There's an old saying in economics that it doesn't matter if what you say is right, what matters is if the right person says it. I was reminded of this line when I read Matt O'Brien's Wonkblog post on the success of the Fed in allowing the unemployment rate to fall below the nearly universally accepted measure of the NAIRU, without having any notable acceleration of inflation.
This is a great history that should be tattooed on the forehead of everyone involved in the current debate on how low the unemployment rate can go without kicking off a wage price spiral. Back in the mid-1990s all right thinking economists thought that the NAIRU was in the neighborhood of 6.0 percent. This meant that if the unemployment rate was below 6.0 percent the inflation rate would begin to increase. And, it would keep increasing as long as the unemployment rate stayed below 6.0 percent.
While there was some difference on the precise number (the usual range went from 5.6 percent to 6.4 percent), there was almost no dispute on the basic point. As O'Brien notes, even Janet Yellen adhered to this view, expressing concerns in 1996 that if the Fed didn't raise interest rates inflation would be a big problem. (Paul Krugman also expressed a similar view at the time.)
Thanks to the eccentricities of Alan Greenspan, the Fed did not raise interest rates. Instead it allowed the unemployment to continue to fall. It fell below 5.0 percent in 1997, it crossed 4.5 in 1998, and reached 4.0 percent as a year-round average. And inflation remained tame. The result was that millions of people had jobs who would not have otherwise. Tens of millions of workers at the middle and bottom of the wage distribution saw substantial real wage gains for the first time in a quarter century.
And, for the folks fixated on budget deficits, we saw a large surplus for the first time in decades. As much as the Clintonites like to boast of their great surpluses, the reality is that the budget would have remained in deficit if Clinton's Fed appointees (Janet Yellen and Lawrence Meyer) had gotten their way. It is only because the Fed allowed the unemployment rate to fall far lower than these folks thought wise that the budget shifted from deficit to surplus. (In 1996 the Congressional Budget Office projected a deficit of $240 billion [2.5 percent of GDP] for 2000. In fact, we ran a surplus of roughly the same amount. According to CBO, the legislative changes over this four year period went a small amount in the wrong direction.)
Anyhow, all of this should be a good reminder that the whole of the economics profession can be completely wrong on the most important issues affecting the economy. But that isn't why I brought you here today.
My main reason for doing this post is that O'Brien reminded me of one of my pet peeves about the NAIRU history. When we didn't see the predicted acceleration of inflation, the surprised economists went running around looking for explanations for why their theory had failed. This issue is discussed at some length in a book by Janet Yellen and Alan Blinder, The Fabulous Decade.
One of the explanations they give for inflation not rising is that there were changes in the measured rate of inflation that lowered the CPI relative to the actual rate of inflation. In other words, because changes in the methodology used to construct the CPI, if the true rate of inflation was 2.5 percent throughout the decade the CPI might show a 3.0 percent rate of inflation in 1994, but a 2.5 percent rate of inflation in 2000. If workers don't recognize that the way to measure the CPI has changed, then they may not adjust their wage expectations accordingly.
This means that if they were expecting wage increases that were 1.0 percentage point above the CPI measure of inflation in 1994, then they were expecting wage increases that were 1.5 percentage points above the true rate of inflation. However if they expected wage increases that were 1.0 percentage point above the measured rate of inflation in 2000, then they were expecting wage increases that were just 1.0 percentage points above the true rate of inflation. Yellen and Blinder argue that this could be a reason that inflation did not take off as they had predicted. (Their measure of inflation uses the same methodology over the whole period.)
I think there can be some plausibility to the Blinder-Yellen story, albeit not very much. The reality is that the measured rate of inflation did not change all that much during this period. The gap between the inflation rate shown by the CPI and the currently used measure peaked at around 0.5 percentage points for six years, from 1988 to 1993. It had previously been less than 0.3 percentage points. The Bureau of Labor Statistics then introduced a series of changes that lowered the gap to less than 0.2 percentage points by 1997 and to 0.0 by 2000. Insofar as the CPI provides a reference point for wage increases, these changes could have some moderating effect, but the impact would be very limited compared to the predicted inflation.
But this gets me back to the my original comment. Yellen and Blinder felt it was important to note the impact of changes in the measurement of inflation on the course of actual inflation in the late 1990s. But this is not the first time we had measurement issues with inflation.
We all know about how we had a wage-price spiral back in the bad old days of the late 1960s and 1970s. That was when our links between the unemployment rate and inflation were supposed to be really tight. Well, it turns out that this was also a period in which we had large gaps between the inflation rate as shown by the CPI and what we would now view as the actual rate of inflation using current methods.
The gaps in those years were several times larger than the gaps that concerned Yellen and Blinder. In 1969 and 1970 the gap between the inflation rates shown by the official CPI and the CPI-UX1 that we now view as more accurate were 1.0 percentage point and 0.9 percentage points, respectively. The gap reached 1.1 percentage points in 1974 and averaged almost 2.0 percentage points in 1979 and 1980, as double digit inflation was ravaging the land.
Not only were the measurement issues far larger in the late 1960s and 1970s, but the official CPI was almost certainly more important for wage setting. The unionization rate was well over 20 percent through this period. Unions would naturally look to the CPI in making their wage demands. And many wage contracts, both union and non-union, were directly linked to the CPI, meaning that any error in the measured rate of inflation would be passed on directly in wages.
In other words, if Yellen and Blinder are at all right in their view of the lower than expected inflation in the late 1990s, then the measurement error in the CPI must explain a substantial portion of the inflation in the late 1960s and 1970s. However as far as I know, no one has looked into this. (I have an old working paper with EPI that did find a strong link.)
Anyhow, one day someone important will look at the impact of measurement error on inflation in the 1960s and 1970s. I suspect that we will then find a less direct link between unemployment and inflation than is currently believed.