Public Health System Credibility and Firm-Level Financial Risk: A Panel Data Analysis

Abstract

Introduction

Public health system credibility may influence economic behavior during health emergencies. Limited evidence examines its association with firm-level financial risk.

Methods

This longitudinal panel study uses data from non-financial A-share listed firms between 2017 and 2024. Firm-level stock return data were linked to regional measures of public health system credibility. Financial risk was measured using stock price crash risk and downside volatility. Fixed-effects regression models were estimated. Mediation analysis was conducted to examine behavioral channels.

Results

Negative public health credibility shocks were associated with higher stock price crash risk and downside volatility. Employee withdrawal and consumer demand reduction were associated with this relationship.

Conclusion

Public health system credibility is associated with firm-level financial risk during health emergencies.

Keywords

public health system credibility financial risk stock price crash risk behavioral withdrawal health governance financial stability G1 G32 I18 D81

1. Introduction

Public health systems play a central role in managing uncertainty during health emergencies. Their function extends beyond the provision of medical services. They also serve as key information providers that shape risk perception and guide behavior at both individual and organizational levels. A growing body of research shows that trust in health authorities is closely associated with public compliance, emotional stability, and social coordination during crises (Siegrist & Zingg, 2014; van der Weerd et al., 2011).^1,2 When such trust is weakened, uncertainty intensifies and behavioral responses become more volatile.

Prior studies in public health emphasize that institutional trust is essential for effective crisis management. Transparent communication and consistent policy signals help stabilize expectations and reduce panic, thereby encouraging cooperative behavior (Quinn et al., 2013; Blair et al., 2017).^3,4 In contrast, frequent data revisions, conflicting messages, or abrupt policy reversals may erode confidence in health systems (Malecki et al., 2021).⁵ These credibility problems can alter how individuals interpret health risks and respond to official guidance, even when objective conditions remain unchanged.

Behavioral responses to health-related uncertainty have been widely documented. During epidemic outbreaks, individuals often reduce mobility, avoid workplaces, and postpone consumption, even in the absence of formal restrictions (Ferguson et al., 2020; Goolsbee & Syverson, 2021).^6,7 Importantly, these responses are not driven solely by infection rates or mortality risks. Public health research suggests that perceived credibility of health information plays an independent role in shaping individual behavior (Lep et al., 2020).⁸ When information is unclear or contested, individuals rely more heavily on personal judgment, which may amplify withdrawal behavior.

Health shocks also generate substantial economic consequences (Mazur et al., 2021)⁹ Existing evidence links epidemics and pandemics to lower economic growth, labor market disruptions, and productivity losses (Bloom & Canning, 2006; McKibbin & Fernando, 2020).^5,10 At the firm level, health-related disturbances affect production continuity, supply chains, and demand conditions (Baker et al., 2020).⁴ These effects often arise indirectly through changes in behavior rather than through direct medical costs or physical damage to firm assets.

Financial markets are particularly sensitive to uncertainty and information quality. A large finance literature documents that unexpected shocks increase volatility and downside risk (Pastor & Veronesi, 2012; Bekaert et al., 2014).^11,12 During health crises, stock markets frequently experience sharp declines and heightened risk, especially in the early stages of outbreaks (Goodell, 2020; Ashraf, 2020).^13,14 Event-based studies further show that market reactions are influenced not only by reported infection numbers, but also by policy announcements and public signals issued by authorities (Alfaro et al., 2020).¹⁵

Although prior research links pandemics to stock returns and volatility across countries (Baker et al., 2020; Mazur et al., 2021),^4,9 most studies focus on the severity of outbreaks, such as confirmed cases or death rates. Comparatively little attention has been paid to the characteristics of public health systems themselves. Research on universal health coverage suggests that stronger health systems can mitigate economic vulnerability during health shocks (Stuckler et al., 2010; Tang et al., 2022).^16,17 However, the role of public health system credibility and trust as sources of firm-level financial risk remains underexplored.

In this study, we distinguish between public health system credibility and credibility shocks. Public health system credibility refers to the stability and consistency of officially disclosed public health information at the regional level. It reflects the extent to which institutional communication follows predictable and coherent patterns over time.

In contrast, public health system credibility shocks refer to deviations from this baseline. These shocks capture changes in the regularity of reporting, the occurrence of data revisions, and the stability of policy communication. As such, credibility represents a level-based institutional characteristic, whereas credibility shocks represent time-varying disturbances in that characteristic.

This distinction is conceptually important. Health severity reflects the objective intensity of a health shock, while policy stringency captures the restrictiveness of government interventions. Media narratives, in contrast, reflect information framing in public discourse. Public health system credibility, as defined here, concerns the reliability of institutional information production itself. Even under moderate health conditions, inconsistent reporting or abrupt data revisions may weaken confidence and increase uncertainty. Conversely, stable and transparent information governance can anchor expectations during severe outbreaks. From an economic perspective, credibility defines the baseline reliability of institutional information, while credibility shocks represent changes in that reliability over time. The empirical analysis focuses on these shocks rather than on the level of credibility itself.

From an organizational perspective, credibility shocks affect firms primarily through labor and consumer channels. Health-related uncertainty influences employee absenteeism, job engagement, and productivity (Johns, 2010; De Vries et al., 2016).^18,19 On the demand side, health concerns reduce in-person consumption and increase precautionary behavior among consumers (Adda, 2016).²⁰ At the firm level, these behavioral responses are transmitted through internal organizational processes rather than through immediate market adjustment. Employee withdrawal disrupts production schedules, delays project execution, and weakens internal monitoring, making it more difficult for managers to assess short-term performance accurately. On the demand side, irregular sales patterns and order cancellations complicate cash flow management and increase forecasting errors. In such environments, negative operational signals may accumulate internally before being fully disclosed to external investors.

