Union Membership Is On The Decline In America

The Bureau of Labor Statistics released new annual data on Wednesday showing a continued decline in union membership rates in 2012. This is the percent of wage and salary workers who were members of a union, and the release shows that union membership fell to 11.3% in 2012 from a level of 11.8% in 2011.

The BLS showed that union membership fell to 14.4 million in 2012. When compared to 1983 as the first year of comparable data, that figure was 20.1% of workers with some 17.7 million union workers counted. We would note that the BLS data is based upon a monthly sample survey of about 60,000 households.

Unionized rates of public-sector workers came to 35.9%. That is more than five times higher than that of private sector workers at 6.6%. Local government workers had the highest union membership rate at 41.7%. Union membership participation rates were higher for men at 12% versus 10.5% for women in 2012. The report also broke out some data by age and by racial group as well. It is of little surprise, but part-time worker union participation rates are less than half that of full-time workers. (FULL DATA)

Some 15.9 million wage and salary workers were represented by a union in 2012, but we would note that this includes the 14.4 million union workers and those 1.6 million workers who report no union affiliation but whose jobs are covered by a union contract. Full-time wage and salary workers had median usual weekly earnings of $943 for union members versus $742 for non-union workers.

There was some interesting data presented by state as well. North Carolina had the lowest union rate of 2.9%, while New York was the highest at 23.2%. Texas had a union membership rate of only 5.7%, which is about one-third as many union members as New York despite Texas having 2.7 million more wage and salary employees.

With this being 60,000 households we were curious as to how accurate the figures would be when factoring in data from states and broader locales. The BLS showed in its methodology that there is about a 90% level of confidence that an estimate based on a sample will differ by no more than 1.6 standard errors from the “true” population value because of sampling error.