Running to Stand Still: Dimensions of Firm Growth

and Governance

Natarajan Balasubramanian Ravi DharwadkarSyracuse University Syracuse University

Jagadeesh SivadasanUniversity of Michigan

ABSTRACT

Using establishment-level micro data on Compustat firms, we develop a novel decomposition

of firm employment growth. Our decomposition reveals that on average, firms in this sample

rely more on growth from acquiring and founding new establishments than on growth of

existing establishments. Importantly, closures and sales of establishments make a significant

negative growth contribution that more than offsets the growth of existing establishments.

We then examine how the quality of governance, as measured by the strength of protection

from takeovers, is associated with the various components of firm growth. In line with the

literature, we find that poor governance is associated with lower firm growth. However, this

lower growth for poorly governed firms appears to be the result of both a lower growth rate

for existing and new establishments, and a significantly greater rate of establishment closures

and sales. We do not find any significant differences in the contribution of acquisitions to

growth. Together, our results suggest that well-governed firms are able to better manage

growth through efficient investments and thus have a more balanced growth portfolio than

poorly governed firms.

JEL Codes: L25, G34.

Keywords: Growth decomposition, empire-building, quiet life, establishment acquisitions,

establishment sales and closures.

Natarajan Balasubramanian: nabalasu@syr.edu; Ravi Dharwdkar: rdharwad@syr.edu; Jagadeesh Sivadasan:

jagadees@umich.edu. We thank Amber Anand, Jeff Harris, David Weinbaum, Johan Wiklund and participants at the Whitman

School Finance Brown Bag Series for their constructive comments. We are grateful to Clint Carter at the Michigan Census

Research Data Center and Arnie Reznek for their help with disclosure requests. The research in this document was conducted

while Natarajan Balasubramanian and Jagadeesh Sivadasan were Census Bureau research associates at the Cornell and Michigan

Census Research Data Centers, respectively. Research results and conclusions expressed are those of the authors, and do not

necessarily indicate concurrence by the Bureau of the Census. The results presented here have been screened to ensure that no

confidential data are revealed.

1 INTRODUCTION

Firm growth is a composite outcome arising from several underlying processes. At the

broadest level, firms can grow either internally or through acquisitions. Thus, two firms

with identical overall growth rates could have different blends of these two components.

Furthermore, internal growth is multidimensional — firms could grow by expanding their

current establishments or by starting new establishments. More importantly, firms could sell

or close their establishments, both of which affect their overall growth. So, for example, two

firms with identical overall growth rates of, say, 5% could arrive at that growth rate very

differently. One firm could generate 4% growth internally and acquire the remaining 1%,

while another firm could generate an internal growth of 1%, acquire 7%, and close and sell

establishments that reduce the growth rate by 3%. Thus, when firm growth is considered in

terms of overall annual growth, it masks the underlying complexities of the growth process,

and limits our understanding of how managers direct their efforts on the various underlying

growth dimensions. Furthermore, examining some of these elements (e.g., acquisitions or

plant closures and sales) while ignoring others provides us with only a partial picture of the

growth process.

In this paper, we use detailed establishment-level U.S. Census Bureau microdata for

Compustat firms and develop a novel granular decomposition of firm employment growth

that extends the traditional high level view of firm growth based on sales, assets, or

employment (e.g., Evans (1987); Demirguc-Kunt & Maksimovic (1998); Beck, Demirguc-

Kunt, & Maksimovic (2005); McLean, Zhang, & Zhao (2012)), and builds on research

about firm boundary management that uses establishment-level data (Maksimovic &

Phillips (2001), Maksimovic, Phillips & Prabhala (2008)). Drawing on the productivity

decomposition literature (e.g., Baily et al. (1992); Gilriches & Regev (1995)), we decompose

aggregate annual firm employment growth into two main components: internal growth of

existing establishments and net growth from entry and exit of establishments. The first

2

component, net internal growth in existing establishments, is further decomposed into three

subcomponents: (1) employment growth in establishments that were at least two years

old; (2) employment growth in establishments that were founded exactly one year ago; and

(3) employment growth in establishments that were acquired exactly one year ago, with

respect to the current time period. Our second component, net growth in employment from

entry and exit of establishments, is similarly decomposed into four subcomponents with

respect to the current time period: (1) due to establishments acquired in the last year; (2)

due to establishments that were founded in the last year; (3) due to establishments sold

in the last year; and, finally, (4) due to establishments closed in the last year (with the

last two components being negative by construction). This decomposition of employment

growth enables us to better examine the different patterns of growth — internal vs. external

growth, expansionary vs. contractionary activities, and growing new establishments vs.

closing old establishments — than more aggregate measures, and also capture differences in

the underlying growth dimensions of firms with similar overall growth rates.

We then use our growth decomposition to address an important gap in our

understanding of the firm growth-governance relationship. Thus far, most studies of firm

growth and governance have taken a high-level view of firm growth along the lines of Ijiri

& Simon’s (1967) firm-growth model — typically in terms of sales growth, asset growth,

or employment growth (e.g., Evans (1987); Demirguc-Kunt & Maksimovic (1998); Beck,

Demirguc-Kunt, & Maksimovic (2005); McLean, Zhang, & Zhao (2012)). For example, in

their seminal study, Gompers, Ishii, & Metrick (2003) use sales growth and provide evidence

of a negative correlation between firm growth and the quality of governance.1 Furthermore,

a majority of the studies have been in the context of mergers and acquisitions (M&A) (e.g.,

Billet, King, & Mauer, 2004; Hackbarth & Morellec, 2008; Moeller, Schlingemann, & Stulz,

2005) and have focused on empire building (e.g., mergers and acquisitions that increase

1Specifically, they found that the average year-on-year growth of a portfolio of poorly governed firms(G-index≥14) had a significantly negative growth rate, while a portfolio of well-governed firms (G-index≤5)was positive but insignificant.

3

firm size) by poorly governed managers. In line with this, Gompers et al. (2003) find that

ineffective governance, as measured by the extent of protection from takeovers, is associated

with increased M&A activity. Paul (2007) finds that certain board characteristics result in

corrective action in the context of bad acquisition bids, thereby constraining (possibly value-

destroying) firm growth. Gaspar, Massa, & Matos (2005) find that short-term institutional

owners are associated with not only worse abnormal returns, but also worse long-term

performance in the mergers. In sum, effective corporate governance appears to be associated

with growth through efficient mergers and acquisitions. However, beyond the M&A context,

to our knowledge, there are no in-depth studies that directly and comprehensively examine

the association between firm growth and governance.2 In particular, the relationship between

governance and internal growth has received very little attention, thus limiting a complete

understanding of the growth-governance relationship.

Some recent research has taken a more fine-grained view of the firm but has generally

focused on individual dimensions of firm growth, for example, the exit of establishments due

to divestitures or closures. Maksimovic and Phillips (2001) find that close to seven percent

of plants change ownership annually through mergers, acquisitions, and asset sales in peak

expansion years, and that the probability of asset sales is related to firm organization and

ex ante efficiency of buyers and sellers. In another study using establishment-level data,

Bertrand and Mullainathan (2003) find that increases in antitakeover protections reduce

managerial propensities to sell and close their existing plants, providing the managers with

2Note though that starting with Gompers et al (2003), a rich literature in finance and economics hasdocumented a positive relationship between market value and the quality of corporate governance (e.g.,Bebchuk & Cohen, 2005; Faleye, 2007; Bebchuk, Cohen, & Ferrell, 2009). Gompers et al. (2003) developedthe concept of a governance index (G-index) using twenty-four corporate governance provisions as a proxyfor the strength of shareholder rights, and demonstrated that an investment strategy that purchased sharesin firms with the strongest shareholder rights (i.e., least use of provisions) and sold shares in firms withthe weakest shareholder rights (i.e., most use of the provisions) earned abnormal returns of 8.5% per year.They also found that each one-point increase in their G-index was associated with a decrease in Tobin’s qof 4.3 percent during the ten-year study period. Similarly, Bebchuk, et al., (2009) investigated the relativeimportance of the various provisions and concluded that a subindex of the G-index, referred to as the E-index comprising the key provisions of staggered boards, limits to shareholder bylaw amendments, poisonpills, golden parachutes, and supermajority requirements, was associated in a similar manner with firmperformance.

4

an opportunity to lead a quiet life (in contrast to the empire building perspective).

In sum, while the literature has undertaken high-level examinations of firm growth

(e.g., sales growth or growth from mergers and acquisitions), and examined plant divestitures

and closures, to our knowledge, there are no studies that simultaneously study the various

dimensions of firm growth as part of a unified framework. This, in turn, implies that we are

largely unaware if, and how, governance is associated with these dimensions of growth.

For instance, both the empire-building (Gompers et al., 2003) and the quiet life views

(Bertrand & Mullainathan, 2003) would suggest that poorly governed firms would engage in

establishment closures to a lesser degree than well-governed firms. Thus far, it appears that

this hypothesis has not been tested.

In this paper, we overcome some of these limitations by using our firm-growth

decomposition to analyze how the various underlying dimensions of growth are associated

with the quality of corporate governance, as measured by the extent of protection from

takeovers. We then complement this analysis by directly testing the hypothesis that poor

governance is associated with inefficient investments. In particular, we focus on recently

acquired and newly founded establishments, and we investigate if their performance, as

measured by their growth and probability of exit, is associated with the quality of corporate

governance.

We use the US Census Bureau’s Longitudinal Business Database, which has data on

employment for all establishments in all (non-agricultural) sectors of the US economy, for

the period 1990 to 2005. Our annual employment growth decomposition reveals that firms in

this sample rely more on growth from net entry (mean of 3.57% per year) than on growth of

existing establishments (mean of 1.53% per year). Importantly, there is evidence of significant

churning of establishments in this sample. The mean net entry of 3.57% is the result of

increases of 4.89% through acquisitions and 3.25% through new establishment openings,

5

and declines of 2.22% through establishment sales and 2.66% through closures.3 Our three-

year (and five-year) growth patterns are broadly similar, and indicate that growth through

new establishments and establishment closures are the second (first) and third (third) most

significant components of growth, something that has not been considered in detail by extant

research.

Turning to the growth-governance association, we examine how the various components

and subcomponents are associated with two measures of corporate governance: a broad G-

index, based on Gompers et al. (2003), and a narrower E-index, based on Bebchuk & Cohen

(2005). Our analyses show that under better governance, growth is not only higher but also

more balanced. The unconditional average growth rate for firms in the lowest quartile of G-

index (i.e., the best-governed firms) is 7.91% per year, while those in the highest quartile grow

at only 2.69% per year. Interestingly, only a very small part of this difference between poorly

governed and well-governed firms is attributable to growth through acquisitions (4.90% vs.

4.95%, respectively) and founding of new establishments (3.00% vs. 3.49%, respectively). A

large share of the difference arises from the inability of poorly governed firms to grow their

existing establishments (0.36% vs. 2.78%, respectively) and higher levels of establishment

exit among such firms (-5.49% vs. -4.09%, respectively).

We repeat the analyses after controlling for several known determinants of firm growth

and industry-year variations at the SIC three-digit level. In particular, we adopt a more

conservative version of the pooled specifications adopted in Bebchuk et al. (2009) (Table IV)

with growth and its underlying components and subcomponents as the dependent variables,

and firm age, Delaware incorporation, return on assets (ROA), leverage, capital expenditure

(capex) to sales, research and development (R&D) to sales, and SIC-3-year fixed effects

as other controls. As before, we find that better governed firms grow faster than poorly

governed firms. Unit increases in the G-index and E-index are associated with 0.2% and 1.1%

3Note that the individual components do not exactly add up to the total due to winsorizing of growthrates at the top and bottom 1%. For more details, refer to the Results section.

