Every January, the U.S. Bureau of Labor Statistics releases new data on how many workers in the United States are members of unions or are covered by union contracts. Today, we learn that union membership continues along a sloping downward trend. In 1983, the share of U.S. workers who were members of union was 20.1 percent. Today that share is down to 10.7 percent.

A low rate of union coverage does not bode well for the future of the middle class. When unions covered a high share of U.S. workers, this supported middle-class incomes. As the U.S. economy became more productive and businesses became richer, the demise of unions leaves U.S. workers without the ability to garner any of the benefits of growth.

Yet, even as the share of people covered by a union contract falls, polls consistently show that people want to be members of unions. According to Gallop, six-in-ten people approve of unions. If unions strengthen the middle class and most approve of them, then why isn’t union coverage on the rise?

First, there are some structural issues. Union density historically was highest within the manufacturing sector. The fall in union coverage for private-sector workers has paralleled the long-term decline in the number of manufacturing jobs in the United States as the share of manufacturing in the U.S. overall gross domestic product has slid. Employment has flowed into industries that were traditionally less unionized.

Why haven’t those industries seen an increase in unionization? In short, it’s become harder to organize a union and deliver a first contract. For workers to be covered by a union, they need to hold an election. That sounds like a fair process, but employers have a lot of leeway to “encourage” workers to vote against the union. Further, even if workers vote yes, it often takes years to get a first contract. Economist John-Paul Ferguson of Stanford Graduate School of Business, found that in the 44 percent of organizing drives where the union won the election (36 percent of all organizing drives), two years later, the workers still did not have a contract.

There also are a couple of interesting trends in union coverage. First, the diversity of union members has increased over time. Where unions have grown, it’s been through organizing workers who look like the U.S. population more generally. Union members are still more likely to be male but increasingly less so. As of 2016, 47 percent of those covered by a union contract are women, up from 35 percent in 1983. The same trend exists for blacks, Asians, and Hispanics and Latinos. In 2000, the first year of data available for these groups, they together comprised 29 percent of workers covered by a union. As of 2016, they now account for 35 percent of covered workers.

Second, union membership remains more common among government employees. About one third—34 percent of public-sector workers—are unionized. This rate is about five times as high as among those in the private sector, where union coverage is only about one-in-sixteen, or 6 percent. Current policy trends, however, may push against this as 27 states—Kentucky most recently—have implemented so-called “right to work” laws that prohibit unions from requiring those covered by a contract to pay full dues while federal policymakers have expressed a desire in recent Congresses to pass a national variant of such legislation.

Decades ago, the path to the middle class was through strong unions. President Trump has promised to “make America great again” in part by bringing back jobs in union-heavy industries—and he did well among union households during the election, particularly in the Midwest. It remains to be seen whether the economic policies of the current administration and Congress will focus on declining unionization rates or will be able to bring the same kinds of economic gains to workers and their families that unions brought.