Warts, test rejections and all

At first sight these two books make strange bedfellows for a reviewer. The first is a collection of fairly technical surveys by some dozen authors of issues in international finance; the second is one economist's essays on the methodology of "macroeconomics" illustrated with accounts of the thought of several practising economists, past and present. But what they have in common is that they both reflect cautiously on what is truly offered by this subject, the study of whole economies' behaviour - which embraces forecasting and policy analysis of money, the exchange rate, budgets and general regulation of the economy: areas that economists, like it or not, get most public exposure on.

Is macroeconomics a "progressive research paradigm" or merely an example of a (relativistic) "community interchange" bet-ween thinkers at one time in one particular culture?

Roger Backhouse struggles to sort out this methodological debate; he notes that leading relativists are willing to exempt science from their charge and considers whether macroeconomics perhaps is enough of a science to be also exempted, and he identifies in macroeconomics since Keynes a "neo-Walrasian research programme" which qualifies in Kuhn's and Lakatos' sense as scientific and, he argues, progressive. It has had its periods of revolution ("rational expectations" - that is, the assumption that people use available information optimally in forming expectations - being the main one) and of normal science.

He explains the programme in the terms of the pioneering Chicago economist, Robert Lucas Jr, as the shedding of "free parameters" in macroeconomic modelling. Early attempts at modelling economies, such as Keynes's own, used parameters that were not derived from optimising behaviour by individuals or firms; they were ad hoc rules of thumb either suggested by the data itself or by rudimentary approximations of what the aggregate of optimal individual behaviour might resemble. Yet the hallmark of economics is the application of the paradigm of homo economicus; these parameters, not being derived from that, are "free".

The application of this paradigm came much earlier in microeconomics, the study of individual and market behaviour, than in macroeconomics where the main problem to be modelled was expectations. They are the Hamlet of Keynes' General Theory but like Hamlet they could not be modelled. Backhouse sympathetically reviews the fumbling efforts of pre-Keynesian economists such as Hobson, Clark and Mitchell to formulate the sort of model Keynes later set out: Clark writes admiringly to Keynes about his mathematical modelling flair. The irony is that Keynes disapproved of anything but the most rudimentary formulae and later ticked off Tinbergen for his efforts at a full model of the Dutch economy.

He may have been right. Lucas argues that it is (computer and mathematics) technology that has liberated economists to model properly. As Backhouse notes, the first article on rational expectations, by John Muth in 1961, did not catch on until a decade later, through the efforts of Lucas and his friends. That was because it was horrendously difficult to apply with the techniques available in the 1960s. Now not only have we worked out how to apply it, we have the computer power to apply it to highly nonlinear models with no free parameters at all.

I therefore think that Backhouse's thesis is right. Further evidence is in the huge industry that this research programme has spawned - surely a clear sign of normal science at work. This brings me to the second book which well illustrates that process but also provides a huge dose of caution to economists making much of a claim to scientific effectiveness.

The problem is summed up by MIT's Paul Krugman in a "what do we know?" piece at the end of the book, carefully edited by Peter Kenen of Princeton University from the proceedings of a conference sponsored by its International Finance Centre. The preceding surveys show "dispiritingly", Krugman points out, that while economists have strenuously applied the rational expectations no-free-parameters methods to both national and international economic behaviour, it has not worked well, at least to date; indeed formal tests reject it almost universally (as documented for example in a survey of the asset market evidence by Liverpool's Mark Taylor) for reasons that are hard to identify because only joint hypotheses are testable.

On the other hand, he observes, the old rule-of-thumb models updated to use a few of our latest techniques are widely used in policy analysis and work quite passably. Krugman has a point. But I do not think it is any more than a well-known phenomenon in technological change: that yesterday's technology, which one has thoroughly learned how to use. works better than today's as-yet-unfamiliar (let alone tomorrow's as-yet-fledgling) methods. The model that Krugman has in mind is basically the sort that underlies our Liverpool model, and for a few years now most other major United Kingdom models.

Aggregate demand relationships are complemented by supply relationships, including a "Phillips curve" relating inflation to the cycle and inflation expectations; different countries' interest rates are set equal once adjusted for expected exchange rate changes and rational expectations are assumed throughout. But these relationships are still a huge advance on practice ten years ago and they incorporate a big chunk of the "neo-Walrasian programme". It is simply that we are not yet very good at applying the newer chunks.

When one goes through the concrete issues tackled in this book - such as international cooperation, monetary union, inflation stabilisation - one can with Krugman proclaim how much we do not yet know. But I found equally an astonishing amount that we do - perhaps surprisingly. Is the glass half empty or half full?

The chapter by Ralph Bryant goes through the large model-based efforts to calculate potential strategies for international cooperation and resulting gains: he concludes that because an effective strategy has to take account of model uncertainty simple exercises may show small gains or even be counterproductive. Fair enough, but at least we know enough not to make the egregious mistake of trying to reinvent Bretton Woods in today's "global economy".

A similar scepticism surrounds the discussion of fixed exchange rates and monetary union that recurs throughout the book (including a chapter by Charles Goodhart of the LSE instructively emphasising the overriding political aspects of money, and an extended panel discussion).

At least flexible exchange rates act as shock absorbers in a world of many national "fiat" moneys. We do not know how large the undoubted microeconomic gains of integrating moneys are but we do have a lot of evidence (largely from those macromodels in use today) about the risks to the economy's stability from artificial fixing such as in the ill-fated Exchange Rate Mechanism.

The book contains a variety of other excellent material, too long to list, on our experience with integration of capital markets, the debt crisis, the opening-up programmes in emerging and transition economies, and the control of inflation by monetary and budget reform. Though there is plenty that has not gone as well as one would have hoped, the general picture is certainly not discouraging to the free market and sound money policy programme that emerges from the neo-Walrasian programme, even if much of the detailed analysis still has huge scope for improvement.

So in conclusion I would argue with Backhouse that macroeconomics is not a cosy relativistic chat, or witchcraft carried out by the community gurus, but - warts, test rejections and all - a programme of research with quite a few policy successes to its credit and an exciting menu of research still to work through.

Patrick Minford is professor of applied economics, University of Liverpool.