Tuesday, August 30, 2016

Reprint Of My June 1996 Letter To The US Treasury On The Need For GDP Indexed US Bonds Included In My Comment On Issuing Inflation Indexed Treasury Bonds

The following is a reprint of a letter I sent to the US Treasury on June 5, 1996, in response to its request for comments and information on inflation indexed bonds. At the time, Treasury was considering publishing a proposal on the details of issuing US government inflation indexed bonds. Included about midway through my comment is my still relevant recommendation for the issuance of GDP indexed bonds and bonds indexed to other economic indicators.

If a market for traded US GDP indexed bonds existed today, there would be much more certainty about the trajectory of short-term and long-term future US economic growth

Milton Recht

June 5, 1996

The Government Securities Regulations Staff
Bureau of the Public Debt
999 E Street NW
Room 515
Washington, DC 20220

Dear Sir:

I am writing to comment on the Treasury Department’s proposal to issue a new type of United States Government bond that adjusts for the effects of inflation. My comments relate to the riskiness, structure, and uses of an inflation indexed treasury bond.


One of the purposes motivating the planned issuance of an inflation protected bond is the protection of investors, both individual and institutional, from the risk of an unanticipated loss of purchasing power due to unforeseen inflation over the life of a bond. Expected inflation, as you well know, is already included in the expected yield of the bond. Implicit in the Treasury Department’s attempt at purchasing power protection is the assumption that protection against inflation shocks reduces the riskiness of the investment security.

Volatility or the standard deviation of the total return of a bond is a yardstick for gauging the risk of a potential decline in market value and a capital loss, and while an inflation indexed treasury bond does not yet trade in this country, estimates can be made of its expected volatility. Using data from Ibbotson, Stock Bonds Bills and Inflation 1996 Yearbook, the standard deviation of the total return of a treasury bond from 1929 to 1995 is 9.2% per year. The standard deviation of the CPI for the same period is 4.6% per year. A real return can be derived by removing inflation from the nominal return, and the standard deviation of the real return series can be measured. Due to the negative correlation between inflation and the computed total real return on a treasury bond, the standard deviation of the computed real return, which is the equivalent of an inflation adjusted bond, is higher than a regular treasury bond and is 10.6%. The implication is that an inflation adjusted bond will be about 15% riskier due to increased volatility than a regular treasury bond. This will most likely be true even if historical CPI has been overstated and is restated. Since inflation has a dampening effect on the total nominal return, and since real returns are more volatile than nominal returns, an inflation bond does not protect against purchasing power loss when the increase risk of capital loss is also considered. Marketing the bond as purchasing power protection without mentioning the increased risk of prior to maturity principal loss is misleading to the average investor, and a misleading statement by a mutual fund. The point is particularly important to mutual fund investors since the funds are constantly rolling over maturing debt into new securities and are maintaining a relatively constant average maturity and duration. An inflation indexed bond also will increase the cost of hedging for those institutions that will hold inflation indexed bonds, and that manage their asset liability mix, such as banks, savings institutions, and insurance companies.


If minimizing risk is one of the goals of issuing an inflation bond, than the choice of an inflation index should be partially determine by an objective measure of risk. The selection of an index, whether CPI, PPI, or GDP Deflator, should be based in part on which one of the many indices minimizes the volatility of the total real return of the bond, while providing an accurate measure of inflation.


All publicly traded securities, particularly those that trade in highly liquid markets with the availability of short-selling, contain in their prices the market’s expectations of the future outcomes of the states of the economy over the life of the bond. This concept of expectations theory and efficient market theory has been well studied and documented and generally applies quite well to the treasury market. The market does not make these expectations readily observable other than by changes in the price of the bond or equity.

The issuance of an inflation pegged security by the government is an important milestone for allowing the observation of the market’s forecast of important economic variables, such as inflation expectations, that impact public policy and debate. By comparing a zero coupon treasury bond with the equivalent maturity zero coupon inflation indexed bond, the market’s expectation of average inflation over that period until maturity can be determined. (Estimations of any tax effects can also be made so that they can be removed to get a truer picture of expected inflation.) This expectation of inflation will represent a politically and statistically unbiased best guess of inflation which will incorporate all available information about the future of the economy, which will quickly respond positively or negatively to anticipated policy changes, and which could be an important tool for the setting of monetary policy. It would therefore be useful to have zero coupon inflation bonds of several different maturities outstanding at any time.


When viewed with the perspective that the issuance of an inflation indexed bond is a method of risk reduction and information gathering for policy formation, the fundamental question is whether inflation is the most important economic variable for risk protection and information gathering, or whether another economic variable as an adjustment factor for a treasury bond would serve more useful purposes and be better for the nation. Such a bond could be issued alone or with an inflation bond.

One of the greater problems for the U.S. in recent times has been the inability to achieve sustainable, strong, yet non-inflationary, economic growth. Strong economic growth, as you know, reduces the deficit problem, creates employment, increases real wages, increases tax revenues, and improves consumer confidence and a sense of well-being. The government and the private sector have been stymied by a lack of theory and knowledge about the best courses of actions to take to facilitate and achieve sustainable and strong growth in the economy.

Furthermore, the greatest risk that a business faces is usually that of a downturn in the economy, i.e. recession. A bond that was linked to the general economy, such as a bond whose interest payments were a fixed percentage of concurrently reported GDP, would act a barometer of future recessions and expansions. Zero coupon nominal GDP bonds in conjunction with zero coupon inflation bonds of the same maturities would allow the production of a yield curve of the future yearly expected growth rates of the economy, similar to the yield to maturity curves that are now produced using treasury bond data. The existence of a market in GDP linked bonds would allow short selling and the development of options on the bonds. Either of which could be used by businesses and individuals as hedges to reduce the effects of economic downturns. The information about future growth rates and the changes in the growth rates as policy changes are anticipated could be used by policy makers, academics, and the Federal Reserve to better understand the forces in our economy and to develop strategies for a long-term non-inflationary period of economic growth.

When shocks do occur to the economy, such as the oil price shocks of the 1970’s, the inflationary effects are much greater than would be anticipated by the price increase in the raw material or labor input. In fact, studies have shown that wage increases do not correlate with or predict future inflation, and that economic predictors based on historical or current data, such as the leading economic indicator index, are poor predictors of changes in the growth rate of the economy. Inflation in many cases, as in the 1970’s, represents a resource allocation problem as the economy makes a fundamental structural shift in its productive capacity. Therefore the risk that concerns individuals and the government is more closely aligned to GDP than an inflation index. Bonds linked to GDP would allow hedging of this risk and provide data on the market’s expectation of future economic growth. Reliable data about future economic growth would allow for better production planning and resource allocation which in themselves would substantially mitigate changes in prices and inflationary concerns.


The Treasury should issue inflation indexed bonds, particularly zero coupons bonds of differing maturities, but it should also recognize that the bonds’ volatility and therefore their risk may be greater than regular treasury bonds. The issuance of inflation indexed bonds is an important milestone for the issuance of bonds that can provide valuable information about the future of the economy, and consideration should be given to bonds linked to other economic variables. Bonds with an index linked to GDP may provide greater risk protection to businesses and individuals than inflation indexed bonds. GDP indexed bonds can provide reliable information about the future growth rates of the economy, and the improved accuracy of the information will lead to better production planning, resource allocation, and monetary policy, which will reduce inflationary pressures in the economy.

Thank you for reading my comments on an inflation adjusted bond. If any clarification is necessary, please feel free to contact me.



Milton Recht