The declining real cost of sen/ants
The real-wage rate itself has a further implication about real inequality. High-income families hired labour directly as household servants and day servants. So the lower the real-wage rates sank before 1820, the cheaper the cost of living the high-income lifestyle.
This key point needs to be quantified as best the data permit. Conversely, the rise in real wages after the 1820s meant that the cost of living of the rich advanced faster than the cost of living for workers. Who hired servants, and in what numbers, can be roughly gauged from a variety of sources. Since servants' nominal wages apparently followed the wage rates for the unskilled agricultural workers (Snell 1985: 25—47, 411—17), which in turn moved similarly with unskilled building wages before 1820, one can use an unskilled wage rate as a rough index for servants' wages.To the extent that upper-class employers paid their servants in kind, one could say that expenditures on servants are already built into the data on other expenditures of the household. That is the case for the English data in Table 6.2, though not for the Duc de Saulx-Tavanes. Yet even for the English data, it makes sense that the pay in kind for servants varied in its nominal value with the nominal wage of servants, more directly than with the prices of the goods and services purchased for them. Accordingly, we have inferred the shares of food and other items implicitly paid to servants by applying low-income consumption weights to total expenditures on servants, and have deducted these from the rich-household expenditures on these other categories in Table 6.2. To the estimated full servant bill—both for those paid in kind and for those paid in cash—we use the unskilled wage rate as the barometer of their unit cost. The unskilled wage rate itself thus becomes a luxury-service price, one that greatly affects our estimates of the trend in living costs for the rich.
6.2