Paul Clist (UEA)

"Nature's Frames, Reference Lotteries and Truly Risky Choice: Evidence from a Ugandan Field Lab"

joint with Arjan Verschoor (UEA) and Ben D'Exelle (UEA)

Abstract

Investment and insurance decisions are conceptually identical; both are choices between a riskier option that has a higher expected value and a less risky alternative that has a lower expected value. Crucially however, they differ in their default. This difference does not stem from an artificial construct but rather nature's own framing. By modelling the default as a reference lottery, we can apply Koszegi & Rabin's (2007) framework to the new domain of truly risky choice (where both prospect(s) and reference are non-degenerate lotteries). Using this theoretical framework we test the effect of altering the reference lottery, mimicking nature. We present a single risky choice problem in one of three ways: where the initial starting position of coins are relatively risky (the insurance treatment), relatively safe (the investment treatment) or neutral. We find significant treatment effects, despite the subtle difference in framing. In the second round of the experiment information on the socially most popular option is given, which we model as a competing reference. We find this has a large effect on behaviour, trumping the initial framing effect. Results from a Maximum Likelihood model are able to explain behaviour in both rounds, with surprisingly consistent parameter estimates recovered from apparently inconsistent behaviour. The experiment was conducted using 292 randomly selected participants from a rural farming community in Eastern Uganda, and results are discussed in the context of two puzzles in development economics: underinsurance and underinvestment in such communities. Recent successes in increasing low levels of insurance/investment are also discussed.