Friday, June 5, 2015

Blahous: "The Social Security Trustees Respectable Projection Record"

Over at e21, Social Security Public Trustee Chuck Blahous writes about the accuracy of the Trustees' projections, in the wake of a two recent papers by several academics claiming that SSA's SSA actuaries, who are very influential in setting the Trustees' assumptions, have made significant and one-sided errors in recent years. Blahous states, "As a currently-serving trustee I have been asked for my view of the Kashin-King-Soneji work. Summarizing very roughly, their factual observations and analyses strike me as essentially correct, though I disagree with many of their interpretative conclusions."

Blahous makes several valuable points. He notes, for instance, that while the Kashin-King-Soneji papers focus on Social Security's mortality assumptions -- that is, projections of how quickly life spans increase -- these are only a small part of the many factors that influence social security's finances. Other factors, in particular the Great Recession, caused much larger errors in projecting Social Security finances.The papers are really one some very narrow demographic issues in the Social Security projection process, but are generalized to the broader conclusions drawn by Social Security's actuaries and trustees.

On the other hand, I can't help but think that some of the qualitative points made by Kashin-Kind-Soneji -- that the processes for setting these assumptions and doing other calculations regarding Social Security finances are too closed, secretive and turf-conscious, and that there's a (perhaps undue) emphasis on not making significant changes to these projections from year to year -- are true and important. From my own experience at SSA, which included several years in the inter-agency working group that helps construct the Trustees Report, it's an odd process. Some of their criticisms are harsh, at times overly so, but they do convey something of the flavor of how things work at SSA.

Finally, the Kashin-King-Soneji papers -- which focus on SSA's longevity assumptions -- reminded of a chart on disability projections contained in 2014 Congressional testimony by Richard Burkhauser, a disability expert at Corner who's also affiliated with the American Enterprise Institute. The chart shows the Trustees' projections of disability beneficiaries as a percentage of the working-age population relative to what actually occurred (the solid black line). These projections go back several decades, far before the post-2000 period the Kashin-King-Soneji papers study, but there is a pattern of underestimating the growth of the disability population. Relief from disability costs, it seems, was always right around the corner. If most of the growth of the DI was from predictable demographic causes, which is the SSA actuaries' argument, you'd think the increase would be more easily foreseeable.


WilliamLarsen said...


Can you provide me with why SSA and CBO like to use GDP in forecasting? My review of GDP does not correlate well to wages and wages are used to determine both future benefits and current revenues.

Andrew G. Biggs said...

I'm not sure they really do use GDP, or at least not in a major way. Most of it will be in projecting growth of the wage base, etc., from which (and other assumptions) you can build up GDP. There are some assumptions in terms of how total wages will change relative to GDP, but those are more indirect assumptions. But including GDP measures makes sense, since you want to get a feel for what your assumptions regarding one part of the economy (earnings) implies for the rest.

Bruce Webb said...

One place they do use GDP, and in a particularly useful way, is not in driving projections (if they use it that way at all)but instead in reporting 75 year and Infinite Future actuarial gaps in a way meaningful to most readers. Of the three reporting metrics: percentage of payroll, PV in dollars, and GDP the latter is the one that allows non specialists to really evaluate the impact.
Table VI.F1.—Unfunded OASDI Obligations Through the Infinite Horizon,
Based on Intermediate Assumptions

Is $10.6 trillion really meaningful here? Not IMHO as useful as 2.7% of payroll, but neither is as useful as 1% of GDP over the 75 year window.

Arne said...

" Relief from disability costs, it seems, was always right around the corner."

This seems an odd characterization of the data in the chart. Until 2005 we see extrapolation that seems like growth will continue at the same rate. They were wrong, the growth rate changed, but they were high the one time the growth rate decreased. With 2005 and later they extrapolate that the growth just can't continue. They were wrong because they could not predict the financial crisis.

All this chart shows is that prediction is hard.

WilliamLarsen said...

I have never attempted to model SS-DI. To me that program seemed to be the smaller problem. IT covered workers who died or who had become disabled. OSHA regulations seemed to decreasing workforce injuries/deaths so my first impression was it would have a decreasing cost relative to labor participation rates. With the SS-DI, one had to prove a loss before a benefit could be paid.

In addition, with the dependent change age reduced to age 18 and the birth rate per woman, at first hand, one would think total family benefit costs would be reduced. However, a quick look at the max family benefit most likely is not reduced with lower birth rates because the family max was quickly maxed out with higher birth rates, leading to no real change in max family benefits.

SS-DI benefit uses a similar benefit formula that in the end pays a higher replacement rate.

With advances in medicine, people live longer with a disability as well as healthcare professionals creating new diagnosis which now lead to a wider incidence of being identified as being disabled than in the past.

Modeling this type of program versus just reaching full retirement age is totally different. Clearly its design for revenue and expenses do not match. Many companies offer private disability insurance. How they manage costs or model their risk v payout I have no idea on. However, maybe SS-DI can look at their methods to see why the private sector seams to be working while the government program is not.

Arne said...

"How they manage costs or model their risk v payout I have no idea on."

They adjust their premiums as their costs change. SS needs to do the same.

WilliamLarsen said...

As for managing costs, it is more than just changing premiums. SS-DI in most cases when a change takes place begins retroactively to all who are in the program. Private disability plans once you are enrolled and are paying a premium is set as is the coverage. If the private sector determines a change is needed such as covering fewer or more areas, then they offer this new product at a specified premium based on risk.

Social Security makes a change covering a new area and it now includes all people working without determining risk. I just do not see SS-DI ever being able to manage costs; Its a government agency created by politicians.