Gan and Vernon’s comment does not resolve the puzzle that we orig-inally posed. Their description of the puzzle is unclear, so we start by restating it. The essence of the matter is this. Imagine two households, one of which is larger than the other, for example, containing the same age and sex composition of people, but with twice as many of everyone. Imagine too that both households have the same level of per capita total house-hold expenditure so that, in the example, the household with twice as many people spends twice as much in total. If there are economies of scale, the second household is better off. While it is possible for it to maintain exactly the same expenditure pattern as the smaller household, with everyone having the same of everything, it can also rearrange its purchases to take advantage of the differential economies of scale in different goods. Because food is a normal good, we would expect the larger household to spend more per capita on food. This is especially so in poor countries, where there are few substitutes for food, so that there is limited opportunity for substituting away from food toward goods with greater economies of scale. The evidence contradicts this prediction. We looked at household survey data from the United States,

Household scale economics are plausibly attributed to shared household public goods that make larger households better off at the same level of per capita resources. larger households should therefore have higher per capita consumption of private goods, such as food, provided that they do not substitute too much toward the effectively cheaper public goods. The evidence shows exactly the opposite. Data from rich and poor countries indicate that, at constant per capita total expenditure, the per capita demand for food decreases with household size and that it does so most in the poorest countries, where substitution should be the least.

The authors investigate nutrition and expenditure in rural Maharashtra in India. They estimate that the elasticity of calorie consumption with respect to total expenditure is 0.3-0.5, a range that is in accord with conventional wisdom. The elasticity declines only slowly with levels of living and is far from the value of zero suggested by a recent revisionist literature. In these Indian data, the calories necessary for a day's activity cost less than 5 percent of the daily wage, which makes it implausible that income is constrained by nutrition rather than the other way around.

The permanent income hypothesis implies that, for any cohort of people, inequality in consumption and income should grow with age, a prediction that is here confirmed using data from eleven years of household survey data from the United States, twenty-two years from Great Britain, and fourteen years from Taiwan. In the permanent income hypothesis, the increase in inequality reflects the cumulative effect of luck on consumption. Other models of intertemporal choice--such as those with strong precautionary motives or liquidity constraints--can limit or even prevent the spread of inequality, as can insurance arrangements that share risk across individuals.

The theoretical basis for measuring child costs is discussed, and detailed consideration is given to two straightforward procedures for calculation, Engel's food share method and Rothbarth's adult good method. Each of these methods embodies different definitions of child costs so that the same empirical evidence can generate quite different estimates depending on the method used. It is shown that true costs are generally overstated by Engel's method and understated by Rothbarth's procedure, although the latter, unlike the former, can provide a sensible starting point for cost measurement. Our estimates from Sri Lankan and Indonesian data suggest that children cost their parents about 30-40 percent of what they spend on themselves.

By buying cheap and selling dear, risk-neutral commodity speculators can smooth commodity prices and induce serial dependence in price even when none would exist under a simple process of supply and demand. Commodity prices are variable and strongly positively correlated from one year to the next. The variability is often explained by supply factors and the autocorrelation by the activities of speculators. The authors show that this explanation is not consistent with the evidence. Speculation can substantially increase autocorrelation for prices that are weakly autocorrelated in its absence but not to the high levels that are observed in the data.