Survey of ratio study methods used by the states
Ratio studies are statistical analyses of property values and assessment data. They are used by nearly all the states to develop numerical estimates of the percentage of current market value at which locally assessed taxable property is assessed and to provide measures of assessment performance. Ratio studies are needed because, despite the fact that nearly all the states have clear statutory standards governing the assessment process (e.g., assessments must be at "fair cash value," at "100 percent of market value," at "50 percent of market value," etc.), the assessments of some local governments may in actual practice fail to meet these standards. If the common standard is met by some local governments and not by others, the equity of programs that use property assessments as a basis (e.g., state aid distribution, apportionment of certain taxes) may be compromised. Further, failure of an assessing jurisdiction to meet the statutory standard may indicate that local taxes are not being administered equitably. The central purpose of a ratio study is thus to monitor local assessing practices, and its results can be used to make appropriate adjustments to the total value of an assessing unit in which such de facto deviations from the standard are detected.1
This report presents information gathered in a survey of ratio study practices used in the 43 states for which data were available. It examines the practices of all states except the following seven: Alaska, California, Delaware, Hawaii, Indiana, Maryland, and Montana. Reasons for excluding these states from the survey are as follows. Ratio studies are not conducted in Delaware, where the state government plays only a minor role in property tax administration. In Maryland, the assessing function is carried out by the state and there is thus no need to verify that assessments in different municipalities are determined on the same basis. The situation is somewhat similar in Montana, which has recently suspended ratio studies, in that its county assessors are essentially state employees administering local assessments. In Hawaii, counties conduct their own ratio studies in lieu of a single state study, and a similar arrangement exists among Alaska's appraisal districts. Indiana has only recently begun to require assessment based on market value, and a ratio study will not be completed until 1997 at the earliest. Development of a methodology to accomplish the study is currently in progress. In California, assessments are based on the value of the property at the time the current owner acquired it, rather than on current market value or a percentage thereof. This practice, instituted with approval of Proposition 13 in 1978, makes California's tax administration system sufficiently different as to preclude meaningful comparisons with New York.
Standards for conducting ratio studies are promulgated by the International Association of Assessing Officers (IAAO), the major professional society in the field of property tax assessments.2 These standards are very comprehensive and form an excellent guide for states in designing studies that satisfy contemporary methodological criteria. According to a survey completed by the Idaho State Tax Commission, officials in approximately three-quarters of the states cited familiarity with IAAO standards in 1992.3
To place the ratio study methods employed by the states in perspective, it is necessary to review their overall approaches to property tax administration. In addition to the valuation standard already mentioned, other important characteristics of the administrative system which can influence the design of the ratio study are the level of government responsible for the assessment function and requirements for periodic updating of assessments.
In 30 out of the 43 states studied, the assessing function is performed at the county level. This level of government, encompassing a greater area than the municipal level, holds the potential for greater efficiency in assessing. There is a broad enough tax base for financing a staff with technical expertise and the larger area of the assessing unit usually insures that there are sufficient sales for monitoring market values. Yet the territory covered is usually contained enough so that the assessment staff can be familiar with local markets and the characteristics of properties, enabling them to estimate values accurately. In those states with county-level assessing, the number of assessing units ranges from a low of 15 in Arizona to a high of 254 in Texas.
In contrast to those that assess at the county level, 13 of the 43 states, primarily those located in the northeast and north-central areas (where the tradition of sub-county local government is strong), assess mainly at the town(ship) and city levels. New York State is a member of this group. The number of assessing units in these states ranges from 39 in Rhode Island to approximately 2,700 in Minnesota. Sub-county assessing in New York State is carried out by 920 towns, 61 cities, and 259 villages, with only two counties performing the assessing function.
For a state agency conducting a ratio study, county-level assessing has decided advantages. The first and most obvious is that there are fewer assessing jurisdictions that must be studied. This can reduce the cost of the ratio study substantially. Another advantage is data availability: just as larger geographic areas provide more market data for use in assessing, they also provide larger pools of data for ratio studies. About half the states studied require that assessments be updated at given time intervals. Update cycles range from two to ten years, with cycles of three to five years the most common. Ratio study results can be used to determine if the cycle is appropriate, and the consistently updated assessments reduce the need for large state adjustments to equalize them.
Having a clear, unambiguous statutory standard according to which local assessments must be determined is also an asset to state agencies charged with conducting ratio studies. Clarity with respect to how assessed value should be determined is a basic prerequisite to evaluating the accuracy of the determination. In all but three states, there is a statutory level of assessing which must be attained by all assessing jurisdictions. For nearly half of the states with statutory levels, the required assessment level is set at 100 percent of market value. In the remaining states, the statutory level is a given percentage of market value, with some having classified assessment systems in which the level of assessment varies by property type.
New York, New Jersey, and Pennsylvania are the only states studied that do not require all assessing jurisdictions to use the same predetermined level of assessment. New York's current mandate requires only that all parcels be assessed at a uniform percentage of full value, but the percentage used need not be declared by a local assessor and it can vary among assessing units. Since the level of assessment need not be declared, it can not be verified. Thus, the focus of New York's ratio study is to estimate the level of assessment in the first instance, adjusting assessment levels that are as low as one percent of market value up to a full market value basis. In contrast, the studies conducted in virtually all of the other states focus on verifying that the statutory level of assessment is adhered to, and making comparatively minor adjustments when this is not the case.
Detailed information for each state, gathered from interviews with state officials in the states that were studied, is presented later in this report. This large quantity of data is summarized in the following sections, with comparisons made to New York's system where appropriate. Table 1, which follows the summary, presents a brief description of each state in terms of the major characteristics of ratio studies and equalization practices. Finally, at the end of this summary, we offer a number of findings and suggestions based on review of other states' practices and knowledge of New York's current system.
The primary goals of the ratio studies conducted by the states are: (a) evaluation of local assessment practices; and (b) making any aggregate value adjustments required to place the assessments of all municipalities on the same value basis. States typically monitor the level and uniformity of assessments in each assessing jurisdiction because, having ceded the constitutional authority to levy property taxes to local governments, they thus have an interest in insuring that the taxes are levied equitably. The common value basis is generally needed for two major purposes that may be grouped under the heading "equalization": (a) distribution of intergovernmental aid on the basis of property wealth; and (b) equitable apportionment of taxes within taxing jurisdictions that encompass more than one assessing jurisdiction.
Secondary purposes for which ratio study results are used vary considerably, and may include: (a) verification of property value to insure that state limits on the taxing or borrowing powers of local governments are adhered to; (b) disclosure of essential assessment information to taxpayers; (c) administration of property tax exemptions and certain assessments; and (d) review of certain property classes such as railroads, utilities, or recently sold properties to insure that discriminatory assessment is not being practiced.
These primary and secondary purposes found among the states are quite similar to the purposes of New York's ratio study efforts. Along with equalization of local assessments for state aid distribution, apportionment of county and school taxes, and monitoring of assessment uniformity, the results of New York's appraisal-based study are also used for estimation of constitutional tax and debt limits, administration of certain exemptions, evidence in assessment appeals, among other purposes. An entirely separate sales-based study is conducted in New York primarily to establish the level of assessment of residential properties for the purpose of assisting taxpayers in assessment appeals. For the past three years, it has also been used to determine the level of assessment uniformity for the purpose of distributing special state aid payments to those municipalities with exemplary assessment practices.
Virtually all the states conduct a single ratio study, with a given periodicity, although in some instances two or more state agencies may use the ratio study data for several different purposes. In four states, Georgia, Massachusetts, Minnesota, and Iowa, the study is modified either in terms of its periodicity or its coverage of property types to accommodate more than one use, e.g., state oversight of statutory compliance versus equalization of assessments for state aid distribution. New York is unique in that it conducts two separate studies: one based on 60,000 to 70,000 appraisals and another based on 80,000 to 100,000 usable sales.
Three-quarters of the states studied conduct annual ratio studies of assessment rolls. Two states conduct quarterly studies, four biennially, and one triennially, with Tennessee on a variable cycle. New York's situation is more difficult to categorize, since two ratio studies of differing frequency are conducted. The appraisal-based study, used for equalization and other key purposes, was triennial for many years but has been biennial in more recent times, with inter-study trending used in years between actual studies. The sales based study is conducted annually.
In some states, local assessing units, especially in metropolitan areas, may conduct their own ratio studies to monitor equity levels and to determine the need for a reassessment. In other instances, studies are undertaken at the county level for the purpose of equalizing assessed values among sub-county assessing units. Since no detailed information is available on these local ratio studies, they are not covered in this report.
The data used in ratio studies consist of the estimated market values of a group of parcels and the assessments applied to these same parcels. Estimated market values may be obtained directly from the marketplace in the case of parcels that have recently been sold, or through appraisals of selected parcels. The assessed values of the parcels are simply obtained from the relevant assessment rolls. Dividing the assessed value of a parcel by its estimated market value yields its assessment ratio. Statistical calculations are then carried out on the ratios to determine their central tendency, their variation, etc. The rolls used by the states as the basis for ratio studies are nearly always the most recent ones available; this is in marked contrast to New York's practice of taking the assessed values from "base year rolls" that may be up to five years older that the market value estimates. The majority of states studied (26) use a combination of sales and appraisals to determine market value. A smaller but significant number (14) use only sales, with Missouri and Nevada using only appraisals. Again, it is somewhat difficult to categorize the New York program, since both an appraisal-based study and a sales-based study are conducted. However, New York, like Missouri and Nevada, uses exclusively appraisals in its equalization program.4
If an attempt is made to rely solely on sales, there are likely to be situations where it is difficult to represent all property classes in proportion to the share of the roll they comprise. For some classes, such as industrial and utility property, there may be few or no sales available. Where municipal-level assessing prevails, there may even be cases where insufficient sales are available to represent the entire roll reliably, let alone particular classes of property. This is the main reason that the majority of states have opted to include appraisals in their ratio studies: to supplement the sales data in those classes and geographic areas where insufficient sales are available. Because this approach provides a better representation of the entire roll than would sales alone, the ratio study becomes more statistically reliable.
The main difficulty states encounter in using appraisals is that they are labor intensive and therefore expensive. To reduce the number of appraisals needed, efforts are sometimes made to augment small samples of sales data by extending the time period from which sales are drawn in those areas and/or property classes where the volume of sales is low. A few states attempt to solve some problems of this nature by combining property classes. These techniques are generally found to be less than satisfactory, however, as evidenced by the fact that the great majority are supplementing sales with appraisals. It is not unusual to hear state officials say that they would like to include even more appraisals than they are currently using if the resources were available to do so.
In addition to the issue of resources, there are other areas of concern in deciding whether to use sales or appraisals in estimating market value. In virtually all areas except resource requirements, sales generally pose greater problems. A key requirement in the use of sales data is identification of the usable sales through elimination of those transactions that occur under conditions which render the price paid a misleading indicator of market value. Since major physical changes can be made to a property between the time the assessment roll was prepared and the date of sale, it is also important that transactions involving recent construction, demolition, or subdivision also be eliminated. A related issue is the fact that sales occur over a period of time in which market conditions may be changing whereas assessments are generally set as of a certain date. Although this issue is often raised, the most frequently mentioned solution of adjusting the sales prices to account for the passage of time is rarely applied in actual practice.
In about half of the states studied, the local assessing jurisdiction does the primary data preparation and verification of usability. Many states have informal discussions with local assessors to determine usability of sales. Generally, assessors are not required to offer proof in support of any changes or deletions they make to the data.
State agencies have varying degrees of authority to check the locally-screened sales data and verify that the procedures used by local governments are valid. In about one-third of the states studied, the state agency has the final authority to determine which sales are usable and it also has the ultimate responsibility for screening. In a few cases, there is no state oversight authority at all, with local assessing jurisdictions having the final responsibility for data clean-up and preparation. A widespread practice among the states is the automatic exclusion of extreme sales ratios that have survived editing procedures based on the strong likelihood that they are data errors or highly irregular transactions (generally called "trimming the distribution;" see the Kansas, Pennsylvania, and Virginia profiles for examples of trimming procedures).
It is possible that sales in some communities may be non-random, e.g., where most of the available sales are from a single new subdivision or a certain major property class is over-represented or under-represented. In cases such as this, it may be necessary to eliminate some sales or to weight them in proportion to the incidence of like properties on the roll. For residential property in relatively urbanized areas, issues of non-randomness and lack of transactions generally do not arise or can easily be overcome with suitable weighting of the data. Typically, states tend to either delete sales of over-represented property types and/or supplement the data with appraisals in under-represented property types rather than making attempts to weight the sales appropriately. Overall, the efforts other states make to insure balanced representation of the entire roll fall considerably short of New York's carefully designed stratified random sample approach.
A key consideration in obtaining evidence of market value, whether from sales ratios or appraisal ratios, is the reporting of market transactions. Most states do have a law for the mandatory disclosure and recording of sales information, and New York State is among them. States not having this level of disclosure must often rely on local assessors to supply the sales data (e.g., Tennessee). In addition to using the information supplied on disclosure forms, many states that rely heavily on sales use follow-up questionnaires or phone calls to one or more of the parties involved in the sale for the purpose of clarifying sale conditions or investigating certain transactions further. This practice, which is heavily used elsewhere for commercial and industrial properties, is rarely used in New York for any property type.
States without disclosure laws cite their absence as a major problem in data collection. A few of these states are forced to rely on data collected by the real estate sales industry, such as multiple listing service publications. In some states, both the sales price and the assessment information are provided by the buyer or seller, as is the case in New York. This practice, which can produce erroneous assessment information, is the exception rather than the rule, however, for it is more typically the responsibility of the assessment office to add the proper assessment information to the disclosure form.
In the majority of states, one year of valid sales information is utilized in a ratio study. A few use other time periods ranging from a minimum of 6 months to a maximum of 48 months in Kansas (where the period will be expanded to this maximum, in one-year increments, until there are enough sales). Periods of less than a year are generally used to reduce the number of sales in populous areas or where sales activity is vigorous. At the other extreme, the extended time periods of more than a year are used for increasing sample size where sales are few, mainly in rural areas.
Many states specify a minimum sample size for a valid ratio study, but there is wide variation regarding what constitutes an appropriate minimum. The typical state sets a minimum sample size by class per assessing unit. Some states make it a percent of the class; others give a minimum number below which the statistics are not considered reliable, with certain rural states having sample sizes as low as five per county. In some states and in classes with a heavy concentration of sales, especially the residential class, only a sample of the usable sales may be used in the ratio study. The main problems concerning sample size in ratio studies which rely only on sales are that the samples may be too small or non-representative of the entire assessment roll.
The states that supplement sales with appraisals usually use a predetermined sample size to decide when appraisals will be used. An example of a state which tries to establish an unbiased sample is Arkansas. In this state, a random sample of sales is selected based on a scatter diagram of all the parcels on the roll, and appraisals are then performed for the under-represented classes. In another state, Georgia, some appraisals are done in all classes, even if there is a large number of sales available, to test for assessment bias by neighborhood or discriminatory assessing of recently sold parcels. In at least one state, a random sample is drawn from the available pool of usable sales. However, while this produces an unbiased sample of sold parcels, it does not necessarily increase the randomness of the data, which is really the desired result. A few states give special treatment to parcels that comprise a large portion of the roll. Some remove them from the sample to prevent undue influence upon the results of the study, but at least one state appraises unsold parcels of this type and includes the appraisals in the study.
In contrast to sales, appraisals allow for greater control of the data, since all data elements are carefully prepared and are generally subject to more than one level of review. Issues such as representativeness, randomness, and data accuracy are more directly under the analyst's control, and appropriate sample size can be directly addressed. On the other hand, difficulty may be experienced in completing a large appraisal workload in a timely manner, or obtaining a budget sufficient to meet appraisal costs. Some may also argue that appraisals are less "objective" since they are opinions of value rather that direct evidence taken from the marketplace. The fact that so many states appear to be successfully using sales data and supplementing it where necessary with appraisals, together with IAAO's endorsement of using both types of data, tend to support the validity of this most common approach.
Ratio estimation methodology
As stated earlier, ratio studies are primarily conducted to determine the extent to which local governments are upholding statutory requirements governing the level and uniformity of assessments and to make appropriate adjustments where they are not. The most important step in the verification process is the calculation of a measure of central tendency from the individual parcel ratios (of assessed value to either sale price or appraised value) that are available. Several alternative statistics can be used to measure central tendency. They would have identical values for a perfect normal distribution of parcel ratios, but they normally diverge in practice since the distribution seldom meets this strict criterion.
The simple mean is a measure familiar to most people who have not studied statistics. It is calculated by summing the ratios of the individual parcels and dividing the resulting sum by the number of ratios. The simple mean is influenced significantly by ratios that are extreme on either end of the distribution. A different type of mean, the weighted mean, is calculated by summing the assessed values, then summing the market values, and finally dividing the first sum by the second. An important feature of the weighted mean is that it is even more heavily influenced by high-value parcels than low-value parcels as compared to the simple mean. The median is calculated by arranging the ratios in ascending or descending order and selecting the middle ratio (or, for an even number of ratios, the average of the two middle ratios) from the array. Very high or very low parcel ratios have no greater influence on the median than do more typical ratios.
Almost two-thirds of the states studied use the median as their principal statistic for testing whether the level of assessment meets the statutory requirement and for equalizing property values when the standard is not met. Generally, states deem that the standard is met if the median falls within a certain, predetermined range around the standard rather than being exactly equal to the standard. The primary reason cited by state officials for relying on the median is that it is the statistic that is least distorted by extreme "outlier" ratios, i.e., it is a reliable indicator of central tendency even when the ratios are not normally distributed.
