Choice of Enterprise Form: Spain, 1886–1936

Choice of Enterprise Form: Spain, 1886–1936 Abstract Every new firm selects a legal form. Organizing as a corporation, a limited company, or a partnership shapes the firm’s access to capital markets, its governance arrangements and tax liabilities, and its treatment in bankruptcy. We use multinomial choice models to estimate the determinants of enterprise form using firm-level data on Spain for the period 1886–1936. Our results support hypotheses drawn from the corporate-finance and ownership literatures; entrepreneurs preferred the corporation for the largest firms and for firms vulnerable to holdup. In 1919, Spain introduced a new legal form, a limited company combining attributes of the corporation and the partnership. This Sociedad de Responsabilidad Limitada (SRL) displaced both corporations and partnerships, and was especially popular for small and medium-sized firms whose owners were unrelated. Counterfactual calculations suggest that few enterprises created prior to 1919 would have chosen the SRL even if it had been available. Entrepreneurs creating a new multi-owner firm face a fundamental contracting decision: the enterprise’s legal form. In most legal regimes today, multi-owner firms can be corporations, limited-liability companies, or partnerships, as well as variants on these basic structures. A firm’s enterprise form shapes its owners’ liability, the firm’s access to public equity markets, and governance matters such as the rights of minority owners. Tax codes often treat corporations differently from other forms. Economists and others have studied each of these issues in isolation, and a theoretical literature studies the margin between partnerships and corporations in some contexts. Empirical work on these choices has, however, been limited. This article uses multinomial choice models with newly-collected micro-data on firm formation to study decisions about legal form in a context that offers both rich data and the absence of some complicating influences: Spain in the period 1886–1936. Using micro-evidence from a single country allows us to abstract from potentially confounding influences such as differences in national banking systems, the organization of equity markets, and the tax treatment of different forms. Most importantly, Spain introduced an entirely new legal form in 1919. The Sociedad con Responsabilidad Limitada (SRL) allowed entrepreneurs to combine some features of the corporation with the partnership’s greater flexibility. We use discrete-choice models to ask which firms preferred this new form as well as conduct counterfactual exercises about what would have happened in the absence of the new form. Spain’s business code offered a discrete set of legal forms from which to choose. The commercial registration system allows us to sample from the universe of multi-owner firms created in that period, and the notarial system yields a set of contracts with consistently-defined provisions. Few other countries systematically collect and retain information on partnerships, limiting empirical investigations to corporations alone; Spain offers a rare opportunity to study decisions across the full range of legal forms. Economists have paid less attention to partnerships than to corporations, but the available theory implies testable hypotheses about the trade-offs among enterprise forms. Chandler (1977) stressed that partnerships could not gather the large amounts of capital needed to take advantage of the new technologies of the nineteenth century. In Spulber (2009)’s model, partnerships self-finance and so have lower capital costs than corporations, but suffer from free-riding because of a lack of concentrated management. The corporation pays higher capital costs than a partnership but has a hierarchical structure that reduces free-riding. For small-scale enterprises the partnership form generates more surplus, but as the firm grows and needs more partners to supply capital, agency costs increase and the corporate form dominates. Spulber’s model thus implies a clear prediction about size and adoption of the corporate form. A different perspective associated with Williamson (1979), Grossman and Hart (1986), and Hart and Moore (1990) treats ownership as central to the firm’s governance structure. Assignment of ownership mitigates contractual incompleteness. Thus, transactions costs have important implications for the choice of the corporate form. The corporation usually has more costly reporting requirements and legal formalities than those required of partnerships, but the corporate form offers two attractive features. Equity investments enjoy limited liability, making it easier to attract capital from many small investors who play no meaningful role in the firm’s management. Corporations also lock-in capital. Partnerships, in contrast, are contracts among individuals. Any partner who wants to withdraw can do so, thereby breaking up the firm. Whether this withdrawal reflects exogenous forces (such as partner mortality) or strategic efforts to obtain better terms (hold-up), the possibility of withdrawal makes the partnership unsuitable for investment in specific assets. Enterprises that require specific investments should prefer the corporate form.1 We have three sets of empirical conclusions concerning the choice of enterprise form. First, we confirm the two basic predictions about the margin between partnerships and corporations as implied by the economics literature. Larger enterprises and businesses with specific capital investments avoided partnership forms. These results are not surprising, but the situation we study provides a rare opportunity to test these hypotheses in a rigorous way using firms from across the full range of sectors. Second, we demonstrate the substitution implications of the SRL’s introduction. The SRL displaced both ordinary partnerships and, less often, corporations. This pattern reflects both the new form’s essential characteristics as well as its flexibility, which allowed entrepreneurs to tailor their SRL to suit their specific needs. Counterfactual calculations demonstrate that had the SRL been introduced in the 1890s, few firms would have taken advantage of the form. The SRL form reflected a new type of business in the 1920s. These new businesses took advantage of new technologies, reached new markets, and reflected the efforts of new types of investor groups. Third, our results demonstrate both the continuing importance of family ties in business as well as the interaction between kin connections and legal form. One important characteristic of investor groups favoring the SRL was their lack of family ties. Family groups continued to favor the ordinary partnership. The SRL’s introduction allowed investors to rely on law, rather than kin ties, to contend with information, incentive, and monitoring problems in the multi-owner enterprise. This result confirms the central contention in the family-firms literature and suggests an unappreciated connection between family control and enterprise contracting choices. Recent decades have seen the introduction of new legal forms for business enterprises in many economies. Ribstein (2010) calls these firms “uncorporations” to underline the essential differences between corporations, older partnership forms, and the new limited-liability companies (LLCs). During the 1990s, every US state introduced a new limited-liability partnership (LLP) form. Many US states have also created or reformed the LLC, which allows owners to obtain the benefits of the corporation in a form more suitable to smaller enterprises (Hansmann et al. 2006). Most then-wealthy countries other than Britain and the United States allowed the LLP from the early 19th century, and forms similar to the LLC appeared starting in Germany in 1892. Development of enterprise forms remains an active interest of policymakers in several contexts. The European Union and its Member States have also considered revising their menu of enterprise forms.2 Economists now widely agree that institutions affect economic performance. A growing literature discusses what constitutes good or bad institutions and searches for statistical relationships between those institutions and economy performance. Much of this literature uses country-level aggregates, regressing indicators such as GDP per capita on variables intended to capture one or more features of the institutional environment. In Acemoğlu and Johnson (2005), for example, the institutional variables are summary indices such as the country’s degree of legal formalism, constraints on its executive, and measures of protection against expropriation. The “law and finance” argument attributed to La Porta et al. (1998) (see also La Porta et al. 2008) uses similar empirical methods but focuses on a narrower issue. These authors divide the world’s countries according to legal family; “common law” countries include Britain and its former colonies, whereas the “civil law” countries include most of Continental Europe as well as the former colonies of France, Portugal, and Spain. The civil-law system’s greater reliance on legal codes and statutory law, as opposed to the case-law and jurisprudence at the heart of the common-law system, they argue, makes the civil-law systems more rigid. La Porta et al. use cross-country regression to argue that civil-law countries have less well-developed investor and creditor protection, less well-developed financial systems, and as a result, enjoy lower rates of economic growth. Lamoreaux (2016) notes that cross-country regressions miss most of the important developments in late-19th century company law.3 This is especially true of the United States, where most relevant law is state-level. Our approach differs from these agendas in two important ways. First, by relying on firm-level data and concrete legal rules, we study the effect of well-defined, specific institutions on the behavior of optimizing agents. Second, the SRL’s introduction allows us to see how those agents respond to a change in the institutional environment, a feature usually absent from the cross-country approach.4 1. Law and Enterprise Form Spain’s business code owes much to French legal traditions, a trait common to many civil-law countries. The 1885 Spanish code allowed firms to organize as ordinary partnerships, limited partnerships, or corporations. The ordinary partnership required unlimited liability of its owners. In a limited partnership, some owners could restrict their liability to the sum they had invested, but could not participate in management. All investors in Spanish corporations had limited liability.5 Another important difference makes the partnership subject to hold-up problems. Partnerships are in practice at-will and so cannot lock in capital effectively; if a partner wanted to withdraw, he could leave and take his investment with him. The only way for an investor to exit a corporation is to sell his shares to someone else. Some legal scholars view capital lock-in as the corporation’s essential feature (see Stout (2005) or Blair (2003)). Lock-in is also a feature of limited-liability companies, including the SRL.6 Civil-law countries, such as Spain, ordinarily required firms to use one of the legal forms specifically described in the law. The 1885 Spanish code, however, quite unusually allowed firms to modify the standard forms, so long as the resulting firm respected the code’s other requirements (§117, 122). Some firms took advantage of this flexibility to adopt features of the new enterprise forms appearing elsewhere in Europe. The practice remained rare, however, until 1919, when the Ministry of Justice issued instructions to commercial registries requiring the recognition of the new SRL. The rules governing the SRL relied on the 1885 code’s partnership rules and case law until 1953, when Spain passed its first legislation governing the SRL.7 The SRL resembled a number of new enterprise forms introduced in Europe in the late 19th and early 20th centuries. These new forms enabled entrepreneurs to create a firm that combined some features of the corporation with some features of the partnership. The SRL’s owners all had limited liability for their investments and their ownership stakes were freely transferrable unless restricted by the articles of association. SRL investments were locked-in. These features all mimicked the corporation. Unlike the corporation, however, the SRL could not list its equity on public markets. The new form also escaped some formalities (such as publicity) required of corporations. An SRL could pay one of its owners to run the firm, but managers could also be outsiders who had no ownership stake. The new form retained the flexibility of partnership contracting, allowing, for example, complex distinctions between cash flow and control rights.8 Debates over legal forms and the desirability of reforming or introducing new forms date at least to the early 19th century. Until the middle of the 19th century, most European countries strictly controlled use of the corporate form, shifting to “general incorporation” only later. (The United States was unusual in that many states introduced general incorporation earlier in the nineteenth century; Hilt 2017.) The second half of the 19th century saw major reforms to the corporation in leading economies, as well as the introduction of the first SRL-like enterprise forms (Guinnane et al. 2007). The United Kingdom’s 1862 Companies Act and later court decisions made the corporation attractive to small enterprises and a 1907 Act formally created the Private Limited Company (PLC). Germany introduced its PLC (the Gesellschaft mit beschränkter Haftung or GmbH) in 1892. Both the British PLC and the German GmbH enjoyed many imitators. Spain’s introduction of the SRL reflects, in part, influence from the outside. Developments in company law during the 19th and 20th centuries have been the basis of sustained attacks on the La Porta et al. (1998) “law and finance” hypothesis. Lamoreaux and Rosenthal (2005) noted that France (a civil-law country) offered entrepreneurs more flexibility than did the United States (a common-law country). Gómez-Galvarriato and Musacchio (2006) show that Mexico, another civil-law country, also enjoyed a more flexible company law than is consistent with the law and finance claim. Guinnane et al. (2007) extended the comparison to include both Britain and Germany (another civil-law country); the German GmbH reflects a fairly radical legal innovation of the type that La Porta et al. (1998) views as unlikely in civil-law countries. Some 19th-century governments, including Spain, published summary information on the choice of legal form. These reports formed the basis for empirical discussions of the topic. More recently, scholars have created firm-level databases from business registries.9 Econometric studies of firm-level choices are, however, recent. The first work of this type may be Gómez-Galvarriato and Musacchio (2004), which uses micro-data from Mexico City at the end of the 19th and early 20th centuries. Other econometric studies of enterprise-form choice include Hilt and O’Banion (2009) (studying the difference between ordinary and limited partnerships in New York State) and Abramitzky et al. (2010) (the same question, but in Brazil).10Nicholas (2015) studies trade-offs among enterprise forms in Japan between 1896 and 1939, stressing that the SRL-like form introduced in 1938 did not replace the corporate form. 2. Data on Choices of Legal Form in Spain To enjoy the commercial code’s protection, a firm registered its formation and provided a summary of its articles of association to the provincial branch of the commercial registry. We have assembled three distinct databases based on this registry. The first source records the total number of firms organized under each legal form, along with the total capital in these firms for every year between 1886 and 1936. These reports (Yearbooks) provide a rare and complete picture of how firms organized.11 The other two sources have firm-level data taken from the firm’s articles of association. The first micro-level source encodes a publication that reports selected information on every Spanish firm that registered in the years 1925, 1926, and 1927. The information available in this source, which we call “Firms Census,” is limited to the variables published in the original source. The second micro database (Firms Sample) comprises a random sample of enterprises from the archives of 15 peninsular provincial commercial registries. For each of these provinces, we randomly selected two firms formed in every year from 1886 to 1936. Firms Sample reflects the limited number of accessible Spanish archives, along with the considerable labor required to code each observation. Many of our results rely entirely on Firms Census, which includes every firm formed in Spain in those years. Our data reflect the firm’s initial characteristics as provided to the commercial registry upon formation. Attributes such as capital stock could change over the firm’s life, though partnerships rarely changed hands. The individuals we consider to be a corporation’s “owners,” however, are those who signed the articles of association. Our sources do not include subsequent performance measures, so we cannot, unfortunately, ask how legal form affected profitability.12 Figure 1 uses the Yearbooks to summarize the distribution of new firms by legal form along with the total number of registrations for all of Spain over our period. The majority of firms in 1886 were ordinary partnerships, with a small but stable fraction organized as limited partnerships. Corporations accounted for a modest share of new firms. The period prior to the SRL’s formalization in 1919 saw the slow decline of the partnership and the equally slow rise of the corporation. The loss of Spain’s remaining colonies in 1899 led to the repatriation of considerable capital and an increased number of new corporations starting in that year. By 1936, only a handful of new companies registered as limited partnerships. The new SRL and the corporation’s growing use made the ordinary partnership considerably less popular than earlier. Figure 1 also reports the total number of annual firm registrations. Spain experienced slow growth in new firms up to World War I, then a short boom in new registrations that ended in the early 1920s. This episode reflects the advantage of being neutral in a Europe at war. Figure 1. View largeDownload slide Distribution of Legal Forms for all Enterprises Registered in Spain, 1886–1936, Along with Total Number of Firm Registrations. Source: Yearbooks database. Notes: Left axis is percentage of firms registering as a given legal form; right axis is the total number of registrations. See text for more discussion of legal forms. Figure 1. View largeDownload slide Distribution of Legal Forms for all Enterprises Registered in Spain, 1886–1936, Along with Total Number of Firm Registrations. Source: Yearbooks database. Notes: Left axis is percentage of firms registering as a given legal form; right axis is the total number of registrations. See text for more discussion of legal forms. Table 1 summarizes three dimensions of our sample firms.13 Measured by capitalization and the number of owners, the SRLs were somewhat larger than the two partnership forms. Corporations were larger than all other forms. We constructed the two “family connections” variables from surnames and internal references (e.g. “and his son …”).14 Owners who are all related preferred the ordinary to the limited partnership. When the owners combined both relatives and nonrelatives, the limited partnership was more common. Table 1 also summarizes the practice of stating a contractual duration in a firm’s articles of association. This was optional, and most firms left it open, but corporations were most likely to include a fixed duration in their articles, and to define it as longer than 10 years. Table 1. Characteristics of Firms by Legal Form   Ordinary partnerships  Limited partnerships    1886–1898  1899–1918  1919–1936  1886–1898  1899–1918  1919–1936  Average capital  58,296  42,279  103,971  150,039  88,382  165,188  Average number of owners  2.68  2.72  2.95  3.43  3.59  3.26  Percentage of firms whose owners are all related  26.8  24.6  32.5  8.8  16.5  13  Percentage of firms for which some owners are related  12.5  12.3  10.2  35.1  20.6  29  Percentage