On January 2, 2005 I sent my market friends a bond market forecast for the year ahead. I thought you might find it fun to read. It illustrates the way I approach the problem of market forecasting
2005 BOND MARKET FORECAST
January 2, 2005
Our first long term bond forecast was published on December 29,1982 when the 10 year treasury note was yielding 10.47%.
The thesis of that forecast was that a new long term cycle in bond prices and interest rates had begun at the September 1981 peak in long term interest rates when the 10 year treasury note was yielding 15.84%. We expected the new cycle to evolve in a way similar to the cycles that had begun at the two previous long term peaks in interest rates in 1857 and in 1920.
We quote from the December, 29, 1982 forecast: "If the bond market follows the average of these two historical cycles then bond prices should be in a generally rising trend for the next 30 years. The peak in bond prices and the low in long term interest rates is not due until the year 2010.....long term bonds are once again a high return, long term investment vehicle and will remain so for a generation to come."
It is our view that this predicted 30 year, long term downward cycle in the 10 year note yield has yet to be completed. This contrasts with today's conventional wisdom which holds that both stocks and bonds are "trash", i.e. both financial instruments should be expected to yield subnormal returns for the indefinite future.
Our own "contrary opinion" is based largely on the belief that the long end of the Treasury market has yet to price in a long term lull in inflation (another contrary opinion of ours!) and that when it does the 10 year notes will be yielding less that 3.00% AND less that the Dow Industrials. ( Currently the 10 year note is yielding 4.27% and the Dow 2.25%.)
Our shorter term forecasts (1 to 3 years ahead) are based largely on hypothesis that the duration and extent of bull and bear markets show a certain consistency over the years. This forecasting technique we developed from a study of the late George Lindsay's methods for forecasting stock prices and it has served us well. We admit that we rely as much upon art as upon science in making these forecasts. They are a series of educated guesses based upon the market's historical record and informed by our own market forecasting experience of nearly forty years.
The current situation in the interest rate markets is unprecedented in the post World War II era of activist montetary policy in the USA. In 2003 the Federal Reserve drove short term interest rates to levels (under 1%) normally associated with depressions and last seen in the US in the 1930's and 1940's. We believe the 2003 low of 0.80% in three month t-bills was well below what could resonably be regarded as normal (perhaps 2.50%) for the level of economic activity at the time. This situation was reflected in the abnormally steep yield curve which in mid-2003 had the 10 year notes yielding 275 basis points more that the 2 year notes.
The interest rate cycle which began from the 2003 lows will therefore look quite different from past cycles associated with economic expansions. This difference will show up largely in yield curve behavior. The curve is stll quite steep as this is being written, but not abnormally so, and our best guess is that the yield curve will continue to flatten until we see 2 year notes yielding less that 3 month bills (an "inverted" curve) while the spread between the 10 year and the 2 year notes will narrow to essentially zero.
Put another way, we think that the short end of the market (securities of less than 2 years maturity) will show steadily rising rates with only occasional multi-month interruptions until sometime in late 2006 or in 2007. On the other hand, the long end of the market (10 year and longer maturities) will on average show steady yields during the same period.
Now for a more detailed look at short term rates. For this we will use the 3 month treasury bill as our interest rate index.
For the past 50 years the 3 month bills have marched in tune with what Lindsay would have termed two long term time periods. The first averages 12 years 4 months from low to high or from high to low, while the second averages 14 years 3 months from low to high or from high to low. To illustrate this principle we observe that the November 6, 2000 high yield of 6.24% occurred 14 years 1 month after the low yeild of 5.05% on October 6, 1986. Moreover, the June 19, 2003 low yield of 0.80% occurred 14 years 3 months after the high yield of 9.10% on March 27, 1989.
The next base point for this calculation is the low yield of 2.61% which occurred on October 1, 1992. Adding 14 years 3 months brings us to January 2007 as the ideal time for the next yield high in 3 month t-bills.
Moves from low to high in 3 month bill yields typically last either an average of 26 months or an average of 42 months. August 2005 is 26 months after the June 2003 low yield of 0.80%. Mooreover, February 2005 is 12 years 4 months after the low yield of October 1992. Therefore we estimate that about May of 2005 the Fed will temporarily suspend its move towards high rates, offering as its reason the assesment that rates at the time are consistent with non-inflationary growth. However, this suspension will last only a few months and will be followed by a move to a tighter monetary policy and this tighening will conitnue to January 20007, 43 months after June 2003.
How high will the 3 month bill yield be in January 2007? The last four trends from lower to higher rates carried the 3 month bills up an average of 340 basis points. This mechanically projects a yield of 4.20% in January 2007. But we must remember that the 2003 low yield was abnormally low and that therefore the current yield upmove should tend to be in the upper part of the range of historical experience. The 1986-89 umove in yields sent 3 month bill rates up 405 basis points and it is this that we choose as our best guess for the current cycle. Thus we expect the 3 month bill yield to be 4.85% in January 2007 (versus 2.25% currently).
We next turn to estimating the yield trend for the 10 year notes.
In this market we find that the 10 year decennial pattern offers some insight, at least if one is careful to use only those precedents which occur in the same part (currently falling ) of the 60 year interest rate cycle. These then would be the years 1984-87 and 1994-1997 since the 60 year interest rate cycle started downward in late 1981.
In our 2004 bond market forecast we estimated that the bear market in the long bond prices which began from the June 2003 low yield of 4.14% would end sometime between July and September 2004 with the high yield on the long bond between 5.88% and 6.04%. This would have been consistent with the 10 year cycle indications of a high yield sometime between May 30 (1984 precedent) and November 7 (1994 precedent), averaging out to about August 20.
In the event the long bond reached a high yield at 5.60% early in May 2004 and has since moved toward lower yields (currently 4.86%). The 10 year note reached a high yield of 4.89% in May 2004 and now yields 4.27%. For the past several months we have been expecting these markets to surpass the high yields reached in May 2004. However, a more careful analysis of the historical data has convinced us that the start of this move up in yields will probably be delayed until the third quarter of 2005.
The first and most obvious clue is that the 10 year pattern predicts that 2005 will be a bullish year for the 10 year note prices and that yields will probably drop until January 2006 (1996 precedent) or April 2006 (1986 precedent).
The long time period for notes averages 13 years 11 months. The May 2004 high yield was 14 years 9 months from the August 1989 low yield. Adding 13 years 11 months to the October 1990 high yield gives us September 2004 for a predicted yield low. However, after a yield high like the May 2004 high yield associated with the long time period of 14 years the market generally declines in yield for 20 months. This would predict a yield low for January 2006. Unfortunately this would be out of the historical range of the long time period which generally does not last more than 14 years 10 months. We therefore compromise on August 2005 as the predicted time for the low yield in the notes.
We next observe that a low yield in the 10 year notes occcurred in October 1993. Adding 13 years 11 months to this date brings us to September 2007 as a projected high yield for the notes. On the other hand, the 10 year pattern predicts high yields either for August 2006 (1996 precedent) or October 2007 (1987 precedent). The average of these last two projections is March 2007, quite comparable to the January 2007 projected high yield in the 3 month bills. Finally, yield rallies in the notes from important lows (such as the one predicted for 2005 or early 2006) average about 15 months in length thus indicating a yield high around November 2006 (15 months after August 2005).
Putting these observations together we make the following deductions. First the drop in yields from the May 2004 highs is not yet complete and will probably continue until August 2005. This date is 18 months after May 2004 and is as close as we can get to the yield lows predicted by the 10 year pattern (January and April 2006) without violating the range of historical experience for the 14 year period.
From the predicted yield low in August 2005 yields will move up for about 15 months (the average duration during the declining phase of the 60 year cycle) until November 2006. We prefer this to the March 2007 date given by the 10 year pattern because the high in short rates (predicted for January 2007) typically follows the high in 10 year yields during the falling part of the 60 year cycle.
How low will the 10 year yield be at the projected August 2005 yield low? We do NOT expect the notes to drop in 2005 below the 3.07% low yield reached in June 2003. Instead we think the 10 year notes will drop to about 3.60% by August 2005 and then rally to 5.20% by November 2006.
From projected yield highs in late 2006 or early 2007 all the interest rate markets should move towards lower yields for about two or three years. That drop in yields should carry the 10 year notes below 3.00% and end the declining phase of the 60 year interest rate cycle that started in 1981.