Printer Friendly

Detroit property tax delinquency: social contract in crisis.

1. INTRODUCTION

The nationwide real estate crisis and ensuing decline in the property tax base have affected local government finances throughout the United States. The housing market in Detroit, Michigan, was hit particularly hard. According to Hodge et al. (2013a), in 2010 the average "arms-length" selling price of a residential parcel with a structure was just $12,700. In 2013 the city's fiscal challenges came to a head when Governor Snyder appointed an emergency financial manager, who subsequently set in motion a filing for bankruptcy. Despite the near complete collapse of the real estate market within the city (1), the property tax is still an important revenue source and the quality of its administration can help or hinder economic and fiscal recovery. Currently, about half of Detroit properties are tax delinquent. The extraordinarily high rate of tax delinquency results from a confluence of factors: city and county officials have failed to fully enforce property tax compliance, particularly for low value, low tax properties; many citizens perceive the tax to be unfair, partly because of the over-assessment of property; and many property owners seem frustrated by the lack of public service provision, so that citizens may not pay taxes because local authorities have failed to provide basic public services such as public safety. (2) In many respects, the basic social contract between the city and its residents seems to be broken.

Detroit is not the only city to experience significant property tax delinquency. A recent report by the Pew Charitable Trusts (2013) highlights emerging challenges with property tax delinquency across the nation, identifying the cities of Flint, Cleveland, and Detroit as localities that are estimated to have lost more than 20% of potential tax revenue to tax delinquency in 2011. All other cities cited in the report are below 10% in lost tax revenues, with most of the 30 cities in the study below 5%. Detroit stands out for its high delinquency rate. Indeed, as discussed by Bahl and Martinez-Vazquez (2007), the collection rate in Detroit is comparable to cities in developing countries where the property tax is used as a local revenue source such as the Philippines (collection rate of 40-50%), Jamaica (collection rate of 40%) and Chile (collection rate of 26%).

Perhaps surprisingly, there is relatively little work on property tax delinquency, even though the related issue of noncompliance with taxes in general has attracted increased attention in recent years. Langsdorf (1973) offers an evaluation of property tax delinquency in the previously declining city of St. Louis, bemoaning a 9% property tax delinquency rate that he attributed to the high rate of property abandonment. He describes a cycle by which the value of property becomes so low that the income generating potential is insufficient for the owner and so the property is abandoned. In such cases, local authorities have no effective means by which to enforce tax compliance. (3) By U.S. historical standards, a 9% delinquency rate is high, but it pales in comparison to the current tax delinquency problem in Detroit. White (1986) also considers the role of property taxes in the decision to abandon property in New York City. Using neighborhood level data on abandonment in the late 1970s, she finds that the abandonment rate is very sensitive to property taxes, with estimates suggesting that a 1% increase in property taxes leads to a 2.1% increase in abandonment. More recently, Bradley (2013) considers tax delinquency in the much wealthier Michigan community of Ann Arbor during the 2006-2009 period, where the delinquency rate was about 8% which is high relative to historical standards. However, his focus was on assessing the role of "saliency" in tax delinquency. Bradley (2013) hypothesizes that new homeowners systematically underestimate the amount of taxes that they must pay and that this error contributes to the financial challenge of paying property taxes on time. Indeed, he finds a small but statistically significant effect of the difference between "anticipated tax payments" and actual taxes due on delinquency. Finally, Waldhart and Reschovsky (2012) explore the role that tax administration can play in tax delinquency. Using data from Wisconsin they show that increasing the number of property tax payment installments from two to three (and thus reducing the amount needed for any single given installment) reduces the delinquency rate by a third, but no additional improvements are observed by increasing the number of installments beyond three.

The Great Recession brought to light vulnerabilities across the globe; Detroit's challenges are a small part of a much larger context. Among industrialized countries, Greece could be called the "poster child" of tax evasion, where about a third of government revenue is lost to tax evasion. In writing about the Greek crisis, Surowiecki (2011) succinctly summarized the tax compliance challenge:

   It isn't just a matter of lax enforcement, though Greek citizens
   also have what social scientists call very low "tax morale." In
   most countries, tax-compliance rates are much higher than a
   calculation of risks would imply. We don't pay our taxes just
   because we're afraid of getting caught; we also feel a
   responsibility to contribute to the common good. But that sense of
   responsibility comes with conditions. We're generally what Swiss
   behavioral economist Benno Torgler calls "social taxpayers": we'll
   chip in as long as we have faith that our fellow-citizens are doing
   the same, and that our government is basically legitimate.
   Countries where people feel that they have some say in how
   government acts, and where there are high levels of trust, tend to
   have high rates of tax compliance.


As we show in our evaluation, Surowiecki's statement could have easily been written about Detroit. While Detroit is an outlier among U.S. cities, it offers important lessons for governments at the sub-national and national levels who must maintain the perception of legitimacy in order to foster high citizen tax morale and tax compliance.

To evaluate tax delinquency in Detroit, we develop a theoretical model of the individual decision to become delinquent on one's property tax payments.

We then estimate this model using detailed parcel-level data from the City of Detroit to examine both the factors that determine the likelihood that a property owner will be tax delinquent and, conditional upon delinquency, the factors that determine how much back taxes are owed. We find important linkages between a number of property characteristics and the likelihood and magnitude of tax delinquency. Properties that are of lower value, are non-homestead (non-owner occupied residential properties), are owned by a financial institution or by a Detroit resident, and are delinquent on water bills are more likely to be delinquent. Turning to the policy variables, our primary interest, estimates suggest that properties with a high statutory tax rate, a high assessed value relative to estimated market value, longer police response times, and for which the tax is less likely to be enforced are more likely to be tax delinquent. These findings can be used by city and county officials to improve compliance.

In the next section we provide an overview of economic and fiscal conditions in Detroit, including a description of property tax policies and guidelines in managing tax delinquent properties. We then develop a theoretical model of individual property tax delinquency, and we estimate this model in the following section. We conclude with a discussion of the implications of these findings for Detroit in the final section.

2. ECONOMIC AND FISCAL CONDITIONS IN DETROIT

Since the middle of the 20th century, the City of Detroit has endured one of the longest and most severe declines of any American city. The causes of this decline range from broad macro-economic trends (globalization, de-industrialization) to idiosyncratic, local issues (corruption, mismanagement). A number of authors have chronicled Detroit's decline (Binelli, 2012; Ryan, 2012; Bomey and Gallagher, 2013). Here we provide only a brief outline of the major factors.

For some time now, Detroit has been shrinking in every dimension except in its land area. Population, employment, resident jobs, businesses and housing units have all declined, often by staggering amounts (Figure 1). During the 1970s, Detroit's population fell by more than 310,000; the total population lost between 1970 and 2010 exceeds 888,000. The number of business establishments in Detroit fell by two-thirds, accompanying a drop of 53% in the number of jobs in Detroit. The number of Detroit residents with jobs fell by more than half.

Trends since 2000 have continued to be negative with population losses accelerating in the last decade, with a net loss of more than 237,000 persons between 2000 and 2010. Estimates from the American Community Survey indicate continued decline. Despite an active program of demolishing dilapidated buildings, the residential vacancy rate increased from 10.4% in 2000 to 28.3% in 2013 (US Department of Housing and Urban Development, 2013).

The numbers of jobs and real earnings in the city have fallen dramatically. Despite attracting several thousand new white-collar jobs to the downtown area, including those jobs generated by Compuware and Quicken Loans, Detroit has experienced a net loss of 74,000 jobs since 2000, including a decline of 10,000 jobs downtown (County Business Patterns, 2000-12). Average real (4) earnings for Detroit residents have fallen from $27,900 in 1999 to $21,400 in 2010 (U.S. Census of Population, 2000; American Community Survey, 2012). The value of Detroit payrolls has declined by 35% in current dollars, from $14.1 billion to $9.2 billion during the same period.

These employment and population declines, combined with the real estate crisis, have taken an enormous toll on the city's property tax base. State law requires that local jurisdictions track state equalized value (SEV) and taxable value (TV) for each parcel, where SEV is equal to one-half of market value. Taxable value is equal to SEV at the time of purchase, but since 1994 its annual growth is constrained to the inflation rate or 5%, whichever is less, for as long as one owns a property. TV is reset to SEV upon sale of the property. Thus, for many long-time property owners, there is a gap between SEV and TV, and a corresponding jurisdiction-wide gap between SEV and TV. The aggregate value of real property as measured by SEV in Detroit peaked at $12.5 billion in 2007, before falling to $7.8 billion in 2012, a decline of more than 37% in just five years, though the work of Hodge et al. (2013a) suggests that SEV has not fallen enough. The aggregate taxable value of real property was as high as $8.6 billion in 2008, but had dropped to $6.9 billion by 2012.

For purposes of our analysis, it is useful to first review property tax law and administration in Detroit. The property tax is one of the primary sources of revenue for Michigan local governments (cities, villages, and townships). (5) Even though the City of Detroit's annual property tax revenues are considerably larger than any other municipalities, Detroit derives just under one-fifth of its general fund revenues from the property tax. The City of Detroit also raises revenue through a local income tax, a utility users' tax, and a wagering tax on revenue from the three casinos in the city. For individual properties, city operating and debt service taxes account for about 44% of the total tax bill for owner occupied housing and about 35% for all other types of property. Local school district operating and debt service millages represent about 20% of the total bill for homeowners and 37% for other properties. The balance is made up of the State Education Tax on non-owner occupied residential property, county, and special district taxes.

Detroit has primary responsibility for administration of the property tax, including assessing and collecting the taxes for all taxing jurisdictions that overlap the city. When property taxes are delinquent, they are turned over to the County Treasurer for collection. In recent years, county government (Wayne County in the case of Detroit) no longer provides full reimbursement to Detroit for uncollected taxes. (6) If property taxes remain delinquent for three years, the County may foreclose and take the property's title. Because of the large volume of tax delinquent properties, Wayne County has insufficient personnel to pursue the foreclosure process for every property, which includes serving notices, inspecting the properties and filings with the County Registrar of Deeds.

Since the passage of Proposal A in 1994, allowable increases in taxable values have been limited to the rate of inflation, as measured by the Bureau of Labor Statistics Consumer Price Index for metropolitan Detroit. From 1994 through 2007, average home prices increased more rapidly than the rate of inflation, creating a gap between home values (as measured by SEV) and taxable values (TV). In 2003, Detroit's aggregate TV was 35% less than aggregate SEV. Beginning in 2008, SEV increases began to decline. A year later, TV in

Detroit also began to decline. (7) In 2012 (the most recent data available), aggregate residential property values in Detroit, as measured by SEV had declined by 46.5% from the peak, whereas residential TV fell by 24.7%. Despite this substantial decline, the TV for many properties remained higher than the apparent market value. Pending the results of a citywide reassessment, the newly elected Mayor of Detroit recently ordered an across the board reduction of five to 20% in taxable values (Guillen and Laitner, 2014).

Although Michigan falls well above the median in various measures of property tax burden, the property tax rate in the City of Detroit is more than twice the statewide average. Homestead property tax rates, at 3.257% of value, are the highest in the nation, about 2.4 times the average of the 50 largest cities in the United States (Minnesota Taxpayers Association, 2012). (8) This estimated tax rate includes the tax savings available to owner occupants of residential properties, but it does not include other forms of property tax relief that are available to specified households.

Property tax advantages in Detroit are available in several forms. Michigan provides "circuit breaker" protection for low-income households through a state income tax credit, and the state allows local governments to offer property tax reductions based on hardship. Also, as noted above, since 1994 a cap on increases in taxable value has limited growth in tax bills to the rate of inflation or 5%, whichever is less (Feldman, Courant, and Drake, 2003). Because TV is reset to SEV when a property is sold, a homeowner's effective tax rate becomes a function of the length of tenure, growth property value, the inflation rate, and the statutory tax rate. Also, owners of homestead (that is, owner-occupied) property are exempt from paying local school operating taxes. Finally, Michigan municipalities may create Neighborhood Enterprise Zones, in which the property tax on qualified residential investments may be reduced by up to half (Hodge et al., 2013b).

The administration of the property tax in Detroit is problematic. On average, properties are over-assessed (Hodge et al., 2013a) by a factor of five or more. A lack of personnel at both the city and the county, as well as outdated data processing systems, are responsible for processing errors, including sending bills to the wrong address and failing to return properties to the tax rolls when they are sold by the city (Macdonald, 2013c).

The property tax foreclosure process in Michigan is also complex and lengthy. Delinquent property taxes are first subject to late charges. After one year of delinquency, the property is forfeited to the county and subject to even higher monthly interest charges. During this two-year period, the owner may redeem the property by paying all outstanding taxes and fees. When property taxes have been delinquent for more than two years, the County Treasurer (Wayne County for Detroit properties) initiates foreclosure proceedings. A show cause hearing is held in the Circuit Court; the judicial foreclosure process extinguishes the rights of former owners of interest, including the former property owner, banks, and mortgage companies. The property remains subject to liens filed by government agencies in relation to the Environmental Protection Act, as well as any recorded or visible easements (Michigan Department of Treasury, 2010). The County Treasurer then holds a public auction of these foreclosed properties. At this auction, properties may be purchased for a minimum bid that equals the unpaid property taxes, plus interest and penalties, with the proceeds distributed proportionately to the taxing jurisdictions. If the property is not sold at the first auction, a second sale is held, with the minimum bid reduced to $500. Property not sold at either auction may be transferred to a public body (city or state), to a state or local land bank, or held for a subsequent auction.

There are a number of factors that can disrupt this process. Generally, if the property is involved in a lawsuit or a bankruptcy, the county will delay the foreclosure action until these issues are resolved. Because of the large number of Detroit properties that have entered the property tax foreclosure process, the Wayne County Treasurer and the Circuit Court lack the personnel to pursue every foreclosure action. Wayne County offered 13,500 properties in the 2011 tax auction; the number rose to 21,350 in 2012 (Anglebrandt, 2013). (9) Because of the growing backlog of tax delinquent properties (currently about 200,000), the county does not foreclose on property owners who owe less than $1,600 in taxes and penalties (MacDonald, 2013b).

