You are on page 1of 16

Journal of Corporate Finance 36 (2016) 174–189

Contents lists available at ScienceDirect

Journal of Corporate Finance


journal homepage: www.elsevier.com/locate/jcorpfin

Political uncertainty and corporate investment: Evidence


from China☆
Heng An a, Yanyan Chen b,⁎, Danglun Luo c, Ting Zhang d
a
Bryan School of Business and Economics, University of North Carolina Greensboro, Greensboro, NC 27402, USA
b
College of Economics and Management, South China Agriculture University, 483 Wushan Road, Guangzhou 510642, PR China
c
Lingnan College, Sun Yat-Sen University, 135 Xingangxi Road, Guangzhou 510275, PR China
d
School of Business Administration, University of Dayton, Dayton, OH 45469, USA

a r t i c l e i n f o a b s t r a c t

Article history: Using hand-collected data on changes of government officials in 277 Chinese cities, we examine
Received 10 March 2015 how political turnover affects corporate investment in a transitional economy. We find that polit-
Received in revised form 23 November 2015 ical turnover leads firms to significantly reduce corporate investment, particularly when the new
Accepted 29 November 2015
official is an outsider appointed by a higher level government. The effect of political turnover on
Available online 9 December 2015
corporate investment is stronger for state-owned enterprises, capital intensive firms, and firms
deemed locally important. Overall, the volatility of corporate investment increases with political
JEL classification: turnover. Finally, the investment decline due to political turnover has significantly negative
G32
impact on the profitability of private firms, but not state-owned firms.
G38
© 2015 Elsevier B.V. All rights reserved.
G18
G15

Keywords:
Political turnover
Corporate investment
Local government officials
Chinese market
Transition economy

1. Introduction

China has achieved remarkable economic growth during the last three decades, despite weak economic intuitions and financial
markets. In stark contrast to Western laisser-faire capitalism, Chinese government plays an active role in promoting economic devel-
opment, stirring salient debate on the relative merits of the government and the market. While some attribute China's economic suc-
cess to the central government, many economists recognize local government as the centerpiece of China's economic revolution.
Although policy makers in Beijing enact laws and regulations, local government officials have a great deal of leeway in how the
laws and policies are enforced in their jurisdiction.1 Qian and Xu (1993) describe China as a decentralized economic system in

☆ This research is supported by National Natural Science Foundation of China under Grant No. 71402058, Ministry of Education Humanities and Social Sciences
Foundation under Grant No. 13YJC790013, Guangdong Social Science Foundation under Grant No. GD14CYJ03, and Guangdong Natural Science Foundation under
Grant No. 2014A030313118. We also thank Gary Tian and conference participants at the 2015 AsianFA conference.
⁎ Corresponding author. Tel./fax: +86 20 85282110.
E-mail addresses: h_an@uncg.edu (H. An), daisycyy5@hotmail.com (Y. Chen), luodl@mail.sysu.edu.cn (D. Luo), tzhang1@udayton.edu (T. Zhang).
1
See Li and Bachman (1989), Walder (1995), and Walder (1996).

http://dx.doi.org/10.1016/j.jcorpfin.2015.11.003
0929-1199/© 2015 Elsevier B.V. All rights reserved.
H. An et al. / Journal of Corporate Finance 36 (2016) 174–189 175

which local governments have significant autonomy in economic matters. Furthermore, local politicians have both strong incentive
and enormous power to boost economic growth in their region.2
First, China's fiscal decentralization, particularly the tax-sharing programs implemented in 1993, encourages local governments
to promote local economy in order to collect more fiscal revenues. Second, the party's cadre evaluation system is based on local
economic performance, which encourages yardstick competition among local officials to generate high growth.3 Cull et al.
(forthcoming) document that local governments in China help firms in their region secure loans from banks, and provide local
firms information about products, technologies and market opportunities.
Bai et al. (2014) characterize China's government-guided capitalism as crony capitalism, in which local government officials play a
central role in economic growth. The implicit arrangement is that local leaders will benefit personally from the success of business
connected to them, which incentivizes officials to utilize their enormous political and economic power to support their local cronies.4
Compared with the grabbing hands of many other developing countries though, politicians in China have delivered impressive
economic growth.5 However, questions remain how sustainable the China model is. While local leaders can provide extraordinary
support for their cronies, they also impose local firms to significant downside risk when the helping hands leave office. For firms in
China, dealing with politicians and the associated political risk is paramount. In this study, we investigate how political uncertainty,
particularly government official turnover, affects corporate investment.
China provides a unique setting to study this issue. First, China's economic growth has relied heavily on fixed asset investment,
which accounts for about 50% of the GDP.6 Moreover, state-owned enterprises (SOEs) dominate the Chinese economy. According
to Liu and Siu (2012), more than 50% of China's fixed asset investment concentrates in the state or the quasi-state sectors. SOEs
have close relationships with the government, and their investments can be sensitive to political uncertainty. Second, government
officials in China are appointed by higher-level Communist party officials, rather than elected by local constituents. The opaque
political system makes government official turnovers unpredictable for most market participants. Given the strong influence of
local leaders on the economy, political turnovers can upend the status quo, creating significant uncertainty for corporations.
Dixit and Pindyck (1994), in their modeling of corporate investment decisions, find that the option value of waiting for better
information increases when firms face uncertainty. Given the irreversibility of capital investment, firms are better off withholding
investment until any uncertainty is resolved.7 In the Chinese situation, uncertainties arise when new government officials bring
changes to existing economic policies. Local firms must wait and learn the new rules of the game. Dixit and Pindyck (1994) suggest
that firms should delay investment until the new rules are known. Therefore, we expect corporate investment to decline when polit-
ical turnover occurs.
To measure political turnover, we hand-collect data on changes in the governments of 277 Chinese cities. This includes the replace-
ment of the mayor, party head, or both at the city level. To measure the degree of political turnover, we calculate the frequency of
changes in the government officials of a city during the sample period. We also distinguish different types of political turnover. The
appointment of an outsider by a higher level government can lead to a more dramatic change in local economic policies. Thus, we
categorize the changes as external appointments or local promotions. We expect a more marked decline in corporate investment
following external appointments.
We find Chinese firms significantly reduce their corporate investment following the replacement of local government officials.
The effect of political turnover is more pronounced for locally important firms, which contribute significantly to the local economy
and thus have more at stake. Similarly, capital intensive firms reduce their investments by a higher degree following political
turnover, as local government officials are often closely involved with these firms, due to their high levels of investment and im-
portance to the local economy.
During the sample period, political turnover is positively related to the volatility of corporate investment. The effect of external ap-
pointment is also more significant than local promotion. Moreover, we take advantage of the unique ownership structure of Chinese
firms by comparing SOEs with non-SOEs. We find that political turnover has significantly more effect on the corporate investment of
SOEs, as they are more politically connected, and thus more sensitive to political risk.
Finally, we investigate the welfare implications of investment decline due to political turnover. Since overinvestment is perceived
to be prevalent in China, less wasteful investment should not hurt firm performance and may even have positive welfare implications.
Therefore, we proceed to examine the impact of investment decline due to political turnover on firm performance. If firms invest
efficiently, then reduced investment will result in missed investment opportunities, which is harmful to firm performance. On the
other hand, if firms overinvest, then less overinvestment may help better align corporate investment with investment opportunities.
Overinvestment is particularly a concern for SOEs, whose top executives are appointed by the government. Besides economic reasons,
SOEs also invest to accomplish social and political goals under government intervention.8 We find that the impact of investment

