Ambitious Economic Models

This week, The Economist published a very interesting special report entitled 'Economic models: Big questions and big numbers'. Here are some excerpts:

We cannot live without big and ambitious economic models. But neither can we entirely trust them.

Among the many gadgets, instruments and artefacts in its care, London's Science Museum holds a peculiar contraption that most resembles the work of a deranged plumber. Yellow tubes connect together a number of tanks and cisterns, around which coloured water can be pumped. Sluices and valves govern the flow of liquid and makeshift meters record the water-levels.

The “plumber” responsible for this device was William Phillips. Educated as an engineer, he later converted to economics. His machine, first built in 1949, is meant to demonstrate the circular flow of income in an economy. It shows how income is siphoned off by taxes, savings and imports, and how demand is re-injected via exports, public spending and investment. At seven feet (2.1 metres) high, it is perhaps the most ingenious and best-loved of economists' big models.

Economists today use computers and software not perspex and piping, but they share Phillips's itch to build models that faithfully mirror the real economy. For each of the big economic questions facing the world (What do we stand to gain from a global trade deal? By how much has expensive oil retarded growth? What might be the economic costs of an avian flu pandemic?) there is a model that will provide a big numerical answer ($520 billion, 1.5% of world GDP, and $4.4 trillion, respectively). Such figures are trotted out far and wide. But can we entirely trust them?

Economic models fall into two broad genres. Macroeconomic models, the distant descendants of Phillips's machine, belong mostly in central banks. They capture the economy's ups and downs, providing a compass for the folks with their hands on the monetary tiller. The second species, known as computable general equilibrium (CGE) models, largely ignore the vagaries of the business cycle. They concentrate instead on the underlying structure of production, shedding light on the long-term repercussions of such things as the Doha trade round, a big tax reform or climate change.
...

Short of good data, and stretched to their computational limits, the early modellers nonetheless had high ambitions. They aimed not merely to understand the economy, but to run it.
...

Such ambitions now seem quaint. In countries not cursed by socialism or war, the market is left to decide what to produce and in what proportions. But the state remains responsible for keeping the overall macroeconomy ticking over. Policymakers are largely indifferent to what is in demand, so long as the tank of demand remains full.
...

These measurements were fed into their models, which in turn guided their policy advice.

In 1958, for example, Phillips showed that for long stretches of British history, high unemployment coincided with low wage inflation, and vice versa. Many macroeconomic models therefore featured a trade-off between the two: doves could choose low unemployment at the expense of high inflation; hawks the opposite.

But in the 1970s these trusted relationships broke down. And in 1976 Robert Lucas, of the University of Chicago, explained why. Such trade-offs, he argued, existed only if no one expected policymakers to exploit them. Unanticipated inflation would erode the real value of wages, making workers cheaper to hire. But if central bankers tried to engineer such a result, by systematically loosening monetary policy, then forward-looking workers would pre-empt them, raising their wage claims in anticipation of higher inflation to come. Cheap money would result in higher prices, leaving unemployment unchanged.

In short, one could not judge how the macroeconomy would respond to a new policy based on its behaviour under the old regime. The “Lucas critique”, as it was called, brought its author fame and a Nobel prize. But it dealt a big blow to the confidence of model-makers. As Christopher Sims of Princeton University has put it, “Use of quantitative models as a guide to real-time policy advice was cast into such deep disrepute that academic research on the topic nearly completely ceased.”
...

In the past decade, a number of central banks—and even the International Monetary Fund (IMF)—have reared a new generation of practical macroeconomic models, all of them sporting microfoundations. First-born was Canada's Quarterly Projection Model in the mid-1990s; its close siblings include the Bank of England Quarterly Model (BEQM) introduced in 2004; the SIGMA model groomed by the Federal Reserve's International Finance Department; and the IMF's new Global Economic Model (GEM). Old hands doubt whether the new microfoundations are quite as secure as they seem—the macroeconomy is surely rather more than the sum of its parts—but no self-respecting theorist can now be seen in public without them.

Stabilising the macroeconomy is only one of the responsibilities of governments in a market economy. They must also raise taxes and most feel the need to impose tariffs, both of which put rocks in the stream of economic life. When they contemplate big changes to these policies, most governments cannot resist turning to CGE models to forewarn them of the consequences.
...

Trade's virtuous effects are of two distinct kinds. First, trade helps countries make the most of what they already have. It frees countries to allocate their resources—whether they be cheap labour, fertile land or educated minds—as efficiently as possible. But, secondly, trade can also allow countries to accumulate resources more quickly. Indeed, the biggest prizes lie in faster growth, not heightened efficiency; in accumulation and innovation, not allocation.

By their nature, CGE models are better suited to capturing the first effect than the second. They provide “before and after” snapshots of the economy at two points in time. They are therefore good at capturing the one-off gains that might arrive from a redeployment of the economy's resources. They are much less good at capturing the continuing gains that result from a faster accumulation of capital, or a quickened pace of productivity growth. Most trade models, indeed, hold productivity fixed.
...

Most empirical exercises confront theory with numbers—they test theories against the data; sometimes they even reject them. CGE models, by contrast, put numbers to theory. If the modeller believes that trade raises productivity and growth, for example, then the model's results will mechanically confirm this. They cannot do otherwise. In another context, Robert Solow, a Nobel prize-winner, has noted the tendency of economists to congratulate themselves for retrieving juicy plums that they themselves planted in the pudding.
...

To be fair, most modellers are quite open about the theoretical principles that underlie their simulations. But to compute an economic model, this theory has to be given concrete form, spelt out in definite algebraic terms. Alfred Marshall, one of the fathers of neo-classical economics, distrusted mathematics for this very reason. To be expressed in mathematical form, he complained, many important economic considerations had to be “clipped and pruned till they resembled the conventional birds and animals of decorative art.” Economic theory gives only the roughest guide to this pruning. It points out, for example, that supply rises when prices increase. But does it rise in a straight line or curve upwards? Perhaps, as prices rise, supply traces out an inverted U-shape or an S-shape?
...

Phillips's pump-action model was, he wrote, meant for “exposition rather than accurate calculation.” But all models should ultimately be seen as pedagogical devices, their calculations a means to the end of helping policymakers think through their decisions. Unfortunately, Phillips's model was rather better at this than many of its more sophisticated successors. It was transparent: you could see through its casing, trace the flow of expenditures through its pipes and watch wealth accumulating in its tanks. Get things wrong and prosperity drained away in front of your eyes. The model was also easy to tinker with: valves could be loosened, sluices opened and taps tightened. It was clear what was governing its results.