The Phillips Curve Is Dead: Why Lower 'Unemployment' No Longer Causes Inflation

by: Long/Short Investments


Like many central banks, the US Federal Reserve holds the statutory dual mandate of creating a harmonious balance between inflation and unemployment.

The Phillips curve states the theoretical inverse relationship between inflation and unemployment, given lower levels of unemployment generally create bargaining power among existing workers, leading to wage increases and inflation.

The demographic situation in the US shows completely flat growth in the prime age working population for the past decade, and past 16 years when controlling for participation rate.

If private non-residential fixed investment is growing output capacity yet we lack the human capital to satisfy this potential, the Phillips curve will not work and inflation will be unaffected.

In addition to other global and domestic disinflationary forces, this has implications for the broader world economy and future returns for financial assets.


Over the past decade, growth in the prime age (25-54) population has been flat. Even if private non-residential fixed investment increases to create additional output capacity in the economy, unfavorable demographics are creating a scenario in which too few workers are being produced.

Consequently, even as the (metrically flawed) headline unemployment rate runs below 5% - ostensible "full employment" - this slack in the economy is creating a situation where inflation isn't being produced due to a lack of substitute workers. This is also a material headwind to real wage growth.

In addition to global and domestic disinflationary forces, which I've discussed in some level of detail in previous posts, this poses a challenge to inflation. This has significance for the global economy, such as a continuous "lower rates for longer" environment, which challenges central banks' ability to rectify economic down-cycles.

For financial assets, if inflation continues to run under expectations as it more or less has for most of the past 35+ years, this will continue to benefit fixed-income returns on a risk-adjusted basis.


The unemployment rate receives a large amount of focus given it provides insight into how economic growth is doing relative to potential growth. If we are at "full employment" - a point at which an additional decrease in the metric causes inflation that exceeds the marginal benefit of less unemployment - the economy should be producing at around its capacity.

The US Federal Reserve and essentially all other central banks closely monitor their countries' unemployment rates to attend to mandates often predicated on maintaining an equitable balance between inflation and unemployment. The U-3 unemployment rate is the most commonly reported figure in the US, with "full employment" considered to be around 4%-5%.

With that said, the unemployment rate merely correlates to how actual GDP might appear relative to potential GDP rather than standing as an ideal proxy.

For one, the unemployment rate has some methodological issues in the way in which its calculated that skews its reliability and ultimate value. It excludes large swaths of the potential labor pool, fails to take into account the number of hours worked, and ignores the quality of the employment itself and total factor productivity. All of these factors are important to take into account when assessing economic performance and matters in which this single metric can't capture.

The employment rate has been falling since 2010, yet no up-spurt has been observed in inflation. Inflation and inflation expectations rose from late-June 2016 through late-January but have been moving down since.

(Source: St. Louis Federal Reserve, as is the attribution for all subsequent images)

The patterns over the past three years have largely been driven by movement in the price of oil (a major input in economic activity) and, most recently, altered expectations on the fiscal policy front.

Over the past three years, the correlation between 10-year breakeven inflation and WTI crude oil prices has been +0.81 due to the volatile nature of the market. However, since January 2003, the correlation has come to just +0.13 (Source: Federal Reserve data).

The U-3 employment rate excludes discouraged workers, including those who want to work but for whatever reason can't find it whether it be personal-/family-related matters, skills mismatch, lack of compelling incentives, and so forth. The figure also counts those who work as little as one hour per week.

At its heart, the U-3 is a bureaucratically formulated metric largely designed to make political administrations look good. Even when the number is horrific, such as 10%, it still subtly suggests that nine out of ten people are productively employed, which is never the case.

Reliance on very broad and imprecise labor utility categorizations will correlate with the question policymakers are trying to answer ("How does current GDP relate to potential GDP?") but will fail to tell the whole story. The question of why ostensible labor market tightness has not translated to inflation in the way central bankers have anticipated in such economies as the US, EU, and Japan suggests that the metrics they're looking at may be fundamentally flawed. This has led to persistent cuts in the terminal overnight rate estimation from 4.25% in 2012 to just 3.00% today.

(Source: Brookings Institute)

In other periods of history, the U-3 was a reliable indicator of when inflation would increase. Below is a graph of seasonally adjusted U-3 plotted against the PCEPI inflation rate measured year-over-year:

The gist is that as unemployment decreases, this generally causes inflation to increase.

A clearer demonstration can be found by subtracting inflation from unemployment. As the business cycle goes on, the difference between the two becomes smaller. Once a recession comes to shed inefficiency, some workers become displaced in the shake-out and down-cycles are inherently disinflationary given the drop in demand. This causes the difference between the two to widen back out.

As for the basic economic theory behind it: At a certain point, as more of the labor pool becomes employed this leads to a saturation in the market. The lack of substitute workers leads to greater bargaining power among existing workers for higher wages. Corporations pass off the cost of higher labor by increasing the costs of products and services, leading to price inflation.

At a certain point, the increase in inflation outweighs any beneficial impact on employment. This causes the central bank to allow some slack to exist in the labor market. Hence there will always be some level of unemployment that is in the best interests of the overall economy.