2
A partial list of studies examining the role played by Chinese local officials in promoting economic growth includes: Montinola et al. (1995), Yu (1997), Duckett
(1998), Duckett (2001), Jin et al. (2005), Maskin et al. (2000), Whiting (2001), Bardhan and Mookherjee (2006), and Huang (2008).
3
Li and Zhou (2005) provide empirical evidence that local government officials are more likely to be promoted if their local economy performs well.
4
The personal benefits range from promotion, as mentioned before, to bribes and private consumption.
5
See Fryer and Shleifer (1997), and Blanchard and Shleifer (2001).
6
Export is another pillar of China's economic growth, but it dropped significantly after the global financial crisis which weakens the demands in developed countries,
leaving investment the most important driver of economic growth.
7
Similar theoretical predictions have been made by Bernanke (1983), Leahy and Whited (1996), and Bloom et al. (2007), among others.
8
Liu and Siu (2012) find that the internal rate of return of private firms in China is 10 percentage points higher than that of an SOE. Chen et al. (2011) find that the
sensitivity of investment to investment opportunities is significantly higher for private firms than SOEs.
176 H. An et al. / Journal of Corporate Finance 36 (2016) 174–189

decline due to political turnover on firm profitability depends on firms' ownership. For private firms, the reduction in investment due
to political turnover has a significantly negative impact on firm profitability. However, for SOEs, the impact is not statistically signif-
icant. The results are consistent with the notion that private firms invest more efficiently than SOEs, and the decrease in investment
due to political turnover results in missed investment opportunities for private firms, which eventually hurts their performance. Given
the outsized contribution of private firms to the economic reform in China, our findings show that political uncertainty takes a toll on
an important growth engine of the world's second largest economy.
The findings of our study present a dynamic view of corporate investment activities when firms face political uncertainty.
More importantly, the empirical results suggest that political uncertainty affects real economic outcomes. This study contributes
to a growing number of investigations into the economic effects of political risk. Our results are in line with the cross-country ev-
idence reported by Julio and Yook (2012), who find that corporate investment declines during election years. An implication of
these studies is that political uncertainty deters economic growth. Given the disproportionate effect of corporate investment on
China's economy, lowering government official turnover can help to mitigate the recent economic slowdown. The study also con-
tributes to the literature examining the role of the Chinese government in promoting economic development.9 Our results high-
light the risks of relying on government for economic growth. Political turnover creates significant risk, which complicates
corporate investment decisions, and may result in missed investment opportunities in the corporate sector.
The remainder of the paper is organized as follows. Section 2 describes the sample and variables; Section 3 reports the main
empirical results; and Section 4 concludes the paper.

2. Data and variables

2.1. Data and sample

We obtain the data in two stages. First, we manually collect the data of government official turnover for specific Chinese cities
provided by the China Economic Information Network.10 We collect data concerning mayoral and city leader changes from various
public resources, such as newspapers, websites, press releases, and other announcements.11
Next, we obtain firm-level accounting data from the Chinese Industrial Enterprises Database (CIED), maintained by China's
National Bureau of Statistics. This contains data on all Chinese industrial firms with sales of over RMB5 million. The database in-
cluded detailed information of over 330,000 enterprises at the end of 2009, including SOEs, non-SOEs, publically listed firms, and
private firms. Their combined revenue accounts for approximately 95% of the total revenue of all industrial firms in China.12
Our initial sample includes all industrial firms in the CIED from 1999 to 2009 that meet the following criteria: (a) the firm must
have valid information on total assets, sales, state ownership, and the location of its headquarter; (b) it is not reported as bankrupt,
merged, nor closed; (c) total assets are greater than RMB100 million; total fixed assets are greater than zero; and depreciation ex-
penses are greater than zero; (d) total debt and total sales are greater than zero; (e) the firm was established after 1900; and (f) it
has at least two years of data available in the database. We obtain 1,062,788 firm-year observations from 2001 to 2009, after
implementing this screening criteria.13 Firms are assigned to a particular city based on the location of their headquarters. Combining
firm-level information with government official turnover data for 277 cities (including 484 heads and 560 mayors), our final sample
consists of 1,020,321 firm-year observations from 2001 to 2009.14

2.2. Measures of political turnover

We construct several variables to measure both the nature and degree of political turnover. First, we use four dummy variables
to indicate the nature of government official turnover: CHG_HEADi,t is equal to 1 if the head of a city is replaced in year t and
0 otherwise; CHG_MAYORi,t is equal to 1 if the mayor is replaced in year t and 0 otherwise; CHGi,t is equal to 1 if either the
head or mayor is changed in year t and 0 otherwise; and CHG_BOTHi,t is equal to 1 if both the head and mayor are replaced in year
t and 0 otherwise. If an official change occurs in the first (second) half of year t, we treat it as occurring in year t (year t + 1).
Second, we use the following two variables to measure the degree of political turnover: CHG_FREQi,t and EXT_APPi,t. CHG_FREQi,t
is the frequency of official changes in a city during the sample period. A higher frequency suggests more political uncertainty in
the city. EXT_APPi,t is equal to 1 when an outsider is appointed by a higher level of government as the new city leader, and 0 when
the new leader is promoted within the city. The effect of external appointments is generally more pronounced, as previously
discussed.

9
See Li and Zhou (2005), Chen et al. (2005), Jin et al. (2005), and Li et al. (2008).
10
The website is http://www.cei.gov.cn.
11
Changes in vice mayor and vice head are not included in our sample because these officials generally do not possess absolute political decision-making power or
authority in the Chinese political system.
12
CIED is the most comprehensive database of Chinese industrial enterprises. A growing number of papers examining economic issues in China have utilized this da-
tabase (e.g., Hsieh and Klenow, 2009; Song et al., 2011). More information about the database can be accessed at http://www.allmyinfo.com/eng/services/index1-1.asp.
13
CIED covers industrial firms from 1999 to 2009. Most variables in the study (e.g., sales growth, capital investment) require data lagged for 2 years, so our sample
period begins in 2001.
14
These do not include the four cities directly controlled by the Chinese central government Beijing, Shanghai, Tianjin, and Chongqing the mayors and heads of these
(equivalent to provinces in China) are of much higher political rank than those of the smaller cities referred to in the study.
H. An et al. / Journal of Corporate Finance 36 (2016) 174–189 177

Table 1
Sample description.
The data on government official turnover at the city level, including the change of mayor and CCP head, are manually collected from government newspapers, websites,
news releases, and other public announcements. Firm-level financial and accounting information, including total assets, fixed assets, corporate investment, sales,
leverage, firm age, and cash holdings is obtained from the Chinese Industrial Enterprises Database, maintained by China National Bureau of Statistics. Firms are
assigned to a particular city based on the location of headquarters. Combing firm-level data with government official turnover data for 277 cities (including 484 heads
and 560 mayors), our final sample includes 1,020,321 firm-year observations from 2001 to 2009. Panel A reports sample distributions by year from 2001 to 2009.
Panel B shows the cities with government official turnover as a percentage of total cities (277 cities). Panel C shows government officials tenure distribution.

Panel A: sample distribution by year

Year No. of observations No. of cities GDP of cities to total GDP (%)

2001 59,864 182 68.27


2002 71,139 220 77.83
2003 86,582 230 79.60
2004 84,149 250 84.79
2005 101,399 262 89.21
2006 166,715 268 90.70
2007 178,698 266 87.82
2008 140,375 269 90.17
2009 131,400 251 83.18
Total 1,020,321 277 /

Panel B: distribution of political turnover by year

Year Head turnover (%) Mayor turnover (%) Head or mayor turnover (%) Head and mayor turnover (%)

2001 31.32 34.62 45.60 20.33


2002 30.45 29.09 42.27 17.27
2003 37.39 44.78 58.70 23.48
2004 18.40 22.80 32.80 8.40
2005 22.14 22.14 33.21 11.07
2006 21.27 27.61 37.69 11.19
2007 27.07 37.59 46.62 18.05
2008 40.52 40.89 55.02 26.39
2009 11.95 16.33 21.91 6.37
Average 26.72 30.65 41.54 15.84

Panel C: distribution of the tenure of government officials

Tenure (No. of years) No. of Heads % No. of mayors %

1 40 8.26 54 9.64
2 94 19.42 137 24.46
3 94 19.42 161 28.75
4 95 19.63 102 18.21
5 88 18.18 65 11.61
6 40 8.26 23 4.11
7 22 4.55 9 1.61
8 4 0.83 4 0.71
9 5 1.03 2 0.36
10 1 0.21 2 0.36
11 1 0.21 1 0.18
2 ⩽ Tenure ⩽ 5 371 76.65 465 83.03
Total 484 100 560 100

2.3. Corporate investment measures and other variables

We use the following three variables to measure the level, change, and volatility of corporate investment: INVi,t is corporate invest-
ment calculated as the firm's net fixed assets in year t, minus the net fixed assets in year t-1, plus depreciation in year t, scaled by total
assets in year t-115; ΔINVi,t is the change in corporate investment, calculated as INVi,t minus INVi,t − 1; and INV_Vi is the volatility of
corporate investment, calculated as the standard deviation of corporate investment scaled by its mean during the sample period.
Other variables used include SOE (a dummy variable indicating whether a firm is an SOE), sales growth, leverage, cash holdings,
firm age, total assets, and GDP growth rate (at the city level). We provide detailed definitions of these in Appendix A.

2.4. Descriptive statistics

Table 1, Panel A reports the distribution of sample observations by year from 2001 to 2009. The number of observations is
on average 113,369 per year, with the most (178,698) in 2007, and the least (59,864) in 2001. Panel B reports the distribution

15
Public firms in China report capital expenditures on their cash-flow statements. However, the CIED does not have this item in its database. Hence, we use this
method to estimate capital expenditures.
178 H. An et al. / Journal of Corporate Finance 36 (2016) 174–189

Fig. 1. Distribution of government-official turnover. This figure shows the distribution of government official turnover from 2001 to 2009. We consider four cases
of government-official turnover: (1) a change of a city's head; (2) a change of a city's mayor; (3) a change of head or mayor; and (4) a change of head and
mayor.

of government official turnover. The average turnover is 26.72% for party heads and 30.65% for city mayors, from 2001 to 2009.
The two largest were in 2003 and 2008, the two years following China's national elections (2002 and 2007). Fig. 1 shows the
pattern. Panel C reports the distribution of tenure of government officials. The majority of heads (76.65%) and mayors (83.03%)
in our sample held their positions between two and five years, concurrent with the fact that Chinese government officials typically
serve five-year terms.
Table 2 reports the descriptive statistics for the major variables used in this study. All continuous variables (Panel A) are
Winsorized at the top and bottom 1% each year to mitigate the effect of outliers. Panel A shows a large variation of corporate
investment across the sample: the average corporate investment accounts for 11.29% of prior-year total assets, with a median
of 2.68% and a standard deviation of 31.53%. The year-over-year change and the volatility of corporate investment also vary greatly
during the sample period. CHG_FREQi, which measures government official turnover frequency, is 0.1889 on average, with a standard
deviation of 0.2390.
Regarding financial characteristics, Panel A shows that the sample firms generally experience high sales growth, with an aver-
age growth rate of 39.14% and a median of 16.66%. They are highly levered, and hold about 21.02% of their total assets as cash.
Average total assets and firm age are RMB9.9628 billion (equivalent to USD1.46 billion) and 11.34 years, respectively.
Panel B shows summary statistics of the dummy variables. More than 60% of the sample firms are SOEs. Turnover due to external
appointments is 56.59%, and the remaining 43.41% is due to local promotions.

Table 2
Summary statistics.
This table presents summary statistics of major continuous variables (Panel A) and dummy variables (Panel B). INVi,t or corporate investment in year t is calculated as
firm net fixed assets in year t, minus net fixed assets in prior year t-1, plus depreciation in year t, scaled by total assets in prior year t-1. ΔINVi,t is the change of corporate
investment in year t, calculated as ΔINVi,t = INVi,t–INVi,t − 1. Corporate investment volatility or INV_Vi = standard deviation of corporate investment divided by the av-
erage corporate investment during the sample period. CHG_FREQi is a measure of the degree of political turnover, calculated as the frequency of government-official
change for a city during the sample period. EXT_APPi,t is a second measure of the degree of political turnover. It equals 1 for an external appointment, and 0 for a local
promotion. The definitions of other variables are provided in Appendix A.

Panel A: continuous variables

Mean SD 25% Mdn 75% N

INVi,t 0.1129 0.3153 0.0003 0.0268 0.1165 1,020,321


ΔINVi,t −0.0086 0.4566 −0.0811 −0.0005 0.0738 1,020,321
INV_Vi 0.1889 0.239 0.0414 0.1016 0.2264 166,348
CHG_FREQi 0.3838 0.1851 0.2500 0.3333 0.5000 166,348
GROWTHi,t − 1 0.3914 0.9436 −0.0306 0.1666 0.4766 1,020,321
LEVi,t − 1 0.5779 0.2777 0.3792 0.5911 0.7775 1,020,321
CASHi,t − 1 0.2102 0.1885 0.0637 0.1591 0.3071 1,020,321
FIRMAGEi,t − 1 11.3444 10.4404 5.0000 8.0000 13.0000 1,020,321
SIZEi,t − 1 9.9628 1.4020 8.9641 9.7823 10.7919 1,020,321
INVi,t − 1 0.1216 0.3367 0.0007 0.0300 0.1243 1,020,321
GDP_Gi,t 0.1373 0.0285 0.1210 0.1400 0.1550 1,020,321
INV_GDPi,t 0.4414 0.1660 0.3173 0.4319 0.5431 1,020,321

Panel B: Dummy variables

Dummy variable = 1 Dummy variable = 0 Total no. of observations

SOEi,t 635,944(62.33%) 384,377(37.67%) 1,020,321(100%)


EXT_APPi,t 219,182(56.59%) 168,116(43.41%) 387,298(100%)
H. An et al. / Journal of Corporate Finance 36 (2016) 174–189 179

Table 3
Univariate analysis of the effect of political change on corporate investment.
Panel A compares the average corporate investment in years with political changes versus the years without official changes. Panel B shows corporate investment sur-
rounding government-official change in year window (−2, +2) where year 0 is defined as the event year. Corporate investment in year t (INVi,t) is calculated as firm net
fixed assets in year t, minus net fixed assets in prior year t-1, plus depreciation in year t, scaled by total assets in prior year t-1. Panel C compares the corporate invest-
ment growth in years with internal promotion vs. external appointment and investment volatility for high vs. low degree of political turnover. The high and low degree
of political turnover is determined using the median level of CHG_FREQi. The definitions of all variables are provided in Appendix A. The symbols *** and ** indicate sig-
nificance at 1% and 5%, respectively.

Panel A: corporate investment in change years vs. none-change years

Head change INVi,t Mayor change INV_Vi,t

CHANGE_HEADi,t = 1 0.1118 CHG_MAYORi,t = 1 0.1098


CHANGE_HEADi,t = 0 0.1133 CHG_MAYORi,t = 0 0.1141
Difference −0.0015** Difference −0.0043***
t-statistic −2.0374 t-statistic −6.1594
CHGi,t = 1 0.1097 CHG_BOTHi,t = 1 0.1135
CHGi,t = 0 0.1149 CHG_BOTHi,t = 0 0.1128
Difference −0.0052*** Difference 0.0007
t-statistic −8.0759 t-statistic 0.8269

Panel B: corporate investment surrounding government-official change

Head change Mayor change

Year—2 0.1212 0.1132


Year—1 0.1111 0.1041
Year 0 0.1073 0.1079
Year +1 0.1581 0.1617
Year +2 0.1339 0.1458

Panel C: comparing different types of political turnover and turnover frequency

Type of change ΔINVi,t High vs. low turnover INV_Vi

Internal promotion −0.0012 CHG_FREQi N median 0.1918


External appointment −0.0064 CHG_FREQi ⩽ median 0.1860
Difference 0.0052 Difference 0.0058
t-statistic 3.5865*** t-statistic 4.9346***

3. Empirical analysis and results

In this section, we present our results on changes in corporate investment in the years of turnover in government officials. We
begin the empirical tests with a univariate analysis, followed by a regression analysis. The effects of both the nature and degree of
political turnover on corporate investment are thoroughly examined. We then proceed to investigate how firm ownership struc-
ture affects the relationship between political turnover and corporate investment. Finally, we conduct a battery of robustness
checks.

3.1. Univariate analysis

We first conduct a univariate analysis to examine how political turnover affects corporate investment. In Panel A of Table 3, we
compare the average corporate investment in years with government official changes with that of years without official changes.

Fig. 2. Corporate investment surrounding city head change. This figure shows corporate investment surrounding government official change in window (−2, +2).
Year 0 is defined as the local official's turnover year t or event year. Corporate investment (INVi,t) in year t is calculated as firm net fixed assets in year t, minus net
fixed assets in prior year t-1, plus depreciation in year t, scaled by total assets in prior year t-1.
180 H. An et al. / Journal of Corporate Finance 36 (2016) 174–189

Table 4
Relationship between political turnover and corporate investment.
This table reports the regression results of corporate investment on political turnover. The dependent variable is corporate investment (INVi,t) in Panel A, and the change
in corporate investment (ΔINVi,t) in Panel B. INVi,t is calculated as firm net fixed assets in year t, minus net fixed assets in prior year t-1, plus depreciation in year t, scaled
by total assets in prior year t-1.ΔINVi,t is the change of corporate investment in year t or ΔINVi,t = INVi,t- INVi,t-1. We use four measures of political turnover: CHG_HEADi,t,
CHG_MAYORi,t, CHGi,t, and CHG_BOTHi,t. Control variables include firm sales growth, leverage, cash holdings, age, size, prior year investment, city GDP growth rate, and
city total investment as a percentage of its GDP. The detailed definitions of all variables are provided in Appendix A. The table also presents the number of observations
and the R2 for each regression. Robust t statistics are corrected for clustering of the residual at the firm level. One asterisk indicates significance at the 10% level, two
indicate significance at the 5% level, and three indicate significance at the 1% level.

