Does going public imply short-termism in investment behavior? Evidence from China

Emerging Markets Review xxx (xxxx) xxxx

Contents lists available at ScienceDirect

Emerging Markets Review journal homepage: www.elsevier.com/locate/emr

Does going public imply short-termism in investment behavior? Evidence from China Zhuangxiong Yu, Jie Zhang, Jie Li

⁎

Institute of Industrial Economics & Institute of Industrial Organization and Regulation, Jinan University, Guangzhou 510632, China

ARTICLE INFO

ABSTRACT

Keywords: Listed firms Short-termism Investment China

Adopting a large panel of Chinese manufacturing firms together with the data of listed firms for the period 1998–2007, this paper aims to examine whether Asker et al.'s (2015) argument on the short-termism demonstrated in the investment behavior of listed firms holds for China. We document that listed firms engage in more investments and respond more to changes in investment opportunities as compared to unlisted ones with similar size and age in the same industries, and Asker et al.'s (2015) short-termism argument only holds for firms facing sufficiently low financing constraints.

JEL codes: D22 D92 G31 G32 G34

1. Introduction Empirical research has established that listed firms' decision makers make distorted investment decisions due to short-termism pressures. Employing US firm-level data, Asker et al. (2015) show that in comparison to unlisted companies, listed companies make less investment and respond less to changes in investment opportunities. This is because listed firms' managers tend to increase current earnings and therefore the current share price by abandoning positive NPV (Net Present Value) projects (Graham et al., 2005). Asker et al. (2015) confirm the existence of short-termism in investment behavior in listed firms and is expected in developed and mature markets (e.g., the United States). However, the key determinants of corporate investment in developed economies are different from those in emerging economies (Bokpin and Onumah, 2009; Chirinko, 1993; Fazzari et al., 1988; Fazzari and Petersen, 1993; Geng and N'Diaye, 2012; Hall and Jorgenson, 1967; Jorgenson, 1971), which may alter the conditions for the occurrence of short-termism. This naturally raises the question: does Asker et al.'s (2015) argument on the short-termism in investment behavior of listed firms hold for emerging economies? Early studies identified the actual output and the availability of finance as the two main factors that determine corporate investment (Jorgenson, 1971). In other words, corporate investment decisions have to take into account both external financing constraints and internal performance. Moreover, financing constraints have more profound impacts on corporate investment in economies with less developed financial systems (Love and Zicchino, 2007). This paper extends the existing research to include China as the largest emerging economy and examine whether the shorttermism of firms' investment behavior also depends on financing constraints. The evidence presented in Allen et al. (2007) documents

⁎

Corresponding author. E-mail address: [email protected] (J. Li).

https://doi.org/10.1016/j.ememar.2019.100672 Received 28 October 2019; Received in revised form 19 November 2019; Accepted 9 December 2019 1566-0141/ © 2019 Elsevier B.V. All rights reserved.

Please cite this article as: Zhuangxiong Yu, Jie Zhang and Jie Li, Emerging Markets Review, https://doi.org/10.1016/j.ememar.2019.100672

Emerging Markets Review xxx (xxxx) xxxx

Z. Yu, et al.

a large but inefficient banking sector that dominates the financial system in China; it was not until the late 1990s that China's stock market emerged as a tool to alleviate corporate financing constraints. Even now, bank loans and stock market listing remain the two major channels of financing for Chinese firms, and the purpose of listing in China is still mainly to ease firms' financing constraints. By contrast, concerning the maturity of the financial system and the size of the stock market, the US financial market is more developed (Shan and Qi, 2006; Wong and Zhou, 2011), and there are many other financing channels available to firms in addition to bank loans and stock market listing. Thus, firms in U.S. may have diverse reasons for listing, e.g., creating listed shares for future acquisitions, allowing insiders to cash out, promoting takeover activities, strategic moves, etc. (Brau and Fawcett, 2006; Chemmanur and Fulghieri, 1999; Maksimovic and Pichler, 2001; Zingales, 1995). Moreover, the financing constraints that affect firms' motives for listing can also affect the investment behavior of listed firms. Rauh (2006) finds a negative correlation between investment and financing constraints and proves that financing constraints have a certain impact on investment decisions, which is consistent with Fazzari and Petersen (1993). Thus, due to differences in the financing constraints of Chinese and US firms, different trends are exhibited in the relationship between listing and investment observed in the two countries. The rationale for this is as follows: Chinese firms are faced with higher financing constraints than US firms. When Chinese firms are listed, they can raise more financing through investment opportunities that they would not have had otherwise. That is why Chinese firms invest more after listing. If a firm already has low financing constraints before listing, then it does not have unexploited investment opportunities. In this case, listing does not increase investment. By constructing a comprehensive indicator of financing constraints, this paper is the first empirical research (to the best of our knowledge) to show that Asker et al.'s (2015) argument does not hold for Chinese listed firms and to reveal that financing constraints can explain this difference. Specifically, the current paper is conducive for the literature in the following aspects. First, in sharp contrast to the existing evidence from the US market (e.g., Asker et al., 2015), we reveal new evidence to document that, as compared with unlisted firms, Chinese listed companies make more investments, and respond more to changes in investment opportunities, especially for those whose unlisted counterparts are faced with high financing constraints. Clearly, these findings enrich the current literature concerning firms' investment behavior in emerging markets. Second, for the first time, we document the channel of financing constraints through which the investment behavior of firms in emerging economies differs from that of firms in developed economies. That is, we demonstrate that the effect of listing on a firm's investment level and investment sensitivity is negative only when the firm's financing constraints are sufficiently low, which complements Asker et al.'s (2015) study, i.e., we show that their argument only holds for those listed firms whose unlisted counterparts are faced with sufficiently low financing constraints. Third, we explore a crucial country feature (i.e., state ownership) by extending previous studies (e.g., Chen et al., 2011; Lin and Tan, 1999) and document that state-owned listed companies respond more to changes in investment opportunities as compared to their unlisted counterparts, indicating greater advantages of going public for Chinese state-owned companies. Therefore, this new finding supplements the literature on corporate investment for firms owned by the state (e.g., Chen et al., 2011; O'Toole et al., 2016). Finally, we also explore the influences of product market competition on the different investment behaviors of listed and unlisted Chinese firms. By extending earlier studies on the subject (e.g., Akdoğu and MacKay, 2008; Laksmana and Yang, 2015), we report new evidence to show that when compared to unlisted firms with similar size in the same industries, firms in competitive industries respond more to changes in investment opportunities. The evidence thus also echoes the findings of Akdoğu and MacKay (2008), who show that companies in competitive industries exhibit higher investment sensitivity than those in non-competitive industries. We organize the rest of the paper as follows. We summarize relevant literature in Section 2. A description of the data and sample structure is conducted in Section 3. We illustrate the empirical model specifications and defines the main variables in Section 4. We then presents main empirical findings in Section 5. We introduces the heterogeneity analysis in Section 6. Section 7 reaches a conclusion. 2. Literature review The research on the impact of short-termism behavior on corporate investment has long fascinated economists, and many scholars agree that myopic short-termism may cause the problem of underinvestment. As argued by Holmstrom (1982), Narayanan (1985), and Miller and Rock (1985), decision makers tend to pursue short-term profits and forgo long-term projects if their information on the investment projects is incomplete. Empirical evidence also confirms that the incentives of pursuing myopic short-termism reduce investment (e.g., Baber et al., 1991; Dechow and Sloan, 1991; Edmans et al., 2014; Ladika and Sautner, 2014; Roychowdhury, 2006). Meanwhile, the literature on agency problems (e.g., Jensen, 1989; Jensen and Meckling, 1976) shows that as compared to unlisted firms, listed companies are more exposed to agency problem, and managers of listed firms are inclined to care more about short-term benefits. Asker et al. (2015) are among the first to investigate the manner in which short-termism and corporate investment behavior correlate with each other empirically. As the result shows, compared with unlisted companies, listed companies make significantly less investments and respond less to changes in investment opportunities, particularly in industries where stock prices maintain the highest sensitivity to earnings news. These results comply with the argument that the pressures of short-termism distort investment decisions. Sheen (2009) uses hand-collected data on listed and unlisted companies in the chemical industry and obtains similar findings to those of Asker et al. (2015).

