This begs the question of what dete

This begs the question of what determines the differences and the movements in the value of voting rights.
This paper develops and tests a model that explains to what extent and when the value of private benefits is reflected in the exchange price of a vote. I show that the value of a vote is determined by two factors: the likelihood that the vote will be pivotal in a contest for control and the price it will fetch in case of such a contest.
I use Grossman and Hart's [1988] and Harris and Raviv's [1988] models to formally link the price of votes and the value of private benefits of control. The voting premium reflects the expectation that voting rights become valuable in case of a battle for control. If an acquisition is contested, the superior voting shares should receive a differential premium that is a function of the size of the private benefits of control and of the proportion of superior voting shares outstanding. However, the probability of a contested acquisition is very much influenced by the existing ownership structure. The ownership structure should therefore affect the voting premium through its effects on the probability of a contested acquisition, as I show in three case studies of unexpected changes in the distribution of ownership.
I then test this relationship on a panel of U. S. companies having two classes of common stock between 1984 and 1990. As a proxy for the cost, and thus the probability, of an acquisition, I use the total market capitalization of a company. As a proxy for the marginal value of a vote in a control contest, I use the Shapley value, which measures the probability that a vote not held by holders of large blocks is pivotal. Both these proxies are statistically significant and can explain 14 percent of the cross-sectional and time series variability of the premium.
Thus far, the analysis is carried out under the assumption that private benefits of control are a similar proportion of the value of future dividends in all the companies considered. Because this proportion is likely to vary across companies, I use the level of cash salary paid to the largest shareholder (in excess of what the company's size would predict) as a proxy for differences in control benefits. This proxy is positively correlated with the unexplained component in the voting premium. This suggests that the value of managerial perquisites is, at least partially, reflected in the price of votes.
Although testing the model is by necessity limited to the small set of companies whose differential-voting stock is traded (so that



WHAT DETERMINES THE VALUE OF CORPORATE VOTES? 1049

the price of voting rights can easily be estimated), the implications of my results are much more far-reaching. The price of any common stock includes a vote component, even if, in most publicly traded companies, this value cannot be directly measured. Al-
though under "normal conditions" the fluctuations in this vote component are small enough to be neglected, this is not always the case. Voting premiums can fluctuate widely, and voting rights can sometimes become as valuable as cash-flow rights. In particular, when an event alters the distribution of ownership or the expecta-
tion of a contested acquisition, a fraction of the private benefits of control may be reflected in the stock price and may affect its variability in a substantial way. This suggests that it is inappropri-
ate to automatically interpret an increase in the price of a stock as an increase in the value of expected future cash flow; it can simply be a change in the price of votes. It also suggests that this paper's methodology might help future research distinguish between the two explanations.
This paper is organized as follows. Section I contains three case studies showing how drastic changes in the ownership struc-
ture affect voting premiums. Section II shows why the voting premium is equal to the expected differential premium in case of a contested acquisition, and links this differential payment to private benefits of control. Section III presents the empirical analysis. Section IV discusses the implications of my findings and concludes.

I. THE EFFECTS OF OWNERSHIP ON THE VOTE VALUE
If the value of a vote is determined by the fact that control can be achieved by assembling a large block of stock in a series of small open-market transactions, then the price of a vote should rise at the announcement of any major and unexpected change in the ownership distribution that would facilitate a takeover. By review-
ing the proxy statements of companies with two classes of common stock that were trading during the 1980s, I identified six cases in which there was an exogenous change in the ownership of the controlling block. In three cases the controlling block was inherited by members of the same family, who succeeded in keeping their voting power united by using trusts. Therefore, I focus on the three other cases, where there were substantial changes in the distribu-
tion of voting power.1 Figures Ia, Ib, and Ic illustrate the behavior


1. Details of all these cases are in Zingales [1992].