1、Journal of Financial Economics 79 (2006) 257292/locate/jfecR2around the world: New theory and new tests$Li Jina, , Stewart C. MyersbaHarvard Business School, USAbMIT Sloan School of Management, USAReceived 9 February 2004; accepted 2 November 2004Available online 9 September 2005Abst

2、ractMorck, Yeung and Yu show that R2 is higher in countries with less developed nancial systems and poorer corporate governance. We show how control rights and information affect the division of risk bearing between managers and investors. Lack of transparency increases R2 by shifting rm-specic risk

3、 to managers. Opaque stocks with high R2s are also more likely to crash, that is, to deliver large negative returns. Using stock returns from 40 stock markets from 1990 to 2001, we nd strong positive relations betweeRn2 and several measures of opaqueness. These measures also explain the frequency of

4、 crashes.r 2005 Elsevier B.V. All rights reserved.JEL classification: G12; G14; G15; G38; N20Keywords: Corporate control; International nancial markets; Firm-specic risks; Information and market efciency; Crashes$We received helpful comments from Campbell Harvey, Simon Johnson, Dong Lee, Randall Mor

5、ck, Tobias Moskowitz, Jeremy Stein, Rene Stulz, Robert Taggart, Sheridan Titman and seminar participants at the University of Alberta, Boston College, Duke University, Harvard Business School, the University of North Carolina, Tulane University, the 2005 American Finance Association Meetings, the 4t

6、h Asian Corporate Governance Conference, the Emerging Market Finance Conference at the University of Virginiaandthe Conference on Globalizationand Financial Servicesin Emerging Economiesatthe World Bank. We thank the Research Computing Services Group at Harvard Business School, especially James Zeil

7、ter, for help with obtaining and understanding the data used in this paper. Anna Yu and Alvaro Vivanco provided helpful research assistance.Corresponding author. Tel.: +1 617 495 5590; fax: +1 617 496 6592.E-mail address: (L. Jin).0304-405X/$-see front matter r 2005 Elsevier B.V. All rig

8、hts reserved. doi:10.1016/j.jneco.2004.11.003258L. Jin, S.C. Myers / Journal of Financial Economics 79 (2006) 2572921. IntroductionMorck, Yeung and Yu (MYY, 2000) show that R2 and other measures of stock- market synchronicity are higher in countries with relatively low per-capita GDP and less develo

9、ped nancial systems. MYY and Campbell et al. (2001) also nd a secular decline in R2s in the United States from 1960 to 1997. These are intriguing and important results, which suggest that we may be able to learn about corporate nance and governance not just from the level of stock prices, or from sh

10、ort-term event studies, but also from the second or higher moments of stock returns.Of course there are many possible explanations for the inverse relation between nancial development and R2s, explanations that have nothing to do with corporate nance or governance. The pattern of R2s could reect hig

11、her macroeconomic risk or lack of diversication across industries in smaller, less-developed countries. MYY control for such effects assiduously. The cross-country pattern in R2s remains.MYY propose that differences in protection for investors property rights could explain the connection between nan

12、cial development and R2. MYY are on the right track, but it turns out that imperfect protection for investors does not affect R2 if the rm is completely transparent. Some degree of opaqueness (lack of transparency) is essential.We show how limited information affects the division of risk bearing bet

13、ween inside managers and outside investors. Insiders capture part of the rms operating cash ows. That is, they extract more cash than they would receive if investors property rights could be completely protected. The limits to capture are based on outside investors perception of the rms cash ow and

14、value. This perception is imperfect. Investors can see some changes in cash ow, but not all changes. When cash ows are higher than investors think, insiders capture increases. When cash ows are lower than investors think, insiders are forced to reduce capture if they want to keep running the rm. Inc

15、reased capture therefore reduces the amount of rm-specic risk absorbed by outside investors. An increase in opaqueness, combined with capture by insiders, leads to lower rm-specic risk for investors and to higher R2s.1.1. Opaqueness and investor protectionIn practice opaqueness and imperfect protect

16、ion of investors property rights go together and probably are mutually reinforcing. Nevertheless, we draw a distinction between opaqueness and poor protection of investors. That distinction is important to our model and tests.Poor protection without opaqueness is not enough to explain high R2s. Cons

17、ider a simple example. Suppose that poor protection of investors property rights allows insiders to capture half of the rms cash ows. Outside investors can see all of the rms cash ows (complete transparency) but cant prevent capture. Therefore the stock-market value of the rm is half its potential v

18、alue. Market value still uctuates as cash ows are realized and the rms overall value is updated, but by only half of the unexpected change in potential value. The percentage changes in market valueL. Jin, S.C. Myers / Journal of Financial Economics 79 (2006) 257292259are not affected by the insiders

