Transcript

Jessie Romero: Hi, I'm Jessie Romero, director of research publications at the Richmond Fed. Thanks for listening to Speaking of the Economy, which you can subscribe to in Apple Podcasts.

I'm excited to be talking today with Mariacristina De Nardi, the Thomas Sargent Professor of Economics at the University of Minnesota. De Nardi has also been an economist at the Chicago and Minneapolis Feds. Her research interests include macroeconomics, public economics, health and wealth.

Today, we'll be discussing how tax policy and Social Security benefits affect women's labor supply. De Nardi recently presented research on this topic at the Richmond Fed's conference, "Household Dynamics at Older Ages."

Mariacristina, thank you so much for being here today.

Mariacristina De Nardi: You're welcome.

Romero: Let's start with a little background. You presented the paper we'll be talking about at a Richmond Fed conference on the economic decisions made by older adults. Why is this an important topic for the Fed to be studying?

De Nardi: The Fed's dual mandate includes promoting maximum employment and stable prices. Employment requires savings to invest in capital so that workers can use that capital — for instance embedded in machines and technology — to produce goods and services. Healthy workers are necessary to use that capital.

As populations are aging across the world, studying health and savings at older ages becomes essential to understand the dynamics of savings and working decisions and, therefore, maximum employment, which is one of the two key mandates for the Fed.

Although we have made much progress over time, better data and techniques allow [us] to better understand several important points in these areas. The first point includes how much people save at older ages and why. Is it because they expect out-of-pocket medical expenses, because they want to leave resources to a surviving spouse after their death, or to leave bequests to their children, for instance? A second point is to what extent government policy — such as Social Security, Medicaid and Medicare — influence savings and labor supply. A third point is how health evolves and why. How expensive is it to be sick? How much do people value being healthy?

More generally, there is very important research being done to provide better answers to these questions and many more, and the Richmond Fed had a truly excellent conference on these topics. This conference allowed researchers to connect with each other but also broadened the audience and informed the audience about these important issues.

Romero: That's a great segue to talk a little bit about the paper you presented at the conference.

In this paper, which you coauthored with Margherita Borella and Fang Yang, you focused on women's labor force participation. Can you tell us a little bit about general trends in women's labor force participation since the mid-20th century?

De Nardi: In the United States, women's labor force participation drastically increased since the 1950s until the mid-1990s. For instance, the participation rate of adult women increased from 33 percent in 1950 to 60 percent in 1997. It then stagnated from 1997 until 2008 and dropped to 57 percent in 2017, which is about the same level that it was seven years earlier. Most of these changes are due to changes in the labor force participation of married women.

Previous work has analyzed the role of various factors explaining the increase until the 1990s, including changes in wages and the labor market, in child care costs, women's education, fertility, and home production. However, little work has been done in studying the role of joint taxation and marital Social Security benefits and how these two programs interact and work together.

Romero: Let's start with talking about taxation. One major difference between the United States and other countries is how it taxes married couples. What is the difference, and how does it tend to affect labor supply?

De Nardi: In the United States, married people can either file taxes jointly or as married but filing separately. However, under the latter option – married people filing separately – you lose a lot of deductions and exemptions. As a result, over 97 percent of married people file taxes jointly.

Married women, which have typically the lowest earning potential in the couple, typically face a higher marginal tax rate when they start to work or they work more. (A marginal tax rate is the tax rate that you pay on the additional income that you earn.) And, a higher marginal tax rate means that when you work, you obtain a lower wage which, in turn, is a disincentive to work.

Romero: How about Social Security benefits? Do those also work differently for single people compared to married people?

De Nardi: Social Security [benefits] for a single person is a function of one's average lifetime earnings and the related Social Security contributions that the person paid on their own lifetime earnings.

In contrast, for a married person there is a choice. The person can either claim Social Security under their own entitlement or can claim half of their spouse's entitlement while their spouse is alive and both are of retirement age. After one spouse dies, the surviving spouse can claim the highest between their own entitlement and the spousal entitlement.

Therefore, married couples have an additional level of choice when claiming Social Security benefits. This means that even if a married woman or a secondary earner never worked or had very low Social Security contributions, they can still obtain a sizeable amount of Social Security benefits after they retire just because they got married.

Romero: Then how do you see that affecting incentives to work?

De Nardi: When married women work less, they don't necessarily obtain lower Social Security benefits. These additional benefits reduce the cost of not working by providing an income flow that is essentially unrelated to their own labor supply.

So, both provisions — a higher marginal tax rate that reduces the incentive to work and marital benefits coming from Social Security reduces the cost of not working — work together in terms of reducing the labor supply of the secondary earner.

Romero: Let's turn to talking about your paper in a little bit more detail. You look at two different cohorts of women and men. What are the basic patterns you observe in each cohort, and are there any major differences between the two groups?

De Nardi: We consider two cohorts of women because we want to understand how policies over time effect the behavior of cohorts that behave rather differently in the past.

Our first cohort is the one born in 1941 to 1945, which we call the "1945 cohort" for short. We picked it because it has, by now, completed a large part of its life cycle and, therefore, we have excellent data both on the working period and the retirement period.

