arXiv Science⌕ Search

arXiv subjects

Xinta Yang

Publications and source records attributed to Xinta Yang.

2 recordsLinked to original sources

Optimal Capital Injections in a Microlending Group under External Shocks

Motivated by static models of group default in microlending, we study a dynamic problem in which the members of a lending group are exposed to external shocks, such as droughts or floods, and the manager of a common emergency fund decides at each shock how much capital to inject into the members who were hit. Shocks arrive at random times, the set of members that a shock hits may depend on which members the previous shock hit, and the losses may spill over to members who were not hit directly. Between shocks, the members' wealth follows a linear stochastic differential equation in which members support each other, and the fund grows with interest and member contributions up to a cap. The cost balances the members' shortfall below a poverty line, a welfare measure that contains a standard poverty measure as a special case, against the administrative cost of the injections and the benefit of keeping money in the fund. We formulate the problem as a Markov decision process embedded at the shock times and prove that the embedded process is Markov with a weakly continuous transition kernel. When the discount rate exceeds a growth rate determined by the wealth dynamics, we show that the expected total cost is finite under every policy, that the finite-horizon optimal values converge to the infinite-horizon one at an explicit rate, and that there exists a stationary optimal injection policy, which can be chosen independent of calendar time.

math.OC↗

Group Survival Probability under Contagion in Microlending

In the context of micro-finance, a group of individuals undertake business projects that may interfere with one another. A contagious default happens if one person's project failure leads to the default of another group member. In this paper, we apply a probabilistic approach to analyze the impact of such contagion among investment group members. Firstly, a general formula is provided to compute the group survival probability with the presence of contagion effect. Then, special cases of this probability model are examined in detail. In particular, we show that if the investment group is homogeneous, defined in the paper, then including more members into the group will eventually lead to default with probability 1. This differs from the non-contagious scenario, where the default probability decreases monotonically with respect to the group size. Afterwards, we provide an upper bound of the optimal group size under the homogeneous setup; so, one can run a linear search within finite time to locate this optimizer.

q-fin.MF↗