6

reductions in overall annual firm growth, respectively. We also find that poor governance is

associated with a lower contribution on each of the two major components of firm growth

with the reduction in the net entry component being 1.5 to 2 times that of the net existing

growth component. An examination of the individual subcomponents of net entry reveals

very interesting and significant differences. We find that poorly governed firms not only

have a lower contribution to growth from setting up new establishments but also have a

greater (negative) contribution from closure and sale of establishments. In particular, a

one-standard deviation increase in G-index (E-index) is associated with a 0.23% (0.17%)

greater negative contribution to growth from establishment sales, and a 0.16% (0.07%)

greater negative contribution to growth from establishment closures. As before, we do not

find any significant differences in the growth contributions from acquisitions.

Next, we examine effectiveness of investments. We find that the performance of

newly acquired and recently founded establishments is systematically and significantly

positively correlated with the quality of corporate governance. Regressions of establishment

performance on indices of governance show that a unit increase in G-index is associated with

a 1.2% reduction in the one-year growth rate of these establishments and a 1.9% higher

probability of the establishment being sold or closed within three years of its founding or

acquisition. The corresponding numbers for a unit increase in E-index are a 1.6% reduction

in the one-year growth rate and a 2.1% higher probability of establishment exit.

Our results are robust to alternative measures of growth including the use of three-

and five-year growth rates. Our establishment-level results are also robust to including

very conservative state-industry-year fixed effects, which control for all time-varying changes

within an industry in a given state. Notwithstanding our fine-grained high-dimensional

fixed effects, it is still possible that the indices of governance are correlated with a firm-

level unobserved variable that is also correlated with growth. Since the governance indices

change very little within a firm over the sample time period, the use of any firm fixed

7

effects specification including the use of difference- or system-GMM (generalized method of

moments) estimators would not be very meaningful.4 To address this concern, we performed

a deeper analysis of acquired establishments. In particular, we examined if firms with

poor governance acquired slow-growing establishments. Establishment-level regressions of

an establishment’s past growth rate on the measures of governance showed statistically

insignificant, near-zero coefficients on both governance measures. Thus, results from this

implicit difference-in-difference analysis strongly suggest that the observed decline in growth

rate among establishments acquired by poorly governed firms occurs after the acquisition,

and thus is likely caused by the poorly governed firm. This test also rules out the possibility

that governance is associated with differences in the selection of acquisition opportunities

based on prior growth trends. Nonetheless, this check focuses only on acquired establishment,

and hence, our other results are better interpreted as how well-governed firms differ from

poorly governed firms than as the causal effects of good governance.

Together, our results and methods make two important contributions to the literature.

First, we offer a novel decomposition of firm growth that combines firm- and establishment

level data and provides a considerably more granular decomposition of firm growth than

examined in prior studies. This decomposition not only provides a better understanding

of overall firm growth but also sheds light on growth-related managerial activities that are

masked when growth is studied at the aggregate level. For instance, one of our salient findings

— that selling and closing establishments has a significant impact on overall firm growth

rate — has generally been ignored by prior research at the firm level. Second, our results

strongly suggest that superior management of firm growth through efficient investments is

an important channel through which better-governed firms create value for shareholders.

In particular, it appears that better governance is associated with greater efficiency of

4However, this also means that the level of governance faced by managers was largely static during thistime period and, accordingly, can potentially be thought of as an institutional constraint faced by managersrather than as a decision variable influenced by managers. Of course, this does not rule out the possibility thatthe factors that originally influenced the choice of governance before the sample time period are correlatedwith growth during the study.

8

investments, which is then reflected in more balanced and higher firm growth. Thus, our

results are somewhat different from the two contrasting implications of governance in the

extant literature — the empire-building found in Gompers et al. (2003) and the quiet life

in Bertrand & Mullainathan (2003). In particular, we do not find any strong evidence that

poorly governed firms grow more through acquisitions, as the empire-building view would

suggest. In contrast to the quiet-life view, we find that establishments belonging to poorly

governed firms have a greater probability of exit. Thus, at least with respect to annual

employment growth, managers in poorly governed firms appear to be “running to stand

still” – their growth performance from acquisitions is more than offset by relatively higher

rate of establishment exits. Finally, our analysis of newly acquired and established plants

uses direct measures of investment performance (growth and exit), which provide a good

complement to prior studies that measure investment performance indirectly (for example,

using abnormal returns).

The paper is organized as follows. In section 2, we derive a decomposition of overall

firm growth. In section 3, we describe our data. In section 4, we present our results. Section

5 discusses and concludes.

2 DECOMPOSING FIRM GROWTH

Consider a firm with several establishments. We can write the firm’s total employment

in period t as:

Firm Employmentt ≡ Et =∑i∈St

Eit

where St denotes the set of establishments in the firm in period t, and Eit represents the

employment in establishment i in period t.

The change in the employment of the firm from period t− k to period t is then given

9

by: ∆Et = Et − Et−k =∑

i∈StEit −

∑i∈St−k

Eit−k

As the first step, we can decompose the set of establishments in period t as follows:

St = Sexisting + Sadditions

where Sexisting is the set of establishments that survived from t−k to t, and Sadditions are new

establishments that entered the firm between t − k and t, including in period t. Similarly,

we can decompose the set of establishments in period t− k in the following way:

St−k = Sexisting + Sexit

where Sexisting, as before, comprises establishments that survived from t−k to t, and Sexit is

the set of establishments that exited from the firm between t− k and t, including in period

t− k.

Together, we now have the following decomposition:

∆Et ≡ Et − Et−k =∑i∈St

Eit −∑

i∈St−k

Eit−k (1)

=∑

i∈Sexisting

(Eit − Eit−k) +

( ∑i∈Sadditions

Eit −∑

i∈Sexit

Eit−k

)(2)

This equation represents the broadest level of decomposition used in our study. The first

term in Equation 2 is the change in employment at the surviving establishments, while the

second term denotes the net increase in employment arising from net entry of establishments.

Now, we can further decompose the first term of Equation 2 into the following groups:

Sexisting = Sold + Snewlast + Sacqlast (3)

where Sold is the set of establishments belonging to the firm that were at least one year old

10

in year t − k, Snewlast is the set of establishments that were established in year t − k, and

Sacqlast is the set of establishments that were acquired by the firm in year t− k.

Similarly, we can decompose the newly added establishments as follows:

Sadditions = Snew + Sacq (4)

where Snew is the set of establishments that were established between t− k and t, and Sacq

is the set of establishments that were acquired between t − k and t. Repeating the above

decomposition for exiting establishments, we get:

Sexit = Sclosed + Ssold (5)

where Sclosed is the set of establishments that were closed between t− k to t, and Ssold is the

set of establishments that were sold between t− k to t.

Using the above decompositions and denoting Eit−Eit−k as ∆Eit, Equation 2 can now

be fully expanded as:

∆Et =∑i∈Sold

∆Eit +∑

i∈Snewlast

∆Eit +∑

i∈Sacqlast

∆Eit

+∑

i∈Snew

∆Eit +∑i∈Sacq

∆Eit

−∑

i∈Ssold

∆Eit −∑

i∈Sclosed

∆Eit (6)

11

Dividing throughout by Et−k, we get the complete growth decomposition:

∆Et

Et−k=

∑i∈Sold

∆Ei

Et−k+

∑i∈Snewlast

∆Ei

Et−k+

∑i∈Sacqlast

∆Ei

Et−k

+

∑i∈Snew

∆Ei

Et−k+

∑i∈Sacq

∆Ei

Et−k

−∑

i∈Ssold∆Ei

Et−k−∑

i∈Sclosed∆Ei

Et−k(7)

To summarize, Equation 7 decomposes firm growth into seven components: (i) growth of old

establishments, (ii) growth of newly acquired establishments, (iii) growth of newly established

establishments, (iv) growth by simply acquiring establishments, (v) growth by setting up new

establishments, (vi) optimizing size by selling some establishments, and (vii) optimizing size

by closing some establishments. This decomposition forms the basis for the first part of our

empirical work.

3 DATA, VARIABLES, EMPIRICS

3.1 Data

The main source of data for this study was the Longitudinal Business Database (LBD) from

1990 to 2005. This dataset, developed and maintained by the U.S. Census Bureau, contains

employment and payroll information on all non-agricultural establishments in the U.S. that

have at least one employee. These data also provide information on ownership, as well as on

industry and geography. No information on sales or assets is included. A concordance file

was used to link these data to Compustat and CRSP, which provided firm-level information

on firm age, state of incorporation, profitability, leverage, capex, and R&D. We collected

governance data from the RiskMetrics (formerly IRRC) database for firms. The RiskMetrics

governance data are not available for all years. We therefore follow Bebchuk and Cohen

(2005) to fill in the missing years. They assume that the governance provisions in any given

12

year are the same as those of the preceding year. For instance, the governance provisions

in 1996 are assumed to be the same as those in 1995. In addition, following Bebchuk and

Cohen (2005), we eliminate firms with a dual-class structure. Following Zhao and Chen

(2008), we also eliminate observations in the financial services and insurance services (SIC

codes 6000-6999). These procedures result in a sample of 16,966 firm-year observations,

corresponding to a total of 2,065,032 establishment-year observations. Descriptive statistics

for the firm-level and establishment-level samples are provided in Table 1.

3.2 Key Variables

The main dependent variable of interest was firm growth. Firm Growth was defined as

the change in the total number of employees from last year divided by the total number of

employees in the last year ( ∆EEt−1

).5 To eliminate the influence of outliers, we winsorsized the

growth variable at 1% on both tails. The two main measures of governance were G-index and

E-index. G-index was defined as the GIM Index (Gompers et al., 2003), which was based

on the presence or absence of 24 corporate governance provisions (categorized as delay,

protection, voting, and other). Higher values of this index represent greater protection from

the corporate control market. The second measure of governance was E-index, a strict subset

of G-index, and a narrower measure of protection. E-index was defined as the Entrenchment

Index proposed by Bebchuk et al. (2009). That index is based on the presence or absence

of six provisions: staggered boards, limits to shareholder bylaw amendments, poison pills,

golden parachutes, supermajority requirements for mergers, and charter amendments. As

with G-index, the higher the E-index, the greater the degree of protection. In line with

Bebchuk et al. (2009), when using E-index in regressions, we included a variable other

provisions index, which was defined as the difference between G-index and E-index.

5As robustness checks, we varied the time period to three and five years. These results are discussed inthe robustness checks section.

13

3.3 Empirics

The first part of our analysis, at the firm level, examined whether the quality of governance

is associated with differences in the composition of firm growth. The second part of the

analysis focused on investment performance (as measured by the performance of acquired

and newly founded establishments) and its relationship to the quality of governance.

3.4 Composition of Firm Growth and Governance

To examine this question, we broadly followed the specification in Bebchuk et al. (2009), with

one regression for each component of firm growth. Specifically, we estimated the following

pooled OLS specification:

∑i∈Sx

∆Ei

Ejt−1

= α1.Gjt + α2.Ojt + α3.yjt−1 + Zjt + νmt + εjt (8)

where x ∈ {old, newlast, acqlast, new, acq, sold, closed}; i denotes establishments belonging

to firm j; G is a measure of governance, either G-index or E-index; O is the other provisions

index included when E-index is the dependent variable; yjt−1 is log lagged employment; Zjt is

the vector of firm-year controls described below; and νmt are SIC-3-year fixed effects.6 Note

that these fixed effects are considerably more conservative than the SIC-2 digit industry-

adjusted dependent variables used in Gompers et al. (2003) and other studies, and control for

all time varying differences across SIC-3 digit industries. Turning to Zjt, following Bebchuk et

al. (2009), we included log company age (from CRSP), a dummy for Delaware incorporation,

ROA, leverage, capex to sales and R&D to sales as other controls.

6The SIC-3 digit industry for a firm was defined as the modal SIC-3 code based on employment.

14

3.5 Investment Performance and Governance

To examine if investment performance is associated with the quality of governance, we used

the following two pooled OLS specifications:

∆Eij

Eijt0−k= α1.Gjt0 + α2.Ojt0 + α3.yijt0 + Zjt0 + νmt0 + εijt0 (9)

where the left-hand side represents the k-year growth of establishment i newly founded or

acquired by firm j in year t0. Thus, the sample consists of one observation per establishment

with all control variables pertaining to the year of acquisition or founding.

The second specification is a linear probability model with high-dimensional fixed

effects, and it examines the probability of establishment exit, defined as the establishment

being closed or sold within a certain period of time. As with the previous specification, we

limit the sample to one observation per establishment, with all control variables based on

the year of acquisition or founding.

Dkij = α1.Gjt0 + α2.Ojt0 + α3.yijt0 + Zjt0 + νmt0 + εijt0 (10)

where Dkij is 1 if the establishment was sold or closed within the first k years of founding

or acquisition and 0 otherwise, and the other terms are as defined earlier.

4 RESULTS

4.1 Overall Firm Growth

Table 2 presents a decomposition of the mean one-year growth in firm employment into

the two broad components — change in employment from existing establishments and

15

change in employment from net entry of new and acquired establishments — and into the

seven subcomponents discussed earlier. In this sample of firms, net entry of establishments

accounts for a major share of firm employment growth (3.57% of 5.42%); existing growth

at existing establishments forms a considerably smaller portion (1.53%).7 Within existing

establishments, on average, establishments that were new last year show the highest growth

(0.98% per year), while establishments that were acquired a year ago exhibit no growth. The

much larger net entry term hides an even larger churning of establishments within the firm.

The mean firm growth in employment from net entry is 3.57%, but this is the result of a gross

addition of about 8.14% from founding new establishments (3.25%) and acquisitions (4.89%)

and a reduction of about 4.88% from closure (-2.22%) and sales (-2.66%) of establishments.

In sum, growth in this sample of firms appears to be the result of not just acquisitions or

founding of new establishments but also very active turnover of establishments. We now

examine if the composition of growth varies with the quality of governance.

4.2 Firm Growth and Governance

Table 3 provides a breakdown of the decomposition presented in Table 2 by the quality of

governance. Focusing on net firm growth (the last row), poorly governed firms exhibit a

small increase of 2.69%, compared to a net increase of 7.91% for the best-governed firms.

This large difference is consistent with the finding in Gompers et al. (2003) that firms with

poor governance have lower growth rates. The patterns are similar when we focus on the two

major components. Poorly governed firms show a significantly smaller increase in growth

from the two main components: growth from existing establishments and growth from net

entry. However, as before, the broad patterns conceal interesting and important differences.

First, well-governed firms have a more balanced growth portfolio of existing establishment

7We also performed similar analyses with the 3-year and 5-year growth; growth from existingestablishments was still much smaller than growth from net entry. The mean growth from existingestabishments was 6.1% and 13.4% for 3-year and 5-year growth. The corresponding mean growth fromnet entry were 19.7% and 48.5%. See the attached (online) Appendix for details.

16

growth and net entry, while most of the growth of poorly governed firms comes from net entry.

Net entry accounts approximately for 2.37% of the 2.69% mean total growth among poorly

governed firms, but this component accounts (approximately) for only 4.70% of the 7.91%

growth among well-governed firms. Second, poorly governed firms do not have a lower net

entry component because they have lower levels of entry of establishments. In fact, the gross

additions are similar. The contribution from acquired establishments is virtually identical for

both these types of firms (4.90% and 4.95%), while the contribution from new establishments

is only about 0.5% lower for poorly governed firms. In contrast, the negative contribution

to growth from sales and closure is almost 1.65% higher for poorly governed firms (-5.74%

vs. 4.09%). Thus, poorly governed firms have a lower growth rate because of their greater

rate of churning of establishments than well-governed firms.8

We test these results more formally in Tables 4-6. Table 4 presents the results of

estimating Equation 8 with one-year firm growth – that is, Et−Et−1

Et−1— as the dependent

variable. Recall that all these regressions include SIC3-year fixed effects, thus eliminating

any confounding effects of time-varying changes at the industry level. The results in Table

3 are in line with the findings in Gompers et al. (2003) and indicate that the quality of

governance is positively associated with firm growth.9 The coefficients on G-index and E-

index are negative and significant throughout. A unit increase in the G-index (at the mean

G-index) is associated with a 0.2% reduction in firm growth. A unit increase in E-index is

associated with a 1.1% reduction in firm growth.10 The coefficients on the control variables

are generally in the expected direction. Profitability, leverage, and capital expenditure are

positively associated with growth, while prior size and firm age are negatively related to

growth.

8We also performed these analyses with the three-year and five-year changes, and found similar patterns.9Note that both measures of governance are constructed such that higher values of the measure indicate

poorer quality of governance.10This negative association persisted with three-year and five-year growth rates, but the coefficients on G-

index were insignificant. The coefficients on G-index in these regressions was about -0.3%. A unit increase inthe E-index was associated with a 4.5% and 7.5% reduction in three and five-year firm growth, respectively.

17

Turning to the two main components of firm growth - growth from existing

establishments and growth from net entry of new and acquired establishments, we reestimate

the regressions in Table 5 with∆Eexisting

Et−1and ∆Enetentry

Et−1as the dependent variables. The

results are similar to those in Table 4. Both these components of firm growth are negatively

associated with the quality of governance. A unit increase in E-index is negatively associated

with both components: 0.4% reduction in growth from existing establishments and 0.6%

reduction in growth from net entry. The coefficients on most of the controls have similar

signs as before, although the coefficients on leverage and capex show interesting differences.

Greater leverage is associated with greater net entry but not with greater growth from

existing establishments. Not surprisingly, capex is strongly associated with growth from

existing establishments but not with growth from net entry. Profitability appears to have a

larger impact on growth from existing establishments than on growth from net entry.

Table 6 presents a further disaggregation of the main components of growth and their

link with governance. Of the three subcomponents of growth in existing establishments,

growth of old establishments and growth of establishments founded last year are positively

associated with the quality of governance (Panel A). As in Table 2, the growth of acquired

establishments does not show any significant association with the quality of governance.

Of the four subcomponents of net entry, growth through founding of new establishments is

positively associated with the quality of governance. In line with the findings in Table 3,

well-governed firms appear to have a significantly smaller negative contribution to growth

from sale and closure of establishments (Panel B).

Together, the results in Tables 4-6 support the thesis that firm growth is positively

related to the quality of governance. More importantly, governance appears to be associated

differently with different subcomponents of growth. As seen by their higher contribution

from existing growth and lower sales and closures, better-governed firms appear to be more

capable of managing growth more effectively than their poorly governed counterparts.

18

4.3 Investment Performance and Governance

The analysis, thus far, has been on growth at the firm level. Since growth is ultimately

determined by the performance of a firm’s investments, we now focus on new and acquired

establishments and attempt to understand the impact of governance at the establishment

level. We use two measures of investment performance: establishment growth and

establishment exit (due to sales or closure). Table 7 presents the results of investment

performance regressed on measures of governance. It is clear from the table that new and

acquired establishments of poorly governed firms tend to have a slower growth rate than those

of well-governed firms. The one-year growth is lower by 1.2-1.6% for every unit increase in

the governance measures, while the three-year growth rate is lower by 1.4-2.9%. The five-year

growth regressions also exhibit negative, but statistically insignificant, coefficients.11 Turning

to exit, Panel B shows that establishments of poorly governed firms are significantly more

likely to exit. The probability of exit within three years after they are established or acquired

is higher by 1.9-2.1% for every unit increase in the governance measures; the corresponding

figure for the five-year exit rate is 1.3-2.1%. Together, these results strongly indicate that

the investment performance of better-governed firms is superior to that of poorly governed

firms.

We then reestimate the regressions in Table 7 after limiting the sample to new

establishments, and present the results in Table 8. The results are similar to those in Table

7, suggesting that well-governed firms invest in (or are able to create) better-performing

establishments. Focusing on the performance of acquired establishments, the results on

establishment growth are broadly similar to those in Tables 7 and 8. Based on the broader

measure of governance, G-index, it appears that the growth rate of acquired establishments

is lower at poorly governed firms. The lower growth rate appears to persist well into the

fifth year after acquisition. Based on the narrower entrenchment measure, the correlation

11One possible explanation for is that the lowest-performing establishments have exited by this time. Thisis consistent with the evidence in Panel B.

19

between the quality of governance and growth rate of acquired establishments appears to be

insignificant. The results on exit are similar to those in Tables 6 and 7 for both measures,

and they suggest that the tendency of poorly governed firms to close their establishments

extends to acquired establishments as well.

Together, the results in Tables 7-9 strongly suggest that controlling for industry-year

factors, establishment size, and other firm characteristics such as firm age and profitability,

well-governed firms exhibit better investment performance than poorly governed firms.

4.4 Robustness Checks

We performed several checks to ensure the robustness of our results. We discuss these below.

4.4.1 Alternative Growth Measures

. We repeated the analyses in Tables 2-6 with with three-year and five-year firm growth and

found results similar to those with one-year growth (results presented in Appendix tables

A1-A8). Specifically, growth from net entry continued to be the dominant way of growing

a firm compared to the growth of existing establishments (19.75% vs. 6.13% respectively,

with three-year growth rates, and 48.51% vs. 13.42%, respectively, with five-year growth

rates). Entry of new establishments was also a significant contributor to firm growth in

addition to acquisitions. Firms also exhibited significant churning of establishments (negative

contributions of 14.05% and 21.16% from sales and closures to three- and five-year growth

rates, respectively). The unconditional average growth rates were significantly higher for

firms in the lowest quartile of G-index than for firms in the highest quartile (three-year

growth rate: 45.57% vs. 13.62%; five-year growth rate: 112.12% vs. 29.41%). As with

one-year growth rates, better governed firms had more balanced growth with a lower ratio

of growth from net entry to total growth. They also had significantly fewer closures and

20

sales (-11.48% vs. -16.47% for three-year growth rates and -17.72% vs. -24.05% for five-year

growth rates). Further, as in the baseline analysis, poorly governed firms do not appear to

have a greater contribution from acquisitions.

Our analyses with the full set of controls show results that are qualitatively similar to

the baseline results presented in Tables 4-6. These results are presented in Tables A5-A8.

While G-index is still negatively related to overall growth, it is statistically insignificant.

E-index, on the other hand, continues to be strongly and negatively correlated with overall

growth. The relation between the governance indices and the individual components and

subcomponents of growth is generally negative, although the results are considerably stronger

when E-index is used as a measure of governance. We continue to find little evidence on the

acquisition-based dimension of growth. Together, these results are in line with our baseline

results: well-governed firms are better able to grow their existing establishments and have a

lower (negative) contribution from establishment sales and closures.

4.4.2 Role of Opportunity Selection

We then checked if firms with poor governance acquire slow-growing establishments. If this

were true, then the main problem with poorly governed firms would be more an inability to

choose appropriate growth opportunities rather than problems with managing growth. Table

10 presents the results from establishment-level regressions of an establishment’s past growth

rate on the measures of governance. All the coefficients on both governance measures are

very close to zero and statistically insignificant. Thus, it appears that the observed decline

in growth rate occurs after the acquisition of the establishment by poorly governed firms,

indicating that the problem with poor governance likely lies in the process of managing

growth. Also note that since we have industry-year fixed effects, our analysis has an implicit

difference-in-difference structure, with the changes in establishment performance of poorly

governed firms compared to changes in establishment performance of well-governed firms.

21

Similarly, we examined whether new establishments were started at a larger scale by

poorly governed firms. This addresses the possibility that establishments of poorly governed

firms may exhibit lower growth since they have a smaller gap between their initial and final

size. The regression coefficients (not presented) were all small and insignificant, suggesting

that the slower growth rate of new establishments is not due to a difference in the initial

scale of the establishment.

