Enclosures and population movements in England, 1700–1830: A methodological comment

Enclosures and population movements in England, 1700–1830: A methodological comment

JOHN A. TOMASKE California State College at Los Angeles In a recent paper, Mr. Lawrence J. White used regression and correlation analysis to test t...

329KB Sizes 0 Downloads 1 Views

JOHN

A. TOMASKE

California State College at Los Angeles

In a recent paper, Mr. Lawrence J. White used regression and correlation analysis to test the proposition that internal migration in eighteenth and early nineteenth century England was related to the process of enclosure.’ This note contends that the methodology and the statistical techniques used are inadequate to expose the relationship which may or may not exist between migration and enclosure. This note also explores the possibility that these techniques are responsible for the statistical results obtained by Mr. White. To test his hypothesis, Mr. White computed estimates of migration from cross-section data for English counties. The regression equations are of the form M i = a + bE, + ui . Mi is defined as the quantity of net migration into or out of the ith county. E, is defined as the number of acres enclosed in the ith county and ui is the disturbance item? The regression coefficients are a and b. A second set of regression estimates uses the product of acres enclosed and the mean population per acre in the ith county as the explanatory variable.$ The study concludes that population migration is not significantly associated with enclosures4 The first criticism is directed at the theoretical model used to explain migration. Mr. White explicitly rejects the notion that the enclosure movement depopulated the English countryside. He not& that rural population increased during this period and that this was ac*I Gould like to thank the referees of this journal for their comments on an earlier draft of this note. Any remaining errors are, of course, the responsibility of the author. ‘Lawrence J. White, “Enclosures and Population Movements in England, 1700-1830, “Explorations in Entreprenrural History 6, no. 2 (Winter ‘19691, pp. 175-186. 21bid., p. 176. 3rbid., pp. 178-180. *Ibid., pp. 175, 184-185.

224

Explorations

in Economic History

companied by substantial rural-to-urban migration. His hypothesis is that despite rural population growth, the enclosures may have somehow forced or pushed a “surplus” rural population off of the land.5 This hypothesis says nothing of the alternatives available to these migrants. In this view, migration takes on the character of an evacuation rather than a response by owners of productive factors sensitive to their own best interests. This “push” hypothesis contrasts sharply with much of the current literature on the determinants of migration6 An important contemporary approach to the study of migration applies capital theory to the migration process.7 Migration is treated as investment in human capital and the decision to migrate is seen as an investment decision. The expected returns to migration consist of differentials in the income streams associated with alternative employment opportunities. The costs of migration consist of the direct costs of travel and the income foregone during moving, the search for employment, and any necessary training period. Migration would occur only when the expected benefits of migration, suitably discounted, exceed the costs of migration. A second approach emphasizes the selectivity of the migration process. Migrants are considered to be risk-taking individuals with high adaptive qualities.8 The human capital approach does not deny the selectivity of the migration process. This would be explained in terms of greater income differentials, longer payoff periods or lower migration costs. In contrast to Mr. White’s hypothesis, these models treat migration as a response to labor market disequilibrium and are consistant with the general theory of resource allocation. These models do not attempt to determine or even define the “key” to migration. The ob5Zbid., pp. 175, 182. 6A summary of earlier work is given in: United Nations, Determinants and Consequences of Population Trends, Ppoulation Studies, no. 17, ST/SOA/Series

A (New York:

Department

of Economics

and Social Stud-

ies, 1953). Recent work of interest to historians is summarized in: John A. Tomaske, “An Economic Enquiry Into the Causes of International Migration” (Ph. D. diss., University of Washington, 1968). rAn extended discussion of this approach is given in: Larry A. Sjaastad, “The Costs and Returns to Human Migration,” The Journal of Political Economy 70, no. 5, part 2 (October 1962), pp. 80-93. See also: Philip Nelson, “Migration, Real Income and Information,” Journal of Regional Science1, no. 2 (Spring 1959), pp. 45-58. %imon Kuznets and Dorothy Thomas et al., Population Redistribution and Economic Growth, The United States, 1870-1950,3 ~01s. (Philadelphia: American Philosophical Society, 1957, 1960, 1964).

Enclosures

and Population

Movements

in England, 1700-1830

22.5

jective is to construct a properly specified model which “explains” a substantial proportion of the variation in the migration variable. The results of other work not incorporated in Mr. White’s study also indicate that his model may be poorly specified. A major omission is the now classic study of English migration by Arthur Redford based on English census data.g A major conclusion of Redford’s study was the importance of distance as a deterent to migration. He found that outward migration decreased as the distance to the destination of the migrants increased. Redford also found that migration was usually by short stages. In addition, the theoretical and statistical model used by Mr. White does not take into account the availability of non-agricultural employment opportunities in the home county. This alternative to migration out of the home county is likely to be of some importance. Deane and Cole estimate that by 1801 approximately 64% of the occupied population was engaged in nonagricultural employment. lo The partial and preliminary results of my research on English intercounty migration of this period suggests that (1) the percentage of the county population resident in urban areas and (2) the natural increase of population, suitably lagged, are important explanatory variables. *l The first variable serves as a rough indicator of non-agricultural employment opportunities while the second variable reflects the relative supply of labor and migrants in the appropriate age groups. Under these circumstances, the ceterus paribus assumptions underlying Mr. White’s model are likely to be violated. Therefore, the statistical relationship between migration and enclosures may be obscured by the effect of variables not included in the mode1.12 ‘Arthur Redford, Labour Migration In England, 1800-1850 (Manchester: The University of Manchester Press, 1962). loPhyllis Deane and W. A. Cole, British Economic Growth, 1688-1959 (Cambridge: Cambridge University Press, 1962), p, 142. “Taken together, these variables “explain” approximately 40% of the variation in emigration rates among English counties with net outmigration, 1800-1830. 12The relationship between enclosure and migration may be even more complex than is implied in this note. Professor Blaug notes that enclosure of waste land increases employment while enclosure which increases the area under wheat cultivation increases the seasonal variability of employment. Mark Blaug, “The Myth of the Old Poor Law and the Making of the New,” The Journal oj’Economic History 23, no. 2 (June 1963) pp. 170-171. The complex relationship between a modern enclosure movement and migration is discussed in: Richard H. Day, “The Economics of Technological Change and the Demise of the Sharecropper,” The American Economic Review 51, no. 3 (June 1967) pp. 427-449.

226

Explorations

in Economic History

The second criticism is directed at the form of the data used in estimating the regression equations. In migration studies, it is common to standardize the migration variable by dividing the migration data by the relevant population base. In addition to facilitating the comparison of the results of different studies, the use of migration rates reduces the effect of population size on the migration variable.13 Other things remaining the same, the size of the population of an English county will directly effect the number of people who can respond to a given stimulus. Technical considerations also indicate that the use of raw data rather than migration rates can result in the residuals having a non-constant variance (heteroscedasticity) and in difficulties with the standard errors of the estimates. Similar logic applies to the use of number of acres enclosed as an explanatory variable. The enclosure of 30,000 acres in a small county such as Rutland (area: 97,273 acres) should have a different significance than the enclosure of the same area in a large county such as Norfolk (area: 1,308,439 acres). The lack of a statistically significant relationship between numbers of migrants and the number of acres enclosed may indicate only that acres enclosed is not associated with the size of county populations and the area of these counties. In a second set of regression estimates, the explanatory variable is reformulated to include the effects of population density.14 This variable is constructed by multiplying the number of acres enclosed in each county by the county population density per acre. The result is an estimate of the number of individuals on the enclosed acreage. Assuming little variation in climate, soil type, technology, product mix and urbanization both within and among counties, this measure may be a reasonably accurate estimate of the number of individuals directly affected by enclosure. In this case migration should be positively associated with this variable. Perhaps the greatest source of error in this estimate stems from intercounty differences in urbanization and non-agricultural employment. In relatively urban counties or those with significant non-agricultural employment, this measure overstates the number of individuals affected by enclosure of agricultural land. This matter is complicated by the fact that in a premodern economy with cottage industry, non-agricultural employment may not be reflected in urbanization. It is, however, likely that local non-agricultural employment, which provides a viable alterna13Philip Nelson, “Migration, Real Income, and Information,” p. 46. ‘?Lawrence J. White, “Enclosures and Population Movements,” pp. 178% 180.

Enclosures

and PopuIation

Movements

in England, 1700-1830

227

tive to migration, may be associated with relatively high population densities. Given these considerations, it is not clear, apriori, how this variable and migration should be related. Finally, it is unfortunate that the available data is net migration rather than gross migration. The model asserts that enclosure induces out-migration. This does not explain the reverse flow of migrants included in the data and caused by influences not included in the model. In terms of this model, the data contain errors of observation which may not be random. This can introduce bias into the estimates of the regression coefficients and may reduce their statistical significance. In any case, the computed correlation coefficients will differ from the true correlation coefficients. The purpose of this note is not to present an alternative model of English internal migration. The objective is merely to indicate that there is sufficient reason to doubt Mr. White’s verdict on the relationship between enclosures and the historically important process of migration and urban growth.