Printer Friendly

The yield curve as a predictor of business cycle turning points.

The Yield Curve as a Predictor of Business Cycle Turning Points

THE YIELD CURVE has long been recognized as a potentially useful tool in the analysis of financial markets. More recently, it has been the focus of renewed interest with the suggestion that it may be one of the key indicators monitored by Fed policy makers. For the most part, however, the emphasis with the yield curve has been as a signal of changes in inflation expectations. What seems to get less attention is that the yield curve may be at least as helpful in predicting changes in real economic activity. If the term structure of interest rates reflects in part the collective forecast of expected inflation, it also must reflect investors' assessments of future real growth and interest rates as well.


Before reviewing the historical data, why might we expect the yield curve to provide some insight into the likely future path of real economic activity? The answer lies in yield curve's reflection of changing short- and long-run expectations of future economic performance by active financial market participants. (1)

Consider initially a business cycle expansion. As the economy recovers from recession, for example, the relative demand for liquidity increases and short rates rise. Part of this rise in short rates is passed to long rates as investors anticipate further increases in output, prices and future short rates. As short rates continue to climb, tightening liquidity (and perhaps other imbalances) reduces the probability that the degree of increases in output, prices and future short rates that were anticipated earlier in the expansion will continue. In essence, short- versus long-run expectations of real interest rates or liquidity premiums begin to shift. As a result, the rise in short rates is not passed into long rates to the same extent as earlier, and the yield curve begins to flatten. As the probability of future weakness grows, the yield spread (the difference between short rates and long rates) widens.

A similar process works in reverse during business down turns. As the recession proceeds, the probability grows that future economic performance will be characterized by strength rather than by weakness. Declines in short rates don't get passed to long rates to the same extent as earlier in the downturn, the yield curve steepens and the yield spread narrows as the economy approaches a business cycle trough.


Studies of interest rates as far back as the mid-1800s provide support for the yield curve as a predictor of business cycle turning points. (2) In spite of the difficulties in studying the yield curve over an extended earlier period of time, several common characteristics emerge from the various studies. In particular it has been found that:

1. Both short and long rates tend to rise during business cycle expansions and to decline during subsequent downturns.

2. Yield curves with negative slopes, i.e., short rates exceeding long rates, have occurred only around business cycle peaks

3. With the exception of unusual behaviors of government yields in the period 1933-45 and 1961-66, from the Civil War to the present short rates have risen more relative to long rates during expansions and fallen more relative to long rates during recessions

The potential relevance of the yield curve in helping to predict business cycle turning points was alluded to in one researcher's conclusion that, "The striking coincidence of timing of . . . (interest rates) and business cycles suggests that the forces that determine the peaks and troughs of business cycles must also play a role in determining those in . . . (interest) rate) cycles of . . . government obligations." (Kessel, p. 66)


In order to assess the yield curve's performance as a forecasting tool in more recent times, we examine in some detail the relationship between short- and long-term yields over the business cycle for the past thirty years.

In Figure 1 we have plotted quarterly bond-equivalent yields on three-month Treasury bills (T-bills) and those on ten-year Treasury bonds (T-bonds) from 1955 through the second quarter of 1989, a period that includes six full-blown recessions with their peaks and troughs indicated by the letters P and T. (3) As illustrated in the chart, yields during this period exhibit the same principal characteristics as those found in studies going back much further in time. In particular, we see from the chart that short and long rates tend to move up and down together, that short rates have exceeded long rates -- an inverted yield curve -- only (but not consistently) at business cycle peaks, and that shorts have risen relative to longs during business cycle expansions and have fallen relative to longs in downturns.

A clearer picture emerges in Figure 2 in which we have plotted the quarterly yield spread -- one proxy for the yield curve -- of the rate on T-bills minus the yield on T-bonds over the same period. Our interest in this chart is in identifying any consistent pattern in this spread that appears to be a precursor to business cycle turning points.

Several patterns are evident in the yield spread in Figure 2. First, the spread exhibits a sustained rise prior to business cycle peaks and a sustained decline before business cycle troughs. Secondly, peaks in the spread have been moving up over time while lows have been moving downward. Finally, an inspection of Figure 2 shows a high degree of symmetry between the rise in the spread in the preceding expansion and how much it declines in the recession that follows. The question now is whether we can extract from this record signals that can serve as useful tools in helping to forecast turning points in economic activity.


