How do you measure guilt? There is no market price for it, and most of the time it has no measurable expression. One would need some rather particular circumstances to find a way to measure guilt. Hongbin Li, Mark Rosenzweig and Junsen Zhang found it.
During the Cultural Revolution in China, many urban families had to send some of their children to work among farmers. Assignments were to a large extend random, and parents had to choose which child to send. As this send-off was viewed as an unjustified punishment, many parents felt guilt and tried to compensate later in life with transfers, as a marriage gift or other. To get a clean data set, Li, Rosenzweig and Zhang set out to recruit twins and compare those who were separated during this period, as presumably everything else prior should have been identical. And it turns out that those that were sent off received substantially more from their parents, even when they were subsequently better off than their siblings. And this result is robust to all sorts of controls, for example party membership (there was indoctrination on farms, and party members were more likely to gain favors in the system, which parents may want to buy into).
I hope they will be using this data set for other studies. Otherwise, this would have been a tremendous effort for only a proof of concept. Because I see little use for the result beyond that.