A year ago, when the 30-year bond yield stood at 7.5%, precisely the yield we had forecast a year earlier, we predicted a bond yield for the end of 1992 between 6% and 6.5%, the lower number being mine, the higher, David Goldman's. Our premise was that the Fed would continue a de facto locking in of the price of gold at $350 an ounce, removing one major uncertainty from the bond market. Obviously, with the long bond trading at 7.3% with only a week to year's end, we can only congratulate ourselves for getting the direction right; the majority of forecasters a year ago saw economic expansion in 1992 increasing bond yields. Where did we go wrong this year and what do we expect for 1993?
Our error, we are sure, is entirely due to our ascribing to the Fed governors a greater weight to $350 gold than was the case during the course of the year. We remain absolutely persuaded that the $350 price is the one which is closest to the Fed's thinking on optimality. It is still, though, only one of several targets that bear upon the group dynamic of Fed deliberations and voting. The skepticism that remains about gold is causing greater errors in monetary policy than we had anticipated, and a higher bond yield. In 1993, we are now repeating our forecast of a year ago, a range of 6%-6.5%, believing considerable progress is now being made in removing skepticism about $350 gold as the optimum target of monetary policy at the Fed, with more to be made in 1993.
This is not simply wishful thinking. In the past year, the Fed governors and the Fed permanent staff have spent considerable time pondering our hypothesis that the bond market is happiest when gold stands around $350 -- that the bond yield rises when the gold price deviates from $350 in either direction. There has been resistance to the hypothesis because the bond yield of course does not always follow that path, there being other factors influencing the behavior of investors. David Goldman and Evan Kalimtgis have spent literally hundreds of hours at their computers building and refining a model of the bond market that demonstrates the validity of our hypothesis. They have now done so to our satisfaction. Their statistical work is currently being vetted by academic econometricians. It is only a matter of time before their pathbreaking discoveries can be used not only to underpin the proclivities of those at the Fed who sympathize with our views, but also to neutralize those who do not.
We have for some time been able to demonstrate an extremely close correlation between $350 gold and falling bond yields over a fairly long stretch of time, 18 months. We have been thus able to say the decisive factor in the past year's bond market performance is the stability of monetary policy -- and that the market judges this according to the stability of the gold price over the last 18 months. This is not good enough to be conclusive on the $350 price of gold as being the market's best guess of an equilibrium price. For our arguments to be conclusive, the statistical correlations have to be demonstrated over any segment of the period under examination. To do this, David and Evan have had to add sufficient dimensions to their model to incorporate the other major risk premiums that comprise the long-term bond yield -- the premiums that hedge against the likelihood of unexpected future changes in the inflation rate. If all such risk premiums were removed, the 30-year bond would now stand around 4%.
Their model of daily changes in the bond yield measures the way the market discounts risk in three different ways. Each of the three ways shows remarkable correlations over an 18-month stretch and little correlation within segments of that stretch. But when laid against each other, the correlations are not only strengthened over the longer period, but are conclusive during any segment!
The approach is unprecedented in the economic literature. It demonstrates that the market remains on the gold standard whether or not central banks take gold into account. Conventional wisdom states that once gold is demonetized it behaves like any other commodity and ceases to provide a measure of the overall price level. On the contrary, the gold price has provided an excellent predictor of the price level during the past two decades. Like many shibboleths of conventional economics, the notion that gold no longer measures the overall price level is everywhere repeated but nowhere tested. Using monthly data for 1971-91, they demonstrate that the cumulative effect of gold price changes over several years predict the consumer price index.
Of the three variables for which the market discounts bond risk, first it observes the volatility of the gold price over an 18-month interval to judge the probability of arbitrary future changes in monetary policy. The gold price measures the balance between supply and demand for liquidity in the marketplace; if the central bank supplies more liquidity than the market requires for transaction purposes, the market will lay off cash into gold and other stores of value at the margin.
Second, the market observes the deviation of the gold price around its best guess of an equilibrium price, and $350 is the one price cited by Fed governors. When gold deviates from this presumed target during a two-week period, the market perceives an increased risk that the Fed will act arbitrarily in the future. The market will be patient for a few weeks if gold jumps up or down, as this can happen as a result of sudden changes in expectations of Fed behavior. When Saddam Hussein invaded Kuwait in August 1990, gold jumped to $410 from $350, as the markets had been taught the Fed would ease to offset rising oil prices during Mideast disturbances. When the Fed did not do so, the price of gold moved back down. The notion of an "equilibrium price" of gold is tricky. In theory, the general price level (as measured by CPI or PPI) will stabilize over time if the gold price remains fixed at any price level. The best price for gold is the one that causes the market to adjust portfolios to zero inflation with the optimal combination of speed and comfort. If the Fed governors had spoken of $340 or $360 gold, either price would have done about as well. Even those at the Fed who discount gold as a guide to monetary policy, though, will admit to nervousness if its price were to shoot up and remain above $400, or if it were to plunge toward $300.
Third, the market observes the behavior of equity prices. Conventional wisdom and the financial press forever observe that because rising stock prices predict a stronger economy -- and the bond market hates a strong economy -- bond prices fall when stocks rise. This may seem to happen on a spot basis, which gives rise to the myth, but over a stretch of time we see that bond prices rise when stocks rise. A bigger economy will produce greater tax revenues to redeem the bonds. Lower economic growth expectations increase the risk that lower tax revenues in the future will bring about political pressures for the Fed to inflate.
Again, any one of these three explanatory variables seems haphazard on a day-to-day basis. It is their interaction that makes the model perform well. Conceptual breakthroughs in the last several weeks have added to our confidence that we are mining new ore in fathoming the behavior of the bond markets. It is most reassuring that our exhaustive analysis is producing extremely high statistical significance for these variables, corresponding to the hypothesis to which intuition and observation led us. We're also confident policymakers in Washington, especially those who have been inclined to our hypothesis, will soon be reassured as well. We will get to a 6.5% bond yield in 1993 or break our picks trying.