of firms that state a contractual duration  91.9  75.3  52.5  85.9  78.3  61.3  Number of firms  224  357  175  56  97  31      Corporations  SRL    1886–1898  1899–1918  1919–1936    1919–1936      Average capital  3,330,246  631,158  2,152,169    119,255    Average number of owners  7.78  8.31  6.68    3.93    Percentage of firms whose owners are all related  2.8  1.7  13.1    10    Percentage of firms for which some owners are related  11.1  25.9  32    25.5    Percentage of firms that state a contractual duration  69.4  58  32.7    36.9    Number of firms  36  109  151    140      Ordinary partnerships  Limited partnerships    1886–1898  1899–1918  1919–1936  1886–1898  1899–1918  1919–1936  Average capital  58,296  42,279  103,971  150,039  88,382  165,188  Average number of owners  2.68  2.72  2.95  3.43  3.59  3.26  Percentage of firms whose owners are all related  26.8  24.6  32.5  8.8  16.5  13  Percentage of firms for which some owners are related  12.5  12.3  10.2  35.1  20.6  29  Percentage of firms that state a contractual duration  91.9  75.3  52.5  85.9  78.3  61.3  Number of firms  224  357  175  56  97  31      Corporations  SRL    1886–1898  1899–1918  1919–1936    1919–1936      Average capital  3,330,246  631,158  2,152,169    119,255    Average number of owners  7.78  8.31  6.68    3.93    Percentage of firms whose owners are all related  2.8  1.7  13.1    10    Percentage of firms for which some owners are related  11.1  25.9  32    25.5    Percentage of firms that state a contractual duration  69.4  58  32.7    36.9    Number of firms  36  109  151    140    Source: Computed from the firms sample database. Online Appendix Table A.3 reports analogous measures for Firms Census, where available. Notes: “Capital” is authorized capital, in current pesetas. In 1900, one pesetas equaled about £.032, or $0.149 (US) (Martín Aceña and Pons 2005: 704). Using a cost-of-living index, one 1900 peseta was worth €3.57  in 2016 (www.measuringworth.com). We determine relationships among owners using kinship term in the articles of association. The definition is limited to the nuclear family. See text for more discussion. There was no SRL form prior to 1919. We group firms into sectors using a standard industrial classification for Spain. For the econometric analysis, we further combine these sectors into four groups: factories, trading enterprises, mining and infrastructure firms, and a (residual) miscellaneous category. (See Online Appendix A Table A.4) Corporations dominate two sectors, mining and the heterogeneous category of agriculture processing, utilities, and construction. These sector classifications reflect the limitations of sample-size as well as an effort to use categories for which all provinces have firms. Within a given sector, firms organized as corporations have larger total capital investments than firms organized in other ways. The SRL was at first most popular in two large sectors: factories and trade. The new form also appealed to a range of professional and service-related firms, such as the liberal professions or travel services. 3. Modeling Firm-Level Decisions about Legal Form Entrepreneurs creating a multi-owner firm choose an enterprise form to minimize contracting problems given the characteristics of the firm and its owners, subject to the legal system’s constraints. We study this decision using multinomial discrete-choice models. We have two questions: How do the firm’s characteristics affect the choice of form? And, how did the expansion of the menu of choices represented by the SRL’s introduction affect these choices? There is no natural ordering for the choice of legal form, so we restrict our attention to unordered choice models. After considering several alternatives, we rely on nested logit (NL) models.15 The NL model can be derived from random-utility maximization, so the estimates imply, for each firm, a rank-ordering of preferences over legal forms. The NL model also reveals the SRL’s impact on other enterprise forms. When Spain introduced the new SRL, some entrepreneurs preferred that form to any of the older alternatives. For others, the new SRL was irrelevant in that entrepreneurs still preferred some alternative. Our approach allows us to study what firms would have done in the absence of the new form, and, less formally, to examine the characteristics of firms forced to make a suboptimal enterprise form choice before the SRL was a possibility. The econometric models require an assumption: that a firm’s attributes are all fixed when its founders decide on the legal form. We thus abstract from the possibility that legal form is decided as part of a negotiation over other firm attributes. Concerns over possible endogeneity of the regressors should focus on the possibility that a potential owner might hold out for limited liability or some other firm attribute, thus implying that the participation of some owners was contingent upon the choice of enterprise form. Given our sources, there is no tractable solution to this endogeneity problem, so we must be cautious in interpreting our results as “causal.” We first estimate the NL model using the Firms Census database. These results provide a useful baseline because the data reflect a census of all firms and because these results allow us to understand the post-SRL world in detail. We then turn to the Firms Sample. We define variables in the two databases identically unless otherwise noted. A brief overview will guide our discussion: First, enterprise size affects choice of legal form in three ways. Spulber (2009), as noted, stresses the trade-off between monitoring and the cost of capital in the choice between a partnership and a corporation. His model implies that larger firms prefer the corporate form. In addition, only corporations can access public equity markets, which eases the task of raising capital for the largest enterprises. (The SRL’s equity shares could not be listed on exchanges.) In addition, the corporation requires formalities that entail fixed costs (such as the publication of a balance sheet), providing another reason why smaller enterprises would prefer another form.16 We capture firm-size effects with the natural log of stated capital and its square. We also include a dummy variable for whether the firm has any unpaid capital. Firms could and did have nominal capital in excess of what owners had paid in. Shareholders remained liable to pay in more capital, as stipulated by the firm’s articles. The difference between nominal and paid-in capital also enhanced the firm’s borrowing ability. Second, firms that contemplated long-term investments in illiquid projects would have been reluctant to organize as partnerships because of potential hold-up problems. We have two proxies for concerns about untimely dissolution issues. First, in some sectors, a firm’s physical capital can easily be liquidated while in others it cannot. Thus, sector serves as a proxy for concerns about this form of hold-up.17 Mining and infrastructure firms are those most engaged in investments of the type that would create untimely dissolution problems. Second, a firm’s articles of association could stipulate that it was open-ended or would last only for a specific period. We construct a dummy variable for whether the firm has a stated duration and define another variable as the interaction of the dummy variable with the number of years stipulated as the firm’s duration.18 Third, we allow family ties among owners to affect the choice of legal form. (We discuss the relationship between enterprise form and the family business in Section 6.) Because it relies on a published source that did not report this information, we do not have the relationship variables for the Firms Census source. For the Firms Sample database, we use the number of owners and family relationships to capture aspects of the legal-form decision that may reflect owner identity. The choice of legal form might also reflect local considerations, which is one reason to use province dummies. Notaries outside the main commercial centers had less experience organizing corporations. In the years 1886 through 1936, a total of eight Spanish provinces did not see the creation of a single corporation. And if one reason to organize a corporation was to tap liquid markets for investors, either privately or by listing on the stock market, a firm located outside financial centers such as Madrid or Barcelona (or places with smaller exchanges, such as Bilbao or Valencia) might think twice about the corporate form. 4. Results The NL model requires judgment about which alternatives are similar and thus belong in the same nest. Figure 2 describes our approach. We group the enterprise forms with limited liability for at least some owners into one nest and place the ordinary partnership in its own, “degenerate” nest. Table 2 reports the NL estimates for the Firms Census model. (Online Appendix Table B.1 reports descriptive statistics for the estimation subsamples.) Several alternative nesting structures seem plausible, but the data reject the other candidates as inconsistent with the random utility-maximization model that underlies NL. The model is consistent with utility maximization only if the nesting (or “dissimilarity”) parameter (which we constrain to “1” for the degenerate nest) lies on the unit interval. Table 2. Nested Logit Model for the Firm Census Database   Limited partnership  Corporation  SRL  Firm has stated duration  0.161  −2.569  −0.213  (0.324)  (0.507)  (0.195)  Duration in years  −0.012  0.147  0.006  (0.024)  (0.031)  (0.015)  Firm has any unpaid capital  1.209  3.340  1.392  (0.641)  (0.515)  (0.438)  Ln (capital)  −0.656  −0.698  −0.278  (0.118)  (0.108)  (0.044)  Ln (capital) squared  0.043  0.064  0.022  (0.006)  (0.009)  (0.004)  Sector   Factory  0.165  −0.848  −0.256  (0.283)  (0.198)  (0.153)   Trade  0.216  −1.250  −0.289  (0.304)  (0.252)  (0.159)   Infrastructure  0.274  0.244  −0.013  (0.372)  (0.208)  (0.220)  Location   Catalonia  0.523  0.651  0.040  (0.186)  (0.153)  (0.138)   Madrid  0.524  0.492  0.376  (0.257)  (0.185)  (0.163)   Basque Country  −0.303  −0.556  0.605  (0.284)  (0.280)  (0.158)  Year firm formed   1926  0.121  0.0511  0.156  (0.170)  (0.130)  (0.113)   1927  −0.373  0.00751  0.312  (0.245)  (0.136)  (0.122)  Dissimilarity parameter  0.915      (0.236)  Number of firms  3142      Log-likelihood  −3058.1043        Limited partnership  Corporation  SRL  Firm has stated duration  0.161  −2.569  −0.213  (0.324)  (0.507)  (0.195)  Duration in years  −0.012  0.147  0.006  (0.024)  (0.031)  (0.015)  Firm has any unpaid capital  1.209  3.340  1.392  (0.641)  (0.515)  (0.438)  Ln (capital)  −0.656  −0.698  −0.278  (0.118)  (0.108)  (0.044)  Ln (capital) squared  0.043  0.064  0.022  (0.006)  (0.009)  (0.004)  Sector   Factory  0.165  −0.848  −0.256  (0.283)  (0.198)  (0.153)   Trade  0.216  −1.250  −0.289  (0.304)  (0.252)  (0.159)   Infrastructure  0.274  0.244  −0.013  (0.372)  (0.208)  (0.220)  Location   Catalonia  0.523  0.651  0.040  (0.186)  (0.153)  (0.138)   Madrid  0.524  0.492  0.376  (0.257)  (0.185)  (0.163)   Basque Country  −0.303  −0.556  0.605  (0.284)  (0.280)  (0.158)  Year firm formed   1926  0.121  0.0511  0.156  (0.170)  (0.130)  (0.113)   1927  −0.373  0.00751  0.312  (0.245)  (0.136)  (0.122)  Dissimilarity parameter  0.915      (0.236)  Number of firms  3142      Log-likelihood  −3058.1043      Note: The estimates are relative to an ordinary partnership. Robust standard errors in parentheses. See Figure 2 for nesting structure. Duration refers to the contractual duration stated in the articles of association. The reference sector is all other firms, the reference location is the rest of Spain, and the reference year is 1925. Figure 2. View largeDownload slide Schematic for Nested-Logit Model Structures. Note: This structure underpins the nested logit (NL) models reported in the text. Figure 2. View largeDownload slide Schematic for Nested-Logit Model Structures. Note: This structure underpins the nested logit (NL) models reported in the text. The estimates are relative to the ordinary partnership. The choice of normalization does not affect estimated probabilities. Because the model has several branches, the individual parameters are even more difficult to interpret than is normally the case for nonlinear models. Wald tests reject the null hypothesis that any of the branches are redundant in the sense that the model cannot distinguish that form from the ordinary partnership. We cannot, however, reject the null that any branch is equal to any other.19 Individual regressors do have substantially different effects for different forms, however.20 The estimates imply a ranking over legal form options for each firm. We compare the “predicted” legal form (the highest-ranked form, according to the model) to the form the enterprise actually took. With more than two outcomes, this is a demanding standard. The predictions generated by the model reported in Table 2 correspond to firm choices in about 58% of cases. This statistic varies considerably across enterprise forms. The model correctly predicts about 77% of ordinary partnerships and 79% of corporations. The limited partnership and SRL are more difficult: the model does not predict any limited partnerships (6% of firms) and only about 16% of SRLs. There are two reasons for the model’s difficulty with the limited partnership. First, they are rare, so we are asking an econometric model to capture the tail of a distribution. Second, the considerations that drive the difference in rankings between ordinary and limited partnerships may be idiosyncratic (and not in the data) and thus not available to the model. Some limited partners were family members who retained an investment but withdrew from active participation in the business; if we knew details of an entrepreneur’s living kin, or the history of this enterprise, for example, the model might better distinguish the limited from the ordinary partnership. We see below that family relationships among owners drives part of the margin between partnerships and SRLs. The Firms Census database lacks this information. With four alternatives, it is also useful to examine the first and second-ranked forms. Under this more forgiving criterion, the model predicts 94% of partnerships, 83% of corporations, and 81% of SRLs. But the limited partnerships remain a problem; in fact, the model implies that for 64% of actual limited partnerships, that form was the firm’s last choice. To examine how the firm’s characteristics affect the choice of legal form, we rely on average marginal effects (AMEs), or the mean change in the predicted probability that firms select a given form when one or more regressors are varied. By construction, the AMEs sum to one for all choices facing a single firm. Some AMEs are large but imprecisely estimated. The model confirms two basic hypotheses from the literature on the trade-off between partnerships and corporations. Increasing the firm’s capitalization from the mean by 10% reduces the probability of forming an ordinary partnership by 0.0989 (with a standard error of 0.0358) and increases the chance of a corporation forming by 0.1592 (0.0565). Considering the SRL in this context introduces an interesting wrinkle: more capital also makes the SRL less attractive relative to the corporation (0.0602 (0.0291)).21 This effect for the SRL may reflect entrepreneurs’ ability to more effectively sort when they do not want to issue tradeable shares; the SRL allows limited liability with locked-in capital at a smaller scale. Put differently, the enterprise turns to the corporation when it needs to raise larger amount of capital from investors who demand liquidity. This consideration would be especially important for the largest enterprises that need to raise capital from all over Spain, and abroad.22 The model also confirms that asset specificity makes corporations more attractive because of the need to lock in capital. Hypothetically, moving an infrastructure firm to the trade sector increases the chance of an ordinary partnership by 0.11 (0.0662) and reduces the probability of the corporation by 0.1239 (0.1119). The year dummies imply that the SRL became more popular over time. Since the Justice Ministry authorized the SRL in 1919, this result implies that it took time for the various actors—business people, notaries, creditors, etc.—to adapt to the new enterprise. Starting a firm in 1927 compared to 1925 raises the probability of the firm choosing the SRL by 0.085, with a standard error of 0.0296. 5. Decisions 1886–1918: the Firms Sample Database We now turn to two parallel models using the Firms Sample database for the period prior to and following the SRL’s introduction. We split the Firms Sample database in this way because the SRL’s introduction in 1919 enlarges the choice set for that year and those after. The Firms Sample database comes from archival sources and has fewer observations than the Firms Census. These data have information on firms that is not available in the Firm Census and also contains firms created before the SRL was a practical option. The estimates from this model supplement our findings based on the Firms Census. We divide our 15 sampled provinces into five groups of three provinces each. To account for trends over the period 1885–1936, we use a linear spline, with the knot set at 1899, the year Spain began to see capital repatriated from its former colonies. We know the number and identity of the owners when the firm was established and use the indicators of family relations discussed previously to infer familial relationships among owners. We enter the total number of owners linearly. A first dummy variable is unity if all owners are related, and another is unity if only some owners appear to be related. The reference category is firms for which the owners are not related. We interact these two dummies with the number of owners, allowing the connection between legal form and family connections to vary with firm size. Table 3 reports estimates for the Firm Sample model before 1919. The information on owner numbers and family relationships improves the model’s ability to distinguish between ordinary partnerships and other forms. The model correctly predicts 77% of firms overall, including 95% of ordinary partnerships and about 70% of corporations. The limited partnership remains problematic, however. Only 8% of actual limited partnerships have this as their predicted first choice, but about 85% have this option as their predicted second choice. This is admittedly a low standard when there are only three options. Table 3. Nested Logit Model, Firm Sample, Pre-1919   Limited partnership  Corporation    Est.  SE  Est.  SE  Firm has stated duration  −0.040  (0.364)  −1.877  (0.749)  Duration in years  −0.021  (0.023)  0.117  (0.048)  Firm has any unpaid capital  0.347  (0.399)  1.608  (0.466)  Number of owners  0.400  (0.112)  0.494  (0.139)  Year  −0.495  (0.285)  0.259  (0.170)  Year spline, 1899 = 0  0.025  (0.016)  0.0618  (0.025)  Ln (capital)  1.010  (1.013)  −1.785  (0.725)  Ln (capital) squared  −0.031  (0.049)  0.102  (0.036)  Sector      Factory  0.044  (0.406)  −1.416  (0.538)      Trade  0.211  (0.417)  −1.937  (0.639)      Infrastructure  −0.354  (0.552)  −0.187  (0.430)  Relationships among owners      All owners are relatives  0.407  (0.524)  −0.336  (0.908)      Some owners are relatives  0.503  (0.693)  −0.395  (1.018)      Some owners are relatives × number of owners  −0.151  (0.171)  −0.105  (0.211)      All owners are relatives × number of owners  −0.438  (0.158)  −0.638  (0.224)  Provinces      Group one  0.107  (0.290)  0.612  (0.421)      Group two  0.294  (0.279)  −0.290  (0.479)      Group three  −0.126  (0.277)  0.081  (0.351)      Group four  −0.181  (0.284)  0.404  (0.368)  Dissimilarity parameter  0.811 (0.385)      Number of firms  870      Log-likelihood  −493.2403        Limited partnership  Corporation    Est.  SE  Est.  SE  Firm has stated duration  −0.040  (0.364)  −1.877  (0.749)  Duration in years  −0.021  (0.023)  0.117  (0.048)  Firm has any unpaid capital  0.347  (0.399)  1.608  (0.466)  Number of owners  0.400  (0.112)  0.494  (0.139)  Year  −0.495  (0.285)  0.259  (0.170)  Year spline, 1899 = 0  0.025  (0.016)  0.0618  (0.025)  Ln (capital)  1.010  (1.013)  −1.785  (0.725)  Ln (capital) squared  −0.031  (0.049)  0.102  (0.036)  Sector      Factory  0.044  (0.406)  −1.416  (0.538)      Trade  0.211  (0.417)  −1.937  (0.639)      Infrastructure  −0.354  (0.552)  −0.187  (0.430)  Relationships among owners      All owners are relatives  0.407  (0.524)  −0.336  (0.908)      Some owners are relatives  0.503  (0.693)  −0.395  (1.018)      Some owners are relatives × number of owners  −0.151  (0.171)  −0.105  (0.211)      All owners are relatives × number of owners  −0.438  (0.158)  −0.638  (0.224)  Provinces      Group one  0.107  (0.290)  0.612  (0.421)      Group two  0.294  (0.279)  −0.290  (0.479)      Group three  −0.126  (0.277)  0.081  (0.351)      Group four  −0.181  (0.284)  0.404  (0.368)  Dissimilarity parameter  0.811 (0.385)      Number of firms  870      Log-likelihood  −493.2403      Note: The estimates are relative to an ordinary partnership. Robust standard errors in parentheses. See Figure 2 for nesting structure, and Table 2 for variable definitions. Province group one consists of Albacete, Murcia, and Almería; Group two, Barcelona, Tarragone, and Alicante; Group three, Vizcaya, Navarra, and A Coruña; Group four, Madrid, Cuenca, and Valladolid. The ommitted group consists of Asturias, Cantabria, and La Rioja. The Firms Sample database has 500 firms in the post-1919 period, which strains our ability to estimate 45 parameters. Unfortunately, only this database has the family variables in the SRL period, so we must accept the possible problems the small sample poses. We report the estimates in Table 4. This post-1919 model predicts 73% of partnerships, 26% of limited partnerships, 76% of corporations, and 57% of SRLs. This is a dramatic improvement for the limited partnership and the SRL; the family composition of the owners allows the model to more accurately account for decisions concerning these two forms. We discuss these in more detail below. Table 4. Nested Logit Model, Firm Sample, Post-1919   Limited partnership  Corporation  SRL    Est.  SE  Est.  SE  Est.  SE  Firm has stated duration  −0.653  (0.756)  −2.697  (1.526)  −1.055  (0.472)  Duration in years  0.059  (0.042)  0.142  (0.084)  0.040  (0.042)  Firm has any unpaid capital  1.488  (0.484)  1.764  (0.683)  0.767  (0.688)  Number of owners  0.126  (0.190)  0.309  (0.232)  0.015  (0.194)  Year formed  −0.320  (0.554)  0.126  (0.271)  −0.211  (0.282)  Year formed spline, 1899 = 0  −0.029  (0.104)  0.053  (0.046)  0.129  (0.048)  Ln (capital)  0.712  (2.094)  −1.305  (1.068)  −0.006  (0.973)  Ln (capital) squared  −0.013  (0.093)  0.086  (0.055)  0.012  (0.050)  Sector      Factory  −1.053  (0.600)  −1.408  (0.670)  −0.673  (0.465)      Trade  −0.277  (0.589)  −1.513  (0.810)  −0.737  (0.434)      Infrastructure  0.257  (0.679)  0.005  (0.606)  −0.284  (0.707)  Relationships among owners      All owners are relatives  0.907  (1.944)  −1.137  (0.772)  −1.662  (0.844)      Some owners are relatives  1.810  (1.269)  1.155  (0.868)  0.081  (0.725)      Some owners are relatives × number of owners  −0.315  (0.286)  −0.231  (0.234)  0.066  (0.207)      All owners are relatives × number of owners  −0.751  (0.668)  −0.177  (0.228)  0.014  (0.249)  Provinces      Group one  0.649  (0.893)  0.285  (0.508)  −0.742  (0.740)      Group two  −0.113  (0.458)  0.212  (0.478)  −0.247  (0.335)      Group three  −0.085  (0.540)  0.0976  (0.487)  0.331  (0.387)      Group four  −0.420  (0.808)  0.337  (0.398)  0.451  (0.351)  Dissimilarity parameter  0.502 (0.520)  Number of firms  498  Log-likelihood  −386.9527    Limited partnership  Corporation  SRL    Est.  SE  Est.  SE  Est.  SE  Firm has stated duration  −0.653  (0.756)  −2.697  (1.526)  −1.055  (0.472)  Duration in years  0.059  (0.042)  0.142  (0.084)  0.040  (0.042)  Firm has any unpaid capital  1.488  (0.484)  1.764  (0.683)  0.767  (0.688)  Number of owners  0.126  (0.190)  0.309  (0.232)  0.015  (0.194)  Year formed  −0.320  (0.554)  0.126  (0.271)  −0.211  (0.282)  Year formed spline, 1899 = 0  −0.029  (0.104)  0.053  (0.046)  0.129  (0.048)  Ln (capital)  0.712  (2.094)  −1.305  (1.068)  −0.006  (0.973)  Ln (capital) squared  −0.013  (0.093)  0.086  (0.055)  0.012  (0.050)  Sector      Factory  −1.053  (0.600)  −1.408  (0.670)  −0.673  (0.465)      Trade  −0.277  (0.589)  −1.513  (0.810)  −0.737  (0.434)      Infrastructure  0.257  (0.679)  0.005  (0.606)  −0.284  (0.707)  Relationships among owners      All owners are relatives  0.907  (1.944)  −1.137  (0.772)  −1.662  (0.844)      Some owners are relatives  1.810  (1.269)  1.155  (0.868)  0.081  (0.725)      Some owners are relatives × number of owners  −0.315  (0.286)  −0.231  (0.234)  0.066  (0.207)      All owners are relatives × number of owners  −0.751  (0.668)  −0.177  (0.228)  0.014  (0.249)  Provinces      Group one  0.649  (0.893)  0.285  (0.508)  −0.742  (0.740)      Group two  −0.113  (0.458)  0.212  (0.478)  −0.247  (0.335)      Group three  −0.085  (0.540)  0.0976  (0.487)  0.331  (0.387)      Group four  −0.420  (0.808)  0.337  (0.398)  0.451  (0.351)  Dissimilarity parameter  0.502 (0.520)  Number of firms  498  Log-likelihood  −386.9527  Note: The estimates are relative to an ordinary partnership. Robust standard errors in parentheses. See Figure 2 for nesting structure, and Table 2 for variable definitions. Province group one consists of Albacete, Murcia, and Almería; Group two, Barcelona, Tarragone, and Alicante; Group three, Vizcaya, Navarra, and A Coruña; Group four, Madrid, Cuenca, and Valladolid. The ommitted group consists of Asturias, Cantabria, and La Rioja. Point estimates for the AMEs in the two Firms Sample models are large but imprecisely estimated. We focus on the estimates for the SRL period as reported in Table 4. Consider two different, two-owner firms. The firms with two related owners are less likely to choose the ordinary partnership (0.0817, standard error 0.086) and more likely to form as a limited partnership (0.077 (0.0578)). The effects are slightly different if we compare a firm with three related owners to a three-person firm for which only two owners are related. In the post-1919 model these effects are larger; a three-person firm with all related owners has a smaller chance (0.2978 (0.1811)) of selecting the ordinary partnership and a similarly larger chance (0.294 (0.2188)) of taking on the limited partnership. 6. The SRL: Using Law to Substitute for Family Ties SRLs formed after 1919 closely resembled the partnership forms, especially the ordinary partnership, in capitalization, average number of owners, and sector. We have shown, however, that SRLs differed from ordinary partnerships in an important respect: SRL owners were less likely to be related. This result suggests an important role for enterprise forms such as the SRL. They provide a legal structure for small firms whose owners lack kin ties. This finding echoes and amplifies a long literature on the economics of family businesses. Economists and economic historians have long known that family groups play central roles in forming and sustaining business enterprises. Families and firms constitute two of Ben-Porath (1980)’s “F connections.” The growing finance literature on family firms usually focuses on questions beyond this article’s scope: how family firms differ from others in managerial succession, in governance, as well as in longevity and other performance measures. Much of that literature also deals with firms larger than our SRLs (or partnerships). As Burkart et al. (2003: 2167) stress, outside the United States and the United Kingdom today, the majority of even publically-traded firms are controlled by a family group. Family groups also control some huge public companies in the United States and the United Kingdom.23 Family enterprises are even more numerous among firms of the size of our partnerships and SRLs. Why do we see family firms? A family group that ties up most of its wealth in a single enterprise forgoes opportunities to diversify. A firm that limits its managers and investors to related individuals forgoes a broader pool of talent and capital, as well as potential connections to suppliers and customers. These costs apply to families owning SMEs (such as our partnerships and SRLs) as well as to families owning the listed corporations of the finance literature. Economists offer several answers to this puzzle. Altruism among family members reduces the incentive to act opportunistically. Family members also share other social relationships that sustain honest behavior within the family business; “Severe misconduct involves not just risk of dismissal from a job but also the risk of ostracism or expulsion from the family, a penalty drastic enough that it is likely to be an effective deterrent to serious malfeasance” (Pollack 1985: 586). Because of implicit insurance obligations and an interest in the family’s continuing prosperity, family members can commit to the firm’s long-term future, James (2006) stresses an argument common to studies of successful family businesses, arguing that a family manager running a family enterprise can more credibly commit to long-term investment projects because dynastic considerations reduce the agency costs that lead nonfamily managers to prefer short-term profits.24 Family control of a business thus mitigates the transactions costs that Williamson (1979) stresses as responsible for firm organization. These arguments imply that in a world of complete information, and the costless writing and enforcement of contracts, going into business with relatives would be uncommon. Family connections, that is, mitigate problems that arise from an imperfect legal system. Nafziger (1969), for example, argued that Nigerian extended families supported business ties the formal legal system would not. This argument also implies that the development of formal institutions, including law, reduces the role of family in business: “Broader institutional development diminishes the value of relationships and connections, and facilitates arm’s-length contracting.”25 The SRL’s introduction does not constitute the deep and broad changes that constitute the development of a formal legal system. Rather, the SRL represented concrete changes to specific rules. The response to the SRL’s introduction, however, shows two ways in which the family ties filled-in for limitations of the partnership form. The first is liability: one can easily see the preference for family members as co-investors if owners face unlimited liability. By allowing investors to participate in the enterprise with limited liability, the new form encouraged participation by individuals who might not know one another well and for which agency problems might be higher than in a family firm. The relative capital contributions of these firms’ owners suggests a greater role for hands-off investors in the SRL than in the ordinary partnership. After 1919, about two-thirds of ordinary partnerships in Firms Sample formed with equal capital contributions from their owners, as compared to about half of SRLs. The SRL was also slightly more likely to have an owner who contributed more than half the firm’s capital, specifically 15% of partnerships versus 20% of SRLs. These differences suggest that the SRL form worked better for enterprises that had some less-involved investors with smaller capital commitments. The SRL’s ability to attract capital and skills from a broader array of individuals implied greater returns that were attractive when the risk was limited to the sum invested.26 The second difference deals with the SRL’s ability to lock-in capital. The ownership literature views the corporation in part as a legal device to lock-in capital. Partnerships, because they are subject to hold-up, require other mechanisms to support long-term investment. For the reasons stressed by the family-firms literature, kinship ties among partners provide that mechanism SRLs offered a new possibility: legal mechanisms for locking-in capital that did not require the additional formalities of a corporation. 7. Robustness Online Appendix B reports several robustness checks for the models reported here. We briefly summarize these checks and refer the reader to the Online Appendix. First, the difference between the Firms Sample and Firms Census results raises the concern that the latter suffer from omitted variables bias, as the Firms Census database lacks the information on owner characteristics that seems so important in the Firms Sample model. We estimate a version of the pre-1919 Firms Sample model that drops the variables that rely on owner characteristics. The point estimates for other variables change little. Second, our analysis takes as given that the legal forms reported by the commercial registry are the object of meaningful choices by entrepreneurs starting a new firm. One might worry that firms cared less about legal form than we think. We address this concern by estimating models in which the real enterprise form is replaced by one of two randomly-assigned placebos. The first exercise randomly assigns enterprise forms with equal probability. The second assigns forms randomly but respects the unconditional distribution of legal forms for the database in question. These models cannot distinguish the four alternative placebo forms at all, in stark contrast to the results reported in Tables 2–4. Finally, the NL model imposes more structure on choices than some alternatives, such as the multinomial probit model (MNP). Online Appendix B reports a version of the MNP model estimated using the Firms Census database. Given the structure of our data, the MNP model is poorly-identified. The model we report does not imply results much different from the NL models used in this article, however. 8. If There Had Never Been an SRL? A CounterFactual What would firms have done, after 1919, had the SRL not been a possibility but nothing else had changed? We pose this question by looking to the next-preferred option for firms that organized as SRLs. The Firm Census sample has 783 SRLs. (The model is reported in Table 2.) We consider first the 126 firms that were SRLs and for which the model assigns the SRL as the first choice. Eighty-four of these firms would have been ordinary partnerships, and 20 would have been corporations. None would have chosen the limited partnership form. Most Spanish entrepreneurs apparently viewed the SRL as a form of partnership with limited liability. A smaller group of firms treated the new form as a close substitute for the corporation, perhaps valuing the SRL’s ability to lock in capital and offer limited liability at a small scale.27 The fact that a firm organized as an SRL after 1919 implies that its organizers preferred that alternative to the other possible forms. Thus, eliminating the SRL alternative implies a loss. Looking at the types of firms forced into the “wrong” alternatives conveys a sense of the implications of a more restrictive menu. In the Firms Census model, the firms that would have been ordinary partnerships instead of SRLs were larger than the typical partnerships, having a median capitalization of 52,400 pesetas compared to 30,800 pesetas for actual partnerships. SRLs that would have been corporations were smaller than the typical corporation, having about 293,000 pesetas median capitalization.28 The post-1919 Firms Sample model implies something similar. Firms that were actual SRLs but counterfactual partnerships had more owners than actual partnerships, and the SRLs were significantly less likely to have related owners (2% versus 36%). If the SRL had not existed, firms would have been forced to shoehorn themselves into a form best suited for a different kind of enterprise. To fully understand the SRL’s impact on firm formation and investment would require comprehensive data on the sole proprietorships that constitute the majority of enterprises in most contexts. Such data are not available.29 This extensive margin between sole-owner and multi-owner firms is especially important given our findings for the SRL. The firm-level econometric models we report focus on the choice of enterprise form for multi-owner businesses. A new legal form might alter the margin between sole proprietorships and multi-owner enterprises. Entrepreneurs lacking appropriate family members, for example, might have foregone a multi-owner firm when the only reasonable option was the ordinary partnership, only setting up a multi-owner firm when the SRL provided a suitable legal structure. Still, we can ask what legal form firms would have chosen had the SRL been available in the 1890s. The estimates for the post-1919 model (Table 4) and the data for the pre-1919 period imply a sharp conclusion: no firm prior to 1919 would rank the SRL option first, and for 60% of all firms, the SRL would have been the last choice. The SRL’s legal establishment after 1919 brought with it a new kind of Spanish enterprise that drew on the possibilities of the new form. Did the SRL’s introduction lead to more multi-owner firms, or to increased investment in such firms? Figure 1 shows that the SRL appeared during a boom in new firm creation, a boom that by the late 1920s had tapered off considerably for enterprise forms other than the SRL. The short period between the SRL’s introduction and the onset of the Great Depression make it impossible to reliably estimate either breaks or trends in totals of firm formation or authorized capital in the period 1920–1936. The Yearbooks panel does provide suggestive evidence, however. We take advantage of the database’s panel structure and regress total capital invested in all other forms on the capital invested in the SRL. With year and province fixed effects as well as lags of the SRL variable, these correlations are strongly positive across a variety of specifications. Capital invested in SRLs might have reduced capital invested in other forms, but less than peseta-for-peseta. Repeating the exercise and focusing on corporations alone yields similar results. Focusing on the ordinary partnership alone yields a different picture: there is essentially no correlation between capital invested in the SRL and capital invested in ordinary partnerships. The same results hold broadly for numbers of firms as opposed to capital.30 9. Conclusions Spanish company law reflected the basic ideas current in most civil-law countries in the 19th century, offering entrepreneurs a choice of ordinary and limited partnerships or corporations. Most firms organized as ordinary partnerships, with corporations common only in particular sectors and for the largest enterprises. Spain added to the menu of enterprise forms when it introduced its version of the private LLC in 1919. The SRL was immediately popular, displacing the ordinary and limited partnerships. Today, the overwhelming majority of new firms in Spain take the form of an SRL.31 Many of our findings confirm earlier results from other contexts. For example, the largest enterprises, especially those subject to untimely dissolution costs, favored the corporation. Three more specific findings demonstrate the importance of studying the full range of legal forms using models that can capture the substitution possibilities at work. First, the SRL displaced both partnerships and corporations. The would-have-been partnerships that became SRLs were larger than the typical partnership, and the reverse is true for corporations. The SRL occupied a true middle ground, offering many of the advantages of incorporation in a framework that maintained the partnership’s simplicity. Second, if the SRL had been introduced in the 1890s, virtually no firm would have adopted it. The multi-owner firms organized in the 1890s differed considerably from the SRL. The SRL’s advocates argued that Spanish business had changed in ways that required a new form, and this finding suggests they were right. Finally, on the margin, the SRL appealed more to investor groups lacking familial ties than did the ordinary partnership. This result addresses a long literature on family business (both in the past and today) that emphasizes the family’s role in mitigating information and incentive problems within the firm. The SRL offered both limited liability and capital lock-in, thus providing a stronger structure that could substitute for family bonds in firm governance. These results have implications beyond Spain. The choice of legal form transcends any particular country or period in the past two centuries. Many countries introduced legal forms similar to the SRL starting at the end of the 19th century. Prominent examples include Germany’s GmbH (1892), the UK’s PLC (1907), and France’s SARL (1925). As noted at the outset, US states have recently expanded the menu of organizational forms they offer to business enterprises. We do not expect that the introduction of a new legal form in New York State, for example, will mimic the patterns we find for Spain. But the new forms we see in recent decades share the feature of offering to smaller firms the possibility of organizing with limited liability without the reporting and other burdens of the corporation and without sacrificing the contractual freedom of the partnership. The underlying issues are the same: the way heterogeneous firms select and adapt an enterprise form to minimize the costs of establishing and running the firm. Supplementary material Supplementary material is available at Journal of Law, Economics, & Organization online. The authors appreciate helpful comments from Cihan Artunç, Benito Arruñada, José Antonio Espín-Sánchez, Amanda Gregg, Fernando Gómez-Pomar, Leslie Hannah, Philip Haile, Eric Hilt, Naomi Lamoreaux, Thomas Mroz, Aldo Musacchio, Tom Nicholas, Jean-Laurent Rosenthal, Francesca Trivellato, Maribel Sáez-Lacave, and Ebonya Washington. The Registros Mercantiles of the provinces represented in the Firms Sample database all provided valuable cooperation in data-collection (Albacete, Alicante, Almería, Barcelona, A Coruña, Cuenca, Madrid, Murcia, Navarra, Sevilla, Tarragona, Toledo, Valladolid, Vizcaya, Zaragoza). Portions of this paper draw on Guinnane’s joint research with Naomi Lamoreaux, Ron Harris, and Jean-Laurent Rosenthal. This research was partially funded by the National Science Foundation, the Yale University Economic Growth Center, the McMillan Center at Yale, the Spanish Ministry of Economy and Competitiveness (HAR2013-42013-R). Footnotes 1. See Hansmann (1996: Chapter 2) for an account that expands on the idea that ownership reflects contracting costs within the firm. A different literature treats partnerships as the optimal organization for firms whose most valuable assets are human capital, focusing on the question of worker incentives and abstracting from many of the issues we consider. Examples include Morrison and Wilhelm (2008) or Levin and Tadelis (2005). A related literature discusses partnerships in the context of moral hazard in team production; see Kaya and Vereshschagina (2014). Cai (2003) extends the Grossman-Hart-Moore ownership model to consider the case where investors can hold both specific and general investments. Partnerships have indeed become uncommon outside of professional organizations, but in our historical context, they were the most common enterprise form and appear in trade, industry, and other sectors where human capital is not the most important asset. 