Further squeezing Harvard was a transaction Summers had pushed it into in 2004, when he successfully argued that the university should engage in a multibillion-dollar interest rate swap with Goldman Sachs and other large banks. Under the terms of the deal, Harvard would pay Goldman a long-term fixed rate while Goldman paid Harvard the Federal Reserve rate. The main goal was to lock in a low rate for future debt, and if the Fed had raised rates, Harvard would have made hundreds of millions. But when the Fed slashed rates to historic lows to try to goose stalled credit markets, the deal turned equally sour for Harvard: By last November, the value of the swaps had fallen to negative $570 million. The university found itself needing to post more collateral to guarantee those swaps, and would ultimately buy its way out of them at an undisclosed cost.
If this story is true, Summers basically created balance sheet duration mismatch, with short assets and long liabilities. This creates negative duration, which is just as risky as more traditional positive duration (i.e., the S&L problem). I wonder if Summers plotted out a worst-case scenario for this transaction, or if he just thought he was really good at forecasting interest rates.