Time vs. State in Insurance: Experimental Evidence from Contract Farming in Kenya

Out­line

In the text­book mod­el of in­sur­ance, in­come is trans­ferred across states of the world, from good states to bad. In prac­tice, how­ev­er, most in­sur­ance prod­ucts also trans­fer in­come across time: the pre­mi­um is paid up­front with cer­tain­ty, and any pay­outs are made in the fu­ture if a bad state oc­curs. As a re­sult, the de­mand for in­sur­ance de­pends not just on risk aver­sion, but also on sev­er­al ad­di­tion­al fac­tors, in­clud­ing liq­uid­i­ty con­straints, in­tertem­po­ral pref­er­ences, and trust. Since these fac­tors can also make it hard­er to smooth con­sump­tion over time, and hence to self-in­sure, charg­ing the pre­mi­um up­front may re­duce de­mand for in­sur­ance pre­cise­ly when the po­ten­tial gains are largest, for ex­am­ple among the poor.

Crop in­sur­ance of­fers large po­ten­tial wel­fare gains

This study pro­vides ex­per­i­men­tal ev­i­dence on the con­se­quences of the trans­fer across time in in­sur­ance, by eval­u­at­ing a crop in­sur­ance prod­uct which elim­i­nates it. Crop in­sur­ance of­fers large po­ten­tial wel­fare gains in de­vel­op­ing coun­tries, as farm­ers face risky in­comes and have in­suf­fi­cient sav­ings to self-in­sure. Yet de­mand for crop in­sur­ance has re­mained per­sis­tent­ly low, in spite of heavy sub­si­dies, prod­uct in­no­va­tion, and mar­ket­ing cam­paigns.

We show that the in­tertem­po­ral trans­fer can help ex­plain low in­sur­ance de­mand, es­pe­cial­ly among the poor, and in a ran­dom­ized con­trol tri­al in Kenya we test a crop in­sur­ance prod­uct which re­moves it. The prod­uct is in­ter­linked with a con­tract farm­ing scheme: as with oth­er in­puts, the buy­er of the crop of­fers the in­sur­ance and deducts the pre­mi­um from farmer rev­enues at har­vest time. The take-up rate is 72%, com­pared to 5% for the stan­dard up­front con­tract, and take-up is high­est among poor­er farm­ers. Ad­di­tion­al ex­per­i­ments and out­comes in­di­cate that liq­uid­i­ty con­straints, present bias, and coun­ter­par­ty risk are all im­por­tant con­straints on the de­mand for stan­dard in­sur­ance. Fi­nal­ly, ev­i­dence from a nat­ur­al ex­per­i­ment in the Unit­ed States, ex­ploit­ing a change in the tim­ing of the pre­mi­um pay­ment for Fed­er­al Crop In­sur­ance, shows that the trans­fer across time also af­fects in­sur­ance adop­tion in de­vel­oped coun­tries.

Re­search Team

Author

Lorenzo Casaburi

Associate Professor of Development Economics

Zurich ZCED

Jack Willis

Co­lum­bia Uni­ver­si­ty