The steady expansion of employment and reduction in unemployment since the recovery from the financial crisis of 2008 and the subsequent Little Depression, even as inflation remained almost continuously between 1.5 and 2% (with only a slight uptick to 3% in 2011), has led many observers to conclude that the negative correlation between inflation and unemployment posited by the Phillips Curve is no longer valid. So, almost a month ago, Greg Mankiw wrote New York Times Sunday Business Section defending the Phillips Curve as an analytical tool that ought to inform monetary policy-making by the Federal Reserve and other monetary authorities.
Mankiw starts with a brief potted history of the Phillips Curve.
The economist George Akerlof, a Nobel laureate and the husband of the former Federal Reserve chair Janet Yellen, once called the Phillips curve “probably the single most important macroeconomic relationship.” So it is worth recalling what the Phillips curve is, why it plays a central role in mainstream economics and why it has so many critics.
The story begins in 1958, when the economist A. W. Phillips published an article reporting an inverse relationship between unemployment and inflation in Britain. He reasoned that when unemployment is high, workers are easy to find, so employers hardly raise wages, if they do so at all.
But when unemployment is low, employers have trouble attracting workers, so they raise wages faster. Inflation in wages soon turns into inflation in the prices of goods and services.
Let’s pause for a moment and think about that explanation. If we translate it into a supply-demand framework in which the equilibrium corresponds to the intersection of a downward-sloping demand for labor curve with an upward-sloping supply of labor curve. The equilibrium is associated with some amount of unemployment inasmuch as there are always some workers transitioning from one job to another. The fewer and the more rapid the transitions, the less unemployment. The farther to the right the intersection of the demand curve and the supply curve, the more workers are employed and the fewer are transitioning between jobs in any time period.
I must note parenthetically that, as I have written recently, a supply-demand framework (aka partial equilibrium analysis) is not really the appropriate way to think about unemployment, because the equilibrium level of wages and the rates of unemployment must be analyzed, as, using different terminology, Keynes argued, in a general equilibrium, not a partial equilibrium, framework. But for ease of exposition, I use the partial equilibrium supply-demand paradigm.
Mankiw’s assertion that when unemployment is low, employers have trouble attracting workers, and therefore have to raise wages to hire, or retain, as many workers as they would like to employ, could be true. And that is the way that Mankiw wants us to focus on the relationship between wages and unemployment. Mankiw is focusing on the demand side as an explanation for low unemployment. But low unemployment could also reflect the eagerness of workers to be employed, even at low wages, so that workers quickly accept job offers rather than searching or holding out for better offers.
This is a classic issue in empirical estimates of demand. If high prices are associated with high output, does that mean that the data show that demand curves are upward-sloping, so that when suppliers raise price, customers increase want to buy more of their products? Obviously not. Suppliers raise their price because at low prices their customers are want to buy more than producers want to sell. So to estimate a demand curve, there must be some way of identifying factors that cause the entire demand curve to shift.
The identification problem in estimating demand curves has been understood since the early years of the 20th century. It is incredible that economists, especially one as steeped in the history of the discipline as Mankiw, talk about the Phillips Curve as if they had never heard of the identification problem.
When you estimate a demand curve without being able to identify shifts in demand, you are not estimating a demand curve, you are estimating a reduced form that combines — and fails to identify or distinguish between — both demand and supply. The Phillips Curve is a reduced form that captures both factors that affect the demand for labor and the supply of labor, though as I mentioned above, talking about the demand for labor and the supply of labor in the normal partial equilibrium sense of those terms is itself misleading and inappropriate.
What is unambiguously true, however, is that whatever else the Phillips Curve may be, it is a reduced form and not a deep structural relationship in an economy. It therefore is of little if any use in helping policy makers figure out whether to tighten or ease monetary policy.
For centuries, economists have understood that inflation is ultimately a monetary phenomenon. They noticed that when the world’s economies operated under a gold standard, gold discoveries resulted in higher prices for goods and services. And when central banks in economies with fiat money created large quantities — Germany in the interwar period, Zimbabwe in 2008, or Venezuela recently — the result was hyperinflation.
But economists also noticed that monetary conditions affect economic activity. Gold discoveries often lead to booming economies, and central banks easing monetary policy usually stimulate production and employment, at least for a while.
The Phillips curve helps explain how inflation and economic activity are related. At every moment, central bankers face a trade-off. They can stimulate production and employment at the cost of higher inflation. Or they can fight inflation at the cost of slower economic growth.
The Phillips curve, a reduced form, a mere correlation revealing no deep or necessary structural relationship between inflation and unemployment, explains nothing. It merely reflects the fact that, under a certain set of conditions, monetary expansion is associated with increased output and employment, and, accordingly, with reduced unemployment. And, under a certain set of conditions, monetary contraction is associated with reduced output and unemployment, and, accordingly, increased unemployment. We know now – and have long known — all about those relationships, without the Phillips Curve.
But under other conditions, high inflation may be associated with non-monetary factors (negative supply shocks) causing falling output and employment. And under still other conditions, low inflation may be associated with rising output and employment. There is no deep structural reason causing low unemployment to be incompatible with low inflation or causing high unemployment to be incompatible with high inflation. To suggest that the Phillips Curve is somehow a necessary relationship rather than a coincidental correlation between inflation and unemployment is a shockingly superficial reading of the evidence betraying an embarrassing misunderstanding of elementary theory.
Mankiw seems to be vaguely aware of his own confusion when he writes the following.
Today, most economists believe there is a trade-off between inflation and unemployment in the sense that actions taken by a central bank push these variables in opposite directions.
Mankiw’s qualification that the trade-off between inflation and unemployment reflects the tendency of rapid monetary expansion to cause output and employment to expand (at least temporarily) and prices to rise is a tacit admission that there is no necessary trade-off between inflation and unemployment. What we refer to as a Phillips Curve is simply the tendency for changes in monetary policy to affect inflation and unemployment in opposite directions, not a necessary structural relationship between inflation and unemployment. This is not rocket science. I can’t understand why Mankiw has trouble understanding that the fact that monetary policy may cause unemployment to rise when it causes inflation to fall and vice versa seems is not the same thing as a necessary inverse relationship between inflation and unemployment.
As a corollary, they also believe there must be a minimum level of unemployment that the economy can sustain without inflation rising too high. But for various reasons, that level fluctuates and is difficult to determine.
The level of unemployment that can be sustained without inflation is both unobservable and subject to change. Monetary policy can stimulate a rapid reduction in unemployment when unemployment is clearly higher than normal, as unemployment falls ongoing monetary expansion carries a risk of increasing inflation. But that doesn’t mean that unemployment cannot continue to fall without triggering an increase in inflation.
Mankiw concludes with the following bit of advice.
The Fed’s job is to balance the competing risks of rising unemployment and rising inflation. Striking just the right balance is never easy. The first step, however, is to recognize that the Phillips curve is always out there lurking.
That is just silly. The risk of increasing inflation is there with or without recognizing that the Phillips curve is out there lurking. To avoid the risk of inflation in a responsible way means using monetary policy to keep the rate of increase in nominal GDP within a reasonably narrow band providing enough room for the normal rate of growth in output with a rate of inflation sufficient to keep nominal interest rates moderately low, but substantially above zero. You don’t need a Phillips curve to figure that out.
Uneasy Money 