The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
f e d e r a l rebetwe l< m J c clew / a n d E C O N O M IC R E V I E W IN THIS ISSUE What Happens When the Unemployment Rate Changes? ............. 3 Defining the Product Market in Commercial Banking ..17 JU N E-JU LY 1972 WHAT HAPPENS WHEN THE UNEMPLOYMENT RATE CHANGES? Robert A. McMillan Would a four percent rate o f unemployment today indicate the same labor market conditions as i t did ten years ago? What about future years? The answers to these questions have im portant implications fo r the nation's policy makers who are responsible fo r making decisions fo r the present and establishing goals fo r the future. This article reports on a study o f the relationships between the aggregate unemployment rate and the unemployment rates o f various labor market components. The existence o f stable, predictable relationships would obviously perm it a greater understanding o f the consequences o f a change in a given unemployment rate by defining the groups who would be affected and the extent to which they would be affected. The first step o f the analysis made use o f statistical techniques to establish the degree to which a given overall unemployment rate is associated with the unemployment rates o f specific labor market components. Historical series fo r 26 labor markets representing age-sex, race, industrial, and occupational classifications were used in the computations. The results indicate that significant statistical relationships do exist between the aggregate unemployment rate and the component rates. For example, i t is possible to estimate the effect o f a decrease from a 5 to a 4 percent unemployment rate on the unemployment rates o f workers in wholesale and retail trade. The results also indicate a change in the structure o f unemployment over time and that some groups would be more seriously affected by changes in the overall unemployment rate now than in the past. I t is possible to identify these groups and their potential unemployment rates at a given aggregate rate. In addition to exploring the relationship between the unemployment rate and the component rates, this study examined the implications o f changes in the overall rate in terms o f how many people are affected, fo r how long, and why. In general, it was found that high rates o f unemployment im ply an extended duration o f unemployment, more people experiencing some unemployment, and a higher proportion o f job losers among the unemployed. 3 ECONOMIC REVIEW W h ile there seems to be a consensus that the present level of unemployment is higher than unemployment from 5 to 4 percent have a disproportionate impact on unemployment rates optimal, there is no consensus over an appropriate of various labor market subgroups, such as blue- unemployment rate.1 The importance that policy collar workers, teenagers, construction workers, makers attach to the attainment of alternative, and and nonwhites? Does a 5 percent aggregate rate often conflicting, goals such as price stability and imply a different composition of unemployment full employment differs. In addition, the problem today than a decade ago? Who is affected and how of choosing appropriate unemployment targets is are they affected? compounded by compositional changes in the labor force over the past decade. These changes have led some observers to suggest that an upward THE NATURE OF UNEM PLOYM ENT RELATIONSHIPS revision in the unemployment goal is appropriate. It would be impossible to draw systematic Before a meaningful choice of an aggregate conclusions about the composition of unemploy unemployment goal can be made, some questions ment if individual labor markets responded errat must be answered about the composition of ically relative to changes in the overall rate of unemployment under varying labor market condi unemployment. Therefore, tions. It must be determined if the same unem establish whether or not consistent, reasonably ployment rate in two different periods implies two stable relationships exist between individual sub it is important to widely divergent sets of labor market conditions, group and if the difference between two unemployment unemployment rate.2 The technique of linear rates, fo r example, 4 and 5 percent, is only one of regression was used to examine the relationship scale or if it also represents major changes in the between the overall rate of unemployment in the composition of unemployment. The purpose of this study is to determine the unemployment rates and the aggregate United States, as measured by the Bureau of Labor Statistics, and unemployment rates of selected implications of the overall unemployment rate on labor market groups. Quarterly unemployment individual labor markets and to answer questions data representing 26 age-sex, industrial, race, and such as: Does a movement in the aggregate rate of occupational classifications and extending as far 2 1 A 4 percent "interim " unemployment goal gained wide acceptance after it was first officially proposed in 1962 by the Council of Economic Advisors. Recently, the Joint Economic Committee of Congress advocated a 4 percent rate as an "interim " goal and a 3 percent rate as a long-term goal. See U. S., Congress, Joint Economic Committee, 1972 J o in t Economic Report, (Washington, D. C.: Government Printing Office, 1972). Others have stated that a 4 percent unemployment rate is a "m yth" as a peacetime norm and that pushing the unemployment rate significantly below 5 percent by monetary and fiscal policy alone is no longer possible without adding signifi cantly to inflationary pressures. See Wall Street Journal, January 21, 1972, p. 26. 4 This study focuses exclusively on the composition of unemployment. Other important aspects of unemploy ment are the effect of changes in unemployment on labor force participation and hours worked. At high rates of unemployment, persons might become "discouraged" after a period of searching for a job and leave the labor force, and many workers might involuntarily work less than full time. For example, see this author, "Discouraged Workers and the Unemployment Rate,” Econom ic Commentary, Federal Reserve Bank of Cleveland, March 27, 1972; Michael L. Wachter, "A Labor Supply Model for Secondary Workers," Review o f Economics and Statistics, May 1972; and Claire Hodge and James Wetzel, "Short Workweeks and Underemployment," M o n th ly Labor Review, September 1967. JU N E-JU LY 1972 back as 1948 were used in the regressions. o The such as police and fire protection, has little results of these regressions and a discussion of relationship some of the technical details are in the Statistical demand for other government services may even Appendix. expand and require more workers during economic All 26 equations are highly significant and to overall economic conditions; contractions than during expansions. indicate a close relationship between the aggregate Also, both white and nonwhite groups show a rate of unemployment and the rates of the various highly subgroups.4 The high R2's in most of the equa unemployment rate. Ninety-seven percent of the tions indicate that the regression equations explain white group's unemployment rate and 84 percent by far the bulk of movements in the unemploy of the nonwhite group's rate could be explained predictable relationship to the overall ment rate of the groups. For example, the R2 of by the equations. However, the standard errors of .95 in the regression for white-collar workers estimate indicate that the unemployment rate for indicates that 95 percent of the variation in their whites conforms more closely with the overall rate unemployment rate is explained statistically by than the rate fo r nonwhites. the aggregate unemployment rate and time trend. Moreover, a highly stable and Finally, unemployment rates fo r selected predictable occupational groups are closely associated with the relationship is indicated between the aggregate overall unemployment rate. The unemployment unemployment rate and rates in manufacturing, rate of the sales workers group showed the least wholesale and retail trade, and the services. In the association w ith the aggregate rate, although the mining and government wage and salary groups, bulk of the variations in the unemployment rate the relationship is not as strong as fo r the other for this group can be explained by the regressions. groups. Only 39 percent of the variation in the unemployment rate of mining workers and 42 percent o f the variation in the unemployment rate IS UNEM PLO YM ENT MORE CONCENTRATED? of government workers can be explained by the Many economists have been concerned about regression equations. Government employment is the changing composition of the labor force and not as responsive to changes in economic condi its implications fo r the composition and level of tions as employment in most private industries. aggregate Demand fo r some types of government services, changes in the composition of unemployment are 3 Unemployment data are compiled by the U. S. Depart ment of Labor, Bureau of Labor Statistics from the monthly C urrent Population Survey of approximately 50,000 households. 4 Even though the periods selected are different, these results are similar to those derived in Paul M. Ryscavage, "Impact of Higher Unemployment on Major Labor Force Groups," M o n th ly Labor Review, March 1970. Also see. Joint Economic Committee, Higher Unem ploym ent Rates, 1957-1960: S tru ctu ra l Transform ation o r Inade quate Demand, Washington, D. C., 1961. unemployment.5 Unless structural 5 For a discussion of the problems of changing labor composition see G. L. Perry, "Changing Labor Markets and Inflation," B rooking Papers on Econom ic A c tiv ity , 3, 1970, pp. 411-441; Robert A. Gordon, "Has Structural Unemployment Worsened," In d ustria l Relations, May 1964, pp. 53-77; Richard G. Lipsey, "Structural and D e fic ie n t-D e m a n d Unemployment Reconsidered," E m ploym ent Policy and the Labor M arket, ed. Arthur M. Ross (Berkeley: University of California Press, 1965), pp. 210-255; and, Robert M. Solow, The Nature and Sources o f Unem ploym ent in the U nited States (Stockholm: Almquist and Wicksell, 1964). 5 ECONOMIC REVIEW taken into consideration, an estimated relationship women searching fo r jobs in that industry; no between the aggregate unemployment rate and the discernible rates fo r individual labor market groups would not workers; and a decrease fo r most other groups. be very accurate as a predictive tool. A time trend trend fo r trade and government No major change is indicated in the unemploy variable, therefore, was included in the regression ment rate for whites relative to the aggregate rate. equations to determine if a given aggregate rate The rate for nonwhites, however, shows some implies different group rates at different points in im p ro v e m e n t time.6 demand (a more enlightened populace, antidis (decrease). Apparently, both Results of the regressions show an improvement crimination laws, etc.) and supply (for example, (decrease) over time in the adult male unemploy more education) considerations have favorably ment rate that is associated with any specific affected nonwhite employment. This may explain aggregate rate (see Statistical Appendix for regres why rates fo r nonwhites that are associated w ith a sion results and an interpretation of the time specific aggregate rate are somewhat lower today variable in the regressions). For the adult female than a decade ago. group, there is no indication of a change in the The regressions indicate that unemployment unemployment rate associated w ith a given aggre has become less concentrated among blue-collar gate rate. The teenage unemployment rate associ workers and more concentrated among white- ated w ith a specific overall rate, however, shows a collar workers. dramatic worsening technical workers and clerical workers experience (increase) over time. For In particular, professional and example, over the last decade, the teenage unem somewhat ployment rate associated with a given aggregate specific aggregate rates now than in the past. This rate increased over 2 percentage points. This is may reflect some difficulties in absorbing the most likely a result of the tremendous increase in increasing supply of workers with more training higher rates of unemployment at the number of teenagers in the labor force since and education in the labor market, as well as a the late 1950's and a somewhat smaller expansion slackened demand, particularly of job opportunities for this group.7 space industries. A t the same time, unemployment in defense and With respect to industrial classifications, the among craftsmen and foremen and laborers has regressions indicate an increase in the unemploy been moderating relative to the aggregate unem ment rates fo r finance, insurance, and real estate ployment rate. workers, relative to a given overall rate (in part because of rapid increases in the number of 0 The time trend variable tells whether or not a particular aggregate rate today implies a different, perhaps higher rate (if the coefficient is positive), for a group than the same rate did at some previous point in time. IMPLICATIONS OF CHANGES IN RATES OF UNEMPLOYMENT The regression equations discussed in the preceding sections have established that stable and predictable relationships exist between the overall unemployment rate and the rates fo r labor market ^Between 1960 and 1971, the number of teenagers in the labor force increased from 4.8 million to 7.4 million. This represented an increase of 54 percent compared with only a 21 percent increase in the total civilian labor force. 6 groups as defined by age, sex, industry, race and occupation. It was also established that the composition of unemployment associated w ith a JUN E-JULY 1972 TABLE I Projected Group Unemployment Rates at Selected Aggregate Rates of Unemployment Projected Rates Based on 5.0% Aggregate Rate for I Q 1973 Based on 4.0% Aggregate Rate for I Q 1974 Point Change Percent Change Age-Sex 1 2 3 Both sexes 16-19 Males 20 and over Females 20 and over 16.7% 3.7 4.8 15.2% 2.7 4.0 1.5 1.0 0.8 -9 .6 % - 2 8 .0 - 1 6 .3 5.5 4.7 9.6 5.4 5.3 4.5 4.9 3.7 2.2 7.3 3.8 3.4 3.3 4.1 1.8 2.5 2.3 1.6 1.9 1.2 0.8 - 3 1 .8 - 5 2 .6 - 2 4 .3 - 2 9 .6 -3 6 .1 - 2 6 .2 - 1 7 .5 Industries Nonagricultural goods production Mining Construction Manufacturing Durables Nondurables Service industries Transportation and public utilities Trade Finance,insurance and real estate Government 2.9 5.4 1.8 4.4 1.1 1.0 - 3 9 .2 - 1 7 .2 3.0 2.3 2.8 1.9 0.2 0.4 -9 .2 - 1 4 .9 Whites Nonwhites 4.4 8.4 3.5 6.8 0.9 1.6 - 2 0 .6 - 1 9 .7 3.4 3.0 0.4 - 1 3 .0 2.4 2.0 0.4 - 1 6 .5 1.2 4.4 3.5 5.6 3.5 7.1 8.9 5.1 0.9 3.8 3.0 3.9 2.3 5.3 6.7 4.3 0.3 0.6 0.5 1.7 1.2 1.8 2.2 0.8 - 2 3 .0 -1 4 .1 - 1 5 .8 - 3 0 .7 -3 5 .1 - 2 5 .5 - 2 4 .9 - 1 6 .0 4 5 6 7 8 9 10 11 12 13 14 Race 15 16 Occupations 17 18 19 20 21 22 23 24 25 26 White collar Professional and techinical Managers, officials and proprietors Clerical Sales Blue collar Craftsmem and foremen Operatives Nonfarm laborers Service Source: Federal Reserve Bank of Cleveland specific overall rate of unemployment today differs from that which accompanied the same overall rate in the past; that is, unemployment is unemployment at a given overall rate and the effect of changes in the aggregate rate. Who Is Affected by a Given Unemployment more concentrated in certain labor markets today Rate? The implications of given aggregate unem than it was form erly, and less concentrated in ployment rates fo r the various labor force groups others. This section deals with the composition of can be seen from the data presented in Table I. 7 ECONOMIC REVIEW Some groups, such as teenagers, construction duction in the adult male unemployment rate is workers, laborers, and nonwhites, have higher than much larger than the reduction in the rate for average unemployment rates, no matter what the teenagers, the p o in t change in the rate for teen aggregate unemployment rate is. Other groups, agers is somewhat larger than the change in the such as adult males and professional and technical adult male rate (1.5 compared w ith 1.0). The workers, For difference in the two measures occurs because the example, the distribution of unemployment based average unemployment rate fo r teenagers is so have lower than average rates. on a hypothetical 5.0 percent aggregate rate for much higher than the adult male rate that the the first quarter of 1973 shows a 16.7 percent rate same percentage change translates into a higher for teenagers, which is more than three times the p o in t change. aggregate rate. Among industries, unemployment percent in the aggregate unemployment rate would A change from 5 percent to 4 rates fo r workers in durable goods and trade are imply approximately a 20 percent reduction for somewhat above the 5 percent aggregate rate, both the white and nonwhite groups, but the point while the rate fo r construction workers is nearly change for nonwhites would be almost twice as twice the overall rate. Conversely, the rates for large as the change fo r whites. g o ve rn m e n t workers and professional and Among the broad workers occupational benefit much groupings, technical workers are less than one-half the overall blue-collar rate. white-collar workers by a reduction in the overall more than Who is Affected by a Change in the Unemploy unemployment rate, as indicated by both measures ment Rate? The effects of changes in the overall of change. Likewise, by both measures, govern unemployment rate on individual labor markets ment and the finance, insurance and real estate can be examined using the regression results. For groups benefit less from a reduction in the overall example, the hypothetical case in Table I presents rate than the construction and manufacturing the projected distribution of unemployment at an group. Of course, it should be pointed out that assumed overall unemployment rate of 5.0 percent these equations imply that the groups benefiting in the first quarter of 1973 and the distribution if the most from a reduction in unemployment are the unemployment rate were reduced to 4.0 also hurt the most by an increase. percent by the first quarter of 1974.8 The table HOW ARE THE UNEMPLOYED AFFECTED? shows that such a reduction would have different impacts on the various labor market groups. For A change in the overall unemployment rate example, such a change (20 percent) in the overall implies changes not only in the specific groups unemployment rate would entail about a 28 experiencing percent reduction in the unemployment rate of number of adult number of weeks they are unemployed, and the males compared with reductions of 10 percent and unemployment, but also in the people who are unemployed, the 16 percent for teens and adult reasons fo r their unemployment. These aspects of females, respectively. While the percentage re- unemployment are examined in the remaining g These dates were selected for purposes of exposition and do not represent a projection of unemployment rates for these periods. sections of this article. Although in many cases, data are limited (for example, data series on reasons fo r unemployment are available fo r only JU N E-JU LY 1972 TABLE II Average Unemployment, Total Number of Individuals with Unemployment, and Average Weeks of Unemployment per Person Experiencing Unemployment 1957-1970 Unemployment Rate Average Number Unemployed Total Number Unemployed During Year Average Weeks of Unemployment Total During Year 4.3 6.8 5.5 5.5 6.7 5.5 5.7 5.2 4.5 3.8 3.8 3.6 3.5 4.9 2,859 4,602 3,740 3,852 4,714 3,912 4,071 3,785 3,365 2,878 2,977 2,816 2,832 4,088 11,423 13,977 11,994 13,950 14,898 15,057 14,010 13,839 12,131 11,387 11,564 11,332 11,744 14,565 13.0 17.1 16.2 14.4 16.5 13.5 15.1 14.2 14.4 13.1 13.4 12.9 12.5 14.6 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 NOTE: Data refer to individuals 16 years and over; the figures for total individuals with unemployment before 1965 have been adjusted to exclude 14 and 1 5 year olds. Sources: U. S. Department of Labor and Federal Reserve Bank of Cleveland the last five years), they do permit some tentative example, when the unemployment rate increased conclusions. In general, it w ill be shown that rising from 3.5 percent in 1969 to 4.9 percent in 1970, rates of unemployment imply a lengthening in the it reflected a rise in the average n u m b er u n e m duration of unemployment, a larger number of ployed from 2.8 to 4.1 m illion persons. During the people experiencing some unemployment, and a same period, the total number of people experi higher proportion of job losers. encing some unemployment rose from 11.7 to Numbers Experiencing Unemployment. During periods of rising unemployment, such 14.6 m illion persons, while the average number of as weeks that they were unemployed rose from 12.5 1957-1958, 1960-1961, and 1969-1970, both the to 14.6. The same pattern is evident in other total number of people experiencing some unem periods of rising unemployment, indicating that ployment and the average number of weeks they higher unemployment is reflected in both the were unemployed increased (see Table II).9 For cj number unemployed as well as the number of The number of unemployed can be looked at in two ways the average number unemployed at a point in time and the total number experiencing one or more periods of unemployment during a year. The total number of unemployed excludes those who did not work at least one week out of the year. The average number of weeks of unemployment refers to a total for a year, rather than the average length of any one period of unemployment. weeks unemployed. Time Spent in Unemployment. Disaggregation of data on "number of weeks unemployed” into Bureau of Labor Statistics' groupings o f 1-4 weeks, 5-14 weeks, and 15 weeks and over shows that, at high unemployment rates, the number of people 9 ECONOMIC REVIEW CHART 1 DISTRIBUTION OF UNEMPLOYMENT BY TOTAL WEEKS OF UNEMPLOYMENT Percent of Total Unemployment 1957 '59 '61 '63 '65 '67 '69 '71 NOTE: Shaded areas indicate years when over 11 million persons experienced some unemployment. Last entry: 1970 Source: U. S. Department of Labor experiencing over 15 weeks unemployment within