A Forecasting Test for the Reliability of Salary Data

Abstract

It has been established in the economics literature that competitive outcomes can be well pre- dicted by using the remuneration of the contesting participants as a measure of their productivity. In this paper we use this idea to construct a forecast test to evaluate a novel, and hitherto unevalu- ated source of salary data. If the source of salary data is without measurement error we can expect that it will provide accurate and efficient forecasts of outcomes — in our case, football matches. Establishing the reliability of such a source of salary information is highly valuable for economists studying a range of labour market outcomes, as this particular source freely provides over 30,000 employee-year observations of salaries across a range of competitions and countries, a much wider sample than has previously been available for analysis. By applying standard testing techniques for the efficiency of forecasts, we find that indeed salary-based forecasts are efficient. We find that salary information at an individual player-level is informative relative to general squad level information, also. We thus argue that the information from this particular source is reliable, and has merit for further academic investigation.

Keywords: sport, wages, forecasting, skills

JEL Classification: C53, L83, Z22