Engel versus Baumol: Accounting for structural change using two centuries of U.S. data☆
Introduction
In the last two centuries, the reallocation of labor out of agriculture has been a dominant feature of structural change and economic growth in the United States. The following figures put this striking transformation into perspective: the share of farm employment in the United States dwindled from 75% in 1800 to less than 3% in 2000, while the share of farm output in GDP declined from 40% in 1840 to slightly above 1% in 2000; see Fig. 1. These secular changes in farm shares took place both when large scale immigration increased the farm population (up until 1910), as well as when the farm population decreased dramatically during the twentieth century.1 What then are the key sources of this transformation?2 Not surprisingly, the existing literature has advanced numerous explanations to account for this complex but nevertheless highly secular process. These range from demand-driven factors to purely technological determinants of structural change. We list the most prominent ones here [see also Gallman (2000, pp. 47, and 50–51)]:
- 1.
A version of Engel’s Law operating on employment shares: as incomes rise, agriculture sheds labor due to the low income elasticity of demand for farm goods.3
- 2.
A version of Baumol’s (1967) “cost disease”: relatively faster productivity growth in agriculture pushes farm workers to produce complementary non-farm goods.4
- 3.
Different capital intensities in production: agricultural production is more conducive to rapid capital deepening, which in turn pulls labor into the more labor intensive non-farm sector.
How much has each of these factors accounted for the structural transformation in the United States in the last two centuries? In this paper we address this central question.5
Before we empirically address this question, however, we need appropriate data. Hence, an important contribution of our paper is collecting and carefully parsing the available data series. Almost all the facts and trends we draw upon in this paper have been extensively documented by economic historians in various contexts. However, this is the first time to our knowledge that all of the data required to address the specific theoretical considerations considered here, over a two hundred year time frame, are brought together in one place. Section 2 documents these historical data and tendencies.
In order to systematically quantify the contribution of individual forces to the declining share of agricultural employment in the United States, we also require a unified accounting framework. Section 3 develops a framework founded in economic theory that disaggregates the changes in the employment share of non-agriculture attributable to each of the three drivers of structural change listed above. While similar in spirit to Solow-style growth accounting in its use of production efficiency conditions, this approach builds on a two-sector model to capture labor reallocation between sectors. Although the final accounting expression is non-linear in its components, all the key drivers of structural change emerge in a clear and economically intuitive way. The framework also highlights the often-overlooked but potentially significant interaction between the differential sectoral productivity growth and differential capital deepening effects.
Section 4 presents the quantitative implementation of this accounting framework. The results show that the combination of an income elasticity of demand for farm goods below one and differential sectoral productivity growth in favor of agriculture has been a very significant determinant of the U.S. labor reallocation process. However, over the last two centuries, the relative contributions of these two effects have changed in important ways. We find that a low income elasticity of demand accounts for the bulk of labor reallocation until about the 1950s, after which differential sectoral productivity growth becomes a key determinant.
Working with two centuries of data also reveals that the contribution of the differential sectoral productivity growth effect differed significantly between the nineteenth and twentieth centuries: during the nineteenth century, productivity growth in agriculture lagged behind that of non-agriculture and as such acted as a headwind that slowed the reallocation of labor out of agriculture. We also find that, although the difference in sectoral factor intensities underlying the differential capital deepening effect has amplified the relative productivity effect, the independent contribution of the capital deepening effect has not significantly contributed to the reallocation of labor out of agriculture in the last two centuries. Our empirical findings therefore suggest that historical interpretations (and theoretical models) that emphasize only one dimension of this process cannot adequately account for the dramatic trends in historical U.S. structural change.
We also examine one important extension of the baseline framework—international trade—in Section 5. Although exports and imports are a relatively small share of U.S. GDP and their overall importance has varied over time, we extend the model to evaluate the significance of international trade in agricultural goods given the United States’s historic openness to trade in this sector. We find that extending the accounting framework to an open economy context does not substantially affect our earlier conclusions.
The sources of structural change we discuss in this paper are ultimately driven by endogenous and exogenous sources of economic growth, such as technological progress and (transitional) capital accumulation.6 The accounting framework we use in this paper is not designed to capture these ultimate sources of structural change. The objective of this exercise is rather to ask whether several reasonable statements about technology and preferences are capable of quantitatively accounting for U.S. structural change. This is reminiscent of the approach taken by Solow (1957) in his classic study on aggregate growth accounting. This paper therefore represents an important intermediate step in a more encompassing (and demanding) exercise that links structural change to all the institutional and market forces that determine observed technological progress. Nevertheless, we conclude in Section 6 by offering our thoughts on the implications of our findings for models of directed technological progress.
Section snippets
Three sources of structural change: basic facts
To illustrate the potential relevance of the three distinct drivers of structural change emphasized by the existing literature for the United States, we begin the analysis with an overview of fundamental trends in the data. First, consider the evidence for low (less than unitary) income elasticity of demand for agricultural goods. We have high quality data from the twentieth century: despite an almost secular rise in per capita food expenditures, the share of food in total expenditures has
An accounting framework for structural change
The conceptual framework we develop below is related to recent work by Caselli and Coleman, 2001, Dennis and İşcan, 2007 who also empirically examine structural change in the United States by emphasizing the decline of the agricultural sector. However, these papers primarily rely on twentieth century data and, as we discuss below, some of the trends identified in these papers do not apply to the nineteenth century. Moreover, these papers either limit their analysis to a few sources of
Data
We rely on numerous data sources to account for the U.S. structural change in the nineteenth and twentieth centuries. See Table 1 for a description of variables and data sources, and Dennis and İşcan (2008) for the details. Our nineteenth century data set builds on the meticulous work of numerous economic historians affiliated with the NBER, and the twentieth century data come mostly from official sources. Not surprisingly, our desire to use both nineteenth and twentieth century data faces
International trade
How does openness to international trade affect our understanding of this historical process? After all the U.S. has historically been a net exporter of farm goods, and historians have long argued that in the absence of foreign demand for United States farm goods, the reallocation of labor out of agriculture would have been faster, especially in the nineteenth century (Gallman, 2000).
To account for the impact of international trade on the reallocation of labor out of agriculture, we start by
Concluding remarks
The structural transformation of the U.S. economy from an agricultural to an industrial base was a rapid and striking event. Surprisingly, very few attempts have been made to quantitatively account for this phenomenon. In this paper, we propose an accounting framework to decompose the reallocation of labor out of agriculture into three of its well-studied sources. We find that the low income elasticity of demand for agricultural goods accounts for much of the labor reallocation until the 1950s,
References (47)
- et al.
Productivity growth and agricultural out-migration in the United States
Structural Change and Economic Dynamics
(2007) - et al.
Capital deepening and nonbalanced economic growth
Journal of Political Economy
(2008) - et al.
Capital deepening and the rise of the factory: the American experience during the nineteenth century
Economic History Review
(2005) - et al.
The farm, the farmer, and the market
Macroeconomics of unbalanced growth: the anatomy of urban crisis
American Economic Review
(1967)- et al.
The U.S. structural transformation and regional convergence: a reinterpretation
Journal of Political Economy
(2001) Income elasticities of demand and the release of labor by agriculture during the British industrial revolution: a further reappraisal
Journal of European Economic History
(1980)- et al.
Hours at work and total factor productivity growth in 19th century U.S. agriculture
Advances in Agricultural Economic History
(2000) - Dennis, Benjamin N., İşcan Talan B., 2008. Data appendix to Engel versus Baumol: accounting for structural change using...
- et al.
Technology and industrialization, 1790–1914
Changes in sectoral composition associated with economic growth
International Economic Review
The Escape from Hunger and Premature Death, 1700–2100: Europe, America and the Third World
The Service Economy
Commodity output 1839–1899
Changes in total U.S. agricultural factor productivity in the nineteenth century
The United States capital stock in the nineteenth century
Economic growth and structural change in the long nineteenth century
The service industries in the nineteenth century
American Agriculture in the Twentieth Century
Productivity Growth, Inflation, and Unemployment
Consistent wholesale price series for the United States, 1860–1990
Engel’s law
Comment on The U.S. structural transformation and regional convergence: a reinterpretation
Journal of Political Economy
Cited by (58)
Accounting for structural transformation in the U.S.
2022, Journal of MacroeconomicsStructural transformation and inequality: The case of South Korea
2022, Economic ModellingUrban industrial transformation patterns under natural resource dependence: A rule mining technique
2021, Energy PolicyCitation Excerpt :In summary, the factors influencing the IST for RCs used in this paper are shown in Table 1. However, it cannot be ignored that other influential factors like energy price (Ma and Yu, 2017), energy efficiency (Yan et al., 2019a), the reallocation and transfer of labour (Dennis and İşcan, 2009) may also influence the IST of RCs. The factors clarified in this paper cannot claim to be exhaustive, as the factor identification in this paper has to obey two additional principles: (i) the research questions and the rule mining technique determine that the considered factors should possess interurban comparability to reflect diverse urban features; and (ii) the considered factors should possess relatively complete measurable data sets at the urban level to ensure the objectivity of rule mining results.
Productivity, relative sectoral prices, and total factor productivity: Theory and evidence
2021, Economic ModellingLeisure time and the sectoral composition of employment
2020, Review of Economic DynamicsStructural transformation and productivity growth in India during 1960–2010
2019, Economic Modelling
- ☆
We thank Chris Hanes, Chris Pissarides, Ranjit Dighe, two anonymous referees, participants at the 2007 North American Econometric Society Winter Meetings, and 2006 Canadian Economic Association Meetings for comments, Laura O’Hearn for editorial comments, and Tom Weiss and Chris Hanes for making their data available in electronic format. Part of this research was conducted while İşcan was visiting Cornell University. The data used in this article are available at http://myweb.dal.ca/tiscan/.