We see greater dynamics, however, w

We see greater dynamics, however, when we split the period and compare 2001 with 2008 (centre model) and 2008 with 2017 (rightmost model). In both subperiods, the trend in standard employment was driven by both compositional and behavioural changes, but with behavioural changes being more important. For example, in the first sub-period, 1.1 percentage points of the 3.7 percentage points increase in the standard employment share can be explained by compositional changes, while 2.6 percentage points remain unexplained, suggesting behavioural changes are more important. When we look at the second panel of the table, we see that changes in the firm size and industry mix contributed to the increase in the standard employment share in the first sub-period. Specifically, had the firm size mix and the industry mix remained at their 2001 levels, the share of standard workers in 2008 would have been 1.9 and 0.3 percentage points lower, respectively. Declining union membership rates, however, have partly offset these trends, reducing the standard employment share by half a percentage point.