While trawling through old issues of the MIT alumni magazine, Technology Review, I came across this vintage 1961 piece by Robert Solow that was distilled from a presentation he made to “principal officers of companies participating in M.I.T.’s Industrial Liaison Program.” It is worth rescuing from the obscurity of an alumni magazine to share with the students of today. Artificial intelligence is nice, but Solow’s natural intelligence is a sight to behold. So, behold!
___________________________
Education, Technology & the GNP
Two-fifths of the growth in our national product came mainly from what goes on in men’s heads
By Robert M. Solow
ROBERT M. SOLOW, received his doctorate from Harvard and has taught at M.I.T. since 1949. This article was part of a recent talk on “Productivity and National Performance” during a meeting of principal officers of companies participating in M.I.T.’s Industrial Liaison Program. Professor Solow’s wit, as well as his scholarship, has won him an enviable reputation at the Institute. He is now serving on the President’s Council of Economic Advisers.
Businessmen think of productivity as increasing when they find a way of reducing the number of men it takes to operate a particular machine without any bad effects on its output, or a way of increasing the output of a machine without additional labor. In addition to such developments, we must include a lot of other things—some of them strange but all of them inevitable—to measure the productivity of the whole national economy.
Imagine the American economic system as having a blank wall around it, so you can’t see what really happens inside. All you can observe is that each year certain primary inputs are fed into this economy. These are goods and services which are not part of the economy’s current output; the main ones are labor services of different kinds and levels of skill, land and natural resources of various qualities, and pre-existing capital equipment—all kinds of new and old buildings, machines, and inventories which represent the cumulated past savings of the population. These things are fed into the economy each year and it grinds away and produces some final output.
The conventional classification of final output is into personal consumption (food, clothing, automobiles, haircuts, toys), private investment (the additions to capital equipment and inventories which will become the primary inputs of next year), and government expenditures on goods and services (items we choose to consume as a community rather than individually, such as Atlas rockets and the services of school teachers). The productivity of the national economy increases whenever this flow of final output grows larger without any corresponding increase in its cost in terms of primary inputs, or whenever final output grows faster than primary inputs.
This is not a simple notion. The final output is an assortment of widely different goods and services, ranging from jet aircraft to left-hand-hitting first base-men. To reduce such heterogeneous items to comparability, the economist has to attach values to them and the values he uses are market prices. A tangible commodity and an intangible service are supposed to be worth what they bring in the open market. The market value of the whole flow of final output is called the Gross National Product (GNP); and when the effects of pure inflation or deflation of the price level have been eliminated, the resulting measure is called the Real Gross National Product.
Millions of Decisions
Valuing things at market prices means that some items which you or I might consider worthless or worse are counted as contributions to GNP. I happen to believe, for instance, that the automatic choke was an invention of the devil. But if an automatic choke on a car enables it to be sold at a higher price, the GNP increases when cars are equipped with them.
Other items such as legislation and justice are not bought and sold on the open market (though there have been some notable exceptions to this), so no current price is quoted. The economist has to value the services of such governmental agencies as courts and legislatures at the wage and salary costs incurred in operating them: The contribution to GNP of the nine Justices of the Supreme Court of the United States is measured by their combined salaries; and the contribution to GNP of the Louisiana State Legislature, engaged largely in frustrating the decisions of the Supreme Court, is valued at its salary bill. Such perversities, fortunately, are the exception rather than the rule, and when the real GNP increases faster than primary input the chances are that the economic system has become more productive.
What we do with the annual flow of GNP depends on millions of private and public decisions. How much of it we consume sensibly or frivolously within families, how much we save and invest to create a higher GNP for our children, how much we spend collectively to provide schools, fire protection, national defense, foreign aid and snow removal—all this is determined by decisions made within families, by the boards of directors of corporations, and by voters and their elected representatives. It is no easy matter to understand why things have happened in the past the way they have, and even harder to predict how these things will turn out in the future. But whatever we want to do with goods and services, they can come only from our National Product. And our private and collective desires to do more in the future can be satisfied only out of growth of the National Product.
In principle, a rational and self-disciplined people could decide to satisfy its desire for more national defense and more private investment and more urban renewal by cutting back its own consumption. But it hardly even qualifies as cynical to suspect that the chances of getting more of those things are considerably improved if we can get them out of increases in National Product without putting pressure on consumption.
In any case it would seem important to try to find out how rapidly America’s real GNP has increased in the past and what the sources of that growth have been. We can hope in this way to make a guess about how much economic growth we can expect in the future even though we do nothing special about it, and some hints about how we might accelerate that growth.
The Sources of Growth
Since 1929, the real GNP has grown on the average at a rate just under 3 per cent per year. That growth was far from smooth—in 1939 our economy had only regained the absolute level of output of 1929 in its long climb out of the depression. But a quantity growing at an average rate of 3 per cent a year will double in size every 24 years, and our present real GNP is about two and a third times what it was in 1929. This generation has about twice as much as the generation before.
In 1929, however, between 70 and 75 per cent of the GNP was used for personal consumption, and that percentage has fallen now to 65 per cent. Private investment has about held its own in percentage terms, and there has been a shift in the use of GNP from personal consumption to collective consumption or governmental expenditures, primarily for defense.
If we are interested in the improvement of our potential standard of living, rather than in military potential or national grandeur or similar things, we should look at GNP per head of population. Since 1929 the population of the United States has grown about one and a fourth per cent per year. It follows that real GNP per head of population has increased by about 1.7 per cent per year, and is now about 60 per cent higher than it was in 1929. A quantity growing at 1.7 per cent per year takes about 42 years to double; by the year 2000, if present trends continue, real output per person may be about twice what it is now.
Even now the annual flow of goods and services available to serve our personal and collective needs could be substantially larger than it is without any economic growth taking place at all: Part of the civilian labor force is unemployed, more is on short time. The steel industry is using only a fraction of its capacity. We are, in short, in a recession and a lot of extra output could be had by improving the short-run state of affairs.
One of the best medicines for long-run growth is short-run prosperity. But leaving aside short-run slack, where can increased real output come from? Since national productivity as we have defined it is a ratio of final output to primary inputs, extra output can come either from increased primary inputs or from higher productivity. The line separating these two sources of growth is not a sharp one.
The Primary Inputs
Let’s consider the primary inputs first. In the absence of territorial changes, the supply of land and natural resources doesn’t change. (Although resources can be discovered and developed, this requires labor and capital.) But the input of labor can increase—and it has. The observed increase in the country’s input of labor has been the resultant of several tendencies, some demographic and some more directly social.
There is an increase in employment when the population grows and also (as has happened since 1929) when the fraction of the population engaged in production increases. But this increase in the labor input has been partially offset by a reduction in hours worked per year. Changes in the demographic composition of the population make a difference, too: The labor supply changes in quality and quantity as the age distribution of the population shifts and as the level of health and strength improves. Social institutions matter, too: The participation of women in the labor force affects its quality, and so does the efficiency with which the skills of women are utilized. And another very important contribution to economic growth lies squarely on the borderline between increases in input and in productivity: This is changes in the education of the labor force.
Shall we think of this as equivalent to an increase in the labor input, a high school graduate counting as contributing “more” labor than a worker who left school in the sixth grade? Or shall we think of a better-educated, better-trained labor force as one of the sources of higher productivity? It does not matter, as long as we keep it straight and recognize how important it is.
The other great primary input is reproducible capital equipment: buildings, machines, inventories, vehicles, and the rest. One important way of getting increased output is by equipping the labor force with more horsepower, more capital, more of the fruits of saving. But here, too, we must draw a hypothetical line between changes in the amount of capital used in industry and improvement in the quality of capital goods. The latter ought properly to be counted as a source of increased productivity; most new processes need to be embodied in a new type of capital equipment, and a high rate of investment increases the speed with which technological inventiveness gets built into production.
Factors in Productivity
Coming now to improved productivity as a source of economic growth, there are at least three general categories: One has already been mentioned—ordinary technological progress, new products, new methods, new ideas, the results of research and inspiration. Secondly, a gain or loss in productivity may stem from improved or worsened economic organization. I have in mind such things as monopolistic restrictions, certain kinds of union rules, barriers to the flow of goods, immobility of labor and resources, and practices imposed either privately or by government. A third possibility, known to Adam Smith, is that the sheer increase in the size of the economic system may yield economies of scale by permitting greater specialization and division of labor and allowing goods to be produced in plants of optimal size. If you believe that the “ultimate” source of productivity is the American Family, or the Democratic Way of Life, or the Republican Party, you may be right. But the ultimate causes must work through one of these proximate sources.
In what proportions has economic growth since 1929 come from each of these sources? This is no simple matter and a lot of guesswork and assumptions have gone into answering this question. I’ve tried my hand at it and so have many others. The figures | am going to give you come from a recent study by Herbert Stein and Edward Denison (both of the research staff, Committee for Economic Development), which is more complete than anything else done along these lines.
Remember that the real GNP has grown by about 3 per cent per year since 1929. About one-third of that growth, or one per cent per year, is attributable to increased employment. (Employment rose about 1.3 per cent per year but without accompanying investment, etc., could only increase output by one per cent per year.) Shorter hours canceled about one-fifth of this growth. So altogether, the increase in labor input was responsible for about eight-tenths of one per cent of the total growth rate of 3 per cent per year. Stein and Denison add to that a contribution of a tenth of one per cent from changes in age and sex composition and better utilization of women workers. Thus we account for growth of nine-tenths of one per cent per year, or 30 per cent of the observed growth rate of 3 per cent.
Investment was the source of growth at the rate of a bit over one-half of one per cent a year, accounting for about 19 per cent of the recorded growth. Hence, about half of the growth rate of real GNP in the last 30 years can be ascribed to increased primary inputs of labor and capital.
Had this been all, our national output would have grown at a rate closer to one and a half per cent per year, and we would be much poorer than we are. The other half of the actual growth rate is a consequence of improved national productivity.
When we look at this remainder closely, we see that about half of it is attributed to the historical increase in the education of the labor force. Real GNP, in more explicit terms, has increased about seven-tenths of one per cent a year since 1929—which is about a quarter of the total rate of growth—in consequence of the prosaic fact that each year’s labor force has been on the average better educated than the previous year’s.
This leaves about 27 per cent of the full growth rate, about eight-tenths of one per cent a year, to be accounted for. Stein and Denison concluded that there was a slight loss of output because of greater restrictions against optimal use of resources, but this didn’t amount to much. They attributed about one-third of one per cent a year to economies of scale. And they credited the rest, amounting to one-half of one per cent a year—a sixth of all the growth we’ve had—to advances in technical knowledge and improvements in actual production practices.
Together, education and improved technology accounted for about 40 per cent of the recorded growth of the last 30 years. Of the 3 per cent per year increase in real GNP, about 1.2 per cent has come mainly from what goes on in people’s heads. The national product per head has grown 1.7 per cent per year and it has grown 1.6 per cent per year per person employed. Of this latter rate of growth, no less than three-quarters is the result of increased education and increased knowledge. That’s something to think about.
On the Other Hand…
Such figures tempt one to leap ahead with predictions and policy prescriptions. But no one can guarantee that the next 30 years will reproduce the pattern. Our real product may not continue to grow at 3 per cent per year, and it may well be that whatever growth we get in the next quarter of a century will come from different wellsprings. In some ways, in fact, this seems likely.
According to Stein and Denison, we have earned a high yield from improving the educational level of the labor force. This process, however, may be subject to diminishing returns; giving everyone a high school education may have added more to the nation’s productivity than sending everybody to college would—hard as this may be for a college professor to swallow. There is, moreover, some limit to the possibility of lengthening the school year and adding to the number of years of schooling.
Eventually, the only way left to increase educational input will be to improve the quality of education. This is desirable in any case, but it will also be costly, and the net economic return to society may be less. If you believe, as I do, that more and better education would be a good thing even on wholly noneconomic grounds, and worth what it will cost in diversion of resources, then the economic benefits are just a bonus.
When we turn to policy-making, it is tempting to accentuate the positive and look only for places where conscious action promises a reward. But policy measures have costs. We must be sure we are prepared to pay the bill.
We do not in our society decide casually that people are spending too little on shoes or too much on hair tonic and commence legislation to set them right. Is there something special about education and research?
I think there is. Knowledge is one of the few things that tends to escape the market. Partly this is because knowledge is usually embedded in people and we do not permit property rights in people.
The public nature of knowledge is even clearer in the case of research. The more “basic” research is, the more the private yield is likely to fall short of the social yield. Under these conditions, even the most smoothly working market economy may tend to under-invest in education and research.
One hears often of technical progress being held up or dissipated now by featherbedding and other restrictive practices by labor organizations. There is little doubt that such practices impose a real burden on society by holding back potential output. But they stem largely from fear of technological unemployment, and this is a real thing, especially in the short run. A democratic society which wishes to eliminate featherbedding has, I think, the responsibility to do what is necessary to ease the transitional burden on displaced workers—by retraining programs, assistance in mobility, and (perhaps even more importantly) maintaining a high level of employment and a brisk labor market.
One of the side effects of the kind of emphasis we have been placing on the productivity side of the ledger is a tendency to underestimate the importance of capital formation in increasing output per head. Most methods for separating out the effects of technical progress and investment fail to allow for the strong interdependence between the two. Sheer replication of capital goods of the same type may indeed have only a weak effect on aggregate output, but increased knowledge in blueprints into hardware may be equally unproductive. If this is so, then for any given rate of acquisition of knowledge, society can within limits speed up its rate of growth by stimulating investment, shifting the age distribution of its stock of capital goods toward the young side, and converting obsolete capacity into the latest models.
One source of extra output available to us right now is to get back to full employment. A high level of economic activity can have long-run effects on productivity—because pressure on capacity stimulates investment, and because lively markets for labor and goods help to break down restrictive practices and resistance to change. It could hardly hurt!
Source: Technology Review, April 1961, pp. 17-19, 34, 36.
Image Source: Portrait of Robert M. Solow used in the Technology Review, April 1961 can be seen in the mitmuseum.mit.edu collection of faculty photographs.