What are the consequences of herdlike behavior?

What are the consequences of herdlike behavior?

Episodes of high volatility in international capital flows to emerging market economies in the 1990s put foreign investors in the limelight—these investors are frequently seen as the culprits behind bouts of instability and the ensuing currency crises. Some have argued that market participants disregarded fundamental economic conditions in emerging economies and instead acted as a “herd”—that is, they reflexively did what other investors were already doing. Widespread herding behavior by a group of investors could worsen volatility and touch off panics that do not reflect the true economic conditions in emerging market countries. In our paper published in 2000, we examined the herding proposition empirically for international funds that invest in emerging economies’ equity markets and found that although herding is more prevalent there than in other markets, it is probably not to blame for the observed volatility of international capital flows.

Herding can be fully rational from the point of view of an individual investor. For example, an investor may have only limited access to direct information about asset returns and therefore needs to learn from the actions of others. Sometimes, the best thing an uninformed investor can do is follow the actions of more informed investors. Alternatively, fund managers’ compensation may be linked to the performance of the fund they manage relative to those of other funds of the same type. This may prompt managers to try to avoid falling too much behind their comparator funds, which would lead them to invest in a portfolio of assets that never diverges very much from their comparator funds’ portfolios.

Herding may sometimes be more apparent than real. For example, foreign investors may appear to act as a herd if they react simultaneously to the same news about fundamentals. In this case, their behavior speeds up the adjustment of prices and is not destabilizing. However, in an efficient market, speedy price adjustment should occur without many actual trades having to take place. Moreover, the question remains as to why international investors would react differently to certain news than domestic investors.

“Contrary to our presumption, there is more herding in the largest stock markets than in the smallest ones.”

What does the evidence show?

Assessing the behavior of international investors in a systematic way is difficult. Most of the available financial information consists of data on prices, and it is difficult to convincingly trace movements in prices to the behavior of particular groups of investors. Moreover, herding by international investors may not have an immediate impact on asset prices but may nevertheless worsen a country’s balance of payments.

We studied the behavior of international investors by exploring a novel data set, compiled by Emerging Market Funds Research, Inc., comprising monthly data on the individual portfolios of about 400 dedicated emerging market equity funds for January 1996–March 1999. The database covers 80 percent of dedicated emerging market funds, and—with an aggregate net asset value of over $100 billion—about 90 percent of the market value of all such funds’ assets. While the period covered is relatively short, it encompasses significant financial crises, enabling us to examine each of these in detail. This database also has the advantage of covering a well-specified class of investors—it makes sense to search for evidence of herding within a subset of investors, because the whole market cannot move in the same direction: for every seller, there must be a buyer.

First, we examined gross and net flows into and out of emerging equity markets in different regions. Chart 1 displays gross and net flows into emerging markets in four major regions for the whole period we examined. Our findings show that funds do not always move in the same direction: large outflows may occur at the same time as large inflows. The chart also suggests, however, that funds have pulled out, on a net basis, just before major crises or periods of turbulence. For Asia, there were sizable net outflows starting in June 1997. For Europe, there was a substantial drop in net inflows in July 1998, just before the Russian crisis occurred. For Latin America, there was a sharp outflow one month before the Brazilian devaluation of January 1999.

Chart 1
Chart 1

Gross and net flows, by region

(million dollars)

Source: Authors’ calculations based on data from Emerging Market Funds Research, Inc.

Chart 2 takes a more detailed look at some important episodes of turbulence that are encompassed by this data set: the Czech Republic’s devaluation of May 1997, Thailand’s crisis of June 1997, Russia’s ruble collapse of August 1998, and Brazil’s floating of the real in January 1999. In general, the emerging market mutual funds in our sample tended to withdraw funds from the affected countries in the month before the crisis. (The exception, interestingly, is Thailand, whose crisis has been blamed by many observers on the behavior of international investors.) The amount of net outflows during crises, however, is relatively small, given the volatility of these flows, and is even smaller if a longer period before the crises—for example, the preceding six months—is considered.

Chart 2
Chart 2

Inflows and outflows around crisis periods

(million dollars)

Source: Authors’ calculations based on data from Emerging Market Funds Research, Inc.Note: Numbers are based on a sample of funds that were in the database throughout the period. The vertical line in each panel marks the month in which the peg of the domestic currency was abandoned.

Perhaps more important, relatively large inflows have persisted even during periods just before the crises, which suggests that not all foreign investors anticipated the approaching events. Furthermore, funds did not withdraw indiscriminately from emerging markets in turbulent periods. For example, while it is true that, viewed in the aggregate, funds withdrew large amounts of capital from Latin America and Asia during the Russian crisis, many funds in our sample that reduced their exposures in Russia increased their investments in Latin America. This contradicts a simplistic view of mutual funds’ behavior and suggests that their portfolio choices need to be examined more closely.

A more systematic, quantitative measure of the degree of herding among funds confirms the casual impression given by Charts 1 and 2. The index of herding originally introduced by Lakonishok, Shleifer, and Vishny (1992) essentially counts how often funds end up on the same side of the market relative to what one would expect if they traded independently and randomly. Except for a trend correction, the measure assumes that in the absence of herding, there would be an equal number of buyers and sellers among funds. Naturally, computation of this index makes sense for only a subset of the market because, in the aggregate, it is impossible for all investors to join a herd and be on the same side of the market.

