THE DIFFICULTIES OF MEASURING CHANGES IN THE LEVEL OF PRICES It is a curious fact that men did not attempt to measure changes in the level of prices until after they had learned to measure such subtle things as the weight of the atmosphere, the velocity of sound, fluctuations of temperature, and the precession of the equinoxes. Their tardiness in attacking that problem is the more strange because price changes had frequently been a subject of acrimonious debate among publicists and a cause of popular agitation. Long before the high development of the credit system and a class of permanent wage earners practical issues of grave importance were raised by the instability of prices, as the disturbances created in sixteenth-century Europe by the inflow of American silver and gold abundantly show. Perhaps disinclination on the part of "natural philosophers" to soil their hands with such vulgar subjects as the price of provisions was partly responsible for the delay; but after all a number of eminently "respectable" men wrote upon economic topics in every generation after the days of Columbus—to go no further back. Nor can the technical difficulties of the problem explain this tardiness; for the mathematical intricacy of index numbers, and even the necessity of allowing for changes in the pure silver content of coins, are obstacles far less formidable than those surmounted long before in other fields of research.
Probably the chief cause of delay was that averages of price fluctuations did not promise to command much confidence after they had been made. The quotations available for use by the early investigators were few in number and often of doubtful accuracy. Carli, for example, dealt with only three commodities; Shuckburg-Evelyn with 12. About the vastly greater number of unrecorded price fluctuations the one firmly established fact was that they exhibited bewildering diversity. Under these circumstances, could an average made from a few samples be accepted as a reliable measure of changes in the general level of prices? And if averages could not be trusted, why trouble to devise a plan of making them? So writers upon prices long contented themselves with statements about the fluctuations of particular commodities, and with indefinite assertions that the purchasing power of money had changed little or changed much.
So, also, when certain bold investigators did finally venture to make index numbers, no one was particularly impressed by the significance of their achievement.
This lack of faith in the validity of averages of price variations was overcome rather slowly, partly in consequence of improvements in business organization. The multiplication of commercial newspapers and the more systematic keeping of private and public records provided a larger and more accurate body of quotations. Improved means of transportation made wholesale prices in the larger cities basic for many local markets.
The grading and standardizing of commodities increased the number of articles which could be accepted as substantially uniform in quality from one year to the next. More important still was the discovery by statisticians that social phenomena of most kinds, though seeming to result from the uncontrolled choice of individuals, yet reveal a striking regularity when studied in large numbers. The demonstration that a formerly unsuspected regularity lay hidden in one set of numerical data after another encouraged economists to believe that the known price variations might after all he fair samples of the more numerous unknown variations. The general similarity of the results reached by different investigators using dissimilar data confirmed this faith. Thus emboldened, economic statisticians devoted much time to extending the scope, and improving the technique of index numbers. And their growing confidence in the trustworthiness of their series was gradually imparted to the public.
Today few, if any, competent judges doubt the validity of index numbers or the substantial accuracy of the results they show when properly from carefully collected data. Indeed the danger at present is rather that the figures published will be taken too absolutely as a complete representation of the facts about price fluctuations. It is therefore well to begin a study of index numbers, not by analyzing the finished series, but by inspecting the actual changes in prices from which they are made, and which they purport to summarize. In no other way, indeed, can the value and the limitations of index numbers be learned.