2. For recent European discussion see European Commission (2011). 3. Musacchio and Turner (2013) stress a broader point, arguing that while ostensibly historical, the “law and finance” hypothesis gets the history wrong. 4. For a cross-country approach to the determinants of firm formation in the period 2003–2005, see Klapper et al. (2010). The World Bank’s “Doing Business” project advocates liberalizing the rules governing formation of new firms (Djankov et al. 2002; Kappler et al. 2006). Critics of the “Doing Business” project argue that the ability to formalize and commit to specific organizational forms enables firms to attract outside funding and to contract with other economic agents. See Arruñada (2007, 2008). 5. Depending on bankruptcy law, liability rules may be default rules. An owner can wave limited liability to, for example, use personal property to secure a business loan. For a firm with unlimited liability to negotiate limited liability, however, is harder. As Posner (2007: 422) notes, “In principle, the enterprise could include in all its contracts with customers and suppliers a clause limiting its liability to the assets of the enterprise. But the negotiation of such waivers would be costly. And it would be utterly impractical to limit most tort liability in this way.” 6. Online Appendix Table A.1 provides an overview of the central features of each legal form for Spain and, for comparison, four large industrial economies. 7. Guinnane and Martínez-Rodríguez (2014) provides more detail on the development of Spanish company law. Martínez-Rodríguez (2016) focuses on the SRL’s formalization in 1919, which reflected a bureaucratic regulation rather than a new statute. The episode illustrates the central role of notaries and the commercial registry, as Arruñada (2010, 2012) has argued. Spanish notaries (like civil-law notaries in general) both advise their clients and certify that contracts meet the law’s requirements. Spain allows firms to organize under the commercial code or the civil code. We focus on the former; firms organized under the civil code were generally smaller and less important for economic performance. 8. Guinnane et al. (2007) call the SRL and forms like it PLLCs, for “private limited-liability companies.” Examples include the German Gesellschaft mit beschränkter Haftung (or GmbH, 1892), the French Société à responsabilité limitée (or SARL, 1925), and the British Private Limited Company (1907). Many jurisdictions created a form of this type, sometimes under the influence of the initial examples. Guinnane and Martínez-Rodríguez (2016) provide more detail on the SRL and the way it was used in the 1920s and 1930s. 9. Martín Aceña (1993), for example, relies on the Madrid commercial registry for the period 1830–1848 to study enterprise-form choice and its relationship to firm size and sector. He credits others with pioneering this use of the source. 10. Gómez-Galvarriato and Musacchio (2004), like two other papers from the project, are unfortunately not published. See Gómez-Galvarriato and Musacchio (2006) as well as Gómez-Galvarriato et al. (2008). 11. Online Appendix A provides additional detail and descriptive statistics. 12. In principle, Spanish firms were also obliged to register their dissolution, but few did. We do not use the implied longevity information because of concerns about selection bias. 13. Table 1 is limited to Firm Sample; Online Appendix Table A.2 reports analogous information for the other sources. 14. Spaniards use both the father’s and mother’s (first) surname, and women do not change their surname at marriage. These customs make it possible to identify relatives with more certainty than would otherwise be the case. Consider the ordinary partnership “Carrasco y Viuda e Hijos de Marsal,” a firm in the Firms Sample database registered in Alicante in 1934 (firm number 2029). The owners are a widow, Adela Carrasco López and her children Ramón, Alfonso, and Adela Marsal Carrasco, as well as Adela’s brother Antonio Carrasco López. These naming practices allow us to identify those who are related through a female line. Our coding relies on kinship terms used in the articles of association (for example, “her son”). We limit the definition to parents, children, siblings, and in-laws. Some of the individuals we consider unrelated may be cousins or more extended kin; since we cannot always identify such relationships, we focus on the narrower but better-measured definition. 15. Online Appendix B discusses alternative modeling approaches and provides robustness checks for our econometric models. 16. All new firms paid the notary who drafted the articles of association and incurred fees from the commercial registry. In addition, the government imposed taxes that depended in part on the firm’s total capital. Tax considerations are an important element in the decision about enterprise form, but extant firms faced a number of taxes that changed over the period we study and bear no simple relation to legal form. 17. Gómez-Galvarriato et al. (2008) study the untimely dissolution of enterprises during the Mexican Revolution (1910–1929), stressing the importance of sector for the firm’s persistence. 18. We top code the duration variable at 20 years. Only a few firms state a longer duration. 19. For the individual branches, the smallest Wald χ2 is 61.64, which with 19 degrees of freedom rejects the null hypothesis at any confidence level. 20. For example, for the null hypothesis that the year effects are the same for the limited partnership and the SRL, the χ2 statistic is 6.89, which with 2 d.f. has a “p-value” of 0.0319. 21. Online Appendix Table B.2 reports AMEs for all enterprise forms. The standard errors are estimated by 200 bootstrap replications. The capital AMEs account for both the log of capital and its square. 22. SRLs could adapt their management and governance as they saw fit. They could have the formal boards of directors and outside managers of the corporation, or could designate an owner to run the firm, like in a partnership. Thus the SRL/corporation margin probably does not reflect governance considerations at a formal level. 23. More recently, scholars have tried to shift the focus to enterprises more of the size of our SRLs and partnerships. The finance literature often stresses the political-economic implications of family-owned business groups, which again is beyond our scope. Mehrotra and Morck (2013: 653) point out that “in many countries, each of a very small number of business families controls a large business group, containing scores, sometimes even hundreds, of seemingly distinct corporations.” We have no indications that single families controlled more than one of the partnerships or SRLs in our database, although we cannot rule out individual exceptions. Recent surveys include Bertrand and Schoar (2006), Mehrotra and Morck (2013), and Casson (1999). Casson focuses on the historical aspect of the question. 24. A different strand in this literature notes the downside of these connections: a family firm may be less likely to fire a poorly-performing employee who is a relative. See Karra et al. (2006). 25. Mehrotra and Morck (2013: 665). As they note, this argument echoes a long tradition in social science. 26. Colli et al. (2003: 36) offer a related view of unlimited liability for the family. 27. This counterfactual only considers the SRLs ranked most-preferred by the model. If we instead consider the next-preferred option after the SRL for all SRLs, even those ranked second or worse by the model, the results are similar. 28. The 52,400 pesetas in 1925 would be worth about €105,000 in 2016. And 293,000 pesetas would thus be €589,000. Computed using the Spanish cost of living series on http://www.measuringworth.com 29. Most sole proprietorships did not enroll in the commercial registry. Other counts of sole proprietorships in Spain imply implausibly low totals. 30. The relationship also holds in first differences and in models estimated for the period 1921–1928 along, thus excluding the Great Depression. The regressions are reported in the Online Appendix. 31. According to the February 2017 data, SRLs accounted for 99.45% of firms and 90.8% of authorized capital. Official figures from the Instituto Nacional de Estadística:http://www.ine.es/daco/daco42/daco424/sm0417.xls. References Abramitzky Ran, Frank Zephyr, Mahajan Aprajit. 2010. “Incentives and Contracts; Partnerships in Rio de Janeiro,” 70 Journal of Economic History  1870– 92. Google Scholar CrossRef Search ADS   Acemoğlu Daron, Johnson Simon. 2005. “Unbundling Institutions,” 113 Journal of Political Economy  949– 95. Google Scholar CrossRef Search ADS   Arruñada Benito. 2007. “Pitfalls to Avoid When Measuring the Institutional Environment: Is ‘Doing Business’ damaging business?” 35 Journal of Comparative Economics  729– 47. Google Scholar CrossRef Search ADS   Arruñada Benito. 2008. “How Doing Business Jeopardizes Institutions Reform,” 10 European Business Organization Law Review  555– 74. Google Scholar CrossRef Search ADS   Arruñada Benito. 2010. “Institutional Support of the Firm: A Theory of Business Registries,” 2 Journal of Legal Analysis  525– 76. Google Scholar CrossRef Search ADS   Arruñada Benito. 2012. Institutional Foundations of Impersonal Exchange. Theory and Policy of Contractual Registries . Chicago: University of Chicago Press. Google Scholar CrossRef Search ADS   Ben-Porath Yoram. 1980. “The F-Connection: Families, Friends, and Firms and the Organization of Exchange,” 6 Population and Development Review  1– 30. Google Scholar CrossRef Search ADS   Bertrand Marianne, Schoar Antoinette. 2006. “The Role of the Family in Family Firms,” 20 Journal of Economic Perspectives  73– 96. Google Scholar CrossRef Search ADS   Blair Margaret. M. 2003. “Locking in Capital: What Corporate Law Achieved for Business Organizers in the Nineteenth Century,” 51 UCLA Law Review  387. Burkart Mike, Panunzi Fausto, Shleifer Andrei. 2003. “Family Firms,” 58 Journal of Finance  2167– 201. Google Scholar CrossRef Search ADS   Cai Hongbin. 2003. “A Theory of Joint Asset Ownership,” Rand Journal of Economics  63– 77. Casson Mark. 1999. “The Economics of the Family Firm,” 47 Scandinavian Economic History Review  10– 23. Google Scholar CrossRef Search ADS   Chandler Alfred. 1977. The Visible Hand: the Managerial Revolution in American Business . Cambridge: Harvard University Press. Colli Andrea, Fernández-Pérez Paloma, Rose Mary B.. 2003. “National Determinants of Family Firm Development?” 4 Enterprise & Society  28– 64. Google Scholar CrossRef Search ADS   Djankov Simeon, La Porta Rafael, Lopez-de-Silanes Florencia, Shleifer Andrei. 2002. “The regulation of entry,” 117 Quarterly Journal of Economics  1– 37. Google Scholar CrossRef Search ADS   European Commission. 2011. Report of the Reflection Group on the Future of EU Company Law . Brussels: European Commission. Gómez-Galvarriato Aurora, Musacchio Aldo. 2004. “Organizational Choice in a French Civil Law Underdeveloped Economy: Partnerships, Corporations and the Chartering of Business in Mexico, 1886 –1910.” Harvard Business School Working Paper 24. Gómez-Galvarriato Aurora,, Musacchio Aldo. 2006. “Larger Menus and Entrepreneurial Appetite: An Empirical Investigation of Organizational Choice in Mexico, 1886-1910.” Paper presented in Session 74 of the XIV International Economic History Conference. Gómez-Galvarriato Aurora, Musacchio Aldo, Parral Rodrigo. 2008. “Political Instability and Untimely Dissolution: Partnerships, Corporations, and the Mexican Revolution, 1910-1929.” Harvard Business School Working Paper 08-092. Grossman Sanford J., Hart Oliver D.. 1986. “The Cost of Benefic of Ownership: A Theory of Vertical and Lateral Integration,” 94 Journal of Political Economy  691– 719. Google Scholar CrossRef Search ADS   Guinnane Timothy, Harris Ron, Lamoreaux Naomi, Rosenthal Jean-Laurent. 2007. “Putting the Corporation in its Place,” 8 Enterprise & Society  687– 729. Google Scholar CrossRef Search ADS   Guinnane Timothy, Martínez-Rodríguez Susana. 2014. “Flexibility in the Spanish Company Law, 1886-1936,” 56 Revista de Historia Industrial  81– 113. Guinnane Timothy, Martínez-Rodríguez Susana 2016. “Instructions Not Included: Spain's Sociedad de Responsabilidad Limitada, 1919–1936”. Working paper. Hansmann Henry. 1996. The Ownership of Enterprise . Cambridge MA: Harvard University Press. Hansmann Henry, Kraakman Reiner, Squire Richard. 2006. “Law and the Rise of the Firm,” 119 Harvard Law Review  1335– 403. Hart Oliver, Moore John. 1990. “Property Rights and the Nature of the Firm,” 98 Journal of Political Economy  1119– 58. Google Scholar CrossRef Search ADS   Hilt Eric,, O’Banion Katherine. 2009. “The Limited Partnership in New York, 1822-1858: Partnerships without Kinship,” 69 Journal of Economic History  615– 45. Google Scholar CrossRef Search ADS   Hilt Eric. 2017. “Corporation Law and the Shift toward Open Access in the Antebellum United States,” in Lamoreaux Naomi R., Wallis John J., eds., Organizations, Civil Society, and the Roots of Development . Chicago: University of Chicago Press (forthcoming). James Harold. 2006. Family Capitalism: Wendels, Haniels, Falcks, and the Continental European Model . Cambridge: Harvard University Press. Kappler Leora, Laeven Luc, Rajan Raghuram. 2006. “Entry Regulation as a Barrier to Enterpreneurship,” 82 Journal of Financial Economics  591– 629. Google Scholar CrossRef Search ADS   Kaya Ayca, Vereshchagina Galina. 2014. “Partnerships Versus Corporations: Moral Hazard, Sorting, and Ownership Structure,” 104 The American Economic Review  291– 307. Google Scholar CrossRef Search ADS PubMed  Klapper Leora, Amit Raphael, Guillén Mauro F.. 2010. “Entrepreneurs and Firm Formation across Countries,” in Lerner Josh, Schoar Antoinette, eds., International Differences in Entrepreneurships . Chicago: University of Chicago Press. Lamoreaux Naomi, 2016. “Corporate Governance and the Expansion of the Democratic Franchise: Beyond Cross-Country Regressions,” 64 Scandinavian Economic History Review  103– 21. Google Scholar CrossRef Search ADS   Lamoreaux Naomi,, Rosenthal Jean-Laurent. 2005. “Legal Regime and Contractual Flexibility: A Comparison of Business’s Organizational Choices in France and the United States during the Era of Industrialization,” 7 American Law and Economics Review  28– 61. Google Scholar CrossRef Search ADS   La Porta Rafael, Lopez-de-Silanes Florencio, Shleifer Andrei. 2008. “The Economic Consequences of Legal Origins,” 46 Journal of Economic Literature  285– 332. Google Scholar CrossRef Search ADS   La Porta Rafael, Lopez-de-Silanes Florencio, Shleifer Andrei, Vishny Robert W., 1998. “Law and Finance,” 106 Journal of Political Economy  1113– 55. Google Scholar CrossRef Search ADS   Levin Jonathan, Tadelis Steven. 2005. “Profit Sharing and the Role of Professional Partnerships,” 120 The Quarterly Journal of Economics  131– 71. Martín Aceña Pablo. 1993. “La Creación de Sociedades en Madrid (1830-1848). Un Analisis del Primer Registro Mercantil,” Documento de trabajo- Universidad de Alcala y Fundación Pública  9303. Martín Aceña Pablo, Pons María Ángeles. 2005. “Sistema monetario y financiero,” in Carreras Albert, Tafunell Xabier, eds., Estadísticas Históricas de España, siglo XIX y XX . Bilbao: Fundación BBVA. Martínez-Rodríguez Susana. 2016. “Creating the Sociedad de Responsabilidad Limitada: the Use of Legal Flexibility in Spanish Company law, 1869-1953,” 90 Business History Review  227– 49. Google Scholar CrossRef Search ADS   Mehrotra Vikas, Morck Randall. 2013. “Entrepreneurship and the Family Firm,” in Constantinides George M., Harris Milton, Stulz Rene M., eds., Handbook of the Economics of Finance  2 649– 81. Entrepreneurship and the Family Firm. Amsterdam: Elsevier. Google Scholar CrossRef Search ADS   Morrison Alan D., Wilhelm William J.Jr. 2008. “The Demise of Investment Banking Partnerships: Theory and Evidence,” 63 Journal of Finance  311– 50. Google Scholar CrossRef Search ADS   Musacchio Aldo, Turner John D., 2013. “Does the Law and Finance Hypothesis Pass the Test of History?” 55 Business History  521– 39. Google Scholar CrossRef Search ADS   Nafziger E. Wayne, 1969. “The Effect of the Nigerian Extended Family on Entrepreneurial Activity,” 18 Economic Development and Cultural Change  25– 33. Google Scholar CrossRef Search ADS   Karra Neri, Tracey Paul, Phillips Nelson, 2006. “Altruism and Agency in the Family Firm: Exploring the Role of Family, Kinship, and Ethnicity,” 30 Entrepreneurship Theory and Practice  861– 77. Google Scholar CrossRef Search ADS   Nicholas Tom. 2015. “The Organization of Enterprise in Japan,” 75 The Journal of Economic History  333– 63. Google Scholar CrossRef Search ADS   Pollack Robert. 1985. “A Transaction Cost Approach to Families and Households,” 23 Journal of Economic Literature  581– 608. Posner Richard A. 2007. Economic Analysis of Law . 7th ed. New York: Wolters. Ribstein Larry E. 2010. The Rise of the Uncorporation . New York: Oxford University Press. Spulber Daniel F. 2009. The Theory of the Firm: Microeconomics with Endogenous Entrepreneurs, Firms, Markets, and Organizations . New York: Cambridge University Press. Google Scholar CrossRef Search ADS   Stout Lynn A. 2005. “On the Nature of Corporations,” Univeristy of Illinois Law Review . 2005: 253. Williamson Oliver. 1979. “Transaction-Cost Economics: The Governance of Contractual Relations,” 22 Journal of Law and Economics  233– 61 Google Scholar CrossRef Search ADS   © The Author(s) 2018. Published by Oxford University Press on behalf of Yale University. All rights reserved. For Permissions, please email: journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png The Journal of Law, Economics, and Organization Oxford University Press