The quantitative results indicate that herding behavior is statistically significant but probably does not have a large economic impact. The overall mean of the herding index is 7.2 percent, implying that for a given country, the number of funds buying or selling was slightly more than 7 percent larger than one would have expected if they had acted independently and randomly. This number is approximately twice as large as the values found in other studies for U.S. institutional investors in the U.S. stock market. It is clearly not as large a figure as would have been expected, however, if one believed that herding by international investors was the main cause of international financial volatility. The measure of herding is fairly stable across regions and over time; moreover, considering all countries, herding does not seem to increase significantly during crises. We did find, however, that there is a positive correlation between herding and stock market volatility across countries. It is also important to highlight that, since we are investigating the behavior of dedicated emerging market funds, we can analyze herding in and out of the emerging market asset class as a whole in only a limited sense.

By looking at differences among funds and countries, one can obtain a better understanding of the herding phenomenon. For example, if it were more costly for fund managers to acquire information about small markets than large ones, herding would probably be more common in the former. Similarly, one might expect offshore investment funds to display different investment patterns than domestic funds, given the lesser regulatory constraints that the former face. Because closed-end funds (which have a fixed number of shares issued at their initial public offering) are not subject to redemptions by their investors, one should—if herding involves mainly individual investors—observe lower herding index numbers among closed-end funds.

The empirical results we obtained confirm some, but not all, of these hypotheses. Contrary to our presumption, there is more herding in the largest stock markets than in the smallest ones. This suggests that herding is not a result of fund managers trying to avoid incurring fixed information costs, which, relative to market size, are lower in the largest markets. Instead, illiquid markets may prevent fund managers from imitating the behavior of others in the smallest stock markets. Offshore funds tend to herd less than other funds, while herding by large, global, and international funds is not very different from the average for all funds. In line with our a priori reasoning, herding is less pronounced among closed-end funds, suggesting that the observed tendency to engage in herding might, to some extent, be traceable to the behavior of individual investors.

Leaders and followers

Who are the leaders and who are the followers within the mutual fund industry? The funds in our sample have different expertise and constraints, and operate under different rules and objectives. These differences may affect how they react to each other’s transactions. One might imagine, for example, that regional or single-country funds are more familiar with the specific economic situations in the countries where they invest and thus play a leading role in triggering inflows or outflows. Similarly, if the acquisition of country-specific information involves fixed costs, smaller funds may be at a disadvantage compared with larger funds and may be induced to follow the latter’s strategies. In addition, open-ended funds may be subject to redemptions by nervous individual investors and be forced to reduce their exposures to certain countries before closed-end funds are. In instances where open-ended funds are important enough to affect market trends, however, closed-end funds would be induced to follow them. It is therefore conceivable that a small fraction of funds may regularly be responsible for starting large stampedes into or out of a country.

We divided funds into four interesting pairs: single-country funds and multiple-country funds, global and international funds and regional and single-country funds, large funds and small funds, and closed-end funds and open-ended funds. Then we used an econometric procedure (vector autoregression) to detect systematic relationships in the timing of transactions by these different groups of investors. The results for regional and single-country funds versus global and international funds yield a result consistent with information transmission in the market: inflows or outflows by regional or single-country funds precede flows of global and international funds into the same country. The results also show that open-ended funds’ flows precede closed-end funds’ investments, which, again, is consistent with the view that herding begins with individual investors.

Trading strategies

Another type of behavior that is not consistent with the underlying economic strength of companies in emerging market economies occurs when investors follow “positive feedback” or “momentum” trading strategies. These basically imply a tendency to buy past strong performers and to sell recent weak performers. To test for the existence of such behavior, we examined whether the degree of herding is related to past returns. If funds follow momentum strategies, we should observe herding to be more pronounced if there were extremely low or high returns for the preceding month. For example, a “sell herd” would form after particularly low returns were experienced for a country during the preceding month. We computed separate herding measures for individual countries for months when the proportion of buyers was distinctly below or above average and investigated the relationship between these developments and the prior performance of individual stock markets. We also computed two measures of excess demand and examined their correlations with prior returns.

The results are, again, mixed. Although there is no clear relation between the herding measures and returns for the preceding month, the results indicate that, on average, funds did tend to buy past winners. There is, however, no evidence that this behavior is accentuated during crises. Such trading based on past returns is at least as pronounced for single-country funds as for funds in the aggregate.

Overall, our results suggest that the behavior of emerging market equity funds is more complex than is often suggested. It is true that this class of investors tends to pull out of emerging market economies just before the outbreak of a crisis, with regional and single-country funds appearing to move first. Although funds show some degree of herding behavior, this reaches only modest proportions and is probably too limited to justify considering funds responsible for starting “stampedes.” Similarly, there is some evidence that funds follow destabilizing feedback strategies—buying past winners and selling past losers—but only to a limited extent. Interestingly, neither herding behavior nor feedback trading is more pronounced during crises. In light of this evidence, at least, the case against dedicated emerging market funds remains to be proven: simple characterizations of this class of investors as “panic prone” do not seem to be appropriate.

At the time of writing, Eduardo Borensztein was Chief of the Strategic Issues Division in the IMF’s Research Department. R. Gaston Gelos was an Economist in the Macroeconomic and Structural Adjustment Division of the IMF’s Research Department.


  • Borensztein, Eduardo, and R. Gaston Gelos, 2000, “A Panic-Prone Pack?: The Behavior of Emerging Market Mutual Funds,IMF Working Paper No. 00/198 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Lakonishok, Josef, Andrei Shleifer, and Robert W. Vishny, 1992, “The Impact of Institutional Trading on Stock Prices,Journal of Financial Economics, Vol. 32 (August), pp. 2343.

    • Search Google Scholar
    • Export Citation
The Impact on Trade, Policy, Labor, and Capital Flows