Friedrich Hayek observed decades ago that predictability of laws and regulations was more important than their perfection. The way this is often interpreted (by conservatives and libertarians) is that part of the evaluation process of any new law or regulation should be to test how much it improves predictability versus how much absolute improvement it is expected to make. (Since a new law by definition introduces at least some element of unpredictability, it starts out in the hole at least on this metric.) While this is a fair interpretation, there are others.
In some cases, laws themselves balance predictability and perfection, and then, too, the former exceeds the latter in importance. A recent article from the Hoover Institute is illustrative. Lueck and Libecap examine different ways of defining property: the old but common way by "metes and bounds" (in which natural landmarks and neighboring property lines are used to define boundaries) and the new one promulgated by the U.S. Congress when setting up rules for development in the West (in which boundaries were defined by strict square grids aligned with lines of latitude and longitude).
A natural experiment turns out to be possible by looking at bordering lands, one area using the rectangular system and the other using metes and bounds, and the results are startling:
We found that, controlling for land and owner characteristics, land values were around 25 percent higher under the rectangular system than under metes and bounds in 1850 and 1860. Further, extending the analysis for 100 years revealed that these land value differences persisted!
To learn why that might be so, we turned first to data on land disputes from Ohio court records. Over the entire nineteenth century, we found that parcels in the VMD had 18 times more land boundary disputes than the rest of Ohio combined. Indeed, the history of the VMD is one of ongoing land conflicts. We then turned to land market activity. Land transactions in the middle of the nineteenth century were about 75 percent greater in the counties adjacent to the VMD than within it.
Something to think about when someone proposes a new rule to take care of some tiny edge case in the law. Before deciding whether it's worth it, we might consider the harm to predictability being done.