Presentation by Diane Coyle at IPEG, University of Manchester, 30 May 2003 PART 1
Let me motivate this talk about how to think like an economist by describing some eminent figures who’d make ideal economists. What kind of people are they? My first candidate isn’t, strictly speaking, a person. Mr Spock, first officer of the starship Enterprise, is ultra-logical, in contrast to the emotional and impulsive Captain Kirk. Spock is the ideal rational man, or rational Vulcan rather, of economic theory. What’s more, he sums up the essence of utility theory as he sacrifices himself to save The Enterprise in the movie The Wrath of Khan. Referring to the self-sacrifice of Sydney Carton from A Tale of Two Cities, he famously explains to Kirk that the needs of the many outweigh the needs of the few the greatest happiness of the greatest number.
My second ideal economist is Charles Babbage. Not only did he open the way to the development of programmable computers, which are such a vital tool in modern economics, he also upheld scientific standards of rationality and accuracy against poetic vagueness. In his poem “The Vision of Sin” Alfred, Lord Tennyson wrote:
Every minute dies a man, Every minute one is born
Babbage wrote to congratulate him on the poem but added: “I need hardly point out to you that this calculation would tend to keep the sum total of the world’s population in a state of perpetual equipoise, whereas it is a well-known fact that the said sum total is constantly on the increase. I would therefore take the liberty of suggesting that in the next edition of your excellent poem the erroneous calculation to which I refer should be corrected as follows:
Every minute dies a man, And one and a sixteenth is born
The actual number is much longer but I believe 1-1/16 will be sufficiently accurate for poetry.”
My third example is Hercule Poirot, Agatha Christie’s precise Belgian detective who solves the most puzzling crimes with a passion for order and a disdain for tangible evidence. He’s a theorist. The purest example of his method is in The Mystery of Hunter’s Lodge, in which he solves a murder in Derbyshire without leaving his London flat, where he’s laid up with flu. As Poirot says: “The true work, it is always done from within. The little grey cells–remember always the little grey cells, mon ami. ”
By contrast, we have some evidence that Keynes would have made a lousy economist by modern academic standards. He said: “When statistics don’t make sense, I find it generally wiser to prefer sense to statistics.” Ambitious young lecturers wouldn’t get many journal articles published if they took that attitude today. In his Essays in Persuasion Keynes in fact blows the gaff on one of the difficulties with economics: “Economists, like other scientists, have chosen the hypothesis from which they set out, and which they offer to beginners, because it is the simplest and not because it is the nearest to the facts.” I think you’ll be getting the idea. Economics is based on the assumption of rational behaviour by standardised individuals. One of the great merits of this assumption is that it makes conventional economic models very tractable. But we know that few human beings live up to the economist’s standard of rationality, as it has been refined and narrowed in practice. Hence the standard criticism of economics: it’s unreal, divorced from real life.
This isn’t only an external critique. Plenty of economists think it’s valid. The University of Notre Dame is splitting its economics department into two, putting the Spocks and Poirots, members of the academic mainstream, into an orthodox department, and all its Marxists, post-Keynesians, historians of economic thought and feminists into another. There’s a growing “Post-Autistic Economics Network” which is encouraging alternatives to the mainstream. There’s nothing new about this such criticisms have been voiced since at least the late 1970s, a significant date I’ll return to later. I’d be sorry to see a schism in my subject because I believe it’s a uniquely important one for public policy and civic life. I want to put forward two arguments about it today. One, that there’s much merit in criticisms of the kind of orthodox economics that dominates the profession, especially in Anglo-Saxon universities. But, two, none of the criticisms is fatal, and in fact we should take care not to devalue the real insights of economics. We need more economic theorising, not less, in public policy.
On the contrary, I believe economics is still a fantastically fruitful intellectual discipline. The version of economics that’s taught to most students is hugely flawed. But the version done by economists who make it through the purgatory of their undergraduate and graduate courses is much richer, and there’s an exciting renaissance in several areas of the subject.
To complicate matters, there are criticisms of economics that are valid, and criticisms that are not. I want to start by dismissing the unimportant ones. A lot of the unpopularity of economics outside the citadel is the price of intellectual integrity. Unpopularity is almost the raison d’etre of economists, who are the dirty realists of the social sciences. Economists are the only people who warn, constantly, about difficult choices and trade-off – summed up in the catch-phrase ‘there’s no such thing as a free lunch’. Choosing one course of action means closing off another.
Indeed, economics is essentially intelligent skepticism, applied to human society and politics. It’s a subject born out of the Enlightenment’s elevation of the power of reason. David Hume described the approach as “an attempt to introduce the experimental method of reasoning into moral subjects”, to quote the subtitle of the ‘Treatise of Human Nature’. Good economics is inspired by what Hume described as “the spirit of accuracy”, which takes all the sciences “nearer their perfection, and renders them more subservient to the interests of society.”
I think this gets to the heart of the suspicion with which people from other disciplines regard economics. It contradicts cherished beliefs, and does so in an aggressively rigorous way that is often difficult to challenge – difficult because the logic and evidence go the economists’ way. For example, great insights in economics can amount to saying that accounting identities must add up (as Thomas Schelling argued). The current and capital accounts of the balance of payments must sum to zero. If you’re a net importer of foreign capital, you’ve got a trade deficit. That makes it impossible for a country to be simultaneously flooded with cheap goods and yet exporting jobs to low wage countries – but protectionists can never get the point.
Some critics of economics simply reject the idea that human society is amenable to the scientific method at all, rather than cultural analysis, (and this kind of chasm isn’t confined to economics – hence the nature versus nurture wars). I don’t have much to say about this rather sterile dualism. Some are just innumerate, suspicious of anything at all involving addition, subtraction, multiplication or division (or as the Mock Turtle in Alice in Wonderland renamed them, ambition, distraction, uglification and derision). It’s easier for the critics to attack the entire subject and its intellectual approach. These kinds of critics aren’t worth taking seriously.
Unfortunately, other criticisms of economics have more substance. The idea we’re taught as students, that the positive and the normative can be kept strictly separate, is nonsense, as many of the people teaching it will recognise. But to our cost as a profession, most practitioners of economics – especially in universities have taken an extreme and ultimately unsustainable view about what it means to be scientific.
The pretence by economists, that it was possible to isolate purely positive questions, has rebounded in the form of a reaction inside the fortress of economics, as well as outside, against the assumption of maximising behaviour by rational individuals. The assumption of rational behaviour gets taken to extremes.
Needless to say, it is a powerful working assumption. Vito Volterra compared it to simplifying assumptions in physics, such as frictionless surfaces and non-extendable lines. People are certainly not totally rational in real life, but if you are going to argue that they consistently behave in ways that are not in their best interests, you¹d better have a convincing explanation for it.
So, for example, whenever the stockmarket crashes economists are widely mocked – often by people who should know better – for assuming investors behave rationally and the financial markets are ‘efficient’. In that case, why was a dot.com company worth $10 billion one day worth only a few hundred million the next? There is no question that psychology and sociology have a crucial role to play in explaining what goes on in the stockmarket .
However, it is always worth pointing out to the generally (but not always) innumerate critics, that share prices are supposed to value the entire future profitability of a company at today’s values. Although the observed volatility is indeed “excessive”, a small change in expectations of profits growth in 10 or 20 years’ time can, as a matter of arithmetic, have a big impact on today’s valuation. What’s more, there is also a deep truth in the economists’ argument that investors are rational and will therefore compete away any lasting profit opportunities. For very few investors manage to beat the market for any length of time.
Similarly, people do not often get married for purely economic reasons. Yet economists’ models in which people choose, or ditch, their partners in order to maximize their financial gains can illuminate trends like single motherhood. Welfare payments and the low potential earning power of inner city fathers go a long way to explain how it became a financially sensible choice as well as consistent with changing social norms. The economic insight illuminates the sociological, and vice versa. The economic explanation is never the only one, but it puts the backbone into the political and sociological explanations.
It’s often argued that the emphasis on formal models of rational behaviour means that the use of complicated mathematical techniques has gone too far. And that the formalisation now required to get anywhere in academic economics is not only spurious when it comes to understanding the world, but also simply puts off many potential students. The figures suggest that students increasingly prefer the realism of business schools, the intellectual interest of cross-disciplinary courses or the genuine technicalities of one of the natural sciences.
For the professional economist, working through the mathematics of a simplified model that isolates a particular issue is still the best way to gain new insights. Edgeworth of the Edgeworth box diagram fame compared the use of mathematical formalism to scaffolding. You need the scaffolding to build the house, but then you need to take it down when the structure’s complete.
That’s a big ‘but’ – the economist has to be able to work out what the solution to the equations means. Paul Krugman has argued that many of the critics of mathematical formalism in the subject are criticising bad economics, and that it’s unfair to criticise any subject because of what its bad practitioners, rather than its good ones, do. That may be true up to a point.
But as somebody who has spent many years trying to communicate economics to a wider audience, I think that the scale of meaningless formalism, bad writing and lack of real thought about the nature of the world in academic economics, does constitute a deep institutional problem in our subject (as indeed in other academic disciplines, especially the social sciences). Economists must be able to explain their results to a wider audience, or there has to be a suspicion they don’t actually understand it themselves. It is, after all, a social science whose findings have a social meaning.
This is why Deirdre McCloskey is right to highlight the “rhetoric” of economics. Every subject offers a narrative framework within which the world is to be interpreted. Our profession has developed an extremely unappealing framework or set of metaphors. It is therefore losing the competitive struggle with other intellectual disciplines as a result. This is why some of the most exciting economics being practiced in universities now is at a slight angle to the mainstream, in economic history or geography, urban studies, or in behavioural finance or experimental economics, for example (although this is also just the way the tides of fashion have flowed in the subject).
There are still far too many pointless papers, which fail the ‘so what?’ test that ought to inform any discipline and inflict atrocious econometrics on their readers, by authors encouraged by the career structures and perverse incentives of a creaking university system. The widely understood fact amongst economists, that probably the majority of the empirical work published in professional journals is flawed to the point of uselessness, is our dirty little secret, not often acknowledged to outsiders. A 1996 study by McCloskey and Ziliak of 10 years’ worth of articles in the American Economic Review, perhaps the premier journal, found that a large majority had used statistical inference incorrectly. Even if you think their standards are too demanding, it’s still true that computers have made it too easy to access data the researcher never bothers to understand, or plot, or even check for errors, and too easy to run too many regressions. Many researchers seem to regard their computers as a substitute for thinking, not as a complement. Not surprisingly, they publish results that are silly, or even damaging.
So I do believe that too much actually-existing economics is unecessarily formalistic, over-reliant on a narrowly positivistic philosophy, and stuffed with really bad econometrics.
How did we get into this mess? I’d pinpoint post-war macroeconomics as being where it all started to go wrong and particularly the ideological clashes that marked the 1970s and early 1980s.
Macro is certainly what gives economists such a bad general press – it’s about Gordon Brown’s forecasts going wrong or Denis Healey turning back from the airport because the IMF had to step in in 1979. The humiliation of macroeconomists by events has cast a long shadow. As a result many of today’s professional economists are very uneasy about making sweeping pronouncements on the future of capitalism or the nature of class struggle in modern societies. They know there is no decisive empirical evidence, no refutable facts, to back up many pronouncements on the macro scale.
