TY - JOUR
AU - Skiti,, Tedi
AB - Abstract In this article, we examine the role of strategic investment in the US broadband industry. In particular, we provide evidence that cable incumbents adjust their investment strategy in response to fiber entry threat and that these deterrence strategies have been successful particularly in intermediate sized markets. We compile data on broadband deployment and exogenous franchise agreements for potential fiber entrants at the most local level in New York State. The results indicate that strategic cable investment may negatively affect optical fiber diffusion. Introduction An important dimension of firm business strategy is responding to new firm entry and entry threat. Entry can affect incumbents’ profitability and survival, thus affecting their business strategies (Luo and Junkunc, 2008; Chang and Wu, 2014), but entry threat alone can also induce incumbents to adjust their strategy, thus affecting market outcomes (Simon, 2005; Seamans, 2012; Prince and Simon, 2014). In response to potential entry, incumbent companies can engage in entry deterrence strategies, a type of endogenous entry barrier that enables incumbents to maintain their competitive position in a market (Bain, 1956). For example, in the context of the US broadband industry, cable companies under threat from new entry from fiber optic companies may adopt a new technology to compete (e.g. offer higher speed internet connections), thereby increasing consumers' switching cost and raising the cost of entry for new entrants. These corporate strategies may be profitable for incumbents and beneficial to consumers (in the short run), but they may negatively affect the diffusion of potentially superior technologies. In this article, we explore the role of cable investment (one form of entry deterrence) in the low adoption rate of optical fiber in the US broadband industry. There has recently been a significant increase in demand for high-speed internet and bandwidth generated by high-definition online video streaming, smart television and other internet-related activities. Cable incumbents, as the dominant incumbents in the industry, invest significantly in their networks in the form of adopting cable technologies (i.e. cable systems) while digital subscriber line (DSL) or fiber firms can offer optical fiber technology which provides significantly higher speeds. Anecdotal evidence1 suggests that cable incumbents adjust their investment behavior in response to fiber entry threat in the US broadband industry. This article is a first attempt to present evidence of strategic investment in the broadband industry to deter fiber entry and examine its implications for optical fiber technology diffusion. If cable incumbents engage in strategic investment, they should adopt new technologies faster in markets where fiber entry deterrence is likely. Although previous studies have examined investment and competition in the broadband industry (Xiao and Orazem, 2011; Nardotto et al., 2015; Briglauer et al., 2018), this is the first paper to show evidence of strategic entry deterrence. Although many theoretical studies have explored strategic investment (Bain, 1956; Spence, 1977; Porter, 1979; Dixit, 1980), empirical evidence remains limited. The lack of empirical literature in strategic investment can be partially explained by the difficulty distinguishing between deterrence and accommodation motives. For instance, incumbents may adopt a new technology because of actual or expected increased demand or—because they anticipate potential entrants will enter the market irrespective of incumbents’ strategies—to prepare for post-entry competition (i.e. accommodate entry). An essential part of measuring the effect on potential entrants is explicitly capturing the markets in which, and time at which, entry is actually deterred. To capture these strategies’ effects on entry, we construct and combine a comprehensive dataset consisting of detailed data on broadband deployment for the state of New York at the most local level. The potential entrants are optical fiber firms with franchise rights making entry decisions. These established rights provide an exogenous variation that directly affects fiber entry likelihood but does not affect cable adoption. Markets in which potential entrants decide not to enter (after controlling for demand) are those where cable incumbents’ investment was successful in deterring entry. We use arguments about the nonmonotonicity of investment strategies to capture entry deterrence (Dafny, 2005; Ellison and Ellison, 2011) and provide evidence of the success of these strategies utilizing a new dataset that allows us to directly measure when entry is deterred. Our data is gathered by two main sources. First, we examine data about technology deployment for cable incumbents and entry decisions by potential fiber entrants at the most local level. Second, we examine data about which markets fiber firms consider as potential markets. We identify the exogenous set of potential fiber entrants using license agreements with local authorities and prior network presence. These potential markets affect fiber entry likelihood but not cable adoption, thus providing exogenous variation in the fiber entry threat. We utilize variation in investment rates by cable incumbents, entry rates by potential fiber entrants, and the exogenous set of potential fiber entrants determined by franchise rights to identify these strategic effects. We provide evidence of nonmonotonic variation of new cable technology adoption in markets where cable incumbents can deter entry. Additionally, we provide evidence that these strategies successfully deterred new optical fiber entry. These results imply that cable investment did not occur only as a response to higher demand for broadband services or competition but also as a way to deter optical fiber diffusion. Therefore, these active corporate strategies directly influence optical fiber deployment. This article is related to and contributes to several streams of literature. First, this is one of the few empirical papers that provides evidence of strategic investment in the broadband industry. We provide evidence that cable investment significantly reduces optical fiber deployment in the US broadband industry. Other studies have focused on competition (Nardotto et al., 2015; Briglauer et al., 2018), demand (Lin and Wu, 2013; Abrardi and Cambini, 2019), or institutional environment (Bouckaert et al., 2010; Briglauer and Cambini, 2018) to explain optical fiber deployment. We instead focus on the strategic behavior of potential fiber entrants before they become actual entrants. This strategic behavior takes into account cable incumbents' investment decisions. Second, even among these empirical studies, this is one of the few empirical papers that shows that strategic investment may be a successful strategy. Previous empirical studies have examined advertising (Ellison and Ellison, 2011), limit pricing (Sweeting et al., 2019), product upgrades (Seamans, 2012; Prince and Simon, 2014), strategic alliances (Goolsbee and Syverson, 2008), and capacity investment (Hall, 1990; Cookson, 2017a) as ways to deter entry in other industries. In addition, this article captures nonprice responses to entry threat as opposed to price (Goolsbee and Syverson, 2008) or advertising (Ellison and Ellison, 2011) strategies. Technology adoption leads to a permanent product upgrade and creates a credible threat mechanism that actual entrants may not even be profitable or survive (Fudenberg and Tirole, 1983, 1985). In broadband, this occurs when cable providers choose when and where to upgrade to new cable systems that significantly increase product quality in a local market. Related literature Broadband is a general purpose technology that directly affects individuals, firms, other industries, and the economy as a whole (Bresnahan and Trajtenberg, 1995). Specifically, broadband availability and adoption may have a strong positive effect on growth (Czernich et al., 2011; Greenstein and McDevitt, 2011; Abrardi and Cambini, 2019; Briglauer and Gugler, 2019), household income (Akerman et al., 2015), labor productivity (Akerman et al., 2015), local employment (Briglauer and Gugler, 2019), and firm productivity (Grimes et al., 2012, Haller and Lyons, 2015). Several studies have examined the various factors that affect broadband investment and adoption.2Lin and Wu (2013) find that demographics (such as household income and education) and available content are determinants of optical fiber adoption. Distaso et al. (2006) examine the intraplatform and interplatform competition in broadband services across EU countries and find that interplatform competition between cable and optical fiber firms promotes adoption. Fourie and de Bijl (2018) find that a moderate level of DSL competition drives fiber penetration. Bouckaert et al. (2010) use country-level data and find that interplatform competition drives broadband penetration. Xiao and Orazem (2011) use the Bresnahan and Reiss (1991, 1994) model to examine entry and competition in the broadband industry and find that sunk costs are a central determinant of new entry. The difference between these studies and ours is that we examine the role of strategic investment to deter fiber rather than the role of competition. There is also extensive literature that examines the relationship between access regulation and competition in broadband diffusion. These studies focus on countries that have some form of access regulation. Theoretical studies such as Kotakorpi (2006) show that access to incumbent's infrastructure may reduce the incentive to invest in new technology, whereas access price may further reduce this incentive. Similarly, Bourreau et al. (2012) show that access price has a positive relationship with entrant investment but an ambiguous effect on incumbent's investment. Briglauer and Gugler (2013) examine the role of government regulation of networks in high-speed internet diffusion. Briglauer (2015) uses EU country-level data and empirically shows that access regulation reduces investment in broadband infrastructure. Nardotto et al. (2015) examine the role of network unbundling in the UK and find that it has a positive impact on broadband penetration, increased new entry, and lower investment by incumbents. Briglauer and Cambini (2018) examine the role of regulated access for potential entrants into telecom incumbents' networks in EU and find that a reduction in access price decreases telecom incumbents' incentive to invest in fiber, but it does not affect cable incumbents' incentive. The above studies focus on Europe, in which the regulatory environment involves infrastructure sharing and access pricing to new entrants. In contrast, US cable incumbents have not been required to provide access to potential entrants since 2003, and thus they rarely do (Nardotto et al., 2015). Potential fiber entrants must build their own network to enter in new markets, which is costly if there is no prior network presence. There is a large theoretical literature on strategic investment (Spence, 1977; Salop, 1979; Lieberman, 1987) and technology adoption (Reinganum, 1981; Fudenberg and Tirole, 1983) but only limited empirical literature. This is partly because of the difficulty of identifying strategic incentives. In one such empirical study, Ellison and Ellison (2011) introduce an approach for identifying strategic investment using a cross-section of markets. Examining the pharmaceutical industry, they find that firms strategically advertise in middle-sized markets prior to patent expiration. They also find a nonmonotonic relation between incumbent investment and entry threat. Dafny (2005) uses similar methods to examine hospital markets and finds that volume growth for procedures is larger in intermediate markets, implying an entry deterrence motive by the incumbents in these markets. In this article, we use their arguments for detection of incumbent cable firms’ strategic behavior in markets under the threat of fiber entry while also considering the effect of these strategies on potential entrants’ decisions. Seamans (2012) empirically examines strategic upgrades in the cable TV industry in relation to municipal entry threat. Instead, we study technology adoption in broadband as it relates to fiber entry and the effects of strategic entry deterrence on market structure. Cookson (2017a) examines the US casino industry and finds that firms use leverage and excessive investment to deter new casino entry. Our paper diverges in that we examine strategic investment in the US broadband industry. There is an extensive literature applying entry models to various industries. Bresnahan and Reiss (1991, 1994) examine isolated local markets and show that observable market characteristics affect entry and market structure and Berry (1992) investigates entry in the airline industry. Goolsbee and Syverson (2008) and Sweeting et al. (2019) investigate incumbents’ incentives in the airline industry and their strategic decision to cut fares in markets where Southwest could potentially enter. We similarly deduce incumbents’ investment decisions in markets where a fiber firm has not yet entered, though we importantly differ by focusing on strategic technology adoption with a new potential entry definition based on entry announcements, network presence or city-level franchise agreements. Industry background Broadband internet connection, the successor of narrowband technology (or dial-up), enables higher-speed data transmission and an “always-on” feature. Notably, the broadband industry constitutes an extremely dynamic economic environment. In 2000, 4.4% of US households had a broadband connection; by 2010, that number had jumped to 68%, and by 2013, 72%.3 Between 2002 and 2012, total (business and residential) fixed connections grew from 19 million to 93 million at a compound annual growth rate of 17% per year.4 There are three main internet technologies: cable, fiber, and DSL. Figure 1 shows that since 2008, the increase in broadband deployment has come mostly from the expansion of cable broadband, whereas the share of DSL broadband has declined. In addition, this figure shows that although optical fiber offers the highest speeds, it has diffused slowly. Figure 1. Open in new tabDownload slide Broadband subscribership evolution in the United States (Source: FCC data). Figure 1. Open in new tabDownload slide Broadband subscribership evolution in the United States (Source: FCC data). Cable internet service providers (henceforth ISP) deploy broadband through their conventional cable TV network with some technical modifications.5 After 1996, cable ISPs started upgrading their network to handle higher data transmission—this became feasible with the Data Over Cable Service Interface Specification (DOCSIS) modem system. The DOCSIS modem system was released in 1997, the DOCSIS 2.0 upgrade in 2001 and DOCSIS 3.0 in 2006. A disadvantage of DOCSIS is that end users share bandwidth, which could reduce internet speeds. Cable incumbents are the successors of cable TV networks based on prior franchise agreements. Optical fiber technology usually offers the highest broadband speeds to its end users through fiber-to-the-home (FTTH). This technology converts electronic signals to light, which is then transmitted through optical fiber. Fiber firms can be previously DSL firms or new fiber firms. Although cable and DSL firms may use optical fiber in their network, only fiber firms connect fiber to the end user. This implies that FTTH is user-specific with no shared bandwidth. DSL technology is deployed by telephone companies, transmits data over the existing copper telephone lines and offers substantially lower speeds. Table 1 illustrates the difference in mean broadband speeds offered by ISPs in New York State. Optical fiber provides the highest speeds in the industry, and there was a significant improvement of broadband speeds when cable firms switched from DOCSIS 2.0 to DOCSIS 3.0. In comparison, DSL technology offers significantly lower speeds compared to both DOCSIS 3.0 and optical fiber. Table 1. Speed comparison across broadband technologies Mean speed Cable DOCSIS 3.0 (mps) 260.88 Cable DOCSIS 2.0 (mps) 17.35 Optical fiber (mps) 404.53 DSL (mps) 7.78 Mean speed Cable DOCSIS 3.0 (mps) 260.88 Cable DOCSIS 2.0 (mps) 17.35 Optical fiber (mps) 404.53 DSL (mps) 7.78 Source: Broadband Map. Open in new tab Table 1. Speed comparison across broadband technologies Mean speed Cable DOCSIS 3.0 (mps) 260.88 Cable DOCSIS 2.0 (mps) 17.35 Optical fiber (mps) 404.53 DSL (mps) 7.78 Mean speed Cable DOCSIS 3.0 (mps) 260.88 Cable DOCSIS 2.0 (mps) 17.35 Optical fiber (mps) 404.53 DSL (mps) 7.78 Source: Broadband Map. Open in new tab The source of cable incumbents’ strategic advantage in the industry is derived from two main sources: first, their prior investment in cable networks provided them with a cost advantage over potential fiber entrants, and second, local entry barriers (e.g. no franchise rights) increase fiber firms’ entry cost, reducing threat of entry for incumbents. Although the Telecommunications Act of 1996 required that incumbents share their infrastructure with potential entrants, the Federal Communications Commission (FCC) has not regulated incumbents' networks since 2003 (Nardotto et al., 2015). Therefore, potential fiber entrants must build their own broadband infrastructure to serve their customers, such as gaining access to utility poles and building permits through license agreements with city and state officials. In general, new