My Public Economics class

Today's subject: social insurance.

xtIt's odd how leading economists, including very good ones such as Jonathan Gruber, the author of the textbook I am using, acquiesce to categories that have no economic substance. They could say that they're deferring to common usage, but don't always do so and may not always realize it.

Case in point: Gruber defines social insurance - following prevailing conventional wisdom - in terms of private insurance, which he says is characterized by (1) premiums, (2) event conditioning (i.e., it's paid when and if X happens), and (3) lack of income conditioning. This supposedly explains why Social Security is "social insurance" but the income tax and welfare system aren't.

In fact, insurance is a kind of bet one places to hedge other bets one is forced by circumstances to make with the aim of directing dollars to states of the world where they are expected to have greater marginal value in utility terms. One doesn't insure against horrific events, such as the death of a child, that don't make the marginal dollar more valuable. One does sometimes insure against good things, such as living longer & thus needing more money, or having an untreatable medical condition that becomes treatable.

Let's look at Social Security and the 3 supposed key features. It doesn't meaningfully have premiums absent arm's length exchange. What it has is the arbitrary designation of certain compelled tax revenues as ostensibly earmarked to pay over the long term for the program. Earmarking is an interesting issue, having to do with the effort to create political binding pre-commitment, but has nothing to do with premiums as such. If they ended the earmarking and folded the Social Security premiums into general revenues, absolutely nothing would change beyond bookkeeping unless (as is plausible) actual political decisions changed.

Event conditioning: OK, the Social Security life annuity addresses life expectancy risk and provides insurance against living "too long." Probably not a mode of insurance that the government has to provide on adverse selection grounds, the usual lead argument for government involvement. Rather, the argument is paternalism if people under-save and under-annuitize with a dollop of externalities if they would get public support in the absence of other resources.

But insofar as you expect to live to retirement age, a retirement benefit is not event-conditioned in the usual insurance sense - it relates to an event that just happens at a certain point, rather than one that is risky.

OK, lack of income conditioning. Well, it is pure formalism to say that Social Security isn't income-conditioned. Even leaving aside the payroll tax financing to focus exclusively on the benefit side, we could make it income-conditioned with no change in substance whatsoever if we simply reshuffled existing fiscal rules so that the effect of partial income taxation of Social Security benefits was formally made part of Social Security.

Also, why would anyone say insurance isn't income-conditioned? All insurance other than automobile insurance isn't automobile accident-conditioned. Likewise, insurance that isn't income insurance isn't income-conditioned. By definition, income insurance is income-conditioned and other insurance isn't.

We don't observe much income insurance outside of the fiscal system due to the adverse selection problem. (People expecting low income would disproportionately sign up.) This is why the government provides it through income taxation and welfare, among other fiscal instruments. As recognized in the Mirrlees tradition in optimal income taxation, income insurance via taxes and welfare is the government insurance example sine qua non.

Only, it doesn't have anything that we arbitrarily label as a premium, so the economic substance doesn't count.