: · .. •: . . ;· ', ,·.··.· .·· ' •\ ."'' ., . VOTING ON TAX ISSUES IN, MONTANA (1986-1994) . by Jiaping Zhu · A thesis submitted· in partial. fulfillment of the requirements for !he degree . of · Master of Science m Applied Economics · '· MONTANA STATE UNIVERSITY Bozeman, Montana May, 1995 11 APPROVAL of a thesis submitted by Jiaping Zhu This thesis has been read by each member of the thesis committee and has been found to be satisfactory regarding content, English usage, format, citations, bibliographic style, and consistency, and is ready for submission to the College of Graduate Studies. Douglas J. Young ·:\)~~f ....,, (Signature) ,!:;-- ;13-7~ Date Approved for the Economics Department Clyde Greer R.~ (Signa e) Approved for the College of Graduate Studies Robert Brown iii STATEMENT OF PERMISSION TO USE In presenting this thesis in partial fulfillment of the requirements for a master's degree at Montana State University-Bozeman, I agree that the Library shall make it available to borrowers under rules of the Library. If I have indicated my intention to copyright this thesis by including a copyright notice page, copying is allowable only for scholarly purposes, consistent with "fair use" as prescribed in the U.S. Copyright Law. Requests for permission for extended quotation from or reproduction of this thesis in whole or in parts may be granted only by the copyright holder. ~J I Pvf2'0~ l .·-·71 Signature (}f\/v' Date J- - 2 L/ - r j' . ------------------~-------- i· iv ACKNOWLEDGMENTS The author gratefully acknowledges the help and guidance of Dr. Douglas Young. His abundant patience, understanding and encouragement added tremendous help to the completion of this thesis. I'd also like to express my thankfulness to my fellow graduate students for their friendship and the many hours of laughter they provided, making a most tough time more bright and colorful. v TABLE OF CONTENTS Page LIST OF TABLES .......... ; ........................................ vii LIST OF FIGURES .................................................. viii ABSTRACT ........................................................ ix 1. INTRODUCTION .................................................. 1 2. THEORETICAL FOUNDATION ...................................... 8 Tax Incidence and Burden Shifting . 1 • •••••••••••••••••••••••••••••••• 8 Property Tax Incidence ....................................... 10 Income Tax Incidence ........................................ 19 Sales Tax Incidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 Government Expenditure ......................................... 31 Voting and the Median Voter Model. ............................... 33 3. ECONOMETRIC SPECIFICATION AND DATA ....................... 35 Econometric Specification ....................... :............... 35 Regression Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 Heteroskedasticity and Weighting .............................. 36 Cross Equation Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 7 Data and Specific Hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 8 Dependent Variables ........................................ 38 Independent Variables and Specific Hypotheses . . . . . . . . . . . . . . . . . . . 40 4. EMPIRICAL RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 Preliminary Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 Coefficient Transformations ................................... 51 Test for Weights ............................................ 52 vi TABLE OF CONTENTS- Continued Page Test for Equivalence for Mill Changes and Reappraisals ............. 53 Disturbance Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 A Simple Model ................................................ 54 A Complex Model .............................................. 57 5. SUMMARY AND CONCLUSIONS .................................. 61 BIBLIOGRAPHY .................................................... 64 APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 vii LIST OF TABLES Table Page . 1. Standard Deductions for Different Groups under Old Law and HB671 ..................................... 6 2. Difference in Percentage Yes Vote to Proposition 13 between Public Employees and Others ........................... 33 3. Counties with Per Capita Tax Base Greater than $10,000 in 1986 and 1990 ............................................ 44 4. Correlation between Education and Income . . . . . . . . . . . . . . . . . . . . . . . . . 49 5. Coefficients and T -ratio Regressing Squared Residuals on Maddala Variance ......................................... 53 6 .. Correlation of Disturbances ....................................... 54 7. Regression Results on CI-27 and I-105 ............................. 57 8. Low Income Coefficients ........................................ 59 viii LIST OF FIGURES Figure Page 1. Effects of a Tax on Prices and Quantities ............................. 9 2. Inelastic Demand - Tax Borne by Consumer . . . . . . . . . . . . . . . . . . . . . . . . . 11 3. Perfectly Elastic Supply -Tax Borne by Consumer . . . . . . . . . . . . . . . . . . . 12 4. Inelastic Supply -Tax Borne by Producers ........................... 13 5. Perfectly Elastic Demand -Tax Borne by Producers .................. 14 6. Short-run and Long-run Elasticity ofDemand and Supply .............. 16 7. Supply and Demand for Local Rental Housing . . . . . . . . . . . . . . . . . . . . . . . 20 8. Demand and Supply of Labor in a State ........................ 21 9. Average Effective Tax Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 10. Percent ofNo Liability ........................................ 25 ix ABSTRACT Montana citizens have used the initiative process to bring six taxation issues to the voters in the last nine years. Only two of these (1-105 and HB671) were confirmed by the electorate, all but one commanded substantial support. This study has focused on property taxes as potentially a root cause of voter dissatisfaction, even if it may sometimes be expressed as disapproval for other taxes, fees or spending. The dominant relationship found here is between voting and reappraisals of residential property. Voters in counties where property values rose more quickly were significantly more likely to support all but one of the citizen initiatives, in comparison with voters in counties with lower rates of property appreciation. In sharp contrast and somewhat unexpectedly, there is relatively little evidence that high property tax rates (mill levies) or rapidly increasing mill rates are significantly related to voting on tax issues. One explanation for these fmdings is that reappraisals are largely outside the control of both citizens and local officials, in contrast to mill levies. Thus, rapid property appreciation results in tax increases which have not been approved through the normal workings of the political process, resulting in citizen frustration and anger. 1 CHAPTER 1 INTRODUCTION Beginning in the first half of the 1980's, Montana's economy suffered a dramatic reversal as agricultural and energy productions were curtailed by drought and low prices. In 1983, tax revenue from Montana's natural resources, notably coal, oil and gas, timber and hard rock mining, comprised 24% of the state's total tax revenue. But natural resources revenues declined from a high of $346 million in 1982 to $116 million in 1994 (in 1994 dollars). At the same time, state general fund expenditures were gradually increasing from the $300 million level at the end of the seventies to the $500 million level by the end of the eighties. It was clear that Montana was headed towards an expenditure-revenue crunch. It was against this backdrop that Montana's tax limitation proposals emerged.1 The property tax has been the most important tax. form in Montana. In 1984 it provided 47 percent of state tax revenues, compared to an average of 33 percent for the 50 states and 30 percent for the 11 western states. U.S. Department of Commerce ranked Montana 2nd in the 50 states in the property tax per $1,000 personal income (USDC, 1985). Property tax was the main burden for Montana's public. Like California's Proposition 13, Montana's tax reform began with property tax limitation in November 1986. 1 See Young, Weaver, and Mathre (1994) for a discussion. 2 This paper analyzes the voting results of Montana's 56 counties on eight tax proposals from 1986 to 1994 and the 1992 Presidential vote. I try to probe the reasons why citizens vote for or against different tax limitation proposals. Who will gain and who will lose? Understanding the causes of the "tax revolt" will provide clues as to where and how outreach educational efforts should be directed, as well as areas that should be addressed by government officials. The rest of this chapter describes the issues on which citizens voted. In chapter 2, I analyze theoretical foundation. In chapter 3, I analyze econometric specifications and data. In chapter 4, I present empirical results. In chapter 5, I discuss the conclusion and briefly describe problems and future research. Constitutional Initiative 27 (CI-27) This 1986 effort proposed to abolish all taxes on real and personal property. It also would have amended the constitution to require that a sales tax could not be imposed without voters' approval. Although California's Proposition 13 rolled back local property tax assessments and placed a ceiling on the tax rates to bind the behavior oflocal government, CI-27 would have gone further by eliminating the property· tax altogether. Property taxes have been the main sources oflocal government revenues. In Montana around 97 percent of property tax goes to local governments and school districts with the remaining 3 percent going to the state government. Elimination of the property tax undoubtedly would have limited the authority/autonomy oflocal governments, and would have exerted adverse influence on the quality of education and service provided by local governments. 3 Initiative 105 (1-1 05) One of only two successful tax reform initiatives to date, I- 105 passed with 55 percent of the total vote in 1986. I-105 froze property taxes at the June 30, 1986 level. Additional amendments by the legislature made it more flexible: Government jurisdictions can increase property taxes which result from new construction and improvements to existing property, and local governments which had suffered a decline of 5 percent or more in tax bases can raise property tax rates for compensation. Because of the shut down of mining operations located mainly in Eastern Montana, nearly half of Montana's counties suffered a reduction of more than 15 percent in tax bases. Since passage ofl-105, 112local ballot issues proposing to increase property taxes were presented to voters in various jurisdictions in Montana. Of these 112 issues, 93 were adopted by the voters. There is a seeming paradox here: The same people who voted for I- 105 later approved tax increasing proposals to finance local government. Perhaps when people know where their tax dollars are spent, they are willing to pay the tax. To some extent local governments gain more trust from Montana' s public than does state government. Although I -105 passed in 1986, property taxes were not actually frozen. By 1990, many Montanans paid increased taxes because of upward reappraisals of their property, and because a new school equalization program increased mill levies. Constitutional Initiative 55 (CI-55) Proponents of CI-55 advocated elimination of all income taxes, property taxes, sales taxes, registration and license fees. All government services in Montana would be paid by a "trade charge" levied on the gross value of every business. and fmancial transaction. 4 According to proponents, a 1 percent "trade charge" would suffice to pay all government expenditures, and this meant a dramatic reduction in peoples' tax burden. There is an obvious dilemma here: If everyone's tax burden was to be reduced, how could total revenues suffice to pay all government expenditures? CI-55 was defeated with a margin of 28 to 72 percent. Senate Bill235 (SB235) In 1993, Montanans experienced the largest property tax increase in the state's history. Property taxes increased $65 million, or 11 percent, over the 1992 levels. Rapid increases in property taxes and perceived shortages of revenue at the state level finally brought SB235. The main points of SB235 were to reduce income and property taxes, establish a sales tax, and increase total revenues. 1) Sales Tax: • A 4% retail sales tax would be imposed on a broad base, excluding only a few items ranging from food, medical services, prescription drugs, utility bills, and real estate. 2) Property Tax Reduction: • There would be a residential homestead exemption of $20,000, which meant that the taxable value of a house would be reduced by $20,000 for all households. • Personal property taxes in business and agricultural machinery and equipment would be cut in half by reducing the taxable rate from 9 percent to 4.5 percent. 5 • Property tax mill levies for public schools would be reduced and state tax revenue would be used to eliminate property tax mill levies for teacher's retirement, public school transportation, and public school debt. 3) Income Tax Reduction .. SB235 would have eliminated the graduated income tax rate that ranged from 2 percent to 11 percent and replace it with a flat tax rate of 6 percent at all income levels. • For poor people, sales tax credits would provide a compensation for the regressive nature of the sales tax. • Renters would receive a $150 credit per person in lieu of property tax relief. • An itemized deduction would be eliminated and replaced with an increased standard deduction ($5,000 for single and $10,000 for a couple) and larger personal exemptions (from $1,450 to $3,500). SB235 was defeated, 75 to 25 percent in June 1993. House Bill 671 (HB67 ~) In 1993, a number of changes to Montana's income taxes were passed by the legislature. These changes were subsequently suspended by a citizen petition (IR-112). The public vote on HB671 in November 1994 determined whether the tax changes would be permanently suspended or approved. HB671 would have increased total revenues about 10 percent, or $71 million, over the 1994-1995 biennium. Following are the main specific provisions ofHB671: • Under HB671, a single tax rate of 6.7 percent would replace graduated tax rates ranging from 2 percent to 11 percent. 6 • Personal exemptions would increase from $1,400 to $2,710. • Standard deductions would be increased as follows · Table 1. Standard Deductions for Different Groups under Old Law and HB671 Single Filers $2,630 $5,000 Head of Household $5,260 $7,500 Married • Itemized deductions would be eliminated. • Exemptions, standard deductions and two-earner deductions would be phased out between $100,000 and $180,000 of federal adjusted gross income. HB671 was turned down by 76% of voters. Its defeat was the second victory for citizens using the initiative process. Constitutional Initiatives 66 and 67 (CI-66 and CI-67) CI-66 would have required a public vote on any new or increased taxes imposed by state or local governments or school districts. Almost all state, county, municipal, and school taxes, and some fees would have been affected. CI -66 defmed a property tax increase to be any increase in revenues beyond that resulting from new construction and improvement. Where property values were rising, either mill rates would have to be reduced or a vote of the people would be required. CI-67 would have amended the Constitution to require that a two-thirds majority of the state legislature and similar super majorities of any local governing body be required to increase a state or local government budget, tax, or fee. It would have made it more difficult 7 for government to raise revenue since a minority of voters could block any proposal to increase revenues. CI-66 and CI-67 tried to "ensure the reasonable restraint of growth of government." At the same time, reduced government spending would inevitably lead to reductions in public services. Both proposals were defeated, but almost half the voters voted for them. Constitutional Amendment 28 (C-28) C-28 would have amended the Montana Constitution to allow property to be assessed for tax purposes at acquisition value rather than market value. Also, C-28 would have allowed the legislature to limit annual increases in the assessed value of property. The measure was passed by the state legislature and referred to the voters, but it was turned down by a 41 to 59 percent margin. 8 CHAPTER2 THEORETICAL FOUNDATION The person who writes the check may not be the person who really bears the burden of the tax. Taxes can be shifted forward or backward. Tax incidence studies the "ultimate" burden of the tax which occurs whenever a particular piece of tax comes to rest on the fi:J?.al payee. To understand voters' behavior, we should distinguish who is the initial payee and who is the final victim. Does the tax burden shift? To whom will it shift? In this chapter, the incidence of major taxes will be explored. My purpose is to develop hypotheses about who would benefit or lose from the various measures on which citizens voted. I also consider who stood to. gain or lose frpm changes on the expenditure side of the budget. Tax Incidence and Burden Shifting I begin my analysis by considering a competitive market. The basic principles of tax incidence may be illustrated by the demand-and-supply diagram (Figure 1 ). Before tax, the equilibrium is at E0• A tax imposed on sellers shifts the supply curve up by the amount of tax. This lowers the quantity consumed and the net price suppliers receive, and raises the price consumers pay. 9 Price P2 -------- ------ Ql QO Supply curve after tax SuppJy ctl.rve Before tax Demand curve Quantity Figure 1 Effects of a Tax on Prices and Quantities 10 The extent to which consumers and producers bear the tax depends on the shapes of the demand and supply curves. Suppose the demand is completely inelastic, or the supply is completely elastic (Figures 2 and 3), then the entire burden is borne by consumers. On the · other hand, when the supply is completely inelastic or demand is completely elastic (Figures 4 and 5), the tax burden will be borne by the producers. In general, the side of a market which is relatively inelastic will bear the larger share of the burden. Those results also hold if a tax is imposed on buyers rather than sellers. That is, the incidence or burden of a tax is independent of the legal liability between buyers and sellers. Property Tax Incidence The incidence of property tax is a matter of some controversy. Although the debate is by no means settled, empirical and theoretical research suggest quite strongly that property tax differentials are capitalized into local property values (Aaron, 1975). It is generally agreed that taxes on the value of bare land-the sites themselves exclusive of application of reproducible capital in the form of grazing, fertilizer, and the like - rest on the owners of the site at the time the tax is initially levied or increased. The tax cannot be shifted since the supply of land is perfectly inelastic (Figure 4). The land tax will be capitalized into a lower price. Conversely, a reduction in property taxes, such as that proposed in CI-27, I-105 and C-28, would be capitalized into higher land prices. Thus land owners would be beneficiaries. Price Pl PO 11 Demand curve QO=Ql Quantity Figure 2 Inelastic Demand- Tax Borne by Consumer 12 Price Demand curve Pl~--~----~~------~~--------------­Supply curve after tax I I PO ~------------------+1 ~~~---------- Supply curve before tax I EO 1 I I I I I I I I I I I I I I I I Quantity Figure 3 Perfectly Elastic Supply - Tax Borne by Consumer 13 Price PO=Pl -----~--------------- Supply curve (Before and after tax) QO=Ql Quantity Figure 4 Inelastic Supply - Tax Borne By Producers 14 Price Supply curve aftertax PO=Pti------------------~~--~~~~--~D~e-m-an~d~c-urv---e I I I I I I I I I I I I I I I I I I Supply curve bbfore tax I I I I Ql QO Quantity Figure 5 Perfectly Elastic Demand- Tax Borne by Producers 15 Analysis of the incidence of property taxes on structures differ from that on land primarily because the supply of structures is not regarded as fixed. According to the traditional view, any amount of capital for improvement in real property is available in the long-run at constant marginal cost determined by the productivity of capital in other uses. Although variability in the supply of structures is only partial in the short -run, users of real property eventually must pay property taxes on structures through higher sale prices or rents (imputed rents, in the case of owner-occupants). After sufficient time, an increase in property taxes will shrink the stock of structures, and force up their rental prices (Figure 6). This theory of tax incidence suggests that families bear property taxes in proportion to their purchase of goods and services produced by taxed structures. The tax on residential property will be borne by occupants (tenants, if it is a rental house). In contrast, the new view suggests that a nationwide, uniform tax on the value of capital goods would be borne in full by owners of capital goods since they would be unable to avoid it either by shifting assets to untaxed sectors or by raising prices (above Figure 4). If property owners maximize their return, the price at which each producer of goods does so is unaffected by a universal tax on the value of the capital asset, based on the assumption that aggregate supplies of land and capital are fixed. As a result, the tax simply reduces the yield to each owner. The burden of such a tax would be distributed in proportion to the ownership of assets. Just as for land, then, the expected beneficiaries of reduced property would be owners of capital. However, in reality, localities have vastly different levels of effective property tax rates. Differences in property tax rates give rise to so-called excise tax effects. Excise effects 16 Price P s p2 S' pO~------~--r-~---------------- pl D' ql q2 q* Quantity Q Figure 6 Short-run and Long-run Elasticity of Demand and Supply The supply of the product varies with price along Sin the short-run. The tax reduces demand from D to D' and causes prices received by suppliers to fall by p0 to p1 and that paid by demanders to rise by p2-p0• The quantity demanded falls to q 2 in short-run. If, with the passage of time, the supply of the service or commodity is completely elastic- that is, none of the services or commodity will be supplied at a price permanently below p0 and any amount demanded will be supplied at that price - demanders pay the tax in full through an increase in the price by the amount p3- p0, and the quantity demanded declines to q1• 17 arise because variations in local tax rates produce variations in the cost of doing business among localities. The final resting place of property taxes that deviate from national average tax levels depend on the type of the market in which businesses operate. For a national market, relatively high or low property taxes cannot be passed onto consumers because local producers must take selling prices as given. For capital used in goods and services that compete mainly in local markets, immobile factors of production will bear the brunt of property tax deviations. At the heart of all incidence analysis lies the assumption that taxes are borne by those who cannot avoid them. In the case of properties, this means that inter-jurisdiction deviations from national level will be borne by the various factors in the system in proportion to their relative immobility. Tray and Fernandez (1979) assume that: 1) except land, all other factors of production (capitals and labors/consumers) are sufficiently mobile to avoid property taxes; and 2) there are usually good substitutes for any given locality. From this they conclude that deviations in property taxes fall on land owners. Benjamin and Kochin (1982) get the same result in a slightly different model. They assume that there are only two factors supplied-one is completely elastic (labor) and one is completely inelastic (land). Then maximization of the rents of the immobile factor requires that the taxes levied on the immobile factor exceed the cost of government services to provide it. The efficient government response to receipt of a windfall is to leave the taxes on mobile resources unchanged and the level and mixture of government expenditure unchanged. 18 Statistics readily support the notion that there is a great deal of movement into and out of localities among families in all income groups. Migration is the result of weighing the costs and benefits associated with alternative locations. Part of the cost of alternative locations is the tax rate and the prices of locally-produced goods and services. In a sense, localities compete for citizens. The competition affects the incidence of taxes on capital use in producing local goods and services. If capital is mobile but people are not, the cost of local goods and services will be higher in higher property tax areas than in lower property tax areas. This is so because capitalists will demand a national net rate of return on their invested funds, and property taxes will be shifted to consumers and laborers in the form of higher prices. If, however, both capital and people are mobile, then migration flows will respond to the prices of goods and services. If competition for localities exists, conventional incidence theory tells us that the more elastic residential demand is for a particular community, or theless elastic the supply curve is for the community, the greater is the proportion of property tax deviations that will be born by factors used in production of that community. If demand is perfectly elastic, or supply is perfectly inelastic, then all property tax (leviations will be borne by local factors of production in proportion to their relative immobility. Referring to the capital gains resulting from business property tax reduction (based on Tray and Fernandez's assumption that within jurisdictions the price of the product and the wage rate oflabor is fixed by the national market) no tax burden/benefit will be shifted forward to consumers or backward to laborers. Local owners of capital will bear all the results for the time being since capital is quite specialized and cannot readily move to other 19 industries immediately. In the long-run, economic profits attract more and more capital to enter, which will drive up the rent price and diminish economic rents to zero. Consistent with our previous analysis, all benefits will fmally flow to the land owners. As compared with other commodities, housing is a local good not subject to national competition. Within a jurisdiction, the supply of rental housing is quite inelastic in the short run. The majority of benefits from a reduction in residential property tax will fall on the house owner, while tenants only receive a small portion of the benefit (Figure 7). Homeowners would be expected to favor property tax reductions more than tenants, ceteris paribus. Income Tax Incidence Federal income taxes are believed to be borne by earners because they are universal and because the national supply of labor is assumed to be perfectly inelastic (Bogart, Bradford, and Williams, 1992) (see Figure 4 above). Neither of these conditions is likely to hold for state income taxes since some states have them and some do not, and the rates of taxation vary greatly among states. Further, the supply of labor to any single state is probably quite elastic. How can state income taxes be allocated among income classes on an empirical basis? The answer lies in the supply and demand discussion. The more inelastic the demand of labor in a state, and the more elastic the supply of labor to that state, the greater will be the proportion of state income taxes borne by local residents (see Figure 8). In the limit, if demand is perfectly inelastic or supply is perfectly elastic, the inter-jurisdictional deviations 20 Proce P s S' p2 --------------------- D qO ql Quantity of Rental Housing Figure 7 Supply and Demand for Local Rental Housing Original equilibrium is at p0 & q0, The reduction in residential property tax shifts out the supply curve to S'. p1 is the price paid by tenants and p2 is the price received by the house owner. p2-p1 is the tax reduction amount. 21 Wage Rate S' W3 s wo ----- ­ Wl I D Ql QO Quantity of Labor Figure 8 Demand and Supply of Labor in a State 22 of income tax from the national level will finally translate into changes in the rental value of sites, just as in the case of property tax differentials. The above analysis implies that, in an open economy, taxes will be borne by relatively immobile factors. Which factors are mobile will depend to a large degree on the size of the unit in question: town, county, or state. The clearest case of an immobile factor is real estate (site). Here we treat sites as the immobile factor, and develop the distributional implications of the assumption that all gains and losses due to fiscal policy shifts are translated into changes in the rental value of sites. Therefore, an increase in the state income tax will not in the long-run affect the post-tax earnings from labor or capital of the people who live there. Rather, it will affect what they (or their employers) will pay for the right to occupy sites in the state and, therefore, what they will pay for real estate. Since we treat state land as a small economy open to a much larger economy with a well-functioning capital market, capital and labor-particularly high-skilled labor-are mobile at least at a significant margin, and will migrate on the basis of their compensation. Reduced income taxes will attract new investment and increased income taxes will drive away the high income group. Here labor and capital are supposed to be perfectly elastic (see Figure 3). The net productivity of Montana as a location will be capitalized into the rental value of its sites. Consistent with my previous analysis, the landowner will finally bear all the benefits or losses from income change. But it takes time for people to migrate in response to the income change. People may tend to vote based on their annual situation rather than their 23 permanent situation or long-run situation.2 Regarding income taxes, the effective income tax rate would be a good criterion. For those whose tax rate is increasing, they would regard themselves being worse off. On the contrary, people with decreased income tax rate would be winners. Rational citizens' voting decisions can be predicted through immediate income tax change before and after tax. Only two of Montana proposals were concerned with the income tax- HB671 and SB235. Both of them would have enlarged personal exemptions and standard deductions and replaced the graduated rate structure by a flat tax rate (see Appendix 1 for the comparison ofHB671 and SB235). One significant difference between the two bills is that the total income tax collected would be reduced by $50 million a year in SB235, while HB671 would increase total income tax collected by $30 million a year. Income taxes would have been reduced for all income groups under SB235 (see Appendix 2). Redistribution under HB671 is quite complicated. According to the data provided by Department of Revenue, I analyze effective tax rate changes of different income scales for the following five groups: Entire household, head of household, married couples filingjointly, married couples filing separately, and single filers (Appendix 3-1,3-2, 3-3, 3-4, and 3-5 and Figures 9 and 10). Before HB671, Montana had 313,000 households filing income returns in 1991. Forty-three thousand households owed no tax. Under the provisions ofHB671, those who owed no tax would have increased to 106,000 households, or 1/3 of those filing tax returns 2.Fiorina Morris (1974) holds that" ... in deciding what is the self-interested choice, people weigh heavily is the economic results of the recent past, presumably on the assumption that this at least is known whereas promises about the future are stains in the wind." ! ~ 24 Figure 9 Average Effective Tax Rate HB871 ·An raga EHactlve Tax Rat .. All Houooholdo 0%~~~~~~~~~~~~~~~~~~ 0 4 I 12 Ul 20 :SO -«1 ISO 80 7010 100120UO ~ Income Bracket (OOD'a) 7% ... ... ... '" ... "' 0% •Hi 0 ... ... HB871 • Avoroga EHacllva Tax Ralaa Married Filing Joint ~ ~ ~ ~ ~ ~ lnl*n•bl'llcbl(OOO'I) HB871 • Avoroga Efactlva TIX Rotto Slnglo Filer 0 • 4 8 12 tiS 20 30 40150 80 70 ISO 1001:10140. Income Br.cket (ODD'a) HB871 • Avaroga EHocUve Tax Rata Head of Houoehold .... ---------------------------------~ '" ... '" "' '" 0 .. ~~~~~~~~~~~~~~~~~~~ o 4 a 12 ta ~·~ 40 ~ M ~ ~ t®t~t~ Income Brackat (ODD'a) HBB71 ·Avera go EHoctlvo Tax Rolo >%r-----------~B==a~rr~lo~d~F=III=n~g~S~o~p~a~r~a=t•~----~~ 8% ,70 eo too 120 t.co 25 Figure 10 Percent of No Liability ""' - ""' .. ""' ~ ""' :l! i ""' ""' ""' "' 0 HB67l-Perc:eat or No U.biUty All Household '\. EJ ' \ \ . " ~\ ~- _.._ . " " " lO ... " 60 lncomeBrukct(OOO'a) HD67l- Pceotage or No Liability Married Filing Joint " EJ ""' . HBB71 - Porconbgo of No Liability Single filer " 100 llO 140 ·~%,-~-----------------------------------. - 10% 20 ~ 40 1:10 00 70 1G 100 120 1•0 • lneoma Braekel (OOD'a) HBB71 • Porconbgo of No Llobl!lty Head of Houaehold HB671 ·Percent or No Liability Married Filing Separotcly 100%~~~-------------------------------, ~- --------------------- ,.,, 01'-'+-H>-*--H-'F-+-'l"¥>.:>-<:>+-+++++-+-H-+·+-H·+·+-+-+ 20 30 40 !10 60 70 80 100 120 140 1D«!mO B~Uir:ct ('000) 26 (Montana Taxpayers Association, 1993). Thus an additional $30 million in income tax would be collected from 63,000 fewer tax payers. In general, single taxpayers and families with two income earners would experience the largest tax increases. Under current law, 106,000 single filers pay $55 million in taxes, for an average tax per return of$519. HB671 would require 71,000 single taxpayers to contribute $65 million, or an average tax per return of $908. Overall, the average tax paid by Montanans who pay income taxes would increase from $1,114 to $1,656 per household (see Appendix 4). In general, a larger percentage reduction in income tax liabilities of the lowest income decile would take place as many of those families would be removed from the income tax rolls completely. Indeed, the lowest income decile would have a net income tax refund instead of a net liability. The effective income tax rates on middle- and upper-income classes would be raised. Because income tax proposals treat the four groups differently (see Appendix 4), the switch point occurs differently in each group. Except single filers, whose switch point occurs at $10,000 per person, income tax begins to increase with an annual income of around $30,000 (see Figure 9). Low income households (those with incomes less than $30,000) would be expected to vote for HB671, while higher income households would be expected to vote against it. Sales Tax Incidence Forty-five states currently levy a general sales tax, the exceptions being Alaska, Delaware, Montana, New Hampshire, and Oregon. Proponents argue that a sales tax is a stable revenue source in both inflationary and deflationary periods. Furthermore, it is easy 27 to collect and manage. The opponents hold that it is a regressive tax and heavy dependence on it will break the fairness of fiscal policy. A general sales tax refers to a tax on commodities sold to consumers for fmal personal consumption, which is equal to a consumption tax. It can also be extended to tax intermediate commodities purchased by various business companies for further production. The distributional effects of sales taxes are commonly determined by comparing the relative effects of a tax increase on families of different income levels (Rosen, 1995). Sometimes family consumption expenditures or a variant thereof is used as the basis of comparison, but income is the most commonly used measure of "ability to pay." If the tax increase .causes a greater percentage increase in total expenditure relative to income for a low-income family than for a high-income family, the tax increase is regressive. If, on the other hand, the tax increse causes a greater percentage increase in total expenditure relative to income for a high-income family than for a low-income family, the tax increase is progressive. Generally speaking, a general sales tax on consumer goods would be similar to a tax on all income with consumer expenditure = sales receipts = factor payment = income. It makes no difference at which point the tax is levied, but the existence of saving and capital formation breaks the equivalent. The tax on income equals a tax on the sales of consumer goods plus a tax on the sale of capital goods. Thus the burden experienced by a household will depend on how its income is divided between consumption and saving. Since the ratio ; of consumption to income falls when moving up the income scale, so does the ratio of tax burden to income. This is the basis on which the tax is said to be regressive. 28 In the empirical study, I analyze data concerning tax burden changes before and after imposition of a 4 percent retail sales tax in Montana for various income groups (provided by Department of Revenue). The tax burden is calculated based on the statutory sales tax rate ( 4% ). According to the Department of Revenue, $252.77 million in sales tax revenue would be obtained. But if the sales tax burdens of all income groups are added, the retail sales tax only generates $124 million. Where would another $128 million of revenue come from? The Department of Revenue's calculation entirely overlooks the part falling on intermediate commodities purchased for further production. The actual distributional effect will be different. Calculation is difficult. It will depend not only on the income and expenditure pattern of those consumers who purchase the commodity for fmal personal consumption, but also on the pattern of intermediate use to which the commodity is put and the income and expenditure pattern of the consumer who ultimately purchases the value of the intermediate products (and the sales tax thereon) embedded in other fmal consumption goods and services. Early attempts to account for the distributional impact of the sales tax burden imposed through coverage of business transactions included in crude efforts of Schaefer (1969) and Daicoff (1980). Schaefer concluded that estimates of the burden of sales taxes would not be affected much by the incorporation of business transactions in the analysis. Daicoff relied on the assumption that the ultimate burden of a tax on a business transaction is divided between consumers of the businesses' products and other sources of income. The result of this assumption is that sales and use taxes were regressive throughout the income scale and members of the lowest income group actually faced a sales and use tax rate in 29 excess of the statutory rate, partly because of the burden shifted to them from business. More recently, Raymond Ring (1989) noted that the burdens of a sales tax depend on the degree to which it falls on intermediate goods or service. "If stockholders or employees bear the business share of a 'consumption tax', the incidence is much different from that of a true consumption tax. Even if the business share is shifted onto consumers its incidence pattern differs from that of the share levied directly onto consumers (Ring, 1989)." Ring estimated that about 40 percent of state sales and use taxes fall on intermediate goods and services. He claims that "the usual approaches of treating the sales tax as exclusively a tax on consumers when allocating the tax burden among income classes will overstate regressivity for states with relatively low consumers' share (Ring, 1989)." In contrast to what he said here, in our calculation the effective tax rate is underestimated because it did not include the business tax falling on consumers. The above analysis is based on a cross-section for a single year rather than a continuous record over time. Among economists, the current year's income may not be a good indicator of the lo~g-term economic status of a family. The consumption in the lowest income class includes those whose income is temporarily depressed and retired persons whose previous income was greater. The consumption-income ratio may be biased upward for these groups. On the other hand, in high income classes, the ratio of aggregate consumption to aggregate income may be held down by the presence of families whose incomes are higher than usual. Statistics for any one year probably make consumption taxes appear more regressive with respect to income than would statistics in which the cumulative or average income and consumption for a period of several years or for a life time were 30 compared. Thus the cross-section statistics make a flat-rate sales tax appear more regressive in lower income classes and less progressive in high income classes than would a long-term comparison for identical families. A hypothesis that has received attention from specialists in recent years is that average ratio of consumption to a family's normal or permanent income is the same at all income levels. If this is true, a flat-rate tax on all consumption will be proportional with respect to normal or permanent income. Until better statistics are compiled, tax regressivity can justifiably be measured on the basis of available single~year data, but with the qualification that the estimates for the lowest and highest incomes are probably less representative than those for middle incomes. Though regressivity has been doubted among economists, for the public and politicians it still is an issue of great concern. Because of this, in SB235 a sales tax refund credit was allowed for low income households ($90 per person for those whose income is less than $13,000). The bill's proponents argued that it would offset the regressivity of the sales tax. But a refund credit would only make the sales tax progressive up to the point where the maximum credit is paid and would place the largest tax burden at the point where the credit phases out. In Table 4 under "A. Sales Tax", the estimated sales taxes are $193 and $191 for incomes less than $5,000 and $5,000-10,000, respectively. Refundable credits would provide $153 and $171 which subsequently makes a net sales tax burden of$40 and $20, respectively. In contrast to the conventional view that sales taxes make the lowest income group worse off, a sales tax with refundable credits as in SB235 would put the maximum burden on the large middle income group. 31 Government Expenditure Until now, our analysis has been limited to the pure tax, unrelated to the level of public goods or services provided by the governmental unit that levies the tax. But the utility-maximizing individual's voting decision is based on the calculation of his marginal benefit and marginal cost, which means he will compare the additional services he receives to the additional tax incurred. Ifbenefits gained from government services are believed to be greater than the taxes paid, a rational individual would be willing to pay taxes. So, before making any predictions of people's voting behavior, it is important to analyze government expenditures. The three primary uses of state and local government expenditures are ·education, health and welfare, and highways. Consequently, members of the following groups might, on average, consider themselves to be beneficiaries of government spending: voters with school age children, voters from low income fami~ies, and government employees. Two groups believed to be recipients of negative benefits from the system are voters from upper income families and homeowners. Nearly 50 percent of property taxes are used to finance local elementary and secondary education, and states enforce compulsory education for all children regardless of individual benefit. It follows that taxes for the support of public education may not be appropriately levied on the individuals who directly consume them. The education of the children of low-income families is one of the most important devices for raising the economic status of families. As educational opportunities are made more widely available, 32 it is reasonable to expect that the major beneficiaries will be families at the low end of the income scale. Referring to welfare expenditure, low income groups receive direct financial benefits from state or local governments in the form of unemployment compensation, pensions, public assistance, or medical aid. Such welfare programs are often criticized by the general public, so they might be expected to be among the most vulnerable government programs if state and local taxes are to be radically cut. Local property taxes are sources for funding education, while state revenues cover most welfare expenditures. Thus parents of school age children would be expected to be more likely to oppose CI-27, I-105, and C-28, while welfare recipients will strongly support HB671 and SB235. Affluent Montanans would benefit the most from a tax reduction and would be the least affected by cuts in public services. Now consider public employees. Public employees would be expected to be strong opponents of the tax revolt if they perceive it as likely to put public employees out of work or to lower their salaries. In California's's Proposition 13, public employees strongly opposed tax revolt and displayed an unusually high voting turnout (see Table 2). 33 Table 2. Difference in Percentage Yes Vote to Proposition 13 between Public and Others May; 1978 40% 56% -16% August, 1978 44% 68% -24% Recall Vote 45% 77% -32% November, 1979 55% 71% -16% 47% 69% -22% Citrin (1985), p. 153 Voting and the Median Voter Model Citizens routinely vote directly on levels of property taxes in many communities. Citizens also elect local representatives to county commissions, school boards, and municipal offices. When exarning data across counties, as is done in this thesis, the following points are noteworthy. • Counties differ in the tax base and in the education and income level of their citizens. They may, therefore, choose different levels of taxes and spending, and relatively high tax rates may reflect these choices. In other words, high tax rates may not be correlated with voter dissatisfaction. • The median vote model implies that (almost) 50 percent of the voters think taxes are too high and another 50 percent think taxes are too low (Rosen, 1995). Thus we should expect that measures to reduce taxes and spending would receive at least a substantial fraction of the voters. 34 • Reappraisals of property may disturb the local political-economic "equilibrium" level of taxes, if not offset by changes in mill rates. In particular, if mill rates are simply "slow'' to respond to changes in property valuation, taxes will be "too high" in counties where values are rising, and citizens may be mclined to vote for tax limitation measures. Also, the state levies about one-quarter of the total mill rate, and these mills do not respond to changes in local valuations. Thus taxes increase in communities where values are rising, even if local mill levies are adjusted downward. 35 CHAPTER3 ECONOMETRIC SPECIFICATION AND DATA Econometric Specification Regression Model The primary data used in this study are the actual votes cast in each of the 56 counties of Montana on eight ballot issues and the 1992 Presidential election. The basic question can be stated explicitly: Can the revealed preferences of citizens, formulated as percent yes vote by counties, be explained by observable characteristics of counties, i.e., by differences between counties in mill levy, the growth rate of assessed value, population growth, ownership, income level, state/local government employees, and so forth? Symbolically, the basic hypothesis can be written as a multiple regression. The outcome on issue i in county k, y ki is assumed to be linearly related to a set of various social and economic characteristics in co~ty k, xjki• 36 or simply, where Yki is the logistic transformation of the proportion Pki• which voted for the measure. Note that Pki is the proportion of yes votes of the county k on issue i, b 0 i is the constant term; bj i is the regression coefficient of the independent variable j; xjk i is the value of independent variable j for county k on the issue i; and uki is the value of the random term; i = 1 , ... , 9, signifying issues; k = 1 , ... , 56, signifying counties; j = 1 , ... n, signifying independent variables. Heteroskedasticity and Weighting According to standard sources, the disturbances are heteroskedastic across counties within each equation: 1 where nki is the total number of people voting on issue i in county k. (See, e.g., Maddala, . p.29). Counties with larger number of voters and/or percentages voting "for" close to 50 percent receive larger weights. 37 However, heteroskedasticity may not exist if the disturbances mainly reflect county level phenomena, rather than variations across individual voters. For example, the disturbances may reflect unobserved factors at the county level, such as the (perceived) quality of the public services and the efficiency with which they are delivered, opinions expressed in local newspapers, etc. To test for heteroskedasticity, the following procedure is performed: 1) Estimate each equation by ordinary least squares and obtain the residuals, eki· 2) Regress the squared residuals on the "Maddala variance" and a constant term, According to the Maddala model, a 0i = 0, and a 1i = 1. On the other hand, the disturbances are homoskedastic with Var(eki)=a0i if a 1i =0. Cross Equation Correlation The equations may also be related through nonzero covariances of the error terms across different issues for a given county. Below, a seemingly unrelated model is created by writing the system of 9 equations as follows: yl 1 0 0 pt ul ... y2 = 0 x 2 ... 0 pz + uz .............. y9 0 0 ... x9 p9 u9 where I E(u 1u 1 ) 38 aiJ o ... o = o aiJ ••• o o o ... aiJ =a 1 lj ani 0 ti ···0 tl 02/ 02/ :··a2l 09/ a92I ... a9l The most efficient estimation method is to apply generalized least-squares to· obtain ~· =(x'Q·1x)-1(x'Q·1y) (Greene, 1993, Chapter 17). Data and Specific Hypotheses Dependent Variables Appendix 5 presents descriptive statistics for each of the issues across counties. An "unweighted mean value" for tax issues is the simple average proportion yes vote. A "weighted mean value" is computed by using total votes of each county as a weight. Basically trivial differences exist among "unweighted" and "weighted" means. Tax revolts have often been associated with property tax changes. According to an oft-cited ACIR study, the property tax has been perceived as the least fair tax form. Like California's Proposition 13 and Massachusetts' Proposition 2 Yz, several of Montana's tax referendums have focused on property tax. CI-27 (1986), I-105 (1986), and C-28 (1994) solely deal with property tax reforms. Even though CI-27 and C-28 were defeated by 56 39 percent and 59 percent of the total voters, more than 40 percent of the voters were in favor of them. CI -66 and CI-67 are not exclusively concerned with property taxes. But the correlation between CI-66, CI-67 and C-28 undoubtedly hints that antagonism toward high property tax leads voters to vote on CI-66 and CI-67 as well (r = 0.59 between C-28 and CI-66; r = 0.63 between C-28 and CI-67). In the following section, I briefly discuss the correlations between nine issues (see Appendix 6). 1. The first three tax proposals are highly correlated (r = 0.52 betwee':l CI-27 and I-105; r = 0.62 between CI-27 and CI-55). ii. The extremely highly correlated issues are CI-66 and CI-67 (r = 0.93). iii. All tax limitation proposals (CI-27, I-105, CI-66, CI-67, and C-28) are consistently positively related. IV. SB235 displays negative weak correlation to other proposals, indicating people who favor comprehensive tax reform solely oppose all other single reforms. People who vote for tax cut proposals were inclined to oppose SB235. v. A vote for Perot is positively correlated with almost all property tax and revenue limitation issues. 40 vi. One unexpected phenomenon is that HB671 is positively correlated with CI-66, CI- 67, and C-28 (r = 0.36, r = 0.39 and r = 0.55). HB671 would have increased total government revenue and CI-66, CI-67, and C-28 are proposed to cut government revenues. Independent Variables and Specific Hypotheses Descriptive statistics on the independent variables are presented in Appendix 7. PVALxw These variables measure the percentage change in the assessed value of residential properties over three years. For example, PV AL8386 measures the average change in the assessed value from 1983 to 1986. In Montana the first reappraisal was completed in 1978. The average assessed value increased by 40 percent. This large increase in market value occurred because property in Montana had not been valued since the late 1950s and early 1960s. Still the 1978 reappraisal only brought property up to market values as they existed in 1972. The second reappraisal cycle was completed in 1986 and raised market values to those that existed in 1982. In 1993, Montana finished its third reappraisal cycle bringing property up to market values in 1992. This partly explains why property tax revolts happened in 1986 and 1994. The weighted average appraised value increased by about 4.3 percent, ranging from -17.6 percent to 45.4 percent, from 1983 to 1986. For years 1987-1990, 1989-1992, and 1991-1994, the weighted average of the percentage change in appraised values (PV AL) are 0 percent, 6.6 percent, and 7.8 percent respectively. Obviously, the statewide assessed value of residential property has been progressively increasing year by year. The following 41 table lists counties which experienced the greatest increases and decreases in assessed values in the various periods. Counties with the Greatest Increases in Assessed Value 1983-1986 Ravalli45% Rosebud35% 1987-1990 Cascade 18% Jefferson 13% Powder River 34% Silver Bow 11% 1989-1992 Cascade26% Gallatin 26% Silver Bow 13% 1991-1994 Lake25% Mineral 22% Flathead 17% Counties with the Greatest Decreases in Assessed Value 1983-1986 1987-1990 1989-1992 1991-1994 Roosevelt -18% Dawson -23% Dawson -12% Liberty -22% Madison -16% Fallon -23% Fallon -12% Sweet Grass -18% Missoula -15% Powder River -23% Powder River -12% Daniel -16% Richland -23% Richland -12% Prairie -16% Wibaux -23% Wibaux -12% In 1989-1992 and 1991-1994, the unweighted average increases are 0.2 percent and 0.9 percent, while the weighted averages are 6.6 percent and 7.8 percent respectively. The differences indicate that a lot of sparsely populated eastern and northern counties experienced decreases in assessed values, while most of the increases occurred in the more highly populated counties. Hypothesis: • Tax revolts are more likely to occur in counties with more rapidly rising appraised values. The assessed value is associated with inflation or deflation in a certain district which can't be controlled by voters. About one-fourth of the mill levy is raised by state government for school equalization funding and state expenditure while the rest is decided by the local public or their representatives. If higher appraised values are offset by lower 42 mill levies, then the property tax bills are kept at a stable level. In this case, there should be no revolt at all, since the mill levy (thus property tax as well) reflects the majority's choice for local government's spending. The trick is that the counties experiencing appreciation in the assessed value may not adjust mill levies immediately. Thus the windfall will fall to county governments. On the other side, counties which have decreases in the assessed value, are more likely to raise mill levies to maintain certain levels of government services. In 1993, the legislature increased county mandatory property taxes from 40 mills to 55 mills and imposed a new 40 mills state levy to help equalize school funding. All these mills cannot be reduced to reflect increased taxable values. This indicates that counties with the higher assessed value are forced to finance counties with the lower assessed value. MILLSxx These variables measure the average mill levy for each year. The weighted mean value is consistently higher than the unweighted one, indicating that counties with higher populations have higher mill levies. The disparity is tremendous, ranging from 91.5 in a resource-rich county in 1986 (Fallon) to 545 in an urban county in 1993 (Silver Bow). The mill levy has historically been determined largely by voters or their representatives (school trustees, county commissioners, etc.) at the local level, and thus may more or less reflect the preferences and constraints of individuals in the various communities. In other words, higher mill levies may reflect choices made by citizens. Nonetheless, high mill levies do result in high property taxes, and thus may be a cause of the tax revolt. 43 Hypothesis: • The tax revolts are more likely to happen in counties with higher mill levies. In 1993, Montana legislature passed a new law to force school districts in Montana to spend about the same amount of money per student in each county. To accomplish this goal, the legislature established minimum and maximum budgets of all schools based primarily on the number of students attending each school. Schools spending below the minimum are required to increase their budgets so that they are spending at least the minimum amount. More than 400 school districts experienced substantial property tax increases. Most counties with lower assessed values were forced to raise high mill levies to reach the minimum school budget. PMILLx:x;yy These variables measure percentage changes in mill levies over three years. For example PMILL8386 measures the percentage change in mill levies from 1983 to 1986. The weighted means are around 10 percent in each three year period. Sparsely populated counties experienced huge changes in mill levies. This is easy to understand: With shutdown of mines and excluding the mining proceeds froni the tax base, total county taxable value dropped dramatically after 1986. This gave rise to an increase in mill levies. In 1983-1986 Blaine county experienced the greatest reduction in mill levies while in 1987- 1990 Big Hom county had 100 percent increase in mill levy. Hypothesis: • The tax revolts are more likely to happen in counties with rapidly rising mill levies (PMILL ). 44 Additional hypothesis is that: • Property tax payers care only about the total tax bill - not its composition between mill rates and the appraised value. If this hypothesis is true, then the coefficients on PV ALxxyy and PMILLxxyy should be equal. But the hypothesis may not be true for reasons given earlier: Local citizens have substantial influence over mill levies, but not over reappraisals. TAXBASExx These variables measure per capita tax base in each year. For example TAXBASE86 measures the total county taxable value divided by the population in 1986. The weighted average per capita tax base dropped by about 1/3 between 1986 and 1990. This was mainly due to the decline of natural resources. Table 3 lists six counties with per capita tax base greater than $10,000 in 1986. Fallon 35,943.00 7.4% 4,446.00 50.3% -87.6% 2 Rosebud 17,795.00 19.4% 16,993.00 17.4% -4.5% 3 Wibuax 17,615.00 18.5% 3,510.00 74.6% -80.1% 4 Powder River 15,631.00 19.6% 2,888.00 91.5% -81.5% 5 Sheridan 15,344.00 17.3% 2,630.00 90.1% -82.9% 6 .00 23.4% 2,368.00 81.3% -77.8% Besides per capita tax base, I calculate the resident-borne tax base which is equal to the total tax base- (utilities+ mining proceeds+ railroad+ airline). It represents the part of total taxes falling directly on residents. TRATIO =resident-borne tax base I total tax base, which can be perceived as the relative price for public goods. 45 Table 3 also lists TRA TIO for each county. In 1986, most resource-rich counties had a TRATIO less than 20 percent; with the lowest being 7.4 percent in Fallon county. In 1990, almost all counties had increased TRATIO, except for Rosebud, which kept its tax base almost at the same level. Hypotheses: • Tax revolts are more likely to happen in a county with a lower per capita tax base. • Tax revolts are more like to happen in a county with a higher TRATIO. MARLENEE This variable is the proportion of yes votes on the Republican congressional candidate in 1992. The unweighted mean is 54.0 and the weighted one is 45.3, which hints that Republicans are more likely to be elected in sparsely populated counties, while Democrats get support mainly from areas with a high density of population. A fundamental difference between the Democratic and Republic parties is in their stands on the government's role in economic and social affairs. Democrats typically advocate an expansive role for the public sector, while Republicans are typically characterized as favoring smaller government and lower taxes. Hypothesis: • People who vote for Republicans are more likely to support tax reductions and oppose tax increases. PSTEMP:xx These variables calculate the numbers of state government employees as a percentage of registered voters in each county. The unweighted mean is around 2.5 and 46 the weighted one is around 3.0 which is consistently higher than the unweighted one in each year. It indicates that state employees are more concentrated in urban counties. Hypothesis: • Changes in tax revenue can affect employment opportunity and the salary levels of state employees. They are more like to favor government expenditures and higher taxes. PLOEMfxx These variables calculate the ratio of local government employees to the registered voters in each county. The unweighted mean is around 9 percent for each year and the weighted one is a little bit lower, indicating the percentage of local employees is higher in sparse areas. Hypothesis: • Local government employees are more likely to vote against tax reductions and support increases. PPOPxxyy These variables measure five-year growth rates of population. The unweighted mean is nearly negative for each period, but the weighted mean is positive except in the period from 1985-1990. This implies that urban areas are growing fast relative to rural areas. The biggest decrease in population happened in 1985-1990. The unweighted mean is -7.6% while the weighted mean is -2.8%. It is easy to understand: Dramatic reductions of mining activities brought about outward migration. In the 1990s, Montana is experiencing large population growth. The weighted average is 5.2 percent for 1988-1993 while the unweighted average is 0.9 percent. 47 Growth and decline affect demands for public services. Tax burdens will rise with growth if new infrastructure costs are borne in part by existing tax payers. The prosperity of the economy in a district is usually linked with the migration trend. This drives up the value of regional property as well. Population growth is positively correlated with the percentage change in the assessed value. Hypothesis: • Property tax revolts are more likely to happen in areas experiencing population growth. TURNOUTxx These variables calculate total votes cast as a percentage of the number of registered voters in every county. There are no big differences between weighted and unweighted turnout ratios, which are above 70 percent in all areas. Usually, if people care about the contents of proposals, or alternatively, if they are interested in proposals, the turnout ratio should be higher. Looking at the whole voting history, the turnout ratio is higher in Presidential elections than in other issues. Hypothesis: • The higher the voter turnout ratio is in one county , the more likely people are to vote against the tax revolt issues. INCOME These variables measure the proportions of low, middle and high income groups. Based on annual income, I divide households into five groups: Low Income Groups: (32%) Middle Income Groups: (55%) High Income Groups: (13%) 48 Less than $5,000 Between $5,000 and $15,000 Between $15,000 to $35,000 Between $35,000 to $50,000 More than $50,000 %of Households 7% 25% 40% 15% 13% Except for the high income group, weighted means are lower than unweighted ones for other income groups. The households' annual income in urban areas is a little lower than that in sparsely populated areas. Generally, property wealth is positively correlated with income. Now we can't draw such a simple statement since, among property owners, there are a certain number of retired people who have property of high value, but live on fixed pension. Even incorporating this type of people, a higher property tax is largely borne by the middle and high income groups, while the sales tax hurts the poorest people. But the sales tax bill (SB235) included a refund credit granted to the poorest people, so that the sales tax burden actually would fall mainly on the middle class. Income tax (HB671) would make the effective tax rate higher for households with an annual incomes of more than $30,000. Hypotheses: • The higher income group is more likely to vote against HB671. • The low income group is inclined to vote for SB235 and HB671, while voting against property tax reforms. • Middle and higher classes favor property tax limitation proposals. 49 PEDL12 This variable measures the percentage of persons over age 25with less than 12 years of education. Lower educated people are highly associated with lower income (see Table 4). Table 4. Correlation between Education and Income Hypothesis: • A county with a high percentage of less educated people is more likely to vote against property tax reductions. PEDCOLL This variable measures the percentage of persons over age 25 with a bachelor or higher degree. The urban citizen seems to have more education than those in the rural areas. The unweighted PEDCOLL is 15.9 while the weighted one is 20.0. A common view holds that higher education leads to a taste for relatively more government spending. But Courant, Gramlich, and Rubinfeld (1980) show that education is nearly unrelated to the government spending. Until now nearly no research shows a strong correlation between education and government spending. On the one hand, a highly educated person might have a strong sense of social responsibility. On the other hand, he or she is positively correlated with a high income class (r = 0.58 between PINCM50 and PEDCOLL) which indicates that he or she is most likely to oppose revenue increases. 50 Hypothesis: • Highly educated people show no obvious relation to the tax revolts. OWNER 'This variable measures the percentage of housing units which are owner occupied. The weighted mean is 67 .4. In the theoretical section, I draw the following conclusion: All property and other taxes fmally fall on local land owners as they are capitalized into lower land values. Hypothesis: • Property tax revolts are more likely to happen in a county with a higher percentage of land owners. WKIDS This variable measures the percentage of families with children under 18. Half of the families belong to this group. Hypothesis: • Families with children under 18 are more likely to vote against tax cut proposals, because they consume local government services (schools). OVER 65 This variable compares the number of citizens over 65 to the total of registered voters. Elderly people are inclined to be net taxpayers and receive few government services. Hypothesis: • Elderly people are more likely to favor property tax cuts. 51 CHAPTER4 EMPIRICAL RESULTS Preliminary Results Coefficient Transformations Since I use the logit model, the coefficients in the regression are difficult to interpret. They can be transformed as follows: Let then log-P-=Px +e (1-p) y=Iog-P­ (1-p) ay 1 ap p(l-p) ap _ap ay ---X- ax ay ax 52 ay =P ax where p = the proportion voting yes on each measure. Thus the transformed coefficients, p(l-p )p, measure the change in the proportion voting yes for a unit increase in an x variable. Appendices 9 and 10 list parameters, t-ratios, and goodness of fit measures for a simple model and a complex model. Both regression coefficients, p, and transformed coefficients are presented. Test for Weights As indicated previously, the equations are first estimated by OLS and then the squared residuals, eki2, are regressed on the weight proposed by Maddala.Table 5 indicates that, except for CI -66, the coefficients of the weight variables are never significantly different from zero, indicating that the residuals are homoskedastic. Instead of running the weighted system equation, I use the results from unweighted equations. T-ratio .01 -.02 -1.2 -1.1 Test for Equivalence for Mill Changes and Reappraisals 53 -.4 .4 2.1 0.11 .1 If flo is accepted, it implies that property tax payers care only about the total tax bill -not its composition between mill rates and the assessed value. I make a joint test, and get the F value: F = 3.88 where DF 1 = 9, DF2 = 342 and a= 0.05 The critical F-value, F* = 1.9, indicating that if F4 PMILL91!>4 PVAL91!>4 TAXBASE!>4 TRATI0!>4 PSTEMP93 PLOEMP93 PPOP8893 TURNOUI'94 PlNCSK PlNCS·lS P1NCIS·3S PlNCMSO PEDL12 PEDCOLL POWNER WKID OVER6S 0.22 ·0.07 ·0.02 0.29 0.32 0.32 -0.27 ·0.23 ·0.45 0.38 ·0.01 0.24 ·0.07 0.24 ·0.29 0.17 ·0.31 0.24 ·0.13 -0.26 0.06 0.05 0.12 0.23 ·0.22 ·0.11 ·0.15 0.00 ·0.21 0.29 0.00 ·0.30 1.00 0.22 0.03 ·0.03 ·0.04 0.05 0.01 ·0.17 0.06 0.16 ·0.24 1.00 ·0.42 0.41 ·0.53 0.63 0.02 0.16 0.29 ·0.24 0.13 1.00 -0.35 0.21 ·0.24 -0.07 .0.26 ·0.16 0.22 ·0.29 0.23 .0.29 0.43 ·0.03 ·0.19 0.11 0.41 ·0.40 .0.17 0.13 0.14 .0.11 1.00 -0.59 0.44 0.03 0.22 0.16 -0.21 0.41 1.00 ·0.52 -0.08 -0.03 -0.27 0.25 ·0.38 1.00 -0.03 0.15 0.33 -0.22 0.18 1.00 0.34 -0.04 -0.50 0.43 1.00 ·0.24 .0.61 0.39 1.00 -0.43 0.11 1.00 ·0.58 1.00 ·0.43 0.51 ·0.38 -0.27 ·0.38 ·0.23 0.58 -0.75 1.00 0.24 0.08 ·0.42 0.20 0.54 ·0.52 0.05 ·0.22 0.04 ·0.23 0.30 ·0.38 ·0.29 0.45 0.29 -0.01 ·0.41 0.08 1.00 -0.59 1.00 ·0.11 .o.u 0.28 0.30 0.26 0.26 -0.57 0.40 ·0.40 0.43 -0.59 1.00 Appendix 9. Regression Coefficients (The Simple Model) PVAL 0.0057•• 0.15 0.0014 0.04 0.016•• 0.3 ·0.002 -0.04 0.0004 0.01 0.006° 0.1 0.011•• 0.3 o.oos•• 0.2 0.011•• 0.3 (2.2) (0.7) (5.0) (·0.7) (0) (1.8) (5.0) (4.0) (3.9) MILLS 0.0019•• 0.05 0.0003 0.01 0.0017- 0.03 ·0.0004 ·0.01 0.001 0.02 o.oot• 0.02 0.0002 0.01 0 0 0.0006 0.02 (4.6) (1.2) (3.2) (•1.2) (1.3) (1.8) (0.37) (0.2) (1.2) PM ILLS ·0.0047" -0.01 ·0.002 .o.o5 0.0029" 0.06 0.0004 0.01 0.003 0.06 ·0.0004 .o.o1 0.002 0.05 0.0013 0.03 ·0.0038° .o.1 (·1.9) (·1.0) (1.9) (0.3) (1.1) (·0.1) (1.3) (0.85) (•1.8) MARLENE PSTEMP PLOEMP PPOP TURNOUT86 CP PlNCLSK ,!::>. PlNC5·15K PlNC15·35K PlNCM50K PEDL12 PEDCOU OWNER WKlDS OVER65 CONSTANT R2 0.24 0.03 0.22 0.07 0.06 0.14 0.30 0.23 0.32 Note: Slgnlftcanc:e levels are lndlc.ated aa follows (two tailed testa): • 10% ••S% T atatiatlcs are in parenthesee Dependent Variable a Log (propertlon (yes) I (1•propertlon (yes)) Appendix 10. Regression Coefficients (The Complex Model) PVAL MillS PMillS PSTEMP PLOEMP PPOP TURNOUT86 PINCLSK PINCS·ISK PINCI5·35K PINCMSOK PEDLI2 PEDCOLL OWNER WKIDS OVI!R65 CONSTANT R2 0.007** (2.8) 0.002** (3.7) ..o.oos• (•1.8) 0.005 (1.1) ..0.018 (·1.3) ..0.02 (•1.0) 0.003 (0.5) 0.003 (0.3) 0.033 (1.3) 0.032* (1.8) 0.002 (0.1) 0.014 (O.S) ..0.005 (..0.36) 0.015 (1.0) 0.018* (1.9) 0.047** (2.9) 0.005 (0.4) -5.3 .. (•2.6) 0.56 0.2 0.1 ..0.1 0.1 ..o.s ..o.s 0.1 0.1 0.8 0.8 0.1 0.4 ..0.1 0.4 o.s 1.2 0.1 0.003• (1.6) 0.0006 (1.4) ..0.001 (..0.5) 0.007- (2.6) ..0.004 (..0.5) 0 (0) 0.013** (3.2) 0,02 (1.3) 0.008 (0.8) 0.011 (0.9) 0.01 (0.6) ..0.006 (..0.8) ..0.002 (..0.2) ..0.002 (..0.3) 0.01 (1.0) 0.013* (1.9) ..0.72 (..0.6) 0.45 0.1 0.02 ..0.03 0.2 ..0.1 0 0.3 ..0.3 o.s 0.2 0.3 0.3 ..0.2 ..0.1 ..0.1 0.3 0.3 o.ots•• (3.2) o.oot• (1.9) 0.002 (0.9) 0.001 (0.3) 0.004 (0.3) ..0.012 (..0.7) 0.002 (0.5) ..0.009" (•1.8) 0.012 (O.S) 0.032" (1.8) 0.02 (1.1) 0.045* (1.6) 0.002 (0.12) 0 (0) 0.014* (1.6) 0.04 .. (2.6) 0.001 (0.6) ·S.S .. (-2.8) 0.44 Note: Significance levels are indicated aa follows (two tailed teats): • 10% •• S% T statistlc::a are in parentheses Dependent Variable = Log (proportion (yea) I (!•proportion (yea)) 0.3 0.02 0.04 0.02 0.1 ..0.3 0.04 ..0.2 0.2 0.6 0.4 0.9 0.04 0 0.3 0.8 0.1 0.001 (0.3) 0 (..0.3) 0.002* (1.7) 0.004• (1.8) ..0.009 (·I. I) ..0.018" (•1.8) ..0.001 (..0.5) ..0.02 (·1.4) 0.001 (0.1) 0 (0) ..0.013 (..0.8) ..0.006 (0.7) ..0.006 (..0.6) 0 (0) 0.006 (0.7) ..0.001 (..0.9) ·1.25 (·1.1) o.so 0.02 0 0.04 0.1 ..0.2 ..0.3 0.2 ..0.4 0.02 0 ..0.2 ..0.1 ..0.1 0 0.1 ..0.1 ..0.011 (·I.S) 0.0008 (0.9) 0.001 (M) 0.016 .. (2.9) 0.03* (1.8) ..0.008 (..0.4) ..0.003 (..0.5) ..0.006 (..0.5) 0.003 (0) 0.009 (0.4) 0.0001 (0) 0.048 (1.3) 0.029" (I.S) 0.037" (1.8) ..0.014 (•1.1) ..0.03 (·1.4) 0.004 (0.3) ·2.4 (..0.9) 0.48 0.02 0.001 (1.3) 0.02 ..0.001 (..0.3) 0.3 ..0.009"" (•2.0) 0.6 0.009 (0.7) ..0.2 0.012 (0.6) ..0.1 0.008 ..0.1 0.1 0.2 0.02 0.9 0.6 0.1 ..0.2 ..0.6 0.1 (1.3) ..0.009 (..0.9) ..0.02 (..0,8) ..0.002 (..0.1) 0.009 (O.S) ..0.015 (..0.5) 0 (0) ..0.014 (..0.8) ..0.009 (oi.O) ..0.001 (0) ..0.014 (•1.1) 0.85 (M) 0.47 ..0.1 0.02 ..0.02 ..0.2 0.2 0.2 0.2 ..0.2 ..0.4 0.008 .. (2.8) 0.001 (1.3) 0.002 (1.3) 0 (0.1) ..0.006 (..0.8) 0.006 (0.5) o.ot•• (2.8) 0.001 (0) ..0.04 0.017 (1.4) 0.2 0.003 (0.3) ..0.3 0.005 (0.3) ..0.3 ..0.032- (·3.2) -0.2 0.02** (3.8) ..0.02 0.002 (0.2) 0 ..0.013* (•1.6) 0.036 (0) 0.70 0.2 0.03 0.05 0 ..0.2 0.2 0.3 ..0.5 0.03 0.4 0.08 0.1 ..o.s ..0.8 o.s 0.1 ..0.3 0.006** (2.5) 0.001 (1.3) 0.001 (0.8) 0.001 (0.5) ..0.008 (•I. I) 0.012 (1.2) 0.009"" (2.6) ..o.ou­ (·2.5) ..0.003 (..0.2) 0.022- (2.0) 0.008 (0.7) 0.015 (0.9) ..0.015" (•1.7) o.ou­ (2.8) 0.005 (0.5) ..0.008 (•1.1) ..0.72 (..0.62) 0.63 0.2 0.03 0.03 0.03 ..0.2 0.3 0.2 ..0.4 ..0.08 0.6 0.2 0.4 ..0.4 ..o.s 0.4 0.1 ..0.2 o.oos•• (2.0) 0 (0.2) ..0.004• (·1.6) ..0.006 (·1.5) ..0.015 (•1.4) 0.029" (1.9) 0.012- (2.4) ..0.002 (..0.2) ..0.009 (..O.S) 0.001 (0) 0.005 (0.3) ..0.005 (..0.2) ..0.013 (•I. I) 0.003 (0.3) 0 (0) 0.001 (O.S) 0.01 (1.3) ..0.61 (..0.4) 0,51 0.2 0 ..0.1 0.15 ..0.4 0.1 0.3 ..0.1 ..0.2 0.03 0.1 ..0.1 ..0.3 0.1 0 0.2 0.3 OJ U1 39.1 37.4 37.7 46.8 53.5 39.4 38.4 37.3 40.4 48.2 51.7 34.7 46.7 38.8 . 51.3 48.4 36.8 46.4 33.3 47.3 56.2 42.0 31.6 39.9 49.9 57.2 48.5 57.1 57.1 54.6 ss.o 52.5 54.2 55.1 57.4 47.9 52.0 37.0 49.6 62.1 53.3 63.8 54.3 50.6 53.7 54.5 53.4 49.8 55.4 54.0 56.3 52.9 56.9 53.5 58.7 57.4 52.9 57.3 64.1 54.2 58.6 46.7 49.7 52.9 52.6 60.4 57.7 52.7 54.6 62.0 56.4 .51.1 55.6 53.9 53.3 57.0 54.3 49.0 55.4 20.1 23.5 19.6 25.1 25.5 15.8 24.3 17.8 22.1 16.5 22.6 28.8 9.9 29.2 33.8 34.2 26.0 23.4 19.2 22.3 21.4 30.4 17.2 30.7 2%.7 22.7 24.2 30.0 21.4 14.7 29.5 20.1 30.5 17.6 20.6 27.1 2%.5 23.4 24.3 21.4 29.6 22.8 18.8 19.9 25.2 13.2 25.5 27.0 24.0 19.1 23.9 19.5 24.5 20.1 26.5 27.5 19.1 25.7 31.6 34.0 24.0 24.7 37.2 13.4 38.6 15.8 28.9 28.0 21.6 36.7 19.1 18.2 26.2 17.1 21.9 25.5 22.1 20.6 36.8 22.4 11.0 24.6 21.2 28.6 15.5 23.3 26.7 24.2 24.7 22.3 21.5 37.8 21.9 36.9 14.7 21.1 19.0 19.2 10.7 13.3 17.0 25.9 30.9 26.9 17.3 24.3 21.2 23.1 27 29 26 27 IS 19 30 25 28 19 24 26 24 u 26 29 14 31 19 25 25 21 21 17 27 25 33 26 20 24 36 34 22 27 11 23 28 20 28 22 18 21 25 24 32 16 30 IS 22 26 26 20 26 20 46 48 41 49 47 49 45 41 42 40 43 43 45 39 61 44 45 51 44 54 46 45 34 54 40 36 64 46 42 42 58 44 48 48 36 42 43 49 46 41 ss 48 46 47 59 35 46 48 41 40 44 41 40 39 48 49 44 52 49 51 49 47 45 40 46 44 46 41 59 47 5I 53 46 ss 49 46 39 54 43 44 64 51 44 45 58 so so 49 44 49 48 48 46 46 56 so 47 48 59 34 47 so 44 46 52 44 44 43 41 39 40 35 42 31 46 39 39 33 31 44 34 34 53 41 28 43 29 so 43 38 34 52 35 40 49 46 36 42 48 45 36 46 33 38 40 31 40 34 39 25 34 38 Sl 23 40 39 40 40 41 34 31 38 6.4 314.0 19.4 10.7 117.4 7.1 14.8 174.7 ·13.5 3.3 230.3 •4.7 10.8 268.2 29.3 -6.6 333.2 37.9 -8.8 412.8 17.3 5.2 268.4 19.9 ·ll-2 408.6 3.8 6.1 324.0 16.8 4.5 320.8 8.3 ll.1 455.9 19.1 1.8 91.5 •7.2 7.2 364.8 34.6 5.7 340.4 15.6 6.4 341.4 2.4 8.4 199.8 -8.3 7.1 22%.8 0.2 3.8 229.7 19.4 ·1.3 341.2 6.7 ·2.4 269.3 2.4 ·7.1 327.0 10.9 4.7 274.2 4.2 14.2 314.6 2.3 13.9 391.1 3.4 0.2 186.8 14.2 .0.4 255.7 10.9 ·16.0 275.9 20.8 12.6 295.7 17.4 -4.5 225.9 13.3 10.2 380.9 14.5 ·15.3 378.5 16.6 10.8 145.3 20.6 6.9 317.8 6.1 5.8 160.1 6.5 9.2 207.8 36.4 4.3 274.7 28 33.9 120.1 42.3 ·8.6 304.9 .o.s 1.1 257.9 8 45.4 291.8 4.7 13.4 14%.4 15.6 ·17.6 170.1 ·3.1 34.6 107.0 •4.2 18.3 219.6 %.5 4.1 129.0 11.7 ·3.5 475.7 4.7 17.4 %62.4 10.1 7.% 284.6 10.6 27.4 301.1 8.7 8.0 160.6 ·2.6 15.1 221.3 -5.5 1.7 218.3 ·3.5 •4.9 259.5 11.4 1734 82.5 4.45 10671 23.4 0.72 10.37 6203 . 24.3 0.7 10.75 3251 59.7 0.56 7.24 3457 44.7 0 . .51 7.43 3339 77.9 0.34 8.18 1168 77.2 1.39 7.74 4940 85.4 0.66 10.74 1%94 79 3.86 9.56 3154 84.2 0.56 10.48 2504 872 35943 1706 1636 1394 5589 4206 4964 2274 2669 2170 36~7 1522 1497 8218 1964 56.6 3.54 8.37 78.9 11.35 5.97 7.4 0.8 11.77 87 3.02 7.65 88.2 1.52 7.36 85.6 8.549 5.24 62.4 0.74 9.52 29.2 1.02 12.08 48.4 0.59 9.84 64.4 1.26 I 0.69 55.8 3.8 7.66 58.5 10.66 6.6 69.5 o.s 8.54 82.4 1.01 6.25 80.6 15.65 6. 97 42.9 0.29 16.03 70.1 0.92 7.86 2939 76.3 o.s 9.43 4021 87.2 1.19 10.24 3557 53.8 1.09 8.27 1516 54.4 2.24 8.6 1474 88.8 4.76 6.09 5171 27.9 0.54 7.73 1550 76.3 0.81 6.31 9429 4927 3294 15631 1984 3594 1126 7620 6585 17795 3472 31.2 49.1 62.7 19.8 49.1 57 87.6 24.7 19.4 16.8 27.2 0.27 15.26 1.12 10.21 0.83 9.4 0.73 ll.58 16.97 7.97 0.68 15.06 0.47 - 5.35 0.86 9.5 1.33 ll.25 0.85 12.46 0.87 8 15344 17.3 0.51 8.63 1092 78.1 2.46 5.83 2738 61.4 0.42 6.41 2259 76.4 0.95 9.28 3005 72.5 0.74 8.53 7906 34.5 1.16 14.3 5325 . 40.1 1.04 10.1 4658 39.6 1.54 8.94 3286 48.1 0.94 8.28 1.7 69.16 ·10.0 0.0 69.57 0.0 9.4 79.72 0.0 4.9 80.50 •7.0 o.o 81.67 ·16.0 -8.7 71.23 18.0 ·1.7 84.55 0.0 0.0 75.92 ·16.0 •7.1 •7.3 ·23.0 ·10.8 0.0 ·2;7 •3.1 o.o 10.7 6.0 8.9 13.0 0.0 ·16.0 2.8 o.o 0.0 0.0 o.o o.o ·1.6 0.0 12.5 13.0 •7.1 0.0 7.3 0.0 5.7 0.0 0.0 o.o 3.9 -4.0 0.0 6.0 •7.4 ·16.0 0.0 o.o 2.8 -4.0 1.7 1.0 0.0 •7.0 .0.0 o.o ·14.3 ·14.0 1.9 ·14.0 o.o o.o -4.0 ·23.0 0.0 0.0 ·10.5 ·16.0 7.8 5.0 ·2.2 ·23.0 8.3 ·14.0 12.8 ·10.0 2.3 •4.0 0.0 84.52. ·16.0 ·9.6 72.25 11.0 10.7 81.12 •7.0 0.0 88.22 0.0 ·1.5 83.73 0.0 5.7 79.86 o.o 0.0 87.52 •7.0 -7.9 77.88 ·16.0 0.0 ._75.02 o.o 2368 64.14 2018 288 15.66 3659 335 32.94 2033 274 -8.97 3676 388 -6.73 ll74 301 11.48 4549 437 1.86 1241 379 19.18 2839 390 6.85 1957 458 7.01 839 186 52.46 4466 347 1.46 1713 399 19.1 1590 362 3. 72 1420 259 18.26 3308 233 •ll.07 1569 235 6.33 5660 331 4.09 2885 3ll 8.74 1682 291 1.75 2794 278 ·0.36 3850 375 16.46 1364 406 5.45 1391 273 3%.52 4111 296 16.54 1851 270 ·0.74 3189 327 10.85 3428 254 16.51 4410 369 16.04 2415 4%6 8.67 1474 301 43.33 336 6.67 293 30.8 244 25.77 323 ll 312 60.82 319 ·1.54 353 22.15 333 13.27 289 65.14 281 47.89 1652 1534 3368 4560 2275 2888 1893 3076 ll98 2068 2340 139 28.7 16993 300 32.74 2766 323 41.46 2630 442 ·12.48 1396 302 18.43 2807 324 10.96 2447 387 20.56 2419 %63 S0.%9 3482 277 22.57 5640 348 28.89 3188 280 11.11 336% 81.3 0.83 13.21 70.2 1.40 11.84 ·5.24 47.4 0.83 7.48 ·5.2 • 74.1 0.47 7.72 -6.05 70.8 0.39 9.15 ·11.59 81.9 1.97 8.61 ·3.25 89.5 0.76 12.61 ·10.62 76.1 4.78 10.72 ·12.05 87.9 0.58 10.52 ·16.07 67.6 4.43 10.0% ·18.79 77.4 14.79 6.10 ·5.85 50.3 0.90 15.00 ·16.14 87.8 3.28 8.39 ·5.6 87.9 1.74 8.14 4.26 84.2 9.66 6.14 4.7 99.9 1.05 10.46 -6.53 61.2 1.07 15.38 5.4 48.2 0.84 10.89 ·17.09 51.1 1.36 10.61 ·5.63 79.7 4.04 8.70 •%.46 63.3 11.91 7.18 ·0.76 73.4 0.48 8.71 -15.48 88.1 1.09 7.54 1.16 77 17.04 7.31 3.03 81.6 0.38 16.63 -8.2 71.9 1.08 7.59 •7.02 75.7 0.48 10.06 1.51 95.3 0.94 9.88 -8.96 50.6 1.17 9.37 ·17.3% 39.9 2.49 9.9% ·7.9% 85.3 5.63 6.11 1. 79 83.8 0.59 7.69 ·14.46 78 0.57 5.90 8.9 98.4 0.61 14.94 ·25.86 48.2 0.86 11.29 ·9.42 84.9 0.79 9.56 -6.77 91.5 0.97 13.30 ·16.4 54.8 21.12 8.45 ·2.65 69.8 0.64 15.78 ·23.17 88.3 0.48 5.94 0.04 83.7 0.82 10.91 -24.54 45 1.50 13.95 ·5.18 17.4 1.06 14.44 ·22.19 33.4 0.92 8.14 •5.77 90.1 0.54 8.69 ·18.41 72.7 2.96 6.38 •1.05 66.4 0.49 7.23 5.42 74.1 0.94 10.21 •4.42 86.7 0.62 9.90 . •3.52 74.8 1.46 15.46 ·11.47 39.6 1.03 10.84 ·1%.6 52.4 1.83 9.69 ·14.18 45.9 0.96 9.79 •2.35 65 73 80 76 80 79 85 76 78 74 82 81 78 78 73 82 62 86 77 75 78 79 76 78 85 78 77 88 78 78 67 8% 73 80 83 77 45 81 70 79 80 71 73 78 82 83 77 81 79 84 83 70 82 4.9 139.0 16.2 17615 18.5 1.01 10.14 ·18.8 89.86 ·23.0 237 60.14 3510 74.6 1.31 10.3l ·14.93 83 8.5 313.8 14.3 1859 Appendix 11. Data for Dependent and Independent Variables 0. 0 360.3 7.29 4.0 2%1.8 69.65 0.0 319.0 43.67 o.o 245.9 1.20 2.0 341.4 43.94 •7.9 295.0 28.81 435.6 ·3.02 0.3 333.1 13.95 •7.9 471.5 9.37 •7.9 375.5 3.47 410.1 15.01 445.9 -4.92 244.6 28.29 371.6 10.28 387.5 2.19 375.5 8.41 293.5 36.30 268.3 ·0.92 269.8 31.49 272.9 -9.74 328.7 6.72 7.4 257.1 ·1.49 o.o 294.6 10.25 4.0 400.0 15.05 4.0 425.8 3.34 0.3 275.9 27.57 0.0 317.4 18.59 8.9 315.5 22.04 •7.9 317.5 ·2.60 o.o 279.7 30.69 1.0 340.2 ·0.98 10.0 460.8 14.23 0.0 318.7 41.40 0.0 359.6 18.33 0.0 339.6 56.64 0.0 262.3 50.12 0.3 328.7 15.86 ·11.9 381.0 96.05 0.0 302.4 ·1.76 •7.9 366.4 13.74 350.0 21.35 286.2 33.18 o.o 261.0 16.89 1847 83.4. 5.60 7.92 2297 80.8 1.00 13.09 5.77 2039 71.9 1.42 13.42 ·1.51 4021 43.1 1.07 7.21 •4.38 2153 75.1 0.41 7.64 0.36 4908 55.6 0.44 12.43 ·11.53 1210 82 1.77 6.69 1.45 4631 89 0.71 12.56 ·5.72 1234 76.6 4.78 8.79 •7.06 2936 88 0.55 11.23 ·17.62 2017 68.4 4.48 10.43 ·13.64 868 76.2 11.66 6.18 1.12 3434 66.7 0.93 12.40 ·14.34 1763 87 3.34 8.90 1.10 1MO 88.4 1.42 7.65 7.95 1500 84.9 8.71 6.00 10.84 3753 97.9 1.61 12.54 -9.94 1611 60.6 0.88 14.86 8.76 5682 48.7 0.75 13.18 ·17.64 3025 52.5 1.16 13.24 -6.22 1684 80.2 3.71 9.04 ·0.03 2819 65.4 8.62 7.44 ·0.47 3954 74 0.50 9.56 -9.52 1424 87.1 0.96 7.92 5.26 1392 77.8 16.81 6.35 5.88 4241 82.4 1564 79.7 3426 73.8 3708 95.8 4523 50.8 2343 43.5 1533 86 1675 83.3 1651 75.1 0.39 18.16 ·2.00 0.86 8.36 •7.01 0.47 9.91 ·15.9% 1.01 9.47 . 6.40 1.32 9.13 ·14.29 2.03 10.51 ·1.49 5.34 6.01 4.98 0.63 8.92 -8.89 0.64 6.57 19.82 3757 . 98.5 0.63 16.19 ·15.83 3985 62.1 0.83 12.27 ·5.57 2395 85 0.85 10.95 ·5.44 3134 93 1.41 16.31 ·11.48 1836 56.1 17.73 7.99 ·1.68 79.22 83.8% 85.68 80.33 85.39 78.38 82.25 8%.3 81.45 80.31 77.21 78.36 13.0 419 12.0 ·1.0 12.0 6.0 68.27 ·11.0 71.27 10.0 77.89 ·13.7 70.48 9.0 76.29 ·16.0 66.85 86.10 74.04 76.13 66.95 61.99 72.73 56.02 78.34 74.40 67.84 75.60 74.63 69.29 75.35 78.43 69.03 73.95 72.29 76.11 74.93 %2.0 63.90 9.7 73.62 ·13.0 67.06 4.0 79.68 2.0 73.20 o.o 75.61 ·11.7 68.80 .9.0 75.13 10.0 209 333 3.7 358 31.1 354 ·2.2 336 13.9 414 ·O.S 375 13.3 466 0.2 43% 10.8 475 13.6 483 -4.9 288 44.7 390 6.8 411 3.8 403 4.9 299 5.3 439 66.3 302 18.9 366 8.9 392 24.4 326 12 338 15 411 5.7 441 7.8 324 6.9 352 9 332 13.7 351 ·7.7 318 22.8 426 13.9 491 8.4 341 33.2 386 14.9 303 ·1.9 300 17.6 396 18.9 466 37.9 354 ·0.3 2287 81.4 1.00 13.09 8.29 2042 71.5 1.42 13.42 ·1.27 3084 55.4 1.07 7.21 2.31 2179 73.7 0.41 7.64 4.14 4571 64.2 0.44 12.43 -4.94 1318 82.7 1.77 6.69 2.72 4540 88.3 0.71 12.56 ·8.26 1293 79.3 4.78 8.79 -5.61 2852 92.7 0.55 11.23 ·17.85 1980 67.8 4.48 10.43 ·10.68 987 78.8 11.66 6.18 %.03 3392 65.4 0.93 12.40 -9.45 1840 87.2 3.34 8.90 %.71 1926 89.4 1.42 7.65 11.34 1644 85.9 8.71 6.00 15.37 4088 100 1.61 12.54 ·9.94 1611 59.2 0.88 14.86 10.17 5466 49.6 0.75 13.18 -13.64 3361 53.5 1.16 13.24 -0.04 1788 80.4 3.71 9.04 0.27 2877 66.5 8.62 7.44 3.61 4109 74.3 0.50 9.56 ·9.84 1773 88.3 0.96 7.92 9.42 1514 78.9 16.81 6.35 8.07 4274 82.7 0.39 18.16 -%.96 1487 82.5 0.86 8.36 -1.76 3710 73 0.47 9.91 10.%7 3776 95.4 1.01 9.47 ·14.80 4724 55.% 1.32 9.13 -9.20 %449 49.2 %.03 10.51 6.06 1648 86.4 5.34 6.01 7.84 1620 82.4 0.63 8.92 ·2.16 1761 75.8 0.64 6.57 24.84 3%97 98.4 0.63 16.19 ·13.83 4062 63.1 0.83 12.27 ·6.85 2449 83.6 0.85 10.95 ·8.06 3195 93.9 1.41 16.31 ·8.50 1934 58.8 17.73 7.99 0.03 33.40 34.58 11.08 11.05 30.8 12.8 29.82 38.91 12.20 7.67 29.6 14.4 24.45 45.00 12.07 9.55 26.1 13.5 30.64 35.90 13.24 11.03 21.9 19.% 83.78 12.78 31.86 34.41 11.07 9.88 24.0 10.8 68.33 7.16 ·22.37 40.52 16.54 13.41 17.1 18.4 82.50 5.10. 23.62 44.56 1%.02 14.69 16.6 16.8 69.59 8.35 26.38 39.97 15.59 9.72 22.9 16.0 76.95 9.98 23.10 44.25 17.14 5.53 25.6 11.5 73.14 6.78 %4.93 38.67 18.%3 11.40 25.5 13.2 73.33 10.91 %5.47 42.65 15.46 5.51 25.5 11.5 85.62 5.13 %0.60 52.2% 14.02 8.03 24.7 10.6 78.77 7.98 %4.71 42.81 14.93 9.56 %%.6 14.5 67.79 6.71 %4.05 38.49 16.88 13.88 17.9 17.2 6%.94 6.90 25.04 39.03 14.75 14.28 9.6 33.8 78.81 7.06 35.97 36.32 1%.74 7.92 27.4 8.8 9.48 33.25 34.68 13.42 9.17 28.0 14.5 7.84 35.4% 37.6% 14.4% 4.70 27.6 14.7 7.41 35.23 37.51 12.44 7.41. 24.1 16.9 7.22 %2.18 37.87 16.57 16.16 21.6 18.1 4.59 18.46 31.77 %2.84 %2.34 18.7 %0.8 8.40 20.61 45.04 13.96 12.00 19.6 19.8 8.31 30.67 38.93 12.90 9.19 22.7 15.7 5.84 %0.60. 35.13 21.03 16.45 12.6 27.8 62.6 61.38 .18.97 62.2 54.59 25.92 74.9 46.84 28.00 73.1 44.99 33.89 77.4 45.23 27.46 63.7 5%.31 28.48 69.3 47.98 27.54 66.9 50.%9 33.23 79.4 43.38 32.60 7%.7 46.51 26.44 72.9 45.%8 40.70 77.0 49.94 25.85 71.5 45.24 35. 70.6 51.54 %7.09 58.5 50.24 17.07 70.9 50.79 27.93 60.9 57.64 24.31 79.1 45.09 27.81 75.4 46.37 30.42 63.1 56.34 %4.57 80.7 52.78 19.60 72.9 42.94 24.22 70.2 49.38 32.95 68.5 54.% 21.33 5.12 %0.72 44.69 15.73 13.73 %2.8 6.80 30.17 39.12 15.34 8.57 14.8 7.31 26.83 39.29 13.56 13.01 15.0 7.13 %7.37 45.03 11.35 9.1% 20.5 7.58 30.34 40.87 14.89 6.32 26.1 5.72 29.44 44.01 13.%0 7.63 26.0 8.%5 %4.40 36.21 15.96 15.18 14.6 13.91 30.28 38.79 9.53 7.49 29.1 7.19 %5.37 41.95 15.12 10.36 18.3 8.96 %9.72 44.81 6.13 10.38 18.1 16.9 71.7 5%.68 26.98 12.5 73.3 Sl.%6 25.12 19.7 68.8 47.32 30.17 14.3 78.% 47.8 25.20 14.4 67.4 43.31 28.38 13.1 72.9 49.49 %5.34 %7.7 60.1 51.86 19.62 11.4 78.1 43.87 34.32 19.3 66.3 46.74 %7.61 17.5 76.1 40.37 21.65 79.85 7.26 %5.78 41.74 14.10 11.12 %5.9 81.%7 4.27 %8.76 37.06 17.35 12.57 26.3 77.00 7.93 %3.30 40.77 14.13 13.88 24.8 77.14 7.82 27.57 43.16 13.76 8.42 23.5 13.1 69.8 15.0 67.9 15.3 73.4 16.6 71.8 3287 71.7 1164 88.7 2067 83.7 2281 45.4 0.68 17.34 ·17.19 73.54 ·16.0 434 20.2 3357 70.7 0.68 17.34 ·16.%5 82.61 10.80 33.81 40.88 9.38 5.13 28.9 13.2 78.9 0.39 5.79 9.08 79.05 71.83 10.0 359 %.9 1326 88.9 0.39 5.79 1%.49 68.59 6.61 %7.81 40.61 15.%5 9.72 20.9 18.% 75.1 0.75 10.93 ·12.96 79.51 64.20 6.0 371 43.8 211% 84.6 0.75 10.93 ·11.87 67.70 6.86 22.97 44.50 15.07 10.60 24.6 13.4 70.7 1.58 15.%4 ·2.87 68.55 59.01 24.7 2367 44.4 1.58 15.24 ·1.23 65.84 10.89 28.91 38.71 13.86 7.62 29.9 . 11.3 63.9 4.0 149.4 52.22 17327 16.8 0.99 14.39 ·14.24 73.43 173 16.1 16905 18.4 0.99 14.39 ·12.50 66.89 6.19 20.20 34.46 21.58 17.58 21.7 13.4 68.8 1.0 321.8 20.63 2806 33.5 289.2 46.74 2646 90.1 550.1 8.21 298.4 15.16 323.6 8.29 0.3 421.3 32.61 0.3 279.5 38.88 0.0 282.0 21.31 354.3 28.89 291.6 22.59 1469 70.6 2913 63.9 2623 71.6 2476 87.6 3528 74.9 5581 41.6 3136 53.9 3419 47.1 0.91 8.89 0.73 76.46 0.61 9.79 ·14.64 84.37 3.01 0.49 6.09 1.62 81.77 8.05 9.32 80.92 0.94 11.60 1.94 0.66 9.76 1.98 1.43 16.14 ·3.56 0.90 11.31 ·10.90 1.75 10.21 -4.56 0.89 11.28 3.41 84.19 85.09 84.51 5.8 3108 34.3 0.91 8.89 8.21 9.64 28.09 47.01 9.49 5.78 24.8 14.8 75.1 35.6 2668 90.1 0.61 9.79 ·14.37 26.66 41.97 14.36 7.92 25.5 11.7 77.0 9.2 1556 71.6 3.01 6.09 4.46 %1.6 3052 64.1 0.49 8.05 11.16 %1.5 15.5 . 24.9 19.6 10.1 2614 70.6 2375 86.7 3552 74 5495 45.5 3190 56.6 0.94 11.60 0.97 0.66 9.76 4.15 1.43 16.14 ·0.86 0.90 11.31 ·2.00 1.75 10.21 ·0.90 27.56 35.49 13.63 14.62 21.7 17.9 70.8 22.37 41.37 17.18 12.06- 21.8 16.9 73.6 %5.67 42.96 15.26 8.92 27.16 38.91 13.71 13.37 27.14 34.49 18.74 14.38 34.59 40.70 8.72 10.47 27.09 38.05 17.00 10.%5 21.1 23.2 22.6 14.9 21.4 20.0 75.9 17.8 73.4 14.0 71.9 13.2 64.6 13.2 71.4 35.82 39.91 13.19 4.08 27.8 10.6 75.3 00 -..] 88 Appendix 12. Percentage Change in Mill Levy, Assessed Value, Taxable Base and Resident-Borne Taxable Value 19.4 8.78 7.29 7.1 100 69.65 ·13.5 64.14 43.67 15.7 -9.5 3.7 ·4.7 15.66 1.20 31.1 6.4 0.0 10.7 ·10.0 14.8 0.0 3.3 0.0 29.3 32.94 43.94 ·2.2 10.8 -7.0 13.9 .o.s 13.3 ·6.6 ·16.0 ·8.8 18.0 5.2 0.0 0.0 4.0 0.0 0.0 2.0 •7.9 26.3 0.3 37.9 -8.97 28.81 17.3 ·6.73 -3.02 19.9 11.48 13.95 3.8 1.86 9.37 0.2 -11.2 ·16.0 •7.9 16.8 19.18 3.47 10.8 6.1 -16.0 •7.9 4.5 -23.0 ·11.9 11.1 0.0 0.0 8.3 6.85 15.01 13.6 19.1 7.01 •4.92 •4.9 -7.2 52.46 28.29 44.7 1.8 ·23.0 ·11.9 34.6 1.46 10.28 6.8 7.2 0.0 0.0 15.6 19.1 2.19 3.8 5.7 6.0 16.6 2.4 3.72 8.41 4.9 6.4 13.0 25.7 ·8.3 18.26 36.30 5.3 8.4 ·16.0 7.1 0.0 •7.9 0.0 0.0 0.0 0.3 7.4 0.0 4.0 4.0 0.3 0.0 8.9 0.2 ·11.07 -0.92 66.3 6.33 31.49 18;9 3.8 o.o 0.0 0.0 6.7 4.09 ·9.74 8.9 ·1.3 2.4 8.74 6.72 •2.4 10.9 1.75 -1.49 4.2 -0.36 10.25 2.3 16.46 15.05 3.4 5.45 3.34 12 lS 5.7 7.8 7.1 4.7 14.2 13.9 14.2 32.52 27.57 6.9 0.2 13.0 0.0 0.0 o.o 0.0 •4.0 6.0 10.9 16.54 18.59 9 ·0.4 20.8 -0.74 22.04 13.7 ·16.0 17.4 10.85 -2.60 13.3 16.51 30.69 14.5 16.04 -0.98 16.6 8.67 14.23 20.6 43.33 41.40 6.1 6.67 18.33 6.5 30.8 56.64 36.4 25.77 50.12 28 11 15.86 7.7 12.6 -16.0 •7.9 22.8 •4.5 o.o 13.9 10.2 •4.0 8.4 ·15.3 1.0 33.2 10.8 -7.0 6.9 o.o 5.8 ·14.0 9.2 ·14.0 18.9 4.3 o.o o.o 1.0 10.0 0.0 0.0 o.o 0.0 0.3 60.82 96.05 37.9 33.9 ·23.0 ·11.9 -o.s -1.54 ·1.76 .o.3 -8.6 o.o o.o 8 22.15 13.74 20.2 1.1 -16.0 •7.9 4.7 13.27 21.35 15.6 65.14 33.18 -3.1 47.89 16.89 2.9 45.4 s.o s.o 43.8 13.4 ·23.0 ·11.9 24.7 ·17.6 -14.0 0.0 ·4.2 28.7 52.22 16.1 34.6 ·10.0 2.5 32.74 20.63 5.8 18.3 -4.0 4.0 1.0 11.7 41.46 46.74 4.7 ·12.48 8.21 10.1 18.43 15.16 10.6 10.96 8.29 8.7 20.56 32.61 ·2.6 50.29 38.88 .s.s 22.57 21.31 ·3.5 . 28.89 28.89 11.4 11.11 22.59 55.45 8.31 9.88 4.1 ·16.0 •7.9 ·3.5 11.0 13.2 17.4 -7.0 0.0 7.2 0.0 0.0 27.4 o.o 8.0 0.0 15.1 -7.0 0.3 0.3 0.0 1.7 ·16.0 ·7 .9 -4.9 o.o 0.0 4.9 ·23.0 ·11.9 8.5 ·10.0 ·3.9 12.5 ·41.2 10.1 o.s -7.9 6.6 ·3 1 4.9 2 -1.9 9.9 ·23.7 5.9 33.5 3.1 1.8 1.4 •4.6 7.9 -4.1 -0.6 0 3.7 ·1.6 •4.3 7.7 -10 -3.8 3.4 3.1 3.5 0.73 0.36 247.44 -0.62 o.s 188.89 2.42 -1.39 ·20.6 -9.07 24.36 65.77 1.35 ·3.06 -9.11 ·21.47 6.47 6.09 0.12 0.98 -0.12 4.8 -0.56 ·0.45 -0.34 -3.67 0.66 2.09 1.41 4.39 0.11 ·1.02 6.43 19.43 1.18 ·2.19 1.35 ·1.9 -l.SS 2.36 1.03 ·23.1 0.7 ·1.9 579.73 32.6 ·2.55 0.62 0.4 2.9 1.9 2.4 0.92 -0.91 ·0.11 0.35 ·2.8 3.1 16.2 1 -0.34 0.57 1.47 -0.33 1.9 5.6 8.3 1.2 -1.64 0.83 1.3 -0.12 14 5.3 ·10.4 -7.1 13.5 2.7 0.4 4.9 0.1 1.4 -1.1 ·6 5.5 3.9 0.9 1.3 2.7 ·1.6 4.4 22.6 0.1 7.7 -so 3.2 .s.8 ·lS.S 8.5 7.4 ·2.6 ·1.9 7.6 0.9 3.7 3.2 5.1 -2.6 1.3 8.2 24 2.6 ·3 0 4 1.4 ·1 7.6 11.5 -7.4 ·12.6 5.3 4.9 -0.6 -0.3 8.5 ·1.9 ·3 1.6 6.9 ·1.5 ·2.8 0 ·2.5 9.4 2 2.8 2 -2.6 1.4 -0.1 0.6 -1.9 5.2 8 3.8 0.7 7.2 -4.3 2.4 -5.4 1.3 0.8 -1 -5.1 -1.6 1.7 1.7 -0.9 7.4 60.1 ·2 2.15 0 109.59 -0.98 -0.33 2.4 -0.41 1.04 0.41 5.4 ·20.65 2.74 ·1.14 2.2 42.83 0.63 0.37 -0.12 0.8 5.6 l.S 1 3,4 ·3.1 0.7 1.2 2.3 -0.7 8.21 3.32 5.61 0.82 6.92 ·1.14 -4.47 1.04 0 1.68 0.81 -0.4 1.84 -0.45 1.29 0.13 90.21 0.98 ·0.12 0.49 2.57 10.85 1.38 2.1 -0.79 ·2.51 ·3.39 9.29 0.52 -0.52 -5.95 0.4 3.54 -26.65 9.02 11.n -3.94 o.82 o.ils 200.36 -0.6 -0.72 2.23 -3.72 0.53 215.38 0.1 -0.51 ·1.83 28.84 0.97 35.41 0.12 ·1.41 362.12 1.64 1.18 11.61 2.37 3.74 22.46 2.72 -3.21 0.8 0.45 -0.34 0 -0.12 131.96 0.89 3.57 -3.45 22.79 0.3 0 -0.44 -6.91 -2.89 2.12 8.14 -3.77 1.41 -3.01 -3.37 -0.56 19.59 1.04 -1.14 116.81 0.13 -0.27 -1.25 5.05 5.05 32.32 2.86 -4.57 2.61 3.9