The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
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. MARCH 1972 BANKING STRUCTURE AND PERFORMANCE: SOME EVIDENCE FROM OHIO R obert F. Ware Numerous studies have examined the relationship between the structure o f banking markets and the pe rfo r mance o f banks in those markets. The assumption generally made in these studies is th a t banks operating in com pet itive ly structured markets w ill produce greater o u tp u t at lower prices. Nearly all o f the investigations employed basically fo u r sets o f variables designed to measure (1) bank performance, (2) banking m arket structure, (3) bank size and efficiency, and (4) economic a c tiv ity in a banking IN THIS ISSUE market. Bank performance was usually measured as an aggregate variable fo r all banks in a specified market. Such measures as the average loan rate and the average service Banking Structure and charge on demand deposits were com m only used. The Performance: Some com petitive structure o f a banking m arket was generally Evidence From O hio . . . 3 approxim ated in these studies by a concentration ratio, w hich measures the percentage o f deposits held by the largest banks in a market. In most cases, bank size and The Structure o f State Revenue .............. 15 efficiency were represented by deposits, costs, and loan p o rtfo lio s; and economic a c tiv ity o f a market, by popu lation and income. 3 ECONOMIC REVIEW The results o f the studies have been generally mixed because of differences in techniques, One study th a t found an insignificant relation ship between bank structure and performance in markets, and variable specifications. Five studies, m etropolitan areas concluded th a t average bank found a relationship between m arket structure and size was the most im p o rta n t banking structure performance, although in some cases the relation determ inant o f local loan rates.4 Moreover, results ship appeared to be relatively small. Kaufman, fo r indicated th a t the number o f banks in a m etro example, "fo u n d the market structure variable politan area was an insignificant determ inant o f consistently significantly various the loan rates. In a recent study o f bank structure measures of bank directions in Texas, the authors also found " th a t variations in predicted by related performance economic to in o th e o ry.' While the the level o f concentration appear to have little relationship was statistically significant, however, impact on the effect o f structure on performance was not perform ance."5 They concluded th a t the results six im p o rta n t measures o f banking strong. Relatively large changes in structure were lend support to the position th a t small shifts in the associated w ith relatively small changes in per structure o f banking markets (such as through the formance. merger o f tw o com peting institu tio n s) do n o t have Phillips also found a statistically significant and an appreciable im pact on performance. Differences in m arket areas and banking laws, positive relationship between interest rates and concentration appeared ratios, although the to be relationship "econom ically sm all.” He however, make it d iffic u lt to apply the structureperformance results from one state to another. concludes th a t "th e w eight o f the evidence is This th a t—w ith structure-perform ance region, the effects o f loan size, bank size, and tim e removed—concentration is study, obtain therefore, examined relationship the in bank Ohio. To a homogeneous sample, the study was positively associated w ith interest rates on business lim ited to effects o f m arket structure on bank loans charged by the banks in these 19 metroo performance in counties th a t are n o t included in politan areas." Standard SMSA). 1 Franklin R. Edwards, “ Concentration in Banking and Its Effects on E cono m ics Business and Loan S ta tistics, Rates," August The 1964; R eview of Franklin R. Edwards, "The Banking Com petition C ontroversy," The N a tio n a l B a n k in g R eview , September 1 9 65; George Kaufm an, "B ank Market Structure and Performance: The tions fo r M etropolitan Statistical This type constant a more Areas (non- o f sampling holds condi across sensitive markets test of and the provides structure- performance relationship than some o f the other studies. This means, however, th a t the study results cannot be generalized to other types o f Evidence from Io w a," The S outh ern E c o n o m ic J o u rn a l, A pril 19 66; Alm arin Phillips, "Evidence on Concentration in Banking Markets and Interest Rates,” F ederal Reserve B u lle tin , June 19 67; Charles T. Taylo r, "Average Interest Charges, The Paul A. Meyer, "Price Discrim ination, Regional Loan Rates, and the Structure of the Banking In d u s try," The J o u rn a l o f Finance, March 19 6 7 , p. 48. Loan M ix, and Measures of C om petition: Sixth Federal Reserve District Experience," The J o u rn a l o f Finance, December 1968. 2 4 5 Donald R. Fraser and Peter S. Rose, "M ore on Banking Structure and Performance: The Evidence from Texas," Jo u rn a l o f F in a n c ia l a n d Q u a n tita tiv e A n a lysis, January Kaufm an, op. c it., p. 438. ^Phillips, op. c it., p. 925. 4 1971, p . 611. 5 Counties in which no city has a population over 50 ,000 . M ARCH 1972 th a t operating performance concepts fo r these firm s are changes in bank structure have little effect on also d iffic u lt to measure. When a firm produces overall bank performance. one product, such as autom obiles or steel, p e rfo r markets. Results of this study indicate mance can be measured in terms o f sim ilar units D E FIN IN G THE M ARKET AREA produced. Banks, however, produce such diverse A m ajor reason fo r restricting this study to nonSMSA defining counties relates to a relevant commercial banks bank are the m arket problem area. m u lti-p ro d u ct services as loans, tru s t services, and demand of deposit accounts. Several performance measures Since must therefore be used to take account o f this firm s, it becomes d iffic u lt, in some cases, to isolate a single variety o f products. In this study, five d iffe re n t performance geographic area th a t includes a large percentage of measures (V j) were used to determine the effect o f all the d iffe re n t products th a t are offered by a banking structure on the performance o f banks bank. For example, the relevant m arket area fo r operating in 57 co u n ty markets in Ohio. demand deposits may be confined to a much The firs t performance measure is the ratio of narrower area than the market fo r commercial total service charges on demand deposits to total loans. demand deposits (V ^ ). This ratio measures the In this study, each o f the 57 nonSMSA counties average price charged fo r a dollar o f demand in Ohio was considered a single geographic market deposits. The assumption was made th a t the more fo r banking services. There are tw o reasons w hy com petitive these counties should approxim ate relevant market market, the lower w ould be the average price fo r the environm ent in a particular areas. First, the banks in the nonSMSA counties the demand deposits. The second performance are, on average, smaller (less than $30 m illio n in variable (X ^) is the ratio o f yearend average net deposits) than banks in m etropolitan areas and operating earnings to average total capital fo r the generally derive fro m 80 to 90 percent o f all types banks in each market. This ratio is one measure of of deposit and loan business from th e ir respective the average p ro fita b ility o f the banks in a market, counties. Secondly, under Ohio banking laws, a and it was assumed th a t a lower average p ro fit rate bank cannot establish branch offices outside o f the w ould county in which the main office is located.7 This com petitive environm ent. aspect o f the law has a tendency to restrict the influence o f a nonurban bank to the county be found The th ird in markets w ith a highly performance measure (Vg) is the ratio o f the total revenue received on loans fo r a year to the average gross loans outstanding at the market. end o f the year. This ratio is intended to reflect PERFORMANCE OF BANKS the average loan price charged by the banks in a Just as the geographic market areas fo r m u lti market, and it was assumed th a t the more product firm s are often d iffic u lt to define, the com petitive the market environm ent, the lower 7 An is the rate w ould be. However, many problems are headquartered in a city where the limits overlap into two involved w ith an aggregate performance measure exception to this law is made if a bank or more counties. The bank may then branch into each of the counties. This situation only existed in tw o of the 57 counties in this study. such as this. First, aggregating all types o f loans fo r the banks in a particular market may conceal the 5 ECONOMIC REVIEW Bank Performance Variables fo r the d iffe re n t customers o f a bank because o f factors, such as compensating balances held by the to tal service charges on demand deposits V1 = bank. This makes the real price o f a loan quite total demand deposits d iffe re n t fro m the average loan price th a t was used net operating earnings V2 = V 3 in total capital most studies. While sample selection can p a rtia lly alleviate the firs t problem, the second = total revenue on loans -----------------------------------gross loans problem can only be remedied by a very intensive _ total interest paid on tim e and savings deposits field survey. In this study, the use o f banks in nonSMSA counties provided a sample o f a fa irly total tim e and savings deposits homogeneous group o f banks (i.e., banks less than Vj- = average price spread (V ^ —V^) $30 m illio n in deposits tend to have sim ilar loan p o rtfo lio s ), which m inim ized the loan aggregation Independent Variables problem. The average loan price should, therefore, total deposits o f tw o largest banks in county be generally representative o f the m a jo rity o f X1 = total deposits o f banks in county banks in the study. = percent change in county population, 1 9 6 0 -1 9 7 0 y _ percent change in county per capita personal income, 3 1 9 5 9 -1 9 6 9 The fo u rth performance variable (V ^) is the ratio o f to ta l interest paid on tim e and savings deposits to the to ta l am ount o f tim e and savings = the per capita retail sales in county v _ deposits held by the banks. This ratio reflects the m anufacturing covered em ploym ent average price the banks had to pay in order to total covered em ploym ent in the county attract tim e and savings deposits. Presumably, if a Xg = number o f savings and loan associations in county market is highly com petitive, the average price y paid fo r the tim e and savings deposits w ould be = total consumer loans --------------------------------7 gross loans higher. The fifth performance measure (Vg) was derived by taking the difference between variables Xg = average deposit size o f bank Vg (average price charged by the banks fo r loans to tal operating expenses X9 = they have made to various customers) and V ^ total assets (average price the banks had to pay to a ttra ct the raw materials in the fo rm o f tim e and savings fact th a t loans have d iffe re n t prices and some deposits banks specialize in a particular type o f loan. Thus, represents a measure o f the average price spread to make the loans). This difference the average price o f a loan may n o t be represen between the so-called o u tp u t price (average loan tative effective rate) and the average price paid fo r inputs (tim e interest rate charged fo r similar type loans varies and savings deposits). It was expected th a t the less of all banks.8 Secondly, the com petitive This problem may partially account fo r the fact th at the results of some of the past studies were very inconclusive. A discussion o f this problem appears in: Alm arin Phillips, "Evidence on Concentration in Banking Markets and Interest Rates," Federal Reserve B u lle tin , June 1967. 6 market, the the environm ent larger w ould be in a particular the difference between V g and V ^ (i.e., the higher the price charged on loans and the lower the price paid fo r the tim e and savings deposits). Even though V ^ is MARCH 1972 dependent upon Vg and V ^, it does perm it an 1960-1970 (X 2 ), (2) the percentage change in observation o f the structure-performance question county per capita personal income fro m 1959 to in a slightly d iffe re n t manner. 1969 (X g), (3) the per capita retail sales in the county (X ^) fo r 1969 and 1970, and (4) manufac DETERM INAN TS OF BANK PERFORMANCE The operating performance tu r in g em ploym ent covered by the State unem ploym ent insurance as a percent o f total of banks was covered em ploym ent in the co u n ty (X ^). Variables assumed to be a fu n ctio n o f banking structure as X 2 and Xg serve as proxies fo r shifts in demand well as other bank and market variables. In most fo r banking services in a market, and X ^ is a p ro xy structure-performance studies, the structure o f a fo r the level o f demand. Variable Xg was used to m arket was generally proxied by a concentration control fo r differences in the level o f industrial ratio, w hich indicates the percentage o f deposits a c tivity among the markets. It was assumed th a t held by the largest bank or banks in the m arket.9 the more highly industrialized markets w ould have In this study, the tw o-bank concentration ratio, or a higher level o f economic a ctivity. the percentage o f deposits held by the tw o largest A sixth m arket variable was also included to banks in the market (X ^), was used to proxy take account o f existing and potential com petition banking structure. The tw o-bank ratio was used provided by other financial in stitu tio n s in each because it provided an accurate picture o f banking market. The p ro xy used fo r this effect was the structure in the nonSMSA county markets th a t number o f savings and loan associations operating were used in the study. The assumption was made in tha t the higher the concentration ratio in any number single m arket, the less com petitive w ould be the operating in a particular market, there w ould be a environm ent in th a t market. significant M arket variables th a t could affect the p e rfo r mance o f banks can generally be put in to the each deposits m arket of (X g). savings am ount and Presumably, if and certain of loan com p e titio n types a large associations of fo r loans in are tim e the markets.10 classification o f "econom ic a c tiv ity ” or "dem and” Three bank variables were used to take account variables. It can be expected that the comparative o f bank operations th a t could alter bank p e rfo r performance o f banks in tw o separate markets mance. One o f these variables is the percent of w ould levels o f total loans held in the consumer loan category by economic a ctivity, or demand fo r banking services, the banks in each market (X y). Even though the be affected by the d iffe rin g even if the banking structure is the same. Four banks in the sample are fa irly homogeneous, this variables were used in this study as proxies fo r variable was intended to help co n tro l fo r d iffe r economic a ctiv ity in the individual markets: (1) ences in loan m ix th a t could have an effect upon the percentage change in county population from 10 g The same values fo r variables X 2 , X 3 , X 5 , and Xg were used in both the 1969 and 1 9 70 regression equations. The number o f banks in a m arket is sometimes used as a structural proxy. However, both Fraser and Rose and Kaufman found that from a statistical view point the concentration ratio and the number of banks in a m arket area were equally good proxies fo r the banking structure. This should not hinder the analysis since these data do not change significantly in one year. The number of savings and loan associations was used in X q because data were not available on the deposits o f savings and loan associations by nonSMSA counties. 7 ECONOMIC REVIEW the performance variables. Another bank variable, a c tivity average bank size (X g ), provided an additional types o f banks operating in the m arket (X y —Xg). control Therefore, fo r d iffe rin g bank behavior among (X 2 —Xg) in the co u n ty as well as the markets. Larger banks tend to behave d iffe re n tly w ith respect to loan m ix and some prices. The V jj = F (X ^ j,...,X g j) where i = performance measure and j = m arket ratio o f average to ta l operating expenses to average total assets fo r the banks in each m arket (Xg) presumably measures, on a m arket level, how e ffic ie n tly banks are managed.11 Some banks may There are 57 separate markets or observations in be operated more e ffic ie n tly than others, w hich the study. The fo llo w in g sections discuss the study w ould have an effect on market performance. In results order to isolate the effect of m arket structure empirical findings; the actual statistical results are upon performance, it is necessary, therefore, to presented in the A ppendix. con tro l fo r the effect th a t other variables, such as in detail. The Table summarizes the Service Charge or Demand Deposits. Results efficie ncy, may have on performance. from TECHNIQUES AND RESULTS average service charge on demand deposits. The this set o f equations fail to indicate th a t bank concentration has an im p o rta n t effect on the The structure-perform ance relationship among relationship between service charges and concen banks headquartered in the 57 nonSMSA counties tra tio n in the 57 nonSMSA counties is weak, w ith in Ohio was investigated using m u ltip le regression the degree o f association insignificant (at the 5 te c h n iq u e . percent level) in both 1969 and 1970. On the Cross-sectional regressions were computed fo r tw o d iffe re n t years, 1969 and 1970, other to observe the structure-perform ance relationships indicated by positive coefficients) may tend to under d iffe re n t m onetary conditions. increase the average service charge, b u t the size of Data fo r the banks in the 57 counties were taken fro m the hand, increases in concentration (as this increase w ould be relatively small. Bank cost and bank size are significantly related December 1969 and December 1970 "R eports o f C o n d itio n " and "R eports o f Earnings and D ivi to dends," as com piled by bank regulatory agencies. variable fo r bank costs is significant (at the 0.1 service charges on demand deposits. The It was assumed th a t the performance variable percent level) in both 1969 and 1970, and the (V ), w hich is the average o f the values o f this positive sign on the coefficients indicates th a t variable fo r all o f the banks in a market, was a banks w ith fu n ctio n o f the concentration (X-|) and economic asset ratios tend to charge more fo r th e ir demand relatively high to ta l operating cost/ deposits. This result may sim ply im p ly th a t the 11 Under certain circumstances the bank cost ratio could less e ffic ie n t banks must, and are able to , charge a be viewed as a bank performance variable. However, in this study the cost ratio is only assumed to be a proxy for the differences in bank efficiency th at may have an effect on bank performance. 13 The estimated change in the average service charge brought about by an increase in the concentration ratio can be com puted by taking the concentration coefficient 12 The year 19 69 was a relatively tight money period while 1 9 70 was a period o f relative m onetary ease. 8 and m ultiplying it by a representative change in the concentration ratio. Summary o f Results o f Structure-Performance Tests fo r NonSMSA Counties in Ohio Independent Variables Performance Variables 1969 Percent C oncentration Change in Ratio Population Percent Change in Income X3 Retail Sales Per Capita X4 Industrialization Ratio N um ber of Savings and Loan Associations Consumer Loans/ Average Bank Gross Loans Size Cost Ratio X5 X 6 X7 X 8 X9 V ^ —Average service charge on demand deposits insignificant insignificant insignificant insignificant insignificant insignificant insignificant significant positive significant positive V j —P ro fit rate insignificant insignificant insignificant insignificant insignificant insignificant significant positive insignificant significant negative V g —Average loan rate insignificant insignificant insignificant significant negative insignificant significant positive insignificant insignificant significant positive V ^ —Average savings rate insignificant insignificant significant positive insignificant insignificant significant positive insignificant significant positive V ,-—Price spread insignificant insignificant insignificant significant negative insignificant insignificant insignificant insignificant insignificant insignificant insignificant insignificant insignificant insignificant insignificant insignificant significant positive significant positive significant negative insignificant significant positive significant positive significant negative X 1 X 2 significant positive 1970 V ^ —Average service charge on demand deposits insignificant significant positive V g —Average loan rate insignificant insignificant significant positive insignificant insignificant significant positive insignificant significant positive significant positive V ^ —Average savings rate insignificant insignificant significant positive significant positive insignificant insignificant insignificant significant positive significant negative insignificant significant positive insignificant significant positive insignificant Pro fit rate V g —Price spread significant positive insignificant insignificant insignificant insignificant insignificant N O T E : Coefficients are significant at the 5 percent critical level. When significant, the sign is indicated. Source: Federal Reserve Bank o f Cleveland ECONOMIC REVIEW higher price fo r their demand deposits. 14 Bank The cost ratio is negatively related to the p ro fit size and the average service charges on demand rate, indicating th a t banks in markets w ith rela deposits positively tiv e ly higher average costs generally have lower related fo r both 1969 and 1970. This means th a t are also significantly and profits. The consumer loan/gross loan variable has those banks located in counties w ith a relatively a positive sign, indicating th a t the banks in the high average deposit size tended to charge more more pro fita b le markets had a larger percentage of fo r th e ir demand deposits than banks in counties consumer loans in th e ir p o rtfo lio . This result is w ith a low average deposit size. This result may consistent w ith im p ly th a t relatively large banks in the study are consumer loans tend to be more p ro fita b le fo r located in less com petitive m arket areas. There banks than some other types o f loans. fore, they were able to im plem ent higher service charges on demand deposits.1 5 other evidence th a t indicates There is no significant im pact on bank p ro fit a b ility fro m the economic a c tiv ity variables fo r The coefficients o f the savings and loan associ 1969. For 1970, however, the population variable ation variables and the fo u r variables fo r economic and the industrialization variable are significantly a ctivity to related to the p ro fit rate.16 The sign on the demand deposit service charges fo r either 1969 or show no significant relationship population variable 1970. banks Net Operating Earnings to Total Capital. The performance variable in this analysis measures in is positive, indicating th a t markets th a t had large increases in population also had high p ro fit rates. The indus tria liza tio n variable has a c o e fficie n t w ith a p ro fita b ility o f a bank related to its to ta l invested negative sign, w hich implies th a t banks operating capital. It was hypothesized that banks w ould be in the more industrialized markets tended to have less p rofitab le in a highly com petitive market. This lower p ro fit rates. This may reflect the fa ct th a t hypothesis was somewhat confirm ed by the test these banks must compete w ith large c ity banks results, b ut the findings are relatively insignificant. fo r the more pro fita b le commercial and industrial For 1969, the concentration variable is significant loans th a t are available in these markets. Bank p ro fita b ility in the 57 nonSMSA counties at the 20 percent level, but is insignificant fo r 1970. The positive sign on the coefficients, however, suggests th a t a high rate o f p ro fit is does n o t appear to have been affected by compe titio n fro m savings and loan in stitutions. Average positively related to a high concentration ratio. On the other hand, bank p ro fita b ility is s ig n ifi Loan Rate. The conclusion th a t emerges fro m the analysis o f variables th a t affect bank loan rates o f the banks under study is th a t the variables—costs and consumer loan/gross lo a n - relationship between bank concentration and the according to equations fo r both 1969 and 1970. loan rate is weak, w hile the relationship between cantly 14 affected by tw o of the three costs, income, and the number o f savings and loan The simple correlation coefficients between bank costs and concentration are —0 .1 8 and —0 .1 4 for 1969 and tionship between concentration and average loan 1970, respectively. 15 This point is substantiated associations is significant. S pecifically, the rela somewhat by positive 16 This result implies th at the economic activity variables correlation coefficients between bank size and concen become more sensitive w ith respect to p ro fit rates during tration o f 0 .2 2 and 0 .1 9 for 1969 and 19 70, respectively. periods of m onetary ease. 10 M ARCH 1972 rates is weak fo r both 1969 and 1970, and the markets th a t contain a relatively large number of impact savings and loan associations had a higher average of an increase in bank concentration appears relatively small; i.e., an increase in the loan rate. This may im p ly that, because savings concentration ratio o f 20 percentage points w ould and loan in stitu tio n s specialize in real estate loans, have only increased the average loan rate in 1970 banks in markets w ith several such in stitu tio n s by approxim ately 0.1 percent.17 tend to concentrate on selling other types o f loans, The bank cost variable is highly significant fo r possibly because the com p e titio n is less intense in both years, and the positive sign on the coe fficie n t these other areas.18 Since the rates on real estate indicates th a t banks w ith relatively high costs also loans charged high average rates on th e ir loans. This consumer and some other types o f loans, the result is sim ilar to the one on average service tendency o f holding fewer real estate loans in a charges on demand deposits and suggests th a t the p o rtfo lio w ould therefore result in a higher average less e fficie n t banks in these markets are able to loan are generally rate fo r lower than the commercial the rates on banks in those markets. Thus, it appears th a t in markets where charge higher rates fo r th e ir services. Per capita retail sales and per capita income there are a relatively large number o f saving and were found to be significantly related to higher loan associations, banks had a higher average loan loan rates, although fo r d iffe re n t years. For 1969, rate. Average Time and Savings Rate. Bank concen the retail sales per capita variable implies th a t counties w ith high retail sales per capita had tra tio n in nonSMSA county markets in Ohio does relatively lower average loan rates. The level of not appear to have had an im p o rta n t impact on retail sales per capita was assumed to p ro xy the the average savings rate paid by banks in those intensity o f economic activity in a m arket; and, markets. In both the 1969 and 1970 equations, therefore, the lower average loan rate w ould be the relationship between concentration and the consistent w ith this assumption. For 1970, the per rate paid on deposits is insignificant. capita income variable indicates increasing demand fo r bank services; and, therefore, it w ould be dire ctly related to the average loan rates. Finally, the operation o f savings and loan associations has a significant and positive re lation Bank costs, however, are highly significant in both equations, and the sign on the coefficients indicates a direct relationship between the average 1Q This relationship savings rate and the cost ratio. implies th a t those banks operating in high average ship w ith bank loan rates. Results o f equations fo r 1969 and 1970 indicate th a t banks operating in 18This result is supported by the correlation coefficients between the number of savings and loan associations in a 17 The average loan rate may have been more precisely measured fo r 1 9 70 than for 1969, which was a tight money period w ith increasing loan rates. Since there are usury ceilings in Ohio, the 1969 average loan rate may not represent the effective loan rate. In 19 70, however, m onetary policy was less restrictive, and loan rates tended to be lower. Therefore, the 1 9 70 average loan rate and the m arket and the percentage of various types of loans held by banks in those markets. There was a negative corre lation between the number of savings and loan associations and the percentage of consumer and comm er cial loans held by banks in these markets. 19 Part of the significance of this relationship is because of effective loan rate would probably be approxim ately at the fact that the total operating costs include the interest the same level. paid on tim e and savings deposits. 11 ECONOMIC REVIEW cost markets paid a relatively high average savings markets. However, the relationship is relatively rate. weak, The level of retail as evidenced by the fa ct th a t a 20 sales per capita is also percentage p o in t increase in the tw o-bank concen significant in both the 1969 and 1970 equations. tra tio n ratio w ould have only increased the average This result implies th a t a high level o f per capita price spread by approxim ately retail sales was associated w ith a high average tim e 1970. .002 percent in and savings deposit rate. Since retail sales per The wide difference in the relationship between capita was viewed as a proxy fo r the level of the price spread and concentration fo r 1969 and economic 1970 a c tiv ity in a market, this result is is most like ly a result o f the type of consistent w ith a high average tim e and savings monetary policy th a t was being pursued in both rate. years. In 1969, policy was restrictive and loan F inally, there is little relationship between the rates and deposit rates were relatively high. number o f savings and loan associations and the However, usury laws in Ohio kept some loans rates rates paid on tim e and savings deposits, according from increasing to th e ir natural level, and Regu to equations fo r 1969 and 1970. This result is a lation Q prohibited the deposit rates fo r increasing little surprising because it could be expected th a t beyond th e ir stated ceilings, causing the spread savings and associations compete d ire ctly between the rates to be distorted. In 1970, as the banks fo r tim e and savings deposits and supply o f credit expanded, both loan rates and w ith loan w ould, therefore, drive up the average savings rate. savings rates fell somewhat fro m th e ir ceiling However, the relationship could be distorted by levels. The demand and supply forces operating in the rate ceilings imposed on both types o f in s titu the tions, which w ould lim it their com p e titio n w ith produce a price spread th a t was essentially undis each othe r.20 This fa cto r could account fo r the torte d . As a result, it can be expected th a t the absence o f a relationship between the number o f relationship savings and average price spread is measured more accurately loan associations and the banks' m arket, therefore, were relatively between concentration free to and the fo r 1970 than fo r 1969. average savings rate. Average Price Spread. The relationship between the average price spread and concentration is The number o f savings and loan associations in the banking markets under study was found to 1970 equation, but is exert some influence on price spread. The positive insignificant in the 1969 equation. Both equations sign on the savings and loan variable indicates th a t show highly significant in the a a relatively large number o f these in stitu tio n s in a relatively large price spread is associated w ith a m arket is associated w ith a large price spread fo r high level o f concentration in the 57 banking the banks in those markets. This is consistent w ith a positive association, im plying th a t the earlier findings on the average loan rate and 20 Savings and loan associations have higher ceiling rates than commercial banks. The Federal average savings rate equations. It could be h y p o th Reserve System, however, lifted some of the rate ceilings in 1 9 70 fo r large esized th a t savings and loan associations do n o t denom ination tim e deposits. This change would probably o ffe r commercial banks as much direct com pe not have a great effect on the present findings since many of the banks in this sample do not offer th at type o f tim e deposit. 12 titio n fo r financial services as m ight be expected. In fact, the presence o f a relatively large number MARCH 1972 o f savings and loan associations in a m arket may performance. First, banking is a regulated provide banks w ith an incentive to concentrate on industry. Many o f the regulations tend to dim inish providing financial services th a t are not offered by the savings and loan associations. structure and bank performance since the market The relationship between price spread and significance of the relationship between is n o t entirely free to determine prices and o u tp u t other variables examined—such as per capita retail and sales—is generally insignificant. The retail sales per in e fficie n t perform er.21 to reward the Secondly, the capita variable is significant in the 1969 equation, e ffic ie n t results from or punish the this study may but insignificant fo r 1970. The negative re lation im ply th a t the approach used to measure bank ship between retail sales per capita and the price performance was n o t su fficie n tly disaggregated to isolate the structure-perform ance relationship as it spread is consistent w ith earlier results. exists SUMMARY AND CONCLUSIONS This study generally concludes fo r banks in these nonSMSA county markets. While the structure o f a m arket affects th a t the some aspects o f bank performance, it may be very structure o f markets (as represented by a two-bank d iffic u lt to detect the extent o f the relationship concentration ratio) is n o t strongly related to the w ith aggregate performance variables. A d ditional aggregate performance o f banks in the nonSMSA research markets in Ohio. The study differs somewhat from performance (mortgage loan rates or business loan other such studies in th a t it was lim ited specif rates instead o f average loan rates) is necessary to ically to nonSMSA co u n ty markets in order to measure precisely how great an impact market obtain structure has on bank performance. a more homogenous sample fo r the using disaggregated variables o f bank oo This type o f research must be com pleted fo r individual states empirical tests. The only specific variable th a t appeared to have and fo r d iffe re n t markets w ith in states before any had a consistent impact on bank performance is generalized statements can be made w ith regard to the bank cost ratio. This variable is significant in th e three o f the fo u r equations fo r both 1969 and performance relationship. 1970, im plying th a t bank efficiency may be an im po rtant determ inant o f the performance o f banks in the nonSMSA county markets. There may be at least tw o reasons w hy most studies o f this type have n o t consistently shown a strong relationship between market structure and bank b a n k in g industry and the structure- 21 For a discussion o f this point see: Alm arin Phillips, "C om petition , Confusion, and Commercial Banking," The Jo u rn a l o f Finance, March 1964. 22 For example see: Donald Jacobs, Business L o a n Costs an d B ank M a rk e t S tru c tu re (New Y o rk: National Bureau of Economic Research, 19 71). 13 APPENDIX TABLE Statistical Results of Structure-Performance Relationship Tests for NonSMSA Counties in Ohio Performance Variables Intercept 1969 Concentration Ratio Percent Change in Population Percent Change in Income Retail Sales Per Capita Industrialization Ratio Number of Savings and Loan Associations Consumer Loans/ Gross Loans Average Bank Size Cost Ratio R2 X1 x2 X3 X4 X5 X6 X7 X8 X9 V j —Average service charge -.29112 X 10~2 on demand deposits .94213 X 10- 3 (.78239) .26268 X 10- 4 (.81942) -.56828 X 10- 6 (-.70743) .73054 X 10- 3 (-1.14822) .29635 X 10~5 (.19977) .16071 X 10- 3 (1.18134) .28650 X 10- 3 (.13477) .66004 X 10- 7 * (2.11502) .17510t (3.61064) 51.4 V 2—Profit rate .13929 .36004 X 10- 1 (1.49472) .34542 X 10~3 (.53865) -.33158 X 1 0 - 5 (-.20634) .15612 X 10 ~ 1 -.6 054 5 X 10- 4 (1.22672) (-.20590) .18304 X 10- 2 (.67260) —1.72600* (-1.77914) 26.1 V^—Average loan rate .44813 X 10—1 .66914 X 10~3 (.22360) .11717 X 10~3 (1.47080) .46376 X 10~6 (.23230) -.28203 X 10- 2 * (-1.78375) V^—Average savings rate .16587 X 10-1 .22414 X 10- 3 .12226 X 10- 3 * (2.00654) (.97930 X 10 ~ 1) .14487 X 10~5 (.94884) .20935 X 1 0 _2# -.3 564 5 X 10- 4 -.7 163 4 X 10- 4 -.8 8 7 1 5 X 10 2 * -.1 979 6 X 10 7 (-.33373) (-.27703) (2.19566) (-1.27582) (1.73115) Vg—Price spread .28226 X 1CT1 .44500 X 10- 3 (.14438) V^—Average service charge -.2 466 6 X 10- 2 on demand deposits .91616 X 10- 3 (.70129) .39042 X 10~4 (1.15806) V j —Profit rate .27117 .26421 X 10~1 (1.16675) .11794 X 10- 2 * (2.01821) .15707 X 10~5 -.3 102 5 X 10- 2 --.55160 X 10- 3 * .20936 X 10~2 (-2.00705) (.79767) (-.35520) (.10351) Vg—Average loan rate .42282 X 10-1 .56482 X 10- 2 (1.48064) .11218 X 10- 3 (1.13958) .45742 X 1 0 " 5* -.19429 X 10- 3 -.49444 X 10—4 .100088 X 10_2# (-1.06799) (2.28170) (-.13205) (1.78961) V4 ~Average savings rate .24809 X 10“ 1 -.2 947 2 X 10~2 (-1.24553) .66502 X 10~4 (1.08903) .35315 X 10~5* .17197 X 10- 2 * -.23025 X 10~4 -.20095 X 10- 3 -.40452 X 10- 2 (-.87745) (-.73270) (1.88431) (-.80178) (2.22741) Vg—Price spread .17472 X 10 ~ 1 .85955 X 10~2 * (2.41178) .45683 X 10- 4 (.49670) -.50878 X 10~5 .98500 X 10- 6 (-.47906) (-.62007 X 10~1) -.4 913 8 X 10_ 2 t (-3.01747) .14805 X 10- 4 (.40529) .50450 X 10- 4 (1.34095) .93011 X 10_ 1 * -.4 794 5 X 10- 6 (2.18729) (-.76802) .58994 X 10_3# -.1 3 8 4 9 X 1 0 3 -.3 2 5 9 2 X 10 8 (1.74493) I[-.26216 X 10—1) (-.42024 X 10- 1 ) .66157 X 10- 3 (1.34095) .87330 X 10- 2 (1.60506) .16536 X 10~7 (.20702) .48180t (3.99762) .50038t (5.42831) 43.2 60.6 -.18575 X 10“ 1 32.8 (-.14964) 1970 NOTE: Figures in parentheses are t-values. * Significant at the 5 percent level, t Significant at the 1 percent level. Source: Federal Reserve Bank of Cleveland -.27094 X 10- 6 -.6 916 5 X 10- 3 (-1.37266) (-.30953) .10427 X 10- 5 (.43664) .53850 X 10- 6 .23438 X 10- 3 (.33964 X 10~1) (1.54791) -.19140 X 10“ 2 -.2 641 8 X 10- 4 (-1.39242) (-.61078) .12097 X 10- 2 t (2.92872) -.54663 X 10- 3 (-.21476) .61106 X I Q ' 7 * (2.05432) .15238t (3.33138) 48.6 .11881 + (2.69298) .28658 X 10- 6 (.55581) —3.64484t (-4.59683) 44.7 .43478t (3.25515) 41.8 .34851 X 10” 7 (.64686) .33504t (4.04385) 49.2 -.29231 X 10- 7 (-.36021) .99741 X 10-1 (.79929) 35.8 .56205 X 10- 8 .56415 X 10- 2 (.75906) (.64710 X 10 1) .96868 X 10- 2 (1.39505) MARCH 1972 THE STRUCTURE OF STATE REVENUE Warren E. Farb INTRO DUCTION level. (Services provided at the state government State governments have been faced w ith both increased operational costs and co n tin u a lly level in some states may be provided by counties or m unicipalities in other states.)1 growing demands fo r public services. As a result, the states have found it necessary to increase tax TA X STRUCTURE institute new taxes. They have also The am ount o f revenue th a t a taxing a u th o rity turned to the Federal Government fo r financial is able to raise is necessarily lim ited by the size o f aid. In the past. Federal aid has been in the form the relevant tax bases. The principal bases are o f grants fo r specific programs; but in the future, income, sales, and wealth (property tax), although some funds may be distributed through a form o f other measurable concepts could be used. A range revenue sharing fo r use largely at the discretion o f in possible rates as well as numerous com binations the fo r o f taxes leads to greatly d iffe rin g tax structures revenue sharing th a t are currently being considered and, consequently, variations in tax e ffo rt among rates and recipient government. The proposals are based on factors such as population, per capita the states. The d iffe re n t tax structures make it income, and tax e ffo rt o f the individual govern v irtu a lly impossible to develop a clear-cut measure m ent u nit. o f e ffo rt. For example, one state may be making a This article discusses the variation in state tax strong e ffo rt in terms o f the wealth base, b u t its structure and tax e ffo rt and d iffe re n t aspects o f e ffo rt may appear weak when compared to the the income base. principal taxes now used by the states, particularly those states in the Fourth D istrict. The measure o f tax e ffo rt discussed in this The possible impacts o f a revenue sharing program article is revenue per $1,000 o f personal income, and state funding o f local schools on the state which tends to remove the effects o f differences in revenue structures are also examined. The dis income levels among states stemming from either cussion and data relate o nly to taxation at the state level (thus excluding taxes imposed by cities, counties, and school districts) and do not account fo r differences in services provided at the state 1 For a more complete study, see: "State and Local Revenues and E xpenditures," E c o n o m ic R eview, Federal Reserve Bank of Cleveland, November 1970. 15 ECONOMIC REVIEW T A B LE I Rank o f States' Revenue Per $1,000 o f Personal Incom e* Selected Revenue Sources 1970 Per Capita Personal Income Connecticut Alaska 3 Nevada 4 New Y o rk 5 California $ 4 ,5 9 5 4 ,4 6 0 4 ,4 5 8 4 ,4 4 2 4 ,2 9 0 1 2 46 South Carolina 47 West Virginia 48 Alabama 49 Arkansas 50 Mississippi General Sales or Gross Receipts Total T ax Revenue 2,6 07 2 ,6 0 3 2 ,5 8 2 2 ,4 8 8 2 ,2 1 8 Hawaii New Mexico V erm o nt Mississippi Delaware 8 8 .2 1 Hawaii Mississippi Washington West Virginia Arizona 5 1 .0 3 4 9 .9 6 4 3 .9 5 42.41 3 8 .07 New Y o rk Oklahoma V erm o nt New Jersey Massachusetts $ 1 1 1 .2 6 9 4 .9 9 9 4 .8 0 92.81 Missouri Nebraska New Jersey Ohio New Hampshire Individual Income Tax $ 5 3 .1 7 4 3 .5 5 4 1 .7 2 38 .38 3 0 .4 3 12.44 11.99 11.97 11 .7 3 7.41 Hawaii Wisconsin Delaware New Y o rk Verm ont $ 3 4 .3 2 3 1 .8 6 3 0 .8 8 3 0 .8 0 3 0 .62 Louisiana New Hampshire Tennessee New Jersey Connecticut 4.61 1.39 1.08 0 .5 8 0 .3 6 * Personal income data are U. S. D epartm ent o f Commerce estimates for calendar year 1969. Source: U. S. Departm ent o f Commerce d iffe re n t populations or levels of per capita income.2 A lthough this measure ignores "w e a lth " and of a ctiv ity , most relatively high per capita incomes do not necessarily have the lowest tax efforts. O f the five recognize income as the major source o f tax revenue. states w ith the highest revenue per $1,000 o f Tax e ffo rt—as measured by revenue per $1,000 personal income, all b u t New Mexico are among personal economic w ith states of levels weak negative correlation, suggesting th a t states incom e—varies greatly among the the five to p states in either income or sales tax states because o f variations in income o r d iffe r efforts. S im ilarly, those states having the lowest ences in tax structure (see Table I). The average overall tax e ffo rt either do n o t use one o f the tw o state personal m ajor types o f state taxation or use them to o n ly a income in 1970, and ranged fro m $38 in New lim ited extent. Ohio, w hich did not have a state Hampshire to $111 in Hawaii. O f the five states income tax in 1970, and New Hampshire, which that ranked highest in per capita personal income does n o t have a sales tax, are examples o f the in 1970, none was among the to p five in tax e ffo rt; on the other hand, o f the lowest ranked form er, and New Jersey is an example o f the o latter. The sources and relative d is trib u tio n o f tax states in per capita income, one state (Mississippi) revenue fo r the five states th a t rank as the highest tax was $67 per $1,000 of ranked among the to p five in revenues collected. in overall tax e ffo rt and the five states th a t rank In general, tax e ffo rt and per capita income show a o the lowest are shown in Table II. Tables I and II widely per $1,000 o f income between the five to p ranked Tax e ffo rt measured by tax per $ 1 ,0 0 0 o f income is a used defin ition , but it does have many short comings and is by no means the only measure o f tax e ffo rt found in economic literature. For a more detailed discussion o f measures o f tax effo rt and tax capacity see: Allen D. Manvel, “ Differences in Fiscal Capacity and E ffo rt: Their Significance for a Federal Revenue Sharing System ," N a tio n a l Tax J o u rn a l, V ol. X X IV , No. 2 (1 9 7 1 ). 16 both illustrate the large disparity in revenues raised states and the five lowest ranked states. On average, revenues raised by the five leading states 3 Pennsylvania instituted a state income tax in 19 71; and Ohio began levying an income tax in 1972. MARCH 1972 TA B LE II Percent D istrib u tio n o f State Tax Revenue 1970 National R ank* 1 2 3 4 5 General Sales and Gross Receipts Tax Individual I ncome Tax Hawaii New Mexico V erm ont Mississippi Delaware 30.9% 13.1 32 .3 9.1 35 .0 M EAN 4 6 Missouri 47 Nebraska 48 New Jersey 49 Ohio 50 New Hampshire Corporation Income Tax 47.8% 3 1 .3 All O th e rt 46 .9 - 0- 4.3% 3 .0 4 .3 4.1 6.9 17.1% 52.7 50.8 39.9 58.1 1 9 .2$ 2 9 .5 $ 9 .1 $ 4 2 .2 $ 15.8 17.0 1.3 - 03.7 4 2 .0 28.7 26.7 38.7 - 0- 2 .6 3 9 .6 51.1 59.3 6 1 .3 9 6 .4 1 2 .6 3 .3 12.7 - 0- 0- * Rank of state based on total 1970 tax revenue per $ 1 ,0 0 0 of personal income, t Other includes selective sales and gross receipts taxes on alcohol, m otor fuel, tobacco, etc., property tax, death and gift taxes, and docum ent and stock transfer taxes. | Adjusted to include only those states imposing the specified tax. Source: U. S. D epartm ent o f Commerce were at least tw ice as large as revenues raised per rapid rise in revenues fro m state income and sales $1,000 o f income fo r the five lowest ranked states. taxes. Because these taxes have a broad base, it is For all f if t y states, the sales tax averaged possible to raise large amounts o f tax revenue $19.12 per $1,000 o f personal income in 1970, or through relatively small increases in the tax rates. about 30 percent o f the average to ta l tax revenue Because the d ollar volume o f sales and income are (Table highly correlated, the fact th a t the measure o f tax III). A n other $12.40 per $1,000 of personal income, or 19 percent o f the average total e ffo rt used here does n o t e x p lic itly allow fo r the tax revenue, was derived from state income taxes.4 sales tax base is n o t lik e ly to seriously bias the N ot only are these tw o taxes the most im p o rta n t expressed relationships. sources o f revenue, b u t they have also been the Sales Tax. The retail sales tax was used by 45 fastest growing in terms o f actual revenue raised. states in 1970.5 Am ong these states, however, During the post-World War II period, many states there are many differences in the application o f instituted these taxes to meet expanding needs fo r the tax. Actual sales tax rates range from 2 percent funds. incomes, upward in Indiana and Oklahoma to 6 percent in Pennsyl revisions in tax rates, and continuous grow th in vania, and the items th a t are subject to the tax retail sales also contributed im p o rta n tly to the vary considerably. The most com m only exempted The rapid grow th in item is barber and beauty parlor services, w hich is 4 The sales and income taxes average $ 2 2 .6 3 and $ 1 4 .6 9 , respectively, per $ 1 , 0 0 0 if only those states levying these taxes are considered. 5 States th at did not have a retail sales tax in 1 9 70 are: Alaska, Delaware, Montana, New Hampshire, and Oregon. 17 ECONOMIC REVIEW TA B LE III Sources and D istrib u tio n o f Revenue Per $1,000 o f Personal Incom e* Average, A ll States 1Q70 Per $ 1 ,0 0 0 of Income Source Total general revenue Intergovernm ental revenue from Federal government Total tax revenue General sales tax Individual income tax Other taxest Other revenue^ Percent of Total Revenue $ 1 0 4 .9 5 1 0 0 .0 2 5 .9 9 6 4 .7 3 19 .12 12 .40 33.21 14 .23 2 4 .8 61.7 Percent of Total Tax Revenue % 1 0 0 .0 % 29 .5 19.2 5 1 .3 13.5 * Personal income data are U. S. Departm ent o f Commerce estimates fo r calendar year 1969. tO th e r taxes include "o th e r" as defined in Table II plus comparable income taxes and property taxes. t Consists o f revenue received from local governments in the form of shared revenues and grants-in-aid, as reimbursed fo r services, or in lieu of taxes. Source: U. S. Departm ent o f Commerce taxed by o n ly seven states. U tilitie s, especially arises, and local transportation, are also exempted fro m the government u nits.7 retail sales tax by many states. Other exemptions range from food and clothing to repair services. In addition to exem pting entire classes o f goods is adaptable to sharing w ith other Income Tax. A state personal income tax is more com plicated to adm inister than the sales tax, b u t most states th a t use the income tax tr y to sales taxation, some states allow keep it as simple as possible. In comparison w ith special taxes and tax rates on specific items. For the Federal income tax, these e ffo rts have been example, successful. from general C onnecticut and some other states exem pt admission charges from the general sales O f the 44 states th a t used a personal income tax, b ut impose a separate admission tax. Tw enty- tax in 1970, V erm ont and Alaska opted fo r the five states allow co u n ty and m unicipal govern simplest o f all methods—a fixed percentage tax ments to impose a sales tax levy in addition to the levy on individual Federal income tax lia b ility . For state sales tax. In most states, the additional tax the other states, com plications are introduced at rate is lim ited to either 0.5 or 1 percent; however, tw o levels: (1) in calculating the tax base and (2) Alaska allows m unicipalities to tax at a rate up to in determ ining the applicable tax rate. 5 percent, and Colorado and New Y o rk allow up to 3 percent. In some states, the d e fin itio n o f income fo r tax purposes is related to one o f the several income Regardless o f its form , however, the sales tax is concepts used in the Federal income tax return, relatively simple to understand and administer. It w hile in other states the tax base is independent o f can also be used to obtain large amounts o f the Federal income tax. States may or may not revenues, is relatively easy to increase if the need 0 a llo w standard or itemized deductions or For a com plete list of exemptions by state see, S tate a n d 7 O f the 2 5 states th at perm it a local sales tax levy in L o c a l Sales Taxes, (N ew Y o rk : Tax Foundation, Inc., addition to the state sales ta x, 19 administer the entire 1 9 70). tax at the state level. 18 MARCH 1972 deductions fo r Federal taxes. A ll states provide fo r The graduated income tax is generally regarded some type o f personal exem ption, but both the to be the most progressive o f the major sources o f size and the rules governing the exem ption vary state tax revenue w ith respect to income. w idely. For example, Maryland allows $800 per income tax person; Mississippi allows $4,000 fo r a single considered to be pro p o rtion a l; and a sales tax, individual and $6,000 fo r a fa m ily; and Wisconsin regressive. However, the various adjustments and allows a tax credit o f $10 per person to be applied alternatives to the tax base th a t are perm itted to the actual tax bill. based on a fla t O An rate generally is under the state laws have drastically altered these The rate structures o f the state personal income general relationships. The fla t rate personal taxes can be classified in to tw o general categories: exem ption, fo r instance, w hich is used in many graduated most popular graduated income taxes, tends to lessen the degree method is the graduated rate structure, and it is o f progressiveness because an individual in a high used by over tw o-third s o f the income taxing tax bracket w ill benefit more from the exem ption states. The New Y o rk income tax structure begins than an individual in a lower tax bracket. W ith w ith a 2 percent rate on the firs t $1,000 o f income respect to the retail sales tax, exemptions can and make the tax less regressive. Therefore, the low and fla t. increases to By far the 14 percent on income over $23,000. The fla t rate tax generally tends to be a income relatively low rate, such as the 2 percent used in benefit from the exem ption o f a necessity such as individual w ould Indiana. A nothe r fo rm o f the fla t rate tax is a food from fixed percentage o f Federal income tax lia b ility , degree o f regressiveness. w hich is used by Alaska and V erm ont. This m ethod, although a fla t rate, tends to tax high receive the greatest the sales tax base, causing a lower Intergovernmental Transfers. A n o th e r m ajor source o f revenue fo r state governments, w hich has incomes more heavily than low incomes because o f been increasing rapidly in recent years, is inter the graduated rates b u ilt in to the Federal tax governmental transfers from the Federal Govern structure. ment. Most o f the funds are cu rre n tly earmarked When the fla t rate income tax is used, revenue fo r specific uses, such as highway construction, can be increased in a manner similar to retail sales education, and welfare. However, there has been tax; all considerable debate concerning the desirability o f th a t is required appropriate however, legislation. is the enactment o f W ith a graduated tax, new schedules must be constructed. Depending on the p riorities o f the state, the allowing the recipient, both governments and individuals, fu ll discretion in spending transferred funds. Under most "revenue sharing" plans, the increase can be evenly spread out over all incomes or concentrated on one or more income levels. W ith an income tax, it is also possible to change the am ount o f total revenue raised by the tax O No other single source o f tax revenue contributes as much as 10 percent of total revenue, and only the corporate income tax contributes as much as 5 percent (see Table II) . In this article, a tax is considered to be w ith o u t changing the tax rate structures. This can progressive if the am ount o f tax paid as a percentage of be accomplished by changing the rules concerning income increases as income increases. If the percentage of exemptions, deductions, and credits or by altering the d e fin itio n o f taxable income. income paid as tax is equal for all income levels, the tax is considered to be proportional; and if the percentage decreases, the tax is considered regressive. 19 ECONOMIC REVIEW recipient government u n it w ould be granted a In view o f the current debate involving the specified share o f designated funds instead o f relationship o f Federal revenue sharing to state tax receiving fixed amounts o f money fo r a specific efforts, tax e ffo rt and Federal transfers to states project. One o f the prime objectives o f such plans were is to transfer Federal tax revenue fro m those areas analysis. Results indicate th a t 21 percent o f the o f the co u n try w ith the least pressing need to areas Federal transfer payments per $1,000 o f personal w ith the sharing greatest relative need. The revenue in Congress plans under consideration w ould make the size o f the grant dependent upon statistically income in depended 1970 on the related were tax by simple distributed efforts regression as if they of the states (measured by the tax paid per $1,000 o f personal a com plicated form ula based on the population, income). If the population o f the state were added income level, and possibly the tax e ffo rt o f the to the regression, an additional 10 percent o f the recipient approved government. by the In the version House Ways and recently Means Federal transfer can be explained. Per capita transfer payments to states were n o t as strongly Com m ittee, an additional allowance is made fo r related the degree o f urbanization, w ith large urban areas transfers per $1,000 o f personal income, although eligible to receive the greatest benefits.9 tax to tax e ffo rt and e ffo rt did account fo r population as were 12 percent o f the In 1970, Federal transfers provided nearly 25 Federal transfers per capita and population an percent o f to ta l state revenue and 40 percent o f additional 4 percent. In general, then, w ith o u t total tax revenue. N ationally, these transfers repre revenue sharing and a specifc form ula fo r the sented $1,000 o f d is trib u tio n o f Federal funds, a state's own tax personal income. The range o f Federal transfers, an average of $25.99 e ffo rt and population were in fact related to the however, was from $82.24 in Alaska to $15.13 in state's revenue per $1,000 o f personal income New Jersey. In Alaska, o nly 8.7 percent o f the from the Federal Government in 1970. To date, general revenue came fro m the Federal Govern revenue sharing proposals have contained form ulas ment, even though the transfers were 120 percent th a t o f to ta l tax p opulation, some aspects o f its tax e ffo rt, and its revenue.10 In per New Jersey, even w ould take in to consideration a state's though the transfers per $1,000 o f income appear level o f personal income in determ ining the alio- to be small, the Federal payments provided 21.7 cation o f funds. percent o f the State's general revenue fro m all th a t revenue sharing w ould n o t replace all o f the sources and 34 percent o f its tax revenue. current Federal revenue transfers to states. 11 It should be noted, however, Property Tax. In 1970, property tax accounted g The $ 5 .3 billion revenue sharing plan agreed to by the fo r o n ly 2.3 percent o f state tax revenue, b u t 84.9 House Ways and Means Com m ittee contains both general percent o f local tax revenue. This tax is relatively and special revenue features. The proposal contains no restrictions on the $ 1 . 8 billion allocated to state govern u n im p o rtan t at the state level, b u t it does provide ments, while $ 3 .5 billion allocated to local government units w ould be restricted to certain types o f spending, including capital outlays, maintenance, and operations. 11 It is likely th at, if only those funds th at are transferred to states at the discretion o f the Federal Governm ent—not depending on matching funds or other fixed programs— 10This is caused by the large am ount of general revenue are studied, the im portance o f the level o f income in the derived through state oil and gas holdings and leases. state would increase. 20 MARCH 1972 local educational funds. Pennsylvania, K entucky, and West V irg in ia —o nly Recent rulings by the California Supreme C ourt the major p o rtio n of K entucky and West V irginia had both a personal and other state supreme courts, however, have income tax and a retail sales tax in 1970.14 Ohio raised the question o f whether or not the property and Pennsylvania rely p rim a rily on a sales tax, tax can be considered an equitable source o f funds although fo r income from its corporate income tax (see Table com m u n ity schools.12 If the "C a lifo rn ia Pennsylvania does receive substantial decision" is upheld, the financing o f public educa IV ). It is, therefore, not surprising th a t Ohio and tio n could become a state function. If this should Pennsylvania receive considerably lower tax occur, the states w ould be required to increase revenue the ir tax revenue, on average, as much as 80 K entucky and West V irginia. As m ight be expected per $1,000 o f personal income than percent, ceteris paribus. An increase o f such large from the previous discussion, K entucky and West propo rtion in state tax revenue could be financed Virginia, w hich ranked 43rd and 47th, respec through a broad-based tax such as the income or tive ly, in per capita personal income among the 50 sales tax. A lthough the additional state taxation states—received more intergovernmental transfers could be offset by lower local property taxes, it is per $1,000 o f personal income (and per capita) u n like ly th a t individuals w ould fin d the changes from the Federal Government than Ohio and offsetting. Many w ould fin d their to ta l state and Pennsylvania, which ranked 15th and 16th, respec local tax burden increased, while others w ould tive ly, find their burden decreased. A lte rn a tive ly, a d istrib u tio n pattern o f Federal transfer payments state may decide to maintain the current property possibly reflects the greater need in the relatively in per capita personal income. This tax structure and to make the state the recipient low income states. This is especially true o f West rather than the local school district. Either method Virginia, which received more than double the w ould require a greatly expanded revenue e ffo rt national average transfer per $1,000 o f personal by the state, b u t w ould perm it equal d istrib u tio n income. o f funds among all schools in the state, thus The 1970 d is trib u tio n o f Federal transfers to elim inating the objections raised by the California Fourth D istrict states can be compared w ith the court. d is trib u tio n th a t w ould result fro m any o f the T A X STRUCTURE AND FEDERAL TRANSFERS IN THE FOURTH FEDERAL RESERVE DISTRICT share o f all intergovernmental transfers fro m the proposed revenue sharing plans by calculating the Federal Government th a t is allocated to each o f the Fourth D is trict states. The most notable O f the fo u r states included w h o lly or partly difference between the 1970 d is trib u tio n pattern w ith in the Fourth Federal Reserve D is tric t—Ohio, and the revenue sharing plan proposed by the 1~2 A d m in istra tio n Several other state courts, including those in Texas, in 1971 is th a t the tw o most New Jersey, have also ruled th at the populated states in the D istrict (Pennsylvania and property tax can no longer be used as the prim ary source Ohio) w ould receive a greater p roportion o f total Minnesota, and of school financing. 13 A ltern atively, the Federal Government may provide the 14 Exactly how the tax burden will shift among individ needed financing required fo r education either through uals depends on what taxes are used, what tax schedules general revenue sharing or earmarked grants. are used, and on how the property tax is administered. 21 TAB LE IV Sources and Distribution of Tax Revenue in the Fourth District States Per $1,000 of Personal Income* 1970 O hio Pennsylvania 100.0% n.a. n.a. $ 7 6 .4 0 2 9 .0 9 1 3 .20 4 .2 9 n.a. 42 Percent of Total Tax Revenue 49 34 n.a. n.a. $ 6 4 .3 2 2 1 .9 6 n.a. 12.27 100.0% 30 38 .7 n.a. n.a. 34.1 n.a. 19.1 21 6 1 .3 n.a. 3 0 .09 4 6 .8 48 2 0 .4 5 21.81: $4 2.41 To tal tax revenue 16.41 Sales tax n.a. Individual income tax n.a. Corporation income tax Revenue from Federal 2 6 .0 0 Other ta xest Revenue from Federal 1 6 .3 8 G overnm ent 100.0% 2 3 .5 $ $ 3 ,7 3 8 15 Nation; Rank $ 3 ,6 5 9 n.a. N ot applicable. * Personal income data are U. S. D ep artm ent o f Commerce estimates for calendar year 1969. t As defined in Table II. | As percent o f to tal general revenue. Source: U. S. D epartm ent o f Commerce Percent of To tal Tax Revenue Tax per $ 1 ,0 0 0 Percent of To tal T ax Revenue Per capita personal income Tax per $ 1 ,0 0 0 National Rank Tax per $ 1 ,0 0 0 West Virginia K entucky 16 National Rank T ax per $ 1 ,0 0 0 Percent of T o tal Tax Revenue 100.0% 38.1 17.3 5.6 19 n.a. $81.31 3 8 .3 8 8.4 6 0.8 2 2 9 .8 2 3 9 .0 n.a. 3 6 .6 6 2 8 .8 $ 17 $ 2 ,8 4 7 15 8 43 National Rank 8 1.0 4 31 n.a. 3 3 .6 5 4 1 .4 n.a. 53.71 3 5 .8 $ 4 7 .2 10.4 $ 2 ,6 0 3 4 47 MARCH 1972 Federal funds, w hile the other tw o states (West the state level o f Government was raised through Virginia and K entucky) w ould receive a smaller tw o broad-based taxes—the personal income tax propo rtion o f funds (see Table IV ). It is like ly, and the retail sales tax. The im plem entation o f however, th a t the programs that are most sensitive these taxes and the to need and low levels o f income w ould continue greatly among the states. Many o f the state income independently taxes of any revenue sharing plan, have resultant tax graduated rate e ffo rt vary structures allowing although the House Ways and Means Com m ittee progressiveness, b u t often the degree o f progres proposal favors those areas w ith the greatest need, siveness particularly deductions, and credits. cities and areas w ith low average incomes. is lessened by income exemptions, Being broad-based, the sales and income taxes In the F ourth D istrict, as w ould be expected, are capable o f raising large amounts o f revenue. the tw o states w ith the largest income bases—Ohio Small increases in the tax rates result in large and Pennsylvania—are also the most capable o f increases in tax revenue. A lthough the property increasing their tax revenue. Neither o f these states tax is also broad-based, other things being equal, had a personal income tax in its base can be increased only through property 1970, although Pennsylvania did receive 19 percent o f its tax revaluation, and require taxes th a t the financing o f education—w hich is currently states is also relatively low; financed p rim a rily recent usually decisions are upheld in higher courts, it is possible these If increases revenue from a corporate income tax. The revenue in approval. rate per $1,000 o f personal income received from sales this is probably because o f the exem ption o f food voter actual state co u rt through property ta x a tio n — and numerous other items. West V irginia and may become a state fu n c tio n requiring an increase, K entucky, however, already rank among the to p on average, o f as much as 80 percent in state tax 15 states in terms o f tax e ffo rt and use both a sales revenue. and in c o m e ta x , as well as a c o r p o r a te in c o m e A nother major source o f state revenue th a t has tax. In spite o f a relatively low retail sales tax rate been growing in im portance is transfers and grants o f 3 percent, West V irginia ranks fo u rth among all from the Federal Government. In coming years, states in revenue per $1,000 o f personal income intergovernmental transfers are lik e ly to increase, from a sales tax because o f its low level o f per although the method o f d is trib u tio n may become capita income. The regressions discussed in the more form alized w ith the advent o f a revenue previous section, however, indicate th a t fo r all 50 sharing or sim ilar program. Congressional pro states the relation between per capita income and p o sa ls consideration tax e ffo rt is weak. distrib u tin g SUMMARY AND CONCLUSIONS state's population, b u t also its income level and currently under fo r Federal funds to the states include provisions th a t take in to consideration n o t o n ly a Nearly one-half o f all the 1970 tax revenue at possibly its tax e ffo rt. 23