‘Step-Skipping’ at the Big Labor-Funded Economic Policy Institute

Evidence in Medicine: Correlation and Causation – Science-Based 

The compensation penalty of “right-to-work” laws | Economic Policy 

When it comes to “step skipping,” Jason “Stanky” Hanky has nothing on the academic apologists for compulsory unionism at the Economic Policy Institute. Image: Actor James Spader in The Apology, an episode of NBC-TV’s Seinfeld originally airing in December 1997.

In economics, controlled experiments are nearly always impractical.  Observations have to made in the real world, rather than under laboratory conditions.

Specifically, assessments of the economic impact of state Right to Work laws, which prohibit the firing of employees for refusal to pay dues or fees to an unwanted union, can practically speaking only be made through studies in which a positive correlation, a negative correlation, or no correlation between Right to Work status and a particular economic growth indicator may be observed.

In such studies a positive correlation means that Right to Work states experience, on average, more growth according to a particular economic indicator than forced-unionism states. A negative correlation means that Right to Work states experience, on average, less growth.  And no correlation means growth in Right to Work and forced-unionism states is, on average, roughly equal.

As the cliche goes, correlation is not causality, but this truism hardly means correlation is irrelevant to the question of whether one phenomenon causes another.  Establishing whether two phenomena are correlated or not is a necessary step toward making a scientific assessment of causation.

Steven Novella, a prominent clinical neurologist and professor at the Yale University School of Medicine, points out in a brief analysis of correlation and causation in medical research, linked above, that in observational studies ” a lack of correlation is easier to interpret than a positive correlation — if there is no correlation between A and B then we can pretty much rule out a causal relationship.”  The “only caveat,” Novella explains, “is that a correlation is being obscured by a factor that we have not accounted for.”  Only if there is a correlation is a “more thorough analysis” even required.

That brings us to a 2011 study by the Big Labor-backed Economic Policy Institute (see the link above) regarding Right to Work laws’ impact on employee wages and benefits.

The authors of the EPI study, which remains relevant today because it is still frequently cited, directly or indirectly, by forced-unionism advocates, purport to have found a “causal relationship” between Right to Work laws and lower wages and benefits.

Though the authors don’t say it, logically speaking to assert that Right to Work laws “cause” lower wages and benefits is also to assert that Right to Work laws “cause” wages and benefits to grow more slowly than they otherwise would

As Novella has reminded us, to prove this you would first have to show that Right to Work laws are correlated with slower growth in wages and benefits.  The correlation could be directly observed, or only observed when one or more “obscuring factors” are accounted for.

And the simple fact is, the EPI study does not even attempt to show such a correlation, with or without controlling factors.

A quick review of data related to compensation growth in the 50 states over the past decade or so indicates that one important reason why the EPI didn’t seek to show such a positive correlation between compulsory unionism and compensation growth is that no such correlation exists.  For example, from 2002 to 2012 inflation-adjusted U.S. Bureau of Economic Analysis statistics show private-sector employee compensation (including wages, salaries, benefits and bonuses) increased by 14.3% in the 22 states that had  Right to Work laws throughout the period, compared to an increase of just 5.4% for forced-unionism states in the aggregate.

That’s a negative correlation, rather than a positive one, between forced-unionism and compensation growth.

Since employment growth was also far slower in forced-unionism states than in Right to Work states from 2002-2012, on a per employee basis forced-unionism states and Right to Work states had roughly the same compensation growth over the decade.  But that still doesn’t show any positive correlation between forced unionism and compensation growth.

In short, as long as the EPI fails to clear even the first evidentiary hurdle for establishing that compulsory unionism is good for employees’ pocketbooks, this “thank tank” should stop pretending to have proven that it is.  It is hard to prove anything conclusively in economics, but considerable evidence points in the opposite direction of what the EPI has asserted, again and again.