Interesting items about Household Studies

Back a hundred years or so, back when home economics was considered a valued topic of study, the household featured as a unit of analysis.

This is shown in the book authored by R. G. D. Allen and A. L. Bowley, Family Expenditure: A Study of Its Variation (London: P. S. King, 1935; LSE Studies in Statistics and Scientific Method, no. 2). It is a short monograph—about 145 pages—not a U.S. government survey volume, though 1935–36 is also the fieldwork year of the huge American Study of Consumer Purchases, which later produced many BLS bulletins titled Family Expenditures in Selected Cities, 1935–36. Those are different works.

Allen was then a young mathematical economist at the London School of Economics (soon famous for the Hicks–Allen theory of demand). Bowley was the senior statistician who had spent decades on cost-of-living and working-class budgets. Together they tried to do something new: treat family budget data as econometrics—measurement of economic behavior—rather than only as social description. Bowley called it an attempt “to apply measurement to economic actions.”

Allen and Bowley’s regressions do not discover a single grocery percentage that every household owes. They show that households, grouped by income, settle food at a regular share of total outlay, and that this share is stable enough, within a tier, to be measured.

Treat the household as a spending unit facing prices. Income arrives; alternative uses of that income are compared—food against rent, clothing, fuel, and the rest. That comparison is a market process in miniature: not a committee allocating “needs,” but a sequence of purchases that rations a limited sum across priced goods. Groceries win a large claim where the budget is tight and a smaller claim where the budget is larger. The level of food spending still rises with income; the portion falls. That is the regularity their Engel lines recover.

What the statistics ferret out is that portion. Fit expenditure on food to income (or to total outlay) across budgets in a reasonably homogeneous group and the slope and intercept give the typical grocery claim at each tier. Scatter around the line is taste, habit, and household composition. The line itself is the market settlement: how much of the budget, at that income, households in that class consistently assign to food after comparing it with everything else they might buy.

So the finding is not that groceries are a fixed tax on family life. It is that, tier by tier, the grocery share is an equilibrium result of priced alternatives, and the regression is the instrument that reads that result off the books.