Financial theory suggests that when adverse information accumulates but is not promptly reflected in prices, firms face elevated stock price crash risk once the information is released (Hong & Stein, 2003; Jin & Myers, 2006).^21,22 Health-related operational disruptions induced by credibility erosion may initially remain opaque to investors. As their effects gradually become visible, stock prices may adjust abruptly, leading to heightened downside risk. This mechanism provides a direct link between public health credibility shocks and firm-level financial fragility.

This study examines whether credibility shocks in public health systems increase financial risk among Chinese listed firms. China offers a suitable research setting due to its centralized health governance structure and substantial regional variation in information disclosure and implementation practices. Using firm-level stock data and regional indicators of public health credibility, this study analyzes stock price crash risk and downside volatility. The focus is not on medical outcomes, but on the economic and financial consequences of health system trust.

By distinguishing between public health system credibility and credibility shocks, this study extends the literature on health-related economic shocks and contributes to research on institutional trust and financial stability.

The remainder of this paper is organized as follows. Section 2 reviews the relevant literature and develops the hypotheses. Section 3 outlines the research design and methodology. Section 4 presents the empirical findings. Section 5 discusses the theoretical contributions, managerial implications, limitations, and directions for future research.

2. Relevant Literature and Hypotheses Development

2.1. Public Health System Credibility and Firm-Level Financial Risk

Public health systems influence economic behavior by shaping how individuals perceive risk. When health authorities provide consistent and reliable information, uncertainty is reduced. Trust encourages individuals to maintain normal work and consumption activities (Siegrist & Zingg, 2014; Quinn et al., 2013).^1,3 In contrast, when credibility is weakened, individuals may question official guidance and rely on personal judgment. This often leads to caution and withdrawal from daily activities (van der Weerd et al., 2011; Malecki et al., 2021).^2,16 Previous research shows that health crises increase uncertainty in financial markets. Stock prices react strongly when information is unclear or contested (Goodell, 2020; Baker et al., 2020).^4,13 However, most studies focus on the severity of health shocks, such as infection rates or mortality. Fewer studies consider whether credibility problems in public health systems themselves act as a source of risk.

From an information perspective, credibility shocks increase uncertainty regarding the reliability of public signals. When official health information becomes less stable, individuals and firms face greater difficulty in forming expectations about future conditions. This increase in uncertainty affects not only perceptions of health risk but also expectations about economic activity.

At the firm level, credibility shocks influence operations through indirect channels. Changes in labor participation and consumer demand may not be immediately observable to external investors. As a result, negative operational signals can accumulate within the firm before being reflected in market prices.

Finance theory suggests that when adverse information is gradually accumulated but not promptly disclosed, stock prices may adjust discontinuously once the information becomes available (Hong & Stein, 2003; Jin & Myers, 2006).^21,22 This mechanism links information opacity to stock price crash risk.

Based on this reasoning, credibility shocks are expected to increase firm-level financial risk.

Hypothesis 1

Firms exposed to public health system credibility shocks experience higher financial risk, as reflected in increased stock price crash risk and downside volatility.

2.2. Employee Behavioral Withdrawal as a Transmission Channel

Public health systems affect economic outcomes mainly through their role in shaping risk-related information. When health authorities provide consistent and transparent communication, uncertainty is reduced and individuals tend to maintain regular work and consumption activities (Siegrist & Zingg, 2014; Quinn et al., 2013).^1,3 Under such conditions, expectations remain relatively stable, which helps limit abrupt behavioral responses to health threats.

When public health system credibility weakens, the way risks are interpreted changes. Inconsistent reporting, frequent revisions, or conflicting messages encourage individuals to rely on personal judgment rather than official guidance (van der Weerd et al., 2011; Malecki et al., 2021).^2,9 This shift raises uncertainty and leads to precautionary behavior that may exceed what objective health conditions would justify.

Financial markets are sensitive to changes in uncertainty and information quality. Prior studies document that stock prices respond sharply when uncertainty rises or when information becomes difficult to interpret (Goodell, 2020; Baker et al., 2020).^4,13 Most existing research, however, concentrates on observable indicators of health shocks, such as infection or mortality rates. The informational role of public health systems themselves has received much less attention.

Employee withdrawal represents an internal operational channel through which credibility shocks affect firms. When public health information is perceived as unstable, employees may reduce workplace participation or adjust their behavior in response to increased uncertainty.

These changes can disrupt production processes and reduce coordination within the firm. Importantly, such disruptions are not immediately observable to external investors. Instead, they affect internal performance signals, including project delays and reduced monitoring effectiveness.

As these negative signals accumulate within the firm, managers may face increased difficulty in assessing short-term performance. If the disclosure of these signals is delayed, the probability of a sudden price adjustment increases(Hong & Stein, 2003; Jin & Myers, 2006).^21,22 This mechanism is consistent with the information accumulation explanation of crash risk.

Therefore, employee withdrawal is expected to mediate the relationship between credibility shocks and firm-level financial risk.This reasoning leads to the expectation that credibility problems in public health systems can increase firm-level financial risk even in the absence of severe health outcomes. Thus, we have proposed the following:

Hypothesis 2

Employee behavioral withdrawal mediates the relationship between public health system credibility shocks and firm-level financial risk.

2.3. Consumer Withdrawal and Demand Reduction as a Transmission Channel

To formalize this mechanism, the analysis distinguishes between information shocks, behavioral responses, and financial outcomes.

Consumer behavior is also sensitive to health-related uncertainty. Economic studies show that individuals reduce mobility and postpone consumption when health risks are perceived to be high (Adda, 2016; Goolsbee & Syverson, 2021).^7,20 These adjustments frequently occur without mandatory restrictions and reflect precautionary responses driven by uncertainty.

