How to Explain Corporate Investment Heterogeneity in China's New Normal: Structural Models with State-Owned Property Rights

This paper studies corporate investment and its structural change by the view of state-owned property right. By constructing dynamic investment decision-making model, we find corporate investment heterogeneity in China due to their different dynamic shifts of objective functions, demonstrated by our simulations. Empirical tests imply that the expansion of investment improves financial performance, but does not play a positive role on solving social employment. POEs expanded investment much more than SOEs did, even in the transitional period, but both of them reduced their investment significantly in New Normal. Mechanisms are explored by 3-stage structural models for non-matched control group and nearest neighbor PSM matched control group. Although investment inefficiency of SOEs are concerned, executive stock ownership and equity finance could be exotic methods to stimulate efficient investment. Investment efficiency of POEs has been recovered in New Normal but POEs have shifted away from ‘profit-driving’ they used to be in the old normal.


Introduction
China's slowdown and its structural transformation have been attracting global attention since the end of 2015. 1 Over-debt and over-investment are regarded as the main causes. In normal economic times, Diego (2013) established that physical investment is the key to China's growth miracle which is exploited by the structural transformation during the period 1952-2006. Knight (2014 also considered high investment as the crucial factor of China's Developmental State, 2 which generated a virtuous circle of rapid growth -high confidence, high investment, high growth, high confidence-and kept it going. In order to exploit the key to China's sustainable growth, we examine the structural change of micro corporate investment behavior in this paper, focusing on the stateowned property rights as a resource of its heterogeneity for Chinese firms. This current paper exploits China's corporate investment behavior by two dimensions, vertical and horizontal, respectively. In the vertical dimension, the structural changes are tested on the evolution over time. The question is whether there is structural transformation of the e↵ects on firm's investment and its micro mechanism. If the answer is yes, what's the di↵erence between China's old normal and new normal. On the other hand, the horizontal dimension is state-owned property right which is a special feature of China's companies. Song et al. (2011) examined the di↵erence on productivity and access to financial markets between SOEs (State-Owned Enterprises) and POEs (Private-Owned Enterprises). What is the di↵erence of corporate investment behavior between them in the current China's slowdown background?
Several articles examined the e↵ect from the financial crisis in 2008, which found a reduction in investment for US companies and European companies. Campello et al. (2010), Campello et al. (2011), Kahle and Stulz (2013) examined capital expenditures fall from credit constraints. There is no signs of investment reduction in China during financial crisis, because there was no bank lending supply shock as US and EU. On the contrary, Chinese state-owned enterprises was funded by a great amount of bank loans from China's economic stimulus package. Liu et al. (forthcoming) found that bank lending became less responsive to firm profitability, the stimulus package and the associate increase in bank loan supply resulted in more resources being allocated to SOEs. Therefore, the credit constraint as US and EU companies faced hasn't appear for Chinese SOEs, but it may still exist for POEs. But SOEs have to take on social responsibilities and their managers may have more concerns on the impact of high risk investment on their job promotion. Our analysis is based on a dynamic model of corporate investment strategy considering di↵erent state-owned property rights. The reason why there is heterogeneity in China's corporate investment is that optimization 1 The very first concern on China's slowdown was revealed by Yellen's speech on the last declining to raise interest rate of FOMC (Federal Open Market Committee) in September, 2015.
2 "Enterprises that were owned or controlled by government has access to a ready supply of bank loans at low rates of interest, the non-state enterprises that were not linked to government were su ciently profitable to be able to rely on their own retained profits. " After analyzing the origins, evolution, incentives and successes of China's developmental state, he questioned on whether the developmental state itself can be maintained and revealed the adverse shock that threatening the solvency of the banking system and the demand for and supply of funds for investment might result in the consequent slowdown in economic growth. Research group on China's economic growth (2014) documented that China's structural slowdown resulted from the triple shocks on investment, employment and learning-by-doing. objective function shifts di↵erently between SOEs and POEs. POE's financial distress boundary coming from credit constraints shifts left after the recent global financial crisis. Some POEs abandon their original objectives of value maximization which results in investment expansion even if they are facing financial constraints and economic downturn. Other POEs reduced their investment due to the concerns on increasing risk. Their investment declines as the same trend as European and American companies did after the great recession. For SOEs, some of them reduced investment even if the central government has pushed them to expand investment over and over, including providing large amounts of fundings. Our theoretical model explains that although SOEs have to take on the social responsibility, their investment strategy also shifts since there are severe agency problem due to the decentralized information and incentive problems as Milton and Raviv (1996) suggested.
Using corporate panel data from 2004 to 2015, this current paper has the following three findings. Second, the real e↵ects of investment expansion considering structural transformation are further examined in this paper. In order to know whether investment expansion promoted sustainable economic growth, I also examined the social performance in addition to financial performance. The empirical tests imply that expansion of investment is able to improve financial performance which is about 3% on average, but no significant evidence on improving social performance. Deeply separating the wheat from the cha↵, although SOEs are better at social performance than POEs are, there is no evidence that investment expansion improves employment problem in either SOEs or POEs. IMF (2016) recently highlighted global downside risks from 7 perspectives, the second of which is right the international ramifications of the economic transition in China. This paper shows the real structural transformation and its e↵ect in China by considering China's special feature, state-owned property rights.
Third, the reason why there are heterogeneity of investment behavior not only in SOEs but also in POEs is that their profit vs. risk trade-o↵ shifts di↵erently. In transitional period, 43.3% of SOEs expanded their investment while the percentage is 51.1% for POEs. The ratio declined to 33.3% and 39.5% for SOEs and POEs in new normal economy, respectively. Using propensity score matching method, the heterogeneity from the impact from agency problem, external financing, and profit vs. risk trade-o↵ implies that POEs have transformed from profit-incentive in old normal economy as Knight (2014) described into risk-averse in the new normal economy. Since the investment expansion is unsustainable and has no significant benefits to employment, China's formal developmental state policy, i.e., virtuous circle of high confidence, high investment and high growth, as Knight (2014) put forward, is going to be broken. Investment expansion, as the main e↵ective factor of China's sustainable economic growth, is unsustainable in the new normal economy, thus China do have some concerns on economic sustainable growth.
In the current paper, we attempt to contribute to the literature on Coase's theory of firm. Grossman and Hart (1986) and Hart and Moore (1990) examined this topic using incomplete contracts. But it is hard to make progress because of the di culty of formalizing haggling costs as Hart (2008) referred to. This topic is facing an emerging improvement currently, one of the important exploring on China's market. Huang et al. (2017) added to the general literature of local information and firm decentralization. Chen and Wen (2017) interpreted China's housing boom as a rational bubble which can crowd out productive capital investment, and capital returns of private firms have larger and more significant predictive power on excess housing returns than do capital returns of SOE. This current paper firstly describes the dynamic process of firm's production and documents the objective for firms in di↵erent state ownership, attempting to examine this topic by analyzing its dynamic shift facing shocks of great recession. The di↵erence from state ownership is described by objective functions and credit constraints. Although China's government set o↵ economic stimulus plan, we still find investment decline not only in POEs but also in SOEs. The firms reducing investment are more than those expanding investment which is implied not only by our theoretical simulation but also by our empirical regressions. Our current paper is helpful to provide evidence that the corporate investment heterogeneity does exist in China and give us a clearer view on the structural change for the research in the future.
The remainder of the paper is organized as follows: Section 2 describes the background and provides an overview of the relevant literatures; Section 3 provides theoretical analysis on dynamic shift of the optimization objectives; Section 4 analyzes the structural changes based on DiD models; Section 5 examines the e↵ect of expanding investment; and in section 6 we discuss the mechanism of China's investment heterogeneity through three perspectives and their structural changes. A short conclusion is given in section 7. simulate the economy in 2010. As a result, the economic rebound has lasted for 18 months and domestic A-share stock market rises 33% from the bottom to the periodic peak. Third round, in order to ease the pressure of economic downturn in 2012, Chinese central bank lowered the depositreserve ratio, and NDRC kept speeding up the approval of the project. But the economic rebound has only lasted for 12 months and market rose 23%. Economy slowed down again in the second quarter of 2013. The fourth new round of economic stimulus was injected in a dose in July, 2013, adding open market operation by central bank. But the economic rebound has lasted for only 6 months and Chinese A-share stock market rose 12%. During this period, the ratio of investment to GDP has risen from 48% in 2010 to over 50% in 2013. It is easy to find that these economic stimulus plans were carried out around the investment but its e↵ective period was getting shorter and shorter. Consequently, skepticism on the sustainability of China's investment and financing shows up, which is also considered as the key factor influencing the stable growth in China economy.  (Zhang (2015)). Lin (2012) thought the China's New Normal wouldn't be immutable. "Infrastructure investing will be a win-win strategy for the developed and developing countries, both now and in the future." Recently, macro economists are emerging to discuss whether China's future GDP growth will be of L-shape or U-shape (Higginsa et al. (2016) ), which also implies that the China economy is facing a turn point. This paper firstly aims to test whether there is any structural change of corporate investment behavior and when it is. Two shocks are considered, including financial crisis and points of transition.

Background and literature
Financial crisis is the first shock we aim to investigate, which results in credit constraints (Campello et al. (2010)). Literatures on US and European companies concluded that companies declined their investment because of liquidity scarce (Campello et al. (2011), Tong andWei (2011)) and demand shrinkage (Kahle and Stulz (2013)). Bianchi and Melosi (2017) highlighted the macroeconomic e↵ects of fiscal shocks before, at and after the onset of Great Recession. Financial crisis or great recession is regarded as the turn point of New Normal (Greenstone et al. (2015)) which found that lending shocks led to economically small declines in both small firm and overall employment.
But China shows a special trend that the growth rate of the total investment in fixed assets was more that 15% from 2008 to 2011, even reached a peak of 30% in 2009 ), Higginsa et al. (2016). Investment expansion is an important part of the economic stimulus plan in China and is the key of steady economic growth. Considering there are many views on the beginning of the New Normal, this paper will test the di↵erent time window one by one in Section 4.
Idiosyncratic risk is the second e↵ect on corporate investment during financial crisis. Bachmann et al. (2013) constructed empirical proxies for time-varying business-level uncertainty. Idiosyncratic risk rises, firm investment falls, and more so when managers own a larger fraction of the firm (Panousi and Papanikolaou (2012)). Gulen and Ion (2016) found a strong negative relationship between firm-level capital investment and the aggregate level of uncertainty. Policy uncertainty can depress corporate investment by inducing precautionary delays due to investment irreversibility.
They found a roughly one-third of the 32% fall in capital investments between 2007 and 2009.
Although we found roughly 20% fall in our SOE and POE subsamples in 2009, which is still higher than the fall of US companies in Gulen and Ion (2016). Bloom (2009) o↵ered a structural framework to analyze the impact of these uncertainty shocks, which found that stock-markets levels generate a much more gradual drop and rebound in activity lasting 2 to 3 years. Therefore, we use the risk from stock market to specify each firm's risk instead of policy uncertainty. In order to examine the e↵ects of the state ownership, we compares the growth of investment in two groups: a group that should be directly a↵ected by state ownership, which is the treated group, and a group that should not be directly a↵ected by state ownership, which is control group. The treated group is from the SOEs in our sample, the control group is from the POEs in the sample. We match each SOE with the firm in POE subsample, using the nearest neighbor propensity score matching method based on secondary industry, province, firm size and age. The red line in Figure 1 is the median of treated group's investment growth, which is much lower than the green line of control group. For POEs, the growth of investment is positive in 2010 and 2011, and stayed negative in 2013 ⇠ 2015, while we find a negative growth of investment for SOEs in majority of the period.

State-owned property rights
State-owned property rights is a special feature of Chinese companies, which is important to their investment (Liu and Siu (2012)). 5 The motivations for investment are concluded by Hart et al. (1996) as reducing cost and improving quality. The managers in SOEs are lack of enthusiasm in both of these two objectives (Shleifer (1998)). Agency costs, low investment e ciency and soft budget constraints lead to under-investment in SOEs (Knyazeva et al. (2013)). Nonperformance loan (NPL) is also a concrete form of investment ine ciency for Chinese SOEs. 6 The domestic academic research reveals the duality of investment in SOEs. On the one hand, the first class of principal-agent problems between shareholder and manager generally exists in SOEs. In fact, state-owned listed companies have been in the absence of the owner for a long time. What the manager pursue is job promotion, job consumption and gray income (Lu et al. (2013)). There is a positive correlation between the increase of operating income and the manager's promotion in central SOE (Yang et al. (2013)). Investment of SOEs depends on the return on their assets (Li  2011)). Therefore, a lot of managers in SOEs do nothing and expect promotion safely. On the other hand, SOEs have to take on the social responsibility to solve the issues in employment, social stability and pension (Lin et al. (2004)), and get more government subsidies than POEs (Kong et al. (2013) even found that the firms in loss tend to obtain more subsidies), which result in the long term existence of soft constraints of budget. SOEs obtained a large number of bank loans through economic stimulus plans after the current financial crisis. They have to take on the responsibility to expand investment and to absorb employment. Overall, there are two kinds of behavior presented by SOEs after the financial crisis, the reduction of investment caused by principal-agent problem and the expansion of investment caused by the political burden.
POEs have an advantage over the SOEs in the aspects of the mechanism of governance (Hu et al. (2005)). Empirical results implied that investment of POEs are more sensitive to the cash flow than SOEs are (Luo (2007)). The behavior of POEs may be consistent with the conclusion of the international literature because of the increasing financing constraints which will reduce their investment because their investment is positively sensitive to their cash flow which went down after the financial crisis. In addition, some of the entrepreneurs in POEs are always seeking stability after having become prosperous. These POEs should have shown a similar trend to the conclusions as Lins et al. (2013). On the contrary, POEs always face the second kind of principal-agent problem between the major shareholders and small shareholders. Entrenchment e↵ect is widespread in SOEs through the cash holdings (Luo (2007)), especially in listed companies. Cash dividends are used as a means to tunnel rather than a means to curtail the agency problem (Huang et al. (2011)). The M&A of POEs and the bond financing market developed rapidly 7 . A large number 7 From the Wind Report in 2013, there were 2136 corporate bonds of unlisted companies, with face value of 2.9 of internet corporation acquire plenty of cash flow and expand investment through equity finance abroad. Thus, principal-agent problem in POEs lead to the expansion of investment. In summary, there also exists two investment behavior of POEs, the reduction of investment caused by financing constraints and the expansion of investment caused by principal-agent problem.
The influence of state-owned property rights is exploited in this paper. State-owned property right's impact on the investment of Chinese enterprises has been checked in a large amount of literatures, but the mechanism is complicated. Recently, Jiang and Kim (2015) provides a modern overview of corporate governance in China which surveys the di↵erence between SOEs and non-SOEs. Some articles revealed that POEs and large SOEs invest more positively and more eager to expansion (Luo (2007)). Is this conclusion still holds after the financial crisis or in the China's New Normal? Or do Chinese SOEs and POEs cut investment as European and American companies did (Campello et al. (2010), Campello et al. (2011) and Lins et al. (2013))?

Theoretical analysis
3.1. Dynamic process of firm's production Baker and Wurgler (2013) pointed out that a complete explanation of financing and investment patterns requires a correct understanding of the beliefs and preferences of managers and investors.
These two sets of agents are supposed to develop unbiased forecasts about future events and use these to make decisions that best serve their own interests. It has been hard to make progress on Coase's theory of firm because of the di culty of formalizing haggling costs. Using contracts as reference points to which parties feel entitled is a substitute for the assumptions of incomplete contracts and ex post bargaining over the surplus that drive the results in Grossman and Hart (1986) and Hart and Moore (1990)). Hart (2008) uses reference points more broadly as the underpinning for a theory of the firm. Because we do not observe this sort of bargaining within real firms, the reference point approach may outlive the existing architecture of the property rights theory of the firm.
We construct dynamic process to examine corporate investment decision. Let K and I denote the level of capital stock and gross investment rate, respectively. Following DeMarzo et al. (2012)'s capital accumulation model, the firm's capital stock K evolves according to where > 0 is the depreciation rate. impact on the economy. Introducing the following uncertainty as Albuquerue and Wang (2008) where 1 > 0 is the constant volatility parameter, Z t is a Brownian motion, and K 0 > 0. Suppose I old is the gross investment rate without structural change, i.e. in the old normal, and I t is the additional change, which will be either positive meaning expansion caused by government policy or negative implying reduction due to risk aversion. Then capital stock will change after the shock, satisfying where volatility parameter 2 is di↵erent from the risk in the old normal, which is indicated as 1 .
If it is in new normal, it is still exogenous, meaning it is constant. But if it is in transitional period, the equilibrium was broken and the new equilibrium is still not constructed, 2 will be endogenous, depending on investment. It is supposed to the increasing function of I t . The firm's operating profit dY t over time increment dt is the net value of capital revenue and producing cost, satisfying the dynamic function: where h denotes capital output ratio, f denotes external financing, and G denotes adjusted cost function including the cost of labor L and the private benefits of controlling shareholders C. The last term G is the monotonic increasing function of investment. External financing f is composed of two parts, credit funding B, and other financing f B which contains new bond financing and equity financing for listed company, which is denoted as D and S, respectively.

POE's optimization
There are two categories of objective function. One is based on profit-driving, The other category results from agency problem. The objective function is supposing it is dY t where is the incentive parameter in He et al. (2014). ⌘ is a parameter determining the degree of risk aversion, 0 < ⌘ < 1. g is the money equivalent cost of managerial input, which is dependent on I t . Therefore, the objective function considering agency problems in the new normal can be written as, The objective function can be regarded as a single decision-making problem on investment by on the assumption that the adjusted cost function G is the increasing function of investment, i.e.,

SOE's optimization
For SOEs, Zhang (1996) analyzed the principal-agency problem of SOEs implying that the real controller is absent in SOEs. Therefore, SOE's decision appeared much similar to insider's optimization, in which the objective function should include social contributions, SC t , such as employment expansion and agent's promotion correlated to the firm's profit. Milton and Raviv (1996) explained that the observed budgeting process is a response to decentralized information and incentive problems. These imperfections can result in underinvestment when capital productivity is high and over-investment when it is low. The specific agency problem they considered is that headquarters allocates capital so as to maximize the value of the firm, but the division manager prefers maximizing their personal consumption. This model can describe the investment decisionmaking process of SOEs by regarding the State-Owned Assets Supervision and Administration Commission (SASAC) or a local SASAC as a headquarter, and regarding Chinese SOE's manager as division manager in Milton and Raviv (1996). Thus, the objective function of SOE is where u denotes the utility function of corporate manager, which is defined by equation (5), and U denotes the utility function of SASAC, which is increasing with profit. We suppose U (SC t ) = e ⇣dYt , where ⇣ is the parameter of social contribution over operating profit, 0 < ⇣ < 1. The last term in P3 indicates the manager's expectation of safe promotion or doing nothing. This also explains why entrepreneur in SOE prefer quiet life (Bertrand andMullainathan (2013), Stein (2003)) rather than empire building.

Credit constraint
State-owned property rights bring the impact on investment from corporate financing in China.
For POEs, the decision-making process is not unlimited. Since the banks are also state owned, the SOEs have no di culties on borrowing from banks. Economic stimulus plan brought them a great amount to new bank loans after the financial crisis. Therefore, on the contrary to the firms in U.S. and EU Campello et al. (2010) and Duchin et al. (2010), there is no credit constraint in China's SOEs.
This ownership discrimination leads to POEs obtain far less finance support from the existing banking system than SOEs do. For POEs, their financial constraint mainly appears as credit constraints (Miao and Wang (2012)), which means the loan can't be less than the market value of its collateral. The stopping time of financial distress, ⌧ , is when the market value of the firm's net assets, M ( K) declines below the overall loans. Therefore, the credit constraint is The above objective functions from P1 to P3 are for the financially healthy firm, therefore, their decision-making process is infinite. The credit constraint Equation (7)

Structural change on objective functions
The dash lines in Figure 2 plot the value functions in old normal, where capital stock K subject to equation (2). The solid lines represent the objective functions after the structural change. Figure   2(a) implies that for those POEs with f < 0, increases (Kahle and Stulz (2013)), the drift ratio of equation (3) decreases, and the uncertainty depends on investment and risk 2 . If the credit constraints of POE tightens, the financial distress boundary will move left, resulting in some of the enterprises with original objective function of P1 shifts to P2. Therefore, such companies expand investment. On the other side, according to equation (4) and equation (6), POEs of P1 will choose lower 2 , shifting from profit-driving to risk-averse, thus, such kind of companies will reduce their investment.
Since SOE is able to obtain the credit support from the government, their objective function does not need to satisfy the credit constraint as Equation (7). Thus, the optimization problem of SOE is unlimited. Figure 2(b) represents the shift of SOE's investment. The first term of objective function P3 comes from the agency problem which equals to Equation P2 plus the cumulative social contribution, which is denoted as P2'. Investment must be expanded in order to reach the same objective function. The last term of Equation (P3) is to ensure no loss instead of value maximization, therefore I < 0. Whether to expand investment or to reduce it depends on the objective function is tending to P2' or P3. In fact, SOE obtains a large number of bank loans due to the economic stimulus plans, i.e., f > 0 which leads to the increase of dY t , and P rob(Y t > 0) goes up according to equation (4). That's why some of the SOEs shift from P2' to P3, thus the number of SOEs reducing investment increases.
Overall, Figure 2 implies that China's investment expansion comes from the dynamic shift of objective functions, due to agency problem for POEs and SOEs. On the other hand, investment decline also exists both in POEs and SOEs.  Figure 3 shows the impact of heterogeneity other than profit-driving objective function, in which the simulations are based on the objective function P2 for POE, and P3 for SOE, respectively. We do see the investment reduction from all of simulation results. We suppose the manager's uncertain wage is increasing with the profit, i.e., w t = dY t , following He et al. (2014). In the simulation, we followed Miao and Wang (2012) to specify the credit constraint as increasing with K ⌧ , i.e.,   (2010)).

Simulations on heterogeneity
The factors that a↵ect corporate investment have been systematically sorted out and defined in Hovakimian (2009) (2009)).
Since SOEs have social responsibility, we also examine the social performance in addition to the financial performance. The social performance is defined as the log of total number of employees.

Structural change from property rights
In order to test whether there is a significant di↵erence of corporate decisions on financing and investment before and after the crisis, Di↵erence-in-Di↵erence analysis is following Lins et al. (2013): The vector it comprises a set of firm-, year-and industry-fixed e↵ects to control for the impact of business cycle fluctuations, and industry belongings, respectively. Decision i,t is a financing decision or an investment decision for company i in year t. SOE is a dummy which equals 1 if the firm is a SOE, or equals 0 if it is a POE. Crisis t is an indicator that takes the value of 1 for the years between 2008 and 2015 and is 0 otherwise. X it refers to a set of firm-specific control variables which include firm size, leverage and market-to-book ratio 8 . The parameter of interest is b, which captures the change in either financing activity or investment activity during the crisis for SOEs or POEs. Standard errors are clustered at the province level in Table 1. Lins et al. (2013) implied that family-controlled firms reduce their investment by 0.52 percent points relative to other firms during the 2008-2009 financial crisis using a sample of more than 8,500 firms from 35 countries, but there is no significant di↵erence in finance decision. We find that the median of investment rate before crisis is 0.2109, and the median of investment rate after crisis is 0.293. Overall sample shows a trend of investment expansion of Chinese enterprises, which is di↵erent from Lins et al. (2013) and other international literatures. An interesting finding is this investment expansion is heterogenous in SOEs and POEs. For POEs, the median of investment rate is 0.2709 before crisis and 0.401 after crisis, which is increasing. On the contrary, for SOEs, the median of investment rate is 0.175 before crisis and 0.167 after crisis, reducing 0.8 percent. Table 1 Panel A represents the characteristics of the finance and investment decisions with di↵erent property right. After the crisis, external financing and investment rate are significantly lower in SOEs than those in POEs, but the cash holdings of SOEs is significantly higher than those of POEs. The investment of SOE is lower than that of POE by 11.7% according to the result of DiD analysis by clustering standard error at provincial level.

New Normal or no New Normal
In order to test whether there is New Normal of China's corporate investment, we investigated the periodical di↵erence based on alternative event windows. Table 1 Panel B adds dummy variable N ewN ormal to exploit the structural changes adjusting Table 11 in Lins et al. (2013). Variable respectively. In the transitional period, the expansion of investment is 28.4% in POEs which is much higher while it is only 2.4% in SOEs. In the New Normal, POEs increase their investment compared to themselves in old normal by 0.3 percentage, while SOEs reduce their investment by 4.6 percent. Gulen and Ion (2016) found that policy uncertainty can a↵ect investment up to eight quarters. We also find China's transitional period is 2 years, too.  Lins et al. (2013). The control variables of the first column regression do not include P rofit and of the third column regression does not include Leverage. * : significant at 0.1 level. ** : significant at 0.05 level. *** : significant at 0.01 level.

Definition of expanding investment
Di↵erent with the international academic literatures that enterprises reduce investment during this financial crisis in other countries, SOEs and POEs in China have an impulse to expanding their investment because of SOE's political burden and POE's agency problem. This section is to investigate the consequence of expanding investment in which corporate performance is considered from two perspectives, financial performance (ROE) and social performance (SocialP er) 9 . The model is which is also clustered standard error at provincial level. The P erf ormance is periodical average of ROE and SocialP er, respectively. The control variable in this model is the secondary classification of manufacturing industry according to the standard of CSRC 10 . Invup is a dummy variable representing investment expansion, which equals 1 if the amount of investment increases and 0 otherwise. The index of investment expansion is as follows 11 : where I 1,i represents the ratio between the average investment after crisis and the average investment before crisis of enterprise i, I 2,i represents the ratio between the average investment in 2011⇠2012 and the average investment in 2004⇠2010,and I 3,i indicates the ratio between the average investment in 2013⇠2015 and the average investment in 2004⇠2010. Table 2 supports the analysis on the dynamic shift of the optimization objective for SOEs and POEs in section 3.

Vertical and horizontal di↵erence
For POEs, there are 492 companies increasing their investment in transitional period, which is 51.1% of the total POEs). Tightening on credit constraint drives SOEs from P1 to P2. In New Normal, with the credit financing constraint's relaxing and the booming of debt and equity financing market, some POEs return to P 1 from P 2 and become risk-averse from profit-motivated by choosing low-risk investments. Table 2 shows there are 380 enterprises whose investment in New Normal stage is higher than that in the old normal, which is 39.5% of the total POEs. For SOEs, there are 237 companies increasing investment in transitional period, and 183 companies SOEs I > 0 2 2 5 1 . 0 5 0 6 . 8 5 1 2 9 8 7 I < 0 2 9 5 -0 . 6 2 4 3 . 2 1 5 2 9 2 2 Total 520 -0.209 5.043 2934 POEs I > 0 2 3 5 1 . 2 2 2 1 4 . 3 5 7 1 0 8 6 I < 0 2 9 5 -0 . increasing investment in the New Normal which is 43.3% and 33.3% of total SOEs, respectively, either of which is less than 50%.
Compare to the old normal basis period, the median of I in transitional period and in the New Normal is -0.155 and -0.499 in SOEs, respectively, which implies that SOEs' investment is decline. Overall, China is su↵ering decrease on investment, only POEs increased their investment in transitional period (the median of I 2,i is 0.029). Therefore, I also got the same conclusion that both SOEs and POEs have decreased investment compared to the old normal, which is consistent with the existing international literature. Table 3 implies that state-owned property rights and investment expansion do matter to corporate performance. The impact from state-owned property rights works through the whole process and it is consistent in each period. SOEs perform worse than POEs do, but the social performance of SOEs is higher than that of POEs, since parameters of SOE in Table 3 are all significantly negative in Panel A, and positive in Panel B. The first part of Panel A in Table 3 shows that the financial performance increases 5% by expanding investment if the event window is simply cut by year 2008. The second and third parts divide the period into three parts, the old normal  Table 3 has showed. Three periodical regression shows that there is no di↵erence in financial performance between transitional period and the old normal, but a significant change (about 3%) in financial performance between the New Normal and the old normal.

Impact of investment expansion on corporate performance
Improving employment is a key indicator of sustainable growth of developing countries, according to development economics. Thus, this paper also tests the impact of the expansion of investment on employment rate to investigate the social performance. Panel B in Table 3 implies that the social performance of SOE is significantly higher than that of POE. SOEs do have a positive e↵ect on solving the employment problem. But the e↵ect from investment expansion is not e cient to improve sustainable growth, and the e↵ect in transitional period leads to a increase in employment rate by 2.5% (parameters are 0.961 and 0.909 respectively).The parameter of Invup is not significant in transitional period (part 2), even reduces 1.1% averagely in the New Normal (part 3).

Benchmark model and structural changes
Our benchmark model is as follows: where control variables following Hovakimian (2009) include market-to-book ratio, sales growth, asset tangibility, leverage, financial slack, dividend, age and firm size. Following Brown and Petersen (2009), lagged variable I i,t 1 is used to reduce endogeneity and lagged e↵ects (Michael and Whited (2012)). Standard errors in Table 4 are clustered at provincial level. In order to consider structural change in the investment behavior, column (4) to column (6) add two dummy variables to divide the period into three periods. Shock equals 1 if the time is after 2011. N ewN ormal equals 1 if the time is after 2013. Otherwise both of them equal 0.
In order to examine the e↵ects of the state-owned property right, we divide the total sample, 1670 manufacturing companies, into treated group (548 SOEs) and control group (964 POEs, non matched). To construct a sample of firms similar, we also match each SOE with the firm in POE subsample. Using the nearest neighbor propensity score matching method based on secondary industry, province, firm size and age, the matched control group is finally consisted of 305 POEs, with AT T = 0.235, standard error=0.103 and t-statistics=-3.174, implying that the investment of treated group (SOEs) is significantly lower than that of the control group (POEs). Table 4 shows that the parameters of CF/K in column 1 and 3 are not significant in the benchmark model while it is significant in column 2 implying that POEs present positive investmentcash flow sensitivity overall, but SOEs and similar POEs matched sample aren't sensitive.
We also investigate the structural changes of the three period in column 4 ⇠ 6. Column 4 shows that investment of SOEs are significantly positive sensitive to cash flow before the financial crisis.
Parameters of CF/K ⇥ Shock and CF/K ⇥ N ewN ormal aren't significant . This implies that SOEs are positively sensitive only in the old normal, neither in the transitional period nor in New Normal. Column 5 and column 6 show that SOEs aren't sensitive in the transitional period while they are positively significant both in old normal and New Normal. The regressions of three-period structure change model reveal that although the investment e ciency reduced during transitional period POEs has recovered in the New Normal. The empirical tests enlightened the concerns on SOEs' investment ine ciency in transitional period and in New Normal.

