Did American inflation fall from its 1970s peak because rational expectations taught that lower inflation cost jobs only fleetingly, when the authorities were consistently committed to it, and everyone knew this? Or was it really that inflation expectations adapted to evidence of lower inflation, engineered by a Fed that in fact ignored the Lucas critique (do not base policy on how people behaved when policy was different)?
These are Thomas Sargent's first questions. The volume is not a chronicle of how inflation dropped. On why US unemployment also fell, little is said.Nothing, either, on Reagan's budget deficits, which, had they been expected to last, could - Sargent once showed - provide the real motor for rapid inflation. Instead, the reader gets a lucid, penetrating inquiry into the "ifs and buts" of rational expectations (RE) and their implications for policy. At their simplest, RE stipulate that how people think inflation is determined is precisely how it is determined, aside from blips that no one can foretell. Blips apart, inflation is set at the point of cut between a pair of lines with rock-hard, common-knowledge positions and slopes.
What if, Sargent asks, these positions or slopes can alter? You have to keep re-estimating them statistically, putting the latest observations into your set of data. If you fear a lasting change has happened, you weigh newer observations more than older ones. This makes your expectations adaptive. You learn. How quickly do your beliefs converge on the new reality? Might your beliefs actually stay wrong? Do public and policy-maker have to have exactly the same beliefs?
The clean division RE make between known relationships and unknowable blips looks odd. Even blips can often be explained with hindsight. What if expectations are "boundedly" rational? Sargent's immensely thoughtful book explores all these issues, and many more. Not all of his queries about RE are new, but the way he deals with them is packed with novel insights. Ironically, it was his writings over 25 years that helped to teach generations of economists the new classical macroeconomic ideas, now orthodox, of which RE form such a crucial part.
This book illustrates Sargent's great curiosity and honesty. Some economists have become worn-out records as they age, stuck in a groove. Not Sargent. If RE beg the question of what the "true" model is, they also typically assume (log)-linearity, becoming uncomfortably opaque or intractable when key relationships are curved. We need (the right sort of) curvature in people's economic environment and wants to pinpoint their best plans for action, and optimands need not be quadratic. Curvature is what many relationships (including the all-important short-run one linking inflation to unemployment) actually display. Another form of curvature is aversion to risk, which must bear on what people expect and do. This reviewer would have loved to see more on the problems curvature causes for RE.
If the questionable foundations of new classical macroeconomics are to be laid bare so probingly, readers might be reminded of the flaws of the old orthodoxy it replaced. They are not here. UK inflation was once fought by food subsidies and pay police. And from a Franco-German-Spanish-Italian perspective, we should admit that inflation reduction has been won, it seems, at the price of more unemployment, especially recently. So the US (or even UK) experience is not universal.
Further, has inflation really been conquered? It is way down, but not out.Even allowing for some quality mismeasurement, it remains strictly positive in the US and Britain (though not in depression-hit Japan). If inflation runs at a steady 1.6 per cent in the next 100 years, prices will rise fivefold, tying the 21st century with the 16th. (The 20th century is the exception: prices rose 14-fold in the US, 48-fold in the UK, 2,000-fold in France and 1,046-fold in Hungary.) A far cry from the 300 years after 1615, when the English price index was truly trendless, and even further from Milton Friedman's "optimum quantity of money", another plank in new classical macro-theory, which prescribes prices falling at the real rate of interest.
Sargent's book does not so much provide a thoroughgoing single alternative to consensus RE-based thinking as a panoply of reasons and suggestions for refining and improving it. This extraordinary volume should find a very broad readership among professional economists, policy-makers, model-builders, forecasters and students. It is a beautifully crafted, deep and (mostly) very accessible work of art. It is without doubt the most important book on macroeconomics since Recursive Methods by Nancy L. Stokey and Robert E. Lucas appeared more than a decade ago; it deserves to be one of the century's most influential books on macroeconomics.
Peter Sinclair is professor of economics, University of Birmingham.
The Conquest of American Inflation
Author - Thomas J. Sargent
ISBN - 0 691 00414 5
Publisher - Princeton University Press
Price - £21.95
Pages - 148