Assessing the predictive power of labour-market indicators of inflation

Abstract

This article examines two different measures of wages as predicators of prices in a Vector Error-Correction (VEC) framework using quarterly data for the United States for the period from 1947Q1 to 2008Q1. Based on cointegration and a series of exogeneity tests, it is found that (i) there is a stable, long-run relationship between the Consumer Price Index (CPI) and the Personal Consumption Expenditures Deflator (PCED) on the one hand and Unit Labour Cost (ULC) and Average Hourly Earnings per Unit of Output (AHE) on the other; (ii) ULC is weakly exogenous for both price indices while the two price indices are weakly exogenous for AHE; (iii) ULC is strongly exogenous for CPI but not for AHE; and (iv) ULC is super exogenous for CPI. Taken together, these findings lead to the conclusion that ULC is a reliable indicator of price inflation but productivity-adjusted hourly earnings is not. Thus, monetary policymakers are justified in using information about the behaviour of ULC in formulating policy actions for achieving the goal of price stability.

Keywords:

• cointegration
• exogeneity
• inflation
• unit labour cost
• causality

JEL Classification:

• C32
• C12
• C52
• E31

Acknowledgement

This research has been financed in part by a grant from the Center for Global Economic Studies at Marquette University.

Notes

¹ These correspond to Equations 5(5) and 6(6) of Gordon (1988, p. 278).

² It should be noted that one may include a deterministic trend, dummy variables for regime changes, as well as predetermined and/or exogenous variables in Equation 5(5).

³ As noted by Gordon (1988) and Mehra (2000), Equation 5(5) essentially reflects markup pricing practice by firms consistent with the basic Phillips-curve relation.

⁴ Note that Equation 5(5) can be normalized on the wage rate to get .

⁵ Throughout this article, details of estimation and hypothesis tests will not be reported to conserve space. The author would be happy to supply them to interested readers on request.

⁶ See Hamilton (1994, pp. 608–12).

⁷ Using a variant of this test based on maximum-likelihood estimates of the long-run parameters of the four alternative price–wage equations, we obtained results that were consistent with those reported here.

⁸ We also tested for weak exogeneity using a slightly different version of the Engle test where we included the residuals and squared residuals from the marginal distribution of wages in the conditional distribution of prices. Wages would be weakly exogenous for prices if the estimated coefficients on these terms are not jointly statistically significant. Performing this test, we found evidence of weak exogeneity of ULC but not AHE – results that are consistent with those reported above.