Interest rate behavior is maddeningly difficult to understand, yet it is critical to nearly every area of the economy, including households, businesses, and governments. What determines interest rates and where they are going? If you think you know the answer, think again. Events in the real world regularly demolish what many market participants take to be a profound understanding of interest rate behavior.

Government Financial Activity Doesn’t Set Rates

Whatever determines interest rates (the main subject of this essay), the level of government bond issue, or the lack thereof, and government deficits or surpluses, are not key factors. As the Japanese government approached its fiscal year with a plan to increase by about 10 percent of GDP (nearly $400 billion) the supply of government bonds outstanding, Japanese interest rates on ten-year bonds rose from 0.7 percent to 2.4 percent, then fell back to 1.3 percent as the fiscal year began. All this happened during a time when the Japanese government was contemplating an issue of bonds that pushed the ratio of its annual deficit and total debt to its GDP to levels twice as high as those decried in the United States during the first half of the 1980s. Don’t forget, of course, that while the U.S. fiscal picture in the early 1980s was deteriorating at what was termed "an alarming pace," U.S. interest rates were falling steadily from their highs in the inflationary apex year of 1980.

The U.S. government’s fiscal picture has improved sharply. The Congressional Budget Office has forecast a possible cumulative federal government surplus of more than $1 trillion over the coming decade. That would reduce the stock of outstanding federal debt to 6.4 percent of GDP in 2009, down sharply from 41 percent today. Still, U.S. long-term interest rates have risen from a low of about 4.7 percent last fall to 6 percent this summer.

In an effort to initiate more positive feelings in the U.S. debt markets, the new Treasury Secretary, Lawrence Summers, perhaps frustrated by the rise in U.S. interest rates in the face of such good news on the budget, or maybe taking a page from the Japanese government’s market guidance songbook, announced a plan to have the government buy back part of the national debt. The plan had an odd and ill-timed ring to it because, by definition, a surplus in the federal budget amounts to a reduction in the national debt, just as a deficit amounts to an increase in the national debt. Presumably, what Secretary Summers had in mind was a plan to have the government use taxpayers’ funds to accelerate re-purchases of debt. But since Treasury bonds are not callable, re-purchases of bonds bearing high interest rates would require an extra expenditure of taxpayers’ money. For example, a U.S. Treasury-issue bond due in November 2006 carries a coupon interest rate of 14 percent and a price per thousand of $1,432. Buying it back would raise the budget deficit by $432 for every $1,000 of face value bought back. A premium has to be paid for a bond that yields more than the going interest rate of about 6 percent.

The Treasury’s discussion of bond buybacks is a minor factor when it comes to determining U.S. interest rates. More substantively, based on the broad experience of Japan in the late 1990s and the United States in the early 1980s, "deteriorating" fiscal positions of central governments, where debt and deficits rise rapidly relative to GDP, can coincide with significant drops in interest rates. During the years since 1993, when Japan has put forward larger and larger spending programs and the budget deficit has exploded, interest rates have fallen from their highs of around 5 percent in 1994 to below 1 percent last fall, and they have since risen to about 1.9 percent on signs of economic recovery in Japan. Meanwhile, in the early 1980s, while U.S. bond yields were volatile, interest rates dropped from their highs of around 15 percent in 1980ÿ81 to a far more benign 8 percent by 1986, amid cries of $300 billion budget deficits "as far as the eye can see."

There are two basic reasons why government finances have so little to do with the determination of interest rates. First, borrowing and lending activities by any single government, even a government as large as the U.S. or Japanese government, are only a small portion of the total supply of the world’s interest rate instruments. And second, most of the movement in interest rates is determined by changes in inflation expectations, which, in turn, operate on perceptions about central bank actions to determine short-term interest rates.

