|Continuing the debate from "The Speed Limit," by Alan S. Blinder, and "Why We Can Grow Faster," by Barry Bluestone and Bennett Harrison (September-October 1997).|
In your recent article ["The Speed Limit: Fact and Fancy in the Growth Debate," TAP, September-October 1997], we very much appreciate your modest, wonderfully humorous, and clearly stated reasoning against betting too heavily on faster growth. We share with you the view that the more capital-friendly tax and regulatory policies advocated by the right (and, increasingly, by the center as well) will not cause the economy to grow faster, in either the short or long run, and would only serve to make the distribution of income between labor and capital even more unequal than it is already. As you will see in this letter, we also share other points of concurrence.
But we apparently have an honest disagreement on the central point. We see many signs of a higher potential growth rate; you see growth continuing at pretty much the same average rate as it has over the past five years or so. We, who wrote The Deindustrialization of America and The Great U-Turn about growing inequality, find it more than a little amusing to be cast as optimists after being dubbed Dr. Doom and Dr. Gloom for so many years! We share your soft heart; we don't think we have gone soft in the head.
In the short run, it would seem that a lot of this debate about maximum feasible potential output growth turns on whether and to what extent the supply of labor will continue to increase. We think that labor force participation (and, more important, hours of desired work) are not fixed; in economists' parlance, they are "endogenous." The expectation of continued stagnant wages and anxiety about unexpected future loss of income are impelling more people to seek work for more hours than used to be the case. This is not a new idea; remember the "added worker effect" that we were all taught when we were graduate students of labor economics, in which we learned how spouses responded to the cyclically changing employment and earnings opportunities of "heads." Times have clearly changed; even the concept of a household "head" is rightfully disappearing from the economist's lexicon. But isn't the idea of endogenous supply response sound at the level of the individual, as well as at the level of the household? We freely admit that if and when there is a substantial increase in real wage rates and improvements in job security, labor supply growth could slow again. But this may be years away and by then other factors, notably productivity improvements, can take up the slack.
You state that "the unemployment rate is about as good a measure of aggregate pressure on capacity as we have." It may indeed be the best we have, but that's not saying too much. It certainly does not make sense to use a misleading indicator as a basis for setting fiscal or monetary policy. Because of fundamental changes in the nature of working time, it is not too farfetched to say that the official unemployment rate is now obsolete and should be retired. Today, with "moonlighting" on the rise at one end of the spectrum and a rise in involuntary part-time employment at the other, the issue of working time is more important than ever. To continue to depend on the official unemployment rate—instead of on working time—as our measure of capacity constraint badly misconstrues reality by understating potential labor supply.
You seem to be farthest from us, however, on the issue of productivity. After reviewing all of the failed attempts by economists to understand the sources of growth and particularly the "productivity paradox" of computers, we have turned like others to the work of economic historians who have shown that productivity gains from new technologies follow a nonlinear path. Initially, a new technological revolution reduces average productivity because it takes time to get the bugs out, because diffusion is often slow, and because all kinds of complementary changes are necessary in organizational structure to take advantage of the new techniques. In a sense, we have been wandering in the information technology desert for close to three decades. Our forebears trudged through similar deserts in the years following the introductions of agricultural innovations, the steam engine, and electricity. Now, with computers, we are at least in sight of the promised land. This allows for a much more sanguine view of growth potential than a linear extrapolation from recent past experience does.
A deeper understanding of the growth process can send a signal to business about expected future opportunities for profit, which in turn induce accelerated investments in innovative equipment and ideas, and encourage "tinkering" and external learning spillovers, all of which, in turn, raise the growth rate of potential output.
A little bit of doom and gloom can get governments and nations to change their behavior, and being hardheaded in the use of theory and statistics is generally a good thing. But in this case, a slow-growth mentality on the part of government can become a self-fulfilling prophecy when it comes to business.
We're not looking for a miracle. We just believe there is enough uncounted labor capacity and enough improved productivity to give us at least another seven-tenths of a percentage point in annual growth into the foreseeable future.
P.S. Just for the record: We, too, would love to see the Dodgers move back to Brooklyn. And while we're at it, how about getting the Giants back to Manhattan? At least that might stop George Steinbrenner from threatening to take the Yankees out of the South Bronx!
Dear Barry and Bennett:
As a long-time resident of New Jersey, I cannot endorse your ridiculous idea to move the Giants to Manhattan—unless, of course, you mean the San Francisco Giants. Other than that, I think our disagreements make pretty thin gruel.
You claim that we might add at least 0.7 percent to the growth rate "into the foreseeable future." I wrote that intelligent policies might conceivably add as much as 0.5 percent "for a time." None of us is signing on to the Dole-Kemp, or even the Business Week, party line. Nonetheless, my best guess is notably less optimistic than yours. We appear to have two disagreements: over labor supply and productivity.
Regarding the first, I of course agree that labor force participation is endogenous. But I always thought it responded more to employment prospects (measured, say, by the unemployment rate) than to real wages. Until very recently, one of the puzzling aspects of the current expansion was the failure of the labor force participation rate to exhibit its usual cyclical bounce. It was 66.7 percent in the first quarter of 1994 and it was still 66.7 percent in the second quarter of 1996. Since then, as you noted in your article ["Why We Can Grow Faster," TAP, September-October 1997], it has been creeping up. It's about time! But I would hesitate to extrapolate this behavior "into the foreseeable future."