Panel A: corporate investment

Variable INVi,t

(1) (2) (3) (4)

CHANGE_HEADi,t −0.006***
(−8.51)
CHANGE_MAYORi,t −0.006***
(−8.79)
CHGi,t −0.005***
(−7.37)
CHG_BOTHi,t −0.011***
(−11.67)
GROWTHi,t-1 0.011*** 0.011*** 0.011*** 0.011***
(19.23) (19.20) −19.21 (19.23)
LEVi,t − 1 0.047*** 0.047*** 0.047*** 0.047***
(15.44) (15.44) −15.43 (15.46)
CASHi,t − 1 0.167*** 0.167*** 0.167*** 0.167***
(54.29) (54.31) −54.31 (54.28)
FIRMAGEi,t − 1 −0.000 −0.000 −0.000 −0.000
(−1.37) (−1.39) (−1.38) (−1.38)
SIZEi,t − 1 −0.214*** −0.214*** −0.214*** −0.213***
(−130.68) (−130.68) (−130.70) (−130.63)
INVi,t − 1 −0.182*** −0.182*** −0.182*** −0.182***
(−94.84) (−94.84) (−94.82) (−94.87)
GDP_Gi,t 0.521*** 0.529*** 0.525*** 0.526***
(22.05) (22.40) −22.2 (22.25)
INV_GDPi,t − 1 0.100*** 0.100*** 0.100*** 0.099**
(17.39) (17.32) -17.44 (17.09)
Year effect Control Control Control Control
Firm effect Control Control Control Control
N 1,020,321 1,020,321 1,020,321 1,020,321
Within R2 0.154 0.154 0.154 0.154

Panel B: change in corporate investment

Variable ΔINVi,t

(1) (2) (3) (4)

CHANGE_HEADi,t −0.006***
(−8.72)
CHANGE_MAYORi,t −0.006***
(−8.82)
CHGi,t −0.005***
(−7.41)
CHG_BOTHi,t −0.011***
(−11.91)
GROWTHi,t − 1 0.011*** 0.011*** 0.011*** 0.011***
(18.99) (18.96) (18.97) (18.99)
LEVi,t − 1 0.046*** 0.046*** 0.046*** 0.046***
(15.03) (15.03) (15.02) (15.05)
CASHi,t − 1 0.172*** 0.172*** 0.172*** 0.172***
(55.24) (55.25) (55.25) (55.22)
FIRMAGEi,t − 1 −0.000 −0.000 −0.000 −0.000
(−1.25) (−1.27) (−1.26) (−1.26)
SIZEi,t − 1 −0.207*** −0.207*** −0.208*** −0.207***
(−128.91) (−128.90) (−128.92) (−128.86)
INVi,t − 1 −1.170*** −1.170*** −1.170*** −1.170***
(−600.72) (−600.68) (−600.69) (−600.75)
GDP_Gi,t 0.520*** 0.528*** 0.524*** 0.524***
(21.96) (22.32) (22.12) (22.16)
INV_GDPi,t 0.095*** 0.094*** 0.095*** 0.093***
(16.62) (16.55) (16.66) (16.32)
Firm effect Control Control Control Control
H. An et al. / Journal of Corporate Finance 36 (2016) 174–189 181

Table 4 (continued)

Panel B: change in corporate investment

Variable ΔINVi,t

(1) (2) (3) (4)

Year effect Control Control Control Control


N 1,020,321 1,020,321 1,020,321 1,020,321
Within R2 0.683 0.683 0.683 0.683

The average corporate investment is 0.1118 when a party head is replaced compared with 0.1133 when no such change occurs.
The difference is significant at the 5% level (t = −2.04). The results are similar when the mayor is changed, and when either the
party head or city mayor is changed.
Panel B of Table 3 shows corporate investment around the government official change window (−2, +2), where year 0 is the
turnover year, or event year. Corporate investment is shown to decline two years before any change in local government officials,
and is at its lowest level during the event year. Fig. 2 shows a similar pattern. These preliminary results show that corporate in-
vestment drops as a response to potential political turnover and will not recover until the uncertainty associated with political
turnover is resolved.
Panel C of Table 3 first compares different types of government official turnover. Internal promotion results in a 0.12% decline
in corporate investment, while an external appointment leads to a 0.64% decline. The difference is also statistically significant
(t = 3.59). We then categorize firms based on their turnover frequency during the sample period. A firm belongs to a high
(low) turnover group if its turnover frequency is above (below) the median turnover frequency of all sample firms. As Panel C
shows, corporate investment volatility is 19.18% for high turnover firms, compared with 18.60% for low turnover firms. The dif-
ference is significant with a t-statistic of 4.93.
To summarize, the results in Table 3 provide preliminary evidence that corporate investment declines when local government
official turnover occurs. The decline is greater for firms experiencing a high turnover frequency. External appointments also have
more effect on corporate investment than local promotions.

3.2. Baseline regression analysis

We use Eq. (1) to examine the effect of political turnover on corporate investment:

INV i;t ¼ α þ β1  TURNOVERi;t þ β2  GROWTH i;t−1 þ β3  LEV i;t−1


þ β4  CASHi;t−1 þ β5  FIRMAGEi;t−1 þ β6  SIZEi;t−1 ð1Þ
þ β7  INV i;t−1 þ β8  GDP Gi;t þ β9  INV GDP i;t þ εi;t

The dependent variable is corporate investment (INVi,t). The variable of interest is political turnover, TURNOVERi,t. We have, as
previously described, four measures of political turnover: CHG_HEADi,t, CHG_MAYORi,t, CHGi,t, and CHG_BOTHi,t. We expect the co-
efficients of these variables to be negative. We follow the literature by controlling for the following variables at the firm level: sale
growth rate, leverage, cash holdings, age, size, and corporate investment in the previous year.16 We also control for the GDP
growth rate of the city, and the investment's share of the GDP.
Panel A of Table 4 shows that the estimated coefficients of the four political turnover measures are all significantly negative at
the 1% level, which suggests that political turnover is negatively related to corporate investment. The estimated coefficients of
both CHG_HEADi,t and CHG_MAYORi,t are −.006; the estimated coefficient of CHGi,t is −.005, shown in column 3; and the estimat-
ed coefficient of CHG_BOTHi,t is −.011 in column 4, the highest among the four coefficients, which suggests that the effect of turn-
over is greatest when both the head and mayor are removed. This indicates that firms reduce their corporate investment more
significantly when both the city head and the mayor change.
The results for the control variables are generally consistent with expectations. Specifically, corporate investment is positively
associated with firm size, growth prospects, and cash holdings, but negatively related to firm age and previous investment. In ad-
dition, corporate investment is larger for firms from cities where the GDP growth rate is higher and the local economy is driven
by fixed asset investment.
In Panel B, the dependent variable is the change in corporate investment. The estimated coefficients of the four political turn-
over measures are all significantly negative, which suggests that changes of local government officials prompt firms to cut their
investment. Overall, these results are consistent with Dixit and Pindyck (1994), who predict that firms withhold investment
under uncertainty to maximize the option value of waiting for more information.

16
Julio and Yook (2012), Baker et al. (2013) and Durnev (2010).
182 H. An et al. / Journal of Corporate Finance 36 (2016) 174–189

3.3. Firms' capital intensity and bargaining power

We document a negative effect of political turnover on corporate investment. However, this effect may vary with different firm
characteristics, such as asset tangibility. In this section we discuss two additional tests, conducted to examine the effect of political
turnover on corporate investment for two types of firms, (1) capital intensive firms and (2) large and influential firms.
First, fixed asset investment of capital-intensive firms tends to be larger and requires government approval. Their investment
is thus more likely to be affected by political uncertainty. To measure capital intensity, we calculate the firm's fixed assets as a
percentage of total assets, and then rank each firm based on this percentage. CAPINTi,t is a dummy variable equal to one if a
firm belongs to a high capital intensity group and zero otherwise. We interact CAPINTi,t with our political turnover variable,
and expect a negative coefficient for the interaction term.
Table 5 reports the regression results. The interaction term between CAPINTi,t and CHGi,t is significantly negative at the 5% level,
shown in column 1, where the dependent variable is INVi,t, and in column 3, where the dependent variable is ΔINVi,t. We obtain
similar results when alternative measures of political turnover are used, such as CHG_HEADi,t, CHG_MAYORi,t, and CHG_BOTHi,t.
These results show that the effect of political turnover on corporate investment is stronger for high capital intensity firms and
weaker for low capital intensity firms.
Second, firms that contribute significantly to the local economy are likely to have political connections; thus, these firms are
more affected by political turnover. Moreover, these large and influential firms might reduce their investment as a bargaining
tool with the aim of obtaining favorable policies from new government officials. To quantify the local influence of a firm, we
first estimate the total revenue of each firm as a percentage of local GDP in each year, and then rank each one based on this per-
centage. BARGAINi,t is a dummy variable equal to one if a firm belongs to a high GDP contribution group, and zero otherwise. We
interact BARGAINi,t with our political turnover measure, and expect a negative coefficient for the interaction variable.
Columns 2 and 4 of Table 5 report the results. The estimated coefficient of the interaction term CHGi,t* BARGAINi,t is negative
and significant at the 1% level in column 2, where the dependent variable is INVi,t, and in column 4, where the dependent variable

Table 5
Firms capital intensity and bargaining power.
This table shows that the effect of political turnover on corporate investment varies with the firms capital intensity and bargaining power. The dependent variable is
corporate investment (INVi,t) in columns 1 and 2, and investment change (ΔINVi,t) in columns 3 and 4. Control variables include firm sales growth, leverage, cash hold-
ings, age, size, prior year investment, city GDP growth rate, and city total investment as a percentage of its GDP. Regressions control for both firm and time fixed-effects.
The table also presents the number of observations and the R2 for each regression. Robust t statistics are corrected for clustering of the residual at the firm level. One
asterisk indicates significance at the 10% level, two indicate significance at the 5% level, and three indicate significance at the 1% level.

Variable INVi,t ΔINVi,t

(1) (2) (3) (4)

CHGi,t −0.003*** −0.001* −0.004*** −0.002*


(−5.91) (−1.68) (−5.87) (−1.90)
CAPINTi,t 0.241*** 0.240***
(170.45) (170.23)
CHGi,t*CAPINTi,t −0.003** −0.003**
(−2.17) (−2.17)
BARGAINi,t 0.092*** 0.090***
(53.44) (52.57)
CHGi,t*BARGAINi,t −0.007*** −0.006***
(−5.31) (−5.10)
GROWTHi,t − 1 0.011*** 0.009*** 0.011*** 0.009***
(21.10) (15.84) (20.75) (15.68)
LEVi,t − 1 0.055*** 0.048*** 0.053*** 0.047***
(18.58) (15.83) (18.12) (15.41)
CASHi,t − 1 0.206*** 0.167*** 0.210*** 0.172***
(69.68) (54.48) (70.31) (55.42)
FIRMAGEi,t − 1 −0.000 −0.000 −0.000 −0.000
(−0.49) (−1.29) (−0.35) (−1.17)
SIZEi,t − 1 −0.218*** −0.220*** −0.211*** −0.214***
(−136.15) (−134.10) (−134.08) (−132.28)
INVi,t − 1 −0.189*** −0.181*** −1.177*** −1.169***
(−101.94) (−94.44) (−619.62) (−600.60)
GDP_Gi,t 0.502*** 0.497*** 0.501*** 0.496***
(21.73) (21.14) (21.64) (21.07)
INV_GDPi,t 0.089*** 0.092*** 0.084*** 0.087***
(15.80) (16.06) (14.99) (15.29)
Year effect Control Control Control Control
Firm effect Control Control Control Control
N 1,020,321 1,020,321 1,020,321 1,020,321
Within R2 0.223 0.159 0.708 0.685
H. An et al. / Journal of Corporate Finance 36 (2016) 174–189 183

Table 6
Turnover types: external appointment vs. local promotion.
This table compares the effect of change to government officials due to external appointment vs.
local promotion. The dependent variable is corporate investment (INVi,t) in column 1 and change of
corporate investment (ΔINVi,t) in column 2. The detailed definitions of all variables are provided in
Appendix A. Regressions control for both firm and time fixed-effects. The table also presents the num-
ber of observations and the R2 for each regression. Robust t statistics are corrected for clustering of the
residual at the firm level. One asterisk indicates significance at the 10% level, two indicate significance
at the 5% level, and three indicate significance at the 1% level.

Variable (1) (2)

INVi,t ΔINVi,t

EXT_APPi,t −0.003** −0.003**


(−2.26) (−2.16)
GROWTHi,t − 1 0.012*** 0.013***
(10.06) (10.60)
LEVi,t − 1 0.038*** 0.036***
(6.35) (6.10)
CASHi,t − 1 0.153*** 0.159***
(24.33) (24.93)
FIRMAGEi,t − 1 −0.000 −0.000
(−1.15) (−1.13)
SIZEi,t − 1 −0.190*** −0.185***
(−64.32) (−63.06)
INVi,t − 1 −0.150*** −1.134***
(−35.08) (−260.42)
GDP_Gi,t 0.565*** 0.569***
(13.02) (13.19)
INV_GDPi,t 0.059*** 0.057***
(5.75 (5.55)
Year effect Control Control
Firm effect Control Control
N 387,298 387,298
Within R2 0.125 0.650

is ΔINVi,t.17 These results show that the effect of political turnover on corporate investment is stronger for firms that are large con-
tributors to a local economy.
Overall, the results reported in Table 5 provide strong and consistent evidence that the effect of changes to local government
officials on corporate investment largely depends on a firm's capital intensity and its bargaining power with local government.
The relationship is significantly greater for firms that are capital intensive and who make significant contributions to the local
economy.

3.4. Turnover types: external appointment vs. local promotion

We have shown that firms reduce their capital expenditure as a response to local government official changes. In this section,
we further distinguish between the types of turnover. We expect the effect of political turnover to be more pronounced if the
local government officials are replaced by outsiders. Table 6 shows only observations when changes in government officials
occur, including both external appointments and local promotion. EXT_APPi,t is a dummy variable indicating the type of the ap-
pointment; this equals one for external appointments and zero for local promotions. The dependent variable is INVi,t in column
1 and ΔINVi,t in column 2. Control variables are the same as in Eq. (1). Table 6 shows that the estimated coefficients of EXT_APPi,t
are significantly negative, which suggests that there is a greater decrease in corporate investment when changes in government
officials are external. These results show that the effect of political turnover on corporate investment is greater for external ap-
pointments, which generally represent a high degree of political uncertainty.

3.5. Turnover frequency and investment volatility

In this section, we calculate the turnover frequency for each firm during the sample period and then regress the volatility of
corporate investment on turnover frequency.

INV V i ¼ α þ β1  CHG FREQ i þ β2  GROWTH V i þ β3  LEV V i


þ β4  CASH V i þ β5  FIRMAGE V i þ β6  SIZE V i ð2Þ
þ β7  GDP V i þ β8  INV GDP V i þ εi

17
We obtain similar results when alternative measures of political turnover are used.
184 H. An et al. / Journal of Corporate Finance 36 (2016) 174–189

Specifically, CHG_FREQi is the number of government official changes that a firm experiences during the sample period. INV_Vi
is the standard deviation of corporate investment divided by the average corporate investment during the sample period. The
control variables are similar to those used in Eq. (1), but converted into their corresponding volatility measures. For example,
sale growth rate (GROWi,t − 1) is replaced by the volatility of growth rate (GROW_Vi), calculated as the standard deviation of
the sale growth rate divided by the average growth rate during the sample period.
The regression results are reported in Table 7. The estimated coefficient of CHG_FREQi is positive and significant at the 1% level,
which is evidence that high turnover frequency is associated with more volatile corporate investment.