firm-city agreements are costly, so many fiber firms prefer to build their networks in areas where they already offer DSL or where they have pre-existing network presence (i.e. dark fiber). DSL and fiber firms’ prior franchise agreements provide us with an exogenous variation of potential fiber entry. This variation, that we use in this article, is exogenous in the sense that occurred before broadband deployment and therefore does not influence cable technology adoption. Each household in the state can usually access only one cable ISP (if any) and/or a DSL ISP. After 2010, there was a considerable switch to a new cable system and fiber entry. The major cable incumbent in New York State is Time Warner Cable while the major DSL and optical fiber ISP is Verizon. Table 2 shows the technology evolution across markets for Time Warner Cable and optical fiber. Specifically, it shows the percentage of markets with DOCSIS 3.0 and optical fiber. This table shows that a significant amount of adoption occurred before 2011 and that there was a significant switch to DOCSIS 3.0 and fiber entry. In particular, in 2011, 69% of markets had at least one cable provider offering DOCSIS 3.0 and 96% in 2013. Additionally, in 2011, 26% of total markets had at least one fiber provider and 34% in 2013. Table 2. Markets with cable DOCSIS and optical fiber in New York State. Obs. is market/t Jun, 2011 Jun, 2012 Jun, 2013 DOCSIS 3.0 6647 7757 9202 % of markets 69% 81% 96% Optical fiber 2523 2976 3295 % of markets 26% 31% 34% Jun, 2011 Jun, 2012 Jun, 2013 DOCSIS 3.0 6647 7757 9202 % of markets 69% 81% 96% Optical fiber 2523 2976 3295 % of markets 26% 31% 34% Open in new tab Table 2. Markets with cable DOCSIS and optical fiber in New York State. Obs. is market/t Jun, 2011 Jun, 2012 Jun, 2013 DOCSIS 3.0 6647 7757 9202 % of markets 69% 81% 96% Optical fiber 2523 2976 3295 % of markets 26% 31% 34% Jun, 2011 Jun, 2012 Jun, 2013 DOCSIS 3.0 6647 7757 9202 % of markets 69% 81% 96% Optical fiber 2523 2976 3295 % of markets 26% 31% 34% Open in new tab In Figures 2 and 3, we provide geographical information about the transition to the cable modem system and optical fiber ISPs. We use the maximum number of ISPs in a local market for a region. Figure 2 illustrates the areas that cable incumbents offer DOCSIS 3.0 while Figure 3 indicates significant fiber entry in central New York State and in urban areas. Figure 2. Open in new tabDownload slide Cable DOCSIS 3.0 adoption evolution in New York State. Figure 2. Open in new tabDownload slide Cable DOCSIS 3.0 adoption evolution in New York State. Figure 3. Open in new tabDownload slide Fiber deployment evolution in New York State. Figure 3. Open in new tabDownload slide Fiber deployment evolution in New York State. Data We use firm-level data and examine firms' strategic behavior before potential entrants become actual entrants. Our approach allows us to identify whether incumbents invest in their networks to deter fiber entry. A microlevel (e.g. firm-level actions and local markets) approach is essential to capture strategic investment instead of country- or industry-level data. We also examine how strategic investment in broadband networks affects optical fiber diffusion and highlight that policy makers need to consider these strategic effects. The dataset used in this article is compiled from two main sources: (i) firm-level investment data for the New York State and (ii) data about which markets potential fiber entrants may actually enter. The first part of the dataset is the US National Broadband Map (henceforth NBM), which is compiled by the FCC,6 NTIA,7 and New York State. This dataset provides firm-level data on broadband technology deployment for the period 2011–2013.8 The time interval is 6 months. The initial dataset is at the census block level. We aggregate the data at the block group level and assume that if an ISP offers a particular broadband technology in a block, it may offer it in the block group as well. The reason for this aggregation is twofold: first, we aim to capture consumer choice set at the local level, and second, there is large variation within a census tract. In the robustness checks, we use an alternative aggregation which does not affect the main results. Additionally, from this dataset we receive information about the market structure such as the number of DSL, fiber, and cable firms. Although the dataset is comprehensive and provides information about the coverage at the most local level, it does have some limitations. First, there is no information about the number of subscribers, prices, or contract promotions. Second, the data are compiled based on ISPs’ responses to a survey.9 Finally, a large number of local markets have already switched to the new cable system. In addition, for the second part of the dataset, we compile information on which markets fiber providers consider a potential market which defines the fiber entry threat. This part includes ISP agreements with local authorities about covering an area, as well as internet archives on prior announcements of which area or cities they will cover or where these firms previously owned dark fiber (or dormant) networks. Therefore, a potential entrant is a fiber ISP who has franchise rights in a market but not offering optical fiber. This dataset also provides an exogenous variation in which markets’ potential fiber entrants can actually enter, and it is compiled at the county level. We assume that if a provider has a license to potentially offer service in a county, then all census block groups in that county are potential markets. Almost 70% of the total local markets in the state have at least one potential fiber entrant. Finally, we match the data with demographics information from the US Census for the local markets to capture consumer heterogeneity in a market. More specifically, we obtain population and aggregate household income. Also, we use population density as population divided by the area of a local market. We provide additional information on the dataset and its construction in the Appendix. Variable description In this section, we define and describe the variables used. The dependent variable—technology adoption—is a dummy capturing when a cable ISP currently offering DOCSIS or DOCSIS 2.0 switches to DOCSIS 3.0 in a local market. This switch is a credible commitment for cable firms since it gives them the ability to offer a much higher-quality product than it could previously, and it cannot be reversed. The dependent variable for fiber entry is a dummy variable capturing when a fiber ISP offers initial service in a local market. Also, we abstract from cable entry in new markets since it was not a common business strategy during the period examined. We use various control variables to measure market characteristics. In particular, we include variables to capture market size and measures of how competitive the market is. We use population density and aggregate household income (both from the US Census) as measures of market size and demand characteristics. High population density is a factor that may reduce broadband deployment cost (FCC, 2018), and household income may be a significant factor of broadband adoption (Abrardi and Cambini, 2019). We also include DSL ISPs as the number of ISPs offering DSL broadband service and Fiber ISPs as firms offering fiber broadband. Moreover, potential entrants is the number of fiber potential entrants that have franchise rights in a specific market (i.e. can enter the market if they choose), entry threat is a dummy for when at least one fiber potential entrant has a franchise rights to a market. Time Warner 3.0 is a dummy indicating whether the major incumbent firm (i.e. Time Warner Cable) has switch to DOCSIS 3.0. Table 3 provides variable definitions and summary statistics. Further, it shows that markets are mostly oligopolistic and have on average a cable incumbent and a DSL firm. Additionally, Table 3 shows that the mean number of potential entrants for each market is 0.69 while the maximum number is 2. The cable adoption and fiber entry rate are 7% and 6%, respectively. Table 3. Variable description and summary statistics. Obs. is market/t Mean SD Median Max Log (population density) 1.51 2.13 −7.73 5.46 Log (population density) Log (aggregate household income) 17.27 0.72 13.22 20.21 Log (sum of all household income) Cable adoption 0.06 0.23 0 1 Dummy, cable incumbent’s switch to DOCSIS 3.0 Fiber entry 0.07 0.31 0 2 Dummy, potential fiber entrant offers service in a new market DSL ISPs 0.77 0.73 1 4 Number of internet service providers offering DSL Cable ISPs 1 0.44 1 2 Number of incumbent cable providers Fiber ISPs 0.49 0.74 0 7 Number of incumbent fiber internet service providers #Potential entrants 0.69 0.70 1 2 Potential fiber entrants with franchise rights for a market Mean SD Median Max Log (population density) 1.51 2.13 −7.73 5.46 Log (population density) Log (aggregate household income) 17.27 0.72 13.22 20.21 Log (sum of all household income) Cable adoption 0.06 0.23 0 1 Dummy, cable incumbent’s switch to DOCSIS 3.0 Fiber entry 0.07 0.31 0 2 Dummy, potential fiber entrant offers service in a new market DSL ISPs 0.77 0.73 1 4 Number of internet service providers offering DSL Cable ISPs 1 0.44 1 2 Number of incumbent cable providers Fiber ISPs 0.49 0.74 0 7 Number of incumbent fiber internet service providers #Potential entrants 0.69 0.70 1 2 Potential fiber entrants with franchise rights for a market Open in new tab Table 3. Variable description and summary statistics. Obs. is market/t Mean SD Median Max Log (population density) 1.51 2.13 −7.73 5.46 Log (population density) Log (aggregate household income) 17.27 0.72 13.22 20.21 Log (sum of all household income) Cable adoption 0.06 0.23 0 1 Dummy, cable incumbent’s switch to DOCSIS 3.0 Fiber entry 0.07 0.31 0 2 Dummy, potential fiber entrant offers service in a new market DSL ISPs 0.77 0.73 1 4 Number of internet service providers offering DSL Cable ISPs 1 0.44 1 2 Number of incumbent cable providers Fiber ISPs 0.49 0.74 0 7 Number of incumbent fiber internet service providers #Potential entrants 0.69 0.70 1 2 Potential fiber entrants with franchise rights for a market Mean SD Median Max Log (population density) 1.51 2.13 −7.73 5.46 Log (population density) Log (aggregate household income) 17.27 0.72 13.22 20.21 Log (sum of all household income) Cable adoption 0.06 0.23 0 1 Dummy, cable incumbent’s switch to DOCSIS 3.0 Fiber entry 0.07 0.31 0 2 Dummy, potential fiber entrant offers service in a new market DSL ISPs 0.77 0.73 1 4 Number of internet service providers offering DSL Cable ISPs 1 0.44 1 2 Number of incumbent cable providers Fiber ISPs 0.49 0.74 0 7 Number of incumbent fiber internet service providers #Potential entrants 0.69 0.70 1 2 Potential fiber entrants with franchise rights for a market Open in new tab Empirical approach Model In this section we discuss our empirical approach to examining whether cable incumbents strategically invest to deter fiber entry. To identify strategic effects, we focus on markets in which Time Warner Cable is the major incumbent, of which there are 9542. Importantly, we use the exogenous variation in potential fiber entrants to predict the fiber entry likelihood. To show that cable incumbents invest to deter we need to find whether new cable technology adoption (henceforth simply cable adoption) varies nonmonotonically with the likelihood of fiber entry. This approach follows Ellison and Ellison’s (2011) argument about identifying incumbent’s strategy in the case of deterrence motives and in which the cable investment exhibits an increase in markets with intermediate attractiveness for potential fiber entrants. For instance, in small markets, fiber entry may be deterred because of inadequate demand, whereas the high demand in large markets will make fiber entry more likely. In intermediate markets, cable incumbents who adjust their investment may be most successful in strategically deterring entry. We examine how potential fiber entry likelihood affects a cable incumbent’s technology adoption decision. We estimate specifications to capture the relationship of predicted cable adoption and fiber entry threat. We use the dummy yc that takes the value 1 when a cable incumbent decides to adopt and estimate a logit model using variation in which markets are under fiber entry threat. We use two stages to examine cable incumbents’ investment strategy in markets under entry threat. In the first stage of the application, we use a logit specification of fiber entry on market characteristics to predict fiber entry. In addition, we allow fiber firms to enter only in a set of potential markets. Their franchise rights to entry provide an exogenous variation of fiber entry likelihood. From this stage, we receive the predicted fiber entry likelihood Pr(Entry). From this specification, we receive the predicted fiber entry likelihood and use it as an explanatory variable for cable adoption decision in the specification below. In the second stage, we use Pr(Entry) as an explanatory variable in cable incumbent’s adoption decision. We use the following the logit specification: P(yi,m,tc=1 | X, Z, m, t) = Λ (αXm,t + β Zm,t +αt + Pr(Entrŷ)m,t + Pr(Entrŷ)m,t2) where i is a cable firm; m is market and t is the time period; Xm,t is a matrix of market characteristics (population and aggregate income) at time t, Zm,t are market structure variables (number of cable, DSL or fiber firms); and αt corresponds to period t fixed effects. The variable Pr(Entry) is the entry likelihood for a potential entrant holding a license to enter in a market. Since we use predicted value of fiber entry at the second stage, we use the method by Murphy and Topel (1985) to adjust the standard errors. Murphy and Topel (1985) derive the asymptotic covariance matrix of the second stage estimates and apply it to a household choice example where the TV perception is the dependent variable. We use the Hardin (2002) application of this approach for our setting. Identification Our identification strategy is based on the argument that fiber entry threat is exogenous from a cable incumbent's perspective since franchise rights for potential fiber entrants are exogenous. Variation in investment rates by cable incumbents and entry decisions by potential fiber entrants as determined by the exercise of franchise rights identify the strategic effects. In particular, the industry background implies that prior network presence by DSL or fiber firms occurred prior to cable broadband deployment. Therefore, the exclusion restriction assumes first, that cable incumbents cannot strategically determine the set of markets the potential fiber entrants can potentially enter in the first stage of our main specification and second, prior network presence is determined before cable adoption decision. Results We present our results in Table 4. We report marginal effects. Column (1) shows that the Entry Threat variable is positive and statistically significant. This implies that in markets with at least one potential fiber entrant, cable firms are 36% more likely to adopt the new technology. Table 4. Second stage: Cable adoption, entry threat and predicted fiber entry likelihood. Markets in which Time Warner is the major cable incumbent. Time fixed effects are included. Standard errors are clustered at the census tract level P (cable adoptsjmt) (1) (2) Marginal effect Marginal effect Log (population densitymt) 0.079**** 0.136* (0.01) (0.08) Log (aggregate household incomemt) 0.132**** 0.101** (0.02) (0.05) DSL ISPsmt 0.002 −0.060** (0.02) (0.03) Fiber ISPsmt 0.081**** 0.244*** (0.02) (0.08) Entry Threatm 0.360**** (0.04) Pr(Entrŷ)m,t 0.604** (0.29) Pr(Entrŷ)m,t2 −0.064** 0.03 Number of obs. 5964 5964 (Pseudo) R-squared 0.502 0.461 P (cable adoptsjmt) (1) (2) Marginal effect Marginal effect Log (population densitymt) 0.079**** 0.136* (0.01) (0.08) Log (aggregate household incomemt) 0.132**** 0.101** (0.02) (0.05) DSL ISPsmt 0.002 −0.060** (0.02) (0.03) Fiber ISPsmt 0.081**** 0.244*** (0.02) (0.08) Entry Threatm 0.360**** (0.04) Pr(Entrŷ)m,t 0.604** (0.29) Pr(Entrŷ)m,t2 −0.064** 0.03 Number of obs. 5964 5964 (Pseudo) R-squared 0.502 0.461 * P < 0.1, ** P < 0.05, *** P < 0.01, **** P < 0.001. Open in new tab Table 4. Second stage: Cable adoption, entry threat and predicted fiber entry likelihood. Markets in which Time Warner is the major cable incumbent. Time fixed effects are included. Standard errors are clustered at the census tract level P (cable adoptsjmt) (1) (2) Marginal effect Marginal effect Log (population densitymt) 0.079**** 0.136* (0.01) (0.08) Log (aggregate household incomemt) 0.132**** 0.101** (0.02) (0.05) DSL ISPsmt 0.002 −0.060** (0.02) (0.03) Fiber ISPsmt 0.081**** 0.244*** (0.02) (0.08) Entry Threatm 0.360**** (0.04) Pr(Entrŷ)m,t 0.604** (0.29) Pr(Entrŷ)m,t2 −0.064** 0.03 Number of obs. 5964 5964 (Pseudo) R-squared 0.502 0.461 P (cable adoptsjmt) (1) (2) Marginal effect Marginal effect Log (population densitymt) 0.079**** 0.136* (0.01) (0.08) Log (aggregate household incomemt) 0.132**** 0.101** (0.02) (0.05) DSL ISPsmt 0.002 −0.060** (0.02) (0.03) Fiber ISPsmt 0.081**** 0.244*** (0.02) (0.08) Entry Threatm 0.360**** (0.04) Pr(Entrŷ)m,t 0.604** (0.29) Pr(Entrŷ)m,t2 −0.064** 0.03 Number of obs. 5964 5964 (Pseudo) R-squared 0.502 0.461 * P < 0.1, ** P < 0.05, *** P < 0.01, **** P < 0.001. Open in new tab Column (2) of Table 4 shows the results of the second stage of our main specification. The results indicate that in markets with moderate attractiveness (as defined by predicted fiber entry), the cable incumbent is more likely