مشوق های مالیاتی و تقاضا برای بیمه عمر: شواهدی از ایتالیا

The theoretical literature suggests that taxation can have a large impact on household portfolio selection and allocation. In this paper we analyze the tax treatment of life insurance, considering the cancellation of tax incentives in Italian life insurance contracts for investors with high marginal tax rates and the introduction of incentives for those with low rates. Using repeated cross-sectional data from 1989 to 1998, we find that the tax reforms had no effect on the decision to invest in life insurance or the amount invested. The likely explanations are the lack of information and lack of commitment to long-term investment.

The theoretical literature suggests that taxation has a potentially large impact on household portfolio selection and allocation. The theory has two central insights: that what matters for investors is the after-tax return on each asset, and that the differing fiscal treatment of the various assets creates wedges in the structure of those returns. In this paper we bring fresh evidence to this literature by studying the portfolio effect of changes in the tax treatment of life insurance, using repeated cross-sectional data. The change that we consider is the cancellation of tax incentives in Italian life insurance contracts for investors with high marginal tax rates and the introduction of incentives for those with low marginal tax rates. Our sample, a decade of microeconomic data on household assets, income and demographic variables, provides a truly unique setting for spotlighting the effect of taxes on household portfolio selection and allocation. A study of this kind raises crucial identification issues. Theory predicts that portfolio choice is affected by household resources and by after-tax interest yields. However, as the after-tax yield on some assets depends on the taxpayer’s marginal income tax rate, which is inherently correlated with the level of income, it is difficult to disentangle genuine variation in after-tax interest rates, for given income, from genuine variations in income, for given after-tax interest yields. For some assets this is actually impossible, because at any point in time all investors face the same rate of return. Despite the identification problems, some empirical studies do document the existence of a link between marginal tax rates and portfolio choice. In general, applied work in this area has estimated the tax rate elasticity of participation in tax-sheltered assets and their portfolio shares controlling for household income, wealth and other demographic variables. The most recent study is Poterba and Samwick (1999), who build on the seminal contributions of Feldstein (1976) and King and Leape (1998). Poterba and Samwick impute marginal tax rates in the Survey of Consumer Finances and estimate probit models for eight broad asset categories. Their results support the view that taxes affect asset selection. Controlling for income and wealth, they find that the probability of individuals’ investing in tax-deferred accounts, equity and bonds is a positive function of the marginal tax rate. To the best of our knowledge, outside the United States the evidence on the role of taxes in shaping household portfolios is limited to the Netherlands (Alessie et al., 1997), Sweden (Agell and Edlin, 1991) and the United Kingdom (Banks and Tanner, 2001). Poterba (2001) reviews these empirical studies and concludes that investors take the tax treatment into account when selecting their asset menu. In all countries the evidence of a link between taxes and portfolio shares is weaker than for asset selection. Identifying the tax effects on portfolio choice is hard in the Italian case as well. Pre-interest income on bank deposits, government bonds, corporate bonds and mutual funds is subject to a flat rate withholding tax in settlement of the tax liability, so the after-tax yields on these assets are identical for all investors. The return on stocks, on the other hand, depends on the marginal tax rate and therefore does display cross-sectional variability. In order to identify the effect of taxes on portfolio choice, however, one needs not only cross-sectional variability in returns but also genuine variation in after-tax yields that is not perfectly correlated with the general income tax rate. Life insurance contracts provide a good opportunity to test the theory of portfolio taxation in Italy. From 1986 to 1992 life insurance premiums were fully tax-deductible up to 1300 euro (2.5 million lire), so that after-tax returns increased with the marginal tax rate. In 1992 the tax incentive for those with marginal tax rates of 27 percent and higher became a flat rate of 27 percent (the previous regime still applied to low-income taxpayers). In 1994 any link with the marginal tax rates was eliminated, and the incentive made proportional to premium payment (a flat rate of 22 percent, lowered to 19 percent in 1998). As a consequence, the extra yield on life insurance policies over non-sheltered financial assets was substantially reduced for the rich in 1992, while in 1994 it was slightly increased for the poor, slightly reduced for the rich, and unchanged for investors in the intermediate tax brackets. These group-specific tax changes provide the ground for our empirical analysis and for the identification of the effect of the tax reform. If the tax reform had an impact on asset selection and portfolio allocation, it should emerge among households in the highest tax brackets in 1992, and among both the lowest and the highest tax brackets after the 1994 reform. The paper is organized as follows. Section 2 explains how the 1992–94 reforms affected the after-tax return on life insurance. Section 3 presents the data used in the empirical analysis, drawn from the 1989–98 Survey of Household Income and Wealth. Section 4 presents the empirical analysis. There turns out to be no detectable effect of the tax reforms on the demand for life insurance as such or on the amount invested. Section 5 suggests various explanations for these findings, such as the role of information, lack of commitment and minimum investment requirements. Section 6 concludes.

6. Conclusions Before 1992 premiums on life insurance retirement plans were fully tax deductible up to an amount of 1300 euro, so that the after-tax return was proportional to the marginal tax rate. In 1992–94 the Italian government implemented a series of reforms whose ultimate effect was to cancel the relation between the after-tax return and the marginal tax rate. In the new regime the tax incentive was proportional to the contribution regardless of the marginal tax rate. In this paper we study the impact of the reform on the decision to purchase life insurance and on the amount invested. Portfolio taxation theory suggests that investors with high tax rates should have reduced their demand for life insurance plans and those with low rates should have increased it. We find no evidence for the theory. Among low-income taxpayers, the incidence of life insurance participation and the amount invested do not change after the reform relative to high-income taxpayers. What we find is an across-the-board increase in life insurance participation in the last decade, not tilted towards low-income investors. This result bears on the substantial literature on the effect of targeted saving incentives, such as those for IRAs and 401(k). Our reading of this literature, as summarized by Poterba et al. (1996) and Engen et al. (1996), is that there is broad consensus that in the United States targeted saving incentives have induced portfolio shifts towards tax-favored assets. Contrary to the US evidence, our study of the Italian data finds little support for the hypothesis that investors respond to tax changes by adjusting their portfolio at either the intensive or extensive margins. The most likely explanations for our findings are reluctance to commit to long-term saving, anticipation of future liquidity constraints, minimum investment requirements, and lack of knowledge of the tax incentives. In this respect, the paper suggests that information and transaction costs are of paramount importance in shaping portfolio selection and allocation.