Evidence of Nominal Wage Rigidity and Wage Setting from Icelandic Microdata

This paper presents new evidence about wage stickiness and the nature of wage setting. We use a unique micro dataset on monthly frequency, covering wages in the Icelandic private sector for the period from 1998-2010, and draw the following conclusions. First, the mean frequency of wage change is 10.8% per month. When weighted for heterogeneity across industries and occupations the result is almost identical; the frequency of change is 10.5% per month. Second, only 0.5% of monthly wage changes are decreases. Third, the mean duration of wage spells is 8.9 months. Onefifth of wage spells last longer than a year while other spells last for one year or shorter. Fourth, wage setting displays strong features of time-dependence: half of all wage changes are synchronised in January, but other adjustments are staggered through the year. Fifth, there is limited evidence of state-dependence: frequency of wage increases, size of increases, frequency of wage decreases and size of decreases do not correlate with inflation. However, both frequency and size of wage decreases have significant correlation with unemployment. Sixth, the hazard function for wages is mostly flat during the first months but has a large twelve-month spike. These facts align with a model of time-dependent wage contracts of fixed duration.