Does the Trade Deficit Destroy American Jobs?

Spencer,

Thanks for the clarification.

IMO, the decline in U.S. manufacturing jobs is much more the result of automation and relaxed work rules. I don't have the data right at hand to support that opinion. But I did earlier supply data showing that direct employment in the auto industry declined sharply even though the number of autos produced in the U.S. has remained the same. I could probably find similar data for other industries, but, quite frankly, I don't have the time right now.

IMO, our trade deficit is as much or more due to increased purchases of U.S. assets than anything else. I don't see anything wrong when Toyota, Nissan, Honda, and BMW invest capital in U.S. auto plants. Or when BP expands and refurbishes its U.S. refinieries. Or when Dutch investors diversify and buy U.S. stocks. I also see no risk in China buying U.S. government debt.

I guess we're always going to disagree. I'm very optimistic about the U.S. economy, and confident the good jobs will remain plentiful over the long run for U.S. workers. You don't seem to share my optimism, and I don't think anything I write will change your mind.

I'm refering to two things. One, the change in the composition of manufacturing output. Second, the break in trend of manufacturing employment.

the original post was that the trade deficit had not impacted employment and manufacturing employment data was presented to support that argument. But if you look at manufacturing employment growth you see a major break in trend at about the time the trade deficit emerged. Prior to the late 1970s-- when we ran a trade surplus manufacturing employment grew. After the late 1970s when the trade deficit expanded manufacturing employment fell.

So the manufacturing employment data actually supports the argument that

trade has cost the US jobs.

But to be complete you need to see if the influx of foreign capital financed the creation of other jobs to offset the apparent negative impact on manufacturing jobs.

I am simply making the point that the data Russ presented does not support the case he is making.

Spencer,

Thanks for the data and graphs. Now that I've seen the data, I'm not sure what to make of it.

First, why do you refer to 2.5% growth as "extremely slow"? If the overall GDP is growing at about 3.2%, and consumers are showing a preference for high tech goods and services, then isn't 2.5% what we should expect for non-tech goods?

Second, I'm a little confused by your post. The first part refers to the growth in manufacturing output. I think the second part refers to the declining growth in manufacturing employment. Forgive me, but I'm not understanding the point of your comment about " a clear break in trend in the late 1970s".

Are you commenting about how productivity increases have reduced the need for factory labor? about how the decline in union power, and the resulting relaxation in work rules, reduced the need for factory workers in some industries? about how factory jobs moving to Southern states reduced union power? about how corporations moved labor-intensive industry offshore?

My guess is that the decline in U.S. manufacturing jobs resulted from all of those factors.

Thanks, that's the best and most clear explanation of the topic I've ever seen.

Go to the Federal reserve industrial production data where one of the ways they publish the data is with this breakdown.

this actually creates an interesting data problem. The Fed weights the industrial production data by the value of nominal output. But since IT prices have been falling so rapidly it roughly offsets the more rapid output growth. So even though IT growth has been roughly 10 times "all other" growth for some 20-30 years its weight in the index has barely changed.

It is a very interesting statistical question to address.

spencer: "you find that since the 1970s the growth of high tech has averaged something on the order of 20% to 25% annual rates while for the all other, or traditional manufacturing the growth has been more on the order of 2.5% annually."

Where did you obtain these numbers? I've been looking at 1977-2005 industry GDP data at the BEA site. It is true the 'Computer and electronic products' industry realized above average growth. But that growth was hardly ten times the growth of other industries. Can you point me to your data source?

Actually, I agree with you that the trade impact on US manufacturing appears to be insignificant. The real story in both manufacturing output and employment is the change in composition and productivity growth. If you break manufacturing output into two components, high tech and all other you find that since the 1970s the growth of high tech has averaged something on the order of 20% to 25% annual rates while for the all other, or traditional manufacturing the growth has been more on the order of 2.5% annually.

How much of the extremely slow growth in all other is a product of import displacement is an interest question.

But my point is that there was a clear break in trend in the late 1970s -- exactly when is difficult to determine because of cyclical factors. But if you look at the growth rate by decade you see little difference between the 1940s, 1950s, 1960s and 1970s. So how did WW II, Korea, and Vietnam cause manufacturing employment to grow at about the same rate in the 1970s as in the earlier decades?

Prof. Roberts, this is an excellent presentation. It actually reminds me of a quote from a Krugman essay on comparative advantage ( http://web.mit.edu/krugman/www/ricardo.htm ):

"During the NAFTA debates I shared a podium with an experienced, highly regarded U.S. trade negotiator, a strong NAFTA suppporter. At one point a member of the audience asked me what I thought the effect of NAFTA would be on the number of jobs in the United States; when I replied 'none', based on the standard arguments, the trade official exploded in anger: 'It's remarks like that which explain why people hate economists!'"

Professor Roberts, thanks for this excellent overview. I plan to share this link with friends and coworkers.

I'm surprised manufacturing employment hasn't fallen more since the mid-70's. Computer power has gotten so cheap - and programming so sophisticated - that nearly every manufacturing task should now be automated. Perhaps the delay is due to lack of engineering and computer progamming talent.

Spencer,

Yes, a comparison of post-1976 and pre-1976 finds that the growth rates were different. The numbers you show are distorted by the huge increases of World War II, Korea and Vietnam. But forget that. You need to show more than a difference in pre-1976 and post-1976 growth rates. You need to show a relationship—if deficits reduce employment in manufacturing, big deficits should cause a bigger decrease than small deficits.

Between 1983 and 2000, manufacturing employment is flat. In 1983, there were 17,363,000 manufacturing workers. In 2000, there were 17,322,000. Shouldn't the massive trade deficits between 1983 and 2000 have had an effect?

Of course, it's possible that the effects of the trade deficit were exactly canceled out by other effects. I don't know of any work that supports the claim.

Actually, if you look at the manufacturing employment data you present there is a clear break in the data around 1976 -- the actual peak is in 1979.

From 1939 to 1976 maufacturing employment grows at a 1.88% average growth rate. After 1976 the average growth rate is -0.67%.

If you do a best fit trend analysis you get a parabola with the peak in the late 1970s.