5.28*  (2.30)  Constant  −0.31  (0.27)  −0.21  (0.53)      −0.29  (0.26)  0.49  (0.45)          N firm-years  581    581    581    581    581    581    581    N Firms  37    37    37    37    37    37    37    RMSE  0.04    0.39    0.40    0.04    0.39    0.40    0.43    Log likelihood  1000    −270    −276    1000    −277    −279    −312    F  34.5     4.46     5.52     37.2     3.19     3.73     3.11     Dependent variables  Absorbed slack   Exploitative innovation   Absorbed slack   Exploratory innovation   Model 15  Model 16  Model 17  Model 18  Model 19  Model 20  Model 21  R&D intensity  0.33**  (0.10)          0.33**  (0.10)              Total assets  0.05*  (0.02)          0.05*  (0.02)              Organizational size  0.02  (0.04)  0.01  (0.06)  0.01  (0.06)  0.02  (0.03)  −0.03  (0.06)  −0.03  (0.05)  −0.12  (0.08)  R&D  0.00  (0.00)  0.06†  (0.04)  0.17*  (0.07)  0.00  (0.01)  0.11**  (0.03)  0.09***  (0.02)  0.06†  (0.03)  Age  −0.06*  (0.03)  −0.05  (0.06)  −0.12  (0.09)  −0.07*  (0.03)  −0.07  (0.13)  0.02  (0.09)  0.12  (0.12)  M&As  0.01  (0.02)  −0.13  (0.13)  −0.09  (0.16)  0.02  (0.02)  0.01  (0.13)  −0.02  (0.15)  −0.01  (0.15)  Attainment discrepancy  0.18  (0.19)  −1.39*  (0.55)  −1.12**  (0.43)  0.18  (0.19)  −0.05  (0.48)  −0.47  (0.45)  −0.68  (0.61)  Size of pipelines  0.00  (0.00)  0.01  (0.01)  0.00  (0.01)  0.00  (0.00)  −0.02  (0.01)  −0.02  (0.01)  −0.03  (0.02)  Price revisions  0.01  (0.03)  −0.20  (0.26)  −0.20  (0.40)  0.01  (0.03)  0.09  (0.42)  −0.10  (0.47)  0.04  (0.49)  Asset growth  0.01  (0.05)  0.23*  (0.09)  0.14†  (0.08)  0.02  (0.05)  0.03  (0.07)  0.02  (0.08)  0.07  (0.12)  Potential slack  0.05  (0.04)  0.07  (0.09)  0.14†  (0.08)  0.05  (0.04)  0.02  (0.08)  0.12  (0.07)  −0.15  (0.16)  Non-NCE pipelines  0.00  (0.00)  −0.02***  (0.01)  −0.03***  (0.01)                  NCE pipelines              0.00  (0.00)  −0.02***  (0.01)  −0.02***  (0.00)  −0.02***  (0.00)  Exploratory innovation  0.00  (0.00)  0.27***  (0.06)  0.28*  (0.12)                  Absorbed slack          −1.28*  (0.52)          −0.15  (0.37)  3.92*  (1.78)  Absorbed slack, squared                          −5.28*  (2.30)  Constant  −0.31  (0.27)  −0.21  (0.53)      −0.29  (0.26)  0.49  (0.45)          N firm-years  581    581    581    581    581    581    581    N Firms  37    37    37    37    37    37    37    RMSE  0.04    0.39    0.40    0.04    0.39    0.40    0.43    Log likelihood  1000    −270    −276    1000    −277    −279    −312    F  34.5     4.46     5.52     37.2     3.19     3.73     3.11     a Robust standard errors adjusted for clustering by firm are in parentheses. Two-tailed tests for all effects. † p <0.1; *p <0.05; **p <0.01; ***p <0.001. Table A2. Results of the panel continuous-updating estimator (CUE) instrumental variable regression analysis for the effects of unabsorbed slack on innovation increasesa Dependent variables  Unabsorbed slack   Exploitative innovation   Unabsorbed slack   Exploratory innovation   Model 22  Model 23  Model 24  Model 25  Model 26  Model 27  Model 28  Model 29  Environmental dynamics  2.25  (1.34)                  2.06  (1.36)          Revenue growth  −1.72†  (0.92)                  −1.73†  (0.91)          Organizational size  −0.54*  (0.22)  0.01  (0.06)  −0.07  (0.08)  −0.35  (0.29)  −0.15  (0.11)  −0.53*  (0.23)  −0.03  (0.06)  0.34  (0.21)  R&D  0.01  (0.07)  0.06†  (0.04)  0.06†  (0.03)  0.06  (0.05)  0.05  (0.03)  −0.05  (0.08)  0.11**  (0.03)  0.14*  (0.06)  Age  0.61  (0.40)  −0.05  (0.06)  0.14  (0.13)  0.62  (0.45)  0.21  (0.16)  0.72†  (0.42)  −0.07  (0.13)  −1.00*  (0.46)  M&As  −0.30  (0.26)  −0.13  (0.13)  −0.17  (0.16)  −0.21  (0.24)  −0.18  (0.17)  −0.43  (0.29)  0.01  (0.13)  0.24  (0.20)  Attainment discrepancy  0.50  (1.82)  −1.39*  (0.55)  −1.18**  (0.42)  −1.49†  (0.82)  −1.22**  (0.47)  0.54  (1.84)  −0.05  (0.48)  −0.48  (0.96)  Size of pipelines  −0.08†  (0.04)  0.01  (0.01)  0.00  (0.01)  −0.01  (0.02)  0.00  (0.01)  −0.06  (0.04)  −0.02  (0.01)  0.01  (0.02)  Price revisions  0.28  (0.64)  −0.20  (0.26)  −0.20  (0.41)  −0.13  (0.50)  −0.18  (0.42)  0.31  (0.65)  0.09  (0.42)  −0.12  (0.63)  Asset growth  0.50*  (0.24)  0.23*  (0.09)  0.22*  (0.09)  0.26†  (0.14)  0.23*  (0.09)  0.47*  (0.23)  0.03  (0.07)  0.36  (0.32)  Potential slack  −0.72†  (0.40)  0.07  (0.09)  −0.01  (0.10)  −0.32  (0.37)  −0.12  (0.14)  −0.76†  (0.42)  0.02  (0.08)  0.46†  (0.27)  Non-NCE pipelines  0.00  (0.02)  −0.02***  (0.01)  −0.02***  (0.01)  −0.02*  (0.01)  −0.02***  (0.01)              NCE pipelines                      0.01  (0.01)  −0.02***  (0.01)  −0.03***  (0.01)  Exploratory innovation  −0.08  (0.10)  0.27***  (0.06)  0.27*  (0.12)  0.29*  (0.12)  0.27*  (0.12)              Unabsorbed slack          −0.12†  (0.07)  −0.56  (0.41)              0.62*  (0.29)  Unabsorbed slack, squared              0.05  (0.06)                  Unabsorbed slack, natural logarithm                  −0.59*  (0.30)              Constant  3.75†  (1.89)  −0.21  (0.53)              3.60†  (2.04)  0.49  (0.45)      N firm-years  581    581    581    581    581    581    581    581    N Firms  37    37    37    37    37    37    37    37    RMSE  0.72    0.39    0.41    0.52    0.42    0.72    0.39    0.61    Log likelihood  −628    −270    −286    −427    −302    −627    −277    −518    F  4.04     4.46     4.45     2.72     3.80     4.20     3.19     1.42     Dependent variables  Unabsorbed slack   Exploitative innovation   Unabsorbed slack   Exploratory innovation   Model 22  Model 23  Model 24  Model 25  Model 26  Model 27  Model 28  Model 29  Environmental dynamics  2.25  (1.34)                  2.06  (1.36)          Revenue growth  −1.72†  (0.92)                  −1.73†  (0.91)          Organizational size  −0.54*  (0.22)  0.01  (0.06)  −0.07  (0.08)  −0.35  (0.29)  −0.15  (0.11)  −0.53*  (0.23)  −0.03  (0.06)  0.34  (0.21)  R&D  0.01  (0.07)  0.06†  (0.04)  0.06†  (0.03)  0.06  (0.05)  0.05  (0.03)  −0.05  (0.08)  0.11**  (0.03)  0.14*  (0.06)  Age  0.61  (0.40)  −0.05  (0.06)  0.14  (0.13)  0.62  (0.45)  0.21  (0.16)  0.72†  (0.42)  −0.07  (0.13)  −1.00*  (0.46)  M&As  −0.30  (0.26)  −0.13  (0.13)  −0.17  (0.16)  −0.21  (0.24)  −0.18  (0.17)  −0.43  (0.29)  0.01  (0.13)  0.24  (0.20)  Attainment discrepancy  0.50  (1.82)  −1.39*  (0.55)  −1.18**  (0.42)  −1.49†  (0.82)  −1.22**  (0.47)  0.54  (1.84)  −0.05  (0.48)  −0.48  (0.96)  Size of pipelines  −0.08†  (0.04)  0.01  (0.01)  0.00  (0.01)  −0.01  (0.02)  0.00  (0.01)  −0.06  (0.04)  −0.02  (0.01)  0.01  (0.02)  Price revisions  0.28  (0.64)  −0.20  (0.26)  −0.20  (0.41)  −0.13  (0.50)  −0.18  (0.42)  0.31  (0.65)  0.09  (0.42)  −0.12  (0.63)  Asset growth  0.50*  (0.24)  0.23*  (0.09)  0.22*  (0.09)  0.26†  (0.14)  0.23*  (0.09)  0.47*  (0.23)  0.03  (0.07)  0.36  (0.32)  Potential slack  −0.72†  (0.40)  0.07  (0.09)  −0.01  (0.10)  −0.32  (0.37)  −0.12  (0.14)  −0.76†  (0.42)  0.02  (0.08)  0.46†  (0.27)  Non-NCE pipelines  0.00  (0.02)  −0.02***  (0.01)  −0.02***  (0.01)  −0.02*  (0.01)  −0.02***  (0.01)              NCE pipelines                      0.01  (0.01)  −0.02***  (0.01)  −0.03***  (0.01)  Exploratory innovation  −0.08  (0.10)  0.27***  (0.06)  0.27*  (0.12)  0.29*  (0.12)  0.27*  (0.12)              Unabsorbed slack          −0.12†  (0.07)  −0.56  (0.41)              0.62*  (0.29)  Unabsorbed slack, squared              0.05  (0.06)                  Unabsorbed slack, natural logarithm                  −0.59*  (0.30)              Constant  3.75†  (1.89)  −0.21  (0.53)              3.60†  (2.04)  0.49  (0.45)      N firm-years  581    581    581    581    581    581    581    581    N Firms  37    37    37    37    37    37    37    37    RMSE  0.72    0.39    0.41    0.52    0.42    0.72    0.39    0.61    Log likelihood  −628    −270    −286    −427    −302    −627    −277    −518    F  4.04     4.46     4.45     2.72     3.80     4.20     3.19     1.42     a Robust standard errors adjusted for clustering by firm are in parentheses. Two-tailed tests for all effects. † p <0.1; *p <0.05; **p <0.01; ***p <0.001. Table A2. Results of the panel continuous-updating estimator (CUE) instrumental variable regression analysis for the effects of unabsorbed slack on innovation increasesa Dependent variables  Unabsorbed slack   Exploitative innovation   Unabsorbed slack   Exploratory innovation   Model 22  Model 23  Model 24  Model 25  Model 26  Model 27  Model 28  Model 29  Environmental dynamics  2.25  (1.34)                  2.06  (1.36)          Revenue growth  −1.72†  (0.92)                  −1.73†  (0.91)          Organizational size  −0.54*  (0.22)  0.01  (0.06)  −0.07  (0.08)  −0.35  (0.29)  −0.15  (0.11)  −0.53*  (0.23)  −0.03  (0.06)  0.34  (0.21)  R&D  0.01  (0.07)  0.06†  (0.04)  0.06†  (0.03)  0.06  (0.05)  0.05  (0.03)  −0.05  (0.08)  0.11**  (0.03)  0.14*  (0.06)  Age  0.61  (0.40)  −0.05  (0.06)  0.14  (0.13)  0.62  (0.45)  0.21  (0.16)  0.72†  (0.42)  −0.07  (0.13)  −1.00*  (0.46)  M&As  −0.30  (0.26)  −0.13  (0.13)  −0.17  (0.16)  −0.21  (0.24)  −0.18  (0.17)  −0.43  (0.29)  0.01  (0.13)  0.24  (0.20)  Attainment discrepancy  0.50  (1.82)  −1.39*  (0.55)  −1.18**  (0.42)  −1.49†  (0.82)  −1.22**  (0.47)  0.54  (1.84)  −0.05  (0.48)  −0.48  (0.96)  Size of pipelines  −0.08†  (0.04)  0.01  (0.01)  0.00  (0.01)  −0.01  (0.02)  0.00  (0.01)  −0.06  (0.04)  −0.02  (0.01)  0.01  (0.02)  Price revisions  0.28  (0.64)  −0.20  (0.26)  −0.20  (0.41)  −0.13  (0.50)  −0.18  (0.42)  0.31  (0.65)  0.09  (0.42)  −0.12  (0.63)  Asset growth  0.50*  (0.24)  0.23*  (0.09)  0.22*  (0.09)  0.26†  (0.14)  0.23*  (0.09)  0.47*  (0.23)  0.03  (0.07)  0.36  (0.32)  Potential slack  −0.72†  (0.40)  0.07  (0.09)  −0.01  (0.10)  −0.32  (0.37)  −0.12  (0.14)  −0.76†  (0.42)  0.02  (0.08)  0.46†  (0.27)  Non-NCE pipelines  0.00  (0.02)  −0.02***  (0.01)  −0.02***  (0.01)  −0.02*  (0.01)  −0.02***  (0.01)              NCE pipelines                      0.01  (0.01)  −0.02***  (0.01)  −0.03***  (0.01)  Exploratory innovation  −0.08  (0.10)  0.27***  (0.06)  0.27*  (0.12)  0.29*  (0.12)  0.27*  (0.12)              Unabsorbed slack          −0.12†  (0.07)  −0.56  (0.41)              0.62*  (0.29)  Unabsorbed slack, squared              0.05  (0.06)                  Unabsorbed slack, natural logarithm                  −0.59*  (0.30)              Constant  3.75†  (1.89)  −0.21  (0.53)              3.60†  (2.04)  0.49  (0.45)      N firm-years  581    581    581    581    581    581    581    581    N Firms  37    37    37    37    37    37    37    37    RMSE  0.72    0.39    0.41    0.52    0.42    0.72    0.39    0.61    Log likelihood  −628    −270    −286    −427    −302    −627    −277    −518    F  4.04     4.46     4.45     2.72     3.80     4.20     3.19     1.42     Dependent variables  Unabsorbed slack   Exploitative innovation   Unabsorbed slack   Exploratory innovation   Model 22  Model 23  Model 24  Model 25  Model 26  Model 27  Model 28  Model 29  Environmental dynamics  2.25  (1.34)                  2.06  (1.36)          Revenue growth  −1.72†  (0.92)                  −1.73†  (0.91)          Organizational size  −0.54*  (0.22)  0.01  (0.06)  −0.07  (0.08)  −0.35  (0.29)  −0.15  (0.11)  −0.53*  (0.23)  −0.03  (0.06)  0.34  (0.21)  R&D  0.01  (0.07)  0.06†  (0.04)  0.06†  (0.03)  0.06  (0.05)  0.05  (0.03)  −0.05  (0.08)  0.11**  (0.03)  0.14*  (0.06)  Age  0.61  (0.40)  −0.05  (0.06)  0.14  (0.13)  0.62  (0.45)  0.21  (0.16)  0.72†  (0.42)  −0.07  (0.13)  −1.00*  (0.46)  M&As  −0.30  (0.26)  −0.13  (0.13)  −0.17  (0.16)  −0.21  (0.24)  −0.18  (0.17)  −0.43  (0.29)  0.01  (0.13)  0.24  (0.20)  Attainment discrepancy  0.50  (1.82)  −1.39*  (0.55)  −1.18**  (0.42)  −1.49†  (0.82)  −1.22**  (0.47)  0.54  (1.84)  −0.05  (0.48)  −0.48  (0.96)  Size of pipelines  −0.08†  (0.04)  0.01  (0.01)  0.00  (0.01)  −0.01  (0.02)  0.00  (0.01)  −0.06  (0.04)  −0.02  (0.01)  0.01  (0.02)  Price revisions  0.28  (0.64)  −0.20  (0.26)  −0.20  (0.41)  −0.13  (0.50)  −0.18  (0.42)  0.31  (0.65)  0.09  (0.42)  −0.12  (0.63)  Asset growth  0.50*  (0.24)  0.23*  (0.09)  0.22*  (0.09)  0.26†  (0.14)  0.23*  (0.09)  0.47*  (0.23)  0.03  (0.07)  0.36  (0.32)  Potential slack  −0.72†  (0.40)  0.07  (0.09)  −0.01  (0.10)  −0.32  (0.37)  −0.12  (0.14)  −0.76†  (0.42)  0.02  (0.08)  0.46†  (0.27)  Non-NCE pipelines  0.00  (0.02)  −0.02***  (0.01)  −0.02***  (0.01)  −0.02*  (0.01)  −0.02***  (0.01)              NCE pipelines                      0.01  (0.01)  −0.02***  (0.01)  −0.03***  (0.01)  Exploratory innovation  −0.08  (0.10)  0.27***  (0.06)  0.27*  (0.12)  0.29*  (0.12)  0.27*  (0.12)              Unabsorbed slack          −0.12†  (0.07)  −0.56  (0.41)              0.62*  (0.29)  Unabsorbed slack, squared              0.05  (0.06)                  Unabsorbed slack, natural logarithm                  −0.59*  (0.30)              Constant  3.75†  (1.89)  −0.21  (0.53)              3.60†  (2.04)  0.49  (0.45)      N firm-years  581    581    581    581    581    581    581    581    N Firms  37    37    37    37    37    37    37    37    RMSE  0.72    0.39    0.41    0.52    0.42    0.72    0.39    0.61    Log likelihood  −628    −270    −286    −427    −302    −627    −277    −518    F  4.04     4.46     4.45     2.72     3.80     4.20     3.19     1.42     a Robust standard errors adjusted for clustering by firm are in parentheses. Two-tailed tests for all effects. † p <0.1; *p <0.05; **p <0.01; ***p <0.001. © The Author(s) 2018. Published by Oxford University Press on behalf of Associazione ICC. All rights reserved. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Industrial and Corporate Change Oxford University Press

Enabling or constraining? Unraveling the influence of organizational slack on innovation

Loading next page...
 
/lp/ou_press/enabling-or-constraining-unraveling-the-influence-of-organizational-EcMsqaliTj
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of Associazione ICC. All rights reserved.
ISSN
0960-6491
eISSN
1464-3650
D.O.I.
10.1093/icc/dtx046
Publisher site
See Article on Publisher Site

Abstract

Abstract We employ theories of organizational search and agency costs to propose a contingency perspective that reconciles mutually contradictory prior findings on the relationship between organizational slack and innovation. First, we argue that influences of organizational slack depend on whether we consider exploitative innovation or exploratory innovation. Further, absorbed slack and unabsorbed slack differ in their forms of relationship with innovation. The ways in which a certain type of innovation is enabled by organizational slack are conditioned by distinct modes of organizational search associated with alternative types of innovation, as well as by the extent to which effective shareholder monitoring is disturbed by different types of organizational slack. An empirical analysis of 37 Japanese pharmaceutical firms’ new product developments over a 20-year period supports our argument. One of the most critical backbones of the organizations literature is that managers’ rationality is bounded (Simon, 1955). Given that managers are not able to compare all alternatives by precisely forecasting their consequences, it is important to maintain certain amount of flexibility to accommodate future adjustments. A typical source of organizational flexibility is organizational slack (Cyert and March, 1963; Bourgeois, 1981), or a class of organizational resource whose usage is left unspecified. However, the empirical results of works examining whether organizational slack benefits organizations are mixed. Particularly, findings from prior work concerning influences of organizational slack on innovation initiatives (or more generally, on risk-taking) are inconclusive at best. Some argue for a positive relationship (Meyer, 1982; Singh, 1986; Greve, 2007), while others argue for a negative relationship (Bromiley, 1991; Wiseman and Bromiley, 1996; Palmer and Wiseman, 1999; Latham and Braun, 2009). Further, some scholars argue for a curvilinear relationship in an inverted U-shape (Nohria and Gulati, 1996; Geiger and Cashen, 2002) as well as in a U-shape (Wiseman and Catanach, 1997), while nonfinding is also reported (Greve, 2003). Upon the examination of differential effects of organizational slack on exploratory innovation and exploitative innovation, Greve (2007) calls for more work with the larger sample to obtain more precise inferences. Some authors also examine the relationship between organizational slack and an important precursor of innovation initiatives, or risk-taking, by theorizing shifts in the locus of attention of decision makers (March and Shapira, 1992; Audia and Greve, 2006; Chen and Miller, 2007; Iyer and Miller, 2008; Lungeanu et al., 2016). Although they advanced the literature by uncovering the mechanism in which organizational slack influences risk-taking, the authors generally assume positive influences of organizational slack, thereby downplaying the possibility that organizational slack may disturb or discourage risk-taking initiatives under certain conditions. Furthermore, as central interests of the authors lie in the mechanism of performance feedback, organizational slack is theorized to moderate the mechanism, rather than to directly influence risk-taking. In this manuscript, we propose a contingency perspective to reconcile these inconclusive research findings—we argue that organizational slack differentially influences exploitative innovation and exploratory innovation (March, 1991). Extending prior work on alternative search modes enables us to theorize that organizational slack exerts negative influences on exploitative innovation, while positive influences on exploratory innovation. By building upon the theory of agency costs (Jensen and Meckling, 1976; Fama, 1980), we also argue that absorbed slack and unabsorbed slack (Singh, 1986) differ in their forms of relationship with innovation, as the former is more difficult to monitor than the latter. Specifically, we hypothesize that a negative association is observed between absorbed slack and exploitative innovation in a linear form, while the association between unabsorbed slack and exploitative innovation is curvilinear, in that the slope reveals diminishing negative effects of unabsorbed slack. As for exploratory innovation, we hypothesize linear positive influences of unabsorbed slack, whereas diminishing positive effects of absorbed slack. Our empirical analysis of the unique data on new pharmaceutical development by the Japanese firms supports our argument. In short, we attempt to advance our understanding of organizational slack by uncovering boundary conditions on whether we observe a positive, a negative, or a curvilinear relationship. We develop our arguments by showing that the agency theory complements the behavioral theory of the firm (Cyert and March, 1963), BTOF hereafter, in a previously unreported manner. Our findings imply that we could also bridge theories of slack and managerial attention (Ocasio, 1997; Joseph and Ocasio, 2012; Li et al., 2013) by indicating that organizational slack regulates attention of decision makers. 1. Theory and hypotheses 1.1 Types of innovation and organizational search Our first contingency refers to the differences between alternative types of innovation influenced by organizational slack. We particularly consider the possibility that organizational slack’s influences may not be identical between exploitative innovation and exploratory innovation (March, 1991; Voss et al., 2008). We define exploitative innovation as innovation initiatives targeted at improving or modifying existing knowledge utilized for current business, while innovation initiatives targeted at identifying or generating new knowledge that is beyond the scope of current business are defined as exploratory innovation (March, 1991; Sørensen and Stuart, 2000; Rosenkopf and Nerkar, 2001; Benner and Tushman, 2002; Katila and Ahuja, 2002; Lee et al., 2003; Piao, 2010). Accordingly, we refer to innovation as initiatives or resource commitments targeted to innovate new products, processes, or businesses, rather than as outcomes achieved by those initiatives. We particularly focus on the different search modes associated with exploitative and exploratory innovation. Exploitative innovation is enabled by problemistic search, or “search that is stimulated by a problem (usually a rather specific one) and is directed toward finding a solution to that problem” (Cyert and March, 1963: 121). Although one may be tempted to associate problemistic search with high risk-taking choices in that departures from the status quo entail some risk, problemistic search leads organizations to select less, rather than more, risky alternatives, e.g. cost cuts, for more reliable and predictable resolution of the problem (Deephouse and Wiseman, 2000; Iyer and Miller, 2008; Bromiley and Washburn, 2011; Kacperczyk et al., 2015). Accordingly, problemistic search “emphasizes relatively immediate refinements in the existing technology, greater efficiency, and discoveries in the near neighborhood of the present activities” (Levinthal and March, 1981: 309). Put differently, problemistic search is the first-order search (Argyris, 1976; Watzlawick et al., 1974), in that it is a reaction in terms of “existing rules” (Levinthal and March, 1981: 309). Therefore, problemistic search is an important precursor of exploitative innovation because problemistic search is characterized with intentional efforts to improve current performance by addressing deficiencies of current knowledge (Levinthal and March, 1981; Stuart and Podolny, 1996; Levinthal, 1997; Knudsen and Levinthal, 2007), or by searching “in the neighborhood of the current alternative” (Cyert and March, 1963: 121). Organizations may also adopt slack search (Levinthal and March, 1981; March, 1981; Nohria and Gulati, 1996; Greve, 2003) in a hope to identify new ideas and knowledge that may be useful for future new business opportunities, rather than for addressing current business requirements. Unlike problemistic search, the locus of slack search is nonlocal (or distant). Put differently, slack search is not tightly linked to current organizational goals because it reflects “the irrelevant wanderings of loosely controlled subunits” (March, 1981: 214). Slack search is also characterized as “second-order” (Watzlawick et al., 1974; Argyris, 1976) search because it enables “changes in performance targets, technological opportunities, search behavior, and knowledge about opportunities” (Levinthal and March, 1981: 310). Although organizations may discover “something of considerable value” (March, 1981: 214) through slack search, the likelihood of such discovery is not high. In other words, slack search entails higher risk compared to problemistic search that is characterized with certain and predictable outcomes. Accordingly, slack search is an important precursor for exploratory innovation that is enabled by new knowledge beyond the scope of current business. Those differences in underlying search modes inform our discussion on differential influences of organizational slack on alternative types of innovation as discussed below. 1.2 Types of slack and agency theory Second, we consider alternative types of organizational slack as a potential contingency because the type of organizational slack may determine its contribution to performance (Lecuona and Reitzig, 2014). One challenge to examine absorbed slack and unabsorbed slack separately is that there is no a priori theory about their differential effects (Wiseman and Catanach, 1997; Singh, 1986). In this manuscript, we argue that the agency theory represents a promising theoretical perspective with which to evaluate how absorbed slack and unabsorbed slack differ in terms of their influences on managerial risk-taking, and thus on subsequent innovation. Accordingly, we argue that absorbed slack and unabsorbed slack differ in terms of the degree that managers versus shareholders have conflicts regarding risk-taking. Absorbed slack is organizational slack that is distributed to particular usages, or “absorbed into the system design as excess costs” (Bourgeois and Singh, 1983: 43). Examples of absorbed slack include excess inventory, excess machine capacity, and indirect staff. It also is denoted as recoverable slack (Bourgeois and Singh, 1983), or low-discretion slack (Sharfman et al., 1988). On the other hand, unabsorbed slack is an alternative type of organizational slack that is excess, liquid, and uncommitted resources in an organization. Unabsorbed slack is also more readily redeployable because it is not assigned to any particular usages (Bourgeois and Singh, 1983; Singh, 1986). The best examples of unabsorbed slack are cash and marketable securities. Scholars also use available slack (Bourgeois and Singh, 1983), or high-discretion slack (Sharfman et al., 1988) to denote unabsorbed slack. The differences between alternative organizational slacks pertain to the degree of conflict of interests between shareholders and managers. Compared to shareholders, managers are less willing to pursue risk-taking initiatives because managers’ well-being is closely tied to the fate of their organizations (Fama, 1980; Hill and Snell, 1989; Baysinger et al., 1991; Francis and Smith, 1995), while shareholders can reduce their exposure to financial risks by diversifying their investments (Fama, 1980). Consequently, managers often choose to benefit themselves by avoiding risk-taking at the cost of shareholders unless appropriate monitoring tools are in place (Jensen and Meckling, 1976; Fama, 1980; Walkling and Long, 1984; Jensen, 1986; Malatesta and Walkling, 1988). We argue that such conflicts of interest between managers and shareholders are particularly relevant to the usage of absorbed slack because monitoring the usage of absorbed slack is more difficult than monitoring the usage of unabsorbed slack. Put differently, although the expected usage may be more narrowly specified for absorbed slack than for unabsorbed slack, that does not mean monitoring the usage of absorbed slack is easier. First, it is very difficult for external stakeholders to identify excess costs, or absorbed slack (Bourgeois and Singh, 1983; Jensen and Meckling, 1976; Love and Nohria, 2005). In fact, it is also difficult (or at least very costly) for managers to precisely identify excess portions of total costs. Furthermore, it is very difficult to monitor whether general and administrative expenses, one of the typical examples of absorbed slack, are properly used because firms do not disclose how much is spent on what. On the other hand, it is relatively easy to monitor the usage of unabsorbed slack because unabsorbed slack can be clearly identified as liquid resources in excess of current business requirements by capturing current assets that exceed current liabilities. For example, investment in marketable securities, a typical example of unabsorbed slack, is fully disclosed for shareholders’ examination of its appropriateness. Consequently, quantifying gains from absorbed slack is very difficult, whereas returns from unabsorbed slack can be quantified relatively easily. These characteristics of absorbed slack pose substantial challenges for shareholders who try to monitor its appropriate usages. In contrast, monitoring the amount or appropriate usages of unabsorbed slack is relatively straightforward. Therefore, we argue that managers may avoid risk-taking to the extent that monitoring by shareholders is difficult, or that more absorbed slack is available. On the other hand, unabsorbed slack does not pose such challenges for shareholders’ monitoring, and therefore shareholders can pressure managers to be more risk-taking in their decisions to innovate. 1.3 Organizational slack and exploitative innovation Building upon the above, we first argue that organizational slack is negatively associated with exploitative innovation (as show in the left-hand side of Figure 1). This is because increases in organizational slack are associated with being less responsive to competitive changes and feeling less urgency to solve current performance problems (Litschert and Bonham, 1978; Yasai-Ardekani, 1986). It is due to the fact that organizations with more organizational slack are less likely to make changes in their technical core because organizational slack buffers organizations from competitive requirements by absorbing environmental variation (Thompson, 1967). Figure 1. View largeDownload slide The relationship between organizational slack and innovation. Figure 1. View largeDownload slide The relationship between organizational slack and innovation. More specifically, managers can use organizational slack to “pay the price” of “a relatively loose fit between” (Litschert and Bonham, 1978: 216) their choices (in terms of organizational design, strategic decisions, and operational initiatives) and those “dictated by contextual variables” (ibid.). For example, with more absorbed slack, organizations can reduce their bankruptcy risk (Reuer and Leiblein, 2000) because they can avoid profit decreases by simply cutting excess costs, or by decreasing absorbed slack (Cyert and March, 1963). Organizational slack is also associated with lower responsiveness to competitive requirements because organizations may accumulate organizational slack to avoid excessive upward adjustment of organizational aspiration (ibid.). More formally, organizations can keep their attainment discrepancy (Levinthal and March, 1981; Lant, 1992), or the discrepancy between a performance target and achieved performance, at a minimum by adjusting realized performance either by decreasing or by increasing organizational slack. Put differently, organizations with more organizational slack are associated with having smaller and less frequent attainment discrepancies. Consequently, an organization’s search for solutions to a performance problem, or a “problemistic search,” is expected to be less intensive when more organizational slack is available (Cyert and March, 1963: 80). Accordingly, organizations are less motivated to improve current performance by innovations that closely address deficiencies of current knowledge, or exploitative innovation, to the extent that organizational slack increases. We further argue that organizational slack’s influences to discourage exploitative innovation should reveal different forms depending on the types of organizational slack, or between absorbed slack and unabsorbed slack (Singh, 1986). Given that monitoring the usage of absorbed slack poses greater challenges to shareholders (as discussed above), we expect that the negative association between organizational slack and exploitative innovation is particularly salient in the case of absorbed slack. Put differently, organizations with more absorbed slack, often characterized with internally oriented resource allocation patterns (Cheng and Kesner, 1997), are less subject to supervision by shareholders who aim to ensure that managers pursue sufficient risk-taking initiatives, thereby showing lower tolerance for risk-taking associated with innovation (Deephouse and Wiseman, 2000; Steensma and Corley, 2001). Our first hypothesis is stated as follows. H1. The association between absorbed slack and exploitative innovation is negative. On the other hand, we expect that the negative association between organizational slack and exploitative innovation grows less explicit as more unabsorbed slack is available. Put differently, the form of relationship between unabsorbed slack and exploitative innovation is nonlinear, in that the negative influences of unabsorbed slack on exploitative innovation weakens (or the slope grows less steep) as unabsorbed slack increases. Like absorbed slack, unabsorbed slack may also allow organizations to be less responsive to current environmental requirements, thereby discouraging exploitative innovation. However, because shareholders can more easily monitor the usage of unabsorbed slack (as discussed above), it would be increasingly difficult for organizations to forego opportunities (or ignore requirements) of exploitative innovation, as available unabsorbed slack grows more substantial and salient. Accordingly, we expect to observe diminishing negative effects of unabsorbed slack on exploitative innovation because the marginal decrease in exploitative innovation diminishes as unabsorbed slack’s buffering effects are offset by effective shareholders monitoring to the extent that more unabsorbed slack is available. Therefore, we propose our second hypothesis as follows. H2. The association between unabsorbed slack and exploitative innovation is curvilinear, in that the former exerts diminishing negative effects on the latter. 