New paper: “Labor Supply Responses to Marginal Social Security Benefits: Evidence from Discontinuities”

The title of this new paper from Jeffrey B. Liebman, Erzo F.P. Luttmer and David G. Seif, all of Harvard, may be a mouthful, but "Labor Supply Responses to Marginal Social Security Benefits: Evidence from Discontinuities" examines an interesting and important question: do participants in the Social Security program understand and respond to the incentives that the benefit formula presents to them? Here's the abstract:

A key question for Social Security reform is whether workers currently perceive the link on the margin between the Social Security taxes they pay and the Social Security benefits they will receive. We estimate the effects of the marginal Social Security benefits that accrue with additional earnings on three measures of labor supply: retirement, hours, and labor earnings. We develop a new approach to identifying these incentive effects by exploiting five provisions in the Social Security benefit rules that generate discontinuities in marginal benefits or non-linearities in marginal benefits that converge to discontinuities as uncertainty about the future is resolved. We find clear evidence that individuals approaching retirement (age 52 and older) respond to the Social Security tax-benefit link on the extensive margin of their labor supply decisions: we estimate that a 10 percent increase in the net-of-tax share reduces the two-year retirement hazard by a statistically significant 2.1 percentage points from a base rate of 15 percent. The evidence with regards to labor supply responses on the intensive margin is more mixed: we estimate that the elasticity of hours with respect to the net-of-tax share is 0.41 and statistically significant, but we do not find a statistically significant earnings elasticity.

The short story is that the authors conclude that we can reject the view that participants are completely unaware and unresponsive to the incentives presented by Social Security. As a result, gains from shifting to a more transparent means of Social Security benefit accumulation – namely, personal accounts – may be smaller than previously believed.

A couple quick thoughts: First, marginal returns paid by Social Security to near-retirees are in general very low. Based on forthcoming work with David Weaver and Gayle Reznick of SSA, we found a median marginal return from an extra year of work of around -50%, which implies that continued work isn't a particularly good deal for most retirees. That said, there is a range of marginal returns, and from this range it should be possible to back out differences in labor force participation, as Liebman, Luttmer and Seif do.

Second, while personal accounts potentially present clearer work incentives, in practice that's often not the case. Most personal accounts plans layer accounts over the current benefit formula (usually amended in some way to achieve solvency). As a result, participants would face all the incentive problems under current law Social Security, as well as any issues raised by the accounts. I suspect most workers would have a difficult time figuring out what their incentives were. As policy, clarifying and improving work incentives under Social Security should be a major priority, but it will take relatively large fixes to the benefit formula to get things right.

Read more!

Larry Lindsey: Cut payroll tax to stimulate economy

Former Federal Reserve Governor and Bush economic advisor – and current AEI visiting scholar – Lawrence Lindsey argues in the Weekly Standard for a 50 percent reduction in the payroll tax with revenue losses filled with a new carbon tax. I've been skeptical of payroll tax cuts before, but there's something here that I find interesting.

I've argued previously that Social Security's future shortfalls are really a function of over-generosity to past participants, who collected something like $17 trillion more in benefits than they paid in taxes. Diamond and Orszag call this the "legacy debt."As a result, current and future participants will have to collect $17 trillion less in benefits than they'll pay in taxes. In other words, this $17 trillion is a "pure tax," meaning that we'll pay that amount in but receive no benefits back in return.

One idea I've been playing with is to separate out the pure tax element of Social Security financing, funding it with a new tax instead of through the payroll tax. The idea here is a) to fund this legacy debt through the most efficient tax possible, which probably isn't a tax on labor; and b) to make clearer to Social Security participants what's happening with their money. While the tax to fund the legacy debt won't bring them any new benefits, the remaining Social Security tax would pay a market rate of return, rather than the below-market returns paid today.

Lindsey's proposal is consistent with this idea. The $17 trillion shortfall is equal to just around 6.2 percent of all future payroll, meaning that Lindsey's idea to cut the $12.4 percent payroll tax in half and fund the legacy debt with a carbon tax is equivalent to funding the Social Security legacy debt through the carbon tax. However, under Lindsay's plan future Social Security benefits would still need to be reduced to a level affordable within 6.2 percent of payroll.

Read more!

More on Social Security as a Ponzi scheme

Business Week's Michael Mandel talks up the Ponzi scheme angle, then proposes his solution:

But there is one enormous difference between Social Security and a Ponzi scheme: Technological change. Over the past century, new technologies have enabled the output of the country to grow much faster than its population. To be more precise, the U.S. population has more than tripled since the early 1900s, while the U.S. economic output has gone up by more than 20 times.

This long track record of technology-powered growth has enabled the enormous rise in living standards in the U.S. and other developed countries. In fact, this increase in productivity—output per worker—is the key fact which gives us our way of life today.

Assuming that technological progress continues over the next 70 years, and output productivity growth continues over the next 70 years, the finances of Social Security are relatively easy to fix. A fairly minor cut in benefits, combined with a relatively small increase in taxes, will bring the system back into balance again. (the latest Social Security report projects a 75-year deficit of $4.3 trillion. That sounds like a lot of money, but over 75 years it's roughly $60 billion a year…not chicken feed, but not overwhelming).

But here's the rub. Ultimately our ability to make good on the "Ponzi-like" nature of Social Security depends on the continued march of technological progress—and in particular, innovation which boosts output and living standards. If we leave the younger generation a good legacy—a sound scientific and technological base, combined with an innovative and flexible economy and an educated workforce—then Social Security is not a Ponzi scheme. The economy grows, and there's more than enough resources for everyone.

But if instead we—the current generation—invest in homes, flat-screen televisions and SUVs, then we don't leave the next generation with the technological "seed corn" they need. If the technological progress slows, then Social Security does turn out to be Ponzi-like—with unfortunate consequences for everyone.

I'm all for technological progress – though unlike Mandel I personally think flat-screen televisions are a pretty mean technological feat – yet Mandel seems to have a bit of a technology fetish, such that it's the cure for everything.

In this case, technological progress – which affects Social Security by increasing productivity, thereby increasing wages – doesn't have that much to offer. The reason is that retirement benefits in the future are based on wages today; if technology raises worker's wages today, it also increases the amount Social Security promises to pay them in the future. On a person-for-person basis, it's roughly a wash.

Technology/productivity/wage growth can help because benefits paid to current retirees aren't indexed to wage; they rise year-to-year only with inflation. So if wages rise, the cost of paying current benefits falls relative to the wage base. Roughly speaking, the system's actuarial deficit – currently around 1.7% of payroll over 75-years – improves on a one-for-one basis with rising wage growth. The baseline rate of wage growth is 1.1%, so this implies that if wages grew at 2.8% then the program would be solvent for 75-years (though not beyond).

The spreadsheet I posted here lets you play around with different economic/demographic assumptions. The key takeaway for me is that we shouldn't bank too much on productivity growth to bail us out of the Social Security problem.

Read more!

Is the Social Security tax regressive once you account for benefits?

It's sometimes argued that, while income taxes are progressive, the progressivity of the total tax code has to be viewed inclusive of payroll taxes, which are either flat (in the case of the 2.9% Medicare tax) or regressive (in the case of the 12.4% Social Security tax, which applies only to the first $106,000 in earnings). It's not always clear from these statements what the net effect is.

A new CBO letter allows for a better view of this, although even this doesn't tell the whole story (as I'll discuss below). The picture below first shows effective income tax rates by income quintile, for people in 2005. As you'd expect, they're pretty progressive, and the poorest 40 percent of Americans pay negative rates through policies like the Earned Income Tax Credit.

The next picture shows effective social insurance – Social Security and Medicare – tax rates, also by income quintile. While rates are somewhat progressive through the fourth quintile, they decline for the top quintile because of the cap on Social Security taxes.