Although signals of business cycle troughs are undoubtedly of less interest while the economy is still in an expansion, the performance of the T-bill/T-bond yield spread illustrated in Figure 2 provides one guideline that may be quite helpful when the time comes to predict the next business cycle trough. As mentioned above, a noticeable degree of symmetry exists between the decline in the yield spread during recessions and the magnitude of its run-up during the preceding expansion. The record of these full cycles is illustrated in Table 1 below, which shows for each of the cycles how much of the preceding increase in the spread had been given up by the time the recession had about run its course. (4)

Although the numbers in this table do vary, one pattern that emerges is the rapid increase in the recovery percent as we move from three to two quarters prior to the trough. While the former averages about 40 percent, the latter is about 60 percent. In addition, a closer look at the table reveals the guideline that whenever approximately 70 percent or more of the preceding rise in the yield spread has retreated, the recession is about to end within one or two quarters.


Unlike its performance in downturns, the yield curve has not displayed nearly as consistent a pattern during expansions. As with the lows getting lower, the (T-bill minus T-bond) yield spread has been peaking at higher and higher values in more recent years. Unlike the signal of a business cycle trough suggested in Table 1, however, no such pattern is evident in the spread as a sign of an impending business cycle peak. Thus, we need to search for a signal of recession that is not invalidated by these increasing peaks in the yield spread.

Although reasons for a widening spread in more recent expansions are easy to hypothesize, constructing measures that properly adjust for such influences is no easy task. One approach is to treat only the more extreme high-side values of the spread as potential signals of recession. Figure 3 illustrates this methodology where "Tight" and "Easy" values of the spread are identified as those lying outside the band of (plus and minus) one-half of the standard deviation from the (1955-1989Q2) mean spread.

Such a classification scheme has been used by this author and others for measures of monetary policy, fiscal policy and credit conditions. (5) As in these other cases, the choice of one-half standard deviation is arbitrary and its usefulness can be judged only by how well it has performed historically. The top series in Figure 3 illustrates the record of this T-bill/T-bond spread and classification scheme during the past thirty-plus years. As evident in the Figure, every full-blown recession has been preceded by a "Tight" yield spread. Moreover, the only time that "Tight" credit has not been followed by a recession (within from six to eighteen months) was in the mid-1960s. While no full-blown recession occurred at that time, a growth recession did take place, with growth in industrial production (I.P.) grinding to a halt. This fact is illustrated in the bottom series in Figure 3, which displays (through 1989Q2) average annualized I.P. growth rates over the four most recent quarters. Although growth in industrial output did not turn negative in 1967, it was only positive by a slim margin.

All in all, when viewed in this way, the yield curve appears to be a valuable tool for use in predicting turning points in business activity. With forecasters seeming to need all the help they can get in the currnet economic environment, we turn to an examination of what the yield curve has to offer in helping to decipher how much longer the current expansion has to run.


The current expansion is one noteworthy for its longevity. In terms of the yield spread, it also has displayed some unusual elements. One is that the yield spread has remained "Easy" for much of this expansion. Another is that, until 1988, the spread exhibited no sustained rise during this recovery and in fact, remained "Easy" except for brief shifts to an intermittent marginally "Neutral" stance in 1986 and 1987. In short, the measure of the yield curve illustrated in Figure 3 has provided a high degree of comfort against a business downturn induced by credit tightening. But should we have been so sanguine? And what about inflation?


One major potential problem with the yield spread displayed above is that it makes no allowance for the impact of transitory inflation, or inflation expected to be a temporary phenomenon. This factor could pose a problem because bouts of inflation or deflation viewed as temporary can have a disproportionate impact on short as opposed to long rates. As a result, an increase in short relative to long rates, for example, could be mistakenly viewed as a signal that investors are looking for an easing in real economic activity, inflation, and future short rates. Thus, it would be helpful to adjust the yield spread for inflation of a more transitory nature. The approach here is to account for temporary versus more permanent inflation by subtracting the difference between Total CPI inflation and that for the CPI less food and energy from the short-long yield spread. [6]

The result of this adjustment is illustrated in Figure 4, which reveals some interesting differences from the unadjusted version in Figure 3. Some peaks in the spread, for example, are now higher (e.g., 1960), while others are lower (e.g., 1973). [7] In the current expansion, the general pattern is unchanged by the inflation adjustment, but there are important differences nevertheless. While both versions of the spread have shown increases in this expansion -- the one beginning in 1986 and the current one -- those in the adjusted version have been more pronounced and suggest that credit conditions have at times been much tighter. In fact, whereas after adjusting for temporary inflation, conditions were marginally "Tight" in 1986-87 and again by year-end 1988, the unadjusted version registered a "Tight" reading in 1989Q1 for what is the first time in this expansion. In spite of the differences between the yield spreads depicted in Figures 3 and 4, by the first quarter of this year both versions were reflecting the same stance of credit conditions that has preceeded each recession in the U.S. economy over the past thirty years.


