WORKING PAPER SERIES

Transcription

1 College of Business Administration University of Rhode Island William A. Orme WORKING PAPER SERIES encouraging creative research Asset Growth and Stock Returns: Evidence from the Pacific-Basin Markets Shaw Chen, Tong Yao, Tong Yu, Ting Zhang 2007/2008 No. 15 This working paper series is intended to facilitate discussion and encourage the exchange of ideas. Inclusion here does not preclude publication elsewhere. It is the original work of the author(s) and subject to copyright regulations. Office of the Dean College of Business Administration Ballentine Hall 7 Lippitt Road Kingston, RI

2 Asset Growth and Stock Returns: Evidence from the Pacific-Basin Markets Shaw Chen, Tong Yao, Tong Yu, Ting Zhang* February 2008 * Chen, Yu and Zhang are from College of Business Administration, University of Rhode Island. Yao is from Department of Finance, Eller College of Management, University of Arizona. s: chenshaw@uri.edu (Chen), yaot@ .arizona.edu (Yao), tongyu@uri.edu (Yu), and tingjohn@mail.uri.edu (Zhang). All errors are our own.

3 Asset Growth and Stock Returns: Evidence from the Pacific-Basin Markets Abstract This study examines the effect of corporate asset growth on stock returns using data on nine equity markets in the Pacific-Basin region. There is a pervasive negative relation between asset growth and subsequent stock returns, suggesting potential inefficiencies of the region s financial systems in allocating capitals and valuing investment opportunities. Yet the relation is weaker relative to the U.S. market. We further examine factors affecting the difference in the magnitude of the asset growth effect across the PACAP markets and the difference between the PACAP region and the U.S., such as the homogeneity of asset growth, persistence in growth and profitability, the overinvestment tendency, and corporate financing activities. JEL Classification: G12 Key Words: Return Predictability; Asset Growth; Pacific-Basin Capital Markets; Market Efficiency 1

4 Asset Growth and Stock Returns: Evidence from the Pacific-Basin Markets I. Introduction The relationship between finance and growth has been long debated in academic research, and there is a growing body of empirical evidence that well-functioning financial system contributes positively to a country s economic growth (see, e.g., Demirgüç-Kunt and Levine (2008) for an extensive survey of the literature). At the micro-economic level, an important channel for capital markets or banking systems to facilitate economic growth is to efficiently coordinate financing and investment activities across firms, to the effect that capital flows from firms with low investment opportunities to firms with highly profitable prospects. However, using the U.S. data, many studies have found evidence at discordance with micro-level financial efficiency: firms experiencing rapid growth by raising external financing, and making capital investments and acquisitions, subsequently have poor operating performance and disappointing stock returns, whereas firms experiencing contraction via divesture, share repurchase, and debt retirement, subsequently report good operating results and high stock returns. 1 Recently, Cooper, Gulen, and Schill (2007, hereafter CGS) summarize these investment and financing effects by a simple measure of total asset growth. They show that in the U.S. market during the period from 1968 to 2003, firms ranked in the top decile of asset growth underperform firms ranked in the bottom decile by more than 20% in stock returns during the year after ranking. They conclude that the evidence suggests significant market inefficiency, possibly resulting from the over-investment tendency by corporate managers and an excessive-extrapolation bias by investors when they value stocks based on firms past growth. In this study, we seek empirical evidence on the relation between asset growth and stock returns in international financial markets specifically, the nine stock markets in the Asian Pacific-Basin region (PACAP), including Japan (a well-developed economy), China (one of the most rapid growing economies), as well as Hong Kong, Taiwan, Korea, Malaysia, Singapore, Thailand, and Indonesia. We analyze the asset 1 See Cooper, Gulen, and Schill (2007) for a brief survey of the large body of empirical literature regarding the effects of firms financing and investment activities on operating performance and stock returns. 2

5 growth effect in these markets for following reasons. First, the Pacific-Basin region has experienced rapid economic growth during recent decades, accompanied by rapid asset growth at firm level and active capital market activities (e.g., Shaffer, 2002). Second, different from the capital market-oriented financial system of the U.S., most PACAP economies have primarily bank-oriented financial systems. These two types of financial systems employ different mechanisms in bonding the interests of firms and investors (Townsend, 1979; Diamond, 1991; Datta, Datta, and Patel, 1998; and Puri, 1996, 1999), which may lead to different investment and financing behavior and dissimilar asset growth effects. Third, corporate ownership and governance characteristics in the PACAP region are very different from the U.S. such as highly concentrated ownership of public corporations and complicated, pyramid-like, systems of corporate control (Claessens, Djankov, and Lang, 2000). The looming corporate governance concern is the expropriation of minority shareholders by controlling shareholders, instead of the conflict of interest between managers and diverse shareholders the latter is believed to be a source of the asset growth effect in the U.S. (Cooper, et al., 2007). Fourth, arguably investors in the U.S. market are more sophisticated especially with a large group of institutional investors than those in the PACAP markets. Finally, economic development, corporate governance, financial system, and investor sophistication also vary substantially across the nine markets within the PACAP region. Such differences from the U.S. and cross-market variations within the region provide us an opportunity to check the robustness of the asset growth effect, and to examine the validity of various economic interpretations on the effect. We find a pervasive asset growth effect in the PACAP markets during the sample period from 1981 to That is, there is a significantly negative relation between firms asset growth and subsequent stock returns. For example, the annualized stock return spreads between the top and bottom stock deciles sorted on total asset growth rate are -7.64% in Japan, -8.64% in China, and % in Hong Kong. Taiwan is the only market with an insignificant relation between asset growth and stock returns. Pooling all stocks in the nine PACAP markets, the annualized return spread between top and bottom asset growth deciles is -7.42%. The stock return differences remain significant over multiple years after portfolio ranking, and are robust to the control of the market, size, 3

6 and book-to-market factors using the Fama-French (1993) three-factor model. There is also an accompanying effect in operating performance: during the subsequent five years, high asset growth firms experience deteriorating accounting return on equity (ROA) while that for low asset growth firms improves. Finally, it is interesting to observe that the overall magnitude of the asset growth effect is weaker in the PACAP region relative to the U.S. during the same sample period, the stock return spread between the top and bottom asset growth deciles is % per year in the U.S. market. The empirical result from the PACAP region suggests potential inefficiencies in deploying financial resources across firms and in valuing investment opportunities. Such micro-level inefficiency is striking in contrast to the spectacular macro-economic growth. One might argue that the high economic growth in the region is achieved by means of, for example, high rates of private savings and physical capital accumulation (certainly related to the development of financial systems), despite inefficient cross-firm capital allocation. However, the weaker magnitude of the effect relative to the U.S. suggests an interpretation that is certainly puzzling: at the margin, the financial systems in the PACAP region seem to be more efficient in capital allocation and asset valuation than that the developed financial markets in U.S. To peel a few layers off this puzzle, we analyze factors affecting the differential magnitude of the asset growth effect across regions and across the nine markets within the PACAP region. We first check, mechanically, whether the weaker asset growth effect in the region is due to less dispersion of firm growth or lower sensitivity of subsequent stock return to asset growth. Indeed, we find that firms in the PACAP region are more homogenous in terms of asset growth. Such homogeneity may be due to the fact that firms in the region are similarly exposed to systematic shocks or face similar financial constraints, but have nothing to do with financial system efficiency. However, our results show that the sensitivity of future stock return to asset growth difference is also significantly lower in the PACAP region. Further, both the degree of growth homogeneity and the sensitivity of return to asset growth help explain the difference of the effect across the nine markets within the region. Why is sensitivity of stock return to asset growth lower in the PACAP region? We find that asset growth is more persistent in the region specifically, despite a pattern of 4

7 declining future growth and profitability (measured by ROA) for high asset-growth firms, the reversal of subsequent growth and profitability in the PACAP region is not as strong as that in the U.S.. As a consequence, suppose investors in U.S. and in PACAP markets extrapolate from past firm growth in the same way when they value stocks, stock valuation in PACAP markets could turn out to be less biased. Still, an open question is why there is higher persistence of growth and profitability in the PACAP region. One possible explanation is that there may be less overinvestment tendency. To examine this hypothesis, we break down asset growth into growth of various asset components. We find that relative to the U.S., in the PACAP markets capital expenditure is less important in driving asset growth, and there is a higher degree of cross-sectional homogeneity in capital expenditure as well as in the growth of other asset components (cash, non-cash current assets, other assets). In both regions, capital expenditure (as well as the growth of other asset components) is negatively correlated with future stock returns. However, stock returns are less sensitive to capital expenditure in the PACAP region relative to the US, and the difference is marginally significant (at the 10% critical level). Therefore, there is some evidence, albeit not very strong, that overinvestment is less severe in the PACAP markets. Finally, we break down asset growth from the financing side of balance sheet, and find strong and interesting results. Consistent with the observation that financial systems in some PACAP economies are less developed and that the banking systems in the region often dominate capital markets in providing capital, PACAP firms rely more on internal financing (i.e., retained earnings) than external financing (i.e., equity and debt financing, and change in operating liabilities) to achieve asset growth; further, when raising external financing, they rely more on debt (likely more in the form of bank loans) than equity. Most interestingly, in the PACAP markets the sensitivity of stock return to equity financing is significantly stronger (more negative), whereas the sensitivities to debt and internal financing the two sources of financing that are more important for PACAP firms are less negative relative to those in the U.S. This is further confirmed by the regression analyses where we show economies with a bank-based financial system and those relying more on debt and internal financing have weaker asset growth effects. These results suggest that sources of financing or at a more fundamental level, 5

8 differences in financial systems are important in explaining the differential asset growth effect. They are consistent with the literature on the role of bank-based financial systems in improving capital allocation and corporate governance (e.g., Diamond, 1984; Puri 1996, 1999; Boot and Thakor, 1997; Beck, Levine and Loayza, 2000). The remainder of the paper is organized as follows. Section II discusses the data and the stock sample. Section III reports our main empirical results on the asset growth effect. Section IV analyzes factors affecting cross-market differences in the asset growth effect. Section V concludes. II. Data The data used in this study comes from two sources. We obtain stock return and accounting data from the Pacific Basin Capital Market Research (PACAP) databases (available via Wharton Research Data Services) for nine markets in the Pacific-Basin region Japan, China, Hong Kong, Taiwan, Malaysia, Korea, Singapore, Thailand, and Indonesia (The original PACAP database covers data for Japan, Hong Kong, Taiwan, Malaysia, Korea, Singapore, Thailand, and Indonesia; the newly available PACAP- CCER database includes data for China). The U.S. stock return data is from CRSP and the U.S. financial statement data is from Compustat. Our sample starts in 1981 since for the data for most PACAP markets begin in The sample end year is market specific due to data availability. The Japanese market has the longest sample period from 1981 to 2004 while the Indonesian stock market has the shortest period from 1990 to The Chinese stock market covers from 1994 to To avoid microstructure issues in measuring returns, we exclude firms with stock price less than the maximum of 100 times minimum tick size at the end of June of the portfolio formation year. 3 In addition, we restrict the sample for the PACAP markets to industrial firms. The U.S data are for the sample period from 1981 to For the U.S. sample, we exclude financial firms and 2 The PACAP data have observations for a subset of markets as early as in However, in early sample years (before 1981), the number of firms in these markets are very small. For instance, Hong Kong had three firms in 1978 and Korea had 11 firms in The minimum tick sizes for PACAP market are as following: JP: 1 JPY; HK: HKD; CH: 0.01 RMB; TW: 0.01 TWD; MA: 0.5 RM; KS: 5 WON; SN: 0.5 SGD; TH: 0.01 BAHT; IND: 5 RP. We accordingly set the following minimum price restriction at time of portfolio formation: JP 100 JPY; HK 1 HKD; CH 1 RMB; TW 1 TWD; MA 500 RM; KS 500 WON; SN 50 SGD; TH 1 BAHT; IND 500 RP. For U.S. the minimum price is 1USD. 6

9 firms that are close-end funds, real estate investment trusts (REITs), an American Depository Receipts (ADRs), and require a stock to have price no less than $1 at the end of June of the portfolio formation year. We first provide an overview of the PACAP stock markets and the U.S. market. For comparison purpose, we convert both price and market capitalization from local currency to US dollar based on exchange rates in December of year t. As shown in Panel A of Table 1, there are substantial differences in the market characteristics between PACAP markets and the U.S. The PACAP stock markets generally have smaller market capitalization (except for Japan), fewer firms, and lower average stock prices. Japan has the largest stock market in terms of both market capitalization and firm number among the PACAP markets, followed by China. The average stock price is $3.95 for the PACAP markets, much lower than $16.11 for the average price of the U.S. stocks. Japan is an exception. Its average stock price is $ In addition, the PACAP markets generally experience higher stock turnover, with the highest turnover of 375% in China over the period of and the second largest turnover of 116% in Taiwan over the period of , compared to the average annual turnover of 19% in U.S for the period of Moreover, the PACAP markets have higher stock returns, but appear to be more volatile than the U.S. market. Indonesia has the highest equally weighted (EW) stock returns (43.26%) and the highest standard deviation (69.50%) as well. Combining the Pacific-Basin stock markets together, the average EW (VW) stock returns are 24.69% (15.11%), with the standard deviation of 49.89% (44.27%), relative to the EW (VW) stock returns of 7.80% (6.58%) with the standard deviation of 16.66% (13.02%) for the U.S. market. Annual firm total asset growth rate (AG) is defined as year-over-year percentage change in total assets. To be specific, following CGS, a firm s assets growth rate for year t is defined as the percentage change in total assets from fiscal year t-2 to fiscal year t-1: Total Assets t 1 Total Assets AG t = Total Assets t 2 t 2 (1) 4 The turnover is measured as annual trading volume scaled by the year-end shares outstanding. The turnover of 375% in the China market, for example, suggests that on average, investors shuffle their portfolio positions almost four times a year. 7

10 Total assets are variable BAL9 in the PACAP and PACAP-CCER databases and Data Item 6 in Compustat for the U.S. market. To compute asset growth rate, we require a firm should not have zero or negative total assets in fiscal year t-2 and t-1. We further winsorize asset growth rate at the top and bottom 1% in each year to control for the influence of outliers. Panel B of Table 1 reports descriptive statistics of the asset growth rates across the Pacific-Basin stock markets and the U.S. stock market. Japan has the lowest equalweighted mean (median) annual asset growth rate of 7.81% (5.56%) while Malaysia has the highest annual asset growth rate with mean (median) of 26.79% (22.11%). The average asset growth rate for the combined PACAP market is about 19.30%, relative to 18.85% for the U.S. market. Additionally, the volatility of the asset growth rate in the PACAP market is similar to that in the U.S. stock market, as measured by the annual average standard deviation of asset growth (30.22% vs %). III. The Asset Growth Effect A. Returns to Sorted Portfolios We first examine the effect of asset growth on subsequent stock returns using sorted portfolios. From 1981 to 2004, at the end of June in each year t, we sort stocks in each market into equal-weighted portfolios based on the asset growth rate for the fiscal year that ends in year t-1. 5 The D10 and D1 portfolios consist of stocks with the highest and the lowest asset growth rates, respectively. The portfolios are rebalanced to have equal weights at the beginning of each month, and held from July of year t to June of year t+1. We compute the time-series averages of monthly returns of decile portfolios and the average return spreads between D10 and D1 portfolios for each of the nine PACAP markets and all PACAP markets including and excluding Japan. The results are in Panel A of Table 2. For comparison purpose, we also report the results for the U.S. market. 6 For 5 The only exception is for Japan. Most Japanese firms have a fiscal year that ends on March 31, and practically all firms publish their financial statements within three months after the fiscal year end. Therefore, we assume that in Japan, asset growth information for the fiscal year ending in March of year t is available at the end of June of year t, and form decile portfolios accordingly. The same approach is used by Chan, Hamao, and Lakonishok (1991) when they examine the value effect in Japan. 6 For the U.S. market, if a stock is delisted during a month, we include the delisting return reported in CRSP when computing the return to the portfolio the stock belongs to. If the delisting is missing, following Shumway (1997), we replace it with -33% if delisting is performance related, and zero otherwise. For the 8

11 portfolios formed within a specific market, we report returns in local currency. For portfolios formed using all stocks in the PACAP region (or using the eight markets except Japan), we adjust stock returns by exchange rate changes so that the reported returns are expressed in US dollar term. Exchanges rates are from Macro-Economic files of PACAP database (and the exchange rate file of the PACAP-CCER database for China). In each of the nine PACAP markets, the portfolio with the lowest asset-growth stocks (D1) earns higher monthly returns than the portfolio with the highest asset-growth stocks (D10). The D10-D1 return spreads are significantly negative in eight out of the nine markets, with Taiwan being the only exception, whose D10-D1 decile spread is negative but statistically insignificant. There are also cross-market variations. The decile return spread is -0.66% for Japan while that in Thailand is as high as -1.63%. We also pool all stocks in the nine markets together to form decile portfolios. The monthly D10- D1 decile spread for the pooled portfolios is -0.64% (t = -3.94). With Japan (the largest stock market in the region) excluded but all the other eight PACAP markets combined, the monthly stock return spread (D10 D1) is -0.56% (t = -3.10%). In the U.S. market, for the same sample period ( ), the monthly D10-D1 return spread for portfolios based on asset growth is -1.96% (t = -9.54), much higher than the -0.64% spread in the PACAP region or -0.56% for the PACAP markets excluding Japan. In other words, although the asset growth effect is pervasive in the PACAP markets, its magnitude is generally weaker than that in the U.S. market. In Panel B of Table 2, we further report the differences in stock returns between the D10 and D1 deciles during each of the five years before portfolio formation (June of year t) and each of the five years after portfolio formation. There is a pattern common to the PACAP markets and the U.S. market. Relative to D1 stocks, D10 stocks typically have higher returns during the several years prior to portfolio formation, but lower returns starting from the portfolio formation year, and in the five years after portfolio formation. This pattern is further illustrated in Figure 1. PACAP data, to our knowledge, no serious pattern of missing delisting returns is reported. In addition, firm delistings are far less frequent. For example, there are only 123 companies delisted from the Tokyo Stock Exchange from 1971 to 1988 (Chan, Hamao, and Lakonishok, 1991). According to Kang, Kim, and Stulz (1999), during the period of 1980 to 1993, 64 firms were delisted from the Tokyo Stock Exchange, corresponding to about four firms per year, and the largest number of delistings is 10 in

12 B. Three-Factor Alphas of Sorted Portfolios To control for the effect of risk and risk premium, in each stock market we use the following Fama French (1993) three-factor model to compute alphas for decile portfolios: R itj RF = α + b RMRF + s SMB + h HML + ε (2) tj ij ij tj i where R itj RF tj is the monthly returns of decile portfolio i in excess of the monthly risk free rate for market j. RMRF tj is the market return in excess of the risk free rate, and SMB tj and HML tj are the monthly size and book-to-market factors for market j. We obtain the three factors of the U.S. market from French website ( and follow the procedures described in Fama and French (1993) to construct the factors for each PACAP market. For each market, we construct size and B/M (book value of equity/market value of equity) portfolios in June of each year t. Size is measured by multiplying stock closing prices with total shares outstanding in June of year t. 7 In constructing B/M, we measure book value of equity by total shareholders equity (BAL21 in both PACAP and PACAP- CCER databases) and market value of equity by stock closing prices times total shares outstanding in December of year t-1. In particular, we sort all stocks into small and big size groups based on the median market capitalization of all stocks in June of year t. We independently sort all stocks into low, median, and high B/M groups based on the 30 percent and 70 percent cutoff points of the book-to-market ratio of all stocks. Six size- B/M portfolios are defined as the intersections of the two size and three B/M groups. The monthly value-weighted average return on each portfolio is then computed. SMB is the difference, in each month, between the simple average of returns on the three small-stock tj ij tj itj 7 A typical Chinese firm may consist of various classes of stock shares, including state shares (those owned by the central or local governments), legal-entity shares (those held by domestic legal entities such as listed companies, state owned enterprises, and banks), and tradable shares. Tradable shares are further classified into A-, B-share, and H-share classes. A-shares are ordinary shares primarily made available to Chinese citizens and institutions, whereas B-shares are primarily made available to foreign investors. H-shares are Chinese company stocks listed in Hong Kong. We construct the China s three-factor model using A shares market value and market value of all shares, respectively. We find these two approaches yield similar results in the asset growth effect. To save space, we report the results using A-share based three-factor model in Table 3. 10

