New Income Comparisons for the late Nineteenth and Early Twentieth Century

• Published date: 01 March 2021
• Author: Marianne Ward,John Devereux
• Date: 01 March 2021
• DOI: http://doi.org/10.1111/roiw.12466

© 2020 International Association for Research in Income and Wealth

222

NEW INCOME COMPARISONS FOR THE LATE NINETEENTH AND

EARLY TWENTIETH CENTURY

by Marianne Ward*

Department of Finance and Economics,The Bill Munday School of Business, St. Edward's University

AND

John devereux

Department of Economics,Queens College, CUNY

We provide current price GDP comparisons covering 13 Western economies for 1872 and 1910. The

current price measures compare income with the prices of the comparison year rather than the fixed

recent prices of earlier approaches. Our work fills a gap in the literature since current price GDP com-

parisons generally do not reach before 1950. As it turns out, the current price results differ from the

familiar Maddison projections in constant 1990 prices as they raise relative income for the U.S. and

Canada while they lower income for Belgium, the Netherlands, Switzerland and the U.K.

JEL Codes: N10, O40, O50

Keywords: growth, convergence, current price income comparisons

1. introduction

Modern economic growth began in Western Europe and spread outwards. The

study of the characteristics of growth—structural change, convergence/divergence

and changes in economic leadership—requires comparable long run GDP. Over

recent years, the monumental work of the late Angus Maddison (1995, 2001, 2007) has

dominated the field. Maddison compared income using 1990 relative prices. Recent

years, however, have witnessed major changes in how economists compare GDP;

see Feenstra et al. (2015). Most notably, the field has moved to using current prices,

that is, to comparing income with prices close in time to the desired comparison year

rather than with the fixed recent prices of earlier approaches. The latest version of

the Penn World Tables (henceforth the PWT) exemplifies this change as it provides a

series for GDP comparing income using multiple current price benchmarks.

This paper provides current price historical GDP comparisons for thirteen

western economies. The benchmarks cover 1872 and 1910. They fill a gap in the

Note: We are grateful to Peter Lindert, Jeffrey Williamson, Leandro Prados de la Escosura,

Herman de Jong, and to all the participants at the GGDC 25th Anniversary Conference in June 2017

for very helpful comments. We also give special thanks to Prasada Rao for help above and beyond the

call of duty.

*Correspondence to: Marianne Ward, Department of Economics, The Bill Munday School of

Business, St. Edward's University, 3001 South Congress Avenue, Austin, TX 78704, USA (mperadoz@

stedwards.edu).

Review of Income and Wealth

Series 67, Number 1, March 2021

DOI: 10.1111/roiw.12466

bs_bs_banner

Review of Income and Wealth, Series 67, Number 1, March 2021

223

© 2020 International Association for Research in Income and Wealth

literature as current price comparisons generally do not reach before 1950. The

results, as it turns out, differ from fixed price projection in significant ways. To

summarize, they raise U.S. and Canadian income while they lower income for

Belgium, the Netherlands, Switzerland and the U.K. The Swiss case stands out.

The fixed price projections show Switzerland as the leading economy for 1910 with

income per capita that is forty percent above the U.S. whereas current price com-

parisons show the Swiss as well down the pack.

As mentioned, traditional approaches compare income over long time spans

using fixed recent prices. Maddison (1995, 2001, 2007) popularized this approach

and his work is carried on by the Maddison Project, see Bolt and Van Zanden

(2014). Maddison compares income as follows. First, he takes a purchasing power

parity (PPP) adjusted GDP comparison for 1990 from the International

Comparison Program (henceforth the ICP). Next, he projects his ICP benchmarks

backwards using GDP growth rates.1 The resulting series are in 1990 prices.

The Maddison GDP projections face three difficult index number problems.

First, they compare income across space for a recent year. Second, they compare

income over long spans for the base country and, finally, they compare income over

long spans for the comparison country. Half a century ago, the distinguished eco-

nomic historian Robert Gallman projected income for France, Britain and the U.S.

from 1950 to 1840 using a 1950 base year comparison. His description of the con-

ceptual difficulties he faced (Gallman, 1966, p. 6) holds for all subsequent efforts:

The procedure has one very important disadvantage: the results are

difficult to interpret. As a first approximation, we are comparing 1840

(circa) national products valued in 1950 prices, since the extrapolated

estimates are in 1950 prices. But the extrapolators are not based on

1950. The price base of a constant price national product series affects

the rate of growth of the series; in general, the earlier the price base,

the higher the rate of growth. Since the base years of the extrapolators

are earlier than 1950, the extrapolated 1840 values are really smaller than

1840 national ’products in 1950 prices. If the extent of "bias" in the three

series were identical, the comparisons would be unaffected, of course;

but there is no good reason for supposing that they are. The date of the

price base differs from series to series. In addition, the extent to which

an early price base raises the rate of growth of a national product series,

compared with a late price base, depends on the extent of changes in the

price structure over time. There is no good reason for believing that the

price structures of the three economies changed at the same pace.

We would add a further difficulty to Gallman’s list. Suppose we project GDP for

two countries to the distant past, say 1870. A change in the base year or a revision

in the underlying GDP series for either economy between 1870 and the base year

will change the 1870 projection. Since such revisions are inevitable, projections are

condemned to forever rewriting history—a point made six decades ago by Simon

1Clark (1940) appears to be the first researcher to use projections for long run income

comparisons.