Introduction
This study examines the impact of macroeconomic conditions and policies on the usefulness of accounting information to predict future profitability in an international setting by comparing manufacturing firms in two countries, China and Japan. During the global financial crisis in 2008 and 2009, countries announced economic stimulus packages to offset the negative impact of financial crisis (Freedman et al., 2009; Prasad & Sorkin, 2009). As the timing, magnitude and choices of stimulus funding varied by country, we test whether such differences in stimulus funding policies affect the persistence of earnings and the role of cash flows in the prediction of future earnings.
To test the potential impact on earnings persistence, we compare manufacturing firms in China and Japan for the fiscal years from 2006 to 2015. During the global financial crisis started in 2008, the governments of China and Japan provided stimulus funding to support the economy and the firms operating in those countries. While the goals of stimulus funding were similar, the effects on the economy varied depending on the macroeconomic conditions and how the funding was provided during the recession period. In November 2008, the Chinese government announced a stimulus package of 4 trillion yuan ($586 billion) to be spent in 2009 and 2010 (Batson, 2008; Naughton, 2009). This amount is equivalent to 12.9% of the 2008 Gross Domestic Product (GDP) of China in current US dollars, $4,558 billion (World Bank, 2016). The 4 trillion yuan stimulus package included the investments in the construction of transportation network and other infrastructure construction projects as well as the ecological protection and technical innovation projects.
As the Chinese government provided a significant amount of stimulus funding to sustain the growth of the economy, the annual growth rates of Chinese GDP were 9.62% and 9.23% in 2008 and 2009, respectively (World Bank, 2016). The sustained growth in national output supports that the stimulus funding worked as intended by boosting the productive activities of firms. In 2008 and 2009, the Japanese government also announced several stimulus packages. In August 2008, an 11.5 trillion yen ($105.8 billion) stimulus package was announced by the Japanese government (Nakamoto, 2008), and this was followed by another package of 5 trillion yen ($51.5 billion) announced in October (Tabuchi, 2008). In December 2008, additional 23 trillion yen ($250 billion) package was announced (Fackler, 2008). The total amount of these stimulus packages announced in 2008 was 39.5 trillion yen ($407.3 billion), which is equivalent to 8.4% of the 2008 GDP of Japan in current US dollars, $4,849 billion (World Bank, 2016).
However, these stimulus packages did not boost the Japanese economy as expected. The annual growth rate of Japanese GDP was -1.04% in 2008 (World Bank, 2016). Because these stimulus packages did not perform as expected, the Japanese government announced another major stimulus package in April 2009 with the total amount of 15 trillion yen ($150 billion) and continued to introduce additional funding of 7 trillion yen ($70 billion) in December 2009 (Terada & Ong, 2011). However, such efforts did not result in positive growth of the economy. Even with such additional stimulus packages provided in 2009, the annual growth rate of Japanese GDP was -5.53% in 2009 (World Bank, 2016).
One of the differences between the Chinese and Japanese stimulus packages analyzed by Prasad and Sorkin (2009) is that the Chinese stimulus packages focused on the increase in spending measures with the tax cut share of 0%. In contrast, the tax cut share of the Japanese stimulus packages was 30%.
With a focus on the effects of stimulus funding provided by the government during the global financial crisis in 2008 and 2009, this paper examines whether the differences in the national economic policies affect the usefulness of current accounting information in the prediction of future profitability. Test results show that current earnings are a better predictor of future earnings in the economy where the negative impact of global financial crisis is more effectively offset by the government stimulus packages than in the economy where such economic policy was less effective in boosting the national output.
This paper contributes to the existing literature in several ways. First, this study provides empirical evidence about how the government stimulus package affects the earnings persistence during the period of financial crisis. Secondly, the results of this study indicate that understanding the macroeconomic conditions and policies of different countries helps users better understand the roles of current accounting variables in the prediction of future profitability.
