Currently, about half of the U.S. labor force is comprised of individuals who are married. And nearly 97 percent of married individuals file their taxes jointly, meaning that the married couple is the so-called “tax unit.” For spouses who earn significantly different amounts of income, one of the benefits of filing jointly is simple: in a progressive system, tax can be saved by income-averaging. Rather than facing two separate marginal tax rates, as would be the case with individual filing, filing jointly means that both spouses face the same marginal rate on their aggregate income. Because the married-filing-jointly brackets are approximately double the individual brackets, the higher-earner’s rate will be lower than her counterfactual individual rate, increasing her return to working. And the lower-earner’s rate will be higher than her counterfactual individual rate, reducing her return to working.
These labor-supply incentives are well-known. What is less-explored, however, is the extent to which the combination of tax unit rules and rate schedule progressivity affects the spouses’ labor supply decisions. A look at Scandinavian countries with highly progressive tax rates presents an intriguing puzzle. One might predict that high tax rates discourage productivity. So how have these countries achieved world-leading employment rates and per-capita output levels? Building on the work of scholars including Borella et al. (2023) (working paper version previously reviewed in Jotwell), a new working paper by Hans A. Holter, Dirk Krueger and Serhiy Stepanchuk (“HKS”) argues that the answer, at least in part, may be hiding in plain sight: the tax unit.
The paper makes three main contributions. First, HKS use cross-country panel data to show that higher tax progressivity combined with higher levels of individual-tax-unit filing are associated with higher employment for both married men and married women. (Pp. 1, 8.) They do this by presenting a model of the tax system in which there are two key parameters: first, the progressivity of the system and, second, its degree of jointness, or the extent to which the marginal tax rate of one spouse depends on the income of the other spouse. (P. 6.) For this latter parameter, there are four options: pure jointness (parameter takes value of zero, one imagines the U.S.’s married-filing-jointly tax unit system would have a value of zero and it does score very close to that (P. 7); no jointness (parameter takes value of one; such as in a pure individual tax unit filing system), positive jointness (parameter takes values between but not including zero and one; this means that the marginal tax rate on one spouse depends positively on the income of the other spouse), and negative jointness (parameter takes values greater than one; this means that the marginal tax rate on one spouse depends negatively on the income of the other spouse). (P. 6.) If you’re wondering how there could be varying levels of positive and negative jointness versus just pure joint vs. pure individual systems, the authors explain:
“Empirically most countries do not have a completely joint or a fully separate tax system, however, but somewhat of a mixed system. One reasons for this is that there are various transfers, such as child support or housing support to low-income individuals that follow the household, whereas the income tax schedule itself is fully separate. Another complication is that there are several tax authorities, and local taxes could be separate while national taxes are joint.” (P. 6.)
To generate estimates for the tax jointness parameter as well as the progressivity parameter, HKS use data from the OECD Tax-Benefit web calculator. (P. 7) (I wish the paper got more into the weeds on exactly how they did this, because the tax jointness parameter is a to-my-knowledge novel contribution of the paper.) And to generate estimates of the outcome variable – employment by gender and marital status – they use data from the European Union Labor Force Survey and the Current Population Survey for Western Europe and the U.S., respectively. (P. 7.) They then run a series of regressions in which they measure the correlation between these employment outcomes and a coefficient of interest that is constructed from interaction between tax progressivity and tax jointness; higher values of this parameter indicate less jointness and more separateness. (P. 8.) For both married men and married women, the coefficient is positive and statistically-significant at the 1 percent level, showing a correlation between high employment and the combination of a highly progressive rate schedule and an individual tax unit.
HKS offer the following intuition for the association: “for married couples, higher tax progressivity together with individual taxation of the two spouses means [a] lower average tax rate for the secondary earner, which gives higher incentives to the secondary earner to enter the labor market. There is also an income effect. The primary earner will bring home less net income with separate progressive taxation, which increases the benefit of having two earners.” (P. 8.) In effect, the substitution effect for the secondary earner and the income effect for the higher earner point in the same direction: more work and less leisure. Contrast this with HKS’s results for a less-separate (i.e., more-joint) system: they find that increased progressivity results in a decrease in labor force participation for both spouses. (P. 28.) In effect, the economic productivity cost of progressive taxation for married couples – the distortions of each spouse’s labor efforts – can be mitigated by a more separate system.
The second contribution of the paper is to examine which tax unit – joint or individual – best supports the government’s ability to raise revenue. (P. 26.) Here, the famous Laffer curve (which suggests that tax revenue will increase on an upward slope over increasing tax rates before decreasing on a downward slope to zero at confiscatory rates) makes an appearance. Specifically, the model simulates tax revenue for different levels of progressivity under each filing regime. (P. 24.) HKS find that, for a given level of progressivity, significantly more revenue can be raised with individual taxation as compared to joint taxation (P. 25) and that the revenue cost of increased progressivity is smaller for individual taxation than for joint taxation. (P. 28.) And, if the policy goal is to maximize revenue, individual filing trounces joint filing here as well: the revenue-maximizing degree of progressivity is twice that of the status quo’s joint-filing U.S. system. Further, the revenue-maximizing average tax rate under individual filing is 42 percent – much higher than our current average rate of 12 percent (P. 25) (which, by the way, is far from being on the downward-sloping part of the Laffer curve according to HKS’s Figure 2!).
These two insights – that an individual filing regime in the presence of a progressive rate schedule can both promote labor supply and maximize revenue – are payoff enough for one paper. But what about the normative goal of economic analysis – the promotion not just of the narrow idea of increased employment, but also of the broader idea of welfare? Would “shifting from the status quo to a separate, individual-based system” yield welfare gains (P. 31)? Here’s where the paper gets extra-complicated; I had to print out a Greek-letter cheat-sheet. HKS provide a life-cycle macro-model in which the population consists of overlapping cohorts, households are one of three types (single men, single women, and married couples), individuals have varying innate abilities to earn income, and human capital can be accumulated through labor force participation at varying intensities. (Pp. 9–10.) They also model the production (firm) side of the economy, public sector spending, and a robust tax mix (consumption, labor and capital taxes). (Pp. 12–14.) They then calibrate the model using a dizzying array of U.S. data sources. (Pp. 17–21.)
With this infrastructure, it is possible for HKS to explore “the tax system that maximizes steady state welfare…of a newborn individual, both for a tax system in which married couples are taxed jointly and under separate taxation”. (P. 31.) They find that under joint taxation, the welfare gains from optimizing the level and progressivity of the tax system are small, and entail lower progressivity. (P. 31.) Under an optimized individual-filing system, however, progressivity increases and welfare gains are substantial: indeed, “[a]lmost all demographic groups benefit from the reform to an individual-based slightly more progressive tax schedule…” (P. 32.) Interestingly, one of the two exceptions is young, married, low-ability women, who enter the labor force under the reform because increased progressivity reduces the net income of their high-ability husbands. (P. 33.) They are worse-off welfare-wise, but the U.S.’s stagnating female labor force participation rate improves and the income distribution within couples becomes more equal. (P. 35.)
Greater women’s labor force participation? More progressivity yielding higher revenues? Increased earnings equity within households? It’s not often that a math-heavy economics paper endorses a reform that accomplishes some of the most important and long-standing goals of the feminist movement. Rethinking the tax unit just might bring together unlikely allies.