13 portfolios (S/L, S/M, and S/H) and the simple average of returns on the three big-stock portfolios (B/L, B/M, and B/H). Similarly, HML is the difference, in each month, between the simple average of returns on the two high-b/m portfolios (S/H and B/H) and the average of returns on the two low-b/m portfolios (S/L and B/L). We calculate the monthly market returns in each stock market as the value-weighted average monthly returns. Finally, we follow prior studies to select proxies for the local risk free rates. 8 The results are in Panel A of Table 3. Except for Taiwan (alpha = -0.59% and t = -0.67), the D10-D1 spreads in three factor alphas are significantly negative in the PACAP markets. For example, the alpha spread (D10 D1) is -0.31% (t = -2.01) in Japan, -0.72% (t = -2.42) in China, and -1.24% (t = -2.02) in Hong Kong. For the combined PACAP market, the alpha spread is -0.67% (t = -3.84) whereas the alpha spread in the U.S. is larger in magnitude, at -1.20% (t = -4.88). In the last two columns of the table, we report the F-statistics of Gibbons, Ross, and Shanken (1989) (referred to as the GRS statistics) for testing the hypothesis that the alphas for the ten decile portfolios sorted on asset growth rate are jointly zero. P(GRS) is the p-value of the GRS statistic. 9 At the ten percent critical level, the GRS test rejects the joint zero alpha hypothesis for five countries: Japan, Hong Kong, Malaysia, Korea, and Indonesia, as well for the combined PACAP market and for the U.S. market. We also examine the persistence of three-factor alpha sorted on asset growth. The result is reported in Panel B of Table 3. The pattern is consistent with that for the return spreads reported in Panel B of Table 2 low asset growth firms earn higher alpha up to five years beyond the sorting years in the PACAP markets and in the U.S. market. 8 To be specific, following Chan, Hamao, and Lakonishok (1991) and Chang, McLeavey, and Rhee (1995), we use 30-day Gensaki rate (1/ /2004) as the risk-free rate for Japan. Consistent with Wang (2004) and Kang et al. (2002), we use the monthly yield of the three-month household deposit interest rate in China as the risk-free rate. Following Chui and Wei (1998), we use (1) the interest rate on time deposits of deposit money banks as the risk free rate in Korea; (2) the central bank rediscount rate as the risk free rate for Taiwan; (3) the commercial banks interbank lending rate as the risk free rate for Thailand; (4) a combination of one month fixed deposit rate paid by commercial banks (1/ /1982) and interbank offer rate on one month deposit (1/ /1999) as the risk free rate for Malaysia; and (5) a combination of HSBC s best lending rate (1/1981 6/1982), one month time deposit rate paid by principal licensed banks (7/1982 6/1988), and interbank offer on one month deposit (7/ /2001) as the risk free rate for Hong Kong. 9 Note that GRS test and the sorted portfolio method serve the different purposes. The GRS F-statistics and p-value examine whether all the ten regression intercepts are equal to zero, while the sorted portfolio method compares the return difference between the extreme deciles (D10-D1) sorted on total asset growth. 11

14 C. Operating Performance In Figure 2, we plot the average annual return on equity (ROA) for each asset growth decile, from five years prior to portfolio formation, to five years after portfolio formation. The asset growth deciles are formed by pooling across all stocks in the nine PACAP markets. An obvious pattern is that operating performance for D10 firms (those with highest past asset growth) quickly deteriorate right after portfolio formation, while operating performance for D1 firms (those with lowest past asset growth) quickly improves. The pattern is consistent with that for stock returns, reported in Table 2 and 3. D. Robustness The patterns reported in Table 3 and 4 continue to hold when we form valueweighted portfolios instead of equal-weighted portfolios. We also find that there are no major changes to our conclusions if we convert all local-currency returns reported in Table 3 and Table 4 into returns based on US dollar. 10 In addition, we calculate portfolio alphas by adding a momentum factor into Fama-French three-factor model. Stocks with the lowest asset growth rate have higher four-factor alphas than stocks with the highest asset growth rate. 11 Furthermore, we separate our sample period into the subperiod before Asian financial crisis (before 1997) and the subperiod after the crisis (after 1998). We then investigate whether there is any significant difference in the asset growth effect during the two subperiods. In table 4, we report, for each subperiod, the decile return spreads and the spreads in the three-factor alphas for the combined PACAP market. As the table shows, lower asset growth firms consistently earn higher subsequent returns than firm with higher asset growth, both before and after Asian financial crisis. We also perform similar analysis within each PACAP market, and do not find significant differences in the asset growth effect after the crisis. 10 The only notable difference is that the asset growth effect becomes significant in Taiwan but insignificant in Indonesia after converting stock returns into U.S. dollar term. 11 There is no momentum effect in Japan, China, Korea, Taiwan, Singapore, or Malaysia, but there is momentum effect in Hong Kong (Hong et al. 2003; Wang 2004). Our analysis confirms this. 12

15 IV. Further Evidence on Cross-Market Differences The results in the above analysis produce a striking implication when linked to the efficiency of financial systems: despite spectacular economic growth in the PACAP region, there are significant inefficiencies in the allocation of financial resources across firms, and in the valuation of investment opportunities. This means that the contribution of financial systems to the economic growth in this region, if any, is more likely in the form of mobilizing savings and facilitate overall physical capital formation, than in the form of improving investment efficiency. Even so, there is a further pattern that is difficult to interpret the weaker asset growth effect in the PACAP markets relative to the US. At the first glance, this seems to imply that the financial system in the PACAP region is more efficient in valuing investment opportunities and in allocating capital, despite the obvious fact that the U.S. financial markets are much more developed and U.S. investors are more sophisticated. In order to have a better perspective for understanding and interpreting the differences in the asset growth effect between the two regions, as well as the differences across the nine PACAP markets, we perform a series of further empirical analysis. A. Homogeneity of Asset Growth First, we check if the differential stock return spread to asset growth deciles is due to less dispersion of firm growth, or lower sensitivity of stock returns to asset growth difference in the PACAP region. The cross-sectional dispersion of firm growth could be due to factors unrelated to market efficiency. For example, firm growth may be more homogeneous because firms are exposed to similar economic shocks, or they face similar financial constraints for expansion both are likely for the PACAP economies. We measure dispersion of firm growth in each market by the difference in average asset growth rates between the bottom and top deciles of firms sorted on asset growth. The results are reported in Table 5. Pooled over all PACAP markets, the D10-D1 spread in asset growth rates is 63.8%, significantly less than the spread in the U.S., 149.2% The dispersion of firm growth can also be measured by the cross-sectional standard deviation of asset growth rate. As shown in Panel B of Table 1, the standard deviation is 30.22% for the PACAP market, quite close to the 30.70% standard deviation for the U.S. market. Regardless of the measure of dispersion, 13

16 There are also variations in the asset growth spreads across the nine PACAP economics, from 42.5% in Japan to 107% in Thailand. We further compute the correlations between asset growth spread and the return spread across the nine markets. The time series average of the correlations is significantly negative, at -7.8%. The evidence suggests that indeed PACAP firm growth is in general more homogeneous relative to the U.S., and the degree of homogeneity is significantly related to the difference in asset growth effect across PACAP markets. Also related to asset growth dispersions, the other cause is that stock returns of different economies may exhibit heterogeneous sensitivities to asset growth. To see this, we measure the sensitivity of stock return to asset growth by the standardized return spread the ratio of decile return spread (D10-D1 difference in stock returns) to the asset growth spread (D10-D1 difference in asset growth rates). The results are also reported in Table 5. Pooled over the nine PACAP markets, the standardized return spread is , significantly less negative than the standardized return spread for the U.S. market, There are also significant variations in the standardized return spreads across the nine PACAP markets, and the correlations across the nine markets between the standardized return spread and the total return spread is also significant. Therefore, homogeneity of firm growth cannot completely explain away the differential effect of asset growth on stock returns, between the U.S. and the PACAP region, or across the PACAP markets; the sensitivity of stock return to asset growth also matters. B. Persistence in Asset Growth What determines the sensitivity of stock return to asset growth? A potential factor is investors over-reaction to past growth information. However, if the extrapolative bias is indeed the cause, then it is really troubling to conclude that such bias is stronger in the more efficient U.S. stock market. Instead, we conjecture that this is due to a somewhat mechanical effect. Suppose that in both the U.S. and PACAP markets investors infer firms future performance from past growth in exactly the same way, but in the PACAP market firm growth is more persistent, then the valuation bias (i.e., stock return spread the bottom line of our analysis here is that the dispersion of asset growth cannot completely explain away the differential effect of asset growth on stock returns. 14

17 between extreme asset growth decile) due to such extrapolation is less in the PACAP market. We perform the following cross-sectional regressions to examine the persistence of firm growth and profitability: AG i,t+τ = α i,τ + β i,τ AG i,t + ε i,t+τ (3) ROA i,t+τ = α i,τ + γ i,τ AG i,t + ε i,t+τ (4) where τ = 1, 2, 3, 4, and 5. Higher β i,τ and γ i,τ mean that asset growth and profitability are more persistent over time. The regression is performed in each year across firms in each market. The time series averages of the coefficients and the time series t-statistics are reported in Table 6. Indeed, for both regressions, the coefficients β i,τ and γ i,τ for the PACAP market tend to be significantly higher than those in the U.S., suggesting higher persistence in growth and profitability in the former. Therefore, the stronger asset growth effect in U.S. may not necessarily suggest stronger extrapolative behavior. Instead, it could be a result of less persistent asset growth rates in U.S. C. Overinvestment Tendency Naturally, the question becomes: why are growth and profitability more persistent in PACAP economies? Here, we examine the role of overinvestment. Cooper et al. (2007) point out that overinvestment is a likely cause of the asset growth effect. Overinvestment typically arises from the conflict of interest between corporate managers and diverse shareholders, and Titman, Wei, and Xie (2004) provide more direct evidence a negative cross-sectional relation between capital expenditure and future stock returns in the U.S. market. In contrast, corporate ownership is often highly concentrated in the PACAP markets, and in such firms, managerial interests are more aligned with controlling shareholders. So far there is no strong empirical evidence that in concentrated firms controlling shareholders engage in empire-building and overinvestment to expropriate minority shareholders. 13 Therefore, it is possible that stronger persistence in 13 The only related evidence, to our knowledge, is Titman, Wei, and Xie (2002), who find that in Japan, there is a negative relation between capital expenditures and subsequent stock returns among keiretsu firms (firms belong to large corporate business groups) but a positive relation among independent firms. 15

18 asset growth and profitability, and the resulting weaker stock return effect, are due to less overinvestment behavior in the PACAP region. To test this hypothesis, we consider the following decomposition of asset growth: Total asset growth (AG) = Cash growth (ΔCash) + Non-cash current assets growth (ΔCAsset) + Net fixed assets growth (ΔPPE) + Other assets growth (ΔOAsset) (5) We further define a variable Capex (capital expenditure), which combines the effect of growth in fixed assets and depreciation. Details for constructing these variables are provided in Appendix A. For each PACAP market, we compute the growth of each asset component above for each stock, and then calculate the averages for each stock decile sorted on asset growth. In Panel A of Table 7, we report the D10-D1 spread of the growth of each asset component. Results for the pooled PACAP market, and for the US, are also reported. In the U.S., capital expenditure is the most important factor driving asset growth. Total asset growth due to capital expenditure (fixed asset growth) is 57.8% (51.3%), while those due to growth in cash, non-cash current assets, and other assets are, respectively, 33.5%, 30.8%, and 31.8%. In the PACAP markets, capital expenditure s contribution to asset growth is lower relative to other asset components. Total asset growth due to capital expenditure (fixed asset growth) is 25.9% (15.8%), while those due to cash, non-cash current assets, and other assets are, respectively, 14.2%, 21.0%, and 12.8%. Capital expenditure s share of asset growth is significantly lower in PACAP markets than in the U.S. market. In addition, the results show that growth of various asset components is more homogeneous in the PACAP market. We further compute the standardized return spread for each asset growth component in each market. To compute the standardized return spread for capital expenditure, for example, we first sort stocks into deciles based on capital expenditure, compute the stock return spread between the top and the bottom stock decile, and then divide the stock return spread by the decile spread in capital expenditure in a way similar to the standardized return spread for asset growth reported in Table 5. Standardized return spreads for other asset components are similarly constructed. 16

19 Essentially, these spreads measure the sensitivities of stock returns to the growth of asset components. The time series averages of standardized return spreads are reported in Panel B of Table 7. For the U.S. market, the standardized return spreads for non-cash current asset growth, fixed assets growth, growth in other assets, and capital expenditure, are all significantly negative. The standardized return spread for cash growth is negative but statistically insignificant. The pattern is consistent with that reported in CGS. For the pooled PACAP market, the pattern is very similar, with significantly negative standardized return spreads for all asset components except for cash. Further, the standardized return spreads for non-cash current asset growth and for fixed assets growth are significantly higher (less negative) than those in the US. Finally, standardized return spread for capital expenditure is also less negative in the PACAP region than in the US, with a t-statistic of 1.83 for the difference (significant at 10% critical level). The results are somewhat supportive of the hypothesis that there is less overinvestment tendency for PACAP firms. However, because capital expenditure plays a less important role in driving asset growth in the PACAP region, and because the difference in stock return sensitivity to capital expenditure is only marginally significant, we consider such evidence relatively weak. 14 More convincing conclusions can be reached by directly examining the relationship between corporate control and governance characteristics and firms overinvestment behavior. However, we do not have ownership and governance data for the PACAP markets to perform such analysis. D. The Financing Effect Many studies show that in the U.S. market, firms raising external financing, either in the form of equity or debt, subsequently have lower stock returns (see, e.g., Bradshaw, Sloan, and Richardson, 2006). The pattern may be related to firms overinvestment tendency, but could also be because corporate managers who have superior information about firm value can time the market when issuing stocks and bonds (e.g., Loughran and Ritter, 1995 and 1997; Baker and Wurgler, 2002; Graham and Harvey, 2001). An 14 In Table 10, when we perform multivariate regression analysis, we find that capital expenditure is not a significant factor affecting the cross-market difference in the asset growth effect. 17

20 important difference between the financial systems in the U.S. and in the PACAP region is that in the former, firms heavily rely on capital markets to raise external financing, while in the latter banking systems are more developed. Since banks have direct access to corporate financial information and they can structure the loan contracts in an effective way to monitor a firm s business operation (Townsend, 1979; Diamond, 1991; Puri, 1996, 1999), loan financing has an important monitoring effect on a firm s business performance, potentially reducing firms overinvestment tendency. Further, information asymmetry between corporate managers or controlling shareholders and lending banks may be less severe, and timing bank loans are theoretically more difficult. 15 For these two reasons, we expect firms in bank-oriented systems are more likely to have a weaker asset growth effect. Finally, financial systems in the PACAP region is on average less developed, and in such economics firms tend to rely more in internal financing, which removes market timing incentives from financing decisions. This may also potentially explain the weaker asset growth effect in the PACAP region. To analyze these hypotheses, we break down asset growth from the financing side of balance sheet: Total asset growth (AG) = Growth of operating liabilities (ΔOpLiab) + Debt financing (ΔDebt) + Equity financing (ΔEquity) + Growth of retained earnings (ΔRE) (6) Details of variable construction are provided in Appendix A. We also define a variable External Financing as the sum of growth of operating liabilities, debt financing, and equity financing. Meanwhile, internal financing refers to the growth of retained earnings. In Panel A of Table 8, we report the contribution of each financing component to the D10-D1 asset growth spread, for U.S., for each PACAP market, and for the pooled PACAP market. In the U.S. market, the asset growth spread due to external financing is 127.9%, vs. 20.5% for internal financing. Within external sources of financing, asset growth spread due to debt financing is 40.3% vs. 52% for equity financing. By contrast, 15 The empirical evidence on the relation between bank loan financing and stock returns is somewhat mixed in the U.S. market. On the one hand, several studies have documented positive stock return response to bank loan announcements; see, e.g., Mikkelson and Partch (1986), James (1987), and Lummer and McConnell (1989). On the other hand, Billett, Flannery, and Garfinkel (2006) find negative long-run stock performance after bank loan financing. 18

21 for the pooled PACAP market, the asset growth spread due to external financing is 41.7%, vs. 22.1% for internal financing. That is, internal financing is relatively more important compared with the U.S. Further, the asset growth spread due to debt financing is 18.5% vs. 7.8% for equity financing equity financing is much less important in driving asset growth in the PACAP markets. There are cross-sectional variations in these statistics; however, in general the spread of each financing component is smaller in the PACAP markets than in the U.S. That is, firms financing activities are more homogenous in the PACAP region. In Panel B of Table 8, we report the standardized return spreads for various financing components. The standardized return spreads are constructed in a way similar to those in Table 7, except that here they are with respect to the financing components, instead of asset components. For the U.S. market, stock returns are not significantly sensitive to growth of operating liabilities, or increase in retained earnings; but the standardized return spreads for both debt financing and equity financing are both significantly negative. For the pooled PACAP market, operating liabilities or increase in retained earnings (internal financing) is not significantly correlated with future returns. More interestingly, stock returns are not significantly to debt financing; on the other hand, the standardized spread for equity financing is significantly negative, and significantly more negative than that in the U.S. These results are by and large consistent with our hypothesis that characteristics of financial systems play an important role in explaining the differential asset growth effect. E. Multivariate Regression Analysis In this section, we jointly examine the various hypotheses proposed earlier on cross-market differences of the asset growth effect using panel data regressions. Before proceeding with details of regression analysis, we first provide relevant information about the financial systems in the nine PACAP markets, which will be used in the regressions. In Table 9, we report the relative importance of equity financing for each market, computed as the mean rank for a market across three variables the ratio of the aggregate stock market capitalization held by minorities to GNP; number of listed domestic firms relative to the population; and number of IPOs relative to the population. 19

22 This measure is developed by La Porta et al. (1997), with higher measures indicating greater importance of the stock market. The numbers are obtained from Table 5 of Pincus et al. (2007). The average relative importance of equity, weighted by a market s aggregate capitalization, of PACAP countries is while the relative importance of equity of the U.S. market is The difference is -6.01, significant at 1% level. We also provide, in the last column of the table, the specific types of financial system (bank or capital market) in each market. We decide whether an economy has a bank or market based financial system based on the importance of equity market. Using the U.S. as a benchmark, if an economy s score of importance of equity is greater (lower) than that of the U.S., we consider that economy as a market (bank)-based system. China and Korea have no data on the importance of equity market. We consider China has a bank-based financial system, as reported by Allen et al. (2005) that China s financial system is dominated by a large banking system that is mainly controlled by the four largest stateowned banks. Korea is also considered to be a bank-based economy, as indicated by Ferris et al. (2003). As reported in Table 8, Hong Kong, Singapore, Malaysia and the U.S. with a market-oriented financial system, equity financing contributes a major portion to total asset growth. We first perform a pooled multivariate regression model, after controlling for the calendar year fixed effect. The regression equations are specified as follows: Ret_Sprd j,t =α+β 1 AG_Sprd j,t +β 2 AG_Perst j,t +β 3 CapEx j,t +β 4 Internal j,t+β 5 ΔEquity j,t +β 6 ΔDebt j,t+β 7 Bank j+ β 8 LGDPRate j,t-1 + ε j,t (7) We regresses monthly return spread on various market-level characteristics, including asset growth spread and persistence, capital expenditure, internal financing, equity financing, debt financing, financial system, and lagged GDP growth rate. 16 We report the regression result (with calendar year effect) in Panel A of Table 10. The coefficient for asset growth spread (AG_Sprd) is (t = -1.98), suggesting that the larger the asset growth spread, the less stock return spread. Recalling that the return 16 We first estimate the correlation between debt financing, equity financing, and a country s financial system (Bank = 1 if a country has a bank-oriented financial system and 0 otherwise). We find a positive correlation between debt financing and a bank-based financial system, and a negative correlation between equity financing and a bank-based financial system. This indicates that firms in markets with a bankoriented financial system tend to grow their total assets more from debt financing side, and grow less from equity financing side. In addition, we find no evidence of multicollinearity. 20

23 spreads are negative across all markets, the less stock return spreads represent strong asset growth effect. Turning to coefficient for market-level asset growth persistence (AG_Perst), we find a significantly negative relation between return spreads and growth persistence (β = , t = -2.17). The result shows that weak asset growth effect is related to weak asset growth persistence (i.e., less negative return spread). The coefficient on market-level internal financing (Internal) is (t = 1.86), indicating that the asset growth effect is stronger for markets with a primary financing resource from internal financing (changes in retained earnings). We also find a significant and positive relation between asset growth effect and an indicator of financial system (Bank) and debt financing (Debt), which suggests that weak asset growth effects tend to occur in markets with a bank-based financial system, with a predominant use of debt financing relative to equity financing. There is a negative relation between CapEx and asset growth rate, but the relation is not significant. The coefficients for the GDP growth rate and equity financing are not significant at a conventional level. An implicit assumption for the above market-level regression analysis is that we assume each market plays an identically important role in the pooled regression analysis (we assign an equal weight for each market in the regression). This assumption may be problematic because some markets (e.g., Japan and China) exhibit important influence on other PACAP markets. Therefore, we conduct a regression analysis on firm level by introducing interaction variables between firm asset growth rate and various asset growth sources. Similar to the market-level analysis specified in equation (7), we include dummy variables to control for the calendar year and the country fixed effect. The regression equation is specified as follows: Ret i,t = α +β 1 AG i,t +β 2 AG*CapEx i,t +β 3 AG*Internal i,t + β 4 AG*ΔEquity i,t + β 5 AG*ΔDebt i,t + ε i,t (8) Panel B of Table 10 presents the regression results after controlling for country and calendar year effects. In model (1) we only include asset growth rate as an explanatory variable. The coefficient for asset growth is (t = -5.71), indicating an inverse relation between asset growth and stock returns. This further confirms our previous results using decile sorting portfolio (Table 2). We then interacting asset growth with various asset growth sources into model (2) to examine the effect of asset growth 21

