My Experiment: A Way Forward for Disparate Disciplines
To tie this theoretical discussion of the differences between Economics and Anthropology back to my personal experience in Uganda, I've conducted a small “experiment.” I borrowed a data set collected in 2008 by the Ford Family Program at the University of Notre Dame to do a SMALL econometric analysis involving some variables I find interesting. I compared the results of this econometric analysis to my personal experience in Uganda. What makes sense in the regressions, and what doesn't? How might my personal experiences suggest alternative explanations? Do I experience any of the limitations of econometric methodologies, such as data sets which don't contain local details? I've put it to the test to experience how econometric and ethnographic approaches differ. Through this, I've gained insight into how Anthropologists and Economists can work together to design participatory programs!
Wage Employment
The first variable I was interested in learning about was the likelihood of someone having a form of non-farm employment. My research on SILC groups in Uganda made me aware that many rural villagers struggle to smooth their income flow, and consequently, consumption, throughout the year because they can only sell their produce twice a year. Many villagers described to me the difficulties they face with cash flow, but one woman from Bukibira put it particularly aptly, saying “it's just difficult to predict or know how much money to save, because I only get money very infrequently, and I don't usually know how quickly to spend it. I'm very careful, but my husband gets upset at me when I have money to spend and do not want to because I want to save for the future.” One strategy villagers can use to smooth their income is to engage in some form of wage employment, and I wanted to see what the determinants of this type of behavior are.
I did several specifications of this model, but I finally settled on this one as most informative.
These are the results I got using the computer program STATA.
Non-farm employment= .3 +.4male +.09education-.04educationxexperience -experience-experience^2-.3average harvest-.2 plot size-.2 # of children
In this model, the terms that are statistically significant are in bold.
How this model agrees with my personal experience:
- I agree that education, all other things equal, would make someone in the Nnindye area more likely to have wage employment. Because opportunities for wage employment are fairly rare around Nnindye, store owners or restaurants have their pick of employees to hire. I interviewed several shop owners who said it was very easy to hire workers. Because demand for these jobs outstrips supply, it makes sense to me that more educated workers hold these jobs.
- I agree that having more experience, or number of years in the work force, decreases the likelihood that someone has wage employment. Almost all of the wage laborers I saw were under the age of 30. There is a cultural expectation that working for somebody else is for young people, and that elders are supposed to be self-sufficient if still working age, or that they should have sons take care of them if extremely elderly.
- It also makes sense to me that the likelihood of having non-farm employment decreases with average size of harvest. The people who most need wage employment are the ones that, often because of poor land quality, cannot earn a good income from farming. As someone's plot size increases, they are less likely to have wage employment, because they would be able to have a more profitable farm.
Flaws of this analysis
Although this econometric study did reinforce my personal experience, the only problem I have is my inability to extend the study to include the more detailed observations I took from the field. My research leads me to believe that people who have business skills, such as experience accounting, interacting with customers, marketing, or investing money, are much more likely to be hired as workers. Several shop owners I interviewed said that because they had a wide selection of potential employees to choose from, they looked for people with experience working in or managing a business.
However, the data set does not include any information on business skills. This supports my earlier critique that econometric techniques sometimes aren't sensitive to local factors. It is very difficult to design a survey that is detailed enough to contain all these local factors that would possibly be interesting to include in an econometric analysis. If the data on these business skills isn't collected by the survey, there's not much further I can go in using my local experience to hone econometric methods.
Trust in Religious Institutions
Trust in religious officials = .1 -.3male+.08education-.2experience -.09experience^2+.07 #children+.05education*experience+.1 food insecure+.3trust in government
What my experience supports:
- I agree that males are less likely to trust religious institutions, because many religious leaders speak out against domestic violence, something that is, unfortunately, deeply imbedded in Ugandan culture and something that I witnessed on a number of occasions in Nnindye. One woman told me her husband does not go to Church because he was reprimanded by a priest for domestic violence.
- People with a greater number of children, and who are “food insecure” would be more likely to trust religious officials, because these people are probably poorer. Maybe extremely poor people need to have more faith, or get more outreach from religious institutions
- My intuition tells me that more educated people are asked to participate more frequently in religious ceremonies. At the masses I attended in Nnindye, all of the people participating in the service, according to the priest, were literate
- People with a greater trust in government may be, well, more trusting people overall. I can't think of a great reason why trust in government would explain trust in religious insitutions other than that this variable is capturing a person's level of trust in higher institutions
Just like the previous variable, my local knowledge tells me that a variable not contained in the data set is important. A priest I spoke with told me that he rarely gets to visit parishioners living far from the church. I have a strong feeling that the level of trust a villager has in the church has something to do with how close they live to the church, or how frequent their contact with religious ministers is. Unfortunately, this is not in the data set, so the econometric methodology is again limited.
Conclusion:
My earlier critiques of econometric methodologies in the context of participatory development largely boil down to inattention to the local. I said that Econometric methodologies, and more specifically, Randomized Control Trials, should be taken with a grain of salt because they cannot say anything interesting about individual or village-level specifics. Designers of participatory programs need this local perspective, so I concluded that economic methods couldn't inform participatory programs.
My experiment, however, softens this critique and suggests a way forward. I saw while conducting this experiment that local knowledge could be very helpful in steering econometric methodologies by helping to structure the economic models used. A good model might be able to incorporate finer details if designed carefully! But the data sets need to include these details, which is where I hit a wall. Although both disciplines will continue to have their respective strengths and weaknesses, I do have hope that economists and anthropologists can collaborate even in designing participatory programs, bringing local perspective to economics and expanding the scope of anthropology. The queen bee of the social sciences needn't stand alone! That's a happy thought if there ever was one!!!!