The effect of city reputation on Chinese corporate risk-taking

City reputation is a valuable asset for the local economy and firms in the contemporary society. However, the impact of city reputation on micro-level firms has been largely overlooked by the literature. This paper uses the National Civilized City (NCC) policy in China as a quasi-natural experiment to enhance city reputation. We employ the DID approach to investigate the relationship between city reputation and corporate risk-taking. The result shows that corporate risk-taking significantly increases following the NCC policy adoption. Moreover, information asymmetry can strengthen the positive impact of city reputation on corporate risk-taking. Channel tests show that city reputation improves financial condition and decreases default risk, leading to improved risk-taking tolerance. Overall, our paper indicates that city reputation is an important mechanism to improve corporate financial performance, providing empirical evidence for local governments to pursue the NCC title.

A study by Weber Shandwick in 2020 found that reputation accounts for 63% of a company's market value on average.This raises a natural question: why is reputation so important for firms?According to the signal theory, reputation can serve as a signal to alleviate information asymmetry and communicate the firm's ability to the stakeholders 1 .Prior literature documents that firms with better reputation can obtain higher market value, better financial performance and more innovation outputs [2][3][4] .In contrast, reputational damage leads to customer loss, revenue decline or even bankruptcy 5,6 .Recognizing the importance of reputation, many firms engage in strategic behaviors to improve their reputation, such as corporate social responsibility activities and charitable donations.
Moreover, firms not only have their own reputations, but also share collective reputation with other firms in the same groups.As a important type of collective reputation, city reputation is the perception of a place by potential investors, businesses, and workers 7,8 .It also reflects the attractiveness and competitiveness of local firms.For example, Silicon Valley in San Francisco is widely recognized as a global center for innovation and technology.The reputation of this city benefits local firms, as it attracts more talent, investment, and collaboration opportunities 9,10 .Despite the great value creation of city reputation for firms, the literature has largely ignored its impact on corporate behavior.In this paper, we use the staggered adoption of the National Civilized City (NCC) policy in China as a natural experiment to examine the impact of city reputation on corporate risk-taking.The NCC award is the highest honor for the civilization level of cities, and many governments pursue it to improve their image and reputation [11][12][13] .Thus, this policy provides an exogenous shock that enhances city reputation.We focus on risk-taking because it is crucial to corporate survival and success 14,15 .
Based on the above analyses, this paper selects the public firms traded on the Chinese A-share market over the period 2001 to 2018 as our sample.We first investigate the relationship between city reputation and corporate risk-taking.Second, we shed light on the channels through which city reputation affects corporate risktaking.Finally, we examine how information asymmetry moderates the impact of city reputation on corporate risk-taking.
Our paper contributes to the literature in the following ways.First, city reputation is a valuable asset for local economy, but measuring it is challenging.Existing research has used the number of tourists 16 or surveys of people's impression of cities 7,8 as a proxy for city reputation.However, these measures may not be objective or independent of local economic conditions.This paper exploits the staggered implementation of the NCC policy as a natural experiment to improve city reputation [11][12][13] .Therefore, our paper provides a reliable and valid method to evaluate the real effect of city reputation.
Second, this paper expands the literature on the effects of city reputation on micro-firms.Prior studies have focused on the impact of city reputation on economic performance 7,8 , air pollution 12 and energy efficiency 11 .However, the implications for corporate business have been largely overlooked.The only related study is by Zhao

