سیاست تقسیم سود، سیگنالینگ، و تخفیف وجوه بسته پایانی

We test the predictions of dividend signaling models using closed-end equity funds that adopt explicit policies committing them to pay minimum dividend yields. These policies represent deliberate attempts to reduce share price undervaluation relative to NAV. Funds that adopt minimum dividend policies experience reductions in their share price discounts, trade at smaller discounts than other funds, earn greater excess returns following policy adoption, and their managers survive longer than other managers do. The results are broadly consistent with the predictions of dividend signaling models, and suggest that high quality closed-end funds can reduce undervaluation via dividend policy.

We test the predictions of dividend signaling models using a sample of closed-end equity funds. Closed-end equity funds are well suited for this task because many of them adopt explicit minimum dividend policies that commit them to pay dividends of at least a specified percentage of net asset value at periodic intervals. If a fund adopts a minimum dividend policy and generates a portfolio return below the minimum dividend yield, the policy causes a partial self-liquidation of the fund. Fund manager compensation is usually specified as a percentage of total net assets so a partial liquidation reduces compensation, and repeated partial liquidations would eventually drive compensation to zero as the fund liquidates completely. In contrast, a fund manager who generates a portfolio return greater than the promised minimum dividend yield maintains or increases the level of total net assets and in turn his compensation. Thus, a fund manager who generates high future portfolio returns would find the adoption of a minimum dividend policy less costly than would a fund manager who generates low future returns. Given the differing costs of minimum dividend policies across fund managers that generate low versus high future portfolio returns, we argue that the policies have potential signaling value. Managers of high discount closed-end funds risk losing their jobs via hostile takeovers, liquidations, or investor pressure. To the extent that closed-end fund discounts reflect (at least in part) expectations of inferior future portfolio returns, managers who anticipate superior future portfolio returns have incentives to signal their type to reduce their funds’ discounts. Indeed, press releases indicate that funds adopt minimum dividend policies as deliberate attempts to reduce the undervaluation of the funds’ market share prices from net asset value (NAV).1 Thus, these closed-end equity funds allow for tests of the predictions of dividend signaling models on a focused sample of funds that seek to explicitly and deliberately signal undervaluation. In addition to being able to focus on those funds that deliberately try to signal undervaluation, there are two other advantages to using closed-end equity funds to test dividend signaling. First, the transparency of closed-end funds makes it easy to compare the relative levels of undervaluation (i.e., the discount of market share price from NAV) both before and after the adoption of a minimum dividend policy, and across funds with and without such policies. It is much more difficult to measure undervaluation for traditional corporate firms. Second, it is straightforward to observe and measure changes in the return performance of these funds’ investment portfolios surrounding the adoption of a minimum dividend policy. If minimum dividend policies have signaling value and if closed-end fund discounts reflect expectations of inferior future portfolio returns, then we should observe smaller discounts for funds that adopt these policies. We find that investors distinguish between funds that adopt what we term strong (minimum dividend yield of at least 10%) and weak (minimum dividend yield below 10%) minimum dividend policies. Closed-end funds that adopt strong policies experience statistically and economically significant reductions in their share price discounts from NAV. 2 We also find that funds with strong policies have a mean discount that is approximately 58% smaller than that of funds without minimum dividend policies; the difference holds when we control for various factors that could be related to minimum dividend policies and that previous studies identify as determinants of discounts. In contrast, the mean discount for funds with weak minimum dividend policies does not differ significantly from the mean discount for funds without minimum dividend policies. Consistent with the prediction that minimum dividend policies signal high future portfolio returns, funds that exhibit greater reductions in their discounts upon policy adoption earn larger post-adoption excess NAV returns. The post-adoption excess NAV returns are significantly positive and are significantly greater than the returns of matched funds. Consistent with the conjecture that managers of high quality funds have incentives to signal, we find that funds and managers with minimum dividend policies survive significantly longer than funds and managers without such policies. Moreover, absent the minimum payout policy, investors would have had difficulty distinguishing high from low quality managers because their funds’ excess returns are indistinguishable from each other in the pre-adoption periods. Our results are broadly consistent with the general predictions of dividend signaling models such as Bhattacharya (1979), John and Williams (1985), and Miller and Rock (1985). However, the costs of signaling differ in our analysis, thus we cannot test various models’ specific predictions about the sources of signaling costs. Our results also provide indirect evidence in support of Ross's (1977) signaling model, in which a financial policy choice acts as a signal because it produces different payoffs under the manager's compensation contract. Our results do not imply that signaling is the only reason that funds or firms pay dividends. Indeed, only 20% of our sample funds explicitly commit to minimum dividends. Interestingly, 20% is close to the proportion of firms in Brav et al., (2005) survey that agree or strongly agree that signaling factors affect their dividend policies.3 We interpret both our evidence and the evidence in Brav, Graham, Harvey, and Michaely as implying that in aggregate, firms pay dividends for a variety of reasons, with only a minority of firms using dividends to signal. If indeed only a minority of dividend-paying firms use dividends to signal, this could explain why prior empirical studies on broad cross-sections of firms generally find mixed or negative results for signaling theories (see Allen and Michaely, 2003, for an excellent review of these studies). Although it is clear that minimum dividend policies convey information to investors, we cannot determine empirically the exact nature of that information. In a traditional dividend signaling framework, the manager who signals is privately informed that the distribution of future payoffs, and thus firm value, are higher than investors think. In this case, minimum dividend policies convey private information about extant managerial ability and expectations of relatively high future returns. Alternatively, minimum dividend policies potentially improve incentives to generate high returns because if returns are sufficiently low, the funds self liquidate, and managers’ compensation falls. In this case, minimum dividend policies convey that the distribution of future payoffs should improve because of the improved managerial incentives. Note that these two possibilities are not mutually exclusive. Unfortunately, the discount reductions, superior post-policy NAV return performance, and greater survival rates that we find for funds with minimum dividend policies are consistent with both possibilities about the nature of the information. Funds claim that they adopt minimum payout policies to signal undervaluation. While our evidence is consistent with that claim, we cannot rule out the possibility that the results we find stem from improved contracting that derives from the policies. Consistent with an improved contracting view, prior studies of closed-end funds find that agency costs and contracting issues affect fund discounts and performance (see Barclay et al., 1993; Chay and Trzcinka, 1999; Coles et al., 2000; Khorana et al., 2002; Del Guericio et al., 2002). Gemmill and Thomas (2002), Ross (2002), and Cherkes (2003) develop nonsignaling-based theoretical arguments that predict closed-end fund discounts should be negatively related to dividends. These arguments potentially represent alternative explanations of our findings. However, the correlation between discounts in our sample and the predicted discounts from their models is only 0.15. Wang (2003) also finds that minimum dividend policies reduce fund discounts, but attributes the effect to reducing investor sentiment exposure. None of the alternative explanations in these theoretical models predicts the superior NAV return performance following the adoption of minimum dividend policies, the greater survival rates for funds with minimum dividend policies, or the significant negative relation between discount changes and post-adoption return performance that we find, all of which support a signaling view of minimum dividend policies. We also present evidence that the relation between discounts and minimum dividend policies is not driven by arbitrage cost effects (see Pontiff, 1996). The next section of the paper explains the link between minimum dividend policies and fund manager compensation, and thus establishes differences in the costs of such policies for fund managers who anticipate low versus high future portfolio returns. Section 3 discusses the data and methods we use, and Section 4 contains all of the empirical results. We conclude in Section 5.

