I remember an interesting theory about golden parachutes in Steven Landsburg's "The Armchair Economist". In a typical publicly traded company, the CEO is actually much more heavily invested in the company than the shareholders. His salary, and probably the major part of his assets, are dependent on the performance of the company. On the other hand, the average share holder is well diversified. Therefore, a CEO will be incentivized to pursue a highly conservative strategy, while the shareholders may wish that he takes more risk. To offset this, the shareholders may make it known that a generous severance package, as well as generous benefits if the company does exceptionally well, are on the table, to simultaneously soften the blow if he takes a bad risk, and sweeten the deal if he takes a good risk.
In other words, the point of executive compensation may not be to produce better results on average, but rather to increase the variance of the results.
In other words, the point of executive compensation may not be to produce better results on average, but rather to increase the variance of the results.