Public health communication plays a central role in shaping consumer confidence. Transparent and consistent information helps limit excessive fear and supports more stable consumption patterns (Siegrist & Zingg, 2014)¹ When health information is viewed as unreliable, consumers may overestimate risk and withdraw from market activities, leading to demand contraction across multiple sectors.

Consumer withdrawal represents an external demand channel through which credibility shocks affect firms. When public health information is perceived as unreliable, consumers may increase precautionary behavior and reduce spending.

For firms, reduced demand affects revenue stability and increases uncertainty in cash flow projections. These effects may develop gradually and are not immediately incorporated into financial reporting. As demand-related signals accumulate, investors may revise expectations once these signals become visible. This process can lead to discontinuous price adjustments and increased downside risk (Hong & Stein, 2003).²¹ Thus, consumer withdrawal is expected to mediate the relationship between credibility shocks and firm-level financial risk.

Hypothesis 3

Consumer withdrawal mediates the relationship between public health system credibility shocks and firm-level financial risk.

3. Method

3.1. Study Design

This study is a longitudinal panel analysis using firm-year observations from 2017 to 2024. The design follows the STROBE reporting guideline for observational studies. The unit of analysis is the firm-year.

3.2. Setting and Data Sources

Firm-level financial and stock return data were obtained from the China Stock Market and Accounting Research (CSMAR) database, which provides standardized financial statements and daily stock return data for Chinese listed firms.

Public health information was collected from official provincial government sources, including publicly released health bulletins, statistical reports, and policy announcements issued by provincial health commissions. These documents were manually compiled and coded to construct measures of reporting regularity, data revisions, and policy communication stability.

Firm-level observations were matched to provincial data based on the registered headquarters location of each firm and the corresponding fiscal year.

3.3. Study Population

The study population includes all non-financial A-share listed firms with identifiable provincial headquarters and available financial data during 2017-2024.

Inclusion criteria:

Listed on the A-share market during the study period

Availability of complete firm-level financial and stock return data

Ability to match to provincial-level public health information

Exclusion criteria:

Classified as financial institutions

Missing key financial variables

Designated as ST or *ST

Incomplete or inconsistent regional identifiers

The final sample consists of approximately 3,200 firms, yielding about 21,000 firm-year observations.

3.4. Variables

3.4.1. Exposure

The primary exposure variable is public health system credibility shock. This variable captures temporal deviations in the stability of institutional public health information reporting.

While public health system credibility refers to the baseline level of consistency in official disclosures, the empirical measure used in this study focuses on shocks to that baseline. Higher values indicate greater instability relative to prior reporting patterns, and are interpreted as negative credibility shocks.

The credibility shock variable reflects institutional-level changes in the consistency and stability of officially disclosed public health information. Specifically, it is constructed based on year-over-year deviations in the regularity of provincial public health statistical releases and policy communications. The measure captures observable fluctuations in formal information reporting practices rather than survey-based trust perceptions, individual attitudes, or media coverage intensity. By construction, it reflects changes in institutional information reliability rather than disease severity or policy stringency.

3.4.2. Behavioral Withdrawal Measures

Employee withdrawal and consumer withdrawal are constructed as firm-level proxy variables capturing observable operational patterns consistent with behavioral responses during periods of heightened public health uncertainty.

Employee withdrawal is proxied by abnormal changes in labor intensity, defined as deviations in firm-level employment scaled by total assets relative to industry-year averages based on CSRC industry classifications. This indicator captures reductions in labor input consistent with workplace adjustment during credibility shocks.

Consumer withdrawal is proxied by abnormal revenue contraction, measured as deviations in firm-level sales growth relative to industry-year averages. This measure reflects demand reductions consistent with precautionary consumption behavior during periods of uncertainty.

Both variables are standardized to have a mean of zero and a standard deviation of one to facilitate comparability across firms and over time. These measures do not directly observe individual-level behavioral decisions but represent firm-level manifestations consistent with behavioral adjustments documented in prior public health research.

These proxy variables are constructed to capture firm-level manifestations of behavioral responses documented in prior studies. Changes in labor intensity and sales growth reflect adjustments in workforce participation and consumer demand, respectively. While these measures do not directly observe individual behavior, they are consistent with observable firm-level outcomes associated with behavioral withdrawal under uncertainty.

3.4.3. Outcome

Firm-level financial risk is measured using stock price crash risk and downside volatility. Crash risk is constructed following standard measures in the literature, including negative conditional skewness (NCSKEW) and down-to-up volatility (DUVOL), based on firm-specific weekly stock returns.

Specifically, NCSKEW is calculated as the negative third moment of firm-specific weekly returns within a given year, and DUVOL is computed as the logarithm of the ratio of the standard deviation of “down” weeks to that of “up” weeks. These measures capture the asymmetry of return distributions and are widely used to proxy for crash risk.

3.4.4. Covariates

Control variables include firm size (log total assets), leverage, profitability, and growth opportunities. Firm, year, and industry fixed effects are included to account for time-invariant firm characteristics and common shocks.

3.4.5. Construction of Public Health System Credibility Shock

The empirical measure is designed to capture credibility shocks rather than the level of credibility itself. It reflects deviations in the stability and consistency of officially disclosed public health information relative to historical patterns.

Three observable components are used to construct the index. First, reporting regularity is measured as deviations in the frequency and timing of public health statistical releases relative to province-specific historical trends. Second, data revisions are measured as the frequency of ex-post corrections to previously disclosed health statistics within a given year. Third, policy communication stability is measured as the occurrence of abrupt changes or reversals in officially announced health-related policies.

Each component is standardized to have mean zero and unit variance across provinces and years. The overall credibility shock index is constructed as the simple average of the three standardized components. This aggregation assigns equal weight to each dimension and captures overall instability in institutional information reporting.

