Right to Work (RTW) laws weaken unions by depriving them of the funding they need to be effective, and workers, both union and non-union alike, in RTW states have lower wages. No one really disputes the first fact—workers in non-RTW states are more than twice as likely (2.4 times) to be in a union or protected by a union contract. And wages in RTW states are far lower—almost 16 percent on average. This isn’t surprising, since RTW’s proponents are anti-union hate groups and business organizations that oppose every effort to help workers organize or raise wages. In fact, their key pitch to legislators (outside of campaign contributions) is that RTW will lower labor costs, improve the “business climate,” and encourage out-of-state businesses to relocate.

So it was surprising to see the Heritage Foundation challenge the notion that RTW has no effect on a state’s wage levels. Yes, they say, wages are lower in RTW states, but it isn’t because of RTW. If true, it would leave proponents with no argument for RTW except its core purpose—weakening unions.

But in fact, it’s not true. EPI senior economist Elise Gould and co-author Will Kimball examined the Heritage report and found it to be deeply flawed. Heritage’s finding depends on statistical tricks—the removal of relevant and standard labor market controls such as the worker’s industry, and the inclusion of nonstandard and irrelevant worker characteristics and state-level amenities. Using only standard and relevant factors in the regression analysis yields a consistent finding: wages in RTW states are 3.1 percent lower than those in non-RTW states, after controlling for a full complement of individual demographic and socioeconomic factors as well as state macroeconomic indicators. This translates into RTW being associated with $1,558 lower annual wages for a typical full-time, full-year worker.

Regression analyses like the one done by Elise Gould are not simple to perform, but they aren’t hard to understand. Starting from the observation that RTW states have 15.8 percent lower wages, on average, than free bargaining states that don’t forbid union fair share agreements—$20.66 vs. $23.93—the researcher tries to determine the precise relationship between the two variables: wages and RTW status. Are wages lower because the cost of living is lower in RTW states, or because educational levels are lower, or because of the age and race of the typical worker? Controlling for each of these factors and every other relevant factor, holding them equal, one can conclude that the residual correlation is the relationship that RTW laws have with wages.

But if one excludes factors, as Heritage did, that are obviously relevant—factors that do affect wage levels such as unemployment rates, the type of industry in the state, union status, whether workers are full-time or part-time—the results will be skewed. Heritage argues that controlling for occupation or industry would hide any positive effect RTW has from attracting manufacturing jobs to a state, but the truth is that there is no evidence that RTW has that effect. Site location surveys show that manufacturers don’t decide where to locate based on whether a state is fair share or RTW.

But the most serious problems with the Heritage approach is its use of idiosyncratic controls that don’t belong in the model. Heritage argues, for example, that wages are lower in RTW states because workers have traded them off for factors such as living in a state that borders an ocean, but there is no evidence that it’s true. Most workers, after all, live in the state in which they were born, and almost no working age household moves to another state moves because of the climate. Only 36 percent move for any job-related reason. And, just because a state borders the ocean, doesn’t mean that everyone lives near the ocean and enjoys that amenity.

The fact is that RTW laws do exactly what their proponents, such as Missouri state Rep. Bill Lant, claim: they lower wages significantly for non-union and union workers alike, by $1,558 on average, according to the best estimate.