Property tax delinquencies effects on revenue volatility in American cities: Examining the outcomes of the Great Recession.

Property tax delinquencies effects on revenue volatility in American cities: Examining the outcomes of the Great Recession. Olha Krupa, Ph.D. Assistant Professor Institute of Public Service Seattle University 901 12 th Ave, P.O. Box 222000, Seattle WA 98122 Email: krupao@seattleu.edu Phone: 206-296-2509 Kenneth A. Kriz Regents Distinguished Professor of Public Finance Director, Kansas Public Finance Center Wichita State University 1845 Fairmount Street Wichita, KS 67260-0155 Email: ken.kriz@wichita.edu Phone: 316.978.6959 Abstract This study presents a stylized interactive model of home prices, property tax collection rates and socioeconomic indicators in the US cities before and during the Great Recession. Using a panel of the 147 largest American cities in 2003-2012, this research explores how differences in incomes, population growth, home price dynamics, and elements of the local property tax systems affect the local government finances in recession periods. Correcting for endogeneity in these interactions by using seemingly unrelated regressions and a three stage least squares method, the study finds that economic downturns result in two effects: the first effect is that the recession has a negative effect on housing prices, which is translated into lower assessed valuations. The second effect compounds the negative fiscal impact by reducing property tax collection rates and increasing delinquencies. However, we find no evidence of a take back reaction from jurisdictions in setting millage rates. The first two effects can induce much larger fiscal impacts on local governments during recessions. Key Words: revenue volatility, property tax, assessed values, collection rates, delinquency, Great Recession.

Introduction Property tax income provides the most significant funding source for local governments in the United States. That income is crucial for essential public services including police and fire departments, and numerous social programs. Because cities are limited in their debt issuance capacity, the stability of income from property taxes is necessary for cities to provide effective municipal services. Numerous factors, including housing prices, foreclosure rates, unemployment, income growth, and delinquency rates in property tax collection influence the extent to which that income varies. Shortfalls in property tax revenue undercut a city s ability to fund these essential public services. During the Great Recession, at both the federal and state levels, taxes on income and on consumption of goods and services were the hardest-hit sources of government revenue. At the local level, property tax revenues did not decline as much: therefore, at least partially compensating for losses in other revenue categories (Ihlanfeldt 2011, Alm et al 2011, Mikesell and Cheoul 2013, Mikesell and Mullins 2013). Early studies of the recession agree that higher property tax assessments and growing tax burdens provided for an outstanding resilience in the property tax collections during the most recent economic downturn. The authors of those studies, however, recommend extensive follow up on their findings assuming that due to a significant time lag between property tax assessments and actual collections this documented robust property tax performance will weaken in the coming years. Among earlier studies of tax income during the Great Recession, there remains a gap in understanding how in the midst of the housing crisis property taxes contributed to local 2

revenues, and whether the earlier studies featured a data period too short to fully capture what happened with property taxation during the recession. Similarly, scholars do not yet understand how the specific interactions of the market and the administrative features of a municipality affect property tax revenues. In an effort to understand those factors on revenue volatility, this research proposes to answer the following questions: 1) How do declining property taxes collection rates, house values, and incomes in combination with employment patterns of city residents interact in the largest urban areas of the United States? 2) What factors make some cities consistently more successful in securing their necessary tax revenue than others, and how do regional and local variations contribute to these differences? 3) In anticipation of the next economic downturn, which policy recommendations could most effectively address and improve property tax administration? This research documents property tax collection trends over a decade of 2003-2012 across the largest American cities, noting any regional variation, explains the interaction effects between property tax collection rates, house values, income, and employment patterns of city residents leading to short-term volatility of local government revenues, and provides policy recommendations to city planners and budget analysts on how these institutional factors may affect their ability to fund public services during recessions. Figure 1 motivates the study by illustrating the relationship between the changing home prices and the relative growth in the significance of the property tax in total revenues of the largest US cities before and during the Great Recession. 3

