Older readers may still remember the Prebisch-Singer thesis: the proposition that developing countries suffered from a "secular" deterioration in their terms of trade vis-à-vis industrial countries, because commodity prices tended to exhibit a long-run decline relative to those of manufactures…. The argument implied that poor countries, and the poor farmers that constituted the bulk of their population, were victims of sustained declines in the price of food, and other primary commodities, of which they were net producers.
Much ink has since been spent discussing the statistical validity of that hypothesis, and the balance of opinion now seems to be predominantly against it. Even prior to the recent food price increases , there was little or no robust evidence of a sustained decline in the commodity terms of trade. Be that as it may, though, isn't it surprising that these days we are all so worried about the damaging effects of rising food prices on the poor? Aren't many of the poor people in poor countries actually farmers, who produce and sell food? Don't rising food prices translate into positive terms of trade shocks for net exporting countries (like most in South America ), and thus to higher aggregate incomes? Why are higher food prices a bad thing for developing countries?
Of course, things are not that simple. Many poor farmers in developing countries are net buyers of food, as are the increasing numbers of poor people who live and work in urban areas. All of these people are likely to suffer from rising food prices, and these effects have been documented in a number of countries: Thailand, Madagascar, and Indonesia to name a few.
Almost without exception, however, empirical analysis of the welfare effect of rising food prices has relied on an application of basic consumer theory to the so-called farm-household model. In many developing countries, most agricultural activity is undertaken by peasant households, who both produce and consume different foodstuffs. Multiplying net budget shares of each such food item by its proportional price change yields a convenient first-order approximation to the welfare effect of changing prices. It is also possible to estimate second-order effects, that account for the fact that consumers tend to substitute away from (and producers towards) more expensive foods.
Powerful though this approach has been, it has some rather obvious limitations. A "serious deficiency", writes Angus Deaton (1997), is "its neglect of repercussions in the labor market. Changes in the price of the basic staple will affect both supply and demand for labor, and these effects can cause first-order modifications to the results". The farm-household approach would miss, for example, any increases in wages or enterprise profits generated by rising prices in commercial agriculture, which is now dominant in countries such as Argentina , Brazil  or Uruguay .
To investigate the potential importance of this additional channel (and at the behest of the Brazil Country Unit), Anna Fruttero, Phillippe Leite, Leonardo Lucchetti and I recently looked at the effects of different pass-through scenarios from food prices to agricultural wages in Brazil,  during the food price increases of 2007-2008. Combining spatially disaggregated consumer price data from Brazil's Census Bureau (IBGE), with detailed consumption, occupation and income information from two household surveys, we computed a more comprehensive measure of the impact of price changes on household well-being. In addition to the standard "expenditure" effect (computed as a compensating variation), we estimated "market and transfer income" effects, relying both on agricultural wage data and on transfer receipt information.
The results are interesting: because Brazil's population is 80% urban, rising food prices did lead to a welfare loss on average, and to some increase in poverty – although this assessment still ignores other general equilibrium or dynamic effects.
Poorest People May Gain from Higher Food Prices
By our estimates, food prices alone contributed to an increase in the incidence of extreme poverty from 11% to 12.3% (under a competitive labor market assumption). If the income effects are ignored, however, and one uses the standard farm-household model that dominates the literature, the increase would have been to 13.5% - almost double! In fact, because some of the poorest people even in such a highly urbanized country as Brazil still work in agriculture – but most as wage earners, rather than family farmers – accounting for likely labor market effects dramatically changes the distributional profile of the impacts: instead of the consistently regressive pattern associated with the expenditure effect, the overall "net" effect is U-shaped. Taken as a group, the poorest are not those who suffer most from rising food prices. (Take a look at the figure below.)
Price Increase Incidence Curve – Net Effect (All Brazil – Alpha = 1)
Source: Ferreira, Fruttero, Leite and Lucchetti (2011): "Rising Food Prices and Household Welfare: Evidence from Brazil in 2008". World Bank PRWP 5652.
I think we learned three main things from this exercise.
First, that rising food prices can indeed be poverty-increasing, even in a large agricultural producer such as a Brazil. The terms of trade gain are real, of course, but the resulting income gains are unevenly distributed, even under generous assumptions about how agricultural labor markets work.
Second, ignoring the labor market effects of rising food prices - which may have been acceptable for countries where agriculture consisted predominantly of family farming - simply will not do in more modern, wage-based agricultural sectors, such as Brazil's.
And finally, precisely because these labor market effects are important, it would be important to go beyond the very simple analysis we have attempted, and to estimate them more carefully. If we do not, we will continue to look for welfare effects of rising food prices where the old family-farm light is. In many places, that is increasingly not where the real action is.