These policies, combined with a collapsing real estate market and poor economic conditions, have resulted in increasing numbers of properties for which property taxes are not being paid. For owners of low-valued, low-tax properties, payment of property taxes is effectively optional since a foreclosure action is unlikely. Even owners of higher valued properties with limited equity may elect to ignore their tax bills, with the expectation that they will be able to repurchase the property for $500 when it is auctioned. In addition, County records indicate that 80% of the properties sold at auction over the past two years are once again delinquent on taxes (Detroit News, 2013). It seems likely that purchasers of many of these properties are absentee owners who plan to reduce their operating expenses and increase their net rental income by never paying taxes. Tax delinquency for Detroit property owners possess the elements of a standard tax evasion decision. In the next section, we apply relevant aspects of the tax evasion literature to the specific property tax delinquency decision in order to develop a model for estimating the determinants of property tax delinquency.

3. THEORETICAL FRAMEWORK AND HYPOTHESES (10)

Consider a property owner who must choose whether to pay the legally due property tax obligations or become delinquent. This "decision" is a complicated one, with multiple and intertemporal, aspects. Recall from our earlier discussion of property tax delinquency in Detroit that the process begins with the owner deciding in the current period whether to pay or not, and the individual faces an uncertain prospect that the city will start the forfeiture/foreclosure process if he chooses to become delinquent. However, even if the owner chooses delinquency in the current period and the forfeiture/foreclosure process begins now, the owner may decide to pay taxes plus fines in a future period. As a result, for an owner to be delinquent, he must choose delinquency both in the current period and in the future period. Our theoretical model incorporates both of these decisions.

We begin with the current period. Assume that a property owner with income I0 owns a parcel with an assessed taxable value of B and a true (market) value of H. The owner is legally obligated to pay property taxes at rate t on the assessed value. The owner derives net imputed rental income at rate r on the property's market value, and he also receives public services that are valued in the amount G. The owner must choose whether to pay property taxes tB in the current period, or risk facing forfeiture and foreclosure. However, there is uncertainty about whether the owner who decides not to pay property taxes will be subject to the forfeiture/foreclosure process, as represented by the probability p. (11) This probability is assumed to be a positive function of the taxable value, or p(B), withp'(B)>0.

If the owner pays all property taxes in the current period, then his certain income is [[I.sub.0]-tB+rH+G], and the certain utility is simply U(I) = U([I.sub.0]-tB+rH+G). If the owner decides to become delinquent in the current period, then his income depends upon whether the forfeiture/foreclosure process occurs. With forfeiture/foreclosure, the individual loses the house and the associated benefits from the house and the public services; the individual is also assumed to bear some "guilt" from his failure to pay the legally due taxes. Denoting the monetary value of this guilt as y, the individual's income becomes [[I.sub.0]-[gamma]].(12) If forfeiture/foreclosure does not occur, then the individual's income is [[I.sub.0] + rH + G]. The expected utility EU(I) of this gamble is EU(I)=p(B)U([I.sub.0] - [gamma])+(1-p(B))U([I.sub.0] + rH + G). The individual compares the certain utility from payment U(I) with the expected utility from delinquency EU(I), and chooses to pay or to become delinquent, depending upon which is greater. (13)

It is straightforward to demonstrate that delinquency in the current period is more likely if: the probability of forfeiture/foreclosure is lower, the tax rate is higher, public services are lower, imputed rental income is lower, and guilt from delinquency is lower.

However, even if the owner chooses to become delinquent in the current period, and even if the forfeiture/foreclosure process starts in the current period, the individual has the option of paying (or not paying) taxes plus fines in the future. Denote the additional fine rate as f and the additional fines as ftB. Then the owner's income if he pays the taxes plus fines is [[I.sub.0] - ftB + rH + G], and his utility in this case is U([I.sub.0] - ftB + rH + G); his income if he chooses to become delinquent (again) is [[I.sub.0] - [gamma]], with associated utility of U([I.sub.0] - [gamma]). The owner compares these two certain utilities, and chooses to become delinquent in the future if the utility from delinquency is greater than the utility from payment. Delinquency in the future period is more likely if: the additional fines are greater, the tax rate is higher, public services are lower, imputed rental income is lower, and guilt from delinquency is lower. These results largely parallel the analysis for the current period. Note that enforcement factors do not play a role in the future delinquency decision.

Of course, determining delinquency as the optimal behavior requires that the owner choose delinquency both in the current period and in the future period. Solving for this optimal behavior requires solving the entire intertemporal decision process for the owner. This problem can be solved via backward induction, starting with the future delinquency decision and then (conditional upon optimal future behavior) solving the current delinquency decision. It can be shown that the basic comparative statics results discussed above apply directly to the entire intertemporal decision process.

Overall, then, this analysis suggests several hypotheses about the factors that affect property tax delinquency, factors that relate to the roles of enforcement, public services and commitment to the community, tax rates, assessed versus market values, and notions of fairness (including the impact of guilt) in the delinquency decision:

Hl-Property owners for whom the tax is not enforced are more likely to be delinquent.

H2-Property owners with lower taxable value are more likely to be delinquent due to lower enforcement for lower taxable values.

H3-Property owners who receive lower public services are more likely to be delinquent.

H4-Non-homestead property owners are more likely to be delinquent.

H5-Property owners facing higher statutory tax rates are more likely to be delinquent.

H6-Property owners with higher assessment ratios are more likely to be delinquent.

In the next section we present our data and methods for testing these hypotheses.

4. DATA AND EMPIRICAL METHODS

The City of Detroit's Assessment Division provided parcel-level data for this research. The raw data for fiscal year 2010 include information for 444,183 real and personal property parcels, of which we focus on residential properties. In total there are 224,262 residential properties, both owner-occupied (homestead) and non-owner occupied. Unfortunately, there was missing information for thousands of properties. In most of these cases, state equalized value (SEV), taxable value (TV), or last sale price were recorded as zero. Omitting these properties results in 157,985 properties that we include in our analysis, of which 77,716 are residential owner-occupied and 80,269 are residential non-owner occupied properties. (14) Summary statistics for all of the variables we consider are provided in Table 1, and definitions for all variables are provided in Appendix Table 1.

There are some significant differences between homestead and non-homestead properties. For example, the delinquency rate is substantially higher for non-homestead properties, and the average statutory tax rate for nonhomestead properties is 19 millage points higher than for homestead properties. This difference is largely due to the homestead exemption, but other tax abatement programs also create variability in the statutory rate across both homestead and non-homestead properties. The average length of ownership for homestead properties is 12 years, substantially more than the 8.5 average for non-homestead properties. With these exceptions, the summary statistics are similar across the two groups (e.g., crime response time, living area, lot size, and age). As shown in Figure 2, there is considerable variability in delinquency rates across the city.

