In a previous post this author looked at how cities and counties that have passed “living wage” ordinances differ from the rest of the U.S. on a variety of measures. We’ll now analyze which variables predict the amount of the increase in the minimum above the federal rate of $7.25 per hour.

Previously a model was run with the percentage of population change as a predictor for the amount of the living wage. Some have challenged the percentage change in the population as an indicator of the economic vitality of an area.

The model below looks at the variables as predictors of the amount of the living wage in a multiple regression model for the 38 cities and counties with living wages. This will aid in finding the most robust predictors of the increase.

**MODEL**

The final model presented below accounts for 38 percent of the variability in the living wage with three predictors giving the final equation: Living Wage ($) = $12.57 + $0.12 *(% uninsured) – $0.59*(% veterans) + $0.14*(% pop change)

This equation states that for a city with zero percent uninsured, no veterans and an unchanged population the living wage would be $12.57 an hour.

For every one percent increase in the uninsured, an increase of 12 cents in the wage is expected. For every one percent increase in the population of veterans the amount of the wage decreases by 59 cents. Finally, for every one percent increase in the population since 2010, the wage is expected to increase by 14 cents.

Of these three predictor variables, the percent of the population who are veterans is the only one that is statistically significant with the others being borderline significant. In a simple linear regression model with only the percentage of veterans as a predictor, 28 percent of the variability is explained.

The scatterplot below shows the negative relationship between the percentage of veterans and the amount of the living wage passed. There is a cluster of cities, mostly in California, that passed a wage of $15 per hour or higher that had zero association with the percentage of veterans. If these cities were excluded, the regression line would be even more negative.

This model suggests that as the percentage of veterans in a city or county increases, the amount of their minimum wage decreases. This result is surprising but one must be careful in inferring a cause and effect relationship among significant variables in a regression model. For example, a high percentage of veterans are homeless.

There could also be some other variable mediating this relationship. Further research in this area is needed.

[Governing] [National Coalition for the Homeless] [Photo courtesy AP/Seth Perlman via The Nation]