One of the most interesting talks from the INFORMS Annual Meeting was the “Operations Research in the Public Sector” session (TB23), sponsored by the Public Programs and Processes section and run by IBM Global Business Services. I previously wrote about IBM’s contribution to using a text-mining algorithm to improve the response time of the Social Security Administration to disability claims (IBM has some more info here), so I won’t write anything new.
Two other talks in the session dealt with the postal system (in the United States and in Canada). I’ve always liked the USPS since it is the only government agency that I can think of that supports itself. The US talk dealt with developing an algorithm for optimally using mail sorting machines that are location dependent. The Canada Post talk was interesting to me (an American) since there are different structural differences between the United States and Canada. For one, Canada is much larger than the US but has a much smaller population. Many of the mail processing stations are in the area of Toronto. In both mail systems, mail must be delivered to everyone in a timely manner, and there are real costs for this kind of equity. The MIP model determined the optimal mix of network, flow, and machinery (IBM has some more info here).
The last talk looked at reverse mortgages, which I had never heard of. Apparently, reverse mortgages are popular with seniors, who want to remain in their house that they own and who need some extra cash. The bank slowly buys the house back, providing cash to the homeowners and allowing the homeowners to remain in the house for the rest of their lifetimes. Obviously, the rates depend on the ages and number of the homeowners (AARP has some more info). IBM developed a better loan repayment model, but there are still many issues with forecasting the uncertainties in the housing market.