top of page

No Place Like Home

Home Bias Theory to Foreign Portfolio Investment in Emerging Markets
Qiyuan Zheng
Wesleyan University

April 2021

Modern technology has allowed investors, especially in developed markets, to gain access to a wealth of information about events that affect equity prices almost instantaneously, ultimately making it more difficult for investors in developed economies to ‘beat the market’. Such markets, where prices fully reflect all available information, are considered to be efficient; according to the efficient market hypothesis, opportunities for arbitrage in efficient markets are scarce, if not impossible. In this context, the typical investor could only generate higher returns by taking on greater risks. If this was the case, inefficient markets would be fundamentally more profitable for the informed investor as arbitrage opportunities are abundant and riskless profit can be made once the investor correctly identifies mispriced assets. This line of reasoning suggests that inefficiency in emerging markets might attract foreign portfolio investment (FPI) inflows, since investors in the developed world would seek to exploit the arbitrage opportunities in those inefficient markets.

Market inefficiency in emerging economies is often at least partially due to poor property rights and weak institutional arrangements, such as unstable and corrupt political systems, not fully as a result of economic fundamentals, such as lack of financial development and domestic investor behavior. If inefficiency in an emerging market were to be largely a result of poor property rights and weak institutions, the ability of foreign investors to properly exploit arbitrage opportunities would be low and the institutional risk borne by investing in the market would be high, as unstable political environments foster volatile asset prices. Under such conditions, one might very well expect inefficient markets to drive away FPI. However, if the institutional quality of an investing environment is held constant and market inefficiencies are a result of economic fundamentals, then one should expect such a market to attract FPI.

This paper finds evidence that, after accounting for a given level of institutional risk, potential simultaneity, and time of information absorption, there is no significant relationship between market inefficiency and FPI. To explain the inconsistency between theory and empirical evidence, I suggest an extension to the equity home bias theory. Since capital is abundant in wealthy nations where markets are efficient, investors that account for a majority of FPI inflows into emerging economies would be more familiar with and thus more optimistic about efficient markets since they more closely resemble their domestic investing environment. If a large enough number of foreign investors show a clear preference for efficient markets, the magnitude of their actions may very well offset that of unbiased investors looking to exploit arbitrage opportunities in inefficient markets. The paper proposes that, while market inefficiency should theoretically attract FPI, holding institutional risk constant, empirical evidence fails to show this relationship because foreign investors from the developed world exhibit a preference for more efficient markets that they are familiar with.

A Survey of Theory and Existing Literature
In documenting market efficiency among developing countries, Morck et al. found that the relationship between GDP per capita and price synchronicity can largely be explained by weak institutional arrangements and poor property rights. Their research not only provides a theoretical framework for this paper but also suggests an important measure of market inefficiency. Institutional shortcomings, especially poor property rights, discourages informed trading in the market as volatile political environments make it difficult for investors to price assets and retain their earnings (Morck 15, 16). The lack of informed investors would increase the magnitude of noise trading’s effect on the market. Since noise traders are uninformed and exhibit poor market timing (the buy high-sell low effect), their actions would not only lead to excess volatility in the market, but also push prices of different stocks to move synchronously away from their fundamental values (De Long 705, 715). Morck et al.’s paper presents empirical evidence supporting the above theory, as the observed relationship between GDP per capita and price synchronicity is rendered insignificant once property rights have been accounted for (Morck 22). This conclusion directly implies that, if protection of property rights is the only factor affecting the level of information reflected in prices, one can use price synchronicity as an appropriate proxy for market efficiency (i.e. better property rights lead to more informed traders, less price synchronicity, and a more efficient market). However, there are certainly other factors affecting the level of information reflected in stock prices. I argue that, in the absence of property rights violations, price synchronicity would remain an appropriate proxy as less synchronous prices would indicate that the market captures higher levels of firm-specific information and has a higher concentration of informed traders. Additionally, the lower prevalence and less developed nature of financial institutions in emerging markets would decrease the number of informed traders and thus increase price synchronicity while lowering market efficiency. In the subsequent analysis, I will refer to the following factors as economic fundamentals affecting market efficiency: domestic investor behavior, prevalence of financial institutions, quality of financial institutions, and technological development. This claim is particularly important, as the paper seeks to differentiate between market inefficiency caused by economic fundamentals and that caused by poor political institutions.

Markets are efficient if “security prices at any time ‘fully reflect’ all available information” (Fama 383). Since efficient markets already reflect “all obviously publicly available information”, it would be very difficult for investors to obtain higher returns without undertaking greater risk (Fama 414). Rational investors who seek to increase returns while lowering risk would then be drawn to less efficient markets where arbitrage opportunities are more easily available. Thus, upon first glance, it seems that inefficient emerging markets would be more attractive to foreign investors, especially those from developed countries with efficient markets. Since most foreign portfolio investment comes from wealthier nations where capital is abundant, one could expect that less efficient markets, those with higher price synchronicity, would generate higher levels of FPI. However, this conjecture fails to consider the factors that lead to market inefficiency in developing countries and the preferences of foreign investors. The following paragraphs present two potential explanations for why emerging markets that are less efficient might fail to attract FPI.

The first explanation comes directly from the work of Morck et al. on price synchronicity in emerging markets. While poor property rights decrease market efficiency, they also increase both the opportunity cost—the time spent gathering information to identify asset mispricing—and risk of arbitrage trading. Political events are also typically much harder to predict in these countries and, given the poor property rights, “risk arbitrageurs who do make correct predictions may not be allowed to keep their earnings, […] especially if the risk arbitrageurs are political outsiders” (Morck 15). Thus, if the observed market inefficiency is largely the result of poor property rights and weak institutional arrangements, the expected relationship between price synchronicity and FPI becomes more complicated. While inefficient markets still present certain arbitrage opportunities for investors, the risk of investing in an environment with poor property rights may very well drive foreign investors away.

Another potential explanation comes from an extension of the equity home bias theory. The traditional equity home bias theory states that investors are more inclined to hold domestic stocks despite the potential benefits of international diversification. This phenomenon was first analyzed in 1991, when French and Poterba found that domestic investors expect returns around 300 basis points higher than foreign investors when looking at the identical market (French 4). The optimism in domestic markets would then lead investors to prefer a domestic stock over an international one, even if the economic values of the two stocks do not differ from each other. The equity home bias theory implies that investors prefer assets they feel more familiar with and such preferences can often offset the actual economic differences between any two assets. This paper argues that investors from developed countries with efficient markets would naturally prefer stocks in more efficient markets of the developing world as they more closely resemble their domestic investing environment. Since most FPI comes from wealthier nations in the developed world, even if one observes an inefficient market in a country with strong property rights and therefore higher chances of arbitrage without bearing institutional risk, such a market might not attract FPI as most foreign investors would prefer to hold assets in efficient markets with which they are more familiar.

Existing literature shows that a simple analysis of FPI and price synchronicity is not enough to uncover the fundamental relationship between market efficiency and FPI inflows. To properly understand whether investors are truly drawn to inefficient markets due to opportunities of arbitrage, one must first take into account the institutional risk inherent in emerging markets. Only after accounting for the protection of property rights can one expect there to be a positive correlation between price synchronicity, essentially a measure of market efficiency, and FPI inflows. Empirical results that do not align with such expectations would be consistent with the story of equity home bias theory, where foreign investors prefer efficient markets as a result of familiarity and resemblance to their domestic markets.
Constructing the Data Set

The paper analyzes nine emerging markets: Brazil, Chile, China, Greece, India, Malaysia, Mexico, Thailand, and Turkey. The choice of these countries is based on their per capita GDP, the size of their domestic equity market, and data availability. The time period observed ranges roughly between 2000 and 2016. I shall note here that the somewhat arbitrary decision to characterize these countries as emerging markets through per capita GDP and choosing countries with sufficient stock listings may introduce sampling bias. A future extension of this paper may be to include a larger number of developing countries and test the robustness of this study by shifting the per capita GDP cutoff used to define emerging markets.
Due to limited resources, this paper uses an approximation for its main variable of interest, price synchronicity. To obtain a price synchronicity index for each country and year, I collected weekly stock returns (between 2000 and 2016) for the companies listed on a popular index of the given country. Table 1 details the exact indexes used to construct the price synchronicity data. Given a country, the price synchronicity of year T is then constructed as follows:
SyncT=w∈Tmax⁡(Upw, Downw)Upw+Downw1NT
The above equation states that for every week w in year T, I calculated the number of stocks that rose in share price (if closing price was higher than opening price), the number of stocks that dropped in share price (if closing price was lower than opening price), and divided the maximum of the two numbers by the total number of stocks that experienced a change in share price. An arithmetic mean is then computed for the given NT weeks in year T. Note that this calculation is based on Morck et al.’s methods of finding price synchronicity, with the denominator constructed to include only stocks with changed share prices to avoid non-trading bias (Morck 5). Given the method of calculation, a price synchronicity of 0.5 would indicate that prices do not move together at all while high price synchronicity (such as 0.9) would indicate an inefficient market. Data for other variables were obtained through the World Bank, Transparency International, and World Integrated Trade Solutions (WITS). The following section will discuss the methodologies and rationale for including each regressor.

