The purpose of this paper is to evaluate the model used by Harris and Kemsley (J. Acc. Res. 37 (1999) 275), Harris et al. (J. Public Econ. 79 (2001) 569) and Collins and Kemsley (Acc. Rev. 75 (2000) 405), hereafter CHHK, and to investigate their empirical results. We demonstrate that the model underlying CHHK is flawed, and show that their interpretation of the data is incorrect. Finally, we find that after controlling for market to book ratio, Harris and Kemsley's first main result vanishes. In total, we reject CHHK's conclusions that equity prices are discounted for shareholder dividend taxes on retained earnings.
The purpose of this paper is to evaluate the model used by Harris and Kemsley (1999), Harris et al. (2001) and Collins and Kemsley (2000) and to investigate their empirical results. We refer to the papers as HK, HHK and CK, respectively, and collectively as CHHK. Using the Ohlson (1995) model, CHHK conclude that shareholder dividend taxes on corporate retained earnings (RE) are fully capitalized in stock price and dividend payout policy does not affect the amount of dividend taxes capitalized in price.1 In other words, these studies claim that shareholder dividend taxes on RE are capitalized in equity prices at the top individual investor statutory tax rate independent of the period of tax deferral.
There are several reasons CHHK's conclusions are considered “implausible” (Shackelford and Shevlin, 2001). First, CHHK's assertion that the effect of dividend taxes on equity values is independent of dividend payout policy is inconsistent with the positive value effects of tax deferral. It is also inconsistent with the fact that there are cheaper mechanisms than dividends to distribute corporate earnings to shareholders. Furthermore, CHHK's conclusion that shareholder dividend taxes are fully capitalized in equity values is inconsistent with the existence of tax clienteles. Similarly, there are non-tax reasons (signaling, agency costs) to expect dividends to play an important role in equity values. For all these reasons, we question the model underlying CHHK's empirical tests, their results and conclusions.
We begin by demonstrating that the Ohlson model (as applied by CHHK) is inappropriate to test for the effect of dividend taxes on equity prices. We also examine CHHK's results using their data and methodology because CHHK's results appear consistent with their predictions. Even though we demonstrate that CHHK's model is flawed, in the interest of future research, we also demonstrate that their results do not hold across tax regimes or in response to minor changes in empirical specification. The fact that the data does not support their predictions provides additional evidence that CHHK's model is flawed and that CHHK's implementation of the model and interpretation of the results is incorrect. We further argue that CHHK's main empirical tests suffer from an omitted correlated variable(s) problem and find that after controlling for market to book ratio, HK's first main result vanishes. In summary, we find that CHHK's model is flawed and that the data does not support their predictions. We therefore reject CHHK's conclusion that equity prices are discounted for shareholder dividend taxes on RE. Concurrent work by Hanlon et al. (2003), hereafter HMS, reaches a similar conclusion.
HMS reformulate HK's version of the Ohlson model and demonstrate mathematically that HK's empirical specification of the model is inappropriate. More specifically, HMS show that HK's ratio of RE to book value variable, which is a key component of HK's model, has no role in firm valuation. After correcting HK's empirical specification for this error, HMS show empirically that RE are not discounted in value relative to contributed capital, as a result of shareholder level dividend taxes, as HK claim.
Our study complements HMS in that we also conclude that CHHK's model is inappropriate to test the hypothesis that stock prices are discounted for shareholder dividend taxes on RE. We do not however deconstruct the Ohlson model. Rather, we note a variety of fundamental reasons that the Ohlson model is not appropriate to test CHHK's hypotheses. We then demonstrate that CHHK's implementation of the model is incorrect, as is their interpretation of the data.
The paper proceeds as follows. In the following section, we demonstrate that the model underlying the CHHK studies is flawed. In Section 3, we show that CHHK's interpretation of the data is erroneous and Section 4 concludes.
We critique the CHHK papers, which conclude that equity prices are discounted for shareholder dividend taxes on retained earning. Like HMS (2003), we demonstrate that CHHK's model is inappropriate for testing CHHK's predictions. We also present numerous data analyses that do not support CHHK's conclusions. In many cases, the results are opposite of CHHK's predictions as often as they support CHHK's predictions. Finally, we find that after controlling for market to book ratio, HK's first main result vanishes. Our empirical results indicate that CHHK's implementation of the model and interpretation of their empirical results is incorrect. In sum, we reject CHHK's conclusion that equity prices are discounted for shareholder dividend taxes on RE. Concurrent work by Hanlon et al. (2003) reaches a similar conclusion.
