Inflation Puzzles

Inflation in the United States -- as measured by the annualized increase in the consumer price index -- was a paltry 0.2 percent in September. This is significantly below the 2 percent informal inflation target of the Federal Reserve, the central bank of the United States. Such low inflation came as a surprise to many. No less a figure than Jeffrey Lacker, president of the Federal Reserve Bank of Richmond, speaking in Philadelphia last Friday (November 1), admitted to being surprised by how low and stable inflation had lately been despite the Fed's vigorous program of asset purchases (paid for with mountains of newly printed cash). For Lacker, the idea that money printing by central banks inevitably leads to rampant inflation was an article of faith, not significantly less worthy of trust than the idea that the sun rises in the east.

But the Great Recession has been a stern teacher. It has taught us that the connection between the rate at which the Fed prints (and spends) money and the rate at which prices rise breaks down when unemployment is above the rate consistent with full employment. Lacker should have relied on an alternative theory: the expectations-augmented Phillips curve.

This theory says that inflation in the short run consists of three parts: expectations-driven inflation, demand-pull inflation, and cost-push inflation. First, when firms expect rapid inflation, they expect the prices they'd have to pay for raw materials and labor to rise rapidly, and consequently they raise their own prices rapidly in order to remain profitable. Second, when -- as a consequence of the heavy pull of demand -- the economy's production is close to its potential, and unemployment is close to the rate consistent with full employment, incipient labor scarcity raises wages rapidly, which in turn leads to rapidly rising prices. And third, if droughts, other productivity failures, and/or rising prices of imported oil lead to rising production costs, prices would again have to rise rapidly in order to ensure profitability. [See this.]

The Phillips curve theory does not say that money printing and asset purchases by a central bank cannot cause inflation. It only says that if money printing is to cause inflation, it would first have to instigate either expectations-driven inflation, or demand-pull inflation, or cost-push inflation. Therefore, the theory says, there is no point in watching the rate at which the Fed is printing money in the hope of predicting inflation. One should instead watch data on inflation expectations, the rate of economic growth, unemployment, oil prices, and so on.

A new NBER working paper ("Is the Phillips Curve Alive and Well After All? Inflation Expectations and the Missing Disinflation", NBER Working Paper No. 19598, http://www.nber.org/papers/w19598) by Olivier Coibion and Yuriy Gorodnichenko argues that with a tweak or two the Phillips curve can explain inflation during the Great Recession quite well. Unlike Lacker who relied on the money-printing-causes-inflation theory and was surprised that inflation during the Great Recession was so much lower than he had expected, Coibion and Gorodnichenko had relied on the Phillips curve and were surprised that inflation was higher than they had expected.

But all was not lost for the Phillips curve. Coibion and Gorodnichenko found that when traditional measures of inflation expectations (such as the "TIPS spread" obtained from financial markets or surveys of professional forecasters) are replaced by the measure of inflation expectations in the Michigan Survey of Consumers, the Phillips curve does well in explaining inflation. During the Great Recession, households surveyed by the Michigan Survey consistently expected higher inflation than the financial markets and the professional forecasters.

Coibion and Gorodnichenko also propose an interesting explanation of why households consistently expected higher inflation than bond traders and forecasters. They argue that households tend to be keenly aware of oil prices, which had risen sharply since 2009. And this sensitivity to prices at the gas pump influences households' expectations of inflation. (As a fan of behavioral economics, I see here the presence of the concept of 'salience.')

More provocatively, Coibion and Gorodnichenko argue that it was actually a stroke of luck that rising oil prices kept inflation expectations -- and, therefore, inflation itself -- above dangerously low levels that could have pulled the US economy into deflation. (In a tangential remark in a soon-to-be-published paper, Sebastien Buttet and I briefly pointed out the other side of this very coin: falling oil prices -- normally a reason for rejoicing -- can push a sluggish economy into a deflationary spiral and cause a lot of mayhem.)

Finally, Coibion and Gorodnichenko made me wonder about the wisdom of the Fed's reliance on the so-called core inflation measure that strips out food and energy prices as a guide to policy making. The Fed's argument is that it needs to predict inflation nine months or so down the road because any policy measure adopted today would take nine months or so to take effect, and, as food and energy prices are extremely volatile, a consumer price index that excludes them would be better at predicting inflation nine months ahead. The Fed's argument makes sense. But Coibion and Gorodnichenko's paper makes me worry a bit. Food and energy prices are highly salient for households. As a result, food and energy prices are likely to have a huge impact on inflation expectations. Therefore, today's food and energy inflation is likely to drive future inflation through the expectations-driven inflation in the Phillips curve. Maybe the Fed should take another look at its position on the usefulness of paying attention to food and energy prices.

To wrap up, the expectations-augmented Phillips curve is a pretty good guide to short-run inflation. Mr. Lacker should take note.