1.4 Organizational slack and exploratory innovation On the other hand, we argue that organizational slack is positively associated with exploratory innovation (as show in the right-hand side of Figure 1). Organizations insulated from current competitive requirements decrease their efforts in exploitative innovation, but they may instead pursue exploratory innovation. As is widely acknowledged, current competitive pressure discourages organizations’ efforts in exploratory innovation (Cooper and Smith, 1992; Christensen and Bower, 1996) because returns from exploratory innovation are uncertain and remote, if any returns are gained at all (March, 1991). Accordingly, organizational slack may enable exploratory innovation by buffering organizations from current competitive environments. In short, the effects of organizational slack are asymmetrical between exploitative and exploratory innovation. More formally, organizational slack allows organizations to satisfice during searches by lowering the threshold for acceptability (Bourgeois, 1981: 36) so that “projects that would not necessarily be approved in a tight budget” are accepted (Cyert and March, 1963: 279). Such resource munificence may be associated with less rigorous evaluation of alternatives, which encourages distant search (Knudsen and Levinthal, 2007). Consequently, “slack provides a source of funds for innovations that would not be approved in the face of scarcity” (Cyert and March, 1963: 279). Put differently, managers’ decisions and actions become more exploratory as more organizational slack is available (March, 1994, 2006; March, 2007). The underlying assumption is that organizational decisions are often outcomes of political concessions among competing managerial coalitions (Cyert and March, 1963). Without uncommitted excess resources, or organizational slack, initiatives with uncertain future consequences are closely scrutinized by competing managerial coalitions before approval is given, if any is given at all. Conversely, organizations with more organizational slack may more likely to accept experimental initiatives that would not be justified based on short-term profit forecasts but that look promising in terms of long-term profit potential. Such an experiment or a “irresponsible” search (Levinthal and March, 1981: 309) that is motivated and enabled by organizational slack is termed “slack search” (Levinthal and March, 1981; Nohria and Gulati, 1996; Greve, 2003); these searches are distinct from “problemistic search” in that the motivation for a slack search is not associated with the need to address a particular performance problem. Instead, slack searches enable organizations to “foster future growth through the development of new and different products or processes” (Souder and Shaver, 2010: 1318) that are not constrained by current competitive requirements. Accordingly, we argue that innovation targeted at identifying or generating new knowledge that is beyond the scope of current business, or exploratory innovation, increases to the extent that more organizational slack is available for slack search. We further argue the positive relationship between organizational slack and exploratory innovation may be particularly salient in the case of unabsorbed slack because shareholders are able to effectively monitor the usage of unabsorbed slack, and then force managers to pursue exploratory innovation vigorously. Accordingly, organizations with more unabsorbed slack are characterized with externally oriented resource allocation patterns (Cheng and Kesner, 1997) that may enable them to search for nonlocal and novel knowledge (Smith et al., 1991; Mishina et al., 2004; Carnabuci and Operti, 2013), or to explore new technologies and markets (Danneels, 2008). H3: The association between unabsorbed slack and exploratory innovation is positive. On the other hand, we argue for diminishing positive effects of absorbed slack on exploratory innovation. Put differently, the form of relationship between absorbed slack and exploratory innovation is nonlinear, in that the positive influences of absorbed slack on exploratory innovation weakens (or the slope grows less steep) as absorbed slack increases. We expect that absorbed slack also allows organization to pursue exploratory innovation by relaxing managerial coalitions’ selective screening. However, marginal increase in exploratory innovation diminishes because managers grow increasingly cautious toward additional risk-taking. Furthermore, shareholders find it increasingly difficult to make sure that their managers pursue sufficient risk-taking initiatives because increases in absorbed slack disturbs shareholders to effectively monitor the usage of absorbed slack (Jensen, 1986; Kim et al., 2008). Consequently, we argue that absorbed slack’s enabling effects on exploratory innovation diminish as available absorbed slack increases. H4: The association between absorbed slack and exploratory innovation is curvilinear, in that the former exerts diminishing positive effects on the latter. 2. Methods 2.1 Sample We tested the hypotheses with data from the Japanese pharmaceutical industry. We particularly leveraged data on their new pharmaceutical products development to operationalize our sample firms’ degree of exploitative, as well as exploratory innovation because new product development is often used as a measure of firms’ innovation (Greve, 2003; Voss et al., 2008; Natividad, 2013). The data on the Japanese pharmaceutical firms’ new products development are appropriate for our study for following two reasons. First, upon the approval of all new ethical drugs, independent specialists determine whether each new pharmaceutical contains an NCE (new chemical entity). This distinction enables us to precisely operationalize exploratory innovation and exploitative innovation because an NCE-based pharmaceutical product represents exploration of new knowledge in the context of new pharmaceutical development, while a non-NCE-based pharmaceutical product captures exploitation of existing knowledge (Bierly and Chakrabarti, 1996; Cardinal, 2001; Dunlap-Hinkler et al., 2010). An NCE is a completely new chemical entity whose medical effects were unknown before. Therefore, finding an NCE requires a search beyond known libraries of active ingredients, while pharmaceutical firms reuse NCEs already approved for medical use to develop non-NCE-based products. One good example of an exploratory pharmaceutical product is Eli Lilly’s Prozac which is based on an NCE, called fluoxetine. Non-NCE version of the same chemical entity is Sarafem, which is an example of an exploitative pharmaceutical product. Fluoxetine was successfully developed as an anti-depressant (Prozac) before Eli Lilly redeveloped it for a different indication of premenstrual dysphoric disorder (Sarafem) upon Prozac’s patent expiration. It is important to note that we are concerned about knowledge underlying the NCE. In particular, we aim to capture the knowledge’s degree of continuity or similarity with the current knowledge base of the organization. Accordingly, the degree of innovativeness or radicalness of benefits enabled by the focal knowledge is irrelevant to our operationalization of exploitative innovation and exploratory innovation. Second, rich data on sample firms’ new product development activities are available. Pharmaceutical firms are required to report on their clinical trial activities to the regulatory agency, which then discloses the information to the public. Leveraging these disclosed data, we are able to measure sample firms’ degree of exploitative, as well as exploratory innovation objectively. A professional medical magazine, called New Current, has been publishing exhaustive lists of pharmaceuticals under development (or pipelines) on a quarterly basis since 1990. The list shows each pharmaceutical firm’s detailed pipeline information, including the name of pipelines, targeted therapeutic indications, stages of clinical trials, and whether each pipeline contains an NCE. Our database consists of 37 Japanese pharmaceutical firms who gained new pharmaceutical approvals during January 2001 to December 2010 in the Japanese market. Combined revenue of these 37 firms represents 58.0% of the total Japanese health-care market as of 2010. We constructed a panel database on these 37 firms over 20 years (from 1991 to 2010). After removing observations due to missing values in at least one variable of interest, we end up with a final data set of 597 firm-years. 2.2 Variables and analysis To test our hypotheses, we constructed a measure of exploitative innovation and exploratory innovation and tested their associations with sample firms’ degree of organizational slack. Natividad (2013) points out the possibility that managers endogenously adjust the amount of organizational slack in response to prior organizational performance. Accordingly, we needed to address the omitted variable problem and the reverse causality problem. Specifically, we employed an instrumental variables estimation (Bascle, 2008; Wooldridge, 2010) with panel data. Furthermore, because panel data include multiple observations per sample firm, observations for the same firm are likely to be correlated. As such, our models applied an instrumental variable method to a data set with possible autocorrelation and heteroskedasticity. Therefore, we chose to employ a continuous-updating estimator (Hansen et al., 1996), which provides us HAC (heteroskedasticity and autocorrelation consistent) estimation. We used a Stata 13 command “xtivreg2” (Schaffer, 2010) with “cue” and “robust” options. Below, we describe variables employed in our models. We used annual data to construct all variables. We also lagged most right-hand-side variables 1 year, so that we can mitigate the possibility of reverse causation. Our dependent variable to test H1 and H2, or exploitative innovation, is operationalized by counting sample firms’ non-NCE-based pipelines at the end of each fiscal year. Because larger firms are likely to be associated with more pipelines, we divided the measure by annual research and development (R&D) expenses. We measured our dependent variable to test H3 and H4, exploratory innovation, in a similar way by counting NCE-based pipelines. Our independent variables are absorbed slack for H1 and H4, unabsorbed slack for H2 and H3, respectively. Following prior works (Bourgeois, 1981; Bourgeois and Singh, 1983; Singh, 1986; Bromiley, 1991; Greve, 2003; Bromiley and Washburn, 2011), we measured absorbed slack by dividing sample firms’ annual selling, general, and administrative expenses with total revenue. As for unabsorbed slack, we divided sample firms’ current assets by current liabilities at the end of each fiscal year. As Natividad (2013) points out, our independent variables (i.e. organizational slack) may be endogenous variables. Accordingly, a set of instrumental variables was employed to obtain fitted values of the independent variables, which were used to estimate our dependent variables (Bascle, 2008; Wooldridge, 2010). We lagged all instrumental variables 1 year to the independent variables, and therefore 2 years to the dependent variables, so that we could minimize the possibility that our instrumental variables influence the dependent variables. As for absorbed slack, we employed R&D intensity and total assets (trillion yen) as instruments. Because absorbed slack is “absorbed into the system design as excess costs” (Bourgeois and Singh, 1983: 43), it is highly conceivable that absorbed slack would be used to buffer absorbing processes or subunits from unpredictable variations. In other words, absorbed slack is not committed to specific purposes, but some processes or subunits define its default usages by absorbing it. Therefore, we argue that organizations are motivated to increase absorbed slack to the extent that they recognize their current processes or subunits as unpredictable. One typical source of unpredictability for pharmaceutical firms’ current operation is R&D. Accordingly, we employed R&D intensity, or annual R&D expenditure divided by total revenue, as a measure of the degree to which sample firms are motivated to increase absorbed slack as a buffer against negative surprises. Furthermore, stabilizing processes or subunits calls for substantial buffer resources to the extent that the focal processes or subunits have deployed larger resources. Therefore, we expect a positive association between total assets and absorbed slack because firms with larger total assets may need more absorbed slack as their buffer. Instruments for unabsorbed slack include environmental dynamics and revenue growth. Unabsorbed slack differs from absorbed slack in that unabsorbed slack is not absorbed into “the system design,” indicating that no organizational decisions are made as to which processes or subunits can claim a privilege to use this portion of uncommitted resources. Put differently, organizations reserve unabsorbed slack for usages beyond current operations. Accordingly, we argue that organizations increase unabsorbed slack to the extent that they anticipate major changes in their competitive environments due to a high degree of environmental dynamics. Our measure of environmental dynamics reflects the volatility of competitive environments. Following Wang and Li (2008), we regressed the Japanese health-care industry sales over a moving window of 30 years preceding the focal year on the year variable, and then used the standard error of the regression coefficient related to a year variable to produce an index of environmental dynamics. Conversely, managers may feel that major changes in their competitive environments are unlikely to the extent that their current business performance is favorable. We operationalized the degree of favorable business performance by revenue growth, or annual growth rate of sample firms’ total revenue, which we expect decreases firms’ unabsorbed slack in a subsequent year. Weak identification tests by Cragg-Donald Wald F statistic (Cragg and Donald, 1993) revealed that we can reject the null hypothesis that our instruments are weak, or only marginally relevant for both sets of instruments. Tests of overidentifying restrictions by Hansen J statistic (Hansen, 1982) indicated that the null hypothesis that all instruments are valid is not rejected. Furthermore, n times the R2 from the first stage of two-stage least squares are much larger than the number of instruments, indicating that two-stage least squares are less biased than ordinary least squares for our models (Murray, 2006). We also employed several time-varying control variables that corresponded to each time period in which sample firms were at risk of innovating. First, organizational size is our sample firms’ number of employees (in thousands). R&D is also employed as a measure of the degree of sample firms’ innovative capacity operationalized by their annual R&D expenditure (billion yen). We also included sample firms’ age to control for effects of sample firms’ senescence. A dummy variable that indicates whether sample firms experienced mergers and acquisitions in a preceding year (M&As) controls for influences of drastic changes in their pipelines. We also employed a measure of sample firms’ attainment discrepancy because it influences their degree of risk-taking. We operationalized it by the discrepancy in ROA (return on assets) between the focal firm and the industry average. Because we measure sample firms’ degree of innovation by counting pipelines, it also is important to control for the degree of development resources intensity. Therefore, we also included average size of pipelines by dividing annual R&D expenditure by total counts of pipelines at the end of each fiscal year. It also may be important to consider influences of government regulations. The Japanese government periodically reviews official reimbursement prices of pharmaceutical products downward. The revisions strongly influence pharmaceutical firms’ new product development strategy. Accordingly, we included a variable that shows the industry average degree of reimbursement price revisions. Managers may be more anxious to raise their assets’ productivity to the extent that their assets increase (Abernathy, 1978). Therefore, we included a measure of annual expansion of total assets, or sample firm’s asset growth, to control for assets size effects on managers’ risk avoidance. Furthermore, because sample firms’ potential slack (Bromiley, 1991; O'brien, 2003) may influence their degree of innovation, each firm’s debt–equity ratio is also employed. Finally, lagged exploratory innovation is also included for models with exploitative innovation as a dependent variable (H1 and H2) because new ideas and knowledge gained through exploratory innovation may enable subsequent exploitative innovation. 3. Results Table 1 shows descriptive statistics and a correlation matrix for all variables employed in our models. Overall, the independent and control variables show considerable variability, and most correlations among the variables range from small to moderate. We also checked the variance inflation factors for all variables, and none of them exceeds 10.0, which is the rule of thumb threshold of potential multicollinearity (Cohen et al., 2003). Table 1. Descriptive statistics and correlations Variables  Mean  Standard deviation  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  1. Exploitative innovation  0.53  0.75                                  2. Exploratory innovation  1.01  1.10  0.64*                                3. R&D intensity  0.12  0.24  0.11*  −0.01                              4. Total assets  0.50  0.61  −0.32*  −0.39*  −0.14*                            5. Environmental dynamics  0.19  0.06  0.00  0.13*  0.06  0.00                          6. Revenue growth  0.03  0.12  −0.09*  −0.03  −0.26*  0.07  0.06                        7. Organizational size  7.46  8.90  −0.30*  −0.39*  −0.18*  0.84*  −0.02  0.05                      8. R&D  26.86  35.85  −0.29*  −0.37*  0.00  0.62*  0.04  0.12*  0.39*                    9. Age  68.79  18.88  −0.08*  −0.09*  −0.31*  −0.05  −0.06  0.08*  −0.01  0.23*                  10. M&As  0.01  0.09  0.05  0.02  0.00  −0.02  −0.03  −0.03  −0.02  0.01  0.06                11. Attainment discrepancy  0.00  0.06  0.11*  0.14*  0.48*  −0.12*  0.01  −0.30*  0.07  −0.27*  −0.22*  −0.02              12. Size of pipelines  1.50  2.20  −0.28*  −0.35*  −0.09*  0.51*  0.03  0.03  0.58*  0.33*  0.01  −0.04  0.00            13. Price revisions  0.04  0.04  0.02  0.10*  −0.05  −0.04  0.01  0.03  −0.04  −0.08  −0.05  −0.05  −0.02  0.01          14. Asset growth  1.04  0.16  −0.01  0.00  0.03  0.06  0.03  0.48*  0.00  0.06  −0.08  −0.01  −0.06  0.01  0.03        15. Potential slack  0.37  0.56  −0.08*  0.03  −0.11*  0.07  0.08  −0.03  0.21*  −0.08*  −0.09*  −0.04  0.26*  0.10*  0.09*  −0.06      16. Absorbed slack  0.31  0.10  0.27*  0.32*  0.33*  −0.46*  0.05  −0.14*  −0.54*  −0.14*  −0.13*  −0.03  0.06  −0.33*  0.02  −0.05  −0.15*    17. Unabsorbed slack  2.93  1.63  0.14*  0.05  0.47*  −0.18*  0.01  −0.11*  −0.36*  0.01  −0.09*  −0.01  −0.08  −0.17*  −0.05  0.05  −0.39*  0.37*  Variables  Mean  Standard deviation  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  1. Exploitative innovation  0.53  0.75                                  2. Exploratory innovation  1.01  1.10  0.64*                                3. R&D intensity  0.12  0.24  0.11*  −0.01                              4. Total assets  0.50  0.61  −0.32*  −0.39*  −0.14*                            5. Environmental dynamics  0.19  0.06  0.00  0.13*  0.06  0.00                          6. Revenue growth  0.03  0.12  −0.09*  −0.03  −0.26*  0.07  0.06                        7. Organizational size  7.46  8.90  −0.30*  −0.39*  −0.18*  0.84*  −0.02  0.05                      8. R&D  26.86  35.85  −0.29*  −0.37*  0.00  0.62*  0.04  0.12*  0.39*                    9. Age  68.79  18.88  −0.08*  −0.09*  −0.31*  −0.05  −0.06  0.08*  −0.01  0.23*                  10. M&As  0.01  0.09  0.05  0.02  0.00  −0.02  −0.03  −0.03  −0.02  0.01  0.06                11. Attainment discrepancy  0.00  0.06  0.11*  0.14*  0.48*  −0.12*  0.01  −0.30*  0.07  −0.27*  −0.22*  −0.02              12. Size of pipelines  1.50  2.20  −0.28*  −0.35*  −0.09*  0.51*  0.03  0.03  0.58*  0.33*  0.01  −0.04  0.00            13. Price revisions  0.04  0.04  0.02  0.10*  −0.05  −0.04  0.01  0.03  −0.04  −0.08  −0.05  −0.05  −0.02  0.01          14. Asset growth  1.04  0.16  −0.01  0.00  0.03  0.06  0.03  0.48*  0.00  0.06  −0.08  −0.01  −0.06  0.01  0.03        15. Potential slack  0.37  0.56  −0.08*  0.03  −0.11*  0.07  0.08  −0.03  0.21*  −0.08*  −0.09*  −0.04  0.26*  0.10*  0.09*  −0.06      16. Absorbed slack  0.31  0.10  0.27*  0.32*  0.33*  −0.46*  0.05  −0.14*  −0.54*  −0.14*  −0.13*  −0.03  0.06  −0.33*  0.02  −0.05  −0.15*    17. Unabsorbed slack  2.93  1.63  0.14*  0.05  0.47*  −0.18*  0.01  −0.11*  −0.36*  0.01  −0.09*  −0.01  −0.08  −0.17*  −0.05  0.05  −0.39*  0.37*  *P <0.05. Table 1. Descriptive statistics and correlations Variables  Mean  Standard deviation  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  1. Exploitative innovation  0.53  0.75                                  2. Exploratory innovation  1.01  1.10  0.64*                                3. R&D intensity  0.12  0.24  0.11*  −0.01                              4. Total assets  0.50  0.61  −0.32*  −0.39*  −0.14*                            5. Environmental dynamics  0.19  0.06  0.00  0.13*  0.06  0.00                          6. Revenue growth  0.03  0.12  −0.09*  −0.03  −0.26*  0.07  0.06                        7. Organizational size  7.46  8.90  −0.30*  −0.39*  −0.18*  0.84*  −0.02  0.05                      8. R&D  26.86  35.85  −0.29*  −0.37*  0.00  0.62*  0.04  0.12*  0.39*                    9. Age  68.79  18.88  −0.08*  −0.09*  −0.31*  −0.05  −0.06  0.08*  −0.01  0.23*                  10. M&As  0.01  0.09  0.05  0.02  0.00  −0.02  −0.03  −0.03  −0.02  0.01  0.06                11. Attainment discrepancy  0.00  0.06  0.11*  0.14*  0.48*  −0.12*  0.01  −0.30*  0.07  −0.27*  −0.22*  −0.02              12. Size of pipelines  1.50  2.20  −0.28*  −0.35*  −0.09*  0.51*  0.03  0.03  0.58*  0.33*  0.01  −0.04  0.00            13. Price revisions  0.04  0.04  0.02  0.10*  −0.05  −0.04  0.01  0.03  −0.04  −0.08  −0.05  −0.05  −0.02  0.01          14. Asset growth  1.04  0.16  −0.01  0.00  0.03  0.06  0.03  0.48*  0.00  0.06  −0.08  −0.01  −0.06  0.01  0.03        15. Potential slack  0.37  0.56  −0.08*  0.03  −0.11*  0.07  0.08  −0.03  0.21*  −0.08*  −0.09*  −0.04  0.26*  0.10*  0.09*  −0.06      16. Absorbed slack  0.31  0.10  0.27*  0.32*  0.33*  −0.46*  0.05  −0.14*  −0.54*  −0.14*  −0.13*  −0.03  0.06  −0.33*  0.02  −0.05  −0.15*    17. Unabsorbed slack  2.93  1.63  0.14*  0.05  0.47*  −0.18*  0.01  −0.11*  −0.36*  0.01  −0.09*  −0.01  −0.08  −0.17*  −0.05  0.05  −0.39*  0.37*  Variables  Mean  Standard deviation  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  1. Exploitative innovation  0.53  0.75                                  2. Exploratory innovation  1.01  1.10  0.64*                                3. R&D intensity  0.12  0.24  0.11*  −0.01                              4. Total assets  0.50  0.61  −0.32*  −0.39*  −0.14*                            5. Environmental dynamics  0.19  0.06  0.00  0.13*  0.06  0.00                          6. Revenue growth  0.03  0.12  −0.09*  −0.03  −0.26*  0.07  0.06                        7. Organizational size  7.46  8.90  −0.30*  −0.39*  −0.18*  0.84*  −0.02  0.05                      8. R&D  26.86  35.85  −0.29*  −0.37*  0.00  0.62*  0.04  0.12*  0.39*                    9. Age  68.79  18.88  −0.08*  −0.09*  −0.31*  −0.05  −0.06  0.08*  −0.01  0.23*                  10. M&As  0.01  0.09  0.05  0.02  0.00  −0.02  −0.03  −0.03  −0.02  0.01  0.06                11. Attainment discrepancy  0.00  0.06  0.11*  0.14*  0.48*  −0.12*  0.01  −0.30*  0.07  −0.27*  −0.22*  −0.02              12. Size of pipelines  1.50  2.20  −0.28*  −0.35*  −0.09*  0.51*  0.03  0.03  0.58*  0.33*  0.01  −0.04  0.00            13. Price revisions  0.04  0.04  0.02  0.10*  −0.05  −0.04  0.01  0.03  −0.04  −0.08  −0.05  −0.05  −0.02  0.01          14. Asset growth  1.04  0.16  −0.01  0.00  0.03  0.06  0.03  0.48*  0.00  0.06  −0.08  −0.01  −0.06  0.01  0.03        15. Potential slack  0.37  0.56  −0.08*  0.03  −0.11*  0.07  0.08  −0.03  0.21*  −0.08*  −0.09*  −0.04  0.26*  0.10*  0.09*  −0.06      16. Absorbed slack  0.31  0.10  0.27*  0.32*  0.33*  −0.46*  0.05  −0.14*  −0.54*  −0.14*  −0.13*  −0.03  0.06  −0.33*  0.02  −0.05  −0.15*    17. Unabsorbed slack  2.93  1.63  0.14*  0.05  0.47*  −0.18*  0.01  −0.11*  −0.36*  0.01  −0.09*  −0.01  −0.08  −0.17*  −0.05  0.05  −0.39*  0.37*  *P <0.05. Table 2 reports the results of our tests of hypotheses on absorbed slack’s influences (H1 and H4). Models 1 and 4 show the first-stage models, where we regress instrumental variables against our independent variable, i.e. absorbed slack. Exogenous variables for exploitative innovation and for exploratory innovation are also included in Models 1 and 4, respectively. All instrumental variables show strong association with our measure of absorbed slack. Models 2 and 5 are models with control variables. We then used fitted values of absorbed slack to estimate their associations with exploitative innovation (Model 3, H1), and with exploratory innovation (Models 6 and 7, H4). Table 3 follows the similar format with alternative independent variable of unabsorbed slack. We examine the association between unabsorbed slack and exploitative innovation (H2) by Models 10 and 11. Model 14 tests the association between unabsorbed slack and exploratory innovation (H3). Table 2. Results of the panel continuous-updating estimator (CUE) instrumental variable regression analysis for the effects of absorbed slack on innovationa Dependent variables  Absorbed slack   Exploitative innovation   Absorbed slack   Exploratory innovation   Model 1  Model 2  Model 3  Model 4  Model 5  Model 6  Model 7  R&D intensity  0.36***  (0.07)          0.36***  (0.07)              Total assets  0.06**  (0.02)          0.06***  (0.02)              Organizational size  0.01  (0.01)  0.10†  (0.05)  0.12**  (0.04)  0.01  (0.01)  0.07  (0.11)  −0.18  (0.17)  −0.13  (0.09)  R&D  0.00  (0.01)  0.00  (0.03)  0.03  (0.02)  0.00  (0.01)  0.13*  (0.06)  0.95*  (0.38)  −0.01  (0.03)  Age  −0.05†  (0.03)  −0.03  (0.20)  −0.09  (0.14)  −0.07**  (0.02)  −1.48**  (0.46)  −2.04***  (0.37)  −1.00***  (0.18)  M&As  0.00  (0.02)  −0.18†  (0.09)  −0.17  (0.13)  0.01  (0.01)  0.04  (0.19)  0.15  (0.19)  0.00  (0.19)  Attainment discrepancy  0.25†  (0.14)  −1.05*  (0.51)  −0.63  (0.44)  0.26†  (0.14)  1.62  (1.32)  1.67  (1.73)  0.13  (1.01)  Size of pipelines  0.00  (0.00)  0.00  (0.00)  0.00  (0.00)  0.00  (0.00)  −0.02  (0.02)  −0.03†  (0.02)  −0.02  (0.01)  Price revisions  0.04  (0.02)  0.26  (0.31)  0.32  (0.48)  0.04†  (0.02)  0.74  (0.45)  0.64  (0.64)  0.56  (0.60)  Asset growth  0.04  (0.04)  −0.16  (0.12)  −0.11  (0.11)  0.04  (0.04)  −0.07  (0.15)  −0.27  (0.23)  0.06  (0.17)  Potential slack  0.01  (0.01)  −0.03  (0.04)  −0.01  (0.05)  0.01  (0.01)  −0.04  (0.08)  −0.07  (0.07)  −0.13  (0.12)  Exploratory innovation  0.01†  (0.00)  0.34***  (0.07)  0.36***  (0.10)                  Absorbed slack          −2.02*  (1.02)          0.67  (1.01)  8.36***  (1.72)  Absorbed slack, squared                          −9.02***  (2.42)  Constant  −0.12*  (0.05)  0.33*  (0.13)      −0.12*  (0.05)  1.02***  (0.16)          N firm-years  597    597    597    597    597    597    597    N Firms  37    37    37    37    37    37    37    RMSE  0.04    0.42    0.44    0.04    0.67    0.83    0.71    Log likelihood  1054    −329    −344    1050    −604    −719    −625    F  25.6    27.4    3.68    25.7    6.08    8.37    8.21    Dependent variables  Absorbed slack   Exploitative innovation   Absorbed slack   Exploratory innovation   Model 1  Model 2  Model 3  Model 4  Model 5  Model 6  Model 7  R&D intensity  0.36***  (0.07)          0.36***  (0.07)              Total assets  0.06**  (0.02)          0.06***  (0.02)              Organizational size  0.01  (0.01)  0.10†  (0.05)  0.12**  (0.04)  0.01  (0.01)  0.07  (0.11)  −0.18  (0.17)  −0.13  (0.09)  R&D  0.00  (0.01)  0.00  (0.03)  0.03  (0.02)  0.00  (0.01)  0.13*  (0.06)  0.95*  (0.38)  −0.01  (0.03)  Age  −0.05†  (0.03)  −0.03  (0.20)  −0.09  (0.14)  −0.07**  (0.02)  −1.48**  (0.46)  −2.04***  (0.37)  −1.00***  (0.18)  M&As  0.00  (0.02)  −0.18†  (0.09)  −0.17  (0.13)  0.01  (0.01)  0.04  (0.19)  0.15  (0.19)  0.00  (0.19)  Attainment discrepancy  0.25†  (0.14)  −1.05*  (0.51)  −0.63  (0.44)  0.26†  (0.14)  1.62  (1.32)  1.67  (1.73)  0.13  (1.01)  Size of pipelines  0.00  (0.00)  0.00  (0.00)  0.00  (0.00)  0.00  (0.00)  −0.02  (0.02)  −0.03†  (0.02)  −0.02  (0.01)  Price revisions  0.04  (0.02)  0.26  (0.31)  0.32  (0.48)  0.04†  (0.02)  0.74  (0.45)  0.64  (0.64)  0.56  (0.60)  Asset growth  0.04  (0.04)  −0.16  (0.12)  −0.11  (0.11)  0.04  (0.04)  −0.07  (0.15)  −0.27  (0.23)  0.06  (0.17)  Potential slack  0.01  (0.01)  −0.03  (0.04)  −0.01  (0.05)  0.01  (0.01)  −0.04  (0.08)  −0.07  (0.07)  −0.13  (0.12)  Exploratory innovation  0.01†  (0.00)  0.34***  (0.07)  0.36***  (0.10)                  Absorbed slack          −2.02*  (1.02)          0.67  (1.01)  8.36***  (1.72)  Absorbed slack, squared                          −9.02***  (2.42)  Constant  −0.12*  (0.05)  0.33*  (0.13)      −0.12*  (0.05)  1.02***  (0.16)          N firm-years  597    597    597    597    597    597    597    N Firms  37    37    37    37    37    37    37    RMSE  0.04    0.42    0.44    0.04    0.67    0.83    0.71    Log likelihood  1054    −329    −344    1050    −604    −719    −625    F  25.6    27.4    3.68    25.7    6.08    8.37    8.21    a Robust standard errors adjusted for clustering by firm are in parentheses. Two-tailed tests for all effects. † P < 0.1, *P < 0.05, **P < 0.01, ***P < 0.001. Table 2. Results of the panel continuous-updating estimator (CUE) instrumental variable regression analysis for the effects of absorbed slack on innovationa Dependent variables  Absorbed slack   Exploitative innovation   Absorbed slack   Exploratory innovation   Model 1  Model 2  Model 3  Model 4  Model 5  Model 6  Model 7  R&D intensity  0.36***  (0.07)          0.36***  (0.07)              Total assets  0.06**  (0.02)          0.06***  (0.02)              Organizational size  0.01  (0.01)  0.10†  (0.05)  0.12**  (0.04)  0.01  (0.01)  0.07  (0.11)  −0.18  (0.17)  −0.13  (0.09)  R&D  0.00  (0.01)  0.00  (0.03)  0.03  (0.02)  0.00  (0.01)  0.13*  (0.06)  0.95*  (0.38)  −0.01  (0.03)  Age  −0.05†  (0.03)  −0.03  (0.20)  −0.09  (0.14)  −0.07**  (0.02)  −1.48**  (0.46)  −2.04***  (0.37)  −1.00***  (0.18)  M&As  0.00  (0.02)  −0.18†  (0.09)  −0.17  (0.13)  0.01  (0.01)  0.04  (0.19)  0.15  (0.19)  0.00  (0.19)  Attainment discrepancy  0.25†  (0.14)  −1.05*  (0.51)  −0.63  (0.44)  0.26†  (0.14)  1.62  (1.32)  1.67  (1.73)  0.13  (1.01)  Size of pipelines  0.00  (0.00)  0.00  (0.00)  0.00  (0.00)  0.00  (0.00)  −0.02  (0.02)  −0.03†  (0.02)  −0.02  (0.01)  Price revisions  0.04  (0.02)  0.26  (0.31)  0.32  (0.48)  0.04†  (0.02)  0.74  (0.45)  0.64  (0.64)  0.56  (0.60)  Asset growth  0.04  (0.04)  −0.16  (0.12)  −0.11  (0.11)  0.04  (0.04)  −0.07  (0.15)  −0.27  (0.23)  0.06  (0.17)  Potential slack  0.01  (0.01)  −0.03  (0.04)  −0.01  (0.05)  0.01  (0.01)  −0.04  (0.08)  −0.07  (0.07)  −0.13  (0.12)  Exploratory innovation  0.01†  (0.00)  0.34***  (0.07)  0.36***  (0.10)                  Absorbed slack          −2.02*  (1.02)          0.67  (1.01)  8.36***  (1.72)  Absorbed slack, squared                          −9.02***  (2.42)  Constant  −0.12*  (0.05)  0.33*  (0.13)      −0.12*  (0.05)  1.02***  (0.16)          N firm-years  597    597    597    597    597    597    597    N Firms  37    37    37    37    37    37    37    RMSE  0.04    0.42    0.44    0.04    0.67    0.83    0.71    Log likelihood  1054    −329    −344    1050    −604    −719    −625    F  25.6    27.4    3.68    25.7    6.08    8.37    8.21    Dependent variables  Absorbed slack   Exploitative innovation   Absorbed slack   Exploratory innovation   Model 1  Model 2  Model 3  Model 4  Model 5  Model 6  Model 7  R&D intensity  0.36***  (0.07)          0.36***  (0.07)              Total assets  0.06**  (0.02)          0.06***  (0.02)              Organizational size  0.01  (0.01)  0.10†  (0.05)  0.12**  (0.04)  0.01  (0.01)  0.07  (0.11)  −0.18  (0.17)  −0.13  (0.09)  R&D  0.00  (0.01)  0.00  (0.03)  0.03  (0.02)  0.00  (0.01)  0.13*  (0.06)  0.95*  (0.38)  −0.01  (0.03)  Age  −0.05†  (0.03)  −0.03  (0.20)  −0.09  (0.14)  −0.07**  (0.02)  −1.48**  (0.46)  −2.04***  (0.37)  −1.00***  (0.18)  M&As  0.00  (0.02)  −0.18†  (0.09)  −0.17  (0.13)  0.01  (0.01)  0.04  (0.19)  0.15  (0.19)  0.00  (0.19)  Attainment discrepancy  0.25†  (0.14)  −1.05*  (0.51)  −0.63  (0.44)  0.26†  (0.14)  1.62  (1.32)  1.67  (1.73)  0.13  (1.01)  Size of pipelines  0.00  (0.00)  0.00  (0.00)  0.00  (0.00)  0.00  (0.00)  −0.02  (0.02)  −0.03†  (0.02)  −0.02  (0.01)  Price revisions  0.04  (0.02)  0.26  (0.31)  0.32  (0.48)  0.04†  (0.02)  0.74  (0.45)  0.64  (0.64)  0.56  (0.60)  Asset growth  0.04  (0.04)  −0.16  (0.12)  −0.11  (0.11)  0.04  (0.04)  −0.07  (0.15)  −0.27  (0.23)  0.06  (0.17)  Potential slack  0.01  (0.01)  −0.03  (0.04)  −0.01  (0.05)  0.01  (0.01)  −0.04  (0.08)  −0.07  (0.07)  −0.13  (0.12)  Exploratory innovation  0.01†  (0.00)  0.34***  (0.07)  0.36***  (0.10)                  Absorbed slack          −2.02*  (1.02)          0.67  (1.01)  8.36***  (1.72)  Absorbed slack, squared                          −9.02***  (2.42)  Constant  −0.12*  (0.05)  0.33*  (0.13)      −0.12*  (0.05)  1.02***  (0.16)          N firm-years  597    597    597    597    597    597    597    N Firms  37    37    37    37    37    37    37    RMSE  0.04    0.42    0.44    0.04    0.67    0.83    0.71    Log likelihood  1054    −329    −344    1050    −604    −719    −625    F  25.6    27.4    3.68    25.7    6.08    8.37    8.21    a Robust standard errors adjusted for clustering by firm are in parentheses. Two-tailed tests for all effects. † P < 0.1, *P < 0.05, **P < 0.01, ***P < 0.001. Table 3. Results of the panel continuous-updating estimator (CUE) instrumental variable regression analysis for the effects of unabsorbed slack on innovationa Dependent variables  Unabsorbed slack   Exploitative innovation   Unabsorbed slack   Exploratory innovation   Model 8  Model 9  Model 10  Model 11  Model 12  Model 13  Model 14  Environmental dynamics  2.14†  (1.10)              2.24†  (1.13)          Revenue growth  −0.74*  (0.29)              −0.75*  (0.29)          Organizational size  −0.51***  (0.13)  0.10†  (0.05)  −0.15  (0.15)  −0.56  (0.36)  −0.51***  (0.13)  0.07  (0.11)  0.28*  (0.12)  R&D  −0.06  (0.13)  0.00  (0.03)  −0.01  (0.04)  −0.05  (0.11)  −0.05  (0.13)  0.13*  (0.06)  0.14**  (0.05)  Age  1.88***  (0.50)  −0.03  (0.20)  0.71  (0.48)  1.68†  (0.88)  1.78***  (0.46)  −1.48**  (0.46)  −2.14***  (0.35)  M&As  −0.25  (0.31)  −0.18†  (0.09)  −0.26  (0.20)  −0.26  (0.45)  −0.19  (0.30)  0.04  (0.19)  0.13  (0.15)  Attainment discrepancy  −0.37  (1.86)  −1.05*  (0.51)  −1.13  (0.78)  −4.18†  (2.15)  −0.30  (1.90)  1.62  (1.32)  1.58  (1.04)  Size of pipelines  0.00  (0.01)  0.00  (0.00)  0.01  (0.01)  0.01  (0.01)  0.00  (0.01)  −0.02  (0.02)  −0.03*  (0.01)  Price revisions  0.82†  (0.46)  0.26  (0.31)  0.54  (0.60)  1.15  (0.96)  0.87†  (0.46)  0.74  (0.45)  0.50  (0.67)  Asset growth  −0.02  (0.34)  −0.16  (0.12)  0.03  (0.21)  −0.34  (0.41)  −0.01  (0.34)  −0.07  (0.15)  0.04  (0.19)  Potential slack  −0.02  (0.08)  −0.03  (0.04)  −0.05  (0.04)  −0.09  (0.09)  −0.02  (0.08)  −0.04  (0.08)  −0.04  (0.05)  Exploratory innovation  0.07  (0.06)  0.34***  (0.07)  0.38***  (0.10)  0.35***  (0.10)              Unabsorbed slack          −0.45†  (0.27)  −1.42*  (0.70)          0.42**  (0.16)  Unabsorbed slack, squared              0.17*  (0.08)              Constant  −0.44  (0.47)  0.33*  (0.13)          −0.39  (0.45)  1.02***  (0.16)      N firm-years  597    597    597    597    597    597    597    N Firms  37    37    37    37    37    37    37    RMSE  0.73    0.42    0.57    0.96    0.73    0.67    0.74    Log likelihood  −652    −329    −492    −801    −653    −604    −648    F  7.48    27.4    2.30    1.94    7.41    6.08    10.1    Dependent variables  Unabsorbed slack   Exploitative innovation   Unabsorbed slack   Exploratory innovation   Model 8  Model 9  Model 10  Model 11  Model 12  Model 13  Model 14  Environmental dynamics  2.14†  (1.10)              2.24†  (1.13)          Revenue growth  −0.74*  (0.29)              −0.75*  (0.29)          Organizational size  −0.51***  (0.13)  0.10†  (0.05)  −0.15  (0.15)  −0.56  (0.36)  −0.51***  (0.13)  0.07  (0.11)  0.28*  (0.12)  R&D  −0.06  (0.13)  0.00  (0.03)  −0.01  (0.04)  −0.05  (0.11)  −0.05  (0.13)  0.13*  (0.06)  0.14**  (0.05)  Age  1.88***  (0.50)  −0.03  (0.20)  0.71  (0.48)  1.68†  (0.88)  1.78***  (0.46)  −1.48**  (0.46)  −2.14***  (0.35)  M&As  −0.25  (0.31)  −0.18†  (0.09)  −0.26  (0.20)  −0.26  (0.45)  −0.19  (0.30)  0.04  (0.19)  0.13  (0.15)  Attainment discrepancy  −0.37  (1.86)  −1.05*  (0.51)  −1.13  (0.78)  −4.18†  (2.15)  −0.30  (1.90)  1.62  (1.32)  1.58  (1.04)  Size of pipelines  0.00  (0.01)  0.00  (0.00)  0.01  (0.01)  0.01  (0.01)  0.00  (0.01)  −0.02  (0.02)  −0.03*  (0.01)  Price revisions  0.82†  (0.46)  0.26  (0.31)  0.54  (0.60)  1.15  (0.96)  0.87†  (0.46)  0.74  (0.45)  0.50  (0.67)  Asset growth  −0.02  (0.34)  −0.16  (0.12)  0.03  (0.21)  −0.34  (0.41)  −0.01  (0.34)  −0.07  (0.15)  0.04  (0.19)  Potential slack  −0.02  (0.08)  −0.03  (0.04)  −0.05  (0.04)  −0.09  (0.09)  −0.02  (0.08)  −0.04  (0.08)  −0.04  (0.05)  Exploratory innovation  0.07  (0.06)  0.34***  (0.07)  0.38***  (0.10)  0.35***  (0.10)              Unabsorbed slack          −0.45†  (0.27)  −1.42*  (0.70)          0.42**  (0.16)  Unabsorbed slack, squared              0.17*  (0.08)              Constant  −0.44  (0.47)  0.33*  (0.13)          −0.39  (0.45)  1.02***  (0.16)      N firm-years  597    597    597    597    597    597    597    N Firms  37    37    37    37    37    37    37    RMSE  0.73    0.42    0.57    0.96    0.73    0.67    0.74    Log likelihood  −652    −329    −492    −801    −653    −604    −648    F  7.48    27.4    2.30    1.94    7.41    6.08    10.1    a Robust standard errors adjusted for clustering by firm are in parentheses. Two-tailed tests for all effects. † P < 0.1, *P < 0.05, **P < 0.01, ***P < 0.001. Table 3. Results of the panel continuous-updating estimator (CUE) instrumental variable regression analysis for the effects of unabsorbed slack on innovationa Dependent variables  Unabsorbed slack   Exploitative innovation   Unabsorbed slack   Exploratory innovation   Model 8  Model 9  Model 10  Model 11  Model 12  Model 13  Model 14  Environmental dynamics  2.14†  (1.10)              2.24†  (1.13)          Revenue growth  −0.74*  (0.29)              −0.75*  (0.29)          Organizational size  −0.51***  (0.13)  0.10†  (0.05)  −0.15  (0.15)  −0.56  (0.36)  −0.51***  (0.13)  0.07  (0.11)  0.28*  (0.12)  R&D  −0.06  (0.13)  0.00  (0.03)  −0.01  (0.04)  −0.05  (0.11)  −0.05  (0.13)  0.13*  (0.06)  0.14**  (0.05)  Age  1.88***  (0.50)  −0.03  (0.20)  0.71  (0.48)  1.68†  (0.88)  1.78***  (0.46)  −1.48**  (0.46)  −2.14***  (0.35)  M&As  −0.25  (0.31)  −0.18†  (0.09)  −0.26  (0.20)  −0.26  (0.45)  −0.19  (0.30)  0.04  (0.19)  0.13  (0.15)  Attainment discrepancy  −0.37  (1.86)  −1.05*  (0.51)  −1.13  (0.78)  −4.18†  (2.15)  −0.30  (1.90)  1.62  (1.32)  1.58  (1.04)  Size of pipelines  0.00  (0.01)  0.00  (0.00)  0.01  (0.01)  0.01  (0.01)  0.00  (0.01)  −0.02  (0.02)  −0.03*  (0.01)  Price revisions  0.82†  (0.46)  0.26  (0.31)  0.54  (0.60)  1.15  (0.96)  0.87†  (0.46)  0.74  (0.45)  0.50  (0.67)  Asset growth  −0.02  (0.34)  −0.16  (0.12)  0.03  (0.21)  −0.34  (0.41)  −0.01  (0.34)  −0.07  (0.15)  0.04  (0.19)  Potential slack  −0.02  (0.08)  −0.03  (0.04)  −0.05  (0.04)  −0.09  (0.09)  −0.02  (0.08)  −0.04  (0.08)  −0.04  (0.05)  Exploratory innovation  0.07  (0.06)  0.34***  (0.07)  0.38***  (0.10)  0.35***  (0.10)              Unabsorbed slack          −0.45†  (0.27)  −1.42*  (0.70)          0.42**  (0.16)  Unabsorbed slack, squared              0.17*  (0.08)              Constant  −0.44  (0.47)  0.33*  (0.13)          −0.39  (0.45)  1.02***  (0.16)      N firm-years  597    597    597    597    597    597    597    N Firms  37    37    37    37    37    37    37    RMSE  0.73    0.42    0.57    0.96    0.73    0.67    0.74    Log likelihood  −652    −329    −492    −801    −653    −604    −648    F  7.48    27.4    2.30    1.94    7.41    6.08    10.1    Dependent variables  Unabsorbed slack   Exploitative innovation   Unabsorbed slack   Exploratory innovation   Model 8  Model 9  Model 10  Model 11  Model 12  Model 13  Model 14  Environmental dynamics  2.14†  (1.10)              2.24†  (1.13)          Revenue growth  −0.74*  (0.29)              −0.75*  (0.29)          Organizational size  −0.51***  (0.13)  0.10†  (0.05)  −0.15  (0.15)  −0.56  (0.36)  −0.51***  (0.13)  0.07  (0.11)  0.28*  (0.12)  R&D  −0.06  (0.13)  0.00  (0.03)  −0.01  (0.04)  −0.05  (0.11)  −0.05  (0.13)  0.13*  (0.06)  0.14**  (0.05)  Age  1.88***  (0.50)  −0.03  (0.20)  0.71  (0.48)  1.68†  (0.88)  1.78***  (0.46)  −1.48**  (0.46)  −2.14***  (0.35)  M&As  −0.25  (0.31)  −0.18†  (0.09)  −0.26  (0.20)  −0.26  (0.45)  −0.19  (0.30)  0.04  (0.19)  0.13  (0.15)  Attainment discrepancy  −0.37  (1.86)  −1.05*  (0.51)  −1.13  (0.78)  −4.18†  (2.15)  −0.30  (1.90)  1.62  (1.32)  1.58  (1.04)  Size of pipelines  0.00  (0.01)  0.00  (0.00)  0.01  (0.01)  0.01  (0.01)  0.00  (0.01)  −0.02  (0.02)  −0.03*  (0.01)  Price revisions  0.82†  (0.46)  0.26  (0.31)  0.54  (0.60)  1.15  (0.96)  0.87†  (0.46)  0.74  (0.45)  0.50  (0.67)  Asset growth  −0.02  (0.34)  −0.16  (0.12)  0.03  (0.21)  −0.34  (0.41)  −0.01  (0.34)  −0.07  (0.15)  0.04  (0.19)  Potential slack  −0.02  (0.08)  −0.03  (0.04)  −0.05  (0.04)  −0.09  (0.09)  −0.02  (0.08)  −0.04  (0.08)  −0.04  (0.05)  Exploratory innovation  0.07  (0.06)  0.34***  (0.07)  0.38***  (0.10)  0.35***  (0.10)              Unabsorbed slack          −0.45†  (0.27)  −1.42*  (0.70)          0.42**  (0.16)  Unabsorbed slack, squared              0.17*  (0.08)              Constant  −0.44  (0.47)  0.33*  (0.13)          −0.39  (0.45)  1.02***  (0.16)      N firm-years  597    597    597    597    597    597    597    N Firms  37    37    37    37    37    37    37    RMSE  0.73    0.42    0.57    0.96    0.73    0.67    0.74    Log likelihood  −652    −329    −492    −801    −653    −604    −648    F  7.48    27.4    2.30    1.94    7.41    6.08    10.1    a Robust standard errors adjusted for clustering by firm are in parentheses. Two-tailed tests for all effects. † P < 0.1, *P < 0.05, **P < 0.01, ***P < 0.001. As Model 3 shows, the coefficient for absorbed slack is negative and significant (P < 0.05), indicating a negative linear association between absorbed slack and exploitative innovation (H1). We also find a support for diminishing positive effects of absorbed slack on exploratory innovation (H4) in Model 7, that shows a positive and significant (P < 0.001) coefficient for the linear term of absorbed slack and negative and significant (P < 0.001) squared term. A Wald test indicates that adding a squared term creates a statistically significant improvement in the fit of the model (P < 0.001), lending a further support for the hypothesized curvilinear relationship. In addition, our hypothesis for diminishing negative effects of unabsorbed slack on exploitative innovation (H2) is also supported by a negative and significant (P < 0.05) as well as positive and significant (P < 0.05) squared term of unabsorbed slack in Model 11. Improvement in the fit of the model with the squared term is shown as statistically significant (P < 0.05) as well. Finally, Model 14 shows a positive and significant (P < 0.01) coefficient for unabsorbed slack, lending a support for our hypothesis on a positive association between unabsorbed slack and exploratory innovation (H3). Figure 2 shows predictive margins of exploitative innovation for ±two standard deviations of organizational slack (while other covariates are held constant at their means). Because xtivreg2 routines of Stata (Schaffer, 2010) does not estimate or report a constant with the fixed-effects model, reporting raw values of predictive margins is quite misleading. Accordingly, we report standardized value of predictive margins by dividing deviations from the predicted value at the mean of organizational slack with standard deviation of exploitative innovation. Figure 3 shows standardized predictive margins of exploratory innovation following the same format. Figure 2 graphically supports H1 and H2 by showing a linear negative relationship between absorbed slack and exploitative innovation (H1), as well as a diminishing negative relationship between unabsorbed slack and exploitative innovation (H2). Figure 3 provides graphical supports for H3 and H4, in that it shows a linear positive relationship between unabsorbed slack and exploratory innovation (H3), as well as a diminishing positive relationship between absorbed slack and exploratory innovation (H4). Figure 2. View largeDownload slide Influences of organizational slack on exploitative innovation. Figure 2. View largeDownload slide Influences of organizational slack on exploitative innovation. Figure 3. View largeDownload slide Influences of organizational slack on exploratory innovation. Figure 3. View largeDownload slide Influences of organizational slack on exploratory innovation. Observed patterns are, although not explicitly hypothesized, consistent with our theoretical arguments in that the organizational slack which is more flexibly redeployed as a readily available buffer, or unabsorbed slack, exerts more salient influences on exploitative innovation than less flexibly redeployable one, i.e. absorbed slack, does. On the other hand, we hypothesized that organizational slack relaxes discipline on resource allocation to exploratory innovation because slack, be it absorbed or unabsorbed, is an indication that each dominant coalition has satisfied its claim for the share of resources as “total necessary payments” (Cyert and March, 1963: 36) to members of respective coalition. Put differently, we have no reasons to expect major differences in the salience of influences on exploratory innovation between absorbed and unabsorbed slack. Our empirical results shown in Figure 3 are consistent with our hypothesized rationale that the sense of adequacy associated with organizational slack regulates the magnitude of influences on exploratory innovation. 4. Robustness tests We checked robustness of our findings by examining two alternative model specifications. First, we added lagged dependent variables to control for any residual unobserved heterogeneity across firms leading to systematic differences in innovation performance, although the fixed-effect estimator may be inconsistent when lagged regressors are introduced in instrumental variables estimation (Cameron and Trivedi, 2009). The results show that all hypotheses are statistically supported, corroborating our original findings. We further examined our results’ robustness by employing annual increases of non-NCE-based pipelines and NCE-based pipelines (rather than raw counts, which we included as controls) as alternative measures of exploitative as well as exploratory innovation, respectively. Firm size differences are taken into account by dividing the measures by annual R&D expenses. Results shown in Appendix Tables A1 and A2 are fully consistent with our original findings, except for Model 25 (Table A2) that examines H2. Because the model reveals a weak identification largely due to the addition of a squared term of unabsorbed slack, we employ an alternative approach to examine the hypothesized curvilinear relationship by including natural logarithm of unabsorbed slack instead of its linear and squared term. The Model 26 (Table A2) shows a statistically significant negative coefficient for this alternative independent variable, indicating that the relationship between unabsorbed slack and exploitative innovation is shown in a diminishing negative slope. Overall, the results show that our findings reported above are robust. 5. Discussion In this concluding section, we discuss our contributions. First, we reconciled mutually contradictory prior findings on organizational slack by proposing a contingency perspective on the relationship between organizational slack and innovation. The form of associations between organizational slack and innovation differs across different types of innovation, as well as across different types of organizational slack. Types of innovation and organizational slack are boundary conditions on whether we observe a positive, a negative, or a curvilinear relationship. Accordingly, our argument reconciles mutually contradictory perspectives on organizational slack provided by BTOF (Cyert and March, 1963) and the agency theory (Jensen and Meckling, 1976; Fama, 1980) in a new way. In prior works, scholars who argue for an inverted U-shaped relationship employ the two theories by grafting them together in that they apply BTOF and the agency theory separately to the inverted U-shape’s left-hand side and right-hand side, respectively (Nohria and Gulati, 1996; Geiger and Cashen, 2002). Specifically, as organizational slack increases, innovation increases because latent goal conflicts between managerial coalitions are resolved (Cyert and March, 1963). However, once organizational slack exceeds a certain optimal level, managerial discipline wanes. Managers then act in self-serving manners by selecting “pet R&D projects” at the cost of “value-added innovations to firms” (Nohria and Gulati, 1996: 1248). Several opportunities to refine the prior works’ theoretical explanation motivated our hypotheses development. First, when they argue that organizational slack increases innovation by resolving latent goal conflicts, they ignore an alternative argument that managers with more organizational slack may reduce diligent effort to pursue immediate innovation opportunities, as they feel they are buffered from the competitive pressure (Thompson, 1967). Because these two arguments are mutually contradictory, it is very difficult to reconcile them unless we consider alternative types of innovation influenced by organizational slack differentially. Second, the authors underplay organizational slack’s disturbing influences on shareholder monitoring, thereby obscuring differences between absorbed slack and unabsorbed slack in influences on the extent to which managers are forced to pursue risk-taking initiatives. We build on the argument for a complementarity between BTOF and the agency theory, and extend it by showing that we need to combine (rather than graft) a theory of search (Cyert and March, 1963; Levinthal and March, 1981; Greve, 2003) and a theory of shareholder monitoring (Jensen and Meckling, 1976; Fama, 1980) when we try to explain differential influences of organizational slack on innovation. We employ the theory of search to argue that organizational slack influences exploitative innovation and exploratory innovation differentially. Predominant influences of absorbed as well as unabsorbed slack are negative for exploitative innovation because organizational slack discourages problemistic search. On the other hand, organizational slack enables exploratory innovation by increasing slack search. Put differently, to the extent that absorbed, as well as unabsorbed slack is available, organizations shift their efforts from problemistic search to slack search, or from exploitative innovation to exploratory innovation. However, alternative slack’s forms of relationship with innovation are not exactly identical. Our argument theorizes about the differences between absorbed slack and unabsorbed slack by adjusting their predominant influences (derived from the search theory) with the differential degree of difficulties associated with monitoring by shareholders. Put differently, given the type of innovation considered, whether the relationship is linear or curvilinear depends on the type of organizational slack examined. We argue that absorbed slack is associated with more substantial monitoring challenges for shareholders than unabsorbed slack is. Monitoring appropriate usages of organizational slack is challenging for external stakeholders to the extent that constraints placed on the usage of organizational slack is idiosyncratic to organizational contexts. Without understanding such idiosyncratic contexts, it is very difficult to judge whether focal slack is appropriately used. Therefore, monitoring the usages of organizational slack is more challenging for shareholders to the extent that the focal slack is absorbed to particular expenses with idiosyncratic usages. On the contrary, unabsorbed slack is not constrained by such idiosyncrasy, thereby enabling more effective shareholder monitoring, as it grows more substantial and salient. Because those influences of organizational slack on shareholder monitoring increase as organizational slack increases, predominant influences of organizational slack on search diminish as more organizational slack is available. Specifically, absorbed slack disturbs efforts of shareholders to monitor their managers’ extent of risk-taking, thereby attenuating its positive relationship with exploratory innovation (H4). On the other hand, unabsorbed slack weakens its negative relationship with exploitative innovation by enabling shareholders to monitor more effectively (H2). In short, the type of slack examined is the second contingency that governs the form of the relationship between organizational slack and innovation. The theory of shareholder monitoring enables us to more appropriately explain influences of organizational slack by distinguishing the two types of organizational slack. As such, it is interesting to pursue implications of our findings on some important research fields related to studies of innovation. First, our findings inform the research on search (Cyert and March, 1963; Rosenkopf and Nerkar, 2001; Katila and Ahuja, 2002; Laursen and Salter, 2006; Lopez-Vega et al., 2016). We build our argument on the premise that search is an important precursor of innovation (Laursen, 2012), and accordingly an important contingency which characterizes differential types of innovation. The prior work on search focuses on the intensity of search (or resultant innovation), while efforts to examine search (or innovation) by acknowledging the existence of several distinct types of search are relatively limited (exceptions include Laursen (2012) or Lopez-Vega et al. (2016)). We employed distinct types of search as contingencies that explain differential relationships between organizational slack and innovation. Empirical supports to our argument that types of search regulate influences of organizational slack on resultant innovation attest to the importance of the distinction between different types of search. Furthermore, a perspective that dichotomizes problemistic search (Cyert and March, 1963) and slack search (Levinthal and March, 1981; Greve, 2003) is informed of the possibility that these two types of search are pursued as substitutes in that organizational slack encourages the latter while discourages the former as slack allows organizational decision makers to shift their locus of attention from attainment discrepancy to distant profit potential. Stated differently, our underlying rationale to relate organizational slack to differential types of search is premised on the attention-based theorizing (Ocasio, 1997) because search presupposes certain focus of attention by organizational decision makers. Accordingly, it also is important to note our findings’ implications for a theory of managerial attention (Ocasio, 1997; Joseph and Ocasio, 2012; Li et al., 2013). We argue that organizational slack provides buffers against environmental changes that influence exploitative innovation and exploratory innovation oppositely. Specifically, buffered organizations are less motivated to address current performance problems by way of exploitative innovation. In contrast, the same buffer encourages organizations to invest in exploratory innovation for possible future performance gains. Put differently, organizational slack influences managers’ temporal orientation (Souder and Shaver, 2010; Bromiley and Washburn, 2011), thereby allowing managers to shift their attention from “current viability” toward “future viability” (Levinthal and March, 1993). Accordingly, our findings indicate that the relationship between organizational slack and innovation initiatives is mediated by the distribution of managers’ attention across short-term concerns and long-term concerns. The prior work (March and Shapira, 1992; Audia and Greve, 2006; Chen and Miller, 2007; Iyer and Miller, 2008; Lungeanu et al., 2016) relates organizational slack to shifts in the focus of attention by showing that slack influences managers’ choice from their aspiration level, survival point, and organizational slack. Our findings extend the prior work by acknowledging more general roles of organizational slack as a regulator of attention (Ocasio, 1997). Specifically, we conjecture that organizational slack regulates the shift in the locus of attention of decision makers across alternative reference points, as well as across alternative timeframes they could employ. Put differently, organizational slack may regulate attention of decision makers beyond the context of performance feedback. As such, our findings call attention of scholars on the potential roles of organizational slack as a general regulator of attention, thereby informing the literature of attention about an influential antecedent of attention. Irrespective of the contributions, there are several limitations. First, we were not able to examine the way in which measures to reduce agency costs moderate the relationship between organizational slack and innovation within the scope of current research. Given that alternative types of organizational slack are associated with different degrees of agency costs, several measures to reduce agency costs—including managerial stock holding (Jensen and Meckling, 1976; Walkling and Long, 1984; Eisenhardt, 1989) and stock concentration (Hill and Snell, 1989; Baysinger et al., 1991; Francis and Smith, 1995)—may attenuate differences observed across different types of organizational slack. For example, we argue that absorbed slack and unabsorbed slack reveal different forms of the relationship with innovation because stakeholders experience more difficulty to monitor absorbed slack than unabsorbed slack. Accordingly, the extensive disclosure of usages of absorbed slack may render such differences in the form of the relationships less salient. One example of such disclosure may include explicitly disclosing the usage of general and administrative expenses because the expenses are generally characterized as highly discretionary. More specifically, it is expected that marginal negative influences of absorbed slack on exploitative innovation, which we hypothesized as constant, gradually diminish to the extent that firms disclose their usages of absorbed slack more extensively. As for the influences of absorbed slack on exploratory innovation, we expect to observe hypothesized diminishing positive influences less explicitly to the extent that firms disclose more extensively. Put differently, negative influences on exploitative innovation are attenuated, while positive influences on exploratory innovation are reinforced. Accordingly, search, and resultant innovation, is expected grow more intensive as more information on the usage of absorbed slack is disclosed. Empirical examination of such influences of increased disclosure on the differences in effects between alternative organizational slack, not only replicates our findings but also opens up one of promising lines of inquiry on the relationship between organizational slack and innovation. Another straightforward avenue to extend our argument would be to apply our findings to the relationship between organizational slack and organizational performance. Furthermore, examining long-term influences of organizational slack on subsequent innovation may be worthwhile. It also is important to note that our results may suffer from the omitted variables bias, as alternative slack variables are separately employed. We had to examine them separately because we felt it more theoretically grounded to model them as endogenous variables with distinct sets of instruments. Although it is difficult to completely deny concerns on omitted variables bias, we employed several measures to alleviate the concerns. First, we employed instrumental variable method, which is recommended when one suspects the existence of omitted variables bias (Bascle, 2008). Second, we also employed fixed-effect models, which effectively remove the influences of firm-specific effects including time invariant portion of organizational slack. Finally, usual caveats of external validity apply as well, i.e. we encourage researchers to replicate our findings in other empirical contexts. For example, in a context where performance-related payment is more widely adopted than our empirical contexts, i.e. Japan, managers are more willing to pursue risk-taking initiatives, thereby rendering differences in effects of absorbed and unabsorbed slack less explicit. Furthermore, as our argument is built on the assumption that innovation is closely associated with organizational search, our findings may not be replicated in a context where innovation is institutionalized. It also may be conceivable that our argument is not applied to firms held by governments or business partners, as these shareholders may not necessarily pursue maximizing financial returns from their investment. It is very difficult to identify organizations with no organizational slack. Furthermore, organizational slack influences various aspects of organizational phenomena, including coalition building (Cyert and March, 1963), political behaviors (Bourgeois and Singh, 1983), risk-taking (Singh, 1986; Bromiley, 1991), competitive responses (Thompson, 1967; Smith et al., 1991), and profitability (Tan and Peng, 2003; George, 2005; Wang et al., 2014). However, our understanding of organizational slack is still limited, fragmented, and confused. We hope our arguments stimulate further discussion on this interesting construct. Acknowledgements The author is grateful for the valuable comments and suggestions of Editor Fredrik Tell, Paul Nightingale, and anonymous reviewers. Errors and omissions remain the author’s own. References Abernathy W. J. ( 1978), The Productivity Dilemma: Roadblock to Innovation in the Automobile Industry . Johns Hopkins University Press: Baltimore, MD. Argyris C. ( 1976), ‘ Single-loop and double-loop models in research on decision making,’ Administrative Science Quarterly , 21( 3), 363– 375. Google Scholar CrossRef Search ADS   Audia P. G., Greve H. R. ( 2006), ‘ Less likely to fail: low performance, firm size, and factory expansion in the shipbuilding industry,’ Management Science , 52( 1), 83– 94. Google Scholar CrossRef Search ADS   Bascle G. ( 2008), ‘ Controlling for endogeneity with instrumental variables in strategic management research,’ Strategic Organization , 6( 3), 285– 327. Google Scholar CrossRef Search ADS   Baysinger B. D., Kosnik R. D., Turk T. A. ( 1991), ‘ Effects of board and ownership structure on corporate R&D strategy,’ Academy of Management Journal , 34( 1), 205– 214. Benner M. J., Tushman M. ( 2002), ‘ Process management and technological innovation: A longitudinal study of the photography and paint industries,’ Administrative Science Quarterly , 47( 4), 676– 706. Google Scholar CrossRef Search ADS   Bierly P., Chakrabarti A. ( 1996), ‘ Generic knowledge strategies in the U.S. pharmaceutical industry,’ Strategic Management Journal , 17( S2), 123– 135. Google Scholar CrossRef Search ADS   Bourgeois L. J.III. ( 1981), ‘ On the measurement of organizational slack,’ Academy of Management Review , 6( 1), 29– 39. Google Scholar CrossRef Search ADS   Bourgeois L. J.III, Singh J. V. ( 1983), ‘ Organizational slack and political behavior among top management teams,’ Academy of Management Proceedings , 1983( 1), 43– 47. Google Scholar CrossRef Search ADS   Bromiley P. ( 1991), ‘ Testing a causal model of corporate risk taking and performance,’ Academy of Management Journal , 34( 1), 37– 59. Bromiley P., Washburn M. ( 2011), ‘ Cost reduction vs innovative search in R&D,’ Journal of Strategy and Management , 4( 3), 196– 214. Google Scholar CrossRef Search ADS   Cameron A. C., Trivedi P. K. ( 2009), Microeconometrics Using Stata , revised ed. Stata Press: College Station. Cardinal L. B. ( 2001), ‘ Technological innovation in the pharmaceutical industry: the use of organizational control in managing research and development,’ Organization Science , 12( 1), 19– 36. Google Scholar CrossRef Search ADS   Carnabuci G., Operti E. ( 2013), ‘ Where do firms' recombinant capabilities come from? Intraorganizational networks, knowledge, and firms' ability to innovate through technological recombination,’ Strategic Management Journal , 34( 13), 1591– 1613. Google Scholar CrossRef Search ADS   Chen W.-R., Miller K. D. ( 2007), ‘ Situational and institutional determinants of firms' R&D search intensity,’ Strategic Management Journal , 28( 4), 369– 381. Google Scholar CrossRef Search ADS   Cheng J. L. C., Kesner I. F. ( 1997), ‘ Organizational slack and response to environmental shifts: The impact of resource allocation patterns,’ Journal of Management , 23( 1), 1– 18. Google Scholar CrossRef Search ADS   Christensen C. M., Bower J. L. ( 1996), ‘ Customer power, strategic investment, and the failure of leading firms,’ Strategic Management Journal , 17( 3), 197– 218. Google Scholar CrossRef Search ADS   Cohen J., Cohen P., West S. G., Aiken L. S. ( 2003), Applied Multiple Regression: Correlation Analysis for the Behavioral Sciences . Lawrence Erlbaum Associates: Mahwah, NJ. Cooper A. C., Smith C. G. ( 1992), ‘ How established firms respond to threatening technologies,’ Academy of Management Executive , 6( 2), 55– 70. Cragg J. G., Donald S. G. ( 1993), ‘ Testing identfiability and specification in instrumental variables models,’ Econometric Theory , 9( 02), 222– 240. Google Scholar CrossRef Search ADS   Cyert R., March J. G. ( 1963), A Behavioral Theory of the Firm . Prentice Hall: Englewood Cliffs, NJ. Danneels E. ( 2008), ‘ Organizational antecedents of second-order competences,’ Strategic Management Journal , 29( 5), 519– 543. Google Scholar CrossRef Search ADS   Deephouse D. L., Wiseman R. M. ( 2000), ‘ Comparing alternative explanations for accounting risk-return relations,’ Journal of Economic Behavior and Organization , 42( 4), 463– 482. Google Scholar CrossRef Search ADS   Dunlap-Hinkler D., Kotabe M., Mudambi R. ( 2010), ‘ A story of breakthrough versus incremental innovation: Corporate entrepreneurship in the global pharmaceutical industry,’ Strategic Entrepreneurship Journal , 4( 2), 106– 127. Google Scholar CrossRef Search ADS   Eisenhardt K. M. ( 1989), ‘ Agency theory: An assessment and review,’ The Academy of Management Review , 14( 1), 57– 74. Google Scholar CrossRef Search ADS   Fama E. F. ( 1980), ‘ Agency problems and the theory of the firm,’ Journal of Political Economy , 88( 2), 288– 307. Google Scholar CrossRef Search ADS   Francis J., Smith A. ( 1995), ‘ Agency costs and innovation some empirical evidence,’ Journal of Accounting and Economics , 19( 2–3), 383– 409. Google Scholar CrossRef Search ADS   Geiger S. W., Cashen L. H. ( 2002), ‘ A multidimensional examination of slack and its impact on innovation,’ Journal of Managerial Issues , 14( 1), 68– 84. George G. ( 2005), ‘ Slack resources and the performance of privately held firms,’ Academy of Management Journal , 48( 4), 661– 676. Google Scholar CrossRef Search ADS   Greve H. R. ( 2003), ‘ A behavioral theory of R&D expenditures and innovations: Evidence from shipbuilding,’ Academy of Management Journal , 46( 6), 685– 702. Greve H. R. ( 2007), ‘ Exploration and exploitation in product innovation,’ Industrial and Corporate Change , 16( 5), 945– 975. Google Scholar CrossRef Search ADS   Hansen L. P. ( 1982), ‘ Large sample properties of generalized method of moments estimators,’ Econometrica , 50( 4), 1029– 1054. Google Scholar CrossRef Search ADS   Hansen L. P., Heaton J., Yaron A. ( 1996), ‘ Finite-sample properties of some alternative GMM estimators,’ Journal of Business and Economic Statistics , 14( 3), 262– 280. Hill C. W. L., Snell S. A. ( 1989), ‘ Effects of ownership structure and control on corporate productivity,’ Academy of Management Journal , 32( 1), 25– 46. Iyer D. N., Miller K. D. ( 2008), ‘ Performance Feedback, Slack, and the Timing of Acquisitions,’ The Academy of Management Journal , 51( 4), 808– 822. Jensen M. C. ( 1986), ‘ Agency costs of free cash flow, corporate finance, and takeovers,’ The American Economic Review , 76( 2), 323– 329. Jensen M. C., Meckling W. H. ( 1976), ‘ Theory of the firm: Managerial behavior, agency costs and ownership structure,’ Journal of Financial Economics , 3( 4), 305– 360. Google Scholar CrossRef Search ADS   Joseph J., Ocasio W. C. ( 2012), ‘ Architecture, attention, and adaptation in the multibusiness firm: General Electric from 1951 to 2001,’ Strategic Management Journal , 33( 6), 633– 660. Google Scholar CrossRef Search ADS   Kacperczyk A., Beckman C. M., Moliterno T. P. ( 2015), ‘ Disentangling risk and change: Internal and external social comparison,’ In the Mutual Fund Industry', Administrative Science Quarterly , 60( 2), 228– 262. Google Scholar CrossRef Search ADS   Katila R., Ahuja G. ( 2002), ‘ Something old, something new: a longitudinal study of search behavior and new product introduction,’ Academy of Management Journal , 45( 6), 1183– 1195. Kim H., Kim H., Lee P. M. ( 2008), ‘ Ownership structure and the relationship between financial slack and R&D investments: evidence from Korean firms,’ Organization Science , 19( 3), 404– 418. Google Scholar CrossRef Search ADS   Knudsen T., Levinthal D. A. ( 2007), ‘ Two faces of search: alternative generation and alternative evaluation,’ Organization Science , 18( 1), 39– 54. Google Scholar CrossRef Search ADS   Lant T. K. ( 1992), ‘ Aspiration level adaptation: an empirical exploration,’ Management Science , 38( 5), 623– 644. Google Scholar CrossRef Search ADS   Latham S. F., Braun M. ( 2009), ‘ Managerial risk, innovation, and organizational decline,’ Journal of Management , 35( 2), 258– 281. Google Scholar CrossRef Search ADS   Laursen K. ( 2012), ‘ Keep searching and you’ll find: what do we know about variety creation through firms’ search activities for innovation?,’ Industrial and Corporate Change , 21( 5), 1181– 1220. Google Scholar CrossRef Search ADS   Laursen K., Salter A. ( 2006), ‘ Open for innovation: the role of openness in explaining innovation performance among U.K. manufacturing firms,’ Strategic Management Journal , 27( 2), 131– 150. Google Scholar CrossRef Search ADS   Lecuona J. R., Reitzig M. ( 2014), ‘ Knowledge worth having in ‘excess’: the value of tacit and firm-specific human resource slack,’ Strategic Management Journal , 35( 7), 954– 973. Google Scholar CrossRef Search ADS   Lee J., Lee J., Lee H. ( 2003), ‘ Exploration and exploitation in the presence of network externalities,’ Management Science , 49( 4), 553– 570. Google Scholar CrossRef Search ADS   Levinthal D. A. ( 1997), ‘ Adaptation on rugged landscapes,’ Management Science , 43( 7), 934– 950. Google Scholar CrossRef Search ADS   Levinthal D. A., March J. G. ( 1981), ‘ A model of adaptive organizational search,’ Journal of Economic Behavior and Organization , 2( 4), 307– 333. Google Scholar CrossRef Search ADS   Levinthal D. A., March J. G. ( 1993), ‘ The myopia of learning,’ Strategic Management Journal , 14( S2), 95– 112. Google Scholar CrossRef Search ADS   Li Q., Maggitti P. G., Smith K. G., Tesluk P. E., Katila R. ( 2013), ‘ Top management attention to innovation: The role of search selection and intensity in new product introductions,’ Academy of Management Journal , 56( 3), 893– 916. Google Scholar CrossRef Search ADS   Litschert R. J., Bonham T. W. ( 1978), ‘ A conceptual model of strategy formation,’ Academy of Management Review , 3( 2), 211– 219. Google Scholar CrossRef Search ADS   Lopez-Vega H., Tell F., Vanhaverbeke W. ( 2016), ‘ Where and how to search? Search paths in open innovation,’ Research Policy , 45( 1), 125– 136. Google Scholar CrossRef Search ADS   Love G. E., Nohria N. ( 2005), ‘ Reducing slack: the performance consequences of downsizing by large industrial firms, 1977–93,’ Strategic Management Journal , 26( 12), 1087– 1108. Google Scholar CrossRef Search ADS   Lungeanu R., Stern I., Zajac E. J. ( 2016), ‘ When do firms change technology-sourcing vehicles? The role of poor innovative performance and financial slack,’ Strategic Management Journal , 37( 5), 855– 869. Google Scholar CrossRef Search ADS   Malatesta P. H., Walkling R. A. ( 1988), ‘ Poison pill securities: Stockholder wealth, profitability, and ownership structure,’ Journal of Financial Economics , 20( 0), 347– 376. Google Scholar CrossRef Search ADS   March J. G. ( 1981), ‘Decisions in organizations and theories of choice’, in Van de Ven A. H., Joyce W. F. (eds), Perspectives on Organization Design and Behavior . Wiley: New York, NY, pp. 205– 248. March J. G. ( 1991), ‘ Exploration and exploitation in organizational learning,’ Organization Science , 2( 1), 71– 87. Google Scholar CrossRef Search ADS   March J. G. ( 1994), A Primer on Decision Making: How Decisions Happen . Free Press: New York, NY. March J. G. ( 2006), ‘ Rationality, Foolishness, and Adaptive Intelligence,’ Strategic Management Journal , 27( 3), 201– 214. Google Scholar CrossRef Search ADS   March J. G. ( 2007), ‘ Perspective—Scholarship, Scholarly Institutions, and Scholarly Communities,’ Organization Science , 18( 3), 537– 542. Google Scholar CrossRef Search ADS   March J. G., Shapira Z. ( 1992), ‘ Variable risk preferences and the focus of attention,’ Psychological Review , 99( 1), 172– 183. Google Scholar CrossRef Search ADS   Meyer A. D. ( 1982), ‘ Adapting to environmental jolts,’ Administrative Science Quarterly , 27( 4), 515– 537. Google Scholar CrossRef Search ADS   Mishina Y., Pollock T. G., Porac J. F. ( 2004), ‘ Are more resources always better for growth? Resource stickiness in market and product expansion,’ Strategic Management Journal , 25( 12), 1179– 1197. Google Scholar CrossRef Search ADS   Murray M. P. ( 2006), ‘ Avoiding invalid instruments and coping with weak instruments,’ Journal of Economic Perspectives , 20( 4), 111– 132. Google Scholar CrossRef Search ADS   Natividad G. ( 2013), ‘ Financial slack, strategy, and competition in movie distribution,’ Organization Science , 24( 3), 846– 864. Google Scholar CrossRef Search ADS   Nohria N., Gulati R. ( 1996), ‘ Is slack good or bad for innovation?,’ Academy of Management Journal , 39( 5), 1245– 1264. Google Scholar CrossRef Search ADS   O'brien J. P. ( 2003), ‘ The capital structure implications of pursuing a strategy of innovation,’ Strategic Management Journal , 24( 5), 415– 431. Google Scholar CrossRef Search ADS   Ocasio W. C. ( 1997), ‘ Towards an attention-based view of the firm,’ Strategic Management Journal , 18( 1), 187– 206. Google Scholar CrossRef Search ADS   Palmer T. B., Wiseman R. M. ( 1999), ‘ Decoupling risk taking from income stream uncertainty: a holistic model of risk,’ Strategic Management Journal , 20( 11), 1037– 1062. Google Scholar CrossRef Search ADS   Piao M. ( 2010), ‘ Thriving in the new: implication of exploration on organizational longevity,’ Journal of Management , 36( 6), 1529– 1554. Google Scholar CrossRef Search ADS   Reuer J. J., Leiblein M. J. ( 2000), ‘ Downside risk implications of multinationality and international joint ventures,’ Academy of Management Journal , 43( 2), 203– 214. Rosenkopf L., Nerkar A. ( 2001), ‘ Beyond local search: boundary-spanning, exploration, and impact in the optical disk industry,’ Strategic Management Journal , 22( 4), 287– 306. Google Scholar CrossRef Search ADS   Sørensen J. B., Stuart T. E. ( 2000), ‘ Aging, obsolescence, and organizational innovation,’ Administrative Science Quarterly , 45( 1), 81– 112. Google Scholar CrossRef Search ADS   Schaffer M. E. ( 2010), ‘xtivreg2: Stata module to perform extended IV/2SLS, GMM and AC/HAC, LIML and k-class regression for panel data models,’ http://ideas.repec.org/c/boc/bocode/s456501.html. Sharfman M. P., Wolf G., Chase R. B., Tansik D. A. ( 1988), ‘ Antecedents of organizational slack,’ Academy of Management Review , 13( 4), 601– 614. Google Scholar CrossRef Search ADS   Simon H. A. ( 1955), ‘ A behavioral model of rational choice,’ The Quarterly Journal of Economics , 69( 1), 99– 118. Google Scholar CrossRef Search ADS   Singh J. V. ( 1986), ‘ Performance, slack, and risk taking in organizational decision making,’ Academy of Management Journal , 29( 3), 562– 585. Smith K. G., Grimm C. M., Gannon M. J., Chen M.-J. ( 1991), ‘ Organizational information processing, competitive responses, and performance in the U.S. domestic airline industry,’ Academy of Management Journal , 34( 1), 60– 85. Souder D., Shaver J. M. ( 2010), ‘ Constraints and incentives for making long horizon corporate investments,’ Strategic Management Journal , 31( 12), 1316– 1336. Google Scholar CrossRef Search ADS   Steensma H. K., Corley K. G. ( 2001), ‘ Organizational context as a moderator of theories on firm boundaries for technology sourcing,’ Academy of Management Journal , 44( 2), 271– 291. Stuart T. E., Podolny J. M. ( 1996), ‘ Local search and the evolution of technological capabilities,’ Strategic Management Journal , 17( S1), 21– 38. Google Scholar CrossRef Search ADS   Tan J., Peng M. W. ( 2003), ‘ Organizational slack and firm performance during economic transitions: Two studies from an emerging economy,’ Strategic Management Journal , 24( 13), 1249– 1263. Google Scholar CrossRef Search ADS   Thompson J. D. ( 1967), Organizations in Action: Social Science Bases of Administrative Theory . McGraw-Hill Inc.: New York, NY. Voss G. B., Sirdeshmukh D., Voss Z. G. ( 2008), ‘ The effects of slack resources and environmental threat on product exploration and exploitation,’ Academy of Management Journal , 51( 1), 147– 164. Google Scholar CrossRef Search ADS   Walkling R. A., Long M. S. ( 1984), ‘ Agency theory, managerial welfare, and takeover bid resistance,’ The RAND Journal of Economics , 15( 1), 54– 68. Google Scholar CrossRef Search ADS   Wang D., Sun D., Yu X., Zhang Y. ( 2014), ‘ The impact of CEO duality and ownership on the relationship between organisational slack and firm performance in China,’ Systems Research and Behavioral Science , 31( 1), 94– 101. Google Scholar CrossRef Search ADS   Wang H., Li J. ( 2008), ‘ Untangling the effects of overexploration and overexploitation on organizational performance: The moderating role of environmental dynamism,’ Journal of Management , 34( 5), 925– 951. Google Scholar CrossRef Search ADS   Watzlawick P., Weakland J. H., Fisch R. ( 1974), Change: Principles of Problem Formation and Problem Resolution . W.W. Norton: New York, NY. Wiseman R. M., Bromiley P. ( 1996), ‘ Toward a model of risk in declining organizations: an empirical examination of risk, performance and decline,’ Organization Science , 7( 5), 524– 543. Google Scholar CrossRef Search ADS   Wiseman R. M., Catanach C. ( 1997), ‘ A longitudinal disaggregation of operational risk under changing regulations: evidence from the savings and loan industry,’ Academy of Management Journal , 40( 4), 799– 830. Wooldridge J. M. ( 2010), Econometric Analysis of Cross Section and Panel Data , 2nd edn. MIT Press: Cambridge, MA. Yasai-Ardekani M. ( 1986), ‘ Structural adaptations to environments,’ Academy of Management Review , 11( 1), 9– 21. Google Scholar CrossRef Search ADS   Appendix Table A1. Results of the panel continuous-updating estimator (CUE) instrumental variable regression analysis for the effects of absorbed slack on innovation increasesa Dependent variables  Absorbed slack   Exploitative innovation   Absorbed slack   Exploratory innovation   Model 15  Model 16  Model 17  Model 18  Model 19  Model 20  Model 21  R&D intensity  0.33**  (0.10)          0.33**  (0.10)              Total assets  0.05*  (0.02)          0.05*  (0.02)              Organizational size  0.02  (0.04)  0.01  (0.06)  0.01  (0.06)  0.02  (0.03)  −0.03  (0.06)  −0.03  (0.05)  −0.12  (0.08)  R&D  0.00  (0.00)  0.06†  (0.04)  0.17*  (0.07)  0.00  (0.01)  0.11**  (0.03)  0.09***  (0.02)  0.06†  (0.03)  Age  −0.06*  (0.03)  −0.05  (0.06)  −0.12  (0.09)  −0.07*  (0.03)  −0.07  (0.13)  0.02  (0.09)  0.12  (0.12)  M&As  0.01  (0.02)  −0.13  (0.13)  −0.09  (0.16)  0.02  (0.02)  0.01  (0.13)  −0.02  (0.15)  −0.01  (0.15)  Attainment discrepancy  0.18  (0.19)  −1.39*  (0.55)  −1.12**  (0.43)  0.18  (0.19)  −0.05  (0.48)  −0.47  (0.45)  −0.68  (0.61)  Size of pipelines  0.00  (0.00)  0.01  (0.01)  0.00  (0.01)  0.00  (0.00)  −0.02  (0.01)  −0.02  (0.01)  −0.03  (0.02)  Price revisions  0.01  (0.03)  −0.20  (0.26)  −0.20  (0.40)  0.01  (0.03)  0.09  (0.42)  −0.10  (0.47)  0.04  (0.49)  Asset growth  0.01  (0.05)  0.23*  (0.09)  0.14†  (0.08)  0.02  (0.05)  0.03  (0.07)  0.02  (0.08)  0.07  (0.12)  Potential slack  0.05  (0.04)  0.07  (0.09)  0.14†  (0.08)  0.05  (0.04)  0.02  (0.08)  0.12  (0.07)  −0.15  (0.16)  Non-NCE pipelines  0.00  (0.00)  −0.02***  (0.01)  −0.03***  (0.01)                  NCE pipelines              0.00  (0.00)  −0.02***  (0.01)  −0.02***  (0.00)  −0.02***  (0.00)  Exploratory innovation  0.00  (0.00)  0.27***  (0.06)  0.28*  (0.12)                  Absorbed slack          −1.28*  (0.52)          −0.15  (0.37)  3.92*  (1.78)  Absorbed slack, squared                          −5.28*  (2.30)  Constant  −0.31  (0.27)  −0.21  (0.53)      −0.29  (0.26)  0.49  (0.45)          N firm-years  581    581    581    581    581    581    581    N Firms  37    37    37    37    37    37    37    RMSE  0.04    0.39    0.40    0.04    0.39    0.40    0.43    Log likelihood  1000    −270    −276    1000    −277    −279    −312    F  34.5     4.46     5.52     37.2     3.19     3.73     3.11     Dependent variables  Absorbed slack   Exploitative innovation   Absorbed slack   Exploratory innovation   Model 15  Model 16  Model 17  Model 18  Model 19  Model 20  Model 21  R&D intensity  0.33**  (0.10)          0.33**  (0.10)              Total assets  0.05*  (0.02)          0.05*  (0.02)              Organizational size  0.02  (0.04)  0.01  (0.06)  0.01  (0.06)  0.02  (0.03)  −0.03  (0.06)  −0.03  (0.05)  −0.12  (0.08)  R&D  0.00  (0.00)  0.06†  (0.04)  0.17*  (0.07)  0.00  (0.01)  0.11**  (0.03)  0.09***  (0.02)  0.06†  (0.03)  Age  −0.06*  (0.03)  −0.05  (0.06)  −0.12  (0.09)  −0.07*  (0.03)  −0.07  (0.13)  0.02  (0.09)  0.12  (0.12)  M&As  0.01  (0.02)  −0.13  (0.13)  −0.09  (0.16)  0.02  (0.02)  0.01  (0.13)  −0.02  (0.15)  −0.01  (0.15)  Attainment discrepancy  0.18  (0.19)  −1.39*  (0.55)  −1.12**  (0.43)  0.18  (0.19)  −0.05  (0.48)  −0.47  (0.45)  −0.68  (0.61)  Size of pipelines  0.00  (0.00)  0.01  (0.01)  0.00  (0.01)  0.00  (0.00)  −0.02  (0.01)  −0.02  (0.01)  −0.03  (0.02)  Price revisions  0.01  (0.03)  −0.20  (0.26)  −0.20  (0.40)  0.01  (0.03)  0.09  (0.42)  −0.10  (0.47)  0.04  (0.49)  Asset growth  0.01  (0.05)  0.23*  (0.09)  0.14†  (0.08)  0.02  (0.05)  0.03  (0.07)  0.02  (0.08)  0.07  (0.12)  Potential slack  0.05  (0.04)  0.07  (0.09)  0.14†  (0.08)  0.05  (0.04)  0.02  (0.08)  0.12  (0.07)  −0.15  (0.16)  Non-NCE pipelines  0.00  (0.00)  −0.02***  (0.01)  −0.03***  (0.01)                  NCE pipelines              0.00  (0.00)  −0.02***  (0.01)  −0.02***  (0.00)  −0.02***  (0.00)  Exploratory innovation  0.00  (0.00)  0.27***  (0.06)  0.28*  (0.12)                  Absorbed slack          −1.28*  (0.52)          −0.15  (0.37)  3.92*  (1.78)  Absorbed slack, squared                          −5.28*  (2.30)  Constant  −0.31  (0.27)  −0.21  (0.53)      −0.29  (0.26)  0.49  (0.45)          N firm-years  581    581    581    581    581    581    581    N Firms  37    37    37    37    37    37    37    RMSE  0.04    0.39    0.40    0.04    0.39    0.40    0.43    Log likelihood  1000    −270    −276    1000    −277    −279    −312    F  34.5     4.46     5.52     37.2     3.19     3.73     3.11     a Robust standard errors adjusted for clustering by firm are in parentheses. Two-tailed tests for all effects. † p <0.1; *p <0.05; **p <0.01; ***p <0.001. Table A1. Results of the panel continuous-updating estimator (CUE) instrumental variable regression analysis for the effects of absorbed slack on innovation increasesa Dependent variables  Absorbed slack   Exploitative innovation   Absorbed slack   Exploratory innovation   Model 15  Model 16  Model 17  Model 18  Model 19  Model 20  Model 21  R&D intensity  0.33**  (0.10)          0.33**  (0.10)              Total assets  0.05*  (0.02)          0.05*  (0.02)              Organizational size  0.02  (0.04)  0.01  (0.06)  0.01  (0.06)  0.02  (0.03)  −0.03  (0.06)  −0.03  (0.05)  −0.12  (0.08)  R&D  0.00  (0.00)  0.06†  (0.04)  0.17*  (0.07)  0.00  (0.01)  0.11**  (0.03)  0.09***  (0.02)  0.06†  (0.03)  Age  −0.06*  (0.03)  −0.05  (0.06)  −0.12  (0.09)  −0.07*  (0.03)  −0.07  (0.13)  0.02  (0.09)  0.12  (0.12)  M&As  0.01  (0.02)  −0.13  (0.13)  −0.09  (0.16)  0.02  (0.02)  0.01  (0.13)  −0.02  (0.15)  −0.01  (0.15)  Attainment discrepancy  0.18  (0.19)  −1.39*  (0.55)  −1.12**  (0.43)  0.18  (0.19)  −0.05  (0.48)  −0.47  (0.45)  −0.68  (0.61)  Size of pipelines  0.00  (0.00)  0.01  (0.01)  0.00  (0.01)  0.00  (0.00)  −0.02  (0.01)  −0.02  (0.01)  −0.03  (0.02)  Price revisions  0.01  (0.03)  −0.20  (0.26)  −0.20  (0.40)  0.01  (0.03)  0.09  (0.42)  −0.10  (0.47)  0.04  (0.49)  Asset growth  0.01  (0.05)  0.23*  (0.09)  0.14†  (0.08)  0.02  (0.05)  0.03  (0.07)  0.02  (0.08)  0.07  (0.12)  Potential slack  0.05  (0.04)  0.07  (0.09)  0.14†  (0.08)  0.05  (0.04)  0.02  (0.08)  0.12  (0.07)  −0.15  (0.16)  Non-NCE pipelines  0.00  (0.00)  −0.02***  (0.01)  −0.03***  (0.01)                  NCE pipelines              0.00  (0.00)  −0.02***  (0.01)  −0.02***  (0.00)  −0.02***  (0.00)  Exploratory innovation  0.00  (0.00)  0.27***  (0.06)  0.28*  (0.12)                  Absorbed slack          −1.28*  (0.52)          −0.15  (0.37)  3.92*  (1.78)  Absorbed slack, squared                          −5.28*  (2.30)  Constant  −0.31  (0.27)  −0.21  (0.53)      −0.29  (0.26)  0.49  (0.45)          N firm-years  581    581    581    581    581    581    581    N Firms  37    37    37    37    37    37    37    RMSE  0.04    0.39    0.40    0.04    0.39    0.40    0.43    Log likelihood  1000    −270    −276    1000    −277    −279    −312    F  34.5     4.46     5.52     37.2     3.19     3.73     3.11     Dependent variables  Absorbed slack   Exploitative innovation   Absorbed slack   Exploratory innovation   Model 15  Model 16  Model 17  Model 18  Model 19  Model 20  Model 21  R&D intensity  0.33**  (0.10)          0.33**  (0.10)              Total assets  0.05*  (0.02)          0.05*  (0.02)              Organizational size  0.02  (0.04)  0.01  (0.06)  0.01  (0.06)  0.02  (0.03)  −0.03  (0.06)  −0.03  (0.05)  −0.12  (0.08)  R&D  0.00  (0.00)  0.06†  (0.04)  0.17*  (0.07)  0.00  (0.01)  0.11**  (0.03)  0.09***  (0.02)  0.06†  (0.03)  Age  −0.06*  (0.03)  −0.05  (0.06)  −0.12  (0.09)  −0.07*  (0.03)  −0.07  (0.13)  0.02  (0.09)  0.12  (0.12)  M&As  0.01  (0.02)  −0.13  (0.13)  −0.09  (0.16)  0.02  (0.02)  0.01  (0.13)  −0.02  (0.15)  −0.01  (0.15)  Attainment discrepancy  0.18  (0.19)  −1.39*  (0.55)  −1.12**  (0.43)  0.18  (0.19)  −0.05  (0.48)  −0.47  (0.45)  −0.68  (0.61)  Size of pipelines  0.00  (0.00)  0.01  (0.01)  0.00  (0.01)  0.00  (0.00)  −0.02  (0.01)  −0.02  (0.01)  −0.03  (0.02)  Price revisions  0.01  (0.03)  −0.20  (0.26)  −0.20  (0.40)  0.01  (0.03)  0.09  (0.42)  −0.10  (0.47)  0.04  (0.49)  Asset growth  0.01  (0.05)  0.23*  (0.09)  0.14†  (0.08)  0.02  (0.05)  0.03  (0.07)  0.02  (0.08)  0.07  (0.12)  Potential slack  0.05  (0.04)  0.07  (0.09)  0.14†  (0.08)  0.05  (0.04)  0.02  (0.08)  0.12  (0.07)  −0.15  (0.16)  Non-NCE pipelines  0.00  (0.00)  −0.02***  (0.01)  −0.03***  (0.01)                  NCE pipelines              0.00  (0.00)  −0.02***  (0.01)  −0.02***  (0.00)  −0.02***  (0.00)  Exploratory innovation  0.00  (0.00)  0.27***  (0.06)  0.28*  (0.12)                  Absorbed slack          −1.28*  (0.52)          −0.15  (0.37)  3.92*  (1.78)  Absorbed slack, squared                          −5.28*  (2.30)  Constant  −0.31  (0.27)  −0.21  (0.53)      −0.29  (0.26)  0.49  (0.45)          N firm-years  581    581    581    581    581    581    581    N Firms  37    37    37    37    37    37    37    RMSE  0.04    0.39    0.40    0.04    0.39    0.40    0.43    Log likelihood  1000    −270    −276    1000    −277    −279    −312    F  34.5     4.46     5.52     37.2     3.19     3.73     3.11     a Robust standard errors adjusted for clustering by firm are in parentheses. Two-tailed tests for all effects. † p <0.1; *p <0.05; **p <0.01; ***p <0.001. Table A2. Results of the panel continuous-updating estimator (CUE) instrumental variable regression analysis for the effects of unabsorbed slack on innovation increasesa Dependent variables  Unabsorbed slack   Exploitative innovation   Unabsorbed slack   Exploratory innovation   Model 22  Model 23  Model 24  Model 25  Model 26  Model 27  Model 28  Model 29  Environmental dynamics  2.25  (1.34)                  2.06  (1.36)          Revenue growth  −1.72†  (0.92)                  −1.73†  (0.91)          Organizational size  −0.54*  (0.22)  0.01  (0.06)  −0.07  (0.08)  −0.35  (0.29)  −0.15  (0.11)  −0.53*  (0.23)  −0.03  (0.06)  0.34  (0.21)  R&D  0.01  (0.07)  0.06†  (0.04)  0.06†  (0.03)  0.06  (0.05)  0.05  (0.03)  −0.05  (0.08)  0.11**  (0.03)  0.14*  (0.06)  Age  0.61  (0.40)  −0.05  (0.06)  0.14  (0.13)  0.62  (0.45)  0.21  (0.16)  0.72†  (0.42)  −0.07  (0.13)  −1.00*  (0.46)  M&As  −0.30  (0.26)  −0.13  (0.13)  −0.17  (0.16)  −0.21  (0.24)  −0.18  (0.17)  −0.43  (0.29)  0.01  (0.13)  0.24  (0.20)  Attainment discrepancy  0.50  (1.82)  −1.39*  (0.55)  −1.18**  (0.42)  −1.49†  (0.82)  −1.22**  (0.47)  0.54  (1.84)  −0.05  (0.48)  −0.48  (0.96)  Size of pipelines  −0.08†  (0.04)  0.01  (0.01)  0.00  (0.01)  −0.01  (0.02)  0.00  (0.01)  −0.06  (0.04)  −0.02  (0.01)  0.01  (0.02)  Price revisions  0.28  (0.64)  −0.20  (0.26)  −0.20  (0.41)  −0.13  (0.50)  −0.18  (0.42)  0.31  (0.65)  0.09  (0.42)  −0.12  (0.63)  Asset growth  0.50*  (0.24)  0.23*  (0.09)  0.22*  (0.09)  0.26†  (0.14)  0.23*  (0.09)  0.47*  (0.23)  0.03  (0.07)  0.36  (0.32)  Potential slack  −0.72†  (0.40)  0.07  (0.09)  −0.01  (0.10)  −0.32  (0.37)  −0.12  (0.14)  −0.76†  (0.42)  0.02  (0.08)  0.46†  (0.27)  Non-NCE pipelines  0.00  (0.02)  −0.02***  (0.01)  −0.02***  (0.01)  −0.02*  (0.01)  −0.02***  (0.01)              NCE pipelines                      0.01  (0.01)  −0.02***  (0.01)  −0.03***  (0.01)  Exploratory innovation  −0.08  (0.10)  0.27***  (0.06)  0.27*  (0.12)  0.29*  (0.12)  0.27*  (0.12)              Unabsorbed slack          −0.12†  (0.07)  −0.56  (0.41)              0.62*  (0.29)  Unabsorbed slack, squared              0.05  (0.06)                  Unabsorbed slack, natural logarithm                  −0.59*  (0.30)              Constant  3.75†  (1.89)  −0.21  (0.53)              3.60†  (2.04)  0.49  (0.45)      N firm-years  581    581    581    581    581    581    581    581    N Firms  37    37    37    37    37    37    37    37    RMSE  0.72    0.39    0.41    0.52    0.42    0.72    0.39    0.61    Log likelihood  −628    −270    −286    −427    −302    −627    −277    −518    F  4.04     4.46     4.45     2.72     3.80     4.20     3.19     1.42     Dependent variables  Unabsorbed slack   Exploitative innovation   Unabsorbed slack   Exploratory innovation   Model 22  Model 23  Model 24  Model 25  Model 26  Model 27  Model 28  Model 29  Environmental dynamics  2.25  (1.34)                  2.06  (1.36)          Revenue growth  −1.72†  (0.92)                  −1.73†  (0.91)          Organizational size  −0.54*  (0.22)  0.01  (0.06)  −0.07  (0.08)  −0.35  (0.29)  −0.15  (0.11)  −0.53*  (0.23)  −0.03  (0.06)  0.34  (0.21)  R&D  0.01  (0.07)  0.06†  (0.04)  0.06†  (0.03)  0.06  (0.05)  0.05  (0.03)  −0.05  (0.08)  0.11**  (0.03)  0.14*  (0.06)  Age  0.61  (0.40)  −0.05  (0.06)  0.14  (0.13)  0.62  (0.45)  0.21  (0.16)  0.72†  (0.42)  −0.07  (0.13)  −1.00*  (0.46)  M&As  −0.30  (0.26)  −0.13  (0.13)  −0.17  (0.16)  −0.21  (0.24)  −0.18  (0.17)  −0.43  (0.29)  0.01  (0.13)  0.24  (0.20)  Attainment discrepancy  0.50  (1.82)  −1.39*  (0.55)  −1.18**  (0.42)  −1.49†  (0.82)  −1.22**  (0.47)  0.54  (1.84)  −0.05  (0.48)  −0.48  (0.96)  Size of pipelines  −0.08†  (0.04)  0.01  (0.01)  0.00  (0.01)  −0.01  (0.02)  0.00  (0.01)  −0.06  (0.04)  −0.02  (0.01)  0.01  (0.02)  Price revisions  0.28  (0.64)  −0.20  (0.26)  −0.20  (0.41)  −0.13  (0.50)  −0.18  (0.42)  0.31  (0.65)  0.09  (0.42)  −0.12  (0.63)  Asset growth  0.50*  (0.24)  0.23*  (0.09)  0.22*  (0.09)  0.26†  (0.14)  0.23*  (0.09)  0.47*  (0.23)  0.03  (0.07)  0.36  (0.32)  Potential slack  −0.72†  (0.40)  0.07  (0.09)  −0.01  (0.10)  −0.32  (0.37)  −0.12  (0.14)  −0.76†  (0.42)  0.02  (0.08)  0.46†  (0.27)  Non-NCE pipelines  0.00  (0.02)  −0.02***  (0.01)  −0.02***  (0.01)  −0.02*  (0.01)  −0.02***  (0.01)              NCE pipelines                      0.01  (0.01)  −0.02***  (0.01)  −0.03***  (0.01)  Exploratory innovation  −0.08  (0.10)  0.27***  (0.06)  0.27*  (0.12)  0.29*  (0.12)  0.27*  (0.12)              Unabsorbed slack          −0.12†  (0.07)  −0.56  (0.41)              0.62*  (0.29)  Unabsorbed slack, squared              0.05  (0.06)                  Unabsorbed slack, natural logarithm                  −0.59*  (0.30)              Constant  3.75†  (1.89)  −0.21  (0.53)              3.60†  (2.04)  0.49  (0.45)      N firm-years  581    581    581    581    581    581    581    581    N Firms  37    37    37    37    37    37    37    37    RMSE  0.72    0.39    0.41    0.52    0.42    0.72    0.39    0.61    Log likelihood  −628    −270    −286    −427    −302    −627    −277    −518    F  4.04     4.46     4.45     2.72     3.80     4.20     3.19     1.42     a Robust standard errors adjusted for clustering by firm are in parentheses. Two-tailed tests for all effects. † p <0.1; *p <0.05; **p <0.01; ***p <0.001. Table A2. Results of the panel continuous-updating estimator (CUE) instrumental variable regression analysis for the effects of unabsorbed slack on innovation increasesa Dependent variables  Unabsorbed slack   Exploitative innovation   Unabsorbed slack   Exploratory innovation   Model 22  Model 23  Model 24  Model 25  Model 26  Model 27  Model 28  Model 29  Environmental dynamics  2.25  (1.34)                  2.06  (1.36)          Revenue growth  −1.72†  (0.92)                  −1.73†  (0.91)          Organizational size  −0.54*  (0.22)  0.01  (0.06)  −0.07  (0.08)  −0.35  (0.29)  −0.15  (0.11)  −0.53*  (0.23)  −0.03  (0.06)  0.34  (0.21)  R&D  0.01  (0.07)  0.06†  (0.04)  0.06†  (0.03)  0.06  (0.05)  0.05  (0.03)  −0.05  (0.08)  0.11**  (0.03)  0.14*  (0.06)  Age  0.61  (0.40)  −0.05  (0.06)  0.14  (0.13)  0.62  (0.45)  0.21  (0.16)  0.72†  (0.42)  −0.07  (0.13)  −1.00*  (0.46)  M&As  −0.30  (0.26)  −0.13  (0.13)  −0.17  (0.16)  −0.21  (0.24)  −0.18  (0.17)  −0.43  (0.29)  0.01  (0.13)  0.24  (0.20)  Attainment discrepancy  0.50  (1.82)  −1.39*  (0.55)  −1.18**  (0.42)  −1.49†  (0.82)  −1.22**  (0.47)  0.54  (1.84)  −0.05  (0.48)  −0.48  (0.96)  Size of pipelines  −0.08†  (0.04)  0.01  (0.01)  0.00  (0.01)  −0.01  (0.02)  0.00  (0.01)  −0.06  (0.04)  −0.02  (0.01)  0.01  (0.02)  Price revisions  0.28  (0.64)  −0.20  (0.26)  −0.20  (0.41)  −0.13  (0.50)  −0.18  (0.42)  0.31  (0.65)  0.09  (0.42)  −0.12  (0.63)  Asset growth  0.50*  (0.24)  0.23*  (0.09)  0.22*  (0.09)  0.26†  (0.14)  0.23*  (0.09)  0.47*  (0.23)  0.03  (0.07)  0.36  (0.32)  Potential slack  −0.72†  (0.40)  0.07  (0.09)  −0.01  (0.10)  −0.32  (0.37)  −0.12  (0.14)  −0.76†  (0.42)  0.02  (0.08)  0.46†  (0.27)  Non-NCE pipelines  0.00  (0.02)  −0.02***  (0.01)  −0.02***  (0.01)  −0.02*  (0.01)  −0.02***  (0.01)              NCE pipelines                      0.01  (0.01)  −0.02***  (0.01)  −0.03***  (0.01)  Exploratory innovation  −0.08  (0.10)  0.27***  (0.06)  0.27*  (0.12)  0.29*  (0.12)  0.27*  (0.12)              Unabsorbed slack          −0.12†  (0.07)  −0.56  (0.41)              0.62*  (0.29)  Unabsorbed slack, squared              0.05  (0.06)                  Unabsorbed slack, natural logarithm                  −0.59*  (0.30)              Constant  3.75†  (1.89)  −0.21  (0.53)              3.60†  (2.04)  0.49  (0.45)      N firm-years  581    581    581    581    581    581    581    581    N Firms  37    37    37    37    37    37    37    37    RMSE  0.72    0.39    0.41    0.52    0.42    0.72    0.39    0.61    Log likelihood  −628    −270    −286    −427    −302    −627    −277    −518    F  4.04     4.46     4.45     2.72     3.80     4.20     3.19     1.42     Dependent variables  Unabsorbed slack   Exploitative innovation   Unabsorbed slack   Exploratory innovation   Model 22  Model 23  Model 24  Model 25  Model 26  Model 27  Model 28  Model 29  Environmental dynamics  2.25  (1.34)                  2.06  (1.36)          Revenue growth  −1.72†  (0.92)                  −1.73†  (0.91)          Organizational size  −0.54*  (0.22)  0.01  (0.06)  −0.07  (0.08)  −0.35  (0.29)  −0.15  (0.11)  −0.53*  (0.23)  −0.03  (0.06)  0.34  (0.21)  R&D  0.01  (0.07)  0.06†  (0.04)  0.06†  (0.03)  0.06  (0.05)  0.05  (0.03)  −0.05  (0.08)  0.11**  (0.03)  0.14*  (0.06)  Age  0.61  (0.40)  −0.05  (0.06)  0.14  (0.13)  0.62  (0.45)  0.21  (0.16)  0.72†  (0.42)  −0.07  (0.13)  −1.00*  (0.46)  M&As  −0.30  (0.26)  −0.13  (0.13)  −0.17  (0.16)  −0.21  (0.24)  −0.18  (0.17)  −0.43  (0.29)  0.01  (0.13)  0.24  (0.20)  Attainment discrepancy  0.50  (1.82)  −1.39*  (0.55)  −1.18**  (0.42)  −1.49†  (0.82)  −1.22**  (0.47)  0.54  (1.84)  −0.05  (0.48)  −0.48  (0.96)  Size of pipelines  −0.08†  (0.04)  0.01  (0.01)  0.00  (0.01)  −0.01  (0.02)  0.00  (0.01)  −0.06  (0.04)  −0.02  (0.01)  0.01  (0.02)  Price revisions  0.28  (0.64)  −0.20  (0.26)  −0.20  (0.41)  −0.13  (0.50)  −0.18  (0.42)  0.31  (0.65)  0.09  (0.42)  −0.12  (0.63)  Asset growth  0.50*  (0.24)  0.23*  (0.09)  0.22*  (0.09)  0.26†  (0.14)  0.23*  (0.09)  0.47*  (0.23)  0.03  (0.07)  0.36  (0.32)  Potential slack  −0.72†  (0.40)  0.07  (0.09)  −0.01  (0.10)  −0.32  (0.37)  −0.12  (0.14)  −0.76†  (0.42)  0.02  (0.08)  0.46†  (0.27)  Non-NCE pipelines  0.00  (0.02)  −0.02***  (0.01)  −0.02***  (0.01)  −0.02*  (0.01)  −0.02***  (0.01)              NCE pipelines                      0.01  (0.01)  −0.02***  (0.01)  −0.03***  (0.01)  Exploratory innovation  −0.08  (0.10)  0.27***  (0.06)  0.27*  (0.12)  0.29*  (0.12)  0.27*  (0.12)              Unabsorbed slack          −0.12†  (0.07)  −0.56  (0.41)              0.62*  (0.29)  Unabsorbed slack, squared              0.05  (0.06)                  Unabsorbed slack, natural logarithm                  −0.59*  (0.30)              Constant  3.75†  (1.89)  −0.21  (0.53)              3.60†  (2.04)  0.49  (0.45)      N firm-years  581    581    581    581    581    581    581    581    N Firms  37    37    37    37    37    37    37    37    RMSE  0.72    0.39    0.41    0.52    0.42    0.72    0.39    0.61    Log likelihood  −628    −270    −286    −427    −302    −627    −277    −518    F  4.04     4.46     4.45     2.72     3.80     4.20     3.19     1.42     a Robust standard errors adjusted for clustering by firm are in parentheses. Two-tailed tests for all effects. † p <0.1; *p <0.05; **p <0.01; ***p <0.001. © The Author(s) 2018. Published by Oxford University Press on behalf of Associazione ICC. All rights reserved. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Industrial and Corporate ChangeOxford University Press

Published: Jan 8, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off