Transfer fees in professional soccer

This paper investigates the determinants of transfer fees for professional soccer players in the European summer (inter-season) market. Prior studies have focused primarily on player characteristics and performance. In this study we consider club characteristics, club performance and transfer modalities as well. In total, thirty-four explanatory variables are considered, including, but not limited to, those identified in previous work.

Utilizing a sample comprising 1,091 players who were transferred to or from a Big Five European soccer league club in either 2015 or 2016, twenty-four variables are found to have a statistically significant relationship (at five percent) with the transfer fee in univariate regressions. However, after examining the characteristics of the data more closely, and taking account of collinearity among the explanatory variables, the number of important factors can be pared down to six: the number of games played with their national team during the season preceding the transfer; the period remaining in the player's contract with the selling club; the championship ranking of the buying club at the end of the season preceding the transfer; the number of international games played with their club during the season preceding the transfer; the number of goals scored and assists made during the season preceding the transfer; and the number of titles won by the player throughout their career.

The ordinary least squares (OLS) model estimated using the above variables produced consistent results across the entire sample and, separately, for each of 2015 and 2016. The variables' relationship with the transfer fee was significant at one percent in all cases, except for the number of career titles won (insignificant in the 2015 results and significant at five percent in the 2016 results) and the number of international games played with their club in the year preceding the transfer (significant at five percent in both the 2015 and 2016 results). The estimated coefficients have the expected sign, which is positive in all cases, except for the championship ranking of the buying club in the previous year, which, as expected, is negative. Adjusted R² values indicate that the model explains a consistent 40-44 percent of the variation in the transfer fee. Finally, of the four groups of variables considered: player characteristics, club characteristics, player performance, and transfer modalities, three are captured in the final model. The exception is player characteristics, which is surprising as they have figured prominently in prior studies.