Recently, I published a piece in the Washington Post that highlighted growing challenges facing the Federal Reserve (and other central banks) as they try to achieve their inflation objectives. The conclusion was that the Phillips Curve—a workhorse model for understanding inflation—is no longer a reliable tool for that purpose. In what follows, I extend the discussion, noting that the Phillips Curve joins a series of discarded approaches unable to explain or forecast inflation.
That is problematic. Without a fundamental understanding of inflation dynamics, central banks are increasingly reliant on the incoming data to guide their decisions. Data dependency, however, comes with risks, particularly because it increases the odds that monetary policy will be late to respond to economic events. Without models to guide them, central banks are at risk of making bigger policy mistakes.
A history of misunderstanding inflation
To begin, it is important to realize that the economics profession has had a tough time for decades in understanding what drives inflation, much less forecasting it.
Milton Friedman famously said that inflation is ‘always and everywhere’ a monetary phenomenon. Yet bursts of inflation can occur for completely different reasons, for example owing to energy or commodity supply shocks. To be sure, persistent inflation requires persistent excess money creation, but there are too many important sub-periods when massive and rapid money growth failed to produce inflation (e.g., Japan for most of the past four decades, or the US following the global financial crisis) to conclude that rapid money growth always causes high inflation.
The monetarist explanation for inflation starts on firm ground, namely the truism that all money in circulation multiplied by its velocity of circulation must equal all nominal transactions in the economy. And if the economy is at full employment and money velocity is stable, then inflation will be proportional to money growth. It is also generally true that where money growth finances budget deficits in countries with checkered histories of price stability, rampant inflation typically ensues.
But in the US and most other advanced economies, monetarist approaches to inflation have largely come up short. In part, that is because the two assumptions of full employment and stable money velocity are frequently violated. In particular, if money demand increases, as it has in Japan since 1990 or frequently does during periods of great economic or financial stress, then rapid money growth not only won’t cause inflation—it is required to prevent deflation!
Nevertheless, there was a time in the late 1980s when many economists, including those at the Fed and various European central banks (e.g., the German Bundesbank or the Swiss National Bank) felt that money velocity was relatively predictable. What ensued was a brief fascination with so-called P-star models (above all at the Fed), based on monetarist approaches to central banking. But in less than a decade those models broke down. They failed to explain broad inflation and growth outcomes, and so they were abandoned.
In their place arrived variants of the Phillips Curve model, which dates its lineage to A.W. Phillips, whose empirical work in the 1950s noted an apparent tradeoff between unemployment and wage inflation, which could spill over into broader price inflation.
The Phillips Curve has appealing logic. For one, wages and employment benefits are the single largest cost for most businesses. Accordingly, when the unemployment rate is low enough to create conditions of rising wages, higher labor costs may prompt companies to raise prices, leading to high inflation. The Phillips Curve is also consistent with the idea that price pressures increase whenever slack in the economy recedes to low levels.
The policy implication of the Phillips Curve is to fight high inflation with high interest rates in order to weaken overall demand and create slack in the economy. Rising joblessness then curbs wage growth and hence price inflation. The aim, simply put, is to throw enough people out of work to slow wage demands that are thought to be the cause of price inflation.
That thinking has recently been evident in the Fed’s approach to fighting inflation today. While carefully avoiding specific reference to the Phillips Curve, Fed Chairman Powell nevertheless made that case at a Brooking Institute speech last November. Powell said,
“We are tightening the stance of policy in order to slow growth in aggregate demand. Slowing demand growth should allow supply to catch up with demand and restore the balance that will yield stable prices over time. Restoring that balance is likely to require a sustained period of below-trend growth.”
He added that:
“In the labor market, demand for workers far exceeds the supply of available workers, and nominal wages have been growing at a pace well above what would be consistent with 2 percent inflation over time. Thus, another condition we are looking for is the restoration of balance between supply and demand in the labor market.”
In no uncertain terms, Powell was emphasizing that inflation resulted from too much demand relative to supply, labor shortages and excessive wage gains. In noting that lower inflation required the ‘restoration of balance between supply and demand in the labor market’, Powell was referencing a Phillips Curve relationship between low unemployment and high inflation.
But now, only two months later, things look very different. Goods prices inflation is rapidly slowing. So, too, are housing costs, suggesting that the biggest component of consumer prices—shelter—is headed lower. Even wage inflation is coming down. Crucially, all those outcomes are taking place without ‘a restoration of balance between supply and demand in the labor market’. That was dramatically underscored by the January employment report, which recorded a surge in US job formation and a drop in the unemployment rate to fresh cyclical and 64-year lows.
Simply put, in sharp contrast to what Powell said in November, inflation is falling without a rise in unemployment.
Um, so much for the Phillips Curve.
Let’s be clear. Without a shred of doubt, falling inflation without rising unemployment is a fantastic development. I like to call it ‘immaculate disinflation’. And if present trends continue, the landing for the economy could be wonderfully soft.
But that still leaves the Fed and other central banks in an awkward position. If neither monetarist nor Phillips Curve approaches explain inflation dynamics, what’s left to guide their policy decisions?
The answer, for now, is ‘data dependency’. Central banks will monitor the incoming data and respond accordingly. If inflation continues to decline toward their targets, policy pauses or even pivots will be in order. If, for some reason, inflation re-accelerates, rate hikes will be back on the table.
But data dependency without a model to offer guidance for the future is akin to driving a car at high speed in dense fog. The driver can only respond to objects that come into sight at the last minute, with little or no visibility about what lies just ahead.
Driving fast is not advisable when visibility is low. Similarly, making monetary policy without reliable tools offering guidance about the future is risky business. Those risks are compounded by the knowledge that today’s decisions may have important consequences far into the future. Stepping on the monetary brakes (or accelerator) only when the evidence arrives is not optimal for the guardians of monetary and macroeconomic stability.
Is there an alternative? Is there a better model for inflation? Not at the moment. But with existing models wanting, the economics profession and central bankers must explore new ways to improve our understanding of inflation dynamics and the business cycle. A lot depends on it—nothing less than the stability of the business cycle.