Business environments and managers' risk appetite in emerging markets

When managers make decisions, they consistently weigh the tradeoff between risk and return (McNamara and Bromiley, 1999). If principal (shareholders) and agent (the managers) incentives are aligned and there is no agency problem, then rational and responsible managers seek to maximize the interests of the firm's shareholders by implementing optimal operational strategies and exercising prudence in investment decisions. Both the internal and external business environments exert influence on managerial decision-making, thereby impacting business risk (Buckley and Casson, 2019). Various factors, including competition within the industry, pressure from narrower profit margins, governmental regulations, social preference shifts and technological advancements, can alter a manager's business decisions and risk tolerance (Nobre et al., 2018).

There is no consensus on how to measure perceptions of risk (Schwarzkopf, 2006), but operating risk is recognized as a crucial component. The general manager shapes business strategy and operational decisions, and a faltering strategy results in diminished performance and adversely affects shareholder wealth (Parnell, 2010). Consequently, managers' risk appetite may fluctuate based on prevailing market and operating conditions (Feng and Wang, 2019). For instance, managing larger assets may incentivize managers to pursue riskier investments, while higher delegation power, performance-based contracts and reduced monitoring may further motivate risk-taking behavior (Huang and Wang, 2015), especially when contract terms are clearly tied to remuneration or promotion (Gul et al., 2003). While many firms utilize options to align managers' interests with shareholders, such contracts may significantly amplify managerial risk-taking (Chen and Ma, 2011). Government policies and regulations can impact the market, prompting not only firms but also consumers to alter their consumption patterns.

Changes in risk levels are reflected in earning quality (García-Teruel et al., 2014). Managers, in their pursuit of enhanced firm performance, may expand businesses and engage with customers of varying quality or extended settlement periods (Pratono, 2018). This risk is evident in higher accruals and lower cash, indicative of lower earning quality (Zeidan and Shapir, 2017). Earning quality serves as a measure of managerial risk-taking, where accruals should align with revenue; significant variations suggest slower cash conversion and a higher-risk strategy. Higher accrual quality may enhance future performance and return realization (Chen et al., 2008).

In 2010, Chinese authorities launched a pilot program allowing short-selling and margin-trading for a small set of 90 constituent stocks (Chang et al., 2014). The list gradually expanded to other stocks and even some exchange-traded funds (ETFs), but remained limited with a uniform threshold for margin calls (Dare Bryan et al., 2010). In 2019, a major regulatory revision eliminated the automatic margin-call threshold and more than doubled the number of eligible stocks (China Regulator Relaxes Rules on Short Selling, 2019). This dramatic policy change created a new institutional setting for managers, providing an opportunity to empirically evaluate how the relaxation of short-selling and margin-trading restrictions affected risk-taking. Managers must still balance shareholder interests, since excessive risk can negatively affect share prices (MacKay and Moeller, 2007). However, margin-trading tends to provide price support from below, which may embolden managers to take on greater risk (Ye et al., 2020). Similarly, the Shanghai-Hong Kong and Shenzhen-Hong Kong Stock Connect programs, introduced in 2014 and 2016, respectively, increased foreign demand for selected shares, further reinforcing managerial incentives toward risk-taking. Focusing on manufacturing firms also allows us to assess how China's unique corporate governance structure – including the prevalence of state-owned enterprises (SOEs) and the common practice of share pledging – shapes these dynamics.

The evolution of corporate governance frameworks has increasingly emphasized performance-based compensation to align managerial incentives with those of shareholders. While this approach is intended to mitigate agency problems, it raises important questions: When firms exhibit strong liquidity and operational performance, do managers remain content with stable outcomes, or are they incentivized to pursue additional rewards through riskier strategies? Moreover, in response to policy-induced shifts in the external business environment, might managers exploit new opportunities by further increasing risk exposure? Such shifts in managerial risk appetite, while potentially rational from an incentive-based perspective, may inadvertently introduce new agency conflicts, reduce shareholder welfare, and challenge the adequacy of simplistic performance-based compensation mechanisms. This research investigates the risk appetite dynamics of listed Chinese manufacturing firms regarding earnings quality, using panel data and the difference-in-difference (DID) method that accounts for major regulatory changes described above. The study explores how Chinese-specific business practices and policy changes shaped risk-taking behaviors.

This research contributes to existing literature in several ways. Firstly, it connects external environmental changes with managerial behavior, providing insights that go beyond the performance-related focus of existing research on earning quality and risk-taking (see, for example, Savaser and Şişli-Ciamarra, 2017). Also, the research differs from past risk measures. The past risk measures usually focus on return on asset (ROA) volatility (Jiang and Chen, 2021; Li et al., 2013a, b; Wen et al., 2021), which could be affected by other non-operating incomes. We use fundamental cash collection and conversion deterioration in firm operations to better exclude the noise. Secondly, unlike previous Chinese margin trade and short sell studies that primarily concentrate on the initial implementation of the pilot program in 2010 through event study methodology, this research examines the effects of the significant 2019 modification to the existing policy, assessing behavioral changes rather than just firm performance. Finally, the correlation established between managers' risk-seeking behavior and different business environments contributes to the theoretical understanding of performance-based contracts, designed to mitigate agency problems by aligning managerial and shareholder interests, even though they may inadvertently heighten managers' risk-seeking tendencies and impact shareholders' interests at critical junctures.

Earnings quality is intricately tied to management and investment decisions (Li, 2011) and directly influenced by how managers choose customer contracts and expedite cash collection post-sales (Dichev et al., 2013). Other stakeholder relationships, such as payments to suppliers, can impact earning quality as well (García-Teruel et al., 2014). Lower earning quality does not necessarily translate to lower profits or heightened industrial competition; it may also result from aggressive business expansion or broader economic growth (Ma and Ma, 2017). There exists a tradeoff between pursuing higher performance and entering more conservative, quick cash turnover contracts with customers (Khuong et al., 2022). In China, “share pledging”, the pledging of shares as collateral for funds for reinvestment in the same firm, has become a common way to finance expansion of listed firms (Li et al., 2022a, b; Feng et al., 2023). Stock pledges can significantly alter the risk-taking attitude of managers and larger shareholders (Huang et al., 2022; Pang and Wang, 2020), potentially undermining corporate governance and even increasing the likelihood of fraud (Han et al., 2023). From the perspective of new capital infusion, the firm needs to expand production after receiving injected funds, potentially lowering earning quality and increasing the risk of stock crashes (Li et al., 2022a, b). Additionally, shareholders have a strong incentive to showcase that the firm can generate larger profits to maintain or increase the firm's value and share performance, avoiding collateral margin calls (Li et al., 2019). Therefore, behaviors aimed at business expansion are associated with an amplification of managers' risk appetite. Hypothesis 1 below articulates how managers' aspirations to expand their enterprise's business impact their risk appetitive in the Chinese market.

SOEs have complex structures and experience a so-called dual agency problem (Xu et al., 2005). The objectives of SOEs often extend beyond profit maximization, focusing on meeting social obligations (Cunningham, 2011). SOEs in the Chinese market differ significantly from those solely providing commercial services on behalf of the government (Lin, 2021). Many SOEs are listed on stock exchanges, permitted by governments to generate substantial profits to satisfy the interests of investors providing funding (Song et al., 2015; Carpenter et al., 2021). Evidence suggests that market-oriented SOEs, supported by government policy and bank funding, can sustain growth with centralized management (Lin and Germain, 2003). However, the causality of such growth is influenced by government support and the efficacy of centralized management (Lo, 2020). These features contribute to managerial compensation in SOEs that resemble those in private firms – often linked to profitability but also tied to bureaucratic promotion pathways (Xin et al., 2019). Managers of SOEs are often appointed by governments and given the government's dominant role as the largest shareholder, they are less likely to be influenced by other investors during board voting elections (Shen et al., 2020). SOEs in the Chinese market typically adhere to policy guidance (Liu et al., 2019), and because profitability is a key performance metric, managers may have greater incentives for risk-taking behavior – especially when seeking promotion (Lou et al., 2021; Zhu et al., 2022; Su et al., 2020).

On one hand, SOE managers may take more risks due to performance-linked promotions, potentially lowering earnings quality, as reflected in Hypothesis 2a. On the other hand, the structure of SOEs may mitigate the effect of share pledging on risk appetite due to government backing. This implicit guarantee means SOEs face fewer credit constraints and may have less need to rely on share pledging in the first place (Cao et al., 2023). Even when share pledging occurs, the associated agency costs are likely to be lower, reducing its influence on risk-taking behavior. Hypothesis 2b reflects this possibility.

Prior to 2010, short-selling and margin-trading were banned in China. This changed with the pilot program of 2010, which initially identified a small number of firms eligible for short-selling and margin-trading. Short selling serves to regulate firms by reducing information asymmetry and enhancing the quality of corporate governance (Jin et al., 2018; Deng and Gao, 2018). After the policy change, market price discovery efficiency increased (Gao and Ding, 2019), leading to reduced earnings management (Jiang and Chen, 2019) and improved investment efficiency (Hu et al., 2020; Wu et al., 2022).

Early market reactions showed negative abnormal returns for newly eligible firms, as investors initially avoided stocks open to short-selling (Chang et al., 2014). Over time, successful indexed firms with stronger performance (e.g. higher Tobin's Q) showed reduced incentives to pursue further investments (Dasilas, 2022). Managers also became less willing to disclose qualitative or soft information, increasing information asymmetry (Xie et al., 2021).

Figure 1 illustrates that margin-trading has consistently outweighed short-selling, even after the 2019 expansion of firm eligibility. This imbalance may be partly explained by the removal of automatic margin-call thresholds in 2019. Because margin-trading remains far more prevalent, we expect that eligibility for these policies is more likely to increase managerial risk-taking than curb it. Managers may respond by increasing risk in pursuit of improved performance, rather than simply expanding market share.

Short-selling introduces external monitoring, which helps distinguish deteriorating earnings quality from earnings management, suggesting instead that the decline may be due to aggressive sales strategies or overexpansion. Recent studies link higher idiosyncratic risk in indexed firms to varied managerial approaches and investor expectations (Gui and Zhu, 2021), pointing again to managerial risk appetite as a key factor.

This study explores how the policy change on short-selling and margin-trading affected those firms identified as eligible for short-selling and margin-trading after the policy change: the average effect of treatment (the policy change) on the treated (the firms identified as eligible).

Table 1 summarizes the hypotheses and the major-related literature.

The accrual level of a firm's earnings is used here as an indicator of risk-taking behavior. Earnings may be affected by different accounting treatments and are subject to manipulation, and estimated cash flows may also be inaccurate (Pornupatham et al., 2023). However, cash flow itself is difficult to manipulate, and managers generally have incentives to reduce accrual levels to demonstrate higher management quality (McInnis and Collins, 2011). A higher level of accruals indicates that a firm is extending its cash conversion cycle and taking longer to collect revenues. Such an outcome is usually the result of managerial risk-taking decisions, including market expansion and increased competition for market share (Datta et al., 2017). Thus, earnings quality uncertainty serves as a proxy for managerial risk-taking, since it reflects operating strategies that involve greater exposure to uncollectable debt or slower revenue realization (Armstrong et al., 2013; Li et al., 2013a, b). If risk-taking behavior is the measured outcome at the accrual level, risk appetite is the underlying mechanism that incentivizes managers to engage in further risk-taking investment strategies. Therefore, this research investigates risk appetite through the effect of a changing business environment on risk-taking.

Following Dechow and Dichev (2002), we define risk-taking as the unexplained variation in earnings quality, Uncertainty. This approach models accruals as a function of past, current and future operating cash flow. If cash flow does not adequately explain variation in accruals, even after controlling for sales growth and the net value of property, plant and equipment, then earnings quality is uncertain and risk is higher.

Equation (1) reflects these associated relationships: Uncertainty is quantified by the absolute value of the residual term in equation (1), $εi,t$⁠, with the quality of earnings decreasing in the magnitude of that estimation error as in Dechow and Dichev (2002).

In Equation (1), accruals for enterprise i, whether private or state-owned, at time t are calculated based on past, current, and future cash flow operations (CFO), the change in sales (⁠$Δsales$⁠), and the net property, plant and equipment (PPE) over total assets of the previous period. The accruals calculation includes a residual, the absolute value of which serves as our measure of risk, Uncertainty.

Furthermore, we adopt an alternative measure proposed by Kothari et al. (2005), as shown in Equation (2), to address potential endogeneity concerns. The details of this method are provided in the Endogeneity and Robustness subsection within the Methodology section below.

This study examines the risk appetite of firm managers in the Chinese market. China is chosen as the primary research context for several reasons: the widespread practice of large shareholders pledging shares to banks, the coexistence of private and SOEs, and the significant 2019 policy reform that expanded margin trading and short selling. Each of these factors has the potential to shape managerial risk preferences.

The data were obtained from the East Money (Choice) database for the period 2017–2021. The sample includes 3,556 manufacturing firms listed on the stock exchange before 2017, yielding a total of 17,780 firm-year observations. To ensure data validity, Augmented Dickey–Fuller (ADF) tests were conducted on the three key variables: the dependent variable Uncertainty, the alternative dependent variable Variation, and the independent variable Pledge. All three variables were found to be stationary. Definitions of the variables are presented in Table 2, and Table 3 reports the descriptive statistics for the sample.

According to prior literature, firm performance is influenced by a range of factors, including managerial incentives, capital structure and dividend policies (Bromiley, 1991). When managers receive incentive contracts and share options as part of their compensation, the convexity effect of such contracts can significantly increase their willingness to take risks, particularly when the firm is exposed to downside (left-tail) performance risk (Gormley et al., 2013). A firm's capital structure, shaped by the cost of debt and leverage dynamics, also affects its risk profile and strategic choices (DeAngelo et al., 2011; La Rocca et al., 2011; Zellweger, 2007). Similarly, dividend distribution policies can influence firm valuation, risk-taking and long-term strategies (Dionne and Ouederni, 2011; Rajverma et al., 2019). To account for these influences, we include capital structure and dividend-related variables as controls, ensuring that the effects of share pledging and policy reforms (such as changes to corporate law) are properly isolated.

Our first and second empirical tests correspond to Hypotheses H1 and H2 (including H2a and H2b). These tests evaluate how corporate governance – specifically, share pledging and state ownership – affect managerial risk-taking behavior. The main variables of interest are the number of shares pledged by the largest shareholder (Pledge) and the state-owned enterprise dummy variable (SOE). To examine whether the governance effects differ when SOEs engage in share pledging, we also include the interaction term between Pledge and SOE. Equations (3) and (4) formalize these relationships.

We expect SOEs to be less likely than private firms to pledge shares, since profit maximization is not their sole objective. Moreover, SOEs typically receive support from local governments, which makes it easier for them to obtain financing by issuing debt directly or borrowing from state-owned banks.

The hypotheses imply the following coefficient expectations:

1. H1: When managers of privately owned enterprises pledge shares, their risk-taking behavior should increase. Thus, $β1$​ is expected to be positive.

2. H2a: Because government protection provides SOE managers with incentives for additional risk-taking, $β2$ should also be positive and statistically significant.

3. H2b: State ownership should reduce the effect of share pledging on risk appetite. Therefore, the interaction coefficient $β6$ is expected to be negative and statistically significant.

Equations (3) and (4) include both time-fixed effects (YEAR) and industry-fixed effects (IND), as well as control variables such as cost of capital (CCAP), dividends paid (DIV), return on equity (ROE) and return on assets (ROA). The error term is denoted by $ε$⁠.

Next, we turn to DID analysis to explore hypothesis H3, which addresses the effects of the 2019 change in the number of firms eligible for short-selling and margin-trading, as well as the removal of margin call thresholds, on risk-taking of those firms identified as eligible for short-selling and margin-trading.

In addition to the share pledging and state-owned enterprise variables analyzed above, the DID analysis includes dummy variables for the post-policy implementation period, POLICY, which takes the value of one for any year after the 2019 policy change, and a dummy variable for treatment, a dummy variable that takes the value of one for those firms designated as eligible for short-selling and margin-trading by policymakers, INDEX. If the 2019 policy change, increasing the number of firms eligible for short-selling and margin-trading as well as removing margin call thresholds, increased risk-taking at all firms, regardless of their eligibility for short-selling and margin-trading after 2019, then the post-policy dummy variable, POLICY, will have a positive and statistically significant coefficient estimate. The main parameter estimate of interest, however, is the coefficient estimate on the interaction term between POLICYxINDEX. That is the “difference-in-difference” term, which measures the average effect of treatment, the 2019 policy change, on the treated, those firms identified as eligible for short-selling and margin-trading under the policy change.

In addition to the DID term, the share pledge amount, PLEDGE, and SOE dummy variable are interacted with the policy treatment to test whether they change the average effect of policy treatment on managers' risk-taking preferences. Since both share pledges and the larger agency problems inherent at SOEs are expected to increase risk preferences even further, these triple interaction terms are expected to have positive and statistically significant coefficient estimates.

