مالیات گرفتن از زنان : تجزیه و تحلیل اقتصاد کلان

Based on well-known evidence on labor supply elasticities, several authors have concluded that women should be taxed at lower rates than men. We evaluate the quantitative implications and merits of this proposition. Relative to the current system of taxation, setting a proportional tax rate on married females equal to 4% (8%) increases output and married female labor force participation by about 3.9% (3.4%) and 6.9% (4.0%), respectively. Gender-based taxes improve welfare and are preferred by a majority of households. Nevertheless, welfare gains are higher when the U.S. tax system is replaced by a proportional, gender-neutral income tax.

Two observations are central to this paper. First, it is well known that the labor supply elasticities of women are larger than those of men, especially when the extensive margin is considered.1 Second, the current U.S. tax system is biased against women's work in the marketplace. Since the U.S. system taxes the income of households and not the income of individuals, for a married woman who considers entering the labor force, her marginal tax rate depends on her husband's income. Given the current levels of marginal tax rates, this is arguably a substantial impediment to labor force participation. These observations have motivated work in the theory of optimal taxation. From standard public-finance principles, the higher labor supply elasticities of women suggest that they should be taxed at lower rates than men. Boskin and Sheshinski (1983) were possibly the first to establish this insight. They focused on the optimal linear-income taxation of two-earner households with exogenously given differences in labor supply elasticities between men and women.2 More recently, Alesina et al. (2011) put forward more forcefully the idea of differential taxation of men and women within a model in which gender differences in labor supply elasticities emerge endogenously. Under parametric restrictions, they conclude that married women should be taxed at lower rates than married men.3 Although the above results are attractive for their policy implications, work in this area has been almost exclusively theoretical, and a quantitative evaluation of the relative merits of differential taxation by gender is still missing. It is an open question what are the expected, quantitative effects associated to changing the current structure of taxation in this direction. In this paper, we fill this void. We ask: What are the aggregate effects of taxing married females at lower rates? What are the welfare implications of these lower tax rates? To answer these questions, we use a model able to capture a number of key cross-sectional observations for the problems at hand. We build a life-cycle model populated by heterogeneous single and married agents. Individuals differ in terms of their labor endowments, which differ both initially and how they evolve over the life cycle. In particular, the labor market productivities of females are endogenous and depend on their labor market histories: not working is costly for females since if they do not work their human capital depreciates. Married households decide if both or only one member should work, in the presence or absence of (costly) children and the structure of the tax system. In this context, changes in the structure of taxation lead to changes in participation rates and aggregate labor supply, and can have large welfare consequences. We calibrate our model to the U.S. economy under the current tax system, taking into account the observed heterogeneity in skill endowments, marital segregation by skill, labor-force participation rates as well as the presence of children and their cost. As we explain in detail in Guner et al. (in press), the parameterized environment is capable of jointly reproducing a host of labor supply observations. The model is consistent with the wage-gender gap and its evolution over the life cycle, female labor force participation by educational attainment, and the pattern of participation rates by women with and without children as they age. This makes the model environment an ideal vehicle to evaluate the consequences of differential taxation by gender. Within the model disciplined by data, we then proceed to study the effects of a tax system that imposes different proportional taxes on the labor earnings of married females. Following Alesina et al. (2011), we will refer to these as gender-based taxes, albeit their particular implementation will be connected to marital status as we explain below. The gender-based taxes that we consider nest as special cases the equal tax rates on men and women. Hence, a virtue of our analysis is that it allows us to separate the effects of differential taxation of married females, from the effects associated to the elimination or reduction of tax progressivity. We consider two implementations of gender-based taxes. First, we consider replacing the U.S. tax system by proportional tax rates on labor earnings of married females that are lower than for the rest (married males, singles). We refer to this case as the broad-base case, as the reduction in tax rates on married females is financed by all other agents. In our second scenario, we first calculate a revenue-neutral proportional tax applied to all agents independent of their gender. We then assign this tax rate to singles, and reduce the tax rates on the labor earnings of married females increasing only the tax rates on married males. We refer to this case as the narrow-base case, as only tax rates on married males are used to achieve revenue neutrality. We find that a shift to proportional tax rates has substantial effects. Replacing the current tax structure by a proportional income tax at a 10.2% rate increases aggregate hours worked by 3.2% and aggregate output by 3.2% across steady states. As marginal tax rates are reduced for majority of households, married females increase their labor market participation by 2.8%. Taking into account changes in labor supply along the extensive as well as the intensive margin, the overall contribution of married females to changes in hours is substantial and amounts to 48.9%. The effects of proportional taxes outlined above are amplified when married females are taxed at lower rates. If taxes on married females are lowered to 4% (8%) in our narrow-base case, output increases by 4.0% (3.5%) and aggregate hours increase by 4.2% (3.6%) across steady states. These findings are driven by the much stronger responses of married females; they increase their participation by 6.9% (4.2%), and contribute 65.8% (56.1%) to the overall changes in hours. This is all not surprising, as tax rates are reduced on the group that reacts the most to tax changes. Similar results hold under our broad-base case. To assess welfare effects from our experiments, we compute transitions between steady states under the assumption of a small-open economy. We find that gender-based taxes lead to a welfare improvement to a majority of households alive at the date when the structure of taxes change. Nevertheless, we find that proportional income tax at a uniform rate dominates gender-based taxes in terms of aggregate welfare gains. While a proportional income tax on all delivers aggregate welfare gains of about 1.1% in consumption terms, a differential tax rate of 4% (8%) on married females implies gains of about 0.4% (0.7%). As we explain in Section 7.1, this result is driven by the effects associated to taxing married men at higher rates as in revenue-neutral tax reforms lower taxes on married females have to be financed by higher taxes on married males. While households where married women have a higher initial labor endowment tend to gain from the shift to gender-based taxes, most married households in our model are those where males have higher labor productivity. This is due to the observed marital sorting by skill, and initial wage gaps. Hence, the higher tax rates on males that accompany the lower rates on females have a net detrimental consequence on the welfare of most married households, and thus on aggregate welfare. These conclusions hold in a variety of robustness checks. Our paper is organized as follows. Section 2 presents a simple example that highlights the effects of differential tax rates on females on labor supply and participation decisions. 3 and 4 present the model economy. Sections 3 discusses calibration.4 In Section 6 we explain in detail the nature of the quantitative experiments that we conduct. Section 7 contains the main findings of the paper. Section 8 analyzes the sensitivity of our results via a number of robustness checks. Finally, Section 9 concludes. 1.1. Current U.S. taxes It is well known that the current U.S. tax system is biased against women's work.5 As we mentioned earlier, this bias originates from the fact that the U.S. tax system taxes the income of households, not the income of individuals. As a result, for a woman who considers entering the labor force, her marginal tax rate depends on her husbands' income. In addition, given the progressivity built in the system, the tax rate on her first dollar of income increases with the household's income (inframarginal income). In related work (Guner et al., 2011), we examine in detail the relationship between taxes effectively paid by households and their income in a large cross-sectional data from the U.S. Internal Revenue Service for the year 2000. Using this data, we estimate effective average tax rates. In Fig. 1, we present the average tax rates and corresponding marginal rates, for a married couple with two children in the year 2000.6 To illustrate the bias against women's work, imagine a married couple in which only the husband works and earns about the mean household income in the U.S. (about $58,375 in the 2000 IRS data). The average and marginal tax rates of this household are about 7.9% and 15.5%, respectively. Hence, the marginal tax rate that the household faces is 15.5% for woman's first hour of work. Together with payroll taxes and the additional child care expenses that the family might face, the combined reduction on the additional income that the female generates can be substantial, leading to disincentives for labor market participation. For higher income households, as Fig. 1 indicates, the disincentives can be much stronger. For a household at twice the level of mean income, the marginal tax rate is about 20.8%, whereas for a household at five times mean income, the marginal tax rate amounts to about 27.8%. Fig. 1 also shows average and marginal tax rates for a single household (head of household) with two children. These households face higher taxes than married ones. Still a married female would be less distorted in terms of her labor supply decisions if she was taxed as an individual at the singles' rates. Consider again a female whose husband earns about the mean household income. Suppose her earnings are about 0.6 times her husband's earnings. If she was taxed as an individual facing the tax schedule of singles, her marginal tax rate would be 13%, whereas if she files jointly with her husband her marginal rate would be 19.2%. Figure options Table 1 presents more detailed information about marginal tax rates faced by married households. The table shows marginal tax rates at different levels of household's income, that changes according to different hypothetical earnings for married female (secondary earner). Using our estimates, this is done for when she is about to enter the labor force, at low earnings (one- half mean income), or at higher earnings (mean income).As we note in Guner et al. (2011), the aforementioned marginal tax rates are lower bounds on the marginal rates faced by married households. This follows from the fact that the marginal tax rates reported are calculated from average tax rates, and taken into account all the inframarginal deductions that households have access to. Effective marginal tax rates are good approximations at low levels of income. At high levels of income, reported marginal tax rates are non-trivially higher than effective marginal rates.7 More broadly, international evidence suggests that differences in taxation might indeed matter for cross-country differences in female labor force participation. Bick and Fuchs-Schuendeln (2011) provide a detailed account of how household incomes are taxed in different OECD countries (both in terms of the unit of taxation and the tax burden on the secondary earners). They study a model of household labor supply and show that differences in taxes can account for a large part of cross-country differences in married female labor supply. Fig. 2, which is based on their analysis, shows the relation between tax burden on secondary earner and married female labor supply, where the tax burden is measured as the ratio of the tax liabilities of a two-earner household to a one-earner one. It is clear that higher taxes on secondary earners are associated with lower female labor force participation. It is also worth noting that low labor force participation of countries like the U.S., Germany, France and Portugal are countries that tax jointly household income.8

A central result from this paper is that, on a measure of aggregate welfare, a shift to gender-based taxes delivers welfare gains, and that a majority of households would support such a change. Nevertheless, these gender-based taxes are dominated by the replacement of the current structure of taxes by a uniform, proportional tax system on all households. Put differently, we found mixed support for gender-based proportional taxes in our model economy. It is worth emphasizing at this point that a central concern in the current paper is the detailed consideration of the female labor supply decision, in order to capture the heterogeneity observed in the data. In doing so, we admittedly have abstracted from some factors that may lead to the optimality of differential taxation by gender, as considered by Alesina et al. (2011). Our results highlight one reason why lower taxes on married females might not improve welfare relative to a simple proportional tax: lower taxes on females have to be financed by higher taxes on married males and taxing high earners in married couples at higher rates can be costly. Since our welfare results stand in contrast with results on the optimality of gender-based taxes, we conclude by relating our model with the model in the aforementioned paper. In both papers, the elasticity of female labor supply is endogenous; in Alesina et al. (2011) it is driven by comparative advantage in home production and career investments, whereas in our model is affected by the participation decision of married females. There are income effects in labor supply in our model, while in their paper, home and market consumption goods enter linearly in preferences. Their model is effectively a static setup, amenable for theoretical analysis, while ours incorporates life-cycle behavior and capital accumulation. A central difference between Alesina et al. (2011) and our paper relates to marriage and the modeling of household decisions. All individuals are married in equilibrium in Alesina et al. (2011), while we explicitly consider married and single people. In particular, we assume that (i) marital status and marital sorting is exogenous to the model, and unlike Alesina et al. (2011), (ii) there is no bargaining affecting household decisions as there is nothing to disagree on. Endogenous marriage coupled with bargaining over the gains from marriage would clearly affect the identity of winners and losers from the shift to gender-based taxes and therefore, the scope and magnitude of welfare gains. Gender-based taxes can also affect incentives to acquire education, which in our model is exogenously given to individuals at the start of the life cycle. Future research should determine whether these features are important enough to overcome our welfare findings.