2

Emerging Markets Review xxx (xxxx) xxxx

Z. Yu, et al.

Jorgenson (1971) uses the accelerator theory to demonstrate that the investment expenditure of enterprises is mainly determined by two factors: the level of actual output and the convenience of external financing. From the perspective of the external financing environment, the maturity between emerging markets and developed markets differs considerably. Note that China's financial system is immature as compared to U.S., and the size of China's stock market is smaller (Shan and Qi, 2006; Wong and Zhou, 2011). As Bokpin and Onumah (2009) emphasize, financial market development exerts a great influence upon the investment decisions of emerging market enterprises. Love (2003) utilizes World Bank data to analyze how financing constraints relates to financial development and finds that the former is diminished by the latter by reducing information asymmetries and contracting imperfections. Most importantly, Rauh (2006) finds a negative correlation between investment and financing constraints and shows that financing constraints have a great impact on investment decisions, consistent with Fazzari and Petersen (1993). Therefore, the existing literature reveals that differences in the external financing environments of the two types of markets may lead to differences in the investment decisions made by Chinese and US firms. Financing constraints are essential in explaining such differences. Meanwhile, the existing literature shows that product market competition impacts corporate investment behavior significantly. Specifically, Holmstrom (1982) reveals that product market competition is more effective than supervision in terms of improving corporate efficiency. Caballero (1991) develops a model to demonstrate that a company facing smaller absolute value of demand elasticity is more likely to give up its investment when facing uncertain investment opportunities as it maintains higher market power in the market. In other words, the lower the market power of a firm (in a more competitive market), the more likely it would be to choose to invest when facing uncertain investment opportunities. The underlying reasons are the following: first, since the flexibility of demand in an imperfectly competitive market is not perfectly flexible, a company's incentive to increase output is less as compared to the case of a perfect competition market. This is because its products' prices would be brought down by output increase, which in turn reduces the direct effect of a positive investment opportunity. Second, in a less competitive market, due to diminishing marginal returns to capital, capital stock exerts great influence upon the marginal value of investment. In this context, an increase in a company's current investment exerts a bad impact upon its future profitability or the value of its future investments. Therefore, the company would tend to invest less, which is the essence for the option value of waiting under irreversibility. As an extension of Caballero (1991), Akdoğu and MacKay (2008) and Kandilov and Leblebicioğlu (2011) conduct empirical research using US and Colombian data, respectively. As is revealed, when facing investment opportunities under uncertainty, companies in non-competitive industries exhibit lower investment sensitivity. Supporting this view, Akdoğu and MacKay (2008) use US data to show that companies in competitive industries possess higher investment sensitivity and are quicker to invest than those in competitive industries. Similarly, using Columbian manufacturing census data, Kandilov and Leblebicioğlu (2011) find that a higher mark-up (implying a higher monopoly power) reduces the sensitivity of investment. In sum, the existing literature illustrates that lower investment sensitivity is more likely to occur in non-competitive markets. Besides product market competition, ownership structure also exerts a great influence upon corporate investment behavior. The existing literature reveals that state-owned enterprises (SOEs) exhibit investment inefficiency due to government intervention and social responsibilities compared with non-state-owned enterprises. As Chen et al. (2011) suggest, SOEs' investment behavior is distorted and their investment efficiency is affected by governmental intervention through the appointment of politically connected managers or majority state ownership. Meanwhile, Lin and Tan (1999) point out that because Chinese SOEs bear greater policy burdens than non SOEs, such as solving social employment problems and coordinating national development strategies, they often need to invest in some nonviable industries or in less-developed regions. However, in the early 1990s, China underwent a large-scale corporatization reform of SOEs, listing many on the stock market. Using a panel of pre- and post-listing data for China, Wang et al. (2004) document that going public significantly lowers the state share, mitigating firms' reliance on debt financing and allowing firms to increase their capital expenditure. They also find that a more balanced ownership structure among large institutional shareholders, or large individual shareholders. Moreover, state shareholders is good for corporate performance. That is, consistent with Xu and Wang (1999), the higher the proportion of other types of shareholders is, the more effective the operation of SOEs will be. Wang et al. (2004) also show that SOEs' operating efficiency correlates positively with the shares held by legal persons (institutional investors), but correlates negatively with the state shares. This is because large institutional person shareholders are incentivized to supervise and control the behavior of the firm's management. In sum, due to the presence of supervision from external shareholders and having more capital to invest, Chinese listed SOEs try to improve their investment efficiency. 3. Data 3.1. Sample Firstly, we use the firm-level accounting data of Chinese A-share companies provided by the CSMAR (China Stock Market and Accounting Research) database that begins in 1998 and ends in 2007. Since the mid-1990s, in order to alleviate financial pressures, through converting debt into stock and restructuring them into limited liability firms or joint stock firms, the Chinese government has intensified its reform of SOEs (Zhai Zhuan Gu) (e.g., Zhu, 1999). This process led to a wave of SOEs and non-SOEs being listed, and 864 A-share firms went public from 1998 to 2007. Investigating the listed firms' investment behavior during this period can better serve our purpose of testing how financing constraints lead to differences in listed and unlisted firms' corporate investment behavior. We exclude those observations that do not match with CASIF industry codes and only retain those firms with at least three consecutive annual observations so that lag variables can be constructed, which permits us to investigate within-firm variation. Finally,

3

Emerging Markets Review xxx (xxxx) xxxx

Z. Yu, et al.

Table 1 Descriptive statistics. Full sample

Matched sample

Unlisted companies

Listed companies

Unlisted companies

Listed companies

Mean Median sd

94.490 16.228 1156.313

4328.000 1318.000 31,218.000

2827.417 1071.041 7713.355

3180.433 1348.838 17,709.744

Mean Median sd

0.398 0.143 19.303

0.811 0.145 37.542

0.271 0.110 3.895

0.339 0.165 3.521

Mean Median sd

2.339 2.214 18.312

2.788 2.313 29.871

2.597 2.497 1.066

3.157 2.311 44.734

Mean Median sd

0.135 0.053 1.808

0.071 0.042 3.088

0.060 0.037 0.107

0.117 0.047 4.691

Book leverage

Mean Median sd

0.745 0.647 7.561

0.690 0.543 7.122

0.645 0.609 0.492

0.730 0.523 10.728

Age

Mean Median sd

6.642 4.000 10.322

9.510 9.000 4.212

14.326 9.000 16.678

8.711 8.000 4.150

Size Total assets($m)

Investment opportunities Sales growth

Predicted Q

Firm characteristics ROA

Notes: the above table shows descriptive statistics for the full sample and the matched sample of listed and unlisted firms in the duration between 1998 and 2007. See Section 3.1 for a detailed description on how the full sample from CSMAR and CASIF database is conducted and Section 3.2 for details of the matching procedure. The table demonstrates standard deviations, medians and means of the key variables adopted in our empirical analysis. For the definitions of the variables and details for their constructions, see Appendix A.

we obtain 11,800 firm-year observations for 1390 listed firms, and the sample begins in 1998 and ends in 2007.1 Secondly, financial data of unlisted firms used in this paper for the period 1998–2007 are taken from CASIF, which contains information on all Chinese state-owned firms and private firms whose annual sales are more than five million RMB. Such a huge dataset for unlisted firms possibly allows us to match data with the CSMAR dataset to achieve our research goal. Before undergoing our empirical analysis, we drop firms without names, fewer than three consecutive annual observations, and those with missing or negative values for total assets, total wages, or total sales. As we have already imported the data of listed firms from CSMAR, to avoid data repetition, we also drop firms that went public during the 1998–2007 period from CASIF. Finally, we obtain 1,432,539 firm-year observations. 3.2. Matching To eliminate the sampling bias between listed and unlisted firms (see Asker et al., 2015; Gao et al., 2013), we match the CSMAR dataset with the CASIF dataset based on firm size, industry, and age. Our results are shown to be robust by matching on additional characteristics (see Section 5.1). Before matching, the size of most of the listed firms is much larger than the unlisted ones. As Table 1 shows, the mean (median) total assets of listed companies in CSMAR is 4328 million RMB (1318 million RMB), which is in sharp contrast to 94.490 million RMB (16.228 million RMB) for unlisted companies in CASIF. The left-side graph of Fig. 1 depicts the size distribution for the two datasets. They overlap, but only to a limited extent. A caliper-based nearest-neighbor method is adopted, plus the addition of firm age as a matching condition, to capture the impact of lifecycle to construct our panel data. Starting in 1998, for listed firms, we first regress assets and age for a public dummy variable using a probit analysis in the same two-digit industry in every year. Then we predict the probability of listing for each firm and select the unlisted firms closest in the probability of listing in the same two-digit industry and the same year, mandating that the ratio of their total assets (TA) be less than 2 (i.e., max(TA_listed, TA_unlisted) / min(TA_listed, TA_unlisted) < 2). If the ratio of their total assets is not suitable, we drop those samples. Once a match formed, to retain the panel structure intact, the matched samples in the following years are kept, which permits us to calculate within-firm investment regressions. We filter those firms that do not have three consecutive annual observations before matching again. After matching, the matched sample contains 5239 listed-firm-year observations and an equal number of unlisted-company-year observations. As the replacement is matched with, 837 listed companies 1 Our findings remain to hold even if we extend our sample period to 2013. However, we choose not to report the results obtained from using the post-2008 CASIF data because the CASIF data after 2008 suffer from the problem of missing values and mismatching between firm names and observations.