Government is not the major borrower in the United States. Corporations are. While total government debt (federal and state) has fallen relative to GDP from a peak of about 65 percent in 1994 to around 57 percent today, corporate debt has risen from 38 percent of GDP in 1994 to over 42 percent today. And since corporate debt is larger than government debt, the stock of total nonfinancial debt to GDP has risen from about 63 percent of GDP in 1993ÿ94 to about 68 percent today. Putting the same thing another way, the sum of U.S. corporate and government debt has grown 5 percent faster than the economy (GDP) during a half decade of rapid GDP growth. Yet interest rates fell during most of that period.

Governments typically do not dominate the supply side of global debt markets, and supply in debt markets does not fully explain interest rates. We must also consider demand conditions in debt markets. Both borrowers and lenders, when they agree on the interest rate on a bond of a certain maturity--say, ten years--are thinking about the conditions that will affect the value of that bond. Those conditions are primarily the return on alternative assets and the expected rate of inflation, because that rate determines the purchasing power of the principal and the interest paid on the bond over its life.

The Fisher Equation and the Real Determinants of Rates

Two factors--the underlying real return on the bond, tied to the real return on alternative assets, and the expected rate of inflation--are the major determinants of the interest rate demanded by borrowers and lenders as they meet in the credit markets. With a large and heterogeneous supply of debt available in the market, on which the maturities range from overnight to upwards of thirty years in the government sector, and upwards of one hundred years in the corporate sector, it turns out that interest rates are far more sensitive to the prospect of inflation than they are to returns on alternative assets like equities, which, in turn, mirror the real returns from investing.

One of the best-known descriptions of the proximate determinants of interest rates is the so-called Fisher Equation, named after the famous economist Irving Fisher, who theorized extensively about interest rates during the first third of this century. The Fisher Equation sets the market or "nominal" interest rate equal to the sum of a so-called real rate of return and anticipated inflation. For precision, the simple Fisher Equation would need to be adjusted for taxes and possible risk premia, but the Fisher Equation in its simplest form is an adequate basis for understanding much of the behavior of interest rates.

Of course, governments can affect the behavior of interest rates in the terms represented by the Fisher Equation to the extent that they can cause changes in expected inflation. The persistence of the belief that larger budget deficits lead to higher interest rates can be traced to cases in history where large budget deficits that have pushed up government debt have forced governments to print money to pay the interest on debt rather than to raise the money by raising taxes. The case of Latin American countries during the inflationary 1980s (and 1990s) comes to mind, as does the case of Russia in the 1990s. Countries with large stocks of debt that cannot be financed by tax revenues have no alternative but to print money, thereby pushing up expected inflation and the rates at which people are willing to lend money denominated in a currency that is expected to depreciate against goods.

The most dramatic cases of high interest rates are associated with the hyperinflationary period of the early 1920s, when the German government, faced with massive reparations payments to the victorious allies, had no alternative but to print money to pay its bills. Of course, by printing money it devalued the currency against goods and thereby caused people to hold less money. The stock of money that citizens are willing to hold constitutes the tax base in an environment where printing money is the means to raise revenue. The inflation rate is the tax rate. As the government pushes up the inflation rate or the tax rate on money holdings, individuals want to hold less and less, and so, to raise an equivalent amount of money, the government has to push up the inflation rate even faster. This is the route to hyperinflation that ends with individuals rushing to spend wheelbarrow loads of virtually worthless government note issue before it depreciates even more rapidly in the next round of inflation.

Once a hyperinflation renders it worthless as a store of value, money can’t serve as a medium of exchange because it can’t store purchasing power over time and space. Such hyperinflationary episodes usually end in a barter economy, in which people elect to use alternative media--like cigarettes or other portable, readily acceptable, fairly standard items--that can serve, however imperfectly, the requirements of a medium of exchange.

Although governments can affect interest rates by affecting the supply of debt instruments and/or inflationary expectations, it is better to view interest rates as part of a broad asset-pricing process: borrowers and lenders come together to agree on an interest rate in the context of considerations about the return on other assets, including equities and real assets, the return on assets denominated in different currencies, and the outlook for inflation. The underlying, riskless, real interest rate is determined by a broad spectrum of factors that set the rate at which it is possible to exchange purchasing power today for purchasing power tomorrow and, simultaneously, the rate at which individuals are willing to exchange purchasing power today for purchasing power tomorrow. In a world where the expected inflation is zero, the rate of exchange between current and future goods (the real interest rate) has typically averaged between 2 and 3 percent. Most variations in interest rates around that level have been determined by changes in inflationary expectations, but the rates may also be altered from time to time by higher expected returns on real investments.

By comparing the movements in market interest rates with the movements in interest rates on instruments that are indexed to inflation (and so protect the investor against inflation by a formula that increases the payment on the bond by the increase in inflation), it is possible to disentangle the movements in interest rates that are a consequence of changes in real returns and those that are attributable to changes in expected inflation. For example, between the start of this year and mid-August, the nominal return on ten-year Treasury notes rose from about 4.6 percent to just over 6 percent. Based on a comparison between straight Treasury notes and indexed notes, which are protected for inflation, the real yield rose from about 3.8 percent to about 4.1 percent, or about thirty basis points. The balance--about 110 basis points in nominal yields--of the total rise of 140 basis points is attributable to a rise in expected inflation from just below 1 percent in January to about 2 percent in mid-August.

It is especially interesting to note that the expected inflation rate has, over the course of the year so far, converged toward the year-over-year CPI inflation rate that has tended also to provide a good proxy as a measure of expected inflation. At the beginning of the year, when the year-over-year CPI inflation rate was about 1.6 percent, the implied expected inflation rate for bondholders earning 4.6 percent on their nominal ten-year Treasuries was about ninety basis points. As the actual year-over-year CPI inflation rate has moved from 1.6 percent to 2 percent, the implied expected inflation rate has moved with it, as have bond yields. It seems that over the course of this year borrowers and lenders have come to believe that the underlying inflation rate has gone from somewhere between 1 percent and 1.5 percent to 2 percent. This belief has been reinforced by a stable year-over-year CPI inflation reading of about 2 percent over the past several months.

The extraordinary behavior of Japanese interest rates is easier to understand with the help of the Fisher Equation. Governments can also affect interest rates with deflationary policies. As the yen appreciated in the fall of 1998 while a global financial crisis emerged, rising Asian deflation and heavy government bond buying pushed Japanese long-term interest rates down to 0.7 percent. When they rose to 2.4 percent by February 1999, the Bank of Japan pushed short-term rates to zero, but the Japanese economy was so weak that deflation expectations remained intact. So rates fell back to 1.3 percent. Subsequent rate rises to 1.9 percent reflect expectations of recovery and reflation in Japan following a surge in first-quarter growth to an 8 percent annual rate.

It is still important to bear in mind that Japan has interest rates on ten-year notes below 2 percent--and presents one of the worst fiscal pictures for an industrial country since World War II. Here, again, government fiscal policy is not the determining factor for Japanese interest rates. It is, rather, the presence of few attractive investment opportunities (low real returns) in Japan, which is still burdened with huge excess capacity. Inflation expectations in Japan are probably close to zero, or about equal to the year-over-year CPI inflation rate. That leaves ten-year notes earning a pure real return of about 1.9 percent--somewhat below the long-term average real return in the 2 to 3 percent range.

While the Fisher Equation helps to explain the past behavior of interest rates, it provides little help to those who must predict the rates’ future behavior. The components of the Fisher Equation--the real rate of return on assets alternative to bonds, and the expected inflation rate--are not directly observable variables. Only by constructing a measure of expected inflation from the year-over-year consumer price index, or from a comparison between market rates and interest rates on indexed bonds, can we decompose interest rate changes into changes in real rates and changes in expected inflation rates. This decomposition can be a useful guide in determining the path of the economy and, in some cases, in providing feedback to policymakers, but it is not a forecasting tool.

The Federal Reserve can affect short-term interest rates by setting the federal funds rate, the rate at which it lends overnight to banks. Short-term rates determine the carrying costs of longer-term interest rate instruments and indicate the Fed’s assessment of where the economy is going. These factors, in turn, may affect people’s calculations of the expected inflation rate or real return. For example, when the Federal Reserve Open Market Committee (FOMC) raised interest rates by twenty-five basis points at their meeting late in June but indicated that they were "neutral" about the future path of the federal funds rate, the bond market rallied, probably on the surmise that the Fed expected inflation to ease. Indeed, the level of expected inflation implied by yields on inflation-indexed Treasury bonds fell from close to 2 percent just before the Fed’s June FOMC meeting to about 1.5 percent by late in July. That was reflected largely in a comparable drop in nominal yields in ten-year Treasuries.

Interest rates can also be a useful guide to an evolving economic scenario when viewed in conjunction with other important economic variables, like exchange rates and stock prices. A good example emerged in the period after a sharp drop in the dollar on July 19. As noted, the Fed’s neutral directive at the FOMC meeting late in June caused U.S. interest rates to fall. Meanwhile, U.S. spending growth continued at a fairly rapid pace, and U.S. stocks continued to rise until July 19, when the dollar fell sharply. The sharp drop in the dollar signaled that foreign lenders were not prepared to provide $25 billion to $30 billion per month in lending to the United States--at interest rates as low as 5.5 percent for ten-year notes--to fund the country’s spending in excess of income. The sharp drop in the dollar that began on July 19 and continued into early August was accompanied by a sharp drop in the stock index that began on July 20 and also continued into early August. Simultaneously, U.S. interest rates rose by about fifty basis points. All these measures were part of an equilibrating process: the market was groping for a combination of stock prices, exchange rates, and interest rates that would clear global markets.

Signs of economic recovery in Europe and Japan created an additional competition for funds outside the United States, and thus added to the upward pressure on U.S. interest rates. A weakening of the dollar pushed up the rate of expected inflation in the United States by shifting upward the demand for U.S. goods by foreigners and by making foreign goods more expensive to U.S. purchasers. It constituted an upward shift of U.S. aggregate demand and supply curves. The resulting rise in expected inflation in the United States simultaneously created the possibility of further Fed tightening and more competition for funds--and thereby put downward pressure on the U.S. stock market.

These adjustments of interest rates, equity prices, and the dollar will continue until markets discover a combination that is consistent with sustainable "clearing" of markets. If interest rates start to fall again and the stock market rises, U.S. spending will rise, and the amount that the United States will have to borrow from the rest of the world will rise as well. If foreign economies grow strong, the terms on which the United States can undertake borrowing from abroad will worsen, and U.S. interest rates will rise.

How Cost-Push Inflation May Affect Rates

We are, as always, brought back to the primary reality that a major determinant of U.S. interest rates, irrespective of government announcements about buybacks and long-ranging discussions about surpluses as far as the eye can see, is expected inflation and its determinants, which include exchange rates, growth, and messages from the Fed. Markets will have to learn some new tricks at the mature stages of a supply-side or an investment-led recovery because they are more accustomed to pricing events associated with the end of a demand-side expansion.

During a typical postwar expansion dominated by the growth of demand, the end comes when demand growth outstrips supply growth and prices of goods and services rise, thereby forcing the Fed to tighten. Although there have been modest signs of rising inflation--the year-over-year CPI inflation rate has gone from about 1.5 percent to about 2 percent over the first eight months of 1999--there has certainly been no surge in inflation. In a supply-side expansion dominated by the rapid growth of investment and characterized by faster growth and lower inflation, such as appeared during 1998, incipient inflationary pressure is more likely to come eventually from the cost side. This phenomenon used to be called cost-push inflation, as opposed to the more typical demand-pull inflation that is associated with too much spending and the higher prices of goods and services.

The U.S. economy during 1999 has begun to display clear signs of accelerating cost-push inflation. We have already mentioned that a weaker dollar entails some cost-push inflation by increasing the dollar cost of foreign goods. It entails a demand-pull element as well by increasing the foreign demand for U.S. goods. Earnings behavior is one place to look for cost-push inflation; the early stages of the production cycle (involving intermediate and crude goods) are another. Average hourly earnings growth is accelerating, especially in the goods sector of the economy. For the three months ending in July, the annualized rise in average hourly earnings is 4.6 percent, up from 3.8 percent in 1998. In the manufacturing sector, the acceleration is even more dramatic. The annualized rate of wage increases in manufacturing in the three months ending in July is 7.5 percent--up sharply from a 1.6 percent pace in 1998. By contrast, wage inflation in the services sector is virtually flat. The employment cost index, the comprehensive quarterly measure of wage inflation released in July, showed a sharp acceleration to a 1.1 percent quarterly rise (up from a 0.4 percent quarterly increase in the first quarter).

Wage growth in the United States started to accelerate after the first quarter of this year. Simultaneously, other pipeline pressures are showing up in the intermediate and crude goods sector that measures price increases of goods used to produce other goods. Intermediate-goods producer price inflation accelerated to a 4.9 percent annual rate in the three months ending in July, up from a negative 3.3 percent rate in 1998. Simultaneously, crude goods inflation reached nearly a 30 percent annual rate in the three months ending in July, after a 16.8 percent rate of deflation in 1998. Clearly, the extraordinarily favorable cost conditions that accompanied last fall’s market panic and weakness in Asia and emerging markets have reversed. Headline inflation numbers like the 2 percent year-over-year CPI number look benign, but less-scrutinized labor cost numbers are beginning to look troublesome.

The fact that the inflation pressures are coming from the cost side in a supply-side, investment-led recovery is not surprising. A high level of investment has pushed up the stock of capital available for labor to work with; in so doing, it has pushed up real wages. As long as the real wage increase is accompanied by a productivity increase, there is no inflationary impact from a rise in real wages associated with a higher capital-labor ratio. However, data on U.S. productivity growth appearing early in August showed that productivity growth dropped to an annualized rate of 1.3 percent in the second quarter, down from rates close to 4 percent in the previous two quarters. With a 5.1 percent annualized growth rate of compensation, unit labor costs began to rise in the second quarter and reached a 3.8 percent annualized rate of increase. That compares unfavorably with a 0.8 percent annualized rise in unit labor costs in the first quarter of 1999 and a drop of 0.4 percent in the fourth quarter of 1998.

Inflation is creeping into the U.S. economy from the cost side, as one would expect after a robust supply-side expansion. Simultaneously, external constraint is beginning to operate on the U.S. economy. Modest recoveries in Japan and Europe in 1999 contrast sharply with weakening conditions in 1998. The recoveries place more demands on global resources, even as they simultaneously stiffen the terms upon which global investors are prepared to lend to the U.S. instead of to Japan and Germany. The latter constraint produces a weaker dollar, which is another source of rising inflationary expectations that can be added to the cost-push pressures currently in the pipeline.

For the near term, these pressures have probably been priced by virtue of the half-percentage-point rise in U.S. long-term interest rates that has occurred since mid-July. But the inevitable sigh of relief, accompanied by a bounce in the stock market and perhaps some short-term drop in interest rates and a bounce in the dollar, will only accelerate the process of overheating the U.S. economy. The seemingly tepid 2.3 percent second-quarter growth rate for GDP doesn’t undermine this scenario. Inventory drawdown reduced the growth rate by a full percentage point, while a fall in net exports contributed close to another negative percentage point. As we move forward, an inventory rebuild is likely, given that consumption growth continued at a 4 percent annual rate during the second quarter.

If the Fed wants to bring the underlying growth rate down by a percentage point, from around 4 percent to between 2.5 percent and 3 percent, 150 basis points of short-term rate increases will be required. It is important to remember that the market rate increases that have occurred since the beginning of the year (and that some say will slow the economy) are largely a consequence of higher anticipated inflation, and hence the expectation of more pricing power. They will not contribute to a slowdown of the economy. The Fed has to push up the rates it controls--the Fed funds rate--by an average of 150 basis points over the course of a year to get the growth rate down by a percentage point. It began on June 30 with an increase of twenty-five basis points and followed with a second increase of twenty-five points on August 24. It still has a good deal of work to do.

John H. Makin is a resident scholar at the American Enterprise Institute.