The long run trend for U.S. employment is up. The unemployment rate fluctuates with the business cycle, but any trend is at best minimal. Still, there are constant complaints that imported goods are destroying jobs. Or perhaps it is just the "good jobs." And what are these good jobs? They are factory jobs were men of modest education can earn high wages and benefits. These jobs are supposed to allow those workers to be part of the middle class.
Surely, this is why the loss of manufacturing employment is counted as a major concern. These middle-class jobs disappear and some of those losing the jobs, or perhaps just their children and grandchildren, must must accept low paying jobs in the service sector. Of course, some that are more ambitious may accept more responsibility and risk, or at the very least, seek more formal education, allowing for more skilled work. Even so, people who are little different in terms of skills and attitudes from those who had "good jobs" in the past must now take substantially worse jobs.
A simple model of unionization has the union increasing wages in the union sector. The quantity of labor demanded by the firms in that sector is lower, reducing employment. The workers who would have worked in the unionized sector seek employment in the nonunion sector. The increase in supply in the nonunion sector lowers wages in that sector. In the simplest model, the workers are identical, so the result is that identical workers earn differential wages depending on their industry. Wages are above the competitive level in the union sector and below the competitive level in the nonunion sector.
In the nonunion sector, the labor market clears. In the union sector, there is a surplus of labor. Workers from the nonunion sector would prefer "good jobs" in the union sector. If the union sector is "manufacturing" and the nonunion sector is "services," then this would explain why manufacturing is identified with "good jobs" that are "scarce" and the service sector are "bad jobs."
Unions took off in the U.S. during the Great Depression. In my view, this was mostly due to money illusion. There was massive deflation during the first part of the thirties, and substantial decreases in nominal wages. While real wages actually increased, workers became very interested in joining unions in order to fight the unfair pay cuts. Federal government policy changed to strongly support unionization, but the workers supported unionization to fight nominal pay cuts despite growing real wages.
As time passed, the unionized workforce became less significant, mostly because the growth of unionization failed to keep up with the growth of the labor force. However, many years ago, someone from "management" once explained that there is little benefit for workers to join a union because employers provide pay and benefits for nonunion jobs that are competitive with union pay and benefits. There appears to be substantial truth to this notion, most obviously in industries and even firms that have both union and nonunion operations. Keeping pay and benefits low in the nonunion shop is just asking for an organization drive and the loss of the election.
This suggests that the proper division in the simple model is not between the union and nonunion sectors, but rather between the "easy to unionize" and "difficult to unionize" sectors. If manufacturing is on the whole easy to unionize and the service sector is difficult to unionize, then manufacturing will provide "good jobs" that pay more than the competitive amount and the service sector will have poor jobs that pay less than the competitive amount.
It is certainly plausible that manufacturing is easy to unionize because of economies of scale. There are also substantial sunk costs, which makes exit difficult, which in turn makes entry risky. In the rest of this post, I will assume manufacturing is easy to unionize and the result is higher than competitive wages in manufacturing. The service sector is difficult to unionize and so results in lower wages.
This ties to trade because it is a way to bypass the inefficiency created by unionization. The reduced employment in manufacturing results in too low output and too high prices. The shift of labor to the nonunion sector results in too high output and too low prices.
By importing manufactured goods, those in the service sector obtain products at lower prices. This raises their real income. The domestic manufacturing industry, which is already too small, reduces output further. However, the need to meet foreign competition lowers their too high prices. The reduction in employment in the manufacturing sector increases the supply of labor to the service sector, resulting in lower wages.
Trade must balance, but it is possible to export services. Tourism is an obvious example, and there are various sorts of financial services that can be provided to foreigners. It is also possible that a net capital inflow could fund imports of manufactured goods. Foreign investment funding an expansion of the service sector would fit in well with this account.
Certainly, this story does not account for all of the U.S. experience in the late twentieth century. The simple model ignores sorting in a labor market where workers are not all the same. Sectors with excessive wages and and a surplus of labor will tend to hire what they perceive to be higher quality workers. To some degree, workers left in the low wage sector may be less productive. Manufacturing output has generally increased in the U.S. and not disappeared. However, the "problem" of a lack of high paying jobs for workers with little education is not solved by a demand for highly-skilled workers in manufacturing.
Still, I think it does tell us something about the "problem" of the loss of "good jobs." That just doesn't make much sense in a competitive labor market. We can image shifts in the share of income going to labor and capital due to changes in trade or technology. These changes could tend to depress real wages. These changes simultaneously expand real output so that the net result is ambiguous. But these processes do not appear to create the phenomenon of the loss of "good jobs" in import competing industries.
If a single industry were unionized or were simply subject to unionization, those working in that sector would almost certainly benefit. They would receive a larger share of a very slightly smaller pie. When all manufacturing is unionized or even subject to unionization, the loss in total efficiency is more substantial. The unionized autoworker pays more for shoes produced by union labor. The expansion of imports similarly has ambiguous effects. The union shoe maker can buy a cheaper Korean car, while the union autoworker can buy cheaper Mexican shoes. Still, the analysis treating "manufacturing" as an aggregate provides some element of truth. Those keeping the unionized or unionizable jobs get cheaper haircuts and the barber pays more for cars and shoes. An expansion of imports allows the barber to get cheaper cars and shoes, even if there are more former autoworkers and shoemakers who want to set up barber shops.
Globally, a pattern of international trade that develops because of unionization is inefficient. World output and income may be higher than without the trade, but it would be higher still if wages in the unionized and unionizable sector were competitive with wages in the service sector. That is, if workers in the service sector did not covet "good jobs" in manufacturing, and workers in manufacturing did not see service sector jobs as undesirable. To the degree this makes the domestic production of manufactured goods more profitable and expands the manufacturing sector at the expense of the service sector, the result would be improved global efficiency.