4

Emerging Markets Review xxx (xxxx) xxxx

Z. Yu, et al.

Fig. 1. Size distribution. Notes: the figures show the size distribution for full sample and the matched sample, respectively. The left graph represents the full sample constructed by listed and unlisted firms from CSMAR and CASIF datasets. The right graph is the matched sample matched by size, industry, and age.

and 767 unlisted companies are included in the sample. 4. Variables and empirical strategy This section describes the definitions of the main variables and the specification of the estimated model. 4.1. Variables 4.1.1. Investment Companies' assets can be increased through either enlarging capacity or purchasing the assets of another firm, which are respectively reflected in their capital expenditures and M&A (Merger and Acquisition) activities. Unlike listed firms, unlisted firms cannot always pay for their M&A with stocks. Thus, compared with listed companies, it is likely for their merger and acquisition to involve relatively more capital expenditures. In CASIF, there is no distinction between M&A expenditures and capital expenditures. In order to avoid estimation bias, we choose the annual change of net fixed assets (Net for short) to capture M&A and capital expenditures. In addition, we construct the annual change of gross investment in non-current assets (Grossnca for short), net investment in non-current assets (Netnca for short), and growth of total assets (Growth for short) to check for robustness. Appendix A provides detailed definitions for these variables. 4.1.2. Investment opportunity As suggested in the prior literature, Tobin's Q, or sales growth, is an appropriate proxy for a firm's investment opportunity (e.g., Asker et al., 2015; Richardson 2006). Tobin's Q is usually referred to as the proportion of a company's market value to its book value of assets. However, because unlisted companies do not go public in stock markets, their market values cannot be obtained directly. Following Campello and Graham (2013), we first regress the Tobin's Q of listed firms on four related variables (net income before extraordinary items, book leverage, ROA [Return on Assets], and sales growth) to obtain their predicted Tobin's Q (predicted Q) and then use them for both listed firms and unlisted firms. Since sales growth can be observed in all types of companies, we prefer to use sales growth to measure investment opportunity. It should be noted that Tobin's Q is just a firm value indicator in the stock market, and the Chinese stock market had only been established for eight years in 1998. Thus, the Chinese stock market system is imperfect, and we mainly use Tobin's Q for robustness checks. 4.1.3. Other firm characteristics In addition to the main variables, we also include some control variables, i.e., ROA, book leverage, and firm age, as shown in Table 1. As we have observed in Table 1, Chinese unlisted firms are older and smaller, but they have smaller debt ratios, lower sales growth, and lower ROA than listed firms after being matched for size, age, and industry. Clearly, the differences between listed and unlisted firms become less after the matching process, which makes the samples comparable. 4.2. Investment level equation In this subsection, we describe the summary statistics of different measures of our dependent variable and specify our main regression model following Asker et al. (2015). Our first result is summarized in Table 2: unlisted companies invest greatly less than listed companies. In Row 1 of Table 2, listed firms increase their net fixed assets by 4.1% of total assets annually on average, compared to 7.8% for unlisted firms. The difference is substantial but reasonable, because most of the unlisted firms are smaller, 5

Emerging Markets Review xxx (xxxx) xxxx

Z. Yu, et al.

Table 2 Unconditional investment levels. Listed

1 2 3 4 5 6 7 8

Full sample Full sample Full sample Full sample Sample matched on: Industry, size, age Industry, size, age Industry, size, age Industry, size, age

Unlisted

Investment measure

Mean

Median

Std. dev

No. of obs.

Mean

Median

Std. dev

No. of obs.

Net Grossnca Netnca Growth

0.041 0.211 0.092 0.161

0.007 0.100 0.034 0.082

0.189 1.830 0.910 0.993

10,409 10,409 10,409 10,409

0.078 0.318 0.118 0.278

−0.002 0.077 0.001 0.068

6.109 8.569 6.176 8.416

1,125,677 1,125,677 1,125,677 1,125,677

Net Grossnca Netnca Growth

0.053 0.180 0.084 0.149

0.017 0.108 0.044 0.092

0.196 0.462 0.238 0.386

4502 4502 4502 4502

0.020 0.109 0.037 0.093

−0.006 0.035 0.001 0.032

0.208 0.678 0.303 0.522

4671 4671 4671 4671

Notes: the above table juxtaposes listed and unlisted companies' unconditional investment levels in the full sample. For details regarding matching approach, please refer to Section 3.2. Appendix A presents variable definitions and details of their construction.

which makes it easier for them to expand their capacity at lower costs as compared to listed firms. Differences are also reflected in net non-current assets, growth of total assets and gross non-current assets. Tiny unlisted firms which annual sales of less than twenty million RMB drive disproportional differences. After we match the sample based on size, age, and industry as shown in Row 5 of Table 2, the mean value of listed firms' net fixed assets increases by 5.3% annually, compared to 2.0% for their unlisted counterparts. Thus, listed firms invest significantly more in expansion. Moreover, when using other measures of investment, we obtain similar results, as shown in Rows 6 to 8. We then move on to estimate the following investment regression:

Ii, t Ai, t

1

Si, t

= Listedi +

Si, t Si, t

1

1

+ Xi, t +

j

+

t

+

i, t ,

(1)

in which t, j, and i index fiscal years, industry, and firm, respectively, and I is one of the investment measures in Table 2. A denotes total assets, S denotes sales, and X represents a matrix of control variables. Listed is a dummy variable that equals one if the firm is listed on the stock market. Control variables include a lag of ROA, a lag of book leverage, the logarithm of asset and the logarithm of age. φj and ηt denote industry fixed effects and year fixed effects, respectively. εit is the error term. To control for common trends, unobserved industry-level heterogeneity is removed through industry fixed effects and year effects are also included. Standard errors are clustered at the company level. 4.3. Sensitivity of investment opportunity equation Following Asker et al. (2015), we examine the different responses to change in investment opportunities between listed firms and unlisted firms by adding interaction terms. The equation is as follows:

Ii, t Ai, t

=

Si, t

1

Si, t Si, t

1

1

+

Listedi ×

Si, t

Si, t Si . t

1

1

+ Zi, t + µi +

t

+

i, t .

(2)

As compared to Eq. (1), Eq. (2) drops industry fixed effects but includes company fixed effects, allowing us to concentrate upon investment sensitivities by calculating within-firm variation in response to within-firm variation in investment opportunities. The interaction term captures the investment sensitivity differences between listed and unlisted companies. Z is a matrix of control variables, which includes a lag of ROA, a lag of book leverage, the interaction item of ROA and Listed, the interaction item of leverage and Listed, and the logarithm of asset and the logarithm of age. μi are firm fixed effects. We use predicted Q to substitute sales growth to represent investment opportunities for robustness checks. 5. Empirical results In this section, we report the baseline results and robustness results for investment level and investment sensitivity using Eqs. (1) and (2), and we discuss the underlying mechanism. 5.1. Baseline results Table 3 presents the estimated results of Eq. (1) based on a matched sample for each investment measure, as described in Table 2. Our results will not be qualitatively changed by including investment opportunities and ROA in the regressions, i.e., compared with listed companies, unlisted companies make less investment and yield a similar magnitude as in the unconditional tests shown in Table 2. Column 1 gives the regression results without control variables and suggests that listed firms invest more for expansion than their unlisted counterparts matched on size and industry, where the coefficient of Listed is positively significant at the 1% level. In 6

Emerging Markets Review xxx (xxxx) xxxx

Z. Yu, et al.

Table 3 Estimation results of investment level. Proxy for opps

Sales growth Matched on: size; industry; age

Listed Opps L.roa

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

Net

Grossnca

Netnca

Growth

Net

Grossnca

Netnca

Growth

0.027*** (0.002) 0.058*** (0.004)

0.051*** (0.006) 0.263*** (0.011)

0.040*** (0.003) 0.077*** (0.005)

0.042*** (0.004) 0.242*** (0.009)

8331 0.091

8349 0.131

8326 0.102

8368 0.167

0.026*** (0.002) 0.052*** (0.004) 0.206*** (0.017) 0.016*** (0.005) 0.008*** (0.001) 0.000 (0.002) 6135 0.139

0.041*** (0.006) 0.227*** (0.012) 0.867*** (0.052) −0.025* (0.013) 0.032*** (0.003) −0.006 (0.005) 6161 0.216

0.037*** (0.003) 0.060*** (0.005) 0.351*** (0.025) 0.006 (0.007) 0.013*** (0.002) 0.001 (0.002) 6138 0.171

0.038*** (0.005) 0.212*** (0.010) 0.755*** (0.040) −0.007 (0.011) 0.026*** (0.003) −0.002 (0.004) 6170 0.261

L.lev lnasset lnage No. obs R2

Notes: this table presents the estimation results of Eq. (1), which investigates the differences regarding investment level between listed and unlisted companies, holding investment chances and profitability unchanged. To be specific, the table compares conditional investment levels in the full sample, our size-plus-industry matched samples, and various variations of our basic matching specifications. Section 3.2 shows the details of the matching procedure. Each column in Table 3 employs the same sample and investment measure as defined in Table 2. For definitions of the variables and details for their construction, see Appendix A. We adopt sales growth to proxy for investment opportunities and ROA to measure profitability. Columns 1 to 4 are the estimation results without control variables for four independent variables, respectively, i.e., Net, Grossnca, Netnca, Growth. Columns 5 to 8 are the estimation results with control variables for four independent variables, respectively. Each regression includes industry and year fixed effects (not reported in the table), and standard errors are clustered at firm level. ***, ** and * are adopted to denote significance at the 1%, 5% and 10% level, respectively. All variables, except assets and age, are winsorized 2.5% in each tail to deal with the effect of outliers.

Columns 2–4, we replace the dependent variable by other measures of investment and obtain similar results, indicating that our results are robust. We then add control variables to the regression and report the estimated results in Columns 5–8. In Table 2, the coefficient of our main independent variable, Listed, in Column 5 is 0.026 and is positive and significant at the 5% level, while the coefficients of Listed in Columns 6–8 are also significantly positive. Hence, the results in Tables 2 and 3 consistently demonstrate unlisted companies invest less than listed ones do, even after controlling for differences in investment opportunities. This finding is consistent with the work of Brau and Fawcett (2006), who suggest that a common reason for going public is to get easier access to less expensive investment capital. On the other hand, this finding also contradicts the findings of Asker et al. (2015), who use US data to show that short-termism exists for US listed firms. Thus, our finding indicates that short-termism does not affect Chinese listed companies in terms of their investment level. Table 4 reports the estimated results of Eq. (2) using sales growth as the measure of investment opportunity. Similar to the structure of Table 3, Columns 1–4 present the regression results for different investment measures without adding control variables, while Columns 5–8 present the results with control variables. As Table 4 demonstrates, the coefficient of the interaction term of investment opportunity and Listed in Column 1 is positive and significant at the 1% level, which means that listed firms' investment decision makers respond more to changes in investment chances. When adding control variables in Column 5, it is found that the difference between listed and unlisted companies is also significantly positive but becomes bigger, indicating listed companies' greater sensitivity to investment opportunity. We also replace the dependent variable with other measures of investment for the robustness tests. The results are reported in Columns 2–4 and Columns 6–8. They show that the interaction term's coefficients are significantly positive again. This implies that listed firms are more aggressive when pursuing and seizing investment opportunities than their unlisted counterparts, and short-termism does not affect listed firms' investment sensitivity. We then move on to use predicted Q as a substitute for sales growth to measure investment opportunity. The estimated results of predicted Q in Table 5 are not as good as those presented in Table 4. Although most coefficients of the interaction terms are significantly positive, the coefficients are insignificant in the case of using the scaled net investment (Net) and net investment of noncurrent assets (Netnca) as dependent variables. This is possibly because China's stock market is still immature and has many deficiencies, such that Tobin's Q cannot perfectly reflect a firm's true value. Meanwhile, it should be noted that the mechanisms through which Tobin's Q and sales growth affect firm value are different. Sales growth can affect the actual capital of firms, whereas Tobin's Q mainly influences firm value in financial markets. However, in spite of these insignificant cases, the results still imply that our regression is robust. Moreover, in order to show the robustness for matching methods, we adjust our matching method to construct our sample and reestimate Eqs. (1) and (2). First, we only employ a caliper-based nearest-neighbor match to construct our panel data, which follows

7

Emerging Markets Review xxx (xxxx) xxxx

Z. Yu, et al.

Table 4 Estimation results of investment sensitivity. Proxy for opps

Sales Growth Matched on: size; industry; age

Opps x Listed L.roa

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

Net

Grossnca

Netnca

Growth

Net

Grossnca

Netnca

Growth

0.026*** (0.005) 0.017** (0.008)

0.167*** (0.016) 0.033 (0.024)

0.038*** (0.007) 0.009 (0.010)

0.161*** (0.014) 0.026 (0.019)

No 8331 0.028

No 8349 0.062

No 8326 0.022

No 8368 0.084

0.028*** (0.006) 0.021** (0.009) 0.113*** (0.037) 0.071 (0.052) −0.007 (0.010) −0.017 (0.015) 0.032*** (0.004) −0.026** (0.011) Yes 6135 0.055

0.152*** (0.020) 0.052* (0.027) 0.457*** (0.113) 0.879*** (0.153) −0.318*** (0.033) −0.185*** (0.047) 0.267*** (0.019) −0.038 (0.033) Yes 6161 0.221

0.029*** (0.008) 0.021* (0.012) 0.210*** (0.048) 0.281*** (0.069) −0.027** (0.014) −0.102*** (0.021) 0.065*** (0.007) −0.025* (0.014) Yes 6138 0.093

0.145*** (0.015) 0.050** (0.021) 0.364*** (0.087) 0.698*** (0.121) −0.276*** (0.027) −0.137*** (0.037) 0.240*** (0.016) −0.022 (0.027) Yes 6170 0.270

L.roa x Listed L.lev L.lev x Listed lnasset lnage Control No. obs R2

Notes: this table presents the results of estimating investment Eq. (2), which employs within-firm variation to study the differences in the sensitivity of investment spending to investment chances between listed and unlisted companies. The main dependent variable is net fixed assets (the annual increase in gross fixed assets scaled by beginning-of-year total assets). Columns 1 to 4 are the results without control variables for four independent variables, respectively, i.e., Net, Grossnca, Netnca, Growth. Columns 5 to 8 are the results with control variables for four independent variables, respectively. Each regression includes industry and year fixed effects (not reported in the table), and standard error are clustered at the firm level. ***, ** and * are adopted to represent significance at the 1%, 5% and 10% level, respectively. All variables, except assets and age, are winsorized 2.5% in each tail to deal with the effect of outliers. Table 5 Predicted Q for investment opportunity. Proxy for opps

Predicted Q Matched on: size; industry; age

Opps x Listed Control No. obs R2

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

Net

Grossnca

Netnca

Growth

Net

Grossnca

Netnca

Growth

0.004 (0.005) −0.002 (0.002) No 8132 0.012

−0.070*** (0.018) 0.030*** (0.006) No 8163 0.021

−0.027*** (0.006) 0.005* (0.003) No 8145 0.012

−0.046*** (0.014) 0.020*** (0.005) No 8167 0.020

0.023*** (0.007) −0.004 (0.002) Yes 5975 0.036

0.017 (0.025) 0.033*** (0.007) Yes 6009 0.189

−0.004 (0.010) 0.003 (0.003) Yes 5991 0.081

0.034 (0.021) 0.025*** (0.006) Yes 6011 0.221

Notes: this table reports the robustness estimation results of Eq. (2) that uses predicted Q to replace sales growth to represent investment opportunity. The main dependent variable is net fixed assets (the annual increase in gross fixed assets scaled by the beginning-of-year total assets). Columns 1 through 4 report the results without control variables for four independent variables. Columns 1 to 4 present the results without control variables for four independent variables, respectively, i.e., Net, Grossnca, Netnca, Growth. Columns 5 to 8 are the results with control variables for four independent variables, respectively. Each regression includes industry and year fixed effects (not presented in the table), and standard error are clustered at the firm level. ***, ** and * are adopted to represent significance at the 1%, 5% and 10% level, respectively. All variables, except assets and age, are winsorized 2.5% in each tail to deal with the effect of outliers.

Asker et al. (2015). Second, we follow Michaely and Roberts (2012) to augment our matching criteria for ROA, book leverage, and sales growth. The robustness tests are presented in Table 6, demonstrating that the selection of matching variables does not alter our basic results, whether for regressions of investment level or investment sensitivity.

