What Do We Know About the Propensity to Save?

by Reihan Salam

Riffing on the same CBO numbers that Josh Barro highlights below, Ezra Klein writes:

Alert readers will have noticed something odd in the CBO’s graph of the harm that different tax cut policies will do to the economy. On the one hand, the harm comes from increasing the amount that we borrow. On the other hand, the extension of all tax cuts — as opposed to those just for income under $250,000 — actually does less damage to the economy, at least under one scenario.

People can — and will — take issue with the CBO’s model here. For instance, it holds demand-side effects constant, so the fact that lower-income folks spend while higher-income folks save isn’t included. [Emphasis added.]

But there is a difference between long-run damage to the economy and short-run damage. We’re very concerned about aggregate demand at the moment, but it’s still true that savings and investment are a driver of long-run growth. And the incentive effects of high marginal tax rates are a very big deal. Even if we leave these perennial issues aside, however, we have new evidence on the propensity to save, which my friend and colleague Arpit Gupta recently wrote about for Economics 21. He cites new research from economists Matthew Shapiro, Claudia Sahm, and Joel Slemrod that draws on responses to the tax incentives included in ARRA:

The report also challenges another key plank of the Administration’s policymaking: that tax cuts delivered to the poor offer more bang-for-the-buck than tax cuts delivered to the rich. This idea underlies a large number of policy choices, from unemployment insurance extensions to extending the Bush tax cuts for the middle-class but not the rich. Yet the authors find this was not the case for the stimulus program. When considering households that report they will respond to the tax cut by “mostly spending,” the researchers found:

Households who report being worse off financially are almost 7 percentage points less likely to report mostly spending the additional income from any given policy

Interestingly, the authors also find comparable effects for households that expect economic conditions to deteriorate in the future:

A household that expected their income to decline by more than 10 percent over the next year has a mostly-spend rate almost 8 percentage points lower than a household that expects their income to be unchanged

This is plausible if one thinks that poor or struggling households choose to save additional income rather than spend it, a seemingly wise decision. Yet that, in turn, implies that the “multiplier” of spending in response to, say, extensions of unemployment insurance may not be as high as many had projected. Expectations of the future loom large in predicting behavior today. Whether future disposable income is expected to be lower due to tax increases or a deteriorating economy, the natural reaction among households is to reduce spending today. 


These results do not suggest that tax cuts, in general, are a bad way to deliver growth in prosperity. The study only examined the impact of temporary, one-time cuts in taxes, as opposed to permanent decreases in marginal tax rates. More permanent cuts in tax rates, especially if spending cuts are used to offset them, induce supply-side responses that raise long-term labor supply and output. Additionally, households facing a long-term decrease in taxes envision a long-run rise in wealth, which should induce a larger consumption response compared to a temporary boost in income.

Nor do these results necessarily suggest that the tax component of the stimulus was a bad idea. If the goal was simply to transfer income to poor or struggling households, the program may have worked as intended. [Emphasis added.]

This research is hardly conclusive. But it’s certainly worthy of note. Rather oddly, it suggests that if tax cuts to the rich might have been a more effective way of increasing spending, as low to moderate income households have been focused on repairing their balance sheets. Arpit continues:

For instance, if the goal were simply to maximize overall spending, a credible case could be made that a more effective stimulus measure would have been a large, one-time tax rebate to affluent households. Of course, for a variety of reasons, such a proposal would almost surely get blocked in Congress. (It’s also worth noting that just like with short-term tax policies aimed at low-income individuals, the permanent income hypothesis does suggest that even these new resources would be invested.)

So do we really care about the propensity to save or is redistribution our goal? Arpit suggests that there might be a conflict between addressing household economic insecurity and boosting spending. Once again, we see thing that good things don’t always go together.

I don’t think the public policy upshot is obvious here. I think an exclusive focus on spending is unwise, and like Mark Thoma I think it’s good for people to repair their balance sheets. But I do think we should be cautious about these issues.