The Predictive Power Of The Yield Curve For Output And Inflation

The predictive power of the yield curve has been at the forefront of recent economic discussions in both the United States and Europe. New research by Arturo Estrella, published in the July 2005 Economic Journal, explains why there are these empirical relationships between real economic activity, inflation and the pattern of yields on bonds of different maturities – and whether they are likely to persist under changing economic conditions.

A large body of empirical evidence since the late 1980s documents a strong predictive relationship between the slope of the yield curve – a line drawn through the effective interest rates on government securities with different maturities – and future real economic activity and inflation. A normal yield curve gently slopes upwards with long-term rates higher than short-term ones, reflecting expectations of positive economic growth and compensation for the higher risks – including unexpected inflation – of investing for a longer period. A flat curve when all maturities have similar yields or an inverted curve when long-term yields fall below shortterm yields – as in the UK today – send signals of economic uncertainty and expectations of slowdown or even recession.

The analysis in this study suggests that the interaction of monetary policy with some features of the real economy gives rise to the predictive power of the yield curve. Though precise numerical formulas may vary with changes in monetary policy or economic parameters, the existence of some form of predictive power for both output and inflation is robust.

In the United States, yield curve inversions and subsequent recessions have occurred hand in hand since the 1960s. The start of a recession typically follows within a year of the low point for the difference between 10-year and 3-month Treasury rates. Recent experience shows that the empirical evidence is of more than historical interest. For example, the National Bureau of Economic Research dated the start of the last recession in the United States to March 2001, following a yield curve inversion that lasted over most of the second half of 2000. The empirical relationship between the yield curve and recessions appears quite robust in the face of ostensibly important changes in the economic, financial, and monetary environment since the 1980s.

The relationship is also evident in the international context: similar evidence has been gathered for various European economies, as well as for Canada and Japan. Especially strong are the results for Germany, the United States and Canada, whereas weaker results are found in the case of Japan.

The research strategy of this article is to construct a model that is sufficiently general as to incorporate the main relevant features of the economy, and yet simple enough to be solved analytically so as to expose the key economic relationships. The model includes components corresponding to the private sector of the economy, the monetary authority, and the financial sector.

Although the analysis is mainly theoretical, the realism of the model is validated empirically by estimating the behavioural equations using US data from 1962 to 2002. Estimates are consistent with both the earlier evidence and the theoretical assumptions of the model. Why does the yield curve predict output and inflation? The model suggests that monetary policy is an important determinant of the predictive power and of the precise numerical form of the predictive relationships. But other features of the real economy are also important.

In the case of inflation predictions, the relationship between economic slack and inflation (the ''Phillips curve'') plays an important role. In addition, if the monetary authority follows specific goals such as targeting inflation or limiting output variability, the full structure of the real sector of the economy becomes relevant in the predictive equations. In general, though precise numerical formulas may vary with changes in monetary policy or economic parameters, the model suggests that the existence of the predictive relationships encountered in the empirical literature for both output and inflation is quite robust.

One interpretation of this result is that qualitative relationships – such as between a yield curve inversion and a subsequent recession – should be fairly persistent, whereas precise quantitative relationships – such as between the yield curve slope and the subsequent rate of real GDP growth or the inflation rate – may vary over time.

''Why Does the Yield Curve Predict Output and Inflation?'' by Arturo Estrella is published in the July 2005 issue of the Economic Journal. Arturo Estrella is Senior Vice President in the Capital Markets Department of the Research and Statistics Group at the Federal Reserve Bank of New York.