Back to Tax Revision Commission — Back to list of research papers
|
Columns DCWatch
Archives Elections Government and People Budget issues Organizations |
Preliminary Report on Aspects of the Real Property Tax System of the District of Columbia by John H. Bowman, Ph.D. with the assistance of Michael E. Bell, Ph.D. and Thomas E. Heinemann November 1997 Table of Contents (The tables are not available on line. Please call (202) 518-7275 for copies of the tables.)
The property tax is the largest single source of revenue for state and local governments in the U.S. In 1994, the most recent year for which the U.S. Census Bureau has published comparable data for the U.S. and the 50 states, state and local property taxes yielded $197 billion of revenues. This represented 31.5 percent of total state and local tax revenues, and 22.3 percent of total state and local own-source revenues. While the absolute amount of state and local property taxes tripled over the last 15 years, its share of tax revenues remained virtually unchanged, while its share of total own-source revenues dropped modestly -- from 24.2 percent in 1979 to 22.3 percent in 1994. The purpose of this section is to describe recent trends in the role of property taxes in state and local finances, their contribution to state and local revenues in the metropolitan area, and in assessed values, rates, collections, and delinquencies in the District of Columbia. The Role of Property Taxes in State and Local Finance While property taxes tripled nationally between 1979 and 1994, they increased even faster in the District of Columbia, Maryland, and Virginia. According to data published by the U.S. Census Bureau, state and local property taxes increased between 1979 and 1994 by 282 percent, 216 percent, and 279 percent in the District of Columbia, Maryland, and Virginia respectively (see Table 1). The pattern of growth, however, varied among the jurisdictions and over time. For example, during the decade from 1979 to 1989, property taxes in Maryland, Virginia, and the nation increased by 124 percent, 185 percent and 120 percent respectively. During the same period, property taxes in the District of Columbia increased by 235 percent, nearly twice the national average. Alternatively, for the period from 1989 to 1994, property taxes in the District increased by only 14 percent while property taxes in Maryland, Virginia, and the U.S. increased by 41 percent, 33 percent, and 38 percent, respectively. While the absolute dollars of state and local property taxes increased in all three jurisdictions over this 15 year period, the relative importance of property taxes experienced somewhat different trends. For example, property taxes accounted for a greater share of state and local tax revenues in 1994 than in 1979 for the District of Columbia, Maryland, and Virginia, and was virtually unchanged for the U.S., albeit the pattern of change was not uniform across jurisdictions and over time. The relative dependence on property taxes for state and local tax revenues increased from 1979 to 1989, and from 1989 to 1994 in both the District and Virginia. Alternatively, the relative dependence on property taxes for state and local tax revenues fell from 1979 to 1989 in Maryland and the U.S., in part as a response to concerns about taxpayer revolts that spread across the U.S. in the wake of Proposition 13 in California. However, the relative dependence on property taxes for state and local tax revenues increased dramatically from 1989 to 1994 in Maryland and the U.S. -- erasing all of the decline that took place during the previous decade. In part, this reflects a continuing effort to shift revenue raising and service delivery responsibilities from state to local governments.(1) Similarly, property taxes accounted for a greater share of state and local own-source revenues in 1994 than in 1979 for the District of Columbia, Maryland, and Virginia, while their relative importance for state and local own-source revenues fell modestly for the nation as a whole. Again, the relative dependence on property taxes increased in Virginia and the District over the entire period, fell from 1979 to 1989 in Maryland and the nation, and increased rapidly in Maryland and the nation from 1989 to 1994. In summary, the District of Columbia increased substantially the absolute amount of property taxes and the relative dependence on property taxes for both total tax revenues and total own-source revenues, while state and local governments in Maryland and the U.S. reduced their reliance on property taxes. During the period 1989 to 1994, however, the District constrained the growth in property tax revenues and its dependence on the tax. This is in sharp contrast to the substantial increase in property tax revenues and its relative importance in generating tax and own-source revenues in Maryland and the U.S. during the same period. These trends have accelerated in the District recently and are the subject of the next section. Property Taxes in the District of Columbia: Recent Trends Data from the District of Columbia Comprehensive Annual Financial Report confirm the general trends identified in the Census data. Property tax levies and total collections increased throughout the mid- to late-1980s and early 1990s. Property tax levies reached a peak of $929 million in 1993 and total collections reached a peak of $889 million the same year. (2) Ignoring 1993, property tax levies and total collections in the District increased at an average annual rate of approximately 12 percent between 1986 and 1992. In Fiscal Year 1994, property tax levies declined by more than 22 percent and total collections fell by over 20 percent -- in significant part because of the inflated values reported for 1993. If the adjusted levies and collections for 1993 are used, property tax levies fell at an average annual rate of 2.0 percent between 1993 and 1996 while total collections fell at an average annual rate of 1.4 percent during the same period (see Table 2).(3) These trends are somewhat exacerbated when property tax levies are adjusted for inflation. Using the Consumer Price Index for metropolitan Washington D.C. to adjust property tax levies for the effects of inflation, real property tax levies in 1987 equaled $403 million dollars and increased every year reaching a peak of $567 million in 1992. Between 1987 and 1992, real property tax levies experienced an average annual growth rate of 7.1 percent. Real property tax levies then fell every year between 1993 and 1996, declining to $439 million in 1996.(4) During this four year period, real property tax levies experienced an average annual decline of 6.2 percent. Real property tax levies in 1996 were the lowest level they have been since 1988 (see Table 2). A primary cause for the decline in property tax levies and total collections was a dramatic decline in taxable assessed values which resulted from the softness of the real estate market which started in 1992. After increasing at an average annual rate of 11.8 percent between 1986 and 1992, taxable assessed values declined 7.7 percent in 1993, 1.3 percent in 1994, 5.0 percent in 1995, before rebounding 1.3 percent in 1996. These trends were driven in large measure by activity in the commercial real estate market. Between 1986 and 1992 the assessed value of commercial real estate increased at an average annual rate of 15.2 percent, compared to just 8.2 percent for the assessed value of residential properties. Then, in 1993, the assessed value of commercial real estate declined by over 12 percent, followed by declines of more than 8 percent in 1994, 3.4 percent in 1995 and 4.7 percent in 1996. The assessed value of residential real estate declined 1.6 percent in 1993 and 6.7 percent in 1995, but was up 7.0 percent in 1994 and rebounded 7.6 percent in 1996 (see Table 3). In addition to the declining assessed value of commercial and residential real estate, property tax rates have not been increased in the District since 1990 -- except for an increase in the rate applied to vacant properties which went from $3.29 per $100 assessed value to $5.00 per $100 assessed value in 1993. Thus, declines in assessed value were automatically translated into declines in levies since rates were held constant. Exacerbating the decline in property tax levies has been an increase in delinquencies.(5) Between 1986 and 1992 collections from current tax levies averaged 97.9 percent, a respectable delinquency rate of just 2.1 percent. However, between 1993 and 1996, the average delinquency rate ballooned to 8.2 percent, and has been near 10 percent in both 1995 and 1996. Between 1992 and 1993 the rate of delinquencies more than doubled -- increasing from 3.0 percent to 6.6 percent. In large part, this is a direct result of changing the end of the tax year from June 30 to September 30. For example, in 1992 the first half of the fiscal year was from July 1 to December 31. The payment of property taxes for that period was due September 15, albeit they were not considered delinquent until December 31. Changing the tax year meant that tax payments for the second half of the year (April 1 to September 30) were still due September 15, but were considered delinquent after September 30. Thus, for anyone who had a mortgage escrow account which paid property taxes between October 1 and December 31, they would not have been delinquent in 1992, but they would have been delinquent in 1993. It is estimated that this technical issue accounts for a large portion of the increase in the delinquency rate between 1992 and 1993. In addition, however, the credibility of the property tax has been under attack since 1992 and as people lose faith in the tax, they may become less responsible in paying their taxes. For example, a new computer assisted mass appraisal (CAMA) was instituted in 1992. However, because of serious technical difficulties, the initial couple of years resulted in many faulty assessments. Appeals increased substantially as the credibility of the tax was put in question. Since taxes under appeal typically are not paid until the issue is resolved, an increase in appeals results in an increase in delinquencies as well. This situation has been exacerbated throughout the mid-1990s as the quality of assessments, measured by the coefficient of dispersion reported annually as part of the assessment-sales ratio studies, deteriorated consistently. As a result, individual property owners may question whether they are paying their fair share of taxes and may become reluctant to pay their taxes in a timely fashion. High-quality administration is a critical factor in influencing society's acceptance of the property tax. Another factor affecting the ability of the District to generate revenues from the local property tax is the trend in tax-exempt property. In 1996, the total assessed value of all real estate in the District was approximately $72.4 billion. Of that total, nearly $30 billion was exempt from the local property tax -- 41.1 percent of the total assessed value. In 1986, 47.5 percent of total assessed value was exempt from property taxation, but that share declined steadily to 37.9 percent in 1991. Since 1991, however, the share of assessed value exempt from property taxation increased steadily to 43.3 percent in 1995 before falling modestly in 1996. Currently, there is a backlog of over 400 applications for properties to receive an exemption for paying property taxes (see Table 4). According to O'Cleireacain, 65 percent of the property exempt from property taxation in the District is owned by the federal government. Another 7 percent of property exempt from property taxation is directly attributable to the federal presence in the District -- foreign government property (e.g., embassies), property exempt by special acts of Congress (e.g. American Association of University Women, the National Education Association, and the National Society of Colonial Dames), and property exempt by executive order of the President (The World Bank, the International Monetary Fund, and the Organization of American States). The remaining 28 percent represents other exempt property uses - churches, universities, non-profit organizations, etc.(6) A direct result of the decline in property tax levies and collections has been a reduction in the property tax burden on citizens in the District. Property tax levies per capita stood at $736.72 in 1987, increasing to a peak of $1,401.36 in 1992.(7) This represents an increase in per capita property tax levies of 90.2 percent over this period. As a result of declines in property tax levies after 1992, per capita property tax levies fell by 8 percent between 1993 and 1996. Decreases in property tax levies were offset somewhat by declines in population over this period. Another, perhaps more meaningful, measure of the property tax burden is property tax levies per $1,000 personal income. In 1987 property tax levies were equal to $37.64 per $1,000 of personal income, increasing to a peak of $48.01 per $1,000 personal income in 1991. This measure of the burden of property taxes then declined every year from 1991 to 1995, falling to $38.85 per $1,000 personal income -- a decline from the peak of over 19 percent. In terms of its claim on personal income, in 1996 the property tax burden in the District was a modest 3 percent higher than in it was 1987. Classification of property for tax purposes means the establishment of differential effective tax rates -- i.e., taxes that are different percentages of market value. Precisely what might be considered classification, however, is a matter of interpretation. Clearly, many different policies produce effective-tax-rate differentials -- e.g., homestead exemptions, use-value assessment of farmland, circuit breakers. The term classification, however, tends to be reserved for schemes that assign all property types or uses to some class. Thus, classification is a comprehensive policy for establishing relative tax loads. By contrast, while a homestead exemption, for example, also creates differential effective tax rates across groups of property owners, its terms at least appear generally to be set without explicit attention to the differentials being created. As a practical matter, the difference between classification and other forms of direct property tax relief tends to be small, almost semantic. From the mid-19th century to the middle of the 20th century, uniformity was the norm for U.S. property taxation. Classification of real property began in Minnesota in 1913, spread to Montana in 1917, and then to West Virginia in 1932. Between 1932 and 1968, no other states adopted comprehensive real property classifications. In that latter year, however, Arizona became the fourth classification state. Since then, over half the states have become part of the trend. In many cases, however, de jure classification was adopted simply to codify, as nearly as possible, the pattern of de facto classification that had emerged over a number of years. The codifications were prompted by judicial orders, actual or feared, to enforce the uniformity standards that traditionally had been part of the legal framework of the tax [Bowman 1987 ??]. Classification in the District of Columbia District law provides for five classes of real property [Sec. 47-813]. Briefly stated, the five classes (with their 1997 tax rates, expressed as percentages of assessed value) are:
In reality, any brief statement of the classes is not particularly informative. The definitional materials of Sec. 47-813 for the latest period - omitting the changed provisions that had applied in earlier years that still are a part of the code section - are presented as an appendix. It occupies several pages. As recently as tax year 1978, all real property in the District was taxed at a single rate, 1.83 percent of assessed value. In 1979, residential property was given a reduced rate of 1.54 percent, and all other real property was taxed at the former rate, 1.83 percent. For tax year 1980, the residential class was split into owner-occupied (class 1) and other (class 2), at rates of 1.22 percent and 1.54 percent, respectively, and all other real estate (class 3) continued to be taxed at 1.83 percent. The three-class system remained through tax year 1985, but for 1986 a fourth class was created by breaking hotels and motels out from other commercial property, thus evolving the old class 3 into classes 3 and 4. The fifth class - vacant property - was added in tax year 1991. Thus, the pattern has been for the number of classes to proliferate. This is consistent with the experience of several states, and illustrates the concern that some critics have voiced about classification. A former tax commissioner of Minnesota, the state that started real property classification in 1913 and that has taken it to a more extreme form than most other states, warned 30 years ago that, once started, there is no logical stopping point for classification [Hatfield 1967, 242]. This same phenomenon has caused some to characterize classification as the first step on a "slippery slope" [Sonstelie 1978, 237]. The tendency for change in classification shows up not only in the number of classes, but also in the relative rates at which the classes are taxed. At 1.54 percent and 1.83 percent of assessed value, the tax rates for tax year 1979, the District's first year of classification, the commercial rate was less than 20 percent higher than the residential rate (1.83/1.54 = 1.19). However, when residential was split into two classes the next year with owner-occupied residences taxed at the reduced rate of 1.22 percent, the differential between the highest- and lowest-taxed classes jumped to 50 percent (1.83/1.22 = 1.5). The following year, tax year 1981, differential jumped again, to 75 percent, as the commercial rate was increased to 2.13 percent. The range of rates actually narrowed for a time when the commercial rate was lowered from 2.13 percent to 2.03 percent, effective in tax year 1985. Creation of a fourth class the next year did not affect the range between the highest and the lowest rates, for it established for hotels and motels a new rate below that for other commercial property, 1.82 percent versus 2.03 percent. For tax year 1991, however, tax rates were changed at both ends. A new, fifth class -- vacant properties -- was created and taxed at 3.29 percent, and the rate for owner-occupied residences was cut to .96 percent. Thus, the differential between the highest and lowest rates rose from 66 percent (2.03/1.22 = 1.66) to 243 percent (3.29/.96 = 3.43). Effective tax year 1992, the class 4 (other commercial) rate was raised from 2.03 percent to 2.15 percent, but this did not affect the range of rates. The last change (at least through the 1997 tax year) came three years later, when the class 5 (vacant) rate was increased to 5.0 percent, widening the differential still more, to 421 percent (5.0/.96 = 5.21). Classification is controversial, even though it is now relatively widespread. A possible explanation of this is an observation by Rolland Hatfield, Minnesota Tax Commissioner, on the eve of the comparative explosion of real property classification in the United States but after 50 years' experience in his state:
These and related considerations are discussed briefly, below. Often, the various criteria point to different choices. The tradeoffs must then be weighed. Equity Concerns Ability to Pay. Classification proponents typically say it promotes equity, or fairness. "In the words of Simeon Leland, an early advocate of classification, 'Different types of property are not possessed of the same capacity for throwing off taxes onto other property and persons.'"(8) Proliferation of types of property, which intensified in the latter part of the 19th century, is said to have undermined the logic of tax uniformity. A key distinction often drawn is between income-producing and non-income-producing property, with the former said to represent greater taxpaying ability. In this view, it is logical to devise classification systems that place lower rates on residential property, and particularly owner-occupied residential property. But are equity considerations clear-cut? In what sense is it fair to impose different levels of tax on different property uses, causing different burdens on properties of equal value? And if some differentials are fair, how large should they be? What has guided the District in going from one class to five, and from differences in rates of less than 20 percent initially to over 420 percent now? Was each variant fair? If the answer is yes, then standards of fairness must change radically in relatively short time periods; otherwise, changes on the order of those that have occurred in the District might be characterized as capricious. But if the standards of fairness do change quickly and sharply, what are they, and how can they be tracked or anticipated? If the answer is no -- i.e., if not all versions of classification in the District have been fair -- that unfair versions came about seems to undercut the argument that equity is served by classification. Hatfield may have been correct, 30 years ago, in observing that classification turns the property tax from a levy on accumulated wealth to a tax based on political influence - or lack thereof, with local homeowners (generally perceived as the voting class) being taxed much more lightly than big businesses, whose owners often are not residents, and therefore not voters, in the jurisdiction imposing the tax.(9) Another consideration is the role of the property tax within the overall tax system. Income taxation bestows favor upon homeowners by excluding the value of the housing services from the definition of income while allowing property taxes and mortgage interest as deductions. This imbalance has been pointed out by many writers (and some politicians), yet it persists. This tilt of the tax system in favor of residences and against income-producing property is compounded by standard classification systems, such as that found in the District of Columbia.
It is at least open to question whether cumulative tax breaks of this sort are, indeed, fair. For example, it might be argued that, to help offset the imbalance of the income tax system, classification of property might well impose higher rates on non-income-producing properties. Benefits Received. So far, discussion of equity has seemed to presume the appropriateness of the ability-to-pay concept of equity. Under that approach, equity requires that taxpayers in essentially equal circumstances should bear essentially equal taxes. Another approach to equity, however, is to consider benefits received. Under this conception, tax payments should be in proportion to benefits. Many view the property tax as basically a benefits tax. There seem to be important exceptions to this, but there is some merit, as well. To the extent that the benefits rationale applies, it seems impossible to justify some of the differences that are found in the District's pattern of tax rates. The treatment of Class 2 residential property seems particularly hard to justify. If the tax is passed on to renter, as many assume, what is the rationale for taxing them 60 percent more heavily (even before counting the effects of the homestead exemption) than owner-occupants? And if the tax is not passed forward, what is the rationale for taxing apartments more lightly than other commercial properties? In summary, the classification system seems not to square well with either standard notion of equity. Perhaps there are other advantages that offset this. Efficiency Efficiency is valued because it implies the avoidance of waste. One way in which taxes can cause inefficiency - i.e., costs in excess of the amount of taxes collected - is by inducing changes in decisions, thus causing otherwise inferior options to be taken because of tax considerations. Economists use the very descriptive term excess burden to describe the costs associated with such non-neutralities, because they are costs above what would have to be imposed to raise a given amount of revenue in a more neutral manner. An example of the excess burden of a tax is the relocation of economic activity elsewhere. Other reports to this Tax Revision Commission deal with the question of the effect of taxes on the location of economic activity. While this report seeks to avoid going over this same ground, the issue must be noted in dealing with the possible effects of classification. As those reports have noted, this is a complex issue. The effect of taxes varies by type of tax, type of business, and the availability of alternatives (the last partially overlaps the type-of-business consideration). In general, though, taxes have been shown to have statistically significant negative effects on activity within jurisdictions with relatively high taxes, all else equal. Moreover, the tax-induced effects tend to be greatest within metropolitan areas, where alternative locations are rather conveniently available. Thus, this should be of some concern to the District's policy makers. The small geographic area occupied by the District contributes to a very open local economy. A very dramatic illustration of this occurred about 20 years ago, following the 1973 oil embargo, when the District gasoline tax, in an effort to offset declining revenues due to diminished taxable gallons sold, was raised to levels above those in the Maryland and Virginia suburbs. The result was a sharp further drop in taxable gallonage sold within the District, and the rate was scaled back. Response to property tax differences is not so quick or, therefore, noticeable. After all, one of the standard reasons given for traditional local reliance upon property taxation has been that the property tax base is less affected by inter-jurisdictional differences than income or consumption taxes might be. Still, reductions in property tax base have been found to be linked to higher rates of tax - and not just higher property tax rates, but rates of other taxes, as well [Ladd and Bradbury 1988]. This poses a dilemma for many central cities, which must compete with suburban areas for residents and businesses. If taxes get very far out of line - and if they are not perceived to be offset by better services or other advantages of being in the city - the long-term fiscal health of the city is jeopardized by higher taxes. In addition to the current level of taxes, however, is the uncertainty about what their future level may be. The rather rapid increase in differentials through classification logically would induce caution on the part of investors, especially since the District is surrounded by jurisdictions with more stable tax systems. Administration and Compliance Costs Another source of inefficiency (a sort of excess burden) is incurred if the actions required to make the tax work impose higher costs than would have been necessary for alternative ways of raising the same revenue. In general, two distinct groups incur costs in making a tax system work: Tax administrators (the government) have to prepare tax rolls and tax bills, and enforce the taxes imposed. In so doing, administrative costs are incurred. Taxpayers also bear some costs, through filling out returns, keeping records, and the like. These costs are compliance costs. Compared to income and sales taxes, property taxation entails relatively low compliance costs. The tax is said to be taxpayer-passive, whereas the income tax (for both the employer and the individual filer) and the sales tax (for the vendor) are taxpayer-active. Although precise figures were not available, classification clearly increases the costs of both administration and compliance. In the case of residential property, for example, when a property goes from being owner-occupied to renter-occupied, it changes from class 1 to class 2. Also, mixed-use properties (of which there are about 3,000) get a blended rate, based upon the relative importance of each use. Changes in these shares should trigger tax-rate changes. The Real Property Tax Administration reports that there constantly are classification changes to be dealt with. Currently, one full-time person (out of a total staff of about 50) is assigned to dealing with these matters, and some time of other staff members also is required. Additional compliance costs also are present. Taxpayers have to determine whether they are being treated correctly and, if not, take steps to try to have made the changes that they feel are appropriate. Other Concerns Certainty. One of the longstanding principles of taxation is that taxes should be certain - i.e., known clearly to the taxpayers. As originally used by Adam Smith in The Wealth of Nations over two centuries ago [1937 edition, 778], this meant precluding capriciousness by the tax collector. Although recent hearings focusing on complexity of the federal income tax and behavior of the Internal Revenue Service indicate that this concern is not dead, the more modern concern for certainty tends to focus, in part, on the disadvantages of frequent changes in tax law. Such changes create uncertainty and thus make long-term planning more difficult and hazardous. Probably few District investors could have foreseen several years ago, when making decisions about investing there, that there would be such pronounced changes in relative tax levels for different types of property as have come about. That history of proliferating tax classes and creating ever-larger differentials may discourage future investment, in part because it makes it impossible to guess what taxes will be in the future, but suggests the likelihood of further change. Fundamental Change in Tax System. One of the topics before this Tax Revision Commission is a split-rate, or graded, property tax, under which land would be taxed more heavily than improvements. The authors of the paper studying this alternative note that the combination of classification and a graded tax (especially the extreme version that would zero-rate improvements, leaving only a land tax) would be an uneasy one, pulling policy in very different directions with regard to neutrality [Schwab and Harris 1997, 24]. If such fundamental change in the property tax were to be given serious consideration, therefore, careful thought would need to be given to the continuation of classification. Summary The foregoing discussion shows that classification scores very poorly on the standard criteria for evaluating taxes. It increases administration costs and, to some extent, compliance costs. As practiced over its first two decades in the District, classification also has brought tremendous uncertainty as to the relative taxation of different uses of real property. This may discourage investment in the District. A related matter is the non-neutrality of classification. Tax-rate differences can cause alternatives that are less attractive before tax consideration to become more attractive once the tax is taken into account; that is, the tax may cause property to be devoted to uses that provide fewer benefits. In the context of the open metropolitan economy of the Washington area, moreover, the contrast between the classified District tax and the uniform taxes of the suburban jurisdictions - which generally have lower rates, especially for non-residential property - makes the District a less attractive location. The main argument for classification is equity, but it turns out that equity - like beauty - is in the eye of the beholder. It would be difficult to gain consensus on the proposition that the classified property tax is fairer than a uniform tax, and even more difficult to get it for any one classification scheme over another. Classification tends to change the property tax from a tax on accumulated asset value to a tax on the lack of political power. Removing the discipline of a uniformity requirement opens the tax code to political maneuvering. Individuals vote, businesses do not, so the classification system favors residential property over business property. If there were one logically "right" or fair set of relative tax levels, we would see neither the changes in relative tax rates that have occurred in the District (and many other classification states) nor the differences among such systems at a given time. Naturally, homeowners tend to feel that the tax system is fairer if it taxes them relatively lightly. Many people define a fair tax as one that burdens someone else. But this view of fairness offers no principles to guide it. If classification exists, with its terms set only by statute, continued jockeying for a more favorable treatment should be expected to lead to changes in the system. The slippery slope, in such circumstances, is a real phenomenon. The discussion has revealed serious flaws with classification. This section presents several recommendations. Some provide a range of alternatives for dealing with the situation; others are independent, and might be adopted in combination with one another. Level of Property Taxation Recommendation 1. The District of Columbia should seek to bring its reliance on property taxation into line with that of other jurisdictions, particularly its Maryland and Virginia neighbors. Materials presented in the section providing an overview of property taxation (Table 1) show that, in the mid-1990s, the District's property tax is high, whether measured as a percent of personal income of residents (about half again as high as in Maryland and Virginia) or as a percentage of own-source state-local revenues (about one-fourth again as high as in these neighboring states). Moreover, the rates applicable to commercial properties in the District are well above those in neighboring areas. Reduced reliance on property taxation could help make possible more fundamental reform of the property tax. Recommendation 2. The District of Columbia should abolish classification of real property, returning to a uniformity rule. While this might be phased in over a period of time - perhaps five to 10 years - a commitment to this change should be made and kept. The advent of classification in the District two decades ago has, indeed, turned out to have been the first step on a slippery slope. The number of classes has gone from two to five, and the differentials have increased tremendously. Initially, the highest-taxed class rate was less than 20 percent above that of the lowest-taxed class, but for tax years 1995-1997, the highest rate was 420 percent above that of the lowest-taxed class. Within the broad category of residential property - initially a single class - renter-occupied property is taxed 60 percent more heavily than owner-occupied property, even before the $30,000 homestead exemption is taken into account. Such differentials have no logical justification. They move the property tax from a tax on accumulated property wealth to a tax on relative lack of political power. Although the differentials have been created in the name of equity, they are inequitable, whether considered from the standpoint of ability to pay or benefits received. Moreover, in the context of the whole tax system, a case might be made for a reverse classification, placing a heavier rate on owner-occupied residential property, which is allowed deductions under the income tax even though the stream of housing benefits created goes untaxed. Beyond the equity considerations, classification also scores poorly on efficiency grounds. Within the District, the tax differentials tend to direct resources into tax-favored uses. The need to classify properties also creates added administrative burdens, particularly for mixed-use properties and for residential properties that change from owner- to renter-occupied, or the reverse. Because the District is a comparatively small island of land surrounded by larger land areas in the Maryland and Virginia suburbs, the District's classification system also creates biases in the choice of location that work against the District. First, the current rates on commercial properties in the District are considerably higher than in the suburban areas, where real property tax rates are uniform across types of real property. Second, the history of change within the District's classification system is itself a concern. An investor considering location within the metropolitan area would have to consider not only the currently-higher tax rates in the District compared to those in the suburbs, but also the likelihood - based on past performance - that the commercial rate(s) might become even higher relative to those for other property types. Is there more of the slippery slope ahead, or has it been sanded and/or leveled? In summary, a return to uniformity would provide a clear, understandable guide to the valuation of and taxation of property - a uniform rate applied against market value. This would remove the determination of property tax treatment from the political arena. It would provide more certainty to property owners and prospective investors, and also would reduce the costs of administration of (and to a lesser extent, compliance with) the District's tax code. This recommendation could be carried out more readily if Recommendation 1, reducing the District's reliance on property taxation, were implemented. This is true because the return to uniform property taxation would increase relative taxes on owner-occupants of residential properties, currently the most-favored group of property owners. Recommendation 3. The District of Columbia should collapse its current five-class real property classification system into a system with only two classes - all residential properties (rental and owner-occupied) in one class, and all other properties in the other. The differential between the classes should be modest, and it should be frozen. This is an alternative to the full uniformity envisioned by Recommendation 2. While a two-class system would not overcome some of the problems with classification - after all, it still would be classification - it would reduce them. With fewer classes, the system would be easier to administer and easier for taxpayers to understand. It also would be somewhat more fair. At least, the differential within the broad category of residential property would be gone (leaving aside, for now, the homestead exemption). With the differential between the classes kept relatively small - say, the commercial rate no more than 50 percent higher than the residential rate - the locational effects of the tax would be reduced. Finally, fixing the differential between the classes so that it is not up for on-going political negotiation would - if believed to be an effective and lasting fix - remove the disincentive to invest in the District represented by the current system with its proliferating classes and spreading differentials. A mere statutory restriction would not be convincing. Whatever one session of City Council adopts, another could repeal or supplant. The initial two-class system was provided for by statute, and it quickly was undone. This recommendation is an acknowledgment of the political difficulty of moving from a system that bestows very large tax-reduction benefits on District homeowners, who also are voters or potential voters. Desirable as a return to uniformity seems on many criteria, it also seems unlikely to happen. Because the rate differentials have become large, it is obvious that returning to uniformity would entail significant tax increases for the properties that now are favored, and decreases for the others. It has been estimated, for example, that ending the District's classification system (and also ending the $30,000 homestead exemption) while maintaining the revenue yield of the tax would result in higher taxes for class one property - over a 90 percent increase - while permitting other classes' taxes to decline anywhere from a negligible fraction of a percent for class 2 to nearly 70 percent for class 5 [Schwab and Harris 1997, 40-42]. A Brookings Institution study published earlier this year recommended property tax changes featuring a two-class system with rates (0.9 percent for residential property, 1.35 percent for other property) that would reduce overall property tax revenue by 27 percent, thus permitting a reduction in property taxes across the board [O'Cleireacain 1997, 11 and 81]. Provided a good way could be found to make up for the revenue loss, this compromise may be a good as could be hoped for. Again, this recommendation could be carried out more readily if Recommendation 1, reducing the District's reliance on property taxation, were implemented, for reasons given in the discussion of Recommendation 2. Recommendation 4. The District of Columbia should bring more certainty to its property classification system. This could be accomplished by freezing the number of classes and the tax-rate differentials among them. This basically is a status quo option. Given what has been said earlier about the current classification system, this clearly is not a very good alternative. However, even if both uniformity (Recommendation 2) and two-class classification (Recommendation 3) were to prove infeasible, some improvement still would be made if the number of classes were fixed, along with the differentials among them. This could remove future political negotiation as a factor in property tax treatment (although it would enshrine the results of past maneuvering). Most importantly, it could remove the uncertainty as to future changes in the property tax system, which should make the District a more attractive place in which to invest. The fundamental problem in such an approach is making any freeze seem real, or permanent. As noted above, a mere statutory change would not be convincing, because it could be negated by a new statute. Recommendation 5. The District of Columbia should set itself on a course that would eliminate property taxation over the next several years. Property taxation is an old institution and, while not without some faults, it has some clear merits. Also, any replacement revenue sources (at least among District own-source revenues) come with their own faults, which likely would be exacerbated by increasing their use enough to make up for property tax repeal. What the District (and many states) now call property taxation, however, deviates substantially from the norms for such taxation. As noted, the classification system tends to convert the tax into a tax on the relative lack of political power. Residential property owner-occupants get greatly reduced taxes because they have the political clout to get them. Effective-tax-rate differences that emerge have no other underlying logic; they cannot be justified by standard tax evaluation criteria. In such a situation, it is a stretch to characterize the system in place as property taxation. True, ownership of property is the proximate cause of being presented a tax bill. But the magnitude of that bill - in comparison to the tax bills of other property owners in other classes - has little to do with the underlying market value of the property. Even before the homestead exemption is taken into account, a $100,000 class 5 property is assessed a tax higher than that for a $500,000 home that is occupied by its owner. And a $100,000 class 4 commercial property is taxed about the same as a $140,000 apartment building (class 2). The current classification system is indefensible by any logical criteria. If District policy makers cannot bring themselves at least to put the brakes on changes in the classification system, they should face up to the fact that they do not believe in property taxation - for what exists, as noted, really is something else - and abandon the pretext of property taxation. Assessment-sales ratio studies relate the sales prices of sold properties to the assessed values of those same properties and, thus, provide a measure of the level of assessments. As part of an effort to inform the public of the operation of the assessment process and the level of their property assessments, the code of the District of Columbia requires annual assessment-sales ratio studies [Sec. 47-823]:
The studies are published in the District of Columbia Register. The most recent of these, the 1997 ratio study, was transmitted from the Real Property Tax Administration in September. After considering the sorts of information commonly produced by ratio studies, some of the specifics of the District's studies are taken up. Then assessment-sales ratio findings for the District are presented. Ratio studies can yield several statistics to evaluate property tax assessment quality - i.e., the level and uniformity of assessments. They entail computing, for each property in a sample of sold properties, the ratio of the assessed value to its sales price; generally, the result is expressed as a percentage. Assessment Level. To illustrate the information that can be obtained, Table AR1 presents hypothetical examples. There, all five hypothetical properties are shown to have sold for the same price, $150,000, while their assessed values ranged from $120,000 to $180,000. The ratio for property A is 80 percent (120/150) and, at the other extreme, the ratio for property E is 120 percent (180/150). Three measures of central tendency are shown in the exhibits at the bottom of the table:
As the example has been constructed, each of these takes a value of 100 percent. Thus, a person looking at only the central tendency measures would conclude that the assessor had done an excellent job - assessed values and sales prices match. Assessment Uniformity I -- Horizontal Equity. Ratio studies almost always go beyond this, however, to calculate a measure of variation in individual ratios around the average (usually median) ratio, to gauge assessment uniformity. The most commonly used uniformity measure in ratio studies is reported to be the coefficient of dispersion, or COD [Eckert 1990, 534]. It divides the average absolute deviation of the individual ratios by the median ratio, and expresses the result as a percentage. Table AR1 shows the absolute deviations in the last column. Because absolute values ignore the signs, both property A (ratio = 80 percent) and property E (ratio = 120 percent) have deviations of 20 from median ratio of 100 percent. The sum of these absolute deviations is shown to be 60, and the average (60/5) is shown to be 12. Because the median ratio is 100, the COD is 12. This is considered relatively good performance [Eckert 1990, 534 and 540]. It tells us that, on average, individual properties are assessed within 12 percent of the median. If all five properties were assessed at the same level, the COD would be zero, for there would be no deviation around the median;(10) thus, a higher number indicates less uniformity. Getting ahead of the story a bit, it is noted that the citywide COD for residential property in the 1997 ratio study is 14.5, a bit higher than the hypothetical example, but still respectable. However, the District's ratio studies consider only the middle 50 percent of the properties when arrayed by assessment level: "The coefficient of dispersion reflects the variation of individual ratios around the median ratio, and indicates how close to the median ratio the middle 50% of the ratios are.(11) This is not the standard approach, but it has been used elsewhere. It is the approach used in Virginia, for example, until just a few years ago. It can make assessment uniformity appear to be much greater than it is. The hypothetical data of Table AR1 do not lend themselves well to using only the middle 50 percent of the ratios, for only five ratios are presented. However, the highest and lowest ratios can be omitted, leaving the middle 60 percent. The truncating in this example, therefore, is less than in the District's practice. But it is sufficient to cut the COD from 12 to 4. Is this a misleading example? Data were not obtained to make comparisons for the District, but such comparisons were made for the 15 Arizona counties earlier [Bell and Bowman 1991a]. For one county, there was almost no difference between the standard COD and the one restricted to the middle 50 percent (termed the coefficient of inter-quartile dispersion), but that county had very high dispersion under each measure. For the other 14 counties, the differences between the standard and inter-quartile measures were rather striking; in 12 of them, the more restricted measure reduced measured dispersion by more than 50 percent, and by as much as 88 percent. Omitting the highest and lowest 25 percent of all sales in the sample might be done to avoid giving too much weight to outliers. Throwing out half of available evidence, however, seems an extremely broad definition of "outliers." It is reasonable to get rid of outliers, or at least to reduce their influence, if they might truly be considered unrepresentative of the population. First, use of the median ratio, rather than the mean, reduces the weight given to outliers. And surely, half of all sales is too great a number by this standard. This practice also might serve as a means of offsetting imperfect screening of sales, on the presumption that sales that did not truly reflect market forces would tend to show up at extremes of the array of assessment ratios (discussed below). This is an overly crude proxy, however. Assessment Uniformity II - Vertical Equity. While the COD permits determination of the extent of nonuniformity in the valuations of properties of equal value (horizontal equity), another measure - the price-related differential (PRD) - considers whether there is a systematic bias in favor of either high- or low-valued properties (vertical equity) (PRD) [Eckert 1990, 539-40]. To calculate the PRD, the mean of the individual ratios is divided by the aggregate ratio. The mean of the ratios gives each property equal weight, regardless of its price, while the aggregate ratio (aggregate assessed value divided by aggregate sales price) gives more weight to properties of higher value. There is no systematic bias in favor of either high- or low-value properties if the PRD = 1.0, but a PRD greater than 1.0 indicates a regressive bias (i.e., a tendency to assess low-value properties at relatively high percentages of market value), and a PRD under 1.0 indicates a progressive bias. As noted, the example in Table AR1 results in a values of 100 percent for all measures of central tendency, and thus produces a PRD exactly equal to one. The District's ratio studies do not report PRD values. This may be a useful measure to add, as it provides a different sort of information about assessment uniformity. Also, it would be relatively simple to add, given the use of computers. Aspects of the District's Ratio Studies As stated on the first page of the 1997 study, "The 1997 real property assessment/sales ratio study compares tax year 1998 preliminary assessments of real property with selling prices of 'arms [sic.] length' sales transactions which occurred during 1996." Arm's-Length Transactions. The use of "arm's-length" transactions is necessary to assure that the sales reflect market forces. "It is important to know whether the transaction was arm's length (between unrelated parties or parties not under abnormal pressure from each other) or resulted from foreclosure, condemnation, or other circumstances in which price was not representative of the market" [Eckert 1990, 27]. According to Real Property Tax Administration officials, assessors qualify the sales for ratio studies, setting aside such transactions as those between parties with the same last name and those with partial interest in the property. Although the qualifying is said not to be detailed - for example, sales between related individuals might not involve parties with the same last name, and some parties with the same last name may not be related - the feeling is that too much effort nonetheless goes into sale qualification, or screening. Up to now, no survey of the parties to the transaction is made. Reportedly, the intention is to develop a form to gather several pertinent pieces of information on each sale, such as relationship (if any) of the buyer and seller, any special financing arrangements, and conveyance of personal property in the sale of real estate. This is estimated to be about a year away. When it is done, it may be possible to improve the screening process, and at the same devote less time of the assessors to it. Time Periods of Sales and Assessments. The timing of the sales relative to the assessments also is important. From the above quote, three different years appear to be involved:
Thus, the 1997 ratio study pertains to sales and assessments made in 1996.(12) This same pattern applies to other District ratio studies - i.e, assessed values determined in the closing months of the year are compared to sales that occurred throughout that year. Because most of the year's sales have occurred before the assessments to which the sales prices are to be compared have been determined, assessments could reflect the sales prices. If this were done, the tendency would be to improve the accuracy of the assessments, evidenced by higher and more uniform assessment-sales ratios. In discussing this with officials of the Real Property Tax Administration, it was indicated that assessors may know of the sales when assessed values are being determined, but that they are not trying to "shoot the sales" in order to come up with high, uniform assessment indicators. Whether this timing sequence is a problem depends upon assessors' access to sales data and the uses to be made of the ratios. The timing sequence is unfortunate to the extent the ratio studies are intended as a means of evaluating the accuracy (quality) of assessor performance, whether by his or her supervisors or by the taxpaying public. More confidence could be had in the accuracy of the assessments of unsold properties if the ratio studies compared the assessed values of record at the time of sale to the sales prices. In many jurisdictions, this is the practice. For example, the 1995 Virginia study (published in 1997) gives the following information: "The 1995 assessment/sales ratios are calculated from a selected statistical sample of all fair market sales of real estate in 1995, . . . For each selected sale in a locality, the assessed value in 1995 is compared to its selling price to calculate an assessment/sales ratio" [Virginia Department of Taxation 1997, 2]. Uses of Ratio Studies. Ratio studies have uses, besides informing the public as to the level and uniformity of assessments relative to market values, many of them internal to the assessment process [see Eckert 1990, 518-19]. For this reason, the published study is not the only one made. Assessors conduct their own studies for such purposes as identifying areas or property types that seem to require special attention to improve the level of assessment, and those for which assessments are good. Such information can help in determining how to deploy available resources to best advantage. Assessors may, for example, calculate ratios for shorter periods of time to get a sense of market trends. They also may calculate ratios pooling sales for closely similar properties that are located in different areas, as a way to augment a small number of sales in one neighborhood.(13) Or sales may be drawn from only part of a neighborhood. It is indicated that - while the published ratio study includes information for each of 56 neighborhoods that have been used for several decades - assessors also look at ratios within the smaller sub-neighborhoods that now are the focus of assessing activities. Because Sec. 47-823(c) requires public access to ratio studies for "each of the districts utilized in making assessments" it would seem to require publishing ratios for more than the 56 neighborhoods. Findings from the District's Ratio Studies The District's annual assessment-sales ratio studies include statistics for five groups of property. These do not, however, exactly match the five property classes. Three of the ratio-study groups are residential -- residential, condominium, and multi-family - and they do not break out into owner-occupied (class 1) and tenant-occupied (class 2) sets. The other two ratio-study groups are commercial and vacant land. The last one, vacant land, is narrower than class 5, which can include improved properties that are vacant. The commercial group is comprised mostly of essentially class 4 properties, but could also include some properties from classes 3 and 5. Citywide Data Table AR2 presents citywide summary statistics for the five groups for the eight most recent years (tax years 1991-1998, or studies performed in calendar years 1990-1997). The first impression gained from inspection of the data is that assessments have been rather good, with median assessment ratios relatively high and tending upward over the period, and with dispersion for most groups within the good range (CODs of 15 or under) for most of the groups. The exceptions tend to be in the groups with relatively few sales (especially multi-family) and those generally regarded as being difficult to assess uniformly (especially vacant land and, to some extent, commercial). After presenting an overview for both assessment level and assessment uniformity, summary observations are made. Assessment Level - Median A/S. For each of three groups - residential, condominium, and commercial - median assessment ratios have been within five percent of the 100 percent target for at least the last six years, and they show some upward trend over the eight years. For example, the median residential ratio rose from under 95 percent in the first two years (tax years 1991 and 1992) to over 99 percent in tax year 1998, and ratios for all the intervening years were between these extremes. The pattern was very similar for condominiums, except the first two years were about a percentage-point lower (under 94 percent) and the 1998 ratio was over 101 percent. For commercial properties, too, the pattern was essentially the same, with the lowest ratios being for the first two years (92 percent for tax year 1991, 94.3 percent for tax year 1992) and the highest - in this case, 104.1 percent - being for the latest year. Between these years, the median commercial ratio was at or near 100 percent in each year. For both multi-family housing and vacant land, the median ratios exhibited more variability over the period, and the upward trend, if present, was less clear, as there were years in which the ratio was less than for the preceding year. For multi-family, the median ratio ranged between 82.8 percent (tax year 1995) and 102.1 percent (tax year 1997), and for vacant land, the range was from 79.4 percent (1991) and 100.3 percent (1997). It should be noted that the number of sales in each category in each year was greater than 30, generally taken as the threshold for statistical reliability, with the exception of multi-family properties in tax year 1997 (26 sales) and vacant land in 1998 (28 sales) In fact, for each of the last four years, there were fewer than 50 sales of vacant land. Except for the cases just noted, every other group had more than 50 sales in each year. The largest numbers were for residential properties, with annual sales ranging between 2,871 (tax year 1995) and 4,432 (tax year 1991). Over 1,000 condominium sales were available for analysis each year, and generally well over that level. Even commercial sales were rather numerous, with annual sales varying from 167 for tax years 1993 and 1994 to 343 for tax year 1991. Assessment Uniformity - CODs. In addition to the median assessment levels generally being high, the coefficients of dispersion also look respectable, as noted above.(14) For the residential group, for example, the highest COD - i.e., the greatest degree of nonuniformity - was 14.7 in tax year 1997, followed by 14.5 in tax year 1998. While their level is relatively low, the upward drift in the residential CODs tends to be disturbing. In each of the first four years in Table AR2 the COD was under 10.0, but it is above that level in each of the last four. The upward progression was broken by the 1998 statistic, but the decline was slight, from 14.7 to 14.5. Although the pattern is less clear-cut for condominiums, the highest CODs are those for the last two years 13.3 and 12.7, respectively); the only ones below 10.0 were for tax years 1994 (6.8) and 1993 (7.9). For each of the other three property groups, however, it is difficult to be sanguine about the degree of assessment uniformity. Generally, they also exhibit the pattern observed above, with CODs for the most recent years being among the highest, if not the highest, in the eight-year period, they started from a higher degree of nonuniformity and went up from there. Among these three groups, only commercial had any CODs below 10.0, and those were for three of the first five years; those for tax years 1997 and 1998 were 29.3 and 31.7, respectively. Multi-family housing's CODs range between 11.5 (tax year 1991) and 29.2 (1997), with three of the eight being less than 20.0 - all within the first four years. Finally, vacant land had only two CODs below 20.0 (18.5 for tax year 1994 and 19.2 for tax year 1993); for each of the last four years, COD values were 28.1 or higher, including readings of 34.9 (1995) and 64.3 (1997). While it often is the case that assessment ratios for vacant land show considerable variation, some of the recent levels suggest basis for concern - particularly since the top and bottom quartiles are ignored in the District's COD statistics. Neighborhood Data The published ratio studies also report, as required by statute, the statistics for each of the assessment neighborhoods. Data are for the 56 neighborhoods defined several decades ago. Each is assigned a numerical code as well as a name, and both identifiers appear in the table. Neighborhoods with fewer than 15 sales have no statistics reported. Because the law of large numbers generally applies to samples of 30 and over, some of the data presented may lack statistical reliability. While a major appraisal handbook states that there is no one right answer to how large the sample must be for reliability, it does note that where the population data are more variable, a larger sample is needed. The relatively high CODs for several areas and types of property suggest 15 is too small a number for reliable statistics [Eckert 1990, 526]. This report presents the data from the 1997 ratio study (for tax year 1998) for those situations in which there were at least 30 arm's-length sales [tables AR3-AR5]. Of the 56 neighborhoods, 30 had at least 30 sales of single-family residences. Table AR3 presents the number of sales, median assessment-sales ratio, and the coefficient of dispersion for each of the 30 neighborhoods, arrayed in descending order of the COD values. These data are summarized in Table AR4, a 20-cell matrix cross-classifying A/S values (grouped into five ranges) and COD values (grouped into four ranges). Six of the 30 neighborhoods have CODs between 7.8 and 10.0, and they are evenly divided between the two highest median A/S groups (95.1-100.0 and 100.1-103.8). Twelve of the 14 neighborhoods with CODs between 10.1 and 15.0 also are divided evenly between these top two A/S ranges. Thus, 18 of the 20 CODs below 15.0 are associated with A/S levels of above 95 percent, and six of the 10 neighborhood residential CODs above 15.0 are for areas whose median assessment levels are below 95 percent. This pattern is in line with the literature on the determinants of property assessment quality, which finds that more uniform assessments tend to be found in areas where the assessment level is higher, all else equal [Bowman and Mikesell 1990]. Moving beyond residential property, there are very few instances in which a neighborhood had at least 30 sales of a given property type. Table AR5 presents the data for the 14 exceptions - 12 for condominiums and two for commercial. Within each property type, the neighborhoods are listed according to descending values of the COD. Of the12 condominium cases, only two have COD values below 10.0, and the range of values is 8.8 to 14.0. CODs for the two cases in which there were sufficient commercial sales yielded CODs of 17.2 and 25.6. Other Evidence on Assessment Quality The relatively good impression of assessment quality gained from examination of the official District ratio studies is at odds with some other views encountered on the District's real property tax assessment system. Washington Post stories from 1996 [as cited in O'Cleireacain 1997, 79] told of large fractions of residential properties being either over- or under-assessed by more than 10 percent, and of irregularities in the Real Property Tax Administration that led to the firing of the top two people there. Similarly, a citizen group reported, among other things in a generally negative report, that the "quality of 1997 real property assessments declined probably due to poor or incompetent management, failure to use the CAMA (computer assisted mass appraisal) system, and errors in property record cards" [Citizens for Fair Assessment 1997, 1]. The Citizens for Fair Assessment (CFA) study is based on the data used in the District's 1996 (tax year 1997) ratio study, supplied to CFA and reworked by that group to provide more detailed information than the official study presents. It presents the distribution of sales across 15 assessment-ratio ranges for each neighborhood and for the city as a whole for both single-family residences and condominiums, and somewhat less data for other types of property, limited by the small number of sales of those properties in many neighborhoods. A summary table pulls together some of the key statistics for the city as a whole [Citizens for Fair Assessment 1997, 25]; their summary data are reproduced here as Table AR6. The CFA data paint a rather bleak picture of assessment quality in the District, at least for tax year 1997. Because that year is not an atypical one among recent years based on data from the District's official ratio studies, the CFA data for 1997 probably are reasonably representative of other years, as well. For at least seven sales in each of the five property types, the assessed value was over 175 percent, and for at least two it was under 45 percent. In fact, the maximum and minimum A/S ratios for each type -- not shown in Table AR6 -- are well beyond the upper and lower bounds of the ranges shown in the table, as shown below [Citizens for Fair Assessment 1997, 25]. The highest ratios in each case are over 200 percent, and generally over 400 percent, while the minimums are below 15 percent for two property types. Thus, the maximum ratio for each type is more than 10 times as high as the minimum ratio, and for vacant land it is a rather amazing 35 times as great.
Keeping in mind that the District excludes the top and bottom 25 percent of sales, arrayed by the assessment-sales ratio, Table AR6 makes it possible to see how much variation in individual assessments is ignored. For residential properties, for example, the middle 50 percent of the sales fall within just 3 of the 15 ranges shown in the table, and only the middle one of those (95 percent to 105 percent) is included in its entirety. Thus, the middle 50 percent of sales cluster rather closely around the median ratio (98.9 percent). This is, of course, what the official COD tells us - 14.7 for tax year 1997 (Table AR2). While some outliers legitimately may be thrown out in conducting a ratio study, ignoring half of all arm's-length sales seems indefensible. The CFA data in Table AR6 make it clear that the ignored sales represent a great deal of disparity, or dispersion around the average assessment ratio. Similar tales are told by the data for the other four property types. The official ratio study statistics clearly overstate the degree of assessment uniformity - which, as noted earlier - was found to be the case when the inter-quartile approach used by the district was applied to Arizona counties [Bell and Bowman 1991a]. Diminished Uniformity over Time. The upward drift of CODs in recent years is troublesome. This occurred before the change in leadership within the Real Property Tax Administration in the fall of 1996. In talking with current officials, however, one possible explanation given was the previous tendency to apply uniform multipliers to assessed values as a means of raising the assessment level and, indeed, the median ratios generally have risen in recent years, as we have seen. This practice, it was suggested, magnified existing dispersion, and thus generated higher CODs. A truly uniform application of a single multiplier within an assessment area would not increase the measured dispersion of assessment within that area, all other things equal. An example of such an application would be to apply a multiplier of 1.25 to the assessed values of all properties in an area with an average assessment ratio of 80 percent, to bring the average to 100 percent (80*1.25 = 100). Comparison of Tables AR1 and AR7 demonstrates this using a simple example. Table AR1 was used earlier to show the calculation of several measures of central tendency (assessment level) and of the coefficient of dispersion (assessment uniformity). The example presented there was constructed to yield assessment ratios of 100 percent as measured by each of three measures of central tendency (the mean and median of the individual ratios, and the aggregate ratio relating the sum of assessed values to the sum of sales prices). Table AR7 applies a multiplier of 0.8 to each of the five assessed values in Table AR1. This reduced each assessment-sales ratio to 80 percent of its former level, and thus lowered the median ratio to 80 percent. These changes reduced the sum of the absolute deviations from the median ratio from 20 in Table AR1 to 16 in Table AR7 and dropped the average absolute deviation from 12 to 9.6 - i.e., to 80 percent of the previous level. With both the numerator and the denominator of the COD formula (average absolute deviation from the median ratio divided by the median ratio) reduced by 20 percent, the coefficient of dispersion was unchanged, at 12 (12/100 = 0.12, and 9.6/80 = 0.12). Of course, not everything else other than the assessed value was unchanged from one year to the next as multipliers were applied to raise assessment levels. Because another year had passed, different groups of sales were used in calculating the two years' assessment ratios. Also, changes in underlying market values continued. If existing assessments are essentially uniform, equal upward or downward percentage changes in assessed values can bring the average level of assessment into the target range, provided underlying values do not change. But assessments that are relatively uniform in one year tend to become less uniform the next year if there is no change in relative assessments, because different properties within an area - and in different geographic areas within a larger jurisdiction - tend to change in value at different rates. This is the reason for periodic reassessments. For citywide ratios, a further consideration is that assessments are done for many relatively small areas within the city, in part to permit more appropriate adjustments to assessed values reflecting different rates of change in market values in different areas. Thus, uniform multipliers within each assessment area generally will not translate into uniform multipliers citywide. Measured Level of Uniformity. The relatively low level of CODs reported in the official ratio studies is not very reassuring, given the manner in which the District calculates the COD. The use of just the inter-quartile range of assessment ratios already has been discussed at some length, and its tendency to understate assessment variability noted. The distribution of ratios for individual properties across 15 assessment-ratio ranges by Citizens for Fair Assessment confirms the suspicion of overstated uniformity by the official measures. Changes Set in Motion. While the quality of real property assessment within the District has not been very good in recent years, the problem has been recognized and acted upon. New leadership of the Real Property Tax Administration was installed a year ago and, as discussed in another section, many changes either have been adopted or are being planned - correction of errors in property records, adoption of more homogeneous assessment areas, addition of audits and other management techniques, etc. - all for the purpose of improving assessment accuracy. The foregoing discussion has highlighted some areas of concern with assessment equity, and particularly the measurement of it. Alternative recommendations flowing from that are presented in this section. Assessment-sales ratio studies are at the heart of this portion of the report. More fundamental aspects of the assessment system, such as triennial assessment, are dealt with in another section. Recommendation 1. The District of Columbia should adopt the more common form of the coefficient of dispersion (COD) based on all, or essentially all, of the sales in the sample, rather than the currently-used inter-quartile variant of the COD. The COD is a standard statistic - the most common, according to the International Association of Assessing Officers (IAAO) - for gauging the uniformity of assessments of similarly situated properties (horizontal equity). However, the inter-quartile variant ignores fully half the sales in the sample of sold properties - the 25 percent at each end of the array. Because so much of the available information is omitted, the current measure understates the degree of assessment nonuniformity. The differences can be quite extreme. The primary author of this study and his colleague have reported on the differences based on data from Arizona counties [Bell and Bowman 1991a]. Using the CODs generated under the standard approach as the baseline, the most extreme understatement of assessment nonuniformity - 85 percent or more - occurred in two counties with CODs very similar to those observed for the District when calculated under the inter-quartile variant that the District uses. The data for five counties are given in the tabular presentation below [Bell and Bowman 1991a, 349], to show that:
Changing to the standard calculation of the COD would provide a more accurate picture of the quality of real property assessment in the District. Recommendation 2. To preserve some comparability of data and allow time for users to adjust to the new COD measure, the current one should be continued alongside the new one during a transition period of three to five years. A new COD calculation procedure would produce statistics that are not comparable to those for prior years. While comparability of data over time is desirable, that concern is not sufficient to warrant retaining the old COD series and not switching to the standard measure. Therefore, the assessment-sales ratio studies published in the District of Columbia Register and otherwise intended for public consumption should calculate CODs under both the new and the old methods for the first three to five years in which the new method is used. This would provide a transition period in which people could get used to the new data series, and develop some understanding of how its numbers relate to the old ones to which they have been accustomed. The alternative would be to recalculate CODs for some earlier years using the new approach, thus creating an historic data series comparable to the new statistics that will be generated in the future. This alternative, however, would involve going back into the data for earlier years. The ability to go back reportedly is quite limited because the data for individual properties are not kept for more than a few years. Because computers are used to calculate the assessment-sales ratio studies, it should be feasible to provide the dual COD statistics for a few years for a very low cost increment Recommendation 3. The District's assessment-sales ratio studies should present the price-related differential (PRD), in addition to the coefficient of dispersion. Although somewhat less common than the COD, the price-related differential should be added to the District's ratio studies, at least for the citywide portion of the studies. The PRD - the mean of the individual ratios divided by the aggregate ratio - provides a summary statistic to indicate whether assessment levels differ across property-value ranges (vertical equity). A value greater than 1.0 indicates systematic bias against low-value properties, while a value under 1.0 indicates a bias against high-value properties. Thus, the PRD conveys information not available from the COD - information that is another indication of the quality of assessment. This statistic could be of use to both taxpayers and the Real Property Tax Administration, and could be generated easily. Information Used in Ratio Studies Recommendation 4. The Real Property Tax Administration should follow through on plans to devise an instrument to provide more information on real property sales to assist in the screening, or qualifying, of sales for inclusion in the assessment-sales ratio studies. Qualification of sales for use in assessment-ratio studies is necessary if the arm's-length criterion is to be applied. Application of that criterion is important because it provides assurance that the ratio studies reflect bona fide market transactions. Discussions with District property tax officials revealed the feeling that too much assessor time is spent in qualifying (also referred to as screening) sales, together with the suggestion that the screening may not result in discarding or adjusting all the sales that should be ignored or adjusted. For this reason, their plan is to develop a paper instrument for collecting pertinent information for all real estate sales that might affect the sales price - e.g., relationship between the buyer and the seller, conveyance of personal property with the real estate, special financing terms. This should be done as soon as possible, for it is important to the validity and reliability of the ratio results. There has been some suggestion that the inter-quartile approach has served, in part, to offset less-than-desirable screening of sales to qualify them for use in the ratio studies. While the ratios for the sales that should be screened out through proper qualification might be expected to be in the tails of the array of ratios, throwing out 50 percent of all sales in the sample - which the inter-quartile approach to the COD does - is an extremely crude adjustment for improper qualification of sales. It can result in ignoring many sales whose ratios are far from the average level, not because the sales were not arm's length, but because of problems in the assessment process, per se. Uncommonly high and low ratios resulting from such problems should be retained in the ratio studies, while non-arm's-length sales should not. Better screening of sales, therefore, is an integral part of making the ratio studies more informative and reliable. Recommendation 5. The Real Property Tax Administration should compare sales price in a given year with the assessed value of record at the time of sale, rather than to the assessed value developed in the year of - and often after - the sale. Assessment-sales ratios, based upon a sample of sold properties, are used to make inferences about the level of assessment of unsold properties. In other words, the ratios serve as an indicator of how accurately assessors are able to estimate market values. High, uniform assessment levels indicated by a properly-designed ratio study (representative sample of arm's-length transactions, assessed values determined in advance of the sales) increase confidence in the quality of assessments, not just for sold properties, but overall. While having the assessments determined after most sales for the year have occurred does not necessarily mean that the assessments have been influenced by known sales prices, the opportunity for this to occur tends to reduce confidence that the ratio studies provide a valid indication of assessor performance. There is no obvious reason for retaining the current practice of using assessed values determined after most of the year's sales have occurred, particularly since assessors calculate other assessment-sales ratios as appropriate to aid in the valuation task. The published ratio study's primary purpose is to inform the public. It could serve this purpose better if procedures were changed to relate sales prices to the assessed values of record at the time of the sale. Use of Assessment-Sales Ratios to Adjust Assessments Recommendation 6. In using ratio studies to adjust the values of individual properties, uniform percentage increments (multipliers) generally should not be applied unless the coefficient of dispersion - properly measured, using the standard approach discussed above - is quite low, say, less than 10. Assessment-sales ratios can be used to develop multipliers to adjust the average level of assessment for a group of properties. If the calculated assessment ratio is 90 percent and the target is 100, then - all else unchanged - application of a multiplier equal to 1.11 (100/90) would increase values enough to raise the average assessment level to 100. This approach, however, leaves any underlying nonuniformity of assessments within the group of properties to which it is applied, so it is not a substitute for reappraisal when there is significant nonuniformity. In using such multipliers, the property group(s) used to derive the multiplier and to which the multiplier is to be applied are important. They could range from very narrow to very broad - anything from, say, single-family housing in one part of the city built prior to 1950 to all types of real estate citywide. In general, the broader the group, the more suspect the use of the uniform multiplier approach. Properties of different types and in different areas tend to change in value at different rates, because of those underlying differences in their attributes. Because the use of uniform multipliers to adjust the overall assessment level does nothing to improve horizontal equity (i.e., to reduce assessment dispersion) in the face of significant nonuniformity, it is appropriate to undertake a more complete revaluation - one that considers specific attributes of individual properties in arriving at new assessed values. Ideally, this might involve a full reappraisal, but equalization of assessment levels also might be accomplished through application of several multipliers derived to reflect underlying property differences that seem to account for the differential market value changes that have occurred. While the threshold COD value that would trigger such revaluation could be set at any of many specific levels, setting it much above 10 would defeat the intent of this recommendation. 1. Michael Bell and Murray Johnston, 1997, Lowering Maryland's State Personal Income Taxes to Stimulate Business Development: Myths and Realities, The Abell Foundation, Baltimore, Maryland, Table 4 and discussion on pp. 15-6. 2. The numbers for 1993 are difficult to interpret. The District changed the end of the tax year from June 30 to September 30 in 1993. Thus, the data for Fiscal Year 1993 really represent five quarters of property tax levies and collections. If one were to take four-fifths of the total levy and collections for 1993, the total levy would be $743 million and total collections would be $711 million. Each would be a decline from their 1992 levels, reflecting the general decline in the real estate market that started in 1992. We use such "adjusted" data for 1993 when analyzing trends in property tax levies and collections. 3. The data in 2 are unaudited data from the District of Columbia Comprehensive Annual Financial Report, Year Ended September 30, 1996 (1995). Data on property tax levies from the CAFR are different from those reported by O'Cleireacain who developed independent estimates of property tax levies by multiplying the net assessed value of taxable property in each property class by the appropriate rate and then summing the levies for each class. Following this methodology, property tax levies fell by an annual average rate of 5 percent between 1993 and 1996 after increasing at an average annual rate of 10.5 percent between 1985 and 1992. See Carol O'Cleireacain, The Orphaned Capital: Adopting the Right Revenues for the District of Columbia, 1997, Washington D.C.: Brookings Institution Press, footnote 39, p. 64. 4. The adjusted value of real property tax levies for 1993 was $497.8 million. 5. The numbers discussed here represent net new delinquencies. That is, we are looking at the ratio of current collections as a percent of current property tax levies. 6. Carol O'Cleireacain, 1997, The Orphaned Capital: Adopting the Right Revenues for the District of Columbia, Washington D.C.: The Brookings Institution Press, p. 57. 7. We have adjusted 1993 data to reflect that fact that as reported in the CAFR it consists of five quarters of data. Taking 80 percent of these figures indicates that the adjusted property taxes per capita would be $1,285.50 and the adjusted property taxes per $1,000 personal income would be $42.18. Thus, property taxes per capita reached a peak of $1,401.36 in 1992 and property taxes per $1,000 personal income reached a peak of $48.01 in 1991. 8. Sonstelie [1978, 235] quoting Leland's 1928 volume, The Classified Property Tax in the United States. 9. Taxing big business - typically corporations - creates a great deal of illusion and uncertainty as to who bears the tax. It seems a common sentiment that corporations ought to be taxed rather heavily because "they can afford it." Somewhat paradoxically, many people also seem to believe that businesses pass their taxes on to their customers. The ultimate burden of a tax on a corporation is difficult to determine, as it varies with the circumstances. Even if no tax shifting occurs, however, and the tax rests with the owners of the corporation on which it was imposed, this may not produce vertical equity because not all stockholders are well-to-do, and a large percentage of stocks now are owned by pension funds. An additional equity concern with corporations is presented when one looks beyond the property tax and notes the double taxation of corporate dividends under income taxes. 10. This would be true no matter what that uniform assessment level was - 100 percent, 200 percent, or 10 percent. 11. Quoted from p. 2 of an advance copy of the 1997 ratio study; emphasis added. The same approach is indicated in studies from prior years. 12. A very good explanation of the tax year is provided by O'Cleireacain [1997, 70-71]. 13. District tax regulation 307.3(b) specifically allows this: "Sales comparisons should be made by property type within an assessment area; Provided, that if sufficient sales data for an assessment area is not available, sales data from other similar areas may be used." 14. More will be said below, but the manner in which the District calculates CODs -- using only the middle 50 percent of sales from the array of sold properties -- should be kept in mind. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Send mail with questions or comments to webmaster@dcwatch.com
Web site copyright ©DCWatch (ISSN 1546-4296)
