14 July 2010

The Efficiency Curse

One of the safest predictions that you can make in economics is that per capita GDP will grow in the medium and long term for the foreseeable future.

Caveats and Theoretical Basis

Sure, there will be brief stumbles that we call recessions. Sure, the precise rate of growth is hard to predict (although, on average, it is fair to say that the higher your per capita GDP is now, the less quickly it is likely to grow). And, sure, you can think of exceptions that prove the rule: domestic wars, nuclear winter, epoch defining natural disasters, a drug resistant disease epidemic as bad as the plague or small pox, and so on. But, even most of the exceptions that prove the rule are likely to be geographically specific. Global warming may lead to declining GDP in low lying coastal areas and ski resorts, but has a far more ambiguous effect in less sensitive areas.

A single global government might produce stagnation (similar to that of China when in turned inward due to domestic political considerations), but probably wouldn't produce actual declining per capita GDP for decades or even centuries.

The certainty of this prediction is not just empirical. It is a function of technology. Technology is an economic arrow of time. Some new technological advances increase efficiency and in turn increase productivity which in turn increases per capita GDP if some quite weak assumptions hold. New technology is continually created in modern society. Technology is not likely to be lost for any sustained period of time in a way that cannot be recreated quickly absent apocalyptic circumstances. A single second rate university library would be enough to reconstruct the vast majority of the scientific and technological developments made to date, and would dramatically speed up replication of those developments for which direct evidence was lost.

The only foreseeable development the could plausibly lead to sustained decline in per capita GDP over a large geographic area with any meaningful probability is peak oil. But, even then, peak oil would probably accelerate, rather than slow the development of technology in other areas. It would not halt the general advance of human society towards the development of technologies that make economic activity more efficient. But, even if oil production fell to 1918 levels by 2018, we would have a much more technologically advanced and economically prosperous society than we did in 1918.


The fundamental structure of our economy is such that not all industries are affected by increasing productivity that flow from technology that increases economic efficiency in the same way.

The Example of Agriculture

Civilization didn't emerge until agriculture created enough excess food production to allow people to think about something other than feeding themselves in the current season. A couple of hundred years ago, 95% of Americans were still subsistence farmers; excess agricultural production only freed up one person in twenty to be anything other than a subsistence farmer.

As agriculture became more efficient, this changed. By 1950, only about 15% of Americans were still farmers, so 85% of the population was available for other pursuits. Now, less than 2% of the work force is engaged in farming, and only about 0.6% of the work force is engaged in fishing, forestry, hunting, trapping, and mining. Thus, less than 2.6% of the work force in engaged in providing the raw materials needed for the rest of the economy. The other 97.4% of the work force is free to go do something else.

This number is likely to dramatically fall over the next few decades. More than half of farmers in today's economy provide only a tiny share of the nation's agricultural output, have very small farms with very low yields, and are run by self-employed farmers with very few employees (many temporary) who are themselves, on average, approaching or above what most people in the non-farming world consider to be retirement age. The number of farmers could drop in half in three decades without having any appreciable impact of agricultural production without any new technology.

But, there are also a whole host of technological developments that could make farming dramatically more efficient than it is today as measured by output per farmer. For example, farm machinery that drives itself using GPS guidance and locally placed guideposts that allow one farmer to work several fields at once after setting up the system initially exists with current technology.

As another example, biotechnology is very close to developing commercially viable perennial wheat. This would make it unnecessary to plant fields every single season, allowing a single farmer to have more fields in production than he or she could possibly plant in planting season, something that the farmer could then tend and harvest with GPS guided farm machinery.

Right now, those developments are protected by patents. But, in three decades, those technologies will be in the public domain, which will probably cause the price of those technologies to fall in the same way that the price of generic drugs do when their parents expire. Mass production, competing technologies that serve the same purpose, and other economic factors will probably drive down prices far earlier.

This mentioning of a couple of emerging agricultural technologies is, of course, not exhaustive. But, the amount of land available for cultivation has been only modestly changed over the last two hundred years.

So, the work being done now by fewer than 3 million people to produce raw materials for the U.S. economy may be done by perhaps as few as a million people in 2040. By then, that may mean that 99.75% of the population is engaged in something other than producing raw materials.

Beyond Agriculture

If it was only agriculture that we were concerned about, this development would hardly be consequential to anyone other than a rural development economist. But, the experience of agriculture, whose efficiency driven declining draw on our human resources freed up people to participate in the industrial revolution, can be generalized.

Off shoring and international wage competition has obscured the fact, but the part of our economy that turns raw materials into goods, the manufacturing industry, has also grown dramatically more efficient over the last century.

Historically, as increasingly efficient agriculture freed up labor, a lot of the labor was devoted to producing goods and building things. The rise of agriculture coincided closely in time with the appearance of ceramics and buildings and monuments, and not much later in time, metallurgy. When we learned to use coal to generate power to drive good production and make more sophisticated goods (that in turn made agriculture much more efficient and made its products much easier to distribute) many generations of people who grew up on farms migrated to cities where they were largely swallowed up by the manufacturing industry's seemingly insatiable demand for workers.

Then, around 1970, a funny thing started to happen. Technology was making manufacturing more and more efficient. Factories that once teamed with workers, started to become lonely places where a handful of people supervised key points of a gigantic system of machines. If your understanding of what a factory is like is derived from what you read in English classes and documentaries from the 1950s and earlier, you should schedule a visit to your local factory (surprisingly many have tours) today.

Pick any factory you choose, a car assembly plant, a tea processing factory, a brewery, a chemical plant, the U.S. Mint. They are all the same. They all produce herculean amounts of goods with astonishingly little labor.

In when Upton Sinclair was writing, a factory was a terribly dangerous place. Today, occupational injuries and deaths at factories are scarcely distinguishable statistically from those in office buildings. Farming, mining and construction are some of the very few careers where you have any significantly elevated chance of suffering a work related injury that doesn't involve a car accident or run of the mill criminal acts that just happen to take place at work.

Also, as the jobs have become safer, and the processes have become more automated, the demand for and pay afforded ordinary unskilled or semiskilled factory workers has fallen to just a little above minimum wage with benefits.

This low paid domestic factory workers, whose per person productivity is nonetheless immense because of the efficiency of the systems that they are a part of, are not paid much because there is an abundant supply of people who are qualified to fill these jobs and very little growth in the number of jobs like this because every few years new technologies make greater automation of relatively low skilled jobs possible.

Right now, about 15% of the work force is employed in manufacturing, construction and the utility industry. Even if you adjust this number for the percentage of workers abroad whose efforts give rise to the U.S. trade deficit by the amount of the trade deficit divided by the total manufacturing, mining, construction and utility industry employment in the United States today, this component of the United States labor force is only 20%-25% or so, and is stagnant at best. About half of the trade deficit comes from imported oil and gas, which have a very high productivity per worker, even abroad. We do import it a lot of goods, but we also export a lot of goods. Overall, the United States is closer to manufacturing self-sufficiency than it is to the mental model of a nation that makes nothing.

As technology advances in the decades to come, it is almost certain that the portion of the total work force employed in manufacturing, construction and the utility industry will shrink.

Capitalism has many flaws, but it remains one of the most powerful systems ever devised to improve economic efficiency on a localized basis.

Incidentally, while industry has not migrated away from oil dependence as completely as the electric utility industry (which uses almost no oil outside Alaska, Hawaii and some other very remote areas), it has dramatically reduced its reliance on oil in the last fifty years due to the rising price of oil, and due to environmental regulations, and it continues to do so. The predominant forms of energy consumed in industry today are electricity, coal, and natural gas, none of which are nearly so close peak production as oil resources. Also, a large share of all manufacturing plants were build in industrial neighborhoods established when freight rail was more important in distributing goods and receiving supplies relative to trucking than it is today, so reverting to freight rail in the event of rising oil prices would be less catastrophic than one might naively assume. It is the retail distribution chain, and not the manufacturing part of the good production and distribution process that is most intensely dependent upon oil prices. Thus, even peak oil is not likely to bring the demise of the manufacturing economy as we know it, and hence its impact on per capita GDP may not be as great as one might suspect.

This impact of technology on the share of manufacturing employment in the workforce has not gone unnoticed. It is customary to describe our current economic era as "post-industrial," and a tour through the American Rust Belt, which has earned that moniker over the last forty years or so should leave no doubt in anyone's mind that manufacturing is no longer the dominant and thriving engine behind the American economy.

Once upon a time, as recently as the second half of the 20th century, a majority of Americans either produced raw materials, or turned those raw materials into tangible goods and buildings. Now, the percentage of people engaged in these pursuits is conservatively (and accounting for off shoring) less than 25%-30%.

This percentage could easily be cut in half in another generation or so. While the manufacturing of tangible goods has already made tremendous progress towards very high levels of efficiency, we are just on the cusp of similar revolutions in the construction industry. Some of the developments we have already seen that have dramatically changed productivity, like the nail gun and the pre-assembled trusses, have gone almost unnoticed. New developments, like prefabrication of large components of a building that can be fitted together with each other on site, creating less waste, in a more controlled environment, with a more consistent and efficient process, are just beginning to work their way into the marketplace.

The Efficiency Curse

So, we now have a 70% and growing share of the workforce available to do all of the things that took a minority of the workforce in the 1950s, and was carried out, to the extent that their work was done at all, by a few percent of the people in workforce a century before then.

These other sectors aren't immune to combined forces of technology and capitalism either. Satellite television threatens to make the often highly skilled cable guy redundant. Satellite radio and the Internet threaten to put the DJ out of work. E-mail and other telecommunications have greatly thinned the ranks of postal workers and the "dead tree" publishing industry. Computers have greatly reduced the need for back office operations people in banks. E-filing has put tens of thousands of people (at least) who used to make copies in law offices all day, unnecessary.

Technology hasn't killed live opera or symphonies or live theater, but it has dramatically widened the economic gap between the people who are performing for millions via movies and television and radio and the recording industry, and those who are performing for hundreds or thousands of audience members in real life.

The bottom line is that the increasingly efficient post-industrial economy does not have a modern equivalent to the manufacturing boom of the industrial era to employ the tens of millions of people that are now longer needed to produce improved versions of all of the goods and buildings that we used to build and more. Capitalist creative destruction can and will create jobs for them, but the kinds of jobs that will be created are far from clear.

So far, three of the main kinds of jobs that have been created numerically, are sales jobs, personal service jobs (including food service, child care, lawn care and retail sales), and health care jobs.

The new health care jobs clearly are creating value, and are also a natural need of our economy as our population ages and technology has created more things that can be done to enhance people's health. This sector isn't immune to the efficiency curse, of course, and improved diagnostic technologies, better drugs, and less intrusive surgical techniques are increasingly making it possible to provide more care with fewer people. But, the other two kinds of jobs pose particular concerns.

Sales jobs are easy to create, because straight commission sales don't directly cost an employer anything. But, the nature of productivity in sales is quite different than in more "normal" industries. In most industries, there is a very linear relationship between work force size and productivity. If you double the number of employees you have, you can produce almost twice as many outputs.

In sales, however, this isn't true. Doubling the number of automobile salespeople in the United States probably will increase the number of automobiles sold in the United States, but only very modestly. Mostly, it will almost cut in half the income of the average automobile salesperson. In very immature industries, more sales people may increase sales proportionately to the number of salespeople, or even disproportionately as customers change their consumer spending portfolio to accommodate the new produce (as for example, was the case in emergence of the cell phone), but in relatively mature industries, the total amount of goods or services sold is not strongly related to the number of people who are out there trying to sell them.

Eventually, someone figures this out and somehow or other reduces the ranks of these salespeople industry wide. Two notable recent examples of this are Big Three automobile dealers, whose ranks have been slashed, and travel agents, whose pay for helping customers buy routine airline tickets has been slashed. At the retail level, the rise of Wal-Mart, at the expense of local retailers who employ more people, is an example of the same phenomena.

These examples probably won't be the last. As increasingly efficient industries elsewhere dump talented people into the high end sales force, one industry after another will purge the ranks.

This leaves a whole lot of people who might have found jobs on a farm, or in a mine, or in a factory, or selling cars, in an earlier day, left in an economy where increased efficiency has reduced the number of people who are needed to meet society's needs.

Some of that excess capacity has gone into management functions to make other industries even more efficient, but that is a rear guard action.

It isn't that the jobs needed by the new economy are uniquely highly skilled. Indeed, technology has deskilled a great many once difficult to do jobs making a far larger percentage of people capable of carrying out these jobs, and the average person's level of education has measurably risen from what it was a half a century ago and continues to do so. There are plenty of fairly low skilled people who have jobs in the new economy. But, the productivity of these fairly low skilled people has risen and therefore reduced the labor force necessary to do the work that they do, as a result of improved technology, much more quickly than the size of the pool of workers has fallen.

Indeed, while the effects of the efficiency curse have been less pronounced in more skilled jobs (because it is possible for more skilled workers to substitute down but not for for less skilled workers to substitute for skilled workers), it has hit there as well.

So what dos that mean? It means that until we come up with better things for them to do, that our highly efficient system of production doesn't need the entire available workforce to do the work that used to be done by it, so it is reverting to having more and more people serve as personal servants or in other very unproductive jobs that capitalism rewards poorly.

The great challenge for our age in the future will be to either break the tie between productivity and resource allocation, because to a significant extent productivity is driven by luck in being associated with productive enterprises that have nothing to do with one's own efforts, or we will develop much deeper class divisions than we previously did when everyone's work efforts were needed to produce what our society, collectively needed.

Of course, there is another way out. We can find worthwhile things for everyone displaced to do with the skills that they have or can acquire. This is what has happened in the past and has resulted in great strides for civilization. But, what that new value creating set of services looks like is very hard to determine looking through a glass darkly from the present to the future.

1 comment:

Kevin Dickson said...

The electronics superhero, Andy Grove, in today's Post, picks right up where your analysis leaves off.

Essentially, large bets are made on the next big things. To keep the American economy healthy, these bets must include American labor content.

My takeaway is, if your new startup doesn't provide jobs for your fellow citizens, they will eventually run out of money to buy your product. So even if you can get a better margin by manufacturing in China, think about the long term health of your company and invest in American jobs.