8

Emerging Markets Review xxx (xxxx) xxxx

Z. Yu, et al.

Table 6 Robustness tests with different methods of matching. Matched on: size; industry

Matched on: size; industry; ROA, lev, sales growth

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

Net

Grossnca

Netnca

Growth

Net

Grossnca

Netnca

Growth

0.014** (0.006) 0.049*** (0.004) Yes 5459 0.139

0.021 (0.017) 0.220*** (0.012) Yes 5474 0.216

0.024*** (0.008) 0.059*** (0.005) Yes 5474 0.178

0.009 (0.014) 0.208*** (0.010) Yes 5496 0.263

0.027*** (0.002) 0.055*** (0.005) Yes 4444 0.127

0.055*** (0.007) 0.240*** (0.015) Yes 4448 0.227

0.039*** (0.003) 0.072*** (0.006) Yes 4432 0.182

0.046*** (0.005) 0.216*** (0.011) Yes 4463 0.267

Panel B: Investment sensitivity Opps 0.017*** 0.159*** (0.006) (0.021) x Listed 0.028*** 0.045 (0.009) (0.028) Control Yes Yes No. obs 5459 5474 2 R 0.056 0.221

0.025*** (0.009) 0.024* (0.012) Yes 5474 0.099

0.153*** (0.017) 0.043* (0.022) Yes 5496 0.273

0.023*** (0.007) 0.031*** (0.012) Yes 4444 0.065

0.119*** (0.023) 0.102*** (0.033) Yes 4448 0.221

0.024** (0.009) 0.032** (0.014) Yes 4432 0.105

0.119*** (0.017) 0.083*** (0.024) Yes 4463 0.275

Panel A: Investment level. Listed Opps Control No. obs R2

Notes: the above table presents the results of estimating investment Eqs. (1) and (2) using other matching methods. Sales growth is employed to measure investment opportunities, and four different independent variables to measure investment. The definitions are the same as in Table 2. For the definitions of variables and details for their construction, see Appendix A. We use sales growth to proxy for investment opportunities and ROA to measure profitability. In Columns 1 to 4, we only use asset as a matching condition in the same industry. Columns 5 to 8 are the results after being matched by size, industry, ROA, book leverage and sales growth. We use probit model to estimate a propensity score for each firm in each year. The listed and unlisted firms with the closet score are classified as a group under no replacement matching way. In Panel A, each regression includes industry and year fixed dummies (not reported in the table). In Panel B, each regression includes firm and year fixed effects (not reported in table). Standard errors are clustered at the firm level. ***, ** and * are adopted to index significance at the 1%, 5% and 10% levels, respectively. All variables are winsorized 2.5% in each tail to deal with the effect of outliers.

5.2. Mechanism: Financing constraints According to the existing literature (Shan and Qi, 2006; Wong and Zhou, 2011), there is a significant difference in the financial development of China and the United States. Moreover, adopting cross-country data, Love (2003) analyzes the manner in which financial development correlates with financing constraints. He finds that better financial development can mitigate the difficulty of financing constraints by contract imperfections and reducing information asymmetries. More importantly, Fazzari and Petersen (1993) and Rauh (2006) emphasize the negative correlation between firms' financing constraints and their investment behavior, which might illustrate the differences in investment behavior between listed and unlisted companies. As mentioned in Section 1, the differences in firms' investment behaviors may be caused by differences in the status of financing constraints in China and the United States. To explore how financing constraints affect firms' investment behavior, it is crucial to identify whether a listed firm's main purpose for going public is to alleviate its financing constraints for investment. The most appropriate method for revealing this is to inspect a firm's financing constraints before listing, and then compare its investment behavior before and after going public. However, since the number of observations appropriate for the above analysis is too few (fewer than 50), we instead use the financial data of the matched unlisted counterparts to proxy for the financing constraints facing the listed firms before going public, upon which we base the grouping of listed firms. We then introduce financing constraints into Eqs. (1) and (2) to test our conjecture, i.e., to test whether the differences in financing constraints lead to the differences in investment behavior between Chinese listed firms and US listed firms. To ensure consistency with the literature on financing constraints, we construct a comprehensive indicator to measure corporate financing constraints. Following Musso and Schiavo (2008) and Bellone et al. (2010), we use multiple indexes, as follows. (1.) Cash ratio. It equals the cash flow scaled by total assets. This measure reflects the level of adequacy of endogenous funds. A high cash ratio means that endogenous funds are relatively adequate, implying better capital liquidity. In other words, a higher cash ratio implies lower financing constraints. (2.) Size. It is the logarithm of total assets. The scale of enterprise assets is often used by banks when considering corporate credit. Large enterprises generally have stronger external financing capabilities than small and medium-sized ones. (3.) Age. It is calculated according to firm age. Firms with a long history can accumulate better credit and have more stable cooperation with external fund providers. Hence, they have stronger financing capacity. (4.) Liquidation ratio. It equals the owner's equity divided by total debt. The liquidation ratio can demonstrate a firm's debt structure and repayment ability. A higher liquidation ratio implies easier access to obtaining loans from banks and facing lower financing constraints. 9

Emerging Markets Review xxx (xxxx) xxxx

Z. Yu, et al.

Table 7 Descriptive statistics of financing constraints. Index

Mean

Median

Max

Min

Std.dev

No. of obs.

1 2 3 4 5 6

Cash ratio Size Age Liquidation ratio Net fixed asset ratio Profitability

1.193 3.000 2.855 2.999 3.000 3.000

1.000 3.000 3.000 3.000 3.000 3.000

5.000 5.000 5.000 5.000 5.000 5.000

1.000 1.000 1.000 1.000 1.000 1.000

0.858 1.414 1.387 1.414 1.414 1.416

10,478 10,478 10,478 10,478 10,478 10,478

Add up 7

Financing constraint

16.060

16.000

29.000

6.000

3.650

10,478

Logarithm 8

Score

2.749

2.773

3.367

1.792

0.239

10,478

Notes: this table reports descriptive statistics for six sub-indices and two final indices of financing constraints for listed firms and unlisted counterparts. See Section 5.2 for a detail description on how we construct those six indices from full sample. We deal with Financing Constraint variable by logarithm. Then we get Score. This table reports means, medians, maximum, minimum, and standard deviations of eight indices in matched sample.

(5.) Ratio of net fixed assets. It equals net fixed assets scaled by total assets. Creditors regard fixed assets as collateral to guarantee debt repayment. A higher ratio of net fixed assets implies lower financing constraints. (6.) Profitability. It equals profit divided by operating revenue. This measure reflects investment opportunity in another way. Profitability is an important factor that could influence financing constraints (Bellone et al., 2010; Kaplan and Zingales, 1997). Higher profitability means lower corporate financing constraints. We score each indicator by quintiles. Based on the sorting of each variable of unlisted firms after matching, we divide them into five intervals and score them from one to five. For example, we give a score of one to the top 20% cash ratio. The higher the indicator score the weaker the financing capacity and the higher the financing constraints. The descriptive statistics are shown in Table 7. Rows 1–6 are the sub-index of financing constraint, adding them up to get the Financing Constraint variable in Row 7, and then we deal with the Financing Constraint variable by logarithm and get the Score variable, which is used to measure the intensity of financing constraints. We also draw a distribution graph of the Score variable in Fig. 2. Clearly, the lower the score, the lower the intensity of financing constraints. Table 8 reports the estimated results including the interactions between the financing constraint (Score) and Listed variables. In Column 3, the coefficient of Listed is −0.054 and significantly negative, while that of the interaction term Score*Listed is 0.029 and significantly positive, indicating an aggregate negative impact of Listed for sufficiently low financing constraints and an aggregate positive impact of Listed for sufficiently high financing constraints. In Column 4, the coefficient of the interaction term Listed*Opps is

Fig. 2. Score distribution. Notes: the figures show the distribution of score after matching, which is used to measure the intensity of financing constraints. The closer the x-axis' left is, the lower the intensity of financial constraints is. 10

Emerging Markets Review xxx (xxxx) xxxx

Z. Yu, et al.

Table 8 Investment performances and financing constraints. proxy for opps

Sales growth Matched on: size; industry; age

Listed Opps Score Score * Listed Opps * Listed

(1)

(2)

(3)

(4)

Net

Net

Net

Net

1.006*** (0.213) −0.081*** (0.024)

−0.188** (0.088) 0.070*** (0.008) −0.068** (0.028) 0.079** (0.032)

0.941*** (0.268) −0.063** (0.026)

−0.336*** (0.099) 0.085*** (0.013) −0.100*** (0.025) 0.133*** (0.036)

Score * Opps Score * Opps * Listed Control Obs. R2

No 7174 0.021

−1.275** (0.502) −0.339*** (0.074) 0.462** (0.182) No 7174 0.025

Yes 5217 0.038

−0.729* (0.406) −0.324*** (0.094) 0.273* (0.143) Yes 5217 0.078

Notes: this table reports descriptive statistics for six sub-indices and two final indices of financing constraints for listed firms and unlisted counterparts. See Section 5.2 for a detail description on how we construct those six indices from full sample. We deal with Financing Constraint variable by logarithm. Then we get Score. This table presents minimum, maximum, medians, means, and standard deviations of eight indices in matched sample.

−0.209 and significantly negative, while that of the three-variable interaction term is 0.081 and significantly positive, indicating that listed firms are more sensitive as compared to unlisted ones in terms of high financing constraints and less sensitive than unlisted firms in terms of low financing constraints. This shows that if a company has a lower financing constraint before listing, going public will not increase the firm's investment or create more sensitivity to changes in investment opportunity, and it may be negatively affected by short-termism to reduce investment level and investment sensitivity, similar to Asker et al. (2015). In sum, our conjecture is confirmed, i.e., listed firms have higher investment level and higher investment sensitivity if their unlisted counterparts are faced with higher financing constraints. Thus, financing constraints can explain the differences in Chinese and US firms' investment behavior. 6. Heterogeneity analysis In this section, we explore the heterogeneous performances of firms with different ownership or in different industries with different competition intensity. 6.1. SOEs versus non-SOEs The differences between Chinese SOEs and non-SOEs is a longstanding research topic, and most of the literature indicates that SOEs are less efficient than non-SOEs (Chen et al., 2011; Lin and Tan, 1999). Thus, when examining the influence of listing on corporate investment behavior, we should consider the impact of ownership structure. However, after going public, due to capital inflows and the entry of external shareholders, SOEs have more capital to invest and promote investment efficiency (Wang et al., 2004; Xu and Wang, 1999). This section is a theoretical extension of the work of Wang et al. (2004) and verifies whether SOEs become more efficient after listing. We use the ultimate controlling owner to divide the sample into two subsamples: SOEs and nonSOEs, and re-match our sample by asset and age in the same industry. Finally, we re-estimate Eqs. (1) and (2). Table 9 presents the results for SOEs and non-SOEs, both in terms of investment level and investment sensitivity.2 Panels A and B imply that the coefficients of the dummy variable Listed are significantly positive in the investment regression for both SOEs and non-SOEs. This implies that the two types of firms increase their investments after listing. By contrast, the difference between SOEs and non-SOEs is significant regarding investment opportunity sensitivity, as shown in Panels C and D. We find that the coefficients of the interaction terms are positive for SOEs in Panel C, while the coefficients are insignificant for non-SOEs in Panel D (except Netnca), implying that listed SOEs respond more to changes in investment opportunities. 2 Our findings concerning the impact of ownership remain to hold even when we add a dummy variable to identify whether a firm is a non-SOE or not in the matched sample to replace the subsample regression in Table 9.

11

Emerging Markets Review xxx (xxxx) xxxx

Z. Yu, et al.

Table 9 SOE and non-SOE. Matched on: size; industry; age; SOE (1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

Net

Grossnca

Netnca

Growth

Net

Grossnca

Netnca

Growth

0.039*** (0.009) 0.225*** (0.017) No 3493 0.117

0.035*** (0.004) 0.064*** (0.007) No 3484 0.115

0.031*** (0.007) 0.210*** (0.014) No 3508 0.163

0.017*** (0.004) 0.059*** (0.007) Yes 2539 0.132

0.026*** (0.010) 0.187*** (0.019) Yes 2546 0.178

0.028*** (0.005) 0.051*** (0.008) Yes 2539 0.178

0.019** (0.008) 0.184*** (0.015) Yes 2555 0.248

Panel B: Investment level of Non-SOE Listed 0.036*** 0.065*** (0.003) (0.008) Opps 0.036*** 0.065*** (0.003) (0.008) Control No No No. obs 3878 3889 2 R 0.127 0.166

0.046*** (0.004) 0.046*** (0.004) No 3883 0.141

0.052*** (0.007) 0.052*** (0.007) No 3896 0.202

0.033*** (0.003) 0.042*** (0.005) Yes 2834 0.173

0.063*** (0.009) 0.237*** (0.017) Yes 2843 0.259

0.043*** (0.005) 0.058*** (0.007) Yes 2834 0.200

0.054*** (0.008) 0.219*** (0.013) Yes 2847 0.302

Panel C: Investment sensitivity of SOE Opps 0.035*** 0.112*** (0.009) (0.027) x Listed 0.033*** 0.094** (0.013) (0.038) Control No No No. obs 3501 3493 R2 0.045 0.050

0.018* (0.010) 0.048*** (0.015) No 3484 0.028

0.124*** (0.020) 0.096*** (0.030) No 3508 0.089

0.023** (0.010) 0.040*** (0.015) Yes 2539 0.064

0.077** (0.030) 0.098** (0.043) Yes 2546 0.179

−0.002 (0.011) 0.061*** (0.018) Yes 2539 0.072

0.077*** (0.020) 0.125*** (0.032) Yes 2555 0.264

Panel D: Investment sensitivity of Non-SOE Opps 0.028*** 0.186*** (0.006) (0.027) x Listed 0.003 −0.003 (0.010) (0.035) Control No No No.obs 3878 3889 R2 0.029 0.066

0.045*** (0.009) 0.001 (0.013) No 3883 0.032

0.182*** (0.021) −0.009 (0.028) No 3896 0.089

0.024*** (0.007) 0.015 (0.012) Yes 2834 0.058

0.202*** (0.029) 0.007 (0.038) Yes 2843 0.242

0.030*** (0.011) 0.029* (0.016) Yes 2834 0.103

0.205*** (0.025) −0.015 (0.031) Yes 2847 0.293

Panel A: Investment level of SOE Listed 0.019*** (0.003) Opps 0.063*** (0.006) Control No No. obs 3501 2 R 0.103

Notes: this table reports the results of Eqs. (1) and (2) using the matched sample. Specifically, we group the sample by state-owned enterprise and non-state-owned enterprise to explore the differences in sensitivity to investment chances and investment level between listed and unlisted companies. Section 3.2 shows the details concerning the matching procedure. The investment measure and same sample as defined in Table 2 are employed in each column in Table 8. For the definitions of the variables and details for their construction, see Appendix A. We use sales growth to proxy for investment opportunities and ROA to measure profitability. Columns 1 to 4 are the results without control variables for four independent variables, respectively, Net, Grossnca, Netnca, Growth. Columns 5 to 8 are the results with control variables for four independent variables, respectively. In Panels A and B, year fixed dummies and industry are included in each regression (not presented in the table). In Panels C and D, year fixed effects and firm are included in each regression (not presented in table). Standard errors are clustered at the firm level. ***, ** and * are adopted to represent significance at the 1%, 5% and 10% level, respectively. All variables, except assets and age, are winsorized 2.5% in each tail to deal with the effect of outliers.

From Panels A and B, corporations tend to increase their level of investment after listing. However, the magnitude of the coefficients in between show that listed non-SOEs invest more than listed SOEs. It is worth nothing that listed SOEs respond more to changes in investment opportunities. As has been well established, Chinese non-SOEs are more efficient than Chinese SOEs in terms of investment (e.g., Bai and Lian, 2013; Dollar and Wei, 2007). According to Bai and Lian (2013), SOEs are motivated to over-invest for two reasons. First, government officials distort SOEs' investment decisions for their own political gain through government intervention, resulting in over-investment. Second, SOEs' managers weaken government supervision through rent-seeking behavior and then maximize the self-interest of managers through over-investment. These two motivations work together to result in over-investment. Our results show that over-investment is not reflected in investment increments, but in investment sensitivity. SOEs' returns of capital is significantly lower, on average, than non-SOEs' (Dollar and Wei, 2007), but SOEs still choose to invest as long as surplus funds exist, even though they know that the return on investment is low. This is also a reflection of SOEs' inefficient investments. 6.2. Product market competition Caballero (1991), Akdoğu and MacKay (2008), and Kandilov and Leblebicioğlu (2011) have shown that product market competition plays a key role in firms' investment decisions. In the existing literature, the HHI (Herfindahl–Hirschman Index) is widely 12

Emerging Markets Review xxx (xxxx) xxxx

Z. Yu, et al.

Table 10 Product market competition. Matched on: size; industry; age; low HHI (1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

Net

Grossnca

Netnca

Growth

Net

Grossnca

Netnca

Growth

Panel A: Investment level of low HHI Listed 0.026*** 0.051*** (0.003) (0.007) Opps 0.077*** 0.275*** (0.006) (0.015) Control No No No. obs 4725 4732 2 R 0.102 0.138

0.037*** (0.004) 0.092*** (0.007) No 4729 0.101

0.041*** (0.006) 0.265*** (0.012) No 4748 0.179

0.024*** (0.003) 0.064*** (0.006) Yes 3434 0.143

0.052*** (0.008) 0.229*** (0.017) Yes 3440 0.225

0.036*** (0.004) 0.068*** (0.008) Yes 3439 0.174

0.046*** (0.007) 0.219*** (0.013) Yes 3449 0.281

Panel B: Investment level of high HHI Listed 0.025*** 0.043*** (0.002) (0.007) Opps 0.055*** 0.274*** (0.004) (0.013) Control No No No. obs 5440 5453 2 R 0.086 0.138

0.040*** (0.003) 0.072*** (0.006) No 5439 0.102

0.035*** (0.006) 0.254*** (0.011) No 5464 0.177

0.024*** (0.003) 0.053*** (0.005) Yes 3979 0.132

0.039*** (0.008) 0.235*** (0.015) Yes 4004 0.209

0.035*** (0.004) 0.059*** (0.007) Yes 3982 0.162

0.032*** (0.006) 0.228*** (0.012) Yes 4017 0.262

Panel C: Investment sensitivity of low HHI Opps 0.026*** 0.123*** (0.009) (0.024) x Listed 0.033*** 0.084** (0.012) (0.035) Control No No No. obs 4725 4732 R2 0.031 0.054

0.024** (0.010) 0.033** (0.014) No 4729 0.023

0.125*** (0.018) 0.088*** (0.028) No 4748 0.083

0.019* (0.010) 0.041*** (0.016) Yes 3434 0.052

0.126*** (0.025) 0.085** (0.036) Yes 3440 0.217

0.018 (0.013) 0.039** (0.018) Yes 3439 0.089

0.118*** (0.018) 0.079*** (0.027) Yes 3449 0.268

Panel D: Investment sensitivity of high HHI Opps 0.037*** 0.224*** (0.006) (0.021) x Listed −0.003 −0.024 (0.009) (0.030) Control No No No. obs 5440 5453 R2 0.029 0.078

0.048*** (0.008) −0.010 (0.012) No 5439 0.024

0.198*** (0.017) −0.002 (0.024) No 5464 0.105

0.046*** (0.008) −0.006 (0.011) Yes 3979 0.062

0.188*** (0.026) 0.011 (0.035) Yes 4004 0.227

0.043*** (0.010) −0.002 (0.015) Yes 3982 0.096

0.178*** (0.020) 0.032 (0.027) Yes 4017 0.304

Notes: this table reports the results of Eqs. (1) and (2) for matched samples. Specifically, we group samples by high competition (low HHI) and low competition (high HHI) to explore the differences in sensitivity to investment chances and investment level between listed and unlisted companies. The HHIs are calculated based on the distribution of firm sales. The HHI groups indicate whether an industry's HHI is above or below the median HHI. Columns 1 to 4 are the results without control variables for four independent variables, respectively, i.e., Net, Grossnca, Netnca, Growth. Columns 5 to 8 are the results with control variables for four independent variables, respectively. Panels A and B are results of investment level. Panels C and D are results of investment sensitivity. Section 3.2 shows the details of the matching procedure. In Panels A and B, each regression includes industry and year fixed dummies (not reported in the table). In Panels C and D, year fixed effects and firm are included in each regression (not presented in table). Standard errors are clustered at the firm level. ***, ** and * are adopted to represent significance at the 1%, 5% and 10% level, respectively. All variables are winsorized 2.5% in each tail to deal with the effect outliers exclude assets and age.

adopted as an index to gauge the degree of product market competition that can comprehensively reflect the size distribution of enterprises or intra-industry competition. Following Akdoğu and MacKay (2008), Giroud and Mueller (2010, 2011), Ammann et al. (2013), and Yu et al. (2017), we also measure product market competition using HHIs. Based on how a company's sales are distributed, the HHIs are calculated. We classify the sample into two groups using the median value of HHIs across industries as the critical point. Industries with HHI values lower than the median are classified as competitive, whereas industries with HHI values higher than the median are classified as non-competitive. Table 10 presents the estimation results. Panels A and B suggest that firms would increase investment after going public. Specially, in Panel A, the coefficients of the List variable are almost higher than they are in Panel B. As to investment sensitivity, the coefficients of the interaction terms in Panel C of Table 10 are significantly positive at the 5% level, while the coefficients of the interaction terms are insignificant in Panel D. As compared to unlisted firms with similar size in the same industries, firms in competitive industries respond more to changes in investment opportunity, which is consistent with Akdoğu and MacKay (2008) and suggests that companies in non-competitive industries have lower investment sensitivity and are slower to make investments than those in competitive industries due to uncertainty. The mechanism is described by Caballero (1991), showing that additional investment will increase output, leading to a decrease in goods' price, and therefore lowering the direct effect of a positive change in price uncertainty. Therefore, in a non-competitive market, the incentive to invest is less sensitive when faced with the opportunity to invest as compared to a competitive market. 13

Emerging Markets Review xxx (xxxx) xxxx

Z. Yu, et al.

7. Conclusion This paper investigates whether short-termism has an impact on Chinese listed firms' investment behavior by comparing the differences between listed firms and their unlisted counterparts in terms of both investment level and investment sensitivity. We find that short-termism does not affect Chinese listed firms' investment decisions, which is in sharp contrast to the evidence provided by Asker et al. (2015) based on the US market. Specifically, we show that going public generates significantly positive effects on both investment level and investment sensitivity, mainly through the channel of financing constraints, i.e., short-termism only distorts listed firms' corporate investment behavior when the financing constraints of their unlisted counterparts are sufficiently low. Moreover, we further our research by considering the impact of ownership structure and product market competition. Our findings suggest that compared with their unlisted counterparts, state-owned listed firms or listed firms in competitive industries responds more to changes in investment opportunities. For SOEs, their investments become more efficient mainly due to the introduction of external shareholder supervision once they go public. This implies that we should properly encourage SOEs to go public and utilize market forces rather than administrative forces (e.g., government regulation) to efficiently supervise SOEs. As for market competition, fiercer competition can enable listed firms to make more efficient investment decisions, which indicates that enhancing the construction of a market-oriented environment can bring about benefits for firms through improvements in their investment behavior. An important implication of our findings is that for firms in emerging markets, such as China, going public is effective in mitigating the pressure of financing constraints for investment and improve investment efficiency. With the deepening of marketization in China, more SOEs should be encouraged to go public because it can improve their investment efficiency through the supervision of external shareholders and the enhancement of management autonomy (Groves et al., 1994; Wang et al., 2004). For firms in competitive industries, relaxing listing conditions is necessary so that firms gradually have easier access to financing in the stock market. In this way, the problem of financing constraints would be alleviated. In addition to financing constraints, many other factors like the quality of the regulation system and risk management, might also lead to differences between listed firms' investment behavior in developed economies and emerging economies. However, these are questions for future research. Acknowledgement This paper was presented at the 2018 Cross Country Perspectives of Finance conferences held in Guangzhou, China and Dar es Salaam, Tanzania. We are most grateful to the Co-Chair of this conference and the anonymous referees, together with the participants at the conference, for the helpful comments and useful suggestions. We acknowledge the financial support from the Key Project of National Social Science Foundation of China (17ZDA047), the Key Project of Ministry of Education (17JZD019), and the Fundamental Research Funds for the Central Universities (19JNKY02). Appendix A. Variable Definitions Total assets is CSMAR item Total Assets (A001000000) and the field name is asset in CASIF. One unit of total assets is 1 million RMB. Gross investment in NCA (Non-Current Assets) refers to the change in total assets (CSMAR data item Total Assets [A001000000] or its CASIF equivalent, Total Assets[asset]) minus the change in current assets (CSMAR data item A001100000 or its CASIF equivalent, current_asset), minus the change in net fixed assets (CSMAR data item A001212000 or its CASIF equivalent, fasset_net), plus the change in gross fixed assets (its CASIF equivalent, fassest), all scaled by beginning-of-year nominal total assets. Net investment is the annual change in net fixed assets (CSMAR data item A001212000 or its CASIF equivalent, fasset_net) scaled by beginning-of-year nominal total assets. Net investment in non-current assets (NCA) is the change in total assets (CSMAR data item Total Assets A001000000 or its CASIF equivalent, asset) minus the change in current assets (CSMAR data item A001100000 or its CASIF equivalent, current_asset) scaled by beginning-of-year nominal total assets. Growth in total assets is the change in total assets scaled by beginning-of-year nominal total assets. Sales growth is the annual percentage increase in sales: Salesi,t/Salesi,t-1 – 1 (using CSMAR item B001100000 or its CASIF equivalent, operating_revenue). Predicted Q is computed as follows. Following Campello and Graham (2013), we regress each listed firm's Tobin's Q on the firm's sales growth, return on assets (ROA), book leverage, and year and industry fixed effects (using the 2-digit industry code). We then use the regression coefficients to generate a predicted Q for each firm—both listed and unlisted ones. ROA is the operating income before depreciation (equals profit plus interest_expense) scaled by beginning-of-year total assets. References Akdoğu, E., MacKay, P., 2008. Investment and competition. J. Financ. Quant. Anal. 43, 299–330. Allen, F., Qian, J., Qian, M., 2007. China’s financial system: Past, present, and future. In: Brandt, L., Rawski, T.G. (Eds.), China’s Great Economic Transformation. Cambridge University Press, pp. 506–568. Ammann, M., Oesch, D., Schmid, M., 2013. Product market competition, corporate governance, and firm value: Evidence from the EU area. Eur. Financ. Manag. 19 (3),

14

Emerging Markets Review xxx (xxxx) xxxx

Z. Yu, et al.

452–469. Asker, J., Farre-Mensa, J., Ljungqvist, A., 2015. Corporate investment and stock market listing: A puzzle? Rev. Financ. Stud. 28 (2), 342–390. Baber, W., Fairfield, P., Haggard, J., 1991. The effect of concern about reported income on discretionary spending decisions: The case of research and development. Account. Rev. 66, 818–829. Bai, J., Lian, L., 2013. Why do state-owned enterprises over-invest? Government intervention or managerial entrenchment. China J. Account. Stud. 1 (3–4), 236–259. Bellone, F., Musso, P., Nesta, L., Schiavo, S., 2010. Financial constraints and firm export behavior. World Econ. 33 (3), 347–373. Bokpin, G.A., Onumah, J.M., 2009. An empirical analysis of the determinants of corporate investment decisions: Evidence from emerging firms. J. Aggress. Maltreat. Trauma 23 (3), 249–267. Brau, J., Fawcett, S., 2006. Initial public offerings: An analysis of theory and practice. J. Financ. 59, 399–436. Caballero, R.J., 1991. On the sign of the investment–uncertainty relationship. Am. Econ. Rev. 81 (1), 279–288. Campello, M., Graham, J., 2013. Do stock prices influence corporate decisions? Evidence from the technology bubble. J. Financ. Econ. 107, 89–110. Chemmanur, T.J., Fulghieri, P., 1999. A theory of the going-public decision. Rev. Financ. Stud. 12, 249–279. Chen, S., Sun, Z., Tang, S., 2011. Government intervention and investment efficiency: Evidence from China. J. Corp. Finan. 17 (2), 259–271. Chirinko, S., 1993. Business fixed investment spending: Modeling strategies，empirical results and policy implications. J. Econ. Literat. 31, 1875–1911. Dechow, P., Sloan, R., 1991. Executive incentives and the horizon problem: An empirical investigation. J. Account. Econ. 14, 51–89. Dollar, D., Wei, S.J., 2007. Das (Wasted) Kapital: Firm Ownership and Investment Efficiency in China. NBER Working Paper. Edmans, A., Fang, V., Lewellen, K., 2014. Equity Vesting and Managerial Myopia. Working Paper London Business School. Fazzari, S.M., Petersen, B.C., 1993. Working capital and fixed investment: New evidence on financing constraints. RAND J. Econ. 24 (3), 328–342. Fazzari, S., Hubbard, G., Petersen, B., 1988. Financing constraints and corporate investment. Brook. Pap. Econ. Activ. (1), 141–195. Gao, H., Harford, J., Li, K., 2013. Determinants of corporate cash policy: A comparison of private and public firms. J. Financ. Econ. 109, 623–639. Geng, N., N’Diaye, P., 2012. Determinants of corporate Investment in China: Evidence from cross-country firm level data. Imf Work. Pap. 12 (80). Giroud, X., Mueller, H., 2010. Does corporate governance matter in competitive industries? J. Financ. Econ. 95 (3), 312–331. Giroud, X., Mueller, H., 2011. Corporate governance, product market competition, and equity prices. J. Financ. 66 (2), 563–600. Graham, J., Harvey, C., Rajgopal, S., 2005. The economic implications of corporate financial reporting. J. Account. Econ. 40, 3–73. Groves, T., Hong, Y., McMillan, J., Naughton, B., 1994. Autonomy and incentives in Chinese state enterprises. Q. J. Econ. 109 (1), 183–209. Hall, R., Jorgenson, D., 1967. Tax policy and investment behavior. Am. Econ. Rev. 58, 391–414. Holmstrom, B., 1982. Moral hazard in teams. Bell J. Econ. 13, 324–340. Jensen, M., 1989. Eclipse of the public corporation. Harv. Busin. Rev. (Sep.–Oct.) 61–74. Jensen, M., Meckling, W., 1976. Theory of the firm: Managerial behavior, agency costs and ownership structure. J. Financ. Econ. 3, 305–360. Jorgenson, D.W., 1971. Econometric studies of investment behavior: A survey. J. Econ. Lit. 9 (4), 1111–1147. Kandilov, I.T., Leblebicioğlu, A., 2011. The impact of exchange rate volatility on plant-level investment: Evidence from Colombia. J. Dev. Econ. 94 (2), 220–230. Kaplan, S.N., Zingales, L., 1997. Do investment-cash flow sensitivities provide useful measures of financing constraints. Q. J. Econ. 112 (1), 169–215. Ladika, T., Sautner, Z., 2014. The Effect of Managerial Short-Termism on Corporate Investment. Working Paper University of Amsterdam. Laksmana, I., Yang, Y., 2015. Product market competition and corporate investment decisions. Rev. Acc. Financ. 14 (2), 128–148. Lin, J., Tan, G., 1999. Policy burdens, accountability and soft budget constraint. Am. Econ. Rev. 89 (2), 426–431. Love, I., 2003. Financial development and financing constraints: International evidence from the structural investment model. Rev. Financ. Stud. 16 (3), 765–791. Love, I., Zicchino, L., 2007. Financial development and dynamic investment behavior: Evidence from panel VAR. Quart. Rev. Econ. Finan. 46 (2), 190–210. Maksimovic, V., Pichler, P., 2001. Technological innovation and initial public offerings. Rev. Financ. Stud. 14, 459–494. Michaely, R., Roberts, M., 2012. Corporate dividend policies: Lessons from private firms. Rev. Financ. Stud. 25, 711–746. Miller, M., Rock, K., 1985. Dividend policy under asymmetric information. J. Financ. 40, 1031–1051. Musso, P., Schiavo, S., 2008. The impact of financial constraints on firm survival and growth. J. Evol. Econ. 18 (2), 135–149. Narayanan, M., 1985. Managerial incentives for short-term results. J. Financ. 40, 1469–1484. O’Toole, C.M., Morgenroth, E., Ha, T., 2016. Investment efficiency, state-owned enterprises and privatization: Evidence from Viet Nam in transition. J. Corp. Finan. 37, 93–108. Rauh, Joshua D., 2006. Investment and financing constraints: Evidence from the funding of corporate pension plans. J. Finan 61 (1), 33–71. Richardson, S., 2006. Over-investment of free cash flow. Soc. Sci. Elect. Publ. 11 (2–3), 159–189. Roychowdhury, S., 2006. Earnings management through real activities manipulation. J. Account. Econ. 42, 335–370. Shan, J., Qi, J., 2006. Does financial development lead economic growth? The case of China. Ann. Econ. Financ. 7 (1), 197. Sheen, A., 2009. Do Public and Private Firms Behave Differently? An Examination of Investment in the Chemical Industry. Working Paper UCLA. Wang, X., Xu, L.C., Zhu, T., 2004. State-owned enterprises going public: The case of China. Econ. Transit. 12 (3), 467–487. Wong, A., Zhou, X., 2011. Development of financial market and economic growth: Review of Hong Kong, China, Japan, the United States and the United Kingdom. Int. J. Econ. Financ. 3 (2), 111–115. Xu, X., Wang, Y., 1999. Ownership structure and corporate governance in Chinese stock companies. China Econ. Rev. 10 (1), 75–98. Yu, Z., Li, J., Yang, J., 2017. Does corporate governance matter in competitive industries? Evidence from China. Pac. Basin Financ. J. 43, 238–255. Zhu, T., 1999. China’s corporatization drive: An evaluation and policy implication. Contemp. Econ. Policy 17 (4), 530–539. Zingales, L., 1995. Insider ownership and the decision to go public. Rev. Econ. Stud. 60, 425–448.

15