Although the latest reading from the yield curve suggests that a recession may be in the offing, there are several reasons to question seriously whether a downturn is imminent. These include:

1. Absence of the kind of gross imbalance in the economy that have typically preceded past recessions

2. The possibility that the historical link between the yield spred and real activity has been broken (dur, for example, to such things as financial deregulation and the role of foreign investors in today's U.S. financial markets)

3. As described below, failure of the ecomomy in the current expansion to respond to "Easy" and "Tight" credit as it has in the past

Because the validity of the first two arguments can only be properly assessed with the passage of time, they are not of much help in deciding now upon the likelihood of a near-term recession. By contrast, if we can identify why real activity in this expansion has not responded to domestic credit conditions as it has in the past, we then would be in a better position to assess the real meaning of the latest stance of those credit conditions.

Unlike the past thirty years when recessions tended to follow "Tight" conditions and accelarations in growht followed "Easy" credit conditions, these patterns have not held in the current expansion. The extremely "Easy" stance in the 1984-85 period was followed by a noticeable slowdown in output growth, while an extremely "Tight" position in 1986 did not bring on a recession as would have been predicted from past patterns. More recently, the "Easy" credit conditions from mid-1987 to mid-1988 have been followed by an easing in growth in industrial output. As for the current period, the "Tight" credit conditions at the end of 1988 and early 1989 would normally lead us to predict a recession sometime later this year. In spite of clear evidence that growth is slowing and fears of recession have intesified, unless we can account for the unusual relationship between the yield spred and real activity since 1982, this latest reading from the yield curve may only serve to confound the outlook for the economy rather than helpign to defog our crystal balls.


One area that immediately comes to mind as exerting a different influence this time is the foreign sector. (8) As is well known, the role of the foreign sector in this expansion has been unique. It is generally agreed, for example, that strong competitive pressures from abroad played a significant part in depressing real output in the U.S. in 1985 and 1986. At the same time, robust export demand has led to unexpectedly strong growth in the U.S. economy in more recent years.

Figure 5 repeats Figure 4 since 1982 and also adds a measure of this foreign sector influence in the form of average annualized growth in real U.S. merchandise exports over the four most recent quarters. As illustrated in the figures and not suprisingly, growth in industrial production accelerated as the economy emerged from the recession and then began to slow as the cyclical thrust subsided. By 1985 and 1986, however, growth had slowed to such an extent that it was flat. In fact, for the first time in any expansion of the past thirty years, growth was this low without the occurrence of either an ordinary or a growth recession.

Why such slow growth in spite of noticeably "Easy" credit conditions during the second half of 1984 and for most of 1985? The foreign sector offers an explanation. As indicated in the figure, export growth was downshifting from its very healthy levels of the past two years to nothing less than a dead stop by mid-1985. This, along with the dissipation of the cyclical rebound, served to weaken domestic growth during that 1985-86 period. Similarly, the "Tight" credit conditions of 1986 and early 1987, which would normally be expected to lead to a slowing in growth shortly thereafter, were countered by the eye-catching increase in export growth with a subsequent surge in domestic output. More recently the "Easy" credit of 1987 and early 1988 was tempered by some slowing in export growth and moderation in the pace of industrial output growth. In effect, the U.S. economy since 1985 has been buffeted by the countervailing influences of domestic credit conditions and foreign demand for U.S. goods.


The question at present then is to what extent the current "Tight" credit conditions are a threat to the expansion. (9) While a look at history would say that they pose a significant threat, a review of more recent experience suggests that the influence of "Tight" credit conditions on real growth may be either tempered or reinforced by demand from abroad. If, for example, for foreign demand were to match its strength of the recent past, it could be expected to mitigate somewhat the depressing effect of domestic credit conditions. Reduced stimulus from abroadf, however, would increase the chances that "Tight" credit conditions would lead to recession.

In short, although the yield curve may serve as a good summary measure of the posture of credit conditions and while history can suggest whether current "Tight" credit poses a threat of recession, the yield cyrve by itself cannot tell us whether or not a near-term recession is in the cards.


For many readers, the foregoing analysis may seem a bit contradictory. On the other hand, it has been demonstrated that the yield curve has proved over the past thirty years to be a highly reliable indicator of impending recoveries and, with certain adjustments, of approaching recessions. At the same time, it has benn argued that the yield curve by itself cannot tell us whether the economy is now about to plunge into a recession. The resolution of this apparent inconsistency lies in the recognition that no single measure, including the yield curve, is capable of telling us if we are about to embark on a recession (or recovery). The most that can be reasonably expected from a good forecasting measure is that it provide a reliable indication that the pace of economic activity is about to change in a noticeable fash ion. From there it is the job of the analyst to assess the role of those key elements that could serve to either moderate or intensify the impact of the forces reflected in that forecasting tool.

