The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
FEDERAL RESERVE BANK OF RICHMOND MONTHLY REVIEW '^Forecasting Accuracy in the Sixties Survey of Time and Savings Deposits The Fifth District Forecasting Accuracy in the Sixties F or all the criticism levied upon it by skeptics, economic forecasting nevertheless plays an important role in determining both government and business policies. One measure of its importance is the mil lions of dollars spent annually in efforts to specify, however roughly, the economy’s future course. A n other is the growing amount of professional time and talent in attempts to improve forecasting tech niques. The development of the computer has given sharp impetus to these latter efforts, and in recent years advanced econometric techniques, coupled with computer simulations, have been brought to bear on the problem. Today, literally hundreds of models are used to predict future economic conditions. Many of these are intricate multiple equation econometric models, while others represent less formal applica tions of professional judgment. One approach to the task of improving economic forecasts is through a comprehensive evaluation of forecasting accuracy. The National Bureau of E co nomic Research, with financial support from some leading industrial firms, has produced extensive literature on techniques for evaluating the validity of economic forecasts. It is through careful and systematic evaluation that forecasting errors and biases can be brought to the foreground. Although forecasting techniques have improved since the early fifties, the most casual examination of forecasting performance reveals a need for further improvement. This study focuses on the accuracy of short-term forecasts of two important economic variables, gross national product (G N P ) and the consumer price index (C P I ) . The data for this evaluation were collected from Business Forecasts, published annually by the Federal Reserve Bank of Richmond, and “ Predictions,” prepared annually by the Federal Reserve Bank of Philadelphia. These publications summarize the annual forecasts of lead ing business firms, educational institutions, research organizations, and individuals. The evaluations that follow are based on the annual changes in G N P and the C PI predicted by forecasters whose efforts are summarized in these publications. Although there is some disagreement concerning the best way to evaluate forecasts, it is preferable in most instances to convert absolute level forecasts into predicted changes before evaluation. Since predic tions are made at different times before revised figures are available, the actual levels of the variables at the time of the forecasts are not known. The evaluation should therefore be based on the accuracy of predicting the changes in the variables rather than predicting the levels themselves. Statistical Concepts O ne of the m ost w idely used concepts for depicting the accuracy of forecasts is the prediction-realization diagram. This diagram shows the actual change in the variable plotted against the predicted change in the variable. Per fect forecasting would be represented by a 45 degree line through the origin. This 45 degree line is called the line of perfect forecasts ( L P F ). The predictions of all forecasters can be plotted on the same chart and visualized in comparison with the L P F (see charts). Particular importance is also given to the mean point, i.e., the point in the prediction-realization dia gram, the coordinates of which are the arithmetic means of the predicted and actual changes. If the mean point lies on the L P F , the average predicted change over the entire time span is equal to the average actual change over the same time span. If the means are significantly different, the mean point will not lie on the L P F and the forecasts are said to be biased; i.e., the forecasts consistently under estimate or overestimate the actual changes. Annual Changes in G N P T h e prediction-realiza tion diagram for G N P indicates that forecasters have been quite inaccurate in predicting annual G N P changes during the past decade. This diagram and the accompanying table analyzing forecasting results show that the range o f predictions for all forecasters included the actual change in only four of the ten years (i.e., the range of annual forecasts represented by the horizontal scatter of points