3.6. Firms' ownership structure

In this section we examine the effects of firm ownership structure on the relationship between corporate investment and po-
litical turnover. A distinctive feature of Chinese firms is concentrated ownership, with the government as their largest shareholder
(Claessens et al., 2000; Sun and Tong, 2003). Appointed by the government, executives of SOEs generally carry bureaucratic ranks.
The aim of SOE managers is often to promote government policies and not necessarily to maximize shareholders' wealth
(Lin et al., 2012). Thus, we create a dummy variable SOE indicating whether a firm is owned and controlled by the government.
Next, we interact SOE with the political turnover measures. We expect political turnover to have a greater effect on SOEs.
Table 8 reports the regression results. In Panel A, the dependent variable is INVi,t in column 1, and ΔINVi,t in column 2. Control
variables are the same as in Eq. (1). In both columns, the coefficient of the interaction term CHGi,t * SOEi,t is significantly negative,
evidence that SOEs reduce corporate investment more than non-SOEs after changes in local government officials.
In Panel B, we use Eq. (2) to compare the investment volatility of SOEs with that of non-SOEs. The dependent variable is the
volatility of corporate investment, INV_Vi, in column 1, and control variables are the same as in Eq. (2). Table 6 shows that the
coefficient of the interaction term CHGt * SOEi,t is significantly negative, evidence that SOEs reduce corporate investment more
than non-SOEs after changes in local government officials.
The coefficient for the interaction term SOEi,t * CHG_FREQi is positive and significant at the 5% level (β = 0.017, t = 2.14). This
suggests that the positive relationship between political turnover frequency and corporate investment volatility is stronger for
SOEs.

3.7. Firm performance and welfare implications

So far, our results show that corporate investment declines following political turnover. In this section, we address the welfare
implications of investment reduction. Because overinvestment is perceived to be prevalent in China, less wasteful investment
should not hurt firm performance and may even have positive welfare implications. Therefore, we proceed to examine the impact
of investment decline due to political turnover on firm performance.

Table 7
Relationship between political turnover frequency and corporate investment volatility.
This table shows the regression results of investment volatility on the government official change fre-
quency (CHG_FREQi). The dependent variable is INV_Vi, calculated as the standard deviation of corporate
investment divided by the average corporate investment in the sample period. The detailed definitions
of all variables are provided in Appendix A. Regressions control for time fixed-effects. The table also pre-
sents the number of observations and the R2 for each regression. Robust t statistics are corrected for clus-
tering of the residual at the firm level. One asterisk indicates significance at the 10% level, two indicate
significance at the 5% level, and three indicate significance at the 1% level.

Variable INV_Vi

CHG_FREQ i 0.011***
(3.65)
GROWTH_Vi −0.000
(−1.33)
LEV_Vi 0.131***
(48.10)
CASH_Vi 0.010***
(3.27)
FIRMAGE_Vi 0.067***
(45.39)
FSIZE_Vi 3.719***
(122.80)
GDP_G_Vi −0.182***
(−33.54)
INV_GDP_Vi −0.034***
(−8.93)
Industry effect Control
N 166,348
Adj. R2 0.289
H. An et al. / Journal of Corporate Finance 36 (2016) 174–189 185

Table 8
The ownership structure of firms: SOE vs. non-SOE.
This table examines the relationship between the firm ownership structure and corporate investment. In Panel A, the dependent variables are corporate investment
(INVi,t) and change of corporate investment (ΔINVi,t); in Panel B, the dependent variable is volatility of corporate investment (INV_Vi). SOE is a dummy variable in-
dicating the firm is a state-owned enterprise. The detailed definitions of all variables are provided in Appendix A. The panels also present the number of observations
and the R2 for each regression. Robust t statistics are reported in both panels. Standard errors are corrected for clustering at the firm level in Panel A. One asterisk
indicates significance at the 10% level, two indicate significance at the 5% level, and three indicate significance at the 1% level.

Panel A: corporate investment

Variable (1) (2)


INVi,t ΔINVi,t

CHGi,t 0.001 0.001


(0.82) (0.90)
SOEi,t 0.024*** 0.024***
(15.34) (15.21)
CHGi,t* SOEi,t −0.009*** −0.009***
(−7.17) (−7.31)
GROWTHi,t − 1 0.011*** 0.011***
(19.31) (19.07)
LEVi,t − 1 0.048*** 0.046***
(15.68) (15.26)
CASHi,t − 1 0.167*** 0.172***
(54.28) (55.22)
FIRMAGEi,t − 1 −0.000* −0.000
(−1.76) (−1.63)
SIZEi,t − 1 −0.215*** −0.208***
(−130.90) (−129.12)
INVi,t − 1 −0.181*** −1.169***
(−94.62) (−600.21)
GDP_Gi,t 0.526*** 0.525***
(22.27) (22.18)
INV_GDPi,t 0.105*** 0.099***
(18.16) (17.40)
Year effect Control Control
Industry effect Control Control
N 1,020,321 1,020,321
Adj. R2 0.043 0.543

Panel B: corporate investment volatility

Variable INV_Vi

CHG_FREQ i 0.020***
(2.97)
SOEi 0.030***
(8.59)
SOEi* CHG_FREQ i 0.017**
(2.14)
GROWTH_Vi −0.000
(−0.70)
LEV_Vi 0.117***
(29.39)
CASH_Vi 0.036***
(7.59)
FIRMAGE_Vi 0.061***
(28.97)
FSIZE_Vi 4.160***
(85.83)
GDP_G_Vi −0.219***
(−28.28)
INV_GDP_Vi 0.033***
(5.73)
Industry effect Control
N 83,739
Adj. R2 0.303

If firms invest efficiently, then the reduction in investment due to political turnover will result in missed investment opportu-
nities, which will ultimately hurt firm performance. On the other hand, if firms overinvest, then less overinvestment may help
better align corporate investment with investment opportunities. Overinvestment is particularly a concern for SOEs, whose top
executives are appointed by the government. Besides economic reasons, SOEs also invest to accomplish social and political
goals under government intervention. Liu and Siu (2012) find that the internal rate of return of private firms in China is 10%
186 H. An et al. / Journal of Corporate Finance 36 (2016) 174–189

Table 9
Firm performance.
This table examines the relationship between the firms profitability and corporate investment. The de-
pendent variable is return on assets. Column 1 reports the result for private firms, whereas column 2
does that for SOEs. The detailed definitions of all variables are provided in Appendix A. The panels also
present the number of observations and the R2 for each regression. Robust t statistics are reported
in both panels. Standard errors are corrected for clustering at the firm level. One asterisk indicates
significance at the 10% level, two indicate significance at the 5% level, and three indicate significance
at the 1% level.

Variable (1) (2)


Private SOE

CHGi,t −0.000 −0.000


(−0.58) (−1.57)
INVi,t 0.007*** −0.001
(3.78) (−0.50)
CHGi,t* INVi,t −0.010*** 0.001
(−3.74) (0.55)
GROWTHi,t − 1 0.007*** 0.006***
(16.27) (15.91)
LEVi,t − 1 0.009*** 0.009***
(4.49) (4.69)
CASHi,t − 1 0.004 0.008***
(1.62) (4.66)
FIRMAGEi,t − 1 −0.000*** −0.000***
(−4.09) (−3.25)
SIZEi,t − 1 −0.035*** −0.040***
(−31.31) (−39.95)
Year effect Control Control
Firm effect Control Control
N 290,037 481,318
Within R2 0.022 0.022

higher than that of an SOE. Chen et al. (2011) find that the sensitivity of investment to investment opportunities is significantly
higher for private firms than SOEs.
In Table 9, we examine the impact of investment decline due to political turnover on firm profitability as measured by return
on assets.18 Column 1 shows the result of private firms, whereas column 2 shows that of SOEs. As Table 9 shows, the correlation
between corporate investment and firm profitability is significantly positive for private firms in column 1, but the correlation
turns negative (not statistically significant at conventional levels) for SOEs in column 2. Moreover, when we interact corporate
investment with political turnover, we find that the interaction term is significantly negative in column 1 for private firms, but
not significant in column 2 for SOEs. The results suggest that private firms invest more efficiently than SOEs, and the decrease
in investment due to political turnover results in missed investment opportunities for private firms, which eventually hurts
their performance. Given the outsized contribution of private firms to the economic reform in China, our findings show that po-
litical uncertainty takes a toll on an important engine of China's economic growth.

3.8. Robustness

In this section, we conduct robustness tests, confirming that our results hold after correcting for potential endogeneity and re-
moving the election effect.

3.8.1. Endogeneity
Typically, the tenure of government officials in China is fixed, and is not affected by local economic conditions. City mayors and
heads normally serve five-year terms, and then either step down or are promoted to higher level positions. However, it is still a
concern that government official turnover at the city level could be driven by local economic performance.19 Thus, our empirical
framework is not completely immune to the potential endogeneity issue, in that the replacement of government officials could be
driven by worsening local economic situations.
To address the potential endogeneity issue, we use an instrument variable (IV) approach. Our three IVs are: local government
officials' tenure, age, and education level prior to the turnover year.20 These affect the probability of government official changes; but
are not directly related to corporate investment. We first regress CHG_HEADi,t (CHG_MAYORi,t) on the three IVs and control variables
used in Eq. (1). Columns 1 and 2 of Table 10 report the first-stage regression results. All three IVs are significantly related to

18
Because most of our sample firms are not listed on stock exchanges, we can't use stock price based firm performance measure such as Tobins Q.
19
It is reported by Li and Zhou (2005) that the likelihood of promotion (termination) of government officials at the provincial level increases (decreases) with local
economic performance.
20
This information is manually collected from the resumes of these heads and mayors.
H. An et al. / Journal of Corporate Finance 36 (2016) 174–189 187

Table 10
Instrument variable regression.
This table shows the 2SLS regression results. The IVs for political turnover are the tenure, age, and education level of the government officials. In the first stage, we re-
gress CHAG_HEADi,t (CHG_MAYORi,t) on the three IVs and control variables. In the second stage, we replace CHAG_HEADi,t (CHG_MAYORi,t) by its predicated value from
the first stage. The detailed definitions of all variables are provided in Appendix A. Regressions control for both firm and time fixed-effects. The table also presents the
number of observations and the R2 for each regression. Robust t statistics are corrected for clustering of the residual at the firm level. One asterisk indicates significance
at the 10% level, two indicate significance at the 5% level, and three indicate significance at the 1% level.

Variable First stage Second stage

CHAG_HEADi,t CHG_MAYORi,t INVi,t ΔINVi,t

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

CHG_HEADi,t −0.020*** −0.021***


(−12.39) (−13.06)
CHG_MAYORi,t −0.039*** −0.038***
(−21.62) (−21.29)
TENURE_Si,t − 1 0.157***
(381.93)
AGE_Si,t − 1 −0.011***
(−45.17)
EDU_Si,t − 1 0.019***
(37.73)
TENURE_Mi,t − 1 0.161***
(351.77)
AGE_Mi,t − 1 −0.011***
(−15.22)
EDU_Mi,t − 1 0.019***
(41.51)
GROWTHi,t − 1 −0.002*** −0.002*** 0.011*** 0.011*** 0.011*** 0.011***
(−4.00) (−5.37) (18.87) (18.91) (18.52) (18.63)
LEVi,t − 1 −0.007*** −0.007 0.048*** 0.048*** 0.046*** 0.046***
(−2.16) (−1.61) (15.14) (15.22) (14.69) (14.72)
CASHi,t − 1 −0.020*** −0.020* 0.168*** 0.168*** 0.173*** 0.173***
(−6.05) (1.69) (53.61) (53.03) (54.46) (53.86)
FIRMAGEi,t − 1 −0.000** −0.000*** −0.000 −0.000* −0.000 −0.000
(−2.17) (−3.60) (−1.46) (−1.67) (−1.39) (−1.62)
SIZEi,t − 1 −0.006*** −0.006*** −0.215*** −0.216*** −0.209*** −0.210***
(4.43) (10.19) (−128.44) (−127.29) (−126.29) (−125.23)
INVi,t − 1 −0.000 −0.000 −0.182*** −0.184*** −1.170*** −1.172***
(−0.33) (−1.26) (−92.99) (−93.86) (−588.26) (−586.94)
GDP_Gi,t −0.797*** −0.797*** 0.555*** 0.543*** 0.554*** 0.541***
(−24.06) (−15.29) (22.38) (22.07) (22.24) (21.94)
INV_GDPi,t −0.004*** −0.004* 0.097*** 0.100*** 0.091*** 0.095***
(−5.26) (1.71) (16.19) (16.73) (15.37) (15.96)
Year effect Control Control Control Control Control Control
Firm effect Control Control Control Control Control Control
N 913089 899962 913089 899962 913089 899962
Centered R2 0.334 0.299 0.154 0.154 0.684 0.685

government official changes. Specifically, city heads and mayors are more likely to be replaced if they are young, have longer tenures,
and have higher levels of education. The p-values of the under identification tests are less than 1%, which show that the IVs are rele-
vant. The Kleibergen–Paap rk Wald F statistics are higher than the Stock–Yogo weak ID test critical values at 10%, which also suggests
that the IVs are not weak.
In the second-stage regression, we replace CHG_HEADi,t (CHG_MAYORi,t) by its predicated value from the first-stage regres-
sion. The results reported in columns 3–6 suggest that the negative relationship between political turnover and corporate invest-
ment continues to hold after correcting for any potential endogeneity bias.

3.8.2. Removing the national-election-year effect


There were two national elections during our sample period (2002 and 2007). To address the concern that firms may adjust
their corporate investment in anticipation of these elections and of any resulting political turnover, we remove the observations in
2002 and 2007. Table 11 shows that, during non-election years, firms still cut their investment when government officials change.

4. Conclusions

Using hand-collected data on government official changes for 277 cities in China, we examine the effect of political turnover
on corporate investment in one of the world's largest developing economies. We construct a rich set of variables that captures
both the nature and degree of political turnover. Our results suggest that the turnover of local government officials significantly
deters corporate investment. This effect is stronger for capital intensive firms and for those who have more bargaining power
188 H. An et al. / Journal of Corporate Finance 36 (2016) 174–189

Table 11
Relationship between political turnover and corporate investment in non-election years.
This table shows the results after removing observations in two election years (2002 and 2007). The detailed definitions of all variables are provided in Appendix A.
Regressions control for both firm and time fixed-effects. The table also presents the number of observations and the within R2 for each regression. Robust t statistics
are corrected for clustering of the residual at the firm level. One asterisk indicates significance at the 10% level, two indicate significance at the 5% level, and three
indicate significance at the 1% level.

Variable INVi,t

(1) (2) (3) (4)

CHG_HEADi,t −0.002**
(−2.11)
CHG_MAYORi,t −0.003***
(−3.59)
CHGi,t −0.002*
(−1.89)
CHG_BOTHi,t −0.006***
(−4.73)
GROWTHi,t − 1 0.013*** 0.013*** 0.013*** 0.013***
(18.29) (18.27) (18.29) (18.28)
LEVi,t − 1 0.047*** 0.047*** 0.047*** 0.047***
(12.65) (12.65) (12.65) (12.66)
CASHi,t − 1 0.164*** 0.164*** 0.164*** 0.164***
(44.44) (44.45) (44.44) (44.44)
FIRMAGEi,t − 1 −0.000 −0.000 −0.000 −0.000
(−0.93) (−0.94) (−0.93) (−0.95)
SIZEi,t − 1 −0.214*** −0.214*** −0.214*** −0.214***
(−111.08) (−111.09) (−111.09) (−111.07)
INVi,t − 1 −0.184*** −0.184*** −0.184*** −0.184***
(−73.39) (−73.39) (−73.39) (−73.39)
GDP_Gi,t 0.395*** 0.397*** 0.396*** 0.395***
(14.23) (14.34) (14.27) (14.27)
INV_GDPi,t 0.113*** 0.113*** 0.113*** 0.112***
(16.08) (16.04) (16.10) (15.95)
Year effect Control Control Control Control
Firm effect Control Control Control Control
N 770,484 770,484 770,484 770,484
Within R2 0.158 0.158 0.158 0.158

with local governments. Compared with local promotions, external appointments have a more pronounced effect on corporate in-
vestment. Furthermore, we find that a high degree of political turnover (as measured by the frequency of political changes) is as-
sociated with more volatile firm-level capital investment. Finally, we investigate the effects of firm ownership structure on the
relationship between corporate investment and political turnover; we find that the effect of political turnover on corporate invest-
ment is stronger for SOEs. These findings provide a dynamic picture of the investment behavior of Chinese firms when faced with
political turnover. The results hold for both univariate analyses and multivariate regressions, and are robust after addressing the
election effect and potential endogeneity.

