Passage
Two neighbouring provinces held farmland under different rules. In one, holdings were registered to individual households; in the other, a village council reallocated strips among households at intervals. Because the two systems met along an administrative line drawn long before either was studied, economists have compared villages within a short distance on each side, reasoning that soil, rainfall and distance to market change continuously across such a line while the tenure rule changes abruptly at it. Any discontinuity in yields or in investment at the boundary can then be credited to tenure. The design is attractive because no researcher assigned the rule: it appears to supply an experiment where none was run.
The reasoning rests on a claim about the boundary that is seldom defended — that the line marks a change in nothing else. Boundaries are drawn rather than discovered. Where one follows a river or a ridge, the feature that recommended it to a surveyor is also a break in drainage, in soil, and in the ease of hauling a cart to market. Where a line was settled by negotiation between two administrations, the negotiating itself responded to the land: the districts worth keeping were kept, and which side a village fell on records somebody's estimate of its value.
A second difficulty accumulates with time. Households can move, and the rule that suits a family with many sons is not the rule that suits a family with capital and few. Over several decades the villages nearest the line are those for which crossing costs least, so the comparison drifts from a comparison of rules toward a comparison of the people each rule attracted. The usual precaution — narrowing the band until only the closest villages are compared — makes this worse rather than better, since it selects the sample most exposed to the sorting.
None of this convicts the method. But the credibility of any particular application rests on evidence from outside the design: a boundary whose position can be shown to have been fixed for reasons unrelated to farming, and settlement records showing that movement was rare. Studies that supply such evidence earn the inference. Studies that name the design and proceed as though the name had supplied it are borrowing a credibility they have not established.