If we add social insurance and income taxes, along with corporate income and excise taxes, the next picture shows total effective federal tax rates. As you can see, even if we add social insurance taxes, the overall tax code is still reasonably progressive (in my view; others may defined reasonable differently).

But below is a chart I've constructed from a different data source, the GEMINI model of Social Security financing. The key issue with Social Security taxes is that while the tax itself is regressive, the benefits are progressive. Since taxes pay for benefits, you want to look at the progressivity of the Social Security program as a whole. The chart below is for individuals retiring in the 2030s, although they would not be much different for people retiring earlier or later (it's just the data I had lying around…). I've calculated effective payroll tax rates by lifetime earnings quintile, which means the payroll tax paid minus the disability and retirement benefits the individual receives. If you received benefits exactly equal to your taxes (plus interest at the government bond rate) then your net tax would be zero. Although net tax rates are always lower than the statutory 12.4% rate, at least for people (as here) who survive to retirement, the distributional picture is very different.

The highest quintile of lifetime earners pays a net tax of around 3 of earnings. This implies that they pay 12.4% of wages while working, but then receive benefits back equal to around 9.4% of wages. While this isn't a great deal – in a fully funded system they'd receive back everything they paid in – it's better than paying 12.4% and getting nothing back.

But notice what happens as lifetime earnings decline. Net tax rates are negative, meaning that (under current law benefits, at least) most folks would get out more in benefits than they pay in taxes. For the lowest earners, net taxes are very negative, around -27 percent. This means that on average these folks receive about three times more in benefits than they pay in taxes. Remember this the next time someone says that the Social Security tax is regressive: sure, it is, but things look very different when benefits are counted into the picture.

As a P.S., why are net tax rates more negative for the middle quintile than for the second quintile? The answer to that is, I don't know – although because this data output mixes both retirement and disability benefits, it should be possible to disaggregate them and get a better feel for things.

Read more!

NBER Digest: Changes in Social Security Rules Affect Retirement Choices

The NBER Digest summarizes a recent paper by Al Gustman (Darmouth) and Tom Steinmeir (Texas Tech) on how changes to Social Security benefit rules have affected retirement behavior:

Changes in Social Security Have Affected Retirement

Social Security rules changes increased full-time work by married men aged 65 to 67 by about 9 percent between 1992 and 2004, encouraged later retirement, promoted the return to full-time work after retiring, and facilitated working part-time after retirement.

In How Changes in Social Security Affect Recent Retirement Trends (NBER Working Paper No. 14105), co-authors Alan Gustman and Thomas Steinmeier find that changes in Social Security rules have changed the shape of retirement. Rule changes increased full-time work by married men aged 65 to 67 by about 9 percent between 1992 and 2004, encouraged later retirement, promoted the return to full-time work after retiring, and facilitated working part-time after retirement. All in all, they account for about one sixth of the increase in labor force participation by 65 to 67 year old married men between 1998 and 2004.

One of the main reasons for enacting the 1983 Social Security reforms in the United States was to increase the labor force participation rate of older workers. In 2000, Congress further expanded work incentives by abolishing the Social Security earnings test for people over the normal retirement age. As a consequence, in 2004 more men over age 65 were working than in earlier years. Overall, between 1998 and 2004 there was a 3.1 percentage point decline in the fraction of 65-to-67-year-old men who were completely retired.

But the experience of younger men was different. Labor force participation rates declined for men aged 50 to 56. And, more men were retired at younger ages in 2004 than in previous years. In 1998, 80.4 percent of men 50 to 56 years old worked full-time. By 2004, only 75.5 percent did so.

To isolate the effect of Social Security rule changes from other factors -- such as the abolition of mandatory retirement ages, changes in employment and compensation policies, rising incomes encouraging early retirement, the stock market boom, and the rising labor force participation rates of women -- the authors estimate a retirement model using Health and Retirement Survey data on 2,231 married men. They then simulate the effects of evolving Social Security rules, assuming that each individual has the work history actually experienced. Individual time preference rates are varied, so that some people respond strongly to delayed incentives and others respond only to incentives that affect current consumption.