to adopt. The coefficients of predicted Pr(Entry) and Pr(Entry)2 are positive (0.604) and negative (−0.064), respectively and statistically significant. The sign of the two variables implies that there is a nonmonotonic relationship between cable adoption and fiber entry likelihood. Moreover, population density, household income positively affect cable adoption since the coefficients are positive (0.136 and 0.101, respectively) and statistically significant. The nonmonotonic relationship between cable adoption and predicter fiber entry implies that cable incumbent adopts faster in markets with an intermediate fiber entry likelihood. If the cable incumbent does not strategically invest, we would observe a linear relationship since higher-demand market implies that it is more profitable to adopt the new technology. Instead, the results indicate that cable firms consider fiber entry likelihood when they make their decisions and this pattern is a concave function of predicted fiber entry. In other words, if deterrence motives were absent, we would expect the cable adoption likelihood to be higher the more attractive the market is for potential fiber entrants. Note that this is not just a competitive effect or preparation for a post-entry market structure since in that case, we should just observe a linear relationship. In Table 5, we also show the results of the first stage in which predicted entry is received. Importantly, we allow fiber entrants to enter only in markets where they have franchise rights and which create an exogenous variation of potential markets. Table 5. First stage: Determinants of fiber entry. Markets that are under threat of fiber entry and in which Time Warner is the major cable incumbent. Marginal effects from the logit specification are reported P (fiber entryjmt) Marginal effect Log (population densitymt) 0.004*** (0.00) Log (aggregate household incomemt) −0.002*** (0.00) DSL ISPsmt 0.004 (0.00) Fiber ISPsmt 0.040*** (0.00) Time FE Yes Firm/county FE Yes Number of obs. 26,346 (Pseudo) R-squared 0.508 P (fiber entryjmt) Marginal effect Log (population densitymt) 0.004*** (0.00) Log (aggregate household incomemt) −0.002*** (0.00) DSL ISPsmt 0.004 (0.00) Fiber ISPsmt 0.040*** (0.00) Time FE Yes Firm/county FE Yes Number of obs. 26,346 (Pseudo) R-squared 0.508 * P < 0.05, ** P < 0.01, *** P < 0.001. Open in new tab Table 5. First stage: Determinants of fiber entry. Markets that are under threat of fiber entry and in which Time Warner is the major cable incumbent. Marginal effects from the logit specification are reported P (fiber entryjmt) Marginal effect Log (population densitymt) 0.004*** (0.00) Log (aggregate household incomemt) −0.002*** (0.00) DSL ISPsmt 0.004 (0.00) Fiber ISPsmt 0.040*** (0.00) Time FE Yes Firm/county FE Yes Number of obs. 26,346 (Pseudo) R-squared 0.508 P (fiber entryjmt) Marginal effect Log (population densitymt) 0.004*** (0.00) Log (aggregate household incomemt) −0.002*** (0.00) DSL ISPsmt 0.004 (0.00) Fiber ISPsmt 0.040*** (0.00) Time FE Yes Firm/county FE Yes Number of obs. 26,346 (Pseudo) R-squared 0.508 * P < 0.05, ** P < 0.01, *** P < 0.001. Open in new tab Was entry deterrence successful? We extend our analysis to show that technology adoption by Time Warner does deter fiber entry. If cable adoption makes fiber entry less likely then it may create an incentive for the cable incumbent to adopt faster to keep the market less competitive. The goal is to determine whether cable incumbents can negatively influence optical fiber diffusion. Each potential fiber entrant may enter only in the set of potential markets. We use the following logit specification: P(yi,mtf = 1 | X, Z, mt) =Λ (αXmt +βZmt + αt +αi,county +γ1TW3.0adoptionmt) is the entry dummy for a fiber firm i in local market m and time t; Xm,t is a matrix of market characteristics (population density and aggregate household income); Zm,t are market structure characteristics (number of DSL firms, fiber firms); and αt corresponds to period t effects, while αi, county corresponds to firm-county fixed effects, respectively. TW 3.0 adoption is a dummy indicating whether Time Warner Cable has adopted the cable system in market m. The results are shown in Table 6, where marginal effects are reported. The results indicate that fiber firms prefer to enter in more dense markets. This market characteristic is statistically significant for all specifications. In addition, the number of existing DSL and fiber ISPs, which can be perceived as a demand size proxy, is positive for all specifications. Importantly, cable adoption has a negative effect on fiber entry. When time fixed effects are included, Time Warner adoption leads to a reduction of the likelihood of fiber entry by −2.8% while including firm/county reduces this likelihood to −2.1%. This effect is negative and statistically significant and implies that active deterrence strategies reduced fiber entry. The mean likelihood of fiber entry is 9%. Table 6. Determinants of fiber entry. Potential markets for each fiber firm are used. Marginal effects from the logit specification are reported. Standard errors are clustered at the census tract level (1) (2) (3) P (fiber entryjmt) Marginal effect Marginal Effect Marginal effect Log (population densitymt) 0.007*** 0.005*** 0.004*** (0.00) (0.00) (0.00) Log (aggregate household incomemt) −0.003** −0.002*** −0.002* (0.00) (0.00) (0.00) DSL ISPsmt 0.050** 0.041*** 0.004 (0.00) (0.00) (0.00) Fiber ISPsmt 0.029*** 0.023*** 0.040*** (0.00) (0.00) (0.00) Time Warner 3.0mt −0.023∧ −0.028* −0.021* (0.01) (0.014) (0.01) Time FE Yes Yes Firm/county FE Yes Number of obs. 38,536 38,536 26,346 (Pseudo) R-squared 0.198 0.365 0.451 (1) (2) (3) P (fiber entryjmt) Marginal effect Marginal Effect Marginal effect Log (population densitymt) 0.007*** 0.005*** 0.004*** (0.00) (0.00) (0.00) Log (aggregate household incomemt) −0.003** −0.002*** −0.002* (0.00) (0.00) (0.00) DSL ISPsmt 0.050** 0.041*** 0.004 (0.00) (0.00) (0.00) Fiber ISPsmt 0.029*** 0.023*** 0.040*** (0.00) (0.00) (0.00) Time Warner 3.0mt −0.023∧ −0.028* −0.021* (0.01) (0.014) (0.01) Time FE Yes Yes Firm/county FE Yes Number of obs. 38,536 38,536 26,346 (Pseudo) R-squared 0.198 0.365 0.451 * P < 0.05, ** P < 0.01, *** P < 0.001, ^P  < 0.1. Open in new tab Table 6. Determinants of fiber entry. Potential markets for each fiber firm are used. Marginal effects from the logit specification are reported. Standard errors are clustered at the census tract level (1) (2) (3) P (fiber entryjmt) Marginal effect Marginal Effect Marginal effect Log (population densitymt) 0.007*** 0.005*** 0.004*** (0.00) (0.00) (0.00) Log (aggregate household incomemt) −0.003** −0.002*** −0.002* (0.00) (0.00) (0.00) DSL ISPsmt 0.050** 0.041*** 0.004 (0.00) (0.00) (0.00) Fiber ISPsmt 0.029*** 0.023*** 0.040*** (0.00) (0.00) (0.00) Time Warner 3.0mt −0.023∧ −0.028* −0.021* (0.01) (0.014) (0.01) Time FE Yes Yes Firm/county FE Yes Number of obs. 38,536 38,536 26,346 (Pseudo) R-squared 0.198 0.365 0.451 (1) (2) (3) P (fiber entryjmt) Marginal effect Marginal Effect Marginal effect Log (population densitymt) 0.007*** 0.005*** 0.004*** (0.00) (0.00) (0.00) Log (aggregate household incomemt) −0.003** −0.002*** −0.002* (0.00) (0.00) (0.00) DSL ISPsmt 0.050** 0.041*** 0.004 (0.00) (0.00) (0.00) Fiber ISPsmt 0.029*** 0.023*** 0.040*** (0.00) (0.00) (0.00) Time Warner 3.0mt −0.023∧ −0.028* −0.021* (0.01) (0.014) (0.01) Time FE Yes Yes Firm/county FE Yes Number of obs. 38,536 38,536 26,346 (Pseudo) R-squared 0.198 0.365 0.451 * P < 0.05, ** P < 0.01, *** P < 0.001, ^P  < 0.1. Open in new tab Robustness checks We provide various tests to examine the robustness of the results. First, we use alternative potential entry definitions by allowing fiber firms to enter neighboring markets. There are technological reasons that may make fiber entry less costly for fiber firms if they already have a presence in neighboring markets. In Table 7, we consider entry in neighboring markets of 20 km from a prior market in which they offered broadband service the previous period. Time Warner Cable adoption effect is negative in all specifications. In Column (3), the effect is negative (−0.001) but not statistically significant. These effects are smaller than those in Table 6 indicating that entry in neighboring markets is less important than the potential market definition we used above. Table 7. Robustness check: Firms’ entry choice set is 20 km. Determinants of fiber entry. Marginal effects from the logit specification are reported. Standard errors are clustered at the census tract level (1) (2) (3) P (fiber entryjmt) Marginal effect Marginal effect Marginal effect Log (population densitymt) −0.001* −0.001* 0.000 (0.00) (0.00) (0.00) Log (aggregate household incomemt) 0.001 0.001 0.001* (0.00) (0.00) (0.00) DSL ISPsmt 0.001 0.000 0.000 (0.00) (0.00) (0.00) Fiber ISPsmt −0.007*** −0.005*** −0.003*** (0.00) (0.00) (0.00) Time Warner 3.0mt −0.017*** −0.014*** −0.001 (0.00) (0.00) (0.00) Time FE Yes Yes Firm/county FE Yes Number of obs. 69,748 69,748 42,812 (Pseudo) R-squared 0.033 0.056 0.257 (1) (2) (3) P (fiber entryjmt) Marginal effect Marginal effect Marginal effect Log (population densitymt) −0.001* −0.001* 0.000 (0.00) (0.00) (0.00) Log (aggregate household incomemt) 0.001 0.001 0.001* (0.00) (0.00) (0.00) DSL ISPsmt 0.001 0.000 0.000 (0.00) (0.00) (0.00) Fiber ISPsmt −0.007*** −0.005*** −0.003*** (0.00) (0.00) (0.00) Time Warner 3.0mt −0.017*** −0.014*** −0.001 (0.00) (0.00) (0.00) Time FE Yes Yes Firm/county FE Yes Number of obs. 69,748 69,748 42,812 (Pseudo) R-squared 0.033 0.056 0.257 * P < 0.05, ** P < 0.01, *** P < 0.001. Open in new tab Table 7. Robustness check: Firms’ entry choice set is 20 km. Determinants of fiber entry. Marginal effects from the logit specification are reported. Standard errors are clustered at the census tract level (1) (2) (3) P (fiber entryjmt) Marginal effect Marginal effect Marginal effect Log (population densitymt) −0.001* −0.001* 0.000 (0.00) (0.00) (0.00) Log (aggregate household incomemt) 0.001 0.001 0.001* (0.00) (0.00) (0.00) DSL ISPsmt 0.001 0.000 0.000 (0.00) (0.00) (0.00) Fiber ISPsmt −0.007*** −0.005*** −0.003*** (0.00) (0.00) (0.00) Time Warner 3.0mt −0.017*** −0.014*** −0.001 (0.00) (0.00) (0.00) Time FE Yes Yes Firm/county FE Yes Number of obs. 69,748 69,748 42,812 (Pseudo) R-squared 0.033 0.056 0.257 (1) (2) (3) P (fiber entryjmt) Marginal effect Marginal effect Marginal effect Log (population densitymt) −0.001* −0.001* 0.000 (0.00) (0.00) (0.00) Log (aggregate household incomemt) 0.001 0.001 0.001* (0.00) (0.00) (0.00) DSL ISPsmt 0.001 0.000 0.000 (0.00) (0.00) (0.00) Fiber ISPsmt −0.007*** −0.005*** −0.003*** (0.00) (0.00) (0.00) Time Warner 3.0mt −0.017*** −0.014*** −0.001 (0.00) (0.00) (0.00) Time FE Yes Yes Firm/county FE Yes Number of obs. 69,748 69,748 42,812 (Pseudo) R-squared 0.033 0.056 0.257 * P < 0.05, ** P < 0.01, *** P < 0.001. Open in new tab Second, we use an alternative market definition. This robustness check addresses the issue whether firms make decisions at a level broader than a census block group. This may happen because firms decide to adopt or enter in more than one market—therefore the decision across the markets is not independent. To address this issue, we aggregate the data at the census tract level (a level above a census block). In Tables 8 and 9, we present the results of this exercise. The results indicate that cable adoption reduces fiber entry and that cable adoption is a nonmonotonic function of entry threat and likelihood. Therefore, the market definition does not determine the main results. Table 8. Robustness check (second stage): A local market is defined a census tract. Cable adoption, entry threat and predicted fiber entry likelihood. Markets in which Time Warner is the major cable incumbent. Time fixed effects are included P (cable adoptsjmt) (1) (2) Marginal effect Marginal effect Log (population densitymt) 0.101**** 0.126**** (0.01) (0.02) Log (aggregate household incomemt) 0.170**** 0.134*** (0.03) (0.05) DSL ISPsmt −0.013 −0.053 (0.03) (0.05) Fiber ISPsmt 0.044 0.167*** (0.03) (0.06) Entry threatm 0.306**** (0.04) Pr(Entrŷ)mt 1.024* (0.54) Pr(Entrŷ)mt2 −0.105* (0.06) Number of obs. 1919 1919 (Pseudo) R-squared 0.471 0.537 P (cable adoptsjmt) (1) (2) Marginal effect Marginal effect Log (population densitymt) 0.101**** 0.126**** (0.01) (0.02) Log (aggregate household incomemt) 0.170**** 0.134*** (0.03) (0.05) DSL ISPsmt −0.013 −0.053 (0.03) (0.05) Fiber ISPsmt 0.044 0.167*** (0.03) (0.06) Entry threatm 0.306**** (0.04) Pr(Entrŷ)mt 1.024* (0.54) Pr(Entrŷ)mt2 −0.105* (0.06) Number of obs. 1919 1919 (Pseudo) R-squared 0.471 0.537 * P < 0.1, ** P < 0.05, *** P < 0.01, **** P < 0.001, ^P < 0.1. Open in new tab Table 8. Robustness check (second stage): A local market is defined a census tract. Cable adoption, entry threat and predicted fiber entry likelihood. Markets in which Time Warner is the major cable incumbent. Time fixed effects are included P (cable adoptsjmt) (1) (2) Marginal effect Marginal effect Log (population densitymt) 0.101**** 0.126**** (0.01) (0.02) Log (aggregate household incomemt) 0.170**** 0.134*** (0.03) (0.05) DSL ISPsmt −0.013 −0.053 (0.03) (0.05) Fiber ISPsmt 0.044 0.167*** (0.03) (0.06) Entry threatm 0.306**** (0.04) Pr(Entrŷ)mt 1.024* (0.54) Pr(Entrŷ)mt2 −0.105* (0.06) Number of obs. 1919 1919 (Pseudo) R-squared 0.471 0.537 P (cable adoptsjmt) (1) (2) Marginal effect Marginal effect Log (population densitymt) 0.101**** 0.126**** (0.01) (0.02) Log (aggregate household incomemt) 0.170**** 0.134*** (0.03) (0.05) DSL ISPsmt −0.013 −0.053 (0.03) (0.05) Fiber ISPsmt 0.044 0.167*** (0.03) (0.06) Entry threatm 0.306**** (0.04) Pr(Entrŷ)mt 1.024* (0.54) Pr(Entrŷ)mt2 −0.105* (0.06) Number of obs. 1919 1919 (Pseudo) R-squared 0.471 0.537 * P < 0.1, ** P < 0.05, *** P < 0.01, **** P < 0.001, ^P < 0.1. Open in new tab Table 9. Robustness check (First stage): A local market is defined a census tract. Determinants of fiber entry. Marginal effects from the logit specification are reported P (fiber entryjmt) Marginal effect Log (population densitymt) 0.011*** (0.00) Log (aggregate household incomemt) −0.001 (0.00) DSL ISPsmt 0.011 (0.01) Fiber ISPsmt 0.096*** (0.00) Time FE Yes Number of obs. 9978 (Pseudo) R-squared 0.455 P (fiber entryjmt) Marginal effect Log (population densitymt) 0.011*** (0.00) Log (aggregate household incomemt) −0.001 (0.00) DSL ISPsmt 0.011 (0.01) Fiber ISPsmt 0.096*** (0.00) Time FE Yes Number of obs. 9978 (Pseudo) R-squared 0.455 * P < 0.05, ** P < 0.01, *** P < 0.001. Open in new tab Table 9. Robustness check (First stage): A local market is defined a census tract. Determinants of fiber entry. Marginal effects from the logit specification are reported P (fiber entryjmt) Marginal effect Log (population densitymt) 0.011*** (0.00) Log (aggregate household incomemt) −0.001 (0.00) DSL ISPsmt 0.011 (0.01) Fiber ISPsmt 0.096*** (0.00) Time FE Yes Number of obs. 9978 (Pseudo) R-squared 0.455 P (fiber entryjmt) Marginal effect Log (population densitymt) 0.011*** (0.00) Log (aggregate household incomemt) −0.001 (0.00) DSL ISPsmt 0.011 (0.01) Fiber ISPsmt 0.096*** (0.00) Time FE Yes Number of obs. 9978 (Pseudo) R-squared 0.455 * P < 0.05, ** P < 0.01, *** P < 0.001. Open in new tab Moreover, in Table 10, we estimate our main two-stage specification simultaneously by using linear probability models for both stages. The goal is to examine whether the results are sensitive to the error distribution assumption. We receive the estimated probability of fiber entry in the first stage and use it in the second stage as an independent variable. This alternative specification does not affect our results since both Pr(Entry) and Pr(Entry)2 are positive (0.037) and negative (−0.021), respectively and statistically significant. Table 10. Simultaneous estimation. Two-stage linear probability model about fiber entry and cable adoption P (cable adoptsjmt) Marginal effect Log (population densitymt) 0.057**** (0.01) Log (aggregate household incomemt) 0.023 (0.02) DSL ISPsmt −0.049**** (0.01) Fiber ISPsmt 0.087**** (0.02) Pr(Entrŷ)mt 0.037** (0.02) Pr(Entrŷ)mt2 −0.021** (0.01) Constant 0.783**** (0.27) Number of obs. 1919 (Pseudo) R-squared 0.523 P (cable adoptsjmt) Marginal effect Log (population densitymt) 0.057**** (0.01) Log (aggregate household incomemt) 0.023 (0.02) DSL ISPsmt −0.049**** (0.01) Fiber ISPsmt 0.087**** (0.02) Pr(Entrŷ)mt 0.037** (0.02) Pr(Entrŷ)mt2 −0.021** (0.01) Constant 0.783**** (0.27) Number of obs. 1919 (Pseudo) R-squared 0.523 ** P < 0.05, *** P < 0.01, **** P < 0.001. Open in new tab Table 10. Simultaneous estimation. Two-stage linear probability model about fiber entry and cable adoption P (cable adoptsjmt) Marginal effect Log (population densitymt) 0.057**** (0.01) Log (aggregate household incomemt) 0.023 (0.02) DSL ISPsmt −0.049**** (0.01) Fiber ISPsmt 0.087**** (0.02) Pr(Entrŷ)mt 0.037** (0.02) Pr(Entrŷ)mt2 −0.021** (0.01) Constant 0.783**** (0.27) Number of obs. 1919 (Pseudo) R-squared 0.523 P (cable adoptsjmt) Marginal effect Log (population densitymt) 0.057**** (0.01) Log (aggregate household incomemt) 0.023 (0.02) DSL ISPsmt −0.049**** (0.01) Fiber ISPsmt 0.087**** (0.02) Pr(Entrŷ)mt 0.037** (0.02) Pr(Entrŷ)mt2 −0.021** (0.01) Constant 0.783**** (0.27) Number of obs. 1919 (Pseudo) R-squared 0.523 ** P < 0.05, *** P < 0.01, **** P < 0.001. Open in new tab Conclusion In this article, we combine data sources and utilize variation in investment rates and the set of potential entrants to identify strategic investment in the broadband industry. We provide evidence that cable investment has deterred optical