While the introduction of these contract
date dummy variables may help to control
for contract date related correlations between
the error term and the independent
variables,the estimated coefficients of these
variables themselves have no obvious economic
meaning. This is because of the nature
of the sample. Recall that I observe
contracts in force in 1979. If we think of the
population as consisting of contracts written
for particular plants (i) in a particular year
(t), we can observe a contract only if