The baseline model's estimates suggest that the value of retirement leisure is increasing by 5.4 percent per year, a relatively low value, suggesting that economic incentives can change work effort in retirement. Poor health increases the value of retirement leisure by approximately the same amount as being seven years older, and individuals may change their perception of retirement after they experience it, with some deciding to go back to work. The simulation results further suggest that differences in Social Security rules have no effect prior to age 62. This means that the decline in labor force participation observed for those in their fifties has another cause.

-- Linda Gorman

Read more!

New paper: Chile's Next Generation Pension Reform

The most recent issue of the Social Security Bulletin includes a new article by Barbara Kritzer of SSA titled "Chile's Next Generation Pension Reform." This is important stuff given how influential Chile's personal accounts-based reforms have been throughout Latin America and as an inspiration for "privatization" in the U.S. Here's the abstract:

In 1981, Chile was the first country to replace its public pay-as-you-go pension system with mandatory individual retirement accounts. This system has become a model for pension reformers around the world. Although Chile's system has undergone many changes since its inception, a number of policy challenges remained, including worker coverage, pension adequacy, gender equity, and administrative fees. A March 2008 comprehensive pension reform law addressed many of these issues by adding a basic universal pension, requiring the self-employed to join the individual account system, encouraging greater competition among pension fund providers, lowering administrative fees, providing financial incentives for women to work longer, allowing widowers to receive a survivor pension, dividing individual account assets between spouses in case of divorce or marriage annulment, expanding voluntary pensions, setting up a new administrative structure, and launching a new financial education program.

Click here to read the whole paper.

Read more!

Off-topic: Unionization and the labor share of GDP

I occasionally go off the Social Security/pensions/aging theme and do some digging on other issues that interest me. Like many Americans, I've been hearing about what the bailouts of the auto industry may require of union wages, benefits and work rules, as well as reading a bit on whether Congress should pass so-called "card check" legislation that would make it easier for workers to unionize.

The basic logic I see for a union is that it improves the bargaining power of labor versus capital (meaning, the stock and bond holders who own a company). Capital can easily shift from state to state or overseas while labor moves only slowly, so you could argue that by banding together workers have more leverage and can extract more from their employers in terms of wages or benefits. So here's a quick test I thought of: compare the rate of unionization within a country with the share of that country's GDP that flows to labor rather than capital. Presumably, if unions are effective at increasing the bargaining power of labor versus company ownership – distinct from, say, benefiting union workers versus non-union workers – then the labor share of GDP should be higher in countries where a larger share of workers are unionized.

As it turns out, that's not really the case among large countries. Here's a chart showing what I found. In some countries, such as Finland and Denmark, around three-quarters of workers are in unions. In other countries, like the U.S., less than 15 percent are union members. The labor share of GDP which ranges from 45 percent in Italy to 60 percent in the U.K. (I calculated the labor shares in the year 2000 from UN data available here (fewer countries are available in more recent years) using a methodology similar to that used in this paper. The unionized share of the workforce is from NationMaster (a very cool site) using OECD data.)

The variation in union membership is a lot larger than the variation in the labor share of GDP, but there's no strong statistical relationship between one and the other. Countries with a larger unionized labor force don't receive a larger share of economic output. The split between workers and owners seems pretty much independent of whether the workers are union-represented or not.

This seems strange given that we know that, in the U.S. at least, unionized jobs pay more than non-unionized ones, even within the same industry. The question is, who is "paying" for these higher wages? It might be other workers, such that unionization might result in a smaller number of higher-paid jobs, although a quick cut through the data doesn't show anything obvious there. In any case, though, I thought the lack of any correlation between unionization and the share of GDP that flows to workers rather than owners is interesting, given how big a policy issue unions have become in recent months. I'm sure there's more I could check, both in terms of my own numbers and existing literature, but this was really a little distraction for a day off.

Read more!