Robert Z. Lawrence is Albert L. Williams Professor of International Trade and Investment at Harvard's Kennedy School of Government.

He served on President Clinton's Council of Economic Advisers from 1999 to 2001, and writes frequently on issues surrounding globalization and trade policy. His latest book, coauthored with the University of Cape Town's Lawrence Edwards, is "Rising Tide: Is Growth in Emerging Economies Good for the United States?" We spoke on the phone Monday morning. A lightly edited transcript follows.

Dylan Matthews: When I posted your graph (above) showing that manufacturing employment has declined at the same rate since the 1960s, one response was to point to the work by David Autor and others suggesting that trade has had a big impact in manufacturing's decline as well. What do you say to that?

Robert Z. Lawrence: I’m not saying that there wasn’t some effect due to trade, especially over the long run. I’m not saying that trade has no effect. The regression line is obviously capturing a lot of things, but it is a line and it’s a remarkably persistent one. The Autor paper focuses on the effect of imports, but there's nothing in the paper that touches on what happened to the trade deficit.

There’s another picture where we calculated the job content of the trade deficit, so you could look at a picture of what would have happened to employment if you add the jobs represented in the trade deficit. If you add the job content of the trade deficit, you find exactly the same decline in employment, even though the level of employment would have been higher. The decline would be exactly the same over the decade. What happened over the decade is that our manufacturing trade imbalance almost doubled. You’d think, gee, that ought to lead to a decline in jobs. But the trade deficit did not change.

The point I’m trying to make is that while we definitely did have some displacement and we have some evidence of the displacement that occurred, it's the trade deficit you should look at, not what goes on in one specific part of the trade balance.

People who get upset by this also say, "Listen, up until 2000 manufacturing employment was essentially flat, and since 2000 it has fallen by 40 percent. It’s an unusual behavior, it’s a new behavior, it has to be something else." But the key point is that in the period during which manufacturing employment was flat, aggregate employment was growing by 2 percent a year. If you do the change in manufacturing employment as a function of the change in total employment, and you stop it in 2000 and run the regression, and then I tell you there was no growth in employment in the 2000s, the regression would tell you that manufacturing employment would fall 4 percent every year. And that's exactly what happened.

Matthews: Productivity increases don't necessarily lead to job losses if they mean more people are buying the goods now that they're cheaper. But people are buying fewer goods and more services these days. What's driving that?

Lawrence: Think about this example. Tell me about your cell phone. How much do you pay for your cell phone, and how much do you pay Verizon for your services? They’re giving you the phone. For some goods there’s complementarity. Your television is great and it’s come down in price, but what you want are more cable services, more entertainment. Some of the most innovative new goods have created a disproportionate demand for services. That’s one element to this.

I don’t believe, by the way, that this is necessarily the case for developing countries. When you’re going through a period when you’re increasing the stock of automobiles in your country, that’s surely going to be a period where the demand for goods will rise, perhaps more quickly than income. Look at the cars we’re going to be buying. The aggregate volume of auto sales in the United States is not going to be rising. We’ve saturated the marketplace. That seems to be the big driver of this.

I think it’s quite analogous to what’s happened with food and farming and therefore with agriculture. When you earn more money, you don't buy that much more food. You say, "Great, my budget can allocate only 13 percent of my spending on food." And in developing countries, it’s 50 percent. In the '90s when we were buying a lot of electronic equipment and computers. Since 2000 the share of spending on equipment has fallen in half.

So that’s what I’m saying. There could be Chinese stories of trade causing this, and there’s lots of things going on. You can’t just pick one cause and say that’s why this whole thing has happened. What people haven’t focused on is what’s happening to the demand for goods. We may not be very good producers, but no one denies our prowess at consumption. That we do well. So coming at it from the goods side, and the demand for goods, that’s the potential market that’s out there at this time. Particularly when it comes to electronics, after Y2K and the dot-com bubble bursting, there was this massive decline in equipment spending.

A manufacturing worker with some really hot metal. (Ty Wright/BLOOMBERG)

Matthews: One of my commenters asked a good question when I posted that graph, to the effect of, "Let's see China on that comparison chart of international manufacturing employment." There's a kernel of a point there, that if manufacturing is getting more productive, you shouldn't need as many workers in China either.

Lawrence: Chinese manufacturing employment is pretty flat. I’ve tried to look at this, and the BLS has a sort of a survey of Chinese employment. It’s basically flat. What’s hard to track is they have urban and they have rural employment, and the series isn’t very consistent.

But here’s something that is easy to get if you wanted to look at it. People have an idea that one of the huge deals is that plants have closed in the United States and moved to the rest of the world. In American companies we’ve had offshoring. Well, we have data on that. Just imagine every job that we know was added throughout all the foreign affiliates of American companies over the last decade was an offshored job. This is as extreme as you could get. You get something like 600,000. According to the BEA, from 1999 until 2009, employment in majority-owned manufacturing affiliates, so these are the U.S. companies, increased by 638,000. Most of which was added in Asia and the Pacific.

Now, in other words, if every job added since 2000 had instead been added in the United States, employment in these companies would have only been 9.25 percent larger. The additional 638,000 jobs amount to 11 percent of the overall loss in manufacturing employment. Surely this is an upper bound because first, U.S. workers are more productive than workers in other countries, and some of the jobs added are servicing foreign markets.

Matthews: But that doesn't include contracting companies. Foxconn isn't a subsidiary of Apple, for instance.

Lawrence: That’s not all the offshoring. What I’m focusing on here is mainly the idea that, because of tax incentives, we’ve got our companies closing down plants here and going abroad. There’s a whole debate here about U.S. tax policy, and the administration’s argument is that we’re subsidizing the movement of plants abroad. People can defer their tax payments by sending money abroad, but it’s really more to that. But your point is well taken. It’s not a total estimate of offshoring.

Matthews: All else being equal, industries where productivity increases tend to also see wage increases. That hasn't happened in manufacturing.

Lawrence: In a book I did three years ago, it was called Blue Collar Blues, I went through and I asked, "How would you explain why it is that average wages have been so flat when output per worker has risen so much? And how do you decompose that deviation?"

And it’s quite a complex story. But how much of that is inequality? Someone’s walking off with the money and the workers aren’t getting it. How much of it is other factors? The answer, up until about 2000, was that to the degree that the worker wasn’t getting the money, college-educated workers were doing well. Since 2000, there is a profit story. So the share of profits has risen. It wasn’t so clear even a few years ago but for the moment, profits are a high share. So that inequality is present.

But one part of it is that when you’re looking at the measure of output-per-worker, you’re looking at what workers produce. When you’re looking at real wages, you’re looking at what workers consume. There's a difference. One big difference is in computers and equipment. When productivity growth is fast in equipment, it’s not a big share of what workers buy. Or take housing. Workers don’t produce all of the housing, though there’s construction. So about a third of the whole difference is the difference between what workers produce and what they consume.

The other part of it had to do with the difference between compensation, which includes benefits and wages, which is generally your take-home pay. You’re looking at what people get in their pockets. When your benefits are part of your compensation, there’s some part of that story as well. A third part is that we’ve been adding more and more educated workers to the workforce. So if you have more college-educated people working, their wages will rise to reflect their higher intellectual capital but there’s no reason why the high school people should get a higher wage.

So I decomposed all of those and tried to come up with some estimates of how much is there for those blue-collar workers to be paid. I think, in recent years, there is a profit story. And then there was another part of this which was the very top end, who have walked off with increasing shares. I was looking just at the wage story. The top 0.1 percent earned 6 percent and by 2000 they were earning 12 percent. There is a story about those workers. Now, people at that level are often paid in stock options, but when you exercise it it shows up in wages. It’s quite mind-boggling, it’s even on your W-2.

Matthews: Any final points you want to make?

Lawrence: The other one that drives me crazy is the idea that all imports represent job loss. And the reason is a failure to realize that the major reason for import growth is domestic production and employment growth. And you still have people out there who look at a trade deficit getting larger and say, "That means jobs are getting lost."

What I liked so much about the Hanson/Autor work is that they distinguished between demand and supply. If the Chinese ability to supply something increases, that can cause job loss in the United States, but the biggest reason for import growth, and the dominant one, is the fact that the U.S. economy is growing rapidly, and that’s good for employment. So a really healthy U.S. economy, particularly with our big government deficit, will have a bigger trade deficit. That will be a healthy, export-driven increase in imports.

They’re completely confused between ex ante and ex post. They use the ex post trade data without distinguishing why the deficit has grown, and if it grows because there’s a shock, you increase imports, but if it increases because you’re spending more at home, that’s going to be associated with positive job growth and what I showed (above) actually is the way in which, in the U.S. case, the dominant driver of imports is our domestic spending. 2/3 of our imports are equipment or intermediate goods we use for production.

Matthews: Really? That high?

Lawrence: Yes, 63 percent or something of our imports are either equipment or intermediate inputs. And another 20 percent is oil. So imported consumer goods are a quite small part of the import picture.