Fundamental Valuation in Seven Asian Countries: Role of Earnings,Book Value,and Dividends

Abstract

We examine the role of earnings, book value, and dividends in examining the valuation of firms in select Asian countries. Besides the usual variables of earnings and book value, inclusion of dividends is motivated by prior use of the variable in the literature, as well as an adaptation of the Ohlson 2001 empirical specification of the valuation model. In our specification, absent credible analysts’ forecasts, as is typical in these markets, dividends together with earnings play the role of “other information” in explaining stock prices. In a large sample of Asian firms from seven Asian countries that lack an active analyst community, we document two key results. First, the model with earnings, book value, and dividends outperforms the earnings capitalization, book value, and a model with earnings and book value together, the traditional benchmarks used in the literature. This is in contrast to Ashbaugh and Olsson, 2002 who find that earnings capitalization is the best model for the international firms. Second, the ability of the model to explain stock valuations does not vary materially over time, thus indicating reasonable consistency across different accounting regimes in these countries that may include International Financial Reporting Standards (IFRS) adaptation at different paces. Our finding highlights the information role of earnings and dividends when other channels of information are blocked.

Keywords

fundamental valuation value relevance earnings book value dividends Ohlson model Asian markets

Introduction

This study uses the insights from the Ohlson’s valuation model to explain stock prices in Asian markets where the information environment is generally less efficient. The role of book value and earnings in equity valuation has been established in literature both in U.S. (Collins et al., 1997) and in the international context¹ (Ashbaugh & Olsson, 2002; Clarkson et al., 2011; Graham & King, 2000). Although intuitive, these studies were done without any explicit reference to the valuation theory, more specifically, without any reference to the “other information” of the residual income valuation model (RIV).²

A key to this treatment of “other information” in the RIV Model is an assumption that the expectation of income is observable through the analysts’ forecasts. Ohlson (2001) shows that if the “other information” is captured by the analysts’ earnings forecasts, it would also become part of the valuation formula. However, for economies where such reliable analysts’ forecasts are not available, the fundamental analysis with valuation model faces a challenge. In this study, we offer some conjecture as to how the dividends may fulfill some of the role of “other information” in such settings. Through our conjectured relationship, we offer some evidence that (a) dividend has some explanatory power beyond book value and earnings in a price level regression, and (b) the role of dividend is somewhat consistent with the notion that the dividend plays an informational role. Whether dividends can offer a comprehensive proxy for “other information” is a subject for further research.

We use the dividend as an additional explanatory variable and offer three possible reasons why the dividend may empirically enter the valuation specification. The first one is to refer to the literature and the construction used by studies such as Hand and Landsman (2005) and Akbar and Stark (2003), who have used dividend as a proxy for “other information” in the basic valuation model. According to these studies, information role of dividend is evident even in the U.S. market where there are many sources of other information, including the analysts’ forecasts.³ The second one is to acknowledge that in the Ohlson (2001) valuation formula, in spite of allowing for the “other information” through the analysts’ forecasts, the current dividend appears in the equation because of its earnings displacement property (Penman & Sougiannis, 1997).⁴ Finally, we also offer an extension of the Ohlson (2001) valuation model that may apply in the case of the Asian countries where consistent and reliable analyst’ earnings forecasts are rare and often do not exist.⁵ In our extension of the valuation model, dividends as well as past earnings are used for earnings forecasts and therefore, play an additional informational role in firm valuation.⁶ Although dividend in our model plays an informational role, it is quite different from that of dividend signaling where the cost associated with dividend payout conveys information. All these possible explanations lead to a positive coefficient for the dividend variable in the valuation equation. Our results verify this assertion.

To be consistent with the earlier literature, we revisit the study done by Ashbaugh and Olsson (2002) (AO hereafter) that investigates the accounting-based valuation of international firms. AO found that the earnings capitalization model is the most descriptive model for non-U.S./non-U.K. firms reporting under International Accounting Standards. We conduct a country-by-country analysis for seven Asian countries (detailed later) with a larger sample to estimate the country-specific valuation parameters. To facilitate comparison, our study uses the Earnings and Book value (Ohlson) model⁷ and our extended formulation based on Ohlson (2001) with dividends, and shows that our extended model (Ohlson + D) with dividends is the most descriptive valuation model across countries and years.

We conduct our analysis with and without the Consumer Price Index (CPI) adjustments that may address the bias in accounting numbers due to inflation over time. To neutralize the effect of inflation, we scale our data by the country CPI index of 2010, set as the base year, and take as 100.0.⁸ Having brought the entire data to the 2010 price level, we now apply the 2010 year-end exchange rate to translate all data to U.S. dollars to facilitate comparison of coefficients.

Although our main focus is the explanatory power of each variable (the model R²), we also attempt to examine the coefficients of the explanatory variables. The coefficients and their weights in Ohlson models are supposed to capture variations in accounting policy choices in these countries. Migration to International Financial Reporting Standards (IFRS) by different countries is conducted with different speed and with different interpretation and adaptations. If IFRS is implemented with too much variation, then the valuation model may demonstrate vastly different weights on the different accounting variables and may exhibit such variation over our time period. Our results show that in spite of possible variations in the implementation of IFRS in different countries, the valuation coefficients of the earnings and book value variables are not that much different for the firms in different countries.⁹

Using the average data of the government bond interest rate for all the countries during our estimation period, we attempt to interpret the coefficients for the pooled sample for all countries and all years in terms of some assumed numbers for the basic valuation parameters. This is to verify if the observed coefficients are consistent with some reasonable parameter values of the basic information variables in the theoretical model. Our example indicates that the fundamental parameters of the persistence of the residual income and the new information (innovation in earnings) are consistent with observed coefficients and may be quite stable across countries. Taken together, our results seem to indicate that regardless of implementation variations, these seven countries are adopting IFRS to create, by and large, comparable financial statements. We also observe that the time pattern of the valuation parameters within our sample period is quite stable, indicating that whatever changes these countries may be adopting in the accounting policy, their adoption of IFRS did not materially affect the valuation parameters across time, affecting the implications and relevance of the valuation models in any fundamental way.

The rest of the study is organized as follows: “Analytical Background and Development of the Model” section outlines the analytical development of the model. “Research Design and Sample” section has the research design and the sample. The results are presented in “Results” section. “Numerical Example” section discusses the parameter estimates and the numerical examples, and “Conclusions and Discussions” section concludes the study.

Analytical Background and Development of the Model

A Returns-Earnings Specification With Analysts’ Forecast as “Other Information”

The residual income model with a first-order auto-regressive process [AR(1)] to characterize the residual income sequence with ω and γ as the persistence parameter and ν, the new “other information” that will influence the future residual income, can be restated in terms of book value and earnings, as follows^10,11:

P_{t} = b_{t} + α_{1} x_{t}^{a} + α_{2} v_{t}

(A-1a)

where

v_{t + 1} = γ v_{t} + {\tilde{ε}}_{2 t + 1}

{\tilde{x}}_{t + 1}^{a} = ω x_{t}^{a} + v_{t} + {\tilde{ε}}_{1 t + 1}

Pt = market value of equity at date t

x_{t}^{a} = x_{t} - (R - 1) b_{t - 1} = Residual Income

b_t = book value at date t

x_t = earnings for the period (t − 1, t)

d_t = net dividend paid at date t

R = 1 + r = the discount rate plus one.

With $v_{t} = E_{t} [x_{t + 1}^{a}] - ω x_{t}^{a} = {\bar{x}}_{t}^{at + 1} - ω x_{t}^{a}$ as can be obtained from the above assumption,

The price equation (A-1a) becomes

P_{t} = b_{t} + (α_{1} - α_{2} ω) x_{t}^{a} + α_{2} {\bar{x}}_{t}^{at + 1}

(A-1b)

or, equivalently,

P_{t} = (1 - k) b_{t} + k (ϕ x_{t} - d_{t}) + α_{2} v_{t}

(A-1c)

where

α_{1} = ω / (R - ω) \geq 0

α_{2} = R / (R - ω) (R - γ) > 0

k = r ω / (R - ω) \geq 0

ϕ = R / r

With both ω and γ less than 1, the above model depicts a decay model of residual income with external shocks from other information that also decays over time. With ω greater than one, it will be a growth process.

Lemma 1: Let $v_{t}$ be defined by $v_{t} = α_{1} x_{t} + α_{2} d_{t} + α_{3} b_{t}$ , that is, the new information is conveyed only by the past financial statement components in some linear combination.

Then,

α_{1} = λ - ω (1 + r)

α_{2} = μ + ω r

α_{3} = - (1 - ω) r

are consistent with the earnings forecast

x_{t + 1} = λ x_{t} + μ d_{t}

(A-2)

where λ and μ are the information parameters of earnings and dividends, respectively.

Proof: See Appendix A.

Lemma 2: The valuation equations (A-1b) and (A-2) yield a valuation function given by:

P_{1} = β_{1} b_{t} + (β_{2} ϕ + \frac{β_{3}}{r} λ) x_{t} + (\frac{μ β_{3}}{r} - β_{2}) d_{t}

(A-3)

where

β_{1} = R (1 - ω) (1 - γ) / (R - ω) (R - γ)

β_{2} = r ω γ / (R - ω) (R - γ)

β_{3} = Rr / (R - ω) (R - γ)

Proof: See Appendix A.

Empirically we estimate the following empirical equation (intercept $α$ used but not represented for parsimony)

P_{1} = θ_{1} b_{t} + θ_{2} x_{t} + θ_{3} d_{t}

where

θ_{1} = β_{1}

θ_{2} = (β_{2} ϕ + \frac{β_{3}}{r} λ)

and

θ_{3} = (\frac{μ β_{3}}{r} - β_{2})

With reasonable assumptions and restrictions on the parameters, the coefficients should be positive. However, as these parameters are unobservable, though likely to be positive, we cannot put any firm expectation on the sign on the coefficient of the dividend term.¹²

Research Design and Sample

Research Design

The conclusions of the AO study are based on a sample of 26 international firms of which only eight are from three different Asian countries (four from Hong Kong, three from Japan, and one from China). To complement the AO study, we choose the Asian countries that have at least 500 listed firms, except China, India, and Japan. Japan is a developed country and informational environment in Japan is comparable to the other developed countries. Both the Chinese and the Indian markets have their own idiosyncrasies that may merit an independent treatment and hence left out from our sample. This screening leaves us with seven countries, Hong Kong, Korea, Singapore, Indonesia, Malaysia, Taiwan, and Thailand, for which data are available and all of whom have adopted IFRS, albeit at different dates. From an empirical standpoint, our study offers a more robust formulation over the AO study. One reason is that our study considers a later time period when more data are available that allows us to examine more than 15,000 firm-years from seven Asian countries from 1998 to 2017, a 20-year period.¹³

Our first purpose is to examine the relative performance of the earnings capitalization (E), book value (BV), and the Ohlson (BV+E) models in seven Asian countries that use IFRS as the basis for their national accounting standards. We adopt the standard econometric method of ordinary least squares (OLS) to address the following question.

Research Question 1: Which of the following three models is the dominant accounting-based valuation model in Hong Kong, Indonesia, Korea, Malaysia, Singapore, Taiwan, and Thailand with the OLS estimates?

(E) P_{it} = α + θ_{1} EP S_{it} + ε_{it}

(Model 1)

(BV) P_{it} = α + θ_{1} BP S_{it} + ε_{it}

(Model 2)

(BV + E) P_{it} = α + θ_{1} BP S_{it} + θ_{2} EP S_{it} + ε_{it}

(Model 3)

where P_it = market price per share of firm i 3 months after the end of accounting period t; EPS_it = earnings per share of firm i for accounting period t; and BPS_it = book value of equity per share of firm i at the end of accounting period t.

The E model expresses price as a multiple of earnings generated from net assets.¹⁴ Collins et al. (1999) present evidence that the E model is mis-specified because of omitting BV: the coefficient on earnings is negatively biased for loss firms and positively biased for profit firms when BV is omitted. However, we retain the E model because of the findings of Ashbaugh and Olsson (2002).

The BV model is based on the idea that price and book value are both measures of the value of stockholders’ equity. Price is the market value and BV is the accounting measure of the net assets (i.e., stockholders’ equity) that generate earnings. Easton and Harris (1991) show that a regression of stock returns on (U.S.) GAAP earnings levels is conceptually the same as a regression of price on book value.

The Ohlson model (BV+E) is based on the Ohlson (1995) Residual Income (RI) model as explained in the “Analytical Background and Development of the Model” section. In this model, the BV becomes a valuation anchor or future expected “normal” earnings analogous to that of a savings bank.¹⁵ The direct application of the RI model requires an estimation of the discount rate, r. Empirical specification of the RI model is fraught with the limitation of the assumed interest rate and/or the cost of capital. Although prior research finds that the choice of discount rate does not affect cross-sectional analyses of the RI model (Francis et al., 2001; Frankel & Lee, 1998), we stay away from any interest rate assumption during our empirical test by using the (BV + E) specification. Given the results of Jan and Ou (1995) and Collins et al. (1999) cited above, we expect the (BV+E) model to dominate either the E or the BV model.¹⁶

The second research question is to consider the role of dividend in accounting-based valuation models. As our extension of the model shows, dividends as well as current earnings may fulfill the role of other information in our characterization. Therefore, in addition to the (BV+E) model, inclusion of dividend allows us to focus on the dividend as an additional explanatory variable. Dividends alone could simply proxy for the “other information” term in the Ohlson model as shown by Hand and Landsman (2005) and Akbar and Stark (2003). Based on these logics, this study compares results from the first section with the model with other information (BV+E+D, hereafter Ohlson+D) to see how the other information component contributes to the explanation of firm valuation. Thus, the second question follows:

Research Question 2: After including dividends, does Ohlson model with other information (Ohlson+D) perform better than the best model identified in Research Question 1 for Hong Kong, Indonesia, Korea, Malaysia, Singapore, Taiwan, and Thailand under OLS estimation method?

We use the following additional model to answer Research Question 2:

(Ohlson + D) P_{it} = α + θ_{1} BP S_{it} + θ_{2} EP S_{it} + θ_{3} DP S_{it} + ε_{it}

(Model 4)

where P_it = market price per share of firm i 3 months after the end of accounting period t; EPS_it = earnings per share of firm i for accounting period t; BPS_it = book value of equity per share of firm i at the end of accounting period t; and DPS_it = dividend per share of firm i for accounting period t.

