Projecting Federal Retirements

Apparently there was some confusion about whether the Office of Personnel Management is staying in the business of projecting federal employee retirement rates. One story said they were getting out, the agency says they're staying in. But who is doing the forecast isn't really the point. How it's being done is. The Office of Management and Budget has explained recently that forecasting retirements has become more difficult, and that more people close to retirement age are staying in the labor force. The complications come both from long-term trends towards working later in life, from the decimation of retirement accounts in the economic crisis, and from unemployment trends.

The thing is, I'd imagine there is a way to model these trends. It's not as if data doesn't exist on retirement rates in prior recessions. It's not as if we haven't had a recession (even if not one of this magnitude) since the Great Depression. I'd be curious to hear more about the models and variables that groups like Pew, the Partnership for Public Service and OPM are using, and where their models are running into trouble. One of the things Peter Orszag did at the Congressional Budget Office before leaving to run the Office of Management and Budget was to soup up the office's modeling and prediction capabilities; maybe they've got some ideas about how to predict federal workforce fluctuations?