What the Swedish Evidence Says Changed
This chapter applies the premium-versus-composition distinction to the IZA study’s Swedish evidence. The focus is not a current occupational ranking but a linked set of relationships from 1996 to 2013: employment growth, raw wage growth, estimated premium growth, initial occupation wages, worker composition, and inequality. The practical judgment is to read the study’s numbers as conditional evidence about how occupational prices and sorting interacted, not as individual wage effects or a universal law.
The study’s setting matters before its results do. It covers Swedish employees aged 18–64, uses contractual monthly wage rates as the price of labor, and analyzes 101 occupations in the three-digit SSYK96 classification. Demographic and employment information comes from administrative sources, while occupation and wage information comes from the Swedish Wage Structure Statistics. Because an occupational classification change in 2014 would disrupt employment trends, the analysis ends in 2013.
The premium signal is stronger than the raw wage signal
The first result concerns employment growth. Across the Swedish occupations, raw mean wage growth has a positive but weak relationship with employment growth. The reported weighted slope is 0.026, with a standard error of 0.01 and an R² of 0.07. The relationship is therefore upward on average, but employment growth explains only a modest share of the cross-occupation pattern in raw mean wages.
The estimated premium relationship is stronger. Premium growth has a reported slope of 0.042, a standard error of 0.009, and an R² of 0.20 when plotted against employment growth. The slope is steeper and the fit is better than for raw mean wages. The paper interprets this as evidence that demand-side factors pushed up premia in expanding occupations, while changes in worker composition muted the same pattern in average wages.
The difference is the empirical payoff from separating price and people. Suppose an occupation expands because employers demand more of its output. A higher premium can attract workers from other occupations who have less occupation-specific experience or lower measured skills than incumbents. The occupation’s price rises, but the average wage of everyone now in the occupation includes the new entrants. The raw mean therefore gives a less direct view of the demand signal than the estimated premium.
The association should not be overread. A positive relationship between employment growth and premium growth is not a causal estimate of a particular demand shock. Employment and wages may respond together to technology, consumer demand, institutional changes, or other forces. Workers may also sort in anticipation of opportunities, and the stayer-based estimate has its own selection assumptions. The evidence supports the interpretation that demand factors are important in this setting; it does not identify one unique cause for every occupation’s movement.
The study finds a second important relationship with initial wages. Premium growth is positively associated with the occupation’s initial mean wage in 1996.