There are at least two reasons why other researchers have not found the significant dependencies
described here and in the October 1987 S&C article on money supply.
First, most of the other workers in this area have used either serial or auto correlation or runs tests (see
for example, P. Cottner, Random Character of Stock Market Prices, MIT Press, 1964). As I pointed out
in the April 1988 S&C issue, there are some significant problems associated with using correlation
techniques on this data. Further, correlation only tests for one form of serial dependency — there are
other forms that would be missed. The same holds true for tests based on runs. Statisticians apparently
cannot agree if runs tests are tests of randomness or independence (see S&C, September 1986).