by Edward S. Knotek II and Saeed Zaman – Federal Reserve Bank of Cleveland
Labor costs and labor compensation have garnered considerable attention from economists in the wake of the financial crisis and recession. Across a range of measures, wage growth slowed sharply during the recession. Recently, wage growth has remained near historically low levels despite improvements in the labor market.
Subdued wage growth has been variously seen as both a cause and a consequence of the slow pace of economic growth and persistently low inflation rates. It also may have contributed to rising inequality. In some forecast narratives, a pickup in wage growth is viewed as a necessary condition for a stronger recovery and rising inflation. In others, it is a natural consequence of a tightening labor market.
This Commentary takes a closer look at the relationships between wages, prices, and economic activity. It finds that the connections among wages, prices, and economic activity are more akin to a tangled web than a straight line. In the United States, wages and prices have tended to move together, and causal relationships are difficult to identify. We do find that wages are sensitive to economic activity and the level of slack in the economy, but our forecasting results suggest that the ability of wages to help predict future inflation is limited. Thus, wages appear to be useful in assessing the current state of labor markets, but not necessarily sufficient for thinking about where the economy and inflation are going.
“Wage†Measures
There are a variety of ways to measure labor compensation and labor costs. Our analysis focuses on three such measures. While these measures have varying coverage, we will generically refer to them as “wages.â€[1]
- Average hourly earnings (AHE) of production and nonsupervisory employees on private nonfarm payrolls. This measure is perhaps the closest of our measures to the concept of wages.
- Compensation per hour (CPH) in the nonfarm business sector. This broader measure of compensation includes not only wages but also bonuses and benefits.
- The Employment Cost Index (ECI)—in particular, we use the ECI measuring compensation of private industry workers. The ECI captures wages, salaries, and benefit costs. The ECI abstracts from changes in the sectoral composition of employment over the business cycle, preventing some of the composition bias that affects other wage measures (see Fee and Schweitzer 2011).[2]
Measures of wage inflation have similar trends, but important differences arise (figure 1). Compensation per hour is notably more volatile than the other wage measures, even when looking at year-over-year growth rates. Abstracting from this volatility, all three wage measures decelerated during the recession, and nominal wage growth has been near 2 percent since the end of the recession—which is quite low by historical standards, and just ahead of inflation during that time. When viewed over a long time span, the upturn in average hourly earnings growth since mid-2012 and the acceleration in the ECI in the second quarter of 2014 are very small.
Cross-Correlations
We first consider the connections between wages and inflation or economic activity using cross-correlations. Cross-correlations allow for a simple examination of the lead or lag structure between two series as well as the strength of the connections between the series. If wage inflation reliably comes ahead of price inflation in the data, then the strongest cross-correlation should be between wage inflation in quarter t and price inflation in some k-th quarter after t.
When working with price inflation and wage inflation, there are important long-run trends and short-term volatility that we would like to remove from our analysis. As a general rule, wages and prices have followed roughly similar long-term trends: both accelerated from the 1960s into the 1970s and then decelerated in the 1980s. The forecasting literature has found gains in inflation forecasting accuracy by specifying inflation in gap form as a deviation from a slow-moving long-run trend (see, e.g., Kozicki and Tinsley 2001, Clark 2011, and Zaman 2013). We borrow from this literature and construct wage inflation gaps as quarterly annualized growth in a particular wage measure less a shared long-run trend.[3] To remove high-frequency fluctuations in food and energy prices in our measure of price inflation, we define the price inflation gap as quarterly annualized core PCE inflation less the shared long-term trend. We then look at cross-correlations between quarterly price inflation gaps at time t+k and quarterly wage inflation gaps at time t.
