142 Metastatistics for the Non-Bayesian Regression Runner
3.7 Case study 2: “union wage premium”
I would now like to consider an econometrically sophisticated literature – the liter-
ature on union wage effects. Two comprehensive and influential surveys are Lewis
(1963, 1986). In these works, Lewis literally cites hundreds of studies attempting to
estimate the causal effect of union status on wages.^61 See also the useful discussion
in Heckman (1990).
3.7.1 Early history
The idea that labor unions might raise wages is one of the oldest debates in eco-
nomics and was one of the earliest motivating examples for the famous “supply
and demand” cross in a study by Jenkin (1870) (see Humphrey, 1992, for a short
history). Ironically, although Jenkin (1868) concluded that the supply and demand
analysis wasn’t particularly relevant for explaining the consequences of union
wage-setting, subsequent neoclassical theorizing in the main focused on the simple
model depicted in Figure 3.6, whereWis the real wage,Lis the quantity of labor,
Dis the employer demand curve,Scis the supply curve without unionization and
Suis the supply curve with unionization. Until Lewis’ influential survey, opinions
diverged between those who believed that unions could rarely control the supply
of labor, such as Milton Friedman, and those that thought they could and therefore
acted to create unemployment, such as Paul Samuelson (see Friedman, 1950).
3.7.2 A battery of severe tests
The analysis of union wage effects has become more sophisticated with the advent
of large micro-datasets, but let me highlight some of the comparisons researchers
WLDWuWc
ScSuFigure 3.6 Union wage effects in the neoclassical model