Yield Curve Is a Graphical

Yield curve is a graphical representation of the yields available on treasury securities going out over time. The slope of the curve can be determined by either examining the entire curve or selecting key points along the curve for study. The yields on these securities reflect the expected rate of return on these products. Because the market sets the prices, and therefore the yields, the yield curve represents the market's expectations of future interest rates. These rates are expected to be set in relation to inflation and economic activity. Thus, treasury rates are a proxy for those two variables. As a result, the slope of the yield curve is expected to be an accurate reflection of the market's view of the future economic situation. Indeed, the slope of the yield curve has been linked in studies to subsequent changes not just in GDP but in consumption, industrial production and investment (Estrella & Trubin, 2006).

The yield curve is a frequently cited leading indicator. Research has indicated that the slope of the yield curve has a "consistent negative relationship" with subsequent economic activity. The lead time required for this relationship to materialize is between four and six quarters. The yield curve has been an effective predictor of recessions in particular, since 1950. The relationship between the yield curve and medium-term economic shifts has been demonstrated to hold true in other industrialized nations as well (Federal Reserve Bank of New York, 2009).

The yield curve at any given point is dependent on the maturities that are selected for analysis. The maturities are selected to match the time horizon being studied. Thus, the yield curve that is an accurate predictor of recessions four to six quarters in the future should typically be of similar maturities (Ibid). But while the yield curve has proven a good predictor in the near future, how far into the future has the yield curve been able to predict the economy?

Research indicates that the slope of the pairing of the three-month and ten-year treasury rates is an accurate predictor over longer time periods. This particular slope is most useful two to six quarters ahead (Estrella & Mishkin, 1996) and it would be reasonable that a ten-year rate would only give prediction power up through ten years, however.

It is worth noting that the predictive power of the yield curve is dependent on the benchmark that is set. Each set of rates is a predictor when weighed against a benchmark that is known as a point at which recession is certain. These benchmarks will different between point sets, so it is important that the proper benchmark be chosen for any given point set (Federal Reserve Bank of New York, 2009).

In order to determine how far out the yield curve can be a predictor, it is important to understand why the yield curve is a predictor. Investor expectations are a contributor to the slope. Those expectations are in part related to another key variable -- monetary policy. Monetary policy, however, is short-term in nature and difficult to estimate several years into the future.

Another influencer of the curve is the markets for different Fed maturities. This variable can distort the yield curve, hindering its ability to act as an economic predictor. Another potential constraint on the outwards bounds of the yield curve as a predictor is that the ten-year bond is the longest one with a consistent, liquid market.

The Fed and the broader market typically use the yield curve as a predictor of recession going out a year, maybe a year and a half. Going further is a more esoteric matter. Plosser and Rouwenhorst (1994) showed that "term structure has significant predictive power for long-term economic growth" in part because it contains information about future real activity that is "independent from information about current of future monetary policy." They argued that the spread could predict accurately up to five years forward, but with the caveat that this predictability was heavily influenced by the spread's ability to predict just two years forward (Dotsey, 1998). Subsequent research on the predicative power of the yield curve focuses on the shortcomings of the model, and on shorter-term predictions, between six months and a year. Further research attempts to deconstruct the predictive power of the yield curve.

What we are left with, then, is the supposition that economic activity can be predicted up to five years into the future with the yield curve. This supposition, however, is weakly supported. Plosser and Rouwenhorst's support of two-year predictability is much stronger. Settling on two years as the outer limit of predictability is reasonable given the lack of evidence to support a further time frame. To achieve such predictability, it may be better to use a spread based on the ten-year and a one year bond, rather than the Fed funds rate or a 3-month T-Bill.