Agency problem
Managers are self-interested. The agency problem will reduce if the manager holds shares. The reason why this paper chose managerial ownership instead of the ratio of state-owned shares is that state-owned shares count only in un-tradable shares, which is a small part of the corporate equities now. This ratio hasn't been revealed in the tradable shares since China set o↵ reforms on split-structure of equities. In addition, the ratio of managerial ownership can be used to examine the agency problem not only in SOEs but also in POEs.
Column 1-3 of Table 5 reveal the impact from executive stock ownership on investment and its structural changes. In old normal, executive stock ownership has positive impact on investment both in SOEs and in POEs. In transitional period, the parameter of Agent⇥N ormal is significantly negative, which reveals a severe agency problem both for SOEs and POEs. In the New Normal,  He et al. (2014) suggest that it is possible to study the problem of over-investment through two perspectives, the uncertainty of profit and risk. Column (7)-(9) of Table 5 represent the impact from profit-risk trade-o↵ on investment and its structural changes. The results of regression show that in the old normal the corporate investment is not sensitive to profit for SOEs while it is significantly positive for POEs. This provides evidence of Knight (2014) that investment of POEs depends on their own corporate profit. However, this relationship has su↵ered structural changes since the recent financial crisis. The parameter of prof it ⇥ Shock of POEs in non-matching group is significantly negative, implying that POEs are no longer driven by profit (as Knight (2014)   Structural changes of investment behavior result from the shift of objective function, which is di↵erent with state-property rights. The ratio of expanding investment for SOEs compared to the old normal is 43.3% in transitional period and 33.3% in the New Normal while this ratio for POEs is 51.1% in transitional period and 39.5% in the New Normal. The financial performance of POEs is higher than that of SOEs, but SOEs have more advantage in social performance, especially in transitional period, improving the employment rate by 2.5%. The expansion of investment in the New Normal significantly improves financial performance but has no positive impact on solving the problem of employment, as a result of which, it is not able to drive the sustainable economic growth e↵ectively.
The analysis of investment-cash flow sensitivity based on the three-period structural changes implies the heterogeneity of SOEs and POEs. First, the empirical results reveal the concerns on SOEs since their investment are insensitive to cash flow both in transitional period and in the New Normal. Second, financial constrains are still one of key problem POEs face in the new normal as they were in the old normal, since their investment is positively sensitive to cash flow in these two periods. Three kinds of heterogeneity are considered in the mechanism of corporate investment behavior, agency problem, external financing and profit vs. risk trade-o↵. I find that the structural change results from executive stock ownership, which still work on relieving agency problem for SOEs while it is no longer an e cient way for POEs in the New Normal. In the second perspective, external financing, debt's driving investment doesn't work any more in the New Normal, even results in matched POEs' shrinking their investment. In the third perspective, profit vs. risk trade-o↵, POEs are no longer profit-driving as Knight (2014) described, neither in the transitional period nor in the New Normal. SOEs show a risk averse investment shrinkage in these two periods even though they got the majority amount the economic support and under China's expanding investment policies.  ROE=the annual return on net assets (in %) SocialP er=log(the amount of employees at the end of year)

Explanatory variables
Cash flow CF/K, where CF is the net cash flows from operating activities and K is the fixed assets at the beginning Property rights SOE, Dummy variable, equals 1 if it is a SOE, or equals 0 if it is a POE.

Agency problem
Agent =the manager's shareholding ratio at the beginning of the year (in %) New debt finance Debt=new debt over the initial fixed assets New equity finance Stock=new stock over the initial fixed assets Profit P rofit=the ratio of operating and operating revenue (in %) Risk Risk=Equity return volatility at the beginning of the year(in %).

Control variables
Market-to-book ratio M/B= market value over book value of company Growth Salegrowth=main business's increasing rate of income at the end of year (in %) Tangible assets T angibility=tangible assets over total assets (in %) Asset-liability ratio Leverage=leverage rate at the beginning of the year (in %) Financial slack F inslack=cash and cash equivalents over the initial fixed assets Dividend Div= the cumulative dividends over the initial fixed assets Age Age=number of years since establishment Size Size=ln(the total assets) Industry Industry: Dummy variable, classifications of manufacturing industry according to CSRC code Notes: We calculate the fluctuation ratio in the way of Logarithmic yield method based on the daily data of stock in 6 months up to January 1. In order to reduce the endogeneity, I use the volatility ratio at the beginning of the year. Fortunately, this index is not a↵ected by the stock market crash at the end of 2015.