September 22, 2014
Robert Samuelson devoted his column to discussing the argument of Northwestern University economist Robert Gordon, who argues that we are destined for a prolonged period of slow growth. Samuelson argues that this could lead to major conflicts over distribution since people will not be able to enjoy rising living standards due to growth.
There are several points worth noting about Gordon’s argument. First our ability to predict productivity growth has been virtually zero. There was a huge slowdown in productivity growth in 1973, a pickup in 1995, and possibly another slowdown in 2005. The profession completely missed the slowdown in 1973 and even forty years later there is no universally accepted explanation of why it occurred. The 1995 speedup also caught most economists by surprise, although there is general agreement that it was due to the spread of computers and the Internet.
The 2005 slowdown is not at all universally accepted. While it could mark the end of the 1995 speedup, it could just be due to the weakness of demand following the collapse of the housing bubble. Here also, no one predicted the slowdown. Given this track record, it is reasonable to question the accuracy of Gordon’s or anyone’s predictions about productivity growth over the long-term future.
It is also important to point out that this view is 180 degrees at odds with the robots taking all our jobs view. The fact that both views can be taken seriously within the economics profession speaks to the state of economics. This would be like a person going to a doctor for a check-up, with the doctor concluding that the patient is seriously obese and must immediately begin a strict diet and exercise regimen. The patient then goes to another doctor for a second opinion. This doctor is concerned about the patient being too thin and prescribes a high calorie diet to allow the patient to put on weight. This is the state of economics’ ability to predict productivity.
There are a few other points worth noting. First, the comparison in Samuelson’s piece of projected growth rates to growth in the 1950s and 1960s is somewhat misleading. The population was growing more rapidly in the 1950s and 1960s as the country was experiencing the baby boom. It is per capita growth, not total growth that matters for living standards. If growth slows in line with slower population growth, this does not hurt living standards. In fact, slower population growth would be associated with an improvement in living standards insofar as it means less stress on the natural environment and the physical infrastructure.
The piece also presents productivity growth as purely a supply-side question. This is not true. Firms have more incentive to expand and adopt new technologies in a period of rapid growth which may create limits on its ability to hire workers and get other essential inputs. The weak productivity growth we have seen since 2005 can be in part explained by the fact that many workers have been forced to accept low paying and low productivity jobs in sectors like restaurants and retail. If there was increased demand in the economy, these workers would move into higher productivity jobs and many of the low productivity jobs would disappear, raising overall productivity.
Finally, it is worth noting that the terms of any possible tradeoff between growth and inequality would change if we are destined to see a long period of very slow productivity growth. The conventional argument is that high tax rates will discourage people from innovating. If Robert Gordon is correct, and there is very little to be gained through innovation in any case, then we need have little fear about discouraging innovation with higher tax rates.
The arithmetic on this is straightforward. If the baseline productivity growth rate is 2.0 percent and a high tax policy reduces this by one third, it would have a substantial impact on living standards. After thirty years productivity would be 49 percent higher instead of more than 81 percent higher. On the other hand, if the baseline rate of productivity growth is just 0.6 percent, then reducing it by a third would lower cumulative productivity growth over a thirty year period from just under 20 percent to just under 13 percent. The implication is that if Gordon is right, we have much less reason to fear that high taxation will crimp economic growth.
Comments