Changes in the value relevance of accounting information over time: Evidence from the emerging market of China

Journal of Contemporary Accounting & Economics 9 (2013) 123–135

Contents lists available at SciVerse ScienceDirect

Journal of Contemporary Accounting & Economics journal homepage: www.elsevier.com/locate/jcae

Changes in the value relevance of accounting information over time: Evidence from the emerging market of China Kevin C.K. Lam a,⇑, Heibatollah Sami b, Haiyan Zhou c a

School of Accountancy, The Chinese University of Hong Kong, Hong Kong Department of Accounting, Lehigh University, Philadelphia, PA 19122, United States c Department of Accounting and Business Law, College of Business Administration, The University of Texas – Pan American, United States b

a r t i c l e

i n f o

Article history: Received 18 August 2011 Revised 13 May 2013 Accepted 5 June 2013 Available online 11 June 2013 Keywords: Value relevance Accounting information Firm characteristics Capital markets Emerging market China

a b s t r a c t We investigate the changes in the value relevance of accounting information among Chinese firms over the past two decades, during which accounting reforms are launched to provide decision makers with increased disclosure and higher quality financial information. We also investigate the factors that differentiate firms showing significant value relevance improvement from firms showing little improvement. We find increases in the value relevance of some financial variables and decreases in others, which suggests that accounting numbers help to explain the pricing process of stock shares although at different levels. In addition, we find that value relevance improvements are more pronounced for smaller firms, firms with lower growth rates, and those with greater asset tangibility. We also document that value relevance improvements are generally lower in an exuberant stock market. These results have implications for a variety of information users and policy makers in emerging countries which are reforming their accounting systems. Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction This paper investigates the changes in the value relevance of accounting information among Chinese firms over the past two decades. Value relevance is defined as the informativeness of financial statements. The higher the value relevance, the more financial statements can be relied upon to make investment decisions and thus the greater the association between financial statement items and firm stock-share prices or returns (Francis and Schipper, 1999; Sami and Zhou, 2004). The purpose of this paper is to provide greater insight into changes in value relevance and the factors that influence value relevance against a background of accounting reforms in a broad sense, rather than charting the causal relationship between accounting reforms and the value relevance of accounting information.1 Our study is primarily motivated by the unique environment of the Chinese emerging market, in which accounting standards, regulations, and practices have experienced revolutionary change over the past two decades.2 Another motivation comes from the mixed results documented in the literature on the value relevance of accounting information based on U.S. data. For instance, whereas some studies find increases in the value relevance of accounting information over time (Collins et al.,

⇑ Corresponding author. E-mail addresses: [email protected] (K.C.K. Lam), [email protected] (H. Sami), [email protected] (H. Zhou). We owe thanks to an anonymous reviewer for addressing these issues. 2 We use a broad definition of accounting and regulatory reforms that goes beyond the introduction of new accounting standards, rules, and regulations by themselves. This broad definition considers implementation issues such as (1) changes in auditing standards, profession, and practice; (2) changes in human capital, in particular the qualifications of accounting personnel; and (3) reforms in the judiciary system and administrative sanctions on the management of public companies and their auditors. 1

1815-5669/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jcae.2013.06.001

124

K.C.K. Lam et al. / Journal of Contemporary Accounting & Economics 9 (2013) 123–135

1997; Landsman and Maydew, 2002), others show that value relevance has in fact been decreasing over time (Lev, 1989; Brown et al., 1999) or changes are in different directions when different accounting items are used (Francis and Schipper, 1999). To shed light on this issue, we use the emerging market of China as an experimental setting, where the majority of publicly listed firms are traditional manufacturing and merchandising firms and where the role of new-economy firms is more limited than in developed markets. We address the topic by studying two research questions. The first is whether, with the accounting and regulatory reforms that have occurred in China in the past two decades, the value relevance of financial information has increased for investors. Our principal measure of value relevance is the change in the ability of the key accounting variables, such as assets, liabilities, and earnings, to explain stock returns and market value. Given the accounting and regulatory reforms of the past two decades, we expect the power of the key accounting variables to explain stock returns and market value to improve over time. The second issue that we investigate is whether factors such as firm size, growth, monitoring by long-term debt holders, cross-listing, state-ownership, and auditor characteristics affect changes in the value relevance of financial information. We use changes in the absolute value of the residuals from four evaluation models to measure improvements in the value relevance of accounting information. Consistent with the literature, we find that earnings level and earnings changes are positively associated with the pricing process in most years. Interestingly, the balance sheet relation does not behave as predicted until after 2002, when total assets are positively and liabilities negatively associated with firm value. Further analysis using the lagged book value of equity as an alternative deflator rather than outstanding shares indicates that the unexpected signs in the earlier years may be the result of dramatic changes in outstanding shares. To test the changes in value relevance at different stages of the accounting and regulatory reforms, we group the regressions into three periods. In all cases, the regressions for the later period are closer to the true models than those for the earlier periods. Overall, the results suggest that the accounting numbers do help to explain the pricing process of stock shares in the emerging market of China, which is consistent with earlier studies (e.g., Sami and Zhou, 2004). We do not find evidence that the value relevance of the models changes uniformly in the same fashion over time. On the contrary, we find that the value relevance of the balance sheet relation increases whereas that of the book value and earnings relation decreases over the period studied. Moreover, we do not find a significant change in the pattern of the value relevance of the earnings relation and the accruals and cash flow relation. This result is consistent with the notion that the usefulness of specific accounting information varies with its necessity at different stages of economic development. Finally, using the change in the absolute value of the residual as a measure of improvements in value relevance, we find that smaller firms, firms with lower growth rates, and those with greater asset tangibility experience more pronounced value relevance improvements. Moreover, the value relevance of accounting information is generally lower when the stock market is more optimistic. The remainder of this paper is organized as follows. Section 2 discusses the background of financial disclosure practice and the institutional setting. A literature review and research questions are presented in Section 3. Section 4 discusses the sample, variables, and empirical method. Section 5 discusses the results and the empirical analyses. Finally, Section 6 presents our conclusions. 2. Background In 1992, China implemented the Accounting Standards for Business Enterprises, which represented the first step in a series of accounting and regulatory reforms in the world’s fastest growing economy. The reforms transformed the Chinese accounting system from a Soviet-style central-planning-oriented system to one that aimed to harmonize with international practices. In the period since 1992, improvements in accounting standards have necessitated the parallel development of auditing standards that are, in various ways, modeled after the standards promulgated by the International Federation of Accountants. Accounting reforms in China went through three stages (Chen et al., 2002). The first stage was from 1992 to 1997, during which all listed A-share firms were required to follow the Experimental Accounting System for Joint Stock Limited Enterprises and the Accounting Standard for Business Enterprises issued in 1992 by the Ministry of Finance (MOF), and the accounting regulations issued by the Chinese Securities Regulatory Commission (CSRC). The second stage was from 1998 to 2000, characterized by the adoption of the Accounting System for Joint Stock Limited Enterprises in 1998. During this period, A-share firms were required to follow the Chinese Accounting Standards (CAS) issued by the MOF and the Accounting Law issued by the State Council in 1995. Financial statements were also more commonly audited by independent auditors during this stage. The third stage of development spanned 2001–2008, and was characterized by greater convergence with international practices.3 The MOF also issued the Accounting System for Business Enterprise (2001), which represented a further step toward 3 There have been several other reforms in the Chinese accounting standards in recent years. On December 6th, 2007, the China Accounting Standards Committee (CASC) and the Hong Kong Institute of Certified Public Accountants (HKICPA) issued a joint declaration on mainland Chinese standards convergence. A separate declaration was also issued relating to auditing standards. China has also tried to converge with the IFRS, but with a focus on the convergence process rather than direct adoption. See China affirms commitment to converge with IFRSs, a press release issued by the IASB on February 15, 2006 (http:// www.ifrs.org/News/Announcements+and+Speeches/China+affirms+commitment+to+converge+with+IFRSs.htm).

K.C.K. Lam et al. / Journal of Contemporary Accounting & Economics 9 (2013) 123–135

125

convergence with the International Financial Reporting Standards (IFRS). For example, inventory valuation at lower of cost or market (LCM) was optional in the 1998 accounting standards but was required in the 2001 version. We examine whether changes in the value relevance of accounting information occurred over time along with these reforms.

3. Literature and research questions There are reasons to believe that the value relevance of Chinese firms’ financial statements has improved over time. Xiao (1999) find evidence of a high level of compliance with accounting regulations in China. Bao and Chow (1999) and Sami and Zhou (2004) investigate the value relevance of accounting information based on domestic Chinese GAAP and that based on international accounting standards.4 Although the international accounting standards had better value relevance, they find that the explanatory power of the earnings and book values for share prices increased over the period of study for both sets of financial information. There are also reasons to suspect that the gains from convergence in China may be limited. The potential impediments include (1) the speculative nature of investment in Chinese markets; (2) a lack of institutional features such as a good legal system to provide incentives for managers to supply credible accounting information (Chan et al., 2007); (3) widespread earnings management activity in emerging markets (Haw et al., 2005); (4) less compliance with standards than expected because the consequences of non-compliance are perceived not to be serious (Street and Gray, 2001); and (5) accountants who are inadequately qualified to perform their tasks (Tang, 2000). Moreover, auditors may not be completely independent from their clients, as often both were spun off from the same governmental units. In sum, the Chinese situation is different from that in the West, which creates a unique opportunity for the application of the value relevance framework. Given the increasingly important effect of intangible assets on a firm’s value in a modern economy (Lev and Zarowin, 1999) and the increasing difficulty in valuing firms (Jenkins, 1994), especially young firms, it would also be interesting to investigate these issues in the emerging market of China. Although it may be argued that the development of accounting in the West has stagnated in past decades, this is not so in China, which has literally built a whole market-oriented accounting system from the ground up since the early 1990s, when a modern accounting framework was virtually non-existent. We thus expect significant increases in the value relevance of financial statements in this emerging market, which leads to our first research question. RQ1: Has there been significant change in the value relevance of accounting information among Chinese firms over time? We also investigate the firm characteristics associated with value relevance improvements, such as firm size, asset tangibility, growth rate, long-term debt monitoring, cross listing status, and ownership structure (Francis and Schipper, 1999; Sami and Zhou, 2004). As accounting reforms are accompanied by changes in institutional, audit, and market environments, we are interested in the roles of auditor type, auditor location, market sentiment, political connection, and institutional environment. Our second research question is as follows. RQ2: What factors contribute to the changes in the value relevance of accounting information among Chinese firms? 4. Research design 4.1. Sample selection The sample selection starts with the entire population of firm-year observations for which price and return data are available in the relevant Chinese databases, including the China Stock Market (CSMAR) and the Taiwan Economic Journal (TEJ) databases, for the period 1994–2008. The sample selection process, which is shown in Table 1, then brings this number down to 11,199 firm-year observations. 4.2. Research design The value relevance literature documents two common approaches to measuring the level of value relevance. The first approach is to examine how much of an abnormal return can be earned from the foreknowledge of financial statement information, which is used by Ou and Penman (1989), Harris and Ohlson (1990), and Alford et al. (1993).5 4 A-shares are denominated in RMB and issued only to Chinese citizens, and B-shares are denominated in U.S. dollars on the Shanghai Stock Exchange or in Hong Kong dollars on the Shenzhen Stock Exchange and before 2001 were issued only to foreign residents. Since February 19, 2001, B-shares have also been sold to domestic investors with foreign currency accounts. As the accounting information in the A- and B-share markets is respectively prepared and audited under the Chinese GAAP and the International Financial Reporting Standards (IFRS), these studies investigate the relative value relevance of accounting information in the two segregated markets. 5 However, as indicated by Francis and Schipper (1999), the implementation of this approach can be difficult and subjective. For instance, in predicting returns, unsubstantiated assumptions are somehow needed on the change in risk over the investing horizon.

126

K.C.K. Lam et al. / Journal of Contemporary Accounting & Economics 9 (2013) 123–135

Table 1 Sample selection. Sampling procedure

Firm-year observations

Firm-year observations with price, return, and financial data from 1994 to 2008 for estimation Less: deletion of the top and bottom 1% of the variables used in the estimation equation Sample size before deleting influential observations Less: deletion of influential observations with a studentized residual of greater than 3.0 or a Cooks’ D statistic of greater than 1.0 Sample for estimation – firm-year observations Distribution of the sample by type of shares Firm-year observations for firms with A shares only Firm-year observations for firms with A and B shares Firm-year observations for firms with A and H shares Firm-year observations for firms with B shares only

13,351 1587 11,764 565 11,199 9289 962 250 698

Note: We follow the procedures used in Francis and Schipper (1999). The full sample consists of all firm-year observations for which price and return data are available for the period 1994–2008 from the CSMAR (GTA) database. To ensure that the results for the estimation of the different models are not sensitive to extreme observations, we eliminate the top and bottom 1% of each variable and remove any observations that the diagnostics of Belsley et al. (1980) indicate are influential observations (i.e., those with a studentized residual of greater than 3.0 or a Cooks’ D statistic of greater than 1.0). These procedures apply to all of the data except the cash flow data. As the cash flow data are only available after 1998, there are fewer observations. We also obtain from the Taiwan Economic Journal database institutional data and data on the locality of the firms. These institutional data do not restrict the sample selection.

The second approach, termed the explanatory power approach, is based on the power of accounting information to explain market value and returns. For instance, earnings are used to explain annual market-adjusted returns, and the book values of assets and liabilities are used to explain the market value of equity. An increase or decrease in relevance over time can be measured by the change in the explanatory power of accounting information (e.g., Francis and Schipper, 1999; Sami and Zhou, 2004). To investigate our first research question, we adopt the explanatory power approach and use the R-squares from the yearly regression of the following models. Earnings relation:

RETj;t ¼ q0;i þ q1;t DEARN LMVj;t þ q2;t EARN LMVj;t

ð1Þ

6

Balance sheet relation :

MV PSj;t ¼ p0;t þ p1;t ASSETS PSj;t þ p2;1 LIAB PSj;t

ð2Þ

7

Book value and earnings relation :

MV LBVj;t ¼ d0;t þ d1;t BV LBVj;t þ d2;t EARN LBVj;t

ð3Þ

Accruals and cash flow relation8:

RETjt ¼ c þ c1 ACCR LMVj;t þ c2 CFO LMVj;t

ð4Þ

where RETjt is the cumulative market-adjusted return on stock shares over the 12-month period ending 3 months after the fiscal year end; EARN_LMVjt the earnings before extraordinary items, deflated by the initial market value of equity in year t; DEARN_LMVjt the change in earnings before extraordinary items, deflated by the market value of equity at the beginning of year t; MV_PSjt the per share market value at the end of year t; ASSETS_PSjt the per share book value of assets at the end of year t; LIAB_PSjt the per share book value of liabilities at the end of year t; MV_LBVjt the market value of equity at the end of year t deflated by the book value of equity at the beginning of year t; BV_LBVjt The firm j’s book value of equity at the end of fiscal year t deflated by the book value of equity at the beginning of year t; EARN_LBVjt the earnings before extraordinary items, deflated by the book value of equity at the beginning of year t; ACCR_LMVjt the accruals deflated by the market value of equity at the beginning of year t. Following Aboody et al. (2002), we compute accruals as (DCA  DCash)  (DCL  DSTD  DTP)  DEP, where DCA represents the change in current assets, DCash is the change in cash or cash equivalents, DCL is the change in current liabilities, DSTD is the change in the debt included in current liabilities, DTP is the change in taxes payable, and DEP is the depreciation and amortization expense; and CFO_LMVjt is the cash flow from operating activities, deflated by the market value of equity at the beginning of year t. 6

We also tried a deflator of the initial book value of equity in Models (2) and (4). The results are similar to those reported here. Scaling by the book value of equity yields a regression model that includes the inverse of the book value of equity and where the intercept can be interpreted as a coefficient on the book value of equity in the unscaled model (e.g., Trueman et al., 2000; Core et al., 2003). We also tried the contemporaneous book value as a deflator, and the results are qualitatively the same as those reported. 8 A similar model is used by Aboody et al. (2002). 7

K.C.K. Lam et al. / Journal of Contemporary Accounting & Economics 9 (2013) 123–135

127

The change in the absolute value of the residuals of the valuation models is used to measure improvements in the value relevance of accounting information. The absolute value of the residuals estimates the portion of the pricing process that is unexplained by accounting information. Thus, the change in the absolute value of the residuals is a measure of improvement in the value relevance of financial information. The absolute values of the residuals are used rather than the residual values themselves because the residuals are expected to have a zero mean and constant variance (Cook and Weisberg, 1982). We use the following model to investigate the second question.

DVAL RELMj;t ¼ a þ b1 SIZEj;t þ b2 GROWj;t þ b3 DEBTj;t þ b4 CLISTj;t þ b5 BIG4j;t þ b6 ALOCj;t þ b7 INSTITj;t þ b8 STATEj;t þ b9 SINDEXj;t þ b10 TANGj;t þ e

ð5Þ

where DVAL_RELMj,t is the change in the inverse value relevance measure, calculated as the absolute value of the residuals from the yearly regressions of Models (1) to (4) in year t minus those in year t  19; SIZEj,t the logarithm of total assets; GROWj,t the market to book value; DEBTj,t the long-term debt to assets ratio; CLISTj,t a dummy that represents the cross listing status of firm j, coded as one if the firm has shares listed in Hong Kong (H-share) or elsewhere overseas, and zero otherwise; BIG4j,t a dummy variable on auditor type, coded as one if a firm is audited by one of the big 4 international auditors or one of their former bodies, and zero otherwise; ALOCj,t a dummy variable on the auditor–client relationship, coded as one if a firm is audited by an auditor from the same province, and zero otherwise; INSTITj,t an institutional factor that measures the economic development of the region in which firm j operates, computed as the logarithm of the per capita GDP of the region10; STATEj,t a measure of a firm’s political connection, measured as the percentage of state-owned shares of firm j; SINDEXj,t the one-year cumulative stock market return on the Shanghai or Shenzhen Stock Exchange on which firm j is listed; and TANGj,t is the asset tangibility index of firm j based on Almeida and Campello (2007).11 We include SIZE, GROW, and DEBT to test the effects of firm size, growth, and monitoring by long-term debt holders on the value relevance. As cross-listed firms (Sami and Zhou, 2008) and Big 4 clients (DeAngelo, 1981; Dopuch and Simunic, 1980) have a lower information asymmetry, we expect DVAL_RELM to be negatively related to both CLIST and BIG4. The institutional environment has a significant impact on auditor reporting behavior in China (Chan et al., 2010). We thus expect negative coefficients on ALOC and INSTIT because a better institutional setting and better knowledge by an auditor about a client help to improve information quality. We expect positive coefficients on SINDEX and STATE because a high level of speculation and strong political connections can reduce the informativeness of accounting information. Finally, we include asset tangibility as a control variable and expect a negative coefficient on TANG (Almeida and Campello, 2007). 5. Empirical results 5.1. Descriptive statistics The descriptive statistics of the sample, including the definitions of the variables and the Pearson bivariate correlation coefficients, are presented in Tables 2 and 3. These results provide preliminary evidence that RET is positively related to EARN_LMV and DEARN_LMV, which are used in the earnings relation model (Model (1)). In addition, the correlations between MV_PS and ASSETS_PS in the balance sheet relation model – Model (2) – are positive and significant. It is noteworthy that the correlation between MV_PS and LIAB_PS is insignificant. Previous studies have documented that politically connected firms obtain preferential access to debt financing in China (Li et al., 2008), and thus this result is not completely surprising. As to the variables used in the book value and earnings relation model – Model (3) – the correlations between MV_LBV and BV_LBV and EARN_LBV are all positive and significant. As expected, ACCR_LMV in the accruals and cash flow model – Model (4) – is negatively correlated with RET, whereas CFO_LMV is positively correlated with RET. The correlations among the independent variables used in the same model are not as high as in Model (1), but are higher than 0.50. Because these are the major variables of the models used in the literature, no attempt is made to exclude any of them here. 5.2. Ordinary least squares (OLS) regressions Table 4 presents the slope coefficients and t-statistics for Models (1) to (4) using yearly data. The first four columns report the regression results for the earnings relation model – Model (1). The yearly ordinary least squares (OLS) regressions show that the model performs reasonably well in all years.12 EARN_LMV is positively and significantly associated with stock returns in all years except for 1994, 1995, 1997, and 1998, and DEARN_LMV is positively and significantly associated with stock returns over all years except for 1994, 1995, 1996, 2001, 2003, and 2006. Overall, the results are consistent with the accounting 9 It is impossible to use the R-squares from the yearly regressions of Models (1) to (4) to construct such a variable, as we can only obtain 15 possible values (from 1994 to 2008) for the R-squares. In addition, the R-squares are only available for each sample in each year and not for each firm in the sample. 10 Source: China NBS Statistical Data. See: http://219.235.129.58/reportView.do?Url=/xmlFiles/21af1011f34f48ca940ec0fe043cdd52.xml&id=f4e9def 81b194d6a8d0c9eb9ece96a0b&bgqDm=20050000. 11 We use the same index as that computed by Almeida and Campello (2007), which measures asset tangibility as: (0.715 * Receivables + 0.547 * Inventory + 0.535 * Fixed Assets + Cash)/Book Value of Equity. 12 All of our regressions have significant F-statistics with p < 0.01, with the exception of that for 1994.

128

K.C.K. Lam et al. / Journal of Contemporary Accounting & Economics 9 (2013) 123–135

Table 2 Variable definitions and descriptive statistics. Variable

Description

Panel A. Definitions of the variables Dependent variables in the four models RET Cumulative market-adjusted return MV_PS Per share market value of the firm’s equity at the end of the fiscal year MV_LBV Market value divided by the lagged book value (book value at the beginning of the year) Explanatory variables in the four models EARN_LMV Earnings before extraordinary items deflated by the market value of equity at the beginning of the fiscal year DEARN_LMV Change in earnings before extraordinary items deflated by the market value of equity at the beginning of the fiscal year ASSETS_PS Per share book value of assets at the end of the fiscal year LIAB_PS Per share book value of liabilities at the end of the fiscal year BV_LBV Book value of equity at the end of the fiscal year deflated by the book value of equity at the beginning of the fiscal year EARN_LBV Earnings before extraordinary items at the end of the fiscal year deflated by the book value of equity at the beginning of the fiscal year ACCR_LMV Accruals deflated by the market value of equity at the beginning of the year CFO_LMV Cash flow from operations deflated by the market value of equity at the beginning of the year Residual (VAL_RELM) and changes in the residual (DVAL_RELM) variables relating to the four models AR_EARN Absolute value of residuals based on the earnings relation (Eq. (1)) AR_BAL Absolute value of residuals based on the balance sheet relation (Eq. (2)) AR_BV_EARN Absolute value of residuals based on the book value and earnings relation (Eq. (3)) AR_AC_CF Absolute value of residuals based on accruals and cash flow (Eq. (4)) DAR_EARN Changes in the absolute value of residuals based on the earnings relation (Eq. (1)) DAR_BAL Changes in the absolute value of residuals based on the balance sheet relation (Eq. (2)) DAR_BV_EARN Changes in the absolute value of residuals based on the book value and earnings relation (Eq. (3)) DAR_AC_CF Changes in the absolute value of residuals based on accruals and cash flow (Eq. (4)) Variables used in the regression of DVAL_RELM SIZE Firm size, defined as the logarithmic (base 10) transformation of total assets GROW Firm growth measured as the market to book ratio DEBT Long-term debt to total assets ratio CLIST 1 if the firm is cross-listed outside of mainland China, and 0 otherwise BIG4 1 if the auditor is one of the Big 4 international auditors (PWC, KPMG, Deloitte, EY), and 0 otherwise ALOC 1 if the auditor is from the same province, and 0 otherwise INSTIT An institutional variable computed as the log of per capita GDP of the province and location SINDEX One-year cumulative stock market return on the Shanghai or Shenzhen stock exchange on which firm j is listed TANG Asset tangibility index (Almeida and Campello, 2007) STATE Percentage of state-ownership of the firm’s shares

Panel B. Descriptive statistics RET Annual market ret MV_PS Market value (price) per share MV_LBV Market value to lagged book value EARN_LMV Earnings to lagged market value DEARN_LMV Changes in earnings ASSETS_PS Assets per share LIAB_PS Liabilities per share BV_LBV Book value to lagged book value EARN_LBV Earnings to lagged book value ACCR_LMV Accruals to lagged market value CFO_LMV Cash from operations to lagged market value

Mean

Std. dev.

Min

Max

Q1

Median

Q3

0.230 8.693 3.572 0.050 0.014 7.238 3.792 1.085 0.068 0.063 0.095

0.803 6.729 3.1340 0.193 0.165 5.258 3.538 0.288 0.170 0.280 0.329

0.763 0.066 0.007 1.067 2.087 0.992 0.216 0.649 1.105 1.640 0.819

3.963 108.700 25.778 2.223 1.407 46.505 32.114 3.121 0.840 0.862 2.067

0.286 4.450 1.583 0.006 0.028 3.999 1.574 1.001 0.018 0.259 0.014

0.058 7.100 2.652 0.026 0.002 5.774 2.792 1.050 0.074 0.036 0.052

0.544 10.890 4.500 0.057 0.016 8.555 4.750 1.134 0.141 0.022 0.144

literature, which empirically shows that earnings and earnings change are significant variables in explaining stock prices and returns (e.g., Burgstahler and Dichev, 1997; Collins et al., 1997). The second four columns report the regression results for the balance sheet model – Model (2). The yearly OLS regressions show that the model performs better in later years than in earlier years. Interestingly, up to 2002, ASSETS_PS and LIAB_PS are often insignificant and sometimes have unexpected significant signs. This may signify the difficulty that many newly listed firms had with accurately valuing their assets and liabilities when they were first publicly listed and when the stock market was just beginning to function. However, after 2002, ASSETS_PS is significantly and positively associated with MV_PS and LIAB_PS is significantly and negatively associated with MV_PS. The third four columns report the regression results for the book value and earnings model – Model (3). The yearly OLS regressions show that the model performs well in all years, as suggested by highly significant F-statistics. BV_LBV is significantly and positively associated with MV_LBV in all years except for 1994. However, the coefficient of EARN_LBV is much less stable, with unexpected negative signs in 2001, 2002, and 2003, and insignificant results in 1995, 2000, 2004, and 2005.

129

K.C.K. Lam et al. / Journal of Contemporary Accounting & Economics 9 (2013) 123–135 Table 3 Pearson correlation matrix.

RET MV_PS MV_LBV EARN_LMV DEARN_LMV ASSETS_PS LIAB_PS BV_LBV EARN_LBV ACCR_LMV CFO_LMV

RET

MV_PS

MV_LBV

EARN_LMV

DEARN_LMV

ASSETS_PS

LIAB_PS

BV_LBV

EARN_LBV

ACCR_LMV

0.376*** 0.320*** 0.186*** 0.116*** 0.038*** 0.026* 0.167*** 0.174*** 0.068*** 0.125***

0.688*** 0.034*** 0.046*** 0.055*** 0.005 0.306*** 0.332*** 0.149*** 0.121***

0.130*** 0.078*** 0.327*** 0.236*** 0.324*** 0.207*** 0.174*** 0.180***

0.186*** 0.350*** 0.204*** 0.246*** 0.424*** 0.286*** 0.480***

0.058** 0.051** 0.122*** 0.176*** 0.032** 0.099***

0.913*** 0.134*** 0.163*** 0.056*** 0.115***

0.061*** 0.072*** 0.056*** 0.059***

0.640*** 0.013 0.093***

0.034*** 0.183***

0.957***

*

Significance at the 0.10 level. Significance at the 0.05 level. *** Significance at the 0.01 level. **

The last four columns report the regression results for the accruals and cash flow model – Model (4). Due to restricted data availability because the Chinese accounting standards only required cash flow data from 1998 onward, we can only present the results for the years after 1997. The yearly OLS regressions show that the model performs reasonably well in all years, as suggested by the significant F-statistics. ACCR_LMV and CFO_LMV are both significantly and positively associated with RET in all of the sample years except for 1998, the first year in which such data were released, and 2000, for which ACCR_LMV is not significant. Overall, the results are consistent with the literature on the value relevance of accruals and cash flow information (e.g., Sloan, 1996; Pfeiffer and Elgers, 1999). Panel A of Table 5 presents the slope coefficients and t-statistics for Models (1)–(4) using sub-period data. The first four columns report the regression results for the earnings relation model – Model (1). The OLS regression results for the pooled sample (at the bottom) and by period show that the model has highly significant F-statistics (especially in 2001–2008 and for the pooled sample) and that EARN_LMV and DEARN_LMV have significantly positive coefficients in all periods. The second four columns report the regression results for the balance sheet model – Model (2). Consistent with the annual regressions, the period OLS regressions show that the model performs much better in the last period than in the earlier two periods. In the periods 1994–1997 and 1998–2000, ASSETS_PS take a negative sign and LIAB_PS a positive sign, which is counterintuitive. In 2001–2008, however, the results are much more reasonable, with ASSETS_PS being significantly positive and LIAB_PS being significantly negative. The third four columns report the period regression results for the book value and earnings model – Model (3). In general, BV_LBV always has the expected positive and significant coefficient whereas EARN_LBV is insignificant in the first and third periods and for the pooled sample. Note also that the R-squares steadily decline over the years. The last column reports the period regression results for the accruals and cash flow model from 1998 onward – Model (4). The model performs well for all periods, and both ACCR_LMV and CFO_LMV are significantly and positively associated with RET in all sample periods and in the pooled sample. To further investigate the seemingly inconsistent results between the increasing value relevance for the balance sheet relation and the declining value relevance for the book value and earnings relation with the earnings items, and the varying signs for the variables in the two models (Table 5), we modify the models for the balance sheet relation and book value and earnings relation by regressing MV_PS on ASSETS_PS, LIAB_PS, EARN_PS, and regressing MV_LBV on ASSETS_LBV, LIAB_LBV, and EARN_LBV. The results are reported in Panel B of Table 5, which confirms the results reported in Panel A.13 In summary, the results are consistent with those of past studies (e.g., Sami and Zhou, 2004; Lev and Zarowin, 1999; Francis and Schipper, 1999) that report a significant relationship between stock prices and the accounting information provided to investors. The Vuong tests that compare the regressions by sub-period are reported in Table 6. Interestingly, for all of the relations, the regressions in the later period (2001–2008) are closer to the true model. Also, as we move from one period to the next, the Vuong tests show that all of the regression models move closer to the true model, indicating an improvement in the performance of the accounting information models over time. The regression of the adjusted R-squares on CNT_YR for each model is presented in Table 7. CNT_YR takes the value of 1 for 1994, 2 for 1995, . . . ,14 for 2008. The results are similar if we use sub-periods rather than CNT_YR. CNT_YR in the balance sheet relation has a positive coefficient whereas that of the book value and earnings relation has a negative coefficient, and both are statistically significant. Table 8 presents the average absolute value of the residuals from Models (1) to (4). Interestingly, the residuals fluctuate every year. Table 9 presents the slope coefficients and t-statistics for Model (5) to explore the determinants of changes in the value relevance of accounting information. The dependent variable is the change in the absolute value of the residuals (DVAL_RELM), which is estimated for Models (1) to (4) and represented by DAR_EARN, DAR_BAL, DAR_BV_EARN, and 13

