Emerging Stock Markets Return Seasonalities: the January Effect and the Tax-Loss Selling Hypothesis*

Transcription

1 Emerging Stock Markets Return Seasonalities: the January Effect and the Tax-Loss Selling Hypothesis* Stilianos Fountas a Konstantinos N. Segredakis b Working Paper No. 37 June 1999 Department of Economics National University of Ireland, Galway *Acknowledgements: We are grateful to Konstantinos Grigoriadis of the International Finance Corporation (World Bank) for providing the data on stock market indices and participants in the Galway Economics Workshop for helpful comments. The usual disclaimer applies. a b Department of Economics, National University of Ireland, Galway, Ireland. steve.fountas@nuigalway.ie. Tel (ext. 2300). Fax: Financial Analyst, M.Sc.

2 Abstract We test for seasonal effects in stock returns, the January effect anomaly and the taxloss selling hypothesis using monthly stock returns in eighteen emerging stock markets for the period Even though considerable evidence for seasonal effects applies in several countries, we find very little evidence in favour of the January effect and the tax-loss selling hypothesis. These results provide some support to the informational efficiency aspect of the market efficiency hypothesis. Keywords: Market efficiency, seasonal effects, January effect. JEL Classification: G14, G15 2

3 1. Introduction Emerging equity markets have attracted the interest of international investors during the past decade. Much of the attraction derives from the outstanding return performance both in local currency and dollar terms. It is noteworthy that, during the period , the mean annual returns in dollar terms exceeded 35% in the following emerging markets: Argentina (61%), Brazil (39%), Chile (51%), Colombia (48%), Mexico (39%), Philippines (66%), Thailand (36%), Turkey (88%), and Venezuela (52%) (Diacogiannis and Segredakis, 1996). The increasing importance of emerging stock markets can be explained by both push factors arising in mature stock markets and pull factors in developing countries. Push factors include the absence of investment oppurtunities and the low returns on investments in the developed equity markets that have led international investment funds to the emerging equity markets. Pull factors include structural and economic reforms in developing countries, including equity market opening, international equity offerings and exchange rate stabilization programs, that have also proved helpful to the rapid acceleration of net foreign purchases of emerging markets shares. It is noteworthy that for Brazil, Mexico, South Korea and Taiwan, equity portfolio inflows increased from 2,000 million dollars in 1990 to 10,500 million dollars in 1992 (Mullin, 1993). Moreover, during the period , the annual growth rate of listed companies, market capitalization and value traded increased for the emerging markets by 7.7%, 28.8% and 53%, respectively, while for the developed countries the same growth rates were 1.16%, 16.6% and 26.3%, respectively (Diacogiannis and Segredakis, 1996). It should also be kept in mind that, although the emerging market countries encompass 85 per cent of the world s population, their economies produce only 20 per cent of the world GDP and their stock markets represent only 13 per cent of the world capitalization (Fasken, 1997). The recent outstanding development of the emerging stock markets, the globalization of equity markets and the interest in international asset diversification, require fund managers to have a close and spherical knowledge of the return generating process in 1

4 the equity markets worldwide. Despite the increasing importance of emerging stock markets, published research in the area is slowly emerging. Given the volume of research already existing using stock market data in mature stock markets, it is interesting to ask whether the evidence that has accumulated on the stylized facts of the mature stock markets, including the existence of seasonality in stock returns, extends to the emerging markets. In the spirit of this need, our study attempts to investigate some monthly seasonal anomalies in the major emerging stock markets. In particular, our study focuses on seasonal monthly return patterns, the January effect and the tax-loss selling hypothesis 1. Evidence in favour of return seasonality would have important implications for investment strategies as it would invalidate the paradigm of the efficient markets hypothesis (EMH). According to this hypothesis, security prices follow a random walk, thus making it imposible to predict future returns on the basis of publicly available information. This is the informational efficiency aspect of the EMH. A large number of studies has tested for the EMH by examining serial correlations in stock prices. An alternative approach is to test for seasonal patterns in stock returns. This is the focus of our paper. Evidence in favour of seasonality in returns implies that informational efficiency does not hold. However, even if information available at the present time can be used to forecast future returns, it does not mean that the investor is guaranteed to earn supernormal profits. If an investor chooses an investment strategy based on the prediction of a regression with a small coefficient of determination, the risk involved is quite large. In addition, transactions costs can be quite large, thus making the investment not worthwhile (Cuthbertson, 1996). The remaining of the paper is organised as follows: Section 2 reviews the literature. Section 3 presents the methodology and the data set employed. Section 4 discusses 1 These are examples of economic anomalies. An economic anomaly is an empirical result inconsistent with the present economics paradigm. Economics is believed to explain most of behaviour by assuming that agents are rational optimizers whose choices are based on stable, well-defined preferences and whose actions take place in a market-clearing environment. Hence, an anomalous result would be difficult to rationalize within the paradigm (Thaler, 1987). 2

