The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
FEDERAL RESERVE BANK OF ST. LOUIS SEPTEMBER 1976 1 ■tl ■r CONTENTS The Unemployment Rate as an Economic Indicator ............................ 2 Development of Electronic Funds Transfer Systems ................ .............. 10 LITTLE Vol. 58, No. 9 The Unemployment Rate as an Economic Indicator JEAN M. LOVATI I) "URING the most recent recession, the unemployment rate rose to its highest level since the depression years of the 1930s, reaching 8.9 percent in May 1975. The unemployment rate has since declined and stood at 7.9 percent in August of this year. Despite this de cline, unemployment is still unacceptably high, with 7.5 million persons re porting that they want to work but have not found jobs. The unemployment picture has been aggravated by increases in the unemployment rate in recent months. Besides contributing to fears that the economic recovery is weaken ing, such increases intensify pressure for adopting policy measures designed to reduce unemployment. Labor Market Trends 1941 1950 1952 1954 1956 1958 1960 1962 1964 1966 196S 1970 1972 1974 1976 SSh h a d e d , are a re as, a s represent re p re se n tp e rio d s ot recessions. Latest d a ta p lotted: 1 976 b a se d on first eight month! MEASUREMENT The basic source of labor market information is the monthly survey data collected by the Bureau of the Census for the Bureau of Labor Statistics. The civilian noninstitutional population is sampled,1 with approximately 50,000 randomly selected households surveyed monthly. With a carefully structured set of questions, information is gathered which results in a determination of the labor force status of the civilian population, age 16 and over. The respondents are never asked specifically if they are employed or un employed, nor are they given the opportunity to de cide their own labor force status. For example, for persons replying that they did not work during the survey week, questions are asked to determine whether they had jobs from which they were absent 'The civilian noninstitutional population includes all persons 16 years of age or older who are not members of the Armed Forces, members of penal or mental institutions, sanitariums, or homes for the aged or infirm. Page 2 or had no jobs whatsoever.2 If the respondents had no jobs, further questions are asked until the sequence of questions which leads to a labor force determination is completed. Household members are classified as employed if they report doing any work at all for pay or profit during the survey week. This includes part-time and temporary work as well as full-time employment, and incorporates unpaid workers who worked at least 15 hours in a family-operated enterprise. Although workers may have been absent from work during the survey week due to illness, vacation, strike, or various personal reasons, they are still considered employed. To be classified as unemployed, persons must report being without work during the survey week, being 2The survey covers the calendar week which includes the 12th of each month, although it is actually conducted in the follow ing week. FEDERAL RESERVE BANK OF ST. LOUIS currently available for work, and having actively sought work at some time in the prior four weeks. Seeking work, in the context of the survey, includes registering at an unemployment office, answering ads, or checking with friends or relatives about job open ings. Persons who are waiting to be called back to a job from which they have been laid off or who are waiting to report to a new job within 30 days are considered unemployed, even though they do not seek other work. The total unemployment figure thus counts more than those who have lost their jobs. It includes per sons who have quit their previous jobs to look for other work, new workers looking for their first jobs, and persons looking for jobs after an absence from the labor force. The unemployment rate expresses the number of persons classified as unemployed as a percentage of the civilian labor force. Household members who are not employed and are not seeking employment are classified as “not in the labor force.” Persons in this category are neither employed nor unemployed. Some persons in this category do not participate in the labor force be cause they are retired, attending school, or engaged in home responsibilities. Others are financially inde pendent, or simply prefer leisure. Still others report SEPTEMBER 1976 that they do not look for work because they believe that no jobs are available for persons with their ex perience or training. This group is often referred to as “discouraged workers.” Due to the large size of the household sample, labor market estimates derived from it are generally regarded as reliable in a statistical sense. The data are subject to some degree of sampling error, how ever, and this must be taken into account when an alyzing changes in the data. For the overall unem ployment rate, the error is relatively small. To be deemed a significant change, the national unemploy ment rate would have to move 0.2 percentage point or more on a monthly basis. There is a relatively high probability that movements in the unemploy ment rate which are smaller than 0.2 percentage point are due to sampling variability, instead of changes in the unemployment situation. Unemploy ment rates for component groups, because they are based on smaller samples, are subject to larger errors. For example, the statistically significant error on month-to-month movements in the teenage unem ployment rate is about 0.9 percentage point, and in the black unemployment rate about 0.8 percentage point.3 INTERPRETATION Productivity, Compensation, and Labor Costs The unemployment statistics represent an attempt to estimate the number of persons who want, but can not find, work. The data are most often used as meas ures of labor force utilization and as indicators of general economic activity. These uses seem reason able (although subject to some limitations) since un employment, in principle, means human resources are idled and productive opportunities are foregone. Others use the data as measures of hardship expe rienced by the unemployed. Hardship resulting from unemployment is influenced by a number of other factors such as the amount of savings, eligibility for unemployment compensation, food stamps, other wel fare programs, the number of household members employed, and the extent of payments from former employers or union funds. Because of the growing importance of these other factors, the unemployment rate tends to overestimate the extent of economic hardship suffered relative to periods in the past.4 3“Unemployment: Measurement Problems and Recent Trends,” U.S. Department of Labor, Bureau of Labor Statistics, Re port 445, 1975. P e rc e n ta ge s a re a n n u a l rates of c h a n g e fo r p e rio d s in dicate d. Latest d a t a plotted: 2 n d q u a rte r 4The unemployment rate for household heads is often used as a more meaningful measure of hardship than the total unem ployment rate, but this rate is also subject to limitations. Page 3 FEDERAL RESERVE BANK OF ST. LOUIS Unem ploym ent Rates* S o u rc e : U.S. D e p a rtm e n t o f L a b o r 'P e r c e n t o f c iv ilia n l a b o r fo rce in g ro u p . L a te st d a t a p lo tted : A u g u s t On the other hand, the unemployment statistics might understate the number of persons without work. Because of the current definition of unemployment, discouraged workers — those who forsake job search because they feel no jobs are available for them — are excluded from the official unemployment statistics. Data collected in recent years indicate that dis couraged workers are primarily youths, women, and blacks. If discouraged workers were added to the seasonally adjusted unemployment rate for 1975, the adjustment would add an average of 1.2 percentage points to the total rate. This merely exemplifies the depressed labor market conditions in 1975. In both the years 1973 and 1974, discouraged workers would have added 0.8 percentage point to the total rate had they been included. Of course, the adjustment in the jobless rate for adult men would be relatively small, but would be greater on the already high rates for women, youths, and blacks. In 1975, the addition of discouraged workers would have increased the unemployment rate for Negro and other races an average of 3 percentage points. One very important difficulty encountered with the unemployment statistics is that questions concern ing wages are not part of the survey. Persons report ing that they believe no jobs are available are not asked at what wage jobs appear to be unavailable. Thus no information is gathered on the extent to which unemployment results from an asking wage which is too high. This situation can result from both Page 4 SEPTEMBER 1976 a generally depressed state of economic activity which has not convinced unemployed workers of the general decrease in their value as factors of produc