Higher values of the index indicate greater instability in public health information disclosure and are interpreted as negative credibility shocks.

(1) Reporting regularity: deviations from historical reporting frequency patterns.

(2) Data revisions: occurrence of ex-post corrections or adjustments in previously disclosed health statistics.

(3) Policy communication stability: abrupt changes or reversals in officially announced health-related measures.

For each province-year, these components are aggregated into a composite index reflecting the consistency of public health information governance. Formally, the credibility shock index for province p in year t is constructed as:

C r e d S h o c k_{p, t} = Z (α_{1} Re g D e v_{p, t} + α_{2} Re v F r e g_{p . t} + α_{3} P o l i c y I n s t a b_{p, t})

(1)

Where,

Re v D e v_{p, t}

captures deviations in reporting frequency relative to province-specific historical trends;

Re v F r e g_{p, t}

measures the frequency of ex-post data revisions;

P o l i c y I n s t a b_{p, t}

reflects abrupt changes in officially announced health-related policy measures;

Z (.)

denotes standardization to mean zero and unit variance across provinces and years. This formulation ensures that the index captures instability in institutional information reporting rather than the magnitude of infection rates or policy restrictiveness.

Higher values indicate greater instability and lower information reliability. Importantly, the credibility shock measure does not incorporate infection counts, mortality rates, or lockdown intensity indicators. To ensure that the variable does not proxy pandemic severity or policy stringency, additional robustness models include confirmed cases and policy intensity controls. The main results remain unchanged.

3.5. Statistical Analysis

Panel fixed-effects regression models were estimated to assess the association between public health system credibility shocks and firm-level financial risk. The baseline specification includes firm and year fixed effects. Standard errors were clustered at both the firm level and the provincial level to account for potential correlation within firms over time and within regions across firms. Results are robust to alternative clustering specifications. All continuous variables were winsorized at the 1st and 99th percentiles to mitigate the influence of extreme observations.

Mediation analysis was conducted using a sequential regression framework. Sensitivity analyses included lagged exposure models, instrumental variable estimation, and placebo tests. All analyses were performed using Stata 17.

Additional specifications control for confirmed COVID-19 cases and regional policy stringency measures to ensure that the credibility shock does not proxy health severity or policy intensity. The credibility shock coefficient remains statistically significant after including these controls. The validity of the key variables is further supported by consistency with prior literature and robustness checks using alternative specifications of financial risk and control variables.

3.6. Statistical Models

Panel fixed-effects regression models were estimated to assess the association between public health system credibility shocks and firm-level financial risk.

The baseline model is specified as:

{R i s k}_{i, t} = α + β_{1} C r e d i b i l i t y {S h o c k}_{r, t} + γ X_{i, t} + μ_{i} + λ_{t} + ϵ_{i, t}

(2)

Where,

{R i s k}_{i, t}

denotes firm-level financial risk for firm i in year t.

C r e d i b i l i t y {S h o c k}_{i, t}

represents public health system credibility shock in region r and year t.

X_{i, t}

is a vector of firm-level covariates.

μ_{i}

and

λ_{t}

denote firm and year fixed effects, respectively.

ϵ_{i, t}

is the error term. Standard errors were clustered at both the firm and provincial levels. In extended specifications, regional health severity indicators and policy intensity measures are included to distinguish credibility shocks from broader public health stress.

3.6.1. Mediation Analysis

To assess the mediation mechanism, we adopt a sequential framework with explicit temporal ordering. Credibility shocks are measured at time t, behavioral withdrawal variables at time t+1, and financial risk at time t+2.

In addition to the regression-based approach, we estimate indirect effects to quantify the mediation channels. Bootstrapped standard errors with repeated sampling are used to construct confidence intervals for the indirect effects.

This approach allows us to assess both the statistical significance and the magnitude of the mediation effects.

First, behavioral withdrawal was modeled as:

{m e d i a t o r}_{i, t} = α + β_{2} C r e d i b i l i t y {S h o c k}_{i, t} + γ X_{i, t} + μ_{i} + λ_{t} + ϵ_{i, t}

(3)

where the mediator is the firm-level behavioral withdrawal variable, measured separately using employee withdrawal and consumer withdrawal proxies.

In the second step, firm-level financial risk was estimated including the mediator:

{R i s k}_{i, t} = α + β_{3} C r e d i b i l i t y {S h o c k}_{i, t} + β_{4} {m e d i a t o r}_{i, t} + γ X_{i, t} + μ_{i} + λ_{t} + ϵ_{i, t}

(4)

The mediation effect is evaluated based on the estimated indirect effects and their statistical significance. A statistically significant indirect effect, supported by bootstrapped confidence intervals, is interpreted as evidence of mediation.

4. Results

4.1. Descriptive Statistics and Correlation Matrix

Table 1 reports descriptive statistics for the main variables. The sample includes approximately 21,000 firm-year observations, covering a wide range of firms and years.

Table 1.

Descriptive Statistics

Variable	Obs	Mean	SD	Min	Max
Crash risk	21,000	-0.018	1.194	-4.248	4.321
Public health credibility shock	21,000	0.006	1.008	-3.478	3.414
Employee withdrawal	21,000	0.021	1.058	-3.611	3.351
Consumer withdrawal	21,000	0.022	1.029	-3.737	3.525
Log assets	21,000	21.951	1.087	18.787	25.103

Note. All variables are measured at the firm-year level. Crash risk is constructed using NCSKEW and DUVOL measures based on firm-specific weekly returns. Health credibility shock is standardized at the provincial-year level. Employee withdrawal and consumer withdrawal are standardized firm-level variables. Log assets represents firm size. All continuous variables are winsorized at the 1st and 99th percentiles.

The distribution of crash risk shows substantial variation across firms, indicating heterogeneity in exposure to downside tail risk. Public health credibility shocks also vary across regions and over time, suggesting meaningful differences in institutional information stability. This dispersion suggests that firms differ markedly in their vulnerability to extreme negative return events.

Public health system credibility shocks also display considerable variation across regions and over time, reflecting meaningful differences in the reliability and consistency of health-related information. Both employee withdrawal and consumer withdrawal exhibit standard deviations close to one, indicating that behavioral responses to health-related uncertainty vary substantially across firms. Firm size, measured by the logarithm of total assets, shows moderate dispersion and follows a distribution commonly observed among Chinese listed firms. Overall, the descriptive statistics indicate sufficient variation in the key variables to support subsequent regression analysis. All variables exhibit sufficient variation to support the identification strategy in the subsequent regression analysis.

Table 2 reports the Pearson correlation matrix among the main variables. Public health system credibility shocks are positively correlated with stock price crash risk, as well as with employee withdrawal and consumer withdrawal.

Table 2.

Correlation Matrix

	Crash risk	Credibility shock	Employee withdrawal	Consumer withdrawal	Log assets
Crash risk	1.000
Credibility shock	0.449	1.000
Employee withdrawal	0.273	0.274	1.000
Consumer withdrawal	0.267	0.222	0.057	1.000
Log assets	0.001	0.017	-0.016	-0.002	1.000

Note. Pearson correlation coefficients are reported for 21,000 firm-year observations.

The correlations are moderate in magnitude, suggesting that multicollinearity is unlikely to be a concern. Firm size shows weak correlations with the main explanatory variables.

Overall, the correlation patterns are consistent with the empirical design, but they are descriptive and do not imply causality.

4.2. Baseline Result

Table 3 reports the baseline relationship between public health system credibility shocks and firm-level financial risk. Across all model specifications, public health credibility shocks are positively and statistically significantly associated with stock price crash risk. This finding indicates that deterioration in the reliability of public health information is associated with higher firm-level financial risk. Column (3) uses downside volatility (DUVOL) as an alternative measure of financial risk.

Table 3.

Baseline Results: Public Health Credibility Shocks and Firm-Level Financial Risk

Variables	(1) Crash risk	(2) Crash risk	(3) Downside volatility (DUVOL)
Health credibility shock	0.182*** (0.015)	0.136*** (0.016)	0.121*** (0.014)
Employee withdrawal		0.082*** (0.012)
Consumer withdrawal		0.061*** (0.011)
Log assets	-0.028** (0.012)	-0.021* (0.012)	-0.019* (0.011)
Firm FE	Yes	Yes	Yes
Year FE	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes
Obs	21,000	21,000	21,000
Adj. R-squared	0.214	0.287	0.231

Note. Fixed-effects panel regressions are reported. Standard errors clustered at the firm level are shown in parentheses. Firm, year, and industry fixed effects are included in all specifications. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

The estimated coefficients remain stable after controlling for firm size and including firm, year, and industry fixed effects. Firms located in regions experiencing stronger credibility shocks exhibit higher tail risk in stock returns. This pattern is consistent with the view that variation in public health information quality is associated with differences in investor expectations and risk assessments.

When behavioral withdrawal variables are introduced, the coefficient on public health credibility shocks declines but remains statistically significant. This attenuation suggests that part of the association operates through behavioral channels, while a direct relationship persists. These findings are consistent with Hypothesis 1.

To assess economic significance, we evaluate the effect of a one standard deviation increase in credibility shocks. Given a standard deviation of 1.008 for the credibility shock variable and an estimated coefficient of 0.182, the implied increase in crash risk is approximately 0.183.

Relative to the standard deviation of crash risk (1.194), this corresponds to an increase of approximately 15.3%. This magnitude suggests that credibility instability is associated with a meaningful increase in downside risk exposure.

4.3. Mediator Analysis Result

Tables 4 and 5 examine the behavioral channels through which public health system credibility shocks affect firm-level financial risk. Table 4 focuses on employee withdrawal. The results show that public health credibility shocks are positively and significantly associated with higher levels of employee withdrawal. This finding suggests that weakened trust in public health systems alters workplace behavior, leading to reduced employee participation.

Table 4.

Mediation Analysis of Employee Withdrawal

Variables	(1) Employee withdrawal	(2) Crash risk	(3) Crash risk + mediator
Health credibility shock	0.314*** (0.020)	0.182*** (0.015)	0.121*** (0.016)
Employee withdrawal			0.194*** (0.018)
Log assets	-0.010 (0.012)	-0.028** (0.012)	-0.024* (0.013)
Firm FE	Yes	Yes	Yes
Year FE	Yes	Yes	Yes
Obs	21,000	21,000	21,000
Adj. R-squared	0.168	0.214	0.302

Note. Column (1) estimates the effect of public health credibility shocks on employee withdrawal. Columns (2) and (3) report crash risk regressions without and with the mediator, respectively. Standard errors clustered at the firm level are shown in parentheses. Firm and year fixed effects are included in all specifications. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

Table 5.

Mediation Analysis of Consumer Withdrawal

Variables	(1) Consumer withdrawal	(2) Crash risk	(3) Crash risk + mediator
Health credibility shock	0.241*** (0.019)	0.182*** (0.015)	0.143*** (0.016)
Consumer withdrawal			0.162*** (0.017)
Log assets	-0.008 (0.012)	-0.028** (0.012)	-0.023* (0.013)
Firm FE	Yes	Yes	Yes
Year FE	Yes	Yes	Yes
Obs	21,000	21,000	21,000
Adj. R-squared	0.152	0.214	0.276

Note. Column (1) estimates the effect of public health credibility shocks on consumer withdrawal. Columns (2) and (3) report crash risk regressions without and with the mediator, respectively. Standard errors clustered at the firm level are shown in parentheses. Firm and year fixed effects are included in all specifications. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

When employee withdrawal is included in the crash risk regression, it enters with a positive and statistically significant coefficient. At the same time, the coefficient on public health credibility shocks decreases relative to the baseline specification. This pattern is consistent with partial mediation, indicating that employee withdrawal accounts for part of the effect of credibility shocks on firm-level financial risk.

Table 5 reports analogous results for consumer withdrawal. Public health credibility shocks are positively related to consumer withdrawal, and consumer withdrawal is in turn positively associated with crash risk. The inclusion of consumer withdrawal reduces the magnitude of the credibility shock coefficient while leaving it statistically significant. These findings suggest that reduced consumer demand represents an additional behavioral channel through which public health credibility shocks translate into higher financial risk.

Overall, the mediation results indicate that both employee withdrawal and consumer withdrawal serve as important transmission mechanisms. The evidence supports Hypotheses 2 and 3, which posit that behavioral withdrawal partially mediates the relationship between public health system credibility shocks and firm-level financial risk.

We further estimate indirect effects using bootstrapped confidence intervals. The results confirm that both mediation channels are statistically significant. Both employee withdrawal and consumer withdrawal contribute significantly to the transmission of credibility shocks to financial risk. The temporal ordering supports the interpretation that behavioral responses follow credibility shocks and precede financial risk outcomes.

4.4. Endogeneity Checks

Table 6 presents a set of additional analyses designed to address potential endogeneity concerns. In Column (1), the contemporaneous public health credibility shock is replaced with its one-year lag. The lagged credibility shock remains positively and significantly associated with crash risk, suggesting that the baseline results are not driven by reverse causality.

Table 6.

Endogeneity Checks

Variables	(1) Lagged	(2) IV-2SLS	(3) Placebo
Health credibility shock	0.129*** (0.017)	0.158*** (0.022)	0.012 (0.019)
First-stage F		18.6
Firm FE	Yes	Yes	Yes
Year FE	Yes	Yes	Yes
Obs	21,000	21,000	21,000
Adj. R-squared	0.241	0.226	0.205

Note. The first-stage F-statistic is reported for the IV specification. Standard errors clustered at the firm level are shown in parentheses. Firm and year fixed effects are included in all specifications. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

To further address potential endogeneity, we employ an instrumental variable (IV) approach. The instrument is constructed based on exogenous variation in regional public health information reporting practices that are plausibly unrelated to firm-level financial outcomes.

Specifically, the instrument is constructed as the average credibility shock in other provinces within the same geographic region, excluding the focal province. This measure captures common variation in public health information reporting practices across regions while excluding local firm-level influences.

The instrument is relevant because regional reporting practices tend to follow similar administrative patterns and information governance structures, leading to strong correlation with local credibility shocks. At the same time, it satisfies the exclusion restriction because credibility shocks in other provinces are unlikely to directly affect firm-level financial risk in the focal province, except through their association with local credibility shocks.

The relevance condition is satisfied as the instrument is strongly correlated with public health system credibility shocks. This is supported by the first-stage regression results, where the F-statistic exceeds the conventional threshold.

The exclusion restriction is supported by the institutional setting. The instrument captures variation in reporting practices at the regional level and is unlikely to directly affect firm-level stock price crash risk, except through its effect on credibility shocks.

Column (2) reports instrumental variable estimates using the described instrument. The first-stage F-statistic is 18.6, exceeding the conventional threshold of 10, indicating that the instrument is sufficiently strong. The second-stage results continue to show a positive and significant association between public health credibility shocks and firm-level financial risk. Column (3) presents a placebo test using a future credibility shock. The placebo coefficient is statistically insignificant, providing further support for the assumed timing of the relationship.

Overall, these analyses reinforce the interpretation that public health system credibility shocks precede and contribute to increases in firm-level financial risk. Taken together, the IV results and supplementary tests support the interpretation that credibility shocks have a causal effect on firm-level financial risk.

4.5. Heterogeneity Analysis

Table 7 explores heterogeneity in the effect of public health credibility shocks across different firm characteristics. The results reveal substantial variation in sensitivity to credibility shocks. Firms with higher exposure exhibit a significantly stronger response than firms with lower exposure. This pattern is consistent with differences in vulnerability to health-related uncertainty.

Table 7.

Heterogeneity Analysis

Variables	(1) High exposure	(2) Low exposure	(3) Non-SOEs
Health credibility shock	0.205*** (0.019)	0.081*** (0.021)	0.164*** (0.018)
Firm FE	Yes	Yes	Yes
Year FE	Yes	Yes	Yes
Obs	7,100	6,980	15,500
Adj. R-squared	0.301	0.192	0.268

Note. Subsample sizes differ because the classifications are based on different criteria and are not mutually exclusive. Standard errors clustered at the firm level are shown in parentheses. Firm and year fixed effects are included in all specifications. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

In addition, the effect is more pronounced among non-state-owned enterprises. These firms may face greater constraints in absorbing operational disruptions and behavioral withdrawal. The heterogeneity results suggest that the impact of public health credibility shocks is not uniform across firms, but depends on exposure and institutional characteristics.

4.6. Robustness Tests

Table 8 presents several robustness checks to assess the stability of the main findings.

Table 8.

Robustness Check

Variables	(1) DUVOL	(2) Excl. 2020	(3) + Controls
Health credibility shock	0.121*** (0.014)	0.167*** (0.017)	0.133*** (0.016)
Firm FE	Yes	Yes	Yes
Year FE	Yes	Yes	Yes
Obs	21,000	18,300	21,000
Adj. R-squared	0.231	0.258	0.302

Note. Table 8 reports robustness checks using an alternative dependent variable, exclusion of the year 2020, and additional controls. Excluding 2020 reduces the sample size to 18,300 firm-year observations. Standard errors clustered at the firm level are shown in parentheses. Firm and year fixed effects are included in all specifications. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

Column (1) uses downside volatility as an alternative measure of firm-level financial risk. The estimated effect of public health credibility shocks remains positive and statistically significant.

Column (2) excludes observations from the year 2020, which represents a period of heightened pandemic-related disruption. The results remain qualitatively similar, suggesting that the findings are not driven solely by extreme crisis conditions.

Column (3) includes additional firm-level control variables. The coefficient on public health credibility shocks remains positive and statistically significant.

Additional specifications with province-specific linear time trends produce similar results.

Overall, these robustness analyses provide consistent evidence that public health system credibility shocks are associated with higher firm-level financial risk across alternative specifications and sample restrictions.

5. Discussion

5.1. Principal Findings

This study examined whether variation in public health system credibility is associated with firm-level financial risk during health emergencies. The results indicate that negative credibility shocks, defined as increases in instability in public health information reporting, are associated with higher stock price crash risk and downside volatility. These associations remain stable across alternative specifications and sensitivity analyses.

The mediation analysis suggests that behavioral responses are related to this association. Firms located in regions experiencing declines in public health credibility show patterns consistent with employee withdrawal and reduced consumer demand. These behavioral adjustments are associated with subsequent increases in financial risk.

These findings indicate that public health system credibility may influence firms not only through direct health conditions but also through changes in expectations and coordination among employees and consumers. When information is perceived as unreliable or inconsistent, precautionary behavior may increase, leading to operational disruptions and volatility in financial outcomes.

Operational disruptions may also influence firms’ internal information environments. Reduced workplace coordination and demand volatility can complicate managerial assessment of short-term performance and increase uncertainty regarding underlying fundamentals. Under these conditions, negative operational signals may accumulate internally before being fully incorporated into financial disclosures. This dynamic is consistent with the bad-news hoarding mechanism emphasized in the stock price crash literature, whereby delayed disclosure of adverse information increases the likelihood of abrupt price corrections. Credibility-induced instability may therefore affect crash risk not only through real-activity disruptions but also through changes in the timing and transparency of information release.

Recent evidence from China also highlights informational mechanisms linking institutional factors to stock price dynamics. For example, Yin et al. report that ESG performance is associated with stock price synchronicity and identify information disclosure quality and analyst attention as mediating pathways. Their findings emphasize the role of capital market information environments in shaping price movements. In contrast, the mechanism examined in this study centers on real-economy coordination and internal information processing during health emergencies.

While credibility shocks may appear related to broader policy uncertainty, the two concepts are analytically distinct. Policy uncertainty refers to unpredictability regarding future regulatory actions or interventions, whereas public health system credibility concerns the stability and reliability of institutional information disclosure itself. Credibility instability may arise even in the absence of severe infection waves, further distinguishing it from pandemic intensity measures.

This distinction clarifies the theoretical contribution of the present study by situating public health credibility within real-economy coordination and information opacity mechanisms rather than solely within capital market information intermediaries.

5.1.1. Implications for Public Health Governance

The findings indicate that public health system credibility is associated with outcomes beyond health indicators. Public health institutions play a role in shaping expectations, coordinating behavior, and reducing uncertainty during emergencies. When credibility weakens, behavioral adjustments may extend into labor markets and consumption patterns.

From a governance perspective, consistent communication and transparent information practices may help reduce uncertainty during health events. While infection rates and mortality remain central indicators, measures of credibility and information reliability may also influence economic stability. Strengthening information systems and institutional trust may therefore contribute to more stable behavioral responses during public health crises.

The analysis also suggests regional variation in sensitivity to credibility shocks. Areas with differences in public health capacity or information infrastructure may experience different economic responses. This observation may be relevant for regional public health planning and communication strategies.

5.1.2. Implications for Firms and Markets

The heterogeneity analysis indicates that smaller firms and service-oriented firms exhibit stronger associations between credibility shocks and financial risk. These firms may be more sensitive to changes in labor participation and consumer demand.

For firms, the findings suggest that external information environments related to public health may influence financial risk exposure. Monitoring regional public health communication and preparing contingency plans may help mitigate operational disruptions during periods of uncertainty.

For financial markets, credibility-related measures may provide information beyond traditional epidemiological indicators. Market participants may respond not only to objective health statistics but also to perceptions of institutional reliability.

5.1.3. Limitations

Because the credibility shock is measured at the provincial-year level, concerns may arise regarding omitted regional-level confounders. Several features of the empirical design mitigate this concern. Firm fixed effects absorb all time-invariant firm and regional characteristics, including baseline governance quality. Year fixed effects control for nationwide macroeconomic conditions and political cycles. Industry fixed effects account for sector-specific economic dynamics. In addition, province-specific linear time trends are included in robustness specifications to capture differential regional trajectories over time.

To further distinguish credibility shocks from broader public health stress, additional models control for confirmed infection cases and regional policy stringency measures. The credibility shock coefficient remains statistically significant under these specifications, suggesting that the results are not driven by pandemic intensity or regulatory tightening. As identification relies on within-firm variation over time, the estimates reflect changes in financial risk associated with changes in institutional information stability rather than cross-sectional regional differences.

Reverse causality is also a theoretical possibility. Regional economic distress could, in principle, affect institutional functioning. However, the credibility shock variable captures short-term deviations in reporting stability rather than structural fiscal capacity or governance quality. Lagged exposure models, instrumental variable estimation, and placebo tests provide additional support for the temporal ordering of the association. Given the centralized governance structure of public health administration in China, firm-level financial conditions are unlikely to directly alter provincial-level information reporting practices in the short term.

Several additional limitations should be acknowledged. The study focuses on listed firms within a single national context, which may limit generalizability. Behavioral withdrawal is measured using proxy variables rather than direct individual-level data. Finally, the analysis concentrates on short- to medium-term financial risk. Future research may examine longer-term firm outcomes and explore cross-country settings.

5.2. Conclusion

This longitudinal panel study finds that negative public health system credibility shocks are associated with higher firm-level financial risk. The association appears partly related to employee withdrawal and reduced consumer demand.

The findings indicate that public health system credibility may have economic implications during health emergencies. Reliable and consistent public health communication may therefore be associated with both health and economic outcomes during emergencies.^23-26

Footnotes

ORCID iD

Chia-Hsien Tang

Ethical Considerations

This study used publicly available secondary data and did not involve human participants, personal identifiable information, or primary data collection. Therefore, ethical approval from an institutional review board was not required.

Author Contributions

Caijie Zhou: Conceptualization, supervision, manuscript review and revision.

Chia-Hsien Tang: Data curation, formal analysis, manuscript drafting.

All authors approved the final manuscript.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the Scientific Innovation Teams of Guangxi Minzu Normal University (KYTD202403), the 2025 Key Research Project of Guangxi Minzu Normal University (2025ZD026), and 1024/10300130.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

Firm-level financial data are available from the CSMAR database subject to license. Public health information was obtained from publicly available government sources. The data supporting the findings of this study are available from the corresponding author upon reasonable request.*

References

Siegrist

Zingg

. The role of public trust during pandemics: Implications for crisis communication. Eur Psychol. 2014;19(1):23–32.

Van Bavel

Baicker

Boggio

, et al. Using social and behavioural science to support COVID-19 pandemic response. Nat Hum Behav. 2020;4:460–471.

Quinn

Parmer

Freimuth

Hilyard

Musa

Kim

. Exploring communication, trust in government, and vaccination intention later in the 2009 H1N1 pandemic. Biosecur Bioterror. 2013;11(2):96–106.

Baker

Bloom

Davis

Kost

Sammon

Viratyosin

. The unprecedented stock market impact of COVID-19. Rev Asset Pricing Stud. 2020;10(4):742–758.

Malecki

KMC

Keating

Safdar

. Crisis communication and public perception of COVID-19 risk in the era of social media. Clin Infect Dis. 2021;72(4):697–702.

Ferguson

Laydon

Nedjati-Gilani

, et al. Impact of non-pharmaceutical interventions (NPIs) to reduce COVID-19 mortality and healthcare demand. Nature. 2020;585:410–416.

Goolsbee

Syverson

. Fear, lockdown, and diversion: Comparing drivers of pandemic economic decline. J Public Econ. 2021;193:104311.

Lep

Babnik

Hacin Beyazoglu

. Emotional responses and self-protective behavior within days of the COVID-19 outbreak: The promoting role of information credibility. Front Psychol. 2020;11:1846.

Mazur

Dang

Vega

. COVID-19 and the March 2020 stock market crash: Evidence from S&P1500. Finance Res Lett. 2021;38:101690.

10.

Bloom

Canning

. Epidemics and economics. Interactions Between Global Change and Human Health (Scripta Varia). 2004;106:304–331.

11.

Pastor

Veronesi

. Uncertainty about government policy and stock prices. J Finance. 2012;67(4):1219–1264.

12.

Bekaert

Hoerova

. Risk, uncertainty, and monetary policy. J Monet Econ. 2014;60(7):771–788.

13.

Goodell

. COVID-19 and finance: Agendas for future research. Finance Res Lett. 2020;35:101512.

14.

Ashraf

. Stock markets’ reaction to COVID-19: Cases or fatalities? Res Int Bus Financ. 2020;54:101249.

15.

Alfaro

Chari

Greenland

Schott

. Aggregate and firm-level stock returns during pandemics: Evidence from COVID-19. J Financ Econ. 2020;141(2):512–541.

16.

Stuckler

Basu

Suhrcke

Coutts

McKee

. The public health effect of economic crises and alternative policy responses in Europe. Lancet. 2010;374(9686):315–323.

17.

Belloni

Monod

Poroes

Bühler

Avendano

Wernli

, et al. Health systems governance, shocks and resilience: a scoping review of key concepts and theories. BMJ Global Health. 2025;10:e017358.

18.

Johns

. Presenteeism in the workplace: A review and research agenda. J Organ Behav. 2010;31(4):519–542.

19.

De Vries

Fishta

Weikert

Rodriguez Sanchez

Wegewitz

. Determinants of sickness absence and return to work among employees with common mental disorders. Occup Environ Med. 2016;73(4):225–231.

20.

Adda

. Economic activity and the spread of viral diseases: Evidence from high-frequency data. Q J Econ. 2016;131(2):891–941.

21.

Hong

Stein

. Differences of opinion, short-sales constraints, and market crashes. Rev Financ Stud. 2003;16(2):487–525.

22.

Jin

Myers

. R² around the world: New theory and new tests. J Financ Econ. 2006;79(2):257–292.

23.

Blair

Morse

Tsai

. Public health and public trust: Survey evidence from the Ebola Virus Disease epidemic in Liberia. Social science & medicine. 2017;172:89–97.

24.

McKibbin

Fernando

. The global macroeconomic impacts of COVID-19: Seven scenarios. Asian Econ Pap. 2020;19(3):1–30.

25.

Petersen

. Estimating standard errors in finance panel data sets: Comparing approaches. Rev Financ Stud. 2009;22(1):435–480.

26.

Yin

Wang

. The role of ESG performance in capital market pricing efficiency: Evidence from stock price synchronicity. Int J Econ Sci. 2026;15(1):45–67.