Paper for Article Critique

American Finance Association

Financial Structure, Acquisition Opportunities, and Firm LocationsAuthor(s): ANDRES ALMAZAN, ADOLFO DE MOTTA, SHERIDAN TITMAN and VAHAP UYSALSource: The Journal of Finance, Vol. 65, No. 2 (APRIL 2010), pp. 529-563Published by: Wiley for the American Finance AssociationStable URL: http://www.jstor.org/stable/25656303 .

Accessed: 28/09/2014 22:32

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

Wiley and American Finance Association are collaborating with JSTOR to digitize, preserve and extend accessto The Journal of Finance.

http://www.jstor.org

This content downloaded from 202.185.34.1 on Sun, 28 Sep 2014 22:32:38 PMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=black

http://www.jstor.org/action/showPublisher?publisherCode=afina

http://www.jstor.org/stable/25656303?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp


THE JOURNAL OF FINANCE VOL. LXV, NO. 2 APRIL 2010

Financial Structure, Acquisition Opportunities, and Firm Locations

ANDRES ALMAZAN, ADOLFO DE MOTTA, SHERIDAN TITMAN, and VAHAP UYSAL*

ABSTRACT This paper investigates the relation between firms' locations and their corporate finance decisions. We develop a model where being located within an industry cluster

increases opportunities to make acquisitions, and to facilitate those acquisitions, firms within clusters maintain more financial slack. Consistent with our model we find that firms located within industry clusters make more acquisitions, and have lower debt

ratios and larger cash balances than their industry peers located outside clusters. We also document that firms in high-tech cities and growing cities maintain more

financial slack. Overall, the evidence suggests that growth opportunities influence firms' financial decisions.

In 1992, the apache corporation, then a small oil and gas firm located in Denver (CO), acquired MW Petroleum, a subsidiary of Amoco Corporation, a

major integrated oil company. This acquisition, which more than doubled the size of its oil and gas reserves, was viewed by Apache's top management as a

major success, and to a large extent defined Apache's future strategy: to grow by acquiring mature oil and gas fields from the major integrated oil firms. To

implement this strategy Apache moved from Denver to Houston (TX), where most of the major oil firms had operations, and reduced its debt ratio to im prove its credit rating. By locating in Houston, Apache's management believed that they would have better access to and knowledge about potential deals.

Maintaining an investment grade bond rating (Apache had a B rating when

they acquired MW Petroleum and reached an A rating a few years later) was

* Andres Almazan is from the University of Texas. Adolfo de Motta is from McGill University.

Sheridan Titman is from the University of Texas and the National Bureau of Economic Research.

Vahap Uysal is from the University of Oklahoma. We would like to thank for helpful comments: Alberto Abadie, Jason Abrevaya, Aydogan Alti, Matthias Buhlmaier, Louis Ederington, Chitru

Fernando, Lorenzo Garlappi, Charles Hadlock, Campbell Harvey (the Editor), Simi Kedia, Scott

Lynn, Bill Megginson, Gordon Phillips, Roberto Rigobon, Pradeep Yadav, an anonymous associate

editor, an anonymous referee, and seminar participants at Baylor University, Boston College, Brigham Young University, UC-Irvine, Columbia University, Drexel University, Fundacion Rafael del Pino, Princeton University, Rutgers University, Texas Christian University, Universidad Carlos

III, University of Essex, University of Michigan, University of Oklahoma, University of Southern

California, UT-Dallas, UT-Austin,Vienna Graduate School of Finance, WFA-Hawaii, and York Uni

versity. Adolfo de Motta thanks IFM2 for its financial support.

529



530 The Journal of Finance?

viewed by Apache's management as essential, since their acquisition strategy required an ability to raise capital on relatively short notice.1

This paper examines whether, as the Apache case suggests, a firm's acquisi tion and financing choices are related to its location. More specifically, we exam ine whether these choices are related to whether the firm is located within an

industry cluster, that is, close to many of its industry peers. To study these is

sues, we start by developing a simple model that describes the relation between a firm's location, financial structure, and acquisition activities. Consistent with

Apache's experience, our model assumes that firms located in industry clusters have more acquisition opportunities but also face greater competition from other potential acquirers. To take advantage of these opportunities the firms in clusters maintain more financial slack, because by doing so they can bid more

aggressively for acquisitions. To test this model, we examine the extent to which firms located in industry

clusters are more acquisitive, the interaction between financial structure, loca

tion, and acquisition activity, and the extent to which firms in clusters maintain more financial slack. Since an industry cluster is somewhat of a nebulous con

cept, our empirical tests examine the robustness of our results with respect to a number of cluster definitions. One set of definitions uses the absolute number of firms within an industry in a metropolitan area. A second set defines clus ters as the proportion of firms in an industry located in the metropolitan area.

Finally, we do an in-depth analysis of the software industry (SIC code 737) since this is an industry with a large number of firms and a very well-defined

industry cluster in Silicon Valley. We find that after controlling for industry affiliation, firms located in clusters

make more acquisitions, which is consistent with the idea that firms in clusters have more opportunities. In addition, as documented in previous research, we find that firms with more financial slack tend to make more acquisitions, which is consistent with firms maintaining more slack when opportunities are greater. More importantly, we show that the positive relation between

acquisition activity and financial slack is stronger in clusters. If we assume

that competition for targets is more intense in clusters then this finding is consistent with our model's implication that debt plays a more important role

when there is competition. Moreover, the fact that this relation is stronger for

acquisitions of public targets and within industry targets, which are likely to attract more competition, provides further support for this implication of the

model. While the evidence on the relation between location and acquisition activity is

consistent with our model, our empirical analysis focuses mainly on the relation

between a firm's choice of financial slack and whether the firm is located in a

cluster. Our results indicate that firms in clusters have less debt and hold more

cash, and these relations continue to hold after controlling for the empirical

1 The above discussion is based on three Harvard Business Cases (MW Petroleum Corp. A, MW Petroleum Corp. B, and Risk Management at Apache) and extensive discussions with Tom

Chambers, Apache's Executive Vice President.



Financial Structure, Acquisition Opportunities, and Firm Locations 531

determinants of capital structure (e.g., Titman and Wessels (1988) and Rajan and Zingales (1995)) and cash holdings (e.g., Opler et al. (1999)) in the extant literature. The relation between financial slack and location is particularly robust and economically significant. After controlling for other determinants of

capital structure and cash holdings, firms located in clusters decrease their net market leverage by 19% and increase their cash holdings by 43% with respect to the sample averages.

These findings, which provide evidence against the null hypothesis that cor

porate finance decisions are independent of location, can be interpreted in at least two ways. First, the evidence is consistent with the idea that loca tion affects acquisition opportunities and that these in turn influence capital structure choices (i.e., a direct-cluster effect). Alternatively, these findings may reflect the possibility that firms that choose to locate either within or outside of clusters have fundamentally different characteristics (i.e., a cluster-selection

effect), and that these characteristics are related to the tendency of firms to make acquisitions as well as their capital structure choices.2

In order to address the possibility that a cluster-selection effect is driving our findings, we replicate our analysis on a sample of firms that have been in a given location for more than 10 years and find that the relation between location and corporate finance choices continues to hold.3 To the extent that the unobserved characteristics that influence a firm's location choice become less

important over time, this evidence suggests that the findings are determined more by a direct-cluster effect than by a cluster-selection effect.

Although our analysis concentrates on industry clusters, there are other ge ographical characteristics that are also likely to affect firms' investment and

acquisition opportunities. For example, firms may find it easier to grow and innovate in regions where other firms are also growing and innovating. Hence, if firms tend to maintain financial slack in anticipation of growth opportunities, then, in addition to the cluster effect, other regional attributes may be related to debt ratios and cash holdings. We examine this issue by considering sev eral measures of economic activity in the metropolitan statistical area (MSA). Specifically, we include MSA growth rates, MSA R&D expenditures, and the

acquisitiveness of firms within the MSA in our debt ratio and cash holdings regressions. The regression results indicate that there is indeed a positive rela tion between the use of financial slack and being located in a growing and more innovative region. However, the cluster effect is significant even after including these other regional variables, suggesting that the cluster effect captures more than just the effect of growing regions.

By linking a firm's location to its corporate finance decisions we contribute to the literature in corporate finance that examines the relation between invest

ment opportunities and financing choices. In this sense, our analysis is related

2 For example, as Almazan, de Motta, and Titman (2007) show, clusters are likely to attract firms with attributes that make them more likely to succeed.

3 Even though firms may originally self-select into different locations, their choice of location can be considered almost permanent. For instance, Pirinsky and Wang (2006) show that in the period from 1992 to 1997 less than 2.4% of firms in Compustat changed their headquarters' location.




to Harford (1999), who shows that firms with more cash do more acquisitions. His empirical tests are motivated by the theoretical literature that suggests that reduced financial slack can limit a firm's ability to fund new investments

(Myers (1977), Myers and Majluf (1984), Jensen (1986), and Hart and Moore (1995)). However, as is also recognized in this literature, firms anticipating good investment opportunities have an incentive to maintain financial slack, which

suggests that the observed relation between cash holdings and acquisition ac

tivity might arise because firms accumulate financial slack when they have

acquisition opportunities. Empirically, although a negative relation between market-to-book ratios?a commonly used proxy for growth opportunities?and debt ratios has been previously established, the direction of causation between

opportunities and financial slack has not been fully resolved. The current lit erature provides evidence of an effect running from debt choices to investment

choices.4 In contrast, by comparing the capital structure choices of firms in

different locations, our tests provide evidence of an effect running from invest ment opportunities to the choice of financial slack. In particular, since cluster and regional growth effects are likely to be directly related to growth opportuni ties but can be viewed as exogenous with respect to the firm's capital structure

choice, the documented relation between geography and financial slack is likely to arise from the effect that the presence of potential opportunities produces on a firm's desire to maintain financial slack. We also contribute to the literature that examines how asset liquidity influ

ences financial structure. Within this literature Williamson (1988) and Shleifer

and Vishny (1992) propose a collateral channel, which suggests that firms are

able to obtain more debt financing when they have assets that are more easily

redeployable. While a liquid asset market allows firms in distress to more eas

ily dispose of their assets, it also makes it easier for firms to grow by acquiring assets. Our finding that clustered firms do more acquisitions is consistent with the idea that clusters increase asset liquidity, and our result that firms in clus

ters maintain more financial slack is consistent with an acquisition channel in

which clustered firms maintain financial slack to take advantage of potential

acquisition opportunities that clusters may offer.

Finally, by using location as a proxy for acquisitions and other growth op

portunities, this paper also contributes to the economic geography literature

that examines how location influences corporate choices.5 For example, pre vious studies have shown that firms in urban clusters are more likely to out

source (Ono (2003)), to vertically disintegrate (Holmes (1999)), and to innovate

(Glaeser et al. (1992)). Also, Landier, Nair, and Wulf (2009) find that geographic

dispersion within the firm affects its labor and divestiture policies, and Kedia,

Panchapagesan, and Uysal (2008) show that physical proximity affects the

4 For instance, Lang, Ofek, and Stulz (1996) document that a conglomerate's noncore division

investments are negatively related to the conglomerate's choice of debt and Lamont (1997) shows

that nonoil subsidiaries of oil companies reduce their investments after the 1986 oil price shock. 5 Issues relating to firm location and industrial clustering have been discussed by economists

since Marshall (1890). See Fujita, Krugman, and Venables (2001), Fujita and Thisse (2002),

Duranton and Puga (2004), and Rosenthal and Strange (2004) for recent reviews of this litera

ture.




ability of firms to complete acquisitions. Like our study, these papers provide evidence that supports the idea that location affects firms' behavior.

The rest of the paper is organized as follows. Section I presents a stylized model and describes the empirical research motivated by it. Section II describes the data and considers some descriptive evidence. Sections III and IV present the main empirical results. Finally, Section V presents our conclusions.

I. Empirical Research Design

A. A Simple Model

This section presents a simple model that motivates our empirical tests. The model is based on the idea that firms may find it useful to have financial slack when they have to compete for acquisitions, which is more likely to be the case when firms are located in clusters.

The model considers two periods, t = 0,1, and two potential acquirers, firms 1 and 2. At t = 0 firm i(i = 1, 2) sets its leverage ratio, 0 < dk < 1. At t = 1, with

probability y, firms have the opportunity to make an acquisition and engage in a second-price sealed bid auction for the target. The value created by such an

acquisition depends on the synergy s, with the target, which is known to the firm at t = 1 but unknown at t = 0. We assume that synergies are independent and uniformly distributed on the interval [s

? 1/2, s + 1/2], where s > 1/2.

