Prepared by

The Northern Trust Company

Two weeks ago in our regular monthly economic and interest rate forecast piece, we predicted that the Fed would stand pat at the September 30 FOMC meeting. Nothing we have seen or heard since has led us to change our minds. So, rather than repeating our arguments in this pre-FOMC meeting commentary, we have decided to share with you some research we have been conducting of late pertaining to the declining inflation rate in recent years. Although research always is a work in progress, our findings to date–in the form of an econometrically-estimated inflation forecasting model–suggest that so-called old paradigm explanations have done an outstanding job in forecasting the behavior of core consumer inflation in this expansion. Moreover, the old paradigm is now suggesting that core consumer inflation will start to drift up in 1998. This is not to say that there are not new elements in this business cycle. These new elements, however, seem elusive when it comes to measurement. Rather, they have tended to be after-the-fact anecdotes. If we can't quantify or predict new paradigm elements, how do we know when they will have run their course? Conversely, if we can quantify old paradigm elements and they are currently doing a good job of explaining the economic world, then we will have a better idea when the trend is likely to change.

Before presenting our inflation forecasting model, we want to give you a flavor of the economic theory underlying it. Although economists agree on very little, one thing that most of us will grant is that inflation is a monetary phenomenon–the result of too much money chasing too few goods. Another thing that most economists agree on is that the lags between changes in money growth trends and inflation trends are measured in years. Despite these presumed areas of agreement, very few economic analysts, in their discussions of recent inflation behavior, have mentioned that the broad money aggregates grew at their slowest sustained rates in the early 1990s and now are approaching growth rates closer to postwar norms. This is illustrated in Chart 1 in which the year-over-year percent change in the annual averages of M3 are plotted. The median year-to-year growth in M3 from 1960 through 1996 has been 8.4%. In contrast to the 1971 to 1985 period, when M3 growth was typically well above this median, the 1986 through 1996 period is characterized by consistently below-median M3 growth. Not only was M3 growth below the median in these 11 consecutive years, but the 1990 through 1994 period contained the slowest M3 growth in the 37-year period charted. What accounted for this unusually weak money growth in the early 1990s? The same thing that accounts for the continuing unusually weak money supply growth in Japan today–a banking system unable to extend credit because of huge loan losses. M3 growth has been trending up consistently since its 1993 nadir. In the first eight months of 1997, M3 is up 7.6% versus the comparable 1996 period. If this growth rate holds up over the remainder of the year, M3 growth in 1997 will have been the highest since 1986's 8.3%. We can't say a priori whether there has been too little money growth, relatively speaking, since 1985, but we can say that absolutely money growth has been weak compared to an historical norm.

As alluded to above, inflation results when money growth is rapid relative to the available supply of goods and services, or the potentially available supply. So, to get a sense as to whether M3 growth since 1985 has been weak, not only absolutely, but relatively, as well, we have compared it with the Congressional Budget Office (CBO) estimates of potential GDP growth. In Chart 2, we have plotted the variable we call “excess M3”–M3 growth minus potential GDP growth. Starting in 1987 and continuing through 1996, “excess M3” has been below its median of 4.8%. The weakest sustained period of “excess M3” behavior occurred in the four years ended 1994. If M3 growth holds to its first eight-month pace over the remainder of this year, 1997 “excess M3” growth will come in around 5.2%–its fastest since 1986's 5.4%. So, in recent years, not only has the absolute growth in the money supply been low, but the relative growth in the supply of money also has been slow. In 1997, however, it appears that an excess supply of money is developing.

*yr./yr. percentage change in M3 minus yr./yr. percentage change in potential GDP

Another relative measure that might have some implications for the behavior of inflation is where real GDP is in relation to its potential. Money might be growing relatively rapidly, but if the level of GDP is far below its potential level, inflationary pressures might be slow to develop due to the slack in resource usage. Conversely, money growth might be relatively slow, but if GDP is far above its potential level, inflationary pressures might mount quickly. Plotted in Chart 3 is the actual level of real GDP as a percent of the CBO-estimated level of potential GDP–the GDP gap. As can be seen in this chart, the GDP gap in 1996–the sixth year of the business expansion–was 100.4%. This was lower than the reading of 100.7% in 1988, six years into that expansion. And the 1996 GDP gap fell well short of the 105.3% registered in 1966, again six years into that expansion. So, in this current expansion, money has been growing relatively slowly, and it was not until 1996 that GDP even got up to its potential. But using our estimates of 1997 real GDP–which assumes second half annualized growth of 3.2%–and using CBO's estimate of potential GDP, the 1997 GDP gap would be 101.8%–the highest since 1973's 103.3%. In contrast to earlier in this expansion, money is now growing considerably faster and the economy is pushing up against capacity constraints as measured by the GDP gap.

*actual level of real GDP as a percent of CBO-estimated potential.

Another factor that might have some bearing on the behavior of inflation is the behavior of the dollar. It is generally thought that a stronger dollar, all else the same, will retard inflation by making imported goods less expensive in dollar terms, by constraining domestic producers of import-competing goods from raising their prices and by relieving pressures on domestic resources due to a weakening in export demand. Year-to-year percent changes in the trade-weighted dollar are plotted in Chart 4. Compared to the first half of the 1980s, the year-to-year changes in the dollar have been small.

Using lagged values of these theoretically-inflation-related variables–“excess M3,” the GDP gap and changes in the trade-weighted dollar–we have estimated a model to forecast the year-to-year changes (in basis points) in the annual inflation rate as measured by the GDP deflator for core personal consumption expenditures. The reason we used the deflator rather than the core CPI to measure inflation is that the core CPI is not a consistent series through time. In recent years, the BLS has made changes in its calculation of the core CPI without incorporating these changes in prior years' calculations. We wanted a consistent series. Because we wanted to know if these “old paradigm” variables could have forecast that inflation in 1996 or 1997 would be lower than it was in 1990 or 1991–a first for a postwar business expansion–we only estimated our model through 1989, and then let this estimated relationship forecast beyond 1989. Had we estimated the model through 1996, we would have been stacking the deck in favor of the old paradigm. Chart 5 contains the actual versus forecast changes in inflation. Based on this old-paradigm relationship between the explanatory variables and changes in core inflation, our model not only forecast that inflation would be falling in the 1990 through 1996 period, but its forecasts of inflation declines were even larger than what actually have occurred. Advocates of the new paradigm say that traditional estimates of potential GDP, such as those of the CBO, are too low because of underestimates of actual productivity growth. And, of course, explanations of anything in terms of the behavior of the money supply have been relegated to the dust bin of economic history. Yet, these two integral elements of our model–CBO estimates of potential GDP and M3 growth–produce estimates of core inflation changes that fit the actual changes almost as tightly as the glove O.J. tried on at his trial.

According to our model, what lies ahead for core consumer inflation? Our model is forecasting that in 1997, core consumer inflation in terms of the personal consumption deflator will be up 10 basis points; in 1998, up 34 basis points. These, of course, are not major moves. But the model does suggest that, at the very least, inflation is more likely to stabilize rather than fall further. The model suggests that the risks ahead are tilted more toward higher inflation than toward deflation.

Note: lagged explanatory variables are growth rates in M3, potential GDP and trade-weighted dollar, levels of GDP and potential GDP, and a constant.

In sum, old paradigm explanations for the behavior of inflation in recent years seem to be more than adequate. This is not to say that there is no validity to new paradigm arguments. But the old paradigm seems to be alive, well, and, in contrast to the new paradigm, measurable. The old paradigm says that the inflation risks are tilted to the upside. From comments made by various Fed officials–namely Governor Meyer and Richmond Fed President Broaddus–we think that there is a growing bloc on the FOMC that sees the inflation risks as our model does. This bloc wants to be preemptive. That's why we see the risks tilted up that the FOMC will act to raise the funds rate by 25 basis points before year- end, but not on Tuesday, September 30.

September 25, 1997Paul L. Kasriel and Asha G. Bangalore

The Northern Trust Company

50 South LaSalle Street, Chicago, Illinois

Consensus National Futures and Financial On Line Index

Added to the WWW 09-26-97
Last updated on 09-27-97

Hosted by:
One Crossroads Place
610 West Maple Ave, Suite WWW
Independence, MO 64050
(816) 252-4080
sysop@kcmo.com

wmeubank@ocp.kcmo.com