Grameen Loan Recovery Indicators Recovering

In February I blogged about how indicators of loan repayment at the Grameen Bank were deteriorating. Grameen publishes indicators monthly and I have copied each new report into my spreadsheet, but have spared you from a new post at each iteration.On Monday, Grameen posted its May report. One development pops out: the two key indicators of repayment have improved at record rates since my post. One of these is the non-recovery rate, the share of payments due in a month that are not paid. It fell from 3.45% in February to 2.89% in May, the steepest 3-month drop in the available recorded history save for what looks like a data glitch in 2002. The other metric is the PAR 30 (portfolio at risk, 30 days), which is the share of outstanding amounts that are owed by people at least a month behind on payments. The PAR 30 plunged more than 1% in the last two months, also a record:One can view this apparent good news with acceptance or suspicion. I do want to be fair to Grameen and avoid automatically accepting the bad news that it discloses to the world, with commendable and unusual transparency, while automatically doubting any good news. It is possible that the effects of the financial crisis are easing in Bangladesh, and that Grameen's thousands of hard-working employees have, with success, redoubled efforts to collect on loans. On the other hand, delinquency indicators are slippery fish. In principle, as it were, a lender can maintain or improve appearances by disbursing new loans that on paper allow old ones to be repaid. If the delinquency statistics are properly computed, they won't be skewed by this trick. (See Rich Rosenberg's classic.) But to my knowledge, the Grameen Bank has never public stated precisely how it computes its delinquency statistics.Whether there is any link between my original post and these record improvements is also a subject on which one can only conjecture. What might clear things up is some elaboration from the bank.