Is the Social Security System in need of reform?
[See also The New York Times: "A Question of Numbers" by Roger Lowenstein]
Though the numbers reported in the press vary, a common figure reported based upon estimates by the Social Security Trustees is that in 2042 the system will be able to pay 73% of scheduled benefits, a number that will fall to 68% in 2078. The Congressional Budget Office also provides estimates, and its figures are more optimistic predicting that 81% of benefits will be available in 2053, and 71% will be available in 2100.
These figures are reported as though they were known with 100% certainty. We hear repeatedly with the appropriate fist pounding that the system either is, or is not in serious trouble depending upon who is talking the loudest at the moment. We are told that if nothing is done we will most certainly (or most certainly will not) face calamity in the future. But do we know this for sure?
Here’s one way to think about it. Given enough time, two lines that are not exactly parallel will grow arbitrarily far apart. Thus, if the distance between the lines represents a surplus or a deficit in Social Security, we can make the surplus or deficit arbitrarily large by simply moving far enough into the future. All that is required is that the initial growth rates be slightly different.
In the Social Security funding debate, these two lines are population and GDP. Growth in population is a key factor determining the demand for benefits in the future and the growth in GDP represents the ability to supply those benefits. All projections regarding Social Security’s ability to pay future benefits depend critically upon estimates of these two growth rates. Very slight changes in these growth rates can have very large effects on projections when you allow those differences to compound over many decades.
To get a sense of the degree of uncertainty, consider the most common estimate reported regarding benefits, the estimate for 2042. That projection was made forty years into the future. Go back forty years, to the mid 1960’s, and ask yourself how well people could have predicted the economy we have today. Or how about the figure that only 68% will be available in 2078? That is a seventy five year ahead projection. If we go back seventy five years from when the forecast was made, which is prior to 1929 and the onset of the Great Depression and several years before the creation of the Social Security system, how well would we have predicted GDP in 2005?
I will leave it to demographers to comment upon uncertainty regarding population projections over the next fifty to one hundred years, but it seems safe to presume that some uncertainty exists. What I can talk about is the uncertainty regarding estimates of GDP growth. The problem economists face is the same as the problem faced by scientists trying to figure out if there is global warming based upon temperature and other data. In the global warming debate, the central problem is to sort cycles of cooling and warming from underlying changes in the trend temperature. If you see temperatures rising over time, is it just another warming-cooling cycle as have occurred repeatedly in the past, or does it represent a fundamental change in the underlying trend for temperatures?
To answer this question, a very long series of temperature data is needed, and the longer the better. Economists must also sort out boom and bust cycles in GDP from the underlying trend to accurately project GDP in the future, and like those studying global warming, very long data series are needed to do this accurately. Unfortunately, since reliable data regarding the economy do not exist prior to 1947, there is considerable uncertainty regarding the division of GDP growth into trend growth and cycles around that trend, and hence considerable uncertainty regarding the underlying trend rate of growth in the economy. The two growing lines, one for population and one for GDP, have very uncertain trajectories and very uncertain implications for the ability to pay benefits in the future.
I do not wish to imply that Social Security will definitely be solvent fifty or one hundred years from now. Similarly, I do not wish to imply that it won’t be. We just don’t know. Forecasts that far into the future are sufficiently uncertain so as to render any firm conclusions regarding the health of Social Security difficult to make. We should keep our radar up because some estimates indicate that there may be bumps in the road ahead. But whether there is sufficient basis to radically reform the system based upon these highly uncertain estimates regarding the magnitude of the problem is a question that is worth asking.
Posted by Mark Thoma on Saturday, March 5, 2005 at 10:21 AM in Economics, Social Security |
Permalink
TrackBack (0)
Comments (0)
You can follow this conversation by subscribing to the comment feed for this post.