T C A Srinivasa-Raghavan: Shapely curves don't matter

OKONOMOS

Just as certain human muscles need to be firm if not actually bulging, yield curves, one has always been told, need to be upward sloping. Flat curves are, therefore, bad news for the economy but not as bad as inverted curves.

The latter presage a recession. In the US, since the mid-1960s, every recession has been preceded by an inverted yield curve.

India has had a flat yield curve for some time now, but without a recession.

If you accept the theory that has developed on the subject, this means that the Reserve Bank of India (RBI) has been sending mixed signals to investors, mainly Indian banks.

If theory is right, it means the banks can't figure out whether interest rates will move significantly up or significantly down.

According to theory, when the yield curve is flat, investors maximise their returns by choosing fixed-income securities with the least risk, for example, government bonds.

This is what the Indian banks have done. They would have been foolish to do otherwise. Clearly, either the banks knew, which means the theory is wrong; or the theory is right and the banks have just been lucky.

The more important question, however, is why does flatness occur and, even more importantly, does it matter?

Or, if you like, why do short- and long-term rates converge? No one really knows. One widely accepted guess, though, is that potential lenders take a dim view of future growth prospects.

But this breaks down when you match it with the Indian and Chinese experiences, the two fastest growing economies in the world "" the latter by a long margin.

In both countries, the yield curves are about as flat as they can get, which according to theory means that investors are pessimistic about their long-term growth prospects.

So how is growth being financed? China is financing it through a massive export surplus and a huge, but unacknowledged, fiscal deficit. But it runs the risk of committing the error that Paul Krugman pointed to in the case of east Asia, namely, how Korea and others used far too much capital.

India, far more sensibly (even if inadvertently) is growing via the services sector and the unorganised sector. Both require far less capital.

The sustainability of these strategies is a different issue altogether. But let me be the first to warn: there cannot be two Chinas supplying the world with manufactures. Nor, indeed, can there be two Indias supplying it with services.

Anyway, be that as it may, if long-term rates fail to tell us much about growth prospects, can we make some sense out of the short-term ones? Received wisdom says no, but luckily a research paper* has come along that says yes, you can.

"The short rate has more predictive power than any term spread", say Andrew Ang of the Columbia Business School, Monika Piazzesi of the Graduate School of Business at Chicago and Min Wei from the Board of Governors of the Federal Reserve in Washington.

They have modelled the dynamics of the yield curve with GDP growth and found that "the entire yield curve, not just the arbitrary maturity used in the construction of the term spread, would have predictive power. Using information across the whole yield curve, rather than just the long maturity segment, may lead to more efficient and more accurate forecasts of GDP."

The authors rule out arbitrage on the grounds that hedge funds and large investment banks take large bond positions that eliminate arbitrage opportunities arising from inconsistent bond prices in the cross-section and with their expected movements over time.

They then build a model in which only a few yields and GDP growth are the variables.

They then go on to show that "the yield curve model can capture the same amount of conditional predictability that is picked up by simple OLS regressions." In other words, there is no signal loss in their method.

They then ask how well GDP growth can be predicted and find that their model has several advantages, which are mostly technical. Most importantly, their model "predicts that the nominal short rate contains more information about GDP growth than any yield spread."

In other words, they are saying, to hell with the slope of the yield curve because "our yield-curve model is driven by the gain in estimation efficiency." If India and China are good examples, they are probably right.

*What Does the Yield Curve Tell us about GDP Growth? NBER Working Paper No. 10672, August 2004