Thoughts on testing out poverty traps

In our development reading group, a fellow student mentioned a dissertation idea to empirically test poverty traps.  Not being well versed in poverty traps or the literature on them, I got to thinking about how one could empirically test for their existence.  Basically, the best and most versatile method I could think of was the following.  Given a theory on how the poverty trap would occur, find two groups that would be differentially affected by some economic event.  For example, food price changes may affect landowners and laborers with no land of their own differently, as land-owners do not have to buy all of their food on the market, whereas laborers do, so if wages don’t increase laborers are worse off relative than the landowners.  My thinking was a using panel-data differencing, you could get an estimate of how the price change affected the two groups differently in the subsequent year.  Then, if you tested further in the future, you should get some measure of the persistence of the effects (obviously, with the tenuous assumption that other things aren’t confounding that estimate further along), and if the effect as a result of the initial change persists far into the future, it could be seen as evidence that the initial event shifted some members of the more affected group into a “trap” out of which they couldn’t escape.  If the differences don’t persist, that would stand as evidence that either the trap doesn’t exist or wasn’t relevant in that circumstance.

The idea is the following, suppose in some country or region, the price of a staple food crop increases substantially.  If laborers are hit harder by this increase and fall into malnutrition as a result, if  a “malnutrition” poverty trap exists, you should expect more of them to fall into that trap than the landowners who have some insurance given their land-holdings.  If this trap exists, you should expect to see its effects to persist into the future.  If not, you wouldn’t.