Our second cohort, which is younger, was born from 1951 to 1955 and we call it the "55 cohort" for short. It has completed a lot of their working period and has much higher participation of married women. As you consider a younger cohort, the challenge is that you have less data. But the advantage is that, because they are younger, it is a cohort that is more informative about the women who are working now.

What we find when we analyze these two cohorts is that for the older cohort, the one born in 1945, we show that married men have the highest participation rate and only slowly decrease their participation starting from age 45, whereas single men decrease their participation much faster. Even more importantly, turning to women, the participation of single women starts about 10 percentage points lower than that of single men and gradually increases until age 50.

Our group of married women has the lowest participation rate. It starts at around 50 percent at age 25, increases to 78 percent between ages 40 and 50, and gradually declines at a rate similar to that of the other group after those ages. Not only [do] married women in this cohort participate much less than single men and then married men, but they also work fewer hours.

When we compare with the younger cohort — the one born in 1955 — we see a large difference in the participation of married women. That is the most notable feature across these two cohorts. In particular, the cohort born in 1955 at age 25 had a participation [rate] of 62 percent compared to 50 percent in the previous cohort. So, in the 10-year period that elapsed between the two cohorts, you had a 12 percentage point increase in participation of married women, which is very large.

Romero: You run several experiments. One is eliminating spousal Social Security benefits. Another is eliminating joint taxation. And then you see what happens when you eliminate both. What effects do these changes have on women's labor force participation?

De Nardi: For the 1945 cohort, we find that the elimination of Social Security spousal and survivor benefits and joint income taxation has large effects, both on participation and savings. In particular, it raises the participation of married women at age 25 by over 20 percentage points — so we go from 50 to 75 percent — and that of single women by five percentage points. At age 45, the participation for these groups is still 15 and three percentage points higher, respectively. These are very large effects.

Romero: Yes, those are very large. Did you see any effect on men's labor force participation?

De Nardi: What we find is that, in response to this large increase in participation of married women, the participation of married men decreases starting at age 60, resulting in a participation rate that is seven percentage points lower by age 65.

Romero: Do you have any thoughts on why that might be?

De Nardi: Yes. What happens is that it is much more advantageous and profitable for women to work because their taxes are lower. Women work more, they increase [their] human capital and they earn much more, so the couple has a much larger income flow.

What is going on is that the men would like to enjoy a little more leisure given that their wives are working much more. The moment during which they want to enjoy leisure is when they're older. When they're older, there is less return to working because as you work more you don't have many years left, therefore any additional human capital and future wages is not so important.

Romero: Okay, that makes sense. Did you find effects on other variables, such as earnings or savings?

De Nardi: Eliminating these policies increases the savings of married couples by 20 percent. The reason is that the elimination of these Social Security marital benefits implies a reduction in income close to retirement. Because women earn more, the couple saves more to smooth income over the life cycle.

Not only [do] savings increase, but also women's earnings increase. They increase because of two reasons, not only because their hours increase but also because their human capital, their labor market experience increases.

Romero: You talked some about the large effects [that] you saw for the 1945 cohort. Did you see similar effects for the 1955 cohort?

De Nardi: The short answer to your question is yes. We were surprised and at the same time we didn't know what to expect.

We found that the effects of these marital provisions on the participation, wages, earnings and savings on the 1955 cohort are also large, thus indicating that the joint taxation and marital provisions of Social Security also have an effect on current cohorts for whom the labor market participation of married women is higher.

If you look into what the tax rates are for joint taxation and single taxation, and how large the Social Security marital benefits can be in old age, this is not surprising. But it's still good to actually go through the analysis and really evaluate how people respond to these incentives.

Romero: You found increases in women's labor force participation and in their human capital and savings. These all sound like good things. But are these gains, are they enough to offset things like the cost of going to work, the loss of spousal benefits and the decline you saw in men's labor force participation?

De Nardi: An important outcome of these reforms that we find is that the elimination of Social Security marital benefits implies a budget surplus. When you use this budget surplus to reduce the payroll tax, you can actually reduce the payroll tax by about two percentage points, which is not a small amount. In terms of welfare, abolishing these marital provisions and, at the same time, lowering the payroll tax has important beneficial effects for many people.

More particularly for the 1945 cohort, these reforms would benefit most couples, all single men and one third of single women, therefore 90 percent of people in this cohort. Who are the people who are losing? As you might imagine, these are people who have no human capital and low earnings capacity. Therefore, for them rebating the tax is not enough to compensate for the loss of Social Security benefits.

In terms of the 1955 cohort, these reforms will benefit most couples, all single men and over three quarters of single women. Single women were already more educated and had higher human capital in this cohort.

Very importantly, you shouldn't only look at the number of people who gain and the number of people who lose, which in this case is small, but also the size of the gains and losses. What is important to notice is that the gains of those who gain are large and the losses of those who lose are small. This is an indicator that you can design a reform that compensates those who lose for this reform while maintaining the gains of those who gain. Therefore, to answer your question, you can design a reform that is welfare improving for everyone, if correctly organized.

Romero: Thank you so much for your time. I really enjoyed our conversation and I learned a lot.

De Nardi: Thank you very much.