You write about longer workweeks. Here again we must be careful to distinguish cycle from trend. Your article used a 1991 base year, which was a cyclical trough. The average workweek of private nonagricultural workers was 34.6 hours in 1989 (a cyclical peak year) and is again averaging 34.6 so far in 1997. History suggests that, in the long run, this series is pointing down, not up.
Regarding productivity, no one can confidently predict the future of what we used to call "the residual" because we couldn't explain its past. Who foresaw the dramatic post-1973 slowdown, for example, or can explain it even now? So no one can smugly rule out a sudden acceleration. But to me, our ignorance argues for conservatism in forecasting—such as predicting a continuation of the trend for the last 24 years—not for assuming that we are "on the verge of a productivity renaissance."
You seem to want to bet your pile of chips on the microchip, which has been around for 25 years without giving productivity a boost. Good luck. There are many reasons why the computer may have failed to deliver on its promises. Some of them apply to the future as well as the past. But hey, what do I know? As I admitted in my article, the big productivity gains from computers may be just around the corner. I'm just not willing to bet on it without seeing some evidence.
You say that history tells us that productivity lags behind major technological innovations by two or three decades. If so, the computer, like the proverbial slumping batter, is "due." But these historical parallels strike me as less than compelling—for two reasons.
First, doesn't the world move a lot faster now than it did in the nineteenth century? Doesn't technology diffuse around the globe much faster? Look how quickly fax machines and cell phones proliferated, or how fast the Internet grew. In an economic sense, then, isn't three decades a lot longer today than it was a century ago? And yet we are still waiting for businesses to learn how to squeeze productivity out of the computer.
Second, as Dan Sichel has suggested, the industrial transformation needed to replace typewriters and mainframes by PCs looks to be far less revolutionary and disruptive than that required to replace steam power by electricity. Information highways are more easily built than railroads. If the transition costs that accompany the microprocessor are far smaller than those that accompanied the dynamo, then PC technology should have diffused much faster than electricity—as it probably did. So why are we still waiting?
One last point of contention: I do not see how you can accuse a Federal Reserve that has let the unemployment rate drift down to 4.8 percent (and is still, as of late August, not raising interest rates!) of having "a powerful monetary policy bias against faster growth." In fact, Alan Greenspan has expressed views on the productivity impact of computers that are far closer to yours than to mine. I hope the three of you are right.
Alan S. Blinder
Don't worry so much about the Jersey Giants. Neither of us has any designs on them. It's those guys in San Francisco we want back. We recognize that these days football is the high productivity sport, but we still like baseball.
That we are more optimistic about growth comes from our reading of recent data on both labor supply and productivity. On the labor supply issue, you point to little growth in the labor force participation rate as one reason for your greater pessimism. We agree that there has been little change over time in this source of labor supply, at least until very recently.
But the important development is in the growth in hours of work among incumbent workers by week and by year. You invoke payroll data to show that there is no cyclically adjusted increase in hours of work. But that's the wrong measure. Data from household surveys show a substantial increase since the last business peak in 1989. Between 1989 and 1995, the average workweek of individual prime-age workers has increased from 40.6 to 40.9 hours—an increase of 0.7 percent. Moreover, this increase in the workweek is faster than in either of the two previous economic recoveries. In addition, weeks worked per year for the same population have increased from 47.1 to 47.5 weeks. This means that the typical prime-age worker is putting in 30 more hours per year than at the previous peak. This increase in labor supply is not related to the unemployment rate; it is related to stagnating wage rates and anticipated job insecurity. As long as wages continue to stagnate, and job instability continues to grow, there will be some pressure on workers to offer more hours when the demand is there.
As for productivity, you reason that accelerating technology cycles should somehow reduce the lags in adoption and diffusion, so that we should have seen the payoff to the information technology revolution by now. But it's just the opposite. Shorter shelf lives on new products and processes—including computer hardware and software—make premature adoption expensive and risky and encourage delays. Moving from one learning curve to the next before getting the full benefit of a new technology therefore adds to the lag in both adoption and diffusion. But, over time, people and organizations do learn. Thus using straight-line projections to forecast productivity leads to a more pessimistic perspective than we think is warranted.
We applaud the fact that Alan Greenspan has prevailed over those within the Fed who would have raised interest rates by now. We hope that he will continue to restrain the natural tendency of central bankers to anticipate ruinous inflation around every corner. We're nervous about the possibility that the Fed will raise interest rates at the first sign that workers are becoming even a little less anxious about their own job security.
But our concern is not solely with monetary policy. The recent budget accord further compromises public investment in precisely those areas most crucial to stimulating long-term output—infrastructure, K-12 education, job training, and civilian research and development. Unfortunately, the Fed is not the only body that can sabotage additional potential growth.
Dear Barry and Bennett,
I think we can wrap this up quickly. The Bureau of Labor Statistics builds its productivity numbers, as you know, by dividing output by an estimate of hours based mainly on establishment data. Had they used, instead, the household data that you recommend, the official numbers would show hours growing faster, but productivity (output per hour) growing even more slowly! Hence the acceleration in productivity needed to make your scenario come true would be even greater than we have spoken of in past letters.
Regarding that acceleration, I think we have agreed to disagree when it comes to the computer and productivity. You see the light; I still see the tunnel. The fundamental point, however, is that all of us are just guessing. No one can know what the future will bring.
Finally, I could not agree more with your sentiments on infrastructure, R&D, education, and training—and I have been saying so for years. Skimping on investment expenditures in order to reduce the budget deficit just doesn't make sense.
Alan S. Blinder