4.4.3 Role of Omitted Input Factors and Productivity

We used employment growth as our measure of growth, primarily due to data constraints.

While correlated with other measures of growth, it does omit other factor inputs, notably

capital.12 This could lead to misleading conclusions if the governance indices were

systematically correlated with the use of capital. For instance, if poorly governed firms

tend to expand their capital stock relatively more than employment, then our analysis will

spuriously suggest a negative correlation between governance and growth. To rule this out,

we performed an in-depth analysis of the relation between productivity and governance. We

restricted the analysis to manufacturing because the techniques for estimating productivity

in this domain are well established. Moreover, we had access to data on capital and other

inputs only for manufacturing. We used two commonly adopted measures of productivity —

labor productivity and total factor productivity (TFP, calculated as the Solow residual) and

estimated the same types of regressions as in Table 4. The results, presented in Table

11, do not show any significant association between governance and productivity. The

coefficients on the governance indices in the labor productivity regressions are virtually

zero and statistically insignificant. Thus, it is unlikely that poorly governed firms are

substituting labor with capital in their growth processes. Similarly, the coefficients in the

TFP regressions are also virtually zero. This largely rules out the effect of variables such

12Based on Compustat data, the correlation between employment growth and asset growth was about78%.

22

as unobserved differences in human capital, skill intensity, patent intensity, adoption of

information technology, etc. In particular, these variables strongly influence TFP, and given

that we do not find any differences on TFP, it is unlikely that those variables are the primary

driver of our results.

Our productivity analysis also allays a broader theoretical concern. In models of

industry equilibrium such as Jovanovic (1982) and Hopenhayn (1992), the ultimate size of a

firm is determined by its productivity. Thus, in these models, differences in growth rates of

firms could possibly be attributed to differences (or changes) in productivity. Put differently,

some firms grow faster than others because they use their resources more efficiently, which

provides them a cost (or differentiation) advantage over their competitors. As a consequence,

in these models, productivity also determines profitability, and hence, the market value of

the firm. Furthermore, these models also imply that more productive firms (and in turn,

their establishments) are less likely to exit. Extending these arguments, it is conceivable that

the lower growth of poorly governed firms and the higher exit rate of their establishments

are the results of their lower productivity. That we find near-zero coefficients in productivity

regressions provides greater confidence that the observed variations in growth and exit rates

are not likely to be the result of productivity differences.

As additional robustness checks, primarily to rule out endogeneity in the estimation

of productivity, we used four other measures of productivity and obtained the same results.

These results are presented in Tables A11.

4.4.4 Broader Measure of Establishment Exit

Our baseline analysis considers only the closure of establishments as exits. As an additional

robustness check, we expanded our definition of establishment exit to include those

establishments that were sold to other firms and repeated the exit analyses in Tables 7-

9. The results, presented in Table A9, are similar to the baseline results and confirm that

23

governance is negatively correlated with this broader definition of exit.

4.4.5 Additional Analysis of Acquisitions

In general, we do not find any significant association between acquisition-based growth and

governance. This result is different from Gompers et al. (2003), which finds a positive

correlation between the number of acquisitions and poor governance. To ensure that our

results are not different solely due to the choice of growth measures, we performed some

additional analyses. In particular, we examined whether the share of acquired (and new)

establishments to the total number of establishments was associated with governance. The

results are presented in Table A10. The correlation between the share of acquired to total

establishments is weak, with only E-index exhibiting a mild positive correlation. In contrast,

the share of new establishments shows a much stronger (and negative) correlation with the

governance indices. Thus, this is in line with our baseline results that poorly governed firms

tend to have a lower growth contribution from new establishments.

4.4.6 Other Checks

We repeated the analyses in Tables 7-10 with state-industry-year fixed effects instead of

state-year fixed effects, thus controlling for time-varying changes within an industry in a

state. Our results remained robust to this inclusion. We also repeated the analyses using

G-index and E-index in the initial year (1990) and found similar results.

5 DISCUSSION AND CONCLUSION

At the broadest level, this paper is a commentary on firm growth. Our results suggest that it

is important to understand overall firm growth as a composite process resulting from several

underlying processes. This is in contrast to most of the literature, where firm growth has

24

been considered broadly at the firm level or in terms of M&A events that result in firm

growth.

Our novel decomposition of firm growth reveals that on average, Compustat firms

rely more on growth through acquisitions and starting new establishments (net entry) than

through growing existing establishments (existing growth). The unconditional contribution

from the former component is two to three times larger than from the latter. However, there

is only a small difference in the contribution from growth through acquisitions and growth

from starting new establishments (4.89% vs. 3.25%, respectively, with one-year growth,

and 28.31% vs. 38.16%, respectively, with five-year growth). Thus, while previous research

has focused on acquisition-based growth, it appears that starting new establishments is

also a significant driver of overall firm growth. Another interesting finding is that selling

and closing their current establishments account for a substantial reduction in overall firm

growth (-4.88% and -21.16% with one- and five-year growth rates, respectively). Although

it is known that firms close and sell establishments, our decomposition provides the first

comprehensive look at the impact of these activities on firm growth.

At the next level, this paper contributes new empirical evidence on how the growth

of a firm is associated with its quality of governance. When these growth-composition data

are partitioned across governance quartiles (Table 3), it appears that the differences among

these groups are not driven by the two growth components related to acquisitions: growth

in recently acquired establishments and growth from acquiring establishments. Compared

to other components, the differences among firms with poor, medium, and good governance

on these two components are minimal (e.g., with one-year growth rates, the sum of these

two components is 5.00%, 4.85%, and 5.02%, respectively). This is an interesting finding, as

a large literature suggests that managers indulge in empire-building under poor governance

regimes. In fact, with longer periods, we find that well-governed firms have a greater

contribution from these two components, thus contradicting the empire-building argument.

25

Instead, we find that poorly governed firms exhibit limited growth their old establishments

(-0.36%, 0.14%, and 0.83% over one, three and five years respectively), while well-governed

firms appear to be able to generate modest growth from these establishments (1.44%, 3.76%,

and 3.22%). In fact, the growth from just old establishments for well-governed firms is

comparable (and with one-year growth rates even higher than) to the total growth of poorly

governed firms. Similar conclusions can be drawn from our subsequent analysis, which

controls for other potential factors, including industry-year effects. Finally, the most salient

result of our study is the link between governance and establishment exit. We find that

poorly governed firms have a much higher negative growth contribution from closing and

selling establishments. This is consistent with the hypothesis that such firms make inefficient

investments that do not result in growth. Further support for this argument is found in the

establishment-level regressions; unit increases in G-index and E-index are associated with a

higher probability of establishment exit (1.3% and 2.1%, respectively) within five years of

founding or acquisition.13

Together, our results suggest that poorly governed firms are able to grow neither

their existing establishments nor their new establishments. At the same time, they are

involved in more selling or closing of establishments. Thus, our findings suggest that the

inability to manage growth effectively may lie at the heart of why poor governance may

reduce market value. Overall, well-governed firms appear to have a more balanced growth

portfolio in contrast to poorly governed firms, which rely much more on acquisitions and

new establishments to stimulate growth.

Our findings relate most directly to Gompers et al. (2003) and the related literature.

In particular, they strongly support the conjecture in Gompers et al. (2003) that poorly

13Furthermore, our results also highlight an interesting difference between the growth of new and acquiredestablishments. Among new establishments, the negative association of governance quality with growth isimmediate; the one-year growth rate and the one-year exit rate are strongly associated with the qualityof governance. The effect is not so pronounced in the case of acquired establishments, with only a smallnegative association between the quality of governance and one-year performance measures.

26

governed firms make inefficient investments.14 Our study also speaks to the broader

literature on empire-building, which argues that poor governance results in increases in

capital expenditures and merger and acquisition activities (Baumol, 1959; Williamson, 1964;

Gompers et al., 2003). These studies argue that such investments, while expanding the

size and scope of the firm, result in lower returns and reduced firm value because they

are inefficient. Our findings, particularly on establishment growth and exit, support this

argument. Nonetheless, not all our results are consistent with the premise of empire-building.

First, our results do not suggest that poorly governed firms are associated with higher

levels of acquisition-based growth. Second, poorly governed firms are also associated with

higher levels of selling and closing establishments. This seems contrary to that managers are

interested in “pure” empire-building when faced with limited threats of takeovers.

Our findings are also somewhat different from those of Bertrand & Mullainathan (2003)

who find that increases in anti-takeover protections provide managers with an opportunity to

enjoy the quiet life alluded to by Hicks (1935). Our data, albeit in a slightly different context,

reveal that managers in poorly governed firms are perhaps as active as their counterparts

in well-governed firms in acquiring and starting new establishments, and they are certainly

more active in closing and selling establishments. Thus, their lower growth is conceivably

not because of their lack of effort, but likely in spite of their higher levels of activity.

Before we turn to the limitations of this study, it is important to reiterate that some of

our results should be primarily interpreted as to how well-governed firms differ from poorly

governed firms, rather than as the causal effects of good governance. Even though our

specifications include fine-grained, high-dimensional fixed effects and several time-varying

firm-level correlates of firm growth, the use of pooled OLS specifications means that it is still

possible that the indices of governance are correlated with a firm-level unobserved variable

that is also correlated with growth. Nonetheless, our analysis of acquired establishments

14Specifically, they used this conjecture to reconcile the findings of higher acquisition activity and lowersales growth among such firms.

27

provides some confidence in the direction of causality. In particular, it does not appear that

poorly governed firms buy slower-growing establishments. Thus, this implicit difference-

in-difference analysis strongly suggests that the observed decline in growth rate among

establishments acquired by poorly governed firms occurs after the acquisition and is thus

likely caused by the poor firm governance.

Beyond the cautionary note on causality, our study has some limitations that deserve

mention. First, profit growth, rather than employment growth, is the immediate determinant

of firm value. Since we did not have data on profitability, assets, or revenues of individual

establishments, we used employment growth. Even though employment growth is correlated

with firm value and growth in other variables such as profits, revenues, and assets,(e.g.,

in many theoretical models of firm value such as Jovanovic (1982) and Hopenhayn (1992),

profitability is a direct function of firm employment or size), there could be within-industry

differences in the choice of capital-labor ratios that add noise to these correlations.15

If these differences are systematically correlated with governance, then our results on

the employment growth-governance association cannot be extended to the profit growth-

governance relationship.16 Second, our data do not include any data on global operations of

these firms. Thus, there could be differences in the mix of foreign and domestic growth that

are not examined in our study.

These limitations also suggest potential extensions. Assessing how these individual

components of growth are associated with other measures of governance would be an

interesting study. More broadly, relating the composition of growth to other firm

characteristics beyond governance is also likely to yield potentially useful insights. Another

potential direction of study is to examine the human capital aspects of firm growth. In

15Note that the use of industry-year fixed effects addresses any potential concerns with industry-widechanges in capital-labor ratios.

16As a limited robustness check, we regressed the log of market value on firm employment growth andthe governance indices along with other controls from Table 4. The coefficient on employment growth wasalways strongly positive suggesting that employment growth is correlated with firm value in our sample evenafter controlling for governance.

28

particular, data on employee characteristics (e.g., age, experience, skills, etc.) for employees

at the various establishments may be valuable in analyzing the underpinnings of firm growth.

To conclude, the link between firm growth and governance has largely been a black

box. This study is a first step to opening that box.

29

REFERENCES

Baily, M. N., Hulten, C., Campbell, D., Bresnahan, T., & Caves, R. E., 1992,

Productivity dynamics in manufacturing plants, Brookings papers on economic activity:

Microeconomics, 187-267.

Baumol, W. J., 1959, Business Behaviour, Value and Growth, New York, Macmillan.

Bebchuk, L. A., and Cohen A., 2005, The costs of entrenched boards, Journal of

Financial Economics, 78(2), 409-433.

Bebchuk, L., Cohen, A., & Ferrell, A., 2009, What matters in corporate governance?,

Review of Financial Studies, 22(2), 783-827.

Beck, T., Demirguc-Kunt, A. S. L. I., & Maksimovic, V., 2005, Financial and legal

constraints to growth: Does firm size matter?, Journal of Finance, 60(1), 137-177.

Becker and Gray, 2009, NBER Productivity database.

Bertrand, M. & Mullainathan, S., 2003, Enjoying the quiet life? Corporate governance

and managerial preferences, Journal of Political Economy, 111(5), 1043-1075.

Billet, M. T., Tao-Hsien, D. K., & David C. Mauer, D. C., 2004, Bondholder wealth

effects in mergers and acquisitions: New evidence from the 1980s and 1990s, Journal of

Finance, 59(1), 107-135.

Demirguc-Kunt, A., & Maksimovic, V., 1998, Law, finance, and firm growth, Journal

of Finance, 53(6), 2107-2137.

Evans, D. S., 1987, Tests of alternative theories of firm growth, Journal of Political

Economy, 95(4), 657-674.

Faleye, O., 2007, Classified boards, firm value, and managerial entrenchment, Journal

of Financial Economics, 83, 501-529.

30

Gaspar, J., Massa, M., & Matos, P., 2005, Shareholder investment horizons and the

market for corporate control, Journal of Financial Economics, 76(1), 135-165.

Gompers, P. Ishii, J., & Metrick, A., 2003, Corporate governance and equity prices,

Quarterly Journal of Economics, 118(1), 107-156.

Griliches, Z., & Regev, H., 1995, Firm productivity in Israeli industry 1979-1988,

Journal of Econometrics, 65(1), 175-203.

Hackbarth, D., & Morellec, E., 2008, Stock returns in mergers and acquisitions, Journal

of Finance, 63(3), 1213-1252.

Hicks, J. R., 1935, Annual survey of economic theory: The theory of monopoly,

Econometrica, January, 1-20.

Hopenhayn, H. A., 1992, Entry, exit, and firm dynamics in long run equilibrium,

Econometrica, 1127-1150.

Ijiri, Y., & Simon, H. A., 1967, A model of business firm growth, Econometrica, 348-

355.

Jovanovic, B., 1982, Selection and the evolution of industry, Econometrica, 649-670.

Levinsohn, J. & Petrin A., 2003, Estimating Production Functions Using Inputs to

Control for Unobservables,Review of Economic Studies, 70(2), 317-342

Maksimovic, V., & Phillips, G. 2001, The market for corporate assets: Who engages in

mergers and asset sales and are there efficiency gains?, Journal of Finance, 56(6), 2019-2064.

Maksimovic, V., Phillips, G. & Prabhala, N.R. 2011, Post-merger restructuring and

the boundaries of the firm, Journal of Financial Economics, 102(2), 317-343

McLean, R. D., Zhang, T., & Zhao, M., 2012, Why does the law matter? Investor

protection and its effects on investment, finance, and growth, Journal of Finance, 67(1),

313-350.

31

Moeller, S. B., Schlingemann, F. P., & Stulz, R. M., 2005, Wealth destruction on a

massive scale? A study of acquiring-firm returns in the recent merger wave, Journal of

Finance, 60(2), 757-782.

Paul, D. L., 2007, Board composition and corrective action: Evidence from corporate

responses to bad acquisition bids, Journal of Finance and Quantitative Analysis, 42(3), 759-

783.

Williamson, O. E., 1964, The Economics of Discretionary Behaviour: Managerial

Objectives in a Theory of the Firm, Englewood Cliffs, NJ, Prentice-Hall.

Zhao, Y., & Chen, K. H., 2008, The influence of takeover protection on earnings

management, Journal of Business Finance & Accounting, 35(34), 347-375.

32

TABLE 1: DESCRIPTIVE STATISTICS

Variable Mean (Std. Dev.) Firm-year

Level Sample Establishment Level Sample

One-year growth (%) 5.42 (37.25)

6.97 (74.43)

G-index 9.21 (2.77)

9.16 (2.66)

E-index 2.18 (1.31)

1.96 (1.29)

Other provisions index 7.03 (2.04)

7.20 (1.98)

Lagged log employment 8.11 (1.55)

2.24 (1.53)

Log firm age 2.99 (0.718)

3.11 (0.862)

Delaware incorporation 0.517 (0.500)

0.493 (0.500)

ROA 0.032 (0.085)

0.036 (0.054)

Leverage 0.190 (0.169)

0.205 (0.172)

Capex to sales 0.057 (0.069)

0.049 (0.065)

R&D to sales 0.016 (0.032)

0.005 (0.017)

N 16966 274268

33

TABLE 2: COMPOSITION OF FIRM GROWTH (OVERALL)

The following table decomposes the mean one-year growth in firm employment into two broad components: growth from existing establishments and growth from net entry of new and acquired establishments. Each of these two components is then decomposed into several subcomponents. ΔEold/Et-1 refers to employment growth in establishments that are at least two years old at year t; ΔEnew_last/Et-1 refers to employment growth in establishments that were founded exactly one year ago; ΔEacq_last/Et-1 refers to employment growth in establishments that were acquired exactly one year ago; ΔEacq refers to employment in establishments that were acquired between years t and t-1; ΔEnew refers to employment in establishments that were founded between years t and t-1; ΔEsold refers to employment in establishments that were sold to another firm between years t and t-1; ΔEclosed refers to employment in establishments that were closed between years t and t-1;

Component Mean Growth

Existing Establishments

ΔEold/Et-1 0.42% (16.72%)

ΔEnew_last/Et-1 0.98% (3.65%)

ΔEacq_last/Et-1 0.00% (3.81%)

ΔEexisting/Et-1 1.53% (19.98%)

Net Entry

ΔEacq/Et-1 4.89% (17.83%)

ΔEnew/Et-1 3.25% (9.31%)

ΔEsold/Et-1 -2.22% (8.57%)

ΔEclosed/Et-1 -2.66% (5.98%)

ΔEnetentry/Et-1 3.57% (25.27%)

Net Firm Growth (ΔE/Et-1) 5.42% (37.25%)

Sample standard deviation in parentheses. N=16,966

34

TABLE 3: COMPOSITION OF FIRM GROWTH AND GOVERNANCE

This table decomposes the mean one-year growth in firm employment (in %) into the same components as in Table 1, by the quality of governance. Poor governance refers to firms in the highest quartile of G-index (N=5537). Good governance refers to firms in the lowest quartile of G-index (N=4773). Medium governance refers to the remaining firms. ***: p<1%; **: p<5%; *: p<10%.

PANEL A Poor Governance Medium Governance

Good Governance

t-statistic (Poor v. Good)

Existing Establishments

ΔEold/Et-1 -0.36 0.34 1.44 5.56***

ΔEnew_last/Et-1 0.85 0.98 1.12 3.89***

ΔEacq_last/Et-1 -0.10 0.00 0.07 2.31**

ΔEexisting/Et-1 0.36 1.62 2.78 6.39***

Net Entry

ΔEacq/Et-1 4.90 4.85 4.95 0.15

ΔEnew/Et-1 3.00 3.29 3.49 2.70***

ΔEsold/Et-1 -2.79 -2.16 -1.65 6.64***

ΔEclosed/Et-1 -2.95 -2.58 -2.44 4.25***

ΔEnetentry/Et-1 2.37 3.75 4.70 4.74*** Net Firm Growth (ΔE/Et-1)

2.69 5.91 7.91 7.41***

35

TABLE 4: FIRM GROWTH AND GOVERNANCE

This table presents pooled OLS regressions of firm growth on measures of managerial governance, along with control variables. Firm growth is measured as the one-year growth in the number of employees, that is, (ΔE/Et-1), and is based on data from the Longitudinal Business Database of the U.S. Census Bureau. Prior Employment for year t is defined as the number of employees at year t-1. Managerial governance is measured using G-index (Gomper, Ishii, & Metrick, 2003) and E-index (Bebchuk, Cohen, & Ferrell, 2008). Other Provisions Index is defined in the context of the independent variable. If G-index is the independent variable, no control for other provisions is included. Where E-index is the independent variable, we measure this as G-index less the E-index. Company age is measured from the first appearance on CRSP. Delaware Inc. is defined as 1 if the company is incorporated in Delaware and 0 otherwise. Leverage is measured as total long-term debt over total assets. ROA is measured as net income over total assets. Capex is defined as capital expenditure over sales. R&D is measured as R&D expenditure divided by sales. All regressions include fixed effects for each SIC3-year combination. Coefficients on these fixed effects are not presented. Standard errors clustered by firm presented in parentheses. ***: p<1%; **: p<5%; *: p<10%. $: Coefficients and standard errors multiplied by 100.

(1) (2) (3) (4) G-index$ -0.396***

(0.130)

-0.244* (0.127)

E-index$

-1.163*** (0.263)

-1.127*** (0.258)

Prior Employment -0.035***

(0.004) -0.036***

(0.004) -0.035*** (0.004)

-0.036*** (0.004)

Other Provisions Index

0.000 (0.001)

0.002 (0.002)

Log (Company Age)

-0.027*** (0.006)

-0.029*** (0.006)

Delaware Inc.

0.008 (0.007)

0.010 (0.007)

ROA

0.445*** (0.049)

0.440*** (0.049)

Leverage

0.072*** (0.027)

0.073*** (0.027)

Capex

0.142** (0.061)

0.142** (0.060)

R&D

-0.048 (0.137)

-0.059 (0.136)

N 16,966 16,966 16,966 16,966 R2 0.28 0.29 0.28 0.29

36

TABLE 5: GROWTH IN EXISTING ESTABLISHMENTS, NET ENTRY AND GOVERNANCE This table presents pooled OLS regressions of the two main components of one-year firm growth --- growth from existing establishments and growth from net entry of new and acquired establishments --- on measures of managerial governance, along with control variables. All independent variable definitions are identical to those in Table 4. All regressions include fixed effects for each SIC3-year combination. Coefficients on these fixed effects are not presented. Standard errors clustered by firm presented in parentheses. ***: p<1%; **: p<5%; *: p<10%.$:Coefficients and standard errors multiplied by 100.

ΔEexisting

Et-1 ΔEexisting

Et-1 ΔEexisting

Et-1 ΔEexisting

Et-1 ΔEnetentry

Et-1 ΔEnetentry

Et-1 ΔEnetentry

Et-1 ΔEnetentry

Et-1 G-index$ -0.184***

(0.064) -0.102

(0.063) -0.223**

(0.093) -0.155*

(0.092)

E-index$ -0.450***

(0.139) -0.380***

(0.136) -0.613***

(0.188) -0.641***

(0.187) Prior Employment -0.011***

(0.002) -0.011***

(0.002) -0.011***

(0.002) -0.011***

(0.002) -0.016***

(0.002) -0.016***

(0.002) -0.016***

(0.002) -0.016***

(0.002)

Other Provisions Index 0.000 (0.001)

0.001 (0.001)

0.000 (0.001)

Log (Company Age) -0.013*** (0.003)

-0.014*** (0.003)

-0.013*** (0.004)

-0.015*** (0.004)

Delaware Inc. 0.001 (0.003)

0.002 (0.001)

0.006 (0.005)

0.007 (0.005)

ROA 0.290*** (0.026)

0.289*** (0.026)

0.138*** (0.031)

0.135*** (0.031)

Leverage 0.012 (0.014)

0.013 (0.014)

0.054*** (0.018)

0.055*** (0.018)

Capex 0.146*** (0.034)

0.146*** (0.033)

0.021 (0.042)

0.020 (0.042)

R&D -0.008 (0.079)

-0.012 (0.079)

-0.007 (0.082)

-0.013 (0.082)

N 16,966 16,966 16,966 16,966 16,966 16,966 16,966 16,966 R2 0.24 0.25 0.26 0.26 0.27 0.27 0.27 0.27

37

TABLE 6: SUBCOMPONENTS OF GROWTH AND GOVERNANCE

Panel A of this table presents coefficients on G-index and E-index from pooled OLS regressions of the three subcomponents of growth from existing establishments on measures of governance. Panel B presents the same for the four subcomponents of growth from net entry. The first and the third columns include only lagged employment as a control. Columns 2 and 4 include the full set of controls from Table 4. All regressions include fixed effects for each SIC3-year combination. Coefficients on these fixed effects and other controls are not presented. Standard errors clustered by firm presented in parentheses. ***: p<1%; **: p<5%; *: p<10%. N=16,966 in all regressions. All coefficients and standard errors multiplied by 100.

PANEL A G-index G-index (with Controls)

E-index E-index (with Controls)

ΔEold

Et-1 -0.127** (0.055)

-0.060 (0.053)

-0.341*** (0.116)

-0.274** (0.113)

ΔEacq_last

Et-1 -0.007 (0.010)

-0.007 (0.010)

-0.004 (0.031)

0.002 (0.031)

ΔEnew_last

Et-1 -0.040***

(0.014) -0.025* (0.014)

-0.058* (0.033)

-0.062* (0.032)

PANEL B G-index G-index (with Controls)

E-index E-index (with Controls)

ΔEacq

Et-1 0.067

(0.072) 0.042

(0.071) -0.163 (0.148)

-0.242 (0.148)

ΔEnew

Et-1 -0.088** (0.038)

-0.055 (0.038)

-0.144* (0.077)

-0.153** (0.076)

ΔEsold

Et-1 -0.087***

(0.031) -0.083** (0.032)

-0.165** (0.069)

-0.133* (0.069)

ΔEclosed

Et-1 -0.050** (0.023)

-0.058*** (0.023)

-0.090* (0.049)

-0.056 (0.049)

38

TABLE 7: INVESTMENT PERFORMANCE AND GOVERNANCE

Panel A of this table presents the coefficients on G-index and E-index from regressions of one, three, and five-year establishment growth on measures of governance. The time period is computed from the year of establishment or acquisition. Only data from the year of establishment or acquisition are used. The sample is limited to establishments that were either new or acquired. Hence, in a given year, any establishments that were older than (or were acquired before) one, three and five years respectively are excluded.

Panel B analyzes closure of establishments at one, three and five-years from the year of establishment or acquisition. The dependent variables in Panel B are dummy variables that are 1 if the establishment will be closed one, three or five years from year t, and 0 otherwise.

The first and the third columns include only establishment employment as a control. Columns 2 and 4 include all the other controls from Table 4. All regressions include fixed effects for each SIC3-state-year combination. Coefficients on these fixed effects and other controls are not presented. Standard errors clustered by firm presented in parentheses.

***: p<1%; **: p<5%; *: p<10%. N=274268, 163626, and 98241, respectively for one, three and five-year growth regressions. N= 2065032, 504134, and 504134, respectively for one, three and five-year exit regressions.

PANEL A G-index G-index (with Controls)

E-index E-index (with Controls)

One-year growth -0.012*** (0.004)

-0.012*** (0.004)

-0.020** (0.009)

-0.016* (0.009)

Three-year growth -0.019***

(0.006) -0.014***

(0.005) -0.039***

(0.013) -0.029** (0.012)

Five-year growth -0.021**

(0.010) -0.013 (0.009)

-0.034* (0.020)

-0.021 (0.016)

PANEL B G-index G-index (with Controls)

E-index E-index (with Controls)

Closure within one year 0.001 (0.001)

0.002* (0.001)

-0.001 (0.002)

0.001 (0.002)

Closure within three years

0.016*** (0.003)

0.019*** (0.003)

0.012** (0.007)

0.017*** (0.007)

Closure within five years

0.013*** (0.003)

0.015*** (0.003)

0.011** (0.006)

0.015*** (0.007)

39

TABLE 8: NEW ESTABLISHMENT PERFORMANCE AND GOVERNANCE

Panel A of this table presents the coefficients on G-index and E-index from regressions of one, three, and five-year establishment growth on measures of governance. The sample is limited to establishments that were founded by the focal firm. The time period is computed from the year of establishment. Only data from the year of establishment are used. Establishments that were founded before one, three and five years respectively are excluded.

Panel B analyzes closure of establishments at one, three, and five-years from the year of establishment or acquisition. The dependent variables in Panel B are dummy variables that are 1 if the establishment will be closed one, three or five years from year t, and 0 otherwise. The sample is limited to establishments that were founded by the focal firm.

The first and the third columns include only establishment employment as a control. Columns 2 and 4 include all the other controls from Table 4. All regressions include fixed effects for each SIC3-state-year combination. Coefficients on these fixed effects and other controls are not presented. Standard errors clustered by firm presented in parentheses.

***: p<1%; **: p<5%; *: p<10%. N= 157719, 98215, and 60794, respectively for one, three, and five-year growth regressions. N= 1322647, 301804, and 301804, respectively for one, three, and five-year exit regressions.

PANEL A G-index G-index (with Controls)

E-index E-index (with Controls)

One-year growth -0.015** (0.006)

-0.013*** (0.005)

-0.024** (0.011)

-0.017* (0.010)

Three-year growth -0.020**

(0.008)

-0.013** (0.006)

-0.042** (0.017)

-0.030** (0.015)

Five-year growth -0.019* (0.010)

-0.010 (0.009)

-0.042* (0.024)

-0.027 (0.021)

PANEL B G-index G-index (with Controls)

E-index E-index (with Controls)

Exit within one year 0.002 (0.001)

0.002* (0.001)

-0.001 (0.002)

-0.001 (0.002)

Exit within three years 0.015*** (0.004)

0.018*** (0.004)

0.012* (0.007)

0.014* (0.008)

Exit within five years 0.014** (0.003)

0.017*** (0.003)

0.010* (0.006)

0.013** (0.006)

40

TABLE 9: ACQUIRED ESTABLISHMENT PERFORMANCE AND GOVERNANCE

Panel A of this table presents the coefficients on G-index and E-index from regressions of one, three, and five-year establishment growth on measures of governance. The sample is limited to establishments that were acquired. The time period is computed from the year of acquisition. Only data from the year of acquisition are used. Establishments that were acquired before one, three and five years respectively are excluded.

Panel B analyzes closure of establishments at one, three, and five-years from the year of establishment or acquisition. The dependent variables in Panel B are dummy variables that are 1 if the establishment will closed one, three, or five years from year t, and 0 otherwise. The sample is limited to establishments that were acquired by the focal firm.

The first and the third columns include only establishment employment as a control. Columns 2 and 4 include all the other controls from Table 4. All regressions include fixed effects for each SIC3-state-year combination. Coefficients on these fixed effects and other controls are not presented. Standard errors clustered by firm presented in parentheses.

***: p<1%; **: p<5%; *: p<10%. N= 116549, 65411, and 37447 respectively for one, three, and five-year growth regressions. N= 742385, 202330, and 202330 respectively for one, three, and five-year exit regressions.

PANEL A G-index G-index (with Controls)

E-index E-index (with Controls)

One-year growth -0.006 (0.005)

-0.005 (0.004)

-0.021 (0.016)

-0.016 (0.010)

Three-year growth -0.019**

(0.009) -0.014* (0.008)

-0.022 (0.020)

-0.001 (0.019)

Five-year growth -0.026***

(0.012) -0.030***

(0.011) 0.004

(0.027) 0.016

(0.028)

PANEL B G-index G-index (with Controls)

E-index E-index (with Controls)

Exit within one year -0.001 (0.001)

-0.001 (0.001)

-0.001 (0.003)

0.001 (0.003)

Exit within three years 0.015*** (0.006)

0.014** (0.006)

0.010 (0.013)

0.020 (0.015)

Exit within five years 0.009** (0.005)

0.009** (0.004)

0.014 (0.011)

0.033*** (0.012)

41

TABLE 10: PAST GROWTH OF ACQUIRED ESTABLISHMENTS AND GOVERNANCE

This table presents the coefficients on G-index and E-index from regressions of one, three, and five-year past establishment growth on measures of governance. The sample is limited to establishments that were acquired. Past establishment growth refers to growth of the establishment prior to its acquisition by the current firm. Hence, one-year past growth refers to the growth in employment in the year prior to the acquisition, three-year past growth refers to the growth in employment in the three years prior to the acquisition, and so on. The first and the third columns include only establishment employment as a control. Columns 2 and 4 include all the other controls from Table 4. All regressions include fixed effects for each SIC3-state-year combination. Coefficients on these fixed effects and other controls are not presented. Standard errors clustered by firm presented in parentheses.

***: p<1%; **: p<5%; *: p<10%. N= 154821, 124956, and 104281, respectively for one, three, and five-year regressions.

G-index G-index

(with Controls) E-index E-index

(with Controls) One-year past growth 0.001

(0.003) 0.001

(0.003) 0.004

(0.006) 0.003

(0.007) Three-year past growth

0.000 (0.004)

0.001 (0.003)

-0.005 (0.014)

-0.003 (0.010)

Five-year past growth 0.004

(0.006) 0.004

(0.007) 0.012

(0.016) 0.005

(0.015)

42

TABLE 11: PRODUCTIVITY AND GOVERNANCE

Panel A of this table presents the coefficients on G-index and E-index from regressions of firm-level productivity on measures of governance. The sample is limited to firms that are primarily in manufacturing (i.e., at least 50% of their employment was in a manufacturing industry). Panel B presents the coefficients on G-index and E-index from regressions of firm-level growth on measures of governance for the same sample. The first and the third columns include no controls. Columns 2 and 4 include all the controls from Table A6. All regressions include fixed effects for each SIC3-year combination. Coefficients on these fixed effects and other controls are not presented. Standard errors clustered by firm presented in parentheses. Labor productivity is defined as log real value of shipments divided by employment. Value of shipments is the sales value deflated using 4-digit SIC industry-specific output deflators (Becker & Gray, 2009). The Solow residual TFP is defined as 𝑇𝑇𝐹𝐹𝑃𝑃𝑖𝑖𝑖𝑖 = 𝑦𝑦𝑖𝑖𝑖𝑖 − 𝛽𝛽𝑚𝑚𝑚𝑚𝑖𝑖𝑖𝑖 − 𝛽𝛽𝑘𝑘𝑘𝑘𝑖𝑖𝑖𝑖 − 𝛽𝛽𝑒𝑒𝑒𝑒𝑖𝑖𝑖𝑖 − 𝛽𝛽𝑛𝑛𝑛𝑛𝑖𝑖𝑖𝑖 − 𝛽𝛽𝑙𝑙𝑙𝑙𝑖𝑖𝑖𝑖 where yit is the log of real value of shipments of firm i in year t, m is log real materials, k is log of real depreciated capital stock, e is log of real energy costs, and n is log of white-collar (non-production) employment and l is log of blue-collar (production) employment. Employment is measured in equivalents of production worker hours and thus adjusts for utilization. The elasticities (β’s) are defined equal to the material share, capital share, energy share, white-collar and blue-collar share of total costs in the 4-digit SIC (1987) industry j to which firm i belongs. ***: p<1%; **: p<5%; *: p<10%. N=5343 in all regressions.