Choice of Enterprise Form: Spain, 1886–1936

Loading next page...
 
/lp/ou_press/choice-of-enterprise-form-spain-1886-1936-LBMHxFPMwO
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of Yale University. All rights reserved. For Permissions, please email: journals.permissions@oup.com
ISSN
8756-6222
eISSN
1465-7341
D.O.I.
10.1093/jleo/ewx015
Publisher site
See Article on Publisher Site

Abstract

Abstract Every new firm selects a legal form. Organizing as a corporation, a limited company, or a partnership shapes the firm’s access to capital markets, its governance arrangements and tax liabilities, and its treatment in bankruptcy. We use multinomial choice models to estimate the determinants of enterprise form using firm-level data on Spain for the period 1886–1936. Our results support hypotheses drawn from the corporate-finance and ownership literatures; entrepreneurs preferred the corporation for the largest firms and for firms vulnerable to holdup. In 1919, Spain introduced a new legal form, a limited company combining attributes of the corporation and the partnership. This Sociedad de Responsabilidad Limitada (SRL) displaced both corporations and partnerships, and was especially popular for small and medium-sized firms whose owners were unrelated. Counterfactual calculations suggest that few enterprises created prior to 1919 would have chosen the SRL even if it had been available. Entrepreneurs creating a new multi-owner firm face a fundamental contracting decision: the enterprise’s legal form. In most legal regimes today, multi-owner firms can be corporations, limited-liability companies, or partnerships, as well as variants on these basic structures. A firm’s enterprise form shapes its owners’ liability, the firm’s access to public equity markets, and governance matters such as the rights of minority owners. Tax codes often treat corporations differently from other forms. Economists and others have studied each of these issues in isolation, and a theoretical literature studies the margin between partnerships and corporations in some contexts. Empirical work on these choices has, however, been limited. This article uses multinomial choice models with newly-collected micro-data on firm formation to study decisions about legal form in a context that offers both rich data and the absence of some complicating influences: Spain in the period 1886–1936. Using micro-evidence from a single country allows us to abstract from potentially confounding influences such as differences in national banking systems, the organization of equity markets, and the tax treatment of different forms. Most importantly, Spain introduced an entirely new legal form in 1919. The Sociedad con Responsabilidad Limitada (SRL) allowed entrepreneurs to combine some features of the corporation with the partnership’s greater flexibility. We use discrete-choice models to ask which firms preferred this new form as well as conduct counterfactual exercises about what would have happened in the absence of the new form. Spain’s business code offered a discrete set of legal forms from which to choose. The commercial registration system allows us to sample from the universe of multi-owner firms created in that period, and the notarial system yields a set of contracts with consistently-defined provisions. Few other countries systematically collect and retain information on partnerships, limiting empirical investigations to corporations alone; Spain offers a rare opportunity to study decisions across the full range of legal forms. Economists have paid less attention to partnerships than to corporations, but the available theory implies testable hypotheses about the trade-offs among enterprise forms. Chandler (1977) stressed that partnerships could not gather the large amounts of capital needed to take advantage of the new technologies of the nineteenth century. In Spulber (2009)’s model, partnerships self-finance and so have lower capital costs than corporations, but suffer from free-riding because of a lack of concentrated management. The corporation pays higher capital costs than a partnership but has a hierarchical structure that reduces free-riding. For small-scale enterprises the partnership form generates more surplus, but as the firm grows and needs more partners to supply capital, agency costs increase and the corporate form dominates. Spulber’s model thus implies a clear prediction about size and adoption of the corporate form. A different perspective associated with Williamson (1979), Grossman and Hart (1986), and Hart and Moore (1990) treats ownership as central to the firm’s governance structure. Assignment of ownership mitigates contractual incompleteness. Thus, transactions costs have important implications for the choice of the corporate form. The corporation usually has more costly reporting requirements and legal formalities than those required of partnerships, but the corporate form offers two attractive features. Equity investments enjoy limited liability, making it easier to attract capital from many small investors who play no meaningful role in the firm’s management. Corporations also lock-in capital. Partnerships, in contrast, are contracts among individuals. Any partner who wants to withdraw can do so, thereby breaking up the firm. Whether this withdrawal reflects exogenous forces (such as partner mortality) or strategic efforts to obtain better terms (hold-up), the possibility of withdrawal makes the partnership unsuitable for investment in specific assets. Enterprises that require specific investments should prefer the corporate form.1 We have three sets of empirical conclusions concerning the choice of enterprise form. First, we confirm the two basic predictions about the margin between partnerships and corporations as implied by the economics literature. Larger enterprises and businesses with specific capital investments avoided partnership forms. These results are not surprising, but the situation we study provides a rare opportunity to test these hypotheses in a rigorous way using firms from across the full range of sectors. Second, we demonstrate the substitution implications of the SRL’s introduction. The SRL displaced both ordinary partnerships and, less often, corporations. This pattern reflects both the new form’s essential characteristics as well as its flexibility, which allowed entrepreneurs to tailor their SRL to suit their specific needs. Counterfactual calculations demonstrate that had the SRL been introduced in the 1890s, few firms would have taken advantage of the form. The SRL form reflected a new type of business in the 1920s. These new businesses took advantage of new technologies, reached new markets, and reflected the efforts of new types of investor groups. Third, our results demonstrate both the continuing importance of family ties in business as well as the interaction between kin connections and legal form. One important characteristic of investor groups favoring the SRL was their lack of family ties. Family groups continued to favor the ordinary partnership. The SRL’s introduction allowed investors to rely on law, rather than kin ties, to contend with information, incentive, and monitoring problems in the multi-owner enterprise. This result confirms the central contention in the family-firms literature and suggests an unappreciated connection between family control and enterprise contracting choices. Recent decades have seen the introduction of new legal forms for business enterprises in many economies. Ribstein (2010) calls these firms “uncorporations” to underline the essential differences between corporations, older partnership forms, and the new limited-liability companies (LLCs). During the 1990s, every US state introduced a new limited-liability partnership (LLP) form. Many US states have also created or reformed the LLC, which allows owners to obtain the benefits of the corporation in a form more suitable to smaller enterprises (Hansmann et al. 2006). Most then-wealthy countries other than Britain and the United States allowed the LLP from the early 19th century, and forms similar to the LLC appeared starting in Germany in 1892. Development of enterprise forms remains an active interest of policymakers in several contexts. The European Union and its Member States have also considered revising their menu of enterprise forms.2 Economists now widely agree that institutions affect economic performance. A growing literature discusses what constitutes good or bad institutions and searches for statistical relationships between those institutions and economy performance. Much of this literature uses country-level aggregates, regressing indicators such as GDP per capita on variables intended to capture one or more features of the institutional environment. In Acemoğlu and Johnson (2005), for example, the institutional variables are summary indices such as the country’s degree of legal formalism, constraints on its executive, and measures of protection against expropriation. The “law and finance” argument attributed to La Porta et al. (1998) (see also La Porta et al. 2008) uses similar empirical methods but focuses on a narrower issue. These authors divide the world’s countries according to legal family; “common law” countries include Britain and its former colonies, whereas the “civil law” countries include most of Continental Europe as well as the former colonies of France, Portugal, and Spain. The civil-law system’s greater reliance on legal codes and statutory law, as opposed to the case-law and jurisprudence at the heart of the common-law system, they argue, makes the civil-law systems more rigid. La Porta et al. use cross-country regression to argue that civil-law countries have less well-developed investor and creditor protection, less well-developed financial systems, and as a result, enjoy lower rates of economic growth. Lamoreaux (2016) notes that cross-country regressions miss most of the important developments in late-19th century company law.3 This is especially true of the United States, where most relevant law is state-level. Our approach differs from these agendas in two important ways. First, by relying on firm-level data and concrete legal rules, we study the effect of well-defined, specific institutions on the behavior of optimizing agents. Second, the SRL’s introduction allows us to see how those agents respond to a change in the institutional environment, a feature usually absent from the cross-country approach.4 1. Law and Enterprise Form Spain’s business code owes much to French legal traditions, a trait common to many civil-law countries. The 1885 Spanish code allowed firms to organize as ordinary partnerships, limited partnerships, or corporations. The ordinary partnership required unlimited liability of its owners. In a limited partnership, some owners could restrict their liability to the sum they had invested, but could not participate in management. All investors in Spanish corporations had limited liability.5 Another important difference makes the partnership subject to hold-up problems. Partnerships are in practice at-will and so cannot lock in capital effectively; if a partner wanted to withdraw, he could leave and take his investment with him. The only way for an investor to exit a corporation is to sell his shares to someone else. Some legal scholars view capital lock-in as the corporation’s essential feature (see Stout (2005) or Blair (2003)). Lock-in is also a feature of limited-liability companies, including the SRL.6 Civil-law countries, such as Spain, ordinarily required firms to use one of the legal forms specifically described in the law. The 1885 Spanish code, however, quite unusually allowed firms to modify the standard forms, so long as the resulting firm respected the code’s other requirements (§117, 122). Some firms took advantage of this flexibility to adopt features of the new enterprise forms appearing elsewhere in Europe. The practice remained rare, however, until 1919, when the Ministry of Justice issued instructions to commercial registries requiring the recognition of the new SRL. The rules governing the SRL relied on the 1885 code’s partnership rules and case law until 1953, when Spain passed its first legislation governing the SRL.7 The SRL resembled a number of new enterprise forms introduced in Europe in the late 19th and early 20th centuries. These new forms enabled entrepreneurs to create a firm that combined some features of the corporation with some features of the partnership. The SRL’s owners all had limited liability for their investments and their ownership stakes were freely transferrable unless restricted by the articles of association. SRL investments were locked-in. These features all mimicked the corporation. Unlike the corporation, however, the SRL could not list its equity on public markets. The new form also escaped some formalities (such as publicity) required of corporations. An SRL could pay one of its owners to run the firm, but managers could also be outsiders who had no ownership stake. The new form retained the flexibility of partnership contracting, allowing, for example, complex distinctions between cash flow and control rights.8 Debates over legal forms and the desirability of reforming or introducing new forms date at least to the early 19th century. Until the middle of the 19th century, most European countries strictly controlled use of the corporate form, shifting to “general incorporation” only later. (The United States was unusual in that many states introduced general incorporation earlier in the nineteenth century; Hilt 2017.) The second half of the 19th century saw major reforms to the corporation in leading economies, as well as the introduction of the first SRL-like enterprise forms (Guinnane et al. 2007). The United Kingdom’s 1862 Companies Act and later court decisions made the corporation attractive to small enterprises and a 1907 Act formally created the Private Limited Company (PLC). Germany introduced its PLC (the Gesellschaft mit beschränkter Haftung or GmbH) in 1892. Both the British PLC and the German GmbH enjoyed many imitators. Spain’s introduction of the SRL reflects, in part, influence from the outside. Developments in company law during the 19th and 20th centuries have been the basis of sustained attacks on the La Porta et al. (1998) “law and finance” hypothesis. Lamoreaux and Rosenthal (2005) noted that France (a civil-law country) offered entrepreneurs more flexibility than did the United States (a common-law country). Gómez-Galvarriato and Musacchio (2006) show that Mexico, another civil-law country, also enjoyed a more flexible company law than is consistent with the law and finance claim. Guinnane et al. (2007) extended the comparison to include both Britain and Germany (another civil-law country); the German GmbH reflects a fairly radical legal innovation of the type that La Porta et al. (1998) views as unlikely in civil-law countries. Some 19th-century governments, including Spain, published summary information on the choice of legal form. These reports formed the basis for empirical discussions of the topic. More recently, scholars have created firm-level databases from business registries.9 Econometric studies of firm-level choices are, however, recent. The first work of this type may be Gómez-Galvarriato and Musacchio (2004), which uses micro-data from Mexico City at the end of the 19th and early 20th centuries. Other econometric studies of enterprise-form choice include Hilt and O’Banion (2009) (studying the difference between ordinary and limited partnerships in New York State) and Abramitzky et al. (2010) (the same question, but in Brazil).10Nicholas (2015) studies trade-offs among enterprise forms in Japan between 1896 and 1939, stressing that the SRL-like form introduced in 1938 did not replace the corporate form. 2. Data on Choices of Legal Form in Spain To enjoy the commercial code’s protection, a firm registered its formation and provided a summary of its articles of association to the provincial branch of the commercial registry. We have assembled three distinct databases based on this registry. The first source records the total number of firms organized under each legal form, along with the total capital in these firms for every year between 1886 and 1936. These reports (Yearbooks) provide a rare and complete picture of how firms organized.11 The other two sources have firm-level data taken from the firm’s articles of association. The first micro-level source encodes a publication that reports selected information on every Spanish firm that registered in the years 1925, 1926, and 1927. The information available in this source, which we call “Firms Census,” is limited to the variables published in the original source. The second micro database (Firms Sample) comprises a random sample of enterprises from the archives of 15 peninsular provincial commercial registries. For each of these provinces, we randomly selected two firms formed in every year from 1886 to 1936. Firms Sample reflects the limited number of accessible Spanish archives, along with the considerable labor required to code each observation. Many of our results rely entirely on Firms Census, which includes every firm formed in Spain in those years. Our data reflect the firm’s initial characteristics as provided to the commercial registry upon formation. Attributes such as capital stock could change over the firm’s life, though partnerships rarely changed hands. The individuals we consider to be a corporation’s “owners,” however, are those who signed the articles of association. Our sources do not include subsequent performance measures, so we cannot, unfortunately, ask how legal form affected profitability.12 Figure 1 uses the Yearbooks to summarize the distribution of new firms by legal form along with the total number of registrations for all of Spain over our period. The majority of firms in 1886 were ordinary partnerships, with a small but stable fraction organized as limited partnerships. Corporations accounted for a modest share of new firms. The period prior to the SRL’s formalization in 1919 saw the slow decline of the partnership and the equally slow rise of the corporation. The loss of Spain’s remaining colonies in 1899 led to the repatriation of considerable capital and an increased number of new corporations starting in that year. By 1936, only a handful of new companies registered as limited partnerships. The new SRL and the corporation’s growing use made the ordinary partnership considerably less popular than earlier. Figure 1 also reports the total number of annual firm registrations. Spain experienced slow growth in new firms up to World War I, then a short boom in new registrations that ended in the early 1920s. This episode reflects the advantage of being neutral in a Europe at war. Figure 1. View largeDownload slide Distribution of Legal Forms for all Enterprises Registered in Spain, 1886–1936, Along with Total Number of Firm Registrations. Source: Yearbooks database. Notes: Left axis is percentage of firms registering as a given legal form; right axis is the total number of registrations. See text for more discussion of legal forms. Figure 1. View largeDownload slide Distribution of Legal Forms for all Enterprises Registered in Spain, 1886–1936, Along with Total Number of Firm Registrations. Source: Yearbooks database. Notes: Left axis is percentage of firms registering as a given legal form; right axis is the total number of registrations. See text for more discussion of legal forms. Table 1 summarizes three dimensions of our sample firms.13 Measured by capitalization and the number of owners, the SRLs were somewhat larger than the two partnership forms. Corporations were larger than all other forms. We constructed the two “family connections” variables from surnames and internal references (e.g. “and his son …”).14 Owners who are all related preferred the ordinary to the limited partnership. When the owners combined both relatives and nonrelatives, the limited partnership was more common. Table 1 also summarizes the practice of stating a contractual duration in a firm’s articles of association. This was optional, and most firms left it open, but corporations were most likely to include a fixed duration in their articles, and to define it as longer than 10 years. Table 1. Characteristics of Firms by Legal Form   Ordinary partnerships  Limited partnerships    1886–1898  1899–1918  1919–1936  1886–1898  1899–1918  1919–1936  Average capital  58,296  42,279  103,971  150,039  88,382  165,188  Average number of owners  2.68  2.72  2.95  3.43  3.59  3.26  Percentage of firms whose owners are all related  26.8  