We thank the reviewer for raising this issue.

130

Table 4 Changes in the value relevance of accounting information over time: yearly regressions. Year

n

Model 1: Earnings relation

38

1995

186

1996

267

1997

322

1998

497

1999

684

2000

679

2001

866

2002

1050

2003

1123

2004

1151

2005

1192

2006

1165

2007

802

2008

1177

Model 3: Book value and earnings relation

Model 4: Accruals and cash flow relation

Intercept

EARN_ LMV

DEARN_LMV

Adj R2 F-stat

Intercept

ASSETS_ PS

LIAB_ PS

Adj R2 F-stat

Intercept

BV_ LBV

EARN_ LBV

Adj R2 F-stat

0.393 9.09*** 0.167 8.86*** 0.741 16.51*** 0.273 9.72*** 0.092 5.34*** 0.186 13.00*** 0.609 38.37*** 0.185 16.02*** 0.213 41.46*** 0.2140 16.29*** 0.156 21.74*** 0.149 20.10*** 0.771 44.02*** 1.442 68.65*** 0.571 118.45***

0.208 0.83 0.034 0.46 0.423 2.46** 0.058 0.32 0.222 1.85** 0.219 2.10** 0.586 7.23*** 0.820 10.93*** 0.115 3.06*** 0.674 11.22*** 0.212 3.99*** 0.471 7.80*** 0.752 5.88*** 0.284 2.76*** 0.300 4.00***

0.283 1.49 0.220 3.77*** 0.277 0.83 0.794 5.93*** 0.654 3.74*** 0.730 4.85*** 0.361 2.69*** 0.552 7.82*** 0.193 5.15*** 0.007 0.08 0.353 4.63*** 0.349 4.11*** 0.148 0.96 0.517 4.45*** 0.169 5.57***

0.06 2.21 0.06 7.14*** 0.03 4.47** 0.10 18.0*** 0.04 11.5*** 0.12 45.99*** 0.07 26.61*** 0.27 161.02*** 0.03 17.41*** 0.16 110.99*** 0.06 38.76*** 0.11 73.40*** 0.05 30.61*** 0.03 14.65*** 0.03 19.84***

6.943 8.52*** 5.273 14.55*** 7.156 11.48*** 8.426 23.33*** 9.829 32.73*** 10.368 36.90*** 13.660 28.91*** 10.330 41.02*** 8.364 42.04*** 5.979 34.51*** 4.217 29.42*** 3.388 24.10*** 4.147 19.30*** 10.940 32.91*** 3.863 19.20***

0.191 1.50 0.136 1.31 0.045 0.42 0.165 1.75* 0.313 3.63*** 0.233 2.89*** 0.365 4.10*** 0.014 0.20 0.035 0.64 0.350 8.24*** 0.470 11.38*** 0.422 11.01*** 0.621 10.21*** 0.429 5.07*** 0.602 10.96***

0.239 1.42 0.203 1.23 0.016 0.09 0.184 1.16 0.291 2.07** 0.238 1.89* 0.281 1.90* 0.010 0.09 0.099 1.15 0.419 6.44*** 0.542 9.09*** 0.490 9.00*** 0.597 7.13*** 0.455 3.64*** 0.590 7.80***

0.00 1.16 0.00 0.86 0.00 0.60 0.00 1.80 0.04 10.16*** 0.01 5.33*** 0.04 15.34*** 0.00 0.18 0.00 1.13 0.06 36.09*** 0.11 71.64*** 0.10 65.29*** 0.11 73.88*** 0.04 16.99*** 0.13 85.03***

1.283 1.35 0.033 0.05 0.214 0.34 0.098 0.16 0.047 0.10 1.601 3.57*** 2.608 4.91*** 1.104 2.83*** 1.314 2.88*** 1.009 2.75*** 0.056 0.20 0.097 0.34 0.085 0.27 1.745 4.37*** 0.380 1.62

0.144 0.18 1.682 2.65*** 2.361 3.83*** 3.051 5.11*** 3.341 7.51*** 2.150 4.81*** 2.379 4.72*** 2.839 7.73*** 1.926 4.24*** 3.444 9.69*** 1.888 6.66*** 1.717 6.00*** 2.224 7.14*** 2.367 6.42*** 1.695 7.41***

2.351 3.48*** 1.174 1.18 2.693 2.72*** 2.493 2.13** 2.049 2.40*** 2.999 3.35*** 0.977 0.87 0.441 0.57 1.357 2.18** 0.880 1.84* 0.415 1.20 0.088 0.28 1.548 3.79*** 1.874 2.83*** 0.726 2.29**

0.23 6.50*** 0.06 6.48*** 0.24 42.93*** 0.25 54.06*** 0.26 88.75*** 0.12 45.69*** 0.06 20.81*** 0.08 37.72*** 0.02 9.56*** 0.11 67.79*** 0.09 60.00*** 0.07 46.24*** 0.13 91.21*** 0.11 48.04*** 0.10 64.70***

Intercept

ACCR_ LMV

CFO_ LMV

Adj R2 F-stat

0.080 4.62*** 0.176 12.29*** 0.604 39.01*** 0.164 13.63*** 0.214 41.75*** 0.136 15.87*** 0.161 22.89*** 0.146 20.18*** 0.789 45.85*** 1.434 66.69*** 0.581 118.97***

0.215 1.46 0.956 8.94*** 0.058 0.45 0.497 4.26*** 0.177 3.29*** 0.877 12.77*** 0.743 9.72*** 0.796 11.30*** 1.172 8.15*** 0.362 2.19** 0.247 2.86***

0.269 2.22** 0.842 9.60*** 0.265 2.45** 0.649 6.46*** 0.167 3.52*** 0.843 14.09*** 0.652 9.72*** 0.718 11.62*** 1.041 8.05*** 0.334 2.29** 0.248 2.89***

0.01 4.38** 0.12 46.38*** 0.12 46.51*** 0.19 98.64*** 0.01 6.51*** 0.16 105.47*** 0.08 47.70*** 0.10 67.50*** 0.06 33.32*** 0.00 2.64* 0.01 4.23**

Note: With abs(RSTUDENT) > 3 deleted; COOK’S D > 1 deleted and outliers (1 percentile and 99 percentile deleted). Note that the 1993 and 1992 annual regressions are not presented because of the small sample sizes (lower than 20). * Significance at the 0.10 level. ** Significance at the 0.05 level. *** Significance at the 0.01 level.

K.C.K. Lam et al. / Journal of Contemporary Accounting & Economics 9 (2013) 123–135

1994

Model 2: Balance sheet relation

Table 5 Changes in the value relevance of accounting information over time: regression results by sub-period. Year

n

Model 1: Earnings relation

Model 2: Balance sheet relation

Intercept EARN_LMV DEARN_LMV Adj R2 F-stat

Intercept ASSET_PS LIAB_PS

Model 3: Book value and earnings relation Model 4: Accruals and cash flow relation Adj R2 F-stat

Intercept BV_LBV

EARN_LBV Adj R2 F-stat

Intercept ACCR_LMV CFO_LMV Adj R2 F-stat

1994– 1997

812

0.316 ***

0.230 **

0.543

0.04

***

1998– 2000

1860

13.08 0.310

2.13 0.433

5.46 0.399

18.99 0.05

2001– 2008

8526

28.66*** 0.047

7.40*** 0.830

4.54*** 0.375

1994– 2008

11,199

6.12*** 0.109

16.71*** 0.699

17.00***

18.05***

7.391 ***

33.41 11.75

***

0.162

0.122 **

*

0.00

1.120 *

2.32 0.02

Panel B. Modified Models (2) and (3) 1994–1997 812 7.539 36.28*** 1998–2000 1860 12.450 64.38*** 2001–2008 8526 6.595 77.06*** 1994–2008 11,199 7.783 99.50***

3.615 ***

0.20

0.80 2.303

103.77*** 0.12

1.77 0.188

45.00*** 59.66*** 0.04 6.001

4.55*** 0.425

2.12** 0.486

6.99*** 0.386

196.60*** 70.18*** 0.04 7.201

18.56*** 0.277

14.62*** 191.80*** 1.73* 0.365 0.02 0.389

21.83*** 1.08 3.213 0.011

384.77*** 0.10

8.99***

241.84*** 91.53***

13.09***

11.73***

27.84*** 0.06

3.15 1.473

17.99*** 5.10*** 0.04 0.227

85.86*** 3.28***

10.68 2.611

0.470

2.14 0.243

Model 2: Balance sheet relation Intercept

***

9.40*** 4.06*** 2.810 0.196

0.306

0.293

0.391

0.04

3.39*** 1.136

5.43*** 1.082

39.75*** 0.03

15.76*** 0.986

17.04*** 0.952

153.50*** 0.03

654.53*** 13.096*** 16.43***

18.20***

182.13***

130.16*** 28.15*** 0.08 0.046 6.15*** 0.092

Model 3: Book value and earnings relation

ASSET_PS

LIAB_PS

EARN_PS

Adj R2 F-stat

Intercept

ASSET_LBV

LIAB_LBV

EARN_LBV

Adj R2 F-stat

0.496 7.60** 0.890 12.94*** 0.038 1.43 0.147 5.92***

0.584 6.14*** 0.897 9.21*** 0.071 1.98* 0.092 2.73***

3.011 10.22*** 4.852 14.34** 2.614 25.73*** 2.952 29.78***

0.12 36.58*** 0.12 81.83*** 0.11 357.98*** 0.09 357.31***

1.310 3.71*** 1.203 4.20*** 0.220 1.69* 0.445 3.78***

3.483 10.39*** 2.426 8.82*** 2.513 19.38*** 3.035 26.16***

3.206 9.29*** 2.016 7.07*** 2.295 17.16*** 2.852 23.86***

0.614 1.13 2.808 4.99*** 0.077 0.43 0.190 1.08

0.22 78.39*** 0.15 108.16*** 0.10 315.49*** 0.11 480.37***

Note: With abs(RSTUDENT) > 3 deleted; COOK’S D > 1 deleted and outliers (1 percentile and 99 percentile deleted). * Significance at the 0.10 level. ** Significance at the 0.05 level. *** Significance at the 0.01 level.

K.C.K. Lam et al. / Journal of Contemporary Accounting & Economics 9 (2013) 123–135

Panel A. Models (1) to (4)

131

132

K.C.K. Lam et al. / Journal of Contemporary Accounting & Economics 9 (2013) 123–135

Table 6 Vuong test of the regressions for different periods. Reg1

Reg2

Vuong statistic

Pr > |Z|

Preferred regression

Earnings relation (Model 1) 1994_1997 1994_1997 1998_2000

1998_2000 2001_2008 2001_2008

8.364 11.000 20.845

<0.0001 <0.0001 <0.0001

1998_2000 2001_2008 2001_2008

Balance sheet relation (Model 2) 1994_1997 1998_2000 1994_1997 2001_2008 1998_2000 2001_2008

10.025 10.697 19.701

<0.0001 <0.0001 <0.0001

1998_2000 2001_2008 2001_2008

Book value and earnings relation (Model 3) 1994_1997 1998_2000 1994_1997 2001_2008 1998_2000 2001_2008

11.225 6.719 12.727

0.0007 <0.0001 <0.0001

1998_2000 2001_2008 2001_2008

Accruals and cash flow relation (Model 4) 1998_2000 2001_2008

20.573

<0.0001

2001_2008

DAR_AC_CF, respectively. Common to all of the relations, the increase in value relevance (represented by a higher DVAL_RELM) is greater for smaller firms, firms with lower growth rates, and those with greater asset tangibility. The value relevance improvements are lower when the stock market sentiment (SINDEX) is high. In two models, firms with a higher level of debt monitoring show greater value relevance improvements. We do not find evidence that cross-listed shares (CLIST), having an auditor from the same province (ALOC), having a Big 4 international auditor (BIG4), state ownership (STATE), or the institutional environment (INSTIT) have significant impacts on the changes in the value relevance of accounting information.

5.3. Sensitivity analyses In this section we report the results of sensitivity analyses. First, in the main tests we focus on the value relevance of accounting information in the domestic markets, yet the sample firms may also issue other types of shares: B-shares, Hshares, or both. Hence, as a robustness test, we perform all of the analyses separately for firms with A-shares only.14 When we restrict our analyses to A-share only firms, the results for all models are very similar to those already reported. Second, past studies find that measures of value relevance may be affected by market inefficiencies, and hence stock prices may become a poor measure of the fundamental value of firms over time due to pervasive noise trading in recent years (Aboody et al., 2002; Fung et al., 2010). As a sensitivity analysis, we replace our value relevance (RETj,t, MV_PSj,t, and MV_LBVj,t) measures with market inefficiency-adjusted measures, following Eq. (2) in Fung et al. (2010). We find that our results are robust to the use of the inefficiency-adjusted measures.15,16 Third, as the results for the institutional variable that we use in the regressions are insignificant, we collect other institutional variables such as the number of listed enterprises per province (or location), the marketization index for China’s provinces, the human development index per province, and the tax revenue per province. We also use dummy variables for large cities such as Shenzhen, Shanghai, and Beijing. The results are similar to the institutional variables used in the original analysis. We also control for the effects of corporate governance features as a sensitivity test. We collect two governance measures: duality (CEO and general manager separation) and the percentage of independent directors, data on both of which are available after 2000. Neither variable has a significant incremental impact on the changes in value relevance. Finally, to remove the effect of survivorship bias, we form a fixed sample with firms for which data are available for 1996–2008. The results are 14 Due to limited annual observations on firms issuing both A- and B-shares, those issuing both A- and H- shares, and those issuing only B-shares, we could not re-estimate all of the models for these groups. 15 The results imply that the Chinese stock market has not shown a steady trend of either increasing or decreasing market efficiency over the past few decades. This notion is not inconsistent with the literature, as we find no study that reports a steadily increasing or decreasing trend of market efficiency. A relevant study, although not directly so, is Haldane (2010), which shows a steadily declining trend for the average holding period of equity investment by investors in the developed markets of the United States, United Kingdom, continental European countries, and Japan. However, using observations from the Shanghai Stock Exchange for the period 2003–2009, Haldane (2010, p. 30) finds no such a clear trend for the average holding period in China. We thank the reviewer for highlighting this issue. 16 There may be some concern that the DVAL_RELM measure reflects not only the value relevance of accounting information, but possibly also market fluctuations. However, the RET variable in Models 1 and 4 is already adjusted for the market effect. To further address this concern, we recomputed the marketadjusted measures for the dependent variables in Model 2 (MV_PS or market value per share) and Model 3 (MV_LBV or market value deflated by the lagged book value). We construct a mean MV_PS and mean MV_LBV each year and perform a regression based on the difference between the actual MV_PS or MV_LBV and the annual mean value. The results do not differ much from those already presented. In addition, to detrend the dependent variables, we use the decile ranks of DVAL_RELM as the dependent variable. The results are similar to those reported in Table 9. We thank the reviewer for raising this issue.

133

K.C.K. Lam et al. / Journal of Contemporary Accounting & Economics 9 (2013) 123–135 Table 7 Regression of the adjusted R-squares across years.

* **

N

Intercept

CNT_YR

Adj R2 and F-stat.

Earnings relation (Model 1)

15

0.084 2.41**

0.002 0.38

0.07 0.14

Balance sheet relation (Model 2)

15

0.025 1.5

Book value and earnings relation (Model 3)

15

0.199 5.24***

0.009 2.10*

0.20 4.40*

Accruals and cash flow relation (Model 4)

11

0.148 2.28**

0.007 1.13

0.03 1.27

0.009 4.62***

0.59 21.32***

Significance at the 0.10 level. Significance at the 0.05 level. Significance at the 0.01 level.

***

Table 8 Average absolute value of the residuals. Year

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

Model 1: Earnings relation

Model 2: Balance sheet relation

Model 3: Book value and earnings relation

Model 4: Accruals and cash flow relation AR_AC_CF

N obs

0.299 0.255 0.523 0.375 0.337 0.313 0.307 0.303 0.706 1.313 0.684

38 186 267 322 497 684 679 866 1053 1123 1151 1192 1165 802 1177

AR_EARN

AR_BAL

AR_BV_EARN

0.560 0.325 0.706 0.338 0.290 0.256 0.513 0.388 0.352 0.329 0.322 0.317 0.680 1.307 0.675

3.164 3.614 2.473 2.721 3.108 3.466 6.514 3.667 2.650 2.571 3.064 3.783 3.468 5.280 3.555

1.514 1.369 0.984 1.413 1.678 1.837 2.662 1.763 1.487 1.275 1.298 1.493 1.340 1.929 1.424

Table 9 Determinants of change in the value relevance of accounting information. Measure of DVAL_RELM

N

Intercept

SIZE

GROW

DEBT

BIG4

CLIST

INSTIT

ALOC

SINDEX

STATE

TANG

Adj R2 (F_stat)

DAR_EARN

8042

DAR_BAL

8042

DAR_BV_EARN

8042

DAR_AC_CF

7998

1.417 8.53*** 6.693 5.49*** 5.918 12.23*** 1.441 8.65***

0.138 8.32*** 0.612 5.03*** 0.558 11.56*** 0.141 8.47***

0.060 18.75*** 0.324 13.71*** 0.245 26.11*** 0.058 17.84***

0.122 1.93* 0.082 0.18 0.156 0.84 0.123 1.94*

0.027 1.28 0.109 0.70 0.047 0.75 0.034 1.57

0.042 1.02 0.106 0.35 0.090 0.75 0.039 0.95

0.011 0.51 0.048 0.30 0.046 0.72 0.011 0.49

0.010 0.55 0.043 0.31 0.030 0.55 0.014 0.77

0.449 33.81*** 0.973 9.99*** 0.254 6.58*** 0.478 35.98***

0.008 1.23 0.049 1.06 0.011 0.58 0.005 0.84

0.060 5.54*** 0.293 3.68*** 0.155 4.90*** 0.057 5.28***

0.16 153.93*** 0.04 30.35*** 0.08 74.30*** 0.17 165.65***

Dependent variable: DVAL_RELM = changes in the absolute value of the residuals obtained from Models (1)–(4). The regression results are obtained after the deletion of the top and bottom 1% of observations for each variable and the observations that the diagnostics of Belsley et al. (1980) indicate are influential observations (i.e., those with a studentized residual of greater than 3.0 or a Cooks’ D statistic of greater than 1.0). ⁄⁄ Significance at the 0.05 level. * Significance at the 0.10 level. *** Significance at the 0.01 level.

similar to those already reported. Hence, we conclude that the effect of survivorship bias does not affect materially our conclusions. 6. Conclusion, discussion, and limitations We investigate changes in the value relevance of accounting information among Chinese firms over the past two decades. Specifically, we propose that the usefulness of accounting information in making decisions differs at different stages of

134

K.C.K. Lam et al. / Journal of Contemporary Accounting & Economics 9 (2013) 123–135