We make three assumptions regarding the effects of leverage. First, we as sume that debt is risk free, which allows us to abstract from wealth trans fers between debt-holders and equity-holders that can arise in acquisitions. Second, we assume that leverage reduces the funds that a firm can raise to finance an acquisition. In particular, we assume that firm i can raise funds

up to (v + Si ?

pdi) to make the acquisition, where v is the value of the target as a stand-alone firm, that is, without the synergies, and p < 1/2 measures the intensity with which debt reduces a firm's ability to raise funds.6 With this specification, a firm's ability to raise funds increases in the value of the

target with the synergy (v + Si) and decreases in its leverage ratio c?. Further more, for simplicity we focus on the case where s > 1/2 + p, which ensures that Si

? pdi > 0 for any potential synergy realization and choice of debt. When this

assumption holds, all targets are ultimately acquired.7 Third, we assume that, besides its effect on acquisitions, debt generates some benefits, for example, tax benefits, on firm value, xdk.

To solve the model, we proceed by backward induction and obtain the bidding strategy at t = 1 if an acquisition opportunity arises. Specifically, when this is

6 This aspect of the model is related to Clayton and Ravid (2002) and Morellec and Zhdanov

(2008), who show that more levered bidders tend to bid less aggressively. The assumption that

p < 1/2 ensures that program (2) below is concave and that its first-order condition, that is, equation (3), characterizes the global optimum.

7 Since the minimum realization of the synergy is Si = s ? 1/2 and the maximum choice of debt is di ? 1, assuming s > 1/2 + p implies Si

- pdi > 0 for all Si, a\. Considering the alternative case,

where debt can impede acquisitions when the realized synergy is small, that is, v ? Si ?

pdi < 0, complicates the presentation but does not affect the main intuitions obtained from the analysis.




true, each firm i bids bi = v + Si ?

pdi and acquires the target for bj if bi > bj (for 7 = 1, 2 and j 96 ?. At ? = 0, firm 1 sets its leverage ratio dk taking into account the effect that

leverage will have on its ability to acquire potential targets at t = 1. In partic ular, firm i solves:

max xdi + yE(v +st- bj | bt > &/)Pr(fo; > bj). (1) di

Since bi = v + Si ?

pdi, and since firm i takes firm/s leverage choice as given, (1) can be expressed as

max xck + Y j_ ̂ ̂jf

- + pd/) ds^

cfc/, (2)

where s* = min{s + \ + p(dy -

di), s + \} and s** = max{sy -

p(dj -

di), s - |}.8

Computing the first-order condition that characterizes the symmetric interior solution (i.e., di = dj), we obtain:9

4* = ?2. (3)

yp2

The following result follows directly from equation (3):

Result 1: The optimal debt ratio, d\, is lower when:

(i) acquisition opportunities are more likely to arise (i.e., ̂ < 0);

(ii) debt has a more negative effect on financing acquisitions (i.e., ̂ < 0);

(iii) the non-M&A debt benefits are smaller (i.e., > 0).

In addition, equation (3) has the following implications about the sensitivity of debt to its determinants:

Result 2: The sensitivity of debt to the non-M&A debt benefits (i.e., ^-), is

lower when:

(i) acquisition opportunities are more likely to arise (i.e., < 0);

(ii) debt has a more negative effect on financing acquisitions (i.e., < 0).

The previous result suggests that the empirical importance of r is reduced in situations in which debt has greater influence on the ability to fund acqui sitions, and when acquisition opportunities are more likely to arise. Formally:

8 The limits of integration s* and s** capture the fact that for certain realizations of si and s2,

competition between firms can be limited because of their differences in leverage. For example, if v = 0, p = 0.25, di = 0.2, and cfe = 0.4, then s* = s + \

and s** = max{s2 -

0.05, s - \}.

In this

case, if S2 < (s -

\) + 0.05 and hence s** = s -

\, firm 1 makes the acquisition regardless of its

realized synergy. This occurs because even for the smallest possible synergy s\ = s - \,

firm 1 bids

more than 2, that is, 61 = (s -

\) - 0.05 > b<i = S2

- 0.1. 9 Solving (2): = ?* , which under symmetry and assuming t < yp2 leads to (3). ~

YP*{\-p\a?-dj\)




|^| = is decreasing in y and p. Intuitively, while a reduction in the cost

of financial distress (or an increase in the tax benefits of debt) tends to increase

leverage, this effect is weakened when having financial flexibility is important (for instance, when acquisition opportunities are more likely to arise).

B. Empirical Implementation

Following insights from the literature on economic geography, we conjecture that firms in clusters have more opportunities to make acquisitions that are

subject to competition. In terms of the model, this corresponds to clustered firms

having a larger y, which can stem from either a larger number of acquisition opportunities, or a larger degree of competition for those opportunities. We start by examining the following hypothesis on the differences in the number of acquisitions of firms inside and outside clusters:

Hypothesis 1: Firms in clusters make more acquisitions than firms located outside clusters.

Next, we examine whether leverage has a negative effect on the ability to make acquisitions, and whether this effect is particularly strong for firms in clusters. Notice, however, that leverage is an endogenous variable in our

model and hence one must interpret the empirical relation between leverage and firms' acquisition activity as stemming from exogenous shocks to lever

age. Formally, this effect is described by the partial derivative of debt on the

probability of making an acquisition. In our model, this probability is given by Pi = y Pv(bi > bj), where Pr(fy > bj)

= 1 - s,- + p(dj -

dk). Hence, = -yp, which is decreasing in y and p. Given our conjecture that y is larger in clusters, that is, there are more opportunities to make acquisitions that are subject to

competition in clusters, Hypothesis 2 follows.

Hypothesis 2: The negative effect of leverage on acquisitions is stronger in clusters.

Our third hypothesis considers the effect of a firm's location on its leverage. We start from equation (3) and consider two effects of clusters on the parame ters of the model:

I. According to Williamson (1988), we assume that firms in clusters have better opportunities to redeploy their assets, which reduces liquidation costs and implies an increase in the non-M&A benefits of debt, that is, an increase in r. Specifically, including this cluster effect with other benefits and costs of debt financing generates the following expression for r:

n

n=a + fcClusteri + PjmKi > ^

where {K-} are the j = 1,.., n determinants of the non-M&A net benefits of debt (i.e., K- is the value that determinant j takes for firm i), f$j is the




coefficient associated with K-, Cluster is a dummy variable that takes a value of one if firm / is located within an industry cluster, and pc > 0

represents the increase in r for clustered firms. II. To capture our conjecture that firms in clusters encounter more acquisition

opportunities, we express yt as follows:

Xi = /x + ficClusteri, (5)

where /xc > 0 measures the acquisition effect on clusters.

Substituting (4) and (5) into (3), we obtain the following specification:

^ = a + pcClusten + y Pj j (6)

(jji + /xc Clusteri)p2 (/i + /zc Cluster\)p2

Since Cluster is a dummy variable, (6) can be rewritten as:

dk = 0o + <f>c Clusten + ]TfyK} -

]T 0, (1 - k){R[ x Cluster^, (7) j j

where 0O = fp, <Pc = -

0o, 4>j = ^, and A. = Notice that, while a

priori the net effect of clusters on leverage is ambiguous, the higher likelihood of acquisitions in clusters /jlc > 0 is a necessary condition for firms in clusters to have lower leverage. This condition, however, is not sufficient since firms in clusters may choose to have more leverage due to the higher redeployability of their assets (i.e., 0C is negative only when \xc > ^).

Hypothesis 3: When the acquisition effect is sufficiently strong (i.e., /jlc > firms in clusters will exhibit lower leverage (i.e., 0C < 0).

Finally, our fourth and last hypothesis relates to the determinants of leverage inside and outside of clusters. As shown in equation (7), for each determinant of leverage ifj, the ratio of the coefficients for firms inside and outside of clus ters is constant and equal to X. Notice that k < 1 (i.e., k = -j?) captures the

amelioration effect described in Result 2.

Hypothesis 4: The ratio of the estimates of the determinants of leverage for firms inside versus outside clusters is constant and smaller than one (i.e., k < 1).

II. Data and Sample Characteristics

We examine firms covered in Compustat and CRSP from 1990 to 2005. Since we are interested in firms with a well-defined location (i.e., firms with a high percentage of their assets and employees located at the firm's corporate head

quarters), we exclude from our sample industries like hotels and restaurant

chains, and concentrate instead on manufacturing firms. Furthermore, because

manufacturing industries are likely to exhibit nation-wide product market




competition, in the interpretation of our results we abstract from product mar

ket competition effects associated with differences in firms' locations.10

Specifically, we consider firms with primary three-digit SIC between 200 and 399 and firms in SIC 737 (Computer Programming and Data Processing).11 Our data set also excludes: (i) firms in Hawaii and Puerto Rico; (ii) firms with sales less than $50 million (in 1990 dollars); and (iii) three-digit SIC industries that have less than 10 firms in any of the sample years. The final

sample includes 21 industries, 16 years, 1,910 firms, and 13,342 firm-year observations. Approximately 80% of our sample belongs to firms classified in SIC 200-399 (manufacturing) and the rest to firms in SIC 737. Variables are windsorized at the bottom and top 1% to limit the effect of outliers.

We focus on the location of firms' headquarters, that is, the Metropolitan Statistical Area (MSA) as specified in the 1990 Census. When the firm's head

quarters are not located in an MSA we consider the county of location instead. With this information, for each firm-year, we construct the following four mea sures of industry clustering based on the geographical proximity of the firms'

headquarters: (i) the number of firms with the same three-digit SIC that are located in a given MSA, Number of Firms; (ii) the number of firms with the same three-digit SIC in an MSA divided by the total number of firms with the same three-digit SIC, Ratio of Firms; (iii) a dummy variable that takes a value of one for firm-years in which a firm's headquarters is located within an MSA that has both 10 or more firms with the same three-digit SIC and at least 3% of the market value of the industry, and zero otherwise, Cluster-Firm; and (iv) a

dummy variable that takes a value of one for firm-years in which a firm's head

quarters is located within an MSA that represents at least 10% of the market value of the firm's industry and that has at least three firms with the same

three-digit SIC, Cluster-MV. To save space, we report the results for Number

of Firms and Ratio of Firms in the Internet Appendix.12 Notice that the above measures define clustering in absolute (Number

of Firms and Cluster-Firm) as well as relative terms (Ratio of Firms and Cluster-MV).13 Finding whether our results are robust across these different

10 This is consistent with Glaeser and Kohlhase (2004), who report that transportation costs for manufacturing goods have fallen by over 90% in the last century, and argue that the world is better characterized as a place where it is essentially free to move goods.

11 Most firms in the Computer Programming and Data Processing Industry manufac ture products (e.g., Microsoft Windows) rather than provide a service. Nonetheless, our re sults are robust to excluding SIC 737. Please see the Internet Appendix, available at

http://www.afajof.org/supplements.asp, for these and other additional descriptive statistics and results not included in the paper.

12 While we require that the firm have at least $50 million in sales to be part of our sample, our cluster measures consider all the firms that are included in Compustat in a given year even if their sales are less than $50 million.

13 Notice, however, that Cluster-Firm takes the value one only when firms in the MSA represent

at least 3% of the industry's market value, and Cluster-MV is set to one when there are at least three firms in the same industry within the MSA. These additional conditions are included to

capture the essence of clustering, that is, a concentration of a sufficient number of firms with relative importance in the industry. Nonetheless, as discussed below, our results do not depend on




dimensions of clustering is particularly important because defining a cluster

empirically can be somewhat subjective. Furthermore, as an additional robust ness check, we provide an analysis of the software industry (SIC 737), which has a large number of firms and a very well-defined cluster in Silicon Valley. The Silicon Valley area in our analysis corresponds to MSA code number 7362 as defined by the 1990 Census, which includes San Francisco, Oakland, and San Jose in California. While, strictly speaking, this area is larger than Silicon

Valley, for consistency with the rest of the paper we use the whole MSA code as the geographical location unit.

Based on Cluster-Firm and Cluster-MV there are 1.27 and 1.55 clusters per industry, respectively. This difference can be explained by the tendency of Cluster-Firm to identify clusters in industries in which there are a large number of firms (e.g., SIC 737), and of Cluster-MV to define clusters around the largest firms within an industry (e.g., SIC 356). There are also differences in

clustering across industries. For instance, according to Cluster-Firm, Computer Related Services Industry (SIC 737) has five clusters in 1990, six clusters in

1999, and five clusters in 2005. In contrast, Motor Vehicles and Motor Vehicle

Equipment (SIC 371) has just one cluster during our entire sample period. In our sample, the median number of firms in the same industry within an

MSA is six, and, according to Cluster-Firm and Cluster-MV, 41% and 31% of the observations correspond to firms located within an industry cluster, respectively. (The correlation between Cluster-Firm and Cluster-MV is 46%.)

On average, according to Cluster-Firm, each MSA has 0.14 industry clusters (there are 190 MSAs in the sample), but there is a significant variation across

MSAs. For example, New York City hosts six industry clusters in 1990 and five in 2005. In contrast, Albuquerque did not have any industry clusters during our sample period. In general, clusters tend to be located in larger MSAs, which

points to the need to control for MSA size in our regressions.14 Table I provides descriptive statistics of our sample for a number of variables

of interest. In particular, for each variable, it reports mean, median, and mean

industry-adjusted values for firms inside and outside clusters (based on both definitions of cluster, that is, Cluster-Firm and Cluster-MV). Details on the def inition and construction of all the variables are reported in the Data Appendix following the text.

According to these figures, while firms inside and outside clusters have sim ilar profitability, firms in clusters have lower market leverage and hold more cash than firms outside clusters. In addition, firms in clusters do more R&D, have fewer tangible assets, and exhibit higher market-to-book ratios. These differences remain significant after subtracting the industry means, and hence

they cannot be fully explained by the fact that industries exhibit different tendencies to cluster. Notice also that after subtracting the industry means,

a particular cluster definition, and are robust to excluding these additional conditions on Cluster

Firm and Cluster-MV. 14 In Table IA.I in the Internet Appendix, Panel A reports the per-year average number of

clusters by industry and Panel B reports the number of clusters by MSA.



p o P &> Of. of. CO

o*. OS o*. O f o

Of. <?-j . Of. oj. <t> OD P p, 3 o o p

Of. o*. o Co Oi CO CD

Table I

Summary Statistics

Table I reports summary statistics of our sample for a number of variables of interest. For each variable, it reports mean values, median values, and industry-adjusted mean values for firms located in clusters and outside clusters according to two definitions of industry cluster: Cluster-Firm and

Cluster-MV. The variable Cluster-Firm takes a value of one for firm-years in which a firm's headquarters is located within an MSA that has both

10 or more firms with the same three-digit SIC and at least 3% of the market value of the industry, and zero otherwise. The variable Cluster-MV

takes a value of one for firm-years in which a firm's headquarters is located within an MSA that represents at least 10% of the market value of the

firm's industry and that has at least three firms with the same three-digit SIC. Details on the definitions and construction of the variables reported in the table are available in the Data Appendix. The ^-stat columns report the ^-statistic of the difference between the industry-adjusted mean value

in cluster and outside cluster (standard errors are robust-clustered by firm). Figures in boldface indicate significance at the 10% level.

Mean Value Median Values Industry Adjusted Mean Value

Cluster-Firm Cluster-MV Cluster-Firm Cluster-MV Cluster-Firm Cluster-MV Whole In Off In Off In Off In Off In Off In Off

Sample Cluster Cluster Cluster Cluster Cluster Cluster Cluster Cluster Cluster Cluster ?-Stat Cluster Cluster ?-Stat

Sales 5.733 5.740 5.729 6.100 5.586 5.305 5.406 5.685 5.293 0.202 -0.095 3.59 0.393 -0.157 6.14 EBITDA/TA 0.159 0.164 0.157 0.165 0.157 0.159 0.150 0.156 0.151 0.004 -0.002 1.14 0.005 -0.002 1.47 Market to book 2.201 2.782 1.927 2.593 2.045 2.067 1.486 1.854 1.549 0.284 -0.134 6.28 0.319 -0.127 6.54

Tangible assets/TA 0.227 0.179 0.250 0.202 0.237 0.151 0.218 0.170 0.203 -0.009 0.004 2.52 -0.013 0.005 3.08

R&D/TA 0.065 0.099 0.049 0.085 0.057 0.090 0.024 0.074 0.032 0.013 -0.006 6.91 0.012 -0.005 6.05

Capital exp. / TA 0.067 0.070 0.066 0.069 0.067 0.054 0.049 0.053 0.050 0.002 -0.001 1.42 0.001 0.000 0.64 Average stock return 0.254 0.318 0.224 0.289 0.240 0.218 0.144 0.196 0.153 0.027 -0.013 3.34 0.022 -0.009 2.56 Cash flow Std. Dev. 0.032 0.040 0.028 0.037 0.030 0.032 0.020 0.028 0.022 0.003 -0.001 3.48 0.003 -0.001 2.95

Age 18.264 15.166 19.725 18.025 18.359 10 15 12 14 -0.783 0.369 1.65 0.299 -0.120 0.56 Book leverage 0.476 0.431 0.497 0.454 0.485 0.396 0.485 0.439 0.467 -0.013 0.006 2.17 -0.010 0.004 1.49

Market leverage 0.322 0.242 0.359 0.274 0.340 0.181 0.316 0.214 0.290 -0.028 0.013 4.70 -0.030 0.012 4.59

Net book leverage 0.285 0.146 0.350 0.203 0.317 0.139 0.396 0.219 0.352 -0.059 0.028 5.85 -0.053 0.021 4.87 Net market leverage 0.227 0.115 0.280 0.159 0.255 0.057 0.251 0.100 0.215 -0.042 0.020 6.13 -0.043 0.017 5.74

Cash/nTA 0.323 0.531 0.225 0.466 0.266 0.312 0.077 0.242 0.101 0.114 -0.054 7.88 0.111 -0.045 7.27

Debt rating 0.035 0.052 0.026 0.057 0.025 0 0 0 0 0.018 -0.008 1.95 0.020 -0.008 1.96

Dividend 0.340 0.212 0.400 0.315 0.350 0 0 0 0 -0.026 0.012 1.73 0.011 -0.004 0.67

Population 15.085 15.888 14.707 15.899 14.760 15.734 14.858 15.767 14.921

Observations 13,324 4,272 9,070 3,810 9,532




Table II

Acquisition Activity Summary Statistics Table II reports summary statistics on firms' acquisition activity. For each variable, it reports firm mean values for firms located in clusters and outside clusters according to two definitions of

industry cluster: Cluster-Firm and Cluster-MV. Details on the definitions and construction of the variables reported in the table are available in the Data Appendix. The ?-stat columns report the ^-statistic of the difference between the mean value in cluster and outside cluster (standard errors are robust-clustered by firm). Figures in boldface indicate significance at the 10% level.

Cluster-Firm Cluster-MV

Whole In Off In Off Sample Cluster Cluster ?-Stat Cluster Cluster ?-Stat

Ratio of acquirers 0.376 0.430 0.350 5.52 0.434 0.352 5.23 Ratio of firm acquirers 0.162 0.215 0.137 6.99 0.213 0.142 6.04 Ratio of asset acquirers 0.293 0.322 0.279 3.24 0.332 0.277 3.85 Ratio of public acquirers 0.066 0.096 0.051 6.08 0.099 0.052 5.81 Ratio of within industry acquirers 0.210 0.268 0.183 6.48 0.250 0.194 4.07 Ratio of local acquirers 0.066 0.129 0.036 12.04 0.134 0.039 11.30

Acquisitions value/TA 0.076 0.100 0.065 5.69 0.091 0.070 3.22 Firm acquisitions value/TA 0.046 0.068 0.035 6.71 0.060 0.040 4.06 Asset acquisitions value/TA 0.024 0.024 0.024 0.09 0.023 0.024 0.72

Public acquisitions value/TA 0.023 0.036 0.017 6.09 0.035 0.018 4.92 Within industry acq. value/TA 0.041 0.058 0.033 5.44 0.048 0.038 2.36 Local acquisitions value/TA 0.009 0.018 0.004 9.04 0.018 0.005 8.07

clustered firms are larger, have more volatile cash flows, and are more likely to have an investment grade debt rating.

III. Acquisition Activity and Firm Location

This section presents evidence that relates a firm's location to its acquisition activity. Gur analysis considers all transactions included in the Securities Data

Corporation's (SDC) Mergers and Acquisitions Database from 1990 to 2005 that are listed as an "acquisition of majority interest, merger, asset acquisition or

acquisition of certain assets." In total, there are 9,348 acquisitions in our sam

ple, and according to Cluster-Firm in 3,659 of these acquisitions the acquirer is located within an industry cluster. The average value per transaction is $181

million. Table II documents that, relative to firms located outside clusters, firms in

clusters are 23% more likely to make acquisitions (e.g., 0.430 vs. 0.350 for

Cluster-Firm), and 50% more likely to acquire another firm (e.g., 0.215 vs.

0.137 for Cluster-Firm). These effects are robust and more pronounced for transactions between firms within the same industry, which is of particular interest since these transactions are likely to be subject to more competition. For instance, according to Cluster-Firm, firms in clusters are 46% more likely to make acquisitions within their industry and 85% more likely to acquire another firm with the same three-digit SIC.




Table II also shows that the per-year value of a firm's acquisitions as a

percentage of its total assets is larger in clusters than outside clusters. Notice that although the volume of asset acquisitions is not statistically different, the volume of firm acquisitions is 50% larger in clusters (i.e., 6.8% vs. 3.5% for Cluster-Firm and 6.0% vs. 4.0% for Cluster-MV).

Table III documents that the effect of a firm's location on its acquisition activ

ity remains significant after controlling for year and industry fixed effects (and

hence, for the possibility that more acquisitive industries exhibit a greater ten

dency to cluster).15 Specifically, being located in a cluster raises the likelihood of making an acquisition by 4.8 percentage points for Cluster-Firm and by 7.4

percentage points for Cluster-MV (an increase of 13% and 20% over the sample average, respectively). The effect is robust across all our clustering measures and tends to be relatively more significant for firm acquisitions, public acqui sitions, and within-industry acquisitions (while, as before, it tends not to be

significant for asset acquisitions).16 For example, firms in clusters (as defined

by Cluster-Firm) are 3.3 percentage points more likely to acquire another pub lic firm, an increase of 50% over the sample average. Finally, Table III also documents that in the software industry, firms located in Silicon Valley are more likely to make an acquisition and have a greater volume of transactions relative to total assets. For instance, firms in Silicon Valley are 12 percentage points more likely to make a within-industry acquisition, an increase of 30% over the sample average in the software industry.

Overall, Tables II and III provide support for Hypothesis 1, which postulates that firms in clusters are more acquisitive than firms outside clusters. Our find

ing that clustering spurs acquisitions is also consistent with existing evidence that geographic proximity facilitates input sharing (e.g., Holmes (1999)), labor

market pooling (e.g., Diamond and Simon (1990), Dumais, Ellison, and Glaeser

(1997), and Costa and Kahn (2000)), and knowledge spillovers (e.g., Jaffe, Trajtenberg, and Henderson (1993) and Audretsch and Feldman (1996)).17

Table IV explores the relation between acquisition activity, leverage, and

clustering. In particular, we investigate whether the negative effect of leverage on acquisitions is stronger in clusters, that is, Hypothesis 2. Panel A reports regressions of a firm's acquisition activity on its financial slack (Net Mar ket Leverage), the interaction between financial slack and our cluster proxy (Cluster-Firm or Cluster-MV), controls for a firm's size (Sales in logs), profitabil ity (EBITDA I TA), stock return (Average Stock Return), growth opportunities

15 The reported regressions in the even-numbered columns of Table III correspond to linear

probability models. As reported in the Internet Appendix, a probit analysis produces very similar results.

16 A number of studies raise doubts on the comprehensiveness and accuracy of the SDC database, caveats that are likely to affect the information on nonpublic transactions. See, for instance, Boone and Mulherin (2007) and references therein. This could partially explain the relative lack of evidence in asset acquisitions as well as the stronger evidence found in public acquisitions.

17 Firms in clusters also tend to sell more assets, and the effect is relatively stronger for within

industry transactions. For instance, after controlling for industry fixed effects and using the Cluster-Firm definition, firms in clusters are two percentage points more likely to sell assets to another firm with the same three-digit SIC, a 33% increase over the sample average.



to P p

Table III Acquisition Activity Regressions Table III documents the effect of a firm's location on its acquisition activity after controlling for year and industry fixed effects. Each numbered column and row corresponds to a different regression. Odd-numbered columns correspond to OLS regressions in which the dependent variable is a dummy variable indicating an acquisition. (Several types of acquisitions are considered: All, Firm, Asset, Public, Within Industry, and Local, as denned in

the Data Appendix.) Even-numbered columns correspond to Tobit regressions in which the dependent variable is the ratio of the firm's acquisition volume during the year divided by the firm's total assets. Reported are the

estimated

coefficients with their ^-statistics in parentheses (standard errors

are robust and clustered by firm). All regressions include year fixed effects and the Cluster-MV and Cluster-Firm regressions also include industry fixed effects. (The Silicon Valley regressions study the software industry, that is, SIC 737.) Details on the definition and construction of the variables

reported in the table are available in the Data Appendix. Figures in boldface indicate significance at the 10% level.

Firm Asset Public Within Industry

All Acquisitions Acquisitions Acquisitions Acquisitions Acquisitions Local Acquisitions

Cluster Definitions (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

(1) Cluster-firm 0.048 0.074 0.049 0.123 0.026 0.009 0.033 0.197 0.039 0.071 0.085 0.249 Observations: 13,342 (3.16) (3.66) (4.32) (4.54) (1.78) (0.82) (4.30) (4.76) (3.06) (3.04) (11.07) (10.88) R2 0.05 0.04 0.03 0.02 0.08 0.05 (2) Cluster-MV 0.074 0.088 0.063 0.140 0.052 0.018 0.043 0.235 0.049 0.075 0.088 0.252

Observations: 13,342 (4.92) (4.52) (5.74) (5.36) (3.66) (1.74) (5.40) (6.10) (4.07) (3.46) (11.41) (12.67) R2 0.05 0.05 0.03 0.03 0.08 0.05 (3) Silicon Valley 0.102 0.191 0.108 0.269

0.026

-0.002 0.080 0.387 0.119 0.179 0.181 0.317 Observations: 2,631 (3.14) (4.35) (3.53) (5.00) (0.85) (0.10) (3.40) (4.80) (3.31) (4.16) (6.43) (7.94) R2 0.03 0.04 0.02 0.02 0.03 0.06



5? a cs OW Of. <x. Cft ok' Ol.

O O Of. <x. OK

oi. P a 3 o o os

Ol. o CO CO

Table IV Acquisition Activity and Leverage Table IV documents the effect of leverage on firms' acquisition activity for firms located inside and outside industry clusters. Each column (in each

panel) corresponds to a different 2SLS regression in which the dependent variable is a dummy variable indicating an acquisition. (Several types of

acquisitions are considered: All, Public, Within Industry, Within Industry & Public, and Local, as defined in the Data Appendix.) In the odd-numbered columns the independent variable "Cluster" corresponds to Cluster-Firm, and in the even-numbered columns it corresponds to Cluster-MV. The independent variables Net Market Leverage, Relative Net Market Leverage, Market to Book, EBITDA ITA, and Sales are lagged one period. In the

regressions in Panel A, we instrument Net Market Leverage at t?1 (and its iteration with the cluster dummy) with R&D / TA and Tangible Assets I TA at t-2, and with the 3-year average profitability (EBITDA/TA) and stock returns from t-4 to t-6 (as well as with the iterations between these variables and the cluster dummy). In the regressions in Panel B, we instrument Relative Net Market Leverage at t? 1 (and its iteration with the

cluster dummy) with R&D / TA and Tangible Assets I TA at t-2, and with

the

3-year

average relative profitability (Relative EBITDA/TA) and average

relative stock return (Relative Stock Return) from t?4 to t?6 (as

well

as with the iterations between these variables and the cluster dummy). All

regressions in Table IV include year and industry fixed effects. Reported are the estimated coefficients with their ^-statistics in parentheses (standard

errors are robust-clustered by firm). Details on the definition and

construction

of the variables reported in the table are available in the Data Appendix.

Figures in boldface indicate significance at the 10% level.

Within Industry Within Industry

All Acquisitions Public Acquisitions Acquisitions Public Acq. Local Acquisitions

_(1)_(2)_(3)_(4)_(5)_(6)_(7)_(8)_(9)_(10)

Panel A: Acquisition Activity and Net Market Leverage

Cluster 0.000 0.024 0.031 0.041 0.024 0.034 0.025 0.030 0.098 0.106

(0.01) (1.07) (2.34) (2.98) (1.21) (1.68) (2.13) (2.46) (6.79) (6.95)

Net market leverage x cluster -0.049 -0.070 -0.153 -0.143 -0.125 -0.146 -0.108 -0.118 -0.196 -0.183

(0.45) (0.80) (2.70) (2.87) (1.41) (2.10) (2.29) (2.87) (3.03) (3.05)

Net market leverage -0.265 -0.173 -0.050 -0.043 -0.218 -0.181 -0.066 -0.074 0.041 0.049

(3.12) (2.01) (1.25) (1.05) (3.26) (2.61) (2.15) (2.35) (1.11) (1.22)

Market to book -0.009 -0.004 0.005 0.006 -0.006 -0.004 0.002 0.002 0.005 0.007

(1.38) (0.70) (1.39) (1.63) (1.08) (0.71) (0.74) (0.67) (1.52) (1.81)

EBITDA/TA 0.260 0.289 0.019 0.019 0.161 0.169 0.025 0.019 0.050 0.047

(4.62) (5.12) (0.68) (0.70) (3.62) (3.74) (1.22) (0.95) (1.73) (1.58) Sales 0.083 0.080 0.039 0.038 0.058 0.057 0.022 0.022 0.013 0.011 (13.65) (12.84) (13.28) (13.02) (11.91) (11.50) (9.63) (9.69) (4.52) (3.68)

Average stock return 0.044 0.044 0.016 0.015 0.031 0.031 0.003 0.002 0.016 0.015

(3.16) (3.14) (2.49) (2.37) (2.85) (2.82) (0.58) (0.49) (2.20) (2.09) Age -0.001 -0.001 0.000 0.000 -0.002 -0.002 0.000 0.000 -0.001 -0.001

(0.99) (1.04) (1.23) (1.30) (3.49) (3.58) (1.55) (1.54) (1.93) (2.29)

Population 0.004 0.003 -0.001 -0.001 -0.002 -0.001 -0.001 0.000 0.007 0.007

(0.65) (0.46) (0.41) (0.58) (0.45) (0.30) (0.51) (0.24) (4.05) (4.51) R2 0.12 0.12 0.07 0.08 0.12 0.12 0.05 0.05 0.06 0.06

(continued)



Table IV?Continued

Within Industry Within Industry

All Acquisitions Public Acquisitions Acquisitions Public Acq. Local Acquisitions

_(1)_(2)_(3)_(4)_(5)_(6)_(7)_(8)_(9)_(10)

Panel B: Acquisition Activity and Relative Net Market Leverage

Cluster 0.013 0.023 0.017 0.020 0.025 0.020 0.017 0.014 0.072 0.073

(0.79) (1.39) (1.92) (2.28) (1.73) (1.37) (2.37) (2.06) (8.02) (7.71)

Relative net market lever, x cluster -0.167 -0.467 -0.245

-0.364 -0.293 -0.540 -0.181 -0.288 -0.333 -0.520

(1.03) (2.62) (2.74) (3.22)

(2.12)

(3.38) (2.40) (2.78) (3.18) (3.59)

Relative net market leverage -0.232 -0.116 -0.043 -0.027 -0.191 -0.146 -0.063 -0.065 0.055 0.074

(2.68) (1.30) (1.06) (0.62) (2.79) (2.00) (1.99) (1.96) (1.46) (1.72)

Market to book -0.007 -0.005 0.005 0.005 -0.005 -0.005 0.002 0.002 0.005 0.005

(1.16) (0.76) (1.40) (1.57) (1.01) (0.84) (0.71) (0.60) (1.33) (1.39)

EBITDA/TA 0.260 0.270 0.014 0.003 0.156 0.142 0.020 0.005 0.046 0.025

(4.56) (4.59) (0.50) (0.11) (3.47) (3.01) (0.96) (0.24) (1.52) (0.77) Sales 0.083 0.082 0.039 0.040 0.058 0.060 0.022 0.023 0.014 0.014 (13.53) (12.68) (13.03) (12.56) (11.72) (11.30) (9.45) (9.21) (4.51) (3.83)

Average stock return 0.047 0.047 0.017 0.017 0.034 0.035 0.004 0.004 0.017 0.017

(3.37) (3.39) (2.71) (2.66) (3.12) (3.13) (0.82) (0.82) (2.36) (2.36)

Age -0.001 -0.001 0.000 0.000 -0.002 -0.002 0.000 0.000 0.000 -0.001 (0.93) (1.23) (1.06) (1.59) (3.31) (3.78) (1.37) (1.92) (1.64) (2.38)

Population 0.004 0.003 -0.001 -0.001 -0.002 -0.001 -0.001 -0.001 0.007 0.007

(0.66) (0.56) (0.40) (0.60) (0.43) (0.27) (0.51) (0.35) (4.08) (4.20) 0.12 0.11 0.07 0.06

0.12

0.1 0.05 0.03 0.06 0.03




(Market to Book), and age (Age), a control for city size (Population in logs), and controls for year and industry fixed effects. We run regressions for all, public, within-industry, and local acquisitions.

Panel B reports similar regressions that include a measure of a firm's fi nancial slack relative to its industry and location peers (Relative Net Market

Leverage). We build this measure by demeaning clustered and nonclustered firms' Net Market Leverage by the industry-year average inside and outside

clusters, respectively. These regressions allow us to test whether a relatively weak financial position reduces the acquisition activity of firms in clusters vis-a-vis outside clusters (i.e., whether the interaction term is negative).

The above regressions test whether firms in clusters are more likely to lose out on acquisitions when they have insufficient financial slack. As our model

illustrates, however, firms in clusters will also find it desirable to maintain more financial slack in anticipation of acquisition opportunities, which implies that there is a feedback effect from acquisition opportunities to the choice of financial slack. To address this endogeneity concern, we estimate a linear prob ability model in which we instrument our financial slack variables (Net Market

Leverage and Relative Net Market Leverage). As instruments, we use lagged R&D expenditures (R&D/TA) and asset tangibility (Tangible Assets/TA), as well as lagged values of stock returns and profitability. The use of lagged values of stock returns and profitability is motivated by recent evidence that shows that firms are slow to readjust their leverage ratios after experiencing prof itability and stock return shocks that move them away from their target debt ratios. We assume that these variables, sufficiently lagged, do not have a direct effect on the current probability of making an acquisition.18

Table IV documents that the acquisition choices of firms tend to be affected more negatively by leverage when they are located in clusters. Specifically, the interaction effect between our cluster proxies (Cluster-Firm and Cluster-MV) and the financial slack measures (Net Market Leverage and Relative Net Market

Leverage) is always negative and significant in almost all specifications. More over, as our model predicts, the effect is especially strong for transactions that are likely to be subject to more intense competition, that is, within-industry, local, and public acquisitions. These results are consistent with Hypothesis 2, which postulates that the acquisition activities of clustered firms are more affected by leverage.19 In addition, the coefficient on Net Market Leverage and Relative Net Market Leverage is always negative (except in local transactions) and significant in 11 of the 20 specifications. This negative relation between

18 Specifically, we instrument leverage at t - 1 with the 3-year average profitability and stock

returns from t ? 4 to t ? 6. See Baker and Wurgler (2002) and Welch (2004) for evidence that stock returns have a persistent effect on capital structure, and Kayhan and Titman (2007) for evidence that firms revert relatively slowly to their capital structures following shocks to profitability and stock returns.

19 It is also noteworthy that the cluster coefficient is positive in all the specifications, and it is

statistically significant for public, within-industry public, and local acquisitions. The evidence on

local acquisitions should be interpreted with caution, however. Our cluster measures depend on the number of firms located within an MSA, which also determines, by definition, the likelihood of

making a local transaction.




leverage and acquisitions is consistent with Harford (1999), who shows that financial slack increases a firm's acquisition activity. Our evidence, however, suggests that the negative effect of leverage on acquisitions, which is stronger for firms located in clusters, may arise because of acquisitions that involve

competition.20 It should be noted that in regressions reported in the Internet Appendix, we

find that the positive relation between financial slack and acquisition activity documented in Table IV holds for stock-financed as well as cash-financed ac

quisitions. This result suggests that financial slack provides firms a strategic advantage in the bidding process even when it is not used. Alternatively, finan

cial slack reduces debt overhang-related problems that may lead more levered firms to pass up value-enhancing acquisitions. In addition, we find that firms tend to reduce their cash holdings and increase their leverage in the aftermath of stock-financed as well as cash-financed acquisitions, which suggests that

having a liquid balance sheet might be important for integrating the acquired firm in the postacquisition period.21

IV. Financial Structure and Cluster Location

This section explores the relation between a firm's location and its capital structure. As we discussed earlier, location can affect a firm's capital structure

choice through two channels. In the first channel, which relates to the argu ments in Williamson (1988) and Shleifer and Vishny (1992), firms will have

greater debt capacity if their assets are more easily redeployable. Hence, if

firms in clusters can dispose of their assets more easily, they should also have

higher debt ratios. In the second channel, firms may want to have lower debt

ratios to take advantage of potential growth opportunities. These opportuni ties can arise, as we illustrate in our model, in the form of more competitive

acquisition opportunities (which are more likely to arise in clusters), or, as we

discuss in Subsection D below, as other types of growth opportunities. In fact, the evidence in Section III, which shows that the negative effect of leverage on

firms' acquisition activity is particularly strong in clusters, is consistent with

the presence of this acquisition effect.

A. Leverage Regressions

Table V examines the relation between capital structure and cluster location

after controlling for other determinants of capital structure previously iden

tified in the literature. In particular, in columns (1) through (9) we regress

20 We also perform an instrumental variables probit model (reported in the Internet Appendix)

where we split the sample into firms inside and outside clusters. Consistent with Table IV, we

find that the marginal effect of net market leverage on the probability of making an acquisition is

negative and tends to be stronger in clusters. 21 There are also reasons why stock-financed acquisitions may be more prevalent in clusters. If

within clusters there is less asymmetric information, firms are more likely to be able to use stock

instead of cash as the method of payment, which would tend to reduce their need for financial

leverage for a given level of acquisition activity.



o o 0 <;?i OS OK Ol . CO

Ol. OK Ol. o f o OK 52 g

ok" Ol. ft 3? oX 3 ? O P OK

Of. o Co Or

Table V Leverage Regressions Table V examines the relation between capital structure and cluster location after controlling for other determinants of capital structure previously

identified in the literature. Three measures of leverage are considered: book, market, and net market leverage. Each column corresponds to a different regression (OLS for columns (1) to (9) and NLLS for (10) to (13)). Sales, EBITDA/TA, Market to Book, Tangible Assets/TA, R&D/TA, and R&D

Dummy are lagged one period. Reported are the estimated coefficients

with

their ^-statistics in parentheses (standard errors are robust-clustered

by firm. For k, the reported ^-statistic corresponds to the test of hypothesis k / 1. All regressions include year fixed effects and the Cluster-MV and Cluster-Firm regressions also include industry fixed effects. (The Silicon Valley regressions study the software industry, that is, SIC 737). Details on

the definition and construction of the variables reported in the table are available in the Data Appendix. Figures in boldface indicate significance at

the 10% level.

Book Leverage Market Leverage Net Market Leverage Market Leverage Net Market Lev. (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)

Cluster-Firm -0.029 -0.021 -0.044 -0.055 -0.060

(3.05) (2.57) (4.39) (3.14) (3.43)

Cluster-MV -0.025 -0.019 -0.036 -0.032 -0.039

(2.74) (2.39) (3.63) (2.82) (3.88)

Silicon Valley -0.038 -0.029 -0.056

(1.78) (2.03) (3.06)

Sales 0.036 0.036 0.036 0.014 0.014 0.016 0.023 0.024 0.033 0.014 0.014 0.024 0.024

(12.65) (12.64) (4.75) (5.06) (5.14) (2.58) (7.56) (7.64) (4.48) (4.69) (5.01) (7.15) (7.45)

EBITDA/TA -0.439 -0.439 -0.414 -0.392 -0.392 -0.340 -0.350 -0.350 -0.279 -0.447 -0.418 -0.391 -0.364

(15.98) (15.99) (8.42) (18.44) (18.44) (9.31) (14.27) (14.23) (6.45) (17.93) (17.97) (13.80) (13.89)

Market to book -0.007 -0.007 -0.011 -0.031 -0.031 -0.021 -0.027 -0.027 -0.016 -0.039 -0.034 -0.034 -0.029

(2.90) (2.91) (2.66) (17.73) (17.62) (8.19) (12.97) (12.90) (4.88) (16.70) (17.01) (13.02) (13.05)

Tangible assets/ TA 0.083 0.083 0.456 0.073 0.072 0.352 0.210 0.209 0.499 0.068 0.067 0.206 0.207

(2.17) (2.13) (4.53) (2.07) (2.04) (3.62) (5.35) (5.31) (4.87) (1.81) (1.83) (4.83) (5.06)

R&D / TA -0.285 -0.295 0.048 -0.512 -0.518 -0.295 -0.701 -0.717 -0.439 -0.585 -0.546 -0.786 -0.741

(4.24) (4.38) (0.42) (10.09) (10.11) (3.43) (11.08) (11.22) (4.05) (9.88) (10.00) (10.73) (11.05)

R&D dummy -0.004 -0.005 0.015 0.021 0.020 0.036 0.026 0.025 0.039 0.015 0.017 0.020 0.022

(0.34) (0.39) (0.59) (1.62) (1.58) (1.61) (1.69) (1.63) (1.43) (1.14) (1.27) (1.25) (1.42)

Average stock return -0.008 -0.009 0.000 -0.092 -0.093 -0.076 -0.072 -0.073 -0.044 -0.103 -0.097 -0.082 -0.075

(1.28) (1.31) (0.03) (18.57) (18.55) (8.29) (12.06) (12.06) (3.78) (17.50) (18.05) (11.93) (11.97)

Population 0.006 0.005 0.016 0.003 0.002 0.014 0.000 -0.001 0.010 0.003 0.003 0.000 0.000

(1.58) (1.45) (1.58) (0.77) (0.73) (1.92) (0.08) (0.17) (1.18) (0.78) (0.83) (0.03) (0.12)

k 0.703 0.828 0.739 0.885 (8.49) (4.28) (5.67) (2.36)

Observations 13,342 13,342 2,631 13,342 13,342 2,631 13,342 13,342 2,631 13,342 13,342 13,342 13,342

R2 0.22 0.22 0.21 0.47 0.48 0.41 0.46 0.46 0.34 0.48 0.48 0.46 0.46




three measures of leverage (i.e., book, market, and net market leverage) on the

following variables: (i) a measure of clustering; (ii) firm size (Sales in logs); (iii)

profitability (EBITDAJTA); (iv) Market to Book, (v) asset tangibility (Tangible Assets/TA); (vi) R&D expenses (R&D/TA); (vii) the firm's average stock returns in the last 3 years (Average Stock Return); (viii) a control for the size of the

MSA in which the firm is located (Population in logs); and (ix) industry and

year fixed effects. We run separate regressions for Cluster-Firm, Cluster-MV, and Silicon Valley (as well as for Number of Firms and Ratio of Firms, which are reported in the Internet Appendix). In all regressions, Sales, EBITDA/TA,

Market to Book, Tangible Assets/TA, and R&D/TA are lagged for one fiscal

year.22 The results in Table V show that after controlling for other factors previously

identified as determinants of leverage, firms in clusters have lower debt ratios than firms outside clusters. The effect is both statistically and economically significant. For instance, firms located in clusters exhibit market leverage and net market leverage ratios that are 4.4 and 3.6 percentage points smaller than those of firms located outside clusters (according to Cluster-Firm and

Cluster-MV, respectively). Since the average net market leverage in our sample is 22.7%, this represents a reduction of 19% (for Cluster-Firm) and 16% (for Cluster-MV) from the average firm in the sample. The results are robust across

different measures of clustering and leverage, and the coefficients on other

determinants of leverage previously identified are all significant and have the

expected sign. In relative terms, the cluster effect (i.e., 4.4 and 3.6 percentage

points) is similar to a one-standard deviation change in other determinants of

leverage. For instance, a one-standard deviation change in Sales, EBITDA/TA, and Market to Book, respectively, produces a change in net market leverage of

3.5, 4.8, and 4.4 percentage points.23 These regressions can help us interpret the negative relation between debt

ratios and acquisitions documented in the previous section. Such a negative relation could be because firms with more opportunities maintain financial

slack, as we illustrate in our model, or because, as suggested by Jensen (1986),

managers may simply find it easier to exploit these opportunities when they have financial slack. However, to the extent that a cluster location is associated

with acquisition and growth opportunities, one can interpret our results as

indicating that firms with greater opportunities maintain lower debt ratios.

Notice that this interpretation relies on the exogeneity of a firm's location, an

issue that we address in Subsection C below.

22 We also include a dummy variable, R&D Dummy, that takes a value of one for firm-year observations in which R&D expenditures are not reported. Notice that our specification is similar

to Rajan and Zingales's (1995) except that we also include a proxy for clustering and three additional

controls: R&D/TA, Population, and Average Stock Return. In controlling for past stock returns we

follow Welch (2004), who shows that a firm's past stock return can be an important determinant of

its leverage ratio. In separate regressions, we include Selling Expenses I Sales and Firm Age and

get almost identical results. 23

Loughran and Schultz (2007) find that firms in rural areas are less likely to issue equity.

Our proxy for MSA size, Population, shows, however, that there is a relatively weak but positive

relation between the size of the urban area and debt ratios.




Table V also reports regressions corresponding to the leverage model pro posed in equation (7). These regressions estimate a constant X that multiplies the estimates of the determinants of leverage inside and outside of industry clusters. Our hypothesis (Hypothesis 4 from Section II) is that there is an amelioration effect in clusters (i.e., X < 1), which means that when the estima tion does not take into account the additional importance of having financial

flexibility in clusters, the determinants of capital structure may appear to be

empirically less important in clusters. Specifically, columns (10) through (13) in Table V report the estimation of equation (7), where ck corresponds to each of the measures of leverage and K- are the above-described control variables (i.e., (i) through (ix)). (Notice that if X = 1 is imposed, this nonlinear specification

would be equivalent to that reported in columns (1) through (9).) Table V documents this amelioration effect in the market leverage regres

sions, which is consistent with Hypothesis 4. For instance, according to Cluster Firm, the effect of the determinants of net market leverage is reduced by 26.1% in clusters, that is, X = 0.739. In addition to the amelioration effect, this second set of regressions also confirms the negative effect of cluster on leverage (i.e., in all four regressions the cluster proxy is negatively related to firm leverage and is significant).

B. Cash Regressions

A comparison of the leverage and the net leverage regressions indicates that firms in clusters vis-a-vis outside clusters tend to hold more cash. In this sec

tion, we explore the relation between firm location and cash holdings in more detail. Similar to our analysis of location and leverage, we estimate two types of regressions. First, we estimate OLS regressions that examine how location affects a firm's cash holdings after controlling for the usual determinants of cash. Specifically, we regress the ratio of cash and marketable securities to total assets minus cash, Cash/nTA, and its logarithm, Log(Cash/nTA), on our

measures of clustering (Cluster-Firm, Cluster-MV, and Silicon Valley) and the

following controls:24 (i) Sales (in logs, a control for firm size); (ii) Market to Book (a proxy for investment opportunities); (iii) R&D/TA (a proxy for expected costs of financial distress); (iv) Capital Exp/TA (a proxy for firms' investment needs); (v) Debt Rating, a dummy that takes a value of one if the firm has long-term debt rated by S&P and zero otherwise (a proxy for the costs of ac

cessing financial markets); (vi) Dividend, a dummy that takes a value of one if the firm pays dividends in that year and zero otherwise (to account for the ability to raise funds by reducing dividends); (vii) Cash Flow Std. Dev. (a mea sure of cash flow volatility); (viii) Average Stock Return, the firm's average stock returns in the last 3 years; (ix) Population (in logs, a control for the

24Both cash variables, Cash/nTA and \og(CashInTA), are constructed by subtracting cash balances from total assets. This is for consistency with the literature, for example, Foley et al. (2007). We obtain identical results when cash balances are not subtracted from total assets.




size of the MSA in which the firm is located); and (x) industry and year fixed effects.25

The regressions reported in Table VI, columns (1) through (6), indicate that after controlling for other variables previously identified as determinants of cash holdings, Cash/nTA is 13.9 and 10.7 percentage points higher for firms in clusters according to Cluster-Firm and Cluster-MV, respectively. Since firms in the sample have, on average, 32.3% Cash/nTA, the cluster effect represents an increase of 43% for Cluster-Firm and 33% for Cluster-MV from the aver

age cash holdings. The effect is statistically significant, and robust across the different measures of clustering and cash balances, including the analysis for Silicon Valley in SIC 737. Among the controls, Sales and Capital Exp/TA de crease cash holdings while Market to Book, R&D/TA, Cash Flow Std. Dev, and

Average Stock Return increase cash balances. With respect to these controls, the cluster effect (13.9 percentage points for Cluster-Firm) is equivalent to a

two-standard deviation change in Sales, a three-standard deviation change in Capital Exp/TA, a one-standard deviation change in Market to Book and

R&D/TA, and a six-standard deviation change in Average Stock Return.

We also estimate a nonlinear model similar to the one proposed by (7) and estimated in Table V:

log (Cash/nTA)i = 0O + <t>c Cluster + ]P fyK} -

]T <pj; (1 -

X)(K( x Cluster^, j j (8)

where K? are the above-described control variables (i.e., (i) through (x)). As in the NLLS leverage regressions from Table V, the estimate of X allows us

to test for the presence of the amelioration effect (i.e., X < 1). The NLLS es

timations of equation (8), which are reported in columns (7) and (9), confirm

that firms in clusters hold more cash. In addition, according to Cluster-Firm, there is evidence of a significant amelioration effect in the determinants of cash

holdings in clusters. Specifically, the amelioration effect, which is statistically

significant at the 1% level, represents a reduction of 15.3% (i.e., X = 0.847) in

the coefficients of the determinants of cash holdings in clusters. (The estimated

amelioration effect for Cluster-MV is 6.5%, i.e., X = 0.935, but in this case it is

not statistically significant.) Overall, the findings in this section are consistent

with the previous results that indicate that firms in clusters maintain more

financial slack.

C. The Endogeneity of Location

Our results up to this point can be summarized as follows: Firms in clusters (i) make more acquisitions; (ii) have acquisition activities that are more negatively

25 Controls (i) through (vii) are those considered in Foley et al. (2007) and, as in their analysis, they are lagged for one fiscal year. As in the leverage regressions, we set R&D equal to zero if

the R&D value is not reported and we include a dummy variable for these observations, R&D

Dummy. See Opler et al. (1999), Dittmar, Mahrt-Smith, and Servaes (2003), Acharya, Almeida,

and Campello (2007), and Harford, Mansi, and Maxwell (2008) for other papers that examine firms'

decisions to hold cash.




Table VI Cash Regressions

Table VI examines the relation between cash holdings and cluster location after controlling for

other determinants of capital structure previously identified in the literature. Two measures of cash

holdings are considered: Cash/nTA and Log(Cash/nTA). Each column corresponds to a different

regression (OLS for columns (1) to (6) and NLLS for (7) and (8)). Sales, Market to Book, R&D/TA, R&D Dummy, Capital Expenditures I TA, and Dividend are lagged one period. Reported are the

estimated coefficients with their ^-statistics clustered in parentheses (standard errors are robust

clustered by firm). For A the reported ^-statistic corresponds to the test of the hypothesis A ̂ 1.

All regressions include year fixed effects and the Cluster-MV and Cluster-Firm regressions also

include industry fixed effects. (The Silicon Valley regressions study the software industry, that

is, SIC 737.) Details on the definition and construction of the variables reported in the table are

available in the Data Appendix. Figures in boldface indicate significance at the 10% level.

Cash/nTA Log (Cash/nTA)

(1) (2) (3) (4) (5) (6) (7) (8)

Cluster-Firm 0.139 0.395 0.634

(7.05) (6.47) (4.83) Cluster-MV 0.107 0.243 0.352

(5.54) (4.02) (2.77) Silicon Valley 0.246 0.500

(4.31) (4.92) Sales -0.046 -0.046 -0.062 -0.073 -0.072 -0.145 -0.077 -0.073

(8.28) (8.33) (2.99) (3.29) (3.22) (3.01) (3.24) (3.20) Market to book 0.079 0.080 0.072 0.218 0.220 0.164 0.240 0.228

(12.71) (12.68) (6.64) (16.13) (16.30) (7.32) (13.98) (15.31) R&D/TA 1.378 1.430 1.342 4.952 5.133 4.418 5.304 5.213

(10.02) (10.40) (5.69) (12.54) (12.83) (6.24) (11.73) (12.52) R&D dummy -0.012 -0.009 0.004 -0.243 -0.237 -0.144 -0.233 -0.236

(0.69) (0.54) (0.10) (2.51) (2.44) (0.86) (2.32) (2.39) Capital Exp./TA -0.783 -0.783 -0.555 -1.565 -1.567 -1.271 -1.672 -1.603

(8.08) (8.03) (2.49) (4.21) (4.19) (1.61) (4.21) (4.17) Debt rating -0.097 -0.087 -0.008 0.008 0.032 0.508 0.030 0.046

(2.85) (2.54) (0.09) (0.04) (0.17) (1.48) (0.14) (0.23) Dividend 0.006 -0.001 -0.039 0.019 -0.001 -0.267 0.040 0.007

(0.36) (0.06) (0.74) (0.27) (0.01) (1.67) (0.53) (0.10) Cash flow Std. Dev. 0.665 0.676 -0.852 2.431 2.522 0.129 2.689 2.661

(1.81) (1.82) (1.46) (2.54) (2.62) (0.08) (2.62) (2.71) Average stock return 0.051 0.052 0.067 0.257 0.260 0.159 0.273 0.263

(3.89) (3.96) (2.18) (6.87) (6.89) (2.27) (6.74) (6.80) Population 0.008 0.012 0.031 0.055 0.073 0.109 0.057 0.072

(1.92) (2.62) (2.05) (2.11) (2.81) (2.02) (2.18) (2.77) A 0.847 0.935

(2.66) (1.13) Observations 13,342 13,342 2,631 13,277 13,277 2,627 13,277 13,277 R2 0.38 0.37 0.28 0.43 0.42 0.33 0.43 0.43

affected by leverage; (iii) have more financial slack (i.e., less leverage and more cash); and (iv) are less sensitive to the usual determinants of financial slack. These findings reject the null hypothesis that a firm's corporate finance decisions are independent of location and, to the extent that location choices are exogenous with respect to the firm's capital structure choice, support the




hypothesis that firms maintain financial slack when they anticipate acquisition opportunities.

A potential concern is that firms with better growth and acquisition opportu nities as well as more financial slack choose to locate in clusters. For instance, Almazan, de Motta, and Titman (2007) show that clusters are likely to attract firms with attributes that make them more likely to succeed. This cluster selection effect can potentially be a serious issue for young firms that have

recently chosen their locations. However, to the extent that unobserved char acteristics that may influence a firm's location choice become less important over time, the observed effect on capital structures of older firms that chose locations many years ago is unlikely to arise because of a cluster-selection ef fect.26 For this reason, it is interesting to explore whether the relation between

capital structure and location for older firms is indeed consistent with what we

observe for the entire sample. In Table VII, we present regressions that replicate our previous analysis

but on a sample of firms that have been public for at least 10 years. As was

the case in the full sample, Panel A documents that firms in clusters make more acquisitions and Panel B shows that leverage reduces the tendency to make acquisitions more in clusters. Finally, Panel C documents that older firms in clusters have lower net leverage and larger cash balances than older firms outside clusters. For instance, according to Cluster-Firm, older firms located in clusters exhibit net market leverage ratios that are 3.6 percentage points smaller than those of older firms located outside clusters (as reported in

Table V, the effect for the whole sample is 4.4 percentage points). Overall, the

consistency of the effects between older firms and the entire sample suggests that the results are not likely to be due to better quality firms locating in clusters.

D. Location, Growth Opportunities, and Financial Slack

Up to this point our focus has been on the greater acquisition opportunities that clusters offer, and how this influences the demand for financial slack.

Our focus on clusters is motivated by the economic geography literature that

suggests that opportunities may be more available to firms that are located close to their industry peers, for instance, due to the importance of input sharing and resource pooling, while our focus on acquisitions is motivated by data

considerations. Since acquisitions must be reported publicly, they provide an

observable example of a growth opportunity that may influence a firm's desire

to hold financial slack. In reality, however, a firm's location can influence its opportunities in a va

riety of ways. For example, firms may find it easier to grow and innovate in

regions with widespread growth and innovation. Furthermore, if firms main

tain financial slack in anticipation of growth opportunities, then in addition

26 Pirinsky and Wang (2006) document that from 1992 to 1997 less than 2.4% of firms in Com

pustat moved their headquarters from one MSA to another (118 out of 5,000 firms did so).




Table VII Older Firm Regressions

Table VII presents regressions on a sample of firms that have been public for at least 10 years. In Panel A (which is equivalent to Table III), each numbered column and row corresponds to a

different regression. Odd-numbered columns in Panel A correspond to OLS regressions in which

the dependent variable is a dummy variable indicating an acquisition. (Several types of acquisitions are considered: All, Public, Within Industry, and Local, as defined in the Data Appendix). Even

numbered columns in Panel A correspond to Tobit regressions in which the dependent variable

is the ratio of the firm's acquisition volume during the year divided by the firm's total assets. In

Panel B (which is equivalent to Table IV), each column corresponds to a different 2SLS regression in which the dependent variable is a dummy variable indicating an acquisition. (Several types of acquisitions are considered: All, Public, Within Industry, and Local, as defined in the Data

Appendix.) All regressions in Panel B include the same controls as the regressions in Table IV, that is, Market to Book, EBITDA/TA, Sales, Average Stock Return, Age, and Population, as well as

Net Market Leverage and Relative Net Market Leverage for the odd- and even-numbered columns,

respectively. We also use the same instruments for net leverage as in Table IV. In Panel C (which is equivalent to Tables V and VI), each column corresponds to a different OLS regression. The

leverage regressions (first four columns in Panel C) include the same controls as in Table V, and

the cash regressions (last four columns in Panel C) include the same controls as in Table VI.

All regressions in Table VII include year and industry fixed effects. Reported are the estimated

coefficients with their ^-statistics in parentheses (standard errors are robust-clustered by firm). Details on the definition and construction of the variables reported in the table are available in

the Data Appendix. Figures in boldface indicate significance at the 10% level.

Panel A: Acquisition Activity

Public Within Industry Local

All Acquisitions Acquisitions Acquisitions Acquisitions (1) (2) (3) (4) (5) (6) (7) (8)

(1) Cluster-Firm 0.048 0.062 0.032 0.141 0.055 0.082 0.090 0.249 Observations: 8,415 (2.27) (2.61) (2.93) (3.21) (3.20) (3.09) (9.07) (9.88)

R2 0.05 0.03 0.080 0.05

(2) Cluster-MV 0.087 0.098 0.048 0.214 0.061 0.094 0.088 0.243 Observations: 8,415 (4.23) (4.15) (4.53) (5.34) (3.92) (3.81) (8.79) (8.83)

R2_O05_O04_O08_O06_ Panel B: Acquisition Activity and Leverage

Public Within Industry Local All Acquisitions Acquisitions Acquisitions Acquisitions

Cluster-Firm -0.02 0.002 0.031 0.016 0.034 0.034 0.121 0.088

(0.62) (0.10) (1.74) (1.36) (1.26) (1.83) (6.52) (7.47) Net market leverage x 0.035 -0.157 -0.129 -0.29

cluster firm (0.24) (2.13) (1.08) (3.52) Relative net market -0.04 -0.207 -0.264 -0.448

leverage x cluster-firm (0.19) (1.74) (1.40) (3.28)

Observations 8,245 8,245 8,245 8,245 8,245 8,245 8,245 8,245 R2 0.13 0.13 0.08 0.08 0.13 0.12 0.07 0.05

(continued)




Table VII?Continued

Panel C: Leverage and Cash Regressions

Net Market

Market Leverage Leverage Cash/nTA Log (Cash/nTA)

Cluster-firm -0.015 -0.036 0.111 0.354

(1.55) (2.70) (4.66) (4.29) Cluster-MV -0.017 -0.031 0.077 0.192

(1.62) (2.45) (3.45) (2.35) Observations 8,415 8,415 8,415 8,415 8,415 8,415 8,377 8,377

R2 0.48 0.48 0.44 0.44 0.36 0.36 0.38 0.38

to the cluster effect, other geographical characteristics that may also be re lated to future opportunities may also influence debt ratios and cash holdings. To test this hypothesis, we consider the following four characteristics: (i) the MSA growth rate (Population Growth); (ii) R&D expenditures within the MSA, which we measure by the ratio of the aggregate R&D expenditures to the firms'

aggregate total assets in the MSA (MSA R&D/MSA TA); (iii) acquisitiveness of firms within the MSA, which we measure by the ratio of the aggregate dollar volume of acquisitions by firms in the MSA to the firms' aggregate total assets in the MSA (MSA Acquisitions/MSA TA); and (iv) the average stock return of firms within the MSA (MSA Stock Returns).

Table VIII documents a significant relation between these regional charac teristics and firms' capital structure choices. Specifically, firms in high R&D and growing MSAs, as well as firms in MSAs with greater acquisition activity and stock returns, have lower leverage and hold more cash. Furthermore, after

controlling for these regional characteristics, the effect of clusters on financial slack continues to be significant and of similar magnitude, which implies that the cluster effect documented in previous regressions is not attributable to other characteristics of MSAs that happen to host industry clusters.27

Documenting the effect of these characteristics provides further evidence that growth opportunities affect firms' financing choices. The current literature on capital structure documents an effect running from financial choices to investment choices (e.g., Lang, Ofek, and Stulz (1996) and Lamont (1997)). In contrast, to the extent that regional growth effects are related to a firm's

growth opportunities but are exogenous with respect to the firm's financial

choice, Table VIII shows an effect running from financial choices to investment

choices.