can also be identified. a year increases dramatically. In periods of high In addition to examining the proportion of unemployment, such as 1958 and the early 1960's, total unemployed in the three categories, it is also the proportion of unemployed with less than 5 important to look at the change in the numbers of weeks of unemployment was low, while in periods unemployed by length of unemployment because of low unemployment, such as 1967-1969, the the two measures may yield somewhat different less-than-5-weeks group typically constituted a conclusions. For example, a large change in the much larger share of the total with unemployment total number w ith unemployment could bespread (see Chart 1). On the other hand, the group with among the categories in such a way as to leave the unemployment over 15 weeks made up a much proportion larger share of the total in high unemployment numbers in each group increased appreciably. The years and a smaller share in low unemployment number of individuals who experience 1-4 weeks years. The proportion of unemployed in the group of unemployment has been trending upward (see the same, although the absolute 5-14 weeks unemployment did not vary Chart 2). This probably reflects both a larger total greatly in the period studied. Therefore, just as the workforce and increases in the number of women composition of those who are unemployed at low and teenagers in the workforce, as women and with and high unemployment rates can be identified, it teenagers have a preponderance of short-term is apparent from this analysis that the amount of unemployment. Since 1957, there has been little time spent in unemployment at high and low rates correspondence between changes in the number 10 J U N E -J U L Y 1972 CHART 2 NUMBER OF PERSONS WITH UNEMPLOYMENT BY NUMBER OF WEEKS UNEMPLOYED Millions of Persons Last entry: 1970 Source: U. S. Department of Labor experiencing unemployment and the number with Although the data are limited to the period 1-4 weeks of unemployment. Conversely, the since 1962, the adult male group appears to number experiencing 15 weeks or more of unem account ployment has moved closely with the total unem 15-weeks-and-over unemployment category as the fo r most of the shifts toward the ployed. This suggests that high levels of unemploy total number w ith unemployment rises and away ment involve a substantially greater number of from that category as total unemployment falls weeks of unemployment fo r the unemployed than (see Table III). The adult female and teenage do low levels.10 groups also contribute to the lengthening in the 10 unemployment structure. For a discussion of long-term unemployment see Susan S. Holland, "Long-Term Unemployment in the 1960's," M o n th ly Labor Review, September 1965, pp. 1069-1076; or Edward J. O'Boyle, "America's Less Fortunate: The Long-Duration Unemployed," M o n th ly Labor Review, April 1970, pp. 35-43. It should be noted that at any unemployment, teenagers with level of unemployment tend to be clustered in the 1-4 weeks group far more than adult men and considerably more than 11 ECONOMIC REVIEW 11 TABLE III many Part-Year Workers* by Number of Weeks Unemployed (Thousands of Workers and Percent Distribution) 1962-1970 comprehensive, regularly published series of data Year Total 1-4 Weeks 5-14 Weeks 15 Weeks and Ovei 16-19 Years Old 1962 1963 1964 1965 1966 1967 1968 1969 1970 1,631 1,457 1,682 1,743 1,720 1,808 1,877 1,824 1,915 36.5% 34.3 39.0 46.5 51.3 53.1 53.0 53.0 38.5 31.7% 34.5 30.7 29.0 27.7 27.9 26.9 28.3 32.4 31.9% 31.2 30.2 24.4 21.0 19.0 20.2 18.3 29.0 7,044 6,308 6,009 4,945 4,236 4,239 3,952 4,353 6,013 19.1 19.7 21.6 24.9 30.4 29.4 34.0 31.5 23.0 38.8 38.4 40.2 39.9 40.6 42.1 39.7 40.9 41.3 42.1 41.9 38.2 35.2 29.1 28.5 26.3 27.6 35.7 3,467 3,302 3,471 2,931 2,888 2,883 2,968 3,008 3,736 unemployment, a on the reasons fo r unemployment covers only the period since 1967. The data, however, do cover periods of tight and slack labor markets. Table IV shows the distribution 1967-1971 by reason fo r unemployment. Less of unemployment for than one-sixth of the average number unemployed in any of those years were people who were working but left their last job. Likewise, less than one-sixth of the unemployed were new entrants. Those who lost their last job made up the largest proportion of the unemployed, with reentrants the in 1970 and 1971, the number of persons who lost their last job increased dramatically as a propor tion of the total unemployed. Among the adult male group, loss of job was far and away the primary reason fo r unemployment. In the adult female group, job losers were also very important, although not to the extent that they Females 20 and Over 1962 1963 1964 1965 1966 1967 1968 1969 1970 for next largest group. When unemployment increased Males 20 and Over 1962 1963 1964 1965 1966 1967 1968 1969 1970 explanations 28.7 27.6 30.5 35.8 40.9 39.9 43.6 42.5 31.6 34.3 32.0 32.4 34.0 30.8 33.3 32.0 31.2 34.4 37.0 40.4 37.1 30.2 28.3 26.7 24.4 26.6 34.0 were in the adult male group. In the teenage group, reentrants and entrants accounted fo r most of the unemployed. * Worked at least one week during year. Source: U. S. Department of Labor 11 adult women. In turn, unemployment of adult women is clustered in the 1-4 weeks group more than unemployment of adult men. Compared with the adult male unemployment group, the adult female group has about the same proportion in the 15-weeks-and-over unemployment category, but a far smaller proportion in the 5-14 weeks category. Reasons for Unemployment. While there are 12 Some economists have suggested that high unemploy ment rates are predominately the result of increased frictions in the market; i.e., that individuals search longer for jobs, often as a result of imperfect information and unrealistic wage expectations. See for example, Armen A. Alchian, “ Information Costs, Pricing, and Resource Unemployment," Western Econom ic Journal, June 1969, pp. 109-128. These authors treat much of the unemploy ment as a voluntary phenomenon. However, in times of high unemployment, declining quit rates and an increasing number of job losers cast doubt on frictional and voluntary unemployment as adequate explanations of the high unemployment rates. JU N E-JU LY 1972 TABLE IV Unemployed Workers by Reasons for Unemployment (Percent Distribution of Annual Averages) 1967-1971 Total Lost last job Left last job Reentered labor force New entrant Males 20 years and over Lost last job Left last job Reentered labor force New entrant Females 20 years and over Lost last job Left last job Reentered labor force New entrant Both Sexes 16-19 years Lost last job Left last job Reentered labor force New entrant 1967 1968 1969 1970 1971 100.0% 100.0% 100.0% 100.0% 100 .0 % 40.9 14.6 31.4 13.2 38.0 15.3 32.3 14.4 35.9 15.4 34.1 14.6 44.3 13.4 30.0 12.3 46.3 11.8 29.4 12.6 100.0 100.0 100.0 100.0 100.0 63.9 15.6 18.3 2.4 60.3 16.8 20.6 2.2 57.7 17.0 22.4 2.8 65.1 12.8 19.4 2.7 66.3 11.4 19.6 2.7 100.0 100.0 100.0 100.0 100.0 36.9 16.5 41.7 5.0 34.6 17.0 42.8 5.6 33.0 16.8 44.8 5.4 40.5 15.9 39.3 4.3 42.2 14.2 39.3 4.3 100.0 100.0 100.0 100.0 100.0 17.6 10.9 34.6 36.9 15.5 11.6 33.5 39.3 14.8 11.8 34.5 38.8 18.1 11.4 34.3 36.3 18.5 9.2 32.5 39.8 Source: U. S. Department of Labor From the data available, it appears that rising only did the number of job losers increase dispro unemployment rates are closely associated with an portionately to the other groups, but the amount increase in job losers among the unemployed. Even of time spent in unemployment by job losers also though the composition of unemployment by increased disproportionately. For example, when reason is different in the teenage, adult male, and unemployment increased by 1.9 m illion persons adult female groups, there was a sizable increase in between 1969 and 1971, some 1.3 m illion were the proportion of job losers in all of these groups job losers. The percentage of job losers who were when unemployment rose between 1969-1971. unemployed 15 weeks or more rose from 17.6% to Based on data since 1968, job losers experi 31.0%. Although the percentage of leavers, enced longer duration unemployment than job entrants, and reentrants with long-term unemploy leavers, new entrants, or reentrants (see Table V). ment also increased, the increase among these In addition, when unemployment increased, not groups was not as dramatic. 13 ECONOMIC REVIEW (1) TABLE V Unemployment by Reason and Length (Thousands of Workers and Percent Distribution of Annual Averages) 1968-1971 Total Unemployed Less Than 5 Weeks 5-14 Weeks 2,817 2,831 4,088 4,993 1,070 1,017 1,809 2,313 15 Weeks and Over 28.8% 29.2 31.5 31.6 14.6% 13.2 16.1 23.7 49.4 50.6 44.6 36.3 31.4 31.8 34.7 32.7 19.2 17.6 20.7 31.0 o f unemployment than the makeup of the unemployed. (2) A high aggregate rate of unemploy distribution of unemployment than a low rate. Also, an increase or decrease in aggre gate unemployment has an uneven impact on the various labor market groups. For example, blue-collar in relative terms, adult males, workers, and manufacturing workers are disproportionately affected by 431 436 549 587 59.6 60.6 57.3 46.3 26.0 27.8 28.5 32.5 14.4 11.7 14.3 21.2 909 965 1,227 1,466 407 413 503 627 changes in aggregate unemployment. (3) Different overall rates of unemploy ment are associated w ith differences in the way the unemployed are affected. A t higher 61.9 62.3 59.4 54.2 27.4 26.4 28.6 29.5 10.7 11.3 12.0 16.3 60.7 60.3 57.1 52.2 27.8 30.8 30.4 31.6 11.6 8.9 12.5 16.3 Never Worked Before 1968 1969 1970 1971 a different ment at a point in time implies a different 56.6% 57.5 52.3 44.7 Reentered Labor Force 1968 1969 1970 1971 implies the age, sex, race, occupation, and industry Left Last Job 1968 1969 1970 1971 today same rate, say, a decade ago, w ith respect to Lost Last Job 1968 1969 1970 1971 specific aggregate rate o f un composition Total Unemployed 1968 1969 1970 1971 A employment average rates of unemployment, both a larger number of people are affected and the number of weeks of unemployment are greater. Moreover, the proportion of those Source: U. S. Department of Labor w ith long-term unemployment rises, while the proportion of those w ith short-term unemployment falls. Finally, the reasons for unemployment differ. For example, at high rates of unemployment, job losers represent a much larger proportion of the unemployed SUMMARY AND CONCLUSIONS than at low rates. This study concludes that there are identifiable The implications of these findings suggest that and reasonably stable relationships between the the consequences of specific aggregate unemploy overall unemployment rate and rates fo r individual ment targets on various labor market groups in labor markets. As a result, the impact of various terms o f aggregate unemployment rates on individual labor unemployment can be predicted w ith a reasonable length, amount, and distribution of market groups can be estimated, and the groups degree of certainty. However, it is important to who are affected the most by a change in the note that the difference between alternative aggre aggregate rate can be ascertained. It was also gate unemployment rates is not simply a matter of fou nd: scale. 14 JU N E-JU LY 1972 STATISTIC AL APPENDIX This Appendix reports the results of regressions .056 on the time variable in the regression for that examine the relationship between the aggre teenage group indicates that for a given aggregate gate unemployment rate and the unemployment rate o f unemployment in any quarter, the teenage rate of various labor market groups. In each case, rate w ill be .056 of a percentage point higher than the dependent variable is the unemployment rate it would have been the previous quarter given the of various labor market groups, classified by age, same aggregate rate. sex, race, industry, and occupation. Each equation has tw o independent variables—the aggregate rate A simple test was constructed to determine if there is a different relationship between the of unemployment and a time trend. The regression aggregate unemployment rate and the rate fo r equations fo r various individual groups at high levels of unem age, sex, and industry employ quarterly data from the first quarter of 1948 ployment from the relationship at low levels. If through the first quarter o f 1972. The regressions this were the case, a regression model in which the for unemployment rates by race use data fo r the aggregate unemployment rate was entered linearly period first quarter of 1954 through first quarter would be a biased, inefficient tool in forecasting 1972, and the occupational data cover the first the composition o f the labor force at various quarter of 1958 through the first quarter of 1972. aggregate unemployment rates. In order to test for The time periods fo r race and occupation were this possible source of bias, the regressions were shorter because data fo r earlier years are not run again w ith the data reordered according to size available. of the aggregate unemployment rate. A careful appears in various check of the Durbin-Watson statistic and the original regressions. The results reported in the pattern of residuals was then made to determine if Table are adjusted fo r serial correlation through the equations were misspecified. In no case was the Cochran-Orcutt transformation. In no case was the Durbin-Watson statistic less than 1.39, and in more than one iteration required, and there were most cases it was 1.80 or higher, indicating that no substantial changes introduced in the coeffi "serial" correlation (which here would imply a cients by. this process. systematic relationship other than a linear one Some serial correlation The coefficient on the aggregate rate of unem ployment in each regression shows the association betw een aggregate unemployment and the dependent variable) was not a problem. However, between movements in the aggregate rate and it should be noted, that at high aggregate rates, movements in individual market unemployment there was some tendency to overestimate the rates. For example, the coefficient of .99 in the teenage rate and to underestimate the white-collar "males 20 and over" regression indicates that a rate. A t low aggregate rates, the regression tended movement of one percentage point in the aggregate to underestimate the manufacturing rate and to rate is associated w ith a .99 percentage point overestimate the construction change in the unemployment rate of adult males. none of these groups exhibit behavior that would rate. To repeat, The coefficient on the time variable measures make it necessary to estimate different equations whether becoming at different aggregate unemployment rates, and among various labor corrections can readily be made fo r the small or not unemployment increasingly concentrated is market groups. As an example, the coefficient of biases discussed. 15 Regression Results of Relationships Between Unemployment Rates of Individual Labor Market Groups and the Aggregate Unemployment Rate Selected Time Periods By Quarters Dependent Variable Unemployment Rate Constant Aggregate Unemployment Rate Time Variable Adjusted R2 F Test DurbinWatson Standard Error o Estimat Age-Sex 1948 I - 1972 I 1 Both sexes 16-19 2 Males 20 and over 3 Females 20 and over 1.54 - 0 .2 7 0.89 1.85 (17.50) 0.99 (44.85) 0.78 (26.87) 0.056 (9.59) - 0 .0 1 3 (-1 0 .0 3 ) 0.001 (0.57) .82 212.80 2.02 .59 .96 1030.15 2.17 .12 .89 365.00 2.09 .19 1.66 (33.37) 2.25 (7.29) 2.21 (17.13) 1.57 (26.41) 1.89 (21.00) 1.09 (21.20) 0.86 (25.76) 1.05 (16.51) 0.92 (26.01) 0.32 (11.93) 0.34 (8.28) -0 .0 1 9 (-4 .4 4 ) -0 .0 5 2 (-3 .4 0 ) - 0 .0 2 8 (-4 .2 1 ) - 0 .0 1 5 (-2 .8 2 ) -0 .0 1 4 (-1 .7 6 ) -0 .0 1 5 (-5 .9 2 ) -0 .0 0 4 (-2 .8 5 ) - 0 .0 2 0 (-6 .8 2 ) 0.001 (.61) 0.008 (7.19) -0 .0 0 1 (-0 .5 5 ) .92 561.15 2.08 .23 .39 30.74 2.20 2.01 .76 151.19 1.99 .79 .88 350.04 2.09 .28 .82 220.97 2.06 .42 .83 234.40 2.07 .33 .87 331.97 1.85 .38 .76 151.70 2.06 .50 .88 341.56 2.03 .26 .69 106.28 1.79 .31 .42 34.30 2.01 .23 0.91 (70.39) 1.58 (18.31) 0.001 (1.43) -0 .0 1 7 (-3 .0 7 ) .97 2535.83 2.24 .07 .84 182.39 2.01 .35 0.52 (32.56) 0.47 (13.25) 0.28 (16.20) 0.70 (25.46) 0.56 (8.70) 1.64 (23.26) 1.15 (23.20) 1.86 (17.55) 2.02 (20.63) 0.78 (14.07) 0.015 (13.80) 0.020 (8.01) 0.003 (2.30) 0.018 (9.69) 0.005 (0.92) -0 .0 1 5 (-2 .4 8 ) - 0 .0 2 0 (-5 .7 3 ) 0.005 (0.48) -0 .0 5 3 (-7 .7 0 ) - 0 .0 0 5 (-1 .3 0 ) .95 539.51 1.61 .11 .77 88.85 1.86 .17 .86 167.40 1.83 .12 .93 341.76 1.67 .18 .63 46.68 2.03 .32 .92 320.29 1.91 .22 .95 494.91 1.90 .27 .86 166.69 2.01 .33 .94 451.15 2.07 .51 .83 135.79 1.84 .24 Industries 1948 I - 1972 I 4 5 Nonagricultural goods producing wage and salary Mining wage and salary 6 Construction wage and salary 7 Manufacturing wage and salary -1 .4 1 8 Durables wage and salary -3 .1 1 9 Nondurables wage and salary 10* Service producing industries wage and salary 11 Transportation and public utilities wage and salary 12 Trade wage and salary - 0 .8 2 - 1 .5 9 1.55 1.01 0.62 - 0 .3 7 0.76 13* Finance, insurance and real estate 14 Government wage and salary 0.43 0.58 Race 1954 I - 1972 I 15 White workers 16 Nonwhite workers - 0 .1 3 2.07 Occupations 1958 I - 1972 I 17* White collar 18 Professional and technical 19* Managers, officials and proprietors 2 0 * Clerical - 0 .4 3 - 1 .2 0 - 0 .1 8 - 0 .3 3 0.74 21 Sales workers 22 Blue collar workers - 1 .3 8 23 Craftsmen and foremen - 1 .0 3 24 Operatives - 2 .7 7 25 Nonfarm laborers 1.87 26 Service workers 1.85 NOTE: Figures in Parentheses are t-values. * Not reestimated. Source: Federal Reserve Bank of Cleveland JU N E-JU LY 1972 DEFINING THE PRODUCT MARKET IN COMMERCIAL BANKING Joel M. Yesley The products or services provided by commercial banks and the markets in which these products are merchandised are viewed in two ways. Under one concept, commercial bank services are considered to complement each other to such a degree that banks operate within a narrow product market—competing only with other banks. The other concept holds that commercial bank services can be considered on an individual basis and that banks operate in a broad product market—competing with banks and other financial insititutions. This article examines these opposing viewpoints in academic literature, court decisions, and statistical studies. Most o f the theoretical and statistical work has supported the broad market concept. The Supreme Court decisions, however, have favored the narrow market concept. The similarity of financial services provided by Furthermore, in this view, customers are limited to nonbank financial institutions and by commercial banks if they wish to satisfy all their financial banks has stimulated a considerable amount of needs at one institution. controversy Therefore, nonbank bankers, financial institutions, which generally provide only administrators of banking regulations, and courts one or two basic deposit or loan products, are not regarding the institutional scope of the product able to compete effectively w ith banks. among economists, markets in which banks operate. Most are of the The definition opinion that the similarities among the individual central services provided by these two general types of significance institutions—w ith only a few exceptions such as markets caused by the entry of new institutions checking and mergers of existing ones. Those who regard accounts—place them in direct issue of in of the product market is a determining structural the competitive changes in banking competition w ith each other. Others, however, banks as multiple product firms that compete in hold a considerably narrower view of the scope of several product markets attach less significance to the product market in commercial banking. This the loss of an independent bank in a specific group contends that banks sell their services in geographic market because of a merger than "packages” rather than individually and, in effect, adherents of the narrow product market concept. produce a single composite product that cannot be The latter group, by excluding the products of duplicated nonbank financial institutions from consideration, by other types of institutions. 17 ECONOMIC REVIEW would regard a bank merger as contributing to a institutions are also briefly reviewed. The article greater concentration of financial resources in a concludes w ith a discussion of possible areas for given market area. further research that should be useful in appraising The report of the Hunt Commission, which was changes in banking structure. made public in December 1971,1 has generated considerable interest and discussion regarding the degree of competition among the various types of depository financial institutions (commercial NATURE AND RECENT GROWTH OF FIN A N C IA L INSTITU TIO N S associations, mutual Nature. Financial intermediaries channel funds from savers to borrowers by purchasing pri savings banks, and credit unions). The Commission mary securities (debt or equity instruments of banks, savings and loan recommended that these financial institutions be nonfinancial permitted to provide a wider range of financial from services and that all depository institutions be divided into two broad classifications: commercial subject to the same reserve requirements, tax banks and nonbank financial institutions. The treatment, interest rate regulations. w ith funds obtained These intermediaries can be The latter category includes depository institutions primary aim of the Report is to improve the such as savings and loan associations, mutual efficiency of the financial system by eliminating savings banks and credit unions, as well as life differences in regulations that may have inhibited insurance companies, investment companies, and open competition among depository institutions. finance companies. The major difference between Implementation of these recommendations would these two types o f intermediaries is in the form of pave debt the and institutions) depositors. way fo r a considerably more that each creates. Commercial banks homogeneous financial structure and significantly substitute enlarge the institutional scope of the product primary market in commercial banking. institutions create various forms of nonmonetary This article reviews the development of the money (deposits and currency) for securities, whereas nonbank financial claims on themselves, such as savings deposits, controversy concerning the product market in shares, equities, and other assets. In addition, commercial banking in both economic literature commercial banks administer and in the courts. In the absence of a universally payments accepted body of economic theory on this vital credits among spending units. mechanism by the national transferring deposit issue, courts have been left to define the business The assets and liabilities of commercial banks of banking. As w ill be illustrated in this article, are, in general, considerably broader than those of decisions between the lower courts and the nonbank financial institutions. Commercial banks generally receive short- and intermediate-term number of recent empirical studies that have deposits (ranging balances analyzed the significance of various influences on accounts payable on demand up to certificates of the deposit Supreme Court demands have for 1 the frequently assets of differed. A depository Hunt Commission, The Report o f the Presidents Commission on Financial Structure and Regulation (Washington, D. C.: Government Printing Office, 1971). 18 w ith from wide-ranging in checking maturities) from individuals and all forms of governmental and business organizations, and they supply funds of varying maturities to an equally diverse group of JU N E-JU LY 1972 DISTRIBUTION OF MORTGAGES BY SOURCE OF LENDING 1970 OTHERS 14.3% SA VING S ASSOCIATIONS 33.5% FEDERA L G O VE R N M E N T AGENCIES 6.7% LIFE INSURANCE COMPANIES 16.4% M U TU A L SAVINGS BANKS 12.9% CO M M ERCIAL BANKS 16.2% Source: U. S. Savings and Loan League borrowers in a number of forms (e.g., rediscounts, lim iting term loans, instalment loans). In addition, banks structures. As of yearend 1970, savings and loan them to the financing of residential provide a number of specialized services, such as associations had allocated just over 85 percent of trust facilities and safe deposit boxes. their total assets to mortgage loans and accounted institutions, for one-third o f the total credit extended to the however, receive the bulk of their funds from Other mortgage markets (see Chart 1). Mutual savings households. The purchases made by nonbank banks, which are nonprofit th rift organizations financial institutions, including those that do not operated solely fo r the benefit o f depositors, also accept deposits, are usually concentrated in one allocate most of their total assets (73 percent as of type yearend 1970) to residential mortgages. of types liquid of asset. depository Credit unions, which concentrate their assets in consumer loans to Specialization o f assets is least pronounced for members, and investment companies (e.g., mutual finance and life insurance companies. There are funds), which concentrate in corporate equities, two types of finance companies: sales finance and are nonbank consumer finance or small loan companies. Sales associations, finance companies make loans, generally on a the most specialized intermediaries. Savings which funds accept and loan may of be short- or intermediate-term basis, for the purchase withdrawn w itho ut notice, are the next most of durable and capital goods by individuals and specialized institutions because of legal restrictions businesses. Consumer finance companies lend only that types generally 19 ECONOMIC REVIEW TABLE I Total Assets of Financial Intermediaries 1950-1970 (Billions of Dollars, Percent Distribution and Growth Rates) Growth Rates Intermediary 1950 Commercial banks Savings and loan associations Mutual savings banks Credit unions Life insurance companies Private pension funds State and local government pension funds Finance companies Open-end investment companies T O TA L $149.9 16.9 22.4 0.9 62.6 6.7 1960 54.1% 6.1 8.1 0.3 22.6 2.4 5.0 9.3 3.3 1.8 3.4 1.2 $277.0 100.0% 1970 $230.9 71.5 40.6 5.0 115.9 38.2 41.0% 12.7 7.2 0.9 20.6 6.8 19.6 24.1 17.0 $562.8 19501960 $516.9 176.2 79.0 15.4 200.5 110.8 3.5 4.3 3.0 100.0% 58.0 60.4 47.6 40.9% 13.9 6.2 1.2 15.9 8.7 4.6 4.8 3.8 4.4% 15.5 6.1 18.7 6.4 19.0 8.4% 9.4 6.9 11.9 5.6 11.2 14.6 10.0 17.8 11.5 9.6 10.9 100.0% $1,264.8 19601970 7.4% 8.4% Source: Board of Governors of the Federal Reserve System, Flow of Funds Account to individuals for a wide range of purposes, over this period. Most nonbank financial including the purchase of goods and services. Both institutions, however, still grew at significantly types of finance companies acquire funds by faster rates. In 1950, about 54 percent of the total issuing commercial paper and long-term bonds and assets of financial intermediaries was accounted by borrowing from banks. Life for by commercial banks (see Table I). By 1960, accumulate funds from this share had declined to 41 percent and has premium payments and issuing stock, and they remained fairly stable since then. A substantial insurance invest short-term companies prim arily in corporate and government part of the decline was caused by the growth of other financial intermediaries, particularly savings bonds. financial and loan associations and life insurance companies. intermediaries have increased fivefold since World Regulatory restrictions on commercial banks and a Growth. The total assets of all War II. Commercial banks, however, have shared shift in public preferences away from demand considerably less in this growth than other types deposits and business loans toward interest-bearing of financial intermediaries. For example, between liquid assets and consumer and mortgage loans 1950 and 1960, the assets of commercial banks have been cited as the major causes of this uneven grew at an average annual rate of 4.4 percent, growth. which is well below the growth rate of other types of financial institutions (see Table I). Over the past decade, the growth of commercial bank assets accelerated to an 8.4 percent annual rate, the average growth rate for all financial institutions 20 2 Jules Backman and Arnold W. Sametz, "Workable Competition in Banking," The B ulletin o f the C. J. Devine In stitu te o f Finance, New York University, No. 22 (November 1962), p. 30. JU N E-JU LY 1972 ECONOMIC THEO RY AND BANKING MARKETS The question of what constitutes the product product market concept of commercial banking. His theoretical model of the banking firm emphasizes relationships of complementarity or market in commercial banking has received a interdependency among financial services provided considerable amount of attention in the literature o on banking structure. Economists generally agree by commercial banks.4 Individual banks encourage deposit retention by assuring deposit customers of that nonbank financial institutions are collectively immediate able to produce close substitutes fo r the individual terms. financial services provided by commercial banks, relationship” w ith the major exception of checking accounts. complementarity However, the concept of the bank as a multiple other loan accommodation and favorable Hodgman uses to the term describe the "customer high degree of between deposits, loans, and financial services, payroll such product firm competing in several markets with administration, various nonbank institutions is not universally collection and investment counsel. as preparation, trust account He asserts that the prevailing concept of the accepted. The Bank as a Single Product Firm. Donald relevant product market in literature pertaining to Hodgman is the major advocate of the single bank regulation is developed prim arily around the lending and investing activities o f commercial banks, rather than the input or deposit market. 3 This question also has important implications in the area of monetary policy. John G. Gurley and Edward S. Shaw assert that control over the stock of money as narrowly defined (currency in circulation plus demand deposits at commercial banks) is not sufficient to stabilize the economy since consumer expenditures are determined by the total stock of liquid assets including those held at nonbank intermediaries. Since they regard these institu tions as direct competitors of banks, they question the effectiveness of monetary policy, especially in infla tionary times when depositors at commercial banks are likely to shift funds into nonbank financial institutions for higher returns. This increases the lending ability of those institutions at a time when monetary policy is attempting to restrict bank loans. See Gurley and Shaw, "Financial Intermediaries and the Saving-lnvestment Process," Journal o f Finance, May 1956 and Money in Theory o f Finance (Washington, D. C.: The Brookings Institution, 1960). On the other hand, Milton Friedman has adopted a considerably narrower view of the product market in commercial banking, asserting that neither mutual savings bank deposits nor savings and loan shares are very close substitutes for time deposits at commercial banks. This conclusion is based mainly on observations of patterns in the growth rates of these types of deposits from 1954 to 1958. See Milton Friedman and Anna Jacobson Schwartz, A M onetary H istory o f the U nited States, 1867-1960 (Princeton: Princeton University Press, 1963). Hodgman argues that this popular conception of the bank as an institutional investor, maximizing the interest return available from a wide range of lending and acceptable investing alternatives risks, is misleading. carrying In Hodgman's opinion, this view, which supports the multiple product concept, ignores the dependence of deposits and loans upon the provision of banking services. He refers to tw o widespread practices in banking that cannot be explained by the prevailing concept. One is nonprice credit rationing, which is frequently adopted by banks in periods o f credit restraint. Banks usually cite a responsibility to long-term deposit customers when declining loan requests from nondepositors instead of quoting unrealistically high rates. This preference for depositors is based on the contribution of service 4 Donald R. Hodgman, Commercial Bank Loan and Investment Policy (Champaign, Illinois: University of Illinois, 1963). 21 ECONOMIC REVIEW income to long-run bank profits rather than a credit were in effect. loans. This study indicated that interest rates charged Hodgman asserts that a bank can charge a lower by banks were the lowest where the purchasing contract rate on a loan to a depositor than to a bank handled all of a town's business instead of maximization of interest returns on nondepositor and still earn a higher effective rate merely holding the proceeds of a note issue as they of return because of charges for complementary were being -spent. Furthermore, concessions on services. Furthermore, a loan to a depositor loan rates tended to be greater as the loan size constitutes a smaller drain on a bank's lending increased, capacity than an identical loan to a nondepositor. expectations of larger and more profitable deposit The other practice that does not support the accounts. It was concluded that "nonprice terms concept of banks as institutional investors is the and tendency of bankers to sell government securities importance in explaining the determinants of at a capital loss to accommodate prime borrowers commercial bank in times of tight money. further, more comprehensive investigation." In view of the close relationship between deposits and loans, Hodgman bases his definition a practice deposit The attributed relationships Bank as pricing a are to of bankers' sufficient behavior to warrant Multiple Product Firm. Economists who view banks as multiple product of the product market on the to ta lity of services firms generally make a distinction between the provided by banks. He does not believe that business and nonbusiness customers of a bank in commercial a n a ly z in g banks compete significantly with th e other types o f financial intermediaries since the interdependencies. latter are not individually capable o f duplicating spokesman the maintains that basic product of commercial banks—the safeguarding and transfer of money—or the wide range of services that comprise the "customer relationship." The customer confirmed in a relationship study of hypothesis 30 was Massachusetts municipalities that borrowed in anticipation of tax revenue.5 The tax anticipation notes issued by Massachusetts towns are generally purchased by commercial banks that also hold deposits of these towns. This group of bank borrowers was considered ideal for testing the hypothesis, since for strength David the of product Alhadeff, a leading broad market concept, The available evidence suggests that package sales are typically made compulsory only for business borrowers (and not to all o f them) in the form of a compensating balance requirement. Significantly, the tie-in sales are restricted to those services (business loans and transactions deposits) for which banks are the dominant or sole suppliers whereas bank services that nonbanks also supply (such as home mortgage loans, consumer loans, and savings accounts) can usually be negotiated separately.7 their notes were homogenous, the nonprice terms of the loan were standardized, and no lines of 5 Neil B. Murphy, A Test o f the Deposit Relationship Hypothesis, Staff Economic Study No. 38 (Washington, D. C.: Board of Governors of the Federal Reserve System, 1970). 22 Qlb id ., p. 59. ^David A. Alhadeff, "Monopolistic Competition and Banking Markets," in M onopolistic C om petition Theory: Studies in Im pact, ed. by Robert C. Kuenne (New York: John Wiley & Sons, Inc., 1967), pp. 364-365. JUN E-JULY 1972 He views commercial banks as department Alhadeff has also asserted that some nonbank stores of finance, producing a number of distinct financial institutions compete with commercial services. He has divided these services into five banks categories: former loans, safekeeping facilities (e.g., for demand deposits, even though the institutions do not offer third-party deposit accounts and safety boxes), investment payment facilities. He divides demand deposits outlets for the public in the form of time and into tw o categories: those held for transactions savings accounts, checking facilities in the form of purposes and those held as liquid reserves to serve demand as deposit accounts, and miscellaneous buffers Although specialized services such as trust facilities. against unexpected commercial banks expenditures. are unique in Alhadeff, along with Clifton Kreps, believes satisfying the transactions needs of the demand that competition between banks and other types depositor, Alhadeff contends that credit unions, o f financial institutions is more significant in savings and loan associations, and mutual savings credit product markets than in the deposit markets, partly because their deposit services tend to be more homogeneous than their lending services. O Both economists have focused their banks provide highly liquid interest bearing deposit substitutes for demand deposits held as a liquid reserves. A more generalized form of this hypothesis was originally proposed by James analyses on the consumer, mortgage, and business Tobin and William J. Baumol, who maintained loan markets. They assert that commercial banks that the volume of idle transactions balances compete w ith a wide range of other institutions, increases at a slower rate than total transactions including finance companies, credit unions, savings and income and is inversely related to the level of and loan associations, mutual savings banks, life interest rates available on alternative assets.9 insurance companies, and a number of Federal agencies in the consumer and mortgage loan markets. Alhadeff and Kreps agree that banks are generally isolated from significant competition in JUDICIAL INTERPRETATIO N OF THE PRODUCT M ARKET IN COMMERCIAL BANKING the business loan market, especially in the small The Federal laws that were designed to loan segment. Alhadeff notes that, other than promote competition in the banking industry have commercial finance companies, nonbank financial not attempted to define the product market in institutions are not very active in this market. commercial banking.10 The three Federal bank However, he does not regard finance companies as regulatory effective competitors Corporation since they specialize in serving marginal risk borrowers, who would likely have d iffic u lty obtaining loans from commercial ban ks. g David A. Alhadeff, "A Reconsideration of Restrictions on Bank Entry," The Q uarterly Journal o f Economics, May 1962, and Clifton H. Kreps, Jr., "Characteristics of Local Banking Competition," Banking and M onetary Studies (Homewood, Illinois: Richard D. Irwin, Inc., 1963). agencies—Federal (FDIC), the Deposit Comptroller Insurance o f the 9 William J. Baumol, "The Transactions Demand for Cash: An Inventory Theoretic Approach," Q uarterly Journal o f Economics, November 1952; and James Tobin, "The Interest-Elasticity of Transactions Demand for Cash," Review o f Economics and Statistics, August 1956. 10For a discussion of these laws see "Federal Laws Regulating Bank Mergers and the Acquisition of Banks by Registered Bank Holding Companies," Econom ic Review, Federal Reserve Bank of Cleveland, January 1971. 23 ECONOMIC REVIEW Currency, and the Board o f Governors of the CO M PETITIVE IMPLICATIONS OF THE BANK MERGER ACT OF 1966 The primary purpose of the Bank Merger A c t o f 1966, which amended the Bank Merger A c t o f Federal more Reserve System—have therefore had a or less free hand competitive significance institutions in ruling in determining the o f nonbank on financial bank mergers. The 1960, is to clarify the application of antitrust laws decisions of these agencies, however, are subject to to bank mergers. It permits some acquisitions review fo r possible violation of anti-trust laws by that under a strict the Department of Justice. If the decisions are application of the competitive standard established challenged, the cases are initially tried in the U. S. would have been illegal in the Clayton A ct, providing any anticompetitive District Courts and may come before the Supreme effects were clearly outweighed Court if an appeal is made. by increased The first case in which the Supreme Court public benefits. The language of Section 7 of the acquisitions of treated the product market as a major issue in business concerns that would substantially lessen evaluating bank competition involved a proposed competition merger between the second and third largest banks Clayton A ct, which prohibits in any line of commerce in any section of the country, was incorporated in the in Philadelphia.11 1966 Merger A ct w ith only one change, the narrow The lower court accepted the product market concept, stressing the exclusion of the phrase "in any line of commerce,” unique scope o f commercial banking services and The Supreme Court has not attached any the high degree o f interdependency among them. significance to the omission and continues to The action to enjoin the merger, however, was regard bank services in total as the relevant line of dismissed on the basis of a finding of no violation commerce of the anti-trust laws. or product market o f commercial The Supreme Court reversed the decision of the banking. The defendant banks in merger cases, however, have argued that Congress intended that lower court, prim arily because o f a difference of the competitive implications of mergers be judged opinion on a m ultiple product basis, which would permit However, the higher court agreed w ith the district the consideration of nonbank financial institutions court's product market definition, mainly because as competitors of banks. The determination of a of the uniqueness and significance of the demand merger's legality would, therefore, require some deposit. the bank's services.* * 1968. 24 the geographic market. (T)hey [commercial banks] alone are permitted by law to accept demand deposits. This distinctive power to accept demand deposits...makes banks the intermediaries in most financial transactions (since transfers of substantial moneys are almost always by check rather than by cash) and, concommitantly, the repositories of very judgment as to the community's need for each of Franklin R. Edwards asserts that the courts were directed to ascertain the overall competitive impacts of mergers on the basis of a weighing of the effects on the separate product lines, including "commercial banking" as a distinct product. See "Bank Mergers and the Public Interest: A Legal and Economic Analysis of the 1966 Bank Merger A ct," The Banking Law Journal, September regarding 11 United States v. Philadelphia National Bank,ef. al. 210 Supp. 348 (1962); 83 S. Ct. 1715 (1963). JUN E-JULY 1972 substantial funds.12 The court individual and corporate decisions. This lower court ruled that savings and loan associations, commercial finance companies, further reasoned that the domi government lending agencies, credit unions, and nance of highly liquid demand deposits in the life insurance companies provided reasonable deposit portfolios of banks made them the most substitutes for many of the financial services important source o f short-term business credit. offered by banks. The court also noted that the Moreover, it found that other bank services are not combined share o f the deposits and loans of all subject to competition from nonbank financial financial institutions in California accounted for institutions, even though the services are similar. by the two banks proposing to merge would be The court mentioned consumer loans and savings only one-half the comparable percentage based deposits as examples, arguing that banks can only obtain funds more cheaply than finance companies difference o f product market perspective was an on commercial bank resources. This and therefore lend at lower rates, and also that important factor inducing the judge to rule against savings deposits of commercial banks enjoyed a the Department of Justice, which did not appeal "settled consumer preference" over those of other the decision. The Provident Bank Case institutions. 1R was the first merger In the Lexington Bank Case (1964),13 the case to reach the Supreme Court under the Bank Supreme Court reaffirmed its narrow concept of Merger A c t o f 1966 in which the definition of the the banking, product market became a major issue. A fter the emphasizing once again both the uniqueness of the case was remanded to the district court for further checking account and the wide scope of banking consideration on procedural grounds, the district services, specifically mentioning deposit boxes, court reversed its original decision and ruled the Christmas Club proposed merger between the fifth and seventh product market in commercial accounts, correspondent bank facilities, collection services, and trust department largest banks services. market was in violation of the law. The court in the four-county Philadelphia The first major merger case to come before a however, rejected the narrow concept o f the district court under the Bank Merger A c t o f 1966 product market held by the Department o f Justice was the Crocker-Anglo Citizens Bank Case (1967), on grounds that by om itting the phrase "in any which involved the fifth and seventh largest banks line of commerce" in the Bank Merger A c t o f in California.14 The district court adopted a 1966, Congress had intended a test of the overall considerably competitive impact of a merger in a number of broader concept of the product market than the Supreme Court had in previous individual product markets. The lower court alluded to the uniqueness of 12 Ibid., p. 326 (Footnote omitted). 13 United States v. First National Bank and Trust Company of Lexington, et. at., 208 Supp. 457 (1962); 84 S. Ct. 1033 (1964). 14 United States v. Crocker-Anglo National Bank, et. al., 263 Supp. 125 (1966); 277 Supp. 133 (1967). demand deposits and the interdependency of commercial bank services, especially for corporate customers. However, the interrelationships of 15 United States v. Provident National Bank, et. al., 262 Supp. 297 (1966); 87 S. Ct. 1088 (1967). 25 ECONOMIC REVIEW these services were not strong enough to preclude The Supreme Court took issue w ith the lower direct competition from other types of financial court regarding the appropriateness of analyzing institutions, which the court limited to mutual competition between commercial banks on a savings banks and savings and loan associations on submarket or separate product basis. It asserted the basis that that the use of submarkets should be confined to m eaningful banks..." "they alone offer direct and commercial cases involving mergers between commercial banks In stressing competition for savings competition to and other types of financial institutions. The deposits, the lower court denied that commercial Court emphasized customers' desires to satisify banks enjoyed a "settled consumer preference," as all their financial needs at a single institution, the Supreme Court had asserted in the Philadelphia Bank Case. The increasing importance making nonbank financial institutions less attractive than banks. It also appeared to accept of this competition in recent years was noted, as Hodgman's notion of the customer relationship in illustrated by the decline in the proportion of referring to interdependencies among the services demand to time deposits at all national banks from provided by banks: 70 to 30 percent in 1960 to a rough equivalency as of the end of 1967. The Supreme Court's most recent affirmation of the narrow concept of the product market occurred in the Phi 11ipsburg National Bank Case (1970). third 17 and The proposed merger was between the fifth largest banks located in the two-city area of Phillipsburg, New Jersey, and Easton, Pennsylvania. Finding that banking as a whole was not the relevant line of commerce, the lower court analyzed competition in terms of individual product lines. Competition between the banks involved in the merger and other types of For some customers, full service banking makes possible access to certain products or services that would otherwise be unavailable to them; the customer w ithout significant collateral, for example, who has patronized a particular bank fo r a variety of financial products and services is more likely to be able to obtain a loan from that bank than from a specialty financial institution to which he turns simply to borrow money. In short, the cluster o f products and services termed commercial banking has economic significance well beyond the various products and services involved.18 financial institutions, such as savings and loan The Court distinguished between small and large associations, credit unions, insurance and finance banks in applying this version of the "customer companies, was emphasized. The lower court also relationship" hypothesis, since this was the first noted that the banks involved in the merger were merger case involving two relatively small (less operated than $30 m illion in total deposits) banks to come more in the manner of savings institutions than larger commercial banks. before it. The Court asserted that customers who would tend to hold relatively small accounts would 1 fi United States v. Provident National District Court Opinion (1968), p. 40. Bank, et. at.. 17United States v. Phillipsburg National Bank and Trust Company, et. at., 306 Supp. 645 (1969); 90 S. Ct. 2035 (1970). 26 be likely to have their checking and savings accounts in the same local bank, even when higher savings 18 U. S. Supreme Court Disallows Bank Merger", The Banking Law Journal, April 1971, p. 353. JUN E-JULY 1972 accommodation o f borrowing needs. On the other DEMAND AND SUPPLY OF FINANCIAL ASSETS hand, the district courts have maintained that the specialized Interest Rate nature of nonbank financial institutions has not significantly hindered their ability to compete w ith commercial banks, especially in the markets for savings deposits and home mortgage loans. STATISTIC AL EVIDENCE While the controversies in economic theory and in the courts over the relevant product market have been unresolved, the statistical evidence strongly supports the broad market concept. This Quantity section reviews five recent statistical studies on the substitutability of time deposits among different Economic theory postulates that the demand curves for financial assets are upward sloping (line types o f depository institutions and one study that d). If it is assumed that the quantity supplied is infinitely elastic at the observed yield (line s), the quantity of a particular asset will then be deter directly tested the theory o f a bank as a single product firm. Four of the five studies on mined solely by the level of demand at the point of intersection with the horizontal supply schedule substitutability and the other study support the broad (point e). product assumption of market the concept. An im plicit studies on deposit interest rates were available elsewhere. This general substitutability banking relationship would prevail because the deposits are infinitely or highly elastic w ith respect convenience factor and the advantages of a good to their "o w n " rate of interest (see Figure).19 One relationship w ith the local banker would be more 19 important for small depositors and borrowers, who would have less access to a variety of financial institutions than larger customers. In summary, the district courts have generally accepted a somewhat broader concept o f the product market in commercial banking than the Supreme Court. The Supreme Court has emphasized both the uniqueness of some of the individual services provided by commercial banks and the high degree of interdependence among them. The Court attributed this interdependence to the desire of bankers to handle all the banking business of their customers and the preference of the public for one-stop banking and the ready is that the supply curves of Least squares multiple regression techniques were generally used in these studies to describe how a dependent variable (to be explained) is related to a number of independent or explanatory variables. A demand equation is estimated to fit the actual observations of the dependent variable in such a way that the sum of the squared deviations between these observations and those predicted by the estimated equation is minimized. The independent variables have generally included an income or wealth constraint on deposit holders, the "ow n" rate of interest of an asset, the yields on substitute assets, and sometimes a variable to measure the transactions or convenience costs associated with the acquisition of assets. These studies have hypothesized that the income and "ow n" price or yield elasticities are positive (i.e., more of a particular asset will be demanded as its yield or the income of the asset holder increase) and that an increase in the yield of an alternative asset will depress the demand for the asset whose relative yield declined (i.e., the assets are substitutes). 27 ECONOMIC REVIEW analyst, who has extensively reviewed empirical paid on these deposits by banks to that paid on studies estimating the substitutability relationships near-bank deposits, whereas near-bank deposit among financial assets, has accepted the validity of growth was negatively related to this ratio by an this im plicit assumption "because the regulation of even higher degree.22 Their estimated demand the savings deposit market by Federal Agencies equations for annual changes in total deposits, and state commissions, as well as the stickiness of demand deposits and time deposits were based on the rates set by deposit-type intermediaries, make pooled data fo r all states for 1951-1961. They it reasonable to assume that individual savings found that almost 30 percent of the variance in institutions determine a yield and accept deposits percent change in time deposits and 27 percent of at that set rate.” 20 the variance in near-bank deposits were explained The first study to employ m ultiple regression by the independent variables. Jerry techniques utilizing observations across geographic L. Jordan areas and over time in exploring the product 21 market was by Edgar Feige. He concluded on substitutability also between found time a strong deposits at commercial banks and savings and loan shares, the basis of the pooled data (observations from all based on states fo r the 1949-1959 period) that demand between the demand for time deposits and the deposits ratio of the yield on savings and loan shares to the were either weak substitutes or a significantly negative relationship independent of savings and loan associations shares yield on time deposits. and mutual savings bank deposits in demand. He based on state data from 1956 through 1966, also also concluded that time deposits at commercial indicated that both mutual savings bank deposits banks and savings and loan shares His regression results, were and savings and loan shares are close substitutes independent of each other and that time deposits for time deposits on the basis of convenience cost, and mutual savings bank deposits were either weak and, to a smaller extent, the relative yield. More substitutes or independent. Finally, Feige did not than half of the variation among the states in per find any strong substitutability or complementary capita holdings of these three types of assets was relationship between savings and loan association consistently shares and mutual savings bank deposits. variables in his demand equations. A different conclusion emerged from the work explained by the independent A variation of Milton Friedman's permanent of Cohen and Kaufman, who found that time income form ulation of the demand for money was deposits at commercial banks and "near-bank" developed by Tong Hun Lee, who attempted to deposits were close substitutes for each other determine because time deposit growth was related positively significant interest rate variable by regression analysis the most affecting the and in a significant degree to the ratio of interest 22 20 George K. Kardouche, The C om petition fo r Savings, Studies in Business Economics No. 107 (New York: National Industrial Conference Board, 1969). 21 Edgar L. Feige, The Demand fo r L iq u id Assets: A Temporal Cross-Section Analysis (Englewood Cliffs, N. J.: Prentice-Hall, Inc., 1964). 28 Bruce C. Cohen and George G. Kaufman, "Factors Determining Bank Deposit Growth by State: An Empirical Analysis," Journal o f Finance, March 1965. OO Jerry L. Jordan, "The Market for Deposit-Type Financial Assets," Project fo r Basic M onetary Studies, W orking Paper No. 8, Federal Reserve Bank of St. Louis, March 1969. JUN E-JULY 1972 demand for money.24 Lee began by fittin g data assets by using a u tility function approach to for generate a number o f indifference curves between 1951-1965 to his demand for money equations, using one interest rate differential on an money and other assets.26 alternative asset (e.g., the difference between the The validity of the single market concept in yield on savings and loan shares and the yield on commercial banking was directly tested by Alan money) at a time as an independent variable in McCall.27 He hypothesized that a bank's loan and each equation. regression demand deposit service prices would be higher and equations using Lee also computed combinations of the the rates on time and savings deposits lower than difference between the yields of savings and loan varying those of another bank in a more competitive shares and money and the difference between the market, other things being equal. McCall employed yields of another alternative asset and money as conventional m ultiple regression techniques to influence independent variables, finally including all the isolate yield differentials in one equation. In all of these structure on pricing performance from that of regressions, the negative coefficients of the yield noncompetitive factors such as bank deposit size, on highly local economic activity, and type of market (e.g., savings and loan shares remained the of competitive market significant, while those of the yield on time city, county). All the banks in his sample group deposits were progressively less significant as more had less than $50 m illion in deposits and were variables were introduced. located in the Ohio-Kentucky region of the Fourth Lee concluded on the basis of his tests that Federal Reserve District. The regression equations savings and loan shares were the closest substitutes indicated that only the service charge rate on for money among time deposits, short-term paper, demand deposits was related to competitive long-term bonds, and equities. Lee also noted that structure, thereby implying that banks compete these empirical results were consistent with the with varying degrees of intensity in different substitution hypothesis of Gurley and Shaw and product markets. McCall concluded that the Baumol-Tobin hypothesis, which emphasized Such inconsistent pricing results clearly support rejection of the single line of commerce hypothesis for small- and medium-sized banks...Specifically, small- and medium-sized banks have m ultiple lines of 9Q commerce and geographic markets... the influence of opportunity cost (i.e., interest foregone by not purchasing alternative assets) in the holding of demand deposits. Using a different approach, V. Karuppan Chetty concluded that time deposits were the closest substitutes for money, followed by mutual savings bank deposits and, finally, savings and loan shares.25 substitution He estimated between the elasticity of money and other liquid 24 Tong Hun Lee, “ Alternative Interest Rates and The Demand for Money: The Empirical Evidence," American Econom ic Review, December 1967. 25V. Karuppan Chetty, "On Measuring the Nearness of Near-Moneys," Am erican Economic Review, June 1969. 26 A utility function approach relates the level of enjoyment or appreciation of a good to the rate of consumption. The indifference curves represent the various combinations of two goods that will yield the same total satisfaction. 27 Alan S. McCall, A Statistical Investigation o f Commercial Banking as a Single Line o f Commerce, Working Paper No. 69-10, (Washington, D. C.: Federal Deposit Insurance Corporation, 1970). 28Ibid., pp. 9-11. 29 ECONOMIC REVIEW TABLE II business Sources of Short-term Funds for Nonfinancial Corporations Selected Years (Percent Distribution) inter-institutional competition (see Tables II and Source markets, is an indication of III). Proponents of the narrow product market approach have sometimes argued that commercial banks have a competitive advantage over other 1963 Open market paper Short-term bank loans Finance company loans Other sources loan 1971 p 1967 n.a. 29.9% 7.5 62.7 12.5% 30.0 n.a. 57.5 1 0.3% 20.7 17.2 51.7 institutions offering credit in these markets, since finance companies must rely on banks for a substantial portion o f their funds. However, data collected by Raymond Goldsmith indicate that p Projected. business Source: Bankers Trust Company, New York companies, and sales finance companies acquired finance companies, personal finance only 24 percent, 20 percent and 12 percent, TABLE III respectively, of their funds from commercial banks Percent Distribution of Selected Financial Assets by Type of Institution Selected Years in Institution 1963 1967 1971 p Competition between commercial the consumer loan market in which finance companies devote about two-thirds of their total 45.5% 28.6 9.1 16.9 39.1% 13.0 15.2 32.6 48.9‘ 10.9 16.3 23.9 13.7 43.4 13.6 15.0 3.4 11.0 14.9 43.8 14.2 12.6 4.5 10.0 funds. With regard to competition in mortgage loan markets, one study concluded that the rates charged by savings and loan associations are very Mortgage Loans on Single-Family Dwelling Units Commercial banks Savings and loan associations Mutual savings banks Life insurance companies U. S. agencies Individuals and others oq banks and finance companies is concentrated in Consumer Credit Commercial banks Nonbank financial corporations* Credit unions Others 1965. 15.1 44.8 13.4 9.5 7.6 9.6 p Projected or preliminary. * Includes finance companies, factors, mortgage com panies, and real estate investment trusts. Source: Bankers Trust Company and Federal Home Loan Bank. Board responsive to competitive pressures from on commercial banks. Another investigator found that increased competition between commercial banks and savings and loan associations in the Chicago area over the 1960 to 1965 period, as a result of a liberalization of regulations on commercial banks, almost completely eliminated rate differences at these institutions.31 29 Raymond W. Goldsmith, Financial In stitu tion s (New York: Random House, 1968). 30 Relatively little statistical work has been done on the degree of loan substitutability between banks and other types o f financial institutions. However, evidence of significant shifts in the share of credit extended by commercial banks in some of these markets, especially the consumer loan and 30 Phoebus J. Dhrymes and Paul J. Taubman, "The Savings and Loan Business: An Empirical Survey," Savings and Residential Financing: 1969 Conference Proceedings, Donald O. Jacobs and Richard T. Pratt, co-editors (Chicago: United States Savings and Loan League, 1969). 31 Allen F. Jung, "Terms on Conventional Mortgage Loans—1965 vs. 1960," N ational Banking Review, Vol. 3, No. 3, March 1966. JU N E-JU LY 1972 CONCLUSION deposit customer of a commercial bank to satisfy The Supreme Court has consistently accepted his other major financial needs at other banks or the narrow product market concept of commercial nonbank banking, although the m ultiple product concept of depend to a large extent upon the importance of the banking firm has received more support from the convenience of one-stop banking to him. He is economists. The latter concept has been indirectly likely to find a similar degree of accommodation financial institutions would probably verified by statistical studies that have revealed a of particular needs at both types of institutions. close substitutability between deposits provided The use of customer surveys would provide more by commercial banks and those of other more direct evidence than statistical studies as to how specialized Few the public evaluates comparable services provided investigators, however, have attempted to estimate by banks and other financial institutions and the directly the strength of inter-relationships among importance of one-stop banking. depository institutions. the services provided by commercial banks. The The improved understanding of the product use of tie-in arrangements, whereby the seller market in commercial banking resulting from such attempts to condition the availability of some of surveys would permit more uniform ity in the the goods or services he provides upon the analyses acceptance acquisitions that are conducted by the regulatory by the customer of others, has of competitive effects of bank generally been outlawed in banking. Nevertheless, agencies, the banks can legally grant more favorable terms on courts. Ideally, a weighting scheme that would loans to customers who agree to accept, or have reflect the ability of nonbank financial institutions accepted, other services. Hodgman's theory of the to compete w ith banks would be devised for each customer major product market in which banks operate. relationship appears plausible for Department o f Justice and the business or corporate customers, who are likely to Factors such as legal constraints (e.g., maximum maintain large demand deposit balances at banks interest rates payable on savings deposits) as well where they expect to as the borrow. However, the lending policies of nonbank financial compensating balance requirement, which is at the institutions core of the customer risks) would be considered in the analysis of relationship, does not generally apply to the competitive strength. This approach would be an average loan customer who may be interested in a improvement over the sole criterion of size that is consumer or mortgage loan. currently of Hodgman's theory The willingness of the typical personal demand (e.g., maximum acceptable credit used in the analysis of competitive effects of structural changes in banking markets. Additional copies of the ECONOMIC REVIEW may be obtained from the Research Department, Federal Reserve Bank of Cleveland, P. O. Box 6387, Cleveland, Ohio 44101. Permission is granted to reproduce any material in this publication providing credit is given. 31