Lies and Statistics

Plenty of difficulties stand in the way of making sense of the economic realities we face at the end of the age of cheap abundant energy. Some of those difficulties are inevitable, to be sure. Our methods of producing goods and services are orders of magnitude more complex than those of previous civilizations, for example, and our economy relies on treating borrowing as wealth to an extent no other society has been harebrained enough to try before; these and other differences make the task of tracing the economic dimensions of the long road of decline and fall ahead of us unavoidably more difficult than they otherwise would be.

Still, there are other sources of difficulty that are entirely voluntary, and I want to talk about some of those self-inflicted blind spots just now. An economy is a system for exchanging goods and services, with all the irreducible variability that this involves. How many potatoes are equal in value to one haircut, for example, depends a good deal on the fact that no two potatoes and no two haircuts are exactly the same, and no two people can be counted on to place quite the same value on either one. Economics, however, is mostly about numbers that measure, in abstract terms, the exchange of potatoes and haircuts (and everything else, of course).

Economists rely implicitly on the claim that those numbers have some meaningful relationship with what’s actually going on when potato farmers get their hair cut and hairdressers order potato salad for lunch. As with any abstraction, a lot gets lost in the process, and sometimes what gets left out proves to be important enough to render the abstraction hopelessly misleading. That risk is hardwired into any process of mathematical modeling, of course, but there are at least two factors that can make it much worse.

The first, of course, is that the numbers can be deliberately juggled to support some agenda that has nothing to do with accurate portrayal of the underlying reality. The second, subtler and even more misleading, is that the presuppositions underlying the model can shape the choice of what’s measured in ways that suppress what’s actually going on in the underlying reality. Combine these two and what you get might best be described as speculative fiction mislabeled as useful data – and the combination of these two is exactly what has happened to the statistics on which too many contemporary economic and political decisions are based.

I suspect most people are aware by now that there’s something seriously askew with the economic statistics cited by government officials and media pundits. Recent rhetoric about “green shoots of recovery” is a case in point. In recent months, I’ve checked in with friends across the US, and nobody seems to be seeing even the slightest suggestion of an upturn in their own businesses or regions. Quite the contrary; all the anecdotal evidence suggests that the Great Recession is tightening its grip on the country as autumn closes in.

There’s a reason for the gap between these reports and the statistics. For decades now, the US government has systematically tinkered with economic figures to make unemployment look lower, inflation milder, and the country more prosperous. The tinkerings in question are perhaps the most enthusiastically bipartisan program in recent memory, encouraged by administrations and congresspeople from both sides of the aisle, and for good reason; life is easier for politicians of every stripe if they can claim to have made the economy work better. As Bernard Gross predicted back in the 1970s, economic indicators have been turned into “economic vindicators” that subordinate information content to public relations gimmickry. These manipulations haven’t been particularly secret, either;visit and you can get the details, along with a nice set of statistics calculated the way the same numbers were done before massaging the figures turned into cosmetic surgery on a scale that would have made the late Michael Jackson gulp in disbelief.

These dubious habits have been duly pilloried in the blogosphere. Still, I'm not at all sure they are as misleading as the second set of distortions I want to discuss. When unemployment figures hold steady or sink modestly, but you and everyone you know are out of a job, it's at least obvious that something has gone haywire. Far more subtle, because less noticeable, are the biases that creep in because people are watching the wrong set of numbers entirely.

Consider the fuss made in economic circles about productivity. When productivity goes up, politicians and executives preen themselves; when it goes down, or even when it doesn't increase as fast as current theory says it ought, the cry goes up for more government largesse to get it rising again. Everyone wants the economy to be more productive, right? The devil, though, has his usual residence among the details, because the statistic used to measure productivity doesn't actually measure how productive the economy is.

Check out A Concise Guide to Macroeconomics by Harvard Business School professor David A. Moss: "The word [productivity] is commonly used as a shorthand for labor productivity, defined as output per worker hour (or, in some cases, as output per worker)." Output, here as always, is measured in dollars – usually, though not always, corrected for inflation – so what "productivity" means in practice is dollars of income per worker hour. Are there ways for a business to cut down on the employee hours per dollar of income without actually becoming more productive in any more meaningful sense? Of course, and most of them have been aggressively pursued in the hope of parading the magic number of a productivity increase before stockholders and the public.

Perhaps the simplest way to increase productivity along these lines is to change over from products that require high inputs of labor per dollar of value to those that require less. As a very rough generalization, manufacturing goods requires more labor input overall than providing services, and the biggest payoff per worker hour of all is in financial services – how much labor does it take, for example, to produce a credit swap with a theoretical value of ten million dollars? An economy that produces more credit swaps and fewer potatoes is in almost any real sense less productive, since the only value credit swaps have is that they can, under certain arbitrary conditions, be converted into funds that can buy concrete goods and services, such as potatoes; by the standards of productivity universal in the industrial world these days, however, replacing potato farmers with whatever you call the people who manufacture credit swaps (other than "bunco artists," that is) counts as an increase in productivity. I suspect this is one reason why the US auto industry got so heavily into finance in the run-up to the recent crash; GMAC's soaring productivity, measured in terms of criminally negligent loans per broker hour, probably did a lot to mask the anemic productivity gains available from the old-fashioned business of making cars.

As important as the misinformation generated by such arbitrary statistical constructs is the void that results because other, arguably more important figures are not being collected at all. In an age that will increasingly be constrained by energy limits, for example, a more useful measure of productivity might be energy productivity – that is, output per barrel of oil equivalent (BOE) of energy consumed. An economy that produces more value with less energy input is arguably an economy better suited to the downslope of Hubbert's peak, and the relative position of different nations, to say nothing of the trendline of their energy productivity over time, would provide useful information to governments, investors, and the general public alike. For all I know, somebody already calculates this figure, but I'm still waiting to see a politician or an executive crowing over the fact that the country now produces 2% more output per unit of energy.

Now it's true that a simplistic measurement of energy productivity would still make the production of credit swaps look like a better deal. This is one of the many places where the distinction already made in these essays between primary, secondary, and tertiary economies becomes crucial. To recap, the primary economy is nature itself, or specifically the natural processes that provide the human economy with about 3/4 of its total value; the secondary economy is the application of human labor to the production of goods and services; and the tertiary economy is the exchange of abstract units of value, such as money and credit, which serve to regulate the distribution of the goods and services produced by the secondary economy.

The economic statistics used today ignore the primary economy completely, measure the secondary economy purely in terms of the tertiary – calculating production in dollars, say, rather than potatoes and haircuts – and focus obsessively on the tertiary. This fixation means that if an economic policy boosts the tertiary economy, it looks like a good thing, even if that policy does actual harm to the secondary or the primary economies, as it very often does these days. Thus the choice of statistics to track isn't a neutral factor, or a simple one; if it echoes inaccurate presuppositions – for example, the fantasy that the human economy is independent of nature – it can feed those presuppositions right back in as a distorting factor into every economic decision we make.

How might this be corrected? One useful option, it seems to me, is to divide up several of the most important economic statistics into primary, secondary, and tertiary factors. (Of course the first step is to get honest numbers in the first place; governments aren't going to do this any time soon, for obvious reasons, but there's no reason why people and organizations outside of government can't make a start.) Consider, as a good example, what might be done with the gross domestic product.

To start with, it's probably a good idea to consider going back to the gross national product; this was quietly dropped in favor of the current measure some years back, because it puts a politically uncomfortable spotlight on America's economic dependence on the rest of the world. Whichever way that decision goes, the statisticians of some imaginary Bureau of Honest Figures might sort things out something like this:

The gross primary product or GPP might be the value of all unprocessed natural products at the moment they enter the economy – oil as it reaches the wellhead, coal as it leaves the mine, grain as it tumbles into the silo, and so on – minus all the costs incurred in drilling, mining, growing, and so on. (Those belong to the secondary economy.)

The gross secondary product or GSP might be the value of all goods and services in the economy, except for raw materials from nature and financial goods and services.

The gross tertiary product or GTP might be the value of all financial goods and services, and all money or money equivalents, produced by the economy.

The value of having these three separate numbers, instead of one gross domestic (or national) product, is that they can be compared to one another, and their shifts relative to one another can be tracked. If the GTP balloons and the other two products stay flat or decline, for example, that doesn't mean the country is getting wealthier; it means that the tertiary economy is inflating, and needs to have some air let out of it before it pops. If the GSP increases while the GPP stays flat, and the cost of extracting natural resources isn't soaring out of sight, then the economy is becoming more efficient at using natural resources, in which case the politicians and executives have good reason to preen themselves in public. Other relative shifts have other messages.

The point that has to be grasped, in this as in so many other contexts, is that the three economies, and the three kinds of wealth they produce, are not interchangeable. Trillions of dollars in credit swaps and derivatives will not keep people from starving in the streets if there's no food being grown and no housing being built, or maintained, or offered for sale or rent. The primary economy is fundamental to survival; the secondary economy is the source of real wealth; the tertiary economy is simply a way of measuring wealth and managing its distribution; and treating these three very different things as though they are one and the same makes rank economic folly almost impossible to avoid.

Now it deserves to be said that the chance that any such statistical scheme will be adopted in the United States under current political and social arrangements is effectively nonexistent. Far too many sacred cows would have to be put out to pasture or rounded up for slaughter first. Still, current political and social arrangements may turn out to be a good deal less permanent than they sometimes seem. What might replace them, here and elsewhere, is a topic I plan on exploring in a future essay here.