The scope of this study is limited to the manufacturing firms in China and Japan. Therefore, the results of this study are limited to those countries and industries. To provide more general test results, it is suggested to apply test models to other industries in future studies. The remainder of this paper is organized as follows. Section 2 discusses prior literature review. The hypothesis development and research design are discussed in section 3. Section 4 describes the sample selection. The results of empirical tests are presented in section 5. A summary and conclusion are provided in Section 6.
Literature Review
In response to the global financial crisis in 2008 and 2009, many countries announced the fiscal stimulus packages. Freedman et al. (2009) examine the effects of fiscal stimulus in five economic areas. Prasad and Sorkin (2009) provide an assessment of the stimulus plans in the G-20 countries. Naughton (2009) analyzes the composition of the Chinese stimulus package investment plan. Chan (2012) examines the response by China to the financial crisis at the global and regional levels.
Filip and Raffournier (2014) document that macroeconomic conditions are associated with the quality of accounting accruals. Using the data from public firms in the European Union, they find that accrual quality increases and earnings management decreases during the financial crisis period, 2008-2009. The association between accounting earnings and GDP is also examined by Konchitchki and Patatoukas (2014). They find that higher aggregate earnings growth is followed by higher growth in GDP.
The economic effect of stimulus package on the GDP of China is estimated by Diao et al. (2012) and McKissack and Xu (2011). By measuring the positive effect of the Chinese stimulus package, Diao et al. (2012) estimate that the annual growth rate of GDP in China could have been as low as 2.9% if the stimulus package were not provided. When 2.9% is compared to the actual GDP growth in 2009, 9.62% (World Bank, 2016), the positive effect of stimulus package on GDP growth rate can be interpreted as 6.72%. McKissack and Xu (2011) estimate the 2 to 3% positive impact of stimulus funding on the level of Chinese GDP in 2009 and 2010.
Earnings persistence is one of key indicators of earnings quality and affected by various financial characteristics of firms (Dechow et al., 2010). The components of earnings, such as accruals and cash flow, are reported to have different persistence in the prediction of future earnings (Sloan, 1996). Dechow and Dichev (2002) report that accrual quality is related with earnings persistence. Richardson et al. (2005) find that earnings persistence is lower for the firms with less reliable accruals. The earnings persistence is also affected by the magnitude of accruals (Dechow & Ge, 2006). Dechow et al. (2008) document that different components of cash report different persistence levels.
One of the factors that affect earnings persistence is the volatility of earnings. Dichev and Tang (2009) find that earnings persistence is higher for the firms that report lower earnings volatility. Test results support that understanding the volatility of earnings helps users better predict future earnings, as the adjusted R2 of the regression of future earnings on current earnings increases for the firms with lower earnings volatility (Dichev & Tang, 2009). Frankel and Litov (2009) introduce additional variables to control for other factors that might affect earnings persistence. They report that negative relationship between earnings volatility and persistence is found after controlling for additional variables, such as ROA change, firm size, the amount of accruals and the existence of loss firms. Clubb and Wu (2014) examine UK firms and document that the adjusted R2 of earnings prediction model is higher for the lower earnings volatility firms.
If the uncertainty caused by the economy level factors, such as the global financial crisis, can be effectively offset by the government provided stimulus funding, firms in that country will experience lower earnings volatility. As earnings become less volatile, earning persistence is expected to be higher in a country that introduces more effective economic stimulus packages during the period of global financial crisis, compared to the firms in another country in which stimulus packages were relatively less effective.
Fairfield et al. (2003) report that growth in net operating assets provides incremental information over current earnings in the prediction of future return on assets. In addition, they separate growth in net operating assets into two components, accruals and growth in long-term net operating assets, and find that both components provide incremental information about future profitability with negative coefficients. Kumar and Krishnan (2008) report that the relative importance of cash flows in firm valuation increases for the firms with more investment opportunities.
In this paper, we test whether and how effective government stimulus funding affects the persistence of earnings, accruals and cash flow.
Hypotheses and Research Design
Hypotheses
During a period of global financial crisis, unexpected economic factors affect the performance of firms and, accordingly, it becomes more difficult to predict the impact of such economic factors on the future performance of firms. As uncertainty increases about the future performance of firms, the ability of current accounting variables, such as earnings and cash flows, to predict future profitability will decrease and earnings persistence in general will be lower during the financial crisis period.
