This Vox article on Alaska’s oil dividend, roughly equivalent to UBI, and its effect on human fertility, also details how social science researchers construct comparison entities when none is available:
Also in 2018, UChicago’s Damon Jones and UPenn’s Ioana Marinescu found that the dividend does not deter people from working, and actually increases part-time work. Jones and Marinescu employed what’s known in the social sciences as a “synthetic control” method. Basically, they combine a number of other states whose patterns of employment, part-time work, and related statistics roughly match Alaska’s in the years before the policy was enacted. None of the states alone is a good comparison, but if you combine them carefully, you can come up with a “synthetic Alaska” for comparison.
The new fertility paper, from Yonzan, Kelly, and Timilsina, also uses a synthetic control design. Because they’re interested in fertility, not employment, they rely on states’ fertility rates, average time between births, and abortion rates to construct synthetic controls whose trends matched those of Alaska before 1982.
Fascinating and clever. Of course …
This is just one study, and there are some reasons for skepticism. Synthetic control studies are useful, but there’s always a risk that the other states that make up the “control” differed from Alaska in ways other than not having the dividend program. For the fertility rate comparison for 15- to 44-year-olds, the synthetic control is a weighted average of mostly Wyoming, a bit of Hawaii, and a very small bit of Washington, DC; these are all obviously quite different places from Alaska in ways that might influence fertility rates. That only matters for the analysis if they started to differ increasingly after 1982 but not before, but it’s hard to rule out that possibility.
Very interesting idea, and the caveats still apply for non-synthetic controls as well.