We use these parcel-level data to estimate two models of delinquency. First, we examine the factors that affect the probability of being tax delinquent, Delinquent. Second, we also estimate the amount by which an owner is delinquent, Delinquent Amount, conditional upon delinquency. These two issues are related in the sense that a property owner first chooses whether or not to pay taxes, and, if not, then the owner chooses by how much. This joint process is estimated simultaneously in order to address the potential bias introduced by the property owners' self-selection into delinquency. We therefore use the procedure proposed by Heckman (1979). (15) In the first step we estimate the delinquency (yes/no) selection equation, represented by:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

where Delinquenti indicates whether the property owner is delinquent (yes = 1, no = 0), Pt is a vector of property characteristics (Homestead Exemption, Statutory Tax Rate, Taxable Value, Years Owned, Assessment Ratio (16), Detroit Owner (17), Size, and Delinquent on Water, and [X.sub.i] is a vector of variables that are excluded from the second-stage outcome equation, as presented below. The variables in X are our instruments; these instruments are a variable measuring precinct level police response times (Police Response Time) and indicators for whether the tax is enforced (No Tax Enforcement), the property is owned by a Detroit resident (Detroit Resident), and bank owned property (Bank Owned Property). (18)

The inclusion of the variables follows directly from theory. However, the variable Delinquent on Water requires further explanation. When a property owner is delinquent on his or her water bill, the back-payment is rolled over into the delinquent property tax bill. Since water is an essential service, we view payment of the water bill as a priority, and we include Delinquent on Water as a proxy for economic hardship of the property owner.

The No Tax Enforcement indicator variable also requires further description. In Wayne County, officials are overloaded and thus do not begin tax foreclosure proceedings until back taxes and penalties exceed a fixed amount, generally $1,500 to $2,000. (19) Further, state law requires that an owner cannot be removed from a tax delinquent property for two years. Thus, an astute property owner of a low tax property knows that he/she has at least two years before tax foreclosure proceedings take place and tax foreclosure forces him/her off the premises. In the estimates reported, the No Tax Enforcement indicator variable equals one for properties with less than a $1,000 tax payment (the owner could continue to live in the home for about two years before tax foreclosure), and zero otherwise. (20)

We are also interested in understanding the factors that determine the amount which a delinquent property owner owes, given that he/she is delinquent. The Heckman (1979) second stage outcome equation is represented by:

Delinquency [Amount.sub.i] = [P.sub.i][alpha] + [[epsilon].sub.i]

where Delinquent Amount is equal to the unpaid tax bill over the 2010 spring and winter 2010 billing cycles. (21) We estimate the selection and outcome equations jointly by maximum likelihood.

5. ESTIMATION RESULTS

Determinants of the Probability of Delinquency

Consider first the estimates from the first stage selection equation as reported in Table 2. In column 1 we present probit estimates of the delinquency decision using all residential properties. Because the factors that determine delinquency may differ across homestead and non-homestead properties, we also present separate estimates for these two sub-groups in columns 2 and 3. Note that there are some differences in specification across columns 1, 2, and 3. In column 1, we include a homestead property indicator variable, but this variable is excluded in the column 2 and 3 sub-group estimates because the sub-groups are defined by this variable. Also, in columns 1 and 3 indicator variables for whether the property owner lives in Detroit or is bank owned are included, but these are excluded in column 2 because the property must be the primary residence of the property owner in order to receive the homestead ex emption. (22)

In column 1 all variables are statistically significant determinants of delinquency. Consider first the control variables. For ease of interpretation, we modify the coefficients from the first stage probit regression so that each coefficient reflects the marginal effect evaluated at the mean. (23) Owners of older homes are more likely to be delinquent. Non-homestead property owners are 20% more likely to be delinquent than their homestead property owner counterparts. Property owners who purchased their property more recently are also more likely to be delinquent than long-time owners. Long-time property owners and homestead owners have lower rates of delinquency, perhaps because these property owners are more likely to have social connections as well as more assets at stake. In addition, we find that Detroit owners, bank owned properties, and owners who are delinquent on the water bill are more likely to be delinquent. These findings are generally consistent with our expectations.

Consider now the factors that are to some degree within the control of policymakers. Recall that our earlier discussion highlighted the fact that local officials are not enforcing tax payments for low valued properties. Our results show that many owners of such properties recognize this and choose not to pay taxes. The coefficient on taxable value is negative and statistically significant, so that lower valued properties are more likely to be delinquent. Also, the No Tax Enforcement variable is statistically significant; that is, controlling for taxable value, properties with less than $1,000 in annual tax payments have a 5.8 percentage point higher probability of being delinquent than their higher tax payment counterparts.

This latter finding is quite robust. In Figure 3, we present alternative coefficients estimates for No Tax Enforcement cut-offs of $600, $800, $1,000, $1,200 and $1,400. Figure 3 shows that the largest coefficient on the No Tax Enforcement indicator variable is when the enforcement cut-off is at the $1,000 tax payment. When the cut-off diverges from $1,000, the coefficient estimate declines and becomes statistically insignificant. That is, when the cutoff diverges from $1,000 the differential in the probability of delinquency between properties above and below the cut-off diminishes. This result provides evidence that failure to enforce tax compliance of the lower valued properties exacerbates the delinquency issue. Note, however, that the size of the coefficient is relatively small; that is, full enforcement would not achieve delinquency rates in Detroit comparable to those of other large cities. Our estimates show that full enforcement would reduce delinquency among property owners by about six percentage points, but the potential improvement would be insufficient to fully eliminate delinquency.

Our estimates also show that delinquency is higher in areas where police response times are longer. According to this estimate, if the City of Detroit were to reduce the police response time from 47 minute to 12 minutes (roughly the national average), tax delinquency would fall by about three percentage points. This finding provides evidence that an owner who receives lower quality public services is less likely to pay taxes. Note that we are not able to include simultaneously both neighborhood indicator variables and the police response time variable because the response times are reported by precinct, which overlaps closely with neighborhoods. One might think inclusion of neighborhood indicator variables would be important to capture neighborhood-specific unobservable factors. However, in other, unreported estimates, we found that inclusion of neighborhood effects does not significantly alter the coefficients on our other explanatory variables. (24) It is important to acknowledge that the crime response variables may be correlated with neighborhood characteristics, and coefficient estimate on the crime response time may capture other unobservable factors.

Turning to tax fairness, the estimates show that properties with higher assessment ratios are more likely to be delinquent. If the average assessment ratio was adjusted downward from five to one (the target set by state law), the citywide delinquency rate would fall by about two percentage points. According to Hodge et al. (2013a), the city has systematically over assessed property by an average factor of nearly five. (25) Currently, the State of Michigan Tax Commission is conducting sample reassessments in Detroit to determine whether a full reassessment of property is warranted. 26 A reassessment would result in a significant decline in property tax revenues, but it would also likely improve tax compliance, a finding that is consistent with that of White (1986).

We also find that properties with high statutory tax rates have higher delinquency rates. A reduction in the tax rate of 20 millage points (a reduction of this amount would bring tax rates in line with the regional average) would reduce delinquency by about 4.5 percentage points.

Our estimates suggest that change in any one policy results in a relatively small impact on the overall delinquency rate, but a combination of policy changes (improved tax enforcement, reduced crime response times, improved assessment practices, lower statutory tax rates) could reduce delinquency by as much as 15 percentage points. Given the 48% delinquency rate, a 15 percentage point reduction does not fully resolve the problem. However, a combination of policy changes could result in meaningful reductions in property tax delinquency.

Turning to a comparison of the homestead and non-homestead property results in columns 2 and 3, we see some differences across the two groups in the determinants of delinquency. In particular, the size of a home is a positive determinant of delinquency among homestead property owners, but it is a negative determinant for non-homestead owners. The other variables have the same signs in both sets of estimates, but note that there are some differences in the magnitudes of the coefficients.

Determinants of the Magnitude of Delinquency

It is also useful to consider the factors that determine the amount by which the property owner is delinquent, given that the owner has made the decision not to pay taxes. These estimates are presented in Table 3. The results in Table 3 largely mirror those presented in Table 2. Column 1 contains the estimates that use the full set of parcels; column 2 presents the estimates for the homestead properties; and the non-homestead estimates are found in column 3.

Consider column 1, which contain the estimates using all residential parcels. With the exception of the house size and age variables, all of the coefficients are statistically significant. Homestead properties have smaller delinquency amounts, whereas properties with higher taxable values, higher statutory tax rates, and properties owned for longer periods have greater delinquency amounts. Finally, property owners who are delinquent on their water bills have higher delinquency amounts.

There is general consistency in estimates across homestead and nonhomestead properties, as presented in columns 2 and 3, respectively, with two notable differences. First, in the non-homestead sample smaller and older homes have lower delinquency amounts, but among homestead properties, smaller and older homes have larger delinquency amounts. Second, the coefficient on water bill delinquency is positive and significant in both estimates, but the coefficient in the homestead sample is almost twice as large, and water bill delinquency contributes to more than half of the total delinquency amount for homestead properties. This result suggests that many delinquent homestead property owners are experiencing financial challenges, and are using delinquency as a source of liquidity. The other results are generally similar.

Note finally that the coefficient on the Inverse Mill's Ratio is statistically significant in two of the three columns, indicating that selection bias is an issue in these estimates. The Heckman procedure is appropriate in this context.

6. CONCLUSIONS

Our estimation results are consistent with the important impact of various policy factors on property tax delinquency, such as enforcement, homestead/non-homestead status, taxable value, statutory tax rate, the assessment ratio, and police response times. It must be acknowledged that our estimation results do not necessarily indicate causality between these policy variables and delinquency. We are also unable to include some potentially important property owner-specific factors in our estimation. Even so, our results clearly indicate an important role for policy.

While the exceptionally high rate of tax delinquency is the result of broad regional/global economic forces, it is also a reflection of citizens' confidence in the city government. Clearly, economic decline, falling property values, and reduced ability of owners to pay taxes have resulted in higher rates of tax delinquency. However, in the context of weak tax enforcement, particularly for low valued properties, individuals may ask themselves why they should pay taxes when services are limited and high statutory tax rates and over-assessment of property for tax purposes result in very high effective tax rates. One interpretation of this finding is that the city is failing in its half of the social contract. Further, 80% of the properties that go into tax foreclosure and are resold at auction are again tax delinquent within two years. Properties that are not resold remain publicly held by city government or a land bank. As a result, the city has accumulated thousands of parcels through abandonment and tax foreclosure. About 25% of the city's land area is owned by the city, county, land bank, or state, and this percentage is growing. Further, the prospects for returning these properties to tax paying private owners are low in the foreseeable future. So what should the City of Detroit do now? Our evaluation suggests that city officials can reduce tax delinquency by improving public safety services, targeting property owners that are more likely to be tax delinquent (non-homestead property owners), and improving basic public services like police protection. Stronger enforcement of delinquency provisions could also generate some additional revenues. However, even full enforcement will not totally resolve delinquency problems. Property values need to recover to provide increased incentive for owners of the lowest valued properties to pay taxes and to improve the fiscal viability of the tax foreclosure process.

What local policy options are there for increasing land values, thereby reducing tax delinquency? One option is to reduce assessments, which are currently much higher than what the law mandates. All else equal, reducing assessments lowers the cost of holding property and thus increases property values. Another possible policy option related to the large number of publicly held parcels is to remove some of this land permanently from the market, which could lift land values.

Understanding the underlying causes of these high rates of property tax delinquency can be helpful in designing strategies to reduce them. The available data are not, however, sufficient to explain why property tax delinquency rates in Detroit and why they are so much higher than in other cities. We can only speculate as to the likely reasons.

One important factor is surely necessity. The high unemployment in Detroit (approaching Depression era levels) and consequent low household incomes, it should not be surprising that property taxes go unpaid. Another consideration in delinquency might be termed "strategic choice." Property owners in Detroit may elect to forego property tax payments to improve their short-term cash flow. Many Detroiters have difficulty obtaining credit and when they can, the cost is often high. The 12 or 18% penalty applied to unpaid taxes is considerably less than what would be charged by payday lenders or even credit card companies. Delinquencies may also be attributed in part to the City's administrative systems. A lack of personnel and up to date information technology clearly have limited the effectiveness of the property tax collection process in Detroit.

We also believe that a break down in the implicit Social Compact between the city government and its residents may be partially responsible for the high delinquency rates. In most cities, residents are willing to pay property taxes because they value the services that the local government provides and there is an understanding that all property owners make their fair contribution to the costs. When this breaks down, either because adequate services are not provided or because a large number of property owners are not making payments (yet continuing to receive services), individual property owners who are able to pay their taxes may be tempted to withhold them. (27) These circumstances are likely to be exacerbated when the tax burden is considered unreasonably high.

It seems likely that these and perhaps other factors contribute in some measure to Detroit's property tax delinquency problems. There is clearly no "magic bullet" that will quickly and completely resolve all of these issues. Rather, a commitment to a long term, comprehensive strategy will be required.

Detroit's challenges have accumulated over decades. Leaders mistook long-term structural shifts in the regional economy for short-term challenges, putting off the required long-term adjustments in such areas as pension contributions and deficits by thinking that these challenges were short-term in nature. However, these leaders were mistaken, and now new leaders must deal with the consequences. The social contract between property owners and the City of Detroit needs restoration; that can be accomplished over time with a dedicated and ongoing effort to improve a broad range of policies as shown in our evaluation. Improvement in policies, in turn, will help to stabilize the economic and population base and bring badly needed resources and jobs to the city. There are, however, no easy or quick fixes.