19、 capture, however. Rate-of-return variance is unchanged. Investors capture half of any value change due to rm-specic information, and also half of any value change due to market risk, that is, market- wide, macroeconomic information. The proportions of rm-specic and local market volatility are unaff

20、ected by insiders capture. R2 is not affected.The story changes when the rm is not completely transparent. Change the example so that outside investors can observe all market-wide information, but only part of rm-specic information. Insiders still capture half of the rms cash ows on average, but the

21、y capture more when the hidden rm-specic information is positive and less when it is negative. Opaqueness therefore requires insiders to absorb some rm-specic variance. The rm-specic variance absorbed by investors is corre- spondingly lower. Of course investors absorb all market riskmacroeconomic in

22、formation is presumably common knowledge. Thus the ratio of market to total risk is increased by opaqueness. Higher R2s are caused by opaqueness, not by poor investor protection.The more opaque the rm, the greater the amount of hidden, rm-specic bad news that may arrive in a given span of time. The

23、amount of bad news that insiders are willing to absorb is limited. If a sufciently long run of bad rm-specic news is encountered, insiders give up and all the bad news comes out at once. Giving up means a large negative outlier in the distribution of returns. Therefore we also predict that stocks in

24、 more opaque countries are more likely to crash, that is, to deliver large negative returns, than stocks in relatively transparent countries.But why should insiders absorb any rm-specic risk on the downside? Why dont they hide the upside and reveal the downside? The answer, of course, is that inside

25、rs would always report bad news even when the true news is good.1 Bad news is credible only when reported at a cost, for example, the cost of hiring credible auditors and opening up the rms books to outsiders or the personal costs borne by insiders if they are ejected when bad news is released. Thes

26、e costs set the strike price of the insiders abandonment option.We have discussed the case of full transparency but limited protection of investorsthe case where insiders can capture cash ow in broad daylight with no impact on R2. Consider the opposite extreme case. Suppose that investors could enfo

27、rce their property rights fully and costlessly whenever they receive information about cash ows or rm value. They obtain every dollar of cash ow or value that is apparent to them. Nevertheless, if the rm is not completely transparent, insiders can still capture unexpected cash ows that are not perce

28、ived by investors. The insiders will soak up some rm-specic risk. Again, the more opaque the rm, the higher its R2.There is only one case in which greater opaqueness does not increase R2. The case is improbable but worth noting for completeness. Imagine an opaque rm run by a saintly manager who alwa

29、ys acts in shareholders interest, never taking a dollar more1We assume that insiders cant be forced to report completely and truthfully by some mechanism for punishment after the fact. If such a mechanism exists, it is evidently not effectiveotherwise rms would be transparent. In real life they are

30、not transparent, especially in developing economies.260L. Jin, S.C. Myers / Journal of Financial Economics 79 (2006) 257292or less than deserved. That manager does not have to soak up any rm-specic risk. All rm-specic good or bad news is absorbed by investors sooner or later, even if they cannot see

31、 the news as it happens.The properties of stock market returns in this case depend on how information is nally released. There are three possibilities. First, if the saintly manager reports everything promptly and credibly, opaqueness is eliminated and returns are not affected. Second, suppose that

32、hidden news is revealed after a stable lag. Then the average amount of rm-specic information released in any period is the same as for a transparent rm. Average rm-specic variance and R2 are not affected by delayed reporting. Third, if a stable lag is implausible, think of good or bad news accumulat

33、ing within the rm until the difference between intrinsic value and share price reaches a critical value. The news would then be released all at once, like a pressure vessel letting off steam. The releases would not affect average, long-run R2s, although we would see long tails in the distribution of

34、 stock returns. (We will control for kurtosis in our tests.)1.2. Summary of predictions and resultsOur theory makes two basic predictions. (1) Other things equal, R2s should be higher in countries where rms are more opaque (less transparent) to outside investors. (2) Crashes, that is, large, negativ

35、e return outliers, should be more common for rms in opaque countries. These are not market-wide crashes, but large, negative, market-adjusted returns on individual stocks.We test our models predictions using returns from 40 stock markets from 1990 to 2001. We conrm that R2 is higher in countries wit

36、h less developed nancial systems, and we nd evidence that R2 has been declining over time internationally. We also nd a positive relation between country-average R2s and several measures of opaqueness. Finally, we show that the frequency of large, negative, rm-specic returns is higher in markets wit

37、h high R2and in countries with less developed nancial markets. The frequency of large negative returns is also positively related to our measures of opaqueness.We do not claim that our model is the exclusive explanation of the differences in R2s across countries or over time. For example, countries

38、with less developed nancial markets are more vulnerable to episodes of political risk, which may translate into increased market risk. Higher market risk obviously generates higher R2s, other things equal. Our model includes this effect, and we control for cross- country differences in market risk i

39、n our tests.There may be differences in R2s within countries that could be traced to reasons other than differences in opaqueness. The tests in this paper are limited to differences in country averages, however.1.3. Prior researchThere is not much prior work on our topic. The two leading articles, M

40、YY (2000) and Campbell et al. (2001) are noted above. MYY (2000) suggest that poorL. Jin, S.C. Myers / Journal of Financial Economics 79 (2006) 257292261protection of investors could make rm-specic information less useful to arbitrageurs, decreasing the number of informed traders relative to noise t

41、raders. If the noise traders herd and trade the market rather than individual stocks, market risk may be higher in less developed nancial markets.2 Thus poor protection of investors could affectR2through two channels. First, poor protection could increase market risk. Second, poor protection could p

42、roxy for more opaqueness, whichshiftsrm-specic riskfromoutsideinvestorstoinsidemanagers. Wefocuson the second channel, but control for market risk.Related papers include Wurgler (2000), who nds that capital is more efciently allocated in countries with better legal protection for minority investors

43、and more rm-specic information in stock returns. Bushman et al. (2003) study two kinds of transparency: nancial transparency (the intensity and timeliness of nancial disclosures, including interpretation and coverage by analysts and the media) and governance transparency (for example, the identity,

44、remuneration and shareholdings of ofcers and directors). They nd that nancial transparency is lower in countries with a high share of state-owned enterprises and in countries where rms are more likely to be harmed by revealing sensitive information to competitors or local governments. Governance tra

45、nsparency is higher in countries with high levels of judicial efciency and common-law legal origin and in countries where stock markets are active and well developed.Durnev et al. (2004) nd that the magnitude of rm-specic variation in stock returns is positively related to the economic efciency of c

46、orporate investment. This result is consistent with our theory if more transparency encourages efcient investment. Durnev et al. (2003) show that stock returns are better predictors of future earnings changes for rms and industries with lower R2s. It seems that transparency and low R2s go together,

47、as we argue in this paper.Active security analysts should make rms more transparent. Chang et al. (2001) demonstrate that there is a wide variation in security-analyst activity across 47 countries. They suggest that transparency is primarily inuenced by countries legal systems and information infras

48、tructure. The organizational structure of rmswhether they operate as groups or conglomerates, for exampleseems to be less important. Our tests will use a measure of transparency based on the dispersion of analysts forecasts.Bris et al. (2003) explore international differences in the cost or feasibil

49、ity of short sales. They nd that restrictions on short selling reduce the amount of cross-sectional variation in equity returns. Their results are consistent with our theory if restrictions on short selling make the rm more opaque.Hong and Stein (2003) and Kirschenheiter and Melamud (2002) work out

50、theories that predict negative skewness (or reduced positive skewness) in returns. They saynothing about R2, however, and their models address different issuesthanconsidered here. Hong and Stein investigate how stock markets work when investors cannot agree. We assume that investors do agree (based

51、on what they can see), but2We see no reason why the effects of noise trading should be conned to market returns, however. De Long et al. (1990), suggest that the risks created by noise traders should be assumed to be market-wide and not rm-specic (p. 707). Their paper only considers a single-asset e

52、conomy, however.262L. Jin, S.C. Myers / Journal of Financial Economics 79 (2006) 257292that their property rights cannot be fully protected. Kirschenheiter and Melamud model nancial reporting, showing how managers rationally smooth earnings but occasionally take a big bath of exaggerated losses. We

53、dont model nancial reporting. We take transparency, and thus the quality of nancial reporting, as given exogenously for each country.Our results can also be compared to OHara (2003), who discusses the effects of public vs. private information on rm value. In that paper, private information gives inf

54、ormed traders an edge in forming optimal portfolios, leaving the uninformed traders with more risk to bear. The uninformed traders demand higher expected returns, thereby decreasing the value of rms that generate less public and more private information. We also distinguish public and private inform

55、ation, but assume that all outside investors are imperfectly informed. All private information is held by inside managers, so long as the inside managers do not give up and release it.Our paper also joins a larger number of more general studies of investor protection, corporate governance and the de

56、velopment of nancial markets around the world. These include La Porta et al. (1998, 2000), La Porta et al. (2002) and Rajan and Zingales (2001).The next section presents our theoretical model. Proofs and technical details are located in Appendix A. Section 3 describes the data, explains the setup of

57、 the empirical tests and presents our results. Section 4 wraps up the paper and notes several issues open for future research.2. A model of control and risk-bearing when outside investors have limited informationWe extend Myers (2000) to situations where outside investors cannot see what rm value re

58、ally is. The rm is partly opaque. If good (bad) news arrives that investors cannot see, inside managers capture more (less) cash ow than if the rm were completely transparent.The information received by investors in a particular rm is a combination of macroeconomic and rm-specic news. But the macroeconomic news can be separated, because it is common to all rms. We therefore assume that outside investors can observe a market factor that drives all stocks returns, as well as some rm-specic info