Figure 1. Home price dynamics and the growth of property tax revenue share in total revenues in the 147 largest US Cities (2003-2012). 0.400 0.25 0.300 0.200 0.100 0.000-0.100 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 0.2 0.15 0.1 0.05-0.200 0 PT/total revenues House prices This study explores these relationships with a data panel for the 147 largest US cities in 2003-2013. We employ Seemingly Unrelated Regressions and Three-stage Least Squares methods to correct for endogeneity among variables. The results broadly support the model where economic downturns have two effects: the first effect is that the economic downturn negatively affects housing prices, which is translated into lower assessed valuations, and the second effect reduces property tax collection rates thereby compounding the negative fiscal impact. Therefore, in areas with substantial economic and housing market downturns, not only is the property tax levy falling, the collection of the levy is falling as well. This can induce much larger fiscal impacts on a jurisdiction. On a positive note, the study finds no evidence of a take back reaction from jurisdictions in setting tax rates. Raising tax rates during an economic downturn could create even more problematic economic conditions. 4

In what follows, the next section describes the literature on the interactions between property tax collections, assessed values, and basic socioeconomic indicators in American cities and counties before and during the Great Recession. Section III outlines the study methods and data. Section V describes the findings gained from the study, while section VII concludes the study with the discussion of results. A Review of the Literature: Revenue Volatility and Municipal Fiscal Health Indeed, local government reliance on the property tax rather than on more elastic revenue sources like income, sales, and excise taxes has so far, in any event helped local governments to avoid some of the more severe difficulties experienced by many other governments in the Great Recession that lasted officially from December 2007 to June 2009. James Alm, 2013. Studies of the Great Recession acknowledge the pivotal role of the property tax in stabilizing local revenue streams. Scholars document the unexpected resilience in property tax revenues during the recession years: not only property tax revenues did not significantly decline during the Great Recession, but also to some extent they have compensated the local governments for the loss of sales and income tax revenues. Because of a considerable time lag between the house price movements and the assessed value adjustments, the unpopular, inelastic, and inconvenient property tax has become the best revenue source during the Great Recession (Mikesell and Mullins, 2010; Ihlanfeldt, 2011; Alm et al., 2011; Alm, 2013; Mikesell and Liu, 2013; Ross et al, 2015). In what follows, we review the studies of the state and local government responses to the Great Recession in the United States. In surveying state and local tax revenue trends in 1989-2009, Mikesell and Mullins (2010) document significant declines in state and local income and sales taxes during the years of 5

the Great Recession. During those years the state personal income tax revenues contracted by 10.3 percent, accompanied by the 4.6 and 2.6 per cent declines in sales tax revenues at the state and local levels. A significant increase in property tax collections (6.3 per cent) stabilized other more volatile local revenue streams. Property tax collections have been the only bright spot in state and local finances during recession years attributing to the modest overall increase in local revenues and superior financial performance of local governments as compared to states. These contemporaneous findings, however, ought to be interpreted with caution due to time lags extending recession effects well beyond the years of the Great Recession. Lutz (2008) examines an effect between house prices and property tax collections and estimates a 3-year time lag between those based on a 1985-2005 Census. In his study the elasticity of local revenues to house price increases is 0.4. Lutz conjectures that the remaining 0.6 of the price effect is absorbed by millage rate reductions and other policy changes. A follow up study by Lutz, Byron, Molloy and Shan (2010) explores the effect of house price declines on state tax revenues based on similar data (1975-2007). These authors document varied state responses to fiscal crisis and study lags between the housing crisis outbreak and state revenue declines. Attributing most of the revenue decline to the Recession, the authors estimates of the effect of the housing market collapse on total state revenues are at fifteen billion dollars, or two percent of the total state revenues. Anticipating Recession, Chernick, Langley and Reschovsky (2011) estimate the effect of the housing market contraction on central city spending at seven percent on average, while the 6

that the hardest-hit inner cities may experience even more significant spending cuts. This study has been done based on a sample of synthetically constructed cities in 1997-2008. Boyd and Dadayan (2009; 2010), Center on Budget and Policy Priorities (2010) and Mikesell and Liu (2013) note considerable heterogeneity in state and local government responses to the Great Recession. This regional variation is attributed to their diverse economic nature, the diversity of the state property tax systems, differences in housing market dynamics, and in socioeconomic indicators. For example, studies indicate that states that had been significantly affected by the Great Recession more promptly resorted to spending cuts (Boyd and Dadayan, 2010). Mikesell and Liu (2013) connect tax assessments, tax revenues, and collection rates into one fiscal system in order to study financial consequences of the Great Recession on one hundred largest US cities over 1999-2011. On a city level, they note, property tax stability depends on property tax base and property tax revenue dynamics. They find that the housing crisis did not immediately result in reduced tax collections due to a considerable time lag. Expecting future revenue interruptions of the otherwise stable property tax collections, this study shows decreases in collection rates in select cities. Although the study finds no evidence of such interruptions in revenue collections just yet, these delinquency rate increases may be a sign of concern. In a study of cities and counties Ihlanfeldt (2011) estimates short-run first-year revenue elasticities to property tax rate changes based on a 1995-2009 Florida sample. These elasticities for cities and counties range between (0.097)-(0.212). Ihlanfeldt conjectures 7

that millage rate increases fully offset revenue shortfalls in Florida counties, while cities stabilize their property tax revenues both with rate increases and spending cuts. Exploring interactions between the effects of the housing crisis and property tax revenues based on a 1994-2008 Florida sample, Doerner and Ihlanfeldt (2010) establish the relationship between per-capita revenues and home values through several channels including assessed values, millage rates, tax base expansion from new construction, and non-tax revenue sources. A significant effect on property tax revenues comes from the tax base, they note, while there is no relationship between home prices and property tax revenues. This paper suggests that assessors better react to home price increases than decreases. However, if home price decreases were long-term and substantial, the assessors would eventually have to respond by lowering assessed valuations. In follow-up study, Ihlanfeldt and Willardsen (2014) examine a sample of Florida cities and counties organized into six groups: the large, mid-size and small cohorts of cities and counties over the period of 1995-2011. That study finds that cities more effectively address revenue shortfalls compared to counties, and they do that via millage rate increases and program cuts. Of all examined Florida cities, the largest ones were able to more fully offset millage rates compared to mid-size and small cities. Alm, Buschman and Sjoquist (2014) find that changes in per capita income, population, and employment levels significantly affect changes in the tax base, while increases in foreclosures negatively affecting property tax revenues through reductions in tax levy. They illustrate the negative impact of new foreclosures on property tax levies, tax bases, and property tax revenues in Georgia between 2006 and 2011. 8

Cromwell and Ihlanfeldt (2015) study the effects of the Great Recession through the lens of government competition literature. The hypothesis of their study is that local government responses depend on the degree of local government monopoly in local jurisdictions. Their elasticity of millage rates to property tax base are (0.27) and (0.34), and to intergovernmental transfers - (0.004) and 0.002 for Florida cities and counties. Again, the city millage rate increases and program expenditure cuts were moderately elastic with respect to revenue shortfalls, while intergovernmental transfer cuts were highly inelastic during the Great Recession. Ross, Yan and Johnson (2015) examine financials of the thirty five largest US cities in the wake of the Great Recession. They find that these cities fared surprisingly well primarily because of the stable revenues from property taxes. As intergovernmental grants and transfer flows dried out during recession years, cities resorted to depleting their net assets and borrowing more. During the Great Recession, these cities increasingly relied on property taxes and even more on deficit financing, previously an unconventional financing vehicle for the municipal governments. The study mentions that cities return to financial stability of pre-recession levels by about 2011. The Model The research model conjectures that declines in local economic conditions, housing prices, and property tax collection rates collectively attribute to revenue volatility and short-term budget shortfalls for local governments. This economic and fiscal effect is further exacerbated by weak economy in recession periods. Figure 2 provides a stylized 9

relationship model between housing market and economic conditions, property tax administration variables, and local government revenues from property taxes. Figure 2. Elements of the property tax system and property tax revenues: the channels of influence. House Prices Socioeconomic Conditions Assessed Values Tax Rates Rate Limitations Property Tax Revenues Assessment Levels Collection Levels These relationships can be expressed via the following system of equations: Houseprice!" = α + β! IncomePerCapita!" + β! Unempl!" + ε! log (AV)!" = γ + β! Houseprice!" + β! Ratio!" + ε! (1) TaxRate!" = δ + β! log (AV)!" + β! IncomePerCapita!" + β! Unempl!" + ε! FirstYearCollections!" = ζ + β! IncomePerCapita!" + β! Unempl!" + ε! where Houseprice!" is the first difference in average home price change; IncomePerCapita!" is the first difference in per capita income; Unempl!" is the first difference the rate of unemployment; log (AV)!" is the first difference in natural logarithms of total assessed values; Ratio!" is the assessment ratio, or the level of assessment in each jurisdiction; TaxRate!" is the first difference in total property tax rates 10

of an assessing jurisdiction, FirstYearCollections!" is the first difference in first-year collection rates of an assessing jurisdiction. α, γ, δ, and ζ are the constant terms. ε1, ε2, ε3, and ε4 are the error terms in the equations. Subscripts i, t identify city and year for each observation. Data and Methodology In order to test the hypotheses laid out above, we obtained data from 147 largest US cities (population over 150,000) over a ten-year period from 2003-2012. The cities were geographically distributed throughout the United States, as shown in Table 1. Western and Southeastern cities have somewhat greater representation in the sample compared to Northeastern and Southwestern cities. Table 1. Geographic Distribution of Sample Cities. Region Number of Cities in Sample Midwest 27 Northeast 12 Southeast 37 Southwest 22 West 49 Total Sample 147 Figure 2 illustrates the regional differences in home price dynamics and growth rates in property tax revenues as a percentage of total revenues based on these data. 11

Figure 2. Regional differences in house price dynamics and property tax revenue growth rates (2003=100%) 1.8 Southwest 1.8 Midwest 1.6 1.6 1.4 1.4 1.2 1.2 1 1 0.8 0.8 0.6 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 0.6 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 1.8 Northeast 1.8 West 1.6 1.6 1.4 1.4 1.2 1.2 1 0.8 0.6 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 1 0.8 0.6 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 New York City 1.8 1.6 1.4 1.2 1 0.8 0.6 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Property tax revenues House prices 12

Table 3 shows the variables in our dataset as well as the sources of the data. Table 3. Variables, Definitions, and Sources of Data. Variable Name Definition Source PTRevenue Property Tax Revenue ($ millions) Comprehensive Annual Financial Reports TotalRevenue Total Revenue ($ millions) Comprehensive Annual Financial Reports IncomePerCapita Per Capita Income US Bureau of the Census Unempl Unemployment Rate US Bureau of Labor Statistics Population City Population US Bureau of the Census AV Assessed Valuation ($ millions) Comprehensive Annual Financial Reports Ratio Average Assessment Ratio PTLevy Property Tax Levy ($ millions) Comprehensive Annual Financial Reports FirstYearCollections Collections of Property Taxes due in Current Year Comprehensive Annual Financial Reports TotalCollections Collections of Property Taxes due in Current and Prior Years Comprehensive Annual Financial Reports TaxRate Millage Rate ($/$1,000 in Assessed Value) Comprehensive Annual Financial Reports CPI Consumer Price Index All US Bureau of Labor Statistics Urban Consumers Houseprice FHFA Home Price Index ($ thousands) US Federal Housing Finance Agency The summary statistics for the data (Table 4) show marked deviation from normality for the revenue and levy variables. As these variables are lower bounded by zero, we estimate our models using the natural log of the values. The economic variables have approximately normal distributions. Population is similarly skewed, from the summary statistics it is clear that the sample mainly consists of mid-to-large sized cities. 13

Table 4. Descriptive Statistics for the Sample. Variable Mean Std. Dev. Minimum Maximum PTRevenue 472.01 2,030.73 2.66 19,513.45 TotalRevenue 1,394.98 6,806.83 33,584.00 67,005.64 IncomePerCapita 33,302.08 10,236.71 9,762.00 79,967.00 Unempl 6.99 3.07 2.00 24.80 Population 455,866.50 784,067.40 152,374.00 8,175,133.00 AV 26,861.90 44,735.28 12.87 411,135.20 Ratio 76.52 32.69 10.48 100.07 PTLevy 341.12 1,501.96 2.40 19,284.50 FirstYearCollections 95.95 4.65 55.15 128.15 TotalCollections 98.32 3.60 60.06 128.30 TaxRate 5.20 9.75 0.06 65.48 CPI 1.13 0.08 1.00 1.25 Houseprice 202.92 48.02 100.08 416.61 Figure 3 shows one additional problem for estimating our model. The mean of the financial variables (in this case assessed valuation) is not stationary over time. The problem is that there is likely to be serial correlation in observations. Therefore, we estimate a model in first differences, relating the change in the financial variables to changes in the predictor variables. This model formulation should also account for potential unobserved heterogeneity since the properties of the estimator resemble a mean difference fixedeffects estimator. 14

Figure 3. Mean of Assessed Valuation. A final issue concerns endogeneity. Previous papers (e.g., Mikesell and Liu, 2013) do not account for endogeneity in the tax rate setting process. Legislative bodies may act to take back some gains in assessed values by adjusting tax rates to achieve policy and/or political goals. Also, as we set forth in our theoretical model, some common exogenous variables exhibit effects not just on a single variable, but on the system as a whole. Therefore, we employ both Seemingly Unrelated Equations and three-stage least squares (3SLS) estimators to account for this endogeneity, specifying the following system: Houseprice!" = α + β! IncomePerCapita!" + β! Unempl!" + ε! log (AV)!" = γ + β! Houseprice!" + β! Ratio!" + ε! (1) TaxRate!" = δ + β! log (AV)!" + β! IncomePerCapita!" + β! Unempl!" + ε! FirstYearCollections!" = ζ + β! IncomePerCapita!" + β! Unempl!" + ε! Results The results for the three-stage least squares estimation of equation (1) are shown in Table 5. These results broadly support the model where economic downturns have two effects. 15

The first effect is that the downturn in the economy has a negative effect on housing prices, which is translated into lower assessed valuations. The second effect compounds the negative fiscal impact. An economic downturn also reduces property tax collection rates. Therefore, in areas with substantial downturns, not only is the property tax levy falling, the collection of the levy is falling. This can induce much larger fiscal impacts on a jurisdiction. We further find no evidence of a take back reaction from jurisdictions in setting tax rates. In the context of our inquiry, this should be seen as positive. Raising tax rates during an economic downturn could create even more problematic economic conditions. One caveat with this finding is that our model as with most other research in this field is a shortterm model only. There may be dynamic effects that reveal themselves over time. This would be a productive line of inquiry for future research. The goodness of fit measures shows that we have a fairly good model of housing prices and a sufficient model of the other dependent variables in the system. One may question the low or even negative R 2 value for the Tax Rate variable. Though this variable is important in the model to assess jurisdictional responses, it is difficult to model this variable. We examined tax rate changes in the sample. In most cases, the annual changes in tax rates were very small, so they are difficult to model. To us, this indicates support for our story, namely that there is no evidence that jurisdictions are changing tax rates in a substantial way in order to affect fiscal outcomes. 16

Table 5. Results of Estimation of Equation (1). Dependent Variable Independent Variable SUR Coefficient (Z-stat.) 3SLS Coefficient (Z-stat.) D.houseprice D.IncomePerCapita (000s) 0.210 0.214 (1.27) (1.29) D.Unempl -8.238-8.235 (18.15)** (18.14)** Constant 3.359 3.351 (4.31)** (4.30)** D.log(AV) D.houseprice 0.002 0.002 (6.77)** (3.60)** Ratio (000s) 0.142 0.141 (0.73) (0.72) Constant 0.030 0.030 (1.93) (1.93) D.TaxRate D.log(AV) 0.178 1.818 (1.69) (0.41) D.IncomePerCapita (000s) -0.591-0.574 (1.50) (0.52) D.Unempl 0.022 0.045 (1.58) (0.78) Constant 0.022-0.053 (0.92) (0.29) D_FirstYearCollections D.IncomePerCapita (000s) -0.085-0.085 (3.24)** (3.24)** D.Unempl -0.103-0.102 (2.26)* (2.24)* Constant 0.136 0.136 (1.74) (1.74) N 994 994 R2 for Dep. Variables (significance of χ2) D.houseprice 0.258** 0.258** D.log(AV) 0.043** 0.043** D.TaxRate 0.007* -0.234 D_FirstYearCollections 0.015* 0.015* * p<0.05; ** p<0.01 17

References: Alm, J. 2013. A convenient truth: Property taxes and revenue stability. Cityscape, 243-245. Alm, J., Buschman, R. D., & Sjoquist, D. L. 2014. Foreclosures and local government revenues from the property tax: The case of Georgia school districts. Regional Science and Urban Economics, 46: 1-11. Alm, James, Robert D. Buschman, and David L. Sjoquist. 2011. Rethinking local government reliance on the property tax." Regional Science and Urban Economics, 41(4): 320-331. Boyd, Donald J. and Lucy Dadayan. 2009. State Tax Decline in Early 2009 Was the Sharpest on Record. State Tax Notes, 54: 305-325. Boyd, Donald J. and Lucy Dadayan. 2010. Revenue declines less severe, but states fiscal crisis is far from over. Rockefeller Institute of Government, State University of New York at Albany. State Revenue Report, 79. Available online at: http://www.rockinst.org/pdf/government_finance/state_revenue_report/2010-04-16- SRR_79.pdf Center on Budget and Policy Priorities. 2010. An Update on State Budget Cuts. Available online at: http://www.cbpp.org/cms/index.cfm?fa=view&id=1214. Chernick, H., Langley, A., & Reschovsky, A. 2011. The impact of the Great Recession and the housing crisis on the financing of America's largest cities. Regional Science and Urban Economics, 41(4): 372-381. Cromwell, Erich and Keith Ihlanfeldt. 2015. Local government responses to exogenous shocks in revenue resources: evidence from Florida. National Tax Journal, 68 (2): 339-376. Doerner, William M. and Keith R. Ihlanfeldt. 2011. House Prices and Local Government Revenues. Regional Science and Urban Economics, 41: 332-342. Ihlanfeldt, K. R., & Willardsen, K. 2014. The millage rate offset and property tax revenue stability. Regional Science and Urban Economics, 46: 167-176. Ihlanfeldt, Kieth. 2011. How do cities and counties respond to changes in their property tax base? The Review of Regional Studies, 41(1): 27-48. Lutz, Byron (2008). The Connection between house price appreciation and property tax revenues. National Tax Journal, 61(3): 555-572. 18

Lutz, Byron, Raven Molloy, and Hui Shan. 2011. The housing crisis and state and local government tax revenue: Five channels. Regional Science and Urban Economics, 41(4), 306-319. Mikesell, John L. and Liu, Cheoul. 2013. Property tax stability: a tax system model of base and revenue dynamics through the great recession and beyond. Public Finance and Management, 13(4): 310-334. Mikesell, John L. and Mullins, Daniel R. 2010. State and local revenue yield and stability in the Great Recession. State Tax Notes, 55: 267-274. Mikesell, John L. and Mullins, Daniel R. 2013. Structural/Institutional Determinants of Variations in Household Property Tax Burdens and Effective Tax Rates within and across Local Governments: the Effect of the Great Recession and Beyond. Conference presentation at the annual conference of the Association of Public Budgeting and Finance, Washington, D.C. 19