This is an excerpt from our most recent Economic Outlook report. To access the full PDF, please click here.
The adherence of central bankers around the world to the Phillips Curve raises a big question. Is that inverse data relationship -- between higher consumer inflation as a consequence of a lower overall unemployment rate -- actually as robust as assumed?
Since William Phillips, a New Zealand Economist, discovered an inverse relationship between the unemployment rate and wage growth in 1958, the concept of the so-called Phillips Curve has gone through a number of transformations over the years. It was often known in rougher terms as an economic growth and inflation relationship. The version that is mostly known today: a loose logic tradeoff between lower unemployment rates and higher core consumer price inflation.
That relationship (with all of its core and seasonal variations) makes intuitive sense. Lower levels of overall unemployment should increase average wages (i.e. the price it takes to fill vacant positions). Yet, it does appear more far-fetched (given numerous frictions in the real world) to expect a clean subsequent increase in labor market prices to lift overall consumer inflation.
In sum, it is not quite clear to what degree this theory is actually supported by empirical evidence. Except for the period during Phillips’ original discovery of the relationship, the correlation seems to fluctuate strongly over the following business cycles. For instance, the seeming lack of any relationship in the most recent cycle years did not provide any support.
The following graph demonstrates how the U.S. unemployment rate and core inflation (as measured by PCE) develops over time. Unemployment follows closely the business cycle pattern. Meanwhile, the consumer inflation rate seems to move in an almost erratic fashion.
A Phillips Curve relationship can become quite visible over some time periods. If an observer focused on a certain sub period -- such as the beginning of 1960s -- the relationship does appear indeed negative. However, look at the beginning of the 1970s. The pattern appears to be the reverse. Both variables seem to move in parallel. So, the Phillips Curve becomes almost similar to the 'Loch Ness Monster.’ That is, if you have the perfect time frame, and look at it from the right angle, you can see it!
Across decades, with different proxies, other Phillips Curve-type relationships can be clearer. For example, U.S. consumer Inflation follows a similar pattern to underlying U.S. long-term interest rates. This makes sense. Long-term rates should incorporate underlying inflation expectations. Indeed, the graph below shows. Higher PCE and higher U.S. 10-yr yields tracked one another until the early 1980s, and then reversed:
Core consumer inflation rates are affected by a number of internal frictions and dislocations, such as labor policies (e.g. the presence or lack of unions). These pressure the wage setting behavior of firms, and in turn, the ability of a lower unemployment rate to translate into higher wages and subsequently higher inflation.
It appears to us, therefore, that a simple model of higher U.S. consumer inflation accompanying a lower U.S. household unemployment rate will fail to explain a lot of the variation in U.S. consumer price inflation over time.
Therefore, without strong empirical evidence to suggest otherwise, we expect the Phillips curve relationship to remain shaky. Hence, forecasting U.S. consumer inflation across the next 2 years remains difficult.