Table VIII also tests for the possibility that the documented cluster effect is stronger in those MSAs with a favorable growth environment. Specifically, the regressions in columns (3) and (4) include an interaction term between

27 For example, the coefficients on Cluster-Firm and Cluster-MV in the net market leverage

regressions of Panel A that include MSA Acquisitions/MSA TA as a control (columns (1) and

(2) under the MSA Acquisitions/MSA TA heading) are -0.038 and -0.031, respectively, and the

corresponding coefficients in Table V are ?0.044 and ?0.036.



o O < ?i CO ok OK ol ol Ol. CO

o*. OK Ol . o f

OK Ol. OK Ol.

p oX a o p OK Oi. o Co Cn Oi Oi

Table

VIII MSA-Specific Controls

Table VIII documents the relation between regional characteristics and

firms'

capital structure choices. Four regional characteristics are considered: (i)

the MSA growth rate (Population Growth); (ii) R&D expenditures

within

the MSA, which we measure by the ratio of the aggregate R&D expenditures

to the firms' aggregate total assets in the MSA (MSA R&D I MSA TA); (Hi) acquisitiveness of firms within the MSA, which we measure by the

ratio of the aggregate dollar volume of acquisitions by firms in the

MSA

to

the firms' aggregate total assets in the MSA (MSA Acquisitions /MSA TA); and (iu) the average stock return of firms within the MSA (MSA Stock Returns). Each column corresponds to a different OLS regression. In

the odd-numbered columns the independent variable "Cluster" corresponds to Cluster-Firm, and in the even-numbered columns it corresponds to

Cluster-MV. The independent variable "Regional Activity"

corresponds

to MSA Acquisitions IMSA TA, MSA Stock Returns, Population Growth, or

MSA R&D I MSA TA according to the corresponding column heading. Regressions in Panel A include the same controls as the leverage regressions in

Table V (to save space these controls are reported in the Internet

Appendix).

That is, each regression in Panel A includes Sales, EBITDA I TA, Market

to Book, Tang. Assets/TA, R&D/TA, R&D Dummy, Average Stock

Return, and Population. Regressions in Panel B include the same controls as the

cash regressions in Table VI (to save space these controls are reported

in the Internet Appendix). That is, each regression in Panel B includes Sales,

Market to Book, R&D/TA, R&D Dummy, Debt Rating, Dividend, Cash Flow

Std. Deviation, Average Stock Return, and Population. All regressions in

Table VIII include year and industry fixed effects. Reported are the

estimated coefficients with their ^-statistics in parentheses (standard errors are

robust-clustered by firm). Details on the definition and construction

of the variables reported in the table are available in the Data Appendix. Figures

in boldface indicate significance at the 10% level.

Panel A: Net Market Leverage Regressions

MSA Acquisitions/ MSA TA _MSA Stock

Returns_

_Population Growth_ _MSA R&D/MSA TA_

(1) (2) (3) (4) (1) (2) (3) (4) (1) (2) (3) (4) (1) (2) (3) (4) Cluster -0.038 -0.031 -0.031 -0.021 -0.042 -0.034 -0.043

-0.026 -0.049 -0.041 -0.045 -0.032 -0.035 -0.028 -0.011 -0.013

(3.80) (3.19) (2.67) (1.83) (4.10) (3.41) (3.08)

(1.80)

(4.85) (4.08) (3.88) (2.85) (3.17) (2.65) (0.66) (1.19)

Regional activity -0.576 -0.608 -0.451 -0.444 -0.103 -0.114 -0.105 -0.100 -0.837 -0.750 -0.768 -0.581 -1.237 -1.413 -0.67 -0.616

(6.52) (6.89) (4.14) (4.14) (4.03) (4.39) (3.36)

(3.33)

(2.06) (1.84) (1.74) (1.34) (2.54) (2.97) (1.04) (1.02)

Cluster x regional -0.286 -0.396 0.005 -0.039 -0.382 -0.896 -1.597 -1.691

activity (1.82) (2.45) (0.14) (1.11) (0.73) (1.48) (1.73) (2.61)

Observations 13,342 13,342 13,342 13,342 13,079 13,079 13,079 13,079 12,073 12,703 12,703 12,703 13,127 13,127 13,127 13,127

R2 0.46 0.46 0.46 0.46 0.46 0.46 0.46 0.46 0.46 0.46 0.46 0.45 0.46 0.46 0.46 0.46

(continued)



Ol Ol p P 51 o

Table VIII?Continued Panel B: Log (Cash/nTA) Regressions MSA Acquisitions/ MSA TA _MSA Stock Returns_ _Population Growth_ _MSA R&D/MSA TA_

(1) (2) (3) (4) (1) (2) (3) (4) (1) (2) (3) (4) (1) (2) (3) (4) Cluster 0.375 0.224 0.312 0.180 0.384 0.239 0.303 0.105 0.414 0.261 0.455 0.314 0.284 0.151 -0.029 -0.023 (6.09) (3.70) (4.47) (2.56) (6.28) (3.95) (3.78) (1.25) (6.70) (4.20) (6.38) (4.51) (4.29) (2.39) (0.28) (0.34)

Regional activity 2.031 2.503 1.014 1.796 0.584 0.707 0.443 0.492 3.912 2.547 4.658 3.612 14.614 16.951 7.132 7.329

(3.37) (4.14) (1.27) (2.29) (3.27) (3.91) (1.99) (2.29) (1.39) (0.90) (1.51) (1.18) (4.91) '(5.81) (1.83) (2.00)

Cluster x regional 2.315 1.694 0.372 0.616 -4.127 -5.616 21.064 20.460

activity (2.29) (1.63) (1.74) (2.75) (1.27) (1.48) (3.65) (5.20)

Observations 13,277 13,277 13,277 13,277 13,021 13,021 13,021 13,021 12,640 12,640 12,640 12,640 13,068 13,068 13,068 13,068

R2 0.43 0.43 0.43 0.43 0.43 0.43 0.43 0.43 0.43 0.43 0.43 0.43 0.44 0.43 0.44 0.44




the cluster proxy and the regional characteristics. As shown in the columns under the MSA Acquisitions/MSA TA heading, the coefficients on Cluster, MSA

Acquisitions IMSA TA, and Cluster x MSA Acquisitions I MSA TA (i.e., the interaction term) are negative and statistically significant in the net market

leverage regressions of Panel A, and positive and also statistically significant in the cash regressions of Panel B. These results confirm that firms located in clusters and in acquisitive MSAs have less leverage and hold more cash, and

they also document that the effect of MSA acquisitiveness on firms' leverage and cash holdings is particularly strong in clusters.28 In addition, we find that the effect of MSA Stock Returns and MSA R&D I MSA TA on firms' cash holdings is stronger in clusters. (For net market leverage, only the effect MSA R&D I MSA TA is significantly stronger in clusters). While the effect of Population Growth is not statistically different inside and outside clusters, overall the results in Table VIII provide evidence that the cluster effect is indeed stronger in those MSAs that have a favorable growth environment.

V. Concluding Remarks

A growing economic geography literature describes how a firm's location can be related to a number of its choices. This paper contributes to that literature

by documenting that firms that are located in industry clusters are involved in more acquisitions and maintain more financial slack than their industry peers that are located away from clusters. Although our main focus is on industry clusters, like Silicon Valley, we also consider other urban characteristics, such as the size of the metropolitan area, the rate of growth of the metropolitan area, and the R&D intensity of the metropolitan area. In this regard, we find that, ceteris paribus, firms in high-tech cities and growing cities tend to maintain

more financial slack. This paper contributes to the corporate finance literature that examines the

determinants of corporate debt ratios and cash holdings. Up to now, the focus of this literature has been on the characteristics of firms, and the extent to which these characteristics correlate with how they are financed. In this paper, we show that firms' locations are also related to their capital structure choices, and we argue that firms in locations that facilitate investment opportunities maintain more financial slack.

As we discussed in the introduction, although the negative relation between acquisitions and debt ratios has been previously established, the direction of causation between opportunities and financial slack has not been fully re solved. Given the arguments in Jensen (1986) and others, it is plausible that this association arises because firms exploit opportunities when they have fi nancial slack rather than maintaining financial slack when they anticipate

28 This evidence, which directly links the lower leverage in cluster to a higher acquisition rate, is also consistent with the evidence provided in Table IV (i.e., with the regressions that show that the negative effect of leverage on the probability of making an acquisition is particularly strong in clusters).




opportunities. However, since cluster and city growth effects are likely to be

directly related to acquisitions and other opportunities, but can be viewed as

exogenous with respect to the firm's capital structure choice, the observed re

lation between geography and financial slack in this study is likely to arise from the effect that the presence of potential opportunities produces on a firm's desire to maintain financial slack.

The findings in this paper, which indicate a link between a firm's location and its growth opportunities, suggest that it may be fruitful to examine how

geography affects other corporate finance choices. Recent research by Kedia and Rajgopal (2009) explores the effects of geography on the compensation of both workers and executives. Perhaps our cluster and MSA characteristics can be used to extend this research to better understand why the use of stock

options differs across MSAs. Similarly, it would be interesting to explore how

MSA characteristics influence governance and the ownership structure of firms.

These and other related research challenges are left for future work.

Data Appendix

(All) Acquisitions refers to all the transactions listed in the SDC-M&A

database as an acquisition of majority interest, merger, asset acqui sition, or acquisition of certain assets. (Please see below for the exact

definitions of Acquisition of Majority Interest, Merger, and Asset Acqui sitions, according to the SDC-M&A database.)

Acquisition of Majority Interest is a transaction in which the acquirer must

have held less than 50% and be seeking to acquire 50% or more (but less

than 100%) of the target company's stock.

Acquisitions Value I TA is the ratio of the total Transaction Value of (All)

Acquisitions made by the firm during the fiscal year to the firm's Total

Assets (Item 6) at the beginning of the fiscal year.

Age is the number of years that the firm has been public. Asset Acquisitions refers to transactions classified as an acquisition of assets

or acquisition of certain assets by the SDC-M&A database. The SDC

M&A database defines acquisition of assets as deals in which the assets

of a company, subsidiary, division, or branch are acquired. Acquisition of certain assets is defined as deals in which sources state that certain

assets of a company, subsidiary, or division are acquired. Asset Acquisitions Value/TA is the ratio of the total Transaction Value of

Asset Acquisitions made by the firm during the fiscal year to the firm's

Total Assets (Item 6) at the beginning of the fiscal year. Asset Sales are all the transactions in which a firm is identified as the ulti

mate parent of the target in asset acquisition or acquisition of certain

assets in the SDC-M&A database.

Average Stock Return is the firm's average Stock Return (as defined below) over the last 3 years.

Book Debt is Total Assets (Item 6) minus Book Equity (as defined below).




Book Equity is defined as Total Assets (Item 6) minus liabilities (Item 181)

plus balance sheet deferred taxes and investment tax credit (Item 35) minus Preferred Stock (as defined below).

Book Leverage is Book Debt over Total Assets (Item 6).

Capital Expenditures I TA is capital expenditures (Item 128) over Total As sets (Item 6).

Cash Flow Std. Deviation is the standard deviation of EBITDA/TA during the sample period.

Cash/nTA is cash and marketable securities (Item 1) over Total Assets (Item 6) minus cash (Item 1).

Cluster-Firm is a dummy variable that takes a value of one for firm-years in which a firm's headquarters is located within an MSA (Metropolitan Statistical Area as identified by the U.S. Census Bureau in 1990) that has 10 or more firms with the same three-digit SIC, and these firms

comprise at least 3% of the Market Value of the industry, and zero otherwise.

Cluster-MV is a dummy variable that takes a value of one for firm-years in which a firm's headquarters is located within an MSA that represents at least 10% of the Market Value of the firm's industry and that has at least three firms with the same three-digit SIC, and zero otherwise.

Debt Rating takes a value of one if the firm has an S&P investment grade long-term debt rating, and zero otherwise.

Dividend takes a value of one if the firm pays a dividend (Item 26), and zero otherwise.

EBITDA/TA is EBITDA (Item 13) over lagged Total Assets (Item 6). Firm Acquisitions refers to transactions classified as Acquisition of Majority

Interest or Merger in the SDC-M&A database. Firm Acquisitions Value/TA is the ratio of the total Transaction Value of

Firm Acquisitions made by the firm during the fiscal year to Total Assets (Item 6) at the beginning of the fiscal year.

Local Acquisitions refers to All Acquisitions in which the acquirer and the

target are located within the same MSA. The target's MSA is determined

by the target's state, city, and/or zip code given in the SDC Database. Local Acquisitions Value/TA is the ratio of the total Transaction Value of

Local Acquisitions made by the firm during the fiscal year to the firm's Total Assets (Item 6) at the beginning of the fiscal year.

Market Equity is common shares outstanding (Item 25) times the stock price (Item 199).

Market Leverage is Book Debt over Market Value (as defined below). Market-to-Book ratio is Market Value (as defined below) over Total Assets

(Item 6). Market Value is defined as liabilities (Item 181) minus balance sheet de

ferred taxes and investment tax credit (Item 35) plus Preferred Stock (as defined below) plus Market Equity (Item 25 x Item 199).

Merger is a transaction in which a combination of businesses takes place or 100% of the stock of a public or private company is acquired.




MSA Acquisitions I MSA TA is the ratio of total dollar volume of All Acqui sitions conducted by firms located in an MSA during the fiscal year to the sum of the Total Assets (Item 6) of the firms in the MSA. For each

firm-year observation, we compute this variable by excluding the firm of interest.

MSA R&D I MSA TA is the ratio of total R&D expenses (Item 46) of all firms located in an MSA to the sum of the Total Assets (Item 6) of the firms in the MSA. For each firm-year observation, we compute this variable by excluding the firm of interest.

MSA Stock Returns is the average stock return of all firms in the MSA over the past 3 years. For each firm-year observation, we compute this variable by excluding the firm of interest.

Net Book Leverage is Book Debt minus cash and marketable securities (Item 1) over Total Assets (Item 6).

Net Cash/TA is cash and marketable securities (Item 1) minus short-term debt (Item 34) over Total Assets (Item 6).

Net Market Leverage is Book Debt minus cash and marketable securities

(Item 1) over Market Value. Number of Firms is the number of firms with the same three-digit SIC within

an MSA.

Population is the natural logarithm of the population of the MSA.

Population Growth is the annual population growth of the MSA.

Preferred Stock is equal to liquidating value (Item 10) if available, or re

demption value (Item 56) if available, or carrying value (Item 130). Public Acquisitions refers to All Acquisitions in which the target (as defined

by the SDC-M&A database) is a public firm. Public Acquisitions Value/TA is the ratio of the total Transaction Value of

Public Acquisitions made by the firm during the fiscal year to the firm's Total Assets (Item 6) at the beginning of the fiscal year.

R&D Dummy is a dummy variable that takes a value of one if Compustat reports R&D expense (Item 46) as missing, and zero otherwise.

R&D/TA is defined as R&D expenses (Item 46) over Total Assets (Item 6). Ratio of Acquirers is the proportion of firms that are listed as acquirers in

acquisitions of majority interest, mergers, asset acquisitions, or acqui sitions of certain assets as defined in the SDC-M&A database.

Ratio of Asset Acquirers is the proportion of firms that are listed as acquirers in an asset acquisition or in an acquisition of certain assets as defined in the SDC-M&A database.

Ratio of Firm Acquirers is the proportion of firms that are listed as acquirers in an acquisition of majority interest or in a merger as defined in the

SDC-M&A database. Ratio of Firms is the number of firms with the same three-digit SIC in an

MSA divided by the total number of firms with the same three-digit SIC.

Ratio of Local Acquirers is the proportion of firms that are listed as ac

quirers in acquisitions of majority interest, mergers, asset acquisitions, or acquisitions of certain assets as defined in the SDC-M&A database,




and in which the acquirer and the target are located within the same

MSA. Ratio of Public Acquirers is the proportion of firms that are listed as acquir

ers in acquisitions of majority interest, mergers, asset acquisitions, or

acquisitions of certain assets as defined in the SDC-M&A database, and in which the target is a public firm.

Relative EBITDA/TA is the average EBITDA/TA minus the industry-year average EBITDA/TA of firms with similar cluster status (according to the definition of cluster use in the corresponding regression, that is, either Cluster-Firm or Cluster-MV).

Relative Net Market Leverage is Net Market Leverage minus the industry year average Net Market Leverage of firms with similar cluster status

(according to the definition of cluster use in the corresponding regres sion, i.e., either Cluster-Firm or Cluster-MV).

Relative Stock Return is Stock Return (as defined below) minus the industry year average Stock Return of firms with similar cluster status (according to the definition of cluster use in the corresponding regression, i.e., either Cluster-Firm or Cluster-MV).

Sales is the natural logarithm of sales (Item 12) in 1990 dollars. Silicon Valley is a dummy variable that takes a value of one if the firm is

located in MSA code 7362 as defined by in the 1990 Census, and zero otherwise. MSA code number 7362 includes San Francisco, Oakland, and San Jose in California.

Stock Acquisitions refers to all-stock financed All Acquisitions. Stock Return is the firm's annual stock return.

Tangible Assets / TA is net property, plant, and equipment (Item 8) over Total Assets (Item 6).

Total Assets (TA) is the book value of assets (Item 6). Transaction Value is total value of consideration paid by the acquirer, ex

cluding fees and expenses. The dollar value includes the amount paid for all common stock, common stock equivalents, preferred stock, debt, options, assets, warrants, and stake purchases made within 6 months of the announcement date of the transaction.

Within Industry Acquisitions refers to All Acquisitions in which the acquirer and the target belong to the same three-digit SIC.

Within Industry Acquisitions Value/TA is the ratio of the total Transaction Value of Within Industry Acquisitions made by the firm during the fiscal year to the firm's Total Assets (Item 6) at the beginning of the fiscal year.

Within Industry Bidder is a dummy variable that takes a value of one if the firm is identified as an acquirer in a Within Industry Acquisition, and zero otherwise.

Within Industry Public Acquisitions refers to All Acquisitions in which the

acquirer and the target belong to the same three-digit SIC, and in which the target is a public firm.

Within Industry Public Acquisitions Value/TA is the ratio of the total Trans action Value of Within Industry Public Acquisitions made by the firm




during the fiscal year to the firm's Total Assets (Item 6) at the beginning of the fiscal year.

REFERENCES

Acharya, Viral, V., Heitor Almeida, and Murillo Campello, 2007, Is cash negative debt? A hedging

perspective on corporate financial policies, Journal of Financial Intermediation 16, 515-554.

Almazan, Andres, Adolfo de Motta, and Sheridan Titman, 2007, Firm location and the creation

and utilization of human capital, Review of Economic Studies 74, 1305-1327.

Audretsch, David B., and Maryann Feldman, 1996, R&D spillovers and the geography of innovation

and production, American Economic Review 86, 630-640.

Baker, Malcolm, and Jeffrey Wurgler, 2002, Market timing and capital structure, Journal of Finance 57, 1-32.

Boone, Audra L., and J. Harold Mulherin, 2007, Do termination provisions truncate the takeover

bidding process? Review of Financial Studies 20, 461^489.

Clayton, Matthew J., and S. Abraham Ravid, 2002, The effect of leverage on bidding behavior:

Theory and evidence from the FCC auctions, Review of Financial Studies 15, 723-750.

Costa, Dora, and Matthew E. Kahn, 2000, Power couples, Quarterly Journal of Economics 116, 1287-1315.

Diamond, Charles A., and Curtis J. Simon, 1990, Industrial specialization and the returns to labor, Journal of Labor Economics 8, 175-201.

Dittmar, Amy, Jan Mahrt-Smith, and Henri Servaes, 2003, International corporate governance and corporate cash holdings, Journal of Financial and Quantitative Analysis 28, 111-133.

Dumais, Guy, Glenn Ellison, and Edward L. Glaeser, 1997, Geographic concentration as a dynamic

process, National Bureau of Economic Research Working Paper no. 6270.

Duranton, Gilles, and Diego Puga, 2004, Microfoundations of urban agglomeration economies, in

J. V. Henderson and J. F. Thisse, eds.: Handbook of Regional and Urban Economics Vol. 4

(Elsevier, Amsterdam).

Foley, C. Fritz, Jay C. Hartzell, Sheridan Titman, and Garry Twite, 2007, Why do firms hold so

much cash? A tax-based explanation, Journal of Financial Economics 86, 579-607.

Fujita, Masahisa, Paul R. Krugman, and Anthony Venables, 2001, The Spatial Economy (MIT

Press, Cambridge, MA).

Fujita, Masahisa, and Jacques-Francois Thisse, 2002, Economics of Agglomeration (Cambridge

University Press, Cambridge, UK).

Glaeser, Edward L., Hedi D. Kallal, Jose A. Scheinkman, and Andrei Shleifer, 1992, Growth in

cities, Journal of Political Economy 100, 1126-1152.

Glaeser, Edward L., and Janet E. Kohlhase, 2004, Cities, regions and the decline of transportation

costs, Papers in Regional Science 83, 197-228.

Harford, Jarrad, 1999, Corporate cash reserves and acquisitions, Journal of Finance 54,1969-1997.

Harford, Jarrad, Sattar A. Mansi, and William F. Maxwell, 2008, Corporate governance and a

firm's cash holdings, Journal of Financial Economics 87, 535-555.

Hart, Oliver, and John Moore, 1995, Debt and seniority: An analysis of the role of hard claims in

constraining management, American Economic Review 85, 567-585.

Holmes, Thomas J., 1999, Localization of industry and vertical disintegration, Review of Economics

and Statistics 81, 314-325.

Jaffe, Adam B., Manuel Trajtenberg, and Rebecca Henderson, 1993, Geographic localization of

knowledge spillovers as evidenced by patent citations, Quarterly Journal of Economics 108,

577-598.

Jensen, Michael C, 1986, Agency costs of free cash flow, corporate finance and takeovers, American

Economic Review 76, 323-329.

Kayhan, Ayla, and Sheridan Titman, 2007, Firms' histories and their capital structures, Journal

of Financial Economics 83, 1-32.

Kedia, Simi, Venkatesh Panchapagesan, and Vahap Uysal, 2008, Geography and acquirer returns,

Journal of Financial Intermediation 17, 256-275.




Kedia, Simi, and Shiva Rajgopal, 2009, Neighborhood matters: The impact of location of broad

based option plans, Journal of Financial Economics 92, 109-127.

Lamont, Owen, 1997, Cash flow and investment: Evidence from internal capital markets, Journal

of Finance 52, 57-82.

Landier, Agustin, Vinay B. Nair, and Julie Wulf, 2009, Tradeoffs in staying close: Corporate

decision-making and geographic dispersion, Review of Financial Studies 22, 1119-1148.

Lang, Larry, Eli Ofek, and Rene Stulz, 1996, Leverage, investment, and firm growth, Journal of Financial Economics 40, 3-29.

Loughran, Tim, and Paul Schultz, 2007, Asymmetric information, firm location, and equity is

suance, Working paper, University of Notre Dame.

Marshall, Alfred, 1890, Principles of Economics (Macmillan, London).

Morellec, Erwan, and Alexei Zhdanov, 2008, Financing and takeovers, Journal of Financial Eco

nomics 87, 556-581.

Myers, Stewart C, 1977, Determinants of corporate borrowing, Journal of Financial Economics 5, 147-175.

Myers, Stewart C, and Nicholas S. Majluf, 1984, Corporate financing and investment decisions

when firms have information that investors do not have, Journal of Financial Economics 13, 187-221.

Ono, Yukako, 2003, Outsourcing business services and the role of central administrative offices, Journal of Urban Economics 53, 377-395.

Opler, Tim, Lee Pinkowitz, Rene Stulz, and Rohan Williamson, 1999, The determinants and im

plications of corporate cash holdings, Journal of Financial Economics 52, 3-46.

Pirinsky, Christo, and Qinghai Wang, 2006, Does corporate headquarters' location matter for stock returns? Journal of Finance 61, 1991-2015.

Rajan, Raghuram G., and Luigi Zingales, 1995, What do we know about capital structure? Some evidence from international data, Journal of Finance 50, 1421-1460.

Rosenthal, Stuart S., and William C. Strange, 2004, Evidence on the nature and sources of agglom eration economics, in J. V. Henderson and J. F. Thisse, eds.: Handbook of Regional and Urban Economics Vol. 4 (Elsevier, Amsterdam).

Shleifer, Andrei, and Robert W. Vishny, 1992, Liquidation values and debt capacity: A market

equilibrium approach, Journal of Finance 47, 1343-1365.

Titman, Sheridan, and Roberto Wessels, 1988, The determinants of capital structure choice, Jour nal of Finance 43, 1-19.

Welch, Ivo, 2004, Capital structure and stock returns, Journal of Political Economy 112, 106-131.

Williamson, Oliver E., 1988, Corporate finance and corporate governance, Journal of Finance 43, 567-592.



Documents

Paper for Article Critique