PANEL A G-index G-index (with Controls)

E-index E-index (with Controls)

Labor Productivity 0.004 (0.010)

0.005 (0.010)

-0.000 (0.024)

0.000 (0.023)

TFP-Solow -0.007

(0.010) -0.005 (0.010)

-0.000 (0.024)

0.002 (0.024)

43

ONLINE APPENDIX

TABLE A1: COMPOSITION OF FIRM GROWTH (3-YEAR)

The following table decomposes the mean 3-year growth in firm employment into two broad components – growth from existing establishments and growth from net entry of new and acquired establishments. Each of these two components is then decomposed into several subcomponents. ΔEold/Et-3 refers to employment growth in establishments that are at least 4 years old at year t; ΔEnew_last/Et-3 refers to employment growth in establishments that were founded exactly 3 years ago; ΔEacq_last/Et-3 refers to employment growth in establishments that were acquired exactly 3 years ago; ΔEacq refers to employment in establishments that were acquired between years t and t-3; ΔEnew refers to employment in establishments that were founded between years t and t-3; ΔEsold refers to employment in establishments that were sold to another firm between years t and t-3; ΔEclosed refers to employment in establishments that were closed between years t and t-3;

Component Mean Growth

Existing Establishments

ΔEold/Et-3 1.60% (23.12%)

ΔEnew_last/Et-3 2.42% (10.80%)

ΔEacq_last/Et-3 1.22% (15.39%)

ΔEexisting/Et-3 6.13% (19.98%)

Net Entry

ΔEacq/Et-3 16.44% (53.22%)

ΔEnew/Et-3 15.94% (38.41%)

ΔEsold/Et-3 -5.23% (14.69%)

ΔEclosed/Et-3 -8.82% (12.92%)

ΔEnetentry/Et-3 19.75% (89.99%)

Net Firm Growth (ΔE/Et-3) 27.98% (122.14%)

$: Sample Standard Deviation in Parentheses. N=16,266

44

TABLE A2: COMPOSITION OF FIRM GROWTH (5-YEAR)

The following table decomposes the mean 5-year growth in firm employment into two broad components – growth from existing establishments and growth from net entry of new and acquired establishments. Each of these two components is then decomposed into several subcomponents. ΔEold/Et-5 refers to employment growth in establishments that are at least 6 years old at year t; ΔEnew_last/Et-5 refers to employment growth in establishments that were founded exactly 5 years ago; ΔEacq_last/Et-5 refers to employment growth in establishments that were acquired exactly 5 years ago; ΔEacq refers to employment in establishments that were acquired between years t and t-5; ΔEnew refers to employment in establishments that were founded between years t and t-5; ΔEsold refers to employment in establishments that were sold to another firm between years t and t-5; ΔEclosed refers to employment in establishments that were closed between years t and t-5;

Component Mean Growth

Existing Establishments

ΔEold/Et-5 1.70% (21.39%)

ΔEnew_last/Et-5 5.97% (27.63%)

ΔEacq_last/Et-5 3.54% (28.31%)

ΔEexisting/Et-5 13.42% (61.77%)

Net Entry

ΔEacq/Et-5 28.31% (89.86%)

ΔEnew/Et-5 38.16% (103.33%)

ΔEsold/Et-5 -7.04% (16.22%)

ΔEclosed/Et-5 -14.12% (17.27%)

ΔEnetentry/Et-5 48.51% (191.69%)

Net Firm Growth (ΔE/Et-5) 65.58% (242.06%)

$: Sample Standard Deviation in Parentheses. N=16,045

45

TABLE A3: COMPOSITION OF FIRM GROWTH AND GOVERNANCE (3-YEAR)

This table decomposes the mean 3-year growth in firm employment (in %) into the same components as in Table A1, by the quality of governance. Poor governance refers to firms in the highest quartile of G-index (N=5325). Good governance refers to firms in the lowest quartile of G-index (N=4602). Medium governance refers to the remaining firms. ***: p<1%; **: p<5%; *: p<10%.

PANEL A Poor Governance Medium Governance

Good Governance

t-statistic (Poor v. Good)

Existing Establishments

ΔEold/Et-3 0.14 1.26 3.76 7.78***

ΔEnew_last/Et-3 1.46 2.61 3.26 8.45***

ΔEacq_last/Et-3 -0.00 0.98 3.00 9.60***

ΔEexisting/Et-3 1.63 5.63 12.04 13.23***

Net Entry

ΔEacq/Et-3 14.24 16.46 18.97 4.39***

ΔEnew/Et-3 12.26 16.26 19.78 9.65***

ΔEsold/Et-3 -6.96 -4.99 -3.55 11.21***

ΔEclosed/Et-3 -9.51 -8.89 -7.93 6.15***

ΔEnetentry/Et-3 10.76 19.88 29.97 10.37*** Net Firm Growth (ΔE/Et-3)

13.62 27.27 45.57 12.61***

46

TABLE A4: COMPOSITION OF FIRM GROWTH AND GOVERNANCE (5-YEAR)

This table decomposes the mean 5-year growth in firm employment (in %) into the same components as in Table A1, by the quality of governance. Poor governance refers to firms in the highest quartile of G-index (N=5322). Good governance refers to firms in the lowest quartile of G-index (N=4504). Medium governance refers to the remaining firms. ***: p<1%; **: p<5%; *: p<10%.

PANEL A Poor Governance Medium Governance

Good Governance

t-statistic (Poor v. Good)

Existing Establishments

ΔEold/Et-5 0.83 1.34 3.22 5.54***

ΔEnew_last/Et-5 2.54 5.96 10.03 13.17***

ΔEacq_last/Et-5 0.49 2.85 8.08 13.04***

ΔEexisting/Et-5 4.51 12.07 25.82 17.09***

Net Entry

ΔEacq/Et-5 22.62 28.06 35.40 6.91***

ΔEnew/Et-5 23.72 38.66 54.52 14.50***

ΔEsold/Et-5 -9.39 -6.65 -4.81 13.55***

ΔEclosed/Et-5 -14.66 -14.53 -12.91 5.13***

ΔEnetentry/Et-5 23.44 47.72 79.23 13.86*** Net Firm Growth (ΔE/Et-5)

29.41 62.83 112.12 16.33***

47

TABLE A5: 3-YEAR FIRM GROWTH AND GOVERNANCE

This table presents pooled OLS regressions of firm growth on measures of managerial governance, along with control variables. Firm growth is measured as the 3-year growth in the number of employees, i.e. (ΔE/Et-3), and is based on data from the Longitudinal Business Database of the US Census Bureau. Prior Employment for year t is defined as the number of employees at year t-3. Managerial governance is measured using G-index (Gomper, Ishii, and Metrick, 2003) and E-index (Bebchuk, Cohen, & Ferrell, 2008). Other Provisions Index is defined in the context of the independent variable. If G-index is the independent variable, no control for other provisions is included. Where E-index is the independent variable, we measure this as G-index less the E-index. Company age is measured from the first appearance on CRSP. Delaware Inc is defined as 1 if the company is incorporated in Delaware and 0 otherwise. Leverage is measured as total long-term debt over total assets. ROA is measured as net income over total assets. Capex is defined as capital expenditure over sales. R&D is measured as R&D expenditure divided by sales. All regressions include fixed effects for each SIC3-year combination. Coefficients on these fixed effects are not presented. Standard errors clustered by firm presented in parentheses. ***: p<1%; **: p<5%; *: p<10%.

(1) (2) (3) (4) G-index -0.010*

(0.006)

-0.003 (0.006)

E-index

-0.042*** (0.011)

-0.045*** (0.011)

Prior Employment -0.242***

(0.021) -0.246***

(0.021) -0.228*** (0.021)

-0.231*** (0.021)

Other Provisions Index

0.007 (0.008)

0.020** (0.008)

Log (Company Age)

-0.142*** (0.026)

-0.153*** (0.026)

Delaware Inc.

0.052* (0.029)

0.061* (0.029)

ROA

0.675*** (0.227)

0.647*** (0.226)

Leverage

0.273** (0.119)

0.282** (0.119)

Capex

0.453* (0.232)

0.451** (0.232)

R&D

0.065 (0.067)

0.068 (0.066)

N 16,266 16,266 16,266 16,266 R-Square 0.39 0.39 0.39 0.40

48

TABLE A6: 5-YEAR FIRM GROWTH AND GOVERNANCE

This table presents pooled OLS regressions of firm growth on measures of managerial governance, along with control variables. Firm growth is measured as the 5-year growth in the number of employees, i.e. (ΔE/Et- 5), and is based on data from the Longitudinal Business Database of the US Census Bureau. Prior Employment for year t is defined as the number of employees at year t-5. Managerial governance is measured using G-index (Gomper, Ishii, and Metrick, 2003) and E-index (Bebchuk, Cohen, & Ferrell, 2008). Other Provisions Index is defined in the context of the independent variable. If G-index is the independent variable, no control for other provisions is included. Where E-index is the independent variable, we measure this as G-index less the E-index. Company age is measured from the first appearance on CRSP. Delaware Inc is defined as 1 if the company is incorporated in Delaware and 0 otherwise. Leverage is measured as total long-term debt over total assets. ROA is measured as net income over total assets. Capex is defined as capital expenditure over sales. R&D is measured as R&D expenditure divided by sales. All regressions include fixed effects for each SIC3-year combination. Coefficients on these fixed effects are not presented. Standard errors clustered by firm presented in parentheses. ***: p<1%; **: p<5%; *: p<10%.

(1) (2) (3) (4) G-index -0.015

(0.012)

-0.003 (0.012)

E-index

-0.069*** (0.026)

-0.075*** (0.026)

Prior Employment -0.638***

(0.049) -0.644***

(0.049) -0.609*** (0.051)

-0.615*** (0.051)

Other Provisions Index

0.015 (0.017)

0.038** (0.016)

Log (Company Age)

-0.251*** (0.060)

-0.270*** (0.061)

Delaware Inc.

0.171*** (0.066)

0.189*** (0.066)

ROA

1.546*** (0.452)

1.500*** (0.452)

Leverage

0.637** (0.261)

0.654** (0.261)

Capex

0.633 (0.470)

0.626 (0.470)

R&D

0.237 (1.175)

0.137 (1.176)

N 16,045 16,045 16,045 16,045 R-Square 0.42 0.42 0.43 0.43

49

TABLE A7: SUBCOMPONENTS OF 3-YEAR GROWTH AND GOVERNANCE

Panel A of this table presents coefficients on G-index and E-index from pooled OLS regressions of the three subcomponents of growth from existing establishments on measures of governance. Panel B presents the same for the four subcomponents of growth from net entry. The first and the third columns include only lagged employment as a control. Columns 2 and 4 include the full set of controls from Table A6. All regressions include fixed effects for each SIC3-year combination. Coefficients on these fixed effects and other controls are not presented. Standard errors clustered by firm presented in parentheses. ***: p<1%; **: p<5%; *: p<10%. N=16,266 in all regressions.

PANEL A G-index G-index (with Controls)

E-index E-index (with Controls)

ΔEold

Et-3 -0.124 (0.110)

-0.027 (0.109)

-0.509** (0.238)

-0.430* (0.233)

ΔEacq_last

Et-3 -0.147** (0.063)

-0.039 (0.062)

-0.384*** (0.139)

-0.389*** (0.137)

ΔEnew_last

Et-3 -0.157***

(0.441) -0.092***

(0.043) -0.297***

(0.103) -0.301***

(0.101) ΔEexistitng

Et-3 -0.479***

(0.160) -0.178 (0.159)

-1.382*** (0.365)

-1.330*** (0.352)

PANEL B G-index G-index (with Controls)

E-index E-index (with Controls)

ΔEacq

Et-3 -0.064 (0.288)

0.100 (0.209)

-0.862 (0.545)

-0.114** (0.054)

ΔEnew

Et-3 -0.429** (0.194)

-0.186 (0.194)

-0.892** (0.398)

-1.104*** (0.386)s

ΔEsold

Et-3 -0.229***

(0.073) -0.216***

(0.075) -0.454***

(0.169) -0.400** (0.168)

ΔEclosed

Et-3 -0.192***

(0.071) -0.215***

(0.070) -0.285* (0.160)

-0.182 (0.156)

ΔEnetentry

Et-3 -0.836* (0.434)

-0.427 (0.429)

-2.611*** (0.859)

-2.906*** (0.840)

50

TABLE A8: SUBCOMPONENTS OF 5-YEAR GROWTH AND GOVERNANCE

Panel A of this table presents coefficients on G-index and E-index from pooled OLS regressions of the three subcomponents of growth from existing establishments on measures of governance. Panel B presents the same for the four subcomponents of growth from net entry. The first and the third columns include only lagged employment as a control. Columns 2 and 4 include the full set of controls from Table A6. All regressions include fixed effects for each SIC3-year combination. Coefficients on these fixed effects and other controls are not presented. Standard errors clustered by firm presented in parentheses. ***: p<1%; **: p<5%; *: p<10%. N=16,045 in all regressions.