Equations (5) and (6) are used to test these relationships.

The “average effect of treatment on the treated” effect is estimated by $β8$ in Equations (5) and (6). If this “average effect of treatment on the treated” is statistically significantly different for firms that have more shares pledged, or for firms that are SOEs, then coefficient $β9$ in Equation (5) and (6) are expected to be statistically significantly different from zero as well.

Our first robustness check is to separate our sample into two sub-samples – firms listed on the Shanghai Exchange and firms listed on the Shenzhen Exchange – in case firm heterogeneity across the two exchanges is skewing our results. In addition to running our analysis separately on these two sub-samples, we include a dummy variable for firms on each exchange that have been identified by authorities as eligible for purchase not only within China by Chinese investors, but also by foreign capital on the Hong Kong exchange. These are firms on each exchange that have been identified as eligible to participate in the Shanghai-Hong Kong Stock Connect program and Shenzhen-Hong Kong Stock Connect program.

We expect that designation as a firm eligible for purchase by foreign capital on the Hong Kong exchange may affect risk-taking since firm managers are likely cognizant that being identified as one of those firms eligible for purchase by foreign capital through the Hong Kong exchange likely results in heightened demand for firm shares, supporting share prices and shielding managers from the negative side effects of higher risk.

Equations (7) and (8) empirically analyze the impact of the variable identified above – share pledging and agency problems inherent at SOEs – on risk taking for firms listed on the Shanghai and Shenzhen exchange, respectively. In addition, equations (7) and (8) analyze the effect on managers' risk-taking when firms listed in the Shanghai and Shenzhen exchanges are identified as firms eligible for purchase through Hong Kong by foreign capital.

Managers, understanding the relationship between the Shanghai and Shenzhen indices and share market prices, are expected to exhibit higher risk-taking behaviors when designated as being eligible for purchase by foreign capital through Hong Kong. Therefore, the coefficient estimates of $β6$ are expected to be positive and statistically significantly different from zero. Additionally, an interaction term is included to address instances of SOEs being eligible for purchase by foreign capital through the Hong Kong exchange $SOEi,t*SHANGHAIi,t$ and $SOEi,t*SHENZHENi,t$ . We expect the coefficient estimates on this interaction term, parameter $β7$ in Equations (7) and (8), to also be positive.

As a final robustness check, we examine our hypotheses using an alternative measure of risk-taking: earnings quality variation. Risk-taking as measured by earnings quality variation is project-based rather than cash flow-based. In this second measure, we follow Kothari et al. (2005) in defining earnings quality variation or risk-taking as Variation. This second measure takes into account the fact that an increase in sales may involve transactions with lower-quality customers, postponing the actual receipt of payment: a long cash conversion cycle. Equation (9) is the same as Equation (2) mentioned above, and it reflects this perspective, where accruals are explained by the expansion in sales, any fixed investments and the unexplained portion remains in the residual. A larger absolute value of the residual term in equation (9), $εi,t$⁠, is indicative of Variation, or decreased earning quality (Kothari et al., 2005).

In Equation (9), accruals are calculated based on the difference between the change in sales (⁠$Δsales$⁠) and the net change in accounts receivables (AR), the net property, plant and equipment (PPE), the return on assets (ROA) and managers' risk-taking behavior, expressed by the error term.

This alternative measure of risk, Variation, is incorporated into the DID analysis explained above. Equations (10) and (11) reflect the policy effect on the discretionary accruals measure of earnings quality and risk.

Equations (10) and (11) extend the framework established in Equations (7) and (8) by introducing a distinct dependent variable. These models incorporate a policy change term $β6$⁠, a treatment term $β7$⁠, an interaction term between the policy change and the treated firms $β8$⁠, the “difference-in-difference” term which measures the average effect of the policy change on the treated firms, as well as a triple interaction term representing the policy treatment applied to firms with share pledges or SOEs $β9$⁠. All DID and treatment estimates as well as coefficient estimates on the triple interaction terms, are expected to be positive.

Table 4 and Table 5 present the results of Equations (3) and (4), using alternative specifications of time-fixed and industry-fixed effects. The coefficients on PLEDGE are positive and statistically significant across all models, indicating that larger share pledges increase uncertainty in earnings quality. In the baseline model (Table 4, Column 3), an additional one million shares pledged increases the firm's uncertainty score by 0.85 units. This heightened uncertainty reflects lower earnings quality and greater managerial risk-taking.

Similarly, the coefficients on SOE are positive and statistically significant in both tables, suggesting that the multi-agency problems inherent in SOEs amplify accrual uncertainty and reduce earnings quality. In the baseline model (Table 4, Column 3), SOEs have, on average, uncertainty scores 487 units higher than non-SOEs. Given that the mean value of Uncertainty is 398 (Table 1), this indicates that SOEs exhibit substantially greater uncertainty relative to their private counterparts.

By contrast, the interaction term PLEDGE × SOE has a negative and statistically significant coefficient, suggesting that the impact of share pledging on risk-taking is mitigated at SOEs. This likely reflects the fact that SOEs have less need to rely on share pledging for financing, given their well-documented preferential access to capital through state support and bank lending.

Finally, the bottom panel of Table 4 reports generalized variance inflation factors (GVIFs) for the baseline specification. All values are only slightly above one, indicating that multicollinearity is not a concern.

Taken together, the results in Table 4 and Table 5 provide strong support for Hypotheses H1, H2a, and H2b.

Figure 2 presents the parallel trends plot of the absolute error term estimated by Equation (1). The treatment group, representing indexed firms, is shown with a blue dashed line, while the control group of non-indexed firms is represented by a red dashed line. Before the treatment year, both groups follow similar trends, but the trajectories diverge afterward, reflecting the treatment effect.

Table 6 reports the results of Equations (5) and (6), which test Hypothesis H3 – that the 2019 policy change increased risk-taking at affected firms. The coefficient on the DIDs interaction term (INDEX × POLICY) is positive and statistically significant in Columns (1) and (2), indicating that risk-taking increased among eligible firms after the policy reform. This confirms that managers of firms entering the margin-trading index adopted riskier strategies, as reflected in higher uncertainty scores.

Columns (3) and (4) further show that share pledges and SOE ownership not only enhance the overall level of risk-taking but, in some cases, outweigh the direct effect of the policy change.

Overall, the evidence presented in Table 6 supports Hypothesis H3, demonstrating that the 2019 policy reform significantly increased managerial risk-taking in firms designated as eligible for margin-trading and short-selling.

To verify the robustness of the results, we conduct several additional analyses. First, we explore possible heterogeneity across firms listed on the Shanghai and Shenzhen exchanges. Within this analysis, we also examine whether designation as a firm eligible for purchase by foreign capital through the Hong Kong exchange affects managerial risk-taking. Secondly, we explore the DID analysis of the 2019 policy change examined above, using an alternative measure of risk-taking and earnings quality.

Tables 7 and 8 report the results of Equations (10) and (11) for two sub-samples: firms listed on the Shanghai exchange and firms listed on the Shenzhen exchange. Each specification includes a dummy variable – SHANGHAI or SHENZHEN – that equals one when the firm is listed on the respective exchange and also designated by policymakers as eligible for purchase by foreign investors through the Hong Kong Stock Connect program. This policy allows foreign capital to trade selected shares listed on the mainland exchanges through arrangements with the Hong Kong exchange.

In both tables, share pledges and state ownership continue to positively contribute to managerial risk-taking, confirming that our main findings are robust to potential heterogeneity across exchanges. The coefficients on SHANGHAI and SHENZHEN are positive and statistically significant, indicating that firms identified as eligible for foreign purchase exhibit higher managerial risk appetite. Moreover, the interaction terms between SOE status and eligibility for foreign purchase are also positive and statistically significant, suggesting that state ownership amplifies the effect of foreign investor access on managerial risk-taking.

Next, we test the robustness of our findings using an alternative measure of risk-taking: earnings quality variation. Figure 3 plots the parallel trends for treatment and control groups. The trends are similar prior to the policy change, but diverge afterward, again reflecting a treatment effect.

We then re-estimate the DIDs models with Variation as the dependent variable. The results, reported in Table 9, are consistent with those obtained using our original measure of uncertainty. In columns (1) and (2), the coefficient estimates on the DIDs term—which captures the average effect of the policy change on treated firms – remain positive and statistically significant. When a triple interaction term between PLEDGE and POLICY × INDEX is included, the treatment effect continues to hold. However, once the interaction with the SOE dummy is added, the treatment effect is absorbed, suggesting that the increase in risk-taking was especially pronounced at SOEs.