To mitigate the negative effects of financial crisis, governments provide stimulus funding. If stimulus funding works as planned, the negative impact of unexpected economic factors will be offset by such funding and uncertainty about future performance will decrease. Such offsetting effects will be greater in the countries where stimulus funding is more effective. Based on this discussion, the following hypotheses are developed as presented in alternative form.
Hypothesis 1: Earnings persistence is lower during a period of financial crisis compared to other periods.
Hypothesis 2: During a period of financial crisis, earnings persistence is higher for the firms operating in a country that provides more effective stimulus funding than for the firms operating in another country that provides relatively less effective stimulus funding.
Research Design
To test these hypotheses, the following regression models are applied:
Test of Hypothesis 1
Model 1: ROAt+1 = β0 + β1CrisisPeriodInd + β2ROAt + β3ROAt*CrisisPeriodInd + εt+1
Where,
ROA = Net income for the period / Average total assets
CrisisPeriodInd = 1, if fiscal year is 2008 or 2009, 0 otherwise
Model 1 tests whether the difference in earnings persistence between the financial crisis period and other periods is statistically significant. As earnings persistence is a measure of how closely current earnings are related to future earnings, β2 in the regression model 1 measures earnings persistence. β3 in model 1 measures the incremental earnings persistence during the financial crisis period. If the firms report lower earnings persistence during the financial crisis period, β3 in model 1 is expected to be negative and statistically significant.
Test of Hypothesis 2
Model 2: ROAt+1 = β0 + β1CHNInd + β2ROAt + β3ROAt*CHNInd + εt+1
Where,
ROA = Net income for the period / Average total assets
CHNInd = 1, if ISO Country Code of Incorporation = CHN, 0 otherwise
Model 2 tests whether the difference in earnings persistence between two countries is statistically significant. As the sample consists of manufacturing firms from China and Japan, β3 in model 2 measures the incremental earnings persistence for the firms in China as compared to the firms in Japan. If the firms in China report higher earnings persistence than those in Japan, β3 in model 2 is expected to be positive and statistically significant. Model 2 is tested for the financial crisis period (fiscal years 2008 and 2009) and other periods in the sample (fiscal years 2006-2007, 2010-2015).
Persistence of Cash Flow from Operating Activities and Accruals
Model 3: ROAt+1 = β0 + β1CHNInd + β2ACCOAt + β3ACCOAt*CHNInd + β4CFOAt + β5CFOAt*CHNInd + εt+1
Where,
ACCOA = Accruals for the period / Average total assets
Accruals = Net Income - Cash flow from operating activities
CFOA = Cash flow from operating activities for the period / Average total assets
In model 3, CFOA represents cash flow from operating activities. Accruals component of net income is calculated by subtracting cash flow from operating activities from net income. Previous research (Dechow & Dichev, 2002; Dechow & Ge, 2006; Dichev & Tang, 2009; Sloan, 1996) reported that cash flow and accruals components of earnings report different persistence.
The regression model 3 tests whether different earnings persistence between China and Japan during the financial crisis period is due to cash flow or accruals component of earnings.
Sample Firms
Sample Selection
The COMPUSTAT Global Fundamentals Annual Industrial Format as provided by the Wharton Research Data Services (WRDS) was used to select the firms in China and Japan. The ISO Country Code of Incorporation variable was used to identify the firms in China and Japan. To measure the effect on manufacturing firms, the firms in the manufacturing industries, defined as the SIC Codes between 2000 and 3999, were included in the sample.
For the fiscal years from 2006 to 2015, 33,624 firm-years reported either CHN or JPN as the country code of incorporation and the SIC Codes between 2000 and 3999. To apply the regression models described above, it was required to have all non-missing values of ROA for the period t+1 and ROA, CFOA and ACCOA for the current period, t. From this sample, 2,720 firm-years were excluded due to missing values for one or more of those variables. For the same variables, the 1st percentile and the 99th percentile were used as criteria to determine extreme values. 1,365 firm-years that reported the values lower than the 1st percentile or greater than the 99th percentile were excluded from the sample. Final sample included 29,539 firm-years for the fiscal years from 2006 to 2015 in the manufacturing industries of China and Japan.
Table 1 reports the breakdown of the sample by country and industry classification in Panel A. Firm-years by fiscal year and country are reported in Panel B of Table 1. 14,007 firm-years are from China and 15,532 firm-years are from Japan. All 29,539 firm-years are from the manufacturing industries with the SIC Codes from 2000 to 3999. Limiting sample to the firms in the manufacturing industries reduces the probability that test results are affected by different industry characteristics of the sample. 5,330 firm-years are from the financial crisis period.
Annual GDP Growth Rate
Annual GDP growth rates in China and Japan are reported in Table 2. GDP growth rates in Japan during 2008 and 2009 were -1.09% and -5.42%, respectively. Negative GDP growth rates provide the evidence that the government stimulus funding was not effective enough to offset the negative impact of financial crisis in Japan. In contrast, GDP growth rates in China during 2008 and 2009 were 9.65% and 9.40%, respectively. While GDP growth rates in China were also lower in 2008 and 2009 than compared to 2007, it is noted that China reported over 9% annual GDP growth rates in 2008 and 2009. This is contrasted with the negative annual GDP growth rates of Japan in 2008 and 2009. This is consistent with that the government stimulus funding was more successful in mitigating the negative impact of the global financial crisis in China than in Japan.
Descriptive Statistics
The mean and standard deviation of test variables are presented in Panels A and B of Table 3. Panel A of Table 3 shows that the mean return on assets for Japanese firms are significantly lower in 2008 and 2009 as compared to those of 2007. Mean return on assets is -0.0006 in 2008 which is a significant decrease from 0.0297 in 2007. Mean return on assets continues to be substantially lower in 2009 for Japanese firms. While the firms in China reported slight decreases in the mean return on assets as compared to those in 2007, the magnitude of decrease is noticeably smaller than for the firms in Japan. The mean return on assets for Chinese firms in 2008 is 0.0381, which represents 71.8% of 0.0531, the mean return on assets in 2007. For the firms in Japan, the mean return on assets in 2008 is -0.0006 as compared to 0.0297 mean return on assets in 2007.
The relatively smaller magnitude in the decrease in return on assets for Chinese firms may be explained as a result of the positive effects of stimulus funding provided by the Chinese government. Panel B of Table 3 reports that the mean cash flows from operating activities increased in 2008 and 2009 from the 2007 level for Chinese firms. Japanese firms reported a decrease in cash flow from operating activities in 2008 as compared to 2007.
Test Results
Test of Hypothesis 1
Table 4 reports the results of the regression model 1. When the regression model 1 is applied to all firm-years including both Chinese and Japanese firms, β3, the coefficient of ROAt*CrisisPeriodInd is negative (-0.129) and significantly different from zero at the 1% p-value level. Negative β3 of the regression model 1 provides evidence that earnings persistence is significantly lower during the financial crisis period compared to other periods in the sample. The results support the Hypothesis 1 that predicts lower earnings persistence during the global financial crisis period.
Test of Hypothesis 2
Table 4 also reports the results of the regression model 1 as applied to Chinese and Japanese firms, separately. For both Chinese and Japanese firms, β3, the coefficient of ROAt*CrisisPeriodInd is negative and significantly different from zero at the 1% p-value level. However, the coefficient is more negative for Japanese firms (-0.228) than for Chinese firms (-0.097). To formally test whether the coefficient for Japanese firms is significantly lower than the coefficient for Chinese firms in a statistical sense, the regression model 2 is applied. In the regression model 2, β3, the coefficient of ROAt*CHNInd captures the difference in the earnings persistence between Japanese and Chinese firms.