tion, and artificial restrictions which prevent the ask ing wage from falling to a level commensurate with productive ability. Only the former case is amenable to the traditional tools of monetary and fiscal policy. Growth in aggregate demand will not aid those whose unemployment results from labor market im perfections. For example, Government mandated wage restrictions, such as the minimum wage, prevent em ployers and workers from coming together to bargain for a mutually acceptable wage. Workers, especially teenagers, often are willing to work at wages below the legal minimum but are prevented from doing so. As a result, unemployment in this group rises. Sim ilarly, unions negotiate wage contracts for all employ ees through collective bargaining and typically set floors on wage offers. If some workers consequently are laid off, they may be willing to work at wage rates below union floors rather than lose their jobs, but find they cannot do so. One alternative is to seek other work, usually at a lower wage. It should be noted that some degree of unemploy ment always prevails as workers change jobs to pur sue better job opportunities. This type of unemploy ment is typically short term, usually lasting five weeks or less. Changes in labor supply and demand condi tions necessitate changes in employment and wages. As workers become informed of better job possibili ties, they make changes across jobs. The unemploy ment resulting from movements of this sort, called frictional unemployment, is expected in a changing economic environment and thus is not of primary concern for national economic stabilization purposes.5 USE OF EMPLOYMENT STATISTICS Despite the popularity of the unemployment statis tics, their use alone does not provide a complete view of the labor market situation. Use of unemployment figures should be complemented by at least employ ment data.6 Too often, unemployment statistics signal different labor market actions when viewed alone and when viewed in conjunction with employment figures. This is usually due to movements between 5For a discussion of the different types of unemployment, see Roger W. Spencer, “The National Plans to Curb Unemploy ment and Inflation,” this Review (April 1973), pp. 2-13. 8Similarly, in analyzing the economic state of the nation, the unemployment situation should be reviewed in conjunction with other measures of economic performance, such as in dustrial production or gross national product. FEDERAL RESERVE BANK OF ST. LOUIS SEPTEMBER 1976 Table 1 A C O M P A R IS O N O F SELECTED LABOR MARKET M E A SU R E S1 Civilian Noninstitutional Population Civilian Labor Force Labor Force Participation Rate Total Employment Unemployment Rate Unem ployed 15 W eeks or More as a Percent of Total Unem ployed 12 Months Before Trough of: 14 . 3 % October 1 94 9 103,361 6 0 ,6 4 6 5 8 .7 % 5 8 ,3 8 7 3 .7 % M a y 1954 106 ,9 1 0 6 2 ,6 1 5 58.6 6 1 ,0 1 9 2.5 11.0 April 1958 111 ,94 3 6 6 ,6 4 7 59.5 6 4 ,0 4 7 3.9 19.8 February 1961 116 ,70 2 6 8 ,9 4 9 59.1 6 5 ,6 2 0 4.8 25.3 Novem ber 1 97 0 1 35 ,23 9 8 1 ,3 9 7 60.2 78,541 3.5 13.7 114,831 68,051 59.3 65,5 2 3 3.7 17.8 147 ,81 6 9 0 ,5 4 9 61.3 8 6 ,033 5.0 18.0 AVERAGE M arch 1975 At Trough of: October 1949 104 ,33 8 62,1 8 5 59.6 5 7 ,2 6 9 7.9 20.3 M a y 1954 108 ,1 8 4 6 3 ,6 7 5 58.9 5 9 ,908 5.9 23.3 April 1958 113 ,41 5 6 7 ,6 4 7 59.6 62,631 7.4 27.7 February 1961 1 1 8 ,2 5 0 7 0 ,4 2 0 59.6 6 5 ,588 6.9 29.3 Novem ber 1970 138 ,05 3 83,4 2 2 60.4 7 8 ,5 3 7 5.9 17.8 1 16,448 6 9 ,4 7 0 59.7 6 4 ,7 8 7 6.8 23.7 150 ,448 9 1 ,8 8 0 61.1 84,1 1 0 8.5 25.5 October 1949 105 ,0 9 6 62,4 2 8 59.4 5 9 ,8 0 3 4.2 20.8 M ay 195 4 109 ,54 4 64,381 58.8 6 1 ,6 3 4 4.3 26.7 April 1958 114 ,9 8 6 6 8 ,3 3 9 59.4 6 4 ,768 5.2 29.1 February 1961 1 1 9 ,36 0 7 0 ,4 0 9 59.0 6 6 ,538 5.5 32.1 Novem ber 1 97 0 140,821 85,0 8 5 60.4 7 9 ,9 4 4 6.0 25.1 117,961 70,1 2 8 59.5 6 6 ,5 3 7 5.0 27.0 1 53.17 8 9 3 ,7 1 9 61.2 86,692 7.5 32.6 AVERAGE March 1975 12 M onths After Trough of: AVERAGE March 1975 1A11 data are in thousands except labor force participation and unemployment rates. All data are seasonally adjusted except civilian noninstitutional population. C h an ge From 1 2 M onths Before Trough to Trough C hange From Trough to 1 2 Months After Trough Unemployment C ivilian Labor Force Employment Rate O vilian Labor Force Employment Unemployment Rate October 1 94 9 2 .5 4 % — 1 .9 1 % 4 .2 % 0 .3 9 % 4 .4 2 % — 3.7 % M a y 1954 1.69 — 1.82 3.4 1.10 2.88 — 1.6 April 1958 1.50 — 2.21 3.5 1.02 3.41 — 2.2 February 1961 2.13 — 0 .05 2.1 — 0.02 1.45 — 1.4 Novem ber 1 97 0 2.49 — 0.01 2.4 1.98 1.79 0.1 2.08 — 1.12 3.1 0.94 2.70 — 1.8 1.47 — 2.23 3.5 1.99 3.06 — 1.0 AVERAGE March 1975 the categories of “unemployed” and “not in the labor force.” A more complete representation and different explanation of the labor market situation may result when movements in employment are also considered. Employment figures have several advantages over the unemployment figures. Employment is a more solid and objective concept than is unemployment. With few exceptions, persons working during the survey week are considered employed. Identification of employment is relatively straightforward and thus easier to measure than unemployment, where classifi cation depends on more provisos and is subject to more uncertainties (such as in determining whether a person is actively seeking work or is currently avail able for work).7 Also, employment figures are more closely tied to output measures than are unemploy ment data. Decisions to look for work, for example, ’ Employment statistics are not without definitional problems, however. Questions concerning the status of part-time and underutilized workers are yet to be resolved. Page 5 FEDERAL RESERVE BANK OF ST. LOUIS SEPTEMBER 1976 Total Employment T ro u g h s= 1 0 0 T ro u g h s= 1 0 0 Seasonally Adjusted Unem ploym ent Rate O C T 49 M A Y 54 APR 58 F E B 61 NOV 70 M A R 75 rage of 1969-72, 1960-62, 1957-59, 1953-55, and 1948-51__ -6 -3 0 3 6 M O N T H S TO A N D FROM TRO U G H S Latest data plotted: A u gu st 1976 M O N T H S TO A N D FROM T R O U G H S Latest data plotted: A u gust 1976 often are based on personal factors, unrelated to economic considerations. CYCLICAL BEHAVIOR Since it takes time for labor market actions to be reflected in the unemployment rate, movements in this rate generally trail economic activity. As the level of economic activity falls, for example, it is usually more feasible for employers to initially adjust work schedules of existing workers than it is to discharge workers employed for some time. Although actions such as these signal the decline in economic activity, they are not picked up by the unemployment data until the workers are laid off. In a recovery, employers are likely to increase hours worked of current em ployees rather than begin hiring procedures. Thus, the unemployment rate tends to understate the extent of a decline in economic activity at the start of the decline and the extent of the recovery later on. In past recessions, the unemployment rate typically has peaked from one to four months after the trough of the general business cycle. Employment and Unemployment The labor market response to the most recent down turn differs substantially from the behavior in other postwar recession/recovery periods.8 Total employ ment in this recovery has already reached pre-reces sion levels. Employment has been increasing since March 1975, after a 5.2 percent annual rate of de cline in the previous five months. This decline repre sents a marked change from the average of employ8The most recent recession is compared with the five postwar recessions which reached their troughs in November 1970, February 1961, April 1958, May 1954, and October 1949. Recessions are identified by their trough months in this article. Page 6 ment movements of the previous five recessions. In these recessions, employment declined at a slower rate, averaging a 1.1 percent drop in the 12 months before the troughs. Employment in the 1970 recession remained fairly constant throughout most of the eco nomic decline, only varying by about 350,000 workers in the 12 months preceding the trough. The 1961 recession also did not register great declines in em ployment. Between June 1960 and February 1961, total employment declined at a 1.3 percent annual rate (see Table I, p. 5, for comparison). Despite the steep decline, employment in the most current period recovered faster than the average of employment in other recession/recovery periods. Total employment increased at a 3.1 percent rate over the year since the 1975 trough, compared to a 2.7 percent average rate of increase for other recovery periods. This difference has been made even more apparent since March of this year. Between March and August 1976, total employment rose at an annual rate of 3.6 percent, widening the gap between employ ment movements in the most current and previous recoveries (see accompanying chart). By August 1976, employment reached 88 million workers, an increase of four million workers in this recovery period. While employment growth has been strong in this recovery period, the unemployment rate has declined only slowly. In August of this year, the unemployment rate was only 0.6 percentage point below the rate at the trough. Compared to the average of previous re coveries, the unemployment rate remains remarkably high. Twelve months after the five prior troughs in economic activity, the unemployment rate had dropped an average of 1.8 percentage points, com FEDERAL RESERVE BANK OF ST. LOUIS SEPTEMBER 1976 spectively. For the same period after the 1961 trough, labor force growth de clined by 0.1 percent. Total Civilian Labor Force as a Percent of Civilian Non-lnstitutional Population (16-64)* Several factors have influenced this labor force growth. A primary consid eration is the increase in labor force participation.10 The participation rate in August reached a high of 61.9 per cent of the population. The labor force participation rate of adult women is particularly noteworthy. As inflation erodes real family income, other house "Both civilian la b o r force a n d non-institutional p o pu la tio n d a ta a re b a se d on the w orking force a g e g ro u p of 16-64. The lab o r force an d hold members, particularly women, are population com ponents a re b a se d on quarterly a ve rage s of monthly data. Total civilian la b o r force d ata are se aso n ally adjusted. S h a d e d are a s represent p eriod s of recessions. likely to be drawn into the labor force. Latest d a ta plotted: 2nd quarter This movement has compounded the ef pared to a 1 percentage point drop in the year fol fect of the trend for increased participation of women. lowing the most recent trough. One year after the By August 1976, the participation rate of women troughs of the 1958 and 1949 recessions, the unem reached 47.4 percent. This is an increase of 4.1 per ployment rate had fallen 2.2 and 3.7 percentage centage points since 1970 and 9.8 percentage points points, respectively (see chart). since 1960. As of August of this year, adult women comprised 36.3 percent of the labor force. Another group accounting for a larger proportion of the labor Labor Force Growth and Unemployment force than heretofore is teenagers. The labor force The high unemployment rate relative to most pre participation rate for workers in the age group 16-19 vious recoveries can be explained in part by different averaged 54.5 percent during 1974 and 1975, the high rates of labor force growth. Generally, the civilian est in recent history. In August of this year, the par labor force grew at a much faster rate during the 1975 ticipation rate for teenagers reached 55.4 percent. The recovery than in other postwar recoveries.9 Between shifting composition of the labor force toward women March 1975 and March 1976 (twelve months after and youths and the higher unemployment rates gen the trough), the civilian labor force grew by 2 per cent. By contrast, in the twelve months following other postwar troughs, growth in the civilian labor force averaged 0.9 percent. In particular, in the twelve months following the April 1958 trough, the civilian labor force grew at half that rate, and during the same period after P e r c e it the 1961 recession, the labor force de clined by 0.02 percent. 10Labor force participation rates measure the percentage of persons in a given population, however defined, that are in the labor force. Selected Civilian Labor Force Participation Rates 11 Q u a rte rly A v e r a g e s of M o nth ly D ata S e a son a lly Adjusted P trC M t In recent months, the differences in labor force growth have become increas ingly more evident. From April to Au gust 1976, the civilian labor force grew at a 3.4 percent annual rate. Labor force growth has averaged a 2.8 percent annual rate over the prior seventeen months. Seventeen months after the low points of the 1958 and 1949 recessions, the civilian labor force had grown by annual rates of 0.9 and 0.4 percent, re9The exception is the recovery from the Novem ber 1970 recession in which the civilian labor force grew at about a two percent rate in the twelve months following the trough. U. Participation rates m e asu re the propo rtio n of persons in each p o pu latio n g ro u p that are in the civilian la b o r force. Sh a d e d are as represent periods of recessions. Latest d a ta plotted: 2nd quarter Page 7 FEDERAL RESERVE BANK OF ST. LOUIS erally experienced by these groups tend to raise the average level of the overall unemployment rate (see chart entitled “Selected Civilian Labor Force Partici pation Rates”). Duration of Unemployment The behavior of the unemployment rate in this recovery may also be related to factors which tend to lengthen the duration of the unemployment period. During the most recent recovery, the percentage of the unemployed without work for 15 weeks or more was larger than in previous recovery periods. At the trough of the 1975 recession, 26 percent of the un employed were jobless for 15 weeks or more, com pared to 18 percent at the trough of the 1970 re cession and 20 percent of the unemployed at the 1949 bough. Nine months after the trough of the most recent recession, 22 percent of the unemployed were jobless for 27 weeks or more, while 11 percent of the unemployed were still jobless after 27 weeks dur ing the 1970 recession. Nine months after the troughs of the 1954 and 1949 recessions, 15 and 11 percent, respectively, of the unemployed still had been with out work for 27 weeks or more. The duration of unemployment depends in part on the cost of remaining unemployed which, in turn, is in fluenced by the current system of unemployment com pensation as well as other benefit programs. By signifi cantly reducing the cost of remaining unemployed, unemployment compensation tends to increase the time devoted to searching for a job and thus lengthens the average duration of unemployment.11 In 1975 Congress broadened the scope of eligi bility for unemployment insurance benefits and provided additional benefits of up to 13 weeks a year for covered unemployed workers who have exhausted their regular benefits. Since workers covered by unemployment insurance have been eli gible for 26 weeks of benefits per year and some states provide an additional 13 or more weeks, this extension by Congress enabled many unemployed workers to receive benefits for one full year or more. Expansion of insurance payments, coupled with supplementary unemployment benefits and other sources of aid such as food stamps, tend to reduce in11Longer search time, however, is not necessarily undesirable. Accepting the first job offer reduces the probability of geting the highest-paying job and lowers one’s wealth compared to what it would have been had one taken longer to find other jobs. Thus, taking more time to search for a job may “pay off” by resulting in higher lifetime earnings. Page 8 SEPTEMBER 1976 Output versus Capacity S t a le Q u a rte rly Totals at A n n u a l Rates R a tio B i l li o n s o f D o l la r s Se a s o n a lly Ad ju ste d B i l lio a s o f D o lla r s R a tio 1600 S e a l* 1600 1500 nom ic Capa city (pre-1974) / ^ ^ H e w E conom ic Ca acilv U1259.4 1200 Real 1972 1973 1974 roduct 1975 1976 1977 1978 U. Econom ic cap ac ity is estim ated to h a v e been decre ase d b y 4.5 percent, reflecting the effects of h ighe r e ne rgy prices, gove rn m e nt resource a n d price control p rogram s, a n d other au tonom ous restrictions on a g g re g a te productivity. N o adjustment is m ad e here to the rate o f grow th of c a p acity, but that too p ro b ab ly h as been re du ced from the 3.6 percent norm ally used. See: De nis S. Karnosky, 'The Link Between M o n e y a n d Prices— 1971-76 "tlune 1976). Latest d ata plotted: 2 n d q u arte r centives to look for work.12 Jobs are available even during periods of highest unemployment; that is, job vacancies do not completely dry up, but wages tend to be unacceptably low in view of alternative pros pects. Workers in effect are discouraged from taking lower paying or less desirable jobs when their cur rent spendable income, comprised of tax-free un employment insurance payments, is not very different from their income after taxes derived from work experience. As a result, unemployment and the unemployment rate tend to decrease more slowly. OUTLOOK Although the unemployment rate has increased in recent months, its rise is apt to convey a misleading picture of the state of the labor market and the under lying strength of the current recovery. Other factors reflect a more encouraging economic picture. Indus trial production, for example, rose at an annual rate of 7.6 percent since January of this year. Real gross national product increased at a 6.7 percent annual rate in the first two quarters of 1976. While the labor force has been growing at an ex ceptional rate, so has employment. Since January 1976, 12See Martin Feldstein, “The Economics of the New Unem ployment,” The Public Interest (Fall 1973). SEPTEMBER FEDERAL RESERVE BANK OF ST. LOUIS total employment rose by 1.8 million workers, a 3.6 percent annual rate of increase. In the past year, em ployment rose by 3.2 percent. Thus, output growth has been such that jobs are being provided not only to those who became unemployed during the reces sion, but also to the growing numbers entering the labor force. Moreover, more people have been voluntarily seek ing other jobs in the past year than was previously the case. Approximately 13 percent of those unemployed in August voluntarily left their jobs to seek better opportunities, compared to 10 percent of the unem ployed eleven months before. Typically, this upward trend signals increased confidence on the part of the job leavers that they will be able to find better jobs. While job leavers are an increasing share of the unemployed, job losers and new entrants/reentrants into the labor force have been a decreasing propor tion of the jobless. The number of job losers among the unemployed declined from 58.5 percent in June 1975 to 49.3 percent in August 1976. New entrants and reentrants together represented 38 percent of the unemployed in August, down two percentage points since January. This downward trend suggests that the increasing rates of participation, and thus the strong labor force growth, may be slowing. If this is 1976 the case, and if output growth continues at generally expected rates, unemployment is likely to decline. SUMMARY With over seven million persons unemployed, the nation is confronted with one of the highest unem ployment rates in recent history. Moreover, when compared to the average of previous postwar recov eries, the unemployment rate has made little progress toward decline. One reason for this failing is that the labor force has grown at a much faster rate in this recovery than in previous ones. New entrants and reentrants have shifted the composition of the labor force to ward more women and youths, groups which have higher unemployment rates. In addition, the relatively high number of unemployed may be maintained due to incentives for prolonged job search provided by an expanded system of unemployment insurance compensation. Despite the high rate of unemployment, relief may be in the offing. Growth in employment has been strong. Confidence in the economy to continue pro viding jobs is evident. If increases in labor force par ticipation rates decelerate, the expanding economy should be able to integrate the currently unemployed into desired jobs. Page 9 Development of Electronic Funds Transfer Systems WILLIAM C. NIBLACK Faster than any but the most optimistic thought possible, electronic banking is sweeping the country — overwhelming the laws that govern banks and other financial institutions, changing dramatically the banking and savings habits of millions of Americans. Ultimately, electronic banking will revolutionize the very concept of money itself and will probably force a profound change in how the Federal Reserve regu lates the nation’s money supply. Certainly it will touch off a flurry of competition for the nation’s finan cial business unlike anything seen before.1 Developments in electronic funds transfer systems ( E F T S ) have long been a popular topic of discussion in banking circles, but as the above quotation from a national business magazine indicates, the subject now enjoys even wider currency. Although it does not ap pear likely that we are on the threshold of “the checkless-cashless society,” recent developments have been so rapid that predictions of future developments and effects of EFTS often seem foolhardy at best. The diversity of views on the subject of E FT S re flects the broad scope of current developments. Many bankers see E FT S as providing an opportunity for initiating new services and reducing costs, thus in creasing profitability. Others, especially those repre senting small banks, view E FT S as being prohibitively costly; they fear that they would not be able to com pete with the larger banks which could afford the necessary computer equipment and, as a result, would be forced out of the market. Thrift institutions view E FT S as a means of obtaining deposits for which only commercial banks have heretofore been permitted to compete and have thus been in the forefront of EFTS developments. Government will probably play a large role in the evolution of EFTS. Court interpretations of existing laws have already shaped the direction of some EFTS developments. New legislation and regulations will 1“Bank Cards Take Over the (August 4, 1975), p. 44. Page 10 Country,” Business Week almost certainly be adopted, but as yet no clear trend in the nature of these changes is discernible. Some of these changes may be dependent on the findings of the National Commission on Electronic Funds Trans fers. This commission, which held its first meeting last February* is due to submit an interim report to Con gress in October. It expects to complete its study and make recommendations by the end of 1977.2 EFTS AND THE PAYMENTS MECHANISM To speak of the electronic funds transfer system is an oversimplification. In fact, E F T developments are proceeding in several directions, with a number of different systems in various phases of development or use. The common factor in these systems is that they speed the transfer of funds by communicating infor mation relating to payments by electronic means rather than by use of paper instruments as is pre dominant today. Thus, E F T systems are designed to replace manual processes with electronic data proces sing and to speed the flow of funds through high speed data transmission. Although E FT S is often considered a revolutionary development with far-reaching effects, E F T develop ments can be viewed as just another step in the evolu tion of the payments mechanism — the system by which resources are transferred from one economic unit to another. This evolution reflects the continu ing effort to improve the efficiency of trading. Money, which is simply a device which facilitates trade, repre sented an improvement over barter in that it reduced transactions costs and thus freed resources for use in the production of other goods and services. Checks came into widespread usage because they offered con siderable advantages over cash; they were easily trans 2“E F T Commission Rebuffs Mitchell’s Plea for Comment on Proposed Changes to Reg. J,” American Banker (March 15, 1976), p. 1. SEPTEMBER FEDERAL RESERVE BANK OF ST. LOUIS ported in any amount, easily transferred between individuals, involved much less danger of loss or theft than cash, and served as proof of payment. Checks thus reduced the transactions costs involved in mak ing many types of money payments. Still, considerable transaction costs are associated with the processing of these paper documents. Furthermore, because of the indirect nature of the check clearing process, there is often a delay in the availability of “good funds” for the payee. These de lays can be costly, especially when large sums are involved. The development of wire transfers, through which banks can effect funds transfers by sending electronic messages rather than paper documents, represented a major improvement in the payments mechanism. One of the large wire transfer networks is operated by the Federal Reserve System, which transfers large volumes of funds for member banks and their customers through its computerized com munications system. The dollar volume of funds trans ferred by the Federal Reserve Communications System in 1975 was almost seven times as large as the amount handled by the Fed’s check clearing system.3 The average transfer is quite large — about $1.8 million in 1975 — but the number of wire transfers is only a small fraction of the number of checks handled. To the extent that wire transfers reduce transaction costs and processing time they can be said to improve the efficiency of the payments mechanism. However, the use of wire transfers has not significantly reduced the vast flow of paper through the payments system. A major reduction in paper volume would require implementation of electronic systems designed to be utilized for smaller retail-oriented payments, such as the systems described below. DESCRIPTION OF SOME EFT SYSTEMS For the sake of clarity and simplicity, E F T systems can be grouped into three categories: teller machines, point-of-sale systems, and automated clearing houses. These systems differ in types of payments handled and in means of processing. Teller Machines Machines through which an individual may conduct various routine banking services can be grouped un der the heading of teller machines. Much of the recent EFTS development has involved these ma chines, which are called customer-bank communica ZAnnual Report of the Hoard of Governors of the Federal Reserve System (1 9 7 5 ), p. 379. 1976 tion terminals (C BC Ts) by the Comptroller of the Currency and remote service units (RSU s) by the Federal Home Loan Bank Board. In principle, these machines can be located either on a bank’s premises or elsewhere, conceivably at great distances from the bank. They may be manned or automatic and vary greatly in complexity, ranging from simple communi cations terminals to more complicated automated teller machines (A TM s). Services which ATMs can typically perform include receiving deposits, dispens ing funds from checking or savings accounts, trans ferring funds between accounts, making credit card advances, and receiving payments. The less compli cated manned terminals handle the communications between the customer and his bank while the receipt or disbursing of funds is physically accomplished by the clerk who operates the terminal. Teller machines are usually accessed by a combina tion of a magnetic stripe card (on which account information is encoded) and a personal identification number which, for security reasons, is known only to the customer. If the device is connected “on-line” to the bank’s computer, the customer’s account is up dated immediately; otherwise, a record maintained in the machine is periodically delivered to the bank for processing. Point-of-Sale Systems On-line systems which allow customers to transfer funds to merchants in order to make purchases are usually called point-of-sale (PO S) systems. Systems of this type may be used for check authorizations and credit card transactions, as well as for so-called “debit card” transactions in which funds are immediately transferred from the purchaser’s account to the mer chant’s account. Conceivably, they could also be used for instantaneously transferring funds between busi nesses and/or individuals, using terminals or push button telephones. Large-scale POS systems generally operate in the following manner. On-line terminals are located at check-out counters or other points of sale. When mak ing a purchase, a customer’s card is inserted into a terminal which “reads” the data encoded on it. Other data concerning the transaction are entered manually by the clerk or through an electronic cash register or products code reader. If the customer’s bank is dif ferent from the store’s bank, a switching and proces sing center (SPC ) connects the computers of the two banks. The computer at the customer’s bank verifies that the card and identification code are valid and that the customer’s account has sufficient funds. The Page 11 FEDERAL RESERVE BANK OF ST. LOUIS customer’s account is debited for the amount of the purchase while the store’s account is credited for the same amount. Both parties to the transaction receive a printed statement at the time of the transaction, and the customer’s regular bank statement contains a de scriptive listing, much like those which many credit cards presently use. Since some 30 percent of personal checks are written to grocery or other retail stores,4 this type of system would allow automation of a sub stantial portion of check payments as well as many payments presently made by cash or credit card. Automated Clearing Houses Another type of E F T system, which is conceptually different from the two systems described above, is the automated clearing house (ACH ). As the name sug gests, an ACH is analogous to a traditional clearing house, in that it represents a system for the interbank clearing of debits and credits. The main difference between automated and conventional clearing houses is that the debit and credit items in an ACH exist in the form of electronic signals, whereas they are paper items in a conventional clearinghouse operation. An ACH is thus not a system for automating the handling and clearing of paper checks. The payment items must enter the system in the form of electronic data, usually computer-generated magnetic tape. To illustrate how an ACH works, consider how a payroll payment could be made directly to an em ployee’s checking or savings account through an ACH. The employee authorizes his employer to make such direct payments, eliminating the need for a check to pass through the employee’s hands, be endorsed, cashed or deposited, and sent through the check clear ing process to the employer’s bank. The employer prepares the payroll data on computer tape and sends it to the company’s bank. The bank directly credits the employee’s account if he is a customer and com bines the data for the remaining payees with those from other employers on a magnetic tape. This con solidated tape is delivered or transmitted to the ACH, where a computer processes all the data for a day in a single run, sorting out all payees for each participat ing bank. Each bank then receives a computer tape (or paper advice if it is not equipped to handle tape) which lists the payees and the amount to be credited to each account. The employer’s accounts are debited by the originating banks. Net settlement among the 4This estimate is based on surveys reported in Research on Improvements of the Payments Mechanism: The Final Re port on Phase I, An Analysis of Payments Transactions and Phase II, Payments Flow Data, Volume 1 of 3 (Atlanta: Georgia Tech Research Institute, 1971), p. 30. 12 Digitized forPage FRASER SEPTEMBER 1976 banks is accomplished in the same manner as with paper checks. ACHs are especially suited for handling recurring payments, such as payroll, social security, or pension payments, or recurring payments made by individuals. Payors would authorize their banks to pay a specified amount to a payee (mortgage lender, insurance com pany, etc.) on a specified date. Parties to these types of payments would receive a descriptive statement documenting the payment. Many types of payments — those where the payee and amount vary — are not amenable to this type of preauthorization. The case of regularly recurring bills of varying amounts, such as utility bills, represents a middle ground between the extremes of identical re curring payments and more or less random payments. The customer could, for example, authorize his bank to pay the amount billed by a specified creditor.5 Since many people are reluctant to give such broad power to their banks, a system called Bill Check has been developed.6 This type of payment allows the customer to control the amount and timing of payments to creditors but still achieves some of the benefits of ACHs. An ACH thus differs considerably from teller ma chines and POS systems. The ACH is essentially a “batch” processing system used for the interbank set tlement of recurring credits and debits, whereas many of the other systems allow instantaneous transfers of funds between the customer and his bank or from the customer to third parties. COSTS OF FUNDS TRANSFERS Costs of the Check-Based Payments System Knowledge of the number of checks written in a year and the cost of processing them is imprecise. However, it is estimated that in 1975 between 25 and ‘’American Express has initiated such a service for credit card customers in California. Based on a preauthorization, the full amount of a credit card bill would be paid from the cus tomer’s bank account to American Express, unless the custo mer objected within a specified period after receiving the bill. See “American Express Will Begin Testing Preauthor ized Payments through CACHA,” Payment Systems News letter (February 1975), p. 6. 6The Bill Check itself is a portion of the bill on which the customer indicates the amount to be paid and signs his name, and then returns to the creditor. This completed form au thorizes a debit from the customer’s account to pay the bill. The creditor transfers the data to computer tape and sends the tape to the ACH for processing. The paper Bill Check is retained by the creditor for its records. See Atlanta Payments Project, Automated Clearing Houses: An In-Depth Analysis (Atlanta: Committee on Paperless Entries, 1974), pp. 35-39. FEDERAL. RESERVE BANK OF ST. LOUIS 30 billion checks were written7 involving total proc essing costs of around $6 billion.8 When a payee cashes or deposits a check through a teller line, lock box, or other arrangement, the first bank receiving the check must encode the amount of the check in magnetic ink character recognition (M IC R ) readable symbols. This completes the encod ing of the check (the bank’s routing number and payor’s account number are already encoded on the check), so most of the remaining processing can be conducted by machines capable of reading the MICR characters. The transit items (those drawn on other banks) are sent to the bank on which they are drawn, either directly or through a clearing house, correspondent bank, or the Federal Reserve’s check collection facilities. When the check returns to the bank against which it is drawn, the writer’s account is debited, and the cancelled check is returned to the writer with the periodic statement. Altogether, the average check is handled some ten times and passes through two and one-third banks.9 Much of the processing described above is now automated. Machines read and sort the encoded check according to destination and perform most of the accounting functions. However, a number of the processing functions have not been amenable to auto mation. Many checks are still handled by tellers, the encoding process requires human handling, and the checks must still be physically transported through the banking system and back to the payor. These manual processes appear to be among the most expensive in the check processing function. The average costs of these various processes were esti mated by the Atlanta Payments Project through sur veys of Atlanta banks.10 The estimated cost of re ceiving an item through a branch was 7.4^ and the cost of proof and encoding an item was 1.3^. To gether, these accounted for nearly 60 percent of the 7According to one estimate more than 24 billion checks were processed in 1974, with the number increasing at a 7.3 per cent annual rate between 1971 and 1974. See R. William Powers, “A Survey of Bank Check Volumes,” Journal of Bank Research (Winter 1976), pp. 245-56. sThe latest estimates of average check processing costs of commercial banks are in the 16-21e range. To these costs must also be added the indirect costs borne by writers and receivers of checks, as well as Federal Reserve expenses for check clearing. See Arthur D. Little, Inc., The Consequences of Electronic Funds Transfers, prepared for the National Science Foundation (1975), p. 51. 11Mark J. Flannery and Dwight M. Jaffee, The Economic Implications of an Electronic Monetary Transfer System (Lexington, Mass: D.C. Heath and Company, 1973), p. 41. 10Atlanta Payments Project, Automated Clearing Houses: An In-Depth Analysis, pp. 218-19. SEPTEMBER 1976 13.9«* cost of processing an “on-us” item (a check de posited in the bank on which it was drawn) and more than 80 percent of the 10.6^ cost of a transit item.11 Given this situation, the prospects of lowering the average cost of check processing materially below its present levels do not appear to be good. Economies of scale in check processing may well be nearly ex hausted. As labor and other costs rise, the average cost of check clearing is also likely to rise. These factors have led many to believe that radical changes in the payments mechanism are necessary if costs are to be kept at present levels or be reduced. If the paper document which carries the payments data can be replaced by an electronic signal, manual han dling can be significantly reduced and the flow of payments data accelerated by high speed data transmission. Cost Characteristics of EFT S Although few cost data are available because of the limited experience with EFTS, some dated but representative cost estimates are available. Most E F T systems involve large total cost, much of which is associated with the expensive computer hardware necessary to operate these systems: computers, ter minals, and communication links. Even teller ma chines, which are among the less expensive types of systems, involve purchase costs that may exceed $40,000 per ATM, depending on features. More complicated systems involve higher total costs. An on-line POS system would involve high costs not only for the banks’ computers, the terminals in stores, and the communications links joining them, but also for the SPC which interconnects the com puters of the different banks. In one study, published four years ago, the Atlanta Payments Project esti mated total costs for a proposed POS system linking the banks in that city.12 These included about $650,000 for SPC processing equipment and $655,000 for SPC development costs, about $1,200 for each terminal and associated communications equipment, and about $54,000 for bank communications interface equipment. n An interesting sidelight of this study is that much of the cost of processing a check is borne by the bank of first deposit rather than the bank on which the check is drawn. Of course, the cashing bank frequently receives new deposits in the process which can compensate it for the costs it bears. '-Atlanta Payments Project, Research on Improvements of the Payments Mechanism: Phase III General Systems Design and Analysis of An Electronic Funds Transfer System, Vol ume 3 of 6 Systems Design and Analysis-Point of Sale System (Atlanta: Georgia Tech Research Institute, 1972), Chapter 8. Page 13 FEDERAL RESERVE BANK OF ST. LOUIS SEPTEMBER 1976 COMPETITIVE QUESTIONS SURROUNDING EFTS E F T S has such a significant potential for changing costs and providing new services that it will likely have far-reaching effects on the structure of and competition within the financial industry. In turn, these competitive forces will doubtless influence the course of E F T S im plementation, reflecting developments within the finan cial industiy and concerns of legislators and regulatory agencies. Many people expect that in the years to come E F T S will be one of the principal areas of competition among banks. The changing cost structure associated with new ways of making payments also has considerable implica tions for the competition between commercial banks and thrift institutions. Competition and Regulation Competition between firms and industries is generally considered a good thing, since it forces the sellers to pass the benefits of technological change on to the consumer in the form of lower prices a n d /o r improved service. From the standpoint of the whole economy, competition also leads to an efficient allocation of resources. In competitive markets, it is a normal form of adjust ment for inefficient firms to incur losses and eventually drop out of the market. However, because of the widely held belief that considerable social costs are involved in bank failures, the depository financial institutions have been subjected to regulation which in general limits competition. Although there have been some recent reg ulatory efforts toward encouraging (o r allowing) more competition, there is a distinct possibility that some limits will be placed on competition through E F T S , especially if increased bank failures or pressures on thrift institu tions are expected to result from such competition. Thus, some efficiency gains which might result from E F T S probably will be foregone in order to avoid increased failures. Effect on Structure Many observers expect that E F T S will lead to major changes in the structure of the banking industry. Since teller machines can provide many routine banking serv ices and are much less expensive than traditional “brick and mortar” branches, many observers believe they will be the “branches” of the future. If so, there would be con siderable competition as banks attempt to gain customers by installing these machines in convenient locations. It has been argued that big banks, through E F T machines, will enter the market areas of smaller banks all across the country and, because of the huge initial costs of E F T sys tems, the smaller banks will not be able to compete. It is not clear, however, that smaller banks would be at a disadvantage in operating these machines. F o r ex Digitized forPage FRASER 14 ample, relatively small banks might find that prudent deployment of these machines would enable them to expand their market areas at a much lower cost than with a conventional branch. These machines could thus prove relatively more beneficial to the smaller banks than to the larger ones which have the capability to op erate a large number of conventional branches. A major regulatory question is involved here: are teller machines branches within the meaning of the M cFadden Act, which subjects branching by national banks to state branching laws, or simply communications devices? If the former is ruled to be the case, deployment of the teller machines would be subject to state branching re strictions and regulatory controls; if the latter is accepted their deployment might be relatively unfettered. All but one of the court decisions rendered to date have found the machines to be branches under the M cFadden Act. Since this Act defines a branch to include “any branch place of business . . . at which deposits are received, or checks paid, or money lent,” banks affected by the rulings have been forced to cease these functions at their remote teller machines which are not located in a permissible branch. Savings and loan associations are not subject to the M cFadden A ct and hence have not been subject to the same limitations as banks. Partly as a result of this apparent regulatory inequality, legislation which would exempt bank teller machines from these branching limi tations has been enacted in some states and is pending in others. Ownership of E F T Systems The question of ownership of E F T systems has fre quently been raised. Some individuals and groups argue that operation of competing E F T systems would be uneconomical and could thus lead to increased bank failures. This argument is based, at least in part, on the belief that E F T systems are “natural monopolies,” much like local public utilities. The essential characteristic of the economic concept of natural monopoly is that economies of scale are so per vasive that average costs decline over the entire extent of demand in the market. One seller of the product could produce a given volume at minimum average cost, whereas two or more sellers would each be inefficiendy small, necessitating higher prices. The issue of ownership of E F T systems thus turns on the empirical question of whether there are extensive economies of scale in such systems. Although there is as yet little evidence on which to make a judgment, some generalizations can perhaps be made. There do appear to be economies of scale in such systems, but it is not at all clear that these economies of scale are so extensive that only one firm could operate a large-scale E F T sys tem efficiently. F o r example, both major bank credit card concerns (which have large networks of participat ing banks and merchants as well as extensive computer systems for authorization of transactions and interchange of charge items among the banks) are presently develop ing POS systems, as are some larger correspondent banks and various other organizations. A retail store could utilize more than one competing system through a single dial-up type terminal and existing telephone lines. The ownership question is applicable to ACHs as well. ACHs have developed similarly to conventional clearinghouses, with only one ACH serving a geograph ical area. All but two operating or planned ACHs are operated by the regional Federal Reserve Banks, the principal justification being that these ACH services are basically extensions of the present check clearing func tion of the Fed. As in the case of POS systems, there is the question of whether competing ACHs would be feasible. The Justice Department, apparendy believing that competition could develop in the provision of automated clearing services, has questioned whether the Federal Reserve System or other governmental agencies should directly operate the ACHs: The greatest danger is that a Federal E F T system — especially one that is artificially low-priced — could deter private competitors from deploying pri vate E F T systems, and inflict severe injury on those who have already begun to operate them . . . Many private entities stand ready to supply the same sort of facilities and services that Reserve Banks offer to financial institutions . . . The Board [of Governors should] . . . play the role of overseer, and leave the practical day-to-day operations of the ACH to the private sector with the marketplace acting as the “regulator.”1 Role of Thrift Institutions The role of thrift institutions in E F T systems is cur rently much at issue. These nonbank institutions have been in the forefront of some E F T developments, notably in teller machines and POS systems. This is partly be cause E F T S provides the thrifts an entry into the pay ments mechanism which legislation has largely denied to them, and partly because some laws and regulations have been less restrictive of E F T S development by thrifts than by banks. The involvement of thrift institutions is currendy at issue in many ACHs. Utilization of ACHs has usually been opened to all banks in an area, but thrift institutions have frequently been denied access, at least on an equal footing with banks. ’ Comments of the United States Department of Justice in the Matter of Regulation J, reprinted in American Banker (June 15, 1976), pp. 4 ff. Originally thrift institutions w ere required to use a “pass through” arrangement in which transfers to and from the thrift institutions were made through a bank which had direct access to the ACH. Of course, this in volved additional costs and some delays for the thrifts. Interim guidelines adopted by the Board of Governors in January allow thrifts to receive direct delivery of credit or debit items from Fed-operated ACHs if the institutions have sufficient volume to warrant it and are on an existing courier route. The guidelines also allow any institution which is a member of an automated clear ing house association to send items to the ACHs. Previ ously only those institutions with demand deposit author ity were allowed to originate these items. The thrifts believe that these guidelines still discriminate against many of them and leave them at a competitive disad vantage relative to the banks with more direct access. It is a long-standing principle of antitrust law that where there is a unique resource such as a clearinghouse, all competing firms should have access to it.2 At present there are legal distinctions between banks and thrifts, a major one being the ability to offer accounts from which customers can make payments to third parties. Demand deposit powers have traditionally been limited to com mercial banks, but thrift institutions have long sought third party payment powers, primarily as a source of deposits. Although thrifts have developed some close sub stitutes for checks (notably NO W accounts) they ap parently hope to bypass the cumbersome and expensive check-based payments system and to enter the payments mechanism directly through E FT S . Many of the services already offered by thrifts serve as close substitutes for third party payment powers. For example, the First Federal-Hinky Dinky TMS system in Nebraska allows customers to deposit or withdraw funds through transfers from or to the supermarket’s First F ed eral account. This is tantamount to a third party POS system. Introduction of services like these has forced banks to offer competing services. The distinction be tween savings accounts at thrift institutions and demand deposits at banks is being blurred, and with it the dis tinction between banks and thrifts themselves. Greater competition among types of financial institu tions can be expected to result from development of E F T S . A blurring of the distinctions between financial institutions and a lowering in price of services (and an increased return on deposits) could be expected, in the absence of restrictive regulation. Consumers stand to benefit as financial institutions compete in offering in centives to customers to shift from checking to E F T services. The economy as a whole could also benefit through a more efficient payments mechanism. -Donald I. Baker, “Antitrust and Automated Banking,” Bank ing Law Journal (September 1973), p. 703-18. Page 15 FEDERAL RESERVE BANK OF ST. LOUIS The Federal Reserve Bank of Cleveland has esti mated total costs of operating a larger POS system, which would link 39 banks in the Fourth Federal Reserve District.13 Assuming that the system would capture 6-7 percent of total retail transactions, in the tenth year the average costs of operating the SPCs alone would be an estimated $3.8 million. Costs of participating banks in the tenth year would be an other $13.9 million, it was estimated. Total costs for the first ten years of operation would exceed $100 million. Many of the costs which are associated with E F T systems do not vary with the number of transactions and thus can be considered as fixed costs. An example is the depreciation of the computers, terminals, and other equipment used in the system, which would be the same regardless of the level of usage. Some other costs probably do not vary significantly with the level of usage either. The costs of operating a computer used solely for an E F T system would probably be little higher at relatively high levels of transactions than at low. There are other costs which do vary with the level of output or usage. For example, labor costs associ ated with operating a manned terminal may be greater at higher levels of usage than at low. The cost of computer time necessary to process teller machine or POS transactions would also depend on the number of transactions. One study estimated the cost of the infinitesimal amount of computer time neces sary to process a POS transaction to be about 1^ per transaction.14 To date, ACHs have utilized an existing computer (in most cases at a Federal Reserve Bank), since a computer run processing a whole day’s accumulated items usually takes an hour or less. The processing time and labor costs do not appear to vary greatly with the number of transactions processed. Estimated costs of operating the Atlanta ACH at various levels of output are shown in Table I. A characteristic of the costs of many E F T systems is evident in this table: the average cost of a transac tion declines significantly as the number of transac tions increases. This is caused by the predominance of relatively fixed costs in most E F T systems. From the standpoint of the Federal Reserve System, which operates most ACHs, a high level of output would 13Arthur D. Little, Inc., The Consequences of Electronic Funds Transfer, pp. 195-8. 14Atlanta Payments Project, Research on Improvements of the Payments Mechanism: Phase III, Vol. 3 of 6, p. 166. Page 16 SEPTEMBER 1976 Table 1 ESTIMATED ACH O P E R A T IN G CO ST S M onthly Transaction Volume M onthly Costs Cost Per Transaction 20 ,000 $ 2 ,1 5 0 1 0.8 « 1 5 2 ,5 0 0 2,650 03.3 2 8 5 ,0 0 0 3,1 5 0 01.8 4 1 7 ,0 0 0 3,6 5 0 01.1 5 5 0 ,0 0 0 4,1 6 0 00.8 Note: These estimates are for the Atlanta ACH. The program used requires about one hour of processing time for any daily vol ume up to 25,000. Thus, some costs will be relatively fixed up to the monthly volume of 550,000 shown in the table. The Project anticipates that software changes could expand the hourly volume to 75 - 100,000 ; operating costs would remain about the same, and the average cost of a transaction would drop to about 0.20. Source: Atlanta Payments Project, Automated Clearing Houses: An In-Depth Analysis, Chap. 10. have to be reached before the average cost of a transaction would fall below the Fed’s average cost of clearing a check, which is about l£. The data on Federal Reserve check processing and ACH costs presented in Table II indicate that, at current vol umes, processing an ACH item is much more expen sive to the Fed than processing a check. Of course, this means that ACHs are presently operating at far less than efficient volumes. Furthermore, potential cost savings to participating banks from making elec tronic payments through an ACH rather than using checks should also be considered. Data from the Atlanta Payments Project suggest that monthly vol umes of more than 160,000 transactions at the Atlanta ACH would result in the average cost of an ACH transaction falling below the average cost of a check.15 Estimates of average costs of POS transactions also suggest that if sufficiently large volume is attained, the average cost of a POS transaction can fall below the average cost of a check. In the Cleveland Fed study, referred to above, the average cost per trans action over a three-year period was estimated to be 11.3<; this would be reduced to 7.1< by the tenth year because of expected economies of scale in operation of the system.16 15Atlanta Payments Project, Automated Clearing Houses: An In-Depth Analysis, p. 229. This estimate assumes no bank marketing costs associated with the ACH operation. 16Arthur D. Little, Inc., The Consequences of Electronic Funds Transfer, pp. 195-98. The Cleveland Fed Study estimates that the POS system would generate cost savings such that the annual savings would exceed annual costs by the seventh year and cumulative savings exceed cumulative costs by the tenth year. This estimate did not include sav ings experienced by merchants as a result of using the POS system. FEDERAL RESERVE BANK OF ST. LOUIS SEPTEMBER 1976 Table II FEDERAL RESERVE PAY M EN T S M E C H A N IS M EXPENSES — C O N V E N T IO N A L C H ECK P R O C E S SIN G Checks Processed A U T O M A T E D C L E A R IN G H O U SE P R O C E S SIN G Total Expense Cost Per Check1 $ 1 2 0 ,5 5 9 ,0 0 5 0 .9 9 5 * (thousands) 1 1 ,4 1 1 ,3 3 7 1975 “ Check Im a g e s" Processed Total Expense Cost per " Im a g e " (thousands) 5,941 $ 4 3 1 ,8 8 3 7 .2 7 1 1This item excludes postage and expressage on items handled by others, which were included in the "Total Expense” category. Source: Board o f Governors o f the Federal Reserve System, Functional Expense Report, 1975 Annual Report. Another cost concept should be considered. This is marginal cost, which is simply the change in total cost which results from increasing output by one unit. Marginal cost is thus not affected by fixed cost but only by variable costs. In the case of E F T systems, the marginal cost would be the change in cost that resulted from conducting one additional transaction. The considerations described above suggest that the marginal cost of an E F T transaction is likely to be very low. In most cases, little or no additional labor is involved, and only an infinitessimal amount of com puter time is used. Thus, an additional transaction should add very little to the cost of the system. In comparison, the marginal cost of a check is probably relatively high because of the labor intensiveness of check processing.17 In summary, many E F T systems will involve large fixed costs but relatively small variable costs. Because of this predominance of fixed costs, average costs of E F T transactions will probably decline over a fairly large range of output. The hypothetical cost curves Hypothetical Costs of Funds Transfers COST NUMBER OF TRANSFERS 17One study estimated the marginal below the average cost estimated nery and Jaffee, The Economic tronic Monetary Transfer System, cost to be 14c, not much in the same study. Flan Implications of an Elec p. 42, footnote 10. presented in the accompanying figure show the aver age cost of an electronic funds transfer to be greater than the average cost of a check at relatively low volumes. Beyond some point, however, the average cost of checks levels off while that of electronic funds transfers continues to fall. The average cost of elec tronic transfers is less than that of checks for relatively large volumes. Although these cost relationships are hypothetical, they demonstrate the general relation ship between the costs of check systems and EFTS. CUSTOMER ACCEPTANCE OF EFTS If the potentially lower unit costs of E FT S are to be realized, a substantial proportion of present payments must be switched to electronic means. Otherwise, banks may find themselves with the worst of both worlds: a large, slow, expensive check handling sys tem, plus a very expensive underutilized E F T net work. Thus, failure to consider customer acceptance of E FT S could have exactly the opposite effects from those hoped for. At present, there seems to be little incentive for customers to change their payments practices from using checks to E F T systems. Surveys have found that individuals are generally satisfied with the present payments system. Using checks provides many bene fits, including unquestioned proof of payment. The surveys also indicate that many individuals, especially those in lower income groups, have a negative atti tude toward many aspects of E F T systems.18 Custo mers tend to fear a “loss of control” over their finances which would result from preauthorized deposits or withdrawals, as well as possible losses of privacy, costly errors, and lack of proof of payment. Thus, in many cases, it may take considerable incentive to induce individuals to shift their payments from checks to electronic means. 18See Arthur D. Little, Inc., The Consequences of Electronic Funds Transfer, pp. 43-46, 253-63. Page 17 FEDERAL RESERVE BANK OF ST. LOUIS Such an incentive could be provided if the cost savings which would result from a fully utilized E F T system were passed on to the customer. However, current regulations reduce the likelihood of banks pro viding such incentives. For example, since banks are prohibited by Regulation Q from paying an explicit return on demand deposits, they usually pay an im plicit return by subsidizing the checking costs of their customers, either totally or in part.19 Thus, the cost to the individual of writing an additional check is usually less than the marginal cost to society of processing the check. As a result, “too many” checks are written, and a greater than optimal quantity of resources is allocated to check processing. As long as banks are prohibited from explicitly paying interest on demand deposits, it does not seem likely that they will charge fees for checking services which approximate the marginal costs of the checks. As a result, the prospects of banks providing sufficient incentives for customers to switch from checking to the lower-marginal-cost electronic funds transfer serv ices are reduced. lfl“ ‘Yields’ on Checking Accounts Rise in Recent Years,” Federal Reserve Bank of Philadelphia Business Review (March 1975), pp. 14-15. Page 18 SEPTEMBER 1976 The underutilization of E F T systems which would stem from the prohibition of interest on demand de posits could be worsened if customers receive addi tional disincentives to use E F T systems. Such disin centives could take the form of high initial prices for E F T services which banks may charge, predicated on the notion of recovering the large fixed costs of the E F T systems relatively quickly. SUMMARY Many of the questions surrounding E F T S cannot yet be answered definitely, since they depend on costs, regulations, and other factors for which there are few or no data. However, it is reasonable to expect marginal costs of making a transfer through E FT S to be less than through the check system; average costs of transfers could also be lower, if a sufficiently large volume is achieved. Such possibilities for reducing costs, other things equal, would encourage banks and other producers to increase their supply of these serv ices relative to checking services. However, such fac tors as the prohibition against paying interest on de mand deposits appears to reduce the incentive for customers to use E F T systems, thus slowing their development. EIGHTH FEDERAL RESERVE DISTRICT HEAD OFFICE, BRAN CH , A N D OTHER CITIES OF O V E R 10,000 POPULATION