Core inflation and employment costs in the US

While the surge in US inflation in recent years is generally attributed to the Covid pandemic and the subsequent Russian invasion of Ukraine, it is often argued that strong wage growth has been preventing inflation from returning to levels compatible with price stability. While this sounds plausible, an alternative view is that wage costs simply reflect past inflation. Since inflation lowers real wages (and raises firms’ profit margins), workers demand compensation in the form of higher nominal wages. Under this view, labour market developments contain little information about future inflation.

To attempt to distinguish between these two views, it is useful to look at the behaviour of core inflation and the Employment Cost Index (ECI) which incorporates wages and salaries and the cost of benefits.¹

Chart 1. Core inflation and wage growth

Source: FRED. Data as at 11 June 2024.

The graph above shows core inflation and ECI growth rates over four quarters. During Covid, core inflation surged far above its pre-Covid level. While it averaged 1.9% between 2002 and 2019, it averaged more than twice that, 4.1%, between 2020 and 2024Q1. In comparison, the growth rate of the ECI averaged 2.6% between 2002 and 2019 and rose to 3.9% after 2020.

If wage growth drives inflation, one would expect past ECI growth to forecast future core inflation. By contrast, if wage growth reflects reactions to past inflation, one would expect the converse to be true.

The table shows the results from formal tests of these hypothesis.² It indicates that during the pre-Covid period there was little relationship between core inflation and the growth rate of the ECI. During the period incorporating Covid, inflation has forecasted future ECI growth. Interestingly, the converse is not true: wage growth has not forecasted future core inflation. These results cast doubts on the idea that wage growth is an important driver of inflation.

Chart 2. Results from hypothesis tests

Source: EFGAM calculations.

But other important factors are missing from this analysis. Productivity is likely to be particularly important and will drive a wedge between wage growth and inflation. To proceed, we estimate a simple forecasting model of quarterly core inflation and the ECI, incorporating productivity growth as an additional explanatory variable.³

The models are estimated on data starting in 2002Q1 ending in 2019Q4. We then use them to explore how well they forecast core inflation and ECI growth since the onset of Covid in 2020Q1 by looking at the prediction errors.⁴

If employment costs were a key driver of the recent surge in inflation, we would expect employment costs to be stronger than predicted by the model and the prediction errors for core inflation to be small.

Conversely, if high core inflation was the cause of rapid ECI growth, then we would expect the prediction errors for core inflation to be large and positive, but the prediction errors for employment costs to be small.

With that in mind we turn to the results.

Chart 3a. Core inflation
(95% confidence band shown for both)

Chart 3b. Employment costs

Source: EFG calculations on FRED data. Data as at 11 June 2024.

The results are clear cut. The graphs show that the prediction errors are positive and often above the 95% confidence band for core inflation, indicating that core inflation was stronger than predicted by the model. However, ECI growth evolved much as expected.⁵

These results suggest that the high ECI growth observed in recent quarters is a response to high inflation and is not a source of it. Since high inflation reduces real wages and thereby provides strong incentives for workers to bargain for higher wages, this finding is not unexpected.⁶

One consequence of these findings is that in judging inflation pressures, it makes better sense to focus on inflation outcomes than employment costs.

¹ See https://go.pardot.com/e/931253/eci-home-htm-/47rf1/384502476/h/jVsgIxoKq--REqR7u4l8II1Ht_SlzJKWjWQnWHaZ8OY
² Granger causality tests, incorporating two lags, are conducted on the quarterly growth rates of the two series. A p-value of 5% is used.
³ The other regressors are four lags of the log-levels (to allow for cointegration) of core prices and the ECI.
⁴ This approach follows Andrade et al., “Is post-pandemic wage growth fueling inflation?”, Current Economic Perspectives, 2024-1, Federal Reserve Bank of Boston.
⁵ One would expect that the prediction errors are equally likely to be positive as negative. However, 14 of the 17 prediction errors for core inflation are positive (the likelihood of getting at least 14 positive errors is 0.6%). However, only 9 of 17 prediction errors for ECI growth are positive (p = 50.0%).
⁶ See also the EFG Infocus on “How high is the risk of a wage-price spiral?” 3 September 2023. https://go.pardot.com/e/931253/-Infocus-wage-price-spiral-pdf/47rf4/384502476/h/jVsgIxoKq--REqR7u4l8II1Ht_SlzJKWjWQnWHaZ8OY