The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
April 1961 LY REVIEW Is the Cottle Cycle Changing? . . page Importance of Size and Other Factors Affecting Bank Costs . . Current Statistics . . . . . 3 page 10 . page 16 FEDERAL RESERVE BANK OF KANSAS UITY Suhscriptions to the Mo THLY HEvmw are available to the public without charge. Additional copies of any issue may be obtained from the Research Department, Federal Reserve Bank of Ka nsas City, Kansas City 6, Missouri. Permission is granted to reproduce any material in this publication. ls the Cattle Cycle Changing? business requires long-range planning. In making such plans, some expectations regarding cattle inventories and prices must b e formulated . Accu rate prediction of cattle numbe rs js nev r 'asy, and it has h 'C ll ·s peciall y diffi ult in r c ' nt y •;us. Th' mos t r ce nt contraction phase of the cattle cycle-1955 to 1958- was shorter in length and smaller in amount th an that of any previous cycle. The current expansion phase of the cycle also has behaved in an unusual manner-numbers increased 1 million last year, as compared with a 6 million increase in the corresponding year of the previous cycle. Contributing to the uncertainty about cattle numbers is the recent revision in annual estimates for 1955 throu gh 1960, which was made after 1959 censu s data became available. The large magnitu<le of th e revision-a reduction of 5.2 per cent in total numbers from the original estimates made for 1960has caused considerable confusion concerning the interpretation of recent livestock forecasts. Historically, cattle inventories have varied cyclically around a long-run, upward trend . In addition, irregular movements have occurred from time to time, resulting from such diverse factors as droughts, wars, supplies of competitive products, and changes in economic activity. The cyclical and irregular movements have b een responsible for much of the instability in the cattle industry. While there is little hope for eliminating all cyclical S UCCESS IN THE CATTLE Monthly Review • April 1961 and irregular movements, it may be possible to minimize them through a better understanding of the nature of changes in cattle numbers. Such a minimization of instability would b ben fj ial to catt] producers, feeders, suppli ers, finan ·ial :1gcncics, mark tin g firms , and ·ons um ,,·s. Trends In the 95 years for which annual estimates are available, total cattle numbers have expanded from a low of 28.6 million to a high of 97.l million, an increase of 240 per cent. The average increase of 729,000 per year tends to be misleading because it includes several types of changes. The long-term trend line in Chart 1 indicates a rate of increase of 560,000 annually for the 1867-1961 period. This probably is a more realistic estimate of th e upward trend, since it is not influ enced by shorter-run irregular movements to the same extent as the preceding estimate. Cattle inventories vary in such a way that it is difficult to fit a linear, long-term trend line to the data. Such a trend is influenced by the d ifferential rates of growth for the periods 1867-1890, 1890-1928, and 1928-1961. In the first period, a trend line fitted by visual inspection indicates an average rate of growth of 1,455,000 h ad per year. A similar trend line for the s cond period shows an annual increas of only ,'380,000, while the third period trend shows a growth rate of 1,090,000 head p r year. Durin g the first period, the fronti er was moving rapidly westward and 3 Is the Cattle Cycle Changing? Chart 1. Total Cattle Numbers and Trend s United States MILLION HEAD JANUARY I , INVENTORIES 100 veal. Strong demand, improvements in productivity, and ample feed supplies in recent years induced the strong growth rate that has prevailed since 1928. II Cyclical Movements 90 80 70 60 50 40 30 Actual Numb ers Short - Period Trends - - LonQ - Term Trend 0 L.L-L_j___J_..J._-'---.J_-'--L-l-1....-1-'-'---'--:-"-:--'----:-:.'-::--'---:': 1867 '80 '90 1900 '10 '20 '30 '40 '50 '61 NOTE : The short-period trends were fitted by vi sual Inspection and the long-term trend was fitted mathemati cally by the least squares method . SOURCE : U. S. Department of Agriculture . cattle production was being expanded in the Great Plains area. This was the p eriod following the Civil War when the first transcontinental railroads were built, the Plains Indians were restricted to reservations, and the vast buffalo herds were exterminated. Toward the end of the period, the homest ad movement was gaining momen tum and the op n rang was giving way to barbed wire and the plow. Cattle production xpanded fast r than demand duri11g the first period, and by 1890, cattle numbers were disproportionately large. During the second period , declining per capita beef consumption was a weakening factor, but it was more than offset by population growth. Consequently, the beef cattle industry was able to continue expanding in this period. During the present period , consumer incomes hav ris n greatly from th <lepre sion lows of the ea rly 1930's and this has contribut d to a substantial incr as ·n per apita onsump tion . In addition , population has increased more rapidly since World War II , further iucreasing the demand for beef and 4 Since 1867, th re have be n sev 11 p riods of incr asing cattle numbers an<l si periods of <leer asing numbers. These mov ments have been largely cyclical, with th length of the cycle being closely related to the time r quired to build up and liquidate a cattle herd. High or in reasing cattl pric s apparntly stimulat produ · rs to buil l up hre <ling hercl s and to h Id fc cl rs to h ·avi ·r w igh ts. This r 'slri -ts ·urr ·nt marketings, whi ·h stimulal ·s furth e r pri · in 'reascs. The cycl, r inforc 'S its ,lf until nough ti m has lapsed for the withholding of animals to be translated into increasing suppli s of b eef and veal. The resulting increase in supplies tends to depress prices and to reverse