Only two of the states surveyed that rely on the simple (or unweighted) mean as the primary measure for equalization purposes. A few states calculate it in addition to other statistics, but do not use it for equalization purposes. This is in keeping with the fact that it is not recommended by IAAO for any major purpose associated with ratio studies.5
In contrast, the weighted mean is used in about one-third of the states. The primary advantage of the weighted mean is that, since it is influenced by a given parcel in proportion to the parcel's value, it facilitates an unbiased estimate of the total market value in an assessing unit even in cases where certain value ranges or property types are assessed (illegally) at ratios significantly above or below the median (i.e., the ratios are correlated with parcel value). This feature is recognized by the IAAO in its Standard on Ratio Studies.6 However, for the weighted mean to function effectively in this capacity, it is essential that the sample of ratios accurately represent the entire roll and any subsections of the roll deemed to be significant value components. Representativeness can be accomplished by a combination of techniques such as weighting individual transactions in relation to the incidence of like properties on the roll and supplementing the sales data with appraisals. If representativeness is not assured, or if the objective is not to estimate the market value of the assessing unit but, for example, the typical level of assessment for purposes of assessment appeals, the median is preferable to the weighted mean.
Based on the above considerations, the basic conceptual design of New York's system for determining equalization rates appears to be in keeping with the best standards of professional practice. Full value is determined through use of a weighted mean. This statistic is calculated from a random sample that is stratified according to property use and value in order to assure representativeness. And the carefully prepared appraisals virtually preclude the possibility of "outliers" that result from erroneous data. The fact that most states use a median instead of weighted mean appears to reflect the following considerations: (a) they are more concerned with verifying compliance with the statutory standard of assessment than with estimating total market value; (b) the problem of discriminatory assessing of certain property classes or value levels may be less prevalent elsewhere than it is in New York; and (c) other states generally make considerably less effort to insure that the entire assessment roll is fully represented in the ratio study data than does New York. Under these circumstances, the median's stability with unbalanced data probably overshadows the weighted mean's ability to estimate full value accurately even under discriminatory assessing.
A few states calculate a confidence interval around the estimated level of assessment to decide whether an assessing jurisdiction's level of performance falls within a reasonable range of the statutory ratio requirement. This test is sometimes based on the assumption that there is a normal distribution in the ratio data, with confidence level testing thus appropriate in instances where weighted mean ratios are used and where sample sizes are 30 or more. However, there are other states that rely on confidence level tests that do not require normally distributed ratios, e.g., the binomial test. A new technique (known as "bootstrapping") has recently been developed in one state (Kansas) to perform confidence testing in median-based assessment performance tests, and using small sample sizes.7 The technique uses the full original sample to generate many sub-samples, through an iterative process known as Monte Carlo simulation. This method is being applied in an attempt to reduce the need to extend time periods for obtaining the requisite number of usable sales, and also to substitute for supplemental appraisals, especially in sparsely populated assessing units.
In addition to measures of central tendency, most ratio studies typically calculate measures of dispersion. The purpose of these measures, also referred to as measures of assessment uniformity, is to determine how much variability (or non-uniformity) there is among assessment ratios. Ideally, there should be none, but it is well recognized that zero variability would be impossible to achieve in actual practice. All of the states studied calculate at least one measure of assessment uniformity. The statistic used most frequently is the same one used in New York -- the coefficient of dispersion (COD) -- because it does not require the assumption that the assessment ratios are normally distributed. Another test of uniformity used by a few states is the coefficient of variation (COV), which does require the assumption of normality. One state uses a relatively unusual statistic called the coefficient of concentration, which measures the extent to which ratios fall within a predetermined range around the median ratio.
About half of the states confine their use of uniformity measures to internal and advisory purposes. Another half apply these measures to actually enforce statutory requirements. The COD standards used are commonly based on IAAO guidelines, with some states having standards that are considerably looser. The reluctance to tighten thresholds of compliance to IAAO standards frequently stems from a concern that there are too few usable sales available for analysis in sparsely settled rural assessing units and that insufficient resources are available for conducting supplemental appraisals.
Nearly half the states surveyed also attempt to measure vertical bias is assessments, i.e., a tendency to assess properties at different ratios based on their value. The principal statistic used to test for this possibility is the Price Related Differential (PRD), also known as the Index of Regressivity. States generally use this statistic for advisory and internal purposes rather than enforcement of statutory requirements, as is the case in New York.
In the development of many states' ratio studies, parcels are stratified into property classes. Classification may reflect statutes that allow assessment of different classes at different percentages of value, or an effort to examine subsections of the roll. Property classes are sometimes further stratified into sub-classes, such as rural-residential or multi -family residential of a given number of units. The most common individual classes distinