Keywords

1 Relationship to MeasuringWorth

MeasuringWorth is a historical website created and headed by Samuel H. Williamson. Sam graciously asked me to be co-founder of the site and to serve as Director of Research, which I did for several years. Under that rubric I developed a deep and sustained interest in long-term economic series. Some of the data series in Part III are available on MeasuringWorth in updated format.

2 Terms of Trade (Chapter 9)

In reviewing Officer (2021), Devereux (2022) states:

Take the external terms of trade. Officer covers commodity and service trade for the entire period, where most work in economic history is for commodity trade. He improves deflators and replaces the fixed weight price indices with a more appropriate deflator. The result is that we now have an external terms of trade series for the U.S. from 1790 to now that is superior to the estimates for other developed economies.

Devereux's most-serious criticism is that “some of the most important series appear only as diagrams—including the external terms of trade and the various price series.” In listing the terms of trade and related series, Table 9.1 provides a partial response.

3 Value of Consumer Bundle

In Officer (2007b) I develop three related U.S. series: value of the consumer bundle (VCB), number of consumer units (CU), and average size of the consumer unit (SZ) annually for 1900–2004. VCB is average annual expenditures per consumer unit. A consumer unit, the entity that makes expenditures decisions, is different from a household. One household is the entirety of persons who occupy a housing unit. There can be more than one consumer unit in a household, and there can be consumer units in a non-household setting, namely, non-institutional “group quarters.” So the number of consumer units exceeds the number of households. Table 12.1 presents the series “value of the consumer bundle” (VCB) and “number of consumer units” (CU).

Table 12.1 Value of consumer bundle and number of consumer units
Full size table

Size of a consumer unit is the number of persons that constitute the unit. Average size of the consumer unit (SZ) is 3.5 1900–1902, 3.4 1903–1917, 3.3 1918, 3.4 1919–1921, 3.3 1922–1933, 3.2 1934–1938, 3.3 1939–1941, 3.2 1942–1962, 3.1 1963–1966, 3.0 1967–1970, 2.9 1971–1974, 2.8 1975–1978, 2.7 1979–1982, 2.6 1983–1991, 2.5 1992–2004.

VCB is denominated in current dollars. To serve as a measure of standard of living over time, VCB needs to be adjusted, performed in Sect. 11.1.2.

One would think that “consumer unit,” which by definition is the decision-making unit for expenditures, would be the preferred entity for economic analysis. However, “household,” the body of people who occupy a dwelling unit, remains the primary concept for historical research. Consider the monumental work of Robert J. Gordon (2016, p. 36), who computes “average household consumption” [AHC] as $983 in 1870. That figure is too high relative to $733 for VCB in 1900 (the earliest year of the series). How can that be explained?

Gordon (2016, pp. 36; 673, note 1) estimates current-dollar per-capita GDP in a roundabout way, adopts a consumption/GDP ratio of 0.76, and applies a five-person average household, resulting in the $983 figure. What is VCB for 1870? Consider a four-step process.

First, recompute AHC for 1870, retaining Gordon's methodology but using a direct source for per-capita GDP: Louis Johnston and Samuel H. Williamson (2021). The result is $744.1 This figure is personal consumption expenditures [PCE] divided by number of households.

Second, correct AHC so the numerator pertains only to the consumer-unit universe. The technique is to multiply AHC by the share of consumer units (population in households plus group-quarters residents) in total resident population (PHGQ/POP, in Adjustment of PCE for consumer-unit universe, in Officer 2007b, Sect. 5). The data exist for Census years, including 1870.2 The multiplicative factor is 0.96, the same as for the year 1900, reducing the figure to $715.

Third, estimate the number of consumer units. The technique “to complete the CU series” in Officer (2007b, Sect. 4), was selected there because the developed synthetic series (SCU) is available annually; but there is a serious question of reliability as one proceeds further into the past. A preferred extrapolator, PHGQ (per note 2), can be employed here, because Census data are all that are required. CU in 1870 is estimated as the 1870/1900 PHGQ ratio times CU in 1900, with result 11,166 thousand.

Fourth, adjust the corrected AHC so the denominator is the number of consumer units rather than the number of households. With the number of households in 1870 at 7471.754 thousand (Ruggles 2006, Table Ae-A, 1950–1970 definition), the corrective multiplicative factor is 7471.754/11,166, about two-thirds, whence estimated VCB in 1870 is $478.

Conclusion: The Gordon figure for average household consumption in 1870 is more than double the VCB for that year! In general, with the number of consumer units exceeding the number of households, average household consumption is an overestimate of the consumer expenditures of decision-making units.

4 Consumer Price Index (Chapter 10)

In Officer (2007a) I generate a new U.S. long-run consumer price index (CPI) that is an improvement over alternatives, the most-important of which is the Historical Statistics series, presented in Lindert and Sutch (2006). The new series is better in several respects. First, it utilizes a neglected but impressive series of Paul H. Douglas (1930) for the 1914–1917 period. Second, it links component series for conceptual consistency and superior reliability. Third, it embodies enhanced computational accuracy and avoids rounding error. Various tests in Officer (2007a, pp. 141, 145–146) are indicative of the superiority of the new series over the Historical Statistics equivalent.

The new CPI series is shown in Table 12.2. This CPI series pertains to the domestic U.S. population; it is distinguished from the CPI series in Sect. 10.1, which applies to foreign travelers in the United States.

Table 12.2 New CPI series
Full size table

The new CPI improves the official consumer price index, but only within a narrow statistical framework. There are biases (and other limitations) of the CPI that remain in both the official and improved series. For discussion of the biases, one can consult Brent R. Moulton (1996) and David E. Lebow and Jeremy B. Rudd (2003). For the historical political context of the CPI, one may read Thomas A. Stapleford (2009), the subject of the book review in Sect. 10.2. The VCB and (improved) CPI interact in Chapter 11.

5 Compensation of Manufacturing Workers (Chapter 11)

5.1 Reception

I was flattered by the comment of Robert E. Hall in the back cover of Two Centuries of Compensation for U.S. Production Workers in Manufacturing (Officer 2009): “Highly valuable to scholars interested in quantitative economic history…An intellectual triumph.” Subsequently, Joshua L. Rosenbloom (2009) begins and ends his review of (Officer 2009) as follows:

I suspect that few people will be tempted to read this slim volume cover to cover. But many of them will find it an extremely valuable reference to which they will return numerous times...Anyone with an interest in the long-run growth of the U.S. economy, or the development of American labor markets will find this book an important and useful reference.”

5.2 Data Series

Rosenbloom (2009) makes the following observations on the book's concluding chapter (which is Chapter 11 of the present work).

Readers who are interested primarily in the bottom line will want to skip directly to the concluding chapter of this volume, in which the author presents his estimates of average hourly compensation and its components, average hourly earnings, and average hourly benefits in both nominal and real terms. The story that these series tell is in one sense not that surprising. Since 1800, there have been huge increases in nominal compensation; although some of this increase is due to changes in the cost-of-living, real compensation has nonetheless increased dramatically in the last 200 years. The series reported here indicate that average hourly compensation adjusted for inflation increased from $0.33 in 1800 to $12.09 in 2006 (both measured in 1982-84 prices), a nearly 37-fold increase. Growth was somewhat slower in the nineteenth century, and accelerated after 1900, but the series then leveled off in the 1980s, and remained essentially flat until the early 2000s.

While the broad outlines of Officer's series are consistent with other sources, the shorter run movements of average hourly compensation differ from those of a number of real wage series available over shorter periods. In particular, it appears that average hourly compensation grew faster than wage series constructed by other scholars for most of the nineteenth century.

John Pencavel (2011, p. 566) observes that my real hourly compensation series (AHCR) “suggests a rise in real hourly compensation between 1890 and 1914 of 36.4%, a figure between Douglas’ and Rees’ but closer to Rees.” He finds that the lower growth in real hourly compensation compared to Rees results from lower growth in nominal compensation (AHC) rather than higher growth in my CPI.

Gordon (2016, p. 279) uses data of Albert Rees (1961) to state: “By 1914 [from 1870], the average nominal manufacturing wage had increased by 30 percent from seventeen cents per hour to twenty-two cents per hour.” Consistent with Rosenbloom's rather than Pencavel's comment, my series shows a growth of 45%.

Notes

  1. 1.

    The product of Johnston-Williamson per-capita GDP ($195.76), the Gordon consumption/GDP ratio (0.76), and Gordon's household size (five).

  2. 2.

    Population in households (PH) in Susan Brower and Steven Ruggles (2006, series Ae85), group-quarters residents (GQ) in Steven Ruggles (2006, p. 1–654, Table Ae-A, 1950–1970 definition), resident population (POP) in Michael R. Haines and Richard Sutch (2006, series Aa9). PHGQ = PH + GQ.