Thesis (Ph. D.)--University of Rochester. Department of Economics, 2019.
The impact of wage stickiness on job separations is a crucial concern for macroeconomists
and policy makers. The measurement of the impact of wage stickiness
on job separations is, however, challenging as finding a good measure of
wage stickiness is difficult.
This dissertation consists of three essays providing novel approaches to
measuring the impact of wage stickiness on job separations. The first chapter
uses elapsed time since the last wage reset and expected duration to the
next wage change to show that wage stickiness affects job separations and also
constructs a model that clarifies the underlying mechanisms. The second chapter
extends the previous model to include job-to-job flows and worker specific
productivity shocks. The third chapter uses differences in wage setting practices
(performance pay vs. fixed pay) to test the impact of wage stickiness on
job separations.
The first chapter tests the impact of wage stickiness on job separations by exploiting
that wage stickiness predicts distinct patterns in job separations with
respect to the timing of wage change. First, the longer a wage remains unchanged,
the more likely it becomes mispriced. This obsolescence of wage
should create more job separations if costly wage adjustment prevents wage resets
(‘wage obsolescence effect’). Second, as a wage approaches its next scheduled
reset, separations due to wage stickiness should decline as harms from amispriced current wage become less important (‘proximity effect’). My estimation
results do show that a longer wage spell generates more separations for
hourly workers, especially for lower-wage hourly workers, but not for salaried
workers. But contrary to the second prediction, proximity to the next wage
change is actually associated with an increase in job separation rates. I interpret
this finding as reflecting imperfect information on match productivity,
with new information spiking before scheduled wage changes. Estimation by
job characteristics supports this claim: the rise in separations toward the next
scheduled wage change periods is only prominent for workers with arguably
high uncertainty about their individual productivity, salaried workers performing
abstract tasks. I confirm the importance of wage stickiness and informational
frictions in a Mortensen-Pissarides type search model. Sticky wages are
necessary to generate the wage obsolescence effect while information frictions
can explain the rise in separations as wage resetting approaches.
The second chapter extends the search and matching model of the first chapter
to include job-to-job flows and worker specific productivity shocks. Under
a standard calibration, the model successfully replicates the wage obsolescence
effects both for job-to-job and job-to-nonemployment separations. The introduction
of worker specific productivity shocks allows the model to generate
quits due to the ongoing wage becoming too low and increases the relative importance
of separations driven by wage stickiness. However, the quantitative
importance of the quits to nonemployment due to wage stickiness is much less
significant than that of the layoffs.
The last chapter exploits differences in wage contract arrangements (fixed
versus performance based) to test the role of wage stickiness on job separations.
Performance based wage contracts align productivity with labor cost, making
wages more flexible. Therefore, if wage stickiness creates job separations, performance
pay should reduce job separations. I provide empirical evidence that performance pay reduces the probability of job separation and makes separations
respond less to cyclical unemployment fluctuations compared to fixed
pay jobs. These findings support the presumption that wage stickiness creates
excess job separations.