The lack of confidence is justified. For all of my professional lifetime there have been competing schools of thought about how the macroeconomy works, a sure sign that nobody actually knows. Thus the clash of Keynesians versus Monetarists, which the Keynesians ultimately lost as it became blatantly obvious that they could not fix growth and inflation as they had been confidently proclaiming up to the late 1970s. But the battle was bitter because it mapped into the protagonists’ political views.
Next came the schism between ‘real business cycle’ or ‘new classical’ or ‘rational expectations’ theorists and the former Keynesians. The first group argued that as people are rational, fluctuations in the level of demand in the economy must reflect supply side changes such as a sudden improvement in technology to which everybody was reacting rationally. The presumption at the heart of the rational expectations idea, that people will not be so foolish as to consistently ignore opportunities to profit or increase their incomes, has something going for it. The marriage of the tools and insight of rational expectations (why would people consistently believe something that’s false?) to a study of the imperfections and market failures of the real world has produced some very fruitful research. Macroeconomics is now as close to consensus as I can remember it being, and real policy lessons have emerged witness the greater stability since around 1995 of growth and inflation.
Whatever the merits of one school or another, though, the point is that if there is scope for distinct schools of thought at all, this is not hard science.
That’s not surprising when you’re grappling to keep on top of whatever happens to be going on right now in the world, whether that’s recession, a technology-driven boom, high inflation, or deflation, or globalisation. Sorting out cause and effect is intrinsically challenging in a complex and changing world, in which events are often over-determined. The data with which we are working are unfolding in real time. The analysis of what is in the public interest often involves trade-offs between specific groups of people and is therefore politically fraught.
What’s more, the empirical evidence can hardly ever be decisive. There isn’t all that much data available to test competing theories: the economy changes so much it doesn’t make much sense to take what happened before 1980, say, as good evidence for what might happen in 2005. At best an economist has perhaps 20 years’ worth of statistics, some available monthly, or almost continuously in the financial markets, but some only quarterly or annual. Given that one month’s price level is really very similar to the next month’s, or one year’s GDP much like the next’s, there isn’t even all that much information in the separate pieces of data.
Macroeconomic forecasting in particular – that is, what the general public thinks economists do – is very tricky, essentially because where the economy gets to in future depends on what millions of us do between now and then. In general, aggregation introduces all sorts of self-fulfilling phenomena. This isn’t a new idea. In a very famous passage in The General Theory Keynes, who was himself a successful investor, said investing in shares was like picking the winner in a beauty contest. You wanted to choose the contestant most likely to appeal to most judges.
This applies not only to stockmarket bubbles but also to booms and recessions. A recession is a collective outcome that can feed on itself in a vicious spiral. It emerges in the way a storm does, if you’re think about the weather as an analogy. Macroeconomic forecasting is really quite like weather forecasting. You can spot short-term tendencies and possibilities, might even be able to predict rain ahead with great confidence, but any greater precision in the forecast will be spurious.
Many forecasters are entirely comfortable with this conclusion, and indeed would quite like to educate the general public that sensible forecasts would be something like: “There’s a 75% chance inflation will be above 2% by the end of next year.” But we seem to prefer the appearance of confidence: “Consumer prices will be rising by 2.4% in 18 months’ time.”
Unfortunately, as models capture a 20-year average experience, and therefore inevitably predict the future will be a lot like the average of the past 20 years, unless the forecaster deliberately overrides the predictions using his or her skill and judgement. So models produce bland pictures of the future. They are very bad at forecasting extreme events like recessions. They are pretty bad anyway whenever the economy changes because what’s new is not incorporated into the estimated equations and just compare the economy of 2000 with 1980 or 1970.
And so over the past two centuries, GDP per capita in the leading economies has grown by about 2% a year – but the standard errors in forecasting this growth are about 2.5%, or in other words in excess of the thing being forecast. The trouble is that events, ‘structural breaks’, keep butting in. Events like the Industrial Revolution, fascism, feminism, and world wars, or innovations like electricity, the internal combustion engine, and the computer.
This all leaves macroeconomic theory in an uncomfortable position. Any textbook will reveal it still rests on formal models of fixed equations – some of which capture genuine insights, but which don’t add up to a tool for usefully analysing or forecasting the economy as a whole because they’re metaphors. Unfortunately, there is nothing to put in its place which would aggregate the behavior of millions of people who are not fully rational, who don’t have complete information about everything, who make short-sighted decisions, who are all different, who sway each other’s behavior, and who live in economies where there are all sorts of obstacles to free and competitive markets.
Since the ideological clashes of the 1970s and 80s, there has been a welcome appreciation amongst economists that it might be a good idea to make far more modest claims to be able to understand and predict on the macro front. There is a greater degree of consensus now about what constitutes bad macro-economic policy, thanks in large part to the disastrous results of putting once-fashionable economic theories into practice.
The professional consensus now can be summed up as: don’t make dumb mistakes. Keep inflation low , because it’s economically damaging, unfair, and voters hate it. Get central bankers, who hate inflation too, to keep it low. A high rate of growth is good, but so is a steady rate of growth. The voters hate boom and bust, or at any rate the bust part of it. That means making sure government borrowing isn’t too high or the government surplus isn’t too big, because the government’s not the point of the economy, businesses and consumers are. There are still heated arguments, of course, about whether interest rates or tax and spending have been set at the right levels. But the argument covers a much narrower range of options than before.
Indeed, expectations of economic forecasts are simultaneously extraordinarily high and very low, for economists are expected to foretell the future with a degree of authority we would never demand of a meteorologist or biologist, and not surprisingly are held in low esteem for often getting it badly wrong. Many forecasts are extremely bad – perhaps inevitably so. But as David Hendry at Nuffield has complained: “When weather forecasts go awry, meteorologists get a new supercomputer; when economists mis-forecast, we get our budgets cut.”
But things are getting better. Modern macroeconometrics does offer techniques for coping, for narrowing the inevitable range of uncertainty attached to any forecast. Some of it is quite simple. First, forecast growth rates, not levels (first difference the data). That makes it less important to get the absolute level right, so any mistake you make here will show up as a one-off ‘blip’ error in the forecast. Secondly, if go one step further (second difference the data), so you are forecasting rate of acceleration, you can similarly neutralise any linear time trend in the series.
Two simple steps can therefore tackle the most basic misspecifications of an econometric equation, getting the wrong intercept and the wrong trend. Another tip is to update estimated equations pretty often, using the most recent data. Many forecasts are in practice based on very old computer models, and the economists driving them deal with forecasts that are obviously going adrift by over-riding the predictions with their own judgement. In the trade, it’s known as adjusting the ‘residuals’ or ‘add-factors’. The variable being forecast can be sorted into two parts, the bit that is actually produced by the equation for that variable, and the residual. In creaky old models the residual can easily be the most important component of a forecast value. Forecasters can also learn from their errors. If an equation turns out to have produced a big error last quarter, the error can be added to the intercept term in the equation for forecasting the next quarter.
In short, economic forecasters have to stop believing or allowing the rest of the world to believe that what they do is set down a simplified but essentially true version of the structure of the economy. Most of them don’t do this yet but best practice will improve over time.
At the theoretical rather than practical level there are also some interesting applications of complexity theory. It just says people (like or ants or molecules) influence each other in the many choices they make.
Fans of complexity theory overdo their claims about undermining all of past economics. Actually, economic thinking can easily take account of many of the phenomena beloved of complexity theorists. For example, increasing returns to scale in certain industries that arise from ‘network effects’ (the price I pay depends on how many of you have already purchased the item), certainly don¹t undermine economics. Respected, and conventional, academic economists are doing a lot of research along these lines, especially in finance, trade theory, economic geography, and industrial organization. CONTINUED IN PART 2