The Sanders case for more spending and faster growth

THE standard case for fiscal stimulus goes like this. In a recession, aggregate demand falls -- everyone is afraid to spend and instead just hoards cash. If the government spends it can prompt people to buy more things with the money they get from the government, which raises demand and gets the economy working again. Of course, this costs money, but the government can borrow the money and pay it back the next time the economy is running on all cylinders.

Stimulus, in other words, is part of a short-term strategy to fill in the gaps in the economy caused by the business cycle. That's the basic idea promoted by the inventor of the concept, John Maynard Keynes. It is also the story embraced by most modern proponents of stimulus, such as Paul Krugman. However, in the recent debate surrounding the economic proposals of presidential candidate Bernie Sanders, a small number of economists have started suggesting a very different justification for stimulus. Their idea: Stimulus does something more fundamental to the economy by raising long-term productivity.

It started with a paper by economist Gerald Friedman of the University of MassachusettsAmherst, which analyzed Sanders' economic plans. Sanders wants a lot more government spending; Friedman says that this spending will raise growth so much that the proposals will pay for themselves. Though Paul Krugman, Austan Goolsbee and other economists have ridiculed this plan as beingimplausible -- the mirror image of failed Republican promises that tax cuts would be self-financing -- there have been a number of defenses as well, including some from very unlikely sources. If Friedman and others are right, it would upend most of mainstream macro, and would force a dramatic reconsideration of economic policy.

But Friedman's paper seems far-fetched because the normal action of stimulus -- putting unemployed people back to work -wouldn't be nearly enough to create the kind of growth Friedman projects. In addition, we would need a huge boost to the growth rate of productivity. Usually we think of productivity gains as coming mainly from technological advancements, something that is very hard for government policy to affect.

The notion that fiscal stimulus, in addition to raising employment, also boosts productivity growth was first suggested in 1949 by a Dutch economist, Petrus Johannes Verdoorn. According to what's known as Verdoorn's law, all you have to do is boost gross domestic product growth -- for example, by fiscal stimulus -- and productivity will soar as well. Friedman explicitly assumes in his paper that you can do this. John Jay College professor J.W. Mason has long entertained the possibility. The idea has even garnered support from Narayana Kocherlakota, former president of the Federal Reserve Bank of Minneapolis. Kocherlakota -- a famously open-minded economist who changed his view about monetary policy in recent years -- writes that there is "an empirical basis" for Verdoorn's Law:

The most striking evidence [ for Verdoorn's law] comes from the Great Depression in the US. Total factor productivity fell dramatically at the beginning of the Depression...Over the following three years, in conjunction with the various forms of demand stimulus undertaken by the Roosevelt administration, TFP grew more than 5% per year faster than normal. This super-normal growth rate of TFP was a key contributing factor to the near double-digit annual growth in real GDP from 1933-37.

This seems to be illustrative of a more general and systematic pattern...a fall in the unemployment rate of 1 percentage point is empirically associated with a 0.9 percentage point increase in TFP growth. Economists have long noted the rapid productivity growth during the Depression. This is usually considered to be coincidental -- the standard story is that humanity just happened to invent a bunch of useful stuff during the '30s. But Verdoorn's law says that no, it was Roosevelt and his stimulus that raised productivity. If that's correct, then stimulus becomes a much more important tool, since its growth-boosting power would be much larger than commonly assumed even by stimulus proponents like Krugman.

But there are a couple of big problems with Verdoorn's law. First, correlation doesn't equal causation; as we found in the '70 with the Phillips curve, which said that as unemployment rose inflation would fall, trying to treat statistical correlations as laws of economics often fails. It's obviously possible for fast productivity growth to cause fast economic growth, rather than the other way around; if a lot of new technology gets invented, business will want to invest and use it to take advantage of the business opportunities that result.