Crowding Out and the Perils of Oversimplified Models

Tom Bozzo

tips the hat to Ken and SvN for a pointer to a nice paper from the Journal of Money, Credit and Banking called “The Deficit Gamble,” by Laurence Ball, Douglas Elmendorf (now CBO director), and N. Gregory Mankiw. Ball, Elmendorf, and Mankiw showed that the government can “with a high probability” run a deficit in the present and roll the resulting debt over indefinitely. In the states of the world where this so-called “Ponzi gamble” (*) works, the result is Pareto-improving; if the gamble fails, then at least some generations are better-off. The authors note that this does not imply that deficits are necessarily “good policy.” (**)

The details of how this works also bear a bit on the “freshwater” versus “saltwater” debate over the stimulus package and “crowding out” of private sector investment. One possibility is that there can be crowding out even in the sense that Fama describes and that may nevertheless be welfare-enhancing. Under risk aversion, a reduction in the variance of consumption can compensate for a reduction in the mean, as in substituting (safe) government debt for risky but higher-return private capital investments. But, more realistically, the portfolio choice is among private investments with a variety of risk/reward features. Ball, Elmendorf, and Mankiw show that under such circumstances, increasing holdings of safe government debt should lead to a portfolio shift towards higher-risk, higher-(mean)-return capital. This formalizes a sort-of “safety nets are good for entrepreneurs” argument. The central issue, though, is that it’s not just the quantity of capital that matters, but also its composition.

In fact, in some respects the Ball, Elmendorf, and Mankiw model may be too pessimistic as to the likelihood of the deficit-spending gamble paying off. As is fairly common in macro modeling (***), the government spending financed by the deficit is assumed to be unproductive. Of course deficit-financed stimuli can be arranged as airdrops of nondurable consumption goods. The actual stimulus package we’re fighting over, however, would direct a good chunk of money on public-sector investments (not necessarily “public goods”). So some of the “crowding out” is a substitution of public for private investment. We can fight over the relative marginal products of various public and private capital investments, but in general those will be positive. I leave knock-on portfolio effects to others, but I’ll assert without proof that the net effect shouldn’t be to increase the probability that the debt rollover would fail.


(*) The term “Ponzi” may be too loaded; the “gamble” involves a series of intergenerational transfers that are not inherently unsustainable in contrast to a “scheme” a la Madoff.

(**) Basic prudence is not revoked. Borrowing “too much” money, for instance, would affect the probability of successfully rolling over the debt.

(***) I recently flipped through the main graduate macro text of my formative years, Blanchard and Fisher’s Lectures on Macroeconomics, and you have to get pretty far along (p. 591 of my edition) to read that “in the equilibrium context some components of government spending may operate as a current input into production.” Wow, what a concept!