economic development that is accompanied by accounting, auditing, and institutional reforms. We further investigate the factors that differentiate firms showing significant value relevance improvement from firms showing little improvement. As expected, we find that earnings level and earnings change are significantly and positively associated with the pricing process in most years. Total assets are positively and liabilities are negatively associated with firm value only after 2002. Book value and earnings are positively associated with firm value, whereas the earnings to lagged book value ratio has less stable coefficients. Finally, accruals and cash flow information remain positively associated with firm value in all of the sample years except for the starting year of 1998. However, we do not find uniform patterns in the changes in value relevance as measured by the R-squares for these relations. For the balance sheet model, the R-squares generally increase over the years, which is consistent with the notion of the increasing value relevance of balance sheet data. For the book value and earnings relation, the R-squares decrease over the years. We find no clear pattern of changes in value relevance for the other two relations. We also investigate the average absolute value of the residuals from Models (1) to (4) as an inverse measure of value relevance, as this value represents the unexplained portion of stock returns and prices. Interestingly, 2000 and 2007 show higher residuals than the other years, and these years are associated with the highest stock returns. This correlation suggests that the value relevance of accounting information is generally lower when the stock market is more optimistic. In all other years, the residual values are relatively modest. Finally, we provide evidence that value relevance improvements are more pronounced for smaller firms, firms with lower growth rates, and those with greater asset tangibility. Our results thus suggest that the gap in accounting quality between larger and smaller firms is narrowing. Our results also corroborate the previous finding (e.g., Lev, 1989; Lev and Zarowin, 1999) that traditional accounting models fail to adequately measure firms that rely mostly on intangible assets and growth potential. This is true even in the emerging market of China. As with similar studies on changes in the value relevance of accounting information (e.g., Francis and Schipper, 1999; Lev and Zarowin, 1999), our study has certain limitations. The purpose of the study is not to document a casual relationship between accounting reforms and the value relevance of accounting information. Rather, the primary purpose is to provide greater insight into the changes in value relevance and the factors that influence it.17 We are unable to provide direct evidence on the impact of a particular accounting reform on the value relevance of accounting information because the accounting reforms in China have been accompanied by many related auditing, regulatory, and human capital reforms. Hence, in referring to accounting reforms in this study, we also include (1) improvements in auditing standards (e.g., Zhou, 2007), and improved independence due to the auditor disaffiliation program of 1997–1998 (Gul et al., 2009); (2) improvements in human capital, especially those relating to the increase in the accounting talent pool, which mitigates the shortage of qualified personnel in the early stages of the reform; and (3) reforms in the judiciary system, particularly the CPA Act of 1994 and the Securities Act of 1998. We do not believe that we can separate each of these reforms and we make no attempt to do so. The findings of our study have interesting implications for a variety of information users. First, our study provides useful information on the value relevance of accounting information in emerging markets. Second, policy makers could use the information to evaluate the effectiveness of their policies and determine future reform directions to improve those policies. Future studies could further explore the market consequences of improving the quality of financial information, such as the effects of increased disclosure on other key capital market characteristics, including stock price synchronicity, information asymmetry, and the cost of equity capital. Acknowledgements We would like to thank the participants at the American Accounting Association 2011 Annual Conference, Canadian Academic Accounting Association 2011 Annual Meeting and Journal of Contemporary Accounting and Economics 2012 Annual Conference, especially Ferddie Gul, Gordon Richardson, Simon Fung and several anonymous reviewers for their helpful comments. References Aboody, D., Hughes, J., Liu, J., 2002. Measuring value relevance in a (possibly) inefficient market. Journal of Accounting Research 40 (4), 965–986. Almeida, H., Campello, M., 2007. Financial constraints, asset tangibility, and corporate investment. Review of Financial Studies 20 (5), 1429–1460. Alford, A., Jones, J., Leftwich, R., Zmijewski, M., 1993. The relative informativeness of accounting disclosures in different countries. Journal of Accounting Research 31, 183–223. Bao, B.H., Chow, L., 1999. The usefulness of earnings and book value for equity valuation in emerging capital markets: evidence from listed companies in the People’s Republic of China. Journal of International Financial Management and Accounting 10 (2), 85–104. Belsley, D., Kuh, E., Welsh, R., 1980. Regression Diagnostics. Wiley, New York. Brown, S.K., Lo, K., Lys, T., 1999. The R2 in accounting research: measuring changes in value relevance over the last four decades. Journal of Accounting and Economics 28 (2), 83–115. Burgstahler, D.C., Dichev, I.D., 1997. Earnings, adaption and equity value. Accounting Review 72 (2), 187–216. Chan, K.H., Lin, K.Z., Zhang, F., 2007. On the association between changes in corporate ownership and changes in auditor quality in a transitional economy. Journal of International Accounting Research 6 (1), 19–36. Chan, K.H., Lin, K.Z., Wong, B., 2010. The impact of government ownership and institutions on the reporting behavior of local auditors in China. Journal of International Accounting Research 9 (2), 1–20. 17

We thank the anonymous reviewer for raising these issues.

K.C.K. Lam et al. / Journal of Contemporary Accounting & Economics 9 (2013) 123–135

135

Chen, S., Sun, Z., Wong, Y., 2002. Evidence from China on whether harmonized accounting standards harmonize accounting practices. Accounting Horizons 16 (3), 183–197. Collins, D., Maydew, E., Weiss, I., 1997. Changes in the value-relevance of earnings and book values over the past forty years. Journal of Accounting and Economics 24, 39–67. Cook, R.D., Weisberg, S., 1982. Residuals and Influence in Regression. Chapman & Hall, New York. Core, J.E., Guay, W.R., Van Buskirk, A., 2003. Market valuations in the new economy: an investigation of what has changed. Journal of Accounting and Economics 34, 43–67. DeAngelo, L., 1981. Auditor size and audit quality. Journal of Accounting and Economics 3 (3), 183–199. Dopuch, N., Simunic, D., 1980. The nature of competition in the auditing profession: a descriptive and normative view. Illinois Auditing Symposium. Fung, S., Su, L., Zhu, X., 2010. Price divergence from fundamental value and the value relevance of accounting information. Contemporary Accounting Research 27 (3), 829–854. Francis, J., Schipper, K., 1999. How financial statements lost their relevance. Journal of Accounting Research 37 (2), 319–352. Gul, F.A., Sami, H., Zhou, H., 2009. The auditor disaffiliation program in China and auditor independence. Auditing: a Journal of Theory and Practice 28 (1), 29–52. Haldane, A.G., 2010. Patience and Finance. Working Paper (Bank of England). Haw, I.M., Qi, D., Wu, D., Wu, W., 2005. Market consequences of earnings management in response to security regulations in China. Contemporary Accounting Research 22 (1), 95–143. Harris, T.S., Ohlson, J.A., 1990. Accounting disclosures and the market’s valuation of oil and gas properties: evaluation of market efficiency and functional fixation. Accounting Review 65 (4), 764–781. Jenkins, E., 1994. An examination highway in need of capital improvements. Journal of Accountancy (May), 77–80. Landsman, W., Maydew, E., 2002. Has the information content of quarterly earnings announcements declined in the past three decades? Journal of Accounting Research 40 (3), 797–808. Lev, B., 1989. On the usefulness of earnings: lessons and directions from two decades of empirical research. Journal of Accounting Research (Suppl.), 153– 182. Lev, B., Zarowin, P., 1999. The boundaries of financial reporting and how to extend them. Journal of Accounting Research 37 (2), 353–385. Li, H., Meng, L., Wang, Q., Zhou, L., 2008. Political connection, financing and firm performance: evidence from Chinese private firms. Journal of Development Economics 87 (2), 283–299. Ou, J.A., Penman, S.H., 1989. Financial statement analysis and the prediction of stock returns. Journal of Accounting and Economics 11 (4), 295–330. Pfeiffer, R.J., Elgers, P.T., 1999. Controlling for lagged stock responses in pricing regressions: an application to the pricing of cash flow and accruals. Journal of Accounting Research 37 (1), 239–247. Sami, H., Zhou, H., 2004. A comparison of value relevance of accounting information in different segments of the Chinese stock market. International Journal of Accounting 39 (4), 403–427. Sami, H., Zhou, H., 2008. The economic consequences of increased disclosure: evidence from cross-listings of Chinese firms. Journal of International Financial Management and Accounting 19 (1), 1–27. Sloan, R., 1996. Do stock prices fully reflect information in accruals and cash flows about future earnings? Accounting Review 71 (3), 289–315. Street, D.L., Gray, S.L., 2001. Observance of International Accounting Standards: Factors Explain Non-compliance. Association of Chartered Certified Accountants (ACCA) Research Report: No. 74. Tang, Y., 2000. Bumpy road leading to internationalization: a review of accounting development in China. Accounting Horizons 14 (1), 93–102. Trueman, B., Wong, M.H.F., Zhang, X., 2000. The eyeballs have it: searching for the value in Internet stocks. Journal of Accounting Research 38, 137–162. Xiao, Z., 1999. Corporate disclosures made by Chinese listed companies. International Journal of Accounting 34 (3), 349–373. Zhou, H., 2007. Auditing standards, increased accounting disclosure, and information asymmetry: evidence from an emerging market. Journal of Accounting and Public Policy 26 (5), 584–620.