Formal insurance and informal reciprocity mechanisms: experimental evidence from Ethiopia.

2024-12-01

2024-12-20

The variable used to measure the difference between treatment and control condition,
that is the availability of formal insurance that can be purchased by the players to partially
cover from the random shock, is the monetary transfer from the provider(s) to the recipient in each turn, in each group. This is the main outcome of interest.

In every group, in each round, there are two providers and one recipient. We
therefore have two transfer variables, s1 and s2, one for each provider in every group.

A second outcome of the experiment, which is going
to be used for additional analysis, is the difference in each round, in each group, between the two monetary
transfers s1 and s2. This is denoted with the letter n.

We design a lab in the field game experiment randomly selecting individuals who do not have
any insurance to cover risks.
The game is designed as a repeated dictator game where a negative income shock is induced,
and investigates the extent to which people increase or reduce their monetary transfers in the
presence of an option to buy insurance. There is a treatment group and a control group. The
selected individuals in both groups are randomly matched in groups of 3 people and play
repeatedly at least for 30 periods (the number of 30 rounds follows a previous paper on the
topic, Lin, Liu, and Meng, 2014). To test the consistency of a rational risk-sharing mechanism,
the information about the duration of the game is not communicated to the players and after
the minimum number of rounds (30), the dictator game just ends. In each round, the players in
the triads are given the same income, and they are equally likely to be hit by a negative
idiosyncratic income shock with probability equal to 1/3. According to this, two players are assigned to
the role of dictators, that is help providers (P), and the player hit by the negative shock is the
recipient (R). The providers can decide to transfer part of their incomes to the recipient, that is
to the player hit by the negative income shock. Both the income assigned, and the income
shock are constant across rounds, and no formal insurance is available for the control group.
In the treatment case, a formal insurance is available for purchase at a price p since the first
round. The insurance covers 50% of the income loss so that there is space left for transfers
exchanging between players and subjects take their purchase decision at the beginning of each
round, before the realization of the random idiosyncratic income shock and their decision is
known to their triad companions.
An important aspect of the game is represented by the identity of the players, which is hidden
throughout the game, and the composition of the triads in each round. We keep the triads
constant (meaning that we do not create randomly new triads each round) so that the risk-sharing mechanism can consolidate based on the risk and other-regarding preferences of the
triad members. The personal identity remains always hidden and the players are just told at the
beginning of the game that they play every round with the same other two players. At the end
of every round, the players can observe the support choices made by the providers and therefore
they can develop their own knowledge about the other players altruism.

Not available

Randomization done in office by a computer.

Individual

No

300 Individuals (estimate)

6000 obs. As the participants play the game in group of three people we have: 100 (n of triads) * 30 (n of rounds) * 2 (in every round in every group we observe 2 transfer choices) = 6000

50 triads in control, 50 triads in treatment.