Tuesday, July 3, 2007

Transaction costs



A future client at our last village bank meeting in Nicaragua.

In grad school, we read Robert Chambers' “Voices of the Poor," which book and its sequels later sparked a World Bank initiative by the same name. Chambers’s point was that even in countries that are extensively studied and researched, the very poorest people tend to be excluded from study and consultation, and rarely heard from by outsiders. Why is this? Well, imagine you are a study team with a limited amount of time and money to spend in a country. You will probably focus your work in large cities; if you do get out of the city, you’ll go to areas that are reachable by paved roads; if you get off the paved road, you’ll go to areas where the roads are passable in rainy season; etc. The poorest of the poor tend to live in areas with much worse transport infrastructure. That’s part of what makes them poor. So reaching one such person for a survey or consultation requires ten times the effort of reaching someone else with a similar demographic profile, but living near a paved road.

The tradeoff between a large sample and a representative sample became very clear to us yesterday, our last day of interviews in Nicaragua.

Although we had thought we’d be going to bank meetings in the city, there was a last-minute change of plans and we ended up going out to rural areas. We got to the office at 7:30, but didn’t end up leaving until 10:00, when the FINCA driver became available. Because the credit officers had left before us, the driver dropped us off in the truck, with the branch coordinator riding along to help find the way.



City girl Lauren does not know quite what to make of this strange, green, growing stuff...

First, we drove about half an hour along the paved highway. Then it was time to turn off onto a dirt road and follow the “directions” to the client’s house, where we promptly got lost. As I mentioned earlier, the “two blocks below Pulperia Martinez, turn left at the mango tree, then ask around” method doesn’t work much better for Nicaraguans than it does for us. First we had to find the pulperia - the first person we asked said (with total certainty in his face), “Ah, yes, two blocks to the right.” We drove - no pulperia. The next person we asked, equally certain: “Ah, yes, two blocks to the left.” We finally found the pulperia and drove, twisting and turning and asking directions every few meters, up a road composed of thick black sand to a small house at the end of the road. “FINCA? What finca? You mean our farm?” Eventually we figured out that we were at the house of the wrong Dona Mercedes Gallo, and headed back, where the truck got stuck in that thick black sand for about 20 minutes before the driver finally figured out how to engage the 4WD. (Just in case you were wondering: Sand acts just like snow, and “gun the engine and spin the wheels real fast” is another approach that doesn’t work any better in Nicaragua.) Finally, we found our first bank meeting. Because most of the clients lived far away, and we had no transport, we could only interview those clients who actually showed up to the meeting: Lauren did 1 interview, I did 2. After a few souvenir photos of New York City Girl Meets Pig, it was off to get lost again, this time waiting at the empalme of something-or-other for a while before giving up and hiking along the highway to the pueblo entrance. After hacking our way through a forest of banana trees dotted with isolated houses to the second bank meeting, we had time for just 1 interview each before the credit officers left for another isolated meeting and we had to return to Leon.

5 interviews, 7 hours of 2 people’s time. Average time spent per 1/2-hour interview: 4.4 person-hours.

There’s just no way to contact poor rural people efficiently. It’s a huge investment of time and energy. Which is why it’s so important that when it does happen, the survey should be extremely well-designed, and some sort of education or outreach should take place along with it. It seems such a huge waste of effort to spend so long reaching these people, only to ask them 100 questions of dubious statistical value. We are all trying to at least act as FINCA “goodwill ambassadors,” and my personal goal has been to make sure every client I talked to understood the concept of an interest rate.

Extrapolating wildly from a single experience like the good statistician I am, I´d guess that the actual work of data collection occupies a relatively small proportion of most researchers' time in rural areas, because most of the work is in getting there in the first place. Given that, wouldn't it make more sense to always do data collection as part of some other project? The UN, World Bank, etc. invest so much money in data-gathering every year. Imagine if, oh, basic sanitation training went along with every such effort. Perhaps a naive thought - but at the moment it makes sense to me.

No comments: