|
Article Excerpt
INTRODUCTION
previous studies show that the risk-adjusted abnormal returns on the ex-dividend dates are positive and significantly related to the dividend yield. However, the interpretation of these findings is controversial (see, e.g., Allen and Michaely (2003), Elton, Gruber, and Blake (2002), and Graham (2006) for a review). The question relates to what category of investors sets ex-day prices and to whether ex-day returns are driven by non-tax-related factors. Elton and Gruber (1970) argue that ex-day share prices are set in such a way that marginal long-term investors are indifferent between buying and/or selling before and after the ex-day. As a consequence, ex-day premium should reflect the differential taxation of dividends and capital gains of these long-term investors, and the magnitude of the drop in ex-day prices relative to dividends will indicate their identity. Other studies suggest that the behavior of share prices on the ex-dividend dates may not be totally related to tax effects. For example, Kalay (1982) shows that, in a world of certainty, investors not subject to the differential taxation of dividends and capital gains, referred to as short-term traders or tax-neutral arbitrageurs, will capture dividends, but since short-term trading is cosily, ex-day returns will be related to transaction costs rather than tax differential. Heath and Jarrow (1988) relax the assumption of risk neutrality and argue that since arbitrageurs are risk averse, their long and short positions cannot be unlimited, and, as a result, ex-day returns will reflect a compensation for their excess risk taking. (1) Bali and Hite (1998) and Frank and Jagannathan (1998) relate ex-day premium to market microstructure effects and suggest that ex-day premium can deviate from one for some stocks to reflect tick size and bid-ask bounce.
The empirical evidence is mixed. For example, Dai and Rydqvist (2007) use data from Norway where companies pay the imputation tax credit and cash dividends at different dates to find that the costly tax arbitrage is observed only in the case of cash dividend payment. Beggs and Skeels (2005) find that in Australia only the last tax reform affected ex-day returns. Moreover, while Bali and Hite (1998) argue that when stock prices are constrained to discrete multiples of 12.5 cents, ex-day prices should drop by the largest tick multiple less than the dividend, regardless of tax considerations, Graham, Michaely, and Roberts (2003), Jakob and Ma (2004) and Cloyd, Li and Weaver (2006) cast doubt on this model by showing that ex-day effects persist even after discreteness constraints were removed by the "decimalization" of U.S. stock markets in 2001.
The purpose of this paper is to contribute to this contending literature by examining the ex-day share price behavior across two countries with different tax and institutional frameworks. The analysis centers on Germany and the United Kingdom (U.K.), where the differences in ownership, trading practices, short-term trading rifles and income and capital gains systems allow for a deeper exploration of the conflicting determinants of ex-day returns. First, unlike Germany, ex-day returns in the U.K., where short-term trading is not allowed, are more likely to be driven by the differential taxation of dividends and capital gains or market microstructure factors. Second, although, both countries adopted during the sample period an imputation system whereby dividends carry tax credits and are taxed at lower rates than capital gains for basic taxpayers and tax-exempt investors, their underlying systems differ significantly. These differences are highlighted by modeling the tax systems in each country and by relating the tax differential between dividends and capital gains to ex-day prices. These theoretical models suggest that in the U.K., where the tax differential between dividends and capital gains is high, ex-day returns should be high, and the reforms that affected each country's respective tax differential between dividends and capital gains altered ex-day returns. During the sample period, three major tax changes in each country are identified, and the hypothesis that tax reforms that resulted in dividends to be taxed more heavily than capital gains will lead to an increase in ex-day returns is tested. These tax reforms offer a unique controlled experiment to test for the impact of taxation on the behavior of share prices on the ex-dividend days. Finally, the impact of the market microstructure effects is tested by analyzing the bid-ask spread around the ex-dividend dates and the magnitude of dividends relative to the tick size. These hypotheses are tested using 10,085 events in the London Stock Exchange and 2,910 in the Frankfurt Borse from 1988 to 2002.
Unlike previous single-country studies, the combination of the comparative analysis of ex-day returns across the two countries and the tax reforms, together with the direct analysis of short-term trading and microstructure effects, highlighted strong differences between the two countries, despite their tax system similarities. In particular, while the results show that the average ex-day returns are positive in both countries, they indicate that in the U.K., where the differential taxation between dividends and capital gains is relatively high, the average ex-day returns over the sample period of 0.75 percent are statistically higher than the 0.57 percent observed in Germany. In addition, ex-day returns are affected by the tax reforms, as, in the U.K., they increased monotonically from 0.38 percent (t = 1.75) to 1.17 percent (t = 5.57) in the post-1999 period when taxes on dividend increased following the elimination of the tax credit, and in Germany, they increased from 0.15 percent to 0.74 percent. The results based on the drop-off ratio and alternative event study methodologies are similar. Overall, these results provide support for the tax hypothesis.
Further analysis shows contrasting differences between the two countries. In particular, while in the U.K., ex-day returns are not affected by short-term trading, in Germany, this activity mitigates the tax impact. For example, unlike previous U.S. evidence, (e.g., Eades, Hess and Kim (1984)), the pre- and post-ex-day abnormal returns in the U.K. are not significant and the relationship between ex-day returns and the various measures of risk and transaction cost effects documented in previous studies is weak, as, unlike Naranjo, Nimalendran, and Ryngaert (2000), the relationship between ex-day returns and dividend yield, firm size and 1/[P.sub.cum], used as a measure of transaction costs, is strongly positive (not negative). These results are even stronger for high-yield stocks, which are likely to be subject to short-term trading (e.g., Naranjo et al. (2000)), implying that the ex-day returns are not subject to dividend capture and suggesting that the U.K. laws against dividend capture are effective and/or the high transaction charges involved in trading in the UK market, particularly the stamp duty, limit arbitrage transactions. In contrast, in Germany, ex-day returns are insignificant when the zero ex-day price change events (i.e., stale prices) are excluded, the pre- (post-) ex-day abnormal returns are positive (negative), the abnormal trading volume is positive (negative) for high(low-)yield stock, and, while the ex-day returns are not related to dividend yield, they are negatively related to the abnormal volume, suggesting that, in line with McDonald (2001), ex-day returns, of particularly high yield and low transaction cost stocks, are affected by short-term trading.
Finally, while in the U.K. the market microstructure effects documented in previous studies (e.g., Bali and Hite (1998)) do not influence strongly ex-day returns, their impact in Germany is significant. In the U.K., although there is some evidence of a positive effect of the abnormal trading volume on ex-day prices, this trading did not lead to an increase in the bid-ask spread; thus, it is not fully consistent with the proposition of Frank and Jagannathan (1998) that some investors dislike dividends and buy stocks on the ex-days and inflate ex-day prices. These results are consistent with the more recent U.S. evidence (e.g., Elton, Gruber, and Blake (2002), Jakob and Ma (2004) and Graham, Michaely, and Roberts (2003)). However, in Germany, the bid-ask spread increases significantly for the highest-yield groups and particularly in the last sub-period, suggesting that short-term traders mitigate the tax effects on ex-day pricing.
The results contribute to the literature of the impact of taxation on dividends in several different ways. First, following the Graham (2006) argument that it would be helpful if there were more research that exploits the rich variation in tax codes around the world, I provide a comparative analysis of the tax systems in Germany and the U.K., and expand prior country-specific studies by modeling the impact of recent tax changes. In Germany, while the overall results are consistent with McDonald (2001), I also show, using a relatively larger sample and more recent data, that the tax effects prevail after the 1997 period, but the results are not too strong as this impact is mitigated by the activities of short-term traders and do not apply when stock prices that did not change on the ex-day are excluded. In the U.K., a number of studies provide some mixed evidence on the impact of taxes on ex-day prices: Poterba and Summers (1984), Lasfer (1995) and Bell and Jenkinson (2002) show that the various tax reforms affected ex-day returns, while Asimakopoulos and Hodgkinson (2001) find that ex-day returns of large companies in the U.K. in 1988-97 are driven by market microstructure factors and not taxation, and Armitage, Hodgkinson, and Partington (2006) and Bond, Devereux, and Klemm (2007) show that the 1997 tax reform did not affect the ex-day drop-off. I provide further evidence of the tax effect, particularly in the post-1998 period. Moreover, I show that the tax reform impact is not limited to high dividend yield firms, suggesting that, in line with Bond et al. (2007), the results of Bell and Jenkinson (2002) that pension funds hold high-yield stock and ex-day share prices of these high-yield stocks were affected by the 1997 elimination of the tax credit is not likely to apply. The rest of the paper is organized as follows. The next section presents the institutional framework and the hypotheses to be tested. The third section discusses the data and the methodology. The fourth section discusses the empirical results. The conclusion is in the fifth section.
INSTITUTIONAL FRAMEWORK
This section builds on previous studies (e.g., Elton and Gruber (1970), Kalay (1982), and Lasfer (1995)) and relates ex-day returns to the tax differential of dividends and capital gains in Germany and the U.K. I provide details of the two tax systems and show that investors across these two countries face relatively different tax discrimination between dividends and capital gains. I model these effects and set up the hypotheses.
Taxes and Ex-day Returns
Elton and Gruber (1970) show that in a risk-neutral world with preferential treatment of capital gains and no restrictions on short-sales, shareholders are indifferent between buying or selling shares before the ex-day at [P.sub.cum] or after the ex-day at [P.sub.ex] if these two days' after-tax returns are equal, i.e.,
[1] [P.sub.cum] -([P.sub.cum] - [P.sub.0])z=[P.sub.ex] - ([P.sub.ex] - [P.sub.0])Z
+ D(1- m),
where [P.sub.0] is the original purchase price, m is the marginal tax rate on...
|
