I posted earlier this week about speculation on food markets by linking to a post on another blog which discussed how speculation and arbitrage may help enhance food security.
In response to my post on speculation, Ed Carr — whose new book just came out this week; I am looking forward to blogging about it in the next few weeks — wrote a post in which he discussed the convergence between the quantitative findings discussed in my paper with Chris Barrett and David Just on food price volatility and his own qualitative findings.
Commenting the empirical finding that price volatility hurts the wealthiest 40 percent of households but benefits the poorest of the poor in our paper on the welfare impacts of food price volatility, Ed writes:
“I would bet my house that the upper 40 percent of the population is that segment of the population living in urban areas and/or wealthy enough to be purchasing large amounts of processed food.”
To which I say, tongue in cheek: “Hand over the keys to my new vacation home in South Carolina, Ed.”
Joking apart, the upper 40 percent of the income distribution in our data are all rural households (we use four rounds of the publicly available Ethiopian Rural Household Survey data), and they are still pretty poor. Contrary to Ed’s conjecture, these households tend to be hurt by price volatility because they are producers and therefore net sellers of most of (if not all) the seven commodities retained for analysis (i.e., coffee, maize, beans, wheat, teff, barley, sorghum).
This is a counterintuitive result, but it makes sense, both theoretically and intuitively. Sandmo’s (1971) classic results states that when producers face uncertainty over the price at which they will be able to sell their output once it is produced, they will underproduce in an effort to hedge against price uncertainty.
But price uncertainty is precisely what we mean by price volatility, and I think Ed may have mistaken rising food prices for food price volatility since his reasoning corresponds to what we know about the welfare impacts of rising food prices. I am putting the more detailed (and lengthy) argument under the fold.
I have written three posts recently about the twin issues of rising food prices, or increases in the mean of the food price distribution, and food price volatility, or the variance of the food price distribution. These are two very different things.
Rising Food Prices…
By “rising food prices,” on the one hand, what we mean is that the price people pay for their food is increasing. This tends to make net sellers of food — producers, or households with enough land to produce an agricultural surplus that they sell at market — better off, but it makes net buyers of food — consumers, or households who do not have enough land to produce an agricultural surplus that they can sell at market — worse off. Economists have known this since Deaton’s (1989) seminal article on the topic.
For example, the price of gasoline in the Raleigh-Durham-Chapel Hill area has been equal to about $3 per gallon in the last 24 hours according to this source. If the price of gasoline increased to $3.25 tomorrow morning, we would talk of rising gasoline prices.
… vs. Food Price Volatility
By “food price volatility,” on the other hand, what we mean is the uncertainty over future price levels. To be sure, people have expectations as to what the future price level will be, but they cannot perfectly forecast future prices, so realized prices can be above or below expected price levels.
To keep with the gasoline example, if the price of gasoline were expected to lie in the $2.75 to $3.25 interval today but in the $2.60 to $3.40 interval tomorrow — notice how I am keeping the mean constant at $3 per gallon, this is actually quite important — we would be talking about an increase in gasoline price volatility. That is, price volatility is related to the degree of uncertainty surrounding future prices, holding the expected price level constant. So to reiterate, rising food prices have to do with the mean food price level, whereas food price volatility has to do with food price variance.
What my coauthors and I found in our paper on price volatility was that in rural Ethiopia, food price volatility makes poor households better off and wealthier households worse off, everything else equal.
Remember how I wrote above that keeping the mean price level constant is important? Statistically speaking, what my coauthors and I have done in our paper was to keep the level of food prices constant (this is what I meant by “everything else equal above”; the technical term for what we study is “mean-preserving spreads” in the joint distribution of seven food prices) while studying the effects of varying degrees of food price volatility on households.
The way we did this is fairly involved, both theoretically and empirically, but the end result is that food price volatility — which has mistakenly been blamed by many people for making the poor worse off in developing countries — actually has a counterintuitive impact on the welfare of rural households. A (very) simplified argument for this counterintuitive result can be found here.