PANEL A G-index G-index (with Controls)

E-index E-index (with Controls)

ΔEold

Et-5 0.111

(0.126) 0.110

(0.126) -0.369 (0.238)

-0.290 (0.237)

ΔEacq_last

Et-5 -0.311** (0.149)

-0.115 (0.143)

-0.642** (0.314)

-0.634** (0.310)

ΔEnew_last

Et-5 -0.259** (0.117)

-0.097 (0.116)

-0.666*** (0.257)

-0.669*** (0.252)

ΔEexistitng

Et-5 -0.339 (0.312)

-0.059 (0.308)

-1.91*** (0.652)

-1.86*** (0.632)

PANEL B G-index G-index (with Controls)

E-index E-index (with Controls)

ΔEacq

Et-5 -0.255 (0.549)

-0.310 (0.543)

-1.20 (1.06)

1.63 (1.06)

ΔEnew

Et-5 -1.01* (0.518)

-0.469 (0.505)

-3.23*** (1.16)

-3.47*** (1.13)

ΔEsold

Et-5 -0.304***

(0.102) -0.293***

(0.103) -0.640***

(0.214) -0.550***

(0.214) ΔEclosed

Et-5 -0.186* (0.110)

-0.224** (0.108)

-0.158 (0.248)

-0.059 (0.243)

ΔEnetentry

Et-5 -0.168* (0.101)

-0.081 (0.010)

-0.503** (0.213)

-0.555*** (0.208)

51

TABLE A9: INVESTMENT PERFORMANCE AND GOVERNANCE (Establishment Exit)

This table analyzes exit of establishments at 1, 3 and 5-years from the year of establishment or acquisition. The dependent variables in Panel B are dummy variables that are 1 if the establishment will be sold or closed 1, 3 or 5 years from year t, and 0 otherwise. Panel A includes all establishments that were acquired or newly established by the focal firm. Panel B is limited to newly founded establishments while the sample in Panel C contains acquired establishments only. The first and the third columns include only establishment employment as a control. Columns 2 and 4 include all the other controls from Table 4. All regressions include fixed effects for each SIC3-state-year combination. Coefficients on these fixed effects and other controls are not presented. Standard errors clustered by firm presented in parentheses. ***: p<1%; **: p<5%; *: p<10%. N=274268, 163626, and 98241 respectively for 1, 3 and 5-year growth regressions. N= 2065032, 504134, and 504134 respectively for 1, 3 and 5-year exit regressions.

PANEL A (Both) G-index G-index (with Controls)

E-index E-index (with Controls)

Exit within one-year 0.002 (0.001)

0.002* (0.001)

0.000 (0.002)

0.000 (0.002)

Exit within 3-years 0.016*** (0.003)

0.019*** (0.003)

0.015** (0.007)

0.021*** (0.007)

Exit within 5-years 0.011*** (0.003)

0.013*** (0.003)

0.013** (0.006)

0.021*** (0.007)

PANEL B (New) G-index G-index (with Controls)

E-index E-index (with Controls)

Exit within one-year 0.003* (0.001)

0.003** (0.001)

0.000 (0.003)

-0.001 (0.003)

Exit within 3-years 0.015*** (0.003)

0.018*** (0.003)

0.016** (0.007)

0.019*** (0.007)

Exit within 5-years 0.012** (0.004)

0.014*** (0.003)

0.013** (0.006)

0.018*** (0.007)

PANEL C (Acquired) G-index G-index (with Controls)

E-index E-index (with Controls)

Exit within one-year -0.002 (0.002)

-0.001 (0.002)

-0.002 (0.003)

-0.001 (0.003)

Exit within 3-years 0.015** (0.006)

0.014** (0.006)

0.010 (0.013)

0.020 (0.015)

Exit within 5-years 0.014** (0.005)

0.009** (0.004)

0.014 (0.011)

0.033*** (0.012)

52

TABLE A10: Additional Analyses Related to Acquisitions

This table presents coefficients on G-index and E-index from pooled OLS regressions of four additional measures of acquisition-related growth on measures of governance. The first and the third columns include only firm employment as a control. Columns 2 and 4 include the full set of controls from Table A6. All regressions include fixed effects for each SIC3-year combination. Coefficients on these fixed effects and other controls are not presented. Standard errors clustered by firm presented in parentheses. ***: p<1%; **: p<5%; *: p<10%. N=16,464 in all regressions.

G-index G-index (with Controls)

E-index E-index (with Controls)

Number of Acquired Establishments to Total

-0.002 (0.002)

-0.001 (0.002)

0.007* (0.004)

0.007* (0.004)

Number of New Establishments to Total

-0.003* (0.002)

-0.004** (0.002)

-0.010** (0.004)

-0.008* (0.004)

53

TABLE A11: PRODUCTIVITY AND GOVERNANCE

Panel A of this table presents the coefficients on G-index and E-index from regressions of firm-level productivity on measures of governance. The sample is limited to firms that are primarily in manufacturing (i.e., at least 50% of their employment was in a manufacturing industry). Panel B presents the coefficients on G-index and E-index from regressions of firm-level growth on measures of governance for the same sample.

The first and the third columns include no controls. Columns 2 and 4 include all the controls from Table A6. All regressions include fixed effects for each SIC3-year combination. Coefficients on these fixed effects and other controls are not presented. Standard errors clustered by firm presented in parentheses.

The definitions of the various productivity measures are provided on the next page.

***: p<1%; **: p<5%; *: p<10%. N=5343 in all regressions.

PANEL A G-index G-index (with Controls)

E-index E-index (with Controls)

TFP- OLS Fixed Effects -0.002 (0.017)

-0.000 (0.017)

-0.009 (0.038)

-0.007 (0.038)

TFP-Translog -0.006

(0.018) -0.004 (0.018)

-0.013 (0.044)

-0.011 (0.044)

TFP-Blundell Bond System GMM

-0.005 (0.012)

-0.003 (0.012)

-0.003 (0.029)

-0.001 (0.028)

TFP-Levinsohn-Petrin 0.002

(0.019) 0.005

(0.019) -0.016 (0.043)

-0.013 (0.043)

PANEL B G-index G-index (with Controls)

E-index E-index (with Controls)

one-year Emp. Growth -0.006*** (0.002)

-0.005*** (0.002)

-0.006 (0.004)

-0.006 (0.004)

3-year Emp. Growth -0.024***

(0.009) -0.023***

(0.008) -0.025 (0.019)

-0.025 (0.019)

5-year Emp. Growth -0.088***

(0.021) -0.085***

(0.020) -0.074* (0.040)

-0.073* (0.040)

54

Construction of Productivity Measures

OLS-FE TFP measures The OLS-FE productivity measure is defined as the residual from an OLS establishment-fixed-effects regression of log real value of shipments on log real materials, log real energy costs, log blue-collar employment, log white-collar employment and log real capital. Thus, this measure addresses potential correlation between unobservable, time-invariant, firm-level factors that may be correlated with input choices.

Levinsohn-Petrin TFP measure We assume a 3-factor Cobb-Douglas value-added production function with the productivity residual comprising two terms: ωit, the component of the productivity shock that is known to the decision-maker before making the choice of inputs, but is unobserved by the econometrician, and ηit, assumed to captures all other deviations from the hypothesized production function. The LP method assumes that the demand for the intermediate input (i.e., real materials in our case) is a function of capital, kit and ωit. Based on some mild assumptions about the firm’s production technology, ωit can be written as a function of kit and the intermediate input. Thus, a first stage regression of value added on labor and a polynomial function of capital and materials, estimates coefficients on labor inputs. To recover the capital coefficient, the LP methodology makes two assumptions. One is that ωit follows a first-order Markov process. Then, assuming that kit is chosen prior to realization of period t shocks, kit is orthogonal to innovations in productivity. Further details are available in Levinsohn and Petrin (2003).

Blundell-Bond system-GMM TFP measure We follow Blundell and Bond (2000), and assume a gross output production function (skilled and unskilled labor, capital, materials and energy as inputs) with a firm fixed effect and an AR1 component in the productivity term. The Arellano-Bond (1991) moment conditions are:

𝐸𝐸�∆𝑥𝑥𝑖𝑖𝑖𝑖−𝑗𝑗∆𝜔𝜔𝑖𝑖𝑖𝑖� = 0, 𝑗𝑗 ≥ 3 where x is a vector of inputs and output, and ωit is the error term, assumed MA(1).

Blundell and Bond (2000) show that with some more assumptions, we can obtain additional moment conditions:

𝐸𝐸�∆𝑥𝑥𝑖𝑖𝑖𝑖−𝑗𝑗(𝜃𝜃𝑖𝑖∗𝜔𝜔𝑖𝑖𝑖𝑖)� = 0 where 𝜃𝜃𝑖𝑖∗ is the modified firm fixed effect. Combining both sets of moment conditions provides the Blundell and Bond (2000) system GMM estimator. Further details are available in Blundell and Bond (2003).

Translog-TFP measure As an alternative to the Cobb-Douglas production function, we consider the following second order translog specification:

𝑦𝑦𝑖𝑖𝑖𝑖 = �𝛽𝛽𝑗𝑗𝑗𝑗

𝑋𝑋𝑖𝑖𝑖𝑖𝑗𝑗 + 𝛽𝛽𝑗𝑗𝑗𝑗𝑋𝑋𝑖𝑖𝑖𝑖

𝑗𝑗 𝑋𝑋𝑖𝑖𝑖𝑖𝑗𝑗 + ��𝛽𝛽𝑗𝑗𝑘𝑘

𝑘𝑘𝑗𝑗≠𝑘𝑘

𝑋𝑋𝑖𝑖𝑖𝑖𝑗𝑗 𝑋𝑋𝑖𝑖𝑖𝑖

𝑘𝑘𝑗𝑗 + 𝑓𝑓𝑖𝑖 + 𝜔𝜔𝑖𝑖𝑖𝑖

where i indexes firms, t years, j and k index the inputs. We use log of real materials, log of real energy costs, log of the real depreciated capital stock, log of the number of production (blue collar) employees and the log of the number of non-production (white collar) employees as the inputs. We use the residuals estimated using OLS with firm fixed effects as TFP measures.

55

TABLE A12: UNWINSORIZED CHANGES IN EMPLOYMENT LEVELS

The following table decomposes the mean 3-year change in firm employment into two broad components – growth from existing establishments and growth from net entry of new and acquired establishments. Each of these two components is then decomposed into several subcomponents. ΔEold refers to employment growth in establishments that are at least 4 years old at year t; ΔEnew_last refers to employment growth in establishments that were founded exactly 3 years ago; ΔEacq_last refers to employment growth in establishments that were acquired exactly 3 years ago; ΔEacq refers to employment in establishments that were acquired between years t and t-3; ΔEnew refers to employment in establishments that were founded between years t and t-3; ΔEsold refers to employment in establishments that were sold to another firm between years t and t-3; ΔEclosed refers to employment in establishments that were closed between years t and t-3; The table illustrates that the decomposition is additive in its components and subcomponents.

Component Mean Change in Levels

Existing Establishments

ΔEold -101.12

ΔEnew_last 144.83

ΔEacq_last -35.39

ΔEexisting 8.32

Net Entry

ΔEacq 907.11

ΔEnew 1,114.37

ΔEsold -783.55

ΔEclosed -882.85

ΔEnetentry 355.08

Net Change in Firm Employment 363.40

56