Hypothesis development
Prior literature offers opposing views on the impact of city reputation on corporate risk-taking.On the one hand, we argue that city reputation can increase corporate risk-taking based on the benefit hypothesis.First, city reputation provides firms with easier resource availability, more investments and better financial performance 7,[23][24][25] .These benefits could alleviate corporate resources constraints and improve financial condition.Second, city reputation can create a form of insurance that prevent more damage when firms experience negative events 26,27 .In this case, we suppose that firm may obtain lower default risk.Since better financial condition and lower default risk could encourage firms to take risky projects [28][29][30] , the benefit hypothesis implies that city reputation leads to higher risk-taking.Therefore, we propose the following hypothesis: H1a City reputation is positively associated with corporate risk-taking.
On the other hand, city reputation may lower corporate risk-taking according to the burden hypothesis.Zavyalova et al. 31 argue that high reputation attracts more stakeholder attention.Mishina et al. 32 and Petkova et al. 33 mention that high reputation can raise stakeholders' expectations about future performance of organizations.Therefore, firms located in the NCC may have less incentive to take risks because they face more market penalties in case of negative events 34,35 .Taken together, the burden hypothesis expects that city reputation is negatively associated with corporate risk-taking.Our hypothesis is as follows: H1b City reputation is negatively associated with corporate risk-taking.

Data
We collect financial data on public firms traded on the A-share market from the China Stock Market & Accounting Research (CSMAR) Database.The data on media coverage, firm location and city-level economic condition come from the Chinese Research Data Services (CNRDS) database.Our sample starts from 2000 because it is the first year that the CNRDS database provides information on firm location.Our primary sample covers the period from 2001 to 2020.The dependent variable is measured by using annual data over three-year overlapping period, our final sample period thus ends in 2018.Since the control variables are lagged by one year in the empirical model, our sample period spans from 2001 to 2018.We then exclude (1) firms in financial industry and (2) firms that are special treatment (ST) and particular transfer (PT).The final sample consists of 2,912 firms and 30,573 firm-year observations.

Measuring corporate risk-taking
Following the literature [36][37][38] , we employ two measures to estimate corporate risk-taking.The first is the volatility of industry-adjusted ROA.We construct RiskI as the standard deviation of the 2-digit SIC industry-adjusted ROA (ROAI) over the window t to t + 2. Therefore, we estimate RiskI as follows:

Empirical model
We employ staggered DID approach to investigate the relationship between city reputation and corporate risktaking.The model is as follows: where Risk i,t is corporate risk-taking of firm i in year t, which is measured by RiskI and RiskIM.Treat i equals one if the city/county where firm i is located has ever adopted the NCC policy.Post t−1 equals one if the city/county where firm i is located is selected as the NCC in year t−1 and zero otherwise.We use the coefficient β to capture the effect of city reputation on corporate risk-taking.Following the literature [38][39][40] , CONTROLS i,t−1 includes firm size (Size), firm leverage (Lev), sales growth (Growth), return on asset (ROA), market-to-book ratio (MB), firm age (Age), capital expenditures (Capex) and the growth rate of GDP (GDPgrowth).Detailed definitions of these variables are provided in Table 1.δ i and δ t denote firm and year fixed effects, respectively.The continuous vari- ables are winsorized at the 1% and 99% levels.Finally, we cluster standard errors at the city-year level.

ZScoreA
The average of ZScore calculated in Eq. ( 8) over the window t to t + 2

LevA
The average of the ratio of total debt scaled by total assets over the window t to t + 2

WCA
The average of the ratio of working capital scaled by total assets over the window t to t + 2

Size
The logarithm of total output

Lev
The ratio of total debt to total assets

Growth
The growth rate of total sales

ROA
The net profit divided by total assets

MB
The ratio of market value to total assets

Age
The logarithm of the number of years since the firm was founded

Capex
The ratio of capital expenditures to total assets

GDPgrowth
The GDP growth rate in the city

Board
The number of directors in log amount

Indep
The ratio of independent directors Dual An indicator variable that equals one if the chairman and CEO are the same person and zero otherwise

Robustness checks
Alternative measures of corporate risk-taking Following Boubakri et al. 39 and Khaw et al. 47 , we use the volatility of ROA over three overlapping years to measure the dependent variable.Specifically, we use ROA to replace ROAI in both model (1) and model (2), and then calculate corporate risk-taking, which are denoted as RiskA and RiskAM.Panel A of Table 4 reports the results of alternative measure of corporate risk-taking.In columns (1) and (2), we find that the coefficients on Treat*Post are both significant at 0.0044 (t value = 3.19) and 0.0083 (t value = 3.20), consistent with the results in Panel A of Table 3.This finding suggests that city reputation still plays a significant role in increasing corporate risk-taking after considering alternative measures of the dependent variable.

Alternative measure of city reputation
According to the list of NCC, many counties participate in competition for the NCC award with prefecture-level cities.It is possible that the city can obtain high reputation when its county wins the NCC title.For example, the county Zhangjiagang was selected as the NCC in 2005, we suppose that its prefecture-level city Suzhou would have an increase in city reputation.We then use this alternative variable Treat*PostA and re-estimate Eq. (3).Panel B of Table 4 shows that the coefficients on Treat*PostA in two columns are both positive and significant at 1% level.Accordingly, these findings indicate that our main result is not driven by alternative measure of city reputation. (

Alternative samples
In our sample, firm characteristics and regional economic conditions are matched at the prefecture-level cities.Therefore, we control for city-level characteristics even for those firms located in counties with the NCC title.One might worry that our main result is driven by counties that won the NCC award earlier than their prefecture-level cities.Following Zhao et al. 13 , we then exclude these observations to ensure robust estimation.Panel C of Table 4 reports the results.In columns ( 1) and ( 2), our main inference remains the same after addressing this issue.To mitigate the concern of regional differences, we exclude firms located in provincial capitals and four municipalities (Beijing, Shanghai, Tianjin and Chongqing).The results are shown in Panel C of Table 4. Columns (3) and (4) show that the association between city reputation and corporate risk-taking is still positive and significant in the restricted sample.These results suggest that our main result is not sensitive to regional differences.

Test of parallel trends assumption
The validity of our main result in Eq. ( 4) is based on the parallel trends assumption.That is, the treatment and control groups should exhibit comparable corporate risk-taking trends before the NCC policy is implemented.To test this assumption, we follow Beck et al. 48and examine the dynamic effect of city reputation on corporate risk-taking based on the following model.
We use the following dummy variables to indicate the timing of the NCC award for firms located in prefecture-level cities: Pre 6 + (six or more years before the award), Pre j (j = 2, 3, 4, 5 years before the award), Current (the year of the award), Post 1 (one year after the award), Post j (j = 2, 3, 4, 5 years after the award), and Post 6 + (six or more years after the award).We assign a value of one to the corresponding variable and zero otherwise.The other variables are the same as in Eq. ( 3).
We present the results in Panel A of Table 5.In column (1), we find that Pre 6 + , Pre 5, Pre 4, Pre 3, Pre 2 play an insignificant role in explaining the increase in RiskI.This finding supports the parallel trends assumption by showing no systematic difference in corporate risk-taking between firms located in NCC and non-NCC.Moreover, column (1) shows that the effect of city reputation emerges one year after gaining the NCC award.Using RiskIM as the dependent variable, our result in column ( 2) is consistent with column (1).Finally, we plot the regression coefficient estimates in Fig. 1 based on the [− 6 + , 6 +] window.Figure 1 shows that our main result is not driven by pre-event period, which further verifies parallel trends assumption.

