Development Roast Giving international development a proper roasting
If people married each other more randomly, poverty levels would be considerably lower than they are now. If we abandoned all current family arrangements and randomly grouped all Bolivians into new families of 5 persons, poverty levels would fall by about 15 percentage points (from the current level of 55% of all households to about 40% of all households). The Gini coefficient measuring inequality would also fall from about 0.70 to 0.55 (1).
But Bolivians do not mix much in marriage. The correlation between partners’ education levels is extremely high at about 0.77, with no signs of falling (2). For comparison, the corresponding number for Germany is 0.52 and for Britain it is 0.41 (3).
But not all Bolivians are equally restricted in their marriage choices. In the department of Santa Cruz the correlation is only 0.69 while in Potosi it is 0.82, with a corresponding difference in poverty rates (see Figure 1).
|
Figure 1: Relationship between marital sorting and poverty, Bolivian departments, 2005. |
 |
|
Source: Author’s calculation based on MECOVI 2005. |
Why such differences?
My first guess was that in the warm regions, where people are more scantily clad, people marry more based on good looks than on education levels, whereas in the highlands, people are so covered in clothing that looks matter little, and you have to choose based on some other criteria.
But I think a better explanation is probably urbanization rates. In rural areas, young people tend to marry one of the neighbors’ kids, which would likely have pretty much the same level of education. In urban areas, on the other hand, the pool of potential partners is vastly larger, and the likelihood of education differentials is larger. There is certainly a very strong negative correlation (-0.61) between marital sorting and urbanization rates (see Figure 2).
| Figure 2: Relationship between marital sorting and poverty, Bolivian departments, 2005 |
 |
(1) Based on results from a counterfactual microsimulation with data from the 2005 MECOVI household survey.
(2) Author’s calculation based on the 6 latest MECOVI household surveys.
(3) Ermicsh, J., M. Francesconi & T. Siedler (2006) “Intergenerational Mobility and Marital Sorting.” The Economic Journal, 116: 659-679.
Yes, I agree that the results you show are actually true. However I think that the fact that there is high correlation in education shouldn’t be seen as bad for social mobility, as pursuing higher education itself can improve ones economic status.
Years ago I tried to explore the same issue, but instead of using the correlation between control variables like education, I explored the factors that could affect the marriage decision across different ethnicities, under a framework of low social mobility due to discrimination. The hypothesis was to see if factors such better education could “break” the mix marriages barrier, thus increasing social mobility. The finds there, of course, were not very positive, possibly due to sorting in aspects that I couldn’t control for, that fall withing the “good looks vs Education” category.
How has the sorting patterns changed in the last decade. Probably that could be a nice idea to follow and check if the characteristics of the Marriage market in Bolivia has changed, and if it has positively affected Social Mobility.