One-fifth of the closed-end equity funds in our sample adopt policies that commit them to pay dividends greater than or equal to a specified percentage of net assets. Funds adopt the policies explicitly and deliberately to reduce undervaluation, which is easily observed in their discounts, so the funds are well suited to test the predictions of dividend signaling models. Funds that adopt strong policies (i.e., minimum dividend yield of 10% or greater) experience significant discount reductions and have smaller discounts than do funds without such policies. The effect is large relative to other factors that affect closed-end fund discounts—the mean discount for funds with strong policies is approximately 58% smaller than the mean discount for funds without minimum dividend policies. Funds with minimum dividend policies also earn risk-adjusted NAV returns that are significantly positive and significantly greater than returns of matched funds following policy adoption. The relation between post-policy excess returns and the policy announcement-period discount reductions implies that the revaluations by investors reflect rational expectations of superior future performance. Funds and fund managers with minimum dividend policies have significantly higher survival rates and NAV return performance that is indistinguishable from other funds in the periods before policy adoption, which suggests that fund managers had incentives to signal. The explicit and deliberate signaling by funds that adopt minimum dividend policies and the resulting effects that we document are consistent with the broad predictions of dividend signaling models although the signaling costs are different in our analysis. Moreover, the superior future performance following the adoption of minimum dividend policies makes it unlikely that the value increases (discount reductions) we observe stem from disciplinary factors, e.g., a free cash flow disgorgement view of dividends as in Jensen (1986). Shareholders should not benefit by forcing portfolio managers to disgorge cash when their portfolios earn positive excess returns. The magnitude of the difference in discounts, 58% lower for funds with strong minimum dividend policies, also seems too large to suggest that the discount reductions are driven by reductions in investors’ transaction costs. We cannot observe effective marginal tax rates for the relevant investors, so we cannot draw unambiguous conclusions about tax-related and dividend-clientele hypotheses. Although funds claim that they adopt the minimum dividend policies to signal undervaluation, it is also possible that the effects that we document stem from improved incentives brought about by the minimum dividend policies. Because only one-fifth of our sample funds adopt minimum dividend policies, we cannot conclude that signaling is the sole reason funds pay dividends. With closed-end funds, it is easy to understand why the other funds pay dividends—positive net income not paid in dividends is taxed at the fund level. Although traditional corporate firms do not face the same restrictions, it is interesting that the proportion of funds in our sample that adopt minimum dividend policies is close to the proportion of traditional firms in Brav et al., (2005) survey who agree or strongly agree that they use dividend policy to make them look better than other firms. Consistent with our evidence, Brav, Graham, Harvey, and Michaely's results imply that some firms use dividends to signal, while others pay dividends for other reasons. If in aggregate firms pay dividends for a variety of reasons but only a minority use dividends to signal, this could explain why prior empirical studies on broad cross-sections of firms generally find mixed or negative results for signaling theories. Our research highlights the value of testing predictions on a more focused sample when multiple motivations potentially exist for a particular financial policy choice.