fiber diffusion in the industry. Our results indicate that cable incumbents adjust their investment strategies not because of demand or competition but in order to limit optical fiber diffusion. This article contributes to various streams of literature. First, we provide an additional factor that may influence optical fiber diffusion in the industry. Optical fiber has been characterized as essential to meet the demand for higher speeds and bandwidth in the industry. We show that incumbents’ strategic behavior must be analyzed to describe the diffusion of new technologies. This also involves analyzing firms' strategic behavior before they enter the market. Second, this article contributes to the burgeoning strategic investment literature. Although there have been significant contributions to the theoretical literature that have examined this issue, to date there has been limited empirical work. Further, the empirical literature has largely focused on pricing responses to entry threat or entry rather than strategic investment. Additionally, among the empirical studies that have found evidence of entry deterrence, there is limited evidence about the success of these strategies. In this article, we show that product innovation strategies can be successful in deterring entry of a superior technology. Finally, this article aims to provide a general framework for how firms react to entry threat and how it affects their strategic behavior. Investment in broadband networks creates a credible mechanism for post-entry competition that potential entrants will consider in their entry decision. The results of this article also contribute to the literature beyond broadband. It adds new evidence on why there may be less new firm entry and slower spatial diffusion of new technologies across local markets by examining endogenous entry barriers. The findings and approach are important for managers, scholars, and practitioners. For managers, the results highlight where product upgrades can successfully deter potential competitors and thereby secure firms’ profitability and survival. Therefore, firms may offer high-quality products that secure their market power in the industry and deter entry. For potential entrants, it indicates where they should exercise their right to enter a market taking into account not only demand conditions but also incumbents’ strategies. For practitioners, it is crucial to know whether firms invest to deter entry or to prepare for postentry competition. If incumbents anticipate that they cannot deter entry, they just prepare for postentry market structure and adopt new technologies faster while new firms enter in a more competitive market. In this case, the threat of entry itself may lead to faster technology diffusion and higher equilibrium quality. On the other hand, deterrence strategies may supply high quality products in the short run, but markets become less competitive because of the deterred entry. There are some issues not discussed in this article that may be incorporated in future work. First, we do not incorporate the role of future demand. Firms forming expectations about future demand may affect their investment strategy and entry. Second, lack of price information does not facilitate a full welfare analysis. Third, future work may examine whether potential entrants can influence the way they receive franchise rights. This will make this process endogenous as potential entrants and incumbents may affect the negotiation with the local authorities. Footnotes 1 “…those paying for 25 megabit service will get 50 megabits, those paying for 50 megabits will get 105, and those paying for 105 will get a whopping 150 megabits…(Comcast, Kansas City, August 2014). 2 For a literature review about broadband investment and adoption see Abrardi and Cambini (2019). 3 “Four years of Broadband Growth Report,” Office of Science and The National Economic Council. 4 FCC yearly reports 2000–2012. 5 For the description of the technical modifications of both DSL and cable networks see Spulber and Yoo (2009). 6 Federal Communications Commission. 7 National Telecommunications and Information Administration. 8 For a discussion about the dataset see Grubesic (2012). 9 It may be expected that firms have an incentive to overreport their coverage. Although this survey is self-reporting, it is easier to detect which technology is deployed than measure the actual speed. Acknowledgments I thank Daniel Xu, Allan Collard-Wexler, James W. Roberts, Christopher Timmins, Peter Arcidiacono, Shane Greenstein, Leslie Marx and John D. Singleton for useful comments and suggestions. Financial support from the NET Institute (www.netinst.org) is gratefully acknowledged. References Abrardi L. Cambini C. ( 2019 ), ‘Ultra-fast broadband investment and adoption: a survey,’ Telecommunications Policy , 43 ( 3 ), 183 – 198 . WorldCat Akerman A. Gaarder I. Mogstad M. ( 2015 ), ‘ The skill complementarity of broadband internet ,’ Quarterly Journal of Economics , 130 ( 4 ), 1781 – 1824 . WorldCat Bain J. S. ( 1956 ), Barriers to New Competition . Cambridge, MA : Harvard University Press . Google Preview WorldCat COPAC Berry S. T. ( 1992 ), ‘ Estimation of a model of entry in the airline industry ,’ Econometrica , 60 ( 4 ), 889 – 917 . WorldCat Bouckaert J. van Dijk T. Verboven F. ( 2010 ), ‘ Access regulation, competition, and broadband penetration: an international study ,’ Telecommunications Policy , 34 ( 11 ), 661 – 671 . WorldCat Bourreau M. Cambini C. Doğan P. ( 2012 ), ‘ Access pricing, competition and incentive to migrate from “old” to “new” technology ,’ International Journal of Industrial Organization , 30 ( 6 ), 713 – 723 . WorldCat Bresnahan T. F. Reiss P. C. ( 1991 ), ‘ Entry and competition in concentrated markets ,’ Journal of Political Economy , 99 ( 5 ), 977 – 1009 . WorldCat Bresnahan T. F. Trajtenberg M. ( 1995 ), ‘ General purpose technologies: “engines of growth”? ,’ Journal of Econometrics , 65 ( 1 ), 83 – 108 . WorldCat Bresnahan T. F. Reiss P. C. ( 1994 ), ‘ Measuring the importance of sunk costs ,’ Annals of Economics and Statistics , 34 , 181 – 217 . WorldCat Briglauer W. ( 2015 ), ‘ How EU sector-specific regulations and competition affect migration from old to new communications infrastructure: recent evidence from EU27 member states ,’ Journal of Regulatory Economics , 48 ( 2 ), 194 – 217 . WorldCat Briglauer W. Cambini C. ( 2018 ), ‘ Does regulation of basic broadband networks affect the adoption of new fiber-based broadband services? ,’ Industrial and Corporate Change , 28 ( 2 ), 219 – 240 . WorldCat Briglauer W. Cambini C. Grajek M. ( 2018 ), ‘ Speeding up the internet: regulation and investment in the European fiber optic infrastructure ,’ International Journal of Industrial Organization , 61 , 613 – 652 . WorldCat Briglauer W. Gugler K. ( 2013 ), ‘ The deployment and penetration of high-speed fiber networks and services: why are EU member states lagging behind?, ’ Telecommunications Policy , 37 ( 10 ), 819 – 835 . WorldCat Briglauer W. Gugler K. ( 2019 ), ‘ Go for gigabit? First evidence on economic benefits of high-speed broadband technologies in Europe ,’ Journal of Common Market Studies , 57 ( 5 ), 1071 – 1090 . WorldCat Chang S.-J. Wu B. ( 2014 ), ‘ Institutional barriers and industry dynamics ,’ Strategic Management Journal , 35 ( 8 ), 1103 – 1123 . WorldCat Cookson J. A. ( 2017a ), ‘ Anticipated entry and entry deterrence: evidence from the American casino industry ,’ Management Science , 64 ( 5 ), 2325 – 2344 . WorldCat Cookson J. A. ( 2017b ), ‘ Leverage and strategic preemption: lessons from entry plans and incumbent investments ,’ Journal of Financial Economics , 123 ( 2 ), 292 – 312 . WorldCat Czernich N. Falck O. Kretschmer T. Woessmann L. ( 2011 ), ‘ Broadband infrastructure and economic growth ,’ Economic Journal , 121 ( 552 ), 505 – 532 . WorldCat Dafny L. S. ( 2005 ), ‘ Games hospitals play: entry deterrence in hospital procedure markets ,’ Journal of Economics and Management Strategy , 14 ( 3 ), 513 – 542 . WorldCat Distaso W. Lupi P. Manenti F. M. ( 2006 ), ‘ Platform competition and broadband uptake: theory and empirical evidence from the European union ,’ Information Economics and Policy , 18 ( 1 ), 87 – 106 . WorldCat Dixit A. ( 1980 ), ‘ The role of investment in entry-deterrence ,’ Economic Journal , 90 ( 357 ), 95 – 106 . WorldCat Ellison G. Ellison S. F. . ( 2011 ), ‘ Strategic entry deterrence and the behavior of pharmaceutical incumbents prior to patent expiration ,’ American Economic Journal: Microeconomics , 3 ( 1 ), 1 – 36 . WorldCat Federal Communications Commission (FCC) ( 2018 ), ‘Broadband deployment report’. Fourie H. de Bijl P. W. J. ( 2018 ), ‘ Race to the top: does competition in the DSL market matter for fiber penetration? ,’ Telecommunications Policy , 42 ( 9 ), 778 – 793 . WorldCat Fudenberg D. Tirole J. ( 1983 ), ‘ Capital as a commitment: strategic investment to deter mobility ,’ Journal of Economic Theory , 31 ( 2 ), 227 – 250 . WorldCat Fudenberg D. Tirole J. ( 1985 ), ‘ Preemption and rent equalization in the adoption of new technology ,’ Review of Economic Studies , 52 ( 3 ), 383 – 401 . WorldCat Goolsbee A. Syverson C. ( 2008 ), ‘ How do incumbents respond to the threat of entry? Evidence from the major airlines ,’ Quarterly Journal of Economics , 123 ( 4 ), 1611 – 1633 . WorldCat Greenstein S. McDevitt R. C. ( 2011 ), ‘ The broadband bonus: estimating broadband internet’s economic value ,’ Telecommunications Policy , 35 ( 7 ), 617 – 632 . WorldCat Grimes A. Ren C. Stevens P. ( 2012 ), ‘ The need for speed: impacts of internet connectivity on firm productivity ,’ Journal of Productivity Analysis , 37 ( 2 ), 187 – 201 . WorldCat Grubesic T. H. ( 2012 ), ‘ The U.S. National Broadband Map: data limitations and implications ,’ Telecommunications Policy , 36 ( 2 ), 113 – 126 . WorldCat Hall E. A. ( 1990 ), ‘ An analysis of preemptive capacity expansion in the titanium dioxide industry ,’ International Journal of Industrial Organization , 8 ( 3 ), 469 – 484 . WorldCat Haller S. A. Lyons S. ( 2015 ), ‘ Broadband adoption and firm productivity: evidence from Irish manufacturing firms ,’ Telecommunications Policy , 39 ( 1 ), 1 – 13 . WorldCat Hardin J. W. ( 2002 ), ‘ The robust variance estimator for two-stage models ,’ Stata Journal , 2 ( 3 ), 253 – 266 . WorldCat Kotakorpi K. ( 2006 ), ‘ Access price regulation, investment and entry in telecommunications ,’ International Journal of Industrial Organization , 24 ( 5 ), 1013 – 1020 . WorldCat Lieberman M. B. ( 1987 ), ‘ Excess capacity as a barrier to entry: an empirical appraisal ,’ Journal of Industrial Economics , 35 ( 4 ), 607 – 627 . WorldCat Lin M.-S. Wu F.-S. ( 2013 ), ‘ Identifying the determinants of broadband adoption by diffusion stage in OECD countries ,’ Telecommunications Policy , 37 ( 4–5 ), 241 – 251 . WorldCat Luo Y. Junkunc M. ( 2008 ), ‘ How private enterprises respond to government bureaucracy in emerging economies: the effects of entrepreneurial type and governance ,’ Strategic Entrepreneurship Journal , 2 ( 2 ), 133 – 153 . WorldCat Murphy K. M. Topel R. H. ( 1985 ), ‘ Estimation and inference in two-step econometric models ,’ Journal of Business & Economic Statistics , 3 ( 4 ), 370 – 379 . WorldCat Nardotto M. Valletti T. Verboven F. ( 2015 ), ‘ Unbundling the incumbent: evidence from UK broadband ,’ Journal of the European Economic Association , 13 ( 2 ), 330 – 362 . WorldCat Porter M. E. ( 1979 ), ‘ How competitive forces shape strategy ,’ Harvard Business Review , 57 ( 2 ), 137 – 145 . WorldCat Prince J. T. Simon D. H. ( 2014 ), ‘ Do incumbents improve service in response to entry? Evidence from airlines’ on-time performance ,’ Management Science , 61 ( 2 ), 372 – 390 . WorldCat Reinganum J. F. ( 1981 ), ‘ On the diffusion of new technology: a game theoretic approach ,’ Review of Economic Studies , 48 ( 3 ), 395 – 405 . WorldCat Salop S. C. ( 1979 ), ‘ Strategic entry deterrence ,’ American Economic Review , 69 ( 2 ), 335 – 338 . WorldCat Seamans R. C. ( 2012 ), ‘ Fighting city hall: entry deterrence and technology upgrades in cable TV markets ,’ Management Science , 58 ( 3 ), 461 – 475 . WorldCat Simon D. ( 2005 ), ‘ Incumbent pricing responses to entry ,’ Strategic Management Journal , 26 ( 13 ), 1229 – 1248 . WorldCat Spence M. ( 1977 ), ‘ Entry capacity, investment, and oligopolistic pricing ,’ Bell Journal of Economics , 8 ( 2 ), 534 – 544 . WorldCat Spulber D. F. Yoo C. S. ( 2009 ), Networks in Telecommunications: Economics and Law . Cambridge University Press . Google Preview WorldCat COPAC Sweeting A. Roberts J. W. Gedge C. ( 2019 ), ‘ A model of dynamic limit pricing with an application to the airline industry ,’ Journal of Political Economy, Forthcoming . WorldCat Xiao M. Orazem P. F. ( 2011 ), ‘ Does the fourth entrant make any difference? Entry and competition in the early U.S. broadband market ,’ International Journal of Industrial Organization , 29 ( 5 ), 547 – 561 . WorldCat Appendix Data construction The first part of the dataset is compiled every 6 months by the US Dept of Commerce, the National Telecommunications and Information Administration, and the State Broadband Initiative for the period June 2010 until December 2014. For this article, we use data for the period June 2011 until December 2013 for wireline broadband. This dataset is compiled at the census block level, which is the lowest possible geographical level (smaller than county and census tract). The dataset provides information about the speed provided by broadband providers, and, the technology used. In addition, it provides the name of the provider, the holding company name, the unique identification number per firm and holding company, state, block identification number, maximum download advertised speed, typical download speed, maximum upload advertised speed, and typical upload speed. The technologies in the dataset include: Cable modem DOCSIS 3.0, DOCSIS 2.0, or other, optical fiber, Asymmetric DSL, Symmetric DSL, Satellite, and Electric Power Line. We keep observations only for Cable modem 3.0 or other, Optical fiber and Asymmetric DSL since these are the dominant technologies in the industry. In addition, the industry does not exhibit significant cross-technology firm variation. This means that firms mostly provide a technology to the end user and make upgrades to their existing network. We aggregate provider information at the block group level. This is a level above block but lower than census tract. The reason is twofold: first, tractability and second, because this is the lowest level for which demographic variables are provided by US Census. We consider five fiber firms as potential entrants. These firms account for more than 90% of total observed fiber entry. For each firm, we describe the data collection process and its potential markets. These data are at the county level. For the period 2011–2013, there is no time variation in the potential markets. Nicholville telephone company We use internet archives (i.e. Wayback Machine) from 2011 to 2014 to check the provider’s operation area. The potential markets for this fiber provider are all markets included in St. Lawrence County. Northland communications This firm is a central New York State provider. We use internet archives (i.e. Wayback Machine) from 2011 to 2014 to check the provider’s dark fiber network. The provider issued announcements about service offers for various areas. The potential markets for this fiber provider are all markets in the following counties: Cortland County, Madison County, Oneida County, and Onondaga County. Trumansburg telephone company We use internet archives (i.e. Wayback Machine) to check the provider’s operation area. The provider issues maps of the area serviced. We used maps between 2011 and 2016 to check the service areas. The potential markets for this fiber provider are all markets in the following counties: Allegany County, Cayuga County, Chemung County, Cortland County, Erie County, Monroe County, Onondaga County, Ontario County, Schuyler County, Seneca County, Steuben County, Tioga County, and Tompkins County. Verizon Verizon is the largest DSL provider in the state. Between 2004 and 2008, it rolled out Verizon Fios in a significant number of markets within the state. After 2011, the firm re-initiated fiber rollout in new markets. For Verizon, we use franchise agreements signed with the state’s cities for fiber rollout—in particular, a 2009 agreement with the New York City. The potential markets for this fiber provider are all markets in the following counties: New York County, Suffolk County, Queens County, Dutchess County, Nassau County, and Kings County. Lightower Lightower is a significant fiber provider offering service mainly to businesses, government, schools, and hospitals. We use internet archives (i.e. Wayback Machine) to access the dark fiber network map and service locations (city level). The potential markets for this fiber provider are all markets in the following counties: Albany County, Bronx County, Kings County, Nassau County, New York County, Queens County, Richmond County, Suffolk County, and Westchester County. © The Author(s) 2019. Published by Oxford University Press on behalf of Associazione ICC. All rights reserved. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Strategic technology adoption and entry deterrence in broadband
JF - Industrial and Corporate Change
DO - 10.1093/icc/dtz057
DA - 2020-06-01
UR - https://www.deepdyve.com/lp/oxford-university-press/strategic-technology-adoption-and-entry-deterrence-in-broadband-mqtT69cp6M
SP - 1
VL - Advance Article
IS - 
DP - DeepDyve
ER -