Arnold Kling  

Notes on the Phillips Curve

Michael Mann: I Don't Argue... The Awful Mill...

1. The original version expressed a trade-off in terms of the level of inflation vs. the level of the unemployment rate. From about 1953-1968, the trade-off was approximately: inflation + unemployment = 7 percent.

2. In the 1970's Phillips Curve modelers "took another derivative." They said that the relationship was between the change in the rate of inflation and the level of the unemployment rate.

3. Also in the 1970s, there emerged a split in the causal interpretation of the Phillips Curve. The Keynesian interpretation was that the causality runs from unemployment to the change in the inflation rate. When unemployment is high, inflation declines. When unemployment is low, inflation rises. The monetarist, or Chicago, interpretation was that the causality runs from a change in the inflation rate to the level of unemployment. When inflation goes up, unemployment is low. When inflation goes down, unemployment is high.

4. James Tobin, in his 1972 presidential address for the American Economic Association, offered a simple story for the Phillips Curve. Suppose that we have two industries, and the wage in industry A needs to rise relative to industry B in order to reach a new full employment equilibrium. If nominal wages are flexible, no problem. If nominal wages are sticky downward, then the only way to get the relative wage to its proper level is for nominal wages in industry A to rise. Thus, to get the industry shift to occur and avoid unemployment, the economy needs inflation.

5. In the most recent recession, my impression is that private sector wages were not completely sticky downward. Public sector wages have been sticky downward, however.

6. I think that if you do not assume downward stickiness of wages, the recent data are difficult to explain. That is, if you just think there is a relationship between the change in the inflation rate and the level of unemployment, then by now the inflation rate in the U.S. ought to be well below zero. However, with downward stickiness of wages, you can argue that the high unemployment is not having much effect on the inflation rate.

7. I believe that nominal wages are sticky downward on a case-by-case basis. That is, the wage of worker X in firm Y is unlikely to be to be cut. However, worker X can get a wage cut by losing a job and switching to a different firm. Moreover, firm Y can cut its wages by lowering the wage it pays to new workers, or by outsourcing some of its work to a lower-wage firm. So I am not so convinced that the case-by-case stickiness implies aggregate stickiness.

8. Since 2008, we have had very high unemployment, and the reduction in inflation has been miniscule. I see this as the worst performance for the Phillips Curve since the 1970s.

Comments and Sharing

CATEGORIES: Macroeconomics

COMMENTS (6 to date)
John Hall writes:

Regarding point 8, I tend to agree. However, most of these Philips Curves are estimated relative to the natural rate of unemployment. So if you're willing to do something fancy (like a Kalman filter) to find the natural rate of unemployment, then it will turn out that your PC is fit well. Of course, this doesn't really provide an answer, since you're effectively solving for what you want to prove.

The other side to this is that the New Keynesian Philips Curve incorporates expected inflation. Since inflation expectations are contained, it would suggest that elevated unemployment may not have moved NKPC inflation forecasts that substantially.

Costard writes:

4. As always the question is, at what cost? Sticky wages serve a purpose: equilibria can be ephemeral. Adjusting to a temporary situation involves a lot of wasted motion. The movement of wages seems fluid enough to an economist, but to a business there are costs involved with a change in the business plan. And how can one be sure that inflation will resolve the imbalance, and not simply reinstate it at a higher price level? There is an assumption at work. Additionally, since an imbalance is itself a (presumably passing) equilibrium, how can we be certain we are choosing a more constant state for a less constant one? An active hand on the wheel doesn't necessarily result in a smooth ride.

7. I think the second half of this is on the money. Wage is not the only benefit to workers but it is the last one to be cut. Other costs are squeezed first... break room coffee, company cars, etc. Perhaps more is asked of employees. With the level of specialization in many industries, and the additional costs imposed by regulation and modern liability, hiring and firing are probably less less attractive relative to other solutions, than they might have been 50 years ago.

8. Nevertheless the man who pretends to know how to fix something, is the man who gets the job.

Glen Smith writes:


"Wage is not the only benefit to workers but it is the last one to be cut."

This is suggestive of one of the most effective tool to sell an employee benefit package back when I was in that business. You can manipulate a benefits package to lower the employee compensation without them realizing it.

Joe Cushing writes:

Sticky does not mean completely stuck. Sticky fits the description of sticky on a case by case basis. If you can't lower individuals wages, going through the process of laying people off and hiring people at lower wages is a lower and sticky process.

Dan Carroll writes:

The rigidity of wages certainly varies by industry and sector - government wages are more rigid due mostly to unionization. However, it also varies by income level: high wage individuals usually have variable salaries (bonuses, commissions, stock-based compensation, profit-based payouts as in partnerships, or self-employment), while low-er wage individuals tend to have fixed salaries. Also, lower wage individuals tend to have personal costs that are more fixed (food, shelter/mortgage, transportation), and therefore are less able to take a wage cut easily. Other rigidities include fixed employee costs such as health benefits, which tend to be a higher percentage the lower the wage (until those benefits are eliminated). Therefore, the fixed costs impose a kind of leverage on cash wages: small cuts translate into big cuts in living standards. Non-wage benefits are much smaller as a percent of employee costs for high wage individuals.

As we know, unemployment is higher among lower wage individuals, both structurally and in the recent recession.

Pension costs introduce a different dynamic into wage rigidity, ultimately increasing rigidity. In the public sector, pension commitments are difficult to scale back and even have upward biases relative to inflation. It is a type of debt that introduces leverage into the equation.

Thus, wage rigidity may primarily be a function of leverage: debt, pension, benefits, and basic living costs, as well as union contracts. Morale and psychology play a role too, but I suspect those are temporary and subject to opportunity cost analysis by the employee.

Floccina writes:

How about the idea that if everyone with a job is overpaid then their is not enough money circulating to pay others so that full employment cannot be reached. The idea would be that after the deceleration in NGDP in 2008 people are more overpaid than usual.

Comments for this entry have been closed
Return to top