Alan Harrison
Abstract
The proposition that nominal wages are rigid downward has recently led
some commentators to question the wisdom of central banks' pursuing a
policy of low price inflation. The argument is that, at very low rates of
inflation, the nominal wage rigidity inhibits the ability of the labour
market to effect adequate adjustment through variations in nominal wages,
and the inability to adjust in turn leads to unemployment. The question
of whether nominal wage rates are indeed downwardly rigid therefore
becomes critical.
Canadian evidence drawn from unionized contract data, reported in Fortin
(1996), has been cited approvingly in this debate (for example, by
Akerlof, Dickens and Perry, 1996), but this evidence looks only at very
recent experience and uses only unconditional tests. Other periods of low
inflation may have much to say on this subject, and more sophisticated
tests are required if we are to distinguish wage settlements that would
have been negative in the absence of wage rigidity from those that would
anyway have left nominal wages unchanged.
My paper investigates these issues, using Canadian data on unionized wage
contracts covering the period from the early 1950s to the present. To
provide context for what follows, I look first at unconditional tests for
rigidity. I then turn to use of a method that allows me to look for
excess mass at a particular point in a distribution of nominal wage
changes.
Specifically, I adopt an approach based on work by Donald, Green and
Paarsch (1995), that has been used by Green and Paarsch (1996) to examine
the effects on the wage distribution of the minimum wage. This approach
has several benefits for my work. First, it allows maximum flexibility in
the fitted density of wage changes; second, it can accommodate covariates
that can affect both the shape and the position of the density; third, the
effects of covariates can vary across segments of the distribution.
I estimate the distribution of nominal wage changes, incorporating
covariates that include the rate of price inflation and measures of
productivity growth. Early results indicate the importance of the latter
variable, and suggest (as one might perhaps expect) that it is the
combination of low inflation and low productivity that is most likely to
generate mass in the distribution of wage changes at the point where the
change is zero. The tests also confirm the presence of nominal wage
rigidity in unionized contracts, though the magnitude of the phenomenon is
much lower than the estimates reported by Fortin (1996).