5 the results obtained from our econometric analysis. Finally, the last section summarizes the conclusions of our paper. 2. Literature Review The seasonal behaviour of stock market monthly returns has been documented in several studies. More specifically, a significant number of studies (e.g., Rozeff and Kinney, 1976; Gultekin and Gultekin, 1983) have provided evidence of January returns that are significantly higher than the returns during the rest of the year, aptly dubbed the January effect anomaly. Rozeff and Kinney (1976), for example, have provided evidence for return seasonality in an equally-weighted index of New York Stock Exchange (NYSE) prices for the period The authors found that the average January monthly return was approximately 3.5 per cent while the average return over the other months was 0.5 per cent. Rogalski and Tinic (1986) supported the finding of a January effect using the equally-weighted index of NYSE and the American Stock Exchange stocks for the period The above, and other studies, attempt to explain the January effect by focusing their interest on market frictions that violate the Capital Asset Pricing Model (CAPM) assumptions. The first explanation of the January effect was provided by the tax-loss selling hypothesis. Wachtel (1942), Branch (1977) and Dyl (1977) were among the first to attempt to explain the January effect by appealing to the tax-loss selling hypothesis. According to this hypothesis, investors wait until the tax-year end to sell their common stock losers, in order to realise capital losses to be set against capital gains in order to reduce tax liability. Therefore, there is a downward pressure on the prices of stocks that have faced a price decline during the year. Consequently, at the beginning of the new tax year, in the absence of selling pressure, the downward pressure on stock prices disappears and the stock prices gain their real market price. This phenomenon generates large abnormal stock returns at the turn of each tax year. Roll (1983) argues that small-sized firms are affected more by the tax-loss selling 3

6 hypothesis than large-sized firms are. Similar results are also presented in the study of Reinganum (1983) concerning the US capital market. Brown et. al. (1983) argue that small-firm stocks are likely candidates for tax-loss selling since these stocks typically have higher price variances and consequently larger probabilities of large price declines. However, Brown et. al. (1983) found that in Australian stock markets small firms obtain an average monthly return of 4 per cent, which presented to be fairly constant across months. However, in several cases, the tax-loss selling hypothesis cannot explain the January effect anomaly. Brown et. al. (1983) provide evidence of monthly stock return seasonality in January and July in Australia, even though the beginning of the tax-year is in July. Berges, McConnell and Schlarbaum (1984) found a January effect in the Toronto Stock Exchange prior 1972 when Canada had no taxes on capital gains. January seasonality in the Toronto Stock Exchange was also presented in the study of Gultekin and Gultekin (1983). They also found July seasonality in Australia (but not the validity of the tax-loss selling hypothesis), April seasonality in UK, as well as a January effect in most of the major industrial countries. Furthermore, the findings of Gultekin and Gultekin (1983) indicate a close association between the monthly return seasonalities and the turn of the tax year. Ho (1990) provides evidence that the taxloss selling hypothesis does not receive considerable support in most of the Asia Pacific markets, since only in three out of the nine Asia Pacific markets, the return of the first month of the tax-year was significantly higher than that for all other months. Another explanation of the January effect suggests that abnormal returns in January are due to the new information provided by the firms at the end of the fiscal year (Rozeff and Kinney, 1976). Note that for many firms, announcements of previous year s financial performance, like accounting earnings, are made in January. Keim (1983) suggests that the January effect affects more the small-firm prices than those of large firms for which the information cost is less. 4

7 One more explanation of the January effect, which is based on the well known size effect, is given by Rogalski and Tinic (1986). They found that small-size firms have significantly higher risk (total, systematic and residual) in the beginning of the year than in the rest of the year. Hence, according to the asset pricing theory, investors should be compensated with a higher return (by eight to nine times according to their findings) for the higher risk that they undertake when investing in stocks of small-size firms. A last explanation of the January effect is based on the existence of a positive January risk-return trade off. Tinic and West (1984) report in their study that in the US the risk premium is positive in January and not significantly different from zero over the other eleven months. Moreover, Corhay et. al. (1987), using the cross sectional analysis of the Fama and MacBeth methodology, report in their study a significantly positive relationship between average portfolio returns and systematic risk only in the month of January, for the case of US and Belgium. As far as the London Stock Exchange is concerned, this risk-return trade off was observed only in the month of April. In addition, the findings of Corhay et. al. (1987) for France indicate that the January risk premium was positive and larger than the risk premium over the rest of the year, but not statistically significant. Furthermore, the authors find that for the case of UK, France and Belgium stock markets, the monthly risk premium seasonality was not a reflection of the monthly return seasonality, while that was the case for the NYSE. The above mentioned studies mainly analyze the monthly return seasonality in developed countries. As far as seasonality in emerging stock markets is concerned, the study by Nassir and Mohammad (1987) provides evidence that in Malaysia, the average January returns were significantly positive and higher than in the other months during the period They also found that the tax-loss selling hypothesis is not supported by the data, a finding consistent with the absence of a capital gains tax in Malaysia. Pang (1988) detected the existence of return seasonality during the months of January, April and December in the Hong-Kong stock market. 5

8 Ho (1990), using daily returns for the period January 1975 to November 1987, found that six out of the eight emerging Asia Pacific stock markets exhibit significantly higher daily returns in January than in other months. These markets include Hong-Kong, Korea, Malaysia, Philippines, Singapore and Taiwan. 3. Econometric Methodology and Data 3.1 Methodology In this section we present the econometric approach used in testing for seasonal effects and the January effect and tax-loss selling hypothesis anomalies. In order to test for statistically significant returns, i.e., monthly seasonal effects, we estimate the following regression R t = α 1 D 1t + α 2 D 2t + α 3 D 3t + α 4 D 4t + α 5 D 5t + α 6 D 6t α 12 D 12t + e t (1) where R t stands for the stock market return at time t, e t is a white noise error term and D 1t,..., D 12t are monthly seasonal dummy variables such that D it = 1, for the i th month = 0, otherwise. To test for the January effect, we perform two types of tests: First, we estimate the following regression R t = c + α 2 D 2t + α 3 D 3t + α 4 D 4t + α 5 D 5t + α 6 D 6t α 12 D 12t + e t (2) where the intercept indicates the average return for January and the coefficient α i, i=2,...,12, indicates the difference in return between January and the i th month of the year. The dummy variables are defined as for equation (1). The null hypotheses are 6

9 that each of the dummy coefficients is zero. Evidence that each of the dummy coefficients is less than zero would be consistent with the January effect. The second approach to test for the January effect (and the tax-loss selling hypothesis) is based on the following regression model R t = c 0 + β 1 D 1t + e t (3) where D 1t = 0, for January and/or the first month of the tax year = 1, otherwise. The intercept c 0 measures the average January return and/or the average return of the first month of the tax year. If β 1 is statistically less than zero, there is evidence in favour of the January effect and/or the tax-loss selling hypothesis. Equation (2) provides more information than equation (3) since it indicates the months for which the average return is smaller than the January average return. 3.2 Data We use weekly and monthly data on stock index returns. The weekly data cover the period January 1989 to December 1996 and the monthly data cover the period January 1987 to December The sample of countries includes Argentina, Chile, Colombia, Greece, India, Jordan, Korea, Malaysia, Mexico, Nigeria, Pakistan, Philippines, Portugal, Taiwan, Thailand, Turkey, Venezuela and Zimbabwe. The stock markets of these countries have been classified as emerging by the International Finance Corporation (IFC). The IFC considers as emerging markets the stock markets in countries with income levels that are classified by the World Bank as low or middle income. The returns have been calculated using the stock market indices provided by the Emerging Markets Data Base constructed by the IFC. All returns include dividend yields and capital gains. 7

10 4. Empirical results Tables 1-4 report our main results based on monthly data. The results with weekly data are, in the majority of cases, very similar and are not reported here due to space limitations but are available upon request from the authors. The estimation method is OLS. We report the estimated coefficients and the absolute values of the t-statistics. The error term has been tested for serial correlation up to twelfth order and heteroskedasticity using the Breusch-Godfrey Lagrange Multiplier and LM(1) tests, respectively. In cases where the error term was not white noise, we created the t- statistics using the Newey-West heteroskedasticity- and autocorrelation-adjusted standard errors (Newey and West, 1987). Bartlett weights with various truncation lags were employed in creating the adjusted standard errors. As the results were robust to the choice of the truncation lag, we report only results for truncation lag set at 4. Tests for seasonality in monthly returns are shown in Table 1. The estimated model is equation (1). Significant seasonal effects apply for all countries in our sample. However, for Jordan, Pakistan, Taiwan and Venezuela the evidence is rather weak as statistical significance applies only at 10%. The strongest evidence of significant monthly returns (at 1% significance level) seems to apply for Chile, Colombia, India, Malaysia, Mexico, Nigeria and Zimbabwe. Table 2 shows results of the estimated regression (2) above using monthly data. According to Table 2, evidence for average January returns exceeding the average returns for some of the rest of the months of the year is provided for Chile, Greece, Korea, Taiwan and Turkey. The results of the estimation of regression model (3) are shown in Table 3. According to these results, evidence in favour of the January effect and the tax-loss selling hypothesis is available only for Chile where the average return in January exceeds the average return over the rest of the year. No evidence for the tax-loss selling hypothesis is available for India and Pakistan where the tax year starts in April and July, respectively. 8

11 Table 4 provides information on the relative size of monthly returns for each of the eighteen emerging stock markets. Column 2 indicates the months for which monthly returns are statistically different from zero, as determined according to the results of Table 1. In column 3 we list the months where average return over the sample falls short of returns obtained during the months listed in column 2. The choice of these months was based on a number of t-tests in regressions including monthly dummy variables (of the form of equation (2) above) where the intercept measures the return received in the respective month of column 2. Similarly, column 4 lists the months where monthly returns exceed those obtained in the respective month listed in column 2. The results of Table 4 provide important information to investors of emerging stock markets as they indicate the months where returns are relatively very large or very small. Emerging stock markets can be broken into various groups based on these results. First, in a few markets, returns in some months differ significantly from returns in most of the other months of the year. For example, this is the case of Chile where January returns are relatively large as they exceed returns in seven other months of the year, Colombia, where December returns are large relative to the rest of the months of the year as they exceed the return in every other month except January and Malaysia where December returns exceed returns in six other months of the year. On the other hand, monthly October returns in Greece are relatively low as they fall behind the returns in six other months of the year. In these countries, investors can take advantage of information about the month of the year when investing in the stock market. Second, in some countries (e.g., Jordan, Pakistan, Taiwan), returns in any single month seem to differ significantly from only a few other months. Finally, for Nigeria monthly returns do not differ significantly over the year and for Thailand, Venezuela and Zimbabwe the evidence shows very little difference in returns across months. Therefore, for these three countries, the month of the year provides very little information to stock market investors that could prove useful in obtaining abnormal returns. 9

12 5. Conclusions This paper has tested for seasonality of monthly stock returns, the tax-loss selling hypothesis and the January effect anomaly in eighteen emerging stock markets. Our results show evidence of monthly seasonality in stock returns. However, we have found very little evidence in favour of the January effect and the tax-loss selling hypothesis. These findings are robust to the data frequency used and provide some support to the informational efficiency aspect of the market efficiency hypothesis. The existence of significant differences in monthly returns in several countries should not necessarily imply that supernormal profits can be made in these markets, at least for some investors that are subject to borrowing constraints and high transactions costs. 10

13 References Berges, A., J.J. McConnell, and G.G. Schlarbaum An Investigation of the Turnof-the-Year Effect, the Small Firm Effect and the Tax-Loss-Selling-Pressure Hypothesis in Canadian Stock Returns, Journal of Finance, 39: Branch, B A Tax-Loss Trading Rule, Journal of Business, 50: Brown, P., D. Keim, A. Kleidon and T. Marsh Stock Return Seasonalities and the Tax-Loss Selling Hypothesis: Analysis of the Arguments and Australian Evidence, Journal of Financial Economics, 12: Corhay, A., G. Hawawini, P. Michel Seasonality in the Risk-Return Relationship: Some International Evidence, Journal of Finance, 42: Cuthbertson, K, 1996, Quantitative Financial Economics. Chichester: Wiley. Diacogiannis, G.P., and K.N. Segredakis The Athens Stock Exchange and the Emerging Equity Markets, Hellenic Banks Association, (In Greek). Dyl, E.A Capital Gains Taxation and Year-End Stock Market Behavior, Journal of Finance, 32: Fasken, H Emerging Markets: Parade of Growth, The International, Gultekin, M.N. and N.B.Gultekin Stock Market Seasonality: International Evidence, Journal of Financial Economics, 12: Ho, Yan-Ki Stock Return Seasonalities in Asia Pacific Markets, Journal of International Financial Management and Accounting, 2: Keim, B.D Size Related Anomalies and Stock Return Seasonalities: Further Empirical Evidence, Journal of Financial Economics, 12: Mullin, J Emerging Equity Markets in the Global Economy, Federal Reserve Bank of New York Quarterly Review, 19: Nassir, A. and S. Mohammad The January Effect of Stocks Traded on the Kuala Lumpur Stock Exchange: An Empirical Analysis, Hong Kong Journal of Business Management, 5: Newey, W. and K. West A Simple Positive Semi-Definite, Heteroskedasticity and Autocorrelation Consistent Covariance Matrix, Econometrica, 55:

14 Pang, Q.K.L An Analysis of the Hong Kong Stock Return Seasonality and Firm Size Anomalies for the Period 1977 to 1986, Hong Kong Journal of Business Management, 6: Reinganum, M The Anomalous Stock Market Behaviour of Small Firms in January: Empirical Tests for Tax-loss Selling Effects, Journal of Financial Economics, 12: Rogalski, R.J. and S. Tinic The January Size Effect: Anomaly or Risk Mismeasurement?, Financial Analysts Journal, 42: Roll, R The-Turn-of-the-Year Effect and the Return Premia of Small Firms, Journal of Portfolio Management, 9: Rozeff, M. and W. Kinney Capital Market Seasonality: The Case of Stock Returns, Journal of Financial Economics, 3: Thaler, R Anomalies: The January Effect, Journal of Economic Perspectives, 1: Tinic, S. and R. West Risk and Return: January versus the Rest of the Year, Journal of Financial Economics, 13: Wachtel, S.B Certain Observations on the Seasonal Movement in Stock Prices, Journal of Business, 15:

15 Table 1: Tests for Seasonal Effects (Monthly Data: January December 1995) Argentina(a) Chile(a) Colombia(a) Greece(a) India(a) Jordan Korea Malaysia Mexico(a) D1.07(.85).08(3.64***) -.02(.15).07(1.89*) -.06(.59) -.02(.98).03(1.37).004(.17).07(1.51) D2.31(1.21).06(2.31*).001(.04).09(1.88*).04(1.16) -.001(.09) -.01(.45).06(2.35**).07(.98) D3.18(1.93*).002(.09) -.003(.17).02(.57).05(.59).01(.66).04(1.54) -.009(.35).08(1.94*) D4.13(2.20**) -.01(.72) -.02(1.85*).04(.72) -.004(.16).01(.66) -.03(1.25).04(1.69*).04(1.13) D5.32(1.45).02(1.03).02(1.67*) -.02(.61) -.02(.52).02(1.00).03(1.23).05(1.89*).12(3.46***) D6.28(.91).05(2.04**).04(1.78*).06(1.20).02(1.25).01(.68) -.02(.89).01(.43).01(.28) D7.07(1.36).02(1.39).03(1.44).05(1.59).07(1.44) -.01(.88).04(1.47).04(1.38).10(2.35**) D8.24(2.11**).05(1.67*).004(.25).04(.95).06(2.64***) -.02(1.32) -.02(.68) -.03(1.28).05(1.33) D9.15(1.45).02(1.19).04 (1.72*).04(.50).04(1.38).02(1.16).02(.95) -.006(.25).006(.21) D (.44).01(.40).04(1.01) -.04(1.96**) -.03(1.07).01(.78).006(2.23**) -.02(.61) -.003(.05) D (1.19).006(0.22).02(.48) -.04(1.51) -.01(.37).004(.25).01(.46) -.02(.62).009(.15) D 12.17(1.95*).06(2.80***).14(2.70***).02(1.30).01(.54).03(1.82*).01(.57).07(2.76***).01(.50) BG 10.28(.59) 14.84(.25) 11.48(.49) LM(1).002(.97) 1.81(.18).001(.97) Table 1 (continued): Tests for Seasonal Effects (Monthly Data: January December 1995) Nigeria(a) Pakistan(a) Philippines(a) Portugal(a) Taiwan Thailand Turkey Venezuela Zimbabwe(a) D1.03(3.37***).12(1.10) -.02(.65).04(1.01).07(1.35).03(.85).18(2.42**).08(1.55).05(1.54) D2.03(3.41***).01(.57).04(1.10).02(.81).10(1.84*).03(.90).06(.75).04(.81).01(.26) D3.03(2.89***) -.002(.15).02(.65).05(1.91*).02(.33).02(.48).001(.02).01(.22) -.005(.20) D4.02(2.05**).02(.93).05(1.62).01(.42).05 (1.02).03(.78).04(.57) -.001(.03).04(2.52**) D5.03(2.03**) -.000(.02).02(.62).02(.93).000(.09).04(1.40).09(1.25).01(.25).05(3.08***) D6.07(2.15**).02(1.42).07(1.45) -.02(2.13**) -.05(.95).04(1.31).14(1.85*) -.01(.22).002(.06) D7.05(2.10**).03(.93).05(2.38**).02(.85).06(1.19).02(.73).16(2.14**) -.009(.16).04(1.75*) D8.04(3.29***) -.02(1.08) -.03(.72).06(1.43).02(.46).001(.05).05(.73).10(1.91*).05(2.36**) D9.02(4.01***).01(.62) -.06(1.56).05(.65).04(.69).004(.13).08(1.08).04(.72).02(1.34) D 10.04(4.79***).02(.90).05(1.45) -.05(1.49) -.05(1.05).004(.12) -.05(.70).09(1.57).04(1.34) D 11.03(3.26***).02(.51) -.01(.61) -.04(1.60).05(.89) -.03(1.06).06(.83) -.02(.30).007(.27) D 12.02(2.40**).06(1.69*).07(1.42) -.03(1.13).02(.35).08(2.36**).04(.54) -.03(.52).03(1.10) BG 8.92(.71) 15.16(.23) 8.59(.74) 14.53(.27) LM(1.31(.58).44(.51) 3.03 (.08).004(.95) ) Note : Absolute t-statistics are in parentheses. ***, ** and * denote significance at 1%, 5% and 10%, respectively. BG stands for the Breusch-Godfrey Lagrange-Multiplier test for autocorrelation which is a chisquare test for autocorrelation up to order 12. The LM(1) test for heteroskedasticity is distributed as chi-square with one degree of freedom. In countries (a) the t-statistics were created using the Newey-West adjusted standard errors (Bartlett weights, truncation lag is 4). In the rest of the countries the OLS standard errors are used. The BG and LM(1) statistics (p-values in parentheses) indicate absence of autocorrelation and heteroskedasticity for the OLS case. 13

16 Table 2: Tests for the January effect (Monthly Data: January December 1995) Argentina(a) Chile(a) Colombia(a) Greece(a) India(a) Jordan Korea Malaysia Mexico(a) Nigeria(a) c.07(.85).08(3.64***) -.02(.15).07(1.89*) -.06(.59) -.02(.99).03(1.37).004(.17).07(1.51).03(3.37***) D2.24(.79) -.03(.65).02(.14).01(.21).11(.84).01(.64) -.05(1.29).06(1.54) -.005(.11) -.003(.33) D3.11(.93) -.08(2.04**).01(.12) -.05(1.11).12(.90).03(1.17).004(.12) -.01(.37).008(.11).002(.19) D4.06(.59) -.10(3.27***) -.001(.05) -.03(.44).06(.53).03(1.16) -.07(1.85*).04(1.08) -.03(.50) -.006(.49) D5.25(1.06) -.07(2.25**).04(.30) -.09(1.81*).04(.38).03(1.41) -.004(.1).04(1.22).05(.78) -.001(.07) D6.21(.66) -.03(.91).06(.46) -.01(.17).09(.80).03(1.18) -.06(1.6).007(.18) -.06(1.06).04(1.19) D7.004(.04) -.06(2.22**).05(.41) -.02(.49).13(1.10).001(.08).002(.07).03(.85).02(.35).02(.74) D8.17(1.19) -.04(1.09).02(.18) -.03(.62).12(1.12) -.005(.23) -.05(1.45) -.04(1.03) -.02(.33).007(.50) D9.08(.56) -.06(1.79*).06 (.51) -.04(.43).1(.92).03(1.52) -.01(.29) -.01(.3) -.07(1.27) -.004(.40) D (.97) -.07(1.93*).06(.47) -.11(2.66***).04(.36).03(1.25).02(.61) -.02(.55) -.08(.89).01(1.04) D (1.22) -.08(2.35**).04(.30) -.12(2.59***).05(.49).02(.87) -.02(.65) -.02(.56) -.07(.73).001(.08) D 12.10(.68) -.02(.81).16(1.27) -.05(1.14).08(.77).04(1.99**) -.02(.56).07(1.83*) -.06(1.08) -.004(.50) BG 10.28(.59) 14.84(.25) 11.48(.49) LM(1).002(.97) 1.81(.18).001(.97) Table 2 (continued): Tests for the January effect (Monthly Data: January December 1995) Pakistan(a) Philippines(a) Portugal(a) Taiwan Thailand Turkey Venezuela Zimbabwe c.12(1.10) -.02(.65).04(1.01).07(1.35).03(.85).18(2.42**).08(1.55).05(1.54) D2 -.10(.89).06(1.13) -.02(.37).03(.35).002(.04) -.12(1.18) -.04(.53) -.04(1.31) D3 -.12(1.13).04(.85).005(.14) -.05(.72) -.01(.26) -.18(1.70*) -.07(.94) -.06(1.39) D4 -.09(.90).07(1.87*) -.03(.62) -.02(.23) -.002(.05) -.14(1.31) -.09(1.12) -.02(.41) D5 -.12(1.10).04(.88) -.02(.37) -.07(.95).02(.39) -.09(.83) -.07(.92) -.01(.18) D6 -.09(.89).09(1.56) -.06(1.53) -.12(1.63).01(.32) -.04(.40) -.10(1.25) -.05(1.16) D7 -.08(.76).07(1.91*) -.02(.42) -.008(.11) -.004(.08) -.02(.19) -.09(1.21) -.01(.28) D8 -.13(1.26) -.005(.10).02(.40) -.05(.63) -.03(.57) -.13(1.19).02(.25) -.01(.13) D9 -.11(.99) -.04(.84).01(.15) -.03(.47) -.02(.51) -.10(.95) -.05(.59) -.03(.77) D (.93).07(1.39) -.09(1.72) -.12(1.70*) -.02(.52) -.23(2.20**).001(.01) -.01(.21) D (1.04).009(.28) -.08(1.56) -.02(.33) -.06(1.35) -.12(1.12) -.10(1.31) -.04(1.25) D (.58).09(1.48) -.07(1.43) -.05(.7).05(1.06) -.14(1.33) -.11(1.47) -.02(.68) BG 8.92(.71) 15.16(.23) 8.59(.74) LM(1).31(.58).44(.51) 3.03(.08) Note : Absolute t-statistics are in parentheses. ***, ** and * denote significance at 1%, 5% and 10%, respectively. BG stands for the Breusch-Godfrey Lagrange-Multiplier test for autocorrelation which is a chisquare test for autocorrelation up to order 12. The LM(1) test for heteroskedasticity is distributed as chi-square with one degree of freedom. In countries (a) the t-statistics were created using the Newey-West adjusted standard errors (Bartlett weights, truncation lag is 4). In the rest of the countries the OLS standard errors are used. The BG and LM(1) statistics (p-values in parentheses) indicate absence of autocorrelation and heteroskedasticity for the OLS case. 14

17 Table 3: Tests for the January Effect and the Tax-loss Selling Hypothesis (Monthly Data: January December 1995) Argentina(a) Chile Colombia(a) Greece(a) India(a) India(b) Jordan(a) Korea Malaysia Mexico(a) c 0.07(.89).08(3.56***) -.02(.15).07(1.99**) -.06(.62) -.004(.08) -.02(.45).03(1.35).004(.17).07(1.59) D1.09(.83) -.06(2.32**).05(.40) -.05(1.28).09(.84).02(.39).02(.65) -.02(.85).01(.46) -.03(.61) BG 16.86(.16) 10.34(.59) 13.33(.35) 14.20(.29) LM(1).11(.74).33(.57) 2.72(.10).005(.94) Nigeria Pakistan(a) Pakistan(c) Philippines Portugal Taiwan Thailand Turkey Venezuela Zimbabwe(a) c 0.03(2.03**).12(1.15) 0.03(.84) -.02(.59).04(1.09).07(1.36).03(.86).18(2.46**).08(1.58).05(1.61) D1.005(.35) -.10(1.02) -.009(.23).04(1.22) -.03(.79) -.05(.89) -.006(.19) -.12(1.55) -.06(1.14) -.03(.92) BG 15.93(.19) 13.27(.35) 7.73(.81) 17.50(.13) 8.83(.72) 16.11(.19) 6.50(.89) 11.02(.53) LM(1).34(.56).02(.90).19(.67).002(.97).34(.59).08(.77).007 (.94).03(.87) Note : Absolute t-statistics are in parentheses. ***, ** and * denote significance at 1%, 5% and 10%, respectively. BG stands for the Breusch-Godfrey Lagrange-Multiplier test for autocorrelation which is a chisquare test for autocorrelation up to order 12. The LM(1) test for heteroskedasticity is distributed as chi-square with one degree of freedom. In countries (a) the t-statistics were created using the Newey-West adjusted standard errors (Bartlett weights, truncation lag is 4). In the rest of the countries the OLS standard errors are used. The BG and LM(1) statistics (p-values in parentheses) indicate absence of autocorrelation and heteroskedasticity for the OLS case. The dummy variable D 1 takes the value 0 in January (the first month of the tax year) with the exception of India (b) and Pakistan (c) where it takes the value 0 in the first month of the tax year, that is, April and July, respectively. 15

18 Table 4: Monthly Return Differences (1) (2) (3) (4) Country Month of statistically significant returns Months that have a mean return that is significantly lower than that of the 16 Months that have a mean return that is significantly higher than that of the month in column (2) month in column (2) Argentina March Oct., Nov. - April Oct., Nov. - August Oct., Nov. - December Oct., Nov. - Chile Jan Apr, Mar, May, July, Sept., Oct., Nov. Febr Mar., Apr. - June April - Aug April - Dec Mar, Apr, Nov. - Colombia Apr - May, June, July, Sept, Dec. May Apr - June Apr - Sept Apr - Dec Febr, Mar, Apr, May, June, July, Aug, Sept, Oct, Nov. Greece Jan May, Oct, Nov - Febr May, Oct, Nov - Oct - Jan, Febr, June, July, Aug, Dec. India Aug Apr, May, Oct, Nov - Jordan Dec. Jan, July, Aug. - Korea Oct. Febr, Apr, June, - Aug. Malaysia Febr Mar, Aug, Sept, Oct, - Nov Apr Aug - May Aug, Oct, Nov - Dec Jan, Mar, Aug, Sept, - Oct, Nov Mexico Mar - - May Apr, June, Sept, Oct, - Dec. July June, Sept. - Nigeria Jan - - Febr - - Mar - - Apr - - May - - June - - July - - Aug - - Sept - - Oct - - Nov - - Dec

19 Pakistan Dec Mar, Aug. - Philippines July Jan, Aug, Sept, - Nov. 17

20 Table 4 (continued): Monthly Return Differences Portugal Mar June, Oct, Nov, Dec. - June - Mar, May, Aug. Taiwan Febr. June, Oct. - Thailand Dec. Nov. - Turkey Jan. Mar, Oct - June Oct - July Oct - Venezuela Aug. Dec. - Zimbabwe Apr - - May Mar - July - - Aug

21 Department of Economics National University of Ireland, Galway Working Paper Series No. 37 June 1999 Emerging Stock Markets Return Seasonalities: the January Effect and the Tax-Loss Selling Hypothesis, Stilianos Fountas and Konstantinos N. Segredakis. No. 36 June 1999 Agricultural entrepreneurs as entrepreneurial partners in landuse management: a policy-based characterization, Scott R. Steele No. 35 June 1999 The Monetary Transmission Mechanism: Evidence and Implications for European Monetary Union, Stilianos Fountas and Agapitos Papagapitos. No. 34 May 1999 Exchange rate pass-through, the terms of trade and the trade balance, Eithne Murphy and Lelio Iapadre. No. 33 May 1999 The Impact of Health Status on the Duration of Unemployment Spells and the Implications for Studies of the Impact of Unemployment on Health Status, Jennifer Stewart. No. 32 December 1998 Subsidies in Irish Fisheries: Saving Rural Ireland?, Vilhjàlmur Wiium. No. 31 October 1998 Has the European Monetary System Led to More Exports? Evidence from Four European Union Countries, Stilianos Fountas and Kyriacos Aristotelous. (Published in Economics Letters, Vol. 62, No. 3, 1999). No. 30 October 1998 Real Interest Rate Parity under Regime Shifts: Evidence for Industrial Countries, Jyh-lin Wu and Stilianos Fountas. No. 29 October 1998 Analyzing Gender-Based Differential Advantage: A Gendered Model of Emerging and Constructed Opportunities, Scott R. Steele. No. 28 September 1998 The Impacts of Transition on the Household in the Provinces of Kazakhstan: The Case of Aktyubinsk Oblast, Pauric Brophy. No. 27 July 1998 A Comparison of the Effects of Decommissioning, Catch Quotas, and Mesh Regulation in Restoring a Depleted Fishery, J. Paul Hillis and Vilhjàlmur Wiium. No. 26 July 1998 The Sensitivity of UK Agricultural Employment to Macroeconomic Variables, Patrick Gaffney. 19

22 No. 25 July 1998 The Economic and Social Costs of Alzheimer's Disease and Related Dementias in Ireland: An Aggregate Analysis, Eamon O'Shea and Siobhán O'Reilly. No. 24 July 1998 Testing for Real Interest Rate Convergence in European Countries, Stilianos Fountas and Jyh-lin Wu. (Published in the Scottish Journal of Political Economy, Vol. 46, No. 2, 1999). No. 23 April 1998 Production, Information and Property Regimes: Efficiency Implications in the Case of Economies of Scope, Scott R. Steele. No. 22 April 1998 An Empirical Analysis of Short-Run and Long-Run Irish Export Functions: Does Exchange Rate Volatility Matter?, Donal Bredin, Stilianos Fountas, Eithne Murphy. No. 21 April 1998 Technology and Intermediation: the Case of Banking, Michael J. Keane and Stilianos Fountas. No. 20 March 1998 Are the US Current Account Deficits Really Sustainable?, Stilianos Fountas and Jyh-lin Wu. (Forthcoming in the International Economic Journal) No. 19 December 1997 Testing for Monetary Policy Convergence in European Countries, Donal Bredin and Stilianos Fountas. (Published in the Journal of Economic Studies, Vol. 25, No. 5, 1998). No. 18 September 1997 New Fields of Employment: Problems and Possibilities in Local and Community Economic Development, Michael J. Keane. No. 17 September 1997 Estimation of the Impact of CAP Reform on the Structure of Farming in Disadvantaged Areas of Ireland, Eithne Murphy and Breda Lally. No. 16 May 1997 Exchange Rate Volatility and Exports: the Case of Ireland, Stilianos Fountas and Donal Bredin. (Published in Applied Economics Letters, Vol. 5, No. 5, 1998) No. 15 May 1997 Tests for Interest Rate Convergence and Structural Breaks in the EMS, Stilianos Fountas and Jyh-lin Wu. (Published in Applied Financial Economics, Vol. 8, No. 1, 1998) No. 14 May 1997 Cointegration Tests of the Profit-Maximising Equilibrium in Greek Manufacturing , Theodore Lianos and Stilianos Fountas. (Published in International Review of Applied Economics, Vol. 11, No. 3, 1997) 20

23 No. 13 April 1997 Agriculture and the Environment in Ireland: Directions for the Future, Eithne Murphy and Breda Lally. (Published in Administration, Vol. 46, No. 1, 1998) No. 12 March 1997 Male Mortality Differentials by Socio-Economic Group in Ireland, Eamon O'Shea. (Published in Social Science and Medicine, Vol.45, No.6, 1997) No. 11 July 1996 Testing for the Sustainability of the Current Account Deficit in Two Industrial Countries, Jyl-Lin Wu, Stilianos Fountas and Show-Lin Chen. (Published in Economics Letters, Vol. 52, No. 2, 1996) No. 10 April 1996 Towards Regional Development Programmes in Russia, Michael Cuddy. No. 9 April 1996 Uncertainty in the General Theory and A Treatise on Probability, Joan O'Connell. No. 8 December 1995 Some Evidence on the Export-Led Growth Hypothesis for Ireland, Stilianos Fountas. (Forthcoming in Applied Economics Letters). No. 7 November 1995 Caring and Theories of Welfare Economics, Eamon O'Shea and Brendan Kennelly. No. 6 September 1995 The Relationship Between Inflation and Wage Growth in the Irish Economy, Stilianos Fountas, Breda Lally and Jyh-Lin Wu. (Forthcoming in Applied Economics Letters) No. 5 September 1995 Quality and Pricing in Tourist Destinations, Michael J. Keane. (Published in Annals of Tourism Research, Vol. 24, No. 1, 1997) No. 4 September 1995 An Empirical Analysis of Inward Foreign Direct Investment Flows in the European Union with Emphasis on the Market Enlargement Hypothesis, Kyriacos Aristotelous, Stilianos Fountas. (Published in the Journal of Common Market Studies, Vol. 30, No. 4, 1996) No. 3 September 1995 The Social Integration of Old People in Ireland, Joe Larragy and Eamon O'Shea. No. 2 September 1995 Caring and Disability in Long-Stay Institutions, Eamon O'Shea and Peter Murray. (Published in the Economic and Social Review, Vol. 28, No. 1, 1997) No. 1 September 1995 Are Greek Budget Deficits `too large'? Stilianos Fountas and Jyh-lin Wu. (Published in Applied Economics Letters, Vol. 3, No. 7, 1996). 21

24 Enquiries: Department of Economics, National University of Ireland, Galway. Tel: , ext Fax: Web: 22