Stock Returns and Output Growth in Emerging and Advanced Economies
Author: Mr. Paolo Mauro

Contributor Notes

Author’s E-Mail Address: Pmauro@imf.org

This paper studies the correlation between output growth and lagged stock returns in a panel of emerging market economies and advanced economies. It finds that the correlation is as strong in emerging market economies as in advanced economies. Asset prices therefore contain valuable information to forecast output also in emerging market economies. Moreover, the paper finds that the strength of the correlation between output growth and lagged stock returns is significantly related to a number of stock market characteristics, such as the number of listed domestic companies and initial public offerings and, especially, a high market capitalization to GDP ratio and English legal origin.

Abstract

This paper studies the correlation between output growth and lagged stock returns in a panel of emerging market economies and advanced economies. It finds that the correlation is as strong in emerging market economies as in advanced economies. Asset prices therefore contain valuable information to forecast output also in emerging market economies. Moreover, the paper finds that the strength of the correlation between output growth and lagged stock returns is significantly related to a number of stock market characteristics, such as the number of listed domestic companies and initial public offerings and, especially, a high market capitalization to GDP ratio and English legal origin.

“There is no sense in building up a new enterprise at a cost greater than that at which a similar existing enterprise can be purchased; whilst there is an inducement to spend on a new project what may seem an extravagant sum, if it can be floated off on the stock exchange at an immediate profit.”

(J. M. Keynes, The General Theory of Employment, Interest and Money, 1936.)

“Cause and effect run from the economy to the stock market, never the reverse.”

(John Kenneth Galbraith, 1955)

“The stock market has predicted nine out of the last five recessions.”

(Paul Samuelson, 1960)

I. Introduction

The recent financial crises and associated output declines experienced by a number of emerging market economies have raised anew the issue of the links between financial and real variables, notably between stock market performance and economic activity. The empirical association between real stock returns and real economic activity has been analyzed in depth in the case of the United States and, to a lesser extent, other developed countries. However, this relationship remains surprisingly unexplored in the case of emerging market economies.2 Moreover, while a number of theories have been proposed to explain this empirical association, cross-country differences in its strength have not been used to discriminate among existing theories.

This paper analyzes the correlation between real stock returns and economic growth in a panel of emerging market and advanced economies. By doing so, it seeks to fill two gaps in the existing literature. First, it examines the extent of this correlation in emerging countries, and compares it to that observed in advanced countries. Second, it uses this larger group of countries to address the more general question of what type of countries tend to display a stronger association between output growth and lagged stock returns, and it relates the results to existing theories.

In the United States, the correlation between real economic activity and lagged real stock returns is positive and both statistically and economically significant. This correlation is well known and has been documented by several studies, including Fischer and Merton (1984), Barro (1990), Fama (1990), and Schwert (1990). A similar relationship holds in Canada (Barro, 1990), Japan, Germany, and the United Kingdom (Mullins and Wadhwani, 1989), the G-7 (Choi et al., 1999), and several other European countries (Asprem, 1989, and Wasserfallen, 1989 and 1990).

Although the correlation is significant and stock returns provide valuable information about future economic activity, they are of course not perfect as a leading indicator, as suggested by Samuelson’s oft-quoted adage reported above. Stock and Watson (1990 and 1998) show that the relationship between stock returns and economic growth has not been stable over time, and that the systematic predictive information of stock returns for future activity is also contained in other financial variables—such as yield spreads between 10-year and 3-month government bonds, or between T-bills and private commercial paper—in the case of the United States.3 Hu (1993) shows that the yield spread between long-term and short-term government bonds is a better predictor of future economic activity than stock market returns in the G-7 countries.

Several theories (reviewed in the next section) have been put forward to provide a rationale for this empirically observed correlation. All theories allow for the possibility that news about future output growth is reflected in stock returns. Some authors (including John Kenneth Galbraith, as in the quote reported above) argue that this is the only mechanism underlying the correlation. This mechanism does not involve stock price changes “causing” changes in output in an economic sense, though of course it does involve Granger-causality, with lagged stock returns predicting output growth. Others argue that, in addition, stock price changes that are not justified by fundamentals may cause changes in output, through mechanisms that typically rely on stock markets being sufficiently developed in a number of respects discussed below.

There are three reasons why analyzing the link between stock returns and output growth is especially interesting for emerging markets. First, leading indicators are relatively scarce in emerging markets. In particular, relatively low financial market liquidity and frequent changes in financial structure imply that other financial variables such as yield spreads are unlikely to be successful predictors of output, and in any case it is usually difficult to identify a relevant yield spread for a sufficiently long sample period. Given the speed with which stock market prices become available, it seems that the stock market could be a helpful leading indicator in forecasting economic growth. Immediately following the onset of the Asian crisis in 1997, there was no consensus on how deep an impact the financial crisis would have on economic growth. At that time, the extent of the decline in stock price indices might have provided useful guidance in answering that question. Second, volatility of returns in emerging markets is higher than in mature markets (Richards, 1996) and the same is true of output. While this might reflect greater noise, it might also yield a wealth of information, even though data on emerging countries are usually available for shorter sample periods than data on advanced countries. Moreover, if one were to find that stock price changes that are not justified by fundamentals really affected output, this would raise a number of policy issues that would be particularly relevant for emerging markets. Third, by expanding the sample of countries to include a number of emerging markets, it is possible systematically to address the question of whether the association between stock returns and output growth is stronger in some types of countries than others, depending not only on their level of economic development, but also on several indicators of the size and liquidity of, and legislation governing, their stock markets.

This paper presents two main findings. First, the empirical association between output growth and lagged stock returns is as strong in emerging countries as in advanced countries at the yearly frequency, and is still significant though somewhat weaker at the quarterly frequency. Second, the association is significantly stronger in countries that have high market capitalization, a large number of listed domestic companies and initial public offerings, and English (or non-French) origin of the regulations governing the stock market. Although all of these country characteristics are correlated, those with the best predictive power for whether a country has a strong association between output growth and stock returns are market capitalization and—a less robust finding—English (or non-French) legal origin of the regulations governing its stock market.

The remainder of this paper is organized as follows. Section II reviews existing theories of the link between stock returns and output growth and relates them to a number of characteristics of countries’ stock markets. Section III describes the data on stock market returns and economic growth and reports the estimation results. Section IV concludes.

II. Theories on the Link Between Stock Returns and Output Growth

The observed correlation between economic growth and stock returns has prompted a debate on the causal direction of the underlying relationship. This section, which draws heavily on Morck, Shleifer, and Vishny (1990), reviews five existing theories of this link and briefly summarizes the empirical literature that tests them. It then conjectures which country characteristics may predict the strength of the association between stock returns and output growth, according to each theory. The theories may be grouped into those according to which stock price movements not reflecting changes in future “fundamentals” cannot predict changes in output (the “passive informant” hypothesis and the “accurate active informant” hypothesis), and those according to which they can (the “faulty active informant”, the “financing” hypothesis, and the “stock market pressure on managers” hypothesis).

1) According to the “passive informant” hypothesis, the only mechanism underlying the correlation between stock returns and output growth is the following. Under the assumptions that stock prices reflect the present discounted value of all future dividends and that dividend growth is related to GDP growth, a correlation between this year’s stock returns and next year’s economic growth arises naturally: if next year’s economic growth is buoyant, news revealed this year will typically be positive, resulting in large stock price increases this year. All theories reviewed below also accept that the above mechanism plays a role, but they leave room for additional mechanisms.

2) Under the “accurate active informant” hypothesis, stock price changes provide managers with information about the market’s expectations of future economic developments. Managers base their investment decisions upon that information, thereby justifying the market’s expectations. The stock market therefore acts as a “sunspot,” bringing about one of several possible self-fulfilling equilibria. In this case, stock price changes turn out to be perfectly correlated with fundamentals.

3) In the “faulty active informant” hypothesis, managers’ decisions about investment are influenced by stock price movements, but managers cannot distinguish between movements reflecting fundamentals and those reflecting market “sentiment”. Stock market movements that are not motivated by fundamentals can therefore mislead managers into overinvesting or underinvesting compared with what later turns out to be warranted by fundamentals.

4) The “financing” hypothesis, based upon Tobin’s q theory (a formalization of Keynes’ reasoning in the quote reported above), argues that when stock prices are high compared to the replacement cost of capital, entrepreneurs are more likely to expand their activities by investing in new physical capital (possibly financed by issuing new shares of their company) rather than by purchasing existing firms on the stock market. Therefore, high stock returns will tend to be followed by high investment and economic growth. There is a debate on whether this mechanism allows scope for irrational movements in stock prices to affect real economic activity, as suggested by Fischer and Merton (1984), or whether rational managers will choose not to act upon changes in financing costs resulting from market sentiment rather than fundamentals, as argued by Blanchard, Rhee, and Summers (1993).

5) Finally, the “stock market pressure on managers” hypothesis suggests that stock price changes can affect investment even if they neither convey information nor change financing costs. If investors hold negative views on a firm’s prospects and drive down its stock price, managers may have to cut their investment projects to protect themselves from the possibility of being fired or taken over. Conversely, if investors are very optimistic about a firm’s prospects and lead its stock price to soar, managers may decide to adopt an aggressive investment strategy to avoid appearing too cautious.4

Although lack of data on fundamentals (i.e., dividends) for a sufficient number of countries prevents direct testing of these theories using the large cross-country data set analyzed in this paper, each of these theories has different implications for the country characteristics that can be expected to be systematically related to a strong link between stock returns and output growth. In particular, the following conjectures can be made. (These are summarized in Table 1, which reports whether each of the five theories would predict a certain country characteristic to lead to a closer association between stock returns and output growth, holding other things equal—including the other characteristics in the table.)

Table 1.

Implications of Various Theories for Strength of Growth-Returns Links

article image

  • Under the “passive informant” hypothesis, most country characteristics are unlikely to predict the strength of the association between stock returns and output growth, because good news about output leads to a capital gain on stocks regardless of country characteristics. One possible exception is the extent to which the companies listed on the stock market are representative of the economy as a whole, which might be proxied by high market capitalization and a large number of listed companies. However it seems that, for most countries in the sample analyzed in this paper, the group of firms listed on the stock market is fairly representative of the economy as a whole.

  • Similarly, under the “active informant” (whether “accurate” or “faulty”) hypothesis, market capitalization might matter, because a larger stock market implies that stock price changes provide information that managers will consider more relevant. Other characteristics, such as the number of initial public offerings or the legal origin of regulations governing the stock market, would not matter under these theories.

  • Under the “financing” hypothesis, countries with well-developed financial markets as proxied by high market capitalization and a large number of listed domestic companies and initial public offerings can be expected to display a stronger link between stock returns and growth. In fact, the financing mechanism is more likely to operate in these countries, because it relies on the possibility of issuing shares or taking over other companies.

  • Under the “stock market pressure on managers” hypothesis, countries in which managers are protected from shareholders, that is—according to La Porta, Lopez-de-Silanes, Shleifer and Vishny (1997, henceforth LLSV)—countries whose stock market regulations are of English origin as opposed to Scandinavian, German, or French origin (in increasing order of managers’ protection), should display a stronger association between stock returns and output growth. In fact, if market sentiment brings about a decline in a given company’s stock price, its managers will be able to continue pursuing a strategy of high investment only in countries where it would be difficult to fire them or to take over their firm.

  • Under all of the theories above, stock market turnover, as a proxy for liquidity, would seem to be a potential determinant of the strength of the association between stock returns and output growth whereas, controlling for the other financial development indicators, the degree of economic development or the distinction between “emerging” and “advanced” countries would not matter.

A final observation is that the discussion above has focused on the investment component of GDP, because of its primary role in accounting for GDP growth developments.5 However, stock price developments can affect consumption as well, through their impact on wealth, and one would expect this mechanism to be higher in countries where stocks constitute a large proportion of consumers’ portfolios—which may be proxied by market capitalization as a share of GDP.6

Testing of these theories has thus far relied almost exclusively on data from the United States. Using both aggregate and firm-level data from the United States, Morck, Shleifer, and Vishny (1990) argue that the stock market is largely a “sideshow” which merely reflects changes in expected output growth. Blanchard, Rhee, and Summers (1992) broadly support this view by showing that stock price movements independent of fundamentals have only a small impact on economic activity.

To my knowledge, the only existing study that exploits cross-country differences in the link between output growth and stock returns is that by Mullins and Wadhwani (1989), who find that the link is stronger in the United States and the United Kingdom than in Japan or Germany. They attribute this difference to the fact that the Anglo-Saxon countries are characterized by a greater possibility of takeovers, lower gearing ratios, more pervasive use of stock option schemes in managers’ remuneration, and a smaller role played by employees in decision making.

A possible reason why previous research has not focused on the link between stock returns and output growth in emerging markets might be a presumption that stock markets are smaller and less liquid in emerging markets than in advanced countries. Although indicators such as the ratio of market capitalization to GDP are, on average, higher in advanced economies than emerging markets, several emerging markets score much higher on these indicators than most advanced economies (Table 2). Indeed, there is considerable variation in market capitalization and turnover ratios, the number of listed domestic companies and initial public offerings, and legal origin among both emerging and advanced countries. This valuable information in used in the next section.

Table 2.

Indicators of Stock Market Development, Size, Liquidity, and Legal Origin

article image
Sources: Emerging Markets Factbook. IFC; International Financial Statistics: IMF; and LLSV (1997).Notes: Mkt Cap is market capitalization. NGDP per cap is nominal GDP per capita. The turnover ratio is total value traded during the year divided by average market capitalization. Listed Cos is the number of listed domestic companies. IPOs/Pop is the ratio of initial public offerings to population (in millions). GBR is English, FRA is French. DEU is German, and SCA is Scandinavian Anti-Dir. Rights is the index of antidirector rights from LLSV (1997)

III. Data and Estimation

This section briefly describes the data7 and reports the estimation results. It analyzes the association between output growth and lagged stock returns in the countries in the sample, focusing on whether its strength is the same for various groups of countries. First, it explores whether there are differences between emerging markets and advanced countries (following the definitions of the Investment Finance Corporation, IFC). The objective is to find out whether stock returns could help predict output growth in emerging countries. Second, it turns to possible differences among countries depending on variables that might be related to the strength of the association according to the theories discussed in Section II. The goal here is to understand the sources of the association rather than necessarily to make use of it to predict output growth.

Data on real stock returns (obtained as the difference between nominal stock returns and consumer price inflation) and real GDP growth are available at an annual frequency for a period of at least 22 years for eight emerging countries and 17 advanced countries. At a quarterly frequency, data on real stock returns and real GDP growth are available for at least ten years for six emerging countries and 18 advanced countries. To expand the number of emerging markets for which quarterly data are available, industrial production growth is also used, adding another seven emerging countries to the sample.

This paper uses stock returns instead of a variable such as q, which might be considered to have more solid theoretical underpinnings as a determinant of output growth, owing to two practical considerations. First, data on the replacement cost of capital, which are needed to construct q, are not available for the sample of countries considered. Second, Barro (1990) and Blanchard, Rhee, and Summers (1992) show that, in the United States, economic growth is more highly correlated with lagged stock returns than with a constructed q variable.

When testing whether the association between output growth and lagged stock returns is the same for different groups of countries, two estimation strategies are used: (i) individual-country regressions, followed by computing the average slope coefficient8 on lagged stock returns and, for univariate regressions, the average R2 coefficients for each group; and (ii) panel regressions allowing for individual-country fixed effects but restricting the slope coefficients to be the same for all countries belonging to the same group.

A. Emerging versus Advanced Countries

Considering the 25 countries for which at least twenty annual observations are available, the univariate correlation between real economic growth and real stock returns (lagged by one year) is positive in all countries except India, and significantly positive in five out of eight emerging markets and ten out of 17 advanced countries (Table 3).9 In the case of the advanced countries, the estimates are similar to those reported by previous studies. The slope coefficient typically ranges between 0.01 and 0.09, and amounts to 0.034 averaging over all countries in the sample. An increase in real stock returns by 10 percentage points is therefore typically associated with higher real economic growth by 0.34 percentage point. In seven out of the 25 countries considered, the R2 coefficient is 0.3 or higher. The average slope coefficient is slightly higher in emerging countries than advanced countries, though this

Table 3.

Output Growth and Lagged Stock Returns, Individual Country Regressions, 1971-98

article image
Sources: MSCI web site, Emerging Markets Factbook, IFC and International Financial Statistics, IMF.Note: Newey-West corrected t -statistics in parentheses.

The average R2 is computed setting the R2 in India to zero.

* indicates that the coefficient is significant at the 5 percent level.** indicates that the coefficient is significant at the 1 percent level.

result is reversed when the regression includes lagged growth as an additional independent variable (see below). Similarly, the R2 coefficient is higher, on average, in emerging countries than in advanced countries (setting the R2 coefficient to zero in the case of India).10

Following the existing literature, this paper focuses on the relationship between stock returns and economic growth by regressing real GDP growth in year t on GDP growth and stock returns in year t-1. While this functional form is intuitively appealing, it might seem somewhat restrictive, raising a number of issues involved in the econometric estimation. In particular, both real GDP and an index of real stock prices have unit roots. In most countries considered, tests of whether there is a long-run relationship between real GDP and real stock prices (based on cointegration using the Johansen approach or on the estimation of parameters in an autoregressive distributive lag regression) fail to yield a clear-cut long-run relationship with sensible coefficients. Appendix Table 14 reports the results obtained by estimating the relationship between the logarithms of real GDP and real stock prices using an unrestricted ARDL(2,2) form. In most cases, the restrictions needed to obtain the conventional form in log-differences used in the literature do not seem to be rejected.

Controlling for lagged economic growth, real economic growth and real stock returns (lagged by one year) at the annual frequency are positively and significantly associated in four out of eight emerging market countries and ten out of 17 advanced countries (Table 4). Again, the only country displaying a negative association between growth and lagged stock returns is India. Using this specification, the coefficient on lagged stock returns is, on average, slightly higher in advanced countries than in emerging market countries.

Table 4.

Output Growth on its Lag and Lagged Stock Returns, Individual Country Regressions, 1971-98

article image
Sources: MSCI web site. Emerging Markets Factbook, IFC and. International Financial Statistics, IMF.Note: Newey-West corrected t-statistics in parentheses.* indicates that the coefficient is significant at the 5 percent level.** indicates that the coefficient is significant at the 1 percent level.

The results obtained with individual-country regressions are confirmed by estimating panel regressions with individual-country intercepts but constraining the slope coefficients to be the same for all countries. In a regression of real economic growth on lagged real stock returns, the estimated slope coefficient amounts to 0.034 and is highly significant (Table 5). When the slope coefficient is allowed to differ between emerging country and advanced country groups (though constrained to be the same for all countries within a group), the coefficient turns out to be higher in emerging countries (0.035) than in advanced countries (0.032), though not significantly so. The results are robust to changes in sample period, e.g., excluding 1997-98 (the years the Asian crisis) or the years up to 1977 for the advanced markets. The slope coefficient is somewhat higher in the first half of the sample period than in the second half, but not significantly so. Controlling for one lag of the dependent variable, the results remain similar, although a well-known disadvantage of panel estimation is that when the regression includes lags of the dependent variable the common slope coefficients are not estimated consistently (Pesaran and Smith, 1995).11 However, an advantage of panel estimation is that it makes it possible to use also the information available from those countries that have a smaller number of observations. When the estimation is conducted using all 41 countries for which at least ten observations are available, the coefficients are once again similar to those reported above. (The estimates are similar when an alternative minimum number of observations is used.)12 Estimation as a system of seemingly-unrelated regressions (SURE) reduces the standard errors of the estimated coefficients considerably (especially for advanced countries).

Table 5.

Panel Estimation of Relationship between Output Growth and Lagged Stock Returns 1/

article image
Sources: MSCI web site, Emerging Markets Factbook, IFC and International Financial Statistics, IMF.

All 25 (20+yr) are the 25 countries (eight emerging and 17 advanced) with more than 20 years of data.

All 41 (10+yr) are the 41 countries with more than ten years of data, t-statistics are reported in parentheses.

Panel estimation shows that lagged stock returns remain significantly associated with output growth in both advanced and emerging countries when controlling for lagged values of other leading indicators, including real short-term13 interest rates, and the real growth rate of both narrow and broad money (Table 6). Narrow and broad money growth are used as controls because they have been found to be useful leading indicators for developing countries (Agenor, McDermott and Prasad, 1999) and there is a large literature on interest rates and the term structure as predictors of output (Estrella and Hardouvelis, 1991). The present paper confirms that low real interest rates and high real money growth tend to be followed by rapid output growth. These associations are significant, although narrow money has more predictive power in emerging countries, whereas broad money has more predictive power in advanced countries.14 The slope coefficient on lagged stock returns is, once again,

Table 6.

Panel Regressions of Output Growth on Lagged Stock Returns and other Leading Indicators 1/

article image
Sources: MSCI web site, Emerging Markets Factbook, IFC and International Financial Statistics, IMF.

All right-hand side variables are in real terms and lagged by one year.

similar in separate panels of advanced and emerging countries. In panels of all 24 countries15 for which data are available for at least twenty years, an interaction term of lagged stock returns times an emerging market dummy is not significant. The results are robust to changes in estimation technique (using individual-year dummies, and SURE) and sample period. On the whole, therefore, stock returns provide useful information to predict output growth, over and above that contained in other variables, not only in advanced countries but also in emerging countries, and the value of this information does not seem to be different in the two types of countries.

The evidence on the relationship between lagged stock returns and output growth at a quarterly frequency confirms the key findings obtained with annual data, though the proportion of countries in which the relationship is significant is lower among emerging market countries than among advanced countries. The correlation is analyzed between real stock returns over the twelve months prior to time t and the growth rate of real GDP or industrial production (depending on data availability) between, in turn, the quarter prior to t and the quarter following t; the two quarters prior to t and the two quarters following t; the three quarters prior to t and the three quarters following t; and the four quarters prior to t and the four quarters following t. An appropriately lagged dependent variable and seasonal dummies are included in the estimates. In other words, the estimated equations are the following:

(400/j)log[(i=1jYi1)/i=1j(Yi)]=α0+α1log[(S/P)j/(S/P)j4)]+β(400/j)log[(i=1jYj+i1)/i=1j(Yji)]+ε

where Y is real GDP (or industrial production), S is the nominal stock price index, P is the consumer price index, and j =1,..4. The correlation tends to be stronger, the longer the forecast horizon. Considering the case of j=1 (quarter-on-quarter growth), the correlation is positive and significant in nine out of 18 advanced countries, and two out of six emerging markets, using real GDP; and 13 out of 18 advanced countries, and four out of 13 emerging markets, using industrial production (Table 7).16

Table 7.

Output Growth at Different Forecasting Horizons, and Lagged Stock Returns

article image
Sources; MSCI web site, Emerging Markets Factbook, IFC and International Financial Statistics, IMF.Notes: The t-statistics are Newey-West corrected. Seasonal dummies and one lag of the dependent variable are included in the set of independent variables.*indicates that the coefficient is significant at the 5 percent level.**indicates that the coefficient is significant at the 1 percent level.

The association between (quarter-on-quarter) output growth and lagged stock returns is robust to controlling for other variables. (All regressions include seasonal dummies and one lag of the dependent variable.) The slope coefficient on lagged stock returns (over the previous year) is significantly different from zero in eight out of 18 advanced countries and three out of six emerging markets, using GDP, and in 13 out of 18 advanced countries and three out of 13 emerging markets using industrial production, controlling for the real growth rate of credit to the private sector (Table 8); and in nine out of 18 advanced countries and three out of six emerging markets, using GDP, and in 15 out of 18 advanced countries and three out of 13 emerging markets using industrial production, controlling for the real growth rate of broad money (Table 9).

Table 8.

Output Growth, Lagged Stock Returns, and Private Credit Growth, Constrained

article image
Sources: MSCI web site, Emerging Markets Factbook. IFC and International Financial Statistics, IMF.Notes: Seasonal dummies and one lag of the dependent variable are included as independent variables. Output growth is quarter-on-quarter; stock returns and private credit growth refer to the full preceding year.*indicates that the coefficient is significant at the 5 percent level.**indicates that the coefficient is significant at the 1 percent level.
Table 9.

Output Growth, Lagged Stock Returns, and Broad Money Growth, Constrained

article image
Sources: MSCI web site, Emerging Markets Factbook, IFC and International Financial Statistics, IMF.Notes: Seasonal dummies and one lag of the dependent variable are included as independent variables. Output growth is quarter-on-quarter; stock returns and broad money growth are in real terms and refer to the full preceding year.*indicates that the coefficient is significant at the 5 percent level.**indicates that the coefficient is significant at the 1 percent level.