The explanatory power (adjusted R²) of each model in OLS measures the strength of the association between the accounting variable(s) and price that we report. The adjusted R² from each of the models represents relative value relevance. However, to draw a conclusion on individual valuation variables, we need to calculate incremental explanatory power of valuation variables. Our study is the first to introduce three accounting variables (book value, earnings, and dividend) in one value relevance model, which is Model (4). Appendix B provides concept (Panel A) and an example of the calculation of Incremental Explanatory Power of Valuation Variables using a Singapore case for the period 2013–2017 (Panel B).

To neutralize the effect of inflation, we conduct our analysis with the Consumer Price Index (CPI) adjustments by scaling our data by the country CPI index of year 2010. We believe that a constant dollar measurement may address any possible bias introduced to the parameter estimates by the currency fluctuations, which, of course, is not uniform in these countries. This leads to our third research question.

Research Question 3: What are the properties of the valuation parameters in all these valuation models for Hong Kong, Indonesia, Korea, Malaysia, Singapore, Taiwan, and Thailand under OLS estimation method?

For the purpose of analyzing the properties of the parameter estimates, we use the average of the coefficients across the 20-year time period. Just for the reason of robustness, we also run a combined regression for the pooled data. We also compare the properties of those parameters to see if our conclusions are robust to the estimation methods. For the test of stability over time, we plot the average of the parameter estimates across time and observe more or less a flat pattern. Limitations of sample size do not allow us to investigate this aspect statistically in a more rigorous way. To relate the observed valuation coefficients to theory, we construct a numerical example and verify that the observed coefficients from the pooled regression may be consistent with some “reasonable” parameter values.

Sample

Our sample covers publicly traded firms in Hong Kong, Indonesia, Korea, Malaysia, Singapore, Taiwan, and Thailand across the period 1998–2017. We chose these countries because these are the only Asian countries (besides China, India, and Japan), which have at least 500 firms listed in their stock exchanges and been ranked among the top 21 market capitalization of the world. We exclude these three countries because China and India have their own idiosyncratic issues with respect to their capital markets and information environments, and Japan is a developed economy and has a comparable information environment with other developed countries. Online Appendix C (reposted on the journal website) indicates that our selected countries, except Taiwan (as the data is not available), represent approximately 18.4% of the world market capitalization (excluding the U.S. market) and are ranked 4th for Hong Kong, 9th for Korea, 17th for Singapore, 19th for Thailand, 20th for Indonesia, and 21st for Malaysia. The stock price and accounting data for this study are from the WRDs Compustat Global Database. The consumer price index and the bond interest rate data have been obtained from the World Bank sources and the Thomson Reuter’s Datastream database, respectively.

We restrict our sample to non-financial firms that report earnings, book value, and dividend and whose shares trade in these seven Asian capital markets. To be included in the sample, a company must contain at least 10 firm-years and have information on price, earnings, book value, and dividend on a per share basis. Observations falling in the top or bottom 1% of price, earnings, book value, and dividend per share in each country are excluded to reduce the effects of outliers on the regression results. As shown in Table 1, Panel A, our final sample gives us a total of 15,458 firm-year observations for each valuation model: 1,886 firm-year for Hong Kong; 2,721 firm-years for Indonesia; 2,323 firm-years for Korea; 3,027 firm-years for Malaysia; 2,407 for Singapore, 3,094 for Taiwan; and 3,145 firm-years for Thailand. We use the 2010 exchange rates (Table 1, Panel B) to convert each variable to 2010 U.S. dollars. Table 1, Panel C reports the Consumer Price Index (CPI) for each sample year for the seven sample countries. We use these CPIs to bring each country data for each year to the year 2010 level. Next, Table 1, Panel D shows the interest rates used for simulation of the theoretical coefficients.

Table 1.

Final Sample, U.S. Exchange Rate, CPI, and Government Bond Interest Rate.

Panel A: Number of Firms (Final Sample).
Year	Hong Kong	Indonesia	Korea	Malaysia	Singapore	Taiwan	Thailand	TOTAL
1998	64	59	44	90	53	12	68	322
1999	69	62	49	97	63	17	79	357
2000	73	64	56	102	71	28	79	394
2001	77	73	82	113	84	39	88	468
2002	82	85	84	125	90	90	94	556
2003	82	101	90	144	101	116	118	634
2004	84	108	96	155	113	128	142	684
2005	86	110	102	164	125	147	158	734
2006	91	124	111	166	133	188	167	813
2007	101	150	128	178	143	199	180	899
2008	103	139	146	176	147	204	187	915
2009	107	175	138	181	148	214	204	963
2010	109	191	145	182	155	220	206	1,002
2011	109	187	165	176	150	219	201	1,006
2012	108	186	160	168	147	214	201	983
2013	109	182	149	167	145	215	197	967
2014	109	183	148	165	140	217	198	962
2015	109	183	143	161	139	213	198	948
2016	108	183	144	158	132	209	192	934
2017	106	176	143	159	128	205	188	917
	1,886	2,721	2,323	3,027	2,407	3,094	3,145	15,458

Panel B: U.S. Exchange Rate of 1 Dollar.
Year	Hong Kong	Indonesia	Korea	Malaysia	Singapore	Taiwan	Thailand
2010	7.77	8930.00	1182.10	3.19	1.36	31.9900	32.22

Panel C: Consumer Price Index, CPI [2010 = 100.00].
The data are reposted on the journal website

Panel D: Summary of Government Bond Interest Rate during 1998–2017.
The data are reposted on the journal website

Source. http://www.Xrate.com.

Table 2 shows descriptive statistics on price, earnings, book value, and dividend per share by country. On average, firms’ shares trade above book value (i.e., mean share price exceeds mean book value). The exceptions are Korea and Singapore while Malaysia’s mean share price and book value are roughly equal. Earnings are positive on average.

Table 2.

Sample Descriptive Statistics.

Country	Variables	N	M	SD	Minimum	Maximum
Hong Kong	PPS	1,886	1.534	2.785	0.001	25.459
	EPS	1,886	0.126	0.268	−1.198	3.273
	BPS	1,886	1.452	2.590	−0.195	26.177
	DPS	1,886	0.046	0.097	0.000	1.061
Indonesia	PPS	2,721	0.180	0.429	0.003	7.860
	EPS	2,721	0.002	0.138	−3.956	0.873
	BPS	2,721	0.085	0.235	−3.706	2.146
	DPS	2,721	0.001	0.004	0.000	0.108
Korea	PPS	2,323	2.152	3.567	0.018	35.613
	EPS	2,323	0.121	0.684	−18.307	7.544
	BPS	2,323	2.589	4.011	−0.367	42.891
	DPS	2,323	0.035	0.083	0.000	1.039
Malaysia	PPS	3,027	0.563	0.664	0.006	6.300
	EPS	3,027	0.029	0.104	−1.626	1.217
	BPS	3,027	0.568	0.458	−1.252	4.463
	DPS	3,027	0.012	0.022	0.000	0.254
Singapore	PPS	2,407	1.035	2.107	0.001	19.770
	EPS	2,407	0.081	0.231	−2.508	2.862
	BPS	2,407	1.097	1.890	−0.104	14.567
	DPS	2,407	0.025	0.060	0.000	0.852
Taiwan	PPS	3,094	0.837	0.872	0.046	8.252
	EPS	3,094	0.045	0.086	−0.769	0.644
	BPS	3,094	0.553	0.273	0.035	3.472
	DPS	3,094	0.007	0.025	0.000	0.337
Thailand	PPS	3,145	0.665	1.204	0.001	15.979
	EPS	3,145	0.034	0.194	−6.094	2.033
	BPS	3,145	0.502	0.863	−3.409	8.153
	DPS	3,145	0.009	0.040	0.000	0.774

Note. All variables above are converted to U.S. dollar currency using the exchange rate of July 31, 2010, and adjusted by its county’s consumer price index, CPI in the same period. PPS = the price per share 3 months after fiscal year end; EPS = the earnings (net income before extraordinary items) per share at fiscal year; BPS = the book value per share at fiscal year; DPS = the dividend per share at fiscal year.

Table 3 reports the Pearson and Spearman correlation matrix results of all variables and shows that PPS and other accounting variables are highly correlated.

Table 3.

Correlation Analysis of Pooled Data.

	PPS	EPS	BPS	DPS
PPS		0.712	0.555	0.891
		(***)	(***)	(***)
EPS	0.817		0.774	0.689
	(***)		(***)	(***)
BPS	0.865	0.808		0.518
	(***)	(***)		(***)
DPS	0.785	0.779	0.761
	(***)	(***)	(***)

Note. This table shows Pearson correlation results on the upper section and Spearman correlation results on the lower section.

***

, **, * are significant levels with two-tailed tests at p < .01, p < .05, and p < .10, respectively.

To facilitate summary reference, the four valuation models and the variable definitions are reproduced in Table 4.

Table 4.

Accounting-Based Valuation Models.

Model 1: Earnings Capitalization (E) PPS_it = α+θ₁EPS_it+ϵModel 2: Book Value (BV) PPS_it = α+θ₁BPS_it+ϵ_itModel 3: Ohlson (Ohlson) PPS_it = α+θ₁BPS_it+θ₂EPS_it+ϵ_itModel 4: Ohlson with other information (Ohlson+D) PPS_it = α+θ₁BPS_it+θ₂EPS_it+θ₃DPS_it+ϵ_itModel 5: Dividend (D) PPS_it = α+θ₁DPS_it+ϵ

Note. All variables above are converted to U.S. dollar currency using the exchange rate of July 31, 2010, and adjusted by its county’s consumer price index, CPI in the same year. Although the dividend model is not our focus for this study, we provide some results based on this model for other purposes including the incremental value relevance calculation of main variables of interest (E, BV, and D) and a summary of dominant models.

Results

Identification of the Most Descriptive Accounting-Based Model

At first, we identify which of the valuation models is the most descriptive model in capturing the firm value. This is our Research Question 1. Table 5 reports for each country and each subperiod of 5 years, the adjusted R² (R² hereafter) from our four empirical valuation models (M1–M4). The model with the maximum R² becomes the most descriptive or the dominant model for that period.

Table 5.

Adjusted R² for All Valuation Models.

	Model 1	Model 2	Model 5	Model 3	Model 4
Adjusted R²	E	BV	D	Ohlson	Ohlson+D	Maximum	1st
Hong Kong
Average	0.558	0.372	0.774	0.560	0.817	0.817	Ohlson+D
Pooled	0.506	0.307	0.795	0.506	0.813	0.813	Ohlson+D
Indonesia
Average	0.307	0.236	0.039	0.339	0.239	0.339	Ohlson
Pooled	0.035	0.155	0.017	0.164	0.163	0.164	Ohlson
Korea
Average	0.186	0.281	0.354	0.346	0.426	0.426	Ohlson+D
Pooled	0.124	0.307	0.409	0.350	0.483	0.483	Ohlson+D
Malaysia
Average	0.296	0.333	0.449	0.422	0.579	0.579	Ohlson+D
Pooled	0.190	0.343	0.433	0.398	0.577	0.577	Ohlson+D
Singapore
Average	0.479	0.636	0.434	0.685	0.715	0.715	Ohlson+D
Pooled	0.476	0.608	0.401	0.661	0.685	0.685	Ohlson+D
Taiwan
Average	0.380	0.308	0.068	0.452	0.466	0.466	Ohlson+D
Pooled	0.400	0.326	0.108	0.477	0.491	0.491	Ohlson+D
Thailand
Average	0.489	0.436	0.152	0.567	0.581	0.581	Ohlson+D
Pooled	0.174	0.399	0.120	0.447	0.449	0.449	Ohlson+D

Note. 1st = the dominant model (highest explanatory power) among the five models. Average = the average of adjusted R² amounts received from each of the four subperiods of 1998–2002, 2003–2007; 2008–2012, and 2013–2017, respectively. Pooled = the adjusted R² amount received from pooled sample during 1998–2017. Full results on each of the 5-year periods are reposted on the journal website.

Based on Average and Pooled results of Table 5, it is important to note that without considering the Ohlson+D model (M4), among the accounting-based models, the Ohlson Model (M3) performs best among the remaining models. When considering the Ohlson+D model (M4), the results indicate that the Ohlson+D model performs best for all countries except Indonesia where the Ohlson Model (M3) still performs best. It is the best model in 4 out of 4 periods in Hong Kong, Korea, Malaysia, and Singapore, 3 out of 4 periods in Thailand, 2 out of 4 periods in Taiwan, and 1 out of 4 periods in Indonesia.

The Role of Dividends in the Valuation Models

To further investigate the incremental contribution of each valuation variable, we use the schematic presentation in Appendix B, Panel A, and run a series of regression tests as detailed in the Appendix B, Panel A. By following the procedure of recursive substitution and subtraction as detailed in Appendix B, Panel A, we calculate the incremental explanatory powers of the three explanatory variables. Example of calculation can be found in Appendix B, Panel B.

As Table 6 outlines, based on the mean results, the incremental explanatory power of the dividend variable (IncrD) is higher than that of the E or the BV variable for Hong Kong, Korea, and Malaysia. We also compare the average incremental explanatory power of the annual regression with that of the pooled sample, reported at the bottom of the panels. The results based on the pooled data are consistent. For example, for Hong Kong, the average incremental R² for Earnings, Book Value, and Dividends are 0.016, 0.003, and 0.257, respectively. The corresponding numbers for the pooled sample is 0.007, 0.000, and 0.307, respectively. By both measures, the dividend variable seems to explain about twice or more the amount of variance than that of the earnings variable in Hong Kong, Korea, and Malaysia. For Singapore, the dividend variable explains about the same amount of variation as the earnings variable, while for Taiwan and Thailand, the dividend variable explains less as compared with the amount explained by the earnings variable.

Table 6.

Incremental Adjusted R² for Each Variable [Ohlson+D Model].

	M3: Ohlson	M4: Ohlson+D	M6	M7	M4 − M7	M4 − M6	M4 − M3	M4 − Incr [E, BV, D]
Adjusted R²	E, BV	E, BV, D	E, D	BV, D	IncrE	IncrBV	IncrD	Common
Hong Kong
Average	0.560	0.817	0.813	0.800	0.016	0.003	0.257	0.540
Pooled	0.506	0.813	0.813	0.806	0.007	0.000	0.307	0.499
Indonesia
Average	0.339	0.239	0.318	0.242	(0.004)	(0.080)	(0.100)	0.423
Pooled	0.164	0.163	0.047	0.155	0.008	0.116	(0.000)	0.039
Korea
Average	0.346	0.426	0.378	0.413	0.014	0.048	0.081	0.284
Pooled	0.350	0.483	0.423	0.474	0.009	0.060	0.133	0.281
Malaysia
Average	0.422	0.579	0.509	0.560	0.018	0.070	0.157	0.334
Pooled	0.398	0.577	0.471	0.567	0.010	0.106	0.179	0.282
Singapore
Average	0.685	0.715	0.593	0.686	0.029	0.122	0.030	0.534
Pooled	0.661	0.685	0.568	0.651	0.034	0.117	0.023	0.511
Taiwan
Average	0.452	0.466	0.399	0.336	0.130	0.067	0.014	0.256
Pooled	0.477	0.491	0.428	0.355	0.135	0.062	0.014	0.279
Thailand
Average	0.567	0.581	0.520	0.475	0.106	0.061	0.014	0.400
Pooled	0.447	0.449	0.241	0.403	0.046	0.208	0.002	0.193

Note. For calculation, we ran valuation models with D only (Model 4), E+D (Model 6), and BV+D (Model 7). But these results are not reported for coefficients and adjusted R² tables. Although Models 4, 6, and 7 are not our focus for this study, we provide some results based on these model for other purposes including the incremental value relevance calculation of main variables of interest (E, BV, and D). Here are Models 6 and 7:

Model 6: Earnings (E) + Dividend (D).

PPS_it = α+θ₁EPS_it+θ₂DPS_it+ϵ.

Model 7: Book Value (BV) + Dividend (D).

PPS_it = α+θ₁BPS_it+θ₂DPS_it+ϵ.

M4 is the total value relevance. Incremental value relevance of E equals M4 − M7. Incremental value relevance of BV equals M4 − M6. Incremental value relevance of D equals M4 − M3. Common is the difference between the total value relevance (M4) and the sum of all incremental value relevance (E, BV, and D).Full results on each of the 5-year periods are reposted on the journal website.

We present in Table 7 the proportion of variation explained by each variable out of the total explainable model R² that are not explained by the common effect. This allows us to normalize across the total R² across different countries and see how the proportions of variances explained by each variable change across different countries. The explanatory powers of the different variables are remarkably stable for each country across the different time periods. The pooled row refers to the explanatory powers from the pooled data of all firm-years in each of the countries. The Average row is calculated using the mean of the four subperiods. Based on the average of the four subperiods, we see that the range of variation explained by the dividend variable is between 86.1% (Hong Kong) and less than 10% (Thailand and Taiwan). This evidence points to the differential informativeness of dividends in different countries. The earnings variable explains between 58.9% (Taiwan) and 7.8% (Malaysia) of the variations and the book value variable explains between 67.1% (Singapore) to 1.8% (Hong Kong) of the variations.

Table 7.

Contribution to Incremental Adjusted R² for Each Variable [Ohlson+D Model].

	Adjusted R²			Incremental adjusted R²			% Incremental adjusted R²
Country	TOTAL [A]	Common [B]	[C] = A − B	E	BV	D	E	BV	D
Hong Kong
Average	0.817	0.540	0.277	0.016	0.003	0.257	0.121	0.018	0.861
Pooled	0.813	0.499	0.315	0.007	0.000	0.307	0.022	0.001	0.976
Indonesia
Average	0.239	0.423	−0.184	(0.004)	(0.080)	(0.100)	0.149	0.359	0.492
Pooled	0.163	0.039	0.124	0.008	0.116	0.000	0.066	0.936	−0.002
Korea
Average	0.426	0.284	0.143	0.014	0.048	0.081	0.088	0.351	0.561
Pooled	0.483	0.281	0.202	0.009	0.060	0.133	0.045	0.296	0.659
Malaysia
Average	0.579	0.334	0.245	0.018	0.070	0.157	0.078	0.298	0.623
Pooled	0.577	0.282	0.295	0.010	0.106	0.179	0.036	0.359	0.606
Singapore
Average	0.715	0.534	0.181	0.029	0.122	0.030	0.162	0.672	0.166
Pooled	0.685	0.511	0.174	0.034	0.117	0.023	0.194	0.671	0.135
Taiwan
Average	0.466	0.256	0.210	0.130	0.067	0.014	0.589	0.334	0.077
Pooled	0.491	0.279	0.212	0.135	0.062	0.014	0.638	0.294	0.067
Thailand
Average	0.581	0.400	0.181	0.106	0.061	0.014	0.566	0.359	0.076
Pooled	0.449	0.193	0.257	0.046	0.208	0.002	0.179	0.811	0.010

Note. All the data are from Table 6. We use them to calculate the percentage of each of the incremental value relevance [E, BV, and D] as compared with the total contribution of incremental value relevance. The total contribution of incremental value relevance [C] is the difference between the TOTAL [A] and the Common [B]. Full results on each of the 5-year periods are reposted on the journal website.

The results on the pooled regressions of the four subperiods are consistent for dividend results, but somewhat different for earnings and book value. In other words, we observe that the range of variation explained by the dividend variable is between 97.6% (Hong Kong) and less than 7% (Thailand and Taiwan, but Indonesia provides a negative value). The earnings variable explains between 63.8% (Taiwan) and 2.2% (Hong Kong) of the variations and the book value variable explains between 67.1% (Singapore) to 0.1% (Hong Kong) of the variations.

Robustness: Identification of an Accounting-Based Dominant Model

To confirm our findings in the “Identification of the Most Descriptive Accounting-Based Model” section, we re-run our regressions to obtain the explanatory power of the four valuation models (E, BV, Ohlson, and Ohlson+D) by country-years. Table 8, Panel A provides the list of dominant models across 20 years for each of the countries (the data across years are not reported) and shows a result summary indicating the Ohlson+D model is overwhelmingly the dominant model for all the countries across the years. The average on the bottom line of this panel indicates the annual averages of the adjusted R² from the 20 years (yearly values of R² are not reported) and it confirms that Ohlson+D model is the dominant model for all countries with the average adjusted R² of 85.1% (highest) for Hong Kong to 38.9% (lowest) for Indonesia.

Table 8.

Summary of Dominant Models across Countries and Years.

Panel A: List of Dominant Models.
Year	Dominant models across years
	Hong Kong	Indonesia	Korea	Malaysia	Singapore	Taiwan	Thailand
1998	Ohlson+D	BV	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson	BV
1999	Ohlson+D	Ohlson	BV	Ohlson+D	Ohlson	Ohlson	Ohlson+D
2000	Ohlson+D	BV	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson	Ohlson+D
2001	Ohlson+D	Ohlson	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson	Ohlson
2002	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson	Ohlson
2003	Ohlson+D	Ohlson+D	D	Ohlson+D	Ohlson+D	Ohlson	Ohlson
2004	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson	Ohlson	Ohlson+D
2005	Ohlson+D	E	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson
2006	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson	Ohlson	Ohlson+D
2007	Ohlson+D	Ohlson	Ohlson	Ohlson+D	Ohlson+D	Ohlson	Ohlson+D
2008	Ohlson+D	Ohlson	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson	Ohlson
2009	Ohlson+D	Ohlson	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D
2010	Ohlson+D	E	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson	Ohlson+D
2011	Ohlson+D	Ohlson	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson
2012	Ohlson+D	Ohlson	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D
2013	Ohlson+D	Ohlson	Ohlson+D	Ohlson+D	Ohlson	Ohlson+D	Ohlson+D
2014	D	Ohlson	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D
2015	Ohlson+D	E	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D
2016	D	Ohlson	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D
2017	D	Ohlson	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D
Average	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D	Ohlson+D
[Adj. R²]	[0.851]	[0.389]	[0.487]	[0.601]	[0.762]	[0.520]	[0.634]

Panel B: Number and Percentage of Dominant Models.
	Number of dominant model
Model	Hong Kong	Indonesia	Korea	Malaysia	Singapore	Taiwan	Thailand
Ohlson+D	17	4	17	20	16	9	13	Years
Ohlson	0	11	1	0	4	11	6	Years
BV	0	2	1	0	0	0	1	Years
E	0	3	0	0	0	0	0	Years
D	3	0	1	0	0	0	0	Years
	20	20	20	20	20	20	20	Years
Ohlson+D%	85%	20%	85%	100%	80%	45%	65%
Ohlson%	0%	55%	5%	0%	20%	55%	30%
BV%	0%	10%	5%	0%	0%	0%	5%
E%	0%	15%	0%	0%	0%	0%	0%
D%	15%	0%	5%	0%	0%	0%	0%
	100%	100%	100%	100%	100%	100%	100%

Table 8, Panel B provides the number and percentage of dominant models based on the sum of dominating years of the models. Out of a total of 140 estimations across the countries and years, 96 out of 140 trials (or 69%) yield Ohlson+D as the dominant model. The results indicate that the Ohlson+D model is the dominant model for Malaysia (100%), Hong Kong and Korea (85%), Singapore (80%), and Thailand (65%). Interestingly, the basic Ohlson model is the dominant model for Taiwan and Indonesia (55%) while the Ohlson+D performs second in these two counties if we base our conclusion on these by-year results. Again, the earnings model seems to perform worst for all countries except Indonesia. These results are in direct contrast to Ashbaugh and Olsson (2002), who concluded that the earnings-based model is the most appropriate valuation model for the international firms using International Accounting Standards (IAS). In contrast, in our sample, using OLS, the E model is never the dominant model for any country other than just three occasions for Indonesia.

Estimates of the Valuation Parameters

We report the valuation coefficients from the annual regression for each country in Table 9 (where panels A-C are reposted on the journal website); Panel A reports the annual regression coefficients for Hong Kong, Indonesia, and Korea. Table 9, Panel B reports the results based on annual regressions for Malaysia, Singapore, and Taiwan, while Panel C reports the results for Thailand and the average for all countries. The annual averages are computed and reported as the last row of each table. Rather than focusing on the valuation coefficients of each year for each country, we discuss the average coefficients across all years for each country. The ranges of values of average parameters for the countries are as follows. For the earnings variable, the coefficient lies between 0.16 (Korea) and 7.04 (Thailand). For the book value variable, the corresponding figures are 0.03 (Korea) and 0.91 (Taiwan). The dividend variable lies between −9.42 (Indonesia) and 22.98 (Hong Kong). Ignoring the two negative coefficients, Indonesia and Taiwan, the next minimum is 1.63 (Korea). Online Figure 1 outlines the annual coefficient of each variable by country and year and overall average coefficient for each variable over all countries along the time dimension. Although a small sample size precludes any detailed statistical analysis, the lines show a reasonable horizontal disposition and indicate more or less stable parameters across the time period. The average coefficients across countries are also remarkably stable over time.

Table 9.

Parameter Estimates of Ohlson+D Model.

Country	Intercept	EPS	BPS	DPS
Hong Kong
Average	0.185	1.414	0.152	22.980
Indonesia
Average	0.063	6.773	0.426	−9.415
Korea
Average	0.752	0.161	0.027	1.630
Malaysia
Average	0.052	1.666	0.456	14.121
Singapore
Average	0.067	3.044	0.562	4.930
Taiwan
Average	0.088	4.656	0.915	−1.772
Thailand
Average	0.139	7.041	0.359	4.265
All countries
Average	0.192	3.537	0.414	5.249

Note. This table presents the average of yearly parameter estimates of Ohlson+D Model for each of the countries. The average of all countries on the last row represents the average of all the average numbers from the seven countries. The full yearly data of parameter estimates of EPS, BPS, and DPS are reposted on the journal website.

We also run a series of regressions on the pooled data of each country. Given our price level and currency adjustments, we do not include any year fixed effects. These coefficient estimates are reported in Table 10. For the pooled sample, a few observations are in order. Except for Thailand and Taiwan, for all the countries, the coefficient of the dividend term is several times that of the coefficient of the earnings term and all the dividend coefficients are positive. However, the coefficient of the dividend term for Indonesia is not significant. Given a small number of countries, we are unable to do any statistical test, but for the seven countries, the evidence indicates a significant informational role of the dividend variable. The dividend coefficients range from 22.000 (Hong Kong) to 1.734 (Thailand). The middle four values are between 14.091 (Malaysia) to 1.925 (Korea). Thus, the evidence indicates a widely varied information environment with respect to other information, while relatively stable accounting valuation parameters.

Table 10.

Parameter Estimates for Four Valuation Models by Country.

	Intercept	EPS	BPS	DPS	F	Adj. R²	N
Hong Kong
E	0.606***	7.388***			1933.5***	0.506	1,886
BV	0.668***		0.596***		836.8***	0.307	1,886
Ohlson	0.600***	7.310***	0.010		966.4***	0.506	1,886
Ohlson+D	0.249***	1.627***	0.040	22.000	2735.6***	0.813	1,886
Indonesia
E	0.179***	0.582***			98.4***	0.035	2,721
BV	0.119***		0.720***		500.7***	0.155	2,721
Ohlson	0.107***	−0.386***	0.868***		267.2***	0.164	2,721
Ohlson+D	0.107***	−0.382***	0.864***	0.681	178.1***	0.163	2,721
Korea
E	1.929***	0.184***			331.1***	0.124	2,323
BV	0.877***	0.000	0.049***		1027.5***	0.307	2,323
Ohlson	0.877***	0.113***	0.044***		625.6***	0.350	2,323
Ohlson+D	0.747***	0.055***	0.026***	1.925***	724.4***	0.483	2,323
Malaysia
E	0.482***	2.792***			712.8***	0.190	3,027
BV	0.080***		0.850***		1580.4***	0.343	3,027
Ohlson	0.111	1.624	0.713		1002.8	0.398	3,027
Ohlson+D	0.066***	0.741***	0.528***	14.091***	1377.8***	0.577	3,027
Singapore
E	0.523***	6.299***			2187.8***	0.476	2,407
BV	0.081***		0.869***		3737.4***	0.608	2,407
Ohlson	0.103***	2.814***	0.641***		2350.9***	0.661	2,407
Ohlson+D	0.074***	2.319***	0.549***	6.916***	1743.8***	0.685	2,407
Taiwan
E	0.546***	6.415***			2065.3***	0.400	3,094
BV	−0.170***		1.822***		1496.8***	0.326	3,094
Ohlson	0.050***	4.650***	1.042***		1408.8***	0.477	3,094
Ohlson+D	0.075***	4.452***	0.959***	4.501***	994.8***	0.491	3,094
Thailand
E	0.578***	2.588***			662.7***	0.174	3,145
BV	0.223***		0.880***		2084.1***	0.399	3,145
Ohlson	0.228***	1.448***	0.772***		1271.2***	0.447	3,145
Ohlson+D	0.232***	1.418***	0.738***	1.734***	856.3***	0.449	3,145

Note. This table reports the parameter estimates for each country using the pooled data of 1998–2017.

For the other variables in the complete model, the coefficients of the E variable for the seven countries are between 4.452 (Taiwan) and −.382 (Indonesia). Leaving aside Indonesia, the rest of the six coefficients of the E variable are between 1.627 for Hong Kong and 0.055 for Korea. For the variable BV, the coefficients range between 0.959 (Taiwan) and 0.026 (Korea). In terms of weightage among the three variables (E, BV, and D), we observe that investors provide the highest weight to D as compared with the other two variables in all countries except Indonesia. When comparing between E and BV, we observe that investors place more weight on E in all countries except Indonesia.

Numerical Example

To gain further insights to the observed coefficients, we constructed a numerical example (Table 11) to see what sort of parameter values would be consistent with the observed coefficients. We collected the average of the 10-year government bond interest rates for all seven countries during the years 1998 to 2018 (see Table 1, Panel D). We also run regressions of current earnings with lagged earnings and lagged dividends to capture the earnings forecast parameters (Table 12). We use these parameters along with interest rates and try different values of the two persistence parameters λ and μ, to come up with scenarios that seem closest to the observed coefficients as reported in Table 11. The numbers reported are illustrative in that they produce “small” absolute errors from the observed coefficients and are used as possible descriptors of the underlying economy. It is not a product of any systematic search to produce “minimum error” estimators, which is of course, possible. It is also possible to empirically verify the characterizations of the persistence parameters but those exercises are left out because they are beyond the scope of the current study.

Table 11.

Simulation of Parameter Values for the Valuation Model With Average Data.

Parameter	Parameter values
	Hong Kong	Indonesia	Korea	Malaysia	Singapore	Taiwan	Thailand
ω	0.675	0.950	0.100	0.900	0.300	0.200	0.100
γ	0.940	0.050	0.100	0.300	0.700	0.900	0.600
R	1.023	1.045	1.028	1.045	1.028	1.043	1.034
λ	0.678	0.554	0.452	0.419	0.596	0.589	0.645
μ	0.621	(0.316)	1.567	1.098	0.931	(0.295)	0.342
β₁	0.692	0.525	0.966	0.677	0.903	0.692	0.918
β₂	(0.504)	(0.023)	(0.000)	(0.112)	(0.025)	(0.064)	(0.005)
β3	0.812	0.497	0.034	0.435	0.122	0.372	0.087
Check sum	1.000	1.000	1.000	1.000	1.000	1.000	1.000
Theoretical coefficients
β₁ (θ1 Book Value)	0.692	0.525	0.966	0.677	0.903	0.692	0.918
β_2.Φ+β₃.λ/r (θ2 Earnings)	1.543	5.599	0.527	1.441	1.659	3.539	1.492
(μβ₃/r)−β₂ (θ3 Dividends)	22.529	(3.471)	1.870	10.734	4.027	(2.488)	0.877
% Incremental R² for D	0.861	0.492	0.561	0.626	0.166	0.077	0.010
Observed coefficients
θ₁ (Book Value)	0.150	0.430	0.030	0.460	0.560	0.910	0.705
θ₂ (Earnings)	1.410	6.770	0.160	1.670	3.040	4.660	1.713
θ₃ (Dividends)	22.980	(9.420)	1.630	14.120	4.930	(1.770)	0.004
Sum of absolute value of errors	1.1256	7.2148	1.5438	3.8316	2.6258	2.0573	1.3075

Where $β_{1} = R (1 - ω) (1 - γ) / (R - ω) (R - γ)$ $β_{2} = - r ω γ / (R - ω) (R - γ)$ $β_{3} = Rr / (R - ω) (R - γ)$

Note. The values of R are obtained from Table 1, Panel D.

The solution of this system is far from unique; however, we refrain from characterizing the solution space or its size because that is not germane to this study. For some countries (Hong Kong, Indonesia, Malaysia), the observed coefficients are consistent with a high level of persistence of Residual (Abnormal) income. The persistence parameter ω is close to 1 for these countries. For Singapore and Taiwan, the persistence parameter of residual income is around 0.2 and 0.3, whereas for Korea and Thailand it seems that the persistence parameter is on the order of 0.1. This pattern is simply indicative and deserves further investigation. The persistence parameter of other information (γ is close to 1 for Hong Kong and Taiwan and is quite high for Singapore and Thailand (0.7 and 0.6, respectively), but quite low for Korea and Indonesia (0.1 and 0.05, respectively). These numbers are consistent with our intuition that the information environments in these countries are often vastly different and so is the market’s tendency to rely on them. The values of earnings coefficient in the earnings forecast parameter λ are remarkably stable across the countries with values between 0.678 and 0.419, indicating the role played by earnings information in these countries. The dividend coefficients (μ) vary widely (−.295 to 1.567), indicating a wide variation in reliance on dividends for making earnings forecast. Although by no means definitive, this example illustrates that the observed coefficients are in some sense, reconcilable to the theory.

Conclusions and Discussions

In this study, we show that dividends play an important role in explaining firm value together with book value and earnings. In the context of the valuation model, the dividends play the role of “other information” together with earnings. This evidence is new, particularly in the context of the Asian markets. Also, the similarities of the valuation parameters across countries indicate the applicability of the valuation model, particularly where earnings forecasts are rare. Our finding of value relevance of dividends complements the findings on information content of dividends so far observed in the literature and the findings of Ashbaugh and Olsson (2002) about the role of accounting numbers in valuation of international firms. Our findings indicate that if dividends as well as past earnings are used for earnings forecasts, they will play an additional informational role in firm valuation.

Our country-by-country results show that all three accounting-based valuation models (E, BV, and BV+E, i.e., Ohlson) perform adequately. We compare the results of the three models with an Ohlson with dividend model (Ohlson+D model). This study finds strongest results for Ohlson+D model and documents that this Ohlson+D model is the dominant model across countries, subperiods, and years. The size and sign of the coefficient of dividend are consistent with a possible informational role played by dividends.

We do not find any evidence reported in the earlier study of Ashbaugh and Olsson (2002) about the dominance of the earnings capitalization model. In retrospect, this is not surprising because the earnings capitalization model ought to be underspecified given the various theoretical valuation models since Ohlson (1995). The dominance of the Ohlson (BV+E) model is therefore consistent with theory as well as previous research on U.S. firms following U.S. GAAP (Collins et al., 1997; Francis et al., 2001; Penman & Sougiannis, 1998). We also offer an extension of the valuation model, where we show that absent analyst forecasts, earnings and dividends together could be used by the market to construct a reasonable earnings forecast that could play a role as part of other information.

Furthermore, as our numerical example illustrates, the possible scenarios of persistence parameters underlying the fundamental variables could be quite different across the different countries. There are some differences in the ways current earnings and dividend combine to give an indication of the future earnings and we believe this is the dimension in which the countries differ most.

Our results also will be of interest to researchers who are interested in the impact of IFRS adoption and use of these numbers for valuation. We also observe that the time pattern of the valuation parameters within our sample period is quite stable, indicating that whatever changes these countries may be adopting in their accounting policy, their adoption of IFRS did not materially affect the valuation parameters across time.

Our evidence is also consistent with the theoretical notion that the valuation parameters should be unaffected by accounting policy choices and ought to be driven by fundamentals such as the persistence of residual income and other information. Our evidence indicates that in spite of apparent inconsistency in adopting IFRS, these countries may be adopting the standards so as to preserve the basic notion of harmonization as intended by the standard setters. Thus, regardless of the extent of IFRS implementation, our evidence indicates applicability of the fundamental valuation model for the seven Asian countries in a fairly uniform way as our regression results and the illustrative numerical example demonstrates. However, to draw a more definitive conclusion on the parameters, one needs to do a more elaborate simulation exercise that is left for future research.

Supplemental Material

2.Asian_Valuation_9July2020_ToWebsite_V3 – Supplemental material for Fundamental Valuation in Seven Asian Countries: Role of Earnings, Book Value, and Dividends

Supplemental material, 2.Asian_Valuation_9July2020_ToWebsite_V3 for Fundamental Valuation in Seven Asian Countries: Role of Earnings, Book Value, and Dividends by Kriengkrai Boonlert-U-Thai, Shahrokh M. Saudagaran and Pradyot K. Sen in Journal of Accounting, Auditing & Finance

Footnotes

Appendix A

Appendix B

Acknowledgements

The authors thank Rajib Doogar, Jim Ohlson, Gary Meek, Wayne Thomas, Ron Barniv, Shimin Chen, John Eichenseher, Li Li Eng, Paquita Davis-Friday, Woo Suk Hwang, Piman Limpaphayom, Thomas Lin, Robert Lipe, Timothy West, and participants at the 2013 AAA Western Region Meeting in San Francisco, USA, the 2013 AAAA Annual Meeting in Penang, the 2014 AAA Annual Meeting in Atlanta, USA, and the 2016 European Accounting Association Congress, Maastricht, for their helpful comments on earlier versions of the study.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this study.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Kriengkrai Boonlert-U-Thai

Pradyot K. Sen

Supplemental Material

Supplemental material for this article is available online.

Data Availability

The data used in this study are available from publicly available research databases.

Notes

References

Abarbanell

Bushee

(1997). Fundamental analysis, future earnings, and stock prices. Journal of Accounting Research, 35, 1–24.

Akbar

Stark

(2003). Deflators, new shareholder cash flows, dividend, capital contributions and estimated models of corporate valuation. Journal of Business Finance and Accounting, 30(9–10), 1211–1233.

Ashbaugh

Olsson

(2002). An exploratory study of the valuation properties of cross-listed firms’ IAS and U.S. GAAP earnings and book value. The Accounting Review, 77(1), 107–126.

Barth

Beaver

Landsman

(1993). Structural analysis of SFAS 87 pension disclosures and their relation to share prices. Financial Analysts Journal (January/February), 49, 18–26.

Barth

Beaver

Landsman

(1998). Relative valuation roles of equity book value and net income as a function of financial health. Journal of Accounting and Economics, 25(1), 1–34.

Bar-Yosef

Callen

Livnat

(1996). Modeling dividends, earnings, and book value equity: An empirical investigation of the Ohlson valuation dynamics. Review of Accounting Studies, 1, 207–224.

Berger

Ofek

Swary

(1996). Investor valuation of the abandonment option. Journal of Financial Economics, 42, 257–287.

Burgstahler

Dichev

(1997). Earnings, adaptation, and equity value. The Accounting Review, 72, 187–215.

Callen

Morel

(2000). A Lintnerian linear accounting valuation model. Journal of Accounting, Auditing & Finance, 15(3), 301–314.

10.

Choi

O’Hanlan

Pope

(2006). Conservative accounting and linear information valuation models. Contemporary Accounting Research, 23(1), 73–101.

11.

Clarkson

Hanna

Richardson

Thompson

(2011). The impact of IFRS adoption on the value relevance of book value and earnings. Journal of Contemporary Accounting and Economics, 7, 1–17.

12.

Collins

Maydew

Weiss

(1997). Changes in the value-relevance of earnings and book values over the past forty years. Journal of Accounting and Economics, 24, 39–67.

13.

Collins

Pincus

Xie

(1999). Equity valuation and negative earnings: The role of book value of equity. The Accounting Review, 74(1), 29–61.

14.

Dechow

Hutton

Sloan

(1999). An empirical assessment of the residual income valuation model. Journal of Accounting and Economics, 26(1–3), 1–34.

15.

Easterday

Sen

Stephan

(2011). Another specification of Ohlson’s “other information” term for the earnings/returns association: Theory and some evidence. Journal of Business Financial and Accounting, 38(9)&(10), 1123–1155.

16.

Easton

Harris

(1991). Earnings as an explanatory variable for returns. Journal of Accounting Research, 29, 19–36.

17.

Feltham

Ohlson

(1995). Valuation and clean surplus accounting for operating and financial activities. Contemporary Accounting Research, 11(2), 689–831.

18.

Francis

Schipper

(1999). Have financial statements lost their relevance? Journal of Accounting Research, 37, 319–352.

19.

Francis

Olsson

Oswald

(2001). Using mechanical earnings and residual income forecasts in equity valuation [Working paper]. Duke University, University of Wisconsin, and London Business School.

20.

Frankel

Lee

(1998). Accounting valuation, market expectation, and cross-sectional stock returns. Journal of Accounting and Economics, 25(3), 283–319.

21.

Graham

King

(2000). Accounting practices and the market valuation of accounting numbers: Evidence from Indonesia, Korea, Malaysia, the Philippines, Taiwan, and Thailand. The International Journal of Accounting, 35(4), 445–470.

22.

Hand

(2001). Discussion of earnings, book values, and dividends in equity valuation: An empirical perspective. Contemporary Accounting Research, 18(1), 121–130.

23.

Hand

Landsman

(2005). The pricing of dividend in equity valuation. Journal of Business Finance and Accounting, 32(3&4), 435–469.

24.

Lee

Lin

(2016). A comparative analysis of accounting based valuation models. Journal of Accounting, Auditing & Finance, 26, 1–15.

25.

Jan

(1995). The role of negative earnings in the valuation of equity stocks [Working paper]. Santa Clara University.

26.

Kothari

S. P.

Zimmerman

(1995). Price and return models. Journal of Accounting and Economics, 20, 155–192.

27.

Lyon

Schroeder

(1992). Firm growth and the valuation relevance of earnings levels, earnings innovations, and dividends. Journal of Accounting, Auditing & Finance, 7(4), 531–551.

28.

Ohlson

(1995). Earnings, book values, and dividend in equity valuation. Contemporary Accounting Research, 11(2), 661–687.

29.

Ohlson

(2001). Earnings, book values, and dividend in equity valuation: An empirical perspective. Contemporary Accounting Research, 18(1), 107–120.

30.

Ohlson

Shroff

(1992). Changes versus levels in earnings as explanatory variables for returns: Some theoretical considerations. Journal of Accounting Research, 30(2), 210–226.

31.

Penman

Sougiannis

(1997). The dividend displacement property and the substitution of anticipated earnings for dividends in equity valuation. The Accounting Review, 72(1), 1–21.

32.

Penman

Sougiannis

(1998). A comparison of dividend, cash flow, and earnings approaches to equity valuation. Contemporary Accounting Research, 15(3), 343–383.

33.

Saudagaran

Diga

(1997a). Accounting regulation in ASEAN: A choice between the global and regional paradigms of harmonization. Journal of International Financial Management and Accounting, 8(1), 1–32.

34.

Saudagaran

Diga

(1997b). Financial reporting in emerging capital markets: Characteristics and policy issues. Accounting Horizons, 11(2), 41–64.

35.

Saudagaran

Diga

(2000). The institutional environment of financial reporting regulation in ASEAN. The International Journal of Accounting, 35(1), 1–26.

36.

Shane

Brous

(2001). Investor and (value line) analyst underreaction to information about future earnings: The corrective role of non-earnings-surprise information. Journal of Accounting Research, 39, 387–404.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.68 MB