« Danny Okrent of the New York Times Jumps the Shark (Why Oh Why Can't We Have a Better Press Corps?) | Main | Why Oh Why Can't We Have a Better Press Corps? (Yet Another National Review Edition) »

June 02, 2005

The Global Savings Glut Argument

Brad Setser has musings on Ben Bernanke's "global savings glut" argument:

Brad Setser's Web Log: Bernanke's global savings glut: Bernanke's global savings glut argument is quite subtle, more subtle than I perhaps have recognized in the past. Bernanke's argument goes beyond the 'US current account deficits don't matter since there is a global savings glut' headline that I occasionally push.

Bernanke takes his argument that there is a global savings glut to its logical conclusion. He argues that without a federal budget deficit, the global savings surplus that was invested in US government debt over the past few years would instead have been invested in other US assets. Real interest rates would be even lower, housing prices would be higher, investment in housing would be higher, and consumption in the US would be even stronger. As a result, the US current account deficit would not be much different:

The weakening of new capital investment after the drop in equity prices did not much change the net effect of the global saving glut on the U.S. current account. The transmission mechanism changed, however, as low real interest rates rather than high stock prices became a principal cause of lower U.S. saving. In particular, during the past few years, the key asset-price effects of the global saving glut appear to have occurred in the market for residential investment, as low mortgage rates have supported record levels of home construction and strong gains in housing prices. Indeed, increases in home values, together with a stock-market recovery that began in 2003, have recently returned the wealth-to-income ratio of U.S. households to 5.4, not far from its peak value of 6.2 in 1999 and above its long-run (1960-2003) average of 4.8. The expansion of U.S. housing wealth, much of it easily accessible to households through cash-out refinancing and home equity lines of credit, has kept the U.S. national saving rate low--and indeed, together with the significant worsening of the federal budget outlook, helped to drive it lower. As U.S. business investment has recently begun a cyclical recovery while residential investment has remained strong, the domestic saving shortfall has continued to widen, implying a rise in the current account deficit and increasing dependence of the United States on capital inflows...

I find it easier to think about these issues if we formalize them. Start with the savings-investment side of the national income identity:

I = Sp - DG + DT

Investment I is necessarily equal to private savings Sp minus the government deficit DG plus the trade deficit DT. And start with the premise that savings outruns investment elsewhere in the world, and this shows us as a large U.S. trade deficit as savers abroad soak up the dollar earnings of foreign exporters and invest them in the United States.

As long as investment is still high--as long as we are still in the dot-com boom--then, Bernanke says, U.S. private savings can be normal. But after the dot-com boom ends and investment falls, the savings-investment identity no longer balances: with lower-than-normal investment (especially business investment) as a share of GDP and a large capital inflow associated with the trade deficit, something else has to change in order to keep the savings-investment identity in balance. One thing that changes is the government deficit. But the rise in the government deficit is not sufficient--the private savings term Sp has to fall as well.

What makes Sp fall? Well:

Sp = Y - T - C = Y - T - (C0 + Cy(Y-T) + CwW(r)) = (1-Cy)(Y-T) - C0 - CwW(r)

Private savings, in Bernanke's implicit model, is equal to a fraction (1-Cy) of disposable income (Y-T), minus a term C0 that depends on consumer confidence, minus a term CwW(r) that tells us that private savings drops when household wealth W increases--and household wealth increases when the interest rate falls.

Thus the chain of causation in Bernanke's model runs roughly like this:

After the end of the dot-com boom, U.S. domestic investment falls as a share of GDP and the capital inflow rises. Excess savings thus appear in the flow-of-funds market. These excess savings push down interest rates. As real interest rates fall wealth--especially household real estate wealth--rises. Rising wealth induces households to cut their savings until the flow-of-funds is back in balance.

In this model the federal budget deficit is a minor player. As Brad Setser says, the implicit prediction is that it does not affect the trade deficit. Had the Bush administration pursued a balanced-budget policy there would still have been a fall in investment and there would still have been a rise in the capital inflow. With a balanced budget, excess supply of savings in the flow-of-funds would have been greater. Interest rates would have fallen further. Housing prices would have risen further. And private savings would have fallen further. After all:

I - DT = Sp - DG

And if DG does not grow as DT grows, then Sp must fall by more--and the only thing that can make Sp fall is an increase in household wealth W, which is the result of falling interest rates r.

This is Bernanke's argument. What do we think of it?

I'm very skeptical. It is of a brand of macro that I think of as one-identity-economics. You take an accounting identity. You assume that certain terms of it are fixed. And you then derive conclusions--in this case, that the growth of the budget deficit has moderated the fall in private savings.

The problem with one-identity-economics lies with the assumption that certain terms in it are fixed. There are lots of channels of adjustment in the world economy, and it is a safe bet that with different levels of interest rates and different levels of wealth we would see different levels of corporate investment and of net exports.

Posted by DeLong at June 2, 2005 12:21 PM