My friend Peter gave us Alan Greenspan’s memoir, The Age of Turbulence, as part of our wedding present, and I’ve finally managed to prise it from Raych’s grasp. I’m only a few chapters in, but in the first couple he outlines his development as an economist.
Strikingly, he is an empiricist through and through. He’s at best uninterested and often hostile to theory, especially macroeconomic theory. He developed an early scepticism even about the predictive power of econometrics beyond the short term and the local, though its statistical techniques help to clarify a complex picture. His approach appears to be a brute-force attention to detail. He started with industrial economics, the steel industry in particular (his early work was as a corporate consultant) and built up a picture from there.
In later years I developed some skill in building quite large econometric models, and came to a deeper appreciation of their uses and, especially, their limitations. Modern, dynamic economies do not stay still long enough to allow for an accurate reading of their underlying structures. Early portrait photographers required their subjects to freeze long enough to get a useful picture; if the subject moved, the photo would blur. So too with econometric models. Econometricians use ad hoc adjustments to the formal structure of their models to create reasonable forecasts. In the trade, it’s called add-factoring a model’s equations; the add-factors are often far more important to the forecast than the results of the equations themselves.
If models have so little predictive power, what use are they? The least-heralded advantage of formal models is simply that the exercise of using them ensures that basic rules of national accounting and economic consistency are being applied to a set of assumptions. And models certainly can help maximise the effectiveness of the few parcels of information that can be assumed with certainty. The more specific and data-rich the model, the more effective it will be. I have always argued that an up-to-date set of the most detailed estimates for the latest available quarter is far more useful for forecasting accuracy than a more sophisticated model structure.
At the same time, of course, the structure of a model is quite important to its success. You can’t (or at least I can’t) draw abstract models out of thin air. They have to be inferred from facts. Abstractions do not float around in my mind, untied to real-world observations. They need an anchor. This is why I strive to ferret out every conceivable observation or fact about a happening. The greater the detail, the more representative the abstract model is likely to be of the real world I seek to understand. [p. 36]
This doesn’t surprise me at all. I’ve thought for quite a while that mainstream economics is supported by an exoskeleton of empiricism rather than a backbone of general equilibrium theory. At least the parts that matter. Attacks on a monolithic ‘neoclassical economics’ – conceived as an abstract, ridiculously unrealistic construct – miss the mark, because outside the academy – and even in much of that – this is not how economics is done. Mostly modern economics is about extrapolating data series from past trends, and drawing maps rather than diagrams.