actually crosses the L P F ). In two of these years, the actual change was near one of the extreme points in the range. Another method of showing the inaccuracy is by comparing the mean of the predicted change with the actual change. In only two years, 1962 and 1967, was the mean close to the actual change. Further more, the variation of observations around the L P F was large in all years except 1962 and 1967. One hazard of forecasting changes in data having upward trends is the tendency to underestimate the A C C U R A C Y STATISTICS FOR SELECTED FO RECASTS OF A N N U A L C H A N G ES IN GNP 1960-1969 (Billions of Dollars) Range of Predictions for A G N P M ean Predicted AGNP M ean Actual AGNP 1969 26.0 - 66.0 52.8 66.7 1968 32.0 - 64.0 54.1 72.2 1967 29.0 - 57.0 45.8 43.6 1966 27.8 - 50.0 41.2 65.0 1965 25.0 - 40.0 33.2 52.5 1964 23.7 - 45.0 32.6 41.9 1963 8.0 - 28.5 17.8 30.2 1962 34.0 - 55.0 41.1 40.2 1961 3 .4 - 1 6 .0 7.9 16.3 1960 1 7 .5 -3 0 .0 24.4 19.9 3.4 - 66.0 39.2 50.3 1960-1969 ‘ Prelim inary estim ate. G RO SS NATIO NAL PRODUCT P R ED IC TIO N -R EA LIZ A TIO N D IAGRAM , 1960-1969 (Billions of Dollars) changes.1 Underestimation bias is clearly seen in this analysis. Seven of the ten means of the pre dicted G N P changes were smaller than the actual changes, that is, the means of the annual predicted changes were to the left of the L P F . Significant overestimation of the predicted change occurred only in 1960. Annual Changes in the CPI Forecasters have had almost as little success in predicting price changes as they have in predicting G N P changes. Again, an analysis of the accompanying diagram and table indicates that predictions were approximately correct in only two years, 1963 and 1964. In five years, predictions were underestimated and in three years they were overestimated. The tendency was to overestimate small changes and underestimate large changes. Dispersion around the L P F was relatively large. Since variations in price increases of one or two percentage points may have widely varying policy implications, the annual predictions of the forecasters would appear to be less accurate than is desirable for policy purposes. Results for the Decade Forecastin g accuracy for the entire decade can be measured by basically the same methods that were used to determine the accuracy in predicting the annual changes. In the case of perfection, all points in the predictionrealization diagram would lie on the L P F . In the diagrams for G N P and the C PI, the scatter of points in general does not fall on the L P F . If a line were constructed through the scatter of points connecting the midpoint of the range o f values for each year, the constructed line would be nonlinear. Nonlinearity of the scatter indicates different degrees of accuracy at different levels of actual changes. The scatter of points on both the G N P and the C PI dia grams tend to lie farther from the L P F for large actual changes in the variables than for small actual changes. In each case, there is a definite tendency to underestimate large changes. Another characteristic of the underestimation of changes is the divergence of the mean of the actual changes for the ten year period from the mean of the predicted changes for the same period. The concept of bias, as previously discussed, refers to the inequality of the two means. For the entire period, the mean actual yearly change in G N P was $50.3 billion and the mean predicted yearly change was $39.2 billion, indicating a substantial under 1See Jacob M incer and V ictor Zarnow itz, “ The Evaluation of Eco nomic Forecasts,” Jacob Mincer (e d .), Econom ic Forecasts and Expectations, (N e w Y o r k : N ational Bureau o f Economic Research, Inc., 1 9 6 9 ), pp. 3-46. estimation bias in predicting annual G N P changes for the decade. The mean actual change in the CPI was 3.2% per year and the mean predicted change was 2.5% per year, again indicating a large under estimation bias in predicting annual C PI changes for the ten year period. A least-squares straight line was fitted to the prediction-realization diagrams for G N P and the CPI. This line, which is the straight line that fits the data better than any other straight line, is shown on each diagram by a broken line. If all predictions were perfect and thus all points fell on the L P F , the leastsquares line would be identical to the L P F in each diagram. The intercept would be zero and the slope of the line would be equal to one. Further more, since all of the points fall on a straight line, there would be perfect correlation of actual and predicted changes. The least-squares line for the G N P and the CPI forecasts did not correspond with the L P F in either diagram, indicating bias. Since unbiased forecasts are more accurate than biased forecasts if the distances between the points in the diagram remain constant, the forecasts can be made more accurate by correcting for the bias. Graphically, removal of the bias can be accomplished by a parallel shift in the least-squares line until it intersects the mean of the actual changes on the L P F . The same results are achieved by subtracting an amount equal to the size of the bias from each point on the leastsquares line. In the G N P diagram, the constant is the difference between the mean of the actual changes and the mean of the predicted changes, or $11.1 billion. The constant for the CPI diagram, found by the same method, is 0.7% . If the economy con tinues to expand and prices continue to rise in a pattern similar to that of the past decade, future re sults could probably be improved in the long run if the forecaster raises his estimates by an amount equal to the size of the bias. Composition of Forecasts R em oval of the bias in forecasting will, in most instances, improve the accuracy of predictions. However, correcting the final results does not indicate the sources of the bias. T o improve the forecasting model, it is often useful to determine the source and magnitude of the forecasting error. Errors in aggregate forecasts are usually the result of individual errors in the various components de termining the aggregate variable. Some of the in dividual component errors are reinforcing and others are offsetting. T w o different forecasts, each of which yields the same aggregate results, could have widely varying policy implications. In general, a forecasting technique that yields two small reinforcing errors in the component parts is superior to a technique that has one large positive error and one large negative error yet gives the same aggregate results. In fact, many policy makers would prefer a technique with small errors in the various elements even though the A C C U R A C Y STATISTICS FOR SELECTED FO RECASTS OF A N N U A L C H A N G ES IN CPI 1960-1969 (Per Cent) Range of Predictions for A C P I M ean Predicted ACPI M ean Actual ACPI 1969 2.8 - 4.0 3.3 5 .4 * 1968 2.8 - 4 .3 3.3 4.2 1967 2.5 - 3.7 3.2 2.8 1966 1.2 - 2.5 1.9 2.9 1965 1 .2 - 2 .0 1.5 1.7 1964 1 .0 - 2 .0 1.4 1.3 1963 0 .7 - 1 .5 1.1 1.2 1962 1 .0 - 2 .5 1.6 1.1 1961 0.6 - 2.0 1.4 2.4 1960 1.4 - 2.0 1.8 1.5 1960-1969 0.6 - 4.3 2.5 3.2 ‘ Prelim inary estim ate. CON SUM ER PRICE INDEX P R ED IC T IO N -R EA LIZ A TIO N D IA G R A M , 1960-1969 (Per Cent) error in the aggregate measure was larger than that given by another method. For example, in 1969, the most accurate aggregate forecast in this sample pre dicted prices to increase by 3.6% and real G N P to increase by 4.8% . Actually, prices increased more than 5% and real G N P growth was less than 3% . Since inflation was the major problem confronting policy makers, those forecasts predicting price in creases of 4.5% to 5.0% might have been more use ful. O f course, the consequence of the trade-off between accuracy in predicting the aggregate and accuracy in predicting the elements is a matter de termined by the use of the forecast. Certainly, one major part of the total forecasting error in G N P predictions has been the error in pre dicting prices. Earlier, it was shown that G N P fore casts were relatively accurate in 1962 and 1967, and that the forecasters overestimated the G N P change only in 1960. A n examination of the price statistics indicates that these were the only years in the decade when the predictions for the C PI were significantly greater than the actual changes in the C PI. Inac curate projections for prices accounted for approxi mately one-third of the overestimation in the average 1960 G N P forecast. The 1962 and 1967 G N P fore casts remained reasonably accurate since the figures, adjusted for price inaccuracies, merely changed from a slight overestimation to a slight underestimation of the actual changes in the two years. In the other seven years, both the real changes in G N P and the changes in the C PI were underestimated. Correc tions for inaccurate price predictions would still result in a significant underestimation of real G N P growth. Another major part of the total forecasting error has been the cumulative error of predicting quarter to quarter changes. Most of the G N P forecasts are made in the third or fourth quarter of the preceding year before final data for that year are available. Errors from inaccurate estimates of base period data may cause cumulative errors in the quarters ahead. Other studies have shown that forecasting errors in crease with the length of the predicted time span. In trend dominated series, such as G N P growth, in creasing reliance on the historical trend will often eliminate some of the downward bias and result in more accurate long-term forecasts. S u m m a ry F orecastin g o f econ om ic aggregates has improved since the Korean W ar. Recently de veloped models are now able to incorporate intricate economic relationships that were “ assumed away” before the era of high-speed computer technology. Data are now available with more accuracy, in greater detail, and at earlier dates than twenty years ago. However, with all these improvements in data and technology, forecasting economic aggregates beyond one or two quarters is very difficult. E co nomic relationships are difficult to determine for the near future and become increasingly complex over longer periods of time. Unforeseen changes in fiscal and monetary policy add to the uncertainty of future events. Nevertheless, the forecaster, as complicated as his task may be, can improve his long-run accuracy. An analysis of annual forecasts for the decade of the sixties indicates a clear tendency to underesti mate changes in G N P and the CPI. Since G N P projections were made in current dollars, under estimation of price changes accounted for part of the error in predicting G N P. However, a distinct down ward bias remained. Since projections for trend dominated series generally contain substantial bias, forecasting accuracy can be improved by greater use of trend projections to reduce the downward bias. Clyde H . Farnsworth, Jr. Survey of Time and Savings Deposits In the early postwar years, bankers displayed a rather neutral attitude toward time and savings deposits, accepting such deposits as were offered but not aggressively seeking them. For the past decade, however, bankers have been more aggressive in their competition. In so far as permitted by the Federal Reserve’s Regulation Q, larger banks, at least, have tended to keep rates paid on time and savings de posits competitive with rates paid by other inter mediaries and with rates available on market instru ments. In addition, they have ingeniously designed many new types of time deposits in order to appeal to diverse public preferences and have advertised their new wares extensively. Reflecting this aggressive attitude, time and sav ings deposits have tended to expand rapidly during periods when banks were able, under Regulation Q ceilings, to compete effectively with market rates. Conversely, outstandings have tended to decline, often precipitously, when Regulation Q ceilings foreclosed effective competition, as was the case throughout 1969 and in early 1970. Under the im pact of strong credit demands and restrictive mone tary policy, market interest rates rose to extremely high levels relative to the Q ceilings, and attrition of time and savings deposits became massive. The ebb and flow of time and savings deposits has important implications for monetary policy. Since the behavior of such deposits depends on the interaction of market interest rates, Regulation Q ceilings, and banker competitiveness, the Federal Reserve has attempted to learn more about the structure and functioning of the market for time and savings deposits. T o obtain timely information, the Federal Reserve System in cooperation with the Federal Deposit Insurance Corporation instituted a quarterly survey of time and savings deposits which provides data on amounts outstanding and rates paid. The survey conducted in the fall of each year covers all insured commercial banks. Information for the other quarters is derived from a sample containing all such banks having $20 million or more of time TABL PERCENTAGE C H A N G ES IN TIME at 84 Fifth C October 3 1 , 1 969 tc C ertificates of Deposit $100,000 c Total Fifth District Num ber of Banks Savings Deposits 84 - 1 .0 - 2.2 - 1.2 7.0 57 27 - 0 .9 - 1 .0 - 0.9 2.4 - 0.2 1.5 - 1 3 .6 8.1 7 22 10 3 34 8 - 5 .1 0.4 0.4 - 1 .1 - 0 .5 - 4 .1 - 0.2 3.1 8.0 1.6 0.2 4.1 - 2.8 0.8 2.9 4.3 0.5 4.7 20.1 26.2 - 2 6 .0 none 6.7 - 6.2 24 12 4 10 5 3 3 - 3 .3 0.2 - 1 .1 1.1 1.0 1.0 - 8 .6 0.3 - 2.7 - 1 1 .9 - 0.4 2.5 - 1 4 .6 - 3.0 - 2.8 3.0 8.9 2.9 2.6 - 1 6 .8 1.9 21.5 6.1 - 2 4 .7 17.2 5.3 0.3 - 6.3 Total Under $100,000 N egotiable Deposit Size of Bank: Less than $100 million $100 million and over By State: District of Colum bia M aryland North C aro lin a South C aro lin a V irgin ia W est V irginia - - By SM SA: W ashington Baltim ore C harlotte Richmond N orfolk C harleston, W est V irgin ia Roanoke *lnclu des C hristm as Clubs and sim ilar accounts. 6 and savings deposits of individuals, partnerships, and corporations, plus a selected number of smaller banks. The most recent universe survey was conducted on October 31, 1969, and the latest sample survey on January 31, 1970. Between these dates, on January 21, the Board of Governors of the Federal Reserve System amended Regulation Q permitting banks to raise rates on most classes of time and sav ings deposits. Because of this regulatory change, the recent sample survey is of special interest. This article describes the results of the January survey of 84 Fifth District member banks and makes com parisons with data from the same 84 banks in October. pull large quantities of funds into the banking sys tem and lead to an upsurge in bank lending. Table I shows in detail the percentage change in savings deposits and the various classes of time deposits between October and January at banks in the Fifth Federal Reserve District. Declines in sav ings deposits were quite general throughout the District at both large and small banks. Since sav ings deposits represent about three-fifths of total time and savings deposits at Fifth District banks, the 1% decline at the sample of 84 banks was quite significant. In absolute terms, savings deposits de clined about $30 million, almost exactly equal to the decline in total time and savings deposits. Hence, for all banks in the sample, time deposits remained roughly unchanged with increases in some categories Continued Attrition A m ou n ts outstanding of being matched by declines in others. savings deposits and most types of time deposits con Interestingly, negotiable C D ’s of $100,000 or more tinued to decline between October and January. at Fifth District banks increased 7% even though The change in Regulation Q came too late in the rates on such deposits were below yields on market period to affect outstandings to any appreciable ex instruments. This contrasts with a 9.2% nationwide tent, and in any case, the new ceilings remained well decline in large C D ’s outstanding. below market rates. The Board did not want to deposits open account, only consumer-type deposits permit banks to raise rates to levels which would increased, presumably because of the proliferation Am ong time 9KM ei AND SA V IN G S DEPOSITS, IPC istrict Banks Jan u ary 3 1 , 1 9 7 0 Time Deposits, O pen Account Sp ecial* Consum er-type O ther under $100,000 :nd over Nonnegotiable Total and over Total Time and Savings Deposits O ther $ 100,000 - 1 9 .6 7.1 - 1 0 .8 30.6 - 3 5 .0 - 3 0 .6 - 0 .6 - 1.7 - 2 1 .7 13.6 5.9 - 2 9 .7 - 9.4 28.5 31.3 - 3 1 .6 - 3 5 .5 - 9.6 - 3 6 .3 0.5 - 0 .8 - 2 5 .2 - 1 0 .4 - 1 8 .8 - 1 6 .5 2.7 - 7 8 .0 3.7 - 1 0 .8 4.4 21.1 15.2 23.2 14.5 - 2 8 .4 - 4.2 - 2.4 - 1 4 .8 - 3 0 .2 4.6 6.5 7.6 none 37.9 35.3 - 9 3 .5 25.5 4.7 none - 1 7 .5 none - 4 3 .8 - 7 0 .2 - 1 5 .0 none - 2 1 .3 none - 2 .5 - 0 .1 - 2 .2 0.6 1.0 - 0 .8 - 2 3 .4 - 8 6 .8 - 1 1 .9 - 3.1 - 1 9 .5 - 7 6 .3 - 7 6 .0 5.9 - 1 4 .5 4.2 3.9 - 5.2 2.7 85.3 9.8 - 2 7 .5 2.7 - 1 6 .9 - 3.6 - 2 3 .8 66.0 76.5 none 4.7 28.3 13.0 7.2 87.0 - 7 3 .4 37.2 6.6 - 6.8 - 1 4 .0 none none - 5 1 .4 none - 1 6 .4 6.3 - 3 7 .9 none none - 1 .3 - 1 .1 - 3 .9 0.8 1.2 - 0 .4 1.8 7 of such deposit accounts and the fact that they have been highly advertised. In broad outline, the experience of both large and small banks was roughly the same. Both experienced small declines in savings deposits and C D ’s, and both had fairly sizable gains in time deposits open account, with the result that total time and savings deposits were little changed. Further, for both size classes of banks the increase in time deposits open account was due entirely to gains in consumer-type time deposits which more than offset declines in all other categories. Both large and small banks ex Table II shows for large denomination C D ’s the number of banks reporting increases in “ most com mon rates” since October 31 and also the distribu tion of banks by most common rate paid. Taken at face value, the table implies that banks in general did not take aggressive advantage of the oppor tunity to raise their rates. Considering the high yield on market instruments relative to previous Q ceilings, the extent and pervasiveness of CD attri tion, and the continued strength of loan demand, it is somewhat surprising that more banks did not re perienced relatively small declines in small denomina port higher rates. tion C D ’s, but there was an apparent difference in plemental survey that many bankers waited until W e have evidence from a sup their experience with respect to large denomination February 1 to raise rates on their savings deposits C D ’s. and consumer-type However, if all large denomination C D ’s are time deposits. Perhaps they lumped together ignoring the negotiability feature, also waited until February 1 before raising rates on which frequently is more a technical than substan other classes of time deposits. tive matter, both classes of banks experienced small of the overlapping of the new and old ceilings, a bank declines. could have raised rates to the new ceilings on all The scatter of pluses and minuses indicates some Furthermore, because maturities of large denomination C D ’s and still have regional diversity in the behavior of time and sav reported the same most common rate that it reported ings deposits, but more striking is the evidence of in October. For example, on January 31 the 6.25% strong trends which transcend regional boundaries. rate on very short maturities very conceivably could For example, the similarity among regions in time have pulled in the most deposits. deposits open account is very arresting. All regions bank would have reported 6.25% as the most com experienced significant increases in consumer-type mon rate, perhaps the same rate it reported on O cto In such a case, the time deposits which in most instances outweighed ber 31. declines in other open account deposits. response to the new Regulation Q ceilings. Also, banks Thus, the table probably understates bank throughout the Fifth District generally had small declines in C D ’s under $100,000 and fairly sharp About half of the banks issuing large negotiable C D ’s reported higher rates, and of these, about half drops in large denomination C D ’s of a nonnegotiable reported that the 7.50% rate on C D ’s maturing in nature. The only obvious regional diversity oc curred in the behavior of large denomination negotia ble C D ’s and of savings deposits. Rates Paid on Large Denomination C D ’s 'Regulation Q Ceilings: In the survey, which was conducted as of January 31, Savings banks were asked to report the most common rate Multiple m aturity time deposits offered on new savings deposits and on time deposits of various kinds. The most common rate was de fined as that rate which generated the largest dollar inflow of deposits during the 30 days preceding the New Ceilings Old Ceilings per cent per cent 4.50 4.00 30-89 days 90 days to 1 year 4.50 5.00 4.00 5.00 *1 year to 2 years 5.50 5.75 5.00 5.00 5.00 5.00 *2 years and over Single maturity time deposits Less than $100,000 reporting date, or, if a rate change was made during 30 days to 1 year 5.00 this 30-day period, that rate prevailing on the survey 1 year to 2 years 2 years and over 5.50 5.75 date which generated the largest dollar inflow. Since the Regulation Q ceilings were raised on most classes of time and savings deposits on January 21,1 banks at least had an opportunity, and probably strong incentive, to change their rates during the last ten days of the month. 8 5.00 $100,000 and over 30-59 days 60-89 days 90-179 days 180 days to 1 year 1 year or more 6.25 6.50 6.75 7.00 7.50 5.50 5.75 6.00 6.25 6.25 ♦These ceilings were raised on March 3, 1970, but made retroactive to January 21, 1970. one year or more was exerting the greatest pulling power. A s expected, higher rates were somewhat more common among large banks than small banks. Interestingly, a larger fraction of banks in Virginia and the District of Columbia reported higher rates than was the case elsewhere. Only about 40% of banks issuing nonnegotiable C D ’s reported higher rates on these instruments, despite the fact that most banks throughout the Fifth District reported significant declines in out standings since the October survey. Again, rate increases tended to be concentrated somewhat more heavily among large banks and banks in Virginia. Rates Paid on Savings Deposits and ConsumerType Time Deposits as of February 2 In order to determine the aggressiveness with which com mercial banks competed for savings deposits and socalled “ consumer-type” time deposits, the Federal Reserve conducted a supplementary survey of rates paid as of February 2. Table III summarizes the results of that survey. The inescapable conclusion is that banks generally, and large banks in particular, jumped at the opportunity to engage in rate compe tition for consumer savings. Almost all large banks raised rates to the new ceilings, and the vast majority of small banks did likewise. In addition, a few banks TABLE II MOST COMMON RATES PAID ON TIME AND SA V IN G S DEPOSITS, IPC, By a Sam ple of Fifth District Banks, Ja n u a ry 31, 1970 TIME C D 'S - $ 1 00,000 AN D O V ER Total Num ber N um ber of Banks of Banks H aving Reporting Rates Particular Type Higher than on of Deposit 1 0 /3 1 /6 9 Interest Rate (%) Linder 6.25 6.25 Total Fifth District 6.50 6.75 7.00 7.50 (Num ber of Banks) N egotiable 31 15 12 19 10 12 8 1 2 1 7 By Deposit Size of Bank: Less than $100 million $100 million and over 1 5 By State: District of Colum bia M arylan d N orth C aro lin a South C a ro lin a V irgin ia W est Virgin ia 3 4 5 2 16 3 10 43 17 1 1 1 By SM SA: W ashington Baltim ore C harlotte Richmond N orfolk Roanoke N onnegotiable Total Fifth District 20 13 By Deposit Size of Bank: Less than $100 million $100 million and over 12 25 18 8 By State: District of Colum bia M arylan d North C a ro lin a South C a ro lin a Virgin ia W est Virgin ia 6 2 7 1 6 3 2 21 10 1 1 By SM SA : W ashington Baltim ore Charlotte Richmond N orfolk Roanoke 15 1 1 6 4 1 9 survey, both because of its timing and because the questionnaire asked for the most common rate rather than the maxium rate being paid. The increase in rates offered by banks on their deposits in conjunction with a general decline in other short-term market rates has created a climate favorable to the revival of intermediation and a turnaround in time and savings deposits outstanding. Jimmie R. Monhollon Jane F. Nelson that were paying rates well below those of other banks raised rates, but not to the new ceilings. Thus, there occurred a general upward adjustment of rates paid for consumer savings. C onclusion T h e supplem ental survey strongly suggests that banks have taken advantage of the lee way under the new Regulation Q ceilings to com pete more aggressively for time and savings deposits. This fact was obscured somewhat by the January 31 TABLE III INTEREST RATES PAID ON SA V IN G S DEPOSITS AND ON SINGLE M ATURITY TIME DEPOSITS OF LESS THAN $100,000 As of Feb ruary 2, 1970 S A V IN G S Total Paying 4 ’/2% Maxim um Paying 4% 84 64 20 57 27 41 23 16 4 7 5 9 DEPOSITS Total Fifth District By Deposit Size of Bank: Less than $100 million $100 million and over By State: District of Colum bia M arylan d North C a ro lin a South C aro lin a Virgin ia W est Virgin ia 22 2 13* 10 10 3 34 2 3 2 ** 6 8 Paying 5V2% Maxim um Paying 5% or Less Not O ffering Instrument M ATURITY O F 1 Y EA R BUT LESS THAN 2 Total Fifth District By Deposit Size of Bank: Less than $100 million $100 million and over By State: District of Colum bia M aryland North C aro lin a South C aro lin a Virgin ia W est Virgin ia 84 68 57 27 43 25 7 6 22 9 10 3 34 10 2 34 8 6 Paying 5%% M aximum 10 1 Paying 5Vi% Paying 5% or Less Not O ffering Instrument M ATURITY O F 2 YEA R S OR M ORE Total Fifth District By Deposit Size of Bank: Less than $100 million $100 million and over By State: District of C olum bia M aryland North C a ro lin a South C aro lin a Virgin ia W est Virgin ia 84 59 18 57 27 35 24 15 3 7 22 10 3 34 5 8 10 8 2 2 30 2 8 6 1 *lnclu des one bank indicating maximum w ould be paid A pril 1, 1970. **ln clu d es one bank paying 414%. 2 1 The Fifth District Personal Income The upward trend of total and per capita per sonal income continued last year in the Fifth District as well as in the nation as a whole. But while per sonal income continued to grow in 1969, the gains became slimmer as the year progressed. In the final months of the year the national growth in personal income failed to keep pace with the rising cost of living. PERSONAL INCOME $ Billions For the United States, the year-to-year growth in total personal income declined from 9.3% in 1968 to 8.3% in 1969; comparable figures for the Fifth District were 9.3% and 8.8% respectively. Within the District, W est Virginia and the District of C o lumbia showed greater advances in 1969 than in 1968. Total personal income increased 6.4% in W est V ir ginia in 1969, up from the 5.8% advance registered in 1968; it rose 8.8% in the District of Columbia in 1969, compared to an 8.0% increase in 1968. Personal income in Maryland, Virginia, and the two Carolinas rose by a smaller percentage in 1969 than the year before. Maryland’s advance of 10.5% in 1969 was down from 11.5% in 1968. For the same periods, Virginia had increases of 9.5% and 1 0.3 % ; North Carolina, 9.4% and 1 0.0% ; and South Carolina, 9.0% and 9.9% . During the past two decades, Maryland, Virginia, North Carolina, and South Carolina have more than tripled their total personal income while the District of Columbia and W est Virginia have doubled theirs. PERSONAL INCOME Per C ap ita Total A v erag e Annual G row th 1969 1949-'69 1949 1969 A verag e A nnual G row th 1949-'69 $ mil. $ mil. per cent $ mil. $ mil. per cent Md. 3,392 15,454 7.9 1,456 4,105 5.3 D. C. 1,700 3,894 4.2 2,107 4,880 4.3 V a. 3,648 15,395 7.5 1,108 3,297 5.6 W. V a. 1,994 4,738 4.4 1,033 2,605 4.7 N. C. 3,675 14,926 7.3 940 2,868 5.7 S. 1,724 6,910 7.2 850 2,567 5.7 16,133 61,317 6.9 1,128 3,236 5.1 205,791 740,761 6.6 1,384 3,669 5.0 C. 5th Dist. U. S. Source: 1949 Data for 1949-1968, Survey of Current Business, U. S. Departm ent of Com m erce. 1969 d ata reprinted from M arch 7, 1970 issue of Business W eek by special permission. Copyrighted (c) 1970 by M cG raw -H ill, Inc. 11 Federal Reserve Bank of Richmond Richmond, V irg in ia 23213 6c Paid Return Requested Richmond, V a . Permit No. 2 The average annual growth rate of 6.9% for the Fifth District compares favorably with the 6.6% for the nation. The average growth rates in four of the District states were above the national average. These were Maryland (7 .9 % ), Virginia (7 .5 % ), North Carolina (7 .3 % ), and South Carolina (7 .2 % ). Growth rates for W est Virginia and the District of Columbia were below the national average, at 4.4% and 4.2% respectively. The Fifth District’s per capita personal income, which is a better measure than total personal income of the economic well-being of the District’s popula tion, has increased at an average yearly rate of 5.1% from 1949 to 1969, compared to a 5.0% rate for the nation. Personal income per person in the Carolinas and Virginia out-paced the nation with average an nual growth rates of 5.7% in each of the Carolinas and 5.6% in Virginia. Maryland and the District of Columbia had growth rates below the 5% national PER CAPITA PERSONAL INCOME average; even so, they maintained the highest levels of per capita income in the District throughout the Dollars period. In 1968, the latest year in which complete break downs are available, 67.5% of total personal income in the United States represented wages and salaries, 24% represented property and proprietors’ income, and 9 % was in the form of transfer payments such as Social Security and welfare payments. In the Fifth District North Carolina, South Carolina, V ir ginia, and Maryland all received a larger percentage of their income in the form of wages and salaries than the national average— 70% , 73% , 74% , and 75% respectively. W ages and salaries represented 66% and 67% of total personal income in the Dis trict of Columbia and W est Virginia. Property and proprietors’ income was a smaller fraction of total personal income in the Fifth District than in the nation, ranging from 19% in Virginia and Maryland to 2 2% in North Carolina. The percentage of personal income from transfer payments was larger than the national average in the District of Columbia *1969 estimated. Source: Survey of Current Business, U. S. Department of Commerce. (1 4 % ) and W est V ir ginia ( 1 3 % ) , but smaller in North Carolina ( 8 % ) , South Carolina ( 8 % ) , Virginia (7 .5 % ) and Mary land (7 .2 % ). M . Grace Haskins
@FedFRASER