Since February 1960, the correlation between the U-3 and PCEPI inflation has been +0.17. The lack of a negative correlation is due to some level of lag in the data - that is, lower unemployment, then higher inflation. When U-3 is correlated with 10-year breakeven inflation, for which data is available from January 2003 forward, this correlation comes to -0.17.

Why is the Phillips curve less effective than it used to be?

Earlier this month, Goldman Sachs wrote in a research note, "In our view, the US economy has now reached full employment and is likely to overshoot meaningfully, a path that has often proven risky. From this perspective, the case for further tightening is strong."

As mentioned, the issue with the U-3 unemployment rate as a proxy for labor market tightness is that's its imprecision and numerous defects fail to provide any focus on the number of labor force dropouts, productivity or working extent of the employment, or the relatively new situation of insufficient growth in the labor force.

On a seasonally adjusted basis, from December 1962 to April 2000 we had a secular climb in the labor force participation rate as a consequence of a favorable demographic upshift. This resulted in an 8.9% increase in this metric. Since April 2000, this has decreased by 4.4%. This is noteworthy and indicative that the headline unemployment rate may be overstating the labor market's strength.

The graph below shows the sum of the year-over-year increase (or decrease) in the number of labor hours worked and productivity, going quarter to quarter. This is what economic growth essentially is.

The graph is choppy, but there's a clear trend toward lower output.

One likely contributor: from March 2001 to the present, the growth in the effective prime age working population - that is, number of prime age individuals multiplied by their employment rate - has stayed precisely flat. From January 1977 to March 2001, there had been a 2.4% year-over-year increase. This means a higher amount of slack - i.e., substitute workers in the economy - than taken into account in tradition economic models, which has caused subsequent overestimation in terminal interest rates and nominal growth.

Based on data from the OECD, as of December 2016, the number of estimated job vacancies in the US was 5.5 million, around the absolute highest seen since the data was first collected in December 2000.

If we transform this data set by dividing it by the unemployment rate, we can observe that this metric suggests the labor market was a bit tighter than what was observed during the late-2001 to late-2007 expansion - when there was also no material inflation increase observed from labor market increases - but not as firm in the business cycle before that.

Estimation of the number of job vacancies in the US is nonetheless not as well statistically accounted for as the number of people unemployed. In a previous era, statisticians would literally measure job vacancies by the square footage of jobs ads in various newspaper publications from around the country. Given the dispersion of employment ads through different mediums, tracking the number of job vacancies in the country has become more difficult.

If we look at the prime age employment rate, it's the weakest it's been in the past four business cycles including the current one.

The current 78.3% prime age employment rate is two percentage points below the October 2006 peak (80.3%) of the previous cycle. This seems small, but given that there are 125.6 million individuals in the 25-54 age range in the US, this comes to what would be the equivalent of another 2.5 million not in the labor force relative to the prior peak holding population constant.

Demographically, we can observe that we reached an inflection point in this curve in the mid-1980s, where steepening non-linear growth transformed into a year-over-year slowing. Over the past ten years, we have been basically flat.

Economic growth is a function of labor productivity plus the growth in the number of hours worked. Expanding the latter is difficult if there is no population growth among the working age or prime age population. This means economic growth is more or less effectively reduced to the growth in labor productivity.

Private non-residential fixed investment recently saw an uptick, but has seen a leap of just 1.7% y/y over the past ten quarters. This generally feeds directly into productivity gains. Given this is a nominal rate, this would actually translate into slightly negative territory in real terms. If we proceed over a longer time horizon - back to Q1 1990 - private non-residential fixed investment has been 4.43%, or a bit above 2% in real terms.

If we have fixed investment creating jobs to any positive extent but a demographic situation that is effectively flat, then the Phillips curve and its theoretical trade-off relationship between inflation and unemployment breaks down.

If job vacancies aren't being adequately filled simply because not enough workers are being produced, then it would be illogical for lower unemployment - however vaguely and bureaucratically it's defined - to create inflation given the lack of workers to fill these jobs in the first place. Phenomena of this nature aren't going to reflect in the U-3 unemployment figure.

Implications for the global economy and financial assets

Aging demographics are a material disinflationary force in the global economy and this is one conduit by which it directly holds down short- and long-term interest rates. If developed economies can't break out of this ongoing rut, monetary policy will become less effective moving forward as laid out in this article. In short, when rates are compressed down as far as they can get both from lowering the nominal overnight rate and from quantitative easing regimens, central bank influence in economic matters becomes less influential through these channels.

Demographic headwinds in developed markets aren't likely to abate in the coming years/decades. With slow overall economic growth a reversal doesn't seem likely in a type of self-perpetuating cycle. This is part of the overarching thesis for assets that thrive in an environment where inflation runs under expectations, such as bonds. On aggregate, bonds have outperformed stocks on a risk-adjusted basis over the past 35 years and I expect this to continue moving forward.

The Phillips curve's influence could return at some point, but the unemployment/inflation relationship is weak at the moment and demographics appears to be the main culprit.

Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

About this article:

Author payment: $35 + $0.01/page view. Authors of PRO articles receive a minimum guaranteed payment of $150-500.
Want to share your opinion on this article? Add a comment.
Disagree with this article? .
To report a factual error in this article, click here