The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
,')'.--. "" • '" •• :'. • ',', " ,', '" .. i. March 16, 1984 Comparing Inflation Forecasts rawly defined money, or M1, expanded by 1 3.4 percent from july 1982 to July 1983 -the highest sustained monetary expan.sion since World War II. The growth of M 1 has since slowed, but in the past inflation has tended to follow the path of M 1 growth with an average lag of 2 to 3 years. The high money growth of 1982 and 1983 in this view ordinarily would raise inflation in 1984. Inflation forecasts for 1984 appear to be subject to more than the usual uncertainty. To a large extent, the differences in these forecasts reflect different theoretical approaches to explaining inflation. In one view, the effects of the rapid monetary growth that occurred during 1982 and 1983 dominate. An alternative view stressesthe influence of higher than normal economic slack. According to the latter view, high monetary growth cannot boost the inflation rate unless it puts pressure on wages and prices by reducing the rate of· unemployment. This approach is not necessarily incompatible with the first since the dynamics of the inflationary process are likely to be as depicted in the slack model. Fastermonetary growth should first produce increases in real aggregate demand that reduce economic slack; but as inflationaccelerates because of this reduction in slack, real money balances fall. The resulting reduction in real aggregate demand would then return the level of economic slack to its trend. The only long-run effect of higher monetary growth would be on inflation, but the mechanism of transmission would be temporary movements in slack. This Letter describes these two models and compares their forecasts for 1984. We show that while the two approaches normally forecast inflation equally well, the model emphasi Zing the effects of economic slack is I ikely to provide a better forecast of inflation over the next year or two. Two views The approach to modeling inflation embodied in most large structural econometric models focuses on the effects of economic slack-particularly slack in the market for labor -and of expectations of future inflation. What labor market participants presumably really care about are anticipated real wages. Thus, money wages adjust by increasing faster relative to anticipatedinflation when labor markets are tight than when they are loose, even though the aggregate level of money wages does not movequickly enough to clear the market for labor in any particular year. Since prices are viewed as being primarily determined by a mark-up over unit labor costs, the implication of these models is that realized inflation tends to be higher when labor markets are tight. When labor markets are neither particularly tight nor loose, the inflation rate tends to be equal to that anticipated. If past movements in monetary growth are the dominant determinant of short-run movements in economic slack, as monetarists believe, the monetary approach has an important advantage for forecasting. A forecast of inflation based upon past monetary growth would implicitly embody about as good a forecast of economic slack as can be made. Formal models We have used econometric equations embodying these two approaches to compare their forecasts of inflation. The equations were estimated over the period from 1964 through 1980. To make a clean test of their comparative forecasting powers, the temporary effects of supply shocks unrelated to current demand conditions were removed. This was done by using the implicit price deflator for personal consumption The other view argues that inflation is basically a monetary phenomenon. Nar1 ,·"r,,.,,·:,, 'fc'';:'·''n.'!' Bank of San '·,·"n, ""." of the' [<,(Jeral Board of (;o\,elnol" ----. _-----_. __ . _-. _----_.._--- -, .... _---_ _-----_. .. .... _--, .. _-- - - - - - - _. Forecasts for 1 984 Beyond the period of estimation, the accuracy of these two approaches to forecasting inflation is very different. The economic slack model tracks the decline in the inflation rate during 1982 and 1983 quite well, with an average absolute error of only 0.<,)of a percentage point. In making these forecasts, the slack model's past predictions of inflation were used in the measure of expected inflation, so that the forecasts depend only upon movements in slack. In contrast, the monetary model overpredicts inflation quite badly in 1 982, with an average absolute error of 1.7 percentage points. Theerrorin 1983 isevenworseat4.0 percentage points. For 1984, the two forecasts continue to diverge, with the monetary model forecasting a 9.2 percent inflation rate from fourth quarterto fourth quarter and the slack model predicting 6.3 percent. expenditures, excluding food and energy, for the measure of prices. In principle, movements in the real exchange value of the dollar could affect consumer prices in a way similar to supply shocks. But statistical tests indicated that this effect was not important in the period of estimation. The economic slack model of inflation contains measures of slack and expected inflation. The central tendency of inflation is the rate of inflation anticipated. Variations in inflation around this central tendency are captured by movements in the slack variable. For the slack, the current unemployment rate for males in the 25-54 age bracket was used. The normal, or non-cyclical, rate of unemployment in this measure has been less affected by demographic shifts over time, making it preferable to the total unemployment rate. Expected inflation is measured by past changes in the price index over the previous 16 quarters. For the monetary equation, M 1 growth of 6 percent, equal to the mid-point of the Federal Reserve's current target range, was assumed for 1 984. The forecast from the slack model is based on the 0.8 of a percentage point reduction in the national unemployment rate for 1 984 predicted by a sample of forecasters polled by the American Statistical Association and the National Bureau of Economic Research, and on the historical relationship between changes in the unemployment rates for the total labor force and males of prime age. Interestingly, this sample of forecasters predicts a 5.4 percent inflation rate for 1 984 (measured by the G N P deflator)-much closer to the forecast of the slack model than that of the monetary model. The monetary model of inflation simply contains current and past monetary growth. It is thus a"reduced-form" relationship that leaves the transmission mechanism relating money to prices implicit in the lag structure. Current and lagged changes in M 1 growth over 16 quarters were used. Within the period of estimation (a portion of which is:shown in the chart), the tWb approaches predictthe inflation rate equally well. The average difference between the actual annualized inflation rate in anyone quarter and the predicted value is 0.7 of a percentage point in each case. Also, to the extent that economic slack affects inflation, its influence appears to have already been captured by past monetary growth within the sample period. This is indicated by the fact that when the unemployment rate is included in the monetarist equation, which contains current and past monetary growth, there is no significant reduction in the prediction error. The slack model is likely to provide a better inflation forecast for 1 984 than the monetary model because it was more accu rate in 1982 and 1 983 and because it produces an inflation forecast closer to the current consensus. Although the monetary model normally has the advantage of not requiring an indepen- 2 - - _ _--- .. .. ...__ •. ... __ __ Inflation Forecasts Deflator for Personal Consumption Expenditures· 12 10 8 6 4 2 Period Forecast Period o" - _ ....... _- - ' - _........ . . Ii. . . . -_L-_. 1978 1979 .. term positive growth trend of 3 percent. As a result, economic slack is now higher than cou Id have been predicted on the basis of prior monetary growth. Moreover, most forecasts for 1984 indicate gradual reductions in slack that would not usually be enough to generate significant increases in inflation. Adrian W. Throop dent forecast of economic slack, forecastsof slack based purely on monetary considerations have not fared well recently due to declines in the income velocity of money, or its rates of turnover. Between the fourth quarter of 1981 and and the first quarter of 1 983, the velocity of M1 dropped at a 5.5percent annual rate, compared to a long- Estimation ..... 1980 1981 1982 . I-_- L. _--I 1983 1984 *(Annualized Percent Change; Food and Energy Excluded) 3 IIPMPH .'. 4Pln • PIUJOjIlP:) • • PPPA<3N • 04 PPI PUOZIJV • P>j5PIV Jr CD) \ill@\\llil Jrw:<d1 @@ 1S{l;:] JrW:@,§@CQ[ BANKING DATA-TWELFTH FEDERAL RESERVE DISTRICT ( Dollar amounts in millions) Selected Assetsand Liabilities Large Commercial Banks Loans, Leases and Investments 1 2 Loans and Leases1 6 Commercial and Industrial Real estate Loans to Individuals Leases U.S. Treasury and Agency Securities 2 Other Securities 2 Total Deposits Demand Deposits Demand Deposits Adjusted) Other Transaction Balances4 Total Non-Transaction Balarices6 Money Market Deposit Accounts-Total Time Deposits in Amounts of $100,000 or more Other Liabilities for Borrowed MoneyS Weekly Averages of Daily Figures ReservePosition, All Reporting Banks Excess Reserves ( +)/Deficiency (- ) Borrowings Net free reserves (+ )/Net borrowed( -) 1 Amount Outstanding 2/29/84 Change from 2/22/84 177,701 157,394 46,262 59,229 26,950 5,006 12,187 8,119 185,994 44,236 28,698 12,004 129,752 2,209 2,267 449 66 125 3 37 - 20 2,020 1,824 1,225 73 122 40,373 94 38,085 20,222 Weekended 2/29/84 - - Change from year ago Dollar Percent - - - - 1,676 2,039 299 330 299 56 319 44 5,002 5,000 2,633 770 768 - - 776 - 11 716 79 2,784 Weekended 2/22/84 NA NA NA - NA NA NA 4.8 6.6 3.3 2.8 5.6 5.6 12.8 2.7 13.1 50.8 42.0 30.1 3.0 9.8 - 1.0 60.5 Comparable year-ago period NA NA NA Includes loss reserves, unearned income, excludes interbank loans 2 Excludes trading account securities u.s. ) Excludes government and depository institution deposits and cash items 4 ATS, N OW, Super N OW and savings accountswith telephone transfers s Includes borrowing via FRB, TI&L notes, Fed Funds, RPs and other sources 6 Includes items not shown separately Editorial commentsmay be addressedto the editor (GregoryTong)or to the author .... Freecopiesof Federal Reservepublications can be obtained from the Public Information Section, FederalReserve Bank of San Francisco,P.O. Box 7702, SanFrancisco94120. Phone(415) 974-2246.