پرداخت مالیات و تخصیص سرمایه گذاری
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|5196||2013||24 صفحه PDF||سفارش دهید||20640 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of Financial Economics, Volume 107, Issue 1, January 2013, Pages 1–24
When corporate payout is taxed, internal equity (retained earnings) is cheaper than external equity (share issues). If there are no perfect substitutes for equity finance, payout taxes may therefore have an effect on the investment of firms. High taxes will favor investment by firms who can finance internally. Using an international panel with many changes in payout taxes, we show that this prediction holds well. Payout taxes have a large impact on the dynamics of corporate investment and growth. Investment is “locked in” in profitable firms when payout is heavily taxed. Thus, apart from any level effects, payout taxes change the allocation of capital.
Corporate payout, in the form of dividends or as repurchases of shares, is subject to taxation in most countries. Such taxes on corporate payout drive a wedge between the cost of internal and external equity (retained earnings and equity issues, respectively). Higher payout taxes raise the cost of capital for firms using external equity to fund investment relative to those using internal equity. Therefore, higher payout taxes are expected to “lock in” investment in profitable firms, at the expense of firms with investment opportunities which would require external equity financing to undertake. The empirical relevance of this simple prediction has not been well tested. Despite the large amount of theoretical and empirical research about the effect of dividend taxes on the level of investment and on the valuation of firms (see, e.g., Auerbach, 1979a, Auerbach, 1979b, Bradford, 1981, Chetty and Saez, 2010, Feldstein, 1970, Guenther and Sansing, 2006, Harberger, 1962, King, 1977, Korinek and Stiglitz, 2009, Poterba and Summers, 1984 and Poterba and Summers, 1985), little is known about the effects of such taxes on the allocation of investment across firms. Yet, the theoretical prediction is very clear: higher payout taxes will increase the wedge between the cost of internal and external equity, and firms with more costly external financing will exhibit greater investment cash flow sensitivities. Put differently, payout taxes favor investment financed by retained earnings over investment financed by equity issues. This can matter for the productivity and nature of investment if (a) debt finance is an imperfect substitute for equity (in other words, if the Miller Modigliani propositions do not hold), (b) different firms have different investment opportunities, (c) the marginal investor is subject to taxation, and (d) firms make equity payouts while the tax is in effect. All these conditions have empirical support.1 But are such frictions important enough for this to matter in practice for investment levels? This paper aims to test the extent to which the “lock in” effect of payout taxes matters empirically. There are several challenges in testing how payout taxes affect the cross-firm allocation of investment. First, large changes in the US tax code are rare. The 2003 tax cut has provided a suitable natural experiment for testing how dividend levels responded to taxes (see Chetty and Saez, 2005 and Brown et al., 2007), but investment is a more challenging dependent variable than dividends, so the experiment may not provide sufficient statistical power for examining investment responses. First, unlike dividends, investment is imperfectly measured by accounting data that, for example, leave out many types of intangible investment such as that in brands and human capital. This means that available empirical proxies (e.g., capital expenditures) are noisy estimates of the true variable of interest. Second, much investment is lumpy and takes time to build, so any response to tax changes is likely slow and more difficult to pinpoint in time. This suggests that a longer time window may be necessary (the payout studies used quarters around the tax change). Third, however, investment is affected by business cycles and other macroeconomic trends, so extending the window around a single policy change introduces more noise from other sources, and may not provide better identification. We address these challenges by using an international dividend and capital gains tax data set covering 25 countries over the 19-year period 1990–2008 (Jacob and Jacob, 2012). This data set contains 15 substantial tax reforms and 67 discrete changes in the dividend or capital gains tax rate. With so many tax changes, we have sufficient variation to study the effects of payout taxes on the investment allocation. We use this tax database to test if the allocation of investment across firms with and without access to internal equity depends on payout taxes. We first run nonparametric (NP) tests that contrast the investment by the two groups of firms around tax reforms. We focus on events where payout taxes changed by at least three percentage points and compare the five years preceding the tax change with the two years following it. There are 15 events with payout tax reductions. The mean tax drop is 9.8 percentage points (median 5.5). There are 14 tax increase events with a tax change of 8.4 percentage points (median 5.6). We sort firms into quintiles of the ratio of cash flow to assets in each country-year cell. We then calculate average investment over lagged assets for each quintile. There is no trend in investment for any of the quintiles during the five-year period preceding the tax events. After the tax cuts, we observe a significant convergence of the investment rate of high and low cash flow firms (top and bottom quintiles). In other words, firms with limited internal equity increase their investment relative to firms with plenty of internal equity. This is consistent with the tax wedge theory, and suggests that low taxes favor firms with limited access to internal equity. In contrast, following increases in payout taxes, there is a divergence of investment of high and low cash flow firms. The estimated effects appear large in both sets of tax reforms. On average, the difference in investment between low and high cash flow firms increases from 5.33% (of assets) to 7.59% following a payout tax increase—a 42% increase. When payout taxes are cut, the difference in investment falls from 7.27% to 5.54%—a decrease by 31%. In other words, for the typical large tax change, a large quantity of investment is estimated to get displaced. When taxes go up, investment flows from firms with limited access to internal equity to those with more internal equity, and vice versa for tax reductions. These nonparametric results are consistent with the predictions of the tax wedge theory. Because the panel data set contains multiple tax change events, we can estimate not just the mean treatment effect of a tax change, but also ranges. Only two (three) of the 15 (14) tax decreases (increases) have difference-in-difference effects that are in conflict with our hypothesis. The other estimates agree with the tax wedge hypothesis, and many point estimates are large: one-third of tax decrease events reduce the difference in the investment rate of high and low cash flow firms by at least 2.5 percentage points. About 40% of the tax raises are associated with a point estimate for the increased wedge between high and low cash flow firms by more than 2.5 percentage points.2 In other words, the effect of tax changes on the relative investment of firms varies quite a bit across events, and is sometimes large. We can also use the individual difference-in-difference point estimates to do nonparametric tests. For example, a sign test of the frequencies with which estimates are positive and negative suggests that we can reject that an increase and a decrease of the investment rate difference are equally likely after a tax increase (decrease) at the 5% (1%) level of statistical significance. We also use linear regressions to produce parametric estimates of the effect of taxes on relative investment of rich and poor firms. Unlike the nonparametric tests, the regressions use data from all years, and can integrate both tax increases and decreases in the same specifications. The two methods also put different weight on observations (equal-weighting tax changes vs. equal-weighting firm-years). For our baseline tests, we regress investment on firm controls, fixed effects for firms and for country-year cells, and the interaction of the payout tax rate with cash flow. Thanks to the panel structure of the data set, we can allow the coefficient on cash flow to vary across countries and years, in essence replicating the identification strategy of the many studies exploiting the 2003 tax cut in the US, but for the whole panel of 25 countries times 19 years. The estimated coefficient for the cash flow⁎tax interaction variable is consistently positive and significant. In other words, the higher payout taxes are, the stronger is the tendency for investment to occur where cash flows are high. As predicted by the tax wedge theory, payout taxes “lock in” investment in firms generating earnings and cash flow. The estimated magnitudes are large. For example, going from the 25th percentile of payout tax (15.0%) to the 75th percentile (32.2%) implies that the effective coefficient on cash flow increases by 0.029, an increase by 33% over the conditional estimate at the 25th percentile. Like the NP results, this implies that payout taxes have an important effect on the allocation of capital across firms. We report extensive robustness tests for these results. For most tests, we report regression results with three alternative tax rates, with similar results. The results also hold for alternative measures of the ability to finance out of internal resources (e.g., net income instead of cash flow), as well as when controlling for the corporate income tax rate and its interaction with cash flow. We also collect economic policy controls from the World Development Indicators. This is to address endogeneity concerns, i.e., to ensure that tax changes are not just fragments of wider structural changes in an economy that change firms' investment behavior around tax reforms. This test shows that payout tax changes appear to have their own very unique and economically significant effect on the allocation of investment (assuming we have identified the relevant set of policy variables). We further ensure robustness to alternative regression frameworks and specification strategies. We also ensure robustness of our model predictions to cash investments, and we address concerns about the potential impact of measurement error in investment opportunities. We also examine cross-sectional differences in the response to taxes. We attempt to identify firms whose marginal source of funding is likely to be external equity based on three measures. First, we consider predicted equity sales. Second, we look at historical equity issuance. We exploit the fact that such issuance is persistent, so that classifying firms by recent equity issuance likely indicates their ability to issue in the future.3 Third, we classify firms within the two lowest quintiles of the age and the size distribution in one country-year as more likely to use external equity (Hadlock and Pierce, 2010). For all three classifications, there is a sizable difference in the effect of taxation on the marginal source of funds for investment between external and internal equity firms. The cash flow coefficient is sensitive to tax rates only for firms that are more likely to use external equity. This confirms the mechanism behind the differential responses of investment to tax rates that we have documented earlier. Finally, we examine how quantities of equity raised respond to taxes. If our identifying assumptions are valid, and if we have identified real variation in the effective taxation as perceived by firms, we would expect to see a drop in equity issuance when taxes go up. We find exactly this: When taxes are high, equity issuance tends to be low. This supports the interpretation that the tax variation we pick up is meaningful. Our results are related to the debate about the impact of payout taxes on the level of investment between the “old view” (Harberger, 1962, Harberger, 1966, Feldstein, 1970 and Poterba and Summers, 1985) and the “new view” (Auerbach, 1979a, Bradford, 1981 and King, 1977). This debate can be understood in terms of different assumptions about the marginal source of investment financing. To simplify, the old view assumes that marginal investment is financed by equity issues, so that payout taxes raise the cost of capital and reduce investment. The new view assumes that marginal investment is financed by retained earnings, so that payout taxes do not reduce investment. In practice, firms are likely to differ in their ability to finance investment with internal resources (e.g., Lamont, 1997). If they do, the tax rate will affect the allocation of investment. Our results have three main implications. First, it appears that payout taxes influence the allocation of capital across firms. High taxes lock in capital in those firms that generate internal cash flows, ahead of those firms that need to raise outside equity. If firms have different investment opportunities, this means that tax rate changes alter the type of investments being made. For example, high payout taxes may favor established industries. We consider the allocation across firms an important topic in itself, but there may also be some suggestive implications for aggregate investment. Our results generally point to the relevance of payout taxes for investment (for a subset of firms). Second, the effect of payout taxes is related to both access to the equity market and governance. Firms that rely on access to equity markets as a source of finance, “old view” firms, are the most affected by tax changes. Firms whose only source of equity finance is internal are less affected by taxes, as predicted by the “new view.” A final source of heterogeneity is governance. Firms where decision makers have low financial stakes are less affected by tax changes, reflecting their propensity to make investment decisions for reasons unrelated to the cost of capital. Third, the relation between cash flow and investment (see, e.g., Fazzari et al., 1988 and Kaplan and Zingales, 1997) appears to partially reflect the difference in the after-tax cost of capital between firms with and without access to inside equity.
نتیجه گیری انگلیسی
Our results have three main implications. First, it appears that payout taxes drive the allocation of capital across firms. High taxes lock in capital in those firms that generate internal cash flows, ahead of those firms that need to raise outside equity. The cost of capital, especially for firms with much equity (low leverage), could be very sensitive to tax rates on payout for some firms, as well as heavily dependent on access to internal equity when payout taxes are high. We estimate economically large effects of payout taxes. This corresponds to recent trends in the amounts raised by various taxes. From 1960 to 2009, the share of corporate income taxes in aggregate US Federal tax receipts fell from 24% to 10% (Internal Revenue Service, 2009). A study by the Department of the Treasury, Office of Tax Analysis suggested that individual income taxes on dividends were 13% of Federal tax receipts in 2005. In other words, payout-related taxes may currently raise more revenue than corporate income taxes in the US. Second, the impact on investment allocation makes payout taxes an important policy tool. If firms have different investment opportunities, our results mean that tax rates change the type of investments being made. Tax policy offers a tool for affecting the access to investment resources by firms without retained earnings. For example, high payout taxes may favor established industries while low taxes favor new industries. Third, the well-known relation between cash flow and investment (see, e.g., Fazzari et al., 1988, Kaplan and Zingales, 1997 and Lamont, 1997) may partially reflect the difference in the cost of capital between firms with and without access to inside equity. Firms invest more if they have easy access to more resources (see, e.g., Lamont, 1997 and Rauh, 2006), especially internal cash flows. Our results point to a tax channel generating (a part of) this effect: having internal cash flows implies a lower after-tax cost of equity capital.