JOB MARKET PAPER

In this paper I study tit-for-tat pairs, i.e. situations where CEOs serve on each others' boards in a way that makes it possible for them to reward (punish) favourable (negative) compensation outcomes partly influenced by the other player. I find that the residuals from predictive regressions of CEO compensation are positively correlated in such pairs, implying that these relationships are indeed relevant. Although I cannot exclude the possibility of outright corruption, the results hold true when allowing large time lags between the two compensation decisions. I therefore lean towards an explanation where the varying cordiality of the individuals' personal relationship, which is likely stable over some years, is the driving factor. The result is robust to the inclusion of various CEO and board centrality measures as well as a dummy soaking up the average effects of tit-for-tat relationships. When studying a sample of non-CEO top executives (whose compensation is typically recommended by the CEO, rather than the board) none of the effects are present. This indicates that the results are not driven by some unobserved characteristic of the firms whose CEOs form tit-for-tat pairs.