the cycle. Since there is no restraint on the rate of liquidation comparable to the restraint on the rate of inventory buildup imposed by the time required for gestation, growth, and fattening, th downward movement can transpire mor · rapidly th an th e upward mov ment. Th compara tive amplitudes and 1 ngths of the yclical movements ar shown in Table 1. Since the first period was not characterized by a complete cyclical movement, it serves primarily as a base for the fol1owing movements. The first "low" ( in 1876) was determined statistically by removing the trend influence as shown in Chart 2. In terms of absolute numbers, there has been littl change in the upward amplitude of the ycles sine I 90 but, in p ' r e ntag t rms, th av rag mer ase in the ·urr nt p 'riod has be n smaller than it was in th e s cond. The downward changes hav be n considerably small r than the upward chan ges and th y have becom incr asingly smaller during the current pe- Is the Cattle Cycle Changing? Table 1. Characteristics of Cyclical Movements in Total Cattle Inventories United States Absolute Change Period Fast growth Year Cycle Position Inve ntory Numbers (Mi llion) 1876 Lowt 36.1 1890 High 60.0 Relative Change* I low High to to high low (Million) +23.9 I Low High to to high low (Per Cent) Low 1904 High - 19.8 Low 14 +29.8 High 16 Low 55.7 14 High 73.0 16 -24.1 Low 16 +25.9 1945 High 10 Low 65.2 1958 High Low 11 7 +27.0 11 85.6 - 10.8 4 76.8 10 +22.8 +19.8 1955 4 -1 3.1 -8.8 1949 6 74.4 +20.4 Ill Fast growt 10 57.3 - 9.2 1938 6 +27. 0 + 17.1 1934 8 - 17.6 - 15.7 1928 8 66.4 + 17.3 1918 High to high 6 49.2 - 10.7 19 12 I (Years) Low to low 20 +17.2 II Slow growth High to low 14 +49.6 - 10.8 1896 Length of Cycle Low to high 96.6 - .4 6 - 5.8 3 9 91.2 * Ea ch rela ive change was calc ul ated by dividing the differen ce between the high and low by the avera ge of the two In order to eliminate th upward bias inheren t in perccn .1ge changP. expre ssions . t The low in 1876 wa ermiOP c1 stati st1rally by removing trend (see Chart 2). SOURCE : U. S. Department of Agriculture . rio<l. Th av rag m n ber of years of the upswings appar .ntly chang d c ry li tl " between th econcl and bird pe iods, while that of the downswings changed cons 'derably. During t e second eriod, the downward ovements increased in length from 6 years to 10 years, whi e iI I e t ir peri d, they have decreased to yea s in length. s a resu lt, th over-all cycles have decreased from an av rag of about 15 years in th second p riod to about 10 y ars in the third . The purely cyclical movements in total cattle numbers, after the trend and irregular movements were removed statistically, are shown in Chart 2. T he eye es have been conMonthly Review • April 1961 tracting both in length and amplitude. The upward deviation from the trend line declined from a maximum of 14 per cent in the 1918 peak to 8 per cent in the 1954 peak. Only the depression peak of 1934 rose above the trend by a smaller amount than the most recent high point of the cycle and, since it began from a much lower level, its total rise was greater. The downward amplitudes have contracted ev n more than the upward amplitudes - from 14 per cent in the 1896 trough to 4 per cent in that of 1959. The over-all amplitudes from low to high and high to low d clin d by about one half from the second to the third period. s Is the Cattle Cycle Changing? Chart 2. Cyclical Movements and Rates of Change In Total Cattle Numbers United States PER CENT +30 I m lI +20 +10 0 - 10 - 20 - - Cyc li c al Movemen ts - - 3 0 '--'--'---'--'---'- 186 7 ' BO '9 0 L 1900 L '1 0 Rotes of Change ~ '2 0 '30 l '40 ' 50 ' 61 NOTE : The cy cli cal mov ement s were computed a perce nt ag e devia ti ons of Inventory numb er from a tr nd curve fill ed 111at11 ematl cally by th e I ast squares meth od. Th e irr gul ar movement s were removed by mean s of a moving av erage . The rate s of change were computed as percentage change s in inventory numbers from year to year . The rate of change in cattle numbers from year to year reveals a great deal about the nature of the cattle inventory cycles. The rates of change depicted in Chart 2 were calculated from the original data and, consequently, reflect trend and irregular movements as well as cyclical variations. The turning points in rates of change have usually preceded the turning points in inv ntory numbers by about 2 years, indicating that th e buildups and liquidations usually begin to lose momentum some time before the turning points of the inventory cycles. The amplitude of the rate-of-change cycle seems to have changed very little, except for the drastic liquidation period in the early 1890's and the Government liquidation program in 1934. Since the rate-of-change amplitude has remained constant, the contraction of the inventory amplitude must be explained by the shorter lengths of the cycles. Inventory Changes by Classes While movements in total cattle numbers are of considerable interest, they tend to obscure many important divergencies in move6 ments among the different classes of cattle . An especially significant difference has occurred between cattle and calves kept fo r milk and those not kept for milk. The proportion of the cows 2 years ol<l and over which were kept for milk d eclined fro m 71 per cent in 1928 to 43 p er cent in 1961. Similarly, the proportions of h eifers 1-2 years old kept for milk dropped from 62 to 42 per cent and calves from 37 to 21 p er cent. This should not be interpreted as a shift from dairy breeds to b eef breeds b ecause much of the ch ange has b een the result of a shift from dual-purpose animals to speciali zed animals. Pd or to World War ll , man y beef cattl e ra is 'rs milk d th eir ·ows durin g the flu sh season and solcl crea m. Both da iry and b eef pro h1 'lion have become more highly p ccializ cl in r cent years and , while fewer b eef and dual-purpose cows are milked , dairy h erds still provide a substantial amount of beef and veal. The growth of specialization in dairy production has been accompanied by a considerable increase in average milk output per cow. Since this increase has occurred during Chart 3. Cyclical Changes In Dairy Cattle Numbers By Classes United States PER CENT +40 +20 - - Cows - - Heifers - - Heifer Calves - 40 1928 '35 '40 '45 '50 '55 '6 1 NOTE : The cyclical movements for each cla ss were computed as percentage deviations cf inventory numbers from a trend lin e fitted mathematically by the least squares method . Is the Cattle Cycle Changing? Chart 4. Cyclical Changes In Beef Cattle Numbers, By Classes United States PER CENT +4 0 +20 ,A 0 - 20 ----- -40 192 8 Co ws H e if er s Calv es St eer s been the least irregular, reflecting the influence of longer-run production plans. In addition, there has been considerable variation between trends for the different classes, with cows having the strongest upward trend and calves being second. Steers and h ifers 1-2 years old have incr ased the least of the group, although both increas d more than any class of dairy cattle. The trend line for all beef cattle indicated an increase of 1,108,950 per year based on a linear regression equation fitted to the 192561 data. A similarly calculated trend for d airy cattl indi at d an in r a of only 7,440 p r year . . in ·c World War 11 , h ·d attl num bers hav mo v cl upwar I even more strongly, whil ' dairy cattl numb rs have de lined . ' 35 NOTE: The cyclical movements for each class were computed as percentage deviations of inventory numbers from a trend line fitted mathematically by the least squares method . a period of slow growth in the demand for milk, fewer and fewer milk cows have been needed. Dairy cattle numbers, which trended upward from 1921 to 1944, have followed a downward trend since then, particularly for milk cows 2 years and older. Chart 3 shows that cycl xist in dairy cattle numbers, but comparison with Chart 4 shows that they differ consid rably from the cycles in b ef cattle numbers. The amplitud of the dairy cattle cycle has been smaller and the turning points have tended to precede those of beef cattle. It is usually assumed that the culling of dairy herds is influenced by slaughter cattle prices and that this causes some similarity in the cyclical patterns. Chart 4 shows that the cyclical movements among the different classes of b ef cattle have tended to coincide in timing and direction but not in amplitude. Steer numbers have been the most irregular, often moving opposite to the others. This probably is a reflection of the single purpose and more ready marketability of steers. Cow numbers have Monthly Review • April 1961 Meat Production and Slaughter Prices Beef and veal production are used to measure production responses because they reflect the influence of variations in slaughter weights as well as slaughter numbers. Cattle and calf prices tend to respond immediately to changes in beef and veal production-a.sChart 5. Deflated Beef Cattle Prices and Per Capita Beef Production United Sta tes PER CE N T OF 19 10 25 0 200 150 100 50 - De flated Pric~s - - Per Ca pi ta Pro du c tion 0 ~'-'-'-...........-'-'-c'-'-'-LJ....L.'-'-'-,'-L--L -'-"-'--'-'-'...J-L.LL..L.L.L..L.J...L.l..,LJ....LL..L.L.L..L-'-LJ 19 10 ' 20 ' 30 '40 '50 ' 60 NOTE : Deflated prices were computed by dividing average prices re ceived by farm ers for beef cattle by the index of prices paid by farmers for comm odi ties used in produ ction . Per capita production was comput ed by di viding total beef produ ction by civilian populat ion . SOURCE : U. S. Departments of Agriculture and Commerce . 7 Is the Cattle Cycle Changing? suming that demand conditions and supplies of substitutes remain the same. Production, for reasons discussed earlier, responds more slowly to the influence of price incentives, although responses tend to be more rapid for veal than for beef production. Chart 5 shows the movements in cattle prices and beef production from 1910 to 1960. The prices were deflated by the index of prices paid by farmers, and per capita production was derived by dividing total beef production by civilian population. Use of this procedure gives a truer picture of the cyclical movements in the beef market. Cyclical movements in b ef cattl pric s w 'r larg ly obscur d b y a generally rising pri e 1' vel from 19.34 to 1951, but a strong y ·Ji al mov m nt s ems to hav d 'veloped since th n. Per capita production of beef has shown a moderate although fairly regular cycle. Calf prices and veal production display a comparatively regular cyclical movement, as shown in Chart 6. This apparently reflects Chart 6. Deflated Calf Prices and Per Capita Veal Production United States PER CENT OF 1910 250 200 50 - - 0efloted Prices - - Per Ca pito Production 0 .............._.........,__......__._.L..L..L.-'-'--'-.LJ...J.....L..L..l-'--UW-1-LLLL.L...L.LL..L.LJ...J.~-'--UL..J...L.LLLL..J 19 10 '20 '30 '40 '60 '50 NOTE : Deflated prices were comput ed by dividing average pri ces received by farmers for ca lves by the index of prices paid by farmers for commo dities used in produ ction. Per capi ta production was co mputed by div id ing total veal produ ction by civilian population . SOURCE : U. S. Departments of Agriculture and Commerce . 8 the greater responsiveness in veal production. The amplitudes of the cyclical movements in veal production and calf prices are similar, whereas those of beef production are considerably smaller than those of cattle prices . Veal and beef production are not, of course, unrelated. The responsiveness of veal production prevents great r variations in beef production b y absorbing much of th e shock of abrupt changes in demand or supply conditions and, in the case of increased calf slaughter, by decreasing the potential supply of beef. The ratio of calf slaughter to the calf crop offers some advance indication of hang s in cattl numbers. Thi · ratio usuall y r a ·h ·s a low and begins to rise about 2 y ars h fore total caltl ' in v 'nlori c · r ac h ,l peak and b gin to d dine. Further, the ratio usually reaches its high and b egins to decrease about 4 years b efore inventories reach a low and begin to rise. Concluding Remarks Livestock numbers display so e regularities of movement but sufficient irregularity exists to make forecasting difficult. Year-toyear predictions often miss by a wide margin and longer-term predictions are especially hazardous. Nevertheless, changes in cattle production require such a long time that producers and capital suppliers must formulate some sort of expectations for as much as 5 to 10 years in advance. In view of this necessity and the sharp fluctuations which occur in livestock prices and feed costs, livestock production and financing are hazardous occupations from an economic standpoint. A statistical projection of the trend curve and cyclical pattern in total cattle numbers is shown in Table 2. The figures ar not a prediction of future cattle numbers but simply a first and second approximation based on an extension of past conditions into the future. The trend shows a figure of 95.3 million for 1958 with a continuous rise to 120.5 million Is the Cattle Cycle Changing? Table 2. Statistical Projection of Total Cattle Numbers United States Year Trend Values* (Million) 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 95.3 97.2 99.1 101.2 103.3 105.5 107.8 110.1 112.6 115.1 117.8 120.5 Cyclical Statistical Inventory Deviations** Estimatest Numbers (Per Cent) (Million) (Million) - 5.3 - 5.0 - 1.9 +3.0 +6.9 +7.9 +6.7 +4.0 - 0.3 - 3.3 - 3.8 - 3.4 90.2 92 .3 97.2 104.2 110.4 113.8 115.0 114.5 112.3 111.3 113.3 116.4 91.2 93.3 96.2 97.1 • Ba se d on a trend curve fitted mathematically by the lea st sq uare s m thod . •• Bas ed on smoothed p r ccntagc r c latlon shlp between Inventory numb r s and tr end value s in l ast full cyc l e. t Ba se d on t rend valu es adjusted for cyclical deviations . in 1969. Superimposing the most recent cycle pattern ( from 1949 to 1957) on these figures indicates 90.2 million for 1958, rising to 115.0 million in 1964, declining to 111.3 million in 1967, and climbing to 116.4 million in 1969. That the pattern of the current cycle differs from that of the previous cycle is shown by Monthly Review • April 1961 comparing th e inventory numbers with the statistical es timates for 1958 throu gh 1961 . The in ventory numbers differed from the stc tistical estimates by only 1 million in each of the first 3 years, but they were 7 million less than the estimate in 1961. This ma pr sage a shorter buildup in this cycle than in th e last. The turning point of th present cycle s ms highly uncertain . If the patt rn of th e most recent cycles were repeated, the p eak would occur in 1964. However, th er is some indication tha t it may occur soo n r. The rate of incr as dropp cl off in J960 and , . in this has usuall y pn.' cedcd a peak in nurnhcrs hy about 2 y 'ars, it incli 'ates a possihlc peak in 1962 or 1963. 'uch a hri ,f h11ild11p wcmld b the shortest on re ord and ther is little in the previous history of cattle cycles to support such an expectation except that the last liquidation was also the shortest on record. In any event, the amplitude of th e cycles has declined in recent years, and if it continues to decline, th e cattle cycle may eventually assume negligible proportion s. 9 Importance of Size and Other Factors Affecting Bank Costs preceding issu es of this Review dealt with the relationship between size and costs at member banks in the Ten th Federal Reserve District. M asurecl as a p er cent of assets, costs at a sample of abou t 270 District member b anks during the period 1956-59 were found to decline significan tly with increasing size. The cost advantages of la rger-scale op rations w ere shown to rdlc t tli ' ability of Jarg 'r banks lo opC'rat' with srnalJer numbers of unpJoy 'Cs per dollar of assets ancl with a high 'r prop rtion of nonofficial employe s to officers. 1 hes cost advan tages seem to stem partly from opportunities to perform ordinary banking functiorts in more efficient ways , and partly from the ability of larger banks to carry on transactions for loans and investments in larger dollar amounts. In the two previous articles, attention was focused on the average relationship b etween bank size ancl costs without consid ri ng the importance of iz in r lation to oth r factor that influ nc bank cost ratios. Is th size of a bank th e dominant chara t ristic influencing its expenses as a p r c nt of a sets, or are other characteristics of greater significance? The first portion of this article presents a discussion of the relative importance of various factors that influence bank cost ratios, and is based on the same statistical analysis used in the earlier articles. second and related topic d eals with chang s in the r lative importance of fa ctors influencing costs during the years 1956 to 1959. This p eriod witnessed strikin g changes in the volume and composition of assets and liabilities of District memb r banks. A study of the ch anging relative importance of various facto rs associated with b ank costs during A 10 RTICLES IN THE TWO these 4 years sheds light on the way in which District banks respond ed to a sharp upswing in th eir loans and deposits and to the spread throu gh th banking sys tem of higher int rest ra tes on tim deposits. Factors Identified as Cost Determinants Methods of statistical analysis do not permit isolation of all of th many factors that accou11t for cliff r ·nc 'S in ·osts among hanks. Sp ' ·i, il c irc111nsla11 -cs tl1at ar' 1rniqu' to an indi vidu ul bank oft '11 a· ·ount for a significant shar ' of th ' cliff rcn e betw en its costs and those of other banks similar in size and in other respects. Moreover, there are some forces responsible for cost differences among banks that cannot easily b e m easured or for which the necessary d ata are unavailable. The statistical method employed to investigate the relationship between bank size and costs also sought to find an association betwe n b ank cos ts and major characteristics of as ts and liabilities for which data ar readily availabl from m mb r bank r 'ports of onclition. A bri f discussion of th reasons for s lectin g the chara t ristics included in the study provides a helpful b ackground for the discussion to follow. The division of a bank's assets among major classes-loans, securities, and cash-is certain to have a significant effect on its costs. The structure of assets by major classes is represented in the analysis by two factors: ( 1 ) th per cent of total ass ts in the form of loans, and ( 2) the per ent of total assets held as s curities other than U. S. Governmen t issues . Given thes two p ercentage fi gures, th e proportion of b ank assets in liquid form ( cash and U.S. Government se urities) is automatically allowed for, since Joans, se- Importance of Size and Other Factors Affecting Bank Costs curities, and cash assets add up to virtually 100 per cent of total assets at almost all banks. Since analysis showed that the distribution of liquid assets between cash and Government securities was not closely related to b ank costs, this possibk consideration can be safely ignor d. There ar' four principal categories of Joans xtended by District banks- loans to businesses, nonguaranteed farm credits, real-estate mortgage loans, and loans to individuals or consumers. The proportion of total loans extended to consum rs was found to exercise a strong upward influence on hank cos t ratios. 011 tl1c other hand , no significant association w;1s discovered between costs and the proportio11 of loans cxl<.'1 1dcd to businesses, to farm ers, or to th e mortgag ' market. 1 his onclusion , which may seem surprising, might well indicate that the administrative costs of making a loan are determined not so much by the type of borrower as by characteristics of the individual loan transaction, particularly the size of loan. Consumer loans are high cost assets because they are small loans and because the bulk of them are repaid in instalments. The average size of other types of Joans d ep end s primarily on the size of bank, and so t ·nds to be refJ 'cted in the cost advantages enjoyed b y larger banks. On the liability side of the balance sh et, time deposits involve substantially larger costs than demand balances, since interest payments are forbidden on demand accounts. The amount of expense incurred on time deposits depends both upon the average rate of interest paid on time accounts and on the percentage of deposits in time accounts. These arc not, however, unrelated characteristics. Indeed , th e association between average rates paid on time deposits and the proportion of deposits in time accounts was so strong that their influ ence on cos ts of th e sample banks could not be separated statistically. Therefore, the latter characteristic Monthly Review • April 1961 alone was employed to represent both influences on costs. The statistical analysis also showed that, among larger banks, the percentage of dem:md deposits in the form of correspondent balances was associated with bank costs. Other thin gs equal , ratios of cos ts to assets tended to be lower for banks with a high perc ntage of interbank demand deposits. Results of an earlier study-pu blished in the July 1960 issue of this Review- suggested that banks with high growth rates tended to have higher ost ratios. The present study confirms this association and yield s additional information 011 the re lationship betwceu hank growth rates and h:111k costs. Direct Influence on Costs Tog ' th r with bank size, the characteristics of assets and liabilities mentioned in the precediug section account for 62 per cent of the variation in ratios of total costs to assets among the sample banks over the period 1956-59. Each factor separately, or directly, accounts for part of the variation, while an additional portion is explained by the joint influ ence of the several factors. The direct influence of each factor provid es the best initial guide to its relative importance as a d terminant of bank cost ratios, and therefore will be discu sed first. Measures of direct influence on cost ratios, as shown in Table 1, can be compared with one another readily, since each measure expresses the percentage of variation in cost ratios among the banks that a particular factor explains. The first four characteristics listed in the table each explain directly from 9 to 14 per cent of the variation in ratios of total costs to total assets among the sample banks. Although each of the four accounts for a slightly different percentage of total cost variation, the differenc s are not large enough to be assigned any important weight. The appropriate inference is that differences among the 11 Importance of Size and Other Factors Table 1. Measures of Direct Influence On Total Cost Ratios Sample of Tenth District Member Banks, 1956-S9 Bank Characteristic 1. Asset size 2. Relative volume of time deposits 3. Percentage of assets in loans 4. Percentage of loans extended to consumers 5. Growth rate of assets, 1956-59 6. Percentage of assets in non-Treasury securities Per Cent of Variance in Total Cost Ratios Explained 13 9 JO 14 2 2 NOrE : The data in the table are based on the function : X1 = f(log X2 , X3 . . . X7 ) , where X 1 Is the ratio of total costs to tot al dsse ts, X:i is asset size in millions , X3 is the r~tio of time to total deposits, X4 Is the ratio of total loans to total assets, Xs is the ratio of non-Trea sury securities to total assets, X6 is the ratio of co nsum er to total loans, and X7 i s the percentage Increase In a se t s, 1956-59, with al l ratio s expressed in percentage term s. Tho function wa s f i tted to data for Individual banks obta ined by ave ragi ng annua l figure s for th e years 1956-59 . The mea ures shown arc sq uare s of the beta (s tandard ized partial re gression) coefficie nts, expressed In percentage term s. banks in size, in the relative amount of tim deposits, in the p rcentage of assets in the form of loans, and in the percentage of loans made to consumers all were of approximately equal significance in explaining differences in total cost ratios. However, the comparative importance of these four factors in accounting for differences in ratios of wages and salaries to total assets was quite different. Bank size, which explain d directly about 28 per nt of th variation in wage and salary ratio , was by far the most important d t rminant. Rates of growth in assets and the percentage of assets held as securities other than U.S. Government issues, the last two characteristics shown in the table, exerted a substantially smaller direct influence on total cost ratios than the other four characteristics. The proportion of demand deposits in the form of interbank accounts - a characteristic not shown in th table - was found to account directly for about 2 per cent of th variation in total cost ratios among sample banks with ov r $25 million in assets. Clearly, then, m asures of dir ct influ ence point to th first four charact ristics in Table 1 as exerting the dominant influence on total cost ratios. 12 The previous articles in this series dis cussed in some detail th e average relationship between size and costs, and it is of in terest to note how cost ratios chang , on th a rage, with changes in the oth r three principal factors affecting costs. The top panel of the chart shows th way in which ratios of total cost to ass ts t nd to rise with high r ratios of time accounts to total deposits, aft r removing the influen e on costs of all other Relationship Between Total Cost Ratios and Characteristics of Assets and Liabilities Sample of Tenth District Member Banks, 19S6-S9 Per Cent Total Costs to Total Assets 3.2 Tl ME DEPOSITS 2.8 2.4 2 .0 1.6 Per Cent, Time Deposits to Total Depos its 0 10 20 30 40 50 60 3.2 TOTAL LOANS 2 .8 2.4 2 .0 1.6 / . Per Cent, Total Loans to Total Assets 0 10 20 30 40 50 60 3 .2 CONSUMER LOANS 2 .8 2.4 2 .0 1.6 Per Cent, Consumer Loans to Total Loans 0 10 20 30 40 50 60 NOTE: The charts are based on the function described In th e note to Table 1. The line in the top panel is obtained by setting vari ables X2 , X4 . . . . X7 at their mean values and then graphically portraying the resulting relation between X 1 and X3 • Lines in th e second and third panels are obtained by an analogous procedure . For each characteristic , the lines are drawn to cover the range of variation which is found among the sample banks . For example, few banks have ratios of total loans to total assets of less than 15 per cent , so the line in the second panel is not extended below that figure. Affecting Bank Costs charact ristics identifi d as cost determinants. The cost ratio rises by .16 percentage points, on the average, for each 10 percentage point increase in the ratio of time to total deposits. Similarly, as displayed in the second and third panels, the total cost ratio rises .22 percentage points for each 10 percentage point increase in the ratio of total loans to total assets, and .16 p 're ·ntage points for each 10 percentage point increase in the proportion of total loans extended to consumers. Joint Influences on Costs Additional insight into the r lativ importan ·c of the various ·haractc risti ·s may he gai ned hy 1·x:1mi11ing their joint infl11cn ·c on ·ost ratios. To clarify th · meaning of joinl influence , it may b' h 'lpful to use a simple il lustration from ano ther field. Suppos a person earning $10,000 gives 10 per cent of his income, or $1,000, for charitabl e purposes. When his income increases to $15,000, he gives 15 per cent to charity, or $2,250. Of the $1,250 rise in his contribution , no more than $500 is accounted for directly by the growth in his income-$500 being the product of the increase in income times the initial contribution rate of 10 per cent. Simi larly, no more than $500 is accoun ted for directly by the in rease in contribution rate, since th 5 percentage point ris in contribution rate times the original incom of $10,000 is $500. The $250 not accounted for directly by either the increase in income or the increase in contribution rate is properly described as the joint influence of both changes. Had the contribution rate dropped to 5 per cent when the income figure rose to $15,000, the direct eff cts would have been plus $500 for the hangc in income and minus $500 for th ' chang in contribution rate, bile the joint effect of both changes vould be minus $250. In a similar manner, characteristics of bank assets and liabilities have both direct and Monthly Review • April 1961 joint effects on costs. For example, sample banks with either high ratios of loans to total assets or high ratios of time to total deposits tended to have higher cost ratios, as noted above. These characteristics, however, arc not independent-usually, banks with relatively high time deposit ratios have above average loan ratios. Con equ ntly, these banks tend to have above av ·rage cost ratios for three reasons: their comparatively high volume of loans, th ir higher percentage of time deposits, and because both loans and time deposits are relatively high. As wilh the dir ct influenc on osts discuss cl ahove, the joint influcnc<.' of any two cost d 'lcnni11a11ts may he expressed in terms of the r ,ta ti ve amount of varialion in cost ratios among hanks that it a · ·ounts for. Table 2 show th joint effect on costs of each pair of characteristics listed in Table 1. A red figure indicates that the joint effect is negative-as in the illustration above when the contribution rate declined. These measures of joint influence disclose several interesting aspects of bank cost experienc . First, the joint influence of bank size and the three other major determinants of costs arc all negative. This resu lts from the fact that, among District member hanks , ratios of tim ' to total deposits , total Joans to total ass ts, and consumer loans to to al loans, all tend generally to increase with larger bank size. Thus , while increasing size is associated with decreasing costs, part of the cost advantage of larger size is offset by changes in the structure of a sets and liabilities which make for higher costs. This should not b e taken to imply, however, that ratios of time to total d eposits , total Joans to assets, ancl consumer Joans to total loans rise continu o11sl with size of bank throughout the full range of bank sizes in th e District. For although the very large t banks in th District tend to have the highest ratios of loans to assets , banks with assets in 13 Importance of Size and Other Factors Table 2. Measures of Joint Influence on Total Cost Ratios Sam ple of Tenth District Member Banks, 1956-59 Per Cent of Variance in Total Cost Ratios Explained NonGrowth Trea sury Total Consumer Rate, SecuriAsset Time Loans 1956-59 ties Size Deposits Loans Asset size Time deposits Total loans Consumer loans Growth rate, 2 1 6 4 4 1 3 2 2 2 1956-59 Non-Treasury securities * Less than 0.5 per cent. A red figure indicates that th e joint effect is negative. NOTE: The data in the table are based on the function desc ribed 1n the note to Table 1. The figure s represent twice th e crossproduct o f the relevant beta coefficients times the si mple corre lation coefficie nt for each pair of varia ble s, expressed in percentage terms . The algebraic su m of th e direct effects shown in Table 1 and the /·oi nl effe t s indicated above I equal to th e sq uare of th mul ipl e corre lation coe ffi cie nt In perce nta ge t erm s (62) excep t for a differen ce due to rounding. th range of $10-$50 million hav th highes t ratios of time deposits and the largest p ercentage of their loans xtended to consumers. Larger banks than this, which are typically downtown banks in larger cities, usually have relatively smaller amounts of time deposits and consumer loans. A second notable feature is the substantial amount of cost variation explained by the joint influence of time deposits and two charact ristics of asset structure. Banks with high ratios of time accounts to total d posits also tend to have a relativ ly large portion of their a sets in loans and a higher-than-averag p rcentage of their loans extended to consumers. Presumably, this reflects the attempt by banks with relatively large amounts of time deposits to search for assets carrying higher yields as a means of covering interest expenses on their time accounts. Differences in total cost ratios among the sampl banks also were accounted for to a consid rable degree by th joint influ nee of their growth rates with oth r cost-cl termining characteristics. In fac t, the joint influence of growth rates and other characteristics explains 8 per cent of the variation in total cost ratios among the sample banks, while the di14 rect influence of growth rates on costs explains but 2 per cent. This implies that the relatively high costs found among the more rapidly growing banks resulted primarily from bank characteristics that are associated with rapid growth. Sample banks whose growth rates were high had relatively high ratios of time accounts to total deposits ( and paid above average rates of int rest on time deposits), high percentages of loans to total assets, and high percentages of consumer loans to total loans. Banks whose growth rates wer higher than av rage over th e years 1956-59 were spread broadl y over all J islri t stat s; th list inc- lucl ccl some clow11lown hanks ,1s well as s11h11rhan hanks , and hanks in rural ·ommu11iti s as well as in urban areas. It thus se ms appropriate to view their favorable growth experience as resulting in considerable measure from management policies conducive to growth-including a willingness to pay higher interest rates to attract time deposits, and the adoption of aggressive policies to accommodate loan customers. Interestingly, the characteristics of assets and liabilities associated with more-than-average growth over the years 19.56-59 also were associated with mor than-av rag ' growth over th e long r p riod from 1947 to 1959. Changing Relative Importance of Cost Influences The years from 1956 to 1959 witnessed significant changes in the comparative importance of the four principal characteristics that account for differences in bank cost ratiosthat is, among bank size, the percentage of deposits in time accounts, the perc ntage of assets in loans, and the proportion of loans extended to consumers. Th se changes are refl ct d ad quately in the measur s of their direct influence on total cost ratios for each of the individual years 1956 through 1959, as shown in Table 3. Affecting Bank Costs Table 3. Measures of Direct Influence on Total Cost Ratios, 1956-59 Sample of Tenth District Member Banks Per Cent of Variance in Total Cost Ratios Explained 1956 1. 2. 3. 4. Asset size 16 Relative volume of time deposits 6 Percentage of assets in loans 13 Percentage of loans extended to consumers 12 1957 1958 1959 16 13 10 7 10 11 8 12 11 15 10 8 NOTE : The figures in the table are squares of the beta (s tandardized partial r egression) coefficients of the fun ctio n described 1n the note to Table 1, fitted to each of the individual years 1956 through 1959. Th . most striking change that took pla 'e was the large in -rcasc in th' rclativ irnporta11<·e of time d ·posits as a clctcrmi11n11t of total ·os t ratios. Between 1956 and 1959, average effective rates of int res t on tirn a counts at the sample banks rose from 1.57 per cent to 2.19 p r c nt, with most of this change taking place after January 1957, when legal maximum rates payable on time deposits were raised from 2 1/2 per cent to 3 per cent. Meanwhile, time accounts increased from 17.3 per cent of total deposits in 1956 to 22.0 p er cent in 1959. To be sure, these changes were widespread among District banks , as well as in other s ctions of the country, hut that <lid not prevent ratios of time to total deposits from be oming a more important factor in explaining ost differences among the banks. A given increase in interest rates on time accounts affects costs most at banks where ratios of time to total deposits are relatively high. Similarly, a given increase in the percentage of time accounts affects total cost ratios most at banks paying higher-thanaverage rates to their time deposit customers. Thus, the result of these changes was a sharp rise bctwe n 1956 and 1959 in the importance of time deposit ratios in accounting for cost differenc s among District members. This increasing significance of time deposit ratios itself tends to lower the measures of relative importance for other characteristics Monthly Review • April 1961 of banks that influence th eir expenses. 1 The declining influence of total loans on costs is, however, too large to be attributed to this influence alone. It results mainly from bank responses to the vigorous upswing in loan volume that took plac from 1956 to 1959. At th group of sample banks in -luded in th study, average loan volume in 1959 was about one-fourth higher than in 1956. The average ratio of loans to total assets among the banks advanced from 32.9 per cen t jn 1956 to 36.3 per cent in 1959. The largest part of this surg in loan volume took place in th e relatively short span of 2 years- from mid - 19.57 to niicl - lD.59. To handle lh<' incr<'asi ng volt11ll(' or loans , il W.IS 11ol IH'CC'SS:H y for tlic hanks lo in -rease proportional( ly their staffs of offi · •rs and crnployc 'S; rather, existing staffs were used more intensivel y, with the result that bank costs became less closely associated with the relative amount of their assets in loans . Consequently, the growth of loans in relation to other assets added to the banks' net earnings rates not only because of the shift to assets with higher gross earnings rates, but also because administrative costs per dollar of loans were held down. o ·hanges of funclam ' ntal significanc arc evident, however, in th relative importance of hank size or in th ~ proportion of loans extended to consumers as cost influences from 1956 to 1959. The measures for consumer loans shown in Table 3 vary erratically from one year to the next, suggesting only that the weight as a cost determinant is more appropriately judged by data that are averaged for several years. The measure for bank size is lower in 1959 than in earlier years because the year witnessed a relatively larger increas 1 This is becaus the effe t is to increase the variance of total costs. Thu s, if the partial regression oefficient of, say, bank size and th variance of bank size are unchanged, while th e variance of total costs is incrca ·ed, th e stan<lardize<l partial regression coefficient of bank size is reduced. 15 Importance of Size and Other Factors Affecting Bank Costs in miscellaneous expenses at large than at small banks. Part of this increase was due to a rise in borrowings among larger banks during 1959; a second part was due to the comparatively larger advance for large than for small banks in non-income tax payments during both 1958 and 1959. The basic advantages of larger-scale operations in banking, which are found in wage and salary exp nses, were just as important in 1959 as they were in 1956. Concluding Remarks The foregoing analysis indicates that although bank size is an important factor affecting tho ability of a hank to operat with low costs in relation to its assets, it does not overshadow other factors. Mcani11gf11l comparisons of cost ratios among banks must giv' att ntion not only to the siz of bank hut also to a variety of other characteristics of their assets and liabilities. The dominant structural characteristics of assets and liabilities that influence costs are directly within the control of bank management. However, the avoidance of high costs by policies such as the selection of assets on which administrative costs are low, or the maintenance of low rates of int rest on time deposits, carries its penalties. It is widely r cognized that gross earnings rates are directly influenced by choices among alternative typ s of assets, but th growth rate of a bank also may be affected significantly by its lending policies and its willingness to attract time deposit customers. From the viewpoint of its influ n c on xpcnses, the int rest of a hank in growing rapidly is ·lcarly vi 1 nt in th e relationship between si:1.c and ·osts. For while th ' irnrm'dial ' result of rapid growth appears to in -r 'as ·ost , th long r-run i1 flu n c is to r due costs by permitting the bank to enjoy the cost advantages of largerscale operations. BANKING IN THE TENTH DISTRICT loans Deposits Reserve District PRICE INDEXES, UNITED STATES Reserve City Country City Country Member Member Member Member Banks Banks Banks Banks and Feb. 1961 Index Jan. 1961 Feb. 1960 Consumer Price Index ( 1947-49 = 100) 127.5 127.4 125.6 Wholesale Price Index ( 1947 - 49 = 100) 120.0 119.8 119.3 Prices Rec'd by Farmers ( 1910-14 = 100) 244 241 233 Prices Paid by Farmers ( 1910-14 = 100 ) 302 301 299 February 1961 Percentage Change From Stales TENTH DISTRICT BUSINESS INDICATORS Jan . Feb. Jan. Feb. Jan . Feb. Jan. Feb. 1961 1960 1961 1960 1961 1960 1961 1960 +4 +e +2 +16 - 2 +6 t +s Colorado +3 +2 t +11 +1 +5 t +6 Kansas +4 +1 t +1 - 2 +10 Tenth F. R. Dist. Missouri * Nebraska New Mexico* Oklahoma * Wyoming +10 -1 +23 +1a t +1 + s - 4 +4 + 2 +: t +21 t +5 - 1 +s ** ** +s +14 ** ** +4 +9 +6 + 18 - 5 +6 +2 + 11 +s ** ** - 2 +3 +1 + a ** ** *Tenth District portion only. t Less than 0.5 per cent. 16 - 2 +1 ** No reserve cities in this sta te. District and Principal Metropolitan Areas Value of Check Payments Value of Department Store Sales Percentage change-1961 from 1960 Feb. Year to date Year Feb. to date +2 +s +11 + 13 +5 +1 +5 Denver Wichita - 10 - 3 - 7 - 7 Tenth F. R. District Kansas City + s 0 +6 0 +2 Omaha - 2 +5 +22 +2s Oklahoma City +1 + s - 6 - 4 +5 +3 0 Tulsa 0