Selling food or housing to themselves

The meaning of price movements depends critically, of course, on whether one is a buyer or a seller. Past measurements of the cost of living and real wages have assumed that households sell everything they make and buy everything they consume.

The first place where the consumption of non-purchases intrudes into the traditional discussion is in the consumption by lower-income rural households of food

and clothing they produce themselves. As Witold Kula and William Hagen have rightly warned, ordinary peasants do not experience anything like a 30% loss of real income when food prices rise 30% relative to the wage rate (Kula 1976; Hagen 1986). In many cases, their real income is wholly unaffected. This clearly important point is hard to quantify. In England, the socio-occupational categories in the social tables of 1688—1803 imply that among families with below- average income the share of home food consumption that was home produced was probably between 4% and 30% of total income.¹⁵ The importance of home production will have been greater in less market-integrated settings. In early modern Europe, this kind of protection against price movements was probably greater in the countryside and to the east. All subsequent discussions of real wages as measures of the real purchasing power of ‘the masses' should acknowledge that real-wage measures overstate the real-income importance of movements in those measured real wages.

Yet the likely magnitude of home production as a share of workers' incomes would not have been as great as the share of a different kind of home production, one that affected the top income groups.

For all the narratives about the cost and condition of workers' housing, very few historical cost-of-living indices have been able to include housing rents. This is not surprising, since housing is more varied and changing in its quality than bread or coal. For housing, as for fashionable luxuries, the variety and change frustrate our attempts to build long-run time series. Yet in a few cases it has proven possible to produce long time series of the rental prices of housing of given quality. We now have fair housing rent series for Holland and Belgium. Philip Hoffman has assembled another index for pre-Revolutionary Paris. In addition, Gregory Clark has now assembled credible housing rent series extending from the mid-seventeenth to the mid-nineteenth centuries for London, for the rest of England, and for England as a whole.¹⁶

Housing rents, as we saw in Table 6.4, rose at least as fast as the cost-of-living index in general up to the early nineteenth century, and they rose even faster across the rest of that century. Over the last five centuries as a whole, housing is probably the one major consumer price that might have kept pace with the wages of labour, when quality is held constant.

If everybody spent the same share of their budget on rented housing then the movements in the relative price of housing would play little role in class differences in cost-of-living trends. Housing's share of a broad concept of total income or expenditures is indeed roughly constant across the income classes in any given year. Economists have noted that the income elasticity of housing is near enough to unity that one could use the value of occupied housing as an index of permanent income.¹⁷ So as far as the use of housing is concerned, its rough proportionality to income might suggest that movements in the real price of housing should not have caused any difference in cost-of-living trends between the rich and the poor.

Yet, the impact of house-rent movements on the different income classes was probably far from neutral, since the rich owned housing and the poor typically

did not. It would clearly be wrong to imagine that a jump in rents would impoverish an affluent household that owned its own home. And if that affluent household owned housing that it rented out to others, the jump in market rents would raise, and not lower, its real income. Many affluent households were in fact exporters, not importers, of residential housing. A rise in real rents would therefore favour the rich.

The fact that housing owners were higher in the income ranks than housing occupants would still not have complicated our accounting for relative real incomes if the historical record of nominal incomes had correctly included all the incomes that owners received from their residential properties, including their own (owned) home. Had that been the case, then any jump in real rents would be recorded correctly as a jump in their nominal income, partly offset by the fact that some of the housing they own was consumed by themselves rather than rented out for income.

That is apparently not the case, however. None of the scholarly estimates of early high incomes makes any explicit reference to an imputation for owner-occupied housing. Typically someone like Gregory King would show awareness of owner-occupied housing only in separate discussion of the nation's housing stock, and would not carry its value over to any tabulation of income by class. Nor, as far as we can tell, did the official returns on income taxation in England and France give this point its due. Our views of the real-income movements of high-income groups are in danger of implicitly assuming that they had to rent all their housing from landlords. We will have stepped into that trap if we take nominal incomes, excluding imputations for owner-occupied housing, and deflate them by a cost-of-living index that implies that these affluent families had to rent all their lodging.

Any era in which rents rose faster than the composite price index on the rest of a lower-class consumer bundle was likely to have been an era in which the cost of living rose even less for the most propertied classes than even the present estimates show.¹⁸

6.3