111.2. Summary StatisticsTable II p

111.2. Summary Statistics
Table II presents summary statistics on the panel data-set. Each year the voting premium is computed as the average of the daily premium defined by equation (9). The grand median of the annual average of the daily voting premiums is equal to 3 percent.6 The grand mean (10.5 percent) is about twice the average level found by Lease, McConnell, and Mikkelson [1983, 1984] (5.4 percent). The difference is partly explained by the different definition of voting premium: Lease, McConnell, and Mikkelson define as voting premium (PA — PB)IPB, where PA is the price of the superior voting shares and PB the price of the inferior voting shares. It should also be noted that Lease, McConnell, and Mikkelson limited their attention to dual classes with identical dividend rights. However, the average premium for companies with equal dividend rights is only slightly larger than the mean of the whole sample.
The level of the voting premium in the United States is comparable to the level found in Sweden (6.5 percent) and the United Kingdom (13.3 percent). However, it is below the value found in Canada (23.3 percent) and Switzerland (about 27 per-
cent), and it is well below those in Israel (45.5 percent) and Italy (81 percent).7
Another important datum is the volume of daily trading. I computed the average daily volume for both classes in each year. Then I defined the relative volume as the ratio between the average trading volume in the superior voting class, divided by the average trading volume in the inferior voting class.8 CRSP data on volumes were available from 1982 for companies listed in the NASDAQ system, but only from 1986 for AMEX and NYSE companies. On average, the volume in the superior stock is slightly less than half (0.44) the volume in the inferior stock. However, the median is only 0.2, and there are huge cross-sectional differences.