Definitions matter. I have been giving considerable thought lately to the confusion surrounding use of the term “universal pension”. This concern surfaced again when I read a paper by Manchester economist Armando Barrientos that draws on work by Harvard economist Martin Feldstein. Barrientos writes
Feldstein (1987: 468–84) considered the welfare effects of categorical versus means tested social pensions.
Armando Barrientos, A., “What is the Role of Social Pensions in Asia?“, Working Paper 351, Asian Development Bank Institute, Tokyo, April 2012, p. 21.
Barrientos goes on to discuss Feldstein’s findings at some length. Barrientos (p. 3) carefully explains that social pensions are “noncontributory pensions because entitlements are not based on employment or on a record of payroll contributions”. By “categorical”, Barrientos means “universal”, “providing benefits to all citizens above a particular age” (p. 9). “Means tested” benefits are provided only to those classified as poor.
This sounded unbelievable to me. I had read much of Feldstein’s work on pensions, and never came across anything on noncontributory pensions, much less on universal versus means-tested pensions. I retrieved the relevant article, and began reading it. Here is the first paragraph:
Every society must solve the problem of supporting those individuals who become too old to work but have not made adequate provision for their own old age by saving when they were young. At the present time, the major industrial countries of the world have responded to this problem by creating social security programs that tax the working population and use the proceeds to provide a “universal” benefit to all retirees regardless of their financial condition.
Martin Feldstein, M., “Should Social Security Benefits be Means Tested?” Journal of Political Economy 95:3 (1987), pp. 468–484.
Already, I was suspicious. In the US, “social security” refers to contributory pensions (tier 2) and “supplemental security income” to social pensions (tier 1). Reading more, it became very clear. Feldstein was referring to contributory pensions, not noncontributory pensions. When he writes that “social security” is “universal”, he doesn’t mean that every elderly resident receives a pension, nor that the size of the pension is a decreasing function of financial wealth. What he means is that the same rules apply to everyone, and that all workers can (and should) participate in the scheme.
Why, then, does Feldstein refer to a social security “tax” on workers. If it is a tax, it is an unusual tax: the more you pay, the larger the benefit. It is widely-known that social security is not very redistributive, and Feldstein does not argue that contributions are a tax in this sense. In fact, in his model he assumes that all workers earn the same wage and pay the same “tax”; they differ only in their propensity to save. Feldstein nonetheless considers social security to be a tax because the implicit rate of return on contributions (equal to the rate of growth of aggregate wages) is lower, on average, than the return on savings invested in the “real” economy (the stock market).
This is nothing new. Feldstein has made the same point in numerous articles. He has been criticised for neglecting investment risk in stocks, and for ignoring the high cost of inflation-indexed, life annuities in financial markets. “Social Security” benefits, after all, increase each year along with prices, whereas private annuities (pensions) generally do not.
Even with these omissions, which tilt the findings against “universal” social security, Feldstein finds that means-tests for benefits may not be such a good idea. Why? Because “a means-tested program … may induce some utility maximizing workers to save nothing” (p. 470). In other words, a means-test is a tax on saving, which induces participants to save less in order to qualify for social security benefits. With flat, noncontributory pensions, it is even easier to show that universality trumps targeted benefits. This is not just a theoretical point. One need only look at Australia to see how means-tested social pensions can distort savings patterns.