Placebo test
Another endogeneity issue is that our main result could be driven by a linear combination of the independent variable.To address this concern, we follow Mao and Zhang 49 and conduct placebo test.In our sample, there are 38.3% firms located in cities with NCC award.We then randomly select 38.3% firms from the full sample as the false treatment groups.Treat*PostR equals one if a firm is assigned to treatment status in the random sample and zero otherwise.We re-estimate Eq. ( 3) and present the results in Panel B of Table 5.In columns (1) and ( 2), we find that city reputation is not significantly associated with corporate risk-taking.This finding suggests that our main inference is not an artifact of the data structure. (5)

Instrumental variable approach
Finally, we use instrumental variable approach to further address the omitted variable concern.Specifically, we use the number of scenic spots (IV) as the instrumental variable.On the one hand, this variable is highly correlated with city reputation.More scenic spots tend to attract a greater number of tourists from other regions, leading to a higher city reputation.On the other hand, scenic spots often are influenced by a city's history and geography, which is exogenous to corporate risk-taking.Then, we collect scenic spots information from the CNRDS database, which is available after 2011.Panel C of Table 5 reports the result of instrumental variable approach.In column (1), we find the coefficient on IV is positive and significant at 1% level.This finding is consistent with our prediction that cities with more scenic spots tend to obtain better reputation.The results in column ( 2) and (3) show that city reputation significantly increases corporate risk-taking.

Channel tests
Our main result suggests that firms located in the NCC have better financial condition and lower default risk, leading to a higher level of risk-taking.In this section, we first test the impact of city reputation on cost of capital.Second, we investigate the relationship between city reputation and default risk.Finally, we examine how information asymmetry moderates the relationship between city reputation and corporate risk-taking.

Cost of capital
First, we use cost of capital to measure corporate financial condition.Koirala et al. 30 claim that lower cost of capital may encourage firms to engage in more risk-taking activities.We then suppose that firms located in cities with the NCC title are more likely to exhibit cheaper equity financing.To test this conjecture, we employ the PEG and MPEG approach proposed by Easton 50 to estimate the cost of capital, denoted as γ PEG and γ MPEG .Parallel to our risk-taking proxies, we follow Ferris et al. 51 and calculate the cost of capital as the average of γ PEG and γ MPEG over the window t to t + 2. As shown in Panel A of Table 6, we find the coefficients on Treat*Post are significant at − 0.0021 (t value = − 2.61) and − 0.0036 (t value = − 3.19) in columns ( 1) and ( 2), indicating that firms headquartered in NCC obtain lower cost of capital than firms in non-NCC.This finding confirms our prediction that city reputation can bring firms with better financial condition and thus increase corporate risk-taking.

Default risk
Second, we examine how city reputation affects default risk.Hilscher and Raviv 28 and Favara et al. 29 document that firms with higher default risk adopt less risk-taking activities.Theoretically, if obtaining the NCC title can provide firms with a form of insurance, we predict that it may reduce the likelihood of default risk.To examine this hypothesis, we follow Altman 52 and use Z-Score to capture default risk.We define ZScoreA as the average of ZScore over three-year overlapping period.We report the result in Panel A of Table 6.This result in column (3)  indicates that firms located in the NCC are indeed more likely to obtain lower default risk.As predicted, these results suggest that city reputation could improve financial condition and reduce default risk, and thus motivate firms to take more risk-taking behaviors.