Appendix A. Variable definitions

CHG_HEADi,t A dummy variable coded one if there is a change of CCP head in the period from July in year t-1 to June in year t for
a city, and zero otherwise.
CHG_MAYORi,t A dummy variable coded one if there is a change of CCP mayor in the period from July in year t-1 to June in year
t for a city, and zero otherwise.
CHGi,t A dummy variable coded one if there is a change of CCP head or mayor in the period from July in year t-1 to June in
year t for a city, and zero otherwise.
CHG_BOTHi,t A dummy variable coded one if there is a change of CCP head and mayor in the period from July in year t-1 to June
in year t for a city, and zero otherwise.
CHG_FREQ i A measure of the degree of political turnover, calculated as the frequency of changes of government officials for a city
during the sample period.
EXT_APPi,t Another measure of the degree of political turnover (coded one for an external appointment and zero for a local
promotion). For an external appointment, an official is appointed from another city by a higher level of government.
Local promotion is the promotion of an official by the government of the same city.
INVi,t Corporate investment in year t; calculated as firm net fixed assets in year t, minus net fixed assets in prior year t-1, plus
depreciation in year t, scaled by total assets in year t-1.
ΔINVi,t Change of corporate investment in year t; ΔINVi,t = INVi,t − INVi,t − 1.
INV_Vi Corporate investment volatility; calculated as the standard deviation of corporate investment divided by the average
corporate investment during the sample period.
H. An et al. / Journal of Corporate Finance 36 (2016) 174–189 189

GROWTHi,t − 1 Sales growth rate of the firm in year t-1, calculated as sales in year t-1 minus sales in year t-2, scaled by sales in year t-2.
LEVi,t − 1 Firm leverage ratio in year t-1, calculated as total debt (including current portion of long term debt) in year t-1 scaled
by total assets in year t-1.
CASHi,t − 1 Cash holding in year t-1, calculated as the current asset in year t-1 minus accounts receivable in year t-1 and inventory
in year t-1, scaled by total assets in year t-1.
FIRMAGEi,t − 1 Firm age at year t-1.
SIZEi,t − 1 The natural logarithm of firm total assets in year t-1.
INVi,t − 1 Corporate investment in year t-1.
GDP_Gi,t A city's GDP growth rate in year t.
INV_GDPi,t A city's total fixed capital investment as a percentage of its GDP in year t.
SOEi,t A dummy variable coded 1 for an SOE, and zero for a non-SOE.
CAPINTi,t A measure of firm capital intensity. We first estimate firm fixed assets as a percentage of total assets for each firm i in
year t; and then rank each firm based on this percentage into high and low capital intensity groups. CAPINT is a dummy
variable equal to one if a firm belongs to a high capital intensity group and zero otherwise.
BARGAINi,t A measure of firm bargaining power with local government. We first estimate the total revenue of firm i as a percentage of
local GDP in year t and then rank each firm based on this percentage into two groups: high and low GDP contribution
groups. BARGAIN is a dummy variable equal to one if a firm belongs to a high GDP contribution group and zero otherwise.

References

Bai, C., Hsieh, C., Song, Z., 2014. Crony Capitalism with Chinese Characteristics. University of Chicago, working paper.
Baker, S., Bloom, N., Davis, S., 2013. Measuring economic policy uncertainty. Chicago Booth Research Paper 75 pages.
Bardhan, P., Mookherjee, D., 2006. The rise of local governments: an overview. In: Bardhan, P., Mookherjee, D. (Eds.), Decentralization and Local Governance in
Developing Countries: A Comparative Perspective. MIT Press, Cambridge, MA, pp. 1–52.
Bernanke, B.S., 1983. Irreversibility, turnover, and cyclical investment. Q. J. Econ. 98 (1), 85–106.
Blanchard, O., Shleifer, A., 2001. Federalism with and without political centralization: China vs. Russia in transitional economics: how much progress? IMF Staff. Pap.
48, 171–179.
Bloom, N., Bond, S., Van Reenen, J., 2007. Turnover and investment dynamics. Rev. Econ. Stud. 74 (2), 391–415.
Chen, Y., Li, H., Zhou, L., 2005. Relative performance evaluation and turnover of provincial leaders in China. Econ. Lett. 88 (3), 421–424.
Chen, S., Sun, Z., Tang, S., Wu, D., 2011. Government intervention and investment efficiency: evidence from China. J. Corp. Finance 17 (2), 259–271.
Claessens, S., Djankov, S., Lang, L., 2000. The separation of ownership and control in East Asian corporations. J. Financ. Econ. 58 (12), 81–112.
Cull, R., Xu, L., Yang, X., Zhou, L., Zhu, T., 2015. Market facilitation by local government and firm efficiency: evidence from China. J. Corp. Finance (forthcoming).
Dixit, A.K., Pindyck, R.S., 1994. Investment under uncertainty. Princeton University Press.
Duckett, J., 1998. The Entrepreneurial State in China: Real Estate and Commerce Departments in Reform Era Tianjin. Routledge, London and New York.
Duckett, J., 2001. Bureaucrats in business, Chinese-style: the lessons of market reform and state entrepreneurialism in the People's Republic of China. World Dev. 29
(1), 23–37.
Durnev, A., 2010. The real effects of political turnover: elections and investment sensitivity to stock prices. 2010 Finance Meeting EUROFIDAI-AFFI.
Fryer, T., Shleifer, A., 1997. The invisible hand and the grabbing hand. Am. Econ. Rev. 87 (2), 354–358.
Hsieh, C., Klenow, P., 2009. Misallocation and manufacturing TFP in China and India. Q. J. Econ. 124 (4), 1403–1448.
Huang, Y., 2008. Capitalism with Chinese Characteristics: Entrepreneurship and the State. Cambridge University Press, Cambridge, UK.
Jin, H., Qian, Y., Weingast, B., 2005. Regional decentralization and fiscal incentives: federalism, Chinese style. J. Public Econ. 89 (9–10), 1719–1742.
Julio, B., Yook, Y., 2012. Political turnover and corporate investment cycles. J. Financ. 67 (1), 45–83.
Leahy, J., Whited, T., 1996. The effect of turnover on investment: some stylized facts. J. Money, Credit, Bank. 28 (1), 64–83.
Li, C., Bachman, D., 1989. Localism, elitism, and immobilism: elite formation and social change in post-Mao China. World Polit. 42 (1), 64–94.
Li, H., Zhou, L., 2005. Political turnover and economic performance: the incentive role of personnel control in China. J. Public Econ. 89 (9–10), 1743–1762.
Li, H., Meng, L., Wang, Q., Zhou, L.-A., 2008. Political Connections, Financing and Firm Performance: Evidence from Chinese Private Firms. J. Dev. Econ 87, 283–299.
Lin, B., Lu, R., Zhang, T., 2012. Tax-induced earnings management in emerging markets: evidence from China. J. Am. Tax. Assoc. 34 (2), 19–44.
Liu, Q., Siu, A., 2012. Institutions and corporate investment: evidence from investment-implied return on capital in China. J. Financ. Quant. Anal. 46 (6), 1831–1863.
Maskin, E., Qian, Y., Xu, C., 2000. Incentives, scale economies, and organization forms. Rev. Econ. Stud. 67 (2), 359–378.
Montinola, G., Qian, Y., Weingast, B., 1995. Federalism, Chinese style: the political basis for economic success in China. World Polit. 48 (1), 50–81.
Qian, Y., Xu, C., 1993. Why China's economic reforms differ: The M-form hierarchy and entry/expansion of the non-state sector. Econ. Transit. 1 (2), 135–170.
Song, Z., Storesletten, K., Zilibotti, F., 2011. Growing like China. Am. Econ. Rev. 101 (1), 196–233.
Sun, Q., Tong, W., 2003. China share issue privatization: the extent of its success. J. Financ. Econ. 70 (2), 183–222.
Walder, A.G., 1995. Career mobility and the communist political order. Am. Sociol. Rev. 60 (3), 309–328.
Walder, A.G., 1996. Local governments as industrial firms: an organizational analysis of China's transitional economy. Am. J. Sociol. 101 (2), 263–301.
Whiting, S., 2001. Power and wealth in rural China: the political economy of institutional changes. Cambridge University Press, Cambridge.
Yu, T.F., 1997. Entrepreneurial state: the role of government in the economic development of the Asian newly industrializing economies. Dev. Policy Rev. 15 (1), 47–64.

You might also like