Analytical Methodology
Given the panel structure of the data, the paper will use a fixed effects model on the country level with robust standard errors to analyze the relationship between market efficiency and foreign portfolio investment. The fixed effects model would allow for the paper to account for unobserved but time-constant differences between countries, therefore yielding a less biased estimate. The fixed effects model was chosen over a random effects model on empirical grounds. The Sargan-difference test of overidentifying restrictions yielded a Sargan-Hansen statistic of 566.9 when applied to a random effects regression with robust standard errors, which indicates a significant level of overidentification in the model.

Initially, I estimated a simple fixed effects model with price synchronicity as the only explanatory variable:
FPIit=α+βSyncit+i+eit (1)

However, the coefficient for the model is difficult to interpret and meaningless to this paper’s purpose. While the paper is primarily interested in exploring the effect of market inefficiency (caused by non-institutional factors) on FPI inflows to emerging markets, the coefficient presented in Equation 1 is theoretically biased downwards as the result of institutional risks present in emerging markets with high price synchronicity. More specifically, one can see that is subject to omitted variable bias because the level of corruption drives up price synchronicity (positive correlation) and discourages foreign investors (negative correlation with FPI). Additionally, could be affected by other confounding variables as a result of selection bias.

To properly identify how FPI is affected by market inefficiency caused by economic fundamentals, one must account for the level of political risk the investor must bear to participate in the market and other confounding variables with the following fixed effects model:
FPIit=α+1Syncit+2Corruptionit+3Xit+i+eit (2)

Note that Corruptionit reflects the Corruption Perception Index of country i in year t, obtained from Transparency International. The author calculated Corruptionit=(100- Corruption Perception Index) so that 0 represents no corruption and 100 is the value for the most corrupt extreme. Ideally, the paper would’ve liked to use the “good government” index from Morck et al.’s work that included factors specific to property rights protection but financial constraints limited the data collection process (Morck 15). Xit is a vector of time-varying country-level characteristics that consists of the following variables: GDP per capita, inflation volatility, market capitalization of domestic companies (in current US dollars), and capital openness. Inflation volatility is calculated by taking the 5-year moving coefficient of variance of each country’s consumer price index (obtained from the World Bank). Capital openness is measured by the standardized version of the Chinn-Ito Index (Chinn). GDP per capita and inflation volatility could both be omitted variables as they both are significant indicators of an economy’s stability and development, in turn affecting confidence levels in foreign investors. Market capitalization indicates both the breadth and depth of the domestic financial markets, as a higher levels of market capitalization would provide more opportunities for foreign investors and increase FPI inflows. Although intuition suggests that capital openness would be a significant factor in affecting foreign investment, empirical evidence from existing literature suggests that capital controls on FDI and FPI have no significant impact on FPI inflows (Li 228, 230). The variable was still included in Equation 2 largely because theory implies that capital controls would increase the opportunity cost for foreign investors to invest in the domestic market and therefore decrease FPI. To test for robustness of the results, I removed the variable from the model and found no significant changes to the parameters of interest. Pre-existing theory and literature suggested that each of the variables included in the vector Xit would be correlated with FPI inflows. Thus, the model should include these variables as controls in the regression to minimize standard errors and account for any sampling bias.

To account for potential information absorption time, I re-estimated Equation 1 and 2 with lagged price synchronicity, inflation volatility, and GDP per capita. If these factors were to exhibit greater cross-year variations than within-year shifts, then the incorporation of the lagged independent variables would allow for the possibility of foreign investors “reacting” to changes in their values in the next time period. I chose to not pursue a fully lagged model because market capitalization and capital controls are present constraints on the foreign investor’s choice set. Further, the corruption index remained period t as well since it measures the level of corruption perceived by the public at time t, which is a direct factor in determining the level of FPI in the same time period. Additionally, a fixed effects lagged-distributed model was also estimated as follows:
FPIit=α+1Syncit+2Syncit-1+3Corruptit+5Xit+6Yit-1+νi+eit (3)
Note that Syncit-1 and Yit-1 are lagged price synchronicity and lagged vector of control variables, hence they are values of country i in year t-1.
The paper conducted further robustness tests by removing 2008 from the estimated models. Figure 1 shows that FPI inflows in observed countries dropped dramatically as a result of the Global Financial Crisis (GFC). Figure 2 shows the average price synchronicity among the observed countries across time and indicates that average price synchronicity varies between 75% and 70% for most of the years with no significant change during the GFC. One can see that the patterns exhibited by the data during the GFC is an aberration caused by an external shock, which could affect both the precision and accuracy of previous estimations. The paper accounts for this by re-estimating all the previous models with a smaller sample size that does not include 2008. Although doing so limits the power of the test, the removal of the outlier (2008) should offer a more accurate estimate of the effect market efficiency has on FPI.
Empirical Results

Before estimating the models specified in the Methodology section, one must return to examine an earlier claim: “price synchronicity would remain an appropriate proxy under [scenarios in which other factors (besides poor property rights) affect the level of information reflected in stock prices] as well, since less synchronous prices would indicate that the market captures higher levels of firm-specific information”. Table 2 Column 1 shows an estimated fixed effects model that captures the relationship between price synchronicity and institutional risk (represented by the Corruption Perception Index). As expected, the coefficient for the Corruption Index is positive, since higher levels of corruption means more institutional risk and thus higher price synchronicity. The estimated coefficient is statistically significant. Most notably, the R-squared for the estimation is only around 8%, indicating that there are certainly other factors, such as market inefficiency due to economic fundamentals and the quality of financial institutions, that affect price synchronicity. Additionally, Table 2 Column 2 shows a negative correlation between corruption and FPI inflows, as expected. Although the relationship is not statistically significant, theory suggests that it would bias the estimates of Equation 2 downwards.

Tables 3, 4, and 5 show the estimated regressions specified in the Methodology section. Table 3 displays the estimates obtained by using the “present” model without any lagged variables while Table 4 shows the results after lagging the appropriate independent variables. Table 5 shows the estimates obtained from the lag distributed model (Equation 3). In Table 3, the first two columns represent Equation 1 and 2 without the lagged independent variables. The first column shows a negative and marginally significant (at the 10% threshold, p=0.054) coefficient for price synchronicity, with the point estimate show approximately $0.605 billion decrease with every 1 percentage point increase of price synchronicity. This is not surprising, as price synchronicity caused by institutional risk would most likely drive foreign investors away. The second column on Table 3 also show a negative and marginally significant coefficient for price synchronicity. Although this coefficient is greater in magnitude ($0.967 billion decrease for every 1 percentage point increase in price synchronicity), it has a larger confidence interval than that of the first model and is less statistically significant (p=0.066). Columns 3 and 4, estimates after removing 2008, exhibits a similar pattern as the coefficient for price synchronicity is negative and statistically significant (at the 5% threshold) when it’s the only regressor in the model but no longer significant once the model accounts for institutional risks and other sources of selection bias (Column 4, Table 3). Table 4 incorporates lagged price synchronicity, inflation volatility, and GDP per capita. Columns 1 and 2 show the results of Equation 1 and 2 when with the lagged independent variables replacing their non-lagged counterparts. Columns 3 and 4 show those same models estimated after removing 2008 from the sample. The results across all columns are consistent in that the coefficient for lagged price synchronicity are all positive but statistically insignificant even after accounting for the downward bias caused by institutional risks.

Interpreting the results of Table 3, the paper finds that market inefficiency (proxied by price synchronicity) drives foreign investors away mostly as a result of the poor property rights that created the inefficient market in the first place. This effect is exhibited by the negative and (marginally) significant coefficient for price synchronicity in Column 1 and 3. Once the model accounts for institutional risk and potential selection bias, market inefficiency remains negatively correlated with FPI inflows but at a less significant level in Column 2 and completely insignificant when 2008 is removed from the sample, as shown in Column 4. If the level of institutional risk does not change and the market becomes more inefficient (price synchronicity rises), theory suggests that more investors would be drawn to the market as they seek to exploit arbitrage opportunities. This theory implies that a model which accounts for institutional risk should generate a positive and significant coefficient for price synchronicity. However, empirical evidence does not support this conjecture and instead illustrates that market inefficiency stemming from causes unrelated to institutional risks either does not significantly affect or decreases the level of FPI inflows, depending on whether year 2008 was included in the sample. Results obtained by the lagged models fit the theory slightly better, as the coefficient for lagged price synchronicity is positive, as shown in Table 4. Both Column 2 and 4 of Table 4 exhibit point estimates that indicate a $0.42 billion rise in FPI inflows per percentage point increase in lagged price synchronicity (decrease in market efficiency). However, this estimate is not statistically significant, which could be a result of the small sample size and thus less power / minimal detectable effect. Further, Column 2 and 4 of Table 4 showed a higher point estimate than Column 1 and 3 of the same table, respectively, fitting the theory that not including corruption as a covariate would downwardly bias our estimate.
Considering the empirical results on both tables (with and without the lagged component), one can see that, once institutional risks are accounted for, changes in market efficiency does not significantly affect FPI inflows. A potential explanation for this phenomenon is an extension of the equity home bias theory, in which foreign investors from the developed world feel more familiar and are thus more optimistic about efficient markets. Hence, market inefficiency (under the same level of institutional risk) can both attract investors through opportunities of arbitrage and drive away investors through its unfamiliar nature. If those effects offset each other, one would observe no significant relationship between market inefficiency and FPI inflows after accounting for institutional risk. That said, it is more probable that Table 4 is the better model, as it lagged certain independent variables that investors would be “reacting” to in period t based on their information in period t-1. To further test this, a lag distributed model was estimated in Table 5 to show that price synchronicity in period t-1 is indeed positively correlated with FPI inflows in period t, even after account for price synchronicity in period t. However, as before, the estimate is not statistically significant, most likely due to a combination of small sample size and the potentially offsetting effect from the equity home bias theory.

Robustness Tests
Another potential explanation for the insignificant coefficient for price synchronicity is a reverse causality chain between FPI inflows and market efficiency. The paper argues that increased FPI inflows can lead to higher market efficiency in emerging economies. As established, a majority of FPI in emerging markets come from developed countries where capital is abundant. These developed countries also have better financial institutions, which help “foreign” investors make more informed decisions than their domestic counterparts in the developing country. If one assumes that most foreign investors in emerging markets are more informed than their domestic counterparts, then an increase in FPI inflow would mean more informed investors in the market and therefore an increase in market efficiency. Incorporating this conjecture, one can obtain three distinct factors that affect market efficiency in emerging economies: level of institutional risk, amount of foreign investment (FPI inflows), and other economic fundamentals. Since the paper is only interested in the relationship between market inefficiency caused by economic fundamentals and FPI inflows, it must account for the first two factors. Equation 2 properly accounts for the level of institutional risk but fails to address the joint relationship between market efficiency and FPI inflows.
I used a three-stage least squares method to estimate the following simultaneous equations system:
FPI=γ+1Sync+2X+3GDP+e2 (4)
Sync=α+1FPI+2Corruption+3(HH Index)+4GDP+e1 (5)
The variable HH Index, the Hirschman Herfindahl Index for exported products (obtained from WITS), was added to account for the level of economic specialization, since more specialized economies tend to experience greater price synchronicity (Morck 9). Additionally, the fixed-effects approach was replaced with a least-squares dummy variable approach by adding a dummy variable for every panel value except one. The results of this model are shown in Table 6, with the coefficients for the panel dummy variables suppressed from the output. The first two columns show the estimates for both endogenous variables (Equation 4 and 5) when using the entire sample size and the second set of columns shows the results after removing 2008 from the sample.

Even after addressing the simultaneity problem in the non-lagged model, the paper fails to find a significant relationship between market efficiency and FPI. Although the coefficients for price synchronicity are positive, as shown by Column 1 and 3 in Table 6, they are statistically insignificant. Additionally, these estimates fail to provide empirical evidence in favor of the claim that FPI leads to more efficient markets in developing countries; both Column 2 and 4 show insignificant positive coefficients for FPI when it’s used as an independent variable in estimating price synchronicity.

As stated previously, theory and empirical evidence (from Table 4) suggest that a model with lagged price synchronicity and corruption index would better capture the causal relationship between market efficiency and FPI. If the lagged model is a better fit and the conjecture of reverse causality remains valid, then changes in FPI inflows in time period t-1 would affect market efficiency of time period t-1, which in turn would influence the FPI inflows of time period t. To put it simply, FPI and market efficiency are two endogenous variables that are both sequentially and jointly determined. A precise and accurate estimate of the effect market inefficiency (when caused by economic fundamentals) has on FPI inflows would then require a model that allows for both sequential and simultaneous relationship between market efficiency and FPI. I estimated such relationship with the following equation:
FPIit=α+1Syncit-1+2Corruptit+3Xit+4Yit-1+5FPIit-1+νi+eit (6)
where FPIit-1 is the lagged FPI variable. This equation would allow us to “parse out” the reverse causality effects on market efficiency caused by changes in FPI inflows. Thus, 1 would be the unbiased estimate if such reverse causality indeed exists. Results in Table 7 show a point estimate of between $0.14 to $0.28 billions of FPI increase per percentage point increase in the lagged price synchronicity (depending on whether or not 2008 is included in the sample). However, this estimate is also not statistically significant, most likely due to the same reasons addressed earlier in the previous section.

Conclusion and Avenues for Future Research:
Due to the lack of available data and time constraints, there are a number of robustness tests and models I wished to estimate but was unable to do so. As mentioned earlier, the paper used an approximation for price synchronicity. Although the approximated values fall somewhat around those provided by Morck et al.’s research (provided on Table 2 of Morck’s article, for year 1995), using the actual price synchronicity of all stocks, as opposed to that of index stocks, in each given country and year would reduce sampling bias in the estimated models. I also wished to estimate the same fixed effects models but replace the Corruption Perception Index with the “good government” index constructed by Morck et al. that more closely represents the institutional risks foreign investors face in emerging markets.

Ultimately, using a fixed effects model and a three-stage least squares estimation of a simultaneous equations system, this paper finds evidence consistent with an extension of the equity home bias theory. Economic theory suggests that, under the same level of institutional risks, inefficient markets should attract foreign investors and therefore increase the level of FPI inflows into an emerging market. Empirical evidence shows that there is no statistically significant relationship between market efficiency and FPI inflows once the protection of property rights has been accounted for. At “best”, empirical evidence suggests a $0.42 billion rise in FPI per percentage point increase in lagged price synchronicity, but the point estimate is statistically insignificant. This paper proposes that one can reconcile the inconsistency between theory and empirical evidence by looking at an extension of the home bias theory, where some foreign investors prefer efficient markets because they more closely represent their domestic investing environment.

Chinn, Menzie D. and Hiro Ito (2006). "What Matters for Financial Development? Capital Controls, Institutions, and Interactions," Journal of Development Economics, vol. 81, no. 1, pp. 163-192.

De Long, J. Bradford, et al. “Noise Trader Risk in Financial Markets.” Journal of Political Economy, vol. 98, no. 4, 1990, pp. 703–738.

Fama, Eugene F. “Efficient Capital Markets: A Review of Theory and Empirical Work.” The Journal of Finance, vol. 25, no. 2, 1970, pp. 383–417.

French, Kenneth R., and James M. Poterba. “Investor Diversification and International Equity Markets.” The American Economic Review, vol. 81, no. 2, 1991, pp. 222–226.

Li, Jie, and Rajan, Ramkishen S., “Do capital controls make gross equity flows to emerging markets less volatile?” Journal of International Money and Finance, vol. 59, pp 220-244

Morck, Randall K. and Yu, Wayne and Yeung, Bernard Yin, "The Information Content of Stock Markets: Why Do Emerging Markets Have Synchronous Stock Price Movements?" Journal of Financial Economics (JFE), vol. 58, no. 1-2, 2000

Wooldridge, Jeffrey M., 1960-. Introductory Econometrics : a Modern Approach. Mason, Ohio :South-Western Cengage Learning, 2012. Print.

bottom of page