Heterogeneous effects
In this section, we examine the moderating role of information asymmetry.Li et al. 26 argue that country reputation may have weaker effect when investors have access to more specific information from other sources.Since information asymmetry could strengthen the power of reputation, we expect that the positive relationship between city reputation and corporate risk-taking is pronounced for firms with higher information asymmetry.
To test this prediction, we use three measures to capture the information asymmetry: analyst following, media coverage and investor interaction.Roulstone 53 and Frankel and Li 54 find that analyst following can improve the informativeness between managers and investors.As such, we suppose that city reputation has a stronger effect on increasing risk-taking for firms with less analyst following.We first calculate analyst following as the log of (1 + the number of analysts).The data on analyst following come from the CSMAR database during 2001 to 2018.Then, our sample is divided into two groups according to the median of analyst following.Panel B of Table 6 shows the results.In columns (1) and ( 2), we find that city reputation demonstrates a significantly positive association with corporate risk-taking.In contrast, the coefficients on Treat*Post are both insignificant in columns (3) and ( 4).These results suggest that the impact of city reputation on corporate risk-taking is more pronounced for firms with less analyst following, which is consistent with our prediction.
Second, we investigate the moderating effect of media coverage.Prior studies document that the media can simplify and explain complex information for market participants, alleviating the mispricing of accounting information [55][56][57] .We thus expect the effect of city reputation to be stronger for firms with less media coverage.Following You et al. 58 , we select eight leading newspapers to capture media coverage.We collect newspaper data from the CNRDS database, spanning the period 2001 to 2018.A firm is viewed as having higher (less) information asymmetry if its media coverage of "Big 8" is lower (greater) than the median of all firms.We then run separate regressions for the groups and present the results in Panel C of Table 6.We find that city reputation exhibits a significant role in risk-taking for firms with less media coverage, but not for more media coverage.
Finally, we test whether the impact of city reputation on corporate risk-taking varies across firms with different investor interaction.Blankespoor 59 and Lee and Zhong 60 suggest that investor interactive platform may significantly reduce information processing costs and improve corporate transparency.In this case, increasing investor interaction may weaken the effect of city reputation on risk-taking tolerance.Following Lee and Zhong 60 , we use the number of replies posted by the firm on the investor interactive platform to measure information asymmetry.The more replies by target firms, the less information asymmetry.The information on investor interaction comes from the CNRDS database over the period 2010 to 2018.We then divide our sample into two

Additional analyses
The effect of corporate governance Prior research documents that better corporate governance can encourage firms to pursue riskier activities, leading to a higher corporate risk-taking 30,61,62 .It is possible that the increasing risk-taking has been attributed to the improvement of corporate governance with the passage of the NCC policy.To exclude this explanation, we follow Deutsch et al. 62 and include the number of directors in log amount (Board), the ratio of independent directors (Indep), and CEO-chairman duality (Dual) in our baseline model.We gather corporate governance data from the CSMAR database, covering the period from 2001 to 2018.We re-estimate Eq. ( 3) and present the results in Panel A of Table 7.In columns ( 1) and ( 2), we find that the role of city reputation in explaining www.nature.com/scientificreports/corporate risk-taking is still significant after controlling for corporate governance.These results suggest that our results are not driven by a higher corporate governance associated with higher city reputation.

City reputation and corporate financial policies
Next, we examine whether firms located in NCC adopt riskier financial policies.Following Ferris et al. 51 , we employ two measures of financial policies.The first is the leverage ratio, which reflects the proportion of debt in the capital structure.The second is the working capital ratio, which proxies for the liquidity of the firm.We estimate these two measures by averaging them over a three-year period from year t to year t + 2, denoted as LevA and WCA respectively.As shown in Panel B of Table 7, in column (1), Treat*Post shows a significantly positive association with LevA, indicating that firm leverage significantly increases after the adoption of the NCC policy.
We also find that city reputation plays a significant role in decreasing working capital in column (2).These results indicate that better city reputation could motivate firms to adopt riskier financial policies by increasing more debt burden and holding less liquid assets.

Industry characteristics
Finally, we investigate the impact of city reputation on corporate risk-taking across different levels of industrial competition.On the one hand, high competition often drives firms to invest more to stay competitive, which may decrease financial resources 63,64 .On the other hand, increased competition can also pose more risks to firms, leading to a greater default risk.We thus suppose that the positive relationship between city reputation and corporate risk-taking is more pronounced for firms with high industrial competition level.Following Jiang et al. 64 , we use the Herfindahl-Hirschman Index (HHI) to capture the industry's competition.An industry is classified as high (low) competition if its HHI exceeds (falls below) the median across all industries.. Panel C of Table 7 reports the results, indicating that the treated firms in highly competitive industries tend to engage in more risk-taking.

Conclusion
In this paper, we examine how city reputation affects corporate risk-taking.Utilizing the staggered adoption of the NCC policy in China, our DID analysis finds that city reputation exhibits a significant role in increasing corporate risk-taking.To address endogeneity issues, we first test the parallel trends assumption.We find that no divergent trend emerges in corporate risk-taking prior to receiving the NCC award.Second, we conduct the placebo test.The results show that our main result is not an artifact of the data structure.Finally, our main inference still holds after using instrumental variable approach to address the omitted variable concern.
In the channel tests, we first examine whether city reputation may motivate firms to adopt risk-taking behavior through improving financial condition and reducing default risk.The results suggest that the treated firms experience a significant decrease in the cost of capital and default risk.Then, we conduct cross-section tests to investigate the moderating role of information asymmetry.We find that the positive relationship between city reputation and corporate risk-taking is more pronounced for firms with less analyst following, less media coverage and less corporate replies.In additional analyses, we confirm that firms located in the NCC pursue riskier financial policies, including more debt burden and less liquid assets.We also find that the impact of city reputation on corporate risk-taking is more pronounced in highly competitive industries.
Overall, this paper sheds light on the important role of city reputation in shaping corporate risk-taking behavior.Our results also indicate that city reputation may be a informal mechanism to alleviate financial constraints

Policy implications
First, the government should recognize the value creation of city reputation and adopt relevant policies to enhance it.In the current economy, city reputation indicates the competitiveness and quality of the local economy.For example, we associate San Francisco with technology, London with finance, and Detroit with automobiles.In this case, city reputation can act as a signal to reduce information asymmetry and attract more capital and human capital.This is in line with our finding, which shows that better city reputation encourages firms to undertake more risky investments and increases corporate risk-taking tolerance.Meanwhile, the government should also emphasize the externality of reputation and prevent activities that may damage city reputation.Second, firms should recognize the important role of city reputation for improving financial performance.Due to underdeveloped financial market, 75% of firms in China suffer from financial constraints, particularly for private and small enterprises 65 .Although the Chinese government has made great efforts to ease financing difficulties, the problem caused by financial constraints remains a significant obstacle for firm development and investment.Our finding suggests that city reputation can decrease cost of capital and default risk.Thus, firms should strive to build and maintain a good reputation to alleviate financial constraint and increase financial stability.
Finally, it is clear that information asymmetry leads to market failure and inefficient resource allocation 66 .However, our finding reveals that city reputation has a stronger positive effect on corporate risk-taking for firms with higher information asymmetry.This suggests that city reputation can act as an information channel to help investors evaluate firm attributes.Therefore, the government should enhance city reputation to reduce information frictions and improve corporate governance.

Limitations
First, our study only covers the public firms listed on the Chinese A-share market.These firms may have more access to financial resources and lower financial risk.Thus, our results may understate the effect of city reputation on corporate risk-taking.In fact, private firms contribute the most to GDP and employment in China 67 .However, these firms face severe financial constraints and high financial costs.Therefore, future research can explore how city reputation influences corporate behavior in these firms.
Second, city reputation is a complex concept that captures how a city is viewed by its residents, visitors, investors, and other stakeholders.Measuring this indicator is challenging and difficult.Although our paper uses the NCC title, the highest honor for the civilization level of cities, as a proxy for city reputation, it may not fully reflect all aspects of city reputation.Finally, we follow the previous literature and use corporate risk-taking to capture a firm's propensity towards engaging in risky activities [36][37][38] .However, it's worth noting that this dependent variable primarily measures the scale of effect rather than the probability of risk.Recent studies have shed light on more detailed and precise methodologies for assessing the likelihood scale of risk [68][69][70] .We will pay close attention to this field in the future.

Table 2
reports summary statistics of the main variables.For the dependent variables, RiskI and RiskIM have means at 0.073 and 0.135, with the standard deviations of 0.234 and 0.412 respectively.The mean of Treat*Post is 0.383, suggesting that 38.3% of firms are located in cities with the NCC award in our sample.The average values