24 sources on the stock return predictability of asset growth rate. The coefficient on interaction item AG*Internal is (t = 3.80), indicating a weakening effect of internal financing on the asset growth effect. That is, asset growth effect tends to be weak if a firm grows its total asset more from internal capital (changes in retained earnings). In addition, the interaction item AG*ΔDebt has the coefficient of (t = 2.04) and AG*ΔEquity has the coefficient of (t = -5.87). This suggests that the weak (strong) asset growth effect tends to occur for firms with a predominant use of debt (equity) financing, consistent with our previous findings on financing effect. Similar to the result of panel A, we do not find a significant coefficient for AG*CapEx. This indicates that asset growth effect is not related to capital expenditure (overinvestment). Taken together, the regress analysis of Table 10 shows that the asset growth effect tends to be weaker in markets (1) that have small asset growth spread; (2) that rely more on internal financing for asset growth, and (3) that have a bank-oriented financial system with a predominant use of debt financing. V. Concluding Remarks We show that there is a pervasive negative relation between asset growth and future stock returns in the nine Pacific basin markets. The evidence suggests that despite rapid economic growth in this region during the recent decades, there are significant inefficiencies in the region s financial system in terms of allocating financial resources across firms and in terms of valuing investment opportunities. Interestingly, this asset growth effect on stock returns is lower in magnitude in the Pacific basin region, relative to the evidence documented for the U.S. market. This seems to suggest that rather puzzling the financial systems in the PACAP region are more efficient. We launch a series empirical analysis on factors affecting cross-region and cross-market differences in the asset growth effect, to peel off a few layers of this puzzle. We find that the weaker asset growth effect in the Pacific Basin region is partially, but not completely, due to higher homogeneity of asset growth. We also find some supportive, although not conclusive, evidence that there is less overinvestment behavior in the Pacific Basin region. But perhaps the most relevant factor for explaining the differential asset growth effect is the characteristics of financial system the reliance on 22

25 internal finance and the dominance of banking system in the region significantly weakens the negative correlation between asset growth and subsequent stock returns. 23

26 Appendix A: Variables Definition Variable PACAP Item COMPUSTAT Item 1. TA: Firm total assets BAL9 DATA6 2. AG: A firm s total asset growth rate for year t is defined as the percentage change in total assets from fiscal year t - 2 to fiscal year t ΔCash: changes in cash and cash equivalents from the fiscal year t-2 to fiscal year t-1, scaled by total assets in the fiscal year t ΔCAsset: changes in current assets from the fiscal year t-2 to fiscal year t-1, scaled by total assets in the fiscal year t ΔPPE: changes in fixed assets from the fiscal year t-2 to fiscal year t-1, scaled by total assets in the fiscal year t CapEx: changes in fixed assets from the fiscal year t-2 to fiscal year t-1 plus the depreciation, scaled by total assets in the fiscal year t ΔOAssets: changes in other assets from the fiscal year t-2 to fiscal year t-1, scaled by total assets in the fiscal year t ΔEquity: changes in equity financing from the fiscal year t-2 to fiscal year t-1, scaled by total assets in the fiscal year t ΔDebt: changes in long-term debt financing (including the current portion of long-term debt) from the fiscal year t- 2 to fiscal year t-1, scaled by total assets in the fiscal year t ΔOpLiab: changes in operating liabilities from the fiscal year t-2 to fiscal year t-1, scaled by total assets in the fiscal year t External financing: changes in operating liabilities, long-term debt, and equity financing from the fiscal year t-2 to fiscal year t-1, scaled by total assets in the fiscal year t Internal financing (ΔRE): changes in retained earnings from the fiscal year t-2 to fiscal year t-1, scaled by total assets in the fiscal year t ROA: Return on total assets defined as firm operating income divided by total assets. 9 BAL 9 BAL9 BAL assets. 9 BAL 1 BAL1 BAL ( BAL 6 BAL1 ) ( BAL6 BAL9 9 BAL 7 BAL7 BAL ( BAL 7 ΔTA ( BAL18 ΔBAL4 ΔTA, where BAL9 is total t 1 BAL7 ) + ΔCash BAL9 ΔCAsset TA BAL1 Depreciation t 1 + BAL19t 1) ( BAL18 + ΔEquity BAL9 + ΔBAL11t 1 + ΔBAL5 BAL9 ΔRE 9 ΔStock TA + ΔDebt TA BAL20 BAL20 BAL INC 5 BAL9 ) ΔPPE ΔDebt + ΔOpliab BAL19 + ΔBAL6 ) DATA 6t 1 DATA6 DATA6t 2 assets. DATA 1 DATA1 DATA ( DATA4 6 DATA 8 DATA8 DATA ( DATA 8 ΔTA ΔDATA130 DATA1 ) ( DATA4 DATA6 6, where DATA6 is total t 1 DATA8 ) + ΔCash DATA6 ΔCAsset TA DATA1 Depreciation + ΔDATA60t 1 + ΔDATA38 DATA6 ΔDATA 9t 1 + ΔDATA34 DATA6 ΔTA ΔEquity ΔRE + ΔDebt TA 6 ΔStock TA DATA36 DATA36 DATA DATA 13 DATA6 ΔPPE ΔDebt + ΔOpliab ΔDATA36 ) 24

27 References Allen, F., and Gale, D., 2000, Comparing financial systems, MIT Press, Cambridge, MA. Allen, F., J. Qian, and M. Qian, 2005, Law, finance, and economic growth in China, Journal of Financial Economics 77 (1), Baker, M. and J. Wurgler, 2002, Market timing and capital structure, Journal of Finance 57, Beck, T., R. Levine, and N. Loayza, 2000, Finance and the sources of growth, Journal of Financial Economics 58, Berglof, E., 1990, Capital structure as a mechanism of control: A comparison of financial systems, in: Masahiko, A., Gustafsson, B., Williamson, O.E. (Eds.), The Firm as a Nexus of Treaties. London: Sage, Billett, M. T., M. J. Flannery, and J. Garfinkel, 2006, Long-Run Underperformance Following External Finance Issues? Working Paper, University of Iowa. Boot, A. W., and A. V. Thakor, 1997, Financial system architecture, Review of Financial Studies 10, Bradshaw, M. T., S.A. Richardson, and R.G. Sloan, 2006, The relation between corporate financing activities, analysts forecasts and stock returns, Journal of Accounting and Economics 42, Chan, C. K., Y. Hamao, and J. Lakonishok, 1991, Fundamentals and stock returns in Japan, Journal of Finance 46 (5), Chang, R. P., McLeavey, D.W., Rhee, S. G., 1995, Short-term abnormal returns of the contrarian strategy in the Japanese stock market, Journal of Business Finance and Accounting 22 (7), Chui, A., S. Titman, and J. Wei, 2003, Momentum, legal systems and ownership structure: An analysis of Asian stock markets, Working Paper, University of Texas. Chui, A., and Wei, J., 1998, Book-to-market, firm size, and the turn-of-the-year effect: Evidence from Pacific-Basin emerging markets, Pacific-Basin Finance Journal 6 (1), Claessens, S., S. Djankov, and L. Lang, 2000, The separation of ownership and control in East Asian corporations, Journal of Financial Economics 58, Comerton-Forde, C., and J. Rydge, 2006, Current state of Asia-Pacific stock exchanges: A critical review of market design, Pacific-Basin Finance Journal 14 (1),

28 Cooper, M.J., H. Gulen, and M.J. Schill, 2007, Asset growth and stock returns, Journal of Finance, forthcoming. Datta, S., M. I. Datta, and A. Patel, 1998, Bank monitoring and the pricing of corporate public debt, Journal of Financial Economics 51, Demirgüç-Kunt, A., and R. Levine, 2008, Finance, financial sector policies, and long-run growth, Working Paper, World Bank - Development Research Group. Diamond, D.W., 1991, Monitoring and reputation: The choice between bank loans and directly placed debt, Journal of Political Economy 99 (4), Fama, E. F. and K. R. French, 1992, The cross-section of expected stock returns, Journal of Finance 47, Fama, E. F. and K. R. French, 1993, Common risk factors in the returns on stocks and bonds, Journal of Financial Economics 33, Fama, E. F. and K. R. French, 1998, Value versus growth: The international evidence, Journal of Finance 53, Fama, E. F. and J. MacBeth, 1973, Risk, return, and equilibrium: Empirical tests, Journal of Political Economy 81, Ferris, S. P. and K. A. Kim, and P. Kitsabunnarat, 2003, The costs (and benefits?) of diversified business groups: The case of Korean chaebols, Journal of Banking and Finance 27, Gibbons, Michael R., S. A. Ross, and J. Shanken, 1989, A Test of the efficiency of a given portfolio, Econometrica 57 (5), Graham, John R., and C. R. Harvey, 2001, The theory and practice of corporate finance: Evidence from the field, Journal of Financial Economics 60 (2-3), Hong, D., Lee, C.M.C., and Swaminathan, B., 2003, Earnings momentum in international markets, Working Paper, Cornell University, Ithaca, NY. James, Christopher, 1987, Some evidence on the uniqueness of bank loans, Journal of Financial Economics 19, Kang, J., Y. Kim, and R. M. Stulz, 1999, The underreaction hypothesis and the new issue puzzle: Evidence from Japan, Review of Financial Studies 12 (3), Kang, J., M. Liu, and S. Ni, 2002, Contrarian and momentum strategies in the China stock market: , Pacific-Basin Finance Journal 10,

29 Lakonishok, J., A. Shleifer, and R. Vishny, 1994, Contrarian Investment, Extrapolation, and Risk, Journal of Finance 49, La Porta, R., Lopez-de-Silanes, F., Shleifer, and A. Vishny, R., 1997, Legal determinants of external finance, Journal of Finance 52, La Porta, R., Lopez-de-Silanes, F., Shleifer, and A. Vishny, R., 2000, Investor protection and corporation governance, Journal of Financial Economics 58, Loughran, T. and J. Ritter, 1995, The new issues puzzle, Journal of Finance 50, Loughran, T. and J. Ritter, 1997, The operating performance of firms conducting seasoned equity offerings, Journal of Finance 52, Lummer, S. L., and J. J. McConnell, 1989, Further evidence on the bank lending process and the capital market response to bank loan agreements, Journal of Financial Economics 25, Mikkelson, W., Partch, M., 1986, Valuation effects of security offering and the issuance process, Journal of Financial Economics 15, Pincus, M., Rajgopal, S., Venkatachalam, and M., 2007, The accrual anomaly: International evidence, Accounting Review 82 (1), Puri, M., 1996, Commercial banks in investment banking: Conflict of interest or certification effect? Journal of Financial Economics 40, Puri, M., 1999, Commercial banks as underwriters: Implications for the going public process, Journal of Financial Economics 54, Rouwenhorst, K. G., 1998, International Momentum Strategies, Journal of Finance 53, Rouwenhorst, K. G., 1999, Local return factors and turnover in emerging stock markets, Journal of Finance 54, Shaffer, S., 2002, Firm size and economic growth, Economics Letters, 76 (2), Shumway, T., 1997, The delisting bias in CRSP data, Journal of Finance 52, Titman, S., K. Wei, and F. Xie, 2002, Corporate groups, capital investments and stock returns in Japan, Working Paper, University of Texas at Austin. Titman, S., K. Wei, and F. Xie, 2004, Capital Investments and Stock Returns, Journal of Financial and Quantitative Analysis 39,

30 Townsend, R., 1979, Optimal contracts and competitive markets with costly state verification, Journal of Economic Theory 21, Wang, C., 2004, Relative strength strategies in China s stock market: , Pacific-Basin Finance Journal 12,

31 Table 1: Summary Statistics This table reports the summary statistics of firm and market characteristics and asset growth rates in the Pacific- Basin capital market (PACAP) and the U.S. market. Except the last column (NUM), Panel A presents time-series average of firm and market characteristics in Japan (JP), China (CH), Hong Kong (HK), Taiwan (TW), Malaysia (MA), Korea (KR), Singapore (SG), Thailand (TH) and Indonesia (IND), the combined PACAP stock market (all PACAP) and the U.S. market during respective sample periods. PRICE is the average stock price of all firms in a market in December of year t. TURN is the average annual trading volume scaled by the shares outstanding in December of year t. RETURN (in percent) is the average annual return of a market during year t. Both equalweighted (EW) and value-weighted (VW) market index return are used. STD is measured as the standard deviation of monthly EW or VW market index returns during year t. CAP is the aggregate market capitalization in December of year t. Both price and market capitalization are converted to US$ based on exchange rates in December of year t between the local currencies and the US$. CAP is denoted in billions of US$. NUM is the number of listed firms that at least appear in the database for one time. Financial firms and firms with stock prices below the maximum of 100 times minimum tick size and 1 unit of local currency are excluded. For the U.S. sample, close-end funds, real estate investment trusts (REIT), and American Depository Receipt (ADR) are excluded. Panel B reports summary statistics of the asset growth rates in each of the PACAP markets, all PACAP and the U.S. market, including means, medians, and standard deviations, of asset growth rates are reported. Asset growth rates are computed as the percentage changes of total assets from fiscal year t-1 to fiscal year t-1. The last column reports the firm-year observations in each market. Panel A: Overview of Pacific-Basin Capital Market (PACAP) and U.S. Market Sample PRICE RETURN STD RETURN TURN Period (US$) (EW) (EW) (VW) STD (VW) CAP (Bil. $) NUM JP ,085 2,486 CH ,371 HK TW MA KR SG TH IND All PACAP ,927 7,461 U.S ,194 7,415 Panel B: Summary Statistics of Asset Growth Rates across PACAP Market and U.S. Market Stock Market Mean (%) Median (%) Standard Deviation (%) # Observations (Firm-Years) JP ,261 CH ,161 HK ,823 TW ,800 MA ,704 KR ,094 SG ,515 TH ,894 IND All PACAP ,203 U.S ,124 29

32 Table 2: Raw Returns across Assets Growth Deciles This table reports time-series averages of monthly raw stock returns in each of the asset growth decile in each of the nine PACAP markets, all PACAP, all PACAP without Japan, and the U.S. market. Panel A reports the timeseries averages of monthly raw returns of the decile portfolios sorted by asset growth rates in year t, defined as the percent change of total assets from fiscal year t-2 to fiscal year t-1. Portfolios are constructed in the end of June of year t, and held unchanged from July of year t to June of year t+1. Returns for portfolios of all PACAP and all PACAP without Japan are computed in terms of U.S. dollars. D10 portfolio contains stocks in the highest decile of asset growth rates and D1 portfolio contains stocks in the lowest decile of asset growth rates denotes the difference in monthly stock return between D10 and D1 portfolios. Panel B reports the monthly raw stock returns spreads (D10-D1) five years before and after portfolio formation in each of the nine PACAP countries, all PACAP, all PACAP without Japan, and the U.S. market. Year -i (+i) reports monthly stock return spread between D10 and D1 i th year before (after) portfolio formation. The stock return for each decile portfolio is the equal-weighted average of the stocks in that portfolio. The t-statistics are reported in the parentheses. Panel A: Monthly Returns across AG-sorted Deciles in 1-Year after Portfolio Formation (%) JP CH HK TW MA KR SG TH IND All PACAP PACAP without JP U.S. D1(Low) D10(High) (-2.76) (-1.82) (-2.09) (-0.28) (-1.88) (-2.78) (-2.47) (-2.94) (-2.03) (-3.94) (-3.10) (-9.54) Panel B: Monthly Return Spreads (D10-D1) within 5-Year around Portfolio Formation Year JP CH HK TW MA KR SG TH IND (-0.55) 0.28 (1.60) 0.28 (1.54) 0.34 (1.80) 0.65 (3.08) (-1.81) (-2.76) (-1.69) (-1.34) (-1.75) 0.38 (1.86) (-0.96) 0.07 (0.16) 0.17 (0.46) 0.31 (0.68) 0.59 (1.16) (-0.12) (-1.82) (-1.83) (-1.59) (-2.02) (-1.82) (-1.09) (-0.17) 0.00 (-0.01) 0.29 (0.59) 1.32 (2.56) (-0.60) (-2.09) (-1.46) (-1.68) (-0.83) (-0.12) 0.41 (0.88) 0.52 (1.01) 1.05 (2.17) 1.21 (2.28) 1.85 (3.13) (-0.11) (-0.28) (-0.02) (-0.01) 0.13 (0.24) 0.51 (0.98) 0.61 (1.24) 0.24 (0.46) 0.22 (0.43) 0.22 (0.43) (-0.04) (-0.85) (-1.88) (-1.82) (-1.50) (-1.45) (-0.88) 0.28 (0.81) 0.46 (1.36) 0.04 (0.12) 0.21 (0.59) 0.14 (0.35) (-2.58) (-2.78) (-1.96) (-2.00) (-2.39) (-1.86) (-2.07) 0.22 (0.54) (-0.11) 0.50 (1.17) 0.49 (1.22) (-1.86) (-2.47) (-0.95) (-1.29) (-1.02) (-1.21) (-1.37) 0.25 (0.48) (-0.49) 0.31 (0.56) 0.69 (1.28) (-1.63) (-2.94) (-1.41) (-1.23) (-2.60) (-0.91) (-0.52) (-1.09) 0.45 (0.41) 1.20 (1.04) 0.63 (0.54) (-0.54) (-2.03) (-0.01) (-1.35) (-0.66) (-0.68) All PACAP (-0.42) 0.18 (1.48) 0.26 (1.74) 0.35 (2.41) 0.75 (4.00) (-2.99) (-3.94) (-2.41) (-2.90) (-3.13) (-3.02) PACAP without JP (-0.95) 0.20 (1.82) 0.23 (1.74) 0.32 (2.37) 0.72 (4.38) (-3.03) (-3.10) (-2.55) (-2.63) (-3.00) (-3.13) U.S (-3.21) (-1.89) 0.07 (2.12) 0.94 (3.60) 1.17 (3.88) (-8.21) (-9.54) (-6.02) (-6.89) (-7.35) (-5.15) 30

33 Table 3: Three-Factor Alphas of Decile Portfolios Sorted on Assets Growth Panel A reports three-factor alphas in decile portfolios sorted by asset growth rates one year after portfolio formation. In July of year t, we sort stocks into deciles portfolios based on asset growth rates in year t, defined as the percentage changes of total assets from fiscal year t-2 to fiscal year t-1. Decile portfolios are constructed in the end of June year t and held unchanged from July of year t to June of year t+1. D10 is the decile with the highest asset growth and D1 is the decile with the lowest asset growth. Alphas for portfolios of all PACAP and all PACAP without Japan are computed in terms of U.S. dollars. Three-factor alpha is the intercept from the time-series regression of equal-weighted average returns of each decile portfolio on the three factors: RMRF t is the market return in excess of the risk free rate; SMB t and HML t are the monthly returns on size and book-to-market factors respectively denotes the difference in monthly stock return between D10 and D1 portfolios. F-statistics and the associated p values of Gibbons, Ross, and Shanken (1989) are reported to test the hypothesis that the regression intercepts for a set of ten portfolios sorted on asset growth rate are all zero. Panel B reports alphas spread (D10-D1) across the asset growth deciles five years before and after portfolio formation. Year -i (+i) shows monthly adjusted stock return spread between the top and bottom asset growth sorted stock deciles i th year before (after) portfolio formation. The t-statistics are reported in the parentheses. Panel A: Alphas across the AG-sorted Deciles in 1-Year after Portfolio Formation (%) JP CH HK TW MA KR SG TH IND All PACAP PACAP without JP U.S. D1(Low) D10(High) (-2.01) (-2.42) (-2.02) (-0.67) (-2.08) (-1.96) (-2.47) (-2.05) (-2.44) (-3.84) (-3.50) (-4.88) F(GRS) p(grs) Panel B: Alpha Spreads (D10-D1) across AG-sorted Deciles within 5-Year around Portfolio Formation (%) Year JP CH HK TW MA KR SG TH IND (1.02) 0.32 (1.94) 0.38 (2.14) 0.46 (2.24) 0.76 (3.80) (-0.70) (-2.01) (-0.55) (-0.45) (-1.02) 0.11 (-0.82) (-2.21) 0.04 (0.69) 0.03 (0.77) 0.16 (0.32) 0.38 (0.69) (-0.34) (-2.42) (-1.63) (-1.59) (-1.12) (-1.05) (-1.32) (-1.77) 0.47 (0.72) 0.33 (0.51) 0.39 (0.59) (-1.41) (-2.02) (-1.67) (-2.03) (-1.37) (-1.26) 0.32 (0.41) 0.43 (0.47) 1.14 (1.09) 0.87 (0.75) 0.67 (0.64) (-0.51) (-2.08) (-0.68) (-0.23) (-0.19) 0.17 (0.18) 0.08 (1.51) 0.03 (0.68) 0.04 (0.68) 0.05 (0.93) (-0.09) (-3.66) (-0.67) (-1.57) (-2.14) (-1.59) (-2.30) 0.17 (0.39) 0.41 (0.95) 0.16 (0.35) 0.20 (0.47) 0.14 (0.32) (-1.92) (-1.96) (-1.38) (-1.24) (-1.60) (-1.96) (-1.95) 0.06 (0.16) 0.13 (0.32) 0.53 (1.25) 0.52 (1.33) (-1.85) (-2.47) (-1.09) (-1.57) (-1.28) (-1.30) 0.57 (0.98) 0.35 (0.57) 0.09 (0.16) 0.20 (2.00) 0.76 (1.34) (-0.46) (-2.05) (-1.65) (-0.81) (-2.40) (-0.77) (-0.32) (-0.39) 0.44 (0.39) 0.92 (1.03) 0.58 (0.49) (-0.19) (-2.44) (-0.11) (-1.26) (-0.68) (-1.19) All PACAP (-0.69) 0.02 (1.32) 0.03 (1.68) 0.03 (2.26) 0.03 (3.62) (-2.42) (-3.84) (-3.12) (-2.56) (-2.56) (-2.32) PACAP without JP (-0.68) 0.01 (1.66) 0.04 (1.96) 0.03 (2.11) 0.04 (2.98) (-2.74) (-3.50) -0.2 (-3.71) (-2.10) (-2.00) (-2.14) U.S (-2.96) (-1.85) 0.08 (2.71) 0.04 (2.65) 0.62 (3.86) (-7.08) (-4.88) (-5.45) (-4.85) (-5.89) (-4.14) 31

34 Table 4: Monthly Returns and Alphas: Before and after Asian Financial Crisis for PACAP Market This table reports monthly stock returns and three-factor alphas across the asset growth deciles for all PACAP markets before and after Asian financial crisis (before 1997 and after 1998). In July of each year t, stocks are ranked into deciles based on assets growth rate, calculated as the percentage change of total assets from fiscal year t-2 to fiscal year t-1. Decile portfolios are formed in the end of June and held unchanged from July of year t to June of year t+1 for each stock decile. D10 portfolio contains stocks in the highest decile of assets growth and D1 portfolio contains stocks in the lowest decile of assets growth. Raw returns are the time series averages of monthly equal-weighted averaged raw stock returns of all PACAP markets, computed in terms of U.S. dollars. Three-factor alphas are the intercepts from time-series regressions of equal-weighted decile portfolio returns on three factors: RMRF t is the market return in excess of the risk free rate; SMB t and HML t are the monthly returns on size and book-to-market factors respectively denotes the return difference between D10 and D1 portfolios. The t-statistics are reported in the parentheses. D1 (Low) D10 (High) 10 1 Raw Return Three-Factor Alpha Before Crisis After Crisis Before Crisis After Crisis (4.75) (1.26) (0.89) (-1.25) (4.78) (1.65) (2.02) (-0.89) (4.88) (1.63) (2.10) (-2.05) (4.71) (1.40) (-1.64) (-1.86) (4.65) (1.33) (-1.59) (-0.99) (4.55) (1.73) (-1.79) (-2.00) (4.31) (1.30) (-1.08) (-1.67) (3.75) (1.35) (-1.72) (-0.69) (3.73) (1.78) (-1.77) (-2.11) (3.68) (1.81) (-2.52) (-3.52) (-3.30) (-1.76) (-2.05) (-2.01) 32

35 Table 5: Asset Growth Spread and Return Spread in the PACAP Market and U.S. Market This table reports the asset growth (AG) difference and standardized stock return spread, and three-factor alpha differences between the highest AG portfolio (D10) and the lowest AG portfolio (D1) for each of nine PACAP markets, all PACAP and the U.S. market. The asset growth rate is the percentage change in total assets from fiscal year t -2 to fiscal year t 1. Decile portfolios are formed in the end of June and held unchanged from July of year t to June of year t+1 for each stock decile. D10 portfolio contains stocks in the highest decile of assets growth and D1 portfolio contains stocks in the lowest decile of assets growth. AG spread of a market is the timeseries average of the differences between equal-weighted averaged asset growth rates of D10 stocks and those of D1 stocks. The standardized return spread of a market is the time series average of the stock return spreads between D10 and D1 portfolios scaled by the AG spreads of that market in the corresponding years. The standardized alpha spread of a market is the time-series averaged alpha spread between D10 and D1 stocks scaled by AG spreads of that market in the corresponding years. Standardized return spreads and alpha spreads for portfolios of all PACAP markets are computed in terms of U.S. dollars. We estimate 3-factor adjusted alpha for each stock based on its prior 12-month returns. The last two rows report the correlation of return spreads and standardized return spreads and the correlation of return spreads and asset growth in the PACAP markets. The t- statistics are reported in the parentheses. Market AG Spread Standardized Return Spread (%) Standardized Alpha Spread JP (-2.12) (-2.03) CH (-2.13) (-1.97) HK (-2.24) (-2.01) TW (0.25) (-0.52) MA (0.07) (-0.68) KR (-3.22) (-2.85) SG (-2.28) (-2.12) TH (-2.19) (-1.86) IND (-1.09) (-0.99) All PACAP (-2.48) (-2.35) U.S (-9.14) (-8.74) Diff (t-stat) (-7.78) (2.60) (2.19) Corr (Return Spread, Std Return Spread) in PACAP markets (84.15) (64.75) Corr (Return Spread, AG Spread) in PACAP markets (-3.55) (-2.87) 33

36 Table 6: Tests of Future Asset Growth and Returns on Assets: PACAP Markets versus U.S. In Model 1 we regresses the current year asset growth on prior 5-year s asset growth in each of the nine PACAP markets, all PACAP, and the U.S. market. The asset growth rate is defined as the percentage change in total assets from fiscal year t -2 to fiscal year t-1. The regression equation is: AG i,t+t = a i,t + β i,t AG i,t + ε i,t, where t = 1, 2, 3, 4, and 5. β i measures the asset growth persistence i th -year after the portfolio formation year. In Model 2, we regress the current year ROA on previous 5-year s asset growth rate. The regression equation is: ROA i,t+t = a i,t + γ i,t AG i,t + ε i,t, where t = 1, 2, 3, 4, and 5. γ i measures the asset growth persistence i th -year after the portfolio formation year. We first perform the time-series regressions for individual firms and then average them crosssectionally. Diff in the last row denotes the difference in the coefficients between all PACAP and the U.S. market. The last row reports the correlation of return spreads and 5-year average asset growth persistence and the correlation of return spreads and 5-year average ROA-AG sensitivity in PACAP markets. The t- statistics are reported in the parentheses. JP CH HK TW MA KR SG TH IND All PACAP U.S. Diff Model 1 Model 2 β i,1 β i,2 β i,3 β i,4 β i,5 γ i,1 γ i,2 γ i,3 γ i,4 γ i, (5.10) (8.07) (6.05) (3.04) (3.46) (7.90) (6.00) (5.01) (4.79) (4.23) (4.91) (5.82) (2.46) (-0.13) (-2.85) (3.28) (1.85) (3.83) (-0.21) (-0.58) (7.37) (4.49) (1.88) (-0.60) (-0.74) (2.18) (2.06) (0.49) (1.77) (1.71) (3.63) (2.58) (1.27) (5.48) (0.30) (3.06) (1.33) (2.93) (2.55) (2.45) (4.38) (4.97) (1.00) (5.21) (-0.67) (-1.71) (-1.34) (-1.06) (-0.49) (0.11) (4.40) (2.22) (1.57) (0.12) (-2.13) (-0.20) (1.65) (0.27) (-0.05) (-0.38) (4.22) (2.77) (-0.30) (0.04) (2.32) (1.85) (1.23) (0.98) (1.27) (1.00) (5.27) (2.94) (0.79) (1.22) (1.21) (-2.54) (-1.28) (-2.23) (-0.52) (-1.85) (3.39) (0.55) (0.35) (1.35) (-0.86) (0.37) (0.18) (0.37) (-0.41) (-1.69) (13.40) (11.60) (9.94) (5.19) (1.32) (5.29) (3.22) (2.20) (2.17) (1.59) (8.62) (5.30) (6.24) (4.67) (3.86) (12.57) (-1.21) (-3.71) (-3.71) (-3.15) (1.87) (2.36) (1.17) (2.02) (1.72) (2.26) (2.71) (3.41) (2.98) (2.46) Corr (Return Spread, 5-Year Average ROA (2.59) AG Sensitivity) in PACAP (2.27) Corr (Return Spread, 5-Year Average AG- Persistence) in PACAP 34

37 Table 7: Analysis of Sources to Total Asset Growth - Investment Side Panel A reports the differences between D10 and D1 portfolios sorted by asset growth rates. In July of year t, we sort stocks into deciles portfolios based on asset growth rates from fiscal year t-2 to fiscal year t-1. D10 is the decile with the highest asset growth and D1 is the decile with the lowest asset growth. Decile portfolios are held unchanged from July of year t to June of year t+1. ΔCash, ΔCAsset, ΔPPE, ΔOAssets, and CapEx are the changes in cash and cash equivalents, current assets, PPE, other assets, and PPE plus depreciation from the fiscal year t-2 to fiscal year t-1, scaled by total assets in the fiscal year t-2. See Appendix A for variables definitions. Diff denotes the difference in AG components between all PACAP and the U.S. market. Panel B of this table reports the standardized return spread corresponding to each investment side items. To compute standardized return spreads, we scale the monthly return spreads by the corresponding asset components. Standardized return spreads for portfolios of all PACAP markets are computed in terms of U.S. dollars. Diff denotes the difference in the standardized return spreads between all PACAP and the U.S. market. The last two rows report the correlation of return spreads and standardized return spreads of specific investment-side items and the correlation of return spreads and spreads of investment-side items. The t- statistics are reported in the parentheses. Panel A: Spreads of Investment-side Items Sorted by Asset Growth ΔCash ΔCAsset ΔPPE ΔOAsset CapEx JP CH HK TW MA KR SG TH IND All PACAP U.S Diff (t-stat) (-1.31) (-1.58) (-2.41) (-1.96) (-2.33) Panel B: Standardized Return Spread (%) ΔCash ΔCAsset ΔPPE ΔOAsset CapEx JP (-1.21) (-1.74) (-2.08) (-1.68) (-1.77) CH (-0.18) (-0.80) (1.09) (-0.62) (-0.55) HK (0.99) (-1.94) (-1.59) (0.16) (-1.59) TW (1.31) (-0.65) (0.27) (0.12) (0.27) MA (-0.27) (0.49) (0.05) (-0.61) (0.05) KR (0.75) (-3.45) (-2.37) (-1.95) (-2.37) SG (-0.84) (-0.90) (-1.35) (-1.13) (-1.35) TH (-0.46) (-1.17) (-0.99) (-3.00) (-0.99) IND (-0.17) (-0.63) (-0.44) (-0.13) (-0.44) All PACAP (-0.25) (-3.65) (-3.05) (-2.85) (-3.22) U.S (-1.53) (-8.24) (-9.45) (-7.02) (-5.39) Diff (t-stat) (1.29) (2.75) (2.14) (1.56) (1.83) Corr (Return Spread, Std Return Spread) in PACAP (13.54) (25.58) (18.22) (18.08) (28.43) (t-stat) Corr (Return Spreads, AG components) in PACAP (t-stat) (-1.11) (-1.80) (-0.76) (-1.34) (-0.68) 35

38 Table 8: Analysis of Sources to Total Asset Growth - Financing Side Panel A of this table reports the differences between D10 and D1 stocks sorted by asset growth rates and the corresponding standardized return spreads. In July of year t, we sort stocks into deciles portfolios based on asset growth rates from fiscal year t-2 to fiscal year t-1. D10 is the decile with the highest asset growth and D1 is the decile with the lowest asset growth. Decile portfolios are held unchanged from July of year t to June of year t+1. ΔOpliab, ΔDebt, ΔEquity, External, and Internal are the changes in operating liabilities, long-term debt, equity financing, external financing, and retained earnings from the fiscal year t-2 to fiscal year t-1, scaled by total assets in the fiscal year t-2. See Appendix A for variables definitions. Diff denotes the difference in AG components between all PACAP and the U.S. market. Panel B reports standardized return spread corresponding to each financing side item. To compute standardized return spreads, we scale the monthly raw return spreads by the corresponding asset components. Standardized return spreads for portfolios of all PACAP markets are computed in terms of U.S. dollars. Diff denotes the difference in the standardized return spreads between all PACAP and the U.S. market. The last two rows report the correlation of return spreads and the standardized return spreads of specific financing-side items and the correlation of return spreads and the spreads of financing-side items. The t-statistics are reported in the parentheses. Panel A: Spreads of Financing-side Components ΔOpliab ΔDebt ΔEquity External ΔRE/Internal JP CH HK TW MA KR SG TH IND All PACAP U.S Diff (t-stat) (-1.48) (-1.95) (-2.41) (-2.84) (0.48) Panel B: Standardized Return Spreads (%) ΔOpliab ΔDebt ΔEquity External ΔRE/Internal JP (-0.60) (1.09) (-1.76) (-1.89) (-1.00) CH (-1.02) (1.09) (0.73) (0.99) (-0.14) HK (-0.36) (-1.09) (-1.33) (-2.90) (-0.85) TW (1.78) (0.26) (-0.37) (-0.88) (1.54) MA (-0.81) (-0.65) (1.41) (-0.52) (0.81) KR (-1.76) (3.95) (-0.30) (-4.09) (-1.01) SG (0.80) (-1.98) (2.01) (-2.09) (0.38) TH (0.41) (-2.27) (-1.20) (-2.78) (-1.31) IND (-0.85) (0.68) (-1.33) (-0.31) (0.49) All PACAP (-1.21) (-0.05) (-3.37) (-1.90) (-0.48) U.S (-1.24) (-8.48) (-5.14) (-8.10) (-1.58) Diff (t-stat) (1.48) (3.01) (-3.03) (1.35) (2.03) Corr (Return Spread, Std Return Spread) in PACAP (18.86) (21.17) (15.83) (15.79) (25.04) (t-stat) Corr (Return Spread, AG component) in PACAP (t-stat) (-0.97) (1.56) (-1.14) (-1.40) (1.91) 36

39 Table 9: Bank-Based versus Market-Based Financial Systems The importance of equity refers to the mean rank for a market across three variables (ratio of the aggregate stock market capitalization held by minorities to GNP; the number of listed domestic firms relative to the population; and number of IPOs relative to the population) with higher scores indicating greater importance of the stock market. Bank-Based is an indicator variable equal to 1 for bank-based financial system, and equal to 0 for market-based financial system. The last row reports the differences for the variables between the PACAP markets and the U.S. market. The t-statistic of the difference is reported in parenthesis. Market Importance of Equity Bank-based JP CH - 1 HK TW MA KR - 1 SG TH IND All PACAP U.S Diff (t-stat) (-6.15) 37

40 Table 10: Regression Analysis of Stock Returns on Asset Growth Determinants Panel A: Regression of Stock Return Spreads on Market-Level Sources of Asset Growth Ret_Sprd j,t =α+β 1 AG_Sprd j,t +β 2 AG_Perst j,t +β 3 CapEx j,t +β 4 Internal j,t+β 5 ΔEquity j,t+β 6 ΔDebt j,t +β 7 Bank j,t+β 8 LGDPRate j,t-1 +ε j,t Variables Estimate Intercept (-0.23) AG_Sprd ** (-1.98) AG_Perst 0.004** (2.17) CapEx (-0.75) Internal 0.003* (1.86) ΔEquity (-0.84) ΔDebt 0.005** (2.44) Bank 0.015** (1.96) LGDPRate (0.45) Year Fixed Effect Yes Adj. R N 178 Panel B: Regression of Stock Returns on Firm-Level Source of Asset Growth Ret i,t = α + β 1 AG i,t + β 2 AG*CapEx i,t +β 3 AG*Internal i,t + β 4 AG*ΔEquity i,t + β 5 AG*ΔDebt i,t + ε i,t Variables Estimate (1) (2) Intercept 1.999*** 1.990*** (40.20) (40.15) AG ** (-5.71) (-2.22) AG*CapEx (-0.42) AG*Internal 0.023*** (3.80) AG*ΔEquity *** (-5.87) AG*ΔDebt 0.018** (2.04) Year Fixed Effect Yes Yes Country Fixed Effect Yes Yes Adj. R N 1,679,510 1,607,458 Panel A reports regression result of stock return spreads (D10 D1) on market-level source of asset growth, after controlling for year effect. The t- statistics are in parentheses. *, ** and *** indicate significance at the 10%, 5%, and 1% levels, respectively. Variables are defined as follows: Ret_Sprd j,t: The average return spreads between the highest and the lowest asset growth decile (D10 D1) for market j at year t. AG_Sprd j,t : The rank of average asset growth spread between the highest and the lowest asset growth decile (D10 D1) for amrket j at year t. We first estimate decile asset growth spread for each market in each year t, and then average asset growth spreads across year to get average market level asset growth spread. We rank market-level asset growth spread in each year into five quintiles, with 1 (5) represents the lowest (highest) rank. 38

41 AG_Perst j,t : CapEx j,t : Internal j,t : ΔEquity j,t : ΔDebt j,t : Bank j : LGDPRate j,t-1 : The asset growth persistence for market j at year t. We first estimate firm-level coefficient on asset growth based on Equation (7), with τ = 1. We then average firm-level coefficient across the market and year t to get market level asset growth persistence. We rank market-level asset growth persistence in each year into five quintiles, with 1 (5) represents the lowest (highest) rank. Capital expenditure for market j at year t. We first estimate the firm-level capital expenditure (changes in firm fixed assets plus depreciation scaled by total assets), and then we average firm-level capital expenditure across the market and year to get market level measurement. We rank market-level capital expenditure in each year into five quintiles, with 1 (5) represents the lowest (highest) rank. Internal capital financing for market j at year t. We first estimate the firm-level year over year changes in retained earnings scaled by total assets, and then we average firm-level internal financing across the market and year to get market level measurement. We rank market-level internal financing in each year into five quintiles, with 1 (5) represents the lowest (highest) rank. Changes in equity financing level for market j at year t. We first estimate the firm-level equity financing (firm common equity scaled by total assets), and then we average firm-level equity financing across the market and year to get market level measurement. We rank market-level equity financing in each year into five quintiles, with 1 (5) represents the lowest (highest) rank. Changes in debt financing level for market j at year t. We first estimate the firmlevel changes in debt financing (year over year changes in firm long-term debt, including current portion of long-term debt, scaled by total assets), and then we average firm-level debt financing across the market and year to get market level measurement. We rank market-level debt financing in each year into five quintiles, with 1 (5) represents the lowest (highest) rank. An indicator variable that takes 1 if a market j has a bank-based financial system and 0 if it has a market-based financial system. Lag GDP growth rate of j at the year t-1. Panel B reports regression result of stock returns on firm-level source of asset growth. The variables are defined as follows: Ret i,t : Monthly stock returns for firm i at year t; AG i,t : Total asset growth rate for firm i at year t; AG i,t *CapEx i,t : The interaction variable between asset growth and capital expenditure. CapEx i,t is capital expenditure for firm i at year t, measured as year over year changes in firm fixed assets plus depreciation scaled by total assets; AG i,t *Internal i,t: The interaction variable between asset growth and internal financing. Internal i,t is internal financing for firm i at year t, measured as year over year changes in firm retained earnings scaled by total assets; AG i,t *ΔEquity i,t : The interaction variable between asset growth and changes in equity financing. ΔEquity i,t is changes in equity financing for firm i at year t, measured as year over year changes in firm common equity scaled by total assets; AG i,t *ΔDebt i,t : The interaction variable between asset growth and changes in debt financing. ΔDebt i,t is changes in debt financing for firm i at year t, measured as changes in firm long-term debt (including current portion of long-term debt) scaled by total assets; 39

42 Figure 1: Plots of Return Spreads in PACAP Markets and U.S. Market 2 CHINA 2 HONGKONG 2 INDONESIA JAPAN SINGPORE PACAP KOREA THAILAND PACAP w/o JAPAN MALAYSIA TAIWAN U.S This figure plots the monthly return differences (in percent) of the stocks in the D10 and D1 asset growth sorted decile portfolios for each of the nine PACAP markets, all PACAP, all PACAP without Japan, and the U.S for 5-year before and after portfolio formation. D10 portfolio contains stocks in the highest decile of assets growth and D1 portfolio contains stocks in the lowest decile of assets growth. Returns for portfolios of all PACAP and all PACAP without Japan are computed in terms of U.S. dollars. 40

43 Figure 2: Plot of ROA across Assets Growth Decile 5-Year Before and After Portfolio Formation Decile Decile 10 This figure plots mean equal-weighted ROA for assets growth deciles in event time for the combined Pacific-Basin Stock Markets. In July of each year t, stocks are sorted into deciles based on assets growth rates and mean equal-weighted return on total assets (defined as operating income divided by total assets) is calculated. D10 portfolio contains stocks in the highest decile of assets growth and D1 portfolio contains stocks in the lowest decile of assets growth. Returns for the decile portfolios are computed in terms of U.S. dollars. We plot ROA across the asset growth decile 5-year around portfolio formation for each stock market and obtain the similar patterns. 41

44 Founded in 1892, the University of Rhode Island is one of eight land, urban, and sea grant universities in the United States. The 1,200-acre rural campus is less than ten miles from Narragansett Bay and highlights its traditions of natural resource, marine and urban related research. There are over 14,000 undergraduate and graduate students enrolled in seven degreegranting colleges representing 48 states and the District of Columbia. More than 500 international students represent 59 different countries. Eighteen percent of the freshman class graduated in the top ten percent of their high school classes. The teaching and research faculty numbers over 600 and the University offers 101 undergraduate programs and 86 advanced degree programs. URI students have received Rhodes, Fulbright, Truman, Goldwater, and Udall scholarships. There are over 80,000 active alumnae. The University of Rhode Island started to offer undergraduate business administration courses in In 1962, the MBA program was introduced and the PhD program began in the mid 1980s. The College of Business Administration is accredited by The AACSB International - The Association to Advance Collegiate Schools of Business in The College of Business enrolls over 1400 undergraduate students and more than 300 graduate students. Mission Our responsibility is to provide strong academic programs that instill excellence, confidence and strong leadership skills in our graduates. Our aim is to (1) promote critical and independent thinking, (2) foster personal responsibility and (3) develop students whose performance and commitment mark them as leaders contributing to the business community and society. The College will serve as a center for business scholarship, creative research and outreach activities to the citizens and institutions of the State of Rhode Island as well as the regional, national and international communities. The creation of this working paper series has been funded by an endowment established by William A. Orme, URI College of Business Administration, Class of 1949 and former head of the General Electric Foundation. This working paper series is intended to permit faculty members to obtain feedback on research activities before the research is submitted to academic and professional journals and professional associations for presentations. An award is presented annually for the most outstanding paper submitted. Ballentine Hall Quadrangle Univ. of Rhode Island Kingston, Rhode Island