Inflation Returns

The Philips Curve, US fiscal expansion and the USD.

The Phillips Curve Is Muted But Alive

Reading William Phillips’ seminal 1958 paper this quarter1, we noted two things:

  1. The Phillips curve is defined as the relationship between unemployment and the change in wages specifically, not broad inflation
  2. The Phillips curve is exponential not linear

In Figure 1 we show the US average annual unemployment and core personal consumption expenditure (‘PCE’) by calendar year since 1961. We can see that the correlation is limited, and what relationship there is slopes upward, the opposite of what Phillips posited. You can data mine all you want with different lags (as we have): the result will be broadly the same.

But it should exist. The most basic economic logic suggests that if the supply of a good or service reduces then its price should rise, assuming demand is constant. In theory, labour markets should be no different to any other.

One explanation is that the relationship between wage growth and inflation is not exact. In Figure 2 we show a scatter of US non-farm unit labour costs and core PCE since 1960. Although there is a good relationship in total (R² of 67%), this is biased up by the nine year period 1974-1982 (these periods coloured blue), when twin oil crises arguably led to a unique environment. If we remove these the fit drops to 27% – still a connection, but a long way from lockstep.

Even for wage inflation, you can struggle to find a relationship in the past two decades. In our view the reason is demographic. If you define the baby-boom generation as those born between 1943 and 19602, then the last 20 years has seen significant retirement from one of the most populous US generational segments. The baby-boomers have been replaced by the children of the gig-economy, the unit labour cost of which is smaller. Illustrating this trend, in 2017 the US working age population (16-65) lost 115,000 members, the first annual decline since 19513. Growth has been trending down since 2000.

We adjust for this in the graph in Figure 3, which shows the Federal Reserve Bank of Atlanta’s (‘Atlanta Fed’) wage data against a broad unemployment gauge. The Atlanta Fed series measures the median worker, and is thus less influenced by the departure of high earning baby-boomers than a mean-based calculation would be. On this basis we can see that there is a strong exponential relationship.

So for us it is likely that the Phillips curve as originally formulated (for wages) is relevant today. And it also seems likely that this wage inflation will feed through to broad inflation, albeit the transmission is imprecise, as we have already discussed.

Phillips’ original work also suggests a curve that is exponential not linear. For Phillips, unemployment moving below 3% was the catalyst for wage inflation to start accelerating much more rapidly. Phillips’ data reflected the British Empire 100 years ago, so we would not want to make assertions about what the equivalent level will be in the US today but, with most measures of unemployment approaching all-time lows (see discussion around Figure 5), we may be getting close.

Finally, we note that the Federal Reserve (‘the Fed’) takes the Phillips curve seriously. At least Janet Yellen did, and the evidence for its existence was a regular feature of her speeches4. Since we last wrote Jay Powell has taken over as Fed Chair and, as we said in our previous quarterly, whilst he might be a bit more hawkish at the margin, we expect him to be a continuity candidate.

Figure 1. US - Inflation and Unemployment (1961 to 31 December 2017)

Source: Man GLG and Bloomberg; as of 31 December 2017.

Figure 2. US Unit Labour Cost Growth and Inflation (1960 to 2018, quarterly data)

Source: Man GLG, Haver Analytics and Bloomberg; as at 31 December 2017.

Figure 3. US Phillips Curve – Underemployment vs. Median Wage Inflation (1998 to 2017)

Source: Man GLG, Atlanta Fed and Bloomberg; as of 31 December 2017.

Figure 4. US Budget Deficit as % GDP. An Unprecedented Expansion Away from Recessionary Periods

Source: Bloomberg with 2018 and 2019 estimates from the US treasury. Years containing recessionary periods are outlined with a blue dashed line. As of April 2018.

Figure 5. US Now at Full Employment

Source: Bureau of Labor Statistics, JOLTS and Man GLG; as of April 2018.

US Fiscal Expansion Is Unprecedented – Another Possible Trigger for Reflation

In Figure 4 we show the US budget deficit as a percentage of GDP, from 1968 to the present, with 2018 and 2019 estimated as per US Treasury forecasts. We can see that, up until the present day, there have never been consecutive years of deficit expansion outside of, or immediately after, a recessionary year.

Now President Trump is accelerating what President Obama began. If Treasury forecasts hold, we are halfway through four years of consecutive deficit expansion, a full nine years into the economic cycle.

The data on whether expanding budget deficits are inflationary is inconclusive, but in this instance we think there is likely to be a link. In Figure 5 we show our favoured measure of spare capacity in the US labour market, being the ratio of the number of unemployed persons to the number of job openings. The metric is at its lowest point so far this century.

Pouring unprecedented stimulus onto an economy which is already running hot may be the trigger that shifts the US economy into the exponential segment of the Phillips curve that we have already discussed.

Figure 6. The DXY Index, February 2015- February 2018. ‘Our currency, your problem’ – US Treasury Secretary John Connally, 1971

Source: Bloomberg; as of 20 March 2018.

The USD Need Not Strengthen

We think that the USD does what the US Executive wants it to do, as we illustrate in Figure 6. Whilst in the short term we may see some technical strengthening (given short speculative positioning in the USD), over a longer horizon we believe that this administration, wedded to shrinking trade deficits, will put a ceiling on appreciation.

Economics 101 would be for this to be inflationary for the US, given the increased cost of imports, and for EM given the reduced cost of external borrowing, which in turn enhances money velocity.


We are not in the business of taking big macroeconomic bets and will not be fundamentally recalibrating the portfolio based on an inflation view. We will be making shifts around the margin, however, and will also be monitoring our inflationary indicators closely in relation to the impact it could have on 60/40-type portfolios, which have become an article of faith for many asset allocators, in our view.


1. A.W. Phillips – The Relation between Unemployment and the Rate of Change of Money Wage Rates in the United Kingdom, 1861-1957 – Economia, Vol.25, No. 100, pp.283-299 – 1958.
2. This is the most commonly accepted definition, as per the work of Neil Howe and William Strauss.
3. Source: US Bureau of Labour Statistics.
4. See, for example, Janet Yellen – Inflation, Uncertainty, and Monetary Policy – Speech to the 59th Annual Meeting of the National Association for Business Economics, Cleveland, Ohio – September 2017.