24.6  32.5  8.8  16.5  13  Percentage of firms for which some owners are related  12.5  12.3  10.2  35.1  20.6  29  Percentage of firms that state a contractual duration  91.9  75.3  52.5  85.9  78.3  61.3  Number of firms  224  357  175  56  97  31      Corporations  SRL    1886–1898  1899–1918  1919–1936    1919–1936      Average capital  3,330,246  631,158  2,152,169    119,255    Average number of owners  7.78  8.31  6.68    3.93    Percentage of firms whose owners are all related  2.8  1.7  13.1    10    Percentage of firms for which some owners are related  11.1  25.9  32    25.5    Percentage of firms that state a contractual duration  69.4  58  32.7    36.9    Number of firms  36  109  151    140      Ordinary partnerships  Limited partnerships    1886–1898  1899–1918  1919–1936  1886–1898  1899–1918  1919–1936  Average capital  58,296  42,279  103,971  150,039  88,382  165,188  Average number of owners  2.68  2.72  2.95  3.43  3.59  3.26  Percentage of firms whose owners are all related  26.8  24.6  32.5  8.8  16.5  13  Percentage of firms for which some owners are related  12.5  12.3  10.2  35.1  20.6  29  Percentage of firms that state a contractual duration  91.9  75.3  52.5  85.9  78.3  61.3  Number of firms  224  357  175  56  97  31      Corporations  SRL    1886–1898  1899–1918  1919–1936    1919–1936      Average capital  3,330,246  631,158  2,152,169    119,255    Average number of owners  7.78  8.31  6.68    3.93    Percentage of firms whose owners are all related  2.8  1.7  13.1    10    Percentage of firms for which some owners are related  11.1  25.9  32    25.5    Percentage of firms that state a contractual duration  69.4  58  32.7    36.9    Number of firms  36  109  151    140    Source: Computed from the firms sample database. Online Appendix Table A.3 reports analogous measures for Firms Census, where available. Notes: “Capital” is authorized capital, in current pesetas. In 1900, one pesetas equaled about £.032, or $0.149 (US) (Martín Aceña and Pons 2005: 704). Using a cost-of-living index, one 1900 peseta was worth €3.57  in 2016 (www.measuringworth.com). We determine relationships among owners using kinship term in the articles of association. The definition is limited to the nuclear family. See text for more discussion. There was no SRL form prior to 1919. We group firms into sectors using a standard industrial classification for Spain. For the econometric analysis, we further combine these sectors into four groups: factories, trading enterprises, mining and infrastructure firms, and a (residual) miscellaneous category. (See Online Appendix A Table A.4) Corporations dominate two sectors, mining and the heterogeneous category of agriculture processing, utilities, and construction. These sector classifications reflect the limitations of sample-size as well as an effort to use categories for which all provinces have firms. Within a given sector, firms organized as corporations have larger total capital investments than firms organized in other ways. The SRL was at first most popular in two large sectors: factories and trade. The new form also appealed to a range of professional and service-related firms, such as the liberal professions or travel services. 3. Modeling Firm-Level Decisions about Legal Form Entrepreneurs creating a multi-owner firm choose an enterprise form to minimize contracting problems given the characteristics of the firm and its owners, subject to the legal system’s constraints. We study this decision using multinomial discrete-choice models. We have two questions: How do the firm’s characteristics affect the choice of form? And, how did the expansion of the menu of choices represented by the SRL’s introduction affect these choices? There is no natural ordering for the choice of legal form, so we restrict our attention to unordered choice models. After considering several alternatives, we rely on nested logit (NL) models.15 The NL model can be derived from random-utility maximization, so the estimates imply, for each firm, a rank-ordering of preferences over legal forms. The NL model also reveals the SRL’s impact on other enterprise forms. When Spain introduced the new SRL, some entrepreneurs preferred that form to any of the older alternatives. For others, the new SRL was irrelevant in that entrepreneurs still preferred some alternative. Our approach allows us to study what firms would have done in the absence of the new form, and, less formally, to examine the characteristics of firms forced to make a suboptimal enterprise form choice before the SRL was a possibility. The econometric models require an assumption: that a firm’s attributes are all fixed when its founders decide on the legal form. We thus abstract from the possibility that legal form is decided as part of a negotiation over other firm attributes. Concerns over possible endogeneity of the regressors should focus on the possibility that a potential owner might hold out for limited liability or some other firm attribute, thus implying that the participation of some owners was contingent upon the choice of enterprise form. Given our sources, there is no tractable solution to this endogeneity problem, so we must be cautious in interpreting our results as “causal.” We first estimate the NL model using the Firms Census database. These results provide a useful baseline because the data reflect a census of all firms and because these results allow us to understand the post-SRL world in detail. We then turn to the Firms Sample. We define variables in the two databases identically unless otherwise noted. A brief overview will guide our discussion: First, enterprise size affects choice of legal form in three ways. Spulber (2009), as noted, stresses the trade-off between monitoring and the cost of capital in the choice between a partnership and a corporation. His model implies that larger firms prefer the corporate form. In addition, only corporations can access public equity markets, which eases the task of raising capital for the largest enterprises. (The SRL’s equity shares could not be listed on exchanges.) In addition, the corporation requires formalities that entail fixed costs (such as the publication of a balance sheet), providing another reason why smaller enterprises would prefer another form.16 We capture firm-size effects with the natural log of stated capital and its square. We also include a dummy variable for whether the firm has any unpaid capital. Firms could and did have nominal capital in excess of what owners had paid in. Shareholders remained liable to pay in more capital, as stipulated by the firm’s articles. The difference between nominal and paid-in capital also enhanced the firm’s borrowing ability. Second, firms that contemplated long-term investments in illiquid projects would have been reluctant to organize as partnerships because of potential hold-up problems. We have two proxies for concerns about untimely dissolution issues. First, in some sectors, a firm’s physical capital can easily be liquidated while in others it cannot. Thus, sector serves as a proxy for concerns about this form of hold-up.17 Mining and infrastructure firms are those most engaged in investments of the type that would create untimely dissolution problems. Second, a firm’s articles of association could stipulate that it was open-ended or would last only for a specific period. We construct a dummy variable for whether the firm has a stated duration and define another variable as the interaction of the dummy variable with the number of years stipulated as the firm’s duration.18 Third, we allow family ties among owners to affect the choice of legal form. (We discuss the relationship between enterprise form and the family business in Section 6.) Because it relies on a published source that did not report this information, we do not have the relationship variables for the Firms Census source. For the Firms Sample database, we use the number of owners and family relationships to capture aspects of the legal-form decision that may reflect owner identity. The choice of legal form might also reflect local considerations, which is one reason to use province dummies. Notaries outside the main commercial centers had less experience organizing corporations. In the years 1886 through 1936, a total of eight Spanish provinces did not see the creation of a single corporation. And if one reason to organize a corporation was to tap liquid markets for investors, either privately or by listing on the stock market, a firm located outside financial centers such as Madrid or Barcelona (or places with smaller exchanges, such as Bilbao or Valencia) might think twice about the corporate form. 4. Results The NL model requires judgment about which alternatives are similar and thus belong in the same nest. Figure 2 describes our approach. We group the enterprise forms with limited liability for at least some owners into one nest and place the ordinary partnership in its own, “degenerate” nest. Table 2 reports the NL estimates for the Firms Census model. (Online Appendix Table B.1 reports descriptive statistics for the estimation subsamples.) Several alternative nesting structures seem plausible, but the data reject the other candidates as inconsistent with the random utility-maximization model that underlies NL. The model is consistent with utility maximization only if the nesting (or “dissimilarity”) parameter (which we constrain to “1” for the degenerate nest) lies on the unit interval. Table 2. Nested Logit Model for the Firm Census Database   Limited partnership  Corporation  SRL  Firm has stated duration  0.161  −2.569  −0.213  (0.324)  (0.507)  (0.195)  Duration in years  −0.012  0.147  0.006  (0.024)  (0.031)  (0.015)  Firm has any unpaid capital  1.209  3.340  1.392  (0.641)  (0.515)  (0.438)  Ln (capital)  −0.656  −0.698  −0.278  (0.118)  (0.108)  (0.044)  Ln (capital) squared  0.043  0.064  0.022  (0.006)  (0.009)  (0.004)  Sector   Factory  0.165  −0.848  −0.256  (0.283)  (0.198)  (0.153)   Trade  0.216  −1.250  −0.289  (0.304)  (0.252)  (0.159)   Infrastructure  0.274  0.244  −0.013  (0.372)  (0.208)  (0.220)  Location   Catalonia  0.523  0.651  0.040  (0.186)  (0.153)  (0.138)   Madrid  0.524  0.492  0.376  (0.257)  (0.185)  (0.163)   Basque Country  −0.303  −0.556  0.605  (0.284)  (0.280)  (0.158)  Year firm formed   1926  0.121  0.0511  0.156  (0.170)  (0.130)  (0.113)   1927  −0.373  0.00751  0.312  (0.245)  (0.136)  (0.122)  Dissimilarity parameter  0.915      (0.236)  Number of firms  3142      Log-likelihood  −3058.1043        Limited partnership  Corporation  SRL  Firm has stated duration  0.161  −2.569  −0.213  (0.324)  (0.507)  (0.195)  Duration in years  −0.012  0.147  0.006  (0.024)  (0.031)  (0.015)  Firm has any unpaid capital  1.209  3.340  1.392  (0.641)  (0.515)  (0.438)  Ln (capital)  −0.656  −0.698  −0.278  (0.118)  (0.108)  (0.044)  Ln (capital) squared  0.043  0.064  0.022  (0.006)  (0.009)  (0.004)  Sector   Factory  0.165  −0.848  −0.256  (0.283)  (0.198)  (0.153)   Trade  0.216  −1.250  −0.289  (0.304)  (0.252)  (0.159)   Infrastructure  0.274  0.244  −0.013  (0.372)  (0.208)  (0.220)  Location   Catalonia  0.523  0.651  0.040  (0.186)  (0.153)  (0.138)   Madrid  0.524  0.492  0.376  (0.257)  (0.185)  (0.163)   Basque Country  −0.303  −0.556  0.605  (0.284)  (0.280)  (0.158)  Year firm formed   1926  0.121  0.0511  0.156  (0.170)  (0.130)  (0.113)   1927  −0.373  0.00751  0.312  (0.245)  (0.136)  (0.122)  Dissimilarity parameter  0.915      (0.236)  Number of firms  3142      Log-likelihood  −3058.1043      Note: The estimates are relative to an ordinary partnership. Robust standard errors in parentheses. See Figure 2 for nesting structure. Duration refers to the contractual duration stated in the articles of association. The reference sector is all other firms, the reference location is the rest of Spain, and the reference year is 1925. Figure 2. View largeDownload slide Schematic for Nested-Logit Model Structures. Note: This structure underpins the nested logit (NL) models reported in the text. Figure 2. View largeDownload slide Schematic for Nested-Logit Model Structures. Note: This structure underpins the nested logit (NL) models reported in the text. The estimates are relative to the ordinary partnership. The choice of normalization does not affect estimated probabilities. Because the model has several branches, the individual parameters are even more difficult to interpret than is normally the case for nonlinear models. Wald tests reject the null hypothesis that any of the branches are redundant in the sense that the model cannot distinguish that form from the ordinary partnership. We cannot, however, reject the null that any branch is equal to any other.19 Individual regressors do have substantially different effects for different forms, however.20 The estimates imply a ranking over legal form options for each firm. We compare the “predicted” legal form (the highest-ranked form, according to the model) to the form the enterprise actually took. With more than two outcomes, this is a demanding standard. The predictions generated by the model reported in Table 2 correspond to firm choices in about 58% of cases. This statistic varies considerably across enterprise forms. The model correctly predicts about 77% of ordinary partnerships and 79% of corporations. The limited partnership and SRL are more difficult: the model does not predict any limited partnerships (6% of firms) and only about 16% of SRLs. There are two reasons for the model’s difficulty with the limited partnership. First, they are rare, so we are asking an econometric model to capture the tail of a distribution. Second, the considerations that drive the difference in rankings between ordinary and limited partnerships may be idiosyncratic (and not in the data) and thus not available to the model. Some limited partners were family members who retained an investment but withdrew from active participation in the business; if we knew details of an entrepreneur’s living kin, or the history of this enterprise, for example, the model might better distinguish the limited from the ordinary partnership. We see below that family relationships among owners drives part of the margin between partnerships and SRLs. The Firms Census database lacks this information. With four alternatives, it is also useful to examine the first and second-ranked forms. Under this more forgiving criterion, the model predicts 94% of partnerships, 83% of corporations, and 81% of SRLs. But the limited partnerships remain a problem; in fact, the model implies that for 64% of actual limited partnerships, that form was the firm’s last choice. To examine how the firm’s characteristics affect the choice of legal form, we rely on average marginal effects (AMEs), or the mean change in the predicted probability that firms select a given form when one or more regressors are varied. By construction, the AMEs sum to one for all choices facing a single firm. Some AMEs are large but imprecisely estimated. The model confirms two basic hypotheses from the literature on the trade-off between partnerships and corporations. Increasing the firm’s capitalization from the mean by 10% reduces the probability of forming an ordinary partnership by 0.0989 (with a standard error of 0.0358) and increases the chance of a corporation forming by 0.1592 (0.0565). Considering the SRL in this context introduces an interesting wrinkle: more capital also makes the SRL less attractive relative to the corporation (0.0602 (0.0291)).21 This effect for the SRL may reflect entrepreneurs’ ability to more effectively sort when they do not want to issue tradeable shares; the SRL allows limited liability with locked-in capital at a smaller scale. Put differently, the enterprise turns to the corporation when it needs to raise larger amount of capital from investors who demand liquidity. This consideration would be especially important for the largest enterprises that need to raise capital from all over Spain, and abroad.22 The model also confirms that asset specificity makes corporations more attractive because of the need to lock in capital. Hypothetically, moving an infrastructure firm to the trade sector increases the chance of an ordinary partnership by 0.11 (0.0662) and reduces the probability of the corporation by 0.1239 (0.1119). The year dummies imply that the SRL became more popular over time. Since the Justice Ministry authorized the SRL in 1919, this result implies that it took time for the various actors—business people, notaries, creditors, etc.—to adapt to the new enterprise. Starting a firm in 1927 compared to 1925 raises the probability of the firm choosing the SRL by 0.085, with a standard error of 0.0296. 5. Decisions 1886–1918: the Firms Sample Database We now turn to two parallel models using the Firms Sample database for the period prior to and following the SRL’s introduction. We split the Firms Sample database in this way because the SRL’s introduction in 1919 enlarges the choice set for that year and those after. The Firms Sample database comes from archival sources and has fewer observations than the Firms Census. These data have information on firms that is not available in the Firm Census and also contains firms created before the SRL was a practical option. The estimates from this model supplement our findings based on the Firms Census. We divide our 15 sampled provinces into five groups of three provinces each. To account for trends over the period 1885–1936, we use a linear spline, with the knot set at 1899, the year Spain began to see capital repatriated from its former colonies. We know the number and identity of the owners when the firm was established and use the indicators of family relations discussed previously to infer familial relationships among owners. We enter the total number of owners linearly. A first dummy variable is unity if all owners are related, and another is unity if only some owners appear to be related. The reference category is firms for which the owners are not related. We interact these two dummies with the number of owners, allowing the connection between legal form and family connections to vary with firm size. Table 3 reports estimates for the Firm Sample model before 1919. The information on owner numbers and family relationships improves the model’s ability to distinguish between ordinary partnerships and other forms. The model correctly predicts 77% of firms overall, including 95% of ordinary partnerships and about 70% of corporations. The limited partnership remains problematic, however. Only 8% of actual limited partnerships have this as their predicted first choice, but about 85% have this option as their predicted second choice. This is admittedly a low standard when there are only three options. Table 3. Nested Logit Model, Firm Sample, Pre-1919   Limited partnership  Corporation    Est.  SE  Est.  SE  Firm has stated duration  −0.040  (0.364)  −1.877  (0.749)  Duration in years  −0.021  (0.023)  0.117  (0.048)  Firm has any unpaid capital  0.347  (0.399)  1.608  (0.466)  Number of owners  0.400  (0.112)  0.494  (0.139)  Year  −0.495  (0.285)  0.259  (0.170)  Year spline, 1899 = 0  0.025  (0.016)  0.0618  (0.025)  Ln (capital)  1.010  (1.013)  −1.785  (0.725)  Ln (capital) squared  −0.031  (0.049)  0.102  (0.036)  Sector      Factory  0.044  (0.406)  −1.416  (0.538)      Trade  0.211  (0.417)  −1.937  (0.639)      Infrastructure  −0.354  (0.552)  −0.187  (0.430)  Relationships among owners      All owners are relatives  0.407  (0.524)  −0.336  (0.908)      Some owners are relatives  0.503  (0.693)  −0.395  (1.018)      Some owners are relatives × number of owners  −0.151  (0.171)  −0.105  (0.211)      All owners are relatives × number of owners  −0.438  (0.158)  −0.638  (0.224)  Provinces      Group one  0.107  (0.290)  0.612  (0.421)      Group two  0.294  (0.279)  −0.290  (0.479)      Group three  −0.126  (0.277)  0.081  (0.351)      Group four  −0.181  (0.284)  0.404  (0.368)  Dissimilarity parameter  0.811 (0.385)      Number of firms  870      Log-likelihood  −493.2403        Limited partnership  Corporation    Est.  SE  Est.  