APPENDIX

Table 1. Variable Names and Definitions

Variable Name              Variable Definition

Delinquent                 Indicator variable equal to 1 if the owner
                           is delinquent on at least two tax bills,
                           and zero otherwise

Delinquent Amount          The total amount by which a property owner
                           is delinquent over two billing cycles
                           (spring and winter)

No Tax Enforcement         Indicator variable equal to 1 if the
                           annual tax payment on the property is less
                           than $1,000, and 0 otherwise

Police Response Time       The police response time in minutes for
                           each of the 13 precincts in Detroit

Size                       Size of residential structure in 1,000s of
                           square feet

Age                        Age of residential structure in ten-year
                           intervals

Homestead Property         Indicator variable equal to 1 if the
                           property owner receives the principal
                           residence exemption, and 0 otherwise (also
                           known and the homestead exemption)

Statutory Tax Rate         Statutory Tax Rate, which is applied to
                           the taxable value of a property

Taxable Value              Taxable value, which grows at an annual
                           rate of 5% or the rate of inflation,
                           whichever is less, for as long as the
                           owner retains ownership

Years Owned                Number of years owned by the current
                           property owner

Assessment Ratio           (2*state equalized value)/(estimated sale
                           price). To generate the estimated sale
                           price, we use a hedonic price regression
                           based on observations on property  sales
                           during the 2008-2010 period. The
                           coefficient estimates generated from this
                           regression are used to generate predicted
                           prices. Once we generate predicted values
                           for the prices, we calculate the
                           assessment ratio.

Detroit Owner              Indicator variable equal to 1 if the owner
                           of the property resides within the City of
                           Detroit, and 0 otherwise

Bank Owned Property        Indicator variable equal to 1 if the
                           property has owned by a bank or other
                           financial institution

Delinquent on Water Bill   Indicator variable equal to one if the
                           property owner is delinquent on the water
                           bill, and zero otherwise

Notes: All data are provided by the City of Detroit Assessment
Division and the City of Detroit.


ACKNOWLEDGEMENTS

We thank the Lincoln Institute of Land Policy for financial support. We also thank Fred Morgan of the City of Detroit Assessment Division for providing detailed parcel-level data and Councilman Kenneth Cockrel for inviting us to evaluate Detroit's property tax base. We are also grateful to several anonymous referees for helpful comments.

REFERENCES

Achen, C. (1986) The Statistical Analysis of Quasi-Experiments. Berkeley, CA: University of California Press.

Allingham, M. G. and A. Sandmo. (1972) Income Tax Evasion: A Theoretical Analysis. Journal of Public Economics 1 (3-4), 323-338.

Alm, J. (2012) Measuring, Explaining, and Controlling Tax Evasion: Lessons from Theory, Field Studies, and Experiments. International Tax and Public Finance 19 (1), 54-77.

Alm, J., G. H. McClelland, and W. D. Schulze. (1999) Changing the Social Norm of Tax Compliance by Voting. Kyklos 52 (2), 141-171.

Alm, J. and B. Torgler. (2011) Do Ethics Matter? Tax Compliance and Morality. Journal of Business Ethics 101 (4), 635-651.

Angelbrandt, G. (2013) Deputy Treasurer: Wayne County "Worst off in Country for Tax Foreclosures. Crain's Detroit Business. August 29.

Bahl, R. and J. Martinez-Vazquez. (2007) The Property Tax in Developing Countries: Current Practice and Prospects. Lincoln Institute of Land Policy Working Paper. Cambridge, MA.

Binelli, M. (2012) Detroit City Is The Place To Be. New York, NY: Metropolitan Books.

Bomey, N. and Gallagher, J. (2013) How Detroit Went Broke. Detroit Free Press. September 14. http://www.freep.com/arti cle/20130915/NEWS01/ 130801004/Detroit-Bankruptcy-history- 1950-debt-pension-revenue

Bradley, Sebastien. (2013) Property Tax Salience and Payment Delinquency. Drexel University Working Paper. Philadelphia, PA. City of Detroit. 2013. Comprehensive Annual Financial Report. http://www.detroitmi.gov/Portals/0/docs/finance/CAFR/Final%202012%20De troit%20Financial%20Statements.pdf

Detroit News. 2013. Most Properties in Wayne Foreclosure Auctions Repeat Cycle of Unpaid Taxes. http://www.detroitnews. com/arti cle/ 20130923/METRO01/309160111/Most-properties-Wayne-foreclosure-auct ions -repeat-cycle-unpaid-taxes

Feldman, N.F., P.N. Courant, P.N. and D.C. Drake. (2003.) The Property Tax in Michigan. In Michigan at the Millennium, edited by C. Ballard, P. N. Courant, D. C. Drake, R. C. Fisher, and E. R. Gerber. East Lansing MI: Michigan State University Press.

Guillen, J. and B. Laitner. (2014) Detroit to Slash Property Tax Bills by 5% to 20%. Detroit Free Press. January 28, p. A1.

Heckman, J. 1979. Sample Selection Bias as a Specification Error. Econometrica 47 (1), 153-161.

Hodge, T., M. Skidmore, G. Sands, and D. McMillen, (2013a) Assessment Inequity in a Declining Housing Market: The Case of Detroit. Michigan State University Working Paper. East Lansing, MI. http://199.169.205.158/ economic/nesg/papers/hodge-skidmore-sands-mcmillen.pdf

Hodge, T., M. Skidmore, G. Sands, and D. McMillan. (2013b) Tax Base Erosion and Inequity from Michigan's Assessment Growth Limit: The Case of Detroit. CESifo Working Paper No. 4098. http://www.econstor.eu/ handle/10419/69527

Kirchler, E. (2007) The Economic Psychology of Tax Behavior. Cambridge, UK: Cambridge University Press.

Kotlikoff, L. (2013) Assessing Fiscal Sustainability. Mercatus Center Report, George Mason University.

Landsdorf, K. (1973) Urban Decay, Property Tax Delinquency: A Solution in St. Louis. The Urban Lawyer 5, 729-748.

MacDonald, C. (2013a) Half of Detroit Property Owners Don't Pay Taxes. The Detroit News, February 12.

Macdonald, C. (2013b) Wayne County Overwhelmed by Foreclosures. The Detroit News, February 21, page 13A.

Macdonald, C. (2013c) Property Values, Taxes often in Error. The Detroit News. February 22, page 1A.

Michigan Department of Treasury. 2010. Real Property Tax Forfeiture and Foreclosure. http://www.michigan.gov/taxes/0,4676,7-238-43535 55601,00 .html

Minnesota Taxpayers Association. (2011) 50-State Property Tax Comparison Study. Cambridge MA: Lincoln Institute of Land Policy.

Muller, J.M. (2013).Bus Rapid Transit: The Answer to Transportation Problems in Megacities?: The example of TransMillenio (Bogata, Columbia in Megacities: Our Global Urban Future, edited by F. Kraas, S. Aggarwal, M. Coy and G. Mertins. New York: Springer Science+Business.

Pew Charitable Trusts. (2013) Delinquent Property Tax in Philadelphia: Stark Challenges and Realistic Goals.

Ryan, B. (2012) Design after Decline. Philadelphia PA: University of Pennsylvania Press.

Sandmo, A. (2012) An Evasive Topic: Theorizing About the Hidden Economy. International Tax and Public Finance 19 (1), 5-24.

Slemrod, J. (2007) Cheating Ourselves: The Economics of Tax Evasion. The Journal of Economic Perspectives 21 (1), 25-48.

Sigelman, L. and Langche Z. (1999) Analyzing Censored and Sample-Selected Data with Tobit and Heckit Models. Political Analysis 8 (2), 167-82.

Surowiecki, J. (2011) Dogger Mania. The New Yorker, July.

Torgler, B. (2007) Tax Compliance and Tax Morale: A Theoretical and Empirical Analysis. Cheltenham, UK: Edward Elgar Publishing.

US Census Bureau. (2012) American Community Survey. Selected Characteristics: Detroit, Michigan, ACS One Year Estimates.

http://factfinder2.census.gov/faces/nav/jsf/pages/community_facts.xhtml#none

US Census Bureau. (2000) Census of Population and Housing. Detroit Michigan. Census Summary File 3. http://factfinder2.census.gov/faces /tableservices/jsf/pages/productview.xhtml?fpt=table

US Census Bureau. (2000-12) County Business Patterns. ZIP Code Business Statistics, Michigan. http://factfinder2.census. gov/faces/tableservices/ jsf/ pages/productview.xhtml?fpt=table

US Department of Housing and Urban Development. (2013) Metropolitan Area Quarterly Vacancy Report Detroit Michigan. http://www.huduser.org/portal/datasets/usps.html

Waldhart, P. and Reschovsky A. (2012) Property Tax Delinquency and the Number of Payment Installments. Public Finance and Management 12 (4), 316-330.

White, M. (1986) Property Taxes and Urban Housing Abandonment. Journal of Urban Economics 20, 312-330.

James Aim, Tulane University

Timothy R. Hodge, Allegheny College

Gary Sands, Wayne State University

Mark Skidmore, Michigan State University

(1.) Our discussion and analysis focuses on the City of Detroit and not the Detroit region.

(2.) See MacDonald (2013a, b, c) for media coverage on the issue.

(3.) In the words of Langsdorf (1973): The economics of property ownership become such that sufficient revenue cannot be generated, especially if the tenants of such property are at the bottom of the economic ladder. Tax payments are the area where the first delinquency occurs. Physical maintenance of property is consequently deferred and decay begins. With deferral of physical maintenance, tenants move out, and the cycle of urban decay is accelerated. When the property no longer generates sufficient income, the owner abandons the property. The cycle constantly repeats itself. Real estate tax collection, of course, depends a large measure upon the ability of the jurisdiction in which the property is located to impose effective sanctions upon the owner for non-payment. A city which has no workable enforcement mechanism can expect to have a geometrical rise in the stock of abandoned and tax delinquent property.

(4.) These values are denominated in 1999 dollars.

(5.) Other jurisdictions, including school districts, special districts (such as library, parks and public transit), county governments, and state government also rely on the property tax.

(6.) Previously, Wayne County reimbursed the City for uncollected taxes, minus a small administrative fee.

(7.) Michigan law allows TV to increase even when SEV decreases, as long as TV is less than SEV.

(8.) Because of Detroit's low property values, it ranks last among the 50 largest cities in terms of average tax payment.

(9.) In both years, less than half of these properties were actually sold.

(10.) Much of this literature is based in some way upon the original work by Allingham and Sandmo (1972). For recent discussions of the tax compliance literature, see Slemrod (2007), Sandmo (2012), and Alm (2012).

(11.) As indicated in our earlier discussion, delinquency starts with forfeiture proceedings, and then progresses to foreclosure proceedings, so that these are in fact separate processes with (likely) different probabilities. Without loss of generality, we combine these two processes into a single one.

(12.) See Alm and Torgler (2011) for a detailed discussion of the role of ethics and morality in the tax compliance decision.

(13.) The Michigan foreclosure process requires three decisions before the final step of judicial foreclosure and auction sale occurs. The model offers a simplified version of the tax foreclosure process while still capturing the essential features of the process.

(14.) Unfortunately, we do not have enough information about the omitted properties to evaluate potential sample selection bias. We note that roughly an equal number of homestead and non-homestead properties were omitted from the sample, offering some assurance that omitted parcels are similar to those in the sample, at least in this dimension.

(15.) See Achen (1986) and Sigelman and Zeng (1999) for theoretical and intuitive discussions on the Heckman procedure.

(16.) As described in the Appendix, the assessment ratio is defined as (2*state equalized value)/(estimated sale price). To generate the estimated sale price, we use a hedonic price regression based on observations for property sales during the 2008-2010 period. The coefficient estimates generated from this regression are used to generate predicted prices. Once we have predicted values for the prices, we calculate the assessment ratio.

(17.) All homestead property owners are "Detroit Owners". However, many landlords own property and have a Detroit address but do not live on the property of interest and thus do not receive a homestead exemption.

(18.) Detroit Resident is excluded from the evaluation of homestead properties because all homestead properties are in principle owned by a person who lives in Detroit. Similarly, Bank Owned Properties is excluded from the evaluation of homestead properties.

(19.) The critical point for enforcement varies from year to year. In 2010, the cut-off was $1,600.

(20.) In addition, see below for a discussion of the robustness of this variable.

(21.) In Michigan property taxes are due twice a year, one in the spring and one in the winter.

(22.) There are, however, a few hundred property owners who live outside the city who receive the homestead exemption. We have no explanation for why such properties were coded as such.

(23.) Specifically, this analysis was conducted using the Stata Heckman estimation command. Unfortunately, Stata does not offer a command to calculate the marginal effects for the first stage probit estimation within the Heckman estimation command. However, Stata has a command to calculate the marginal effects from the standard probit estimation procedure. We therefore generate and report the marginal effects of each covariate at mean values from the probit estimation of the first stage regression. These calculated marginal effects offer an appropriate interpretation for the indicator variables. However, the calculation for discrete (noninstantaneous) changes in continuous variables requires that we determine the change in probability of delinquency at different values of each continuous variable.

(24.) All alternative estimates are available upon request.

(25.) In 2010, the average sale price of arm's length transactions was $12,700, and yet the average assessed value of these same properties for tax purposes was $56,100. State law requires the ratio of assessed value to sale price to be one; the current average ratio for recently sold properties is nearly five.

(26.) The State Tax Commission is currently conducting an independent review of assessment practices, and in September 2013 city officials announced that over the next three to five years all properties within the city would be reassessed. For a detailed analysis of Detroit assessment practices, see Hodge, et al. (2013a).

(27.) An example of this can be seen in Bogota, Columbia during the 1980s and early 1990s, when that city was in the grip of the drug cartels and the local government barely functioned, the property tax delinquency rate was in excess of 70 percent. Subsequently, with the restoration of effective local government, delinquency has declined to more normal levels (Muller, 2013).

Table 1. Summary Statistics

Variable                   Full Sample       Homestead

                             Mea      SD      Mean     SD

Delinquent (yes = 1,       0.542     0.498   0.410    0.493
  no = 0)
Delinquent Amount          1,115     1,401   712.2    1,172
No Tax Enforcement         0.181     0.385   0.212    0.415
  (yes = 1, no = 0)
Police Response Time       47.41     8.053   48.01    7.718
  (minutes)
Homestead Property         0.492     0.500   --       --
  (yes = 1, no = 0)
Statutory Tax Rate         64.85     19.25   55.04    14.72
  (tax payment/SEV)
Taxable Value ($1,000s)    21.58     11.14   22.89    11.75
Years Owned                10.07     6.148   11.81    5.527
Assessment Ratio           10.41     3.892   11.37    3.682
Detroit Owner (yes = 1,    0.778     0.415   --       --
  no = 0)
Size (per 1,000 sq. ft.)   1.159     0.544   1.100    0.386
Age (Decades)              6.789     1.415   6.606    1.384
Bank Owned Property        0.045     0.206   --       --
  (yes = 1, no = 0)
Delinquent on Water        0.244     0.429   0.202    0.402
  (yes = 1, no = 0)
Number of Observations     157,985           77,716

Variable                   Non-homestead

                            Mean     SD

Delinquent (yes = 1,       0.670    0.470
  no = 0)
Delinquent Amount          1,506    1,491
No Tax Enforcement         0.141    0.348
  (yes = 1, no = 0)
Police Response Time       46.82    8.322
  (minutes)
Homestead Property         --       --
  (yes = 1, no = 0)
Statutory Tax Rate         74.37    18.32
  (tax payment/SEV)
Taxable Value ($1,000s)    20.26    10.28
Years Owned                8.382    6.247
Assessment Ratio           9.479    3.863
Detroit Owner (yes = 1,    0.591    0.492
  no = 0)
Size (per 1,000 sq. ft.)   1.215    0.656
Age (Decades)              6.966    1.421
Bank Owned Property        0.075    0.264
  (yes = 1, no = 0)
Delinquent on Water        0.284    0.451
  (yes = 1, no = 0)
Number of Observations     80,269

Table 2. Heckman First Stage Selection Estimation

                                                   Homestead
Independent Variable              All Properties   Properties

No Tax Enforcement                0.058 ***        0.0549 ***
(yes = 1, no = 0)                 (12.98)          (9.61)
Police Response Time (minutes)    0.001 ***        0.001 ***
                                  (6.25)           (5.66)
Homestead Property                -0.182 ***       --
(yes = 1, no = 0)                 (-50.86)
Statutory Tax Rate                0.003 ***        0.002 ***
                                  (19.98)          (7.97)
Taxable Value (per $1,000)        -0.006 ***       -0.006 ***
                                  (-25.02)         (-18.39)
Years Owned                       -0.015 ***       -0.013 ***
                                  (-53.61)         (-33.61)
Assessment Ratio                  0.007 ***        0.007 ***
                                  (10.31)          (7.81)
Detroit Owner (yes = 1, no = 0)   0.032 ***        --
                                  (8.31)
Size (per 1,000 square feet)      -0.003           0.040 ***
                                  (-0.99)          (6.69)
Age (Decades)                     0.004 ***        -0.002
                                  (3.370)          (-1.39)
Bank Owned Property               0.099 ***        --
(yes = 1, no = 0)                 (10.58)
Delinquent on Water               0.203 ***        0.245 ***
(yes = 1, no = 0)                 (65.23)          (57.40)
Number of Observations            157,985          77,716
Number of Censored Observations   72,559           46,022
Pseudo [R.sup.2]                  0.107            0.059

                                  Non-
                                  Homestead
Independent Variable              Properties

No Tax Enforcement                0.063 ***
(yes = 1, no = 0)                 (10.08)
Police Response Time (minutes)    0.001 ***
                                  (3.66)
Homestead Property                --
(yes = 1, no = 0)
Statutory Tax Rate                0.003 ***
                                  (19.26)
Taxable Value (per $1,000)        -0.006 ***
                                  (-19.50)
Years Owned                       -0.0169 ***
                                  (-42.84)
Assessment Ratio                  0.007 ***
                                  (7.38)
Detroit Owner (yes = 1, no = 0)   0.027 ***
                                  (6.91)
Size (per 1,000 square feet)      -0.016 ***
                                  (-4.00)
Age (Decades)                     0.079 ***
                                  (4.60)
Bank Owned Property               0.155 ***
(yes = 1, no = 0)                 (13.79)
Delinquent on Water               0.1489 ***
(yes = 1, no = 0)                 (37.30)
Number of Observations            80,269
Number of Censored Observations   26,537
Pseudo [R.sup.2]                  0.064

Notes: Coefficients represent marginal effects evaluated at mean
values. Numbers in parentheses are z-values. * Indicates
significant at the 90% confidence level. ** Indicates significant at
the 95% confidence level. *** Indicates significant at the 99%
confidence level.

Table 3: Heckman Second Stage Estimation

Dependent Variable:
ln(Delinquent Amount)
                                          Homestead    Non-Homestead
Independent Variable     All Properties   Properties   Properties

                         0.002            -0.092 ***   0.036 ***
Size (per 1,000          (0.25)           (-4.11)      (3.68)
  square feet)
                         0.008 **         0.021 ***    0.001
Age (Decades)            (2.42)           (3.29)       (0.19)
                         -0 197 ***       --           --
Homestead Property       (-9.28)
  (yes=1, no=0)
                         0.007 ***        0.008 ***    0.006 ***
Statutory Tax Rate       (26.88)          (16.17)      (18.33)
                         0.039 ***        0.0375 ***   0.039 ***
Taxable Value            (54.67)          (30.42)      (44.33)
  (per $1,000)
                         0.011 ***        0.002        0.015 ***
Years Owned              (5.33)           (0.74)       (5.27)
Delinquent on Water      0.468 ***        0.769 ***    0.360 ***
  (yes=1, no=0)
                         (22.29)          (18.75)      (17.81)
                         -0.388 ***       0.049        -0.576 ***
Lambda (InverseMill's    (-5.18)          (0.49)       (-5.83)
  Ratio)
Wald [Chi.sup.2]         22,247           8,253        14,162
Number of Observations   157,985          77,716       80,269

Notes: Estimations include a set of neighborhood indicator variables
that are not reported here. Numbers in parentheses are z-values. *
Indicates significant at the 90% confidence level. ** Indicates
significant at the 95% confidence level. *** Indicates significant
at the 99% confidence level.
COPYRIGHT 2014 Southern Public Administration Education Foundation, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2014 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Alm, James; Hodge, Timothy R.; Sands, Gary; Skidmore, Mark
Publication:Public Finance and Management
Article Type:Report
Geographic Code:1USA
Date:Jun 22, 2014
Words:10238
Previous Article:Symposium on public finance after the crisis: U.S.A. introduction.
Next Article:Blind, broke, and bedlam: differentiating fiscal stress from bankruptcy in California.
Topics:

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |