I just came across a new World Bank paper by Brian Dillon and Chris Barrett about agricultural factor markets in Africa. Dillon and Barrett have written an "old school" development paper, meaning it reminds me of papers written in the 1980's and 1990's. They use observational data, and appeal to theory to motivate their empirical identification in an attempt to answer a classic development question. (This is going to be really distressing nomenclature to the development economists from that era, who were "new-school" when the field moved beyond the "old school" work by Lewis, Schultz, and others from the 1960's and 1970's. Sucks to grow up, I guess.)

They test for "household separation", which means that the consumption decisions of a household (i.e. how much or what to eat) are completely divorced from the production decisions (i.e. how much or what to plant). If there are complete markets, then household separation is expected to hold. If you live in a modern Western country then you face effectively complete markets, and your decision about what to consume (a new TV!) is completely unaffected by what you do at work (accounts payable!). All that matters is that you make money producing, and then you spend that money on goods and services.

If markets are missing completely, or so unreliable as to effectively be missing, then household separation fails. The extreme case is easiest to think of. If a household is completely autarkic, and can trade with no one else, then it can only consume what it produces. The two decisions are inseparable. If they want a new TV, then they'd better have a source of rare earth elements in their back yard and a passion for soldering.

The importance of knowing if household separation holds or not is that it tells us something fundamentally important about why a developing area is poor. If separation fails because markets fail to exist, then this is like saying there is a lot of latent economic activity that is not being realized. Household A and household B would be better off if one specialized in raising goats and the other in rice, but because the markets for goats and rice don't exist they each have to raise both. It's a basic comparative advantage argument; households can be made better off (just like we argue countries can) by trading with each other. If you want to enact policies or take action to assist these households, then promoting functioning markets in outputs and inputs is the way to go. How do you do that? It could be by making institutional changes (making property rights clear so land can be easily exchanged), or providing insurance (so people can take the risk of producing for the market), or it might be as simple as paving a road (so that transport costs are low enough to trade with the next town over). It could be by promoting better information flows; one of my favorite development papers is by Robert Jensen about the positive effect of mobile phone introduction on the efficiency of the fish market in Kerala, India.

If separation holds, and by implication markets are functioning relatively well, then the implications for development are different. Households are able to buy and sell, so they are probably taking advantage of nearly all the gains from trade available. They can likely benefit more from increased investments in factors of production than households without access to markets. Here's what I mean. When markets are (reasonably close to) complete, then endowing a household with a tractor and the technical training to keep it running is a huge change in their productivity. They can use it on their own farm and they can rent it out to others. It may make sense for them to become a full-time tractor operator in their village because with complete markets they know they can buy what they need using the proceeds of the tractor business.

In a non-separating economy without markets, giving a family a tractor and technical training is kind of a waste. If they're well off for a developing country they've got 2 hectares of land (about 5 acres or just under 4 football fields). Yes, the tractor can alleviate some of the workload, but it's overkill. You'd spend more than half your time just turning the tractor around trying to plow that small of a space. Moreover, the tractor would sit idle for the vast majority of the year. Without markets, most investments and improvements in technology are not worth it.

Okay, back to Dillon and Barrett. They use data from the Living Standard Measurement Surveys from 5 Sub-Saharan African countries to test for household separation. The logic of the test comes from a classic paper by Benjamin (1992). If markets are functioning efficiently, then the characteristics of your household (age, gender, education, number of kids) should be unrelated to the input usage on your farm. The production decision (I need three workers to plant rice) is separated from your household characteristics (I have five workers in my household). Those extra workers in your household can go work on someone else's farm if markets exist. (And if you don't have enough household labor, you can hire in laborers).

What they find is that across all five countries (Ethiopia, Malawi, Niger, Tanzania, Uganda) separation fails, meaning that there are significant failures in factor and/or output markets. In every country, the size of the household is always significantly related to the amount of labor used on that household's land. It doesn't matter whether the household is led by a male or a female, or where the household is located within the country. The failure of separation appears to be a general failure. One thing to note is that this doesn't mean households fail to participate in markets at all; Dillon and Barrett find that nearly all households do make some transactions in labor and land markets. It is likely that the actual transactions costs (fixed costs, time costs, etc..) of participating in the market are so high that people don't undertake all the trades that they would otherwise make.

So what does this mean? It means that development assistance is likely to be most effective if it promotes making markets more efficient; perhaps through encouraging better information flows like in the Indian fish-market example. Pure investment strategies (let's give everyone a bag of fertilizer!) are unlikely to be as effective without the markets in place that allow people to take advantage of that investment.

Research like this is particularly valuable to people like me who want to study growth and development from a macro perspective. It reminds us (beats us over the head with?) that "developing countries" and "developing economies" are not the same thing. Market failures mean that developing countries are really a collection of myriad small economies, and therefore we need to be careful in thinking about things at too aggregate a level. This is also an example of where I think the "institutions" literature could really add value. Can we provide better theories or models of what precisely it means for markets to "fail"? What particular institutional details are important for markets to work efficiently?