SE  Firm has stated duration  −0.040  (0.364)  −1.877  (0.749)  Duration in years  −0.021  (0.023)  0.117  (0.048)  Firm has any unpaid capital  0.347  (0.399)  1.608  (0.466)  Number of owners  0.400  (0.112)  0.494  (0.139)  Year  −0.495  (0.285)  0.259  (0.170)  Year spline, 1899 = 0  0.025  (0.016)  0.0618  (0.025)  Ln (capital)  1.010  (1.013)  −1.785  (0.725)  Ln (capital) squared  −0.031  (0.049)  0.102  (0.036)  Sector      Factory  0.044  (0.406)  −1.416  (0.538)      Trade  0.211  (0.417)  −1.937  (0.639)      Infrastructure  −0.354  (0.552)  −0.187  (0.430)  Relationships among owners      All owners are relatives  0.407  (0.524)  −0.336  (0.908)      Some owners are relatives  0.503  (0.693)  −0.395  (1.018)      Some owners are relatives × number of owners  −0.151  (0.171)  −0.105  (0.211)      All owners are relatives × number of owners  −0.438  (0.158)  −0.638  (0.224)  Provinces      Group one  0.107  (0.290)  0.612  (0.421)      Group two  0.294  (0.279)  −0.290  (0.479)      Group three  −0.126  (0.277)  0.081  (0.351)      Group four  −0.181  (0.284)  0.404  (0.368)  Dissimilarity parameter  0.811 (0.385)      Number of firms  870      Log-likelihood  −493.2403      Note: The estimates are relative to an ordinary partnership. Robust standard errors in parentheses. See Figure 2 for nesting structure, and Table 2 for variable definitions. Province group one consists of Albacete, Murcia, and Almería; Group two, Barcelona, Tarragone, and Alicante; Group three, Vizcaya, Navarra, and A Coruña; Group four, Madrid, Cuenca, and Valladolid. The ommitted group consists of Asturias, Cantabria, and La Rioja. The Firms Sample database has 500 firms in the post-1919 period, which strains our ability to estimate 45 parameters. Unfortunately, only this database has the family variables in the SRL period, so we must accept the possible problems the small sample poses. We report the estimates in Table 4. This post-1919 model predicts 73% of partnerships, 26% of limited partnerships, 76% of corporations, and 57% of SRLs. This is a dramatic improvement for the limited partnership and the SRL; the family composition of the owners allows the model to more accurately account for decisions concerning these two forms. We discuss these in more detail below. Table 4. Nested Logit Model, Firm Sample, Post-1919   Limited partnership  Corporation  SRL    Est.  SE  Est.  SE  Est.  SE  Firm has stated duration  −0.653  (0.756)  −2.697  (1.526)  −1.055  (0.472)  Duration in years  0.059  (0.042)  0.142  (0.084)  0.040  (0.042)  Firm has any unpaid capital  1.488  (0.484)  1.764  (0.683)  0.767  (0.688)  Number of owners  0.126  (0.190)  0.309  (0.232)  0.015  (0.194)  Year formed  −0.320  (0.554)  0.126  (0.271)  −0.211  (0.282)  Year formed spline, 1899 = 0  −0.029  (0.104)  0.053  (0.046)  0.129  (0.048)  Ln (capital)  0.712  (2.094)  −1.305  (1.068)  −0.006  (0.973)  Ln (capital) squared  −0.013  (0.093)  0.086  (0.055)  0.012  (0.050)  Sector      Factory  −1.053  (0.600)  −1.408  (0.670)  −0.673  (0.465)      Trade  −0.277  (0.589)  −1.513  (0.810)  −0.737  (0.434)      Infrastructure  0.257  (0.679)  0.005  (0.606)  −0.284  (0.707)  Relationships among owners      All owners are relatives  0.907  (1.944)  −1.137  (0.772)  −1.662  (0.844)      Some owners are relatives  1.810  (1.269)  1.155  (0.868)  0.081  (0.725)      Some owners are relatives × number of owners  −0.315  (0.286)  −0.231  (0.234)  0.066  (0.207)      All owners are relatives × number of owners  −0.751  (0.668)  −0.177  (0.228)  0.014  (0.249)  Provinces      Group one  0.649  (0.893)  0.285  (0.508)  −0.742  (0.740)      Group two  −0.113  (0.458)  0.212  (0.478)  −0.247  (0.335)      Group three  −0.085  (0.540)  0.0976  (0.487)  0.331  (0.387)      Group four  −0.420  (0.808)  0.337  (0.398)  0.451  (0.351)  Dissimilarity parameter  0.502 (0.520)  Number of firms  498  Log-likelihood  −386.9527    Limited partnership  Corporation  SRL    Est.  SE  Est.  SE  Est.  SE  Firm has stated duration  −0.653  (0.756)  −2.697  (1.526)  −1.055  (0.472)  Duration in years  0.059  (0.042)  0.142  (0.084)  0.040  (0.042)  Firm has any unpaid capital  1.488  (0.484)  1.764  (0.683)  0.767  (0.688)  Number of owners  0.126  (0.190)  0.309  (0.232)  0.015  (0.194)  Year formed  −0.320  (0.554)  0.126  (0.271)  −0.211  (0.282)  Year formed spline, 1899 = 0  −0.029  (0.104)  0.053  (0.046)  0.129  (0.048)  Ln (capital)  0.712  (2.094)  −1.305  (1.068)  −0.006  (0.973)  Ln (capital) squared  −0.013  (0.093)  0.086  (0.055)  0.012  (0.050)  Sector      Factory  −1.053  (0.600)  −1.408  (0.670)  −0.673  (0.465)      Trade  −0.277  (0.589)  −1.513  (0.810)  −0.737  (0.434)      Infrastructure  0.257  (0.679)  0.005  (0.606)  −0.284  (0.707)  Relationships among owners      All owners are relatives  0.907  (1.944)  −1.137  (0.772)  −1.662  (0.844)      Some owners are relatives  1.810  (1.269)  1.155  (0.868)  0.081  (0.725)      Some owners are relatives × number of owners  −0.315  (0.286)  −0.231  (0.234)  0.066  (0.207)      All owners are relatives × number of owners  −0.751  (0.668)  −0.177  (0.228)  0.014  (0.249)  Provinces      Group one  0.649  (0.893)  0.285  (0.508)  −0.742  (0.740)      Group two  −0.113  (0.458)  0.212  (0.478)  −0.247  (0.335)      Group three  −0.085  (0.540)  0.0976  (0.487)  0.331  (0.387)      Group four  −0.420  (0.808)  0.337  (0.398)  0.451  (0.351)  Dissimilarity parameter  0.502 (0.520)  Number of firms  498  Log-likelihood  −386.9527  Note: The estimates are relative to an ordinary partnership. Robust standard errors in parentheses. See Figure 2 for nesting structure, and Table 2 for variable definitions. Province group one consists of Albacete, Murcia, and Almería; Group two, Barcelona, Tarragone, and Alicante; Group three, Vizcaya, Navarra, and A Coruña; Group four, Madrid, Cuenca, and Valladolid. The ommitted group consists of Asturias, Cantabria, and La Rioja. Point estimates for the AMEs in the two Firms Sample models are large but imprecisely estimated. We focus on the estimates for the SRL period as reported in Table 4. Consider two different, two-owner firms. The firms with two related owners are less likely to choose the ordinary partnership (0.0817, standard error 0.086) and more likely to form as a limited partnership (0.077 (0.0578)). The effects are slightly different if we compare a firm with three related owners to a three-person firm for which only two owners are related. In the post-1919 model these effects are larger; a three-person firm with all related owners has a smaller chance (0.2978 (0.1811)) of selecting the ordinary partnership and a similarly larger chance (0.294 (0.2188)) of taking on the limited partnership. 6. The SRL: Using Law to Substitute for Family Ties SRLs formed after 1919 closely resembled the partnership forms, especially the ordinary partnership, in capitalization, average number of owners, and sector. We have shown, however, that SRLs differed from ordinary partnerships in an important respect: SRL owners were less likely to be related. This result suggests an important role for enterprise forms such as the SRL. They provide a legal structure for small firms whose owners lack kin ties. This finding echoes and amplifies a long literature on the economics of family businesses. Economists and economic historians have long known that family groups play central roles in forming and sustaining business enterprises. Families and firms constitute two of Ben-Porath (1980)’s “F connections.” The growing finance literature on family firms usually focuses on questions beyond this article’s scope: how family firms differ from others in managerial succession, in governance, as well as in longevity and other performance measures. Much of that literature also deals with firms larger than our SRLs (or partnerships). As Burkart et al. (2003: 2167) stress, outside the United States and the United Kingdom today, the majority of even publically-traded firms are controlled by a family group. Family groups also control some huge public companies in the United States and the United Kingdom.23 Family enterprises are even more numerous among firms of the size of our partnerships and SRLs. Why do we see family firms? A family group that ties up most of its wealth in a single enterprise forgoes opportunities to diversify. A firm that limits its managers and investors to related individuals forgoes a broader pool of talent and capital, as well as potential connections to suppliers and customers. These costs apply to families owning SMEs (such as our partnerships and SRLs) as well as to families owning the listed corporations of the finance literature. Economists offer several answers to this puzzle. Altruism among family members reduces the incentive to act opportunistically. Family members also share other social relationships that sustain honest behavior within the family business; “Severe misconduct involves not just risk of dismissal from a job but also the risk of ostracism or expulsion from the family, a penalty drastic enough that it is likely to be an effective deterrent to serious malfeasance” (Pollack 1985: 586). Because of implicit insurance obligations and an interest in the family’s continuing prosperity, family members can commit to the firm’s long-term future, James (2006) stresses an argument common to studies of successful family businesses, arguing that a family manager running a family enterprise can more credibly commit to long-term investment projects because dynastic considerations reduce the agency costs that lead nonfamily managers to prefer short-term profits.24 Family control of a business thus mitigates the transactions costs that Williamson (1979) stresses as responsible for firm organization. These arguments imply that in a world of complete information, and the costless writing and enforcement of contracts, going into business with relatives would be uncommon. Family connections, that is, mitigate problems that arise from an imperfect legal system. Nafziger (1969), for example, argued that Nigerian extended families supported business ties the formal legal system would not. This argument also implies that the development of formal institutions, including law, reduces the role of family in business: “Broader institutional development diminishes the value of relationships and connections, and facilitates arm’s-length contracting.”25 The SRL’s introduction does not constitute the deep and broad changes that constitute the development of a formal legal system. Rather, the SRL represented concrete changes to specific rules. The response to the SRL’s introduction, however, shows two ways in which the family ties filled-in for limitations of the partnership form. The first is liability: one can easily see the preference for family members as co-investors if owners face unlimited liability. By allowing investors to participate in the enterprise with limited liability, the new form encouraged participation by individuals who might not know one another well and for which agency problems might be higher than in a family firm. The relative capital contributions of these firms’ owners suggests a greater role for hands-off investors in the SRL than in the ordinary partnership. After 1919, about two-thirds of ordinary partnerships in Firms Sample formed with equal capital contributions from their owners, as compared to about half of SRLs. The SRL was also slightly more likely to have an owner who contributed more than half the firm’s capital, specifically 15% of partnerships versus 20% of SRLs. These differences suggest that the SRL form worked better for enterprises that had some less-involved investors with smaller capital commitments. The SRL’s ability to attract capital and skills from a broader array of individuals implied greater returns that were attractive when the risk was limited to the sum invested.26 The second difference deals with the SRL’s ability to lock-in capital. The ownership literature views the corporation in part as a legal device to lock-in capital. Partnerships, because they are subject to hold-up, require other mechanisms to support long-term investment. For the reasons stressed by the family-firms literature, kinship ties among partners provide that mechanism SRLs offered a new possibility: legal mechanisms for locking-in capital that did not require the additional formalities of a corporation. 7. Robustness Online Appendix B reports several robustness checks for the models reported here. We briefly summarize these checks and refer the reader to the Online Appendix. First, the difference between the Firms Sample and Firms Census results raises the concern that the latter suffer from omitted variables bias, as the Firms Census database lacks the information on owner characteristics that seems so important in the Firms Sample model. We estimate a version of the pre-1919 Firms Sample model that drops the variables that rely on owner characteristics. The point estimates for other variables change little. Second, our analysis takes as given that the legal forms reported by the commercial registry are the object of meaningful choices by entrepreneurs starting a new firm. One might worry that firms cared less about legal form than we think. We address this concern by estimating models in which the real enterprise form is replaced by one of two randomly-assigned placebos. The first exercise randomly assigns enterprise forms with equal probability. The second assigns forms randomly but respects the unconditional distribution of legal forms for the database in question. These models cannot distinguish the four alternative placebo forms at all, in stark contrast to the results reported in Tables 2–4. Finally, the NL model imposes more structure on choices than some alternatives, such as the multinomial probit model (MNP). Online Appendix B reports a version of the MNP model estimated using the Firms Census database. Given the structure of our data, the MNP model is poorly-identified. The model we report does not imply results much different from the NL models used in this article, however. 8. If There Had Never Been an SRL? A CounterFactual What would firms have done, after 1919, had the SRL not been a possibility but nothing else had changed? We pose this question by looking to the next-preferred option for firms that organized as SRLs. The Firm Census sample has 783 SRLs. (The model is reported in Table 2.) We consider first the 126 firms that were SRLs and for which the model assigns the SRL as the first choice. Eighty-four of these firms would have been ordinary partnerships, and 20 would have been corporations. None would have chosen the limited partnership form. Most Spanish entrepreneurs apparently viewed the SRL as a form of partnership with limited liability. A smaller group of firms treated the new form as a close substitute for the corporation, perhaps valuing the SRL’s ability to lock in capital and offer limited liability at a small scale.27 The fact that a firm organized as an SRL after 1919 implies that its organizers preferred that alternative to the other possible forms. Thus, eliminating the SRL alternative implies a loss. Looking at the types of firms forced into the “wrong” alternatives conveys a sense of the implications of a more restrictive menu. In the Firms Census model, the firms that would have been ordinary partnerships instead of SRLs were larger than the typical partnerships, having a median capitalization of 52,400 pesetas compared to 30,800 pesetas for actual partnerships. SRLs that would have been corporations were smaller than the typical corporation, having about 293,000 pesetas median capitalization.28 The post-1919 Firms Sample model implies something similar. Firms that were actual SRLs but counterfactual partnerships had more owners than actual partnerships, and the SRLs were significantly less likely to have related owners (2% versus 36%). If the SRL had not existed, firms would have been forced to shoehorn themselves into a form best suited for a different kind of enterprise. To fully understand the SRL’s impact on firm formation and investment would require comprehensive data on the sole proprietorships that constitute the majority of enterprises in most contexts. Such data are not available.29 This extensive margin between sole-owner and multi-owner firms is especially important given our findings for the SRL. The firm-level econometric models we report focus on the choice of enterprise form for multi-owner businesses. A new legal form might alter the margin between sole proprietorships and multi-owner enterprises. Entrepreneurs lacking appropriate family members, for example, might have foregone a multi-owner firm when the only reasonable option was the ordinary partnership, only setting up a multi-owner firm when the SRL provided a suitable legal structure. Still, we can ask what legal form firms would have chosen had the SRL been available in the 1890s. The estimates for the post-1919 model (Table 4) and the data for the pre-1919 period imply a sharp conclusion: no firm prior to 1919 would rank the SRL option first, and for 60% of all firms, the SRL would have been the last choice. The SRL’s legal establishment after 1919 brought with it a new kind of Spanish enterprise that drew on the possibilities of the new form. Did the SRL’s introduction lead to more multi-owner firms, or to increased investment in such firms? Figure 1 shows that the SRL appeared during a boom in new firm creation, a boom that by the late 1920s had tapered off considerably for enterprise forms other than the SRL. The short period between the SRL’s introduction and the onset of the Great Depression make it impossible to reliably estimate either breaks or trends in totals of firm formation or authorized capital in the period 1920–1936. The Yearbooks panel does provide suggestive evidence, however. We take advantage of the database’s panel structure and regress total capital invested in all other forms on the capital invested in the SRL. With year and province fixed effects as well as lags of the SRL variable, these correlations are strongly positive across a variety of specifications. Capital invested in SRLs might have reduced capital invested in other forms, but less than peseta-for-peseta. Repeating the exercise and focusing on corporations alone yields similar results. Focusing on the ordinary partnership alone yields a different picture: there is essentially no correlation between capital invested in the SRL and capital invested in ordinary partnerships. The same results hold broadly for numbers of firms as opposed to capital.30 9. Conclusions Spanish company law reflected the basic ideas current in most civil-law countries in the 19th century, offering entrepreneurs a choice of ordinary and limited partnerships or corporations. Most firms organized as ordinary partnerships, with corporations common only in particular sectors and for the largest enterprises. Spain added to the menu of enterprise forms when it introduced its version of the private LLC in 1919. The SRL was immediately popular, displacing the ordinary and limited partnerships. Today, the overwhelming majority of new firms in Spain take the form of an SRL.31 Many of our findings confirm earlier results from other contexts. For example, the largest enterprises, especially those subject to untimely dissolution costs, favored the corporation. Three more specific findings demonstrate the importance of studying the full range of legal forms using models that can capture the substitution possibilities at work. First, the SRL displaced both partnerships and corporations. The would-have-been partnerships that became SRLs were larger than the typical partnership, and the reverse is true for corporations. The SRL occupied a true middle ground, offering many of the advantages of incorporation in a framework that maintained the partnership’s simplicity. Second, if the SRL had been introduced in the 1890s, virtually no firm would have adopted it. The multi-owner firms organized in the 1890s differed considerably from the SRL. The SRL’s advocates argued that Spanish business had changed in ways that required a new form, and this finding suggests they were right. Finally, on the margin, the SRL appealed more to investor groups lacking familial ties than did the ordinary partnership. This result addresses a long literature on family business (both in the past and today) that emphasizes the family’s role in mitigating information and incentive problems within the firm. The SRL offered both limited liability and capital lock-in, thus providing a stronger structure that could substitute for family bonds in firm governance. These results have implications beyond Spain. The choice of legal form transcends any particular country or period in the past two centuries. Many countries introduced legal forms similar to the SRL starting at the end of the 19th century. Prominent examples include Germany’s GmbH (1892), the UK’s PLC (1907), and France’s SARL (1925). As noted at the outset, US states have recently expanded the menu of organizational forms they offer to business enterprises. We do not expect that the introduction of a new legal form in New York State, for example, will mimic the patterns we find for Spain. But the new forms we see in recent decades share the feature of offering to smaller firms the possibility of organizing with limited liability without the reporting and other burdens of the corporation and without sacrificing the contractual freedom of the partnership. The underlying issues are the same: the way heterogeneous firms select and adapt an enterprise form to minimize the costs of establishing and running the firm. Supplementary material Supplementary material is available at Journal of Law, Economics, & Organization online. The authors appreciate helpful comments from Cihan Artunç, Benito Arruñada, José Antonio Espín-Sánchez, Amanda Gregg, Fernando Gómez-Pomar, Leslie Hannah, Philip Haile, Eric Hilt, Naomi Lamoreaux, Thomas Mroz, Aldo Musacchio, Tom Nicholas, Jean-Laurent Rosenthal, Francesca Trivellato, Maribel Sáez-Lacave, and Ebonya Washington. The Registros Mercantiles of the provinces represented in the Firms Sample database all provided valuable cooperation in data-collection (Albacete, Alicante, Almería, Barcelona, A Coruña, Cuenca, Madrid, Murcia, Navarra, Sevilla, Tarragona, Toledo, Valladolid, Vizcaya, Zaragoza). Portions of this paper draw on Guinnane’s joint research with Naomi Lamoreaux, Ron Harris, and Jean-Laurent Rosenthal. This research was partially funded by the National Science Foundation, the Yale University Economic Growth Center, the McMillan Center at Yale, the Spanish Ministry of Economy and Competitiveness (HAR2013-42013-R). Footnotes 1. See Hansmann (1996: Chapter 2) for an account that expands on the idea that ownership reflects contracting costs within the firm. A different literature treats partnerships as the optimal organization for firms whose most valuable assets are human capital, focusing on the question of worker incentives and abstracting from many of the issues we consider. Examples include Morrison and Wilhelm (2008) or Levin and Tadelis (2005). A related literature discusses partnerships in the context of moral hazard in team production; see Kaya and Vereshschagina (2014). Cai (2003) extends the Grossman-Hart-Moore ownership model to consider the case where investors can hold both specific and general investments. Partnerships have indeed become uncommon outside of professional organizations, but in our historical context, they were the most common enterprise form and appear in trade, industry, and other sectors where human capital is not the most important asset. 2. For recent European discussion see European Commission (2011). 3. Musacchio and Turner (2013) stress a broader point, arguing that while ostensibly historical, the “law and finance” hypothesis gets the history wrong. 4. For a cross-country approach to the determinants of firm formation in the period 2003–2005, see Klapper et al. (2010). The World Bank’s “Doing Business” project advocates liberalizing the rules governing formation of new firms (Djankov et al. 2002; Kappler et al. 2006). Critics of the “Doing Business” project argue that the ability to formalize and commit to specific organizational forms enables firms to attract outside funding and to contract with other economic agents. See Arruñada (2007, 2008). 5. Depending on bankruptcy law, liability rules may be default rules. An owner can wave limited liability to, for example, use personal property to secure a business loan. For a firm with unlimited liability to negotiate limited liability, however, is harder. As Posner (2007: 422) notes, “In principle, the enterprise could include in all its contracts with customers and suppliers a clause limiting its liability to the assets of the enterprise. But the negotiation of such waivers would be costly. And it would be utterly impractical to limit most tort liability in this way.” 6. Online Appendix Table A.1 provides an overview of the central features of each legal form for Spain and, for comparison, four large industrial economies. 7. Guinnane and Martínez-Rodríguez (2014) provides more detail on the development of Spanish company law. Martínez-Rodríguez (2016) focuses on the SRL’s formalization in 1919, which reflected a bureaucratic regulation rather than a new statute. The episode illustrates the central role of notaries and the commercial registry, as Arruñada (2010, 2012) has argued. Spanish notaries (like civil-law notaries in general) both advise their clients and certify that contracts meet the law’s requirements. Spain allows firms to organize under the commercial code or the civil code. We focus on the former; firms organized under the civil code were generally smaller and less important for economic performance. 8. Guinnane et al. (2007) call the SRL and forms like it PLLCs, for “private limited-liability companies.” Examples include the German Gesellschaft mit beschränkter Haftung (or GmbH, 1892), the French Société à responsabilité limitée (or SARL, 1925), and the British Private Limited Company (1907). Many jurisdictions created a form of this type, sometimes under the influence of the initial examples. Guinnane and Martínez-Rodríguez (2016) provide more detail on the SRL and the way it was used in the 1920s and 1930s. 9. Martín Aceña (1993), for example, relies on the Madrid commercial registry for the period 1830–1848 to study enterprise-form choice and its relationship to firm size and sector. He credits others with pioneering this use of the source. 10. Gómez-Galvarriato and Musacchio (2004), like two other papers from the project, are unfortunately not published. See Gómez-Galvarriato and Musacchio (2006) as well as Gómez-Galvarriato et al. (2008). 11. Online Appendix A provides additional detail and descriptive statistics. 12. In principle, Spanish firms were also obliged to register their dissolution, but few did. We do not use the implied longevity information because of concerns about selection bias. 13. Table 1 is limited to Firm Sample; Online Appendix Table A.2 reports analogous information for the other sources. 14. Spaniards use both the father’s and mother’s (first) surname, and women do not change their surname at marriage. These customs make it possible to identify relatives with more certainty than would otherwise be the case. Consider the ordinary partnership “Carrasco y Viuda e Hijos de Marsal,” a firm in the Firms Sample database registered in Alicante in 1934 (firm number 2029). The owners are a widow, Adela Carrasco López and her children Ramón, Alfonso, and Adela Marsal Carrasco, as well as Adela’s brother Antonio Carrasco López. These naming practices allow us to identify those who are related through a female line. Our coding relies on kinship terms used in the articles of association (for example, “her son”). We limit the definition to parents, children, siblings, and in-laws. Some of the individuals we consider unrelated may be cousins or more extended kin; since we cannot always identify such relationships, we focus on the narrower but better-measured definition. 15. Online Appendix B discusses alternative modeling approaches and provides robustness checks for our econometric models. 16. All new firms paid the notary who drafted the articles of association and incurred fees from the commercial registry. In addition, the government imposed taxes that depended in part on the firm’s total capital. Tax considerations are an important element in the decision about enterprise form, but extant firms faced a number of taxes that changed over the period we study and bear no simple relation to legal form. 17. Gómez-Galvarriato et al. (2008) study the untimely dissolution of enterprises during the Mexican Revolution (1910–1929), stressing the importance of sector for the firm’s persistence. 18. We top code the duration variable at 20 years. Only a few firms state a longer duration. 19. For the individual branches, the smallest Wald χ2 is 61.64, which with 19 degrees of freedom rejects the null hypothesis at any confidence level. 20. For example, for the null hypothesis that the year effects are the same for the limited partnership and the SRL, the χ2 statistic is 6.89, which with 2 d.f. has a “p-value” of 0.0319. 21. Online Appendix Table B.2 reports AMEs for all enterprise forms. The standard errors are estimated by 200 bootstrap replications. The capital AMEs account for both the log of capital and its square. 22. SRLs could adapt their management and governance as they saw fit. They could have the formal boards of directors and outside managers of the corporation, or could designate an owner to run the firm, like in a partnership. Thus the SRL/corporation margin probably does not reflect governance considerations at a formal level. 23. More recently, scholars have tried to shift the focus to enterprises more of the size of our SRLs and partnerships. The finance literature often stresses the political-economic implications of family-owned business groups, which again is beyond our scope. Mehrotra and Morck (2013: 653) point out that “in many countries, each of a very small number of business families controls a large business group, containing scores, sometimes even hundreds, of seemingly distinct corporations.” We have no indications that single families controlled more than one of the partnerships or SRLs in our database, although we cannot rule out individual exceptions. Recent surveys include Bertrand and Schoar (2006), Mehrotra and Morck (2013), and Casson (1999). Casson focuses on the historical aspect of the question. 24. A different strand in this literature notes the downside of these connections: a family firm may be less likely to fire a poorly-performing employee who is a relative. See Karra et al. (2006). 25. Mehrotra and Morck (2013: 665). As they note, this argument echoes a long tradition in social science. 26. Colli et al. (2003: 36) offer a related view of unlimited liability for the family. 27. This counterfactual only considers the SRLs ranked most-preferred by the model. If we instead consider the next-preferred option after the SRL for all SRLs, even those ranked second or worse by the model, the results are similar. 28. The 52,400 pesetas in 1925 would be worth about €105,000 in 2016. And 293,000 pesetas would thus be €589,000. Computed using the Spanish cost of living series on http://www.measuringworth.com 29. Most sole proprietorships did not enroll in the commercial registry. Other counts of sole proprietorships in Spain imply implausibly low totals. 30. The relationship also holds in first differences and in models estimated for the period 1921–1928 along, thus excluding the Great Depression. The regressions are reported in the Online Appendix. 31. According to the February 2017 data, SRLs accounted for 99.45% of firms and 90.8% of authorized capital. Official figures from the Instituto Nacional de Estadística:http://www.ine.es/daco/daco42/daco424/sm0417.xls. References Abramitzky Ran, Frank Zephyr, Mahajan Aprajit. 2010. “Incentives and Contracts; Partnerships in Rio de Janeiro,” 70 Journal of Economic History  1870– 92. Google Scholar CrossRef Search ADS   Acemoğlu Daron, Johnson Simon. 2005. “Unbundling Institutions,” 113 Journal of Political Economy  949– 95. Google Scholar CrossRef Search ADS   Arruñada Benito. 2007. “Pitfalls to Avoid When Measuring the Institutional Environment: Is ‘Doing Business’ damaging business?” 35 Journal of Comparative Economics  729– 47. Google Scholar CrossRef Search ADS   Arruñada Benito. 2008. “How Doing Business Jeopardizes Institutions Reform,” 10 European Business Organization Law Review  555– 74. Google Scholar CrossRef Search ADS   Arruñada Benito. 2010. “Institutional Support of the Firm: A Theory of Business Registries,” 2 Journal of Legal Analysis  525– 76. Google Scholar CrossRef Search ADS   Arruñada Benito. 2012. Institutional Foundations of Impersonal Exchange. Theory and Policy of Contractual Registries . Chicago: University of Chicago Press. Google Scholar CrossRef Search ADS   Ben-Porath Yoram. 1980. “The F-Connection: Families, Friends, and Firms and the Organization of Exchange,” 6 Population and Development Review  1– 30. Google Scholar CrossRef Search ADS   Bertrand Marianne, Schoar Antoinette. 2006. “The Role of the Family in Family Firms,” 20 Journal of Economic Perspectives  73– 96. Google Scholar CrossRef Search ADS   Blair Margaret. M. 2003. “Locking in Capital: What Corporate Law Achieved for Business Organizers in the Nineteenth Century,” 51 UCLA Law Review  387. Burkart Mike, Panunzi Fausto, Shleifer Andrei. 2003. “Family Firms,” 58 Journal of Finance  2167– 201. Google Scholar CrossRef Search ADS   Cai Hongbin. 2003. “A Theory of Joint Asset Ownership,” Rand Journal of Economics  63– 77. Casson Mark. 1999. “The Economics of the Family Firm,” 47 Scandinavian Economic History Review  10– 23. Google Scholar CrossRef Search ADS   Chandler Alfred. 1977. The Visible Hand: the Managerial Revolution in American Business . Cambridge: Harvard University Press. Colli Andrea, Fernández-Pérez Paloma, Rose Mary B.. 2003. “National Determinants of Family Firm Development?” 4 Enterprise & Society  28– 64. Google Scholar CrossRef Search ADS   Djankov Simeon, La Porta Rafael, Lopez-de-Silanes Florencia, Shleifer Andrei. 2002. “The regulation of entry,” 117 Quarterly Journal of Economics  1– 37. Google Scholar CrossRef Search ADS   European Commission. 2011. Report of the Reflection Group on the Future of EU Company Law . Brussels: European Commission. Gómez-Galvarriato Aurora, Musacchio Aldo. 2004. “Organizational Choice in a French Civil Law Underdeveloped Economy: Partnerships, Corporations and the Chartering of Business in Mexico, 1886 –1910.” Harvard Business School Working Paper 24. Gómez-Galvarriato Aurora,, Musacchio Aldo. 2006. “Larger Menus and Entrepreneurial Appetite: An Empirical Investigation of Organizational Choice in Mexico, 1886-1910.” Paper presented in Session 74 of the XIV International Economic History Conference. Gómez-Galvarriato Aurora, Musacchio Aldo, Parral Rodrigo. 2008. “Political Instability and Untimely Dissolution: Partnerships, Corporations, and the Mexican Revolution, 1910-1929.” Harvard Business School Working Paper 08-092. Grossman Sanford J., Hart Oliver D.. 1986. “The Cost of Benefic of Ownership: A Theory of Vertical and Lateral Integration,” 94 Journal of Political Economy  691– 719. Google Scholar CrossRef Search ADS   Guinnane Timothy, Harris Ron, Lamoreaux Naomi, Rosenthal Jean-Laurent. 2007. “Putting the Corporation in its Place,” 8 Enterprise & Society  687– 729. Google Scholar CrossRef Search ADS   Guinnane Timothy, Martínez-Rodríguez Susana. 2014. “Flexibility in the Spanish Company Law, 1886-1936,” 56 Revista de Historia Industrial  81– 113. Guinnane Timothy, Martínez-Rodríguez Susana 2016. “Instructions Not Included: Spain's Sociedad de Responsabilidad Limitada, 1919–1936”. Working paper. Hansmann Henry. 1996. The Ownership of Enterprise . Cambridge MA: Harvard University Press. Hansmann Henry, Kraakman Reiner, Squire Richard. 2006. “Law and the Rise of the Firm,” 119 Harvard Law Review  1335– 403. Hart Oliver, Moore John. 1990. “Property Rights and the Nature of the Firm,” 98 Journal of Political Economy  1119– 58. Google Scholar CrossRef Search ADS   Hilt Eric,, O’Banion Katherine. 2009. “The Limited Partnership in New York, 1822-1858: Partnerships without Kinship,” 69 Journal of Economic History  615– 45. Google Scholar CrossRef Search ADS   Hilt Eric. 2017. “Corporation Law and the Shift toward Open Access in the Antebellum United States,” in Lamoreaux Naomi R., Wallis John J., eds., Organizations, Civil Society, and the Roots of Development . Chicago: University of Chicago Press (forthcoming). James Harold. 2006. Family Capitalism: Wendels, Haniels, Falcks, and the Continental European Model . Cambridge: Harvard University Press. Kappler Leora, Laeven Luc, Rajan Raghuram. 2006. “Entry Regulation as a Barrier to Enterpreneurship,” 82 Journal of Financial Economics  591– 629. Google Scholar CrossRef Search ADS   Kaya Ayca, Vereshchagina Galina. 2014. “Partnerships Versus Corporations: Moral Hazard, Sorting, and Ownership Structure,” 104 The American Economic Review  291– 307. Google Scholar CrossRef Search ADS PubMed  Klapper Leora, Amit Raphael, Guillén Mauro F.. 2010. “Entrepreneurs and Firm Formation across Countries,” in Lerner Josh, Schoar Antoinette, eds., International Differences in Entrepreneurships . Chicago: University of Chicago Press. Lamoreaux Naomi, 2016. “Corporate Governance and the Expansion of the Democratic Franchise: Beyond Cross-Country Regressions,” 64 Scandinavian Economic History Review  103– 21. Google Scholar CrossRef Search ADS   Lamoreaux Naomi,, Rosenthal Jean-Laurent. 2005. “Legal Regime and Contractual Flexibility: A Comparison of Business’s Organizational Choices in France and the United States during the Era of Industrialization,” 7 American Law and Economics Review  28– 61. Google Scholar CrossRef Search ADS   La Porta Rafael, Lopez-de-Silanes Florencio, Shleifer Andrei. 2008. “The Economic Consequences of Legal Origins,” 46 Journal of Economic Literature  285– 332. Google Scholar CrossRef Search ADS   La Porta Rafael, Lopez-de-Silanes Florencio, Shleifer Andrei, Vishny Robert W., 1998. “Law and Finance,” 106 Journal of Political Economy  1113– 55. Google Scholar CrossRef Search ADS   Levin Jonathan, Tadelis Steven. 2005. “Profit Sharing and the Role of Professional Partnerships,” 120 The Quarterly Journal of Economics  131– 71. Martín Aceña Pablo. 1993. “La Creación de Sociedades en Madrid (1830-1848). Un Analisis del Primer Registro Mercantil,” Documento de trabajo- Universidad de Alcala y Fundación Pública  9303. Martín Aceña Pablo, Pons María Ángeles. 2005. “Sistema monetario y financiero,” in Carreras Albert, Tafunell Xabier, eds., Estadísticas Históricas de España, siglo XIX y XX . Bilbao: Fundación BBVA. Martínez-Rodríguez Susana. 2016. “Creating the Sociedad de Responsabilidad Limitada: the Use of Legal Flexibility in Spanish Company law, 1869-1953,” 90 Business History Review  227– 49. Google Scholar CrossRef Search ADS   Mehrotra Vikas, Morck Randall. 2013. “Entrepreneurship and the Family Firm,” in Constantinides George M., Harris Milton, Stulz Rene M., eds., Handbook of the Economics of Finance  2 649– 81. Entrepreneurship and the Family Firm. Amsterdam: Elsevier. Google Scholar CrossRef Search ADS   Morrison Alan D., Wilhelm William J.Jr. 2008. “The Demise of Investment Banking Partnerships: Theory and Evidence,” 63 Journal of Finance  311– 50. Google Scholar CrossRef Search ADS   Musacchio Aldo, Turner John D., 2013. “Does the Law and Finance Hypothesis Pass the Test of History?” 55 Business History  521– 39. Google Scholar CrossRef Search ADS   Nafziger E. Wayne, 1969. “The Effect of the Nigerian Extended Family on Entrepreneurial Activity,” 18 Economic Development and Cultural Change  25– 33. Google Scholar CrossRef Search ADS   Karra Neri, Tracey Paul, Phillips Nelson, 2006. “Altruism and Agency in the Family Firm: Exploring the Role of Family, Kinship, and Ethnicity,” 30 Entrepreneurship Theory and Practice  861– 77. Google Scholar CrossRef Search ADS   Nicholas Tom. 2015. “The Organization of Enterprise in Japan,” 75 The Journal of Economic History  333– 63. Google Scholar CrossRef Search ADS   Pollack Robert. 1985. “A Transaction Cost Approach to Families and Households,” 23 Journal of Economic Literature  581– 608. Posner Richard A. 2007. Economic Analysis of Law . 7th ed. New York: Wolters. Ribstein Larry E. 2010. The Rise of the Uncorporation . New York: Oxford University Press. Spulber Daniel F. 2009. The Theory of the Firm: Microeconomics with Endogenous Entrepreneurs, Firms, Markets, and Organizations . New York: Cambridge University Press. Google Scholar CrossRef Search ADS   Stout Lynn A. 2005. “On the Nature of Corporations,” Univeristy of Illinois Law Review . 2005: 253. Williamson Oliver. 1979. “Transaction-Cost Economics: The Governance of Contractual Relations,” 22 Journal of Law and Economics  233– 61 Google Scholar CrossRef Search ADS   © The Author(s) 2018. Published by Oxford University Press on behalf of Yale University. All rights reserved. For Permissions, please email: journals.permissions@oup.com

Journal

The Journal of Law, Economics, and OrganizationOxford University Press

Published: Mar 1, 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 12 million articles from more than
10,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Unlimited reading

Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere.

Stay up to date

Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates.

Organize your research

It’s easy to organize your research with our built-in tools.

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