Table 5 reports the results of the regression model 2 as applied to the financial crisis period and other periods. β3, the coefficient of ROAt*CHNInd is positive (0.124) and significant at the 1% p-value level during the financial crisis period. The results are consistent with the Hypothesis 2. Next question about this result is whether higher earnings persistence for Chinese firms is due to the positive effects of stimulus funding provided by Chinese government. If higher earnings persistence is due to factors other than the effects of stimulus funding, similar results are expected be found in all periods, not only during the financial crisis period. However, the coefficient of ROAt*CHNInd is not significantly different from zero during other periods, including fiscal years 2006-2007 and 2010-2015. The higher earnings persistence of Chinese firms, as compared to Japanese firms is reported during the financial crisis period only. The results reported in Table 5 support that higher earnings persistence of Chinese firms is primarily due to the positive effects of stimulus funding, but not due to other structural differences between two countries.
The results reported in Tables 4 and 5 indicate that uncertain macro-economic conditions caused by the global financial crisis affect earnings persistence negatively. Earnings persistence during the financial crisis period is lower than that of other periods. Such negative impact of uncertainties can be offset by effective stimulus funding provided by the government. If the government of a country provides more effective stimulus funding, earnings persistence will be higher for the firms in the country compared to the firms in another country that provides less effective stimulus funding. The positive effects of stimulus package not only sustain higher growth in GDP, but also help firms report higher earnings persistence by reducing uncertainties caused by negative impact of the global financial crisis.
This study identifies the years 2008 and 2009 as a financial crisis period. However, the positive effects of government stimulus funding might have lasted beyond this period. Current test models do not capture the effects of stimulus funding after 2009, which is a limitation of this study. Testing the effects of stimulus funding beyond 2009 in different test models will be an extension of this study.
Persistence of Accruals and Cash Flow
Table 6 reports the results of the regression model 3 that tests the relative information contents of accruals and cash flow components of earnings. During the financial crisis period, the coefficients of both ACCOAt*CHNInd and CFOAt*CHNInd are positive and significantly different from zero. The results show that Chinese firms reported higher persistence in accruals and cash flow components of earnings than those of Japanese firms during the financial crisis period. Higher earnings persistence of Chinese firms during the financial crisis period is due to higher persistence of both accruals and cash flow components earnings.
Sensitivity Analysis
Table 7 reports the results of sensitivity analysis of the earnings persistence difference between Chinese and Japanese firms during the financial crisis period. In the regression model 4, additional variables were added to control for the effects of other factors that might affect earnings persistence. After controlling for additional variables, such as cash flow from operating activities, total revenue and the leverage ratio as measured by total liabilities divided by total assets, Chinese firms still report higher earnings persistence during the financial crisis period with the coefficient of ROAt*CHNInd as 0.134, significant at the 1% level.
Table 8 reports the results of the regression model 5 that allows the control variables to vary by country. Adding additional control variables in the regression model 5 did not affect the higher earnings persistence for Chinese firms during the financial crisis period. The results reported in Tables 7 and 8 support that higher earnings persistence of Chinese firms than that of Japanese firms during the financial crisis period is not sensitive to the additional control variables included in the regression model.
Summary and Conclusion
This paper examines whether macroeconomic conditions and policies of different countries affect the earnings persistence. To offset the negative impact of financial crisis, governments provided economic stimulus funding. The effects of stimulus packages on the economy varied depending on the macroeconomic conditions and the amount of funding. According to the World Bank data, GDP of China continued to grow in 2008 and 2009, while Japan experienced negative GDP growth rates during the same period.
We test whether such differences in the amounts and effects of stimulus funding between China and Japan affected the persistence of earnings, accruals and cash flows differently. If the economic structures have effects on the earnings persistence in these two countries, we expect the consistent earnings persistence pattern between China and Japan during all test periods. However, the test results in this study show that earnings persistence is higher for Chinese firms than for Japanese firms during the financial crisis period only.
We interpret that such results support that earnings of Chinese firms are more persistent during the financial crisis period, because of the positive effects of stimulus funding provided by Chinese government. Findings of this study suggest that understanding macroeconomic conditions and policies of different countries helps users better understand the roles of current accounting variables in the prediction of future profitability. The results of this paper suggest that managers and investors need to consider how macroeconomic policy at the economy level affects the financial analysis of manufacturing firms.