W ,- _fl. smz‘m n; . ..~_. “ ~ ’75:: ,I‘ 09 DJ |II» ' EL.) I’I-I‘EI I ‘ 'J'UH [M II I ..\ I.- 0'», tjlb‘”| H 'I]V"i’l '(I I"-l‘ "')' \‘\~1‘I .33: 3'; I?!“ QEE‘E‘ a :1.“ ~.' ' I49” 1| I I? x 4' 5902)]; HIV}: L‘.’ 5L? ._I.I\I.'\.I'. am ..I ( impl— 5’ >3. II . h!" |‘ I ' ‘ kw? ‘ L L “_ {III-1ft. f6‘x: \L 421.15.. . . . I ‘a‘-}{:' 9:3)... ,4 134% {"1“ °‘|“‘ 5"le "T Mi ”IS: LL o~::’f;':(_ ,. ‘11.“ . .2‘ .d I; . “If I \u , ‘J .59-5'1- ‘ - .¢ ~52?" 5‘ a' hp‘ L .' I9: I5 - J‘ 7"I'.l_- éfiw‘; I .5..' 315;: I fr‘IIa’ 0“‘h‘\“0-{‘ I .rta‘. ‘t‘ mum”? fiL£APIrI 9 y-J'IJ :75 I1 . {A ,FIIXEH'IUZIQ: ”In I r 0 ' I‘IIU'Q‘II‘I) '. ”125-.” L1“: . I I“) 30} ' J. ‘4‘ ' I 'I I I‘ .3.” II ll'. ‘ T M l ‘ "II: I E I.‘ . 1.": Am ‘ J5“ u '.'."f.I" "I Eur 10 I ‘II, ‘ ' ugwfifll" ‘i" n‘ Hwy-“f”; ' o ‘LI ’ I ‘ “ ‘ ‘ ’ L -. I Ia“ IN.“ I - “ . I , . ‘ ‘ I I. , .-.-I.I..{I.I,-.~I-,., I - I...» I .,=_ .I I .‘U I; .I .. M . - I, T_ i) '—!“,.‘".J .'. ':. IQ \: ‘1. ":~"1I'LL“‘"'J-"I‘ . ‘A'1‘EI‘iIJ a ‘,-'-‘.>” fig '5: '_ ‘.‘ Unfit ".u 'JTI‘ I'|(0'. 'I‘ N“. .l'll‘ " ~ LK'.".W“‘.I". ‘3)“:ka V I 'HrZVI ‘1 .» .1’ "05' l’ In ‘.I " ‘I-h‘u ' ‘. 4' . .I/ ”5-3: Ann "-3 3 ‘1‘0: " I'.' II "cf-1‘ I E ' II]: ' ’ :P“ ' {I . " ‘3‘.- MIA-A l'. ‘I‘O:“‘I "‘ I ~" I I‘. 7? E. M u- ""’."."v.- ..‘.-\~,hv J~ .71 In.” - "I? _. \ “w I ‘ K J? ‘ I; \".’.'r . I. . -' . )4 ,I L, .. . - <" I‘ :"I .J' 0.. .' “4...“; ., .,%,I.-~ v ,, “,3 ‘ - ...": 3.x. I - "II ‘ ‘ , r"“ 1" NOV Ilk I .1 A III ‘7‘ I v4 . ;‘~I.I‘“'£VL'I‘I CUZ'I’ ‘- ‘l I,H'N:T‘ Sigh? I L”. :3“; ‘. . . . . y ‘ " . I . ( . ‘ .I hf‘ ’- I . ‘ ‘-" 3.0. . K .' “"0 .II ‘r’r I -‘ «.5. . . u _' ., . _. rI ‘ I . ' I'-- I‘- ‘ ' I. ":Itt‘IIr.‘ ”II E ’W "" ‘ w I -' """1""~'I‘ M1 was? _ I" " ; « |I I“. . ‘0‘ .‘IINI‘ I I ._ I ‘ I A" . Uta J‘I‘ 1‘ pu| ..I I "I v” ‘ "‘1' I'- .' . I. ‘ :E I'. I" .41-'I,',.‘I‘. E. ‘ 0- “’ -.‘:I 'I, .. u”'l""'”" '; I..-JI'_‘. "' . I. I. I '. . I. II M . .‘. _. “ ' llilJ‘ ' k (' of “V M1. ': W .' .. ‘ ‘ ‘ . 0'“ x “‘1 {'EZII'S: I . r‘4r‘ H‘ . ' ‘ I."'I‘I'I ‘ ‘ ' . 7" . I" 1' I. 1*. I‘M“ ' .. m' III-‘7'.” . I I _‘|‘ ' ’ ‘. ' I 4- ‘IIII '5, I' 11”.)”le .JA“ fill.“I .":) (“I‘VE , I "."ij ‘ it'fljfi'; '1 ' 'r, {NI I1'(!.%$IWII! :I” h " “‘I I ‘fimw'll’l If” I“ I. ‘ f‘ Ilp «II‘IHIII 59"" ,. . 'I:.:.I. 'IIIIIIII I'.' !'_.'I.' m. ”Muir mummmulluluunlmnunwwymn L 3 1293 006 THESlS Fun-n... .. m' ‘ L [BR A R Y iii “ Egan State " University This is to certify that the thesis entitled Municipal Bond Credit Ratings: Regional Patterns and Spatial Correlates presented by Donald J. Zeigler has been accepted towards fulfillment of the requirements for Ph.D . degree in Geography ajor professor I)ate :§?%/:;€;/<;%7 0-7639 OVERDUE FINES; ' "Ir 25¢ per day per item Ramps Lramv MATERIALS: F Place in book return to remove 5 chum from circulation records MUNICIPAL BOND CREDIT RATINGS: REGIONAL PATTERNS AND SPATIAL CORRELATES By Donald Jay Zeigler A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Geography 1980 ABSTRACT ‘MUNICIPAL BOND CREDIT RATINGS: REGIONAL PATTERNS AND SPATIAL CORRELATES By Donald Jay Zeigler The general obligation credit ratings assigned by Moody's Investors Service to central cities on the bond market are analyzed in this re- search in an effort to delineate the regional bond rating patterns which have evolved over the 1960 to 1980 period and to determine the degree of correlation between bond ratings and selected social, demographic, and geopolitical characteristics of cities in the various rating cate- gories. Many of these selected characteristics are subsequently used to determine how successfully central city credit ratings may be predicted by employing step-wise multiple regression and multiple discriminant models as classification tools. The hypotheses under investigation are that the bond ratings of central cities exhibit meaningful spatial and temporal variability which parallels the decline of the American manu- facturing belt and the rise of the peripheral amenity belts, and that non-financial characteristics of central cities may be used to predict credit standing because they are important correlates of financial well- being. This study departs from previous investigations of credit ratings in that it is the first to focus on the regional aspects of the ratings and the first to use only non-financial correlates in multivariate models. A nationwide set of 354 central cities, comprising all central cities on the municipal bond market during the past two decades, serves as the study area for this analysis. reg 196 deer Can: of 1 of 1 the int: more cept exam high 38th area rate, stOCk nth Donald Jay Zeigler Graphic, cartographic, and statistical analyses of the changing regional patterns of central city bond ratings reveal that over the 1960 to 1980 period the Northeast Census region experienced a dramatic decline in credit standing, the South a dramatic improvement, the North Central little net change, and the West only slight improvement. Cities of the North Central emerge as the highest rated in the nation. Cities of the Middle Atlantic and the East South Central divisions emerge as the lowest rated. Over the decade of the 19703 location has become an increasingly better predictor of credit ratings indicating a trend toward more homogeneous bond rating regions and an increasingly regional per- ception of the American metropolitan system. Of the individual correlates of municipal bond ratings which are examined using crosstabulation analysis and analysis of variance, the highest degrees of association are to be found with city pOpulation size, metropolitan growth rate, annexation activity, percent of a metropolitan area's population in the central city, per capita income, unemployment rate, percent of population black, and the growth and age of the housing stock. Several multivariate social indices are also closely associated with municipal bond ratings, particularly a per capita needs index and several quality of life indices. These associations indicate not only that social characteristics of cities are important correlates of the ratings but also that the assignment of credit ratings discriminates against the neediest cities. A comparison of the multiple regression and discriminant analyses reveals that the latter yields a superior equation for the prediction of bond ratings. By utilizing a selected set of demographic, social, and geo‘ reg can has ign: thet ratj betn bei: Donald Jay Zeigler geopolitical variables and by running separate analyses for each Census region, the discriminant models are able to correctly predict 61 per- cent of central city credit ratings, 3 higher predictive accuracy than has been found in other discriminant analyses using only financial var- iables. This finding suggests that non-financial variables, heretofore ignored in studies of creditworthiness, need to be incorporated into theoretical models of financial well-being. Within the context of the nascent field of urban financial geography, the study of municipal bond ratings illustrates the need to devote more attention to the interface between the social well-being of city populations and the financial well- being of governmental units. C) Copyright by DONALD JAY ZEIGLER 1980 thank advic other and e tiCUla Prepa, CWPUt ACKNOWLEDGMENTS I would like to take this opportunity to extend a special word of thanks to my adviser, Stanley D. Brunn, for his expert guidance and advice in the preparation of this dissertation and for the countless other ways in which he made my doctoral program a stimulating, rewarding, and enjoyable experience. Also deserving of acknowledgment and thanks are the other members of my guidance committee, Joe T. Darden, Ian M. Matley, and Edward Cupoli, for their helpful discussions, comments, and criticisms during the course of my research. For her continuing support in general and her clerical help in par- ticular I want to extend my loving thanks to my wife Debbie. For the preparation of the graphics which appear herein I extend my appreciation and gratitude to Donald K. Emminger. And finally, for the computer ser- vices which made much of this analysis possible, I want to thank the Computer Center at Michigan State Universtiy for a truly efficient and reliable operation. ii TABLE OF C LIST OF TABLES . . . . . . . . . . . LIST OF FIGURES . . . . . . . . . . . ONTENTS CHAPTER I: INTRODUCTION TO THE PROBLEM . . . Background to the Problem . . . . What is a Credit Rating? . . What do Municipal Credit Rati What is the Bond Market? . . Statement of the Problem . . . . Significance of the Study . . . . Urban Financial Geography . Money Flows and Allocations Geography of the Urban Future Regional Analysis . . . . . CHAPTER II: RESEARCH METHODOLOGIES Hypotheses . . . . . . . . . . . Study Area . . . . . . . . . . . Variables and Data Sources . . . Bond Ratings . . . . . . . Demographic and Geopolitical Social Variables . . . . . . Financial Variables . . . Methods of Analysis . . . . . . Cartographic and Other Graphi Statistical Methods . . . . CHAPTER III: REVIEW OF THE LITERATURE Summaries and Critiques of the Municipal Bond ngs Measure? 0 O O O O O O O O O O O O O C O O O 0 Variables c Methods . Rating System Studies on the Correlates of Municipal Bond Ratings . . . References to the Regional Patterning of Municipal Bond Ratings . . . . . . . . . . . . . CHAPTER IV: SPATIAL AND TEMPORAL DIMENSIONS OF MUNICIPAL CREDIT RATINGS: 1960-1980 Regional Patterns in 1980 , . . . C Q Q I Q Q 0 Changing Patterns of Central City Credit Ratings, 1960 to 1980 I Q O O O O O O O I I O O 0 _iii 22 22 23 26 26 33 34 34 35 35 35 39 39 42 48 51 52 58 CFAF cm: A Graphic Analysis of Regional Change . . . . . . . . A Statistical Analysis of Regional Change . . . . . Bond Rating Dynamics of the Gilt Edge Central Cities, 1960- 1980 O O O O O O O O O O O O O O O O O O O I O O O O O O 0 Bond Rating Dynamics of the Grit Edge Central Cities, 1960- 1980 O O O O Q I O O O C O I O O O O O O O O O O I O O O 0 Summary . . . . . . . . . . . . . . . . . . . . . . . . . CHAPTER V: SOCIAL, DEMOGRAPHIC, AND GEOPOLITICAL CORRELATES OF MUNICIPAL CREDIT RATINGS . . . . . . . . Population Size . . . . . . . . . . . . . . . . . . . . . . Population Growth Rate . . . . . . . . . . . . . . . . . . Metropolitan Geopolitical Organization . . . . . . . . . . Selected Social Characteristics of Central Cities . . . . . The Per Capita Needs Index . . . . . . . . . . . . . . . Selected Quality of Life Indices . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . CHAPTER VI: GEOGRAPHIC PREDICTORS OF MUNICIPAL CREDIT RATINGS: REGRESSION AND DISCRIMINANT ANALYSES . . . . . . . A Multiple Regression Analysis of Municipal Bond Ratings . A Multiple Discriminant Analysis of Municipal Bond Ratings suma ry O C O O O O O O O O O I O O O O O O O O O O O O 0 CHAPTER VII: SUMMARY AND CONCLUSIONS . . . . . . . . . . . . Multivariate Analyses of Municipal Bond Ratings . . . . . . Spatial Correlates of Municipal Bond Ratings . . . . . . . Regional Patterns of Municipal Bond Ratings . . . . . . . . Directions for Future Research . . . . . . . . . . . . . . conCIUSion O O O O O O O O O O O O O O O O I O O O O O O 0 APPENDIX A: MUNICIPAL BOND CREDIT RATINGS BY CITY . . . . . . . APPENDIX B: CENSUS REGIONS AND CENSUS DIVISIONS . . . . . . . . BIBLIWRAPHY . O O O Q 0 O C O 0 O O O O C . I Q Q O O Q 0 0 . . iv 60 70 74 86 88 91 92 99 102 108 115 121 128 133 135 147 159 162 163 165 168 170 174 176 206 207 LIST OF TABLES Table 1. Long-Term State and Local Debt, 1944-1977 . . . . . . . 6 Table 2. Regional Distribution of Central Cities . . . . . . . . 25 Table 3. Distribution of Central Cities by Population Size, 1976 O O O O O O O O I O O O O O O O O O O O O O O O I I 27 Table 4. Sources of Data for Central Cities and Metropolitan Areas 0 O O O O O O O 0 O O O O O O I O O O O O I O O O 28 Table 5. Municipal Bond Rating Research Using Regression and Discriminant Analysis . . . . . . . . . . . . . . . . . 44 Table 6. The Distribution of Central City Bond Ratings by Census Region: 1980 . . . . . . . . . . . . . . . . . . 53 Table 7. The Distribution of Central City Bond Ratings by Census Division: 1980 . . . . . . . . . . . . . . . . . 54 Table 8. The Regional Distribution of Central Cities in Each Rating Category: 1980 . . . . . . . . . . . . . . . . . 56 Table 9. A Crosstabulation Analysis of Bond Ratings and Census Region: 1970-1980 c o u o o o o o o o o o o o o o o o o 72 Table 10. A Crosstabulation Analysis of Bond Ratings and Census DiViSion: 1970—1980 0 o o o o o o o o o o o o o o o o o 73 Table 11. Bond Ratings and City Population Size . . . . . . . . . 94 Table 12. Bond Ratings and Metropolitan Population Size . . . . . 94 Table 13. Bond Ratings and‘Metropolitan Growth . . . . . . . . . . 103 Table 14. Central City Dominance . . . . . . . . . . . . . . . . . 106 Table 15. Bond Ratings and Social Characteristics of Central Cities: A Comparison of Group Means . . . . . . . . . . 110 Table 16. Components of the Per Capita Needs Index . . . . . . . . 116 Table 17. Crosstabulation Analysis of Bond Ratings and the Per Capita Needs Index, 1970—1980 . . . . . . . . . . . . . 118 V la la TaE lat Tab Tab: Tabl Tabl Iabl Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. A2. Crosstabulation Analyses of Bond Ratings and State Qllality Of Life Indices o o o 0 o 0 Q o o o g Q o 9 Crosstabulation Analysis of Bond Ratings and Liu's Metropolitan Quality of Life Index . . . . . . . . . Crosstabulation Analysis of Bond Ratings and Subcomponents of Liu's Metropolitan Quality of Life Index 0 O O O O C O O I O O O O O O O O O O Crosstabulation Analysis of Bond Ratings and Zeigler's Quality of Life Index . . . . . . . . . . Summary of Variables Associated With Bond Ratings During the 19703 . . . . . . . . . . . . . . . . . Predictor Variables Used in Regression and Discriminant Analyses . . . . . . . . . . . . . Results of the Nationwide Regression Analyses, 1974 Results of the Nationwide Regression Analyses, 1980 Classification Results of the Nationwide Regression Analyses, 1974 . . . . . . . . . . . . . . . . . . . Classification Results of the Nationwide Regression Analyses, 1980 O O O O O O O O O O O O O O O O O O 0 Results of the Regional Regression Analyses, 1974 and 1980 O O O O O O C Q C O O O O O O O O O O O O 0 Results of the Nationwide Discriminant Analyses, 1974 and 1980 O O C O O O O O O O O O O O O O O 0 Classification Results of the Nationwide Discriminant Analysis, 1974 . . . . . . . . . . . Classification Results of the Nationwide Discriminant Analysis, 1980 . . . . . . . . . . . . Results of the Regional Discriminant Analyses, 1974 and 1980 o o o o o o o o o o o o o o o o o o 0 Municipal Bond Credit Ratings by City . . . . . . Census Regions and Census Divisions . . . . . . . . vi 123 126 127 129 131 136 138 139 141 142 144 148 152 153 158 176 206 Figm FigUl Figur Figur Figurc Figure Figure Figure Pigure Figure PiEUre FiéUre 3 Figure 1 The l Figure 1. Figure 2. Figure 3. Figure 4. Figure 5. Figure 6. Figure 7. Figure 8. Figure 9. Figure 10. Figure 11. Figure 12. Figure 13. Figure 14. Figure 15. LIST OF FIGURES ‘Moody's Municipal Bond Credit Ratings . . . . . . . Direct and Indirect Determinants of Municipal Bond Credit Ratings . . . . The Gilt Edge and Grit Edge Central Cities, 1980 . . Number of Central Cities in Each Bond Rating Category, 1960 and 1967 . . Number of Central Cities in Each Bond Rating Category, 1970, 1975, 1980 . o o o o o o o o o o o o A Regional and Temporal Comparison of Central City Bond Ratings . . . . . . . . The Gilt Edge and Grit Edge Central Cities, 1960-1980 0 o o o o o o O 0 Changes in Central City Bond 1970-1980 0 o o o o . o o 0 Rating Profiles of Gilt Edge Northeast . . . . . . . . . Rating Profiles of Gilt Edge North Central . . . . . . . Rating Profiles of Gilt Edge South and West . . . . . . . Rating Profiles of Grit Edge New York and New England . . Rating Profiles of Grit Edge Pennsylvania and New Jersey Rating Profiles of Grit Edge the North Central . . . . . Rating Profiles of Grit Edge the South Atlantic . . . . . vii Ratings, 1960-1980 and Cities in 12 57 61 62 64 66 69 75 76 77 78 79 80 81 Yigu: Pigur Pigur Pigar Figun Figure Figure Figure Figure 16. Rating Profiles of Grit Edge Cities in the East South Central . . . . . . . . . . . . . . . 82 Figure 17. Rating Profiles of Grit Edge Cities in the West South Central . . . . . . . . . . . . . . . 83 Figure 18. Rating Profiles of Grit Edge Cities in the weSt ' O O O O Q ' O 9 Q 0 O O C O O O I I O C 9 85 Figure 19. Bond Ratings and City Population Size, 1976 . . . . 95 Figure 20. Temporal Bond Rating Profiles of the Nation's Ten Largest Cities . . . . . . . . . . . . . . . . . 97 Figure 21. Bond Ratings and Metropolitan Population Size . . . 98 Figure 22. Bond Ratings and the Per Capita Needs Index, 1980 . 119 Figure 23. Overrated and Underrated Cities in the Nationwide Discriminant Analysis, 1974 . . . . . . . . . . . . 155 viii 1i lv‘h is: inc nit tie “En: tal CHAPTER I INTRODUCTION TO THE PROBLEM When local governmental units in the United States need to acquire large sums of money in order to fund major capital improvements such as bridges, highways, parking facilities, or schools, they may venture onto the securities market by authorizing the issuance of bonds. A bond may be defined as: an interest bearing certificate of debt, being one of a series constituting a loan made to or an obligation of a government or business corporation, a formal promise by the borrower to pay to the lender a certain sum of money at a fixed future date with or without security, and signed and sealed by the maker. (Garcia, 1962, 79) The sale of bonds enables incorporated municipalities and other pub- lic agencies to borrow money from private institutions and individuals who have funds to invest. "Municipals" is the term applied to any bond issued by a state or local governmental unit. When these securities are purchased by investors, they earn interest which is exempt from federal income taxes. This tax-exempt property is the primary attraction of mu- nicipals, in competition with corporate bonds, as investment opportuni- ties. Bond issues, general taxation, and intergovernmental transfer pay- ments constitute the three main sources of revenue for local governmen- tal units. Bond issues, however, differ from the other two methods of l IEV ofi Up‘ fai: 2 revenue generation in that communities which issue bonds assume a burden of debt that must be paid off, with interest, over a span of time ranging up to fifty years into the future. When this debt is backed by the full faith, credit, and taxing power of the municipality, the bonds are known as "general obligations" or "GOs." In 1977, total long term local debt amounted to $157 billion of which $95 billion was full faith and credit debt; the remainder was nonguaranteed (The Tax Foundation, 1979, 253). Because the number of municipalities with outstanding bond issues numbers in the tens of thousands, investors would find it difficult to investigate and pass judgment upon each municipal bond issue were it not for the services of the the two national credit rating agencies, Moody's Investors Service and Standard and Poor's Corporation. Credit ratings assigned by these two companies provide potential buyers with an indica- tion of the relative soundness of municipalities' bond issues as long- term securities. These ratings, expressed by an alphanumeric code to be discussed later, provide a third party's evaluation of a municipality's prospects for the future. Lower credit ratings are likely to undermine a city's ability to attract investment dollars on the bond market and usually result in higher interest costs on the money borrowed. Interest paid on debt, or debt service, amounted to an expenditure of $7.9 bil- lion by local governments in 1977 (The Tax Foundation, 1979, 231). This figure is the aggregate price localities paid for the use of funds pro- vided by the private sector of the American economy. This study is concerned not with the operation of the bond market but with the ratings assigned by Moody's Investors Service to municipal- ities issuing general obligation bonds. According to Moody's, ratings are not the outcome of a fixed statistical formula but instead "include 3 the recognition of many non-statistical factors" (Moody's, 1979, v). As a result, the specific characteristics that are utilized by the rating agencies and communicated to the investing public have been Open to ques- tion as evidenced by the testimony before the U.S. Congressional commi- tees (U.S. Congress, 1967-1968 and 1977), the critical evaluation of the ratings by a Twentieth Century Fund Task Force (Twentieth Century Fund Task Force, 1974), and the range of explanations about what credit rat- ings actually measure which appears in the literature. Because the credit ratings of general obligation bonds provide an index for comparing one city with another, they lend themselves well to geographic analysis. This study attempts to delineate the regional pat- terning of municipal bond ratings for central cities of Standard Metro- politan Statistical Areas (SMSAs) over the past twenty years and to de- fine the distinguishing demographic, geopolitical, and social correlates of the ratings for cities in the different bond rating categories. While it is impossible for anyone but Moody's or other ratings agencies to say exactly what considerations are encapsulated in the ratings, it is still possible to explore the dimensions of regional variation in non-finan- cial characteristics which may be communicated by the ratings. With this in mind, it is also the aim of this investigation to determine how much of the spatial variation in assigned credit ratings of central cit- ies may be accounted for by reference to variables in the demographic, geopolitical, and social realms and to determine which of these factors best discriminate among central cities in various rating categories. These non-financial variables have largely been ignored in previous at- tempts to develop statistical models for predicting the ratings. E“ .n H Pr t1 4 Background to the Problem What is a Credit Rating? The credit ratings assigned to long—term general obligations are or- dinal level measures of the future financial prospects of local govern- mental units as assessed by private enterprise or, more specifically, by small groups of investment analysts employed by Moody's and Standard and Poor's. Both agencies are based in New York City. Any municipality in the United States may apply to either agency on a fee-for-service basis and receive a credit rating, provided that all required information is supplied to the rating agency. Fees are computed according to the pop- ulation of the borrowing unit and the amount of work involved in proces- sing the rating. In 1974 fees ranged from $650 for a municipality with fewer than 10,000 inhabitants to $1,750-$2,500 for a municipality with more than a million inhabitants (Moody's, 1974, 16). Prior to 1970, no fee was charged. The rating system itself was devised by John Moody who began rating railroad securities in 1909 and eventually applied his system to the full range of corporate and municipal securities as well. ‘Moody's intent was "to provide investors with a simple system of gradations by which the relative investment qualities of bonds may be noted" (Moody's, 1978,v). The first published municipal bond rating appeared in 1918 (Ellinwood, 1957, 74), more than century after the first municipal bond was known to have been issued by an American city, New York City in 1812 (Greenberg, 1977, 339). Standard and Poor's did not begin rating municipal bonds until 1950. Since World War 11, state and local government debt has grown much more rapidly than private debt, federal debt, or the gross national is nu re hav an- ”BO: the tiOn judg: haVe Posed will inter 340) are a; are av anothe O Clal p DESS C, the 5 product (Hastie, 1972, 1729). The increase in long-term debt outstanding is presented in Table 1 for both state and local governments. As the number of bond issues has increased to some 92,000 as of 1967 (U.S. Cong- ress, 1967, 3), it has become increasingly difficult for investors to ac- quire their own information on the creditworthiness of all municipalities in which they might want to invest. This has heightened the importance and widespread utilization of the rating agencies' evaluations. Ratings have increasingly come to guide the buying habits of investors who need an information broker's services. As Smith (1979, 330) has noted: "Bonds are not avoided or acquired because of the ratings, but because of the characteristics reflected in the ratings." Bonds which are not rated at all attract few bidders. Rather than objective, quantitative measurements of financial condi- tions, municipal credit ratings have been and continue to be qualitative judgments about the creditworthiness of local governmental units which have issued "full faith and credit" bonds. In essence, ratings are sup- posed to communicate to investors the likelihood that the municipality will be able "to earn or to raise by taxation sufficient funds to pay the interest and principal on its debt" when they fall due (Greenberg, 1977, 340). The ratings admittedly represent judgments about the future and are appealing to the geographer because of the small scale at which they are available. Ratings provide an index for comparing one city with another and for measuring improvement or degradation in a city's finan- cial prospects over time as perceived by the investment community. The ratings which are used by Moody's to indicate the creditworthi- ness of an obligation range from Aaa ("triple A"), the highest grade, to C, the lowest. A description of the eleven rating categories is provided 6 Table l. LONG-TERM STATE AND LOCAL DEBT, 1944-1977 Percentage Increase Total State Local Debt Since Last Census Year Debt Debt Total Full Faith in Total in Local and Credit Debt Debt 1944 $ 17,323 $ 2,786 $ 14,537 $12,605 1952 28,720 6,640 22,080 18,480 65.8 51.9 1957 50,845 13,522 37,323 26,087 77.0 69.0 1962 77,543 21,612 55,931 38,008 52.5 49.9 1967 107,621 31,185 76,436 49,204 38.8 36.7 1972 158,781 50,542 108,239 70,585 47.5 41.6 1977 246,816 87,184 159,632 93,496 55.4 47.5 Source: U.S. Census of Governments, 1944-1977. in F1 inves disti perch two de there dium g critic: credit which u categor 58: Mesa“ bonds an Edge" ci the mUni '0 rEfer the term, to Cities was model many of t gritty Cl" be USEd t} in Figure 1. These ratings may be divided into two broad categories: investment grade (Aaa to Baa) and subinvestment grade (Ba to C). The distinction is based on the grades which commercial banks are legally permitted to buy. Almost all central cities in the U.S. during the past two decades have fallen into the investment grade category. Until 1968, there were only four investment quality ratings but in that year the me- dium grade category was expanded to four ratings, largely in reSponse to criticism which resulted from the Congressional hearings on municipal credit in 1967-68 (U.S. Congress, 1967-1968). The two additional ratings which were added to distinguish the best credit risks in the A and Baa categories were called A-l ("A one") and Baa-1 ("B double A one"). Securities which are rated Aaa are termed gilt edge investments by investment analysts. Cities which have issued full faith and credit bonds and which have gained a Aaa rating may therefore be termed "gilt edge" cities since the credit rating is based on the creditworthiness of the municipality. Although there is no comparable term popularly used to refer to securities rated at the lowest end of the investment range, the term "grit edge" has been coined and adopted in this study to refer to cities with Baa (not Baa-1) ratings and below. The term grit edge was modeled after Proctor and Matuszeski's Gritty Cities (1978) since so many of the lowest rated cities seem to exhibit the characteristics of gritty cities as described by the authors. Gilt edge and grit edge will be used throughout this treatise to refer to the extremes of the bond rating continuum and to the cities which hold those ratings. What Do Municipal Credit Ratings Measure? Credit ratings are designed to measure the creditworthiness of a jpaxticular bond issue, which means, in the case of general obligation .Hmacmz ucmEcum>ou cam Hmmwuwssz m.%vooz "mouaomv .mwcfiumm ufimouo wcom Hmafioacsz m.>vooz Aufia> .a .aema .H ouawwh wawvcmum ucoaumo>cfi Hmmu wcacfimuum uo>o mo muooamoua noon hamamuuxm uazmmmv ca :mumo moonwmv swan m E“ m>HumH=ooam mo uasmwmv a“ on was monuma £03m mwcfivamum uoom moo ucoEumo>cw manmufimmv mo muwumfiumuomumno song vousmmm Hams uoc opsusm “mucoEmHm m>fiumanooam o>ms ou powwow mm mnzHumem moaumauouomumxo ucmEumo>CH wcfimcmumuso xomH “wafixoma on has mucmEon w>wuomuoua camuumo “monsoon hauooa no: monomuoua hanmfin umsufimz mum Human LOWER ouausu onu ca mEauoEom uaoEuHmaEa Ow >uHHanHuaoomam m umowwnm nofina ucommua on has mucmEmHo “mousnfiuuum acmEumo>ce manmuo>mm xams mmommom UPPER mn=« mo omuwov umoHHmam “muaamsv ummm mm< mnzH bor deb com the Gen: prer ager ized tury Sue. trihu those the 0 EGHer Plana Serti and s ray 0 istlc assig emplo hell, that ; bonds, the creditworthiness of the particular municipality backing the debt with its full faith, credit, and taxing power. Nevertheless, "no complaint is more frequently voiced than the lack of clarity about what the ratings actually measure," according to the report of the Twentieth Century Fund Task Force on Municipal Bond Credit Ratings (1974, 3). The premier recommendation of the Task Force, in fact, was that the rating agencies should more explicitly define the considerations which are util- ized in measuring creditworthiness and assigning ratings (Twentieth Cen- tury Fund Task Force, 1974, 6—8). In theory, credit ratings are supposed to measure credit risk, i.e., the risk that a municipality will not be able to meet its payments on in- terest and principal when they fall due over the course of the bond is- sue. What an evaluation of this risk depends on, however, has been at— tributed to different factors by different authors. At one extreme are those who see a rating only as an indication of financial well-being. At the other are those who see bond ratings as more holistic indicators of general financial, economic, political, and social well-being. Some ex- planations are almost overly simplistic as exemplified by Robinson's as- sertion that "bond ratings are based on financial information" (Aronson and Schwartz, 1975, 235). Other explanations acknowledge a broader ar— ray of factors but still focus only on financial and economic character- istics. An introductory banking textbook, for instance, attributes the assignment of credit ratings to the diversity of industry, stability of employment, and debt load that has been incurred by a community (Camp- bell, 1978, 111). Likewise, White's literature survey states simply that rating services look at a government's assets, debts, and financial practices (White, 1977, 2). Even Sullivan's superb report on bond rating relati debt 8 As Cha predic tively Mt long ac social, things" 19605 t' a more ; Ships, a (1970, 1 been a r judgment foremOSt believes 10 ratings as grant/tax mechanisms views them as a direct function of the relationship between debt outstanding and an estimation of the amount of debt which the municipality can repay on schedule (Sullivan, 1976, 3). As Chapter 3 will demonstrate, however, studies which have sought to predict credit ratings using only financial correlates have been rela- tively unsuccessful. Moody's, as verbalized by a vice-president of the corporation, has long acknowledged that "in the appraisal of long term risks, economic, social, and political trends and tendencies are considered, among other things" (Ellinwood, 1957, 75). Packer (1968, 95) noted during the late 19603 that "the focus has shifted away from preoccupation with default to a more inclusive analysis of economic trends, intergovernmental relation- ships, and other factors that affect an issue's market standing." Moak (1970, 163) augments the previous statement by noting that "there has been a new awakening to the importance of key social factors in reaching judgments as to community credit." More recently, one of the country's foremost experts on credit ratings has summarized the factors which he believes influence the ratings as follows: In the case of general obligation bonds, prime importance is attached to various measures of debt to wealth, popu- lation, and governmental revenues. The economic base of a community, the stage of its development, its sociologi- cal character, and the quality of its government are also leading factors. Last, the analysts examine the exact nature and strength of the legal obligation that the bonds represent. (Petersen, 1974, 83) This diversity of explanations underscores the need to further investi- gate exactly what the ratings measure and the extent to which they may be used as indices and forecasts of municipal well-being in more than finan- cial terms. In schematic form, some general categories of factors which infil char; infll teris which howeve graphi Correl the be the 00 faCtOr uUltiv 11 influence the rating process are diagrammed in Figure 2. The financial characteristics of the community are shown to be of major significance in influencing the rating process. It is the financial and economic charac— teristics which are generally requested by Moody's in the application which must be filed in order to be assigned a rating and which are most prominently displayed in the prospectus (or official statement) which a city publishes in readying itself to issue bonds. Futhermore, while there is considerable disagreement about the components of the full spec- trum of factors that are considered by Moody's, virtually everyone ac- knowledges that financial characteristics are the most important in as— sessing creditworthiness. "Information pertaining directly to debt and debt burden, the traditional measures, and to financial Operating per- formance tend to be the most important items to investors" (Petersen, 1979, 46). Acknowledging the logical importance of financial characteristics, however, is not equivalent to dismissing the importance of social, demo- graphic, and political characteristics. Such factors may be important correlates of the ratings either because they are directly examined by the bond analysts or because they influence the financial complexion of the community. In either case, the influence of such non-traditional factors on the ratings may make it possible to use credit ratings as multivariate indices of city well-being. Other factors which are shown in Figure 2 to influence credit rat- ings are the persuasiveness of the community vis-advis the rating agen- cies, the background and attitudes of the financial analysts, the his- tory of payment and default on municipal obligations, and managerial ex- pertise. Management factors listed by Moak (1970, 165-167) which 12 Persuasiveness of the Community vis—a-vis the Rating Agencies Background and Attitudes of the Financial Analysts M U N I C I P A L B O N D R A T I N G S Managerial and Administrative Efficiency History of Payment and Default on Debt Obligations Financial and Economic Characteristics of the Community Social Characteristics of the Community Political Characteristics of the Community Demographic Characteristics of the Community Figure 2. Direct and Indirect Determinants of Municipal Bond Credit Ratings. 01 f1 13 exercise an important influence on the evaluation of creditworthiness in- clude the overall government structure, the degree to which government is well-administered, organization for financial administration, excellence of budgetary practices, effective capital planning and programming, qual- ity of revenue administration, the revenue base, the revenue calendar, the reputation for prudent financial management, the condition of the physical plant, and contingent liabilities. Social scientists have devised elaborate multivariate indices to analyze the character and function of urban places, but few of these indices have had more than academic applicability. Examples of some very valuable indices devised by social scientists which have yet to be employed in a problem-solving capacity are the various quality of life indices, hardship indices, industrial diversity indices, and housing quality indices. On the other hand, the credit rating, which has been devised for the practical purpose of guiding private investment deci- sions has been largely ignored by social scientists. A3 Rabinowitz (1969, 136) has noted, ratings have commanded little research attention at least in part because "the combination of lack of data, lack of the— ory, and incredible variation among local governments has limited pro- fessional writings on the subject." Yet, the credit rating seems to provide a commentary on some of the issues which social scientists are addressing, such things as the financial plight of the central cities, the inequities of a private marketplace which discriminates against the :most socially distressed cities, the decline of the Frostbelt and the rise of the Sunbelt, and the American urban future. While credit rat- ings are meant to be a guide to the future financial well-being of cit- ies, it must be understood that the concept of financial well-being l4 rests on a broader social, economic, and political foundation than sim- ply debt ratios and tax delinquency rates. In short, while credit rat— ings are imprecisely defined, they may be thought of as indices of cred- itworthiness, financial well-being, and confidence in the American urban future. The dimensions of these concepts, however, remain to be more fully elaborated and theoretically justified. What is the Bond Market? The bond market is an over—the-counter exchange between the users of funds and the suppliers of funds. Since the users of funds may be either private or public institutions, bonds issued by municipalities and other public authorities must directly compete with bonds issued by corpora- tions for the fixed amount of investment capital which is available at any one time. Whereas the yields of corporate bonds are higher than those of municipals, the tax-exempt feature of municipals commends their purchase to individuals and institutions in higher income tax brackets. Municipals are exempt from federal income taxes and from state taxes if held within the state where they were issued. The suppliers of funds on the municipal market are essentially limited to only three groups: com- mercial banks, insurance companies, and wealthy individuals and personal trust funds. Commercial banks, in fact, hold over half the long-term municipal debt in the United States (Dougall, 1973, 167). There are essentially two major types of municipals: (1) general obligation bonds, and (2) limited liability bonds, of which revenue ‘bonds are the most important type. General obligations are backed by the full faith, credit, and taxing power of the municipality, while re- venue bonds are backed only by the revenue generated by the facilities :financed by the bond issue. Prior to World war II most local bonds were us sl US ti ma by er. b0: neg hOL rat QUE Und thi art Fina lopE tere FEaT 15 sold as general obligations but since, sales of revenue bonds have risen to nearly fifty percent of total municipal sales (Forbes and Petersen, 1976, 45). Bonds of either type are usually the preferred means of funding major capital improvements at the local level in order to assure user-benefit equity. As Steiss (1975, 5) puts it: "a public facility should be financed so that the burden does not fall to one generation of users but is spread over the life of the facility." Only general obligation bonds are under consideration in this re- search effort since their ratings alone may be interpreted as evalua- tions of the creditworthiness of the municipalities issuing them. No matter how many separate general obligation bond issues have been floated by a single municipality, the rating assigned to the municipality's gen— eral obligations remains the same for all issues. General obligation bonds are generally sold by a competitive bidding process (as opposed to negotiated sale). Bids are submitted to the municipality by investment houses, commercial banks, and syndicates; the bid which offers the lowest rate of interest is accepted. The bidding process is strongly influ- enced by the ratings assigned to an issue. Bond issues with lower ratings understandably attract fewer bidders when the issue is put up for sale; this results in higher interest costs. Bond issues with higher ratings attract more bidders with the result that interest rates are lower. To illustrate the importance of credit ratings, New York City's Director of Finance testified before a Congressional Committee that when Moody's lowered New York's rating from A to Baa in the mid-19603, the extra in- terest cost was $2.5 million per year per issue or about $20 million per year total extra cost (U.S. Congress, 1967-68, 20-21). The Twentieth Century Fund Task Force (1974, 2) confirmed New York's experience as a gene tior volv inve fina agai with with 16 general principle by noting that each rating step down involves an addi- tional 0.1 to 0.3 percent in interest costs. ' or purchased, by a bond dealer When new issues are "underwritten,' the primary market has fulfilled its function. The dealer may then re- offer the bonds for sale on the secondary, or trading, market which in- volves a much larger number of transactions. The secondary market allows investors to convert bonds into cash before the bonds have reached their final date of maturity, a time which may be several decades away. Here again, ratings influence the demand for a particular bond issue. Bonds with higher ratings are more easily and profitably liquidated than bonds with lower ratings. It should be pointed out, however, that Moody's stresses the fact that ratings are not recommendations to buy or sell (Moody's, 1978, v). The attractiveness of a bond to a potential buyer may depend on other factors that are not involved in the ratings such as yield, date of maturity, or tax considerations. Nevertheless, bond rat- ings do have a demonstrable effect on the operation of the secondary as well as the primary market and therefore on the flow of funds between the private and public sectors of the American economy. Statement of the Problem Determining what municipal credit ratings reveal about central cities in the United States and how they are patterned in space are the basic problems under investigation in this analysis. The primary objective of this study is to examine the interrelationships between bond ratings as rassigned to central cities in the United States by Moody's Investors Service, and the regional demographic, social, and geopolitical charac- teristics of these cities and their metropolitan areas. Specifically, the purposes of this research are fourfold: 17 (1) To define and interpret the regional patterning of central city bond ratings as it has varied over the past twenty years; (2) To analyze the relationships between central city bond rat- ings and selected demographic, geopolitical, and social variables; (3) To develop a model expressing the probability that a cen- tral city will have a particular bond rating given knowledge about its demographic, ge0politica1, and social structure; (4) To determine what factors best discriminate among central cities in the various rating categories. Significance of the Study The financial troubles experienced by New York City in 1975, coupled with the increasing popularity of municipal bonds, has led to a rapid ex- pansion of municipal bond research on individual issues and municipalities (Madrick, 1977, 81). At a larger scale, "much of the research in munici— pal debt has involved studies of how the market might be changed to im- prove the efficiency of tax-exemption as an implicit subsidy to state and local borrowers” (Petersen, 1979, 46-47). In general, research on munici- pal bonds has been relatively restricted in scope to individual municipal- ities and some of the injustices and inequities of the bond market. This study broadens the base of the municipal bond literature by examining bond ratings from a social time-space perspective. It will contribute to four areas of contemporary concern to the geographic and social science community: (1) urban financial geography, (2) the study of money flows and allocations, (3) the geography of the urban future, and (4) the re- gional analysis of the United States. 18 Urban Financial Geography "The problems facing the bond analysts suggest how little is truly known by state officials, credit analysts, budget—makers, politicians, political scientists, economists, and planners about our complex urban- izing environment and the conditions under which local government can best be operated" (Rabinowitz, 1969, 77). By virtue of its intent to explain the spatial and temporal variations in assigned credit ratings, this study may be seen as a contribution to the nascent field of urban financial geography which may be defined as the study of the spatial pat— terns of revenue generation and allocation and their impacts on the urban environment and the American metropolitan system. Geographers have pio- neered in the effort to understand the economic, social, and physical structure of cities and metropolitan areas but have never explored the linkages between these structural characteristics and financial well- being. Just as the socioeconomic and physical make-up of a city affect the way in which a city raises and spends public revenues, so the various methods of revenue generation and allocation have an impact on the geo- graphic environment, including both people and places. The preperty tax, for instance, has had a dramatic effect on the condition of city neigh- borhoods; whatever comes to replace the property tax as we know it will also have a demonstrable effect of the use of land and the well-being of people. Similarly, the user fees levied on public services have a direct impact on place utilities, that is, the use which city residents make of public facilities. Likewise, the social geography of the city greatly influences the spending of public funds and, in so doing, often confers ‘benefits on one part of the city at the expense of another. Many of the financial regulations, methods of taxation, and the intracity allocation 19 of funds decided upon by local governmental units have discernable spatial impacts which have yet to be studied by geographers. These and other top- ics comprise the emerging field of urban financial geography which should command more attention in the fiscally conservative decade ahead. Because patterns of city expenditures and fiscal problems are closely related to the underlying character of the city and its inhabitants, re- search in urban finance may find a comfortable niche in geography and the other social and behavioral sciences. Hirsch, for instance, outlines three reasons for the increasing difficulty which central cities are hav- ing in financing urban public services: (1) Central cities have been havens for the poor and disadvantaged minorities; (2) central cities are characterized by aging physical structures, congestion, outmigration of high income groups and industry, and diseconomies of scale; (3) central cities are the victim of metr0politan governmental fragmentation which has produced major spillovers of social costs and benefits (Hirsch, 1971a, 5). It is obvious, therefore, that fiscal distress (or well-being) is clearly associated with the social and political characteristics of governmental units and that relationships between city finance and the total urban environment need to be more fully examined. Money Flows and Allocations One of the major subfields of urban financial geography is the study of money flows and allocations at both the intraurban and interurban scales. While the present research undertaking is primarily an examina— tion of the "point pattern" of bond ratings, the next step in the research Process should be an examination of the "flow pattern" of investment capital as it is affected by the ratings. The flow of capital from pri- vate institutions and wealthy individauls into public coffers is, in part, fl no Co: 20 determined by assigned credit ratings. Even more important, however, is the reverse flow of money paid out of the public purse for debt service. Since credit ratings "exert substantial influence on the cost of capital to state and local governments" (Twentieth Century Fund Task Force, 1974, 2), they discriminate against some cities while benefiting others. Just as the equitable allocation of federal funds to America's urban areas needs to be undertaken with great care, public policy should be consider— ed to make more equitable the flow of funds between the public and private sectors of the American economy as regulated by the bond market. In re- lated areas of financial geography, geographers have investigated money flows and allocations with reference to federal housing assistance and mortgage lending practices (Harvey and Chatterjee, 1974; Boddy, 1976; Cox, 1978; Darden, 1977, 1980; and Dingemans, 1979). Geography of the Urban Future Credit ratings are acknowledged to be judgments about the future. When they are cartographically displayed, the distribution of cities in various rating classes may be seen as maps of confidence in the American urban future as perceived by investors and credit analysts. Credit rat- ings therefore comprise predictions about the future of American cities. Because the ratings serve to direct the flow of private funds, however, their function as an index of future well-being may turn out to be a self-fulfilling prophesy. Nevertheless, they provide one of the few, perhaps only, operational indices of the urban future as viewed from the arena of the national capital market, a group of borrowers and lenders ‘wdth considerable vested interest in the future of American cities. Regional Analysis Credit ratings may be used as a criterion for regional analysis and regionalization at the national, state, and metropolitan scales. Their 21 variability over time and over space should parallel many of the trends taking place in the United States with respect to the economic, social, and political complexion of the country. For those interested in defin- ing the distinctive regional character of the United States, therefore, credit ratings may be another component in assessing unique regional iden- tities. This study comprises a maiden attempt at moving in this direc- tion at the national scale. In the chapters which follow, the regional patterning and spatial correlates of municipal bond ratings over the past twenty years will be the focus of attention. Chapter 2 presents the working hypotheses around which this research is centered; it also delineates the study area, sets forth the variables used, and describes the methods of analysis, both graphic and statistical. Chapter 3 is a review of the literature and Chapter 4 presents a time-space analysis of central city credit ratings. Chapter 5 examines the spatial correlates of assigned credit ratings and Chapter 6 presents the results of two multivariate models relating credit- worthiness to the demographic, social, and geopolitical characteristics of central cities. The final chapter comprises the summary and conclu- sions and offers recommendations for further research. 1 tr . re Fi Vh 5t] 8111' CHAPTER II RESEARCH METHODOLOGIES In light of the problem discussed in Chapter 1, the working hypo- theses under investigation, the study area of the research, the variables and data sources, and the methods of analysis are put forth in the pre- sent chapter. Hypotheses Two broad working hypotheses are offered to guide the course of this research. These hypotheses are, of necessity, broad for several reasons. First, there are few macrospatial financial analyses of urban systems on which more specific hypotheses may be based. Second, the absence of a unifying theory makes it impossible to use a purely deductive approach. Third, the employment of more specific hypotheses would needlessly re- strict the range of variables to be employed. The hypotheses which have guided the course of this research follow: (1) Bond ratings of central cities exhibit meaningful spatial and temporal variability; their geographic distribution, as it has changed over time, provides a map of confidence in the American urban future as perceived by investors and parallels the decline of the American manufacturing belt, particularly the industrial Northeast, and the rise of the 22 23 peripheral amenity belts. (2) Social, ge0political, and demographic characteristics of central cities in the United States are correlated with assigned credit ratings whether or not they have been taken into consideration by the rating agencies. StudyiArea The central cities of Standard MetrOpolitan Statistical Areas (SMSAs) in the United States, as recognized by the U.S. Bureau of the Census, serve as the subjects of this investigation, Central cities are those ap- pearing in the official SMSA titles. These are the same cities which serve to anchor the American metropolitan system, give focus to the sys- tem's component metropolises, and generate many of the impulses whose impacts are felt nationwide. Of the more than 80,000 units of local gov- ernment in the United States, central cities comprise one of the most conspicuous and important sets of municipalities at the national scale. As of January 1980, there were 382 officially designated central cities (excluding four in Puerto Rico) in the United States. Out of this number, 343 were assigned credit ratings by Moody's Investors Service. In 1960, only 306 of the 382 cities were rated, indicating that over the course of the last two decades an increasing number of cities have felt the need to solicit a credit rating in order to market their municipal obligations. During the same two-decade period, however, 28 of the 382 officially designated central cities had never been assigned credit rat- ings by Moody's.1 These unrated cities were predominantly in the South; 1Fort Smith, Pine Bluff, Springdale, and Texarkana, Arkansas; Garden Grove, Lompoc, Seaside, and Simi Valley, California; Washington, D.C.; Lakeland, Panama City, Pensacola, and Winter Haven, Florida; Kankakee and Rantoul, Illinois; Bloomington and West Lafayette, Indiana; H0pkinsville, 24 they comprised many of the smaller central cities in the U.S. Their elimination reduces the number of data points used in this analysis to 354 cities. For any given year, however, additional cities may have gone unrated. In 1960, for instance, there were a total of seventy-six unrat- ed cities and in 1980, the number had been reduced to forty. The fact that some cities do not have credit ratings assigned by Moody's may be accounted for by any one of the following reasons: (1) The city may never have needed to borrow funds on a long-term basis; (2) the city may have depended on local institutions or the regional capital market to supply it with long-term credit; (3) the amount of money borrowed on the national capital market may have been so little that a rating was not assigned; (4) the city may have requested a rating but may not have sup- plied the investment service with all of the information needed to assign a credit rating; (5) the city may not have solicited a credit rating, preferring to sell unrated bonds on the local capital market; or (6) the city may have decided that Standard and Poor's services were more desira- ble than Mbody's. The study area, therefore, comprises a nationwide set of all central cities, as defined by the U.S. Department of Commerce in 1980, which have carried a municipal credit rating by Moody's sometime during the preceding score of years. The regional distribution of these 354 cities is presented in Table 2. Only two states, Vermont and Wyoming, and the District of Columbia are unrepresented in the study area. Otherwise, the location of the cities ranges nationwide. In terms of population size, 1(Cont'd) Lexington, and Owensboro, Kentucky; Baton Rouge and laafayette, Louisiana; Muskegon Heights and Norton Shores, Michigan; Moss ‘Point, Mississippi; San Benito, Texas; and Weirton, West Virginia. REGIONAL DISTRIBUTION OF CENTRAL CITIES 25 Table 2. Census Region and Number of Census Division Central Cities in Each Region Relative Frequency (in percent) Iflortheast 78 New England Middle Atlantic North Central 94 East North Central West North Central South 124 South Atlantic East South Central West South Central west 58 Mountain Pacific TOTAL 35 43 66 28 57 21 46 18 40 354 22.0 26.6 35.0 16.4 9.9 12.1 18.6 7.9 16.1 5.9 13.0 5.1 11.3 100.0 Igource: Compiled by author. 26 as presented in Table 3, the central cities used in this analysis ranged from 13,400 to 7,838,000 in population as of 1976. Despite the pOpular conception of a central city as one with 50,000 or more inhabitants, more than a quarter of the 354 cities had fewer than 50,000 in 1976. In gen- eral, these smaller central cities are companions of larger cities which actually qualified their areas for metropolitan status, components or twin-city SMSAs, or the primary central cities of SMSAs designated under the revised p0pulation criteria of 1974. Variables and Data Sources Municipal credit ratings, as they have been assigned to central cit- ies in the United States by Moody's Investors Service, serve as the de- pendent variables in the regression and discriminant analyses to be em— ployed. Independent variables have been selected to portray the location, age, demographic, geopolitical, social, and financial character of cities and their metropolitan areas. The social variables were chosen to por- tray the well—being of people and their living environment. As such, many of the social variables do overlap economic variables, particularly measures of well-being such as per capita income and unemployment rate. The number of variables employed could have been greater or fewer had additional data been readily available. All of the variables used in this investigation are presented in Table 4 along with the sources from which they were drawn. M Ratings The most convenient and timely source of general obligation bond ra tings is the monthly publication Moody's Bond Record. The January isSue of each volume is designated "Year End Edition" and it is from this number that bond ratings have been taken. This publication first I I 1976 F 13, 25, so, 100 250, 500 1.000, r~srl ”Cute! 27 Table 3. DISTRIBUTION OF CENTRAL CITIES BY POPULATION SIZE, 1976 Number of Central Relative 3L9976 POpulation Size Cities in Each Frequency Size Class (in percent) .13,400 to 25,000 11 3.1 25,100 to 50,000 86 24.3 50,100 to 100,000 121 34,2 100,100 to 250,000 81 22.9 250,100 to 500,000 31 8.8 500,100 to 1,000,000 18 5.1 1, 000,100 to 7,838,000 6 1.7 TOTAL 354 100.0 Source: U.S. Bureau of the Census, "Population Estimates and Projec- tions," Series P-25, nos. 740-789, 1979. 28 Table 4. SOURCES OF DATA FOR CENTRAL CITIES AND METROPOLITAN AREAS ——1 Variable Year Data Source BOND RATINGS (CL) Moody's General Obliga— 1960 Moody's Bond Record tion Bond Ratings through Vol. 27 (1960) - Vol. 47 (1980) 1980 "Year-end Edition," January DEI’IOGRAPHIC VARIABLES (2) City Population 1960 The World Almanac 1980 (3) City Population 1970 Same as (2) (4) City Population 1976 U.S. Bureau of the Census, Population Estimates and Projections, Series P-25, nos. 740-789 (5) SMSA Population 1970 U.S. Bureau of the Census, Population Estimates and Projections, Series P-25, no. 739 ( 6) SMSA Population 1976 Same as (5) C 7) City Population Growth 1960-1970 Derived from (2) and (3) C 8) City Population Growth 1970-1976 Derived from (3) and (4) C 9 ) City Population Growth 1960-1976 Derived from (2) and (4) (10) SMSA Population Growth 1970-1976 Derived from (5) and (6) GEOPOLITICAL VARIABLES (ll ) Percent of Urbanized Area 1970 U.S. Bureau of the Census, IPopulation in Central City Population of Urbanized Areas Established Since the 1970 Census, Table 2 (12 ) Percent of SMSA Population 1970 Derived from (3) and (5) in.the Central City ( l3) Derived from (4) and (6) IPercent of SMSA Population 1976 in the Central City 29 Table 4 (Cont'd). Variable Year Data Source (14) Territory Annexed (more 1970—1977 U.S. Bureau of the Census, than 20 square kilometers) Boundary and Annexation Rank Survey 1970-1977, Table 4 (15) Geopolitical Fragmentation 1970—1972 Derived from (3), (S), and the Index number of local governments Rank per 100,000 population in the SMSA as reported in U.S. Bureau of the Census, Census of Governments 1972 SOCIAL VARIABLES (16) Percent of the Population 1960 County and City Data Book 1962 Foreign Born (17) Percent of the Population 1970 County and City Data Book 1972 of Foreign Stock (18) Percent of Population 1960 Same as (16) Nonwhite (19) Percent of Population 1960 County and City Data Book 1967 Black (20) Percent of Population 1970 County and City Data Book 1977 Black (21) Difference in Percent of 1960-1970 Derived from (19) and (20) Population Black (22) Median School Years Com— 1960 Same as (16) pleted (population over 25 years of age) (23) Median School Years Com- 1970 Same as (17) pleted (population over 25 years of age) (24) Difference in Median 1960-1970 Derived from (22) and (23) School Years Completed (25) Percent of Civilian Labor 1960 Same as (16) Force Unemployed (26) Percent of Civilian Labor 1970 Same as (17) Force Unemployed 30 Table 4 (Cont'd). W ‘fariable Year Data Source (:27) Difference in Civilian 1960-1970 Derived from (25) and (26) Unemployment Rate (28) Percent of Civilian Labor 1970 Same as (17) Force Professional and Managerial (29) Per Capita Income 1959 Same as (19) (30) Per Capita Income 1969 Same as (17) (31) Per Capita Income 1974 Municipal Year Book 1977 (32) Per Capita Income Growth 1959-1969 Derived from (29) and (30) (1313) Per Capita Income Growth 1969-1974 Derived from (30) and (31) (34) Per Capita Income Growth 1959-1974 Derived from (29) and (31) (35) Percent of A11 Families 1969 Same as (17) Below Low Income Level (36) Percent of Occupied Hous- 1960 Same as (19) ing Units Owner Occupied (37) Percent of Occupied Hous- 1970 Same as (20) ing Units Owner Occupied (38) Difference in Occupied 1960-1970 Derived from (36) and (37) Housing Units Owner Occupied (39) Percent Change in 1960-1970 Same as (17) Housing Stock (40) Per Capita Needs Index 1970 U.S. Department of H.U.D., Rank An Evaluation of the Community Development Block Grant Formula by Harold Bunce (:‘331-) Liu's Metropolitan 1970 Midwest Research Institute, Quality of Life Index Quality_of Life Indicators in the U.S. Metropolitan Areas, 1970, 1975 31 Table 4 (Cont'd). \Iariable Year Data Source (42) Zeigler's MetrOpolitan 1970 Unpublished M.A. Thesis, Quality of Life Index "Selected Quality of Life In- dicators and Demographic Char- acteristics of SMSAs in the U.S.," 1976 (143) Liu's State Quality of 1970 Midwest Research Institute, Life Index The Quality of Life in the United States 1970, 1973 (44) Smith's State Quality of 1970 The Geography of Social Well— Life Index Being in the U.S., 1973 (45) Wilson's State Quality of 1970 Midwest Research Institute, Life Index Quality of Life in the United States, 1974 (46) Percent of Housing Units 1970 Same as (17) in Structures Built Prior to 1950 IPJEDJANCIAL VARIABLES (47) General Revenue 1975-1976 Municipal Year Book 1977 (48) Revenue Per Capita 1975-1976 Derived from (4) and (47) (49) Percent of General Revenue 1975-1976 Same as (47) from Own Sources (50) Cross Outstanding Debt 1975-1976 Same as (47) (51) Outstanding Debt Per 1975-1976 Derived from (4) and (50) Capita C52) Percent of Outstanding 1975-1976 Same as (47) Debt Non-guaranteed (53) Ratio of General Revenue 1975-1976 Derived from (48) and (50) to Outstanding Debt \ Sout'ce: Compiled by author. 32 appeared in 1931 as Moody's Bond Ratings and in 1936 changed its name to goody's Bond Record. Bond ratings have also been reported in the annual Moody's Municipal and Government Manual since 1918. Whereas Moody's Bond Record reports only bond ratings, the Municipal and Government Manual pre— sents a full assessment of bonded debt, tax effort, and details of parti- cular bond issues for communities on the bond market. In addition to towns and cities, ratings are reported for states, townships, counties, salaec1al districts, and other public authorities. The variety of small geographic areas for which ratings are available make possible intra- metropolitan studies of financial well-being as well as broad national studies. A problem with the isolation of credit ratings as dependent variables for a given year is that there is no ready way of ascertaining when the rating for a particular city was last reviewed. A set of city ratings may therefore represent evaluations from any number of years in the past rather than an assessment from the perspective of a single point in time. Since the mid-19703, however, Moody's has made it a point to maintain an active data file on the cities it evaluates in order to make sure bond ratings accurately reflect the status and performance of the c ornmunity . The only other widely-used rating of municipal creditworthiness is Standard and Poor's whose credit ratings are reported monthly in The Mcipal Bond Selector. It is common practice for municipalities to apply for a rating from each agency. Because Moody's has rated more than twice as many central cities as has Standard and Poor's, Moody's cIredit ratings provide a more substantial data set for geographic anal- ysis at the national scale. When Moody's rating differs from Standard and Poor's, a community is said to have a "split rating," but a Con- gressional committee investigating public facility financing found that 33 70 percent of the ratings issued by the two rating services were identi- cal (U.S. Congress, 1967-1968, 3). Whether there are any systematic dif- ferences between communities with identical versus split ratings has yet t:o be investigated. _Izemographic and Geopolitical Variables Demographic variables have been employed to determine the impact of (:zity and metropolitan population size and growth on municipal credit rat- ings. Geopolitical variables which reflect the impact of political boun- ciziries on the structure of metropolitan areas were selected as a comple— ment to the demographic variables in order to determine how central city dominance and geOpolitical fragmentation affect assigned credit ratings. Proportion of the urbanized area population and the SMSA population re- siding in the central city have been used as a measure of central city dominance. City annexation activity has been used as a measure of the ability of central cities to capitalize on growth on the periphery. Cities which annexed twenty or more square kilometers during the period 1970-1977 numbered sixty-five. The Geopolitical Fragmentation Index (GFI) is a single value calculated by dividing the number of local governments Per 100,000 population in the SMSA by the proportion of the SMSA popula— tion living in the central cities. A GFI was calculated for each SMSA reported in the U.S. Census of Governments 1972. Annexation activity and the GFI were used as ordinal measures of the geopolitical character of a c OImnunity . Type of government (mayor council, council manager, commission, etc.) was considered as an additional political variable but a preliminary in- vestigation did not reveal any significant relationship between superla- tive or inferior bond ratings and type of government. Sanders (1979, 107), 34 in a study of all cities with more than 10,000 inhabitants, also found that there were no significant differences in ratings between munici- palities with mayor-council or council manager form of government. He (iid find that cities with a commission form of government (a distinct minority) were rated the lowest of all cities. _S_oc ial Variables There are two basic types of social variables which have been em- ployed in this research: unidimensional variables and multidimensional (or index) variables. The unidimensional variables describe selected (:IIaracteristics of the city population and their magnitude or rate of change over time. Basic characteristics refer to the percent of popula- tion black, nonwhite, or foreign born; median educational attainment; unemployment rate; per capita income; and percent of occupied housing units owner occupied. The multidimensional variables comprise three (llléility of life indices computed for states, two quality of life indices computed for metropolitan areas, and one per capita needs index computed for cities. All six will be discussed in detail in Chapter 5. Financial Variables Although the primary purpose of this research is to investigate the relationship between bond ratings and the demographic, social, and geo- POlitical character of central cities, several financial variables were Collected as a basis for comparison. Revenue per capita, outstanding debt per capita, and general revenue as a percentage of gross outstanding debt have traditionally been used, along with several other variables, as measures of fiscal capacity. These financial variables have figured in t0 the following analyses only as a basis for comparison in the multi- variate models derived in Chapter 6. 35 Methods of Analysis Both graphic and statistical methods have been used to describe and analyze the regional variation and spatial correlates of assigned credit ratings over the past twenty years, 1960-1980. Cartographic and Other Graphic Methods The regional patterning of central city bond ratings has been de— lineated by graphing the bond rating spectrum for each of the four Census regions and by mapping those central cities (1) in the highest and lowest bond rating categories for five~year intervals between 1960 and 1980, and (2) those which have moved up and down by more than one rating category during the 1960-1980 and the 1970-1980 periods. In addition, time graphs have been prepared to demonstrate changes in bond ratings over the 1960 to 1980 period. These graphs have been constructed for all central cities which have been rated Aaa at any time during the past twenty years and for all central cities which have had Baa or lower bond ratings at any time during the two-decade period. These temporal profiles of forty—four gilt edge and seventy-nine grit edge cities are analyzed by regional groupings in order to demonstrate the degree of regional homogeneity of credit rating histories. A time graph is also presented to depict the credit standing of the nation's ten largest cities, according to the 1976 population estimates, over the twenty-year span. Statistical Methods Four basic statistical methods are employed to elucidate the charac- teristics of financial well-being and to determine the relationship be- tween bond ratings and the underlying social, geopolitical, and demo- graphic characteristics of central cities: (1) Crosstabulation analysis 36 using the Chi-square test of statistical significance, (2) analysis of variance, (3) multiple regression analysis, and (4) multiple discriminant analysis. The computer programs provided by the Statistical Package for the Social Sciences (Nie et a1., 1975) were selected for use in conduct— ing the statistical analyses; they are identified in the source statements for tables presenting results of the analyses. Crosstabulation analysis has been used to test the relationship be- tween bond ratings and all discrete independent variables such as the quality of life, geopolitical fragmentation, and per capita needs indices, regional groupings of cities, and the rank in territory annexed to the central city. Results of these analyses indicate the extent to which the various independent variables predict or vary with bond ratings. Multiple regression and discriminant analysis have both been used to test the relationship between bond ratings and the previously discussed set of continuous independent random variables. The specific hypothesis tested by the multiple regression exercise is that municipal bond ratings depend, at least in part, on certain given characteristics of the issuing municipality. The regression model derived from the set of independent variables is then used to predict the bond ratings of central cities in the sample. The specific hypotheses tested by the discriminant analysis is that the probability distribution of the independent random variables is the same for each of the bond rating categories. The use of discrim- inant analysis makes it possible to identify the dimensions of variation which best discriminate among the various rating classes of central cit- ies. Discriminant functions have been identified for the nation as a whole and for each of the four Census regions to determine whether dif— ferent variables emerge as being important in classifying the financial 37 well-being of central cities in different parts of the country. Throughout this investigation a series of descriptive and inferen- tial statistical analyses are used. The cities under study, however, do not constitute a sample from a larger population to which conclusions may be generalized. A11 central cities with assigned credit ratings have been included in the study area and no effort has been made to predict the credit ratings of other cities in the United States. A justification for the use of inferential statistics in this study is therefore in order. For the purpose of this research, central cities do not represent a sam- ple of all cities, but the ratings assigned to central cities represent a sample of all possible ratings which the bond analysts at Moody's could have assigned to the central cities of the nation. The ultimate aim of the inferential statistical analyses, therefore, is to garner support for the underlying hypothesis that the assignment of credit ratings is not a random one but instead is influenced, whether directly or indirectly, by regional location, characteristics of social well-being, and the specific demographic and geopolitical character of the cities being evaluated. Both multiple regression and discriminant analysis have been used by other authors investigating bond ratings, both municipal and corporate. Horrigan (1966), Pogue and Soldofsky (1969), Horton (1970), West (1970 and 1973), Bahl (1971), and Rubinfeld (1971 and 1973) have used multiple regression analysis in their investigations. Carleton and Learner (1969), Finches and Mingo (1973), Rubinfeld (1973), Morton (1975-76), and llichel (1977) have all used discriminant analysis. Of those aforemen- tioned studies which have focused on municipal bond ratings rather than (Harporate bond ratings, explanatory variables have been almost exclu- sively financial and economic while the regional dimensions of credit 38 standing have been almost completely ignored. As stated above, this study departs from previous studies in that it attempts to link the con- cept of financial well-being to non-financial characteristics of munici— palities and to elucidate the regional patterning of financial well-being as measured by credit ratings and their correlates. CHAPTER III REVIEW OF THE LITERATURE Social scientists have almost completely ignored the subject of credit ratings and only a limited number of formal investigations have been done by the financial community. These studies have generally ig- nored the spatial patterning of credit ratings or have treated regional- ity as a peripheral component of multivariate models. Similarly, the search for correlates of municipal credit ratings has been hampered by considerable confusion over what the ratings actually measure and has focused primarily on financial and economic characteristics of communities. Summaries and Critiques of the Municipal Bond Rating System Background statements on the rating and marketing of municipal bonds may be found in more comprehensive works which put the subject of municipal credit within the framework of either local public financial administration or investment analysis. The public finance literature emphasizes the selling of bonds as debt instruments while the investment literature emphasizes the buying of bonds as securities. Steiss (1975), for instance, treats the subject of municipal bonds within the overall «context of local capital facilities planning and debt administration. Sinujar treatises exist as part of the municipal finance community's 39 40 trade literature: Aronson and Schwartz (1975), in a book published by the International City Management Association, and Moak (1970), in a book pub- lished by the Municipal Finance Officers Association, both offer a hand- book of financial practices, including debt administration, for local gov- ernments. Moak provides perhaps the most comprehensive list of factors considered by investors and rating agencies in determining preferences for and prices of municipal bonds. These factors are listed under four main headings: (1) amount and nature of the debt and debt services require- ments, (2) economy of the community and the region of which it is a part, (3) social factors, and (4) management of the local government (Moak, 1970, 157-171). Aronson and Schwartz (1975, 235), as cited previously, say only that "bond ratings are based on financial information." Authors which treat the subject of municipal bonds as investments within the broader context of investment analysis include Amling (1974), Campbell (1978), Christy and Clendenin (1978), Dougall (1973), Gup (1979), Mendelsohn and Robbins (1976), and Stevenson and Jennings (1976). These are supplemented by the investment community's trade literature as exem- plified by Davis (1958), Calvert (1965 and 1969), Drott (1971), Moody's Investors Service (1974), and Smith (1979). The volume edited by Calvert and published by the Investment Bankers Association provides probably the best technical summary of municipal credit and the bond market. Smith (1979), in a volume published by Moody's, provides another excellent and very comprehensive summary of the meaning of credit risk. Bond ratings in particular are also elaborated upon by Burke (1968), Ellinwood (1957), Greenberg (1977), Hoffland (1972), Kirk (1967), Matteson (1968), Packer (1968), Riehle (1968), Rose (1975), and Tyler (1962). Burke (1968, 171) attributes bond ratings to a sound financial program, an 41 ability to pay bills as they fall due, and the role of management. To the extent that ratings are based on the ability to pay, Ellinwood (1957, 75) attributes them to the extent of community income and/or reserve wealth, commercial and residential resources, and what he calls economic geography. Riehle (1968, 72), a Moody's vice president, cites only two major consider- ations, an economic base factor and a management factor, both of which he elaborates upon. During the 1970s a Task Force was commissioned by the Twentieth Cen- tury Fund to study municipal bond ratings; another was commissioned to study the municipal bond market (Twentieth Century Fund Task Force, 1974 and 1976). Their reports and the background papers which accompany them (Petersen, 1974; Forbes and Peterson, 1976) are liberally referenced to the municipal bond literature and provide a useful guide thereto. The report of the Bond Market Task Force provides a critique on the bond mar- ket as an efficient and equitable resource allocation mechanism. The most important recommendation of the Bond Market Task Force was that the market should be broadened to appeal to a wider variety of investors (not just commercial banks, insurance companies, and wealthy individuals) by creat- ing a voluntary taxable bond option. Secondly, the report recommended that in order to reduce the supply of bonds, tax-exempt financing be elim- inated as a means of funding industrial development, pollution control facilities, and the acquisition of facilities for private firms. The Bond Rating Task Force strongly criticized the current rating system because of the ambiguity over what assigned credit ratings measure. frhe study recommended that the factors and their relative weights be made public and that a National Data Bank be created to collect and disseminate reliable and timely information on local governmental units. Since the 42 Bond Rating Task Force report a greater burden of responsibility has been placed on municipalities to supply the rating services with accurate and timely information and to update that information on a regular basis. In addition, Madrick (1977, 81) reports that there has been a rapid expansion of municipal bond research on the part of brokerage firms and financial institutions. To date, however, no action has been taken in the estab- lishment of a National Data Bank and no definitive criteria have been issued by Moody's to define the requisites of a high credit rating. The two Twentieth Century Fund reports supplement two major contri- butions made to the municipal bond literature during the 19603 by Hempel (1967) and Rabinowitz (1969). Hempel focused on the problem of measuring municipal bond quality and in so doing provided a critique on the operation of the capital market and a history of municipal bond defaults. Rabinowitz treated many of the problems investors face as participants in the capital market. He also discussed the issue of objectively measuring bond quality and concluded that while the present bond rating system needs to be improv- ed it should not be standardized (Rabinowitz, 1969, 75). Studies on the Correlates of Municipal Bond Ratings Five studies have appeared in the financial research literature aimed at determining the correlates of Moody's municipal bond ratings. They have all been prompted by an interest in predicting assigned credit rat- ings using multivariate models. The studies by Carleton and Learner (1969), Horton (1970), and Michel (1977) all attempted to build a predic- tive model based on a set of independent variables, primarily financial amid economic, using either multiple regression or discriminant analysis. (harleton and Learner's independent variables, which are presented in Table 43 5 along with variables used in the other studies as well, were chosen on the basis of being commonly used and readily available; their model was able to correctly classify only 54 percent of a holdout sample into five rating categories (Carleton and Learner, 1969, 760). Morton (1976, 80) was able to correctly classify 58 percent of cities in his original sample into one of four Moody's bond rating categories using a discriminant model and primarily financial variables. Michel (1977, 595) chose independent variables which had been cited in the literature as the most important determinants of municipal bond ratings; his model was able to correctly classify only 58 percnet of a holdout sample. Michel concluded his analy- sis by asserting that "probably the most important reason for poor clas- sification accuracy is that the variables most frequently used to charac- terize risk are not reflective of economic reality" and that "classifica- prediction is relatively ineffective using traditional measures of munici- pal risk assessment" (Michel, 1977, 597). Horton, unlike either of the previous two studies, attempted only to distinguish between the charac— teristics of investment quality (Baa and above) and subinvestment quality municipals; his best regression model was able to classify only 80 percent of a holdout sample into the correct grade (Horton, 1970, 36). In another study, Rubinfeld (1973) sought to relate municipal credit ratings to municipal bond yields and in so doing to determine what commun- ity characteristics are important indicators of the ratings. His sample of 128 cities and towns is restricted to New England; it is the only one of the five studies listed in Table 5 which is geographically restricted :in scope. The author uses both regression analysis and multiple discrim- 1inant analysis to predict bond ratings. He did not work with a holdout salnple but instead applied his predictive models to the original sample ooaeeon ea An.e a my «and .nmdaooa atom 44 antenna NHH new An.apoozv oea aoaoowaaeo Howe voom m.uoom was pudendum was mamaamdm “.m.= osu ow mvooz oumoEoo was uofiomua ow ucmcfiEHuomao moaned Hmuucoo uofimz Iamum was m.%oooz .dOuuoz Aoasaosom one antenna neon panama .oaaas .eoowdaandz .awos HaoHuNGoH .nmaaoom uofioouo ou com: on and moeumu :u«3 mamaamcm 3oz uaooxov mowuwo ocom cowumwfiano nnma Hmfiocmoaw HHms So: moaeuouoo oe ucmcfiawuomfin cmofiuoa< umowuma on Handsou m.oooz .Hosofiz mwaaumu econ wdwuofioouo MOw Hoooa mammamdo manna «Hosea OmmH .mwcHumm m aoao>mo ou was moaofih uodmmm IEwuomao w coed ca wNH mocmawdm 3oz ocom aowumwwano mnma mwcfiumu coon we oaaeuouoo oa Immuwou maneuasz :fi mczou was mowufio Hmuocou m.>oooz .oHDHCwnsm mofiuow maaemw Hmcwwfiuo noumo odosumo>sfinom one uses mammamcm moo ca omH “.m.D mwceumm lumo>aw ousH meson woodman moans scammouwou can we wuowuumeo odom coaummwano omma lumomm you oocuoe m aoHo>oo oh oaofiuaaz was moaufiamawoacaz kuoamo m.%oooz .oouuom meson ooumuds ou oHnmo onEmm Hmcawwuo leaned on was mwcfiumu m.%oooz > m.omm «no :4 Hme “.m.a mwcfiumm mama oumowadoo vasos £0453 Eoumzm moan: mamaamcm mom :4 muoauumfio ocom dowumwfiaoo .uocumoq wcwuoom Hmofiumfiumum m a0H0>oo oe ucmcfiefiuumfio was mofiufiamafiofidoz Homosoo m.%oooz can couoHumo cowumwaumo>aH ouma w mooum onu mo o>Huoohoo hwoaooocumz mou< moaum mo uoofioam Amvuosua< mHmWAou Hmuou mo N no moo4> lumm Hmcomuoo mow monouwocmmxm woco>ou kuou mo N on once Emma uponm ammo wcaoamfiuo>o mEoocH mHHEmm amaooz oEousH muwdmo Mom mumum oEouca Hmuoa oEoocH mufiamo Hum ommmuucfi mEOUCH muwomo Hon N mmmouoaw oeooca N mum» coauomaaoo xmu odouuoo undo mo woo ocmawcm 3oz s4 cowumuoa souumas noon Jonas :4 owcmso N ooam> mono odomefiumo \mousuaocooxo udosuuo>ou mmxmu Hmuou «o N on mmxmu owuomaaooco msco>ou HmuOu mo N no mcoeumwwano ocom :oamcom oscm>wu Hmuou mo N no mucoshma oumeoz coeumoam> oommommm\uowo um: uoouwa muuomsnuwmmmz :4 sowumoog xufiowamo\uomm moumum Houumm moumum mouoom Hosomluouma oumu doauomHHoo any cowumaaaoa mo woo coeumaaaoo\uoon oeoocfi mafiamu pom mowufi Icaseoo ooucowuo umfiusoe oomm> owmmommm oooaw\xme msam> mauu\unmn mofi>umm undo «o N on was oumum cam Hmuooom oofi>uom undo mo N on moco>om m=Hm> oommommm no N no anon mufiamo use osco>mm moaned non undo wouooaaooda moxmu no N muuoaouo mo coaumsam> Hash undo mo woo coHumaoaoa mo woo cowumaoaoa\uooo cowum=Hm> commommm\unoo ooawooee Hooeom UDHG> fiwmmwmwm\u00n— aoawanaw Hooaom acuuoz Horne: namedaoam :Ouuom mocummq was commando moanmwum> zuoumcmamxm Anvwotua< Ap.odoov m manna 46 from which the models had been derived. He was able to correctly predict 67 percent of the ratings using the regression procedure and 68 percent using the discriminant function procedure. This success rate would un- doubtedly have been lower had the author tested it on an independently selected holdout sample. A study done by Morton and McLeavey (1978) departs from the five aforementioned studies in both method and objectives. Rather than trying to predict bond ratings, they propose an "objective and mathematically defensible" method of rating general obligation bonds. They derive groups of similar cities from a cluster analysis of 171 major cities in the United States using twenty-six independent variables which are theoretically re- lated to municipal bond quality. They suggest that cluster analysis be the first stage in a two-stage rating process which objectively assigns credit ratings on the basis of quantifiable variables unless non-quanti- fiable factors may be cited to justify other ratings. Parallel studies of corporate and industrial bond ratings have been done with the objective of being able to predict the assigned ratings by calling on a vector of accounting ratios. Horrigan (1966), Pogue and Soldofsky (1969), and West (1970) use regression techniques in order to determine the characteristics of a firm which determine the credit rating assigned by Moody's Finches and Mingo (1973) use factor analysis and dis- criminant analysis to accomplish the asme purpose. While the specific sub— ject matter of these studies on corporate bond ratings is peripheral to the present undertaking, their research demonstrates the appropriateness of using the same methodologies for studying both corporate and municipal bond ratings. As evidenced by the variables listed for each study in Table 5, the 47 most commonly ascribed correlates of municipal bond ratings are financial and economic characteristics, and to a lesser extent demographic charac- teristics of the issuing municipality. The rather disappointing results of these studies seem to indicate that traditional indicators of credit- worthiness may be outweighed by other factors in the assignment of credit ratings by Moody's. Such results should not be surprising given the sub- jectivity of the ratings acknowledged by Moody's as early as 1931 when they published the following in the first issue of Moody's Bond Ratings: The ratings on all bond issues, of every class, are based on a scientific formula [sicg] which, after exhaustive research, has been proven to be the most accurate guide for determining correct investment values for bond risks. . . . The so-called "statistical rating" is then put to the test with various non-statistical factors which af- fect the investment value of nearly all bonds to a more or less degree; the ultimate result of such test giving us the "final rating." (Moody's, 1931, 4) Today Moody's makes no pretense of using a "scientific formula" and maintains that the evaluation of credit risk is as much an art as a science. As Kirk (1967, 10) has put it: "The job of the rating analyst is to re- view all the factors--and those factors vary from municipality to munici- pality-to find those specific factors in the 'life' of the community which would over a period of years. . .result in these particular bonds being in a dangerous position." The correlates of municipal bond credit ratings do not and would not appear to be consistent from one city to another. Brunn and Zeigler (1979) examined a series of demographic, geopoliti- cal, and social characteristics of the highest rated and lowest rated cen- ‘tral cities in the United States and found strong relationships between time ratings and pOpulation size and growth rate, the proportion of the metropolitan area's population living in the central city, and a per capita 48 needs index which comprised a vector of social attributes of the cities. Another study which has explored the policy implications of bond ratings as they affect municipal borrowing cost, has also found an inverse cor- relation with city need: Sullivan (1976) discovered a tendency to assign lower bond ratings to the neediest communities and higher bond ratings to the wealthiest. In essence, concludes Sullivan (1976, 46), "bond ratings tend to penalize those least in a position to finance debt obligations." Supportive of further research on the social correlates of financial well- being is Sullivan's finding that of the ten municipal need proxies employ— ed in his analysis, the three social variables (per capita income, percen— tage of families below the poverty level, and employment rate) were the most powerful in proving that ratings discriminate against the neediest cities. His financial variables (per capita expenditures, tax rates, local tax effort, etc.) were far less effective. References to the Regional Patterning of Municipal Bond Ratings Interest in municipal bonds is usually limited to the bond yields of individual municipalities or to some of the imperfections of the bond mar- ket itself. Little has been written on the spatial patterning of bond ratings as an indication of the financial future of the American metrOpol- itan system. Only the pioneering study by Brunn and Zeigler (1979) has approached municipal bond ratings from a geographic perspective. Their research provided an analysis of the regional patterning of Moody's muni- Cipal bond ratings in 1979 and included maps of the gilt edge and grit edge Ilentral cities of the United States. The present investigation builds on time Brunn and Zeigler study by expanding the temporal dimensions of the inves duced to sup bond r econonj the fa: prefer: lens w} 8XCElle with at MQHdelg locatic Conside mUnicip all par tailor Siderat In Moody's ings an. Cities . and mos: quately to the E In his a HOteS th 49 investigation, by increasing the number of independent variables intro- duced into the analysis, and by utilizing a number of statistical methods to supplement the cartographic analysis of the data. There is ample evidence in the finance literature to indicate that bond ratings are a regionally patterned phenomenon. Moak (1970) lists the economy of the community and the region of which it is a part as one of the factors considered by investors and rating agencies in determining preferences. He goes on to acknowledge that there are regional image prob- lems which are likely to extend to the bonds of a local government with an excellent actual credit situation . . . because of 'guilt by association' with an overlapping unit" (Moak, 1970, 173-174). Investment analysts Mendelsohn and Robbins (1976, 529) also acknowledge that the community's location, particularly with respect to regional growth, is an important consideration in evaluating a municipal obligation. Since the volume of municipals in circulation on the market is so large and originates from all parts of the country, investors have ample opportunity to regionally tailor their purchasing habits based on both rational and irrational con- siderations. In a study of twenty-five of the largest cities in the United States Moody's Investors Service (1977) found some relationship between bond rat- ings and location but concluded that "while the young and fastest growing cities tend to be highly rated and lowest rated cities are among the oldest and most stagnant, neither geographic location or stage of development ade— quately explains assigned ratings" (Moody's, 1977, 1). Other references to the spatial patterning of municipal bond ratings are made by Sanders irl his analysis of U.S. cities with 10,000 or more population in which he nates that New England communities have the highest bond ratings with 56 parcel the l: Engla! Massa< other found predic Arkans Caroli group Connec and It Qualit {Y the Changi H Eeanin Cial c raphi the de 918% Dentin SeVera found is, an! allothe: regulal of the 50 percent rated Aa or better and that the solid South and mid-Atlantic have the lowest ratings (Sanders, 1977, 107). In Rubinfeld's study of New England towns and cities, he noted that of all the New England states, Massachusetts' communities received higher ratings than communities in other New England states (Rubinfeld, 1973, 24). Similarly, Horton (1970) found that the state in which a community was located was one of the best predictors of bond quality. States in Horton's poorer group were Alabama, Arkansas, Florida, Georgia, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee, Minnesota, and New Jersey. States in the better group were Maine, New Hampshire, Vermont, Massachusetts, Rhode Island, Connecticut, New York, Pennsylvania, Wisconsin, Michigan, Ohio, Illinois, and Indiana. Horton was only trying to discriminate between investment quality and subinvestment quality ratings, however, and he did not speci- fy the basis of his state groupings. Nor did he attempt to chart the changing regional character of credit evaluation over a period of years. Why is it reasonable to suspect that credit ratings should exhibit meaningful patterns of geographic variation? Simply because the finan- cial characteristics of cities, as they are correlated with diverse demo- graphic, social, and political characteristics, vary considerably from the declining cities of the Northeast to the growth centers of Texas and elsewhere, especially over the past two decades. Peterson (1976), com- menting on the urban financial predicament, asserted that the future of several old, industrial cities was so precarious that no one could be found to buy their bonds. Since the urban financial predicament has been, is, and will be significantly different from one region of the country to another, one might suspect that bond ratings would also exhibit regional regularities. It is these regional regularities which serve as the focus of the chapter to follow. rat: a1 a fac1 Per: InUn: be 1 Pat; Cen 8013 The Ove 3115 and of Pro CHAPTER IV SPATIAL AND TEMPORAL DIMENSIONS OF MUNICIPAL CREDIT RATINGS: 1960-1980 One of the working hypotheses of this research endeavor is that bond ratings of central cities in the United States exhibit patterns of region- al and temporal variation which parallel the decline of the American manu- facturing belt, particularly the industrial Northeast, and the rise of the peripheral amenity belts. In this chapter, the regional patterning of municipal credit ratings as it has varied over time (1960 to 1980) will be explored at the scale of Census region and Census division; the 37th parallel as proposed by Sale (1975) has been used as a divide between Sun- belt and Frostbelt. In a general sense, the Northeast and North Central Census regions may be equated with the American manufacturing belt and the South and West regions may be equated with the peripheral amenity belts. The bond ratings assigned by Moody's to central cities in the United States over the past twenty years provide the raw material for this regional analysis.2 In the two chapters which follow, the city characteristics 2All central cities which have been rated at any time between 1960 311d 1980 are included in Appendix A which is a year-by-year enumeration (If bond ratings for all central cities in the study area. Appendix B pzrovides a table of all states which comprise the four regions and nine 51 52 which are related to the regional patterning of bond ratings are analyzed. Regional Patterns in 1980 Central city bond ratings as of January 1980 are aggregated by Census region in Table 6. Striking regional contrasts between rating categories are apparent from an examination of the relative percentages of central cities in each rating category. The modal rating for the Northeast in 1980 was Baa, with 23 percent of the seventy-seven rated central cities having been assigned to this rating category. In contrast, the modal rat- ing for the North Central and West was Aa, which accounted for approxi- mately 50 percent of all 147 rated central cities in both regions. In the South, the modal rating for 1980 was A-l with 35 percent of all 118 rated central cities in the region. At a more refined scale, all but two of the Census divisions of the United States exhibited the same modal category as the regions to which they belong. A breakdown by Census division is presented in Table 7. The region with the sharpest contrast between divisions is the Northeast where central cities in New England exhibit markedly better ratings than central cities in the Middle Atlantic division. The Aa category accounted for 29 percent of rated central cities in the New England states while the Baa category accounted for 33 percent of rated central cities in the Middle Atlantic states. These are the modal categories in each division. The only other Census division to deviate markedly from the region in which it is located was the East South Central where over one—third of all central 2(Cont'd) geographic divisions of the United States as defined by the U.S. Bureau of the Census. Many of the analyses to follow will utilize these statistical areas. Ce Re No ‘1 Ce So 53 Table 6. THE DISTRIBUTION OF CENTRAL CITY BOND RATINGS BY CENSUS REGION: 1980 Number of Central Cities by Rating Category Census (Percent of Rated Central Cities in a Region) Region Aaa Aa A-l A Baa-l Baa B Caa Northeast 2 16 l4 l7 9 18 l 0 (3) (21) (18) (22) (12) (23) (l) (0) North 13 45 20 9 2 l O 1 Central (14) (50) (22) (10) (2) (l) (O) (1) South 5 28 41 28 6 10 0 0 (4) (24) (35) (24) (5) (8) (0) (0) West 6 27 10 12 0 1 O O (11) (48) (18) (21) (0) (2) (0) (0) TOTAL 26 116 85 66 17 3O 1 1 Source: Moody's Bond Record, January 1980. 54 Table 7. THE DISTRIBUTION OF CENTRAL CITY BOND RATINGS BY CENSUS DIVISION: 1980 Number of Central Cities by Rating Category Census (Percent of Rated Central Cities in a Division) Division Aaa Aa A-l A Baa-l Baa B Caa New 2 10 8 8 3 4 O 0 England (6) (29) (23) (23) (9) (ll) (0) (0) Middle 0 6 6 9 6 l4 1 0 Atlantic (0) (l4) (14) (21) (14) (33) (2) (0) East North 6 33 14 8 l 1 O 1 Central (9) (58) (22) (13) (2) (2) (O) (2) West North 7 12 6 l l 0 0 0 Central (26) (44) (22) (4) (4) (0) (0) (0) South 2 15 23 8 1 4 O 0 Atlantic (4) (28) (43) (15) (2) (8) (0) (0) East South 0 5 4 7 2 2 0 0 Central (0) (25) (20) (35) (10) (10) (O) (0) West South 3 8 14 13 3 4 0 0 Central (7) (18) (31) (29) (7) (9) (O) (0) Mountain 1 10 3 3 0 1 0 0 (6) (56) (17) (17) (0) (6) (0) (0) Pacific 5 l7 7 9 0 0 O 0 (13) (45) (18) (24) (0) (O) (O) (0) TOTAL 26 116 85 66 17 30 l 1 Source: Moody's Bond Record, January 1980. 55 cities were rated A, making it the most common rating in the division. The Middle Atlantic and the East South Central divisions, in fact, are the only Census divisions to completely lack any Aaa rated central cities. The spatial distribution of bond ratings in 1980, as Tables 6 and 7 reveal, was not a random one but one that exhibited some distinct regional concentrations. If the percent of all central cities in a particular rat- ing category which fall in a single Census region may be used as an index of regional concentration, central cities at the extremes of the rating scale exhibited the most pronounced patterns of regional alignment. The regionality of bond ratings is presented in Table 8 which indicates what percentage of all central cities with a given bond rating are accounted for by a particular region. Whereas the North Central region accounted for only 27 percent of all rated central Cities in 1980, 50 percent of the Aaa-rated cities fell in that region. Only the grit edge cities (those with Baa ratings and below) were more regionally concentrated than the gilt edge cities in 1980. Almost 60 percent of the grit edge cities were located in the Northeast region even though the region contained only 23 percent of all rated cen- tral cities in the country. Similarly, though the association is not quite as pronounced, the Aa-rated cities were most heavily concentrated in the North Central region and the Baa-l rated cities were most heavily concentrated in the Northeast. The remaining two categories, A and A-1, were most heavily concentrated in the South but their distribution tends to more closely approximate the overall distribution of central cities in the entire study area. The location of the twenty-six gilt edge and thirty-two grit edge central cities in 1980 is displayed in Figure 3. The North Central region 56 Table 8. THE REGIONAL DISTRIBUTION OF CENTRAL CITIES IN EACH RATING CATEGORY: 1980 Number of Distribution by Census Region (in percent) Bond Rating Central Clties Northeast North South West Central Aaa 26 7.7 50.0 19.2 23.1 Aa 116 13.8 38.8 24.1 23.3 A-l 85 16.5 23.5 48.2 11.8 A 66 25.8 13.6 42.4 18.2 Baa-l 17 52.9 11.8 35.3 0.0 Baa, B, 32 59.4 6.3 31.3 3.1 & Caa TOTAL 342 22.5 26.6 34.5 16.4 Source: Moody's Bond Record, January 1980. 57 .omoa .noaoao Houoaoo owns oats poo owns oaao are .m ouawfim .53 38 =6 2.: 80 .oo 68 o .53 .25 :6 2.: 23 o . O uuurrlxlult w u. C o .7 ....... . x m {air .a .. O c o ’0' c It a u s. u 0004 . .. III on. ION!!!I\\ u u m .- \\ lilo-I \QV‘ !! s W n u .g a. o t- r ......... root-.. . - \....--------..... ...... e 7,, \|!!! a, . lllll l! . \o! \3 (N [or u m .. a!“ / (\k .. r. n .. .. x , \\\ A o I ,4 ............ a a N ,x a . IIIIIIII a .o 1...... a . .. a a. ..... t. . K o . m. . moo o . l. o . . o . .. - m 2.-.. f-) .. r- .. . o * !!!\ O a a. llllllllll A .. annlblo a I!!!!!! o O !!!!!!!!!! _ 00!.1! e . - o . .- . --.--.,. M. x o u m’!!!! .. K. II!!! o; I!!! N! !!!!!!!!!!!! L. !!!!!! !tr!!la. \m (x a. t. 3 a. a. a. .- ~l"-'| P .. / as .c ... 58 clearly stands out for its heavy concentration of Aaa ratings and for its near absence of grit edge ratings; only Detroit and Cleveland carried a Baa or lower rating in 1980. Within the North Central, the state of Iowa emerges as the focus of gilt edge ratings, accounting for four of the twenty-six, a total exceeding that of any other state. Outside the North Central, only Oregon, California, Utah, Texas, North Carolina, Connecticut, and Maine are represented by central cities with Aaa bond ratings. Those cities with Baa and lower ratings, on the other hand, are heavily concen- trated in the Northeast and include many of the gritty cities of Pennsyl- vania, New Jersey, and New York such as Johnstown, Scranton, Passaic, Buffalo, and Troy. All other central cities with grit edge ratings in 1980 were located in the Sunbelt. With respect to the regional cleavage between Sunbelt and Frostbelt in 1980, it can be seen that the most striking contrast in bond ratings was not between cities north and south of the 37th parallel. Instead, it was between the cities of the Northeast and the North Central regions of the country, that is, within the region designated in this analysis as the American manufacturing belt. Similarly, within the peripheral amenity belt there is a strong contrast between the South and the West. There consequently appears to be a stronger contrast between east and west with- in the manufacturing belt and within the peripheral amenity belt than be- tween Sunbelt and Frostbelt. Changing Patterns of Central City Credit Ratings, 1960 to 1980 The credit ratings assigned to municipalities on the bond market by MCody's Investors Service are not revised at regular intervals nor, as evidenced by the rating histories presented in Appendix A, are they 59 subject to frequent revisions. Nevertheless, during the two-decade period under study, only eighty-seven out of a total of 271 central cities which were rated for the entire period, had experienced no change in ratings from year to year during the period. During any given year, however, few- er than one in twenty cities had their credit ratings changed from the previous year. Only in 1968 when 18 percent of all central cities experi— enced a change in rating and during the 1972 to 1977 period when 6 to 12 percent per year experienced a change, did more than one in twenty cities change ratings in a single year. The large number of changes in 1968 probably resulted from the heavy criticism of the ratings which followed the downgrading of New York City's bonds in 1965 and which reached a peak with the Congressional hearings on municipal bond finance (U.S. Congress, 1967-68) in 1967 and 1968. The large number of changes during the mid- 705 may be eXplained by the many changes to new rating categories, A-1 and Baa—l, and perhaps by the impact of newly released 1970 Census data. Moody's acknowledgs that bonds of lower grades are more likely to experience rating changes than bonds of higher grades (Moody's, 1978, v). This policy, combined with the initiation of the A-1 and Baa-1 categories to designate the better credit risks among A and Baa rated bonds, has nur- tured a general trend toward higher ratings over the 1960 to 1980 period. Of the 296 central cities which were rated in both 1960 and 1980, 38 per- cent experienced a net improvement in rating and only 24 percent experi- enced a net decline. The remainder experienced no net change in credit status. Similarly, during the 1970 to 1980 period, more cities improved in the ratings than declined. A second noteworthy trend over the two— decade span has been the increase in the proportion of the 382 central cities which carry credit ratings. Only 80 percent of the total were 60 rated in 1960 while 90 percnet were rated in 1980. The graphs in Figures 4 and 5 display absolute changes in the popular- ity of bond rating classes during the 1960 to 1980 period. A comparison of the 1960 and the 1967 graph in Figure 4 reveals a decline in the num- ber of unrated central cities, an increase in the number and relative pro- portion of Aa, A, and Baa rated central cities, and a decline in the num— ber of cities rated at the extremes of creditworthiness, i.e., Aaa and Ba. Beginning in 1968, the number of rating categories was increased by two, A-1 and Baa-1, and by 1970, as revealed in Figure 5, these new ratings accounted for almost 18 percent of the ratings assigned to central city general obligations. By 1980, the relative proportion of A-1 and Baa-l ratings had increased to 30 percent. In contrasting the early 1970s with the later 1970s, the period from 1970 to 1975 was characterized by an over- all improvement in bond ratings: Cities rated Aaa, Aa, and A-1 increased in number and relative percentage, while cities rated A, Baa-l, Baa, and Ba decreased in number and relative percentage. The 1975 to 1980 period, on the other hand, saw no net change in the number of cities rated Aaa, a decline in the number and relative percentage of cities rated Aa, and an increase in the number and relative percentage of cities rated Baa-l, Baa and below. This negative trend may have been related to the recession and inflation which severely affected central city fiscal well-being during this period. During the entire decade, the number of central cities not rated varied only slightly. A Graphic Analysis of Regional Change Aggregating rated central cities by Census region for the years 1960, 1970, and 1980, as illustrated in Figure 6, reveals that over the past twenty years (1) the Northeast has experienced a dramatic decline in credit 61 a IOSO COM"! cm» 0 I Number Au M! A I“ In No! In!“ ‘004 180‘ CHM. Noun 0! Conn! 8 l An In A In In No! no!“ Figure 4. Number of Central Cities in Each Bond Rating Category, 1960 and 1967. 62 1401 , .. 1970 110‘ Cities d 8 1 00‘ .0- o! Conmll 40d Numb" J ' / /, /, // Add An A-I A loo-l loo Do No! Rand 1401 . |975 920‘ cum ‘00‘ Cum“ 8 j c 8 l Number Au An A- A loo-l Inc II II! no!“ Figure 5. Number of Central Cities in Each Bond Rating Category, 1970, 1975, and 1980. 63 100‘ .- I980 120‘ CH“! 1004 .04 Cums! 004 of Number 3L 3 . I . I“ . 7' : 1 ~ \WV K . \\\ . A00 A0 A-l A Boo-I Boo Do No! Rom! Figure 5 (Cont'd). 64 NORTHEAST 1.00 NORTH CENTRAL 1.10 10.0 SOUTH 1.10 10.0 WEST Men! of loud Canal cm.- in tact County [:JDIIIII A loo-I Boo Boalonr Figure 6. A Regional and Temporal Comparison of Central City Bond Ratings. 65 standing, (2) the South a dramatic improvement, (3) the North Central lit— tle net change, and (4) the West only a slight improvement. The most con- sistently highly rated region has been the North Central with over 60 per— cent of its central cities enjoying either Aaa or Aa standing in 1960, 1970, and 1980. In the Northeast, on the other hand, the number of Aaa and Ad rated central cities has shrunk from about 50 percent in 1960 to 28 percent in 1980. In addition, the modal bond rating category for the Northeast changed from A in 1960 and 1970 to Baa in 1980. In the South it changed from A in 1960 and 1970 to A-1 in 1980. The modal category for the North Central and West has been Aa for all three years. The maps presented in Figure 7 depict changes in the regional dis- tribution of gilt edge and grit edge central cities at five year inter- vals during the 1960 to 1980 period. In 1960 it can be seen that the Northeast was as generously endowed with Aaa rated central cities as the North Central region. An examination of the succeeding maps shows in the Northeast a continuous attrition of cities in the gilt edge category. By 1980, the only Northeastern central cities to be included in the Aaa cate- gory by Moody's were Portland, Maine, and Stamford, Connecticut. Portland is one of only eight central cities in the United States which has retain- ed a Aaa rating for the entire two decades and Stamford, located on Con— necticut's "gold coast," is the only Northeastern city to be elevated to a Aaa rating during the period. Northeastern cities which lost their Aaa rating during the period include such cities as Buffalo, Rochester, and Syracuse in New York; Hartford in Connecticut; and Harrisburg and Lancaster in Pennsylvania. In 1960, 1965, and 1970, the only gilt edge central cities outside the North Central and Northeast were Salt Lake City, Utah, and Richmond, 66 REGIONAL PATTERNS OF CENTRAL CITY BOND RATINGS I960 " I980 0 Add o 800 8 lower Figure 7. The Gilt Edge and Grit Edge Central Cities, 1960-1980. 67 Virginia. The 1975 map reveals, however, a blossoming of the Sunbelt and West Coast with two cities for North Carolina, three for Texas, two for California, and two for Oregon showing up in the gilt edge category. Be- tween 1975 and 1980, the number of gilt edge central cities remained the same, but the Northeast lost Hartford, Connecticut, while California gain- ed Sacramento. The number of gilt edge central cities remained the same in the North Central region but Iowa lost Sioux City while Minnesota gain— ed Rochester. While the number of central cities in the Northeast with gilt edge ratings has been steadily declining, the number of cities in the grit edge category (Baa and below) has been steadily increasing. In 1960, the grit edge cities of the Northeast were confined to the eastern metrOpolises of the region. From this hearth, their number has steadily moved westward, extending all the way to Detroit; their density has also increased. At the same time, the number of grit edge cities in the South has steadily diminished and their number in 1980 was confined to central cities in Florida, Alabama, Mississippi, Louisiana, and Texas. In the West, where grit edge central cities never numbered more than five, there was an ini- tial increase in their number during the early 19605 followed by a gradual decline. Three basic trends in the regional patterning of central city bond ratings over the two decade period from 1960 to 1980 may be defined: (1) the decline of Northeastern central cities, a decline underway since at least 1960, (2) the improvement of Sunbelt and West Coast central cities, particularly during the 19703, and (3) the stability of North Central cen- tral cities which have remained the most highly ranked in the nation. Using Sale's 37th parallel as the divide between Sunbelt and Frostbelt 68 reveals that the strongest cleavage along this line was in 1960. At that time, the Sunbelt was a decidedly homogeneous region of low credit stand- ing while the Frostbelt's central cities enjoyed an overall high evalua- tion of their creditworthiness. The increase of grit edge cities in the Frostbelt and the increase of gilt edge cities in the Sunbelt since then has largely blurred the regional dichotomy. By 1980 the most striking dichotomy between extremely high and extremely low credit ratings was not between Sunbelt and Frostbelt but within the Frostbelt itself. A line passing between Lansing and Detroit, Michigan, and between Fort Wayne, Indiana, and Cleveland, Ohio, splits the Frostbelt into two almost per- fectly homogeneous regions of credit quality as evidenced in the 1980 map displayed in Figure 7. The long-term trend, assuming the pattern of the 19703 continues, seems to be toward a reversal of the Sunbelt-Frostbelt cleavage that existed in 1960. The Frostbelt is becoming increasingly characterized by central cities with low credit ratings while Sunbelt central cities are attaining increasingly higher credit ratings. This trend is confirmed by comparing the maps which portray the gilt edge and grit edge cities in 1960 and 1980 (Figure 7). Further evidence is provided by Figure 8 which depicts all central cities which have moved up and down in credit standing by more than one rating during the 1960 to 1980 period and during the sec- ond half of that period, 1970 to 1980. The Sunbelt stands out on those maps as a region of improving creditworthiness while the Northeastern por— tion of the Frostbelt stands out as a region of decidedly declining credit- worthiness. The map of changes in credit ratings during the 19703 reveals that only a single Sunbelt city moved down by two or more rating classes during the 1970 to 1980 period. Twenty-six Frostbelt cities, however, 69 l960-I980 O I foo I ~ . \ <- _ 5 a {""”"’ M l9 70-l980 T.— g. 1: - IA-..- - . A A A A . w . . -Lxfi) A . «3 .-¢.1:I: . A A \-.».. o A v A am a . . m o o u . u . a .o n V . . g u n s o c - _ s u .- u - 1000‘s “ u u A N A A A A. A A A w::t . u . u “- NI" ....... ‘- .- I I o u 1.”... u llllll . . . . n — u . IOIIA - UIIPII‘I’II"- . Tr IIIIIIIIIIIIIIIIIIII P . . P — '0‘. a o d ..... '0'--- Q _ . ""-“'-- A . .. A .A 1A . J . .. A u . u. . . . . . . . . . A . o . . o . .. . . . n . . . . D o n D Q Q Q A .r ........ A m. u. A A. A a . JI oooooo I . . aaaaaaaa J. ~I ...... d0!“ .- "SS “ 0m DI ..... 9. o ....... ts ... ... f 00' W More ... \ \ \ . !|!\!!i‘ A. A. o . fl ' I“!\ .. u. .-.. ...... .. A... . .- ...-\ r- .- . of} .. 0!.qu .- W n ooooooo .. lllllll g . I’ll . !II N m a" j s. A. \\\ 3 2 w T K a \\ e. m n a 2 A s I'll \\\ . w v P I \\\ s. I(\ o D D u u e. !l!ll \\ A . I! o o o a o A. o . Changes in Central City Bond Ratings, 1960-1980 and 1970- 1980. 0 Down 3 or more roflnqs :- Up 2 voting: 0 0m 2 refines - Up 3 or non mflnos Figure 8. 70 declined by two or more ratings during the same period. By the same token, so long as the American manufacturing belt is de- fined as comprising the Northeast and North Central regions of the United States, its internal pattern of central city bond ratings continues to be very heterogeneous. As the eastern portion of the manufacturing belt grades into the agricultural and less densely populated interior, central city bond ratings improve considerably. Likewise, the peripheral amenity belt as of 1980 still contains a heterogeneous pattern of bond ratings with both extremely high and extremely low rated central cities. In terms of changing bond ratings (Figure 8) the peripheral amenity belt exhibits a much more homogeneous complexion in that both the West and the South Census regions completely lack cities which have fallen two or more rat- ing categories over the past decade. A Statistical Anaiysis of Regional Change How successfully can geographic region be used to predict bond rat— ings of central cities? To test the hypothesis that there has been a significant correlation between the bond ratings assigned to central cit— ies and their locations, crosstabulation analysis has been employed. Bond rating categories have been set up as the dependent variable and both Census region and Census division as the independent variables for each year between 1970 and 1980. For this analysis and for the multiple regression analysis to follow, bond ratings are being treated as an interval level random variable. The assumption that the intervals between bond rating categories are equiva- lent has been made by Bahl (1971) in his study of municipal creditworthi- ness, and by Horrigan (1966), West (1970), and others in their regression analyses of corporate bond ratings. Census region and Census division are 71 both nominal level variables. With a nominal level and an interval level random variable, the cor- relation ratio represented by eta—squared is an apprOpriate measure of association in crosstabulation analysis (Mueller, Schuessler, and Costner, 1970, 325—333). Eta varies between zero and one depending on how differ- ent the means are in each of the nominal categories, in this case in each of the four Census regions or the nine Census divisions. Eta-squared may be interpreted as the proportion of the variance accounted for by the in— dependent variable, region or division. The Chi-square test statistic indicates a highly significant statis- tical association between bond ratings and both region and division (Ta- bles 9 and 10). In each year between 1970 and 1980, the pattern of cor— relation which emerged from the crosstabulation analysis would have been expected by chance only once in more than 10,000 times as evidenced by the .0000 level of statistical significance. The values of eta and eta— squared, as enumerated for regions in Table 9 and for divisions in Table 10, indicate the strength of association between the dependent and the independent variables. As revealed by the values of eta-squared in Table 9, the maximum proportion of the variance which may be accounted for by Census region is 17.8 percent in both 1979 and 1980. Since 1972 the trend in explained variance has been upward indicating that region is becoming a better predictor of bond ratings. Using a finer geographic mesh, that of Census division, the predictive power of location is increased to 22.6 percent for 1980. Again the proportion of variance which may be explain- ed by location has been upward since 1972. The values of eta—squared for Census division are presented in Table 10. If regional boundaries were altered to conform with spatial bond rating patterns, rather than using 72 Table 9. A CROSSTABULATION ANALYSIS OF BOND RATINGS AND CENSUS REGION: 1970-1980 Level of Significance Degree of Association Year Chi Degrees Sifnif— Eta Eta- square of icance Squared Freedom 1970 58.3 15 .0000 .35129 .123 1971 57.4 15 .0000 .35568 .127 1972 54.3 15 .0000 .33519 .112 1973 60.3 15 .0000 .34611 .120 1974 67.9 15 .0000 .35750 .128 1975 - 61.2 15 .0000 .36903 .136 1976 54.3 15 .0000 .35613 .127 1977 63.0 15 .0000 .38567 .149 1978 71.7 15 .0000 .40497 .164 1979 80.2 15 .0000 .41948 .178 1980 80.0 15 .0000 .42161 ,178 Source: SPSS, CROSSTABS. 73 Table 10. A CROSSTABULATION ANALYSIS OF BOND RATINGS AND CENSUS DIVISION: 1970-1980 Level of Significance Degree of Association Year Chi Degrees Signif— Eta Eta- square of icance Squared Freedom 1970 90.1 40 .0000 .41007 .168 1971 93.6 40 .0000 .41868 .175 1972 103.5 40 .0000 .40668 .165 1973 100.7 40 .0000 .41141 .169 1974 106.1 40 .0000 .43434 .189 1975 106.0 40 .0000 .44554 .199 1976 111.8 40 .0000 .45163 .204 1977 114.4 40 .0000 .45645 ,208 1978 116.7 40 .0000 .46740 .218 1979 122.0 40 .0000 .47225 .223 1980 117.8 40 .0000 .47542 .226 Source: SPSS, CROSSTABS. 74 pre-established regional divisions, it is likely that the explained vari- ance would be higher. Bond Rating Dynamics of the Gilt Edge Central Citiesi 1960-1980 Over the past score of years, forty-four central cities in the United States have found themselves assigned to the Aaa rating category at one time or another. These cities are singled out by name in Figures 9, 10, and 11 in which their credit rating histories are charted at yearly in— tervals for the period between 1960 and 1980. The entire set of gilt edge cities has been divided up into regions with Northeastern central cities displayed in Figure 9, North Central in Figure 10, and South and West in Figure 11. Each composite regional profile evidences a distinctly differ- ent pattern of bond rating dynamics. In the Northeast, there were thir- teen central cities which were rated Aaa in 1960 and only two with that rating in 1980. In the South and West the pattern is completely reversed: In 1960 there were only two central cities with Aaa ratings and in 1980 there were eleven. The most dramatic decline in bond rating since 1960 was among cities which began the 1960-1980 period with a gilt edge rating and plummeted to the Baa category by 1980. Only Harrisburg, Pennsylvania, and Buffalo, New York, experienced this dramatic but negative total tran- sition. No central city has moved up to the Aaa category from a rating less than A, and only four have made the transition from A to Aaa during the past two decades. All but one of the four, Rochester, Minnesota, have been Sunbelt cities. Only one central city in the Northeast and one central city in the west retained a Aaa bond rating for the entire two-decade period. No .central city in the South may claim twenty-one years of Aaa status. This MW Nurflonl «has!» Incas!" .. 3 Btldopor! - Aaa human \ \ \ .\ \I \‘H u u \I \I I I I If Stamford u \\ u u \‘ I \ SprInqh'old Ad New Haven Woven!" I Nerlden Buffalo Hurrlsburg A-I ' A | Boa-I NORTHEAST Isa ITIIIIiI 111T 111111; N O O O E- s s as i :e '3‘ § § § '-.: Figure 9. Rating Profiles of Gilt Edge Cities in the Northeast. 76 nguaue Fort Wayne Kalamazoo Milwaukee Omaha Ifinneqpofie Sioux City Loafing Aaa lads H rggee Cedar Rapids lmdkuguxfls Ao Hadleon Dayton Euanekm Davenport I I] Deslolnee A-l NORTH CENTRAL A etr I’ l 1 1 1 1 1*‘1 r’ 1 1 1 1f 1’ 1 1 1 1 1 1 g 8 i i i i a a Figure 10. Rating Profiles of Gilt Edge Cities in the North Central. 77 L l J J l L I l L l l l l J l 14 l l l I 53» Lake Clty Aaa Rléltntond to: Angel“ JI Ralel J rl t , . Auetln _____ allae Aa Portland Salem San Francleco I] III horamento A-l A SOUTH 3 WEST Houston — recon mm reu- weed nee-A «roe ten-u ten - me A ten-l teeo- Figure 11. Rating Profiles of Gilt Edge Cities in the South and West. 78 LllllJlllliJlJlellJJ Aaa Aa lullalo A4 I auetuclil I Love" A I. New 1 Baad Boston Baa I gardck Tray Ba 8 NEWYORKG NEWENGLAND Caa 'IljljjTl'Ijz‘I‘;l:lg 2 a a 3 r 3 a a a a : Figure 12. Rating Profiles of Grit Edge Cities in New York and New England . Add -~ eflnm _!fldmflon Scranton Wadelphla A Ede figumrk Paeealc -aanu Baa-l v] Flfluuon I thruugghx Lon Bnnmh Iaa 'k 1 [ Harteten Aflonuc (Nty may I NEW JERSEY& . PENNSYLVANIA a Aahgylfluk ;] gTLTlTT1llglll11l1LT§ ; 3 a i 2 E s a; -=. 5 e Figure 13. Rating Profiles of Grit Edge Cities in Pennsylvania and New Jersey. 80 Ad 11!! Paul Cleveland it. Cloud Noorhead I Qetroit NORTH CENTRAL 1002 d 1004-4 1000-1 1000‘ 1070-1 1072- 1.74- d 9.70 cl ”7. .1 ”lo-4 Figure 14. Rating Profiles of Grit Edge Cities in the North Central. 81 Aa A-I l l 1 I 1 ti 1L 1 1.41 l J, l l .l l l 1,.1 1 11 /‘.VI II Vifliuxm iiffiumnbu Ahmyoood I l I loco Ratonl I, I ’ West Palm :Oeach ' fortygpro: Qaynma Bradenton Nolbourne ,7 x7 SOUTH ATLANTIC or T L I l I : I 1 1 L I l I I I 1 II. I 3 Q i 2 E Q i E E E E E E Figure 15. Rating Profiles of Grit Edge Cities in the South Atlantic. 82 ~ A0 A.. [/ Johnson City % I 8004 I I/ ’ n N /’/ [/l ’0’ / I jgumflfle ,1) Tuucoloota ,’/, fluntsvme I!” II I ’1 II I I” III, 300 T 1’,’ // Guinon 1’ J Ill ['lgobflo el/ gnu" I figdmmum Amman L Toscogoola 80 EAST SOUTH CENTRAL ; ‘1 l ’1' l I £ I ’1 1 L I I I ‘1 I 1 I 1 I 1% N a z s a a a a 5 § '5 : Figure 16. Rating Profiles of Grit Edge Cities in the East South Central. 83 Figure 17. Rating Profiles of Grit Edge Cities in the West South Central. A-l J Port Arthur fl ,7 grunge ([7 I Totarkano ’ ' “cranium II gflgmn dlnbur h Duo {1 Silvana loam. ggylnon Iv ’ Ikounmfluc on Charles Iuondfli No. Little Rock ] ’ I Phorr A 41040 8 WEST SOUTH CENTRAL o sou-4 1.“- 9.... 1010-4 1071-4 1074- 1010-1 an- 1002‘ Figure 17 .. 85 Ao A-l 1111111111111111111 L I [I / / 1 ! “our” / // // l/ I/ / influflud [A 'Anohihn Anchors” .Foirfiold ] “Vogue WEST :rl'J'L'L'; z'11g':§ §§§!§§§§§§2 Figure 18. Rating Profiles of Grit Edge Cities in the West. 86 paucity of consistently high ranking cities contrasts sharply with the North Central where six central cities have been rated Aaa for the past two decades. Four more in the region were removed from the Aaa category during the 19603 but had been reinstated by 1980. The North Central re- gion's pattern of Aaa dynamics, as seen in Figure 13, has been character- ized by decline during the 19605 with half of the cities rated Aaa in 1960 dropping out of the gilt edge category by 1968. During the 19703 in the North Central region, however, there was a reversal of the credit evalua- tion of many cities as seven central cities moved into the gilt edge cate- gory from lower classes and only one city dropped to a lower rating. Bond Rating Dynamics of the Grit Edge Central Cities, 1960—1980 The grit edge central cities are those whose general obligation bonds are rated Baa or below. Baa-1 rated cities, a category used only since 1968, are not included in the grit edge category. In actuality, few cen- tral cities, only fourteen during the past twenty years, have been rated Ba or below with the result that most of the grit edge cities carry Baa ratings. All cities which have fallen into the grit edge category during the past two decades are profiled in Figures 12 through 18. They are grouped by either Census region, Census division, or groups of states within a Census division. The conspicuous point of contrast between the temporal profiles of the grit edge cities and the gilt edge cities is that there have been al— most twice as many grit edge cities as gilt edge cities over the past two decades. Seventy-nine central cities have found themselves in the Baa rating category or below between 1960 and 1980. When these cities are 87 grouped by region, distinct regional profiles emerge just as they did with the Aaa cities. The Northeastern central cities are illustrated in Figures 12 and 13. These two graphs portray the declining evaluation of credit quality in the Northeast. While there were only eight Northeastern central cities with Baa or lower bond ratings in 1960, that number had increased to nineteen by 1980. Most of these had dropped into the Baa category from an A rating. Only one central city, Pawtucket, Rhode Island, began the decade of the 19603 with a Baa rating and ended the decade of the 19703 with a Aa rat- ing. Long Branch, New Jersey, was the only central city that began with a Ba rating and ended with a A rating, and only one other, Jersey City, New Jersey, began with a Baa rating and ended with a Baa-1 rating. Another conspicuous feature of Figures 12 and 13 is the numerous changes in credit ratings for so many central cities. This makes many of the graphs appear like a maze. Very few grit edge central cities in the Northeast which have had their ratings revised upward during the two- decade period have retained those higher ratings. The period of substan- tial downgrading in credit quality began in New Jersey and Pennsylvania during the later 19605; New York and New England did not experience the initiation of such a decline until the mid-1970s. The remarkable point of contrast between the Northeast's temporal profiles and the North Central's, as portrayed in Figure 14, is the far fewer number of cities in the latter region which have ever been rated Baa or below. The pattern of grit edge ratings in Figure 14 parallels the pattern of gilt edge ratings in Figure 10 in that the period through the early 19703 was one of overall movement down while the remainder of the decade was one of overall movement up in credit evaluation. The only 88 exceptions to this generalization are Detroit and Cleveland, the North Central's only two grit edge cities in 1980. In contrast to the temporal profiles of grit edge cities in the North- east, the temporal profiles of grit edge cities in the South, as portrayed in Figures 15, 16, and 17, and the West in Figure 18, reveal a pattern of movement out of the grit edge category. The South began the 19603 with thirty-two grit edge central cities and ended the 19703 with only nine. Similarly, the West had five central cities rated Baa or below in 1960 but only one, Las Vegas, in 1980. No city in either region had made the com- plete transition to gilt edge status, but Jackson, Mississippi, and Albu- querque, New Mexico, had improved to a Aa rating and thirteen more had risen to A—1 standing. Also noteworthy is that not a single city in ei- ther the South or the West dropped permanently into the Baa category from a higher rating and only one central city, Galveston, Texas was reassigned to the Baa category after a brief period of higher credit standing. Summary The regional patterning of central city bond ratings in 1980 has re- vealed a strong cleavage between the highly rated cities of the North Cen- tral region and the lowly rated cities of the Northeast. This cleavage makes it impossible to characterize the Frostbelt as a homogeneous bond rating region. The peripheral amenity belt was also characterized by an east-west split as central cities in the West have been assigned ratings in the highest classes while Southern central cities reveal a diversity of ratings spanning the entire investment grade rating continuum. The most highly rated Census division in 1980 was the West North Central, while the most lowly rated division was the Middle Atlantic. 89 Between 1960 and 1980 there has evolved a general trend toward higher ratings. At the regional scale, Northeastern central cities experienced a dramatic decline in bond ratings, the South a dramatic improvement, the North Central little net change, and the West only slight improvement. The most consistently high rated cities have been those in the North Cen- tral region. These trends are verified by the regional distributions of all bond rating categories and particularly the gilt edge, Aaa, and grit edge, Baa and below, categories of bond ratings which have been the most regionally concentrated. Maps of those cities which have improved and declined in the ratings between 1960 and 1980 show the Sunbelt to be a region of improving credit standing, while the industrialized Northeast stands out as a region of decidedly declining credit evaluation. The statistical analysis of regional change in credit ratings revealed a sta- tistically significant association between bond ratings and Census region and division, an association which has increased over time or that in 1980, Census divisions alone explained 22.6 percent of the variation in credit ratings nationwide. The temporal profiles which were devised to depict the bond rating histories of the gilt edge and grit edge cities over the past twenty years illustrate once again the distinctive regional character of central city bond rating dynamics. The temporal profiles for Northeastern gilt edge cities show a continuous attrition of cities rated Aaa, while those for Southern and Western gilt edge cities show a movement into Aaa category, and those for North Central gilt edge cities a pattern of fairly consis- tent high ratings. The grit edge cities evidence bond rating histories which are just the opposite, though subject to more frequent changes. Grit edge cities in the Northeast have been shown to have dropped into the 90 Baa and below rating categories since the late 19603, while grit edge cities in the South and West have been shown to have moved rapidly out of the lower ratings and into higher ones. The North Central region has had only a very few grit edge cities; those which dropped into the Baa cate- gory during the two—decade span have tended to rebound to higher ratings. The bond ratings have therefore been shown to be a highly regional phenomenon which have changed their geographic complexion considerably over the past score of years. The spatial correlates of municipal bond ratings which will be discussed in the next chapter may be called upon, in fact, to help explain these regional tendencies. CHAPTER V SOCIAL, DEMOGRAPHIC, AND GEOPOLITICAL CORRELATES OF MUNICIPAL CREDIT RATINGS The objectives of the present chapter are twofold: (l) to establish some of the univariate relationships between assigned credit ratings on the one hand and selected demographic, geopolitical, and social charac- teristics on the other, and (2) to determine the degree of association between bond ratings and several multivariate statistical indices measur— ing geopolitical fragmentation, city need, and quality of life. The two methods of analysis used to test the significance of the relationship be- tween bond ratings and the above selected variables are (1) crosstabulation analysis using gamma as a measure of association and the Chi-square test statistic, and (2) one-way analysis of variance to test the differences between group means of normally distributed variables.3 Because so many of the variables examined in this chapter are highly regional in charac- ter, the explanation for the regional patterns delineated in Chapter 4 is provided here. Furthermore, the present chapter lays the foundation for 3In general, crosstabulation analysis was used on discrete variables and variables such as population size which are continuous but not nor- mally distributed. Analysis of variance was used on variables which are continuous and which appear to be normally distributed. 91 92 the multivariate regression and discriminant models to be developed in the chapter which follows. Population Size As presented in Chapter 2, the central cities under investigation ranged from 17,000 to almost 7.5 million inhabitants in 1976 according to the population estimates of the U.S. Bureau of the Census. With respect to the impact of population size on assigned credit ratings, Horton (1970, 32) has summarized the traditional wisdom in this regard as follows: The population of a community is likely to influence the rating of its bonds in a number of ways. Larger communi- ties tend to have more specialized and experienced finan- cial staff and management, and a larger community is likely to have greater economic diversity than a smaller one and thus is able to better withstand fluctuations in economic conditions. Size in itself may allow a larger municipality to withstand financial difficulties which a smaller com- munity could not. There is also the consideration that the larger community may be more able to depend upon being bailed out of financial difficulties by higher levels of government. Mendelson and Robbins (1976, 536) add another factor to account for the predicted relationship between bond ratings and city size by noting that "available data will probably be more comprehensive for larger com- munities, thereby facilitating the analysis [of creditworthiness]." Sanders' empirical investigation of 1976 bond ratings for all American cities with 10,000 or more inhabitants verified the fact that "with the notable exception of New York City, large cities receive the highest rat- ings" (Sanders, 1979, 107). Because of the theoretical and empirical evidence which does suggest that higher ratings are assigned to larger cities, population size has been the most common non-financial factor in— corporated into models designed to predict municipal bond ratings (Table 5 of Chapter 3). 93 The relationship between bond ratings and city size is tested in this study by using crosstabulation analysis and Chi-square statistic. All cities under investigation were divided into population size quintiles based on their 1960, 1970, and 1976 populations. These quintiles were then crosstabulated with four categories of bond ratings, Aaa, Aa, A/A-l, and Baa/Baa-l and below. Only four rating categories were used in order to make comparable the 1960, 1970, and 1976 matrices. For all three years the Chi-square test statistic indicated a highly significant association between city size and bond rating category. These significance levels are listed in Table 11. This association, as it ex— isted in 1976, is graphed in Figure 19. It can be seen that with every step down in bond rating, the proportion of large cities decreases and the proportion of small cities increases with striking regularity. As evident in the figure, there were no Aaa central cities in the smallest population quintile in 1976. Nor were there in 1960 or 1970. In addition, despite the fact that one-quarter of all rated central cities fell below 50,000 population, a common cut-off for metropolitan status, none of the Aaa cities and only 20 percent of the Aa cities fell below the 50,000 threshold in 1976. These findings support the contention that there is a minimum population size which appears to be a necessary, albeit insuf- ficient, condition for being assigned the highest credit rating, Aaa. A large population size, however, is no guarantee of a superior credit eval- uation as witnessed by the 12.5 percent of the Baa-l and lower rated cit- ies which were in the largest population size quintile in 1976. In 1980 the outstanding examples of large cities with poor credit ratings were New York City and Cleveland, Ohio, the only two central cities in the coun- try rated at the sub-investment level. 94 Table 11. BOND RATINGS AND CITY POPULATION SIZE Y r Raw Degrees Signif— Gamm ea Chi-square of Freedom icance a 1960 36.7 12 .0002 .23918 1970 46.9 12 .0000 .31738 1976 44.5 12 .0000 .39280 Source: SPSS, CROSSTABS. Table 12. BOND RATINGS AND METROPOLITAN POPULATION SIZE Raw Degrees Signif- Year Chi-square of Freedom icance Gamma 1970 24.6 12 .0167 .18021 1976 23.8 12 .0216 .21712 Source: SPSS, CROSSTABS. 95 100 80 SO 40 20 O Aocl Ao A/A-l Boo-I Blower I CO CO CO 100 PERCENT OF CITIES IN EACH POPULATION SIZE OUINTILE 1 O 20 . . _ g ' 5.. ‘.'.‘._.>. I -'-'o ‘t‘u' 'lu' . .;.‘ ' . 3 _. ‘3. - A . , .-. . .‘ 13.0 44.0 61.9 97. 5 1 83.1 -43.9 -S1.8 -O7.4 483 O -7422.8 City Population Sizo Uh thousands) Figure 19. Bond Ratings and City Population Size, 1976. 96 Another statistic reported in Table 11 is the gamma statistic. Gam- ma is the best measure of association between two variables which are both rank ordered by categories; other measures such as lambda and tau are not as well suited and may provide a "misleading summary" of the association (Mueller, Schuessler, Costner, 1970, 279). Gamma may vary between posi- tive one and negative one. In the table the sign of the gamma statistic is positive indicating a direct correlation between bond ratings and popu- lation size, i.e., as city p0pulation increases the bond rating also has a tendency to increase. The magnitude of the gamma statistic may be in- terpreted as the probability of correctly predicting the ordering of a pair of cities on the bond rating variable once the ordering of the cities on the population size variable is known. While the gamma value for the association in 1960 is somewhat weak, a value of .39280 in 1976 indicates a fairly strong association between the two variables. The trend toward a higher gamma over time also indicates that city pepulation size may be a better predictor of bond ratings in the future. Credit ratings of the nation's ten largest cities in 1976 are graphed for the 1960 to 1980 period in Figure 20. The temporal profiles of these cities indicate that region must be taken into consideration before it can be asserted that the nation's largest cities themselves are evaluated as the best credit risks. While Sanders, as cited previously, mentioned only New York City (Sanders, 1979, 107) as an exception to the foregoing rule, the graphical portrayal in Figure 20 indicates that the list of exceptions comprises more than one city, particularly in the Northeast. During the 19603, mega-cities in the South and West improved in the ratings while those located in the Northeast declined (Figure 20). In the North Central region, Detroit, like its Northeastern counterparts, 97 Figure 20. Temporal Bond Rating Profiles of the Nation's Ten Largest Cities. 98 IOO Aaa SO 60 4O 20 O Aa A/A-l Baa-I 8 lower L Figure 21. 20 40 SO CO 100 PERCENT OF CITIES IN EACH METROPOLITAN SIZE OUINTILE 42S.l 7851 —138.0 -245.0 -426.0 -785.0 -9527.0 CID 69.1 1 38 1 24 5.1 Metropolitan Population Size fin thousands) Bond Ratings and Metropolitan Population Size, 1976. 99 dropped in the ratings to a Baa, and Chicago, in keeping with the superior credit evaluations of many other cities in the North Central, improved in the ratings to a Aa. In the Northeast and North Central regions it has been the smaller (e.g., Portland, Maine, and Dubuque, Iowa) and larger 'medium-sized cities (e.g., Omaha, Nebraska, and Minneapolis, Minnesota) which have been rated Aaa fairly consistently. In the South and West, the largest cities (e.g., Los Angeles, Dallas, Houston) were the first to move into the Aaa rating category. In point of fact, the average size of Aaa cities in the South and West was 666,700 in 1976, while the average size in the Northeast and North Central was only 224,700. Both means fall into the largest population size quintile but a wide gulf separates the city means when divided on a regional basis. In addition to the size of the central city, the size of the entire metropolitan area is also examined with respect to its relationship with assigned credit ratings. Given the strong correlation between city size and metropolitan size, it is not surprising that the Chi-square test sta- tistic computed from the crosstabulation analyses in 1970 and 1976 indi- cated a significant relationship between metropolitan population size and bond ratings. Chi-square values, significance levels, and the gamma sta- tistics are listed in Table 12. As evidenced by the graph in Figure 21 and the gamma values in the table, this relationship was not as strong as the one with city population size though the same general trend presented in Figure 19 is still in evidence. Population Growth Rate The relationship between city finances and population growth rate has been demonstrated by both Peterson (1976) and Muller (1975a and 1975b). Peterson found that among cities with over 500,000 papulation, per capita 100 governmental expenditures were more than 70 percent greater in declining cities than in growing cities (Peterson, 1976, 48-50). Muller also assem— bled data for cities with over 500,000 pOpulation and found that per capi- ta outlays for local services averaged 46 percent higher for declining cities when compared with growing cities (Muller, 1975b, 36). No regional dimensions were considered in either study, however. Given these demon- strated correlations between fiscal characteristics and population change, it is reasonable to suspect that declining cities would be assigned lower credit ratings than growing cities, with the caveat that rapid growth de— riving from a narrow economic base or such industries as tourism would probably not be perceived as deserving of high quality ratings. In Sanders' empirical investigation of municipal credit ratings, population change from 1970 to 1975 was found to have only a limited asso- ciation with 1976 bond ratings (Sanders, 1979, 107). The results of the data set investigated in this study for the most part confirm Sanders' finding that population growth rate of the central city is not a good pre- dictor of bond ratings. As with population size, crosstabulation analysis is used to test the relationship between annual city growth rate (divided into quintiles) and bond ratings for the periods 1960—1970, 1970-1976, and 1960-1976. Significance levels based on the Chi-square statistic were found to be .204, .637, and .957 respectively for the three periods. None of the three are significant at the .01 or even the .05 level, results that are not surprising given the fact that almost all cities, with the exception of some growth centers in the Sunbelt, have entered a period of slow or negative population growth. Between 1970 and 1976, for instance, 47 jpercent of all rated central cities experienced an absolute loss in POpulation and 24 percent more grew by less than 1.5 percent annually. 101 In general, cities in the various bond rating categories seem to take on the growth character of the region in which they are located. That is, in the Frostbelt, both Aaa and Baa cities are likely to be de- clining; whereas, in the Sunbelt they are both likely to be growing. In 1976, Aaa cities in the Northeast and North Central regions experienced an annual growth rate of -.66 percent and Aaa cities in the South and West experienced an annual increase of 1.17 percent during the 1970 to 1976 period. Similarly, central cities rated Baa-1 and below were likely to be growing in the South and West, where their average annual growth rate from 1970 to 1976 was 2.12 percent, and declining in the Northeast and North Central at -.81 percent per year during the same period. While these figures mirror the national cleavage in growth rates, they seem to indicate that there is not a simple or direct relationship between bond ratings and city growth. Metropolitan pOpulation growth between 1970 and 1976 was also divided into quintiles and crosstabulated with bond ratings for 1976. No statis- tically significant relationship appeared; the significance level was .5604. Consequently, an alternative crosstabulation was designed. Only 14 percent of all SMSAs under investigation lost pOpulation during the six year period and an equal number grew by more than 18 percent. While the initial corsstabulation analysis using metropolitan growth quintiles fail- ed to reveal any direct positive relationship between metropolitan growth and bond ratings, it was suspected that either negative population growth in the metropolitan area or very rapid population growth may have an impact on bond ratings. The results of the crosstabulation analysis, which was performed using three categories of metropolitan growth and four bond 102 rating categories for 1976, are listed in Table 13.4 A Chi-square test yielded an alpha value of .0038 which indicates a highly significant re- lationship. As expected, both Aaa and Aa cities were underrepresented among both the declining cities and the rapidly growing cities. Similar— ly, the medium and substandard grade ratings are overrepresented at the extremes of the metropolitan growth continuum. There consequently appears to be a definite reluctance on the part of investment analysts to assign high ratings to cities in either declining or rapidly growing metropolitan areas. Consequently, the regional patterns of slow growth and rapid growth metropolitan areas discussed by Phillips and Brunn (1978) and by Zeigler (1980) may serve as good precursors of regional bond rating pat- terns. In 1980 the only two central cities rated at the subinvestment level were New York City and Cleveland, Ohio, both of which experienced nega- tive city growth and negative metropolitan growth during the 19703. Also in 1980, one-third of all central cities which were declining themselves and which were located in declining metropolitan areas carried a rating of Aaa or Aa. For cities with positive city and positive metropolitan growth trends during the 19703 the figure was not that much higher, only 42 per- cent. Metropolitan Geopolitical Organization Metropolitan geopolitical organization refers to the structure im- posed on a metropolitan area by the political boundaries which partition 4In Table 13 expected frequencies are given in parentheses under the observed frequencies. Each cell enclosed in a rectangle is one in which the observed frequency is less than the expected frequency. 103 Table 13. BOND RATINGS AND METROPOLITAN GROWTH Bond Rating Frequencies, 1976 Total SMSA Growth, (Expected Frequencies) Number of 1970-1976 Central Aaa Aa A/A-l Baa/Baa-l Cities and Below Zero or Negative 1 16 23 7 47 Population (4) (17) (22) (5) Growth Positive 24 98 105 16 243 Population (19) (88) (112) (23) Growth Rapid 2 9 28 9 48 Population (4) (18) (22) (5) Growth 27 123 156 32 338 Source: SPSS, CROSSTABS. 104 the region into an interlocking and overlapping jigsaw of municipal juris- dictions. The central city or cities which anchor a metropolitan area may be bounded by very constricting and inflexible boundaries or they may be granted the power to periodically adjust their corporate limits and annex adjacent territory. The power of annexation permits a central city to take advantage of new growth on the periphery by increasing the size of the city and its financial base. In general, annexation has a positive impact on the financial well-being of a city. A case study of Richmond, Virginia, by Muller and Dawson (1973, 1976), for instance, concluded that "from the perspective of aging central cities, there is little doubt that annexation is fiscally beneficial when viewed over the longer run" (Muller and Dawson, 1976, 80). During the period from 1970 to 1977, sixty-five central cities an- nexed more than twenty square kilometers of territory (U.S. Bureau of the Census, 1979a, 18). Sixty-one of those cities were rated by Moody's in 1978 and of those sixty-one, 61 percent carried either a Aaa or a Aa rat- ing; only one was rated below A. Nine gilt edge cities were among the sixty-one; these nine comprised one-third of all Aaa cities in 1978. The overall high rating of the most actively annexing central cities suggests that annexation is one of the geopolitical characteristics of cities which has a potentially favorable impact on the city's financial future. A. high credit rating, in turn, may save the city millions of dollars in debt service. As a practical application, such potential savings should be included in any cost-benefit analysis of the annexation process. One of the major components of metropolitan geOpolitical organiza- tion is central city dominance as measured by the percentage of the total ‘metropolitan or urbanized area population living in the central city. 105 Central city dominance during the 19703 is delineated in Table 14. The average percentages of central city dominance are grouped according to four bond rating categories. The most outstanding contrast in the percen- tage of the SMSA population living in the central city is between the Aaa rated and the Baa/Baa-l and lower rated cities in 1970 and 1976. In 1970, 37.0 percent of the SMSA population in metropolitan areas with gilt edge central cities lived in the central city whereas only 27.5 percent of the SMSA population in metropolitan areas with Baa/Baa-l and lower rated cit- ies lived in the central city. The gap separating the most highly rated and the most lowly rated central cities increased from 9.5 to 11.5 per- centage points between 1970 and 1976, indicating that cities which dom- inate their metropolitan areas have been favored in assigning Aaa bond ratings during the period. In addition, by 1976 there had evolved an even more regular relationship between average central city dominance and bond ratings. With each successively higher step up the bond rating scale, the mean proportion of the SMSA population living in the central city increas- es. This trend suggests that variables measuring central city dominance are becoming more closely associated with credit standing. In the future, therefore, it would be reasonable to predict that central cities which cannot dominate their metropolitan areas will be increasingly disadvan- taged when compared with cities which can maintain their dominance. By inference, this trend also suggests that central city dominance will have an increasingly stronger impact on city budgets and financial well-being. An analysis of variance test on the four bond rating groups in Table 14 indicated a significant difference among the 1976 group means at the .002 level. Another similar test on the ratio of central city to urban- ized area population in 1970 proved to be significant at the .013 level, 106 Table 14. CENTRAL CITY DOMINANCE Averages by Band Rating Category Average of Variable All Rated (Bond Rating Aaa Aa A/A-l Baa/Baa-l Cities Year) and Lower (Significance) Percent of Ur— 55.9 60.0 55.5 44.7 55.6 banized Area (.013) Population Liv— ing in the Gen- tral City, 1970 (BR = 1970) Percent of SMSA 37.0 35.0 36.4 27.5 34.5 Population Liv- (.056) ing in the Gen- tral City, 1970 (BR = 1970 Percent of SMSA 40,1 35.5 32.8 28.6 34.0 Population Liv- (.002) ing in the Gen- tral City, 1976 (BR = 1976) Source: SPSS, BREAKDOWN. 107 while the ratio of central city to SMSA population proved to be significant at the .056 level. An opinion about the tendency to downgrade cities vic- timized by their political geography, of which a goodly number are in the Northeast, is offered by Packer (1968, 95): Geographical boundaries have in recent years tend- ed to isolate those with the greatest need for pub- ic services and the least ability to pay for them. Apart from the social issues involved, the approach of simply downgrading the bond ratings of large central cities in response to those changes may well be an oversimplification, in view of the pivotal role these cities play in the economies of their metropolitan areas and states. Another primary component of metropolitan geopolitical organization is the number of local governmental units per 100,000 population in the SMSA. A geopolitical fragmentation index (GFI), discussed in greater de- tail by Zeigler (1980), has been computed for each of the 264 SMSAs in- cluded in the 1972 Census of Governments. In essence, the GFI combines the two major characteristics of metrOpolitan geopolitical organization, that is, the proportion of the SMSA population living in the central city or cities and the number of local governments (excluding special districts) per 100,000 papulation. The formula used for the computation of the index is: Number of units of local government per 100,000 population GFI = Percent of SMSA pOpulation living in the central city The underlying assumption behind the index is that geopolitical fragmen— tation is directly proportional to the number of governmental units per 100,000 papulation in the SMSA and inversely proportional to the percent of the SMSA papulation living in the central city(s). In essence the importance of the jurisdictional fragmentation ratio in the numerator is discounted as the preportion of the population living in the central city increases. 108 Scores on the geopolitical fragmentation index were rank ordered and the 264 SMSAs were divided into quintiles. Central cities were as- signed the quintile rank of the SMSAs in which they are located. These quintiles were then subjected to crosstabulation analysis with bond rat- ings for the entire decade of the 19703. Significance levels from the crosstabulation analyses ranged from .05 in 1975 to .87 in 1980 but even in 1975, the only year when alpha dropped below the .05 level, the gamma value of the association was only .041 indicating an almost nonexistent relationship. In addition, there was no identifiable trend upwards or downwards over the decade from an examination of the crosstabulation ma- trices, or their associated significance levels or gamma values. Central city dominance alone, the quantity appearing in the denominator of the GFI formula, proved to be a much better predictor of bond ratings than the GFI. Such a finding is understandable given the fact that the second var- iable introduced into the GFI calculations is descriptive of the metro- politan area as a whole rather than the central cities which are the en— tities being rated by the credit agencies. Selected Social Characteristics of Central Cities The social characteristics of central cities may influence the rating process in either or both of two ways. They may be examined directly by analysts and investors or they may affect the financial well-being of the municipality and through that medium indirectly affect credit ratings. One recent investments textbook highlights the direct impact of social characteristics on bond ratings by urging potential investors in the mun- icipal bond market to ask themselves the following questions: "Does the 109 population contain a substantial percentage of native—born, educated, in- come-tax paying, propertied citizens?" (Christy and Clendenin, 1978, 509). Indirectly, the financial success of a community and its prospects for the future rest on the ability of its residents to pay the necessary taxes to finance present and future obligations of the local public sector. This ability depends on such social factors as income, employment, and the de- mand for public services by groups which may not be financially capable of supporting them. In an attempt to identify the individual social dimensions of varia- tion along which central cities in the various bond rating categories differentiate themselves, group means on variables pertinent to income, employment, education, race and ethnicity, and housing have been examined and compared. The results of this breakdown by four major bond rating categories are displayed in Table 15. An analysis of variance test was performed on each variable in the table to determine the statistical sig- nificance of the differences among the group means. Most variables con- sider the status of a city in 1960 or 1970. More recent Census data are unavailable at the city scale for any of the variables except income. Many of the differences in means among the bond rating categories in Table 15 proved to be significant at the .05 level; a number were even significant at the .01 level. As might be suspected, per capita income in both 1969 and 1974 exhibited a strong relationship with bond ratings. The highest rated cities were the wealthiest and the lowest rated cities were the poorest, an assertion also confirmed by the differences among the proportion of all families living below poverty level in 1969 for each category. As Hirsch (1971a) has noted, "the presence of so many poor people is a major factor, perhaps the major factor, in the central 110 Table 15. BOND RATINGS AND SOCIAL CHARACTERISTICS OF CENTRAL CITIES: A COMPARISON OF GROUP MEANS Average of Variable Averages by Band Rating Category (Bond Rating w A11 Rated Year) Aaa Aa A/A-l Baa/Baa-l Cities and Lower (Significance) INCOME Per Capita $2034 $2019 $1858 $2146 $1972 Income, 1960 (.387) (BR = 1960) Per Capita $3131 $3166 $3007 $2837 $3044 Income, 1970 (.000) (BR = 1970) Per Capita $4832 $4569 $4474 $4068 $4476 Income, 1974 (.000) (BR = 1974) Income Growth, 54.3% 58.2% 59.4% 64.8% 59.4% 1959-1969 (.050) (BR 8 1970) Income Growth, 44.5% 44.9% 48.8% 47.4% 47.0% 1969-1974 (.013) (BR = 1974) Families Below 11.3% 9.5% 11.4% 15.2% 11.4% Poverty Level, (.000) 1970 (BR = 1970) EMPLOYMENT Unemployment 4.9% 5.1% 5.3% 6.5% 5.4% Rate, 1960 (.000) BR = 1960) Unemployment 4.3% 4.8% 4.6% 5.4% 4.8% Rate, 1970 (.001) (BR 8 1970) 111 Table 15 (Cont'd). Variable Averages by Band Rating Category Average of (Bond Rating All Rated Year) Aaa Aa A/A-l Baa/Baa—l Cities and Lower (Significance) EDUCATION Median School 10.8 10.9 10.7 10.5 10.8 Years Completed, (.091) 1960 (BR = 1960) Median School 11.8 11.9 11.8 11.3 11.8 Years Completed, (.001) 1970 (BR I 1970) RACE AND ETHNICITY Black Popula- 8.7% 8.9% 14.0% 18.1% 12.3% tion, 1960 (.000) (BR = 1960) Black Popula- 10.9% 11.7% 12.7% 16.9% 12.9% tion, 1970 (.063) (BR = 1970) Foreign Born Pop- 6.5% 5.7% 5.3% 5.7% 5.7% ulation, 1960 (.674) (BR I 1960) Foreign Stock 19.0% 16.2% 15.6% 19.8% 16.7% Population, 1970 (.142) (BR I 1970) HOUSING Owner Occupied 53.4% 58.5% 56.6% 55.9% 56.9% Housing, 1960 (.135) (BR I 1960) Owner Occupied 51.9% 56.8% 57.5% 54.8% 56.5% Housing, 1970 (.130) (BR I 1970) 112 Table 15 (Cont'd). Variable Averages by Bond Rating Category Average of ‘ All Rated #22:? Rating Aaa Aa A/A—l Baa/Baa-l Cities and Lower (Significance) Change in 7.1% 19.1% 21.9% 33.1% 21.8% Housing Stock (.024) 1960—1970 (BR = 1970) Housing Units 73.1% 61.3% 55.0% 52.8% 57.8% in Pre-1950 (.000) Structures, 1970 (BR I 1970) Source: SPSS, BREAKDOWN. 113 city's fiscal plight." When income growth between 1959 and 1969 and between 1969 and 1974 were compared, it was found that the lowest rated cities had faster rates of income growth than the highest rated cities. This inverse Correlation may, in part, simply be attributable to the fact that in a growing economy, cities with small bases are likely to experience higher average rates of growth over a period of time than are cities with large bases. Projected over time, this trend appears to auger well for the financial future of the poorer cities with rapid rates of income growth. Already in the Sun- belt are many cities, which were rated Baa or below in 1970, that have subsequently moved up and out of the grit edge category. Employment levels also exhibited the expected relationship with bond ratings, with the highest rated cities in both 1960 and 1970 experiencing the lowest rates of unemployment and the lowest rated cities the highest rates. Educational attainment, on the other hand, as measured by the me- dian number of school years completed by the population 25 years of age and older, failed to exhibit a statistically significant pattern in 1960, but it did in 1970 when cities rated Baa/Baa-l and lower fell far below the overall mean in median educational attainment. The average figures for the other band rating categories hovered close to the mean of all rated cities in 1970. In terms of race and ethnicity, the percent of the population black exhibited a much stronger correlation with bond ratings than did ethni- city as measured by percent of the population foreign born or of foreign stock. In both 1960 and 1970, those cities which had a higher proportion of black residents were not rated as high as those cities whose popula- tions were more predominantly white. 114 The housing variables exhibited the most surprising relationship with bond ratings. It was expected that the proportion of all occupied housing units which were owner occupied would be highest in cities with Aaa ratings and lowest in cities at the apposite end of the rating con- tinuum. While the differences among means of the four rating groups was not statistically significant, it is nevertheless surprising that in both 1960 and 1970, the Aaa cities had the lowest percentage of housing units owner occupied. It is not surprising, however, that in both years the Baa—l and lower rated cities also ranked below the mean of all rated cit- ies. The other two remaining housing variables measured the growth and age of the housing stock. In both cases the differences among group means were statistically significant. As for change in housing stock, it was the lowest rated cities which experienced the most change and highest rated cities which experienced the least. This finding suggests a pre- ference for stability among the credit analysts. Closely associated with housing growth was the percentage of housing units in structures built prior to 1950. The Aaa cities had the oldest housing stock and the Baa/ Baa-l and lower rated cities the youngest. A possible explanation for this pattern is provided by Sanders (1979, 107) who notes that "young cities need vast sums for new streets, sewers, and basic infrastructure. Their needs may exceed their fiscal ability and their governmental com- petence." In summarizing the association of selected social variables with cen- tral city bond ratings, central cities with high credit ratings as of 1970 had high per capita incomes, comparatively low rates of income growth, fewer than an average number of families below poverty level, low 115 unemployment, relatively high median educational attainment, relatively low percentages of the population black, and relatively slow rates of growth in housing stock. Characteristics of cities with lower credit ratings were, on the whole, just the opposite. In light of these find- ings, it is not surprising that the multivariate per capita needs index and the quality of life indices to be discussed next exhibit strong cor- relations with bond ratings. The Per Capita Needs Index Developed in the U.S. Department of Housing and Urban Development (1976) by Harold Bunce as a tool for evaluating the distribution of Com- munity Development Block Grant funds, the Per Capita Needs Index (PCNI) comprises a vector of attributes related to poverty, urban blight, and neighborhood instability at the city scale. Thirteen variables were fac- tor analyzed and five factors were derived, weighted, and consolidated into the final standardized index value. The specific variables used in the formulation of the index and the dimensions of variation which evolv- ed from the factor analysis are enumerated in Table 16. Four-hundred- thirty-five "entitlement cities" were rank—ordered on the basis of the PCNI. To test the proposition that there is a direct correlation between city need and general obligation bond ratings, the central cities of SMSAs were extracted from the list of 431 cities and divided into quintiles. Crosstabulation analysis was used to test the significance and magnitude of the correlation between urban need and bond ratings. While most of the variables used to compute the PCNI were derived from the 1970 Census, city scores on the index were compared to their band ratings for each year between 1970 and 1980. The assumption is that the 1970 Census stat- istics continued to influence the decisions of rating analysts well into 116 Table 16. COMPONENTS OF THE PER CAPITA NEEDS INDEX Variables Subjected to Factor Analysis Factors Defined and Weighted Persons Aged 65 and Over Crime Rate Nonwhite Population Persons Over 25 with Less than a High School Education Female Headed Families Below the Poverty Level Poor Persons Under 18 Persons Below Poverty Level Housing Units Lacking One or More Plumbing Facilities Occupied Housing Units With More Than 1.01 Persons per Room Unemployed Persons Housing Units Built Before 1939 Persons Per Square Mile Owner-Occupied Houses Poverty (.35) Age of Housing Stock (.25) Density (.20) Crime and Unemployment (.10) Lack of Economic Opportunity (.10) Source: Department of Housing and Urban Development, An Evaluation of the Community Development Block Grant Formula by Harold Bunce, 1976, pp. 49-51. 117 the 19703. The results of the crosstabulation analysis are summarized in Table 17. Between 1970 and 1980 the association between bond ratings and the PCNI became increasingly significant statistically. In 1970 the signifi- cance level was only .0726 but by 1980 it had risen to .0000 meaning that the association portrayed in the crosstabulation matrix could be expected to occur by chance fewer than one in ten thousand times. This finding supports the hypothesis that variables related to city need are influen- tial in assigning bond ratings to central cities. The gamma statistics in Table 17 also reveal a steadily increasing degree of association between bond ratings and city need over the ten- year period. Without exception, the magnitude of gamma increases with each succeeding year of the decade. In 1980, a gamma value of .31475 in- dicates a fairly strong degree of association. The negative sign, pre- sent throughout the decade. indicates a negative correlation, that is, cities in the low category of need have higher bond ratings than cities in the higher category of need. Changes in the matrices from year to year are the result of cities being either upgraded or downgraded in their band ratings. The trend in correlation over the course of the decade, therefore, indicates the downgrading of needy cities and the upgrading of the least needy. The relationship between bond ratings and the PCNI in 1980 is graph- ically displayed in Figure 22. The inverse correlation between the two variables is remarkably consistent through the fourth quintile of city need. That is, as city need measured on the PCNI decreases, the propor— tion of Aaa and As cities increases and the proportion of Baa/Baa-l and lower rated cities decreases and eventually disappears. In the fifth 118 Table 17. CROSSTABULATION ANALYSIS OF BOND RATINGS AND THE PER CAPITA NEEDS INDEX, 1970-1980 Year Raw Degrees Signif- Gamma Chi-square of Freedom icance 1970 24.8 16 .0726 .03817 1971 26.5 16 .0472 .05667 1972 30.7 16 .0148 .08196 1973 28.2 16 .0296 .13427 1974 35.7 16 .0032 .14628 1975 34.5 16 .0046 .18517 1976 35.8 16 .0031 .22268 1977 41.3 16 .0005 .22722 1978 39.5 16 .0009 .25732 1979 54.5 16 .0000 .29949 1980 58.2 16 .0000 .31475 Source: SPSS, CROSSTABS. 119 .ommH JOE; wvmaz 3595 Hum 93 van mwcwumm mean .3 auawwm .23. a Team < 0( 00¢ >¢oouhio 01:3: xo smmoo. CASH .mumm acmssoansmca Hmeoawmm amuse. caaauooma .mumm uamasoHQEmca as moamnmmaan nus: «Hood. «Aaa .msouaH muaamo pom ooo. man. man. aoammmnwmm NSSHo. osmd .coauaaaaom sumo we won mNmNO. «aaanossa .nuaouu ascocm mmmso. csoa .coaumasaom mmu< causeway: as some Hanucmu MNQMO. OhmH .meDuufihum OmmleHQ a.“ mafia—D wCfimSO—m mmHDwfihw> emmco. ohma .mumm uaaahoaaaaca Hmaowwmm amNso. casanooma .osmm aamssoaaamas ca auamuommao “songs: eased. sham .msouaH muaamo use coo. Ann. mmm. coamamuwom m CH mucosa m a coco coauoawm sea a“ moanoaum> Imacwfim vaumann< NM «mma .mmmwaauu< coaumxmcc< shame. cams .coaumasaom semen nmqqo. csma .ooam unassoHQEosa moanmfium> mmmqo. Ommauooma .oumm unashoaaemab aw moaauommwa Hacofiwam ounwo. umaozuuoz :uwz canoe. «Aaa .msoueH muaomo sum ooo. was. mms. aoammauwwm oosoo. ohma .wcawaom swansuuo gonzo Hooao. Assauoama .sua>auo< coaumxmaa< enumo. Onoa .mumm u:oE%oaan:D NMANo. Gama .coaumeaaom suao Co was magma. osma .coaumasaom sumac mmHnmaum> onmqo. chad .cowumasoom xooum owfiouom Hmaowwom momso. osoanoeAH .mumm unmesoHQEmaa ca mocmummuaa “songs: «mama. «Boa .osooaH muaono pom oco. mam. Haw. coamnouwom M as . I wanna“ M o coco aofiumaom osu a“ moannaum> luaawam mommafio< mm oon .mmqu Hmconmm nqu :OHmmmummmv AmmHanum> HMGOHwom uaosqu cOHmmouwomv oaom mwcHumm poem mouooaxm mwaHumm coon vmuoonxm Hmauo< oan .mmmWH HnsOHwom squ :0Hmnmuwmmv AmoHanum> Hmaonom uaosqu GOHmmouwumv noon mwcHumm wcom oouuoaxm nwsHuom ocom.vmuooaxm Hmauo< ome .mflmema Noumpom scams messages can. owaa ouHmo. oNAH .aoaumaaaom xuoum awamnom SNHNo. onauoosa .mumm acmsNoHEEmc: ca mocmnmmcaa Assoc. onH .coaumfisaom Naao no was Hmuucmo s.as mamma. oNsanoemH .aoaumaaaom sumam ca moamummman own. «Noe nunoz Nomso. oNsauooaa .mumm ucaesoaasmcs ca museummmaa «some. case .mumm acmesoHEEmca mNHNH. oNoH .coHumHaaoa amuHHoaouuoE mo wOH ~.wm mswoa. «Noe .maooaH madame sum cam. omsa mmsoo. oNaauoosH .wcamsom umaasuuo amaze as mucoumunaa moose. onH .coaumasaom mms< consumes: on some Hanucmo smmso. ossauooma .mumm unwesoaasmaa ca moamnommaa o.m~ massa. oNsH .mumm acmesoflssmaa mom. «AAA ummmnusoz emanammmau NHuomuuoo m aH moHuHu a mono cOHumscm onu cH mmHanum> NM you» aOmem No unmouom ome mam «NmH .mmeHauu< coaumxmaca moeNH. onH .aoHuMHaaom HmHuommnmz can HocOHmmomoum wmoma. oNaauosaa .muwm samesoaaamca ca mucmpmuoaa m.os Newsa. eNoH .coaumaseom Nuao to mom «we. ommH wsNHH. oNaauoosH .mumm assess coaumflsaom Noam omooa. onH .mmusuuanum ommaumum an mafia: wcamsom ooNNo. onH .GOHumHaaom HmHuowmawz mam HchHmmowoum m.Nm omoms. oaaauooma .coaumfisaom xumHm ca muamumcuan mms. «Nam Emma ommmo. oNoHuooaH .xooum wcamsom ca mwcmso mmomo. onauoomH .uumm acmasoHQEmcp ca mocunmaman mmHmH. onH .mumm unmENoHasman «.ms oHooN. oNsH .coaumaaaom some to was Hmm. omma mmNso. oNsH .mmnsuuauum omaatmom as some: wcamso: osHNo. onauoomH .oumm uamsNoHaEmaa an mocmamccao SMNHN. oNaa .aoaumasaom sumo mo woe o.as smomm. oNaa .mumm acmesoaaamaa mam. «Nma suaom smauammmeo hHuomuuou m :H maHuHo a coco GOHumsvm may :H moHanum> Nm umaw aOmem mo ufioouom .Nw.ucouv mN mHnma 146 that bond rating analysts weight factors differently in different regions or that the social, demographic, and geopolitical characteristics of cit- ies in different parts of the country impact differently on the financial well-being of the city. In all likelihood, it is a combination of both factors which accounts for these regional constrasts. Using the regression equation to predict bond ratings for 1974 and 1980 in each of the four Census regions resulted in a considerable im— provement over the nationwide analysis even though only four variables were used in each regional regression equation. In only one region, the North Central in 1980, however, did the number of correctly predicted ratings exceed 50 percent. In all cases, however, the classification accuracy of the regional equation improved between 1974 and 1980, by as little as 3.8 percent in the South and by as much as 15.2 percent in the Northeast. The differences in prediction accuracy indicate that social, demographic, and geopolitical characteristics of cities are more closely associated with (or are perceived to be more closely associated with) financial well-being in some parts of the country than in other parts. Moreover, the differences in prediction accuracy at the regional scale between 1974 and 1980 indicate that social, demographic, and geopolitical variables are becoming better predictors most rapidly in the Northeast and least rapidly in the South. Overall, the regional breakdown made it possible to correctly predict 46 percent of central city bond ratings in 1980, a 10 percent improvement over the nationwide regression analysis which yielded a predictive accuracy of only 36.1 percent. 147 A Multiple Discriminant Analysis of Municipal Bond Ratings Just as in the regression analysis, a stepwise discriminant analysis procedure has been used to construct the discriminant model and to pre- dict central city bond ratings. In the stepwise procedure, variables enter the equation on the basis of their discriminating power. The step- wise selection criterion used in this analysis was the procedure which minimized Wilkes' lambda by maximizing the differences between group cen- troids. The number of discriminant functions in each of the analyses was limited to two because in each analysis the significance of Wilkes' lambda exceeded the .05 significance level when more than two factors were identified. The location of individual central cities on each of the two factors was then used to predict bond ratings of the central cit- ies. These predictions also took into consideration the prior probabili- ties of being assigned to a particular rating category based on the ac- tual distribution of ratings in the original sample. The results of the nationwide discriminant analyses for 1974 and 1980 are presented in Table 29 which identifies the variables selected for inclusion in the discriminant equation, the standardized discrimin- ant function coefficients for each variable, and both the eigenvalues and percent of variance accounted for by each function. The standard- ized discriminant function coefficients measure the relative contribu- tion of each variable to each function; the unstandardized analogs of these scores are the values used in the computational discriminant for— mula. The sign of the standardized discriminant function coefficient reveals whether the variable is positively or negatively associated with the function. In 1974, it can be seen that the primary discriminant 148 Home.I «ammo.- oNaa .mmnauosuum omaaumna as wages wcamsom ~ssss.- Sasso.- oNaa .coaumasaom Hmaummmamz was Hmcoammmmonm mamas. NHNHM. osaa .mumm unmasoHQEmaa NHmNs. mmmso. oNsH .ucmsaamuu< Hmaoaumusem anaemz omNHH.u oeNHs. oNaa .coaumasaom sooum smashes Am.omv Aa.mov Nmamm. HmooN. ease .aoaumasaom amuaHosonumz to was ~mom~. seams. osas~.H- amasa.au onH .aoaumaaaom same we won omsa mNoMs.u HHoNH.u stHnmomH .nusonw meooaH muaamo sum sswm~.an Nuosm.u onH .coaumasaom mo unashamv COHumsvm onu :H mMHpmHnm> ham» mosHo>some nuaoHonmmou :OHuocom usmcHEHuomHa vONHpumocoum owaH 02¢ onH .mmme no ucmonmmv magmauanumou aofiuucsm :oHumsam was as mmHanum> new» mosHo>aome usmcHaHuomHo voNHouvaoum .Ac.ucoov mN mHan 150 function was basically determined by the relationships between central city, urbanized area, and metropolitan papulation size. Unemployment variables also exercised an important influence on the first function. In 1980, the demographic variables also proved to be the most influential discriminating variables but the racial-ethnic factors moved into place with the unemployment factors as the second most important group of vari- ables. This indicates that as cities have been reassigned to rating groups during the decade of the 19703, the racial-ethnic factors have become more important characteristics in determining the financial well- being, either perceived or real, of central cities in the United States. This finding confirms the trend established in the previous chapter con— cerning the relationship between bond ratings and the Per Capita Needs Index which was found to be an increasingly better predictor of bond ratings with each successive Year during the 19703. It appears, there- fore, that variables indicative of social structure may come to replace those indicative of economic structure as the major criteria of credit- worthiness. The eigenvalue associated with each function in the table denotes its relative ability to separate the groups. The percent of variance, which is also reported in the table, is based on the eigenvalue and is a measure of the relative importance of the function. It can be seen that the total percent of variance accounted for by both functions in 1974 is greater than the percent of variance accounted for by the 1980 equation indicating that the 1974 groups were more clearly differentiated along the discriminating variables than the 1980 groups. In both 1974 and 1980 regional variables, recoded as dummy variables, were injected into the analysis on a separate run; their discriminating power proved to be 151 either negative or negligible. In 1974, for instance, the addition of dummy regional variables reduced the discriminating power of the two main functions from 88 to 81 percent. In 1980 the discriminating power of the two main functions remained almost unchanged after the regional variables were included. Rather than indicating a lack of regional correlation, however, the poor performance of the regional variables simply indicates that dummy variables are not well suited to discriminant analysis. In 1974, as well, the five financial ratios were added to the original dis- criminant variable list on a separate run for comparative purposes. Only two of the financial ratios, percent of debt nonguaranteed and log of revenue per capita, were actually selected for inclusion in the discri- minant equation. Their inclusion reduced the percent of variance ex- plained by the two main discriminant functions from 88 to 86 percent. This finding supports the results of the regression discussed earlier in which financial variables were used. Once again, financial ratios have proven to be ineffective predictors of bond ratings and add virtu- ally no power to the discrimination made possible by social variables alone. Following the initial discriminant analyses, the equations derived were used to predict bond ratings for the central cities under investi- gation. The classification results for 1974 and 1980 are presented in Tables 30 and 31, which display the correlation between observed and ex- pected bond ratings. In 1974, 52.8 percent of the central cities were classified correctly and in 1980, 48.7 percent were. These results re- present a vast improvement over the predictions made by the regression equation, but the number of cities incorrectly classified betrays the wide range of variability within each band rating group. These findings 152 Table 30. CLASSIFICATION RESULTS OF THE NATIONWIDE DISCRIMINANT ANALYSIS, 1974 Actual Expected Bond Ratings Bond Ratings Aaa Aa A-l A Baa-1 Baa Aaa 3 19 O O O 0 Aa 2 89 0 11 O 3 A-l 0 31 0 17 0 4 A 0 26 0 36 0 1 Baa-1 0 4 O 2 O 1 Baa O 3 O 5 0 16 Total: 273 Correctly Predicted: 144 (52.8 percent) Source: SPSS, DISCRIMINANT. 153 Table 31. CLASSIFICATION RESULTS OF THE NATIONWIDE DISCRIMINANT ANALYSIS, 1980 Actual Expected Bond Ratings Bond Ratings Aaa Aa A-l A Baa-l Baa Aaa 9 17 O O O O Aa 4 74 13 l 0 3 A-l 0 28 20 ll 2 4 A 0 9 21 16 0 5 Baa—1 O 3 2 4 4 3 Baa 1 2 3 5 1 12 Total: 277 Correctly Predicted: 135 (48.7 percent) Source: SPSS, DISCRIMINANT. 154 and the results of the discriminant analyses done by others call into question the assignment of central cities to the various bond rating groups by the rating agencies. If the differences within each group are greater than the differences between groups, an easily supported hypothe- sis, bond ratings themselves may not represent meaningful or easily in— terpreted categories of creditworthiness. When the crosstabulated results of the 1974 and the 1980 analyses are compared, however, it is of significance that the A-1 and Baa—1 cate- gories have begun to differentiate themselves from their nearest neigh- bors. The discriminant equation for 1974 assigned not a single central city to either the A-l or Baa—1 category despite the fact that 22 per- cent of all cities classified carried one of these two ratings. In 1980, by contrast, the discriminant equation was able to make assignments to these two groups. It is obvious that even though the A-1 and Baa-1 cate- gories were initiated in 1968, by 1974 they still had not distinguished themselves from the four major investment grade classifications along the social, demographic, and geopolitical dimensions under investigation in this analysis. By 1980, however, these groups had begun to take on separate distinguishing characteristics that separated them from other groups with the result that 31 percent of the A-l cities and one-quarter of the Baa-1 cities were correctly classified by the discriminant equa- tion. As these groups continue to set themselves apart from their near- est neighbors, it should be possible to increase the predictive accuracy of discriminant models. The cities which were overrated and underrated by the 1974 discrimi- nant equation are mapped in Figure 23. On this map, closed circles re- present cities whose actual bond ratings were lower than their predicted 155 .mu uname .eNmH .anhHoad unnaHaHuoan ovHaaOHunz on“. 5" «3:0 maggots: can woumuuog 3:6 .228 ozotoue: o . 0.3-0 _ON~COO 3‘0550’0 O O O C O o «H. I .. o u u u I. . . . .. . k . O .. . r--- i .. J, . tonttutllttttifl iiiiiiiii ‘v . a/ n . u IIIIIIIII ‘P w I’ . a L m , n . . / . . . 1 ..I Ila - 5 I'll- a“ .- u II H l. O Plll llll . o gs to .m tltgoonu O . a. O .. a . . ....... 9-55-1 .. 3L--. .. .1 .. Hit... . . . O H cool! 0 c I. llllll . u I 0 wt llllllllllll L.~ ttttt lTltloo \s 6’ /""' 3A. j 156 band ratings (underrated cities), and open circles represent cities whose actual bond ratings were higher than their predicted bond ratings (overrated cities). Because almost half of the central cities included in the exercise were incorrectly classified, it is difficult to detect any strong regional patterns. Several observationsbased on the spatial patterning of the overrated and underrated cities may be made, however. First, most of the Aaa cit- ies show up on the map because the discriminant equation was decidedly unsuccessful in predicting Aaa bond ratings, suggesting the inclusion of a wide variety of central cities in this rating category. Second, the West shows up as the region with the fewest underrated and overrated cit- ies indicating that in this region, social, demographic, and geopolitical characteristics of cities are most closely related to bond ratings. Third, underrated cities show up conspicuously in the manufacturing belt extend— ing eastward from Wisconsin and Illinois to New Hampshire and Massachu- setts. Given knowledge of these cities' social, demographic, and geo— political structure, one would expect their credit ratings to be higher than they actually are. Fourth, many of the largest cities in the North— east and North Central regions appear as underrated cities, possibly in- dicating that their size and associated social characteristics generate certain negative externalities which adversely impact bond ratings even though they are not measured in this analysis. Fifth, the South appears as a region of both overrated and underrated cities indicating the state of flux of credit ratings in this part of the country. In the separate analyses conducted for each of the four Census re- gions, the predictive power of the discriminant equations computed for 1974 and 1980 was, in most cases, greatly improved. The results of the 157 regional discriminant analyses are presented in Table 32. Only in the Northeast in 1980 was the discriminant model unable to correctly predict more than half of the assigned credit ratings correctly; in 1974 the Northeast also fared the worst among the four regions, with only 51.6 percent of its central cities correctly classified. The relatively low prediction accuracy for the Northeast indicates that each bond rating category is poorly differentiated from the others in terms of the char- acteristics examined. The Northeast region of the country, therefore, seems to be the one in which the social, demographic, and geopolitical variables have the least to do with the evaluation of financial well- being by credit analysts. The region which yields the best results in the discriminant analysis is the West, which had the highest percentage of its central city bond ratings correctly predicted in 1974 and in 1980. In 1974, in fact, 87.5 percent of central cities in the West were car- rectly classified by the discriminant equations. The West therefore emerges as the region in which the social, demographic, and geopolitical characteristics of cities are the most closely associated with bond rat- ings. Overall, the regional breakdown made it possible to correctly classify 61.2 percent of the central cities in 1974 and 58.8 percent in 1980. The fact that separate regional analyses greatly improve on the classification accuracy of the nationwide discriminant equations suggests two conclusions, either or both of which may hold. First, region by it- self is directly considered by the rating analysts with the result that some parts of the United States are discriminated against while others are favored on the basis of "regional image." Second, additional varia- bles which are highly regional in nature are examined by the bond analysts 158 .sz mo ucoouomv moaHm>aame owaH nz< «NmH .mflm>Hxo H4 04 > 92 ouqu toqx uumtmu (mow oo Hm P“U>UUJmH4D—U HUJHOUQI—K ZHQVII. “HUQMJGh4u ZH4z DKWOO‘CO: HI” QJ>KZFUOI¥¢KDZ¢D

02¢ K‘Qt—jmlflmm-IUUUOKOUH)ZUJOQUJKZ ZDUVJJOD U:KXUDQWUOJJQP=KOK<°H(ILK>H U‘°(>U>W° P—l >uumowmI-ch34szu .lU-J hhhh>muoummmmoxtzzzzzmzaammuommmz‘32: WI‘QO‘OHWDC’W‘DI‘QG‘O FlNlOQ'IDWI‘QGOHSUlOQ’ID‘DI‘QO‘OH nnnno C'VC'G’C’Q’Q‘Q‘G’IDIDIDIDIDIDIDIDIDIDOWQQQ‘D‘DWW‘DFF STATE BR1960 8R1961 BR1962 BR1963 BR1964 BR1965 BR1966 CNW 1m ¢< < << an: < <‘ I!!!“ ( ¢<<<<< <t>hhm £2 Hzuo<fiu4¢ouo¢m Hmh<<> ¢ ¢Ht cox< m: z 4>N<< (umuuu: 0‘ OHNI’O C’ID 0‘ GO GO GO Hv-Iq-IHHI-i 179 << C(< ‘ < <(<¢‘¢<(<<<<< ¢ <<<< 11¢ ( < ‘ < <( ><<<<3HuU¢n ZOUU‘I<€D.HD 42‘10-2)KLU4hJ‘P<£2l-O.HQI¥£24 HhJuLL¢ZOcxxuXH¢¢04HO¥O¢I¢OUH4 HO-¢tu 2&2 ¢Qo>mmOh3CLUIDU¥Z mm3¢4MOH4l—>4JO:I zc z o o 2 come h ILHuJ-QUO-l—H HZNCDOUII 0.40-34 22 Oman/122F040) QO<20Uh4H u42w¢w4200 440 JHQOZ<>zzoom7533717317772>>>>>r>>>>>>uuu tzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz m < h m>o Unuo>m4xaHm >o¢3><41m 20 <>H4 $01 001: ommm> azuu<>uu4¢ zwzmmzuzzautzmoHtx< th¢hw3I u4o4¢tz<<DI-¢H(<(OHDH ¢>mmw44zw¢ mzzu4>>zzm2xom D<ZO4m4u4 < oa—zuz (m-r-mcun‘t-au: uuuum>m< CWOQHHHZ-lz 4:22040 HUWCCWHZMH DHH4IZI>UHH PUHUOIFU¢4ZQJhDC¥4XP>I¢DUI¢HDP¢ZQXQI4¢P ¢4¢4PFMH¢NIZHl~C¢44¢JOK<4UUmm3uumzmmm0011xtz<<< \DNQO‘OHNNCW‘OT‘QO‘DHNDCDWN CDU‘OF‘NVOC'ID‘UFQU‘OF‘NVOG' ¢¢ Q'C'IDIDIDIOIDIDIDIDIDIDQQQ Q‘OQOVDVD‘OFFN FFFFFFFQQQQQ NNNNNNNNNNNNNNNNNNNNNNNNNNNNN NNNNN NNNNN STATE BR1960 BR1961 BR1962 BR1963 BR1964 BR1965 BR1966 GUY MM C C CC C C C C C a: C C CC CC C CKC c44m4umm44wzmhz 4¢42H 4¢D4¢zm< WWPPFPJ mouwncm OHHfiHHH nnnnnnn 185 C C C C «amt: C C CCCK a: C CCC“ CC C CCCC CZ22CCCCCCCCCCCQZCZCCCCCCZZCCCCCCCCCCCC C C C C MIKMC C C CCCK C C CC¢¢KCCC C CCCC CZZZCCCCCCCCCCCQZCZCCCCCCZZCCCCCCCCCCCC C C Cx C C KKKC C CC CCKIKCCC C CC CC CZZZCCCCCCCCCCCQZCZCCCCCCZZCCCCCC CCCCCC C C C CCC CMKC C CCCKZM C KCCOCKCCC C CCCC CCZZCCCC CC CCCCCQZZZCCCZCCZZCCC CCC CCCCC C C C C CCC CKOCC C CCCKKK C ¢CC¢¢CCC C CCCC CCZZCCCCCCCCCCCQZZZCCCZCCZZCCCCCCCCCCCC C C C CCC CKIKC C CCC¢¢¢ C KCCCMCCC C CCCC CCZZCCCCC‘CCC CCmZZZCCCZCCCZCCC CCC CC (CCC < C C CCC CZKC C C(CCKCKK C CCCCKCCC C CCCC CCZZCCCCCCCCCCCQZZZCCCZCCCZCCCCCCCCCCCC xt-O-l-l—>>>HHHHHHHHHHHH FDDDD>>>>>>>>>>>3>333321313123311333333 hub 4mm 339 x 00 O-t-cncbz H<4 u I4 1 C xHZ433meZuul—H2uu mmmzo (mHszxI o .— (MUUQMAUJLOxCPIC—lZCCuZKHO-PA >.LU)ODVJMJH H2 ngr-Jzzoazm:ouuu4hUOUUHOC>MHMQux¢mmh>oza3¥O H4 04 >- 02 ocDUZt-O IOCX 1408th CHOU CO H“ PZCU>UO4¢HZFO HmHOuJCDI-K ‘Hom; mnuom4wF4CxHUZmOZQXEZ>U ZH—I‘IDOOD‘IDCH ZCCKOOOOCC ZXHC (mu.c9:x:zo—<&I-u.4z<¢nu.u.44a:tzoooxm HNIOC'ID \ohwmo—owncmohmmodwncmmhm «own-0 HHHHH HHNNN NNNNNN C HCC C CCCCCCC C HCH C CCCCCCC C HCH C CCCCCCC C FTCH C CCCCCCC C CH C CCCCC CC C C C CCCCC CC C C C C((CCCC Z :mmcoqmzm .J>- QJ>¢ZPUOI¥¢DZGCP CD¢>OZ¢ OCCUPVQCUQCJUOUOKQUH>ZQAUCDUZKC ZDMUJJO: UIKthaamch—ICDI—DILOKCOHCLL¢>H UCOC>M>WO UJO—J >ummowmhumrqaztu- HAD-I t—PPPCUOUUFKDKKCCUJJJOOPC HOCCUCI-b-COCU mmmm>>muomwammoztzzzzzmzaammuommwz'at \DF QO‘OHN [0610 ohmefiNnc-mohmmonwncmohmmon m0 nnq- d'd' Cd'd- c-d' c-crmmmmmmmm mmmooooomm \Dmhh STATE BR1967 BR1968 BR1969 BR1970 BR1971 BR1972 BR1973 CNW um '4 C C HF! C (KC CHv-I C CCCH H CCF‘CC CCCmCszCCCCCCCCCCCCCCCCCCC H C C HH C CC CHH C CCCH CCHCC CCCmCszCCCCCCCCCCCCCCCCCCC < C «a < ¢¢ CHH C CC <4 CHCC CCCmCszCCCCCCCCCCCCCCCCCCC C C HH C ¢< CHH C C((H CCHCC CCCmCCZmCCCCCCCC<<a:>-r-o-m ‘23 H403UJ¢HUQC HZUWCCWJZZWZZZ-IUZQCl-ZCH LL22 tCQC—IQDCCCDDOCOWOIUCHtmxxHC C-I mazhmmaoau>2v~o<~t04¢ouu¢m HKFCCCH JPDOCCQUJIIUQOUOOQK ICON/1h...- ICCC UIWI GmUUDIZQ mow): wm¢mwv~ma~°HNn¢m 0":an HN'OC'IDCJI‘Q NFFFFBFFQQQQQQQQQQU‘U‘O‘C\O\U‘OU‘U‘ C C CCC n CCCCCCC C C CCC H CCCCCCC < C CCC H CCZCCCC C C CCC H- CCCCCCC’ C C CCC H CCCCCCC C C CCC CCCCCCC C m CCC zCCCgCC 2222222 HHHHHo-u-O zu Au OCth 0 WU¢>> C 00*ng o U112)»: OUXC xx 2 A>h< C CUMULLLDI 0‘ QHNIO C'ID mcococo HHHHdH 189 H CC CCC C C C CCCCCCCCCCCCCHCH CCCCH HCC C CC C CCCCCCCCCCCCCCCCCCCDQQCCCCCCmCCIDCCCCCC CC CCC C C CCCCCCCCCCCCCKCH CCCCH HCC C C C C CCCCCC CCCCCCCZCCCCmmmCCCCCCQCCCCCCCCC CC CCC C C CCCCCCCKCCCCCKCH CCCCH HCC C C C CCCCCCCZCCCCCZCCCCmmmCCCCCCmCCCCCCCCC CC CC C C CCCCCKCCCCCCCHCH CCCCH HCC C C C C CCCCCZCCCCCCCCCCCCmmmCCCCCCmCCCCCCCCC CC CCC C C CCCCCmCCCCCCCHC CCCCH c-QCC C C C C CCCCCZCCCCCCCCCCCCmmmCCCCCCmCCCCCCCCC CC CCC < C CCmCCcCCCCCCC C CC CC CC C C C C CCZCCZCCCCCCCCCCCCmmmCCCCCCmCCCCCCCCC CC CCC C C CCCCCmCCCCCCC C CCCC CC C C C C CCCCCZCCCCCCCCCCCCmmmCCCCCC(DCCCCCCCCC ZZZZZZCCCCCCCwmw>>CCCCCuumoCDHuum ‘UUUZJLCCCLHD AZCP£>ZIUJUZFCiZr-CLHCDIXZZ-l Ct>nmmm¢zooxmuxHCmCUO¢>mmJHoxo¢IxouHJ HOCQma: MIDDummx40-tx momaHhhhuUJUU>th (:sz mag>mm0hzmuzzuxz mmJ¢4w0H4h>4103 (C(DOMNMCUDHCCOHWOJCO)IDmOCoxxCHCOCuJO Hx.11:mo—uuooamzat-JCACAIZmCammmmummIz4.1.: whmmOv-‘N'O can WQOOHNVO «n \Ul‘ma‘Oc-INVO can \DFOO‘OF‘N OOOOdHHHHHw-OHc-co-GNNNNNN Nwwmnnnnnnnnnnccc dddfidHHHfiAdHHHHHc-OHHH HHv-u-cv-c «av-0.4.4.4 HHfiq-‘flv-OH STATE BR1967 BR1968 BR1969 BR1970 BR1971 BR1972 BR1973 GUY um H a C CC CC CCC CCCHCCCHCCCCCCmCHCCCCCCKH CCCCCCCmCCCCCC CZCCCmCmmmZC C CC CC C C CCHCCCc-OCCCCC CMCCCCCC C011 CC CCCCC CDCCCCCC CZCCCmCm CZZC H C CCC KCCCCCC ZCmCmmm C CC C CCCHCCCF‘CCCCC CKC CC CC CCCCC QCCCCCC CZCCC H C CC CC C CCC CCCHCCCHC CCCCCCCCCCCCCCCC CCCCCCCmCCCCCC CZC CCmCmCmmm H C C CC CCC CCCHCCCH‘CCCCCKCCC-t-_JH UJZmaaDJzoo .140 )HQozCCu AcCC>I1777777777777715I>>>>>>>>>>>>UUU ZZZZZZZZZZ2222222222222222222222222222 m C P Q>O OCH m 224 oz P XUU FM th 42 Q m PPZ KJ ZUOCHZZU to KUF UM 46h A; U UC>HNO>QJ¥DHW >0CD’CJCOXW ND- H20 CCC) ID¢U@FU H¢¢C¢IUFJGZICm>IU 20 C>HJ k0: GUI: OKWQ>CQMUP¢ZUDC@k> @IUUC>UUJ¢ ZCmZmedHHmZJJ mh¢>U2002LZJDUII¢°HI¢C thCwCCWPKJUOHUJC¢PWDI OJOJCIZCCDUfiszzmllmh>CCmmU2E¢¢meDCmu oFQmOCNnCmohwmonwncmmhwmonmncmmhmmcnwn FFFFQQOQQQQQQQ¢0$$O$OQOOOOGOOOOQOOHHHH fififlddfififlddfifiddddfidddHHfiHNNNNNNNNNNNNNN STATE BR1967 BR1968 BR1969 BR1970 BR1971 BR1972 BR1973 CITY 192 CH CCCHCF‘H CC CCKH HHKCF‘ CCCCCCCCCCCCCCCCC2CCCC2CC CH CCCo-nCv-h-c CC CC H HHgCo-i CCCCCCCCCCCCCCCCCCCCCCZCC CH CCCHCHfi CC CC H HHHC CCCCCCCCCCCCCCCCCCCCCCCCC CH CCCHCHH CC CC H v-h-OHC CCCCCCCCCCCCCCCCCCCCCCCCC CH CCCHCHH CC CC Hazv-u-ov-oC CCCCCCCCCCCCCCCCCCCZCCCCC C CCC C CC CC 0: H C CCCCCCCCCCCCCCCCCCCZCCCCC C CCC C CCCCC C CCCCCCCCCCCCCCCCCCCCCCCCC UUUUUUUUCDOIIIIIIIIIIIIII 22222222222000OOOOOOOOOOO >COH g—ux a) C20) 2 .JCOH meOIoJU .2 2C: . O uo—t-u. zr-‘UHLQZCK a: zzamzCr- Zuni-um Cl—OZ HHmCOOZOHUtOH—I Hum—l2 :uo—uruz touchu>3hmHCCmHoH a:>-mmw.n_|zm¢ ¢22UJ>>ZIKZ¢O¢ DCCKHCHHHCKXCHJOCJCHOCCHO- gnawIzzjmmmCuUUUQUI—lJIIZm C’ IDWFWU‘ CHNHCMQI‘NU‘OH “"0010 QFQ H HF. F004"! NN NNNNN NNNF‘H’) KNOWN) MOON) N NN NNN NN NNNNN NNNNN NNNN NNN K C CZCCC CC C C CCCCC CC C C CCCCC CC C C- CCCCC CC' C C' CCCCCCC- C C CCCCCCC C C CCCCCCC STATE BR1967 BR1968 BR1969 BR1970 BR1971 BR1972 BR1973 CITY 192 CFO CCCHCHH CC CCmv-O He-«sz-I CCCCCCCCCCCCCCCCCZCCCCZCC CH CCCHCu-u-o CC CC H «.4ng CCCCCCCCCCCCCCCCCCCCCCZCC CH CCCv-flCv-IH CC CC H HHHC CCCCCCCCCCCCCCCCCCCCCCCCC CH CCCHCHH CC CC H v-Iv-Qv-‘C CCCCCCCCCCCCCCCCCCCCCCCCC Cv-O CCCHCHH CC CC Hmv-«HHC CCCCCCCCCCCCCCCCCCCZCCCCC C CCC C CC CC 0: H C CCCCCCCCCCCCCCCCCCCZCCCCC C CCC C CCCCC C CCCCCCCCCCCCCCCCCCCCCCCCC UUUUUUUUQOOIIIIIIIIIIIIII 22222222222000OOOOOOOOOOO >COH n—ux a) C20) 2 .JCOH UHIDOICDJU 2 2C: . O ut—t-u. ZD—ZWILOZCO: ¢ zzamZCo- th—mm CI-OZ HHmCOOZOHUZOH-l O-HLUJZ IMO—MIMI: IOmOhU>DhmHCCmHQH ¢>muwaqzm¢ mzzu4>>tx¢zmo¢ DCCQCHCHHHCKXCH—IOCJCHOCCHQ amwOImnzmmmCuuuuouIAJtttm C mmhmmoawmo-mohwmo—o NWC'ID \OFQ F. HF. “HHNN NNNNN NNN'O'O'OWIOIO [0'0” N NNNNN NN NNNNN NNNNN NNNN NNN I: C C2CCC CC C C CCCCC CC C C CCCCCCC C C- CCCCCCC' C C' CCCCCCC- C C CCCCCCC C C CCCCCCC mOHNF1¢ID ”CC CCQ'C' NNN NNNN KB «I! H H C CCCC CC C CCKC dCCCCCCCHv-OCC C o-o HCv-Ia: KCHC H CC CC CCZCCCCmCmmCmCCmmCCCCCCCCzCZCCmCCCCCCCC H H H C CCCC CC C CCIKC v-OCCCCCCCHHCCKC H Fle-Ig “Co-0‘ H CC CC CCZCCCCmCmmCmCCQGZCCCCCCCzCZCCmCCCCCCCC a a H a C C C CC C C CCCC HCCCC CCHH C¢C a HCH¢ CHC H CC CC CCmCCCCmCmCCmCCCmZCCCCCCCzCCCCmCCCCCCCC -o «I c-c .4 H .4 C C C CC C C C CCCH ”CCCC CCo-h-I CKC HCHCv-ca: Cv-OC H CC CC CCmCCCCmCmCCmCCCQZCCCmCCCZCCCCmCCCCCCCC c-c H H H F. C C CC C C CCCH HCHCCmCCv-u-a C C HCHCc-m: CHC v-O CC CC (CmCCCCCCQZCCDCCCmCCCCGCCCZCCCCQCCCCCCCC C C C C C C CCC C C CxC C C C C C C C CC C CCmCCCCCszCCCCCmCCCCmCCCzCCCCmCCCCCCCC C C C C C C CCC C CCmC C CC C C a: C C CCC C CCmCCCCCCmZCCCCCCCCCCMCCCZCCCCmCCCDCCCCC KKKCCCCCCCCCCCCCCCCHHHUUUQUQDZZZZZzzxxx ooommmmmmmammagaaamzmxmmmmmmmhhphhphpph o H:- u :DzahJch 2mm uz hCJumHCOr-c «.1 .1 O z ILOC: moor-uauo I XM¥VQHHJZUILZ>ZOJ¢04UJ C wr—zun: cm—n—mumzp—mC uoumm>¢C CmOQHHHZA‘ .JtZZOJO HUMCCWHZUH DHHJZZCPOXPXWW>I>UHH huHuthm¢AZUJhQC¥J¥h>2¢DuImHDhm20xmzth «Am—lI-P-mHCNIZHI-CKJJKJOOCCAUUCQOCCIZOSUP-0C”) OCCLJJUCKCCOCLHNUHHOCmCIOa: (LCHIJOHCUCCDID :1.meCmuu:taamm¢m11>amzuu¢zw¢moufixxzzCCC \DFQO‘OHNWCID‘DFCDU‘OHNWCIDO"QO‘OHNVOC’ID \DT‘Q’O‘OHNDC’ c-c-cc-mmmmmmmmmmooxmosoooo ‘DOFFFFT‘N FFT‘FQQQQQ NNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNN STATE BR1967 BR1968 BR1969 BR1970 BR1971 BR1972 BR1973 CITY HM C C CC CC C C HC HC CFOCCCCCCHCCH C H CmCCCCCmCCmmCmmCCmCCCmCC C C CC CC C CC HC v-iC CHCCCCCCHCCH CC H CmCCCCCmCCmmCmmCCmCCmmCC H C C CC CC C CCC HC CHCCCHCCHCCH CCCH CQCCCCCQCCmmCmmCCmCCmmmC H C C CC CC C CCC C v-oC C CCCHCCHCCH CCCH CmCCCCCmCCmmCmmCCmCCmmmC H C C CC CC C CCC C C C CCCHCCHCCH CCC CGJCCCCCIDCCmm CmmCCmCCmmmC C C CC CC C CCC C a: C C CCC CC C CCC CmCZCCCmCCmmCmmCCmCCmmmC C C CC C C CCC < m C C CC CC C CCC CmCZCCCmCCmmCmmCCmCCmmmC xxxxxxxxxxxxxxxxxxxxxxxx Phhhhhhhhhhhhhhhhhhhhhhh p—H u U) IOU 3 0-0 2) u ZKOO-I-(DZZ U229 «.1 CU) {DWV’UDmmW‘QUQO-OUUZCU Cu 822uDCmgCouHhuo>OdeCD¢ a DJCJQJH a.J>_lV).JuJQm.J_lmZ¢D-2 CO>J¢42H .JKDJIZZCDCOUCCKC UKKOOCMDJI—CCUHCODUHDCID mmmuuoouuuwzzxAJJzzoommm mwfimmDHNnc'mWmeOHNnd'ln \Dhm mmmaocommmmmmmmmmoooooooco NNNNNNNNNNNNNNNnnnnnnnnn C (HHH CCCIDCCC C Cv-h-Ov-I CCCQCCC C HH CCCIDCCC C C v-OH CCCIDCCC- H C H CC>>>HHHHHHHHHHHH F2323>>>>>>>>>>>3>333333):. 3.1-3331333333:- um 4mm Dan I go hhmwz HCJ w u a C tzdazxmozuuo-quu mmmzo CmHCmZXI O P CUOUQUJWIOXCPJCJ‘CCmZKHhhd >Imonmum HZ O—JIl-JIZOGCUDZOUUUJl-Ct:JHMJUHUZWWOWCOZK INS) CLUU U-Ut-IZQIKZII—XOHMD-XUJO UUOmHZXO-flu UOUO—IJZQZJCPIZUC UZUCOUXCZKUQJDUZZUDJIQQ HQ¢¢UOUOHOC>UHU1CCID§IQUCOCCUCCHWC: accumuzzazzm¢¢¢>uxxmmh>uzmszuofiggxtomm \DFQO‘C’F‘NMCWWFQU‘OF‘NMCIDwFQO‘CHN'OQ'ID ‘0" amen-mun d" HHHHNNNNNNNNNNVOWVOMVOWnnnMCCCCCCCCC'C'IDU'MOIDID nnnnnnnnnnnnnnnnnnnnnnn nnnnnnn nnnnnnnnn STATE BR1974 BR1975 BR1976 BR1977 BR1978 BR1979 BR1980 CITY 196 H C C C C Cv-IHC CHHu-MKHCIKCCCCC CH CC (DCCTDCCCCCCCCCZCCZCCCCCCCCCCC H C C C C CHHC CHHdgHCzCCCCC CH CC mCCmCCCCCCCCCZCCZCCCCCCCCCCC H C C C C CHHC HOCHCZHCCCCCCCC CH CC C mCCmCCCCCCCZCZCCZCCCCCCCCCCC .4 C C C C CH C CflMHmHCmCCCCC C". (C OCCmCCCCCCCZCZCCZCCCCCCCC‘CC C C C C C C CF‘ZF‘QCF‘CKCCCCC CH CC QCCmCCCCCCCZCZCCZCCCCCCCCCCC C C C C C Cdmdgu-ICKCCCC C (H CC mCCmCCCCCCCZCZCCZCCCCCCCCCCC C C C C C C CC xdmCCmCCCCC CH CC mCCmCCCC (DCCZCZQCZCCCCCCCCCCC AAJJAAJJxNmexCCCCCCCCCCCCCC ¥O H—l U-l >- :32 Cebqu IOCX “0::th CHOU DO HU O-‘Zu>lu0_l¢H¢l-O HWHOuwt-fl: ZHQUJI WHHQWJQCOZOhth¢&ZmZ(/)U meafiC Himmt-Hi-UIUCDU HIMOfiCDLDCUl-CJCCUO: 2100233201000>PJC¥HUZMQZQXFZ>U ZHJCQUODZIDCH ZCCmoOOOCCZXHC CmmeIZD-CQJ-ILAZCCDILIL—ldttZOOOKU) HNIO can whoa“: “NM CID \DFQO‘O “NF, cm ‘Dl‘m «Huh-I “—1 HHv-h-ON NNNNN NNN C C CCC C CCCCCCC C C CCC C CCCCCCC‘ C C CCC C CCCCC CC~ C C CCC C CCCCCCC C CCC C CCCCCCC CCCCC CCCCCCC CCCCCCC 0000000 0 mm 0 ma: “HACK CCUQOZC 22621Cm HKUC .JUHmZCC CmQLLCI-I- mmmmmmm 0‘ OHNM d’lfl Nnnnn HMO 197 F" H <£ < < < < C (<1 ¢¢¢(¢( H<€H H (<<< ”(1(H(H(H( <<<<Z Damoo¢zhuoI¥¢zzwozm mzuaomu¢mz zawmnua thxuaomu0449»3u0¢u u¢o<>u>wo 040-4 >meomthmIJJJtu~ H.104 hhph>muouwnmmoztzzzzz«Jazmmuoumwzdt whmm Odwncmwhmm‘Dv-‘Nn ¢In whmmo'cwncm whmmc'" nnnm¢¢¢¢cercan:¢mmmmmmmmmm~n~o~n~nm~n~n~o~omt~t~ STATE BR1974 BR1975 BR1976 BR1977 BR1978 BR1979 BR1980 CITY MB HHHH «an (Hfl <<<<<< <<<<<<< <<<<t>hhm 23 HZHO> ¢ ¢HI OU¥< at: g J>P<< «mummux O‘DHNVOG'W moooooo u-Iv-Ic-IHFH-O fl” HH H < «(<4 €(4 ( ( € < < .(<(¢¢((<<<<<(H<( «(><<<<aHuum zoum‘zqdez A‘‘J-U.Jlu¢v-HUmu¢20cxazmxH¢m4HOXO¢I¢OUHJ Ho<0m¢ nuzanumux4mx «OUDHl-r—F-UUJUuJ>-¢IU £3qu «00>(anI—ICLUIDUXZ mmeJwOH4h>JIOI 20(DUUUU 2 D O 2 come 0- uHuJ—muu—u—H HZN‘DOMJ m-r-._IH uJZMMLDJZOU 4.10 JHQO£<117'3‘311'31‘)17511712>>>>>>> >>>>>UUU ZZZ22222222222222222222222222222222222 m < h m>c uo<3><4IU ZU <>HJ mu: cox: oumo> mzmu<>uu4¢ ZCWZZWQJHHKZJ: wh¢>uzmmzuzzaozI¢oHI¢< PHI(u<‘WhmJUOHUJ<(“(deJHDJUOOOU>¢PWDI mac4¢12<4m034222mummh>(«mmmzammmmh3DFKH<~mmcp.l_12ma: m22u4>>zx¢2¢a¢ D<<¢H204m4u4 C (DO-‘mz VJPF-mow‘t-UJC UQQwD>CC CWOQHHHZJZ 43220.10 Huw<1>UU-0H hunOIPumAZQAhOJmDuImnahxzoxmxdmh «AmqhhmmmNItham3uuczmxm001xxt2<<< mpmmuHNncmmhmmOHNn¢mohmmUHNn¢mmpmmOHwnc tcccmmmmmmmmmmmooomooomohhhphhhhphmmmmm NNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNN STATE BR1974 BR1975 BR1976 BR1977 BR1978 BR1979 BR1980 CNN flM H C C C C C C HCHHCC HCC C CHC C HCHH CmCCCCCCCCmCCCmCCCC CC (DCC H < C C C < HCHHCC H< C (H4 < H¢ H OJCUJ¢¢ ‘3 DJCJQJHZQJ>JWJUQQJJWZMP2 CO>-.J¢_.IZH .11 KD-JKZGJCOUCCKC UKKOOCUC-fl-CCQHCODUHDmIo mmmUUQOUUKQIIXJJJZIDOQQU) murmmoHqu-m ohmmOH NI‘JC'ID ohm cocoaoaoaoasmmmosasmmmmcc cocoa oo NN NNNNNNNNNNNNNnnnnnnnnn C HHH CCCCCC¢ C Flo-0H CCCCCCC C HHH CCCCCCC' C HHH CCCCCCC- ( HHH ((CCCCC' H C C CHHH CCCmCCC H C C CHHH CCCIDCCC xxxxxxx hhhhhhh H 2H 22 CU o>>>HHHHHHHHHHHH r—DDDD>>>>>>>>>>>3>33233133331333313333:l um Hum 330 I Do hhmwz HCJ u u a < xmzdaameZUUPHZMU momzo CmHCWZXI o h CNDUQUJWLOXCPJCJZCszmHFPJ >xmcamuH HZ CD..|It—JI20mm:oquAh-CIZJHLJJUHuzmmOmCOZz zut> (Lulu gut—IZmaczzr-xOHazr-quO mucmHJXHl-J UOUOJJIQZIKFMUC MZUCOUXszlJQADuZZUQJIUO. Hui¢¢COC0>UOMOHOC>MHMQCCIDCIQUCCCUCCHVJCD 3000400IIJZZQQIK¢>U¥¢VJWP>UIOJICGJUQ‘DXJttOM(D \DFCDm9HNF)¢m\D'~ma\°HONOC'IDQFQO‘OHN'OC'IDsoFQU‘Ov-INN3Q' HHHHmchNNNNNNNnnnnnnnnnnq-c- c-c-cc-c-c-c-c-mmmmm mnnnnnnnnnnnnnnn nnnnnnnnnnnnnnnnnnnnnnn APPENDIX B CENSUS REGIONS AND CENSUS DIVISIONS Census Region Northeast North Central South West 206 APPENDIX B Table A2. CENSUS REGIONS AND CENSUS DIVISIONS Census Division New England Middle Atlantic East North Central West North Central South Atlantic East South Central West South Central Mountain Pacific Component States Maine, New Hampshire, Vermont,‘Massachusetts, Connecticut, Rhode Island New York, New Jersey, Pennsylvania Ohio, Michigan, Indiana, Illinois, Wisconsin Minnesota, Iowa, Missouri, Kansas, Nebraska, South Dakota, North Dakota Delaware, Maryland, Virginia, west Virginia, North Carolina, South Carolina, Georgia, Florida Kentucky, Tennessee, Mississippi, Alabama Louisiana, Arkansas, Oklahoma, Texas Mbntana, Wyoming, Idaho, Nevada, Utah, Colorado, Arizona, New Mexico Washington, Oregon, California, Alaska, Hawaii BIBLIOGRAPHY BIBLIOGRAPHY Advisory Commission on Intergovernmental Relations (1977). Trends in Metropolitan America. Washington, D.C.: U.S. Government Printing Office, February. Allardice, David R. (1974). "Municipal Bonds in the Housing Market," Economic Perspectives, July/August, pp. 16-23. Amling, Frederick (1974). Investments: An Introduction to Analysis and Management. 3rd ed. Englewood Cliffs, N.J.: Prentice-Hall. Aronson, J. Richard, and Eli Schwartz, eds. (1975). ”Management Policies in Local Government Finance. Washington, D.C.: International City Management Association. Bahl, Roy W. (1971). "Measuring the Creditworthiness of State and Local Governments: Municipal Bond Ratings," Proceedings of the National Tax Association, pp. 600-622. Bahl, Roy W.; Bernard Jump, Jr.; and Larry Schroeder (1979). "Federal Policy and the Fiscal Outlook for Cities" in L. K. Hubbell, ed., Fiscal Crisis in American Cities. Cambridge, Mass.: Ballinger Publishing, pp. 3-26. Bannister, Geoffrey (1976a). "Space-Time Components of Urban Population Change," Economic Geography 52: 228-240. Bannister, Geoffrey (1976b). "Towards a Model of Impulse Transmissions for an Urban System," Environment and Planning A 8: 385-394. Bennett, R. J. (1974). "Process Identification for Time Series MOdelling in Urban and Regional Planning," Regional Studies 8: 157-174. Bennett, R. J. (1975). "The Representation and Identification of Spatio- Temporal Systems: An Example of Population Diffusion in North West England," Transactions of the Institute of British Geogra- phers 66: 73-94. Berry, Brian J. L. (1980). "Inner City Futures: An American Dilemma Revisited," Transactions of the Institute of British Geographers 5: 1-28. Boddy, M. J. (1976). "The Structure of Mortgage Finance: Building So- cieties and the British Social Formation," Transactions of the Institute of British Geggraphers 1: 58-71 207 208 Brunn, Stanley D., and Donald J. Zeigler (1979). "The Spatial Correlates of Municipal Bonds: A Geography of Assigned Credit Ratings," Paper presented at the 75th annual meeting of the Association of American Geographers, Philadelphia, Pa., April 23. Burke, D. Barlow, Jr. (1968). "Rating General Obligation Bonds," Municipal Finance 40 (May): 169-176. Calvert, Gordon L., ed. (1965). Fundamentals of Municipal Bonds. 4th ed. Washington, D.C.: Investment Bankers Association of America. Calvert, Gordon L., ed. (1969). Fundamentals of Municipal Bonds. 7th ed. Washington, D.C.: Investment Bankers Association of America. Campbell, Colin D. and Rosemary (1978). An Introduction to Money and Banking. 3rd ed. Hinsdale, Ill.: Dryden Press. Carleton, Willard T., and Eugene M. Lerner (1969). "Statistical Credit Scoring of Municipal Bonds," Journal of Money, Credit, and Banking 1 (November): 750-764. Christy, George A., and John C. Clendenin (1978). Introduction to In- vestments. New York: McGraw-Hill. Cox, Kevin R. (1978). "Local Interests and Urban Processess in Market Societies," in his Urbanization and Conflict in Market Societies. Chicago: Maaroufa Press, pp. 94-108. Darden, Joe T. (1977). "Redlining: A Concept for Study by Urban-Social Geographers," in H. A. and M. K. Winters, eds., Applications of Geographic Research. East Lansing: Michigan State University, Department of Geography, pp. 29-44. Darden, Joe T. (1980). "Lending Practices and Policies Affecting the American MetrOpolitan System," in S. D. Brunn and J. 0. Wheeler, eds., The American Metropolitan System: Present and Future. New York: Wiley. Davis, Edmund H. (1958). Of the People, By the People, For the People. Chicago: John Nuveen and Company. Dingemans, Dennis (1979). "Redlining and Mortgage Lending in Sacramento," Annals of the Association of American Geographers 69 (June): 225-239. Dougall, Herbert E. (1973). Investments. 9th ed. Englewood Cliffs, N.J.: Prentice-Hall, pp. 81—108. Drott, Peter E. (1971). The Professional Investor's Guide to the Muni- cipal Bond Market. Memphis, Tenn.: Municipal Market Publishing. 209 Ellinwood, David M. (1957). "Factors in Municipal Bond Ratings," Public Management 39 (April): 74-77. Forbes, Ronald W., and John E. Petersen (1976). "Background Paper," in Twentieth Century Fund Task Force on the Municipal Bond Market, Building a Broader Market. New York: McGraw—Hill, pp. 27-174. Garcia, F. L. (1962). Glenn C. Munn's Encyclopedia of Banking and Fin- ance. 6th ed. Boston: Bankers Publishing Company. Greenberg, Ronald David. (1977). "Municipal Securities: Some Basic Principles and Practices," The Urban Lawyer 9 (Spring): 338-363. Gup, Benton E. (1979). The Basics of Investing. New York: John Wiley & Sons. Haggett, Peter; Andrew D. Cliff, and Allan Frey (1977). Locational Analysis in Human Geggraphy. 2nd ed. New York: Halsted Press. Harvey, David, and Lata Chatterjee (1974). "Absolute Rent and the Struc- turing of Space by Governmental and Financial Institutions," Antipode 6: 22-36. Hastie, K. Larry (1972). "Determinants of Municipal Bond Yields," Journal of Financial and Quantitative Analysis 7 (June): 1729- 1748. Hempel, George H. (1967). Measures of MMnicipal Bond Quality. Ann Arbor: Bureau of Business Research, University of Michigan. Hirsch, Werner Z. (1971a). "The Fiscal Plight: Causes and Remedies," in Fiscal Pressures on the Central Ci_y. New York: Praeger, pp. 3—40. Hirsch, Werner 2., ed. (1971b). Fiscal Pressures on the Central City. New York: Praeger. Hoffland, D. L. (1972). "Price-Rating Structure of the Municipal Bond Market," Financial Analysts Journal 28 (March): 65-70. Horton, Joseph J., Jr. (1969). "A Statistical Rating Index for Munici- pal Bonds," Financial Analysts Journal 25 (March-April): 72-75. Horton, Joseph J., Jr. (1970). "Statistical Classification of Municipal Bonds," Journal of Bank Research, Autumn, pp. 29-40. Horrigan, James 0. (1966). "The Determination of Long-Term Credit Stand- ing with Financial Ratios," in Empirical Research in Accounting, Selected Studies, 1966, Supplement to Journal of Accounting Re- search 4: 44-62. Hubbell, L. Kenneth, ed. (1979). Fiscal Crisis in American Cities: The Federal Respgnse. Cambridge, Mass.: Ballinger Publishing. 210 International City Management Association (1977). Municipal Year Book 1977. Washington, D.C.: International City Management Association. Kirk, John (1967). "What's What in the World of Municipal Bond Ratings?" Banking, August, pp. 8-10. King, Leslie; E. Casetti, and D. Jeffrey (1969). "Economic Impulse in a Regional System of Cities: A Study of Spatial Interaction," Region- al Studies 3: 213-218. Liu, Ben-Chieh (1973). The Quality of Life in the United States 1970. Kansas City, Mo.: Midwest Research Institute. Liu, Ben-Chieh (1975). Quality of Life Indicators in the U.S. Metropoli- tan Areas, 1970. Kansas City, Mo.: Midwest Research Institute. Liu, Ben-Chieh (1976). Quality of Life Indicators in U.S. Metropolitan Areas: A Statistical Analysis. New York: Praeger. Madrick, Jeffrey (1977). "Municipal Research is a Growth Industry," Business Week, March 14, p. 81. Matteson, Archibald C. (1968). "In Defense of General Obligation Munici- pal Bond Credit Ratings," Financial Analysts Journal 24 (July): 99- 103. Mendelson, Morris, and Sydney Robbins (1976). Investment Analysis and Securities Markets. New York: Basic Books. Michel, Allen J. (1977). "Municipal Bond Ratings: A Discriminant Analy— sis Approach," Journal of Financial and Quantitative Analysis 12 (November): 587-598. Moak, Lennox L. (1970). Administration of Local Government Debt. Chicago: Municipal Finance Officers Association. Moody's Investors Service, Inc. (1960-1980). Moody's Bond Record: Cor- porates, Convertibles, Governments, Municipals. New York: Moody's Investors Service. Moody's Investors Service, Inc. (1974). Pitfalls in Issuing Municipal Bonds. New York: Moody's Investors Service. Moody's Investors Service, Inc. (1977). Moody's Analytical Overview of 25 Leading U.S. Cities. New York: Moody's Investors Service. Moody's Investors Service, Inc. (1978). Moody's Municipal and Government Manual 1978. New York: Moody's Investors Service. Moody's Investors Service, 1km; (1979). Moody's Municipal and Government Manual 1979. New York: Moody's Investors Service. 211 Morton, T. Gregory (1975-76). "Comparative Analysis of Moody's and Stan- dard and Poor's Municipal Bond Ratings," Review of Business and Eco- nomic Research 11 (Winter): 74-81. Morton, T. Gregory, and Dennis McLeavey (1978). "A Cluster Analysis of Municipal Bond Ratings," Review of Business and Economic Research 13 (Spring): 92—99. Mueller, John H.; Karl F. Schuessler; and Herbert L. Costner (1970). Statistical Reasoning in Sociology. 2nd ed. Boston: Houghton Mifflin. Muller, Thomas (1975a). "The Declining and Growing Metropolis--A Fiscal Comparison," in G. Sternleib and J. Hughes, eds., Post-Industrial America: Metropolitan Decline and Inter-Regional Job Shifts. New Brunswick, N.J.: Rutgers University, Center for Urban Policy Re- search, pp. 197-220. Muller, Thomas (1975b). Growing and Declining Urban Areas: A Fiscal Comparison. Washington, D.C.: The Urban Institute, November. Muller, Thomas, and Grace Dawson (1973). The Impact of Annexation on City Finances: A Case Study in Richmond, Virginia. Washington, D.C.: The Urban Institute, May. Muller, Thomas, and Grace Dawson (1976). The Economic Effects of Annexa- tion: A Second Case Study in Richmond, Virginia. Washington, D.C.: The Urban Institute, April. Nie, Norman H.; C. Hadlai Hull; Jean G. Jenkins; Karin Steinbrenner; and Dale H. Bent (1975). Statistical Package for the Social Sciences. 2nd ed. New York: McGraw-Hill. Newspaper Enterprise Association (1979). The World Almanac and Book of Facts 1980. New York: Newspaper Enterprise Association. Norton, R. D. (1979). City Life-Cycles and American Urban Policy. New York: Academic Press. Packer, Stephen B. (1968). "Municipal Bond Ratings," Financial Analysts Journal 24 (July): 93-97. Petersen, John E. (1974). "Background Paper," in Twentieth Century Fund Task Force, The Rating Game. New York: The Twentieth Century Fund, pp. 23-151. Petersen, John E. (1979). "Current Research in State and Local Govern- ment Debt Policy and Management," Governmental Finance 8 (March- June): 45-48. Peterson, George E. (1976). "Finance," in W. Gorham and N. Glazer, eds., The Urban Predicament. Washington, D.C.: The Urban Institute, pp. 212 Phillips, Philip D., and Stanley D. Brunn (1978). "Slow Growth: A New Epoch of American Metropolitan Evolution," Geographical Review 68: 274-292. Finches, George E., and Kent A. Mingo (1973). "A Multivariate Analysis of Industrial Bond Ratings," Journal of Finance 28 (March): 1-18. Pogue, Thomas F., and Robert M. Soldofsky (1969). "What's in a Bond Rating?" Journal of Financial and Quantitative Analysis 4 (June): 201-228. Pred, Allan (1977). Cityj§ystems in Advanced Economies. New York: John Wiley. Proctor, Mary, and Bill Matuszeski (1978). Gritty Cities. Philadelphia: Temple University Press. Rabinowitz, Alan (1969). Municipal Bond Finance and Administration: A Practical Guide to the Analysis of Tax-Exempt Securities. New York: Wiley--Interscience. Reilly, F. K., and M. D. Joehnk (1976). "Association Between Market- Determined Risk Measures for Bonds and Bond Ratings," Journal of Finance 31: 1387-1403. Riehle, Robert C. (1968). "Moody's Municipal Ratings," Financial Analysts Journal 24 (May): 71-73. Rose, Sanford (1975). "The Trouble With Municipal Bonds is not Just New York," Fortune (December): 104-107, 176-182. Rubinfeld, Daniel (1971). "Credit Ratings, Bond Defaults and Municipal Borrowing Costs: A New England Study," Proceedings of the National Tax Association, pp. 331-352. Rubinfeld, Daniel (1973). "Credit Ratings and the Market for General Obligation Municipal Bonds," National Tax Journal 26 (March): 17-28. Sale, Kirkpatrick (1975). Power Shift: The Rise of the Southern Rim and Its Challenge to the Eastern Establishment. New York: Random House. Sanders, Heywood T. (1979). "Governmental Structure in American Cities," in Municipal Year Book 1979. Washington, D.C.: International City Management Association, pp. 97-110. Smith, David M. (1973). The Geogrsphy of Social Well—Being in the United States. McGraw-Hill Problems Series in Geography. New York: McGraw-Hill. Smith, Wade S. (1979). The Appgaisal of Municipal Credit Risk. New York: Moody's Investors Service. Steiss, Alan Walter (1974). Models for the Analysis and Planning of Urban Systems. Lexington, Mass.: Lexington Books. 213 Steiss, Alan Walter (1975). Local Government Finance: Capital Facili— ties Planning and Debt Administration. Lexington, Mass.: Lexing- ton Books. Stevenson, Richard A., and Edward H. Jennings (1976). Fundamentals of Investments. St. Paul: West Publishing Company. Stewart, Jane (1977). Municipal Bonds, Municipal Finance. Exchange Bibliography no. 1339. Monticello, Ill.: Council of Planning Librarians. Sullivan, Patrick (1976). Municipsl Bond Ratings Viewed as Implicit Grant/Tax Mechanisms. Occasional Paper no. 30. Syracuse, N.Y.: Maxwell School of Citizenship and Public Affairs, Syracuse Univer- sity. The Tax Foundation (1979). Facts and Figures on Government Finance. 20th ed. Washington, D.C.: The Tax Foundation. Terrell, Henry S. (1971). "The Fiscal Impact of Nonwhites," in Fiscal Pressures on the Central City. New York: Praeger, pp. 144-240. Twentieth Century Fund Task Force on Municipal Bond Credit Ratings (1974). The Rating_Game. New York: The Twentieth Century Fund. Twentieth Century Fund Task Force on the Municipal Bond Market (1976). Building a Broader Market. New York: McGraw-Hill. Tyler, Walter H. (1962). "Municipal Bond Ratings," Michigan Municipal Review 35 (November): 285-287. U.S. Bureau of the Census (1944-1977). Census of Governments. Washing- ton, D.C.: U.S. Government Printing Office. U.S. Bureau of the Census (1962). County and City Data Book 1962. Washington, D.C.: U.S. Government Printing Office. U.S. Bureau of the Census (1967). County and City Data Book 1967. Washington, D.C.: U.S. Government Printing Office. U.S. Bureau of the Census (1972). Census of Population: 1970, Vol. 1, Characteristics of the Population, part A. Washington, D.C.: U.S. Government Printing Office. U.S. Bureau of the Census (1973). County and City Data Book 1972. Washington, D.C.: U.S. Government Printing Office. U.S. Bureau of the Census (1976). Population of Urbanized Areas Estab- lished Since the 1970 Census, for the United States: 1970. Sup- plementary Report, 1970 Census of Population. Washington, D.C.: U.S. Government Printing Office. U.S. Bureau of the Census (1978a). County and City Data Book 1977. Washington, D.C.: U.S. Government Printing Office. 214 U.S. Bureau of the Census (1978b). "Estimates of the POpulation of Coun- ties and Metropolitan Areas: July 1, 1975 and 1976,"_Pppulation Estimates and Projection, Series P-ZS, no. 739. Washington, D.C.: U.S. Government Printing Office, November. U.S. Bureau of the Census (1979a). Boundary and Annexation Survey 1970- 1977. Report GE30-3. Washington, D.C.: U.S. Government Printing Office, August. U.S. Bureau of the Census (1979b). "1976 POpulation Estimates and 1975 and Revised 1974 Per Capita Income Estimates for Counties and In- corporated Places," Population Estimates and Projections, Series P-25, nos. 740-789. Washington, D.C.: U.S. Government Printing Office, January. U.S. Congress, Joint Economic Committee (1967-1968). FinancingtMuni— cipal Facilities. Hearings. 2 vol. washington, D.C.: U.S. Government Printing Office. U.S. Congress, Joint Economic Committee (1977). Financing Municipal Needs. Hearings. Washington, D.C.: U.S. Government Printing Office. U.S. Department of Housing and Urban Development (1976). An Evaluation of the Community Develppment Block Grant Formula by Harold Bunce. Washington, D.C.: Department of Housing and Urban Development. Weicher, J. C. (1972). "The Effects of Metropolitan Political Fragmen- tation on Central City Budgets," in D. C. Sweet, ed., Models of Urban Structure. Lexington, Mass.: Lexington Books. West, Richard R. (1970). "An Alternate Approach to Predicting Corporate Bond Ratings," Journal of Accounting Research 8 (Spring): 118-125. West, Richard R. (1973). "Bond Ratings, Bond Yields, and Financial Regu— lation: Some Findings," Journal of Law and Economics 16 (April): White, Anthony G. (1977). Municipal Securities: 1. General Obligation Bonds. Exchange Bibliography no. 1411. Monticello, 111.: Council of Planning Librarians. Wilson, John O. (1969). _Quality of Life in the United States: An Ex- cursion into the New Frontier of Socio-Economic Indicators. Kansas City, Mo.: Midwest Research Institute. Zeigler, Donald J. (1980). "Geopolitical Fragmentation and the Pattern of Growth and Need," in S. D. Brunn and J. 0. Wheeler, eds., The American Mettopolitan System: Present and Future. New York: Wiley. Zeigler, Donald J. (1976). "Selected Quality of Life Indicators and Demographic Characteristics of Standard Metropolitan Statistical Areas in the United States," Unpublished M.A. Thesis, University of Rhode Island.