It is a great pleasure to meet with a group such as this, devoted to-or at least interested in-the study of what is actually going on in the real world, with the object of drawing useful conclusions from the facts. In my opinion, as a useful way of spending one's time this compares favorably with the popular pastime of arm-chair pontification with total reliance on a sort of mystic inner guidance developed during a lifetime of defending the indefensible from assaults of change. Simulated cases studied in the classroom and business games on the computer are certainly good fun and, up to a point, are perhaps instructive. But I welcome the evidence of a growing concern with studying the actual facts, rather than even the most ingenious simulations thereof. In such areas as financial communication, we cannot (and I am sure will never be able to) match the controlled conditions which the chemist or physicist can often create in his lab for experimentation. This may be illustrated, in a simple-minded sort of way, by contrasting the control of heat in a lab experiment with the control of inflation in the economy. Presumably by controlling the flow of fuel, temperatures can be raised in a lab experiment absolutely predictably. However, while inflation is certainly closely related to money supply, the effect on prices of a given increase in money supply is nothing like so predictable. The price rise depends on many noneconomic factors-for in dealing with people, all sorts of political considerations inevitably enter the picture-and I sometimes wonder whether our difficulties in controlling inflation do not stem from the fact that we tend to think of inflation as an economic rather than as a political problem. Certainly in Canada, during World War II, we virtually halted inflation by appointing a Prices Czar who announced it was not to take place. A special case of course, but one which dramatically illustrates the principle. But if we cannot hope to match the controlled observations of the

A bstract-A nonparametric time series test is applied to the buyers' and sellers' (BLS) prices collected by George Stigler and James Kindahl. Although BLS pnrces are more ngid, the two senes of pnrces are sufficiently similar in behavior dunrng two business cycles so that one can reject the null that each is generated by different stochastic processes. Moreover, each series reacts similarly in favor of the administered price hypothesis: in recessions and in the initial year of recovery, concentrated pnrces are more rigid and perversely flexible than unconcentrated pnrces, which are characterized by random movements and changes in the direction of real demand.

Abstract Reviews the book "Readings in the Concept and Measurement of Income," edited by R.H. Parker and G.C. Harcourt.