To the extent that investors wish to compare our 5.6% estimate for 10-year S&P 500 total returns with the 2.7% yield on 10-year Treasuries, it is important to recognize that the higher 10-year expected return in the S&P 500 comes with a several-fold increase in risk, particularly over a shorter horizon. Moreover, the divergence between the two figures tends to expand, not contract, during economic downturns. Stocks are not cheap, and to the extent they may outperform bonds over the next 10 years, it will most likely be with extreme discomfort. My impression is that near-term risk in the stock market is very high, while the risk for bonds is more long-term in nature. As with economic data, expected outcomes should not be treated as if they are deliverable with exquisite precision. Long-term returns are not achieved in a straight line.
Observations on massive mortgage refinancing
One of the policy suggestions bouncing around recently involves refinancing all of the mortgages held by Fannie Mae and Freddie Mac. This policy would essentially repay Fannie and Freddie security holders at par (face value), and replace those mortgage pools with freshly issued securities. At present, longer term agency securities typically trade at prices higher than face value because the underlying mortgages were set at interest rates well above the interest rates available now, so a massive refinancing would involve immediate losses to existing holders of agency debt.
Unfortunately, the primary loser would be the Federal Reserve, whose holdings are predominantly long-term mortgage bonds, mostly at prices significantly higher than face (with the possible exception of any newly issued debt having below-market coupons). From my perspective, massive refinancings would have been a better idea before Bernanke and Geithner put us on the hook for losses on agency debt that never had an explicit government guarantee. In effect, the old holders of those agency securities have already walked away with the cash, so the Fed would take the loss in the event of refinancing. It would also undoubtedly be the buyer of the freshly issued refinancing bonds.
Existing borrowers in those mortgage pools would benefit from a reduced interest burden, but as a significant percentage of those homes are worth less than the mortgage principal, it would still leave a great deal of foreclosure risk in the system. In any event, a massive refinancing would entail a reduction in interest rates to homeowners at prevailing levels, with a major loss to the Federal Reserve, which already paid above-par to the previous owners of those Fannie and Freddie securities.
Of course, existing holders of above-par agency securities would also take a loss, and to the extent that they are holding hedged positions (long agency, short other long-duration instruments), it follows that they would have to reposition by absorbing the loss and buying newly issued agencies, or alternatively, cover their hedge by purchasing long-duration instruments (primarily long-term Treasury debt).
As a result, the full impact of a massive refinancing would be a) a major loss to the Fed, since prior owners of that agency debt already received full value; b) a loss to agency debt-holders who still own securities above par; and c) a possible spike in the price of long-term Treasury bonds as private market participants respond to the abrupt shortening of the asset side of their portfolios. Of course, it's possible that the Fed could simultaneously sell Treasuries and buy even more agency debt, which would take the pressure off of the Treasury market and load the Fed's balance sheet with even more underwater mortgage debt. But in any event, it should be observed that the primary loser in the event of a massive refinancing would be the Fed, and by extension, the public. It would be a large transfer of wealth from the general public to specific mortgage borrowers. Again, this would not be monetary policy - it would be pure fiscal policy, and any such action should be subject to Congressional approval.
