Thursday, July 9, 2009

Capacity utilization and inflation

As one who believes that inflation is a monetary phenomenon, I have argued that the Fed's model of how inflation works, which is based on the same principles as the Phillips Curve theory of inflation, is flawed. The Phillips Curve theory posits an inverse relationship between unemployment rates and inflation—in order to reduce inflation an economy needs to accept higher unemployment, and vice versa. The Fed's model (which enjoys broad support in the financial market) assumes that weak economic growth leads to idle resources (e.g., extra capacity, higher unemployment), and those idle resources put downward pressure on prices because workers and producers have to lower their price if they want to stay in business. The corollary to this observation is that strong growth results in fewer idle resources, and so workers and producers can demand higher prices because resources become scarcer.

In defense of those models, many observe that inflation typically falls during and after recessions, and typically rises in advance of recessions; recessions create lots of idle resources, and that is what pushes prices down. One monetarist argument against this observation is that recessions are typically caused by a tightening of monetary policy, so it is only natural that inflation should subsequently fall, while economic booms are typically associated with relatively easy monetary policy, which then allows inflation to rise. Monetarists therefore say that the Fed and the Phillips Curvers are confusing correlation (declining inflation and economic weakness/low capacity utilization) with causation (declining inflation is not caused by increased idle resources, but rather by tight money).

As this chart shows, there was a very strong correlation between capacity utilization and inflation from the early 1970s through the early 1980s. (Note that the red line, capacity utilization, is shifted 17 months to the right, so the fact that the two lines move tightly together suggests that the level of capacity utilization predicts inflation by 17 months. Note also that I am using core inflation, to eliminate the impact of dramatic swings in energy prices.) This correlation breaks down almost completely, however, starting in the mid-1980s. Why? My answer is that it's due to the Fed's zeal in combating inflation. The Fed was very slow to react to signs of inflation in the 197os, always tightening too little and too late. The Fed became more proactive in the 1980s and thus was much quicker to tighten in response to a stronger economy. Indeed, the Fed became aggressively tight in the latter half of the 1990s, fearing that the economy was so strong that it was "overheating." As a result, inflation fell throughout the 90s even though economic growth was accelerating and capacity utilization was very high.

The positive correlation between capacity utilization and inflation began to return, however, in the 2000s. Why? Because the Fed reverted to being reactive, much as they were in the 1970s. As a result of being too tight in the late 1990s, deflationary pressures surfaced in the early 2000s that were then met by aggressive easing. The Fed was then very slow to tighten (2003-2005), even though inflation was rising, and now they are in panicked easing mode because the economy is perceived to be extremely weak and thus deflation risk is assumed to be very real.

As a counterpart to my interpretation of events, I suggest you have a look at a similar post on EconompicData which has a chart that paints a very different picture than my chart. He argues that the change in capacity utilization has always been a good predictor (by 6 months) of inflation. I'm not all that impressed by the fit of the two lines on his chart (sometimes they move together, and sometimes they don't), and I don't think there is a logical reason to expect a strong fit in the first place.

Here's why: Idle resources and high unemployment may indeed depress the prices of some things, and may cause some workers to accept lower wages. But inflation is a condition in which all prices rise, not just some. So whatever reduction in price pressures we see as a result of rising unemployment and falling capacity utilization are not necessarily going to result in all prices falling. Sometimes a decline in capacity utilization will result in falling inflation, but not always. What's really driving inflation is monetary policy, as I've argued above.

The important thing to focus on today is that while the level of economic activity overall has fallen rather significantly from where it was a year ago, the amount of money circulating in the economy has risen significantly. Money is now abundant, whereas goods and services are relatively scarce. When the public's demand for money declines—something I think may already be underway—then we will have a surplus of money and a reduced supply of goods and services, and that is the classic recipe for rising inflation.

The proof will of course be in the pudding. If core inflation doesn't decline significantly in the next 17 months, then the Fed's theory of inflation will be left with very little in the way of empirical support, and it is already skating on thin ice.


DaleW said...


Do you know roughly what capacity is being measured in capacity utilization? Manufacturing is such a small part of the US economy (mining and manufacturing wage and salary accruals = 12.6% of national wage and salary accruals, according to NIPA table 6.3D. for 2007) while Services PCE = 44% of GDP vs. in Q1 2008 vs. 28.6% 40 years ago (NIPA table Table 1.1.5.). Is the capacity of nurses, doctors, software engineers, analysts and PMs, etc. really measured for capacity utilization? Also, I would guess the capacity and utilization components in today's economy rely much more than ever on global capacity and global utilization, not US capacity and US utilization. I would be interested in yours thoughts.

Anonymous said...

Hey Scott,

You're absolutely right. Inflation (and deflation) has always been a monetary phenomenon. And just like the classicalists (now renamed supply siders) I look to gold as a real time indicator of how far off the dollar is from its 10year moving average. If the dollar price to gold stays at these levels ~$900, inflation will be 5.5% a year for the next 10-15years. Of course, we know that CPI will never register that as its been so hacked to pieces by the government that it is a shadow of its former self (watch out with your TIPS, it wont register the full inflation).


Scott Grannis said...

Dale: Capacity utilization is a nebulous concept to begin with, and all the Fed can do is estimate what it is. Also it now represents only about 30% of the economy (i.e., goods production). It is however a good proxy for "slack" in the economy, but I wouldn't want to base policy decisions on it.

Scott Grannis said...

PacificAllen: I share your belief in the importance of gold, but I think that the CPI does do a pretty good job of measuring inflation over longer periods. In fact, I agree with the Boskin study that says the CPI tends to overstate inflation somewhat--about 0.5% per year.

Mark A. Sadowski said...

It's important to remember that inflation is always and everywhere not a monetary phenomenon at all but rather a phenomenon related to labor underutilization. You'll see that I'm right in the months ahead.