Income variability is hugely important. The salient factor about the incomes of poor people isn’t just that they’re low; it’s that they vary seasonally and daily, which makes regular payments for anything difficult. Commitment savings devices are even better worth pursuing in this circumstance, because the combination of poverty and a variable income can be tragic when it comes to large welfare-related expenses, such as school fees or healthcare. Pursuing “opt-in” savings strategies (such as paying people to open accounts) is a good option. So is designing loan repayment systems with A) realistic assumptions of the timeframe of income generation, and B) flexible payment options.
In a sense, “emergency” microloans are just a more expensive form of saving for the poor. Rather than saving in advance and earning interest, they have to save after the fact and pay interest. The incentives of the poor (not reducing current consumption for future expenditures) and the banks (taking in interest instead of paying it out) are currently aligned on this issue, but it results in a suboptimal equilibrium. Or, actually, I wonder how these incentives might vary between formal banks, and microfinance organizations (often non-profits) which may acquire their capital from different sources.