Small farmers in Massachusetts did most of their business without money of exchange. They maintained a dense web of mutual charge accounts among themselves in a system that has been called bookkeeping barter. Changing monetary values in their account books closely matched the movement of money of exchange throughout the Atlantic world. During the price revolution of the eighteenth century, similar rates of inflation appeared in both “imaginary money” and “real money.” See Winifred Rothenberg, From Market Places to a Market Economy; The Transformation of Rural Massachusetts, 1750–1850 (Chicago, 1992); idem, “The Market and Massachusetts Farmers, 1750–1855,” Journal of Economic History 41 (1981) 283–314; idem, “A Price Index for Rural Massachusetts, 1750–1855,” ibid. 39 (1979) 975–1001; and for bookkeeping barter see Baxter, House of Hancock, 17–21.
Excellent works of reference on monetary systems include Peter Spufford (with the assistance of Wendy Wilkinson and Sarah Tolley), Handbook of Medieval Exchange (London, 1986); and John J. McCusker, Money and Exchange in Europe and America, 1600–1775: A Handbook (Chapel Hill, 1978). Another work on exchange in Europe from the late fifteenth century is in progress by Frank C. Spooner.
APPENDIX H
Nominal Prices and Silver Equivalents
How should prices be represented? What units should be used? Most scholars measure prices in standard monetary equivalents. That conventional practice has been followed in this work, with a few exceptions noted in this appendix. But other social and agricultural historians have sometimes reckoned prices differently in an effort to remove the effect of monetary fluctuations, and in particular to control for the effect of monetary debasement.
The silver-content in money of exchange was frequently altered by public authorities and private individuals. Monarchs and mint–masters changed the amount of precious metal in their coins: sometimes by debasements which reduced the content of precious metal; other times by recoinages which increased it. England’s Edward III, for example, repeatedly shrank the silver content of an English penny: twenty-two grains in 1334, twenty in 1344, eighteen in 1351. Henry IV reduced it further to fifteen in 1411, and Edward IV took it down to twelve grains in 1464. Other kings went the other way. Henry VII, founder of a new Tudor dynasty, wished to establish the legitimacy of his reign by improving its coinage in silver content, technical excellence, and artistic merit. His son Henry VIII reversed that policy. In the words of historian Charles Oman, he converted “the finest, best executed and most handsome coinage in Europe” into “the most disreputable money that had been seen since the days of Stephen—the gold heavily alloyed, the so-called silver ill-struck and turning black and brown as the base metal came to the surface.” (Charles Oman, Coinage of England [Oxford, 1931, 244]; Glyn Davies, A History of Money [Cardiff, 1994], 192–93).
Private individuals also debased gold and silver coins that passed through their hands. The crudest and most common method was to clip, shave, or file away part of the metal and pass what remained as if it were the intact coin. This ancient practice is the reason why modern coins are still minted with a distinctive pattern around their edges. A more subtle method of debasement was to wash or “sweat” a coin, so as to remove some of its gold or silver by chemical means. The most laborious technique was to cut the coin through its edge into two narrow discs, remove the center, and rejoin them. A merchant, money-changer or even small storekeeper in the early modern era had to keep his own scales and use them with great care.
In early projects of price history, some scholars tried to correct for monetary instability of this sort by reckoning prices not in monetary units but in grams or grains of pure silver. The pathbreaking British price historian Thorold Rogers did this. His example was followed by German agrarian historian Wilhelm Abel, who computed his grain prices in kilograms of pure silver. Abel was mainly interested in harvest conditions, which he wished to study by a method that would remove the effect of currency debasements and recoinages.
Other price historians have followed Rogers and Abel, notably Fernand Braudel and Frank Spooner. But most have not done so. Increasingly, price historians work with nominal monetary units. One of the most meticulous of medieval price scholars, David L. Farmer, explains the reason why. “I have not followed J. E. T. Rogers and others in attempting to express medieval prices in terms of constant weights of silver,” he wrote. “Such exercises ignore the value of silver relative to the stock in the economy in which it circulates” (“Prices and Wages, 1350–1500,” in Joan Thirsk, ed., Agrarian History of England and Wales, III, 441).
Scholars will continue to disagree on this problem. This book supplies estimates by both methods for the price-revolutions of the Middle Ages, as well as for the sixteenth century and the eighteenth century, so that readers may judge the result. They will find that the two methods of representing prices make little difference for an understanding of long-term secular change.
Farmer himself attempted to measure the effect of debasements and recoinages more directly, and found that the many English debasements of silver pennies between 1334 and 1464 had little impact on long secular trends in price levels. He concluded that changes in 1344 and 1351 “were followed by livestock prices slightly higher than usual for a year or two after each devaluation… . But later changes in the weight of silver in the penny seem to have had little effect on prices” (ibid., 440–41).
Altogether, indicators of the timing, direction, and spatial diffusion of major price-movements yield broadly similar results, no matter whether one uses the prevailing gold and silver currencies of the time or their equivalents in pure silver. Patterns of short-term fluctuation in commodity prices around the secular trend were more apt to show the effect of debasements; but the trend itself, as well as price relatives, wage-price movements, and the movement of rent and interest are much the same by the two methods.
APPENDIX I
Returns to Capital: Interest Rates as Historical Indicators
As a measure of changing returns to capital, the empirical indicator used throughout this inquiry is the annual rate of interest as it has changed through time. Here I have followed the work of Sidney Homer, an American lawyer and investment counselor who worked in the securities market for many years, and made it his hobby to study the history of interest rates throughout the world, which he did with great care. Many scholars and leaders in the American securities industry lent their expertise to his project. Among them were Henry Kaufman, Arthur Burns, and Marshall Dunn (Sidney Homer, A History of Interest Rates (1963, 2d. ed., New Brunswick, N.J., 1977).
From the broad range of materials that Homer collected, I have tried to assemble a set of indicators that have six qualities in common. First, they are specific to a time and place. Second, they are high-grade securities, issued either by leading governments, or by established private institutions. Third, they are securities that are actively bought and sold in financial markets. Fourth, market yields are preferred to nominal yields. Fifth, a range of securities is used wherever possible: long and short, public and private. Sixth, where possible they have been drawn from multiple national economies.
With a few exceptions, data that meet these criteria may be found from the fifteenth century to the present, but not earlier. I have not been able to make much headway on the movement of interest in the medieval price revolution. Scattered scraps of evidence suggest that the patterns were similar to subsequent great waves, but more work needs to be done on this question.
Other questions of high complexity will come quickly to mind. It would be good to know more about the relation between price revolutions and capital-formation, capital-accumulation, and patterns of change in the structure and function of capital markets. All this must be left to later inquiries and larger books.
APPENDIX J
Returns to Labor: Real Wages and Living Standards
A difficult problem in this inquiry is to find a method of measuring returns to labor through four price revolutions. The most simple and straightforward
way is to compute real wages: that is, money wages adjusted by an index of consumer prices. The result of this computation is yet another index, commonly expressed as a ratio of the purchasing power of wages in any given year to their purchasing power in a single benchmark year. This solution has been standard for many generations and is employed throughout this work.
Many scholars have criticized the use of real wages for this purpose. They have done so with good reason. Economists and historians agree that even the most refined indices of real wages are not in themselves an accurate measure of returns to labor. They are even less satisfactory as an indicator of changing standards of living. Here are a few of many problems.
First, standard series of money wages tend to have structural distortions in wage coverage itself. Long-running wage series tend to bias the inquiry toward workers whose employment is more stable than that of the labor force as a whole. This distortion was specially strong in the late medieval and early modern historiography. In twentieth century statistics, the same bias is still present, but not so strong. Its net effect in a study of secular change is to understate long-term improvement of wages before the twentieth century.
Second, wage series tend to omit unreported earnings in the “gray economy.” As wages are increasingly taxed in many nations, and employment is subject to regulations of growing complexity, the gray labor market has grown larger during the twentieth century. Many of these unreported jobs tend to be more poorly paid than those that are reported. In studies of secular change, this problem causes an underestimate of growth in aggregate returns to labor, but an overestimate of average hourly money wages and real wages in the twentieth century.
Third, real wages are commonly computed only from money wages, and take no account of income in kind. A large part of returns to labor in the medieval and early modern periods consisted of income in kind. This assumption, to my knowledge, has never been tested empirically for long periods. The problem can only be solved by the use of personal documents (diaries, private accounts, etc.), which are limited to literate populations. In any case, a bias toward money wages understates returns to labor in every period. The magnitude of this distortion is greatest in earlier periods; the effect is to overstate long-term improvement in returns to labor by excluding a form of income that was relatively larger in the past.
Fourth, wage series in themselves tell us nothing about the extent of unemployment or underemployment. Returns to labor should properly include not only hourly or daily earnings but also the changing proportion of hours and days actually worked. Some twentieth-century studies have introduced corrections for this problem. Stanley Lebergott compiled a series of average annual returns of employees in the United States. He adjusted money earnings for unemployment, then deflated both series by consumer prices. The result was two series: real wages of workers when employed, and real wages of workers “after deduction for unemployment.” But his correction did not fully account for underemployment, as distinct from unemployment. See Stanley Lebergott, Manpower in Economic Growth: The American Record since 1800 (New York, 1964). Both unemployment and underemployment were widespread in the past. Many scholars believe that underemployment in particular was much greater in earlier periods than it is today. Its forms have changed through time. In eighteenth-century France, for example, laborers were often not able to work on religious feast days. This problem has not seemed important to secular scholars, but each year there were 111 feast days in France under the old regime. See George E. Rudé, “Prices, Wages and Popular Movements in Paris during the French Revolution,” Economic History Review 2d ser. 6 (1953–54), 248n.
Fifth, feminists rightly complain of a strong gender-bias in wage indices, which commonly omit the work of women who are not formally in the labor market. How does one estimate the real wages of housewives? Their inclusion poses difficult problems of measurement, and their omission leads to heavy overstatement of real wages per worker. The same problem exists for the unpaid but often very arduous labor of other household members. As more women enter the work force, and fewer children work within the family, the secular effect of this bias is to understate the improvement of real wages in the past century.
Sixth, wage series do not tell us enough about actual living conditions and the standard of living as they have changed through time. There are two problems here. One is conceptual: how is one to define a standard of living? The second is empirical: how should it be measured? Two very able Scottish historians sum up: “We should reiterate the point that any study of the standard of the standard of living is beset with very substantial technical difficulties for the historian, that the study of wages makes up only part of it, and the study of male wages a smaller part still. Income is earned in several ways, and by all the household, so the only fully legitimate way into the problem is through the examination of a total household economy” (A. J. S. Gibson and T. C. Smout, Prices, Food and Wages in Scotland, 1550–1780 [Cambridge, 1995], 356).
But this requirement creates other problems. The “examination of a total household economy” is fraught with difficulty. The evidence itself is much less than total, especially for medieval households. Problems of inference are abundant. Estimates have often been distorted by gross ideological biases in the “standard of living” debate that has raged in economic and social history for many years.
For excellent discussions of the problem see Christopher Dyer, Standards of Living in the Later Middle Ages; Social Change in England, c. 1200–1520 (Cambridge, 1989); and D. Woodward, “Wage Rates and Living Standards in Pre-Industrial England,” Past and Present 91 (1981) 28–45.
For the time being, it is necessary to confine this inquiry to real wages alone, but the limits of this indicator should be clearly understood. It refers only to the purchasing power of money wages for a fixed unit of time, without regard to unemployment, underemployment, unpaid labor, the gray market or the total household economy. It tells us only how the purchasing power of a fixed unit of labor changed through time in terms of a basket of prices. Future inquiries will undoubtedly be able to do better, but for the present this is as far as we can go in a study of long-term secular change in returns to labor.
APPENDIX K
Measures of Wealth and Income Distribution
Many different statistical methods have been used to measure the distribution of income and wealth. They present complex problems of bias in their construction, and are not easily compared with one another.
Most common and straight-forward are what might be called “upper quantile” methods. They estimate the size-shares of the richest I percent or 5 percent or 10 percent, or other top quantiles of the population. Another common measure of a similar type is the Pareto distribution, which calculates the slope of the upper tail of wealth or income holders. These techniques tell us much about patterns of distribution at the top of a wealth-order, but the coverage of all these approaches is biased toward the most affluent classes in a society.
Another favorite device is the Lorenz curve, which is created by plotting the cumulative distribution of wealth for an entire population on the x axis, against the cumulative proportion of the population holding that wealth on the y axis. If wealth is perfectly equal in its distribution, the result is a straight diagonal line, showing that 25 percent of the population owns 25 percent of the wealth, 50 percent owns 50 percent, etc. Where inequality exists, the plot becomes a curve, which moves away from the diagonal line as inequality increases.
Many methods have been invented for summarizing the shape of a Lorenz curve in a single statistic. Chief among them is the Gini ratio, which measures the area between the line of equality and the curve of inequality, as a ratio of the total area below the line. Where perfect equality exists, the line of equality and the curve coincide, and the Gini ratio is zero. Where perfect inequality exists (that is, the upper unit owns everything), the Gini ratio approaches 1.00. In general, the geometry of a Gini ratio tends to give more representation to middling groups, and le
ss to the bottom and top.
Another measure of inequality has been invented by British economist A. B. Atkinson to correct these biases of coverage. It is an index that includes a constant which can be set at different levels, so as to give more or less weight to upper groups, or lower ones. In common practice, the constants are arbitrarily set at several different values and multiple results are given so as to provide different perspectives. Atkinson’s index has become popular among economists, but it is rarely used by historians because it is not as accessible to general readers.
Other measures include the coefficient of variation, various applications of the standard deviation, mean/median ratios, and mean deviations. These tools are very crude and lacking in resolution.
Which measure should one use? A pluralist solution is adopted here, so as to combine clarity and comprehension. Where possible, this inquiry seeks to combine for any given distribution a Lorenz curve, a Gini ratio, and an attached table that lists the proportion of wealth held by each decile of the population. This combination (which can be compressed into a very small space) supplies easily accessible data for the top, middle and lower strata, and also gives the most widely used single summary statistic. The presentation also uses both tabular and graphic representations. The result combines clarity, precision, accessibility and comprehension for different readers.
Unhappily it cannot be used in every instance because of source limits. In some cases only Gini ratios, top quantile shares and zero holders are available.
APPENDIX L
Price Revolutions and Inequality
Why are some people rich and others poor? What is the cause of material inequality? How has inequality changed through time? What, if anything, can or should we do about it? These eternal questions have given rise to many models of inequality, which differ both as theoretical propositions and empirical descriptions. Several leading models might be summarized in a few sentences, and then compared with evidence that we have found in this inquiry.
Great Wave Page 30