Societal trust and information timeliness: international evidence

Abstract

We investigate the association between societal trust and the timeliness with which value-relevant information is incorporated into stock price with a large sample of observations collected from 38 countries between 2000 and 2019. We find that in high trust countries, positive (negative) value-relevant information is impounded into stock price in a less (more) timely manner, mitigating imbalanced optimistically biased information timeliness (IOBIT). Results from robustness checks lend strong support to our inferences. Further analyses suggest that the relation between trust and IOBIT is more pronounced in emerging countries and in countries with weak investor protection, and lower levels of religiosity. The relation between trust and IOBIT is less pronounced in firms with strong corporate governance and lower managerial ability. Taken together, our study highlights the significant role that trust plays in affecting the efficiency of capital markets.

KEYWORDS:

• Trust
• information timeliness
• imbalanced optimistically biased information timeliness (IOBIT)
• corporate governance
• investor protection

JEL Classifications:

• G14
• G15
• G28
• G30

1. Introduction

Previous studies postulate that societal trust plays a crucial role in the interaction between managers and external investors in the presence of incomplete contract and moral hazard, because individuals in high trust societies spend less to protect themselves from being exploited in economic transactions (Guiso, Sapienza, and Zingales 2008; Pevzner, Xie, and Xin 2015).¹ Prior research reports evidence that trust affects a broad spectrum of social and economic outcomes. For example, the literature documents a positive relationship between societal trust and levels of economic growth and social efficiency (La Porta et al. 1997), international trade and investment (Guiso, Sapienza, and Zingales 2009), merger volume and synergy gains (Ahern, Daminelli, and Fracassi 2015), investor reaction to corporate earnings announcement (Pevzner, Xie, and Xin 2015), the likelihood and frequency of management earnings forecasts (Guan et al. 2020), and a negative association between trust and corporate tax avoidance (Kanagaretnam et al. 2018). In this study, we examine whether the timeliness with which value-relevant information is incorporated into stock price is related to the level of societal trust in the country where the firm is domiciled.

There is a consensus in the literature that managers may strategically withhold or delay the disclosure of bad news to protect their future career, maintain the esteem of peers and maximize their equity-based compensation (Kim, Li, and Zhang 2011; Kothari, Shu, and Wysocki 2009). Li (2008) shows that managers obfuscate the disclosure of unfavorable earnings news by making the annual reports difficult to read. When firms are performing well, managers are willing to be more forthcoming in the disclosure of information (Lang and Lundholm 1993; Schrand and Walther 2000). Such disclosure practices result in market reaction being biased towards good news relative to bad news, which is captured by the imbalanced optimistically biased information timeliness (IOBIT) (cf. Zhang et al. 2019).

In this paper, we study the role of societal trust in promoting more balanced information timeliness across countries. As discussed above, managers and controlling shareholders may choose to release particular news to the market in a more or less timely manner, but we still know little about the circumstances under which the motivation of doing so will be strengthened or alleviated. For example, Beekes et al. (2016) fail to find supporting evidence that better corporate governance is associated with improved timeliness of bad news. This would suggest the existence of un-identified factors that attenuate the impact of corporate governance on managers’ disclosure decisions and/or market participants’ responses. Zhang et al. (2019) highlight the differential effects of formal institutions of a country on the balance of timeliness of good news and bad news. Their analysis calls for a further investigation into the benefit–cost trade-off associated with formal and informal institutions in affecting information timeliness and the conditional factors in altering this trade-off around the world.

Prior studies show that in high trust environment, managers are less incentivized to behave opportunistically and conceal negative news, because investors in high trust environment are less likely to attribute unfavorable outcomes to managerial opportunism and penalize managers for unsatisfactory performance (Hilary and Huang 2023). Consequently, managers in high trust countries are expected to be forthcoming in the disclosure of negative information, leading to a timely price discovery of negative information. A greater increase in the timeliness of bad news relative to that of good news would mitigate IOBIT. The literature also suggests that trust and formal institutions are substitutes (Aghion et al. 2010; Kanagaretnam et al. 2018; Pevzner, Xie, and Xin 2015). This implies that the association between trust and IOBIT could be attenuated in the presence of strong formal institution. In this study, we consider both formal and informal institutions including market development, investor protection standards, and religiosity.

We examine the relation between trust and information timeliness in a sample of over 118,000 firm-year observations collected from 38 countries between 2000 and 2019. Following previous studies (Kanagaretnam et al. 2018; La Porta et al. 1997; Pevzner, Xie, and Xin 2015), we measure a country’s level of trust by its citizens’ average response to the following question in the World Values Surveys (WVS): ‘Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people?’ To measure the timeliness of price discovery, we use the metrics developed in Beekes et al. (2016). The metrics are built on prior work (e.g. Alford et al. 1993; Ball and Brown 1968; Beekes and Brown 2006) by distinguishing two types of timeliness: timeliness of good news and timeliness of bad news. The Beekes et al. measures (discussed in more detail in Section 3.1) capture the timely manner of forward-looking information being incorporated into stock price throughout a fiscal year. The main dependent variable of interest is IOBIT, calculated as the ratio of timeliness of good news to timeliness of bad news. The changes in the value of IOBIT indicate the development of balance of information timeliness.

We find that in high trust countries, positive (negative) value-relevant information is impounded into stock price in a less (more) timely manner, which leads to less optimistically biased information timeliness. Our results are robust after controlling for a set of country- and firm-specific characteristics. We interpret the findings as consistent with managers in more trusting environment being straightforward in the disclosure of bad news and investors being more willing to respond vigorously to the disclosure that is perceived to be credible. The concern that trust could be a proxy for the omitted variables applies to the majority of studies that focus on the cross-sectional variation between countries, and ours is not an exception. To alleviate such concern, we use a country’s ethnic diversity measured by ethnolinguistic fraction as an instrumental variable (IV) and find consistent results. Furthermore, we show that the relation between trust and IOBIT is more pronounced in emerging countries, countries with weak investor protection standards, and countries with lower levels of religiosity. In contrast, the association between trust and IOBIT is less pronounced for firms with strong corporate governance and firms governed by more capable managers.

Our contribution to the literature is two-fold. First, we add to the enquiry on the economic consequences of societal trust. To the best of our knowledge, only one published study explores the role played by societal trust in facilitating the capitalization of value-relevant information into stock price. Pevzner, Xie, and Xin (2015) explore the effect of trust on investor reactions to corporate earnings announcements, one of the most important channels of communication between a firm’s managers or insiders and outside investors. Based on a large sample of firm-year observations across 25 countries, they find that investors’ reactions to earnings announcements are significantly higher in more trusting countries.

Different from Pevzner, Xie, and Xin (2015), we are interested in how societal trust reshapes the asymmetric information timeliness reflected by managers’ strategically withholding the disclosure of bad news but quickly revealing good news to the market (Kim, Li, and Zhang 2011; Kothari, Shu, and Wysocki 2009; Lang and Lundholm 1993; Schrand and Walther 2000). We find that societal trust delays the price discovery of positive information but facilitates the incorporation of negative information into market price, thereby mitigating the asymmetric information timeliness that is biased towards good news. In addition, Pevzner, Xie, and Xin (2015) adopt the traditional event study method and use the market models of stock returns to calculate the abnormal return and trading volume in the 3-day short window around earnings announcement dates relative to those in the non-event period. One issue of the informational efficiency measures based on the market models is that high/low fundamental correlation across sample countries/industries could cause biased results (Morck, Yeung, and Yu 2000). Another potential problem of their approach is that the information released around earnings release dates might not be complete because financial report is not a highly timely venue for the provision of value-relevant information, since most of the information has been reported on traditional and social media before the earnings announcement day (Ball and Brown 1968; Blankespoor, Miller, and White 2014). Our model-free information timeliness measure captures the different speed of incorporating good news and bad news into stock price over a 1-year window. Finally, Pevzner, Xie, and Xin (2015) argue from the perspective of investors and state that ‘financial reporting by firms in more trusting countries is perceived by investors as more credible’. In this paper, we argue from the perspective that societal trust mitigates managerial incentives to behave opportunistically and conceal negative news.

Second, our study highlights the role of a country’s informal institution in determining capital market efficiency. The formation and accumulation of trust, which is an integral part of informal institution, is related to a country’s history, politics, religion, and other social infrastructure (Guiso, Sapienza, and Zingales 2004). Managers in high trust environment are more likely to be forthcoming in disclosing negative information to the market, which results in lower information asymmetry and more efficient allocation of financial resources. Although early studies such as Pevzner, Xie, and Xin (2015) show that trust is positively associated with the effectiveness of information transmission in the capital market, our research provides new insights by documenting that the relation between trust and IOBIT is less pronounced in countries with better formal and informal institution (i.e. better market development, stronger investor protection, and higher levels of religiosity), which implies a substitutive relation between formal and informal institutions. Such findings point to the significance of controlling informal institutions when researchers analyze the effects of formal institution on business and economic outcomes in a cross-country setting, especially when the expected results are not found.²

The remainder of this paper is structured as follows: Section 2 reviews the related literature and develops testable hypotheses. Section 3 describes sample, variable definition, and empirical models. Section 4 discusses the main findings. The final section concludes.

2. Related literature and hypothesis development

2.1. Societal trust and price discovery of value-relevant news

Our work belongs to the strand of literature that examine the impact of a country’s characteristics such as institutions and financial market openness on corporate disclosure and information environment (e.g. Alford et al. 1993; Ali and Hwang 2000; Bae, Bailey, and Mao 2006; Beekes et al. 2016; Landsman, Maydew, and Thornock 2012; Zhang et al. 2019). However, evidence on the effect of national culture (e.g. societal trust) is still limited. Social capital is defined as the propensity of members of a society to cooperate to realize socially desirable outcomes (Putnam 1993). Paldam (2000) posits that there are three families of social capital concept: (1) trust, (2) ease of cooperation, and (3) network. According to Putnam (2000), the social capital of a society can be measured by the civic engagement of its members, which enhances collaboration among members of society and engenders reciprocity: in the anticipation that members of a society will receive help when needed, they are more willing to take actions that would not appear in their self-interests.

Knack and Keefer (1997) are among the first to measure social capital with trust based on the World Values Surveys (WVS) question: ‘Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people?’ They show that trust is positively related to economic growth and investment.³ La Porta et al. (1997) also measure social capital with trust based on the WVS question and find that trust is related to GDP growth, the size of the largest firms in the economy, tax compliance, and lack of corruption. They further document that trust is lower in countries with dominant hierarchical religions. In a similar vein, Zak and Knack (2001) show that trust is negatively (positively) related to income inequality and corruption (contract enforceability and investors’ rights).

An important work by Glaeser et al. (2000) distinguishes between trust and trustworthiness. Trust reflects how agents believe others will behave towards them. In high trust societies, agents expect others to act in a cooperative manner and not being taken advantage of by others. Trustworthiness, on the other hand, captures whether agents are indeed worthy of being trusted and behave in the collaborative way. It is predicted that trust and trustworthiness are positively correlated because if one has high level of trust in other people, but others turn out to be untrustworthy, then one will eventually have reduced level of trust in others.

Trust is related to another broad concept: culture. Guiso, Sapienza, and Zingales (2006) define culture as ‘those customary belief and value that ethnic, religious, and social groups transmit fairly unchanged from generation to generation’. Guiso, Sapienza, and Zingales (2009) use trust as a measure of culture and show that bilateral trust between European countries affects portfolio investment and direct investment, after controlling for country-specific variables. Until recently the literature has been silent on the association between trust and information timeliness, and our study aims to fill this research gap.

Trust could affect information timeliness through two channels. On one hand, in high trust societies, investors are more likely to assign a lower probability to mangers behaving opportunistically and manipulating financial results (Pevzner, Xie, and Xin 2015). As a result, investors perceive value-relevant news of a firm as more credible and thus respond to the information in a more timely manner. It is also likely that investors pay less attention to firms’ news because they view managers are trustworthy and forthright (Guiso, Sapienza, and Zingales 2008). If the latter is the case, one might find a negative effect of trust on the speed of price discovery of information. On the other hand, managers in high trust societies are more likely to reciprocate the trust that is placed in them and are more cautious and sensitive to fairness considerations when they deal with major corporate issues (Kanagaretnam et al. 2018). The reciprocation of trust may result in a rigorous verification of information prior to disclosure and a timely announcement when the information is verified to avoid legal and ethical problems.

With respect to the asymmetric disclosure of corporate news, prior research finds that managers may strategically withhold or delay the disclosure of bad news but quickly reveal good news to investors. Doing so enables managers to protect their careers (e.g. promotion and employment opportunities) and equity-based compensation (Kim, Li, and Zhang 2011; Kothari, Shu, and Wysocki 2009). In countries with higher level of trust, managers have less incentive to hide bad news from the market because investors are less likely to attribute unfavorable outcomes to managerial opportunism and penalize managers for unsatisfactory performance. For example, Hilary and Huang (2023) find that firms located in areas with higher trust utilize lower-powered executive compensation and are less likely to dismiss CEOs for poor performance, suggesting that trust can be a substitute for formal governance mechanism in encouraging risk-taking. Consistently, Li, Wang, and Wang (2017) find that firms headquartered in regions of high social trust tend to have smaller stock price crash risks, which implies that managers withhold bad news to a less extent in high trust environments.

When managers release news to the market in high trust countries, it is likely that investors’ limited information processing capacity makes it impossible for them to pay attention to the complete set of favorable and unfavorable news.⁴ Because the market generally requires a lower degree of verification for recognition of bad news relative to good news (Basu 1997), on average investors may attend to bad news first, which leads to a fast incorporation of such news into stock price. Due to investors’ limited cognitive capability in processing information, good news might not be attended by all the investors, which results in a partial and slow impounding of good news into stock price. In summary, in high trust countries, a greater increase in the timeliness of bad news relative to that of good news leads to a decrease in imbalanced optimistically biased information timeliness (IOBIT). H1 is formulated as follows:

H1: Societal trust mitigates the imbalanced optimistically biased information timeliness.

2.2. Moderating role of country-level characteristics

Imbalanced and optimistically biased information environment can be alleviated in developed countries where institutional quality is good and sophisticated investors enhance market efficiency through informed trading (Admati and Pfleiderer 2000; Edmans and Manso 2011; Ferreira, Ferreira, and Raposo 2011; Kacperczyk, Sundaresan, and Wang 2021; La Porta et al. 1998). In contrast, emerging markets are more likely to suffer poor information timeliness and serious IOBIT due to weak institutional environment, poor disclosure quality, and lack of sophisticated market participants, which implies that high levels of societal trust may be more critical in protecting outside investors by reducing the expropriation and private benefit consumption of managers and controlling shareholders (Aslan and Kumar 2014; Zhang et al. 2017; Zhang, Piesse, and Filatotchev 2015). Consequently, the potential benefit–cost trade-off on information timeliness associated with societal trust is amplified in emerging markets. Thus we test the following hypothesis:

H2a: The impact of societal trust in mitigating imbalanced optimistically biased information timeliness is more pronounced in emerging countries.

Studies on trust document that trust and formal institutions are substitutes as well. For example, Pevzner, Xie, and Xin (2015) find that the impact of trust on market reaction to corporate earnings announcements is weaker when investor protection is stronger. Kanagaretnam et al. (2018) find that the relation between trust and tax avoidance is less pronounced when the legal institutions in a country are stronger. Guan et al. (2020) find that the impact of trust on the issuance of management earnings forecasts is less (more) pronounced for firms from countries with stronger (weaker) formal institutions. Strong investor protection helps to ‘level the field’ by imposing high legal and regulatory standards on firm information disclosure (Admati and Pfleiderer 2000; Aslan and Kumar 2014; Leuz, Nanda, and Wysocki 2003), especially on forward-looking information disclosure (Baginski, Hassell, and Kimbrough 2002). To protect investors from being misled by forward-looking information, strong investor protection allows strict information verification to ensure its credibility. It also strengthens the threat of legal or regulatory penalties for failure to release unfavorable information in a timely manner. Therefore, strong investor protection standards are expected to reduce the demand on trust to improve information timeliness. In other words, trust plays a more important role in mitigating IOBIT in countries where investor protection standards are weak. Thus we test the following hypothesis:

H2b: The impact of societal trust in mitigating imbalanced optimistically biased information timeliness is more pronounced when investor protection is weak.

Previous studies show that religiosity reduces criminal activity because individuals with strong religious belief are more likely to exhibit self-control and have ethical intention (Vitell 2009). For example, Grullon, Kanatas, and Weston (2009) report that firms located in countries with higher level of religiosity are less likely to be targets of class action security lawsuits, engage in backdating options, grant excessive compensation package to executives and practice aggressive earnings management. Callen and Fang (2015) find that higher levels of religiosity lower future stock price crash risk, which is consistent with the view that religion helps to reduce bad-news-hoarding activities by managers. Therefore, we expect that managers in high religiosity countries have reduced incentive to conceal bad news from the market, because doing so generates strong cognitive dissonance and discomfort (Akerlof 1980). In other words, religiosity may substitute trust in alleviating imbalanced optimistically biased information timeliness (IOBIT), which suggests that the effect of trust on IOBIT is more salient in countries with lower level of religiosity. Thus we test the following hypothesis:

H2c: The impact of societal trust in mitigating imbalanced optimistically biased information timeliness is more pronounced when the level of religiosity is lower.

2.3. Moderating role of firm-level characteristics

It is well documented that high-quality corporate governance mitigates principal–agent conflicts by monitoring and disciplining management within a firm, which reduces the minority investors’ reliance on market discipline (Gorton, Huang, and Kang 2017; He et al. 2013; Jensen and Meckling 1976; La Porta et al. 1998). When corporate governance is stronger, managers are more likely to be forthcoming in the disclosure of value-relevant news, particularly bad news (Beekes and Brown 2006; Haß, Vergauwe, and Zhang 2014). Moreover, managers and controlling shareholders are likely to adopt better governance practices only if their share of increased firm value is greater than the loss of value from expropriation and private benefit consumption (Fauver et al. 2017). Our view is that for poorly-governed firms, high levels of societal trust may help protect outside investors by reducing the expropriation and private benefit consumption of managers and controlling shareholders due to the reciprocation of trust as discussed earlier. As a result, the possible delay in discourse of bad news might be alleviated, and good news would be rigorously verified prior to disclosure. For good-governed firms, managers and controlling shareholders have been effectively monitored by internal governance mechanisms so that the effects of external factors (e.g. culture) are likely to be attenuated. Thus we expect that societal trust and corporate governance quality are substitutive with each other in affecting IOBIT. We test the following hypothesis:

H2d: The impact of societal trust in mitigating imbalanced optimistically biased information timeliness is less pronounced when corporate governance quality is higher.

Last but not the least, prior research concludes that capable managers are more knowledgeable of business operation, are more efficient in utilizing available resources and have better judgement of industry trends (Demerjion, Lev, and McVay 2012). Firms led by more capable managers exhibit better market and operating performance (Aktas, Louca, and Petmezas 2019; Chen, Podolski, and Veeraraghavan 2015; Doukas and Zhang 2020). We expect the effect of societal trust on IOBIT to be more significant in firms governed by talented and skilled managers, because such managers have a better understanding of the benefits of corporate transparency and will be straightforward in disclosing value-relevant information (including negative news) to the market, while investors tend to attach higher credibility to information released by these firms. Thus we test the following hypothesis:

H2e: The impact of societal trust in mitigating imbalanced optimistically biased information timeliness is more pronounced when managerial ability is higher.

3. Data and variables

In this section, we describe the sample construction and variables used in this study. Our sample includes public firms from 38 countries for the period 2000–2019.⁵ To reduce survivorship bias, we include companies delisted during the sample period. We exclude firms with missing data in societal trust, information timeliness or control variables. We also exclude firms in financial sectors with standard industrial classification codes 6011–6799 due to their unique accounting and financial characteristics. The selection process results in a final sample of 21,019 unique firms and 118,553 firm-year observations in the baseline regression model. We obtain share price and financial data of public firms from the Refinitiv Datastream and Refinitiv Worldscope. Data for constructing societal trust and religiosity of a country is from the World Values Survey. Major corporate governance reform years are collected from the study of Fauver et al. (2017) and documents of local stock exchanges. Data for investor protection standards and other country-level financial and macroeconomic variables is from the World Bank. Table 1 reports the sample distribution by country and year.

Table 1. Sample composition, average societal trust, and major corporate governance reform year by country.

Country	# of firms	# of firm-year observations	Average societal trust	Major corporate governance reform years	Capital market development
Argentina	16	72	0.202	2001	Emerging
Australia	956	4764	0.521	2004	Developed
Brazil	175	1164	0.073	2002	Emerging
Canada	1151	4329	0.443	2004	Developed
Chile	55	370	0.157	2001	Emerging
China	2587	11,954	0.641	2001	Emerging
Colombia	3	14	0.083	2001	Emerging
Egypt	47	182	0.135	2002	Emerging
Finland	82	347	0.588	2004	Developed
France	313	1077	0.187	2003	Developed
Germany	468	3331	0.408	2002	Developed
Greece	16	43	0.084	2002	Developed
Hong Kong	718	3748	0.434	2005	Developed
India	896	3864	0.203	2002	Emerging
Indonesia	191	801	0.306	2007	Emerging
Israel	15	26	0.235	2000	Developed
Italy	169	570	0.292	2006	Developed
Japan	1989	15,489	0.389	2002	Developed
Korea	1246	5014	0.309	1999	Emerging
Malaysia	564	3272	0.120	2001	Emerging
Mexico	82	406	0.130	2001	Emerging
Netherlands	83	539	0.549	2004	Developed
New Zealand	75	382	0.556	2003	Developed
Norway	111	346	0.663	2005	Developed
Pakistan	46	105	0.242	2002	Emerging
Peru	13	76	0.077	2005	Emerging
Philippines	56	187	0.072	2002	Emerging
Poland	186	699	0.220	2002	Emerging
Russia	91	457	0.271	2002	Emerging
Singapore	386	1506	0.284	2003	Developed
South Africa	145	712	0.197	2002	Emerging
Spain	110	787	0.236	2006	Developed
Sweden	206	1319	0.659	2006	Developed
Switzerland	129	503	0.511	2002	Developed
Thailand	312	1569	0.352	2002	Emerging
Turkey	163	945	0.129	2002	Emerging
United Kingdom	1133	3494	0.304	1998	Developed
United States	6035	44,090	0.384	2003	Developed
Total	21,019	118,553

This table shows the numbers of firms and firm-year observations that have both firm-level and country-level data by country. Average societal trust shows the average societal trust scores of four waves of the World Value Survey (WVS) between 1999 and 2020. Years of major corporate governance reform are collected from the work of Fauver et al. (2017) and documents of local stock exchanges. Information on capital market development is obtained from MSCI Developed Markets Index and MSCI Emerging Markets Index.

3.1. Measuring information timeliness

Beekes and Brown (2006) propose a novel intra-year timeliness metric in the spirit of Ball and Brown (1968) and Alford et al. (1993). The metric assesses how accurately a firm's share price ($P_t$), observed at daily intervals throughout the year, approximates its terminal value ($P_0$) 10 trading days after final earnings numbers have been released.⁶ The measure reflects the net effect of all value-relevant information impounded by all market participants in share price over a year. In this paper, we adopt the approach developed by Beekes et al. (2016), which extends the Beekes and Brown’s measure by separating timeliness of difference types of information. Specifically, we calculate two measures of timeliness, namely timeliness of good news (TGOOD) and timeliness of bad news (TBAD).

The two timeliness measures are calculated for each fiscal year from 2000 to 2019. Within each estimation ‘window’, the ending day ($t=0$) is 14 calendar days after the release day of final earnings numbers, and the staring day ($t=s$) is 365 calendar days prior to the ending day. To calculate timeliness of good news (TGOOD), we first identify the third quartile of a share’s positive unadjusted daily log returns, $r_t>0,t=s,\dots ,0$, where $s$ and $0$ stand for starting day and ending day in an estimation ‘window’ respectively.⁷ We then construct a market-adjusted daily log return series, $r_t^{\ast }>0$. If the day’s return is negative, we set the good news return on that day to zero. We next create a series of cumulative good news returns, $C_t^G$, by setting $C_{-365}^G=0$ and cumulating the daily market-adjusted log return series $C_t^G=C_{t-1}^G+r_t^{\ast }$ from day –364 to day $0$. $C_0^G$ represents the aggregation of all non-negative log returns during the year. The timeliness of good news is thereby calculated as (1) $\mathrm{TGOOD}=\left(\left(\sum _{\text{t}=-365}^{\text{t}=-1}(C_0^G-C_t^G)/C_0^G\right)-0.5\right)/365$(1) The equivalent procedure is adopted to estimate timeliness of bad news (TBAD).⁸ $C_0^G$ and $C_t^G$ in Equation (1) are replaced with $C_0^B$ and $C_t^B$ to represent bad news inputs. The longer it takes a firm’s cumulative returns to capture all positive (negative) value-relevant information and converge to its ‘final’ cumulative return $C_0^G$ ($C_0^B$), the larger is the value of TGOOD (TBAD). A high value for TGOOD (TBAD) thus indicates low intra-year timeliness.

The intuition behind the timeliness measures is straightforward. Suppose share price increased dramatically on a major piece of good news in relation to its current year’s earnings on the first day of trading (day –364), and then simply tracked the market index for the remaining 364 days. In this case, $C_0^G-C_{-364}^G$ is small and subsequent $C_0^G-C_t^G$ for the year would also be small assuming only limited numbers of good news remain to contribute to the cumulative positive return. Thus the speed of adjustment is at its maximum and the value of TGOOD is low. Timeliness of bad news (TBAD) is defined in a similar fashion. The imbalanced and optimistically biased information timeliness (IOBIT) is measured by the ratio of timeliness of good news (TGOOD) divided by timeliness of bad news (TBAD). The IOBIT is thus: (2) $\mathrm{IOBIT}=\mathrm{TGOOD}/\mathrm{TBAD}$(2) When good and bad value-relevant information is incorporated into share price in a similar timely manner, IOBIT should be close to 1, thus balanced information timeliness. If disclosure of bad news is delayed relative to that of good news as indicated by Kothari, Shu, and Wysocki (2009), IOBIT should be below 1, suggesting an imbalanced and optimistically biased information timeliness. A decreasing timeliness of good news with an increasing timeliness of bad news, a greater decrease in timeliness of good news than in timeliness of bad news, or a greater increase in timeliness of bad news than in timeliness of good news, will lead to an increase in the value of IOBIT. Higher values of IOBIT indicate less optimistically biased information timeliness.

3.2. Measuring societal trust

Following previous studies such as Pevzner, Xie, and Xin (2015) and Kanagaretnam et al. (2018), we construct our societal trust score (TRUST) based on the results the World Value Survey (WVS). The surveys were conducted in seven waves from 1981 to 2020. We use the surveys in Wave 4 to Wave 7 because they provide coverage of most sample countries and years. The surveys cover four time windows including 1999–2004, 2005–2009, 2010–2014, and 2017–2020. TRUST is estimated based on responses to the WVS question: ‘Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people?’ The two possible answers are ‘Most people can be trusted’ and ‘Need to be very careful’. We recode the response to this question to 1 if a survey participant reports that most people can be trusted and 0 otherwise. TRUST score is calculated as the mean of the responses for each country-wave. Within a wave, a country is allocated the same score for all years covered by the wave. To fill the 2 years gap between waves 6 and 7, the score of 2015 takes the value of wave 6 (2010–2014). The score of 2016 takes the value of wave 7 (2017–2020). Higher values indicate higher societal trust. As shown in Table 1, Norway has the highest average score of TRUST (0.663) and Philippines has the lowest average score of TRUST (0.072). Societal trust scores by wave and country are presented in Appendix A.

3.3. Country and firm characteristics

The effects of societal trust could be a reflection of those of other relevant country-level characteristics such as culture and religion. For example, it is likely that managers behave more opportunistically in countries with higher levels of collectivism (i.e. the opposite of individualism) (Zhang, Liang, and Sun 2013). Callen and Fang (2015) find that religiosity decreases stock price crash risk because managers in more religious countries are less likely to behave opportunistically. To alleviate the impacts of other cultural and religious variables, we include religiosity (RELIG), individualism (IND), and uncertainty avoidance (UAI). Religiosity scores of a country are estimated based on responses to the WVS question: ‘Apart from weddings, funerals and christenings, about how often do you attend religious services these days?’ We recode the response to this question to 1 if a survey participant answers ‘once a week’ and 0 otherwise. We then calculate the mean of the response for each country-wave. The score is calculated for each wave of the WVS. Within a wave, the score is calculated once and applies to all country-years covered by the wave. Scores of 2015 use the values of wave 6 (2010–2014). Scores of 2016 use the values of wave 7 (2017–2020). Higher values indicate higher levels of religiosity. Scores of individualism (IND) and uncertainty avoidance (UAI) are collected from the Hofstede’s website and rescaled to be between 0 and 1.

We construct a country-level corporate governance dummy variable (CGRF) that is equal to 1 if a country-year is in or after the year when a major corporate governance reform became effective in the country, and 0 otherwise. We obtain the information on major corporate governance reforms from the study by Fauver et al. (2017) and local stock exchange regulators.⁹ The corporate governance reform covers components such as board independence, audit committee and auditor independence, and separation of the chairman and CEO positions. Effective years of major corporate governance reforms are reported in Table 1. Most of our sample countries launch their major corporate governance reforms in the 2000s. The earliest effective year is from the UK, which is featured by the introduction of the Combined Code in 1998.¹⁰

To measure investor protection standard of a country (IPS), we use the rule of law index from the World Governance Indicators. The WGI institutional indicators are found to be related to firm operations and value (Nguyen, Locke, and Reddy 2015; Van Essen, Engelen, and Carney 2013; Zhang et al. 2022). The annual rule of law index captures perceptions of the extent to which agents have confidence in and abide by the rules of society, and in particular the quality of contract enforcement, property rights, the police, and the courts, as well as the likelihood of crime and violence. We rescaled the index to be between 0 and 1, with greater values indicating better national investor protection standards.

As suggested by prior literature (Beekes et al. 2016; Landsman, Maydew, and Thornock 2012; Leuz, Nanda, and Wysocki 2003; Pevzner, Xie, and Xin 2015; Zhang et al. 2019; Zhang et al. 2023), it is likely that the variation in information timeliness depends on market differences between countries. We include several variables to control for the economic, informational, and institutional environment of a country. These variables include information dissemination (INFOR), stock market development (STOCK), inflation (INFL), and GDP growth (GDPG). In addition to country-level factors, we also include firm-level control variables which may affect information timeliness. Firm characteristics may either affect information timeliness directly, or exert their effects through influencing other factors such as corporate governance (Aggarwal et al. 2011; Beekes et al. 2016; Beekes and Brown 2006; Ferreira, Ferreira, and Raposo 2011; Guan et al. 2020; Lev and Penman 1990; Zhang et al. 2019; Zhang et al. 2023).¹¹ The firm-level controls we use include firm size (SIZE), earnings management (EM), ownership concentration (CLOSE), profitability (PROFIT), financial leverage (LEV), Tobin’s q (TQ), cash holdings (CASH), capital expenditure (CAPEX), share turnover (TURN), stock return volatility (VOL), and number of analysts following the stock (ANALYST). Detailed definitions of variables and data resources are provided in Table 2.

Table 2. Variable definitions.

Variable	Acronym	Definition	Data source
Panel A: Timeliness variables
Imbalanced optimistically biased information timeliness	IOBIT	Imbalanced, optimistically biased information timeliness, measured with the ratio of timeliness of good news (TGOOD) divided by timeliness of bad news (TBAD)	Refinitiv Datastream & Worldscope
			Authors’ calculation
Timeliness of good news	TGOOD	The timeliness of good news, based on daily market-adjusted log returns $r_t^{\ast }>0$, as estimated in Equation (1)	Refinitiv Datastream & Worldscope
			Authors’ own calculation
Timeliness of bad news	TBAD	The timeliness of bad news, based on daily market-adjusted log returns $r_t^{\ast }>0$, estimated with the equivalent procedure shown in Equation (1)	Refinitiv Datastream & Worldscope
			Authors’ calculation
Panel B: Country-level variables
Societal trust	TRUST	Societal trust scores of a country are estimated based on responses to the WVS question: ‘Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people?’ We recode the response to this question to one if a survey participant reports that most people can be trusted and zero otherwise. We then calculate the mean of the response for each country-wave. The score is calculated for each wave of the WVS. Within a wave, the score is calculated once and applies to all country-years covered by the wave. Scores of 2015 use the values of Wave 6 (2010–2014). Scores of 2016 use the values of Wave 7 (2017–2020). Higher values indicate higher societal trust	Four waves of the World Value Survey (WVS) in 1999–2004, 2005–2009, 2010–2014, and 2017–2022
			Authors’ calculation
Religiosity	RELIG	Religiosity scores of a country are estimated based on responses to the WVS question: ‘Apart from weddings, funerals and christenings, about how often do you attend religious services these days?’ We recode the response to this question to one if a survey participant answers ‘once a week’ and zero otherwise. We then calculate the mean of the response for each country-wave. The score is calculated for each wave of the WVS. Within a wave, the score is calculated once and applies to all country-years covered by the wave. Scores of 2015 use the values of Wave 6 (2010–2014). Scores of 2016 use the values of Wave 7 (2017–2020). Higher values indicate higher levels of religiosity	Four waves of the World Value Survey (WVS) in 1999–2004, 2005–2009, 2010–2014, and 2017–2022
			Authors’ calculation
Individualism	IND	Indicates the extent to which individuals are integrated into groups. Individualism score ranges from 0 (collectivism) to 100 (very individualistic). The score is rescaled to be between 0 and 1	Hofstede’s website: www.hofstede-insights.com
Uncertainty avoidance	UAI	Indicates the extent to which people feel comfortable or uncomfortable with uncertainty and ambiguity and try to avoid such situations. Uncertainty avoidance score ranges from 0 (no fear of uncertainty) to 100 (very concerned about uncertainty). The score is rescaled to be between 0 and 1	Hofstede’s website: www.hofstede-insights.com
Corporate governance reform	CGRF	A dummy variable that is equal to one if a country-year is in or after the year when a major corporate governance reform became effective in the country, and zero otherwise	Fauver et al. (2017) & Local stock exchanges
Investor protection standards	IPS	Measured by the rule of law index from the World Governance Indicators (WGIs). The annual index captures perceptions of the extent to which agents have confidence in and abide by the rules of society, and in particular the quality of contract enforcement, property rights, the police, and the courts, as well as the likelihood of crime and violence. The index is rescaled to be between 0 and 1	WGI from the World Bank
Information dissemination	INFOR	Number of individuals using the internet divided by population	WDI from the World Bank
Stock market development	STOCK	Annual market capitalization of listed domestic companies divided by GDP	WDI from the World Bank
Inflation	INFL	Inflation is measured by GDP deflator	WDI from the World Bank
GDP growth	GDPG	Annual GDP growth rate	WDI from the World Bank
Ethnolinguistic fractionalization	ELF	The probability that two people drawn at random from a country’s population will not belong to the same ethnolinguistic group. A higher ELF index indicates a more fragmented country	Mauro (1995)
Panel C: Firm-level variables
Firm size	SIZE	The natural logarithm of market capitalization of a firm in US dollars (Worldscope item 07210)	Refinitiv Worldscope
Earnings management	EM	A measure of accrual-based earnings management which is calculated using the modified Jones model (Dechow, Sloan, and Sweeney 1995). See Appendix B for estimation procedure	Refinitiv Worldscope
			Authors’ calculation
Closely held shares	CLOSE	A measure of ownership concentration which is the ratio of number of shares held by insiders to number of common shares outstanding	Refinitiv Worldscope
Profitability	PROFIT	Earnings before interest, taxes and depreciation (Worldscope item 18198) divided by total assets (Worldscope item 02999)	Refinitiv Worldscope
Leverage	LEV	Total debt (Worldscope item 03255) divided by total assets (Worldscope item 02999)	Refinitiv Worldscope
Tobin’s q	TQ	Total assets (Worldscope item 02999) minus book value of equity (Worldscope item 03501) plus market value of equity (Worldscope item 08001) divided by total assets	Refinitiv Worldscope
Cash	CASH	Cash and short-term investments (Worldscope item 02001) divided by total assets (Worldscope item 02999)	Refinitiv Worldscope
Capital expenditure	CAPEX	Capital expenditures (Worldscope item 04601) divided by total assets (Worldscope item 02999)	Refinitiv Worldscope
Share turnover	TURN	Share trading volume (Datastream item VO) divided by adjusted shares outstanding (Datastream items NOSH/AF)	Refinitiv Datastream
Stock return volatility	VOL	The standard deviation of daily stock returns over the 360 days prior to fiscal year end dates	Refinitiv Datastream
Analyst following	ANALYST	The natural logarithm of number of analysts following a firm in a fiscal year	I/B/E/S
Corporate governance quality	CGQ	Average corporate governance score based on Chung, Elder, and Kim (2010). See Appendix C for corporate governance standards relating to financial and operating transparency. CGQ is calculated as the mean of annual ratios of the CGQ scores according to the 22 governance standards divided by the full score of 22. Missing information in a year is coded as zero	Refinitiv Eikon
Historical stock return	RET	Annualized market-adjusted stock return in the past 2 years	Refinitiv Datastream & Worldscope
			Authors’ calculation

Table 3 provides descriptive statistics for variables used in main tests. All time-varying variables are winsorized at the top and bottom 1% to control for outliers. Among major variables of interest, imbalanced and optimistically biased information timeliness (IOBIT) ranges from 0.712 to 1.428, with a mean and median of 1.014 and 1.003, and a standard deviation of 0.143. Timeliness of good news (TGOOD) ranges from 0.322 to 0.690, with a mean and median of 0.503 and 0.503, and a standard deviation of 0.076. Timeliness of bad news (TBAD) ranges from 0.322 to 0.689, with a mean and median of 0.501 and 0.500, and a standard deviation of 0.077. The statistics are generally in line with previous studies including Beekes et al. (2016) and Zhang et al. (2019). For example, Beekes et al. (2016) construct a sample of 23 countries from 2003 to 2008, and they report a timeliness of good (bad) news with a mean and median of 0.52 (0.52) and 0.52 (0.51), and a standard deviation of 0.08 (0.09). Societal trust (TRUST) ranges from 0.066 to 0.663, with a mean and median of 0.392 and 0.388, and a standard deviation of 0.131. The statistics are comparable to those reported by Guan et al. (2020), Kanagaretnam et al. (2018) and Pevzner, Xie, and Xin (2015).

Table 3. Summary statistics (N = 118,553).

Variables	Mean	Median	Standard deviation	Min	Max
IOBIT	1.014	1.003	0.143	0.712	1.428
TGOOD	0.503	0.503	0.076	0.322	0.690
TBAD	0.501	0.500	0.077	0.322	0.689
TRUST	0.392	0.388	0.131	0.066	0.663
RELIG	0.152	0.187	0.104	0.011	0.395
IND	0.637	0.744	0.374	0.000	1.000
UAI	0.494	0.413	0.231	0.000	1.000
CGRF	0.920	1.000	0.272	0.000	1.000
IPS	0.702	0.820	0.233	0.142	0.937
INFOR	0.670	0.714	0.204	0.044	0.960
STOCK	1.311	1.052	1.643	0.219	11.247
INFL	2.105	1.884	2.255	−1.895	10.526
GDPG	0.028	0.026	0.027	−0.053	0.097
SIZE	13.388	13.370	1.786	9.334	17.894
EM	0.090	−0.005	0.724	−2.123	4.823
CLOSE	0.346	0.329	0.256	0.000	0.901
PROFIT	0.079	0.102	0.173	−0.825	0.406
LEV	0.216	0.197	0.181	0.000	0.699
TQ	1.827	1.374	1.405	0.236	8.699
CASH	0.186	0.122	0.192	0.001	0.899
CAPEX	0.053	0.036	0.057	0.001	0.316
TURN	1.626	1.021	1.843	0.008	10.213
VOL	0.024	0.021	0.012	0.008	0.073
ANALYST	3.742	3.850	1.182	0.693	5.878

This table reports summary statistics of variables for the full sample. All time-varying variables are winsorized at the top and bottom 1% to control for outliers. For further details on variable measurement and data sources, see Table 2.

3.4. Correlation analysis

Table 4 provides the Pearson correlation coefficients of the main variables. We find that TRUST is positively correlated with IOBIT (0.02), suggesting that high trust societies present less optimistically biased information timeliness. The correlation between TGOOD and TBAD is 0.59, suggesting that these two variables capture similar but different types of information reflected in share price. TBAD is negatively correlated with TRUST (−0.02), suggesting that high levels of societal trust improve timeliness of bad news. We rely on multivariate regressions in the following sections to test our hypotheses by controlling for other variables that could affect information timeliness.

Table 4. Correlation matrix (N = 118,553).

		[2]	[3]	[4]	[5]	[6]	[7]	[8]	[9]	[10]	[11]	[12]	[13]	[14]	[15]	[16]	[17]	[18]	[19]	[20]	[21]	[22]	[23]	[24]
IOBIT	[1]	0.41*	−0.47*	0.02*	0.01*	0.03*	−0.01*	−0.01*	0.03*	−0.02*	0.01	0.01	−0.02*	−0.06*	−0.01*	−0.01*	−0.05*	0.01	−0.08*	0.01	−0.02*	−0.03*	0.01	0.01
TGOOD	[2]		0.59*	0.01	0.01*	0.05*	−0.02*	0.05*	0.04*	0.06*	−0.01*	0.05*	0.02*	−0.05*	0.04*	−0.03*	−0.06*	0.04*	−0.09*	−0.03*	0.01*	−0.08*	−0.19*	0.02*
TBAD	[3]			−0.02*	−0.01*	0.03*	−0.01*	0.06*	0.03*	0.07*	−0.01	0.06*	0.04*	0.01*	0.04*	−0.02*	−0.01*	0.04*	−0.01*	−0.03*	0.03*	−0.06*	−0.20*	0.02*
TRUST	[4]				0.35*	−0.04*	−0.31*	0.03*	0.05*	0.15*	0.02*	−0.23*	0.20*	0.11*	−0.01	−0.07*	−0.06*	−0.03*	0.08*	0.05*	0.01	0.21*	0.06*	−0.01*
RELIG	[5]					0.53*	−0.26*	−0.27*	0.16*	−0.22*	−0.05*	0.29*	−0.07*	−0.05*	0.11*	−0.22*	−0.06*	0.03*	0.07*	0.02*	0.04*	0.03*	0.17*	0.15*
IND	[6]						−0.15*	−0.16*	0.59*	0.28*	−0.10*	−0.07*	−0.43*	−0.03*	0.12*	−0.49*	−0.16*	−0.04*	0.03*	0.07*	−0.01	0.03*	0.17*	0.19*
UAI	[7]							0.05*	0.11*	0.20*	−0.27*	−0.27*	−0.40*	−0.03*	−0.05*	0.05*	0.06*	−0.01	−0.16*	−0.03*	−0.06*	−0.14*	−0.11*	−0.07*
CGRF	[8]								−0.10*	0.28*	0.04*	0.01*	0.08*	0.09*	0.01	0.03*	0.05*	−0.01*	0.01*	0.01	−0.02*	0.03*	−0.26*	−0.01*
IPS	[9]									0.56*	0.23*	−0.40*	−0.59*	−0.08*	0.07*	−0.39*	−0.13*	−0.08*	−0.09*	0.08*	−0.05*	−0.11*	0.09*	0.12*
INFOR	[10]										0.15*	−0.47*	−0.45*	0.03*	0.02*	−0.28*	−0.11*	−0.07*	−0.06*	0.09*	−0.11*	0.02*	−0.08*	0.06*
STOCK	[11]											−0.02*	−0.01*	0.05*	−0.02*	0.07*	−0.01	−0.02*	−0.02*	0.04*	−0.02*	−0.08*	−0.01	0.04*
INFL	[12]												0.37*	−0.02*	−0.02*	0.16*	0.08*	0.06*	0.01*	−0.06*	0.12*	−0.04*	−0.04*	−0.01
GDPG	[13]													0.06*	−0.09*	0.22*	0.08*	0.04*	0.13*	−0.03*	0.09*	0.12*	−0.08*	−0.08*
SIZE	[14]														0.01*	−0.18*	0.31*	0.13*	0.21*	−0.14*	0.02*	0.13*	−0.38*	0.57*
EM	[15]															−0.06*	0.03*	0.02*	−0.01*	−0.02*	0.02*	0.05*	−0.01	0.03*
CLOSE	[16]																0.10*	0.01	−0.07*	−0.04*	0.04*	−0.26*	−0.03*	−0.27*
PROFIT	[17]																	0.03*	−0.01*	−0.36*	0.13*	−0.08*	−0.39*	0.16*
LEV	[18]																		−0.23*	−0.42*	0.11*	0.02*	−0.11*	0.09*
TQ	[19]																			0.37*	−0.01*	0.18*	0.11*	0.10*
CASH	[20]																				−0.19*	0.14*	0.31*	−0.07*
CAPEX	[21]																					−0.01	0.01*	0.04*
TURN	[22]																						0.25*	0.12*
VOL	[23]																							−0.21*
ANALYST	[24]

This table reports the Pearson correlations among variables. For further details on variable measurement and data sources, see Table 2. All time-varying variables are winsorized at the top and bottom 1% to control for outliers. * indicates that the correlation is significant at least at the 5% level (two-tailed test). For further details on variable measurement and data sources, see Table 2.

4. The effects of societal trust on information timeliness

4.1. The direct effects of societal trust on information timeliness

To examine the effects of societal trust on timeliness, we estimate the regression model shown in Equation (3). The dependent variable is the balance between timeliness of good news and timeliness of bad news, as measured by IOBIT. The main independent variable of interest is societal trust of a country (TRUST). We control for firm- and country-level characteristics as discussed in Section 3.3. Because we have time-invariant variables in our analysis, which prevent us from controlling firm and country fixed effects, we follow the estimation approach of Pevzner, Xie, and Xin (2015), Kanagaretnam et al. (2018) and Guan et al. (2020), to include year and industry fixed effects and address firm-level fixed effects by clustering standard error at firm level. The main model specification is shown as follows: (3) $\mathrm{IOBI}{T}_{i,t}=\alpha +\mathrm{\beta TRUS}{T}_{c,t}+\mathrm{\phi CONTROLS}+\mathrm{YEAR}\mathrm{\_FE}+\mathrm{INDUSTRY}\mathrm{\_FE}+{\epsilon }_{i,t}$(3) where $i$, $c$ and $t$ stand for firm, country and year respectively. $\mathrm{CONTROLS}$ includes firm- and country-level control variables.¹² $\mathrm{YEAR}\mathrm{\_FE}$ stands for year fixed effects. $\mathrm{INDUSTRY}\mathrm{\_FE}$ stands for industry fixed effects. Based on Hypothesis 1, we expect higher societal trust to be associated with less optimistically biased information timeliness, and thus we expect β to be positive. To gain better insights, we also examine the effects of societal trust on timeliness of good news (TGOOD) and timeliness of bad news (TBAD) respectively.

Table 5 presents the results of estimating Equation (3) and robustness checks. Model 1 shows that TRUST is significantly and positively related to IOBIT at the 1% level (β = 0.055, p < 0.01). This finding indicates that societal trust mitigates optimistically biased information timeliness. Hypothesis 1 is thus supported. The effect is also economically significant: all else being equal, a one standard deviation increase in TRUST increases the IOBIT by 0.007.¹³ This is equal to an increase of 0.69% in IOBIT for an average firm with a mean IOBIT of 1.014. In Model 2, we control for country fixed effects which are not captured by the country-specific characteristics as included in the baseline regression. To avoid multicollinearity, we exclude time-invariant country-level variables including IND and UAI. In Model 3, we estimate standard errors with firm and year clustering because there could be both within-firm autocorrelation and within-year autocorrelation in the errors. Given the short period of sample periods (i.e. 20 years), we adopted the approach used by Eun, Wang, and Xiao (2015), which is based on Cameron, Gelbach, and Miller (2008, 2011). Specifically, we estimate t values with firm clustering and year clustering bootstrapping for each independent variable in the baseline model. In Model 4, we re-estimate our baseline model after removing firms from China, Japan and the USA, which contribute more than 60% of the total firm-year observations to our sample. In Model 5, we use an alternative trust measure, which is developed by Falk et al. (2018). The measure is estimated based on responses to the survey question: ‘I assume that people have only the best intentions’. To establish causality, we estimate the baseline model using a propensity score matched sample in Model 6. We match a firm from a high societal trust country with a firm from a low societal trust country, based on firm-level characteristics including SIZE, EM, CLOSE, PROFIT, LEV, TQ, CASH, CAPEX, TURN, VOL and ANALYST.¹⁴ Our propensity score model uses one to one matching with a radius/caliper of 0.1. The matching process yields a sample of 11,870 firm-year observations with 5935 observations from high trust countries and 5935 observations from low trust countries. In Model 7, we use 1-year lagged explanatory variables instead of contemporaneous ones in order to alleviate potential forward-looking bias. In the above robustness checks, we continue to find that TRUST is significantly positively related to IOBIT, which is consistent with the result of the baseline model.

Table 5. Societal trust and imbalanced optimistically biased information timeliness.

	Imbalanced optimistically biased information timeliness (IOBIT)
	Baseline model	Country fixed effects	Firm and year clustering	Excluding Chinese, Japanese & U.S. firms	Alternative trust measure	PSM Sample	One-year lagged explanatory variables
Dependent variable:	(1)	(2)	(3)	(4)	(5)	(6)	(7)
TRUST	0.055***	0.041***	0.055***	0.022***	0.021***	0.047***	0.041***
	(10.30)	(4.37)	(5.71)	(3.41)	(5.78)	(4.26)	(6.60)
RELIG	0.065***	0.028**	0.065**	0.042***	0.028***	0.011*	0.021**
	(7.59)	(2.23)	(1.98)	(3.89)	(3.93)	(1.83)	(2.19)
IND	0.011***		0.011	0.026***	0.007*	0.024***	−0.003
	(4.01)		(0.95)	(7.22)	(1.93)	(3.53)	(−0.97)
UAI	0.004		0.004	0.028***	0.001	0.026***	−0.004
	(1.23)		(0.31)	(7.44)	(0.22)	(2.95)	(−1.19)
CGRF	0.027***	0.028***	0.027	−0.010**	0.027***	−0.006	0.013***
	(7.90)	(7.78)	(1.21)	(−2.18)	(7.67)	(−0.73)	(3.23)
IPS	−0.002	0.111***	−0.002	−0.009	0.016**	0.029**	0.047***
	(−0.28)	(5.95)	(−0.09)	(−1.07)	(2.18)	(2.17)	(6.96)
INFOR	−0.022***	−0.119***	−0.022*	0.012*	−0.020***	−0.024**	−0.048***
	(−4.18)	(−11.78)	(−1.78)	(1.85)	(−3.33)	(−2.15)	(−7.52)
STOCK	0.001***	−0.002	0.001	0.003***	−0.007***	0.005	0.001**
	(3.34)	(−1.41)	(1.57)	(9.45)	(−3.19)	(1.31)	(2.02)
INFL	0.001	0.001	0.001	−0.001	0.001	−0.002***	0.002***
	(1.18)	(1.35)	(0.21)	(−0.66)	(0.12)	(−3.28)	(6.27)
GDPG	0.063*	0.085**	0.063	0.235***	0.130***	0.458***	−0.055
	(1.80)	(2.07)	(0.27)	(5.45)	(3.39)	(5.36)	(−1.35)
SIZE	−0.006***	−0.007***	−0.006***	−0.010***	−0.006***	−0.003***	−0.001**
	(−16.20)	(−16.54)	(−8.79)	(−16.51)	(−14.44)	(−3.33)	(−2.12)
EM	−0.002***	−0.002***	−0.002**	0.003	−0.002***	−0.012***	−0.001
	(−3.38)	(−3.12)	(−2.22)	(0.59)	(−3.08)	(−3.33)	(−0.12)
CLOSE	−0.006***	−0.004*	−0.006	−0.001	−0.008***	−0.013**	−0.004*
	(−2.78)	(−1.91)	(−1.24)	(−0.09)	(−3.63)	(−2.07)	(−1.78)
PROFIT	−0.038***	−0.037***	−0.038***	−0.020***	−0.041***	0.034*	−0.001
	(−13.19)	(−12.58)	(−7.08)	(−3.60)	(−13.97)	(1.95)	(−0.21)
LEV	−0.001	0.001	−0.001	−0.007*	0.001	0.007	−0.005*
	(−0.10)	(0.12)	(−0.06)	(−1.67)	(0.46)	(0.85)	(−1.74)
TQ	−0.008***	−0.008***	−0.008***	−0.009***	−0.007***	−0.012***	−0.001***
	(−20.65)	(−20.60)	(−7.21)	(−12.72)	(−20.06)	(−7.51)	(−3.76)
CASH	0.012***	0.013***	0.012*	0.019***	0.011***	0.032***	0.001
	(4.02)	(4.24)	(1.87)	(3.56)	(3.65)	(2.65)	(0.01)
CAPEX	−0.021***	−0.016*	−0.021	−0.036***	−0.017**	−0.056**	−0.028***
	(−2.61)	(−1.93)	(−1.25)	(−3.29)	(−2.08)	(−2.20)	(−2.98)
TURN	−0.001*	−0.001***	−0.001	0.001	−0.001*	−0.001	−0.001
	(−1.77)	(−2.59)	(−0.60)	(0.05)	(−1.69)	(−0.17)	(−0.18)
VOL	−0.618***	−0.595***	−0.618**	−0.712***	−0.610***	−0.078	−0.249***
	(−12.57)	(−12.01)	(−2.25)	(−8.06)	(−12.09)	(−0.40)	(−4.39)
ANALYST	0.007***	0.008***	0.007***	0.008***	0.007***	0.003*	0.002***
	(13.81)	(14.37)	(5.83)	(11.09)	(12.30)	(1.77)	(3.34)
Number of observations	118,553	118,553	118,553	47,020	109,778	11,870	88,748
Country fixed effects	No	Yes	No	No	No	No	No
Industry fixed effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Adjusted R-squared	0.137	0.140	0.137	0.151	0.138	0.162	0.125

This table presents the regression results of the effects of societal trust levels on information timeliness. The dependent variable is the imbalanced optimistically biased information timeliness (IOBIT). Model 1 reports results of the baseline regression model shown in Equation (3). In Model 2, we control for country fixed effects. In Model 3, we report t-statistics of the regression coefficients based on the standard errors estimated with firm clustering and year clustering bootstrapping with 1000 iterations. Model 4 presents the results excluding Chinese, Japanese and U.S. firms. In Model 5, we use an alternative measure of societal trust. In Model 6, we estimate the baseline regression in a propensity score matched (PSM) sample. In Model 7, we use 1-year lagged explanatory variables. All continuous variables are winsorized at the top and bottom 1% to control for outliers. Except for Model 3, the estimation method is ordinary least squares (OLS) with standard errors clustered by firms. The t-statistics are reported in parentheses. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively (two-tailed tests). For further details on variable measurement and data sources, see Table 2.

With respect to control variables, we find that high levels of religiosity (RELIG) mitigate optimistically biased information timeliness (IOBIT) in all regression models. The coefficients of individualism (IND) and GDP growth (GDPG) are significant in most regression models. At firm level, we find that IOBIT is significantly and negatively related to firm size (SIZE) and Tobin’s q (TQ), while significantly and positively related to cash holdings (CASH) and analyst following (ANALYST). Earnings management (EM), profitability (PROFIT), capital expenditure (CAPEX) and stock return volatility (VOL) also help to explain the variation of IOBIT in most models. The results are generally in line with previously documented evidences (Beekes et al. 2016; Beekes and Brown 2006; Callen and Fang 2015; Zhang et al. 2019; Zhang et al. 2023).

We also estimate the baseline model using timeliness of goodness (TGOOD) and timeliness of bad news (TBAD) as dependent variables respectively. As shown in Models 1–2 of Table 6, TRUST is significantly and positively related to TGOOD. Models 3–4 show that TRUST is significantly and negatively related to TBAD. These findings suggest that in more trusting countries positive value-relevant news is reflected in share price in a less timely manner, and negative value-relevant news is incorporated into share price in a more timely manner. The combined effects on timeliness of good news and bad news result in a less optimistically biased information timeliness.

Table 6. Societal trust, timeliness of good news, and timeliness of bad news.

	TGOOD		TBAD
Dependent variable:	(1)	(2)	(3)	(4)
TRUST	0.022***	0.017***	−0.008***	−0.007**
	(8.17)	(4.80)	(−3.31)	(−2.21)
RELIG	0.019***	0.011***	−0.014***	−0.004**
	(5.04)	(2.95)	(−3.92)	(−2.34)
IND	0.018***		0.015***
	(13.45)		(11.21)
UAI	−0.001		−0.002
	(−0.70)		(−1.13)
CGRF	0.006***	0.005***	−0.007***	−0.008***
	(4.08)	(3.26)	(−4.30)	(−4.72)
IPS	−0.021***	−0.032***	−0.015***	−0.083***
	(−7.60)	(−3.58)	(−5.62)	(−9.24)
INFOR	0.008***	0.002	0.016***	0.057***
	(3.21)	(0.50)	(6.39)	(11.91)
STOCK	0.001***	−0.003***	0.001***	−0.002***
	(6.93)	(−4.38)	(3.73)	(−3.74)
INFL	−0.001***	−0.001*	−0.001**	−0.001
	(−3.17)	(−1.83)	(−2.03)	(−0.96)
GDPG	0.086***	0.131***	0.055***	0.096***
	(5.02)	(6.40)	(3.46)	(5.14)
SIZE	−0.006***	−0.006***	−0.003***	−0.003***
	(−32.73)	(−32.69)	(−14.57)	(−14.20)
EM	0.001***	0.001***	0.002***	0.002***
	(3.72)	(3.49)	(6.74)	(6.32)
CLOSE	−0.006***	−0.005***	−0.003***	−0.003***
	(−5.56)	(−4.95)	(−3.00)	(−3.19)
PROFIT	−0.046***	−0.044***	−0.025***	−0.025***
	(−31.83)	(−30.77)	(−17.44)	(−17.09)
LEV	0.010***	0.010***	0.011***	0.010***
	(8.17)	(7.44)	(9.06)	(8.05)
TQ	−0.002***	−0.003***	0.001***	0.001***
	(−13.90)	(−14.29)	(8.02)	(7.66)
CASH	0.003**	0.003**	−0.002	−0.002*
	(2.33)	(2.18)	(−1.40)	(−1.82)
CAPEX	0.019***	0.022***	0.029***	0.030***
	(4.67)	(5.47)	(7.56)	(7.67)
TURN	−0.002***	−0.002***	−0.002***	−0.002***
	(−14.15)	(−14.87)	(−13.47)	(−13.30)
VOL	−1.344***	−1.333***	−1.06***	−1.058***
	(−52.94)	(−51.81)	(−42.80)	(−42.21)
ANALYST	0.005***	0.006***	0.002***	0.002***
	(21.21)	(20.97)	(6.71)	(6.02)
Number of observations	118,553	118,553	118,553	118,553
Country fixed effects	No	Yes	No	Yes
Industry fixed effects	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes
Adjusted R-squared	0.165	0.167	0.168	0.170

This table presents the regression results of the effects of societal trust on timeliness of good news (TGOOD) and timeliness of bad news (TBAD). Models 1–2 report results of TGOOD. Models 3–4 present results of TBAD. Country fixed effects are included in Model 2 and Model 4. All continuous variables are winsorized at the top and bottom 1% to control for outliers. The estimation method is ordinary least squares (OLS) with standard errors clustered by firms. The t-statistics are reported in parentheses. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively (two-tailed tests). For further details on variable measurement and data sources, see Table 2.

4.2. Robustness checks and additional tests

The OLS estimation prevents us from addressing the endogeneity concern due to omitted explanatory variables because it is difficult to control for all possible variables that could affect societal trust. If the omitted variable is also associated with information timeliness, then our documented results on the association between societal trust and information timeliness will be biased. In the baseline models, we have included controls for country-level institutional and cultural variables, such as investor protection standards and religiosity, and thus it is less likely to find an omitted country-level variable that explains both country-level societal trust and information timeliness in our analyses. Nevertheless, we attempt to address potential omitted correlated variable concerns by employing instrumental variable estimation. The IV estimation requires proper selection of instrumental variables that are correlated with the independent variable of interest while being uncorrelated with the error term (Kennedy 2003, 159). We use an index of ethnolinguistic fractionalization (ENTHIC) in a country as an instrument. The ENTHIC index measures the probability that two people drawn at random from a country’s population will not belong to the same ethnolinguistic group. A higher ENTHIC index indicates a more fragmented country. Leigh (2006) argues that ethnic diversity is associated with lower levels of trust because diverse communities find it more challenging to enforce a system of social norms and hence are less likely to trust one another. Ethnicity and diversity are invariant for a country over long periods and hence more likely to be exogenous to economic variables (Guiso, Sapienza, and Zingales 2006; Mauro 1995). Hence we could assume that ethnolinguistic heterogeneity, as a proxy for ethnic diversity, is correlated with societal trust and has no apparent relation to information timeliness. Data on the ENTHIC index is collected from Mauro (1995). We use the Hansen J test to confirm the validity of instrument: the Hansen’s J statistic is 0, indicating that the instrument is valid. The regression coefficients are estimated by two-stage least square regressions. As shown in Model 1 of Table 7, ENTHIC is significantly and negatively related to TRUST, which suggests that ethnic diversity reduces societal trust. Regarding the effects of TRUST on IOBIT, our inference remains unchanged as shown in Model 2 of Table 7.¹⁵

Table 7. Two-stage least square regressions of imbalanced optimistically biased information timeliness (IOBIT) on trust.

	First stage dependent variable: TRUST	Second stage dependent variable: IOBIT
	(1)	(2)
TRUST		0.537***
		(4.32)
ENTHIC	−0.001***
	(−20.73)
RELIG	0.456***	0.241***
	(39.22)	(3.95)
IND	0.094***	−0.029**
	(55.54)	(−2.47)
UAI	−0.050***	0.021***
	(−25.45)	(3.30)
CGRF	−0.055***	0.052***
	(−33.69)	(6.55)
IPS	0.280***	−0.139***
	(60.51)	(−3.78)
INFOR	−0.069***	0.003
	(−22.36)	(0.28)
STOCK	0.002***	−0.001
	(13.00)	(−0.03)
INFL	−0.002***	−0.001
	(−12.35)	(−0.02)
GDPG	0.198***	−0.018
	(10.97)	(−0.41)
SIZE	−0.001***	−0.007***
	(−3.49)	(−13.46)
EM	0.003***	−0.003***
	(11.04)	(−4.39)
CLOSE	−0.012***	0.005*
	(−11.33)	(1.89)
PROFIT	0.005***	−0.041***
	(3.52)	(−13.10)
LEV	0.001	−0.001
	(0.78)	(−0.17)
TQ	−0.001***	−0.006***
	(−7.86)	(−12.16)
CASH	−0.002	0.009***
	(−1.44)	(2.76)
CAPEX	0.027***	−0.043***
	(6.44)	(−4.35)
TURN	−0.001**	0.001***
	(−2.11)	(3.25)
VOL	0.538***	−0.932***
	(22.45)	(−11.04)
ANALYST	0.004***	0.006***
	(14.64)	(7.29)
Number of observations	105,443	105,443
Hansen’s J statistic	0.00
Industry fixed effects	Yes	Yes
Year fixed effects	Yes	Yes
Adjusted R-squared	0.570	0.141

This table reports the results from 2SLS regressions of Equation (3), where we estimate a fitted value of societal trust in the first stage, and then estimate Equation (3) in the second stage with the fitted value of societal trust as the key independent variable. The instrument is the ethnolinguistic fractionalization (ENTHIC) in a country. We used the Hansen J-test to check the validity of instrument used in our estimations, where a rejection of the null hypothesis casts doubts on validity. All continuous variables are winsorized at the top and bottom 1% to control for outliers. Standard errors are clustered by firms. The t-statistics are reported in parentheses. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively (two-tailed tests). For further details on variable measurement and data sources, see Table 2.

Our panel data spans a period of 20 years, which witnessed many policy changes and economic events. To check the effect of societal trust with alleviated impact from exogenous shocks, we employ the tests on two subsamples. The first subsample is between 2002 and 2006, which is after the dot-com bubble burst in 2001 and before the 2008 subprime mortgage crisis. The second subsample is between 2010 and 2019, which is after the subprime mortgage crisis. Panel A of Table 8 reports the regression results for the two subsamples. We continue to find that societal trust improves overall balance in timeliness of good news and bad news.

Table 8. Societal trust and information timeliness: sub-period tests and the impact of 2007–2008 financial crisis.

Panel A: Sub-period tests
	2002–2006		2010–2019
	(1)	(2)	(3)	(4)
TRUST	0.058***	0.140**	0.120***	0.117***
	(5.03)	(2.10)	(12.90)	(3.54)
Number of observations	25,139	25,139	66,429	66,429
Control variables	Yes	Yes	Yes	Yes
Country fixed effects	No	Yes	No	Yes
Industry fixed effects	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes
Adjusted R-squared	0.129	0.147	0.135	0.141
Panel B: Impact of 2007–2008 financial crisis
	(1)	(2)
TRUST	0.055***	0.040***
	(9.68)	(3.33)
CRISIS	−0.027**	−0.052***
	(−2.47)	(−9.82)
TRUST × CRISIS	0.022***	0.012***
	(3.98)	(3.09)
Number of observations	118,553	118,553
Control variables	Yes	Yes
Country fixed effects	No	Yes
Industry fixed effects	Yes	Yes
Year fixed effects	Yes	Yes
Adjusted R-squared	0.137	0.140

Models 1–2 of Panel A report the results using the subsample of 2002–2006. Models 3–4 of Panel A report the results using the subsample of 2010–2019. Models 1 and 3 of Panel A estimate the baseline regression model without controlling for country fixed effects. Models 2 and 4 of Panel A include country fixed effects. Models 1–2 of Panel B present the interactive effects between societal trust and a financial crisis dummy variable. Model 1 of Panel B does not include country fixed effects. Model 2 of Panel B includes country fixed effects. CRISIS is a dummy variable that is equal to 1 if year is either 2007 or 2008, and 0 otherwise. All continuous variables are winsorized at the top and bottom 1% to control for outliers. The estimation method is ordinary least squares (OLS) with standard errors clustered by firms. The t-statistics are reported in parentheses. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively (two-tailed tests). For further details on variable measurement and data sources, see Table 2.

To further confirm the cross-sectional explanatory power of societal trust, we test the interactive effects of trust and the 2007–2008 subprime mortgage crisis. Societal trust could become more important in a financial crisis. A crisis could force investors to recognize and take account of weaknesses in corporate governance (Mitton 2002). Rajan and Zingales (1998) also argue that investors tend to ignore weaknesses of firms while the economy performs well, but quickly pull out once a crisis begins. Thus, in a financial crisis, managers could be forced to take bad news more seriously and get them released in a timelier manner in high trust countries. Regarding the treatment of good news during a financial crisis, managers in high trust countries could be more conservative in taking positive news and make extra efforts to verify them before release. To test the interactive effects of trust and financial crisis, we create a dummy variable (CRISIS) that is equal to 1 if year is either 2007 or 2008, and 0 otherwise. Panel B of Table 8 reports the results of regressions with interaction terms. We find that the positive effects of TRUST on IOBIT are strengthened during the financial crisis period, which are consistent with our predictions.¹⁶

4.3. The interactive effects of societal trust on information timeliness

In this section, we test Hypotheses H2a to H2e and examine whether the effect of trust on imbalanced optimistically biased information timeliness varies with moderating variables including market development, investor protection standards (IPS), religiosity (RELIG), corporate governance quality (CGQ) and managerial ability.

Table 9 reports the results of interactive effects of trust and three country-level characteristics including market development, investor protection standards (IPS) and religiosity (RELIG). Panel A of Table 9 presents the results from the developed markets countries and emerging markets countries subsamples. The market classification according to the MSCI World Index and the MSCI Emerging Markets Index is shown in Table 1. The result suggests that the effect of trust on IOBIT is more pronounced in emerging markets countries (0.093, p < 0.01) than in developed markets countries (0.007, p < 0.1). Panel B of Table 9 shows the results from the high and low investor protection standards (IPS) subsamples. A firm is classified in the ‘High’ investor protection standards subsample if the value of its country’s IPS index is above the sample mean. A firm is classified in the ‘Low’ investor protection standards subsample if the value of its country’s IPS index is below the sample mean. The result suggests that the effect of trust on IOBIT is more pronounced in the low investor protection standards subsample (0.102, p < 0.01) than in the high investor protection standards subsample (0.017, p < 0.1). Panel C of Table 9 provides the results from the high and low religiosity (RELIG) subsamples. A firm is classified in the ‘High’ religiosity subsample if the value of its country’s RELIG index is above the sample mean. A firm is classified in the ‘Low’ religiosity subsample if the value of its country’s RELIG index is below the sample mean. The result suggests that the effect of trust on IOBIT is more pronounced in the low religiosity subsample (0.054, p < 0.01) than in the high religiosity subsample (0.023, p < 0.1). Our hypotheses H2a to H2c are therefore supported.

Table 9. Interactive effects with levels of development, investor protection standards, and religiosity.

Panel A: Interaction between levels of development and societal trust
	Developed markets countries	Emerging markets countries
	(1)	(2)
TRUST	0.007*	0.093***
	(1.77)	(10.18)
Number of observations	86,877	31,676
Adjusted R-squared	0.129	0.189
Control variables	Yes	Yes
Industry fixed effects	Yes	Yes
Year fixed effects	Yes	Yes
Comparing coefficients of TRUST:	Chi-square = 45.74 p-value = 0.00
Panel B: Interaction between investor protection standards and societal trust
	High IPS	Low IPS
	(1)	(2)
TRUST	0.017*	0.102***
	(1.83)	(11.71)
Number of observations	87,150	31,403
Adjusted R-squared	0.129	0.191
Control variables	Yes	Yes
Industry fixed effects	Yes	Yes
Year fixed effects	Yes	Yes
Comparing coefficients of TRUST:	Chi-square = 40.78, p-value = 0.00
Panel C: Interaction between religiosity and societal trust
	High RELIG	Low RELIG
	(1)	(2)
TRUST	0.023*	0.054***
	(1.77)	(6.38)
Number of observations	62,154	56,399
Adjusted R-squared	0.130	0.162
Control variables	Yes	Yes
Industry fixed effects	Yes	Yes
Year fixed effects	Yes	Yes
Comparing coefficients of TRUST:	Chi-square = 11.97, p-value = 0.00

Panel A presents the results from the developed markets countries versus emerging markets countries subsamples. A country is classified as developed or emerging country according to the MSCI World Index and the MSCI Emerging Markets Index. Panel B reports the results of examining the effects of societal trust on IOBIT for different firms based on investor protection standards (IPS). High IPS is the subsample including firm-year observations whose investor protection standards index is above the sample mean. Low IPS is the subsample including firm-year observations whose investor protection standards index is below the sample mean. Panel C reports the results of examining the effects of societal trust on IOBIT for different firms based on religiosity (RELIG). High RELIG is the subsample including firm-year observations whose religiosity ratio is above the sample mean. Low RELIG is the subsample including firm-year observations whose religiosity ratio is below the sample mean. For brevity, we only report the estimates of TRUST. All continuous variables are winsorized at the top and bottom 1%. Standard errors are clustered at the firm level. The t-statistics are reported in parentheses. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively (two-tailed tests).

Table 10 reports the results of interactive effects of trust and firm-level characteristics including corporate governance quality (CGQ) and managerial ability. Panel A of Table 10 presents the results from the high and low corporate governance quality (CGQ) subsamples. To measure corporate governance quality, we adopt the approach of Chung, Elder, and Kim (2010) and construct a firm-level score with 22 underlying governance characteristics.¹⁷ A firm is classified in the ‘High’ corporate governance quality subsample if the value of its average CGQ score is above the sample mean. A firm is classified in the ‘Low’ corporate governance quality subsample if the value of its average CGQ score is below the sample mean. The result suggests that the effect of trust on IOBIT is more pronounced in the low corporate governance quality subsample (0.075, p < 0.01) than in the high corporate governance quality subsample (0.018, p < 0.1). The hypothesis H2d is therefore supported.

Table 10. Interactive effects with corporate governance quality and managerial ability.

Panel A: Interaction between corporate governance quality and societal trust
	High CGQ	Low CGQ
	(1)	(2)
TRUST	0.018*	0.075***
	(1.69)	(8.92)
Number of observations	44,980	45,326
Adjusted R-squared	0.128	0.153
Control variables	Yes	Yes
Industry fixed effects	Yes	Yes
Year fixed effects	Yes	Yes
Comparing coefficients of TRUST:	Chi-square = 18.56, p-value = 0.00
Panel B: Interaction between historical market-adjusted stock returns and societal trust
	High RET	Low RET
	(1)	(2)
TRUST	0.065***	0.038***
	(7.12)	(5.05)
Number of observations	49,088	58,392
Adjusted R-squared	0.144	0.143
Control variables	Yes	Yes
Industry fixed effects	Yes	Yes
Year fixed effects	Yes	Yes
Comparing coefficients of TRUST:	Chi-square = 5.32, p-value = 0.02
Panel C: Interaction between profitability and societal trust
	High PROFIT	Low PROFIT
	(1)	(2)
TRUST	0.060***	0.022**
	(9.43)	(2.27)
Number of observations	74,941	43,612
Adjusted R-squared	0.139	0.138
Control variables	Yes	Yes
Industry fixed effects	Yes	Yes
Year fixed effects	Yes	Yes
Comparing coefficients of TRUST:	Chi-square = 11.69, p-value = 0.00

Panel A reports results of examining the effects of societal trust on IOBIT for different firms based on firm-level corporate governance quality (CGQ). High CGQ is the subsample including firm-year observations whose average corporate governance score is above the sample mean. Low CGQ is the subsample including firm-year observations whose average corporate governance score is below the sample mean. Panel B reports the results of examining the effects of societal trust on IOBIT for different firms based on historical market-adjusted stock return (RET). High RET is the subsample including firm-year observations whose annualized market-adjusted stock return in the past 2 years is above the sample mean. Low RET is the subsample including firm-year observations whose annualized market-adjusted stock return in the past 2 years is below the sample mean. Panel C reports the results of examining effects of societal trust on IOBIT for different firms based on profitability (PROFIT). High PROFIT is the subsample including firm-year observations whose average industry-adjusted profitability in the past 2 years is above the sample mean. Low PROFIT is the subsample including firm-year observations whose average industry-adjusted profitability in the past 2 years is below the sample mean. For brevity, we only report the estimates of TRUST. All continuous variables are winsorized at the top and bottom 1%. Standard errors are clustered at the firm level. The t-statistics are reported in parentheses. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively (two-tailed tests).

Panel B and Panel C of Table 10 show the results from the high and low levels of managerial ability. We calculate annualized market-adjusted stock returns in the past 2 years (RET) and industry-adjusted profitability (PROFIT) as measures of managerial ability (Demerjion, Lev, and McVay 2012). A firm is classified in the ‘High’ managerial ability subsample if the value of its historical stock returns (profitability) is above the sample mean. A firm is classified in the ‘Low’ managerial ability subsample if the value of its historical stock returns (profitability) is below the sample mean. The results suggest that the effect of trust on IOBIT is more pronounced in the high managerial ability subsample than in the low managerial ability subsample. For example, the coefficient of TRUST is significantly higher for the high RET subsample (0.065, p < 0.01) than for the low RET subsample (0.038, p < 0.01), which is supported by the coefficient equality test (i.e. Chi-square = 5.32 and p = 0.02). Our interpretation is that more capable managers have enhanced understanding of the benefit of corporate transparency and are likely to be straightforward in disclosing value-relevant information (including negative news) to the market. The hypothesis H2e is therefore supported.

5. Discussion and conclusion

Based on the analysis of over 118,553 firm-year observations collected from 38 countries between 2000 and 2019, we test the association between societal trust and the timeliness with which value-relevant information is incorporated into stock price. We find that in high trust countries, positive (negative) value-relevant information is impounded into stock price in a less (more) timely manner, leading to a less optimistically biased information timeliness. We interpret the findings as consistent with managers in more trusting environment being forthcoming in the disclosure of bad news and investors being more willing to respond vigorously to the disclosure that is perceived to be more credible. Evidence from 2SLS in which we instrument trust by a country’s ethnic diversity lends strong support to our inference. Further analyses suggest that the relation between trust and IOBIT is more pronounced in emerging countries and in countries with inferior investor protection, and lower levels of religiosity. In contrast, the relation between trust and IOBIT is less pronounced in firms with strong corporate governance and firms led by less capable managers. Taken together, our study highlights the significant role informal institutions (i.e. trust) play in determining the efficiency of capital markets.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The authors confirm that the data supporting the findings of this study are available within the article.

Additional information

Notes on contributors

Qiyu Zhang

Qiyu Zhang is an Assistant Professor in Banking and Finance at University of Sussex Business School. Qiyu obtained his PhD from Trinity College Dublin. Qiyu's research interests include international financial market integration, financial market efficiency, empirical research in corporate finance, and corporate governance. He has published on peer-reviewed journals including Journal of Banking and Finance, Journal of Business Finance and Accounting, Corporate Governance: An International Review, International Review of Financial Analysis, European Journal of Finance, Accounting and Finance, and Emerging Markets Review.

Rong Ding

Rong Ding is a Professor of Accounting at NEOMA Business School. Rong's research expertise is in financial reporting, auditing and market-based tax research, and he has published on peer-reviewed journals including Journal of Accounting and Public Policy, Abacus, Journal of Business Research, Journal of Financial Markets, Journal of Management Accounting Research, Regional Studies, Management Accounting Research, and Journal of International Financial Markets, Institutions and Money, among others. Rong was in the scientific committee of 40th to 42nd European Accounting Association Annual Congress. Rong frequently provides review service for journals such as British Accounting Review, Corporate Governance: an International Review, Journal of Corporate Finance, Journal of Business Ethics, and European Journal of Finance.

Wenhong Ding

Wenhong Ding is an Assistant Professor at Department of Accounting, Control & Legal Affairs in NEOMA Business School. She received her PhD from the Chinese University of Hong Kong in 2019. Her empirical research focuses on several aspects. Firstly, she studies extensive topics in corporate governance, for example, what measures are used to evaluate CEO performance, how geographic dispersion of board members influence their function. Secondly, she has several projects on urban agglomeration and its effects on public firms' accounting disclosure and auditing quality. Finally, she works on firms' interaction with new technologies, such as blockchain technique. She has published papers in internationally reputable journals. She attends regularly academic conferences such as EAA, AAA, MIT Asia.

Wenbin Cao

Wenbin Cao is an Associate Professor of Finance at NEOMA. He obtained his Ph.D. from the Price College of Business at the University of Oklahoma in 2016. He teaches data analytics in finance, managerial economics, corporate finance, and operational risk. His main research interest is in corporate finance. He also diversifies his research in various fields. He has published in the Journal of Economic Dynamics and Control, International Review of Law and Economics, the Journal of Retailing, and Physica A: Statistical Mechanics and its Applications. He has presented his papers at influential conferences, including the China International Conference in Finance, and the Financial Management Association Annual Meetings. He is currently the Head of MSC Financial Markets and Technologies.

Notes

1 According to Guiso, Sapienza, and Zingales (2008), trust reflects the subjective probability individuals attribute to the possibility of being cheated in a society.

2 For example, Beekes et al. (2016) document that better corporate governance is associated with less timely share prices in a sample of 24 OECD countries. The unexpected result could arise from the fact that informal institutions are not controlled in the analysis, the effect of which weakens or even reverses the influence of formal governance metrics.

3 Knack and Keefer (1997) also measure social capital with civic norms based on response to survey questions related to the deemed opportunities of certain behavior in a society such as cheating on taxes or claiming government benefit without being entitled to it. Civic norms measure is also found to be positively associated with growth and investment.

4 Hirshleifer, Lim, and Teoh (2011) investigate the implication of limited investor attention on the association between return and earnings-related information, and find that some investors do not attend to the earnings-related information (i.e. information contained in earnings announcement), and would stick to their outdated expectation on earnings.

5 The countries included in our study are developed and emerging markets defined by the MSCI World Index and the MSCI Emerging Markets Index. The sample countries must also have data on societal trust and other country-level control variables.

6 Ball and Brown (1968) suggest that the annual income report is not a highly timely medium, since most of its content (about 85–90%) is captured by more prompt media. It is therefore justified to calculate the differences between the “final” price $P_0$ and previous prices $P_t$.

7 We follow Beekes et al. (2016) and choose the third quartile as the filter to mitigate undue noise (e.g. due to bid-ask bounce) in identifying the nature of forward-looking information.

8 To facilitate international comparisons, timeliness is measured in calendar time because the number of trading days in a year differs by country. Prices are forward-filled on days when the market was closed (e.g. on weekends and holidays), or when there was no trading in the stock. We set the ending date to be 14 days after the earnings release date because the market may need time to absorb information (Beaver 1968). We follow Beekes et al. (2016) to include the 0.5 adjustment to recognize that the flow of information is reflected in returns over the day. Our results remain unchanged without the adjustment.

9 We consult documents released by local stock exchange regulators for New Zealand, Russia and South Africa. We identify the major board reforms and the year in which they were adopted.

10 In 1998, the Cadbury report and the Greenbury report were brought together to form the Combined Code in the UK.

11 Firm characteristics such as firm size, financial leverage, profitability, growth opportunity, cash holdings and capital expenditure are found to be related to corporate governance quality (Aggarwal et al. 2011; Ferreira, Ferreira, and Raposo 2011). Corporate governance may affect information timeliness (Beekes and Brown 2006; Beekes et al. 2016).

12 We test the contemporaneous effects of explanatory variables on IOBIT. The seminal work of Ball and Brown (1968) finds that annual income reports are not a timely source of earnings-related information, because most of the value-relevant component of earnings (85%–90%) has already been captured by more timely media. According to the Ball and Brown’s study, we think that the information contained in explanatory variables has been largely captured by the market before the information is formally released. Hence these variables can properly explain the gradual incorporation of value-relevant information into the stock price during a year. Prior literature (e.g. Beekes and Brown 2006; Landsman, Maydew, and Thornock 2012; Pevzner, Xie, and Xin 2015; Beekes et al. 2016) also test contemporaneous effect of firm- and country-level time-varying variables on timeliness or other price-related variables.

13 The magnitude of impact of TRUST on IOBIT is calculated as 0.055 (coefficient on TRUST in Table 5) × 0.131 (the sample standard deviation of TRUST in Table 3) ÷ 1.014 (the sample mean of IOBIT in Table 3) = 0.007.

14 We split the 38 sample countries into two subgroups. The high (low) societal trust group includes top (bottom) 19 countries according to their average scores of societal trust.

15 For brevity, we do not report the 2SLS-IV results for timeliness of good news and timeliness of bad news respectively. The findings in Table 6 are robust with the instrumental variable estimation. The results are available upon request.

16 For brevity, we do not report the regression results for timeliness of good news and timeliness of bad news. The results are available upon request.

17 The selected governance items cover major aspects of corporate governance of a firm including audit, board of directors and committees, stock ownership compensation of directors, and provisions in the firm’s charter and bylaws. See Appendix C for details. If a firm meets a characteristic successfully in a given year, it will score 1 point and 0 otherwise. We weight all characteristics equally to obtain total CGQ index for a year. Average CGQ index is then calculated to represent the overall quality of a firm during the sample period.

Appendices

Appendix A

Table

Country	WVS wave 4 (1999–2004)	WVS wave 5 (2005–2009)	WVS wave 6 (2010–2014)	WVS wave 7 (2017–2020)	Average societal trust	Average change in societal trust
Argentina	0.159	0.169	0.232	0.207	0.192	10.9%
Australia	–	0.482	0.544	0.540	0.522	6.1%
Brazil	–	0.092	0.066	0.066	0.075	−14.1%
Canada	0.370	0.421	–	0.495	0.429	15.7%
Chile	0.230	0.124	0.128	0.143	0.156	−10.4%
China	0.545	0.524	0.644	0.654	0.592	6.9%
Colombia	–	0.145	0.041	0.045	0.077	−31.0%
Egypt	0.379	0.184	0.206	0.074	0.211	−34.5%
Finland	–	0.588	–	–	0.588	–
France	–	0.187	–	–	0.187	–
Germany	–	0.341	0.425	0.460	0.409	16.4%
Greece	–	–	–	0.084	0.084	–
Hong Kong	–	0.411	0.483	0.395	0.430	−0.4%
India	0.410	0.233	0.176	–	0.273	−33.8%
Indonesia	0.516	0.425	–	0.052	0.331	−52.7%
Israel	0.235	–	–	–	0.235	–
Italy	–	0.292	–	–	0.292	–
Japan	0.431	0.391	0.388	0.356	0.392	−6.1%
Korea	0.273	0.302	0.297	0.329	0.300	6.6%
Malaysia	–	0.088	0.085	0.196	0.123	63.6%
Mexico	0.218	0.156	0.124	0.103	0.150	−22.0%
Netherlands	–	0.445	0.674	–	0.560	51.5%
New Zealand	–	0.512	0.568	0.595	0.558	7.8%
Norway	–	0.742	–	–	0.742	–
Pakistan	0.308	–	0.239	0.235	0.261	−12.0%
Peru	0.107	0.064	0.083	0.053	0.077	−15.5%
Philippines	0.086	–	0.028	0.053	0.056	10.9%
Poland	–	0.195	0.228	–	0.212	16.9%
Russia	–	0.267	0.292	0.239	0.266	−4.4%
Singapore	0.147	–	0.385	0.340	0.291	75.1%
South Africa	0.131	–	0.236	–	0.184	29.1%
Spain	0.340	0.199	0.195	–	0.245	−21.7%
Sweden	0.663	0.680	0.648	–	0.664	−1.1%
Switzerland	–	0.511	–	–	0.511	–
Thailand	–	0.415	0.326	0.314	0.352	−12.6%
Turkey	0.193	0.048	0.124	0.143	0.127	32.8%
United Kingdom	–	0.304	–	–	0.304	–
United States	0.363	0.396	0.382	0.397	0.385	3.16%

This table shows the societal trust score by the WVS Wave and country.

Appendix B.

Estimation of the accrual-based earnings management measure by Dechow, Sloan, and Sweeney (1995)

The Jones (1991) model is first estimated to obtain estimated values of ${\beta }_1$, ${\beta }_2$ and ${\beta }_3$ using other firms in the same industry. The model is shown as follows: (B1) $\frac{T{A}_{i,t}}{A_{i,t-1}}={\beta }_1\frac{1}{A_{i,t-1}}+{\beta }_2\frac{\mathrm{\Delta }\mathrm{RE}{V}_{i,t}}{A_{i,t-1}}+{\beta }_3\frac{\mathrm{PP}{E}_{i,t}}{A_{i,t-1}}+{\epsilon }_{i,t}$(B1) where $\text{T}{\text{A}}_{\text{i},\text{t}}$ denotes total accruals for firm i in year t; ${\text{A}}_{\text{i},\text{t}-1}$ denotes total assets for firm i in year t–1; $\mathrm{\Delta }\text{RE}{\text{V}}_{\text{i},\text{t}}$ denotes change in revenue for firm i in year t; $\text{PP}{\text{E}}_{\text{i},\text{t}}$ denotes property, plant, and equipment for firm i in year t. $\text{T}{\text{A}}_{\text{i},\text{t}}$ is calculated as follows: (B2) $T{A}_{i,t}=(\mathrm{\Delta C}{A}_{i,t}-\mathrm{\Delta CAS}{H}_{i,t})-(\mathrm{\Delta C}{L}_{i,t}-\mathrm{\Delta ST}{D}_{i,t})-\mathrm{DE}{P}_{i,t}$(B2) where $\mathrm{\Delta }\text{C}{\text{A}}_{\text{i},\text{t}}$ denotes the change in current assets for firm i in year t; $\mathrm{\Delta }\text{CAS}{\text{H}}_{\text{i},\text{t}}$ denotes the change in cash and cash equivalents for firm i in year t; $\mathrm{\Delta }\text{C}{\text{L}}_{\text{i},\text{t}}$ denotes the change in current liabilities for firm i in year t; $\mathrm{\Delta }\text{ST}{\text{D}}_{\text{i},\text{t}}$ denotes the change in short-term debt for firm i in year t; $\text{DE}{\text{P}}_{\text{i},\text{t}}$ denotes the depreciation and amortization for firm i in year t.

Dechow, Sloan, and Sweeney (1995) propose the modified Jones model to calculate normal accruals as follows: (B3) $\frac{\mathrm{NT}{A}_{i,t}}{A_{i,t-1}}=\widehat{{\beta }_1}\frac{1}{A_{i,t-1}}+\widehat{{\beta }_2}\frac{\mathrm{\Delta }\mathrm{RE}{V}_{i,t}-\mathrm{\Delta RE}{C}_{i,t}}{A_{i,t-1}}+\widehat{{\beta }_3}\frac{\mathrm{PP}{E}_{i,t}}{A_{i,t-1}}+{\epsilon }_{i,t}$(B3) where $\frac{\text{NT}{\text{A}}_{\text{i},\text{t}}}{{\text{A}}_{\text{i},\text{t}-1}}$ denotes fitted normal accruals relative to total assets for firm i in year t; $\mathrm{\Delta }\text{RE}{\text{C}}_{\text{i},\text{t}}$ denotes the change in receivables for firm i in year t; $\widehat{{\beta }_1}$, $\widehat{{\beta }_2}$ and $\widehat{{\beta }_3}$ represent estimated values from the model (B1).

The accrual-based earnings management measure is calculated as the difference between total accruals relative to total assets and the fitted value of normal accruals relative to total assets from model (B3).

Appendix C.

Corporate governance index construction method.

Table

Item no.	Aspect of CG	Thomson data item	Our construction
1	Audit	Audit Committee Independence	1 if audit committee independence is true, and 0 otherwise
2	Board	Strictly Independent Board Members	1 if strictly independent board members are more than 50% of the board directors, and 0 otherwise
3	Board	Nomination Committee Independence	1 if nomination committee independence is true, and 0 otherwise
4	Board	Compensation Committee Independence	1 if compensation committee is compromised solely of independent outside directors, and 0 otherwise
5	Board	Committee Meetings Attendance Average	1 if committee meeting attendance average is >0 (i.e. meets at least once during the year), and 0 otherwise
6	Board	Staggered Board Structure	1 If staggered board structure is false, and 0 otherwise
7	Board	Board Size	1 if board size > 5 and < 16, and 0 otherwise
8	Board	Elimination of Cumulative Voting Rights	1 if elimination of cumulative voting rights is false, and 0 otherwise
9	Board	CEO Board Member	1 if CEO serves on no more than two additional boards of other public companies, and 0 otherwise
10	Board	Chairman is ex-CEO	1 if chairman is not ex-CEO, and 0 otherwise
11	Board	CEO-Chairman Separation	1 if CEO and Chairman is separated, and 0 otherwise
12	Board	Disclosure of Board Guidelines	1 if any of the following policies is available publicly, and 0 otherwise
			• Board Independence
			• Board Diversity
			• Board Experience
			• Executive Compensation Performance
			• Executive Compensation ESG Performance
			• Executive Retention
13	Charter	Poison Pill	1 if there is no poison pill provision, and 0 otherwise
14	Charter	Supermajority Vote Requirement	1 if Supermajority Vote Requirement is false, and 0 otherwise
15	Charter	Shareholder Approval Significant Transactions	1 if Shareholder Approval Significant Transactions is true, and 0 otherwise
16	Charter	Written Consent Requirements	1 if Written Consent Requirements is true, and 0 otherwise
17	Charter	Limited Shareholder Rights to Call Meetings	1 if Limited Shareholder Rights to Call Meetings is false, and 0 otherwise
18	Charter	Unlimited Authorized Capital or Blank Check	1 if Unlimited Authorized Capital or Blank Check is false, and 0 otherwise
19	Compensation	Board Member Long Term Compensation Incentives	1 if Board Member Long Term Compensation Incentives > 0, and 0 otherwise
20	Ownership	Executive Compensation Long Term Objectives	1 if Executive Compensation Long Term Objectives is true, and 0 otherwise
21	Ownership	Shareholders Approval Stock Compensation Plan	1 if Shareholders Approval Stock Compensation Plan is true, and 0 otherwise
22	Anti-Takeover Devices	Anti-Takeover Devices	1 if there is no anti-takeover device, and 0 otherwise

This table shows the construction method for the corporate governance quality index. Following Chung, Elder, and Kim (2010), we adopt 22 CG standards relating to financial and operating transparency. The CG standards are taken from data compiled by Thomson Eikon.