China is one of the most famine-affected countries in the world and has even been referred to as “a land of famine” by scholars (Mallory, 1926). Historically, the Great Chinese Famine of 1959–1961 was the most recent famine in China. It had a significant impact on the population, economy, and social development at that time. Figure 1 presents the average mortality rates in every province over the 3 years before, during, and after the famine. From the figure, we can infer the following: first, this famine had a wide impact. During the famine, the average mortality rate of all provinces in China showed a significant upward trend, indicating that almost all provinces and regions were affected (Chen and Yang, 2015). Second, the impact was different in different regions. For example, during the famine, the average mortality rate in Anhui, Guizhou, and Sichuan provinces exceeded 3%, which was significantly higher than that in other provinces. Third, the impact faded rapidly, as the average mortality rate of 3 years after the famine returned to the rate of 3 years before it.

The Great Chinese Famine of 1959–1961 was one of several extraordinary natural disasters in China for which more data are available. It had a profound impact on both the social and economic development of the region. Related topics have been discussed by researchers in the fields of sociology, economics, and finance. Sociological researchers focused on the long-term effects of the famine on personal growth and found that the starvation and abnormal deaths during the Great Chinese Famine had an irreversible impact on the future of survivors, resulting in smaller populations, lower body weights, and reduced labor force supply (Chen and Zhou, 2007; Meng and Qian, 2009). Further, the health status of the generation who experienced the famine was also relatively poor (Ma, 2011).

Studies on the long-term effects of the Great Chinese Famine on macroeconomic and individual behaviors have accumulated in the literature, providing rich evidence. For example, Gooch (2017) found that although the state of the country returned to how it was 3 years before the famine within 3 years after it, this event had a significant negative impact on the economic growth of the country in the long run. Moreover, due to the memory of the famine, individuals who experienced the early-life famine tend to pay closer attention to food storage (Cao, 2005), exhibiting a greater tendency to save (Harbaugh, 2004). Wang et al. (2015) found that experiencing the famine early in life had a major impact on individuals' self-employment choices: it made such individuals more risk-averse, leading them to avoid starting up a business.

Finance scholars have also examined the impact of the Great Famine experience on corporate behavior. O'Sullivan et al. (2021) found that an experience of trauma early in a CEO's life is positively associated with corporate social performance. Xu and Ma (2022) found that firms with CEOs who experienced early-life poverty are associated with more socially responsible activities and fewer socially irresponsible activities. Zhang (2017) showed that CEOs who experienced the famine had a higher risk aversion tendency, which manifested itself in lower debt levels, more cash holdings, and lower probability of engaging in M&A activities. Zhou et al. (2022) demonstrated that CEOs' early-life adversity experiences have a significantly positive effect on firm internationalization. Concerning accounting and finance policies, Feng and Johansson (2018) found that companies with CEOs having lived through the famine are associated with more conservative financial, investment, and cash-holding policies, a lower likelihood of unethical behavior, and better firm performance during economic downturns. Hu et al. (2019) found that CEOs who experienced the Great Famine (1959–1961) hold higher amounts of cash and increased cash sensitivity. Hu et al. (2020) found that accounting conservatism was significantly higher for companies whose CEOs have experienced the Great Chinese Famine. Yao et al. (2020) found firms led by top executives who experienced the Great Famine in early life are less likely to conduct fraudulent financial reporting. Long et al. (2020) found that companies whose CEOs experienced the “Great Chinese Famine” in early life have lower stock price crash risk than those with CEOs who did not experience the famine. Using the Korean War as an exogenous shock, Choi and Jung (2021) found that firms managed by war-exposed CEOs have greater information transparency than firms managed by CEOs who are not war-exposed.

From existing studies, we know that the Great Chinese Famine left an indelible imprint on the psychological growth of individuals who experienced the disaster, which would affect individuals' future behavioral decisions (Zhang, 2017; Hu et al., 2019). Drawing on the related studies, we predict that CEOs' famine experience has the following two effects on corporate earnings quality.

On the one hand, the painful memory of the Great Chinese Famine makes a CEO more risk-averse, which could improve corporate earnings quality. This is the “risk aversion effect.” The existing literature found that early disaster experience has a long-term physical and psychological impact on CEOs, leading to a higher degree of risk aversion. This could affect corporate earnings quality in various ways. Corporate earnings information is the cornerstone of the efficient operation of capital markets, on which investors and information intermediaries' decision making are heavily dependent (Mao et al., 2013). For corporate CEOs, although low-quality earnings information disclosure can increase investor information asymmetry, which may be beneficial for managers' opportunistic behavior and obtaining excess returns, it may also lead to a higher uncertainty and litigation risk (Field et al., 2005). According to the imprint theory, individuals are imprinted by adverse early-life experiences (Marquis and Tilcsik, 2013), famine-experienced CEOs are more risk-averse and precautionary about uncertainties (Zhang, 2017; Hu et al., 2019), and thus, they are more likely to overstate the possible risk and understate the benefit of earnings management, which would reduce the incentive to earnings management and result in a high quality of earnings disclosure.

On the other hand, the background and causes of the famine may increase CEOs' knowledge and cognition, and the famine learning effect may also improve corporate earnings quality. Previous literature has found that the catch-up strategy employed during the Great Leap Forward and the associated over-purchase of food were important causes of abnormal deaths during the Great Famine (Kung and Lin, 2003). Psychological research has shown that individual decision making is rooted in the memory of past experiences and conditioned reflexive learning, which form a sort of knowledge guide for current actions (Freud, 1915; Schlag, 1998). The famine learning effect may be recognized in CEOs' reflection on the negative consequences of the exaggerated behavior (e.g., catch-up strategy) during the Great Leap Forward, which ultimately affects their motivation to manipulate corporate earnings. Feng and Johansson (2018) found that CEOs' early-life famine experience increased corporate conservatism, and such CEOs rarely cheated. CEOs who are influenced by the famine learning effect can be expected to pursue less earnings manipulation and have relatively higher requirements for the quality of their corporate information disclosures, which may improve corporate earnings quality.

Therefore, the risk prevention effect and the famine learning effect produced by the famine experience may lead the CEO to have higher requirements for corporate earnings quality. Accordingly, we formulate our first hypothesis, which is presented as follows:

Differences in cognitive ability can be found between different age groups. Psychological research indicates that adolescence is the most crucial phase in the formation of individual thinking and values. During this period, experiences with clothing, food, cold, and warmth play a key role in the formation of individual qualities, psychological tendencies, and personality structure (Elder Jr et al., 1991; Becker, 1992). Since adolescence is the key stage in the formation of cognitive abilities and personality traits, the imprint left by the Great Chinese Famine at this stage may have a larger and more profound impact on personal behavior. By distinguishing between the birth cohorts of family heads, Cheng and Zhang (2011) found that famine experience during adolescence has a greater impact on an individual's propensity to save. Feng and Johansson (2018) showed that famine experience during CEOs' childhood or adolescence has a more pronounced impact on their corporate decision making than famine experience in other life stages. Based on the above studies, we expect that CEOs who experienced the Great Chinese Famine during adolescence would have a more pronounced effect on corporate earnings quality than those who experienced it in other ages. We propose the second hypothesis as follows:

Our original sample includes all companies listed on the Chinese A-share market, and we obtained financial data from the China Stock Market and Accounting Research (CSMAR) database. The CSMAR database contains financial and corporate governance information from 1999 to date. However, our sample was limited to 2000 to 2015 because observations for some test variables in 1999 are missing from the database. Our sample ending range is 2015 because the CEOs' background information that we manually collected has been limited to 2015.

We applied the following exclusion criteria: samples from the financial and insurance industry; ST and ^*ST listed companies; samples whose CEOs were of a non-Chinese nationality or born in Taiwan, Hong Kong, Macau, or Tibet; and samples with missing financial information or missing place of CEO's birth. To control the influence of outliers, all continuous variables were winsorized at the 1 and 99% quantiles. Ultimately, we obtained 23,974 enterprise-year samples.

It should be noted that age is an important factor for managers' risk preference and corporate earnings quality. For example, older managers are more inclined to avoid risk than younger ones (Yim, 2013; Serfling, 2014), and as such, the reported earnings quality of their corporations tends to be higher (Huang et al., 2012). Since CEOs who experienced the Great Chinese Famine tend to be older, age may be an alternative explanation for our hypothesis. To address this issue, we set up a treatment group and a control group, which were distinguished by whether the CEO experienced the famine and the degree of famine intensity in the region where the CEO was born. We used a DID method to test the effect of CEOs' famine experience on corporate earnings quality. The model is as follows:

In model (3), EQ indicates corporate earnings quality; famine experience is the independent variable that is proxied by Famine^*EDR (EDD); Ctrls are the control variables as described above. Further, due to the economic cycle, industry environment, regional factors, and CEO growth environment may affect the validity of the regression estimation. Thus, in order to control the different fixed effects, we added five dummy variables to our model: year (Y), industry (I), province (P), CEO birthplace (BP), and CEO birth year (BY).

It is worth noting that although the famine experience is an interaction term for Famine and EDR (EDD), these two variables do not change with time and province. While we controlled the fixed effects of BY and BP, we did not add the variables Famine and EDR (EDD) to the model again. In model (3), our variable of interest is Famine^*EDR (EDD). If our hypothesis is true, we can expect that the coefficient of β₂ will be negative and significant even for CEOs who experienced the famine during the same age as different intensities of famine in different regions may lead to differences in earnings quality.

Table 1 reports the descriptive statistics of the main variables in this paper. The average earnings quality (EQ) is 0.047, and the median is 0.036, but the standard deviation is 0.051, indicating a large difference in earnings quality in the sample. The average value of Famine is 0.419, indicating that about 42% of the CEOs in the sample experienced the Great Chinese Famine. The average EDR is 0.591%, which indicates that the average excess mortality rate during the 3-year famine is close to 0.6%. Given China's large population, this number indicates a painful and profound impact. The standard deviation of the EDR is 0.534, which is close to the average, indicating that the impact of the famine differed greatly in different provinces. The descriptive statistics of other control variables are mostly consistent with prior studies and have not been described in detail in this paper.

Based on the intensity of the famine in the region where the CEO was born, we further described the association between CEOs' famine experience and the distribution of the earnings quality. The results are given in Table 2. In the table, the first column shows the distribution of CEO birth areas in our sample, and the second column shows the population EDR in each region from 1959 to 1961. As shown in the table, EQ is relatively higher among CEOs who were born in provinces with higher mortality, indicating the average earnings quality is slightly lower for CEO who did not experience the famine.

To intuitively describe the relationship between famine experience and earnings quality, we plotted the excess mortality rate in a CEO's birth place during the famine and the company's earnings quality based on descriptive statistical analyses. As shown in Figure 2, the straight line represented by Famine = 1 is lower than the straight line of Famine = 0, indicating that famine experience improves corporate earnings quality. Regardless of Famine = 1 or 0, corporate earnings quality increases with an increase in the impact of the famine (the trend line slopes downward to the right). Compared with Famine = 0 companies, the downward trend represented by Famine = 1 companies is more obvious. These results suggest that when a CEO experiences Great Famine, their company's earnings quality improved significantly, and the more severe the famine, if experienced, the higher the earnings quality.

Before regression analyses, we performed a univariate test on earnings quality to check whether CEOs experienced the famine and the extent of famine intensity in CEOs' birthplace (EDR). The results are given in Table 3. For CEOs who did not experience the Great Famine (Famine = 0), no significant difference was seen in the mean and median of EQ between the EDD = 0 and EDD = 1 groups. However, for CEOs who experienced the famine (Famine = 1), the average EQ and median EQ in the EDD = 1 group were 0.038 and 0.036, respectively, which were significantly lower than those in the EDD = 0 group. This indicates that CEOs' famine experience significantly improved earnings quality. We also tested the DID results for EQ in the famine and EDD groups. The difference of the mean and median EQ was −0.008 and −0.002, respectively, both of which are significant at the 10% level at least. These results indicate that whether a CEO experienced the famine and the extent to which he experienced it have a significant impact on earnings quality.

Table 4 reports the baseline regression results for the impact of CEOs' famine experience on corporate earnings quality. In columns (1) and (3), we did not include the control variables but controlled the fixed effect of industry, year, province, CEO birthplace, and CEO birth year. The coefficient of the interaction term Famine^*EDR is −0.352 and that of the interaction term Famine^*EDD is −0.007; both are significant at least at the 5% level. In columns (2) and (4), all firm-level and CEO-level control variables were controlled in the regresses. The coefficient of Famine^*EDR was significantly negative at the 5% level in column (2), and Famine^*EDD in column (4) was significantly negative at the 1% level. It should be noted that since the fixed effect of the birthplace and birth year of the CEO was considered, the influence of famine experience expressed by Famine and the intensity of famine expressed by EDR (EDD) were absorbed by the fixed effect of birth year and birthplace, and their coefficients were omitted due to collinearity (similar to subsequent regressions). Overall, the regression results given in Table 4 show that CEOs' famine experience significantly improved earnings quality of the company, which is in line with our basic expectations and supports H1.

The control variables for the size of the company (Size), the return on assets (ROA), the nature of property rights (SOE), audited by one of the Big 4 or 5 audit firms (Audit), and CEOs' education degree (Degree) were significantly negatively correlated with EQ, indicating that when an enterprise is larger, its return on assets is higher, it is state-owned, it is audited by a major accounting firm, or the CEO has a higher degree, the quality of corporate earnings is relatively higher. Financial leverage (Lev), company growth (Grow), the largest shareholder's ratio of shares (First), and the age of a CEO are significantly positively related to EQ, indicating that when companies have a higher debt ratio, they experience a higher speed of revenue growth and have more shareholding by the largest shareholder; likewise, when the CEO is relatively old, the earnings quality is worse. The results for the control variables are basically consistent with the findings of previous research (Fan and Wong, 2002; Lennox et al., 2016).

According to H2, the period of adolescence is a key stage for individuals' formation of cognitive abilities and personality traits, so the influence of famine on a CEO at this stage is more profound, and as such, the impact on earnings quality is more obvious. To test this hypothesis, we referred to Cheng and Zhang (2011) and further divided the age at which famine was experienced into infancy (<3 years), childhood (3–6 years), adolescence (7–17 years), and adult years (18 years and older) and named as Cohort₁, Cohort₂, Cohort₃, and Cohort₄, respectively. The regression results given in Table 5 show that the famine experience of CEOs of the difference cohorts has different effects on corporate earnings quality.

As shown in column (1) of Table 5, the coefficients of the interaction terms Cohort3*EDR were significantly negatively correlated with EQ at the 1% level, but the coefficients of Cohort1*EDR, CohortCohort2*EDR, and Cohort4*EDR did not have statistical significance. The estimated coefficients of Cohort^*EDD (column 2) for each interactive variable in infancy, childhood, and adolescence are similar to those in the first column.

To compare whether the above differences in coefficients are statistically significant, we performed t-tests on the estimated coefficient value after regression. The statistical results are reported in the last three rows of each column in Table 5. We found that the coefficients ofCohort3*EDR were significantly bigger than those for other three interactions. This shows that the impact of CEOs' famine experience in adolescence is significantly greater than those for experience in infancy, childhood, and adulthood. In summation, the results of Table 5 indicate that CEOs' famine experience in adolescence had a significantly greater effect on corporate earnings quality, supporting H2.

We mainly used DID to test our hypotheses. However, there is a very important prerequisite for using this method, the equilibrium trend hypothesis. Without this, the DID method cannot effectively alleviate potential endogeneity problems. Accordingly, we referred to the practices in the previous studies and designed the following model to test the equilibrium:

where Cohort₋₁ denotes CEOs born 3 years after the famine, 1962–1964; Cohort₋₂ denotes CEOs born during the fourth to sixth years after the famine, 1965–1967. In model (4), we used CEOs born after 1967 as the benchmark for the equilibrium trend test. The coefficient for Cohort^*EDR (EDD) represents CEOs who were born during these two periods after the famine. Assuming equilibrium trends, we expected that the coefficients of Cohort^*EDR (EDD) should not be significantly different from 0.

Table 6 reports the results of regression tests using model (4). Columns (1) and (2) used Famine^*EDR and Famine^*EDD as variables of interest, respectively. The results in these columns indicate that the coefficients of Cohort-2*EDR (Cohort-2*EDD) and Cohort-1*EDR (Cohort-1*EDD) were not significant, suggesting that CEOs born after the famine have no significant differences in terms of influence on corporate earnings quality. These results are in line with the assumption about common trends in DID estimation methods. More importantly, for CEOs born before the Great Famine, the coefficients of Famine^*EDR (Famine^*EDR) are all significantly negative at the 5% level, indicating that CEOs' famine experience are positively associated with corporate earnings quality.

Fixed factors of a company that do not change over time, such as honesty in the corporate culture, may also cause differences in corporate earnings quality. To control for the impact of factors that may be omitted in the regression results, we further test the main hypothesis by controlling firm fixed effects. The results are reported in the first two columns of Table 7. After controlling for these fixed effects, the coefficient for the interaction of Famine^*EDR (Famine^*EDD) remained significant and negative at the 5% level, which is consistent with earlier findings, indicating that the results were not sensitive to the firm-level factors that do not change over time.

Other factors that do not change over time may impact the conclusions of this study, such as cultural differences between different regions. To resolve this issue, we matched the provinces with higher-level EDR due to the famine with neighboring provinces with lower-level EDR. Specifically, we selected three pairs of matched samples from Shandong and Shanxi, Henan and Hebei, and Guangxi and Guangdong. These regions differ little in geography and culture, so we could use it to control some unidentified macro-level factors. The results based on the samples from the six provinces are shown in the last two columns of Table 7. After excluding the samples from other provinces, Famine^*EDR (Famine^*EDD) remained significantly negative at least at the 5% level, which is consistent with the previous findings.

Differences in company characteristics, the CEO, and the region where the company is located may also cause bias. We use propensity score matching (PSM) to test the basic hypothesis. The procedures are as follows: the financial characteristics and CEO characteristics of our sample were assessed for whether the CEO experienced a famine (Famine = 1) and whether famine was severe in the region where the CEO was born (EDD = 1). A one-to-one closest-match procedure was performed to obtain two sets of highly similar samples and to eliminate differences in the earnings quality due to differences in characteristics. Subsequently, the samples were regressed. If the CEO's famine experience continues to have a positive effect on corporate earnings quality, the aforementioned indicators can be ruled out as causing differences in the corporate earnings quality.

Panel A in Table 8 reports the differences in characteristic variables between the corporate samples before and after PSM. Before PSM, the mean values of most feature variables are significantly different at the 10% level. However, after the one-to-one match, except for CEO age (Age), the mean differences in other characteristic variables were no longer significant, indicating that there are no differences between the treatment group and the control group. Panel B in Table 8 reports the regression results using the PSM samples. The coefficient of the CEO's famine experience variable (Famine^*EDR and Famine^*EDD) was still significantly negative, which is consistent with the earlier research conclusions. The matching results based on PSM showed that the promotion effect for the CEO's famine experience on the quality of corporate earnings was not due to differences in characteristic variables.

Risk prevention effects assume that the early-life famine experience of CEOs makes them more risk-sensitive and risk-averse. Low-quality earnings information disclosure may lead to a higher risk of litigation. Therefore, CEO's manipulation of earnings information is more cautious, and the quality of corresponding earnings information disclosure is thus relatively high. Based on the above analysis, if the CEO's early famine experience affected the corporate earnings quality mainly through risk prevention effects, it can be expected that when a company's low-quality earnings information disclosure incurs relatively higher risks, the CEO's risk prevention motivation would be relatively stronger, and the association between the CEO's famine experience and corporate earnings quality should be more obvious.

We measured risk intensity in terms of investor protection and whether the company was cross-listed. Regarding investor protection, the existing literature shows that in regions with high levels of investor protection, laws and regulations are more easily met, so the degree of coordination between laws and market mechanisms is relatively high, and investors' awareness of protections is more visible and intense (Klapper and Love, 2004). In this context, the potential risk intensity for low-quality information disclosure is higher. For cross-listed companies, the regulatory environment of cross-listed companies is more stringent, and the awareness of investor protections is also higher (Ke et al., 2015). This means that cross-listed companies face more risk in relation to low-quality earnings information disclosure.

Based on the above considerations, we used the investor protection index (IPI) from the China Marketization Index Report prepared by Fan et al. (2011) to measure the degree of investor protection. Meanwhile, we used whether the company is cross-listed in both A and H shares (AH) to measure cross-listing. We divided our sample into two groups to test the risk prevention effects of CEOs' experience of the Great Famine. Table 12 reports the regression results. In the first four columns, the coefficients of Famine^*EDR and Famine^*EDD are negatively associated with EQ in two low-IPI groups (where IPI is less than the value of the median), while in the high-IPI group, the Famine^*EDR is not significant but Famine^*EDD is significant at the 10% level. In the last four columns, the results show that Famine^*EDR and Famine^*EDD are all negative at least at the 10% level in the AH list group and only A list group. We further tested the coefficient differences between these divided groups using the chi-square method. The results indicated that the absolute value of the coefficient of Famine^*EDR and Famine^*EDD in the high-IPI (AH list) group is significantly larger than that for the low-IPI (A list only) group. These results suggested that when the risk intensity for low-quality information disclosure is higher, CEOs' famine experience has a more significant effect on the corporate earnings quality, which is consistent with our expectations regarding the risk prevention effect.

The famine learning effect holds that the epochal characteristics and reasons behind the occurrence of the Great Famine increase the knowledge of famine-experienced CEOs. Under the influence of the learning effect, the probability of the CEO to implement earnings manipulation and disclose untrue earnings information is lower, resulting in higher quality earnings information disclosure. Based on the above analysis, if CEOs' early famine experience affected the quality of corporate earnings, mainly through the famine learning effect, it can be expected that when the corporate CEOs' learning ability and knowledge accumulation of the famine event are relatively higher, the association between CEOs' famine experience and corporate earnings quality should be more obvious.

We used CEO education and professional background to measure CEOs' learning ability and knowledge accumulation related to the early famine event. Jiang et al. (2016) found that when the CEO has more education, the secretary's financial experience plays a larger role in information disclosure, and earnings information of the company is of higher quality. Following their research, we expected that when the CEO's education is higher, the learning effect of the famine event should be more obvious. When the professional background of the CEO is in economics or management, the CEO can better understand the reasons for the occurrence of the Great Famine and could learn more from the disaster.

Based on CEO education and professional background data compiled in the CSMAR database, we manually collected samples to supplement the missing data as much as possible. To measure CEO education (Degree), we followed Jiang et al. (2016) again and divided academic education into five levels: high school, junior college, undergraduate degree, masters, and doctorate levels; then, we defined CEOs with a master's or doctorate degree as the high-degree group, and the remaining as the low-degree group. Meanwhile, when the CEOs' major was in economics or management, we assigned these CEOs to the E&M group; otherwise, they were assigned to the non-E&M group. Table 13 reports the results of the CEOs' famine learning effect. In the first four columns, the coefficient of the interaction term Famine^*EDR (Famine^*EDD) is −0.832 (−0.104) in the high-degree group and is significant at the 1% level, while in the low-degree group, the coefficient is not significant. In the remaining four columns, the coefficients of Famine^*EDR and Famine^*EDD in the E&M group are negatively significant at the 5 and 1% levels, respectively, while in the non-E&M group, the Famine^*EDR is not significant, but Famine^*EDD is significant at the 5% level. Using the chi-square method, we further test the differences of coefficients between these two groups. We found that the absolute value of the coefficient for Famine^*EDR and Famine^*EDD in the high-degree (E&M) group is significantly larger than that for the low-degree (non-E&M) group. In summation, these results indicated that when the CEO's learning ability and knowledge accumulation are relatively higher, the famine experience has a more significant effect on corporate earnings quality, which hold against the expectations of the famine learning effect.

Using the exogenous event of the Great Chinese Famine as the starting point, this research examined the impact of CEOs' experience of this famine on corporate earnings quality. This research found that CEOs who experienced this event led to higher earnings quality for their companies. We also showed that this effect is more pronounced for CEOs who experienced this famine during their childhood or adolescence. On the one hand, CEOs who experienced the famine showed a higher risk aversion behavior. When the degree of investor protection is high and the company is cross-listed, the positive effects of the famine experience on earnings quality are more obvious. On the other hand, CEOs show a certain learning effect related to the disaster. More education or professional background in economics or management led to significantly better corporate earnings quality for companies with famine-experienced CEOs. The above findings prove that the CEOs' famine experiences affect the earnings quality through the risk aversion effect and famine learning effect. These conclusions were validated through a series of robustness tests.

The results of the current study are consistent with those of previous research, and this study makes several contributions to the literature. First, taking CEOs' heterogeneity into account, our paper provides empirical evidence that CEOs' early-life famine experience is significantly positively associated with corporate earnings quality, thereby further enriching and supplementing the relevant areas of the literature on corporate earnings quality. Second, due to the exogenous and unpredictable nature of the Great Famine, our paper also provides a more robust estimate of the impact of CEOs' early-life experiences on corporate decision making, thereby addressing the selection bias in the existing literature that considered financial experience, overseas experience, and military experience of CEOs. Third, we explored the mechanism of CEOs' famine experiences affecting the earnings quality, and we found that the risk aversion effect and famine learning effect are the two mechanisms of CEOs' famine experiences affecting earnings quality. These findings enrich the literature on the long-term economic consequences of the Great Chinese Famine.

We can draw some practical implications from our research results. The impact of informal institutions on corporate governance is attracting increased attention. Our paper indicated that CEOs' famine experience operates in a similar manner as an implicit contract, according to which managers seek to produce high-quality outputs in their decision-making process. When an enterprise operates in an environment of high uncertainty, this implicit contract improves governance.