What the yield curve is telling us in the present case is that domestic credit conditions tightened dramatically during 1988 and by teh first half of this year had become "Tight" to a degree othat has led almost unfailingly to recessions in the past. We conclude from this that domestic growth is likely to slow appreciably as 1989 unfolds and that the economy faces a real threat of recession. Whether such weakness will in fact turn into recession, however, cannot be adequately determined from the yield spread alone, but will depend upon other elements such as the Fed's reaction to signs of a slowing economy, the vulnerability of various sectors of the economy to slower growth in income and employment, and the impact from other influences such as fiscal policy and the foreign sector.

Because of the role that foreign demand has played in the recovery to date (and also based on the assumptions that fiscal policy is hamstrung by deficit concerns and that weak spots such as high debt ratios would exacerbate but not trigger a downturn), we focused on the foreign sector as being a real key to how the current slowing plays out. According to this view then, the ultimate answer to the question of recession or no recession hinges on the perceived outlook for growth in real U.s. exports.

Undoubtedly, such a conclusion may leave some analysts disappointed that the yield curve by itself did not provide the ultimate answer they were seeking. What must be recognized, however, is that the yield curve alerted us to the threat of recession and motivated a clooser look at current conditions compared to those preceding past downturns. It is difficult to imagine what more could reasonably be demanded of any single forecasting measure.


(1) For a description of why the yield curve or spread might be a good indicator of monetary policy, see Robert D. Laurent, "An Interest Rate-Based Indicator of Monetary Policy," Economic Perspectives, Federal Reserve Bank of Chicago, (January/February 1988), pp. 3-14. An empirical analysis of the informational content of the yield curve can be found in Frederic S. Mishkin, "The Information in the Term Structure: Some Further Results," Journal of Applied Econometrics, 3 (October-December 1988), pp. 307-314.

(2) See, for example, Phillip Cagan, "Changes In The Cyclical Behavior Of Interest Rates," REview of Economic and Statistics, Volume XLVIII, No. 3, August 1966; Reuben A. Kessell, "The Cyclical Behavior Of The Term Structure Of Interest Rates," Occasional Paper No. 91, National Bureau of Economic Research, New York, 1965; and John H. Wood, "Do Yield Curves Normally Slope Up? The Term Structure Of Interest Rates, 1862-1982," Economic Perspectives, Federal Reserve Bank Of Chicago, (July/August 1983), pp. 17-23.

(3) Ten-year Treasury bonds were chosen for this analysis because they provide a longer continous series than either the twenty-year to thirty-year bonds. Nevertheless, the range of data prevents us from computing the yield spread for the full expansion prior to the 1957-58 recession.

(4) The preceding increase used here is that of the longest sustained rise in the spread prior to each recession. Although six of these yield spread cycles occur in the period under review, only four are included in Table 1. In particular, two of those cycles in which unusual government interference in the financial markets affected the spread between short- and long-term interest rates have been excluded. These include "Operation Twist" in the first half of the 1960s and credit controls in 1980. The former affected the symmetry in the yield spread for 1969-70, while credit controls instituted under President Carter in the first half of 1980 affected the pattern of the spread in conjunction with the 1981-82 downturn.

(5) See, for example, Keith M. Carlson, "The Mix of Monetary and Fiscal policies: Conventional Wisdom Vs. Empirical Reality," Review, Federal Reserve Bank of St. Louis, (October 1982), pp. 7-21 and Howard Keen, Jr., "Summary Measures of Economic Policy and Credit Conditions as Early Warning Forecasting Tools," Business Economics, (October 1985), pp. 38-43.

(6) Inflation rates are measured here as a four-quarter moving average in the year-over-year change in prices.

(7) It is interesting to note that when adjusting for inflation, the 1973 recession triggered by a negative supply shock had less severe credit tightening than any of the others under review.

(8) For the assertion that things are different this time based on different inflationary expectations, see Frederick T. Furlong, "The Yield Curve And Recession," Weekly Letter, Federal Reserve Bank of San Francisco, (March 10, 1989).

(9) For examples of analyses of yield spreads in the current environment, see Robert D. Laurent, "Testing the 'Spread'," Economic Perspectives, Federal REserve Bank of Chicago, (July/August 1989), pp. 22-34; E. J. Stevens, "Is There a Message in the Yield Curve?" Economic Commentary, Federal Reserve Bank of Cleveland, (March 15, 1989); Jack Willoughby, "Dangerous Shapes," Forbes, (January 23, 1989), pp. 39-40; David A. Wyss, "Are The Financial Markets Warning of Recession?," Review of the U.S. Economy, DRI/McGraw-Hill, (March 1989), pp. 13-16; and "Yield Curves: New Twists And What They Mean," Economic Week, Volume 17, No. 1, Citicorp, (January 3, 1989).
COPYRIGHT 1989 The National Association for Business Economists
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1989 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Keen, Howard
Publication:Business Economics
Date:Oct 1, 1989
Previous Article:What Europe 1992 means for U.S. business.
Next Article:Wall Street economists: are they worth their salt?

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters