فرضیه های فروش کاهش مالیات، نقدینگی بازار و فشار قیمت در آستانه تغییر سال
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|13473||2003||26 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of Financial Markets, Volume 6, Issue 1, January 2003, Pages 73–98
In this paper, we use intra-day data for all stocks listed on the ISSM and provide new and direct evidence consistent with the tax-loss selling hypothesis. We find that (a) there is abnormal selling pressure prior to the year-end for stocks that have experienced large capital losses in the current and prior years (b) investors delay realizing capital gain by postponing the sale of capital gain stocks until after the new year (c) there is a significant decrease in the average trade size for stocks with large capital losses before the year-end and for stocks with capital gains in the new year, which suggests that individuals, rather than institutional investors, are the major sellers around the year-end (d) the tax-loss selling hypothesis, and not firm size or share price, is the fundamental explanation for abnormal January returns. Further, small or low share priced firms with capital gains do not experience abnormal returns in January. However, conditional on capital losses, small or low share priced firms magnify the turn-of-the-year effect (e) On average, the increase in selling activity adversely affects market liquidity by increasing bid-ask spreads and reducing depths. (f) The tax-loss selling pressure not only causes the price to be at the bid at the year-end, it also temporarily depresses the equilibrium price indicating the short run demand curve is not perfectly elastic (g) the year-end buying activity suggests that large investors buy capital loss stocks prior to the year-end to take advantage of the temporarily depressed price and capital gain stocks after the new year to reinvest the proceeds of the tax-loss selling.
Rozeff and Kinney (1976) are the first to document that mean equity returns in January are higher than in other months. Further, Keim (1983) and Roll (1983) document that the high January equity returns are concentrated within a narrow window that extends from the last trading day in December to the first few days in January of the following year. This seasonal pattern in stock returns, which Roll (1983) refers to as “turn-of-the-year effect”, is found to be most pronounced in small firms (Keim, 1983; Reinganum, 1983; Blume and Stambaugh, 1983) or in low share price firms (Jaffe et al., 1989; Bhardwaj and Brooks, 1992) or in both. Several hypotheses have been put forward to explain the abnormal January returns. Among them are the risk changes (Tinic and West, 1984; Ritter and Chopra, 1989), information changes (Seyhun, 1988) and institutional window dressing (Bildersee and Kahn, 1987; Lakonishok et al., 1991; Eakins and Sewell, 1994). However, the explanation that has received the most attention in the literature to date is the tax-loss selling hypothesis of Givoly and Ovadia (1983), Reinganum (1983) and Roll (1983). This hypothesis posits that investors who are motivated to reduce their year-end tax liability sell stocks that have experienced a decline in price over the year and use the realized losses to offset capital gains. These transactions produce a selling pressure that results in an increase in the number of trades at the bid price (i.e., sell trades), causing a decline in year-end stock prices. At the beginning of the new year, the selling pressure abates and stocks resume trading at their equilibrium price. The tax-loss motivated trading activity around the turn-of-the-year therefore results in positive abnormal stock returns in the new year. Several studies have tested the tax-loss selling explanation by examining the return patterns around the year-end. However, the results presented in these papers do not provide conclusive support of this hypothesis. For example, Roll's (1983) finding of a negative relation between the prior year's returns and the turn-of-the year returns is consistent with both of the tax-loss selling and institutional window dressing hypotheses.1 Further, Brown et al. (1983) and Kato and Schallheim (1985) examine the turn-of-the-year effect for Australia and Japan respectively and document a similar stock return pattern in January even though these countries having different tax laws and tax-year ends.2 Other studies examine year-end trading activity and provide evidence consistent with the tax-loss selling hypothesis. For example, Dyl (1977) finds that firms with the greatest decline in price experience the largest increase in year-end trading volume while Keim (1989) documents that there is an increase in selling pressure prior to the year-end, especially for firms in the smallest price deciles. Dyl (1977) however does not examine whether the increase in trading volume is caused by higher buying activity or by an increase in selling activity while Keim (1989) fails to relate the selling pressure to the tax-loss potential of equities. Therefore, the results of these studies do not provide direct testing of the fundamental implication of the tax-loss selling hypothesis of whether a relation exists between the tax-loss potential of stocks and selling activity immediately prior to the year-end and if this relation then disappears in the new year. Further, while these results indirectly support the tax-loss selling hypothesis, they are also consistent with the predictions of the institutional window-dressing hypothesis, which predicts higher year-end selling and trading activity driven by institutional managers rebalancing their portfolios. The primary objective of this study is to present new and direct empirical testing of the tax-loss selling hypothesis by examining the trading pattern of all stocks listed on the Institute for the Study of Securities Markets (ISSM) database between 1986 and 1992. The intra-day transaction and quote information obtained from the dataset lets us explicitly classify each trade as either a buy or a sell. This allows us to conduct a more precise test of the tax-loss selling hypothesis than Keim (1989) who uses only the daily closing price to classify the entire day's trades as either buys or sells. Further, we divide firms into quartiles based on their tax-loss potential measured as the decline in share price from the highest point reached between January and the end of November to the last trading day in November. This classification scheme controls for the deficiencies in Dyl's (1977) and Keim's (1989) studies and allows us to directly test the fundamental implications of the tax-loss selling hypothesis that (a) there is an abnormal selling pressure prior to the year-end for stocks with capital losses and that this pressure is positively related to stocks’ tax-loss potential, (b) the selling pressure for firms with the largest tax-loss potential cease in the new year when the motivation to realize losses disappears, and (c) there is selling pressure in the new year for stocks that have increased in value from investors motivated to reduce their current year's tax liability by delaying realizing capital gains. While the first two implications are consistent with both the tax-loss selling and the institutional window-dressing hypotheses, the last implication is predicted by the tax-loss selling hypothesis only. The intra-day transactional data enables us to further distinguish between the tax-loss selling hypothesis and the institutional window-dressing hypothesis by determining the source of the year-end trading activity. Because individual investors, faced with capital constraints, are more likely to trade in smaller lots relative to institutional investors, we use the average trade size to identify the dominant trader at year-end. Thus, a decrease in trade size at the year-end is consistent with tax-loss selling as opposed to institutional trading. Sias and Starks (1997) also test this issue by analyzing institutional ownership data. However, the sample in their study is restricted to only 139 firms during the 1990–1991 year-end period. The authors themselves note that the trading activity during their sample period could be influenced by the “increasing tensions between United States and Iraq just prior to the Gulf war”. In our paper, we examine all stocks listed on the ISSM database over a 6-year period, which allows us to draw stronger conclusions about the source of year-end trading activity than prior studies. Finally, the methodology of our study allows us to provide new evidence supporting the tax-loss selling hypothesis by examining whether the trading activity of stocks classified on short-term and long-term capital losses is consistent with the prediction of the tax-loss selling hypothesis as suggested by Poterba and Weisbenner (2001). The finding of seasonal trading pattern around the year-end also motivates us to reinvestigate the issue of whether tax-loss selling explains abnormal January returns and whether the size (share price) effect continues to be significant after controlling for tax-loss selling. Keim (1989) finds that relative to high share price firms, low share price firms experience higher returns at the turn-of-the-year after controlling for bid-ask fluctuation. Reinganum (1983) limits his attention to firms in the highest and lowest quartiles of capital losses and concludes that both tax-loss selling and firm size are significant. Because firms that experience capital losses over the year are more likely to be small firms or firms with low share prices any test of the tax-loss selling hypothesis must control for these firm size and share price effects. Previous studies by Keim (1989) and Reinganum (1983) do not explicitly account for the tax-loss potential of stocks and firm size. In this paper, we not only directly control for the firm size (share price), but we also include it as an interactive term with the firm's tax-loss potential to capture the possibility that firm size (share price) might magnify the tax-loss selling effect. This study also presents a more comprehensive examination of the tax-loss selling hypothesis by extending the literature in two ways. First, we examine the impact of tax-loss selling motivated trading activity on market liquidity. An increase in year-end trading might make it more costly to match order flow causing an increase in bid ask spreads. Similarly, increased year-end trading might cause a share's market depth to be exceeded, lowering liquidity. Alternatively, year-end trading might have no significant impact on market liquidity if market makers view these transactions as being informationally neutral and are therefore willing to maintain spreads and increase depth. Therefore, the influence of year-end trading on market liquidity remains an unresolved issue, open to new insights from the analysis of transactional data. We also attempt to provide a preliminary microstructure explanation as to why small firms experience the highest year-end returns. Our second set of contributions focus on the impact of tax-induced abnormal trading activity on equilibrium prices. The price pressure hypothesis posits that an increase in trading activity will affect a stock's equilibrium price. Evidence consistent with this hypothesis is significant because it offers investors a potential profitable trading opportunity and also violates a fundamental assumption of asset pricing theories that the demand curve for a firm's shares is perfect elastic (i.e., horizontal). While prior studies papers have tested this hypothesis, the transactions examined by these studies are either often contaminated by the release of economically relevant information or limited to a few additions, deletions or revisions in the composition of a stock index.3 In this study we examine the impact of predictable (and hence devoid of information) but abnormal trading activity induced by tax-loss selling on year-end prices. Further, we conduct cross-sectional tests on all firms over several year-ends using bid (ask) to bid (ask) returns that eliminates the possibility that a price change is due to bid-ask bounce. Our study therefore offers a sharper test of the price pressure hypothesis with higher statistical power and allows us to draw stronger conclusions than those reported by earlier researchers. We find strong evidence consistent with the tax-loss selling hypothesis. We observe a significant increase in abnormal selling activity prior to the turn-of-the-year and this increase is highest for stocks that experience the largest capital losses during the year. After the turn-of-the-year, there is no evidence of abnormal selling pressure for these stocks. For firms that have experienced capital gains however, we find an increase in abnormal selling pressure in the new year. We also find that distinguishing between short-term (current year) and long-term (prior year) losses provides additional evidence that investors’ trading strategies are consistent with the goal of reducing their tax liability. We observe a decrease in the average trade size for stocks with large capital losses before the year-end and for stocks with capital gains in the new year. These findings imply that capital constrained individuals, as opposed to institutions, are the primary cause of the trading pattern at year-end. The results of our regression analyses suggest that the tax-loss selling hypothesis is the primary explanation for the abnormal January returns. The significant positive relation between tax-loss potential and turn-of-the-year returns holds even after controlling for firm size and share price effects. While firm size and share price are insignificant by themselves in explaining the abnormal January returns, they become significant only when interacted with a measure of capital loss. This implies that for firms with no capital losses, smaller size (lower share price) firms do not yield higher returns than larger size (higher share price) firms at the year-end. However, conditional on experiencing a decline in value, smaller firm size (or lower share price) significantly magnifies the impact of tax-loss selling on turn-of-the-year returns. This result, coupled with the dominance of individuals in year-end trading, provide strong evidence for the argument that tax-loss selling accounts for price and size effects around the turn-of-the- year. We also find evidence that tax-loss trading significantly impacts the market microstructure of a firm's securities. On average, there is an increase in the bid-ask spread coupled with a decrease in market depth around the year-end, thus adversely influencing market liquidity. However, we observe a significant increase in bid depth for firms with the largest tax-loss potential, but this increase is insufficient to meet the increased selling volume for these firms induced by tax-loss trading. The shortfall in bid depth further magnifies the tax-loss selling effect for these firms. Moreover, the insufficiency is most severe for small firms who therefore experience a greater tax-loss selling effect than larger firms. We also find evidence that the tax-related increase in trading activity around the turn-of-the-year affects equilibrium prices. There is a significant reduction in average stock prices at the year-end with the decline most pronounced for stocks with large capital losses. These results hold even after controlling for price movement attributable to bid-ask bounce. These findings suggest that stocks are not perfect substitutes and that their demand curves are not perfectly elastic in the short-run. However, a trading strategy of buying large capital loss stocks prior to the year-end and selling them after the new year fails to yield abnormal profit net of transaction costs. Finally, an examination of investors buying activity leads to the following interesting findings. First, there is an increase in buying activity prior to the year-end, especially for the shares of large capital-loss firms. This is consistent with Roll's (1983) hypothesis that some investors attempt to profit from the highly predictable seasonal price pattern induced by tax-loss selling. Second, there is also an increase in buying activity at the start of the new year that supports Ritter's (1988) hypothesis that investors who engage in tax-loss selling delay reinvesting (i.e., “park their proceeds”) until after the new year. However, the abnormal buying activity is primarily concentrated on stocks experiencing a capital gain. We organize this paper into four sections. In the following section we discuss our data and methodology. In Section 3 we present the results from an empirical examination of our hypotheses. We provide a brief summary and conclusion in Section 4.
نتیجه گیری انگلیسی
This studypresents a comprehensive test of the implications of the tax-loss selling hypothesis by analyzing transaction data over a 6-year sample period and provides new and direct results consistent with that hypothesis. We find an abnormal increase in selling activityprior to the year-end for firms that have experienced a decline in share price over the year. The abnormal selling pressure for these stocks disappears in January. For firms with capital gains however, we observe abnormal selling activityin the new year. This is consistent with the fundamental implication of the tax-loss selling hypothesis that investors who are motivated to reduce their tax liabilities, sell capital loss stocks prior to the year-end to realize losses and then delay the sale of capital gains stocks until the new year to postpone the recognition of capital gains. An examination of trading activityof stocks with short-term and long- term losses is also consistent with the implications of the tax-loss selling hypothesis. We also test to see if the trading pattern observed in Januarymight be explained byinstitutional window-dressing rather than individual tax-loss trading. We find a reduction in the average size of sell side transactions during the turn-of-the-year period. Since individual investors are more capital constrained than institutions and hence are more likelyto trade in smaller lots, we conclude similar to Griffiths and White (1993), Sias and Starks (1997), and Poterba and Weisbenner (1998), that individual investors dominate year-end trading. This studyalso makes a significant contribution in explaining the well- documented January anomaly. We find that the tax-loss selling hypothesis, as opposed to firm size or share price effects, is the primaryexplanation for the January anomaly. Firm size/share price become significant only when theyare interacted with measures of tax-loss potential, implying that for stocks with capital losses, smaller firm size (lower share price) will magnifythe turn-of-the-year effect. We find no evidence that small/low price firms with capital gains earn significantlyhigher returns than large/high price firms. Further, this study examines the impact of year-end trading on market liquidity. Overall, we find an increase in bid-ask spreads and a decrease in market depth at year-end, suggesting that tax-loss trading has an adverse effect on market liquidity. For firms with the largest tax-loss potential, however, we observe an increase in bid- depth but this increase is insufficient to satisfythe increased trading volume motivated bytax-loss selling. This causes a decline in relative liquidityfor these firms that amplifies the tax-loss selling effect. Further, because this insufficiencyis mostsevere for small firms, these firms experience the largest turn-of-the-year effect. These findings provide a microstructure explanation to Schultz’s (1985) question as to why small firms have a greater tax-loss effect than large firms. This paper also examines the presence of price pressure effects produced bytax- loss trading. We find that the non-information based selling at year-end temporarily depresses equilibrium equityprices even after controlling for bid-ask bounce. This finding suggests that stocks are not perfect substitutes for one another and consequentlythat the short-run demand curve for equityis not perfectlyelastic. Similar to Harris and Gurel (1986) and Shleifer (1986), we conclude that price pressure is present in the capital market. However, an investor cannot systematically earn meaningful economic returns after controlling for transaction costs from a strategyof buying tax-loss candidate at year-end and selling them at the beginning of the new year. Finally, this paper provides insights into the nature of trading behavior over the year-end period. Consistent with Roll’s (1983) conjecture, we find abnormal buying activityof capital-loss stocks byinstitutional investors reflecting an attempt to exploit temporaryprice depressions. Further, we find an increase in buying activity subsequent to the new year, which supports Ritter’s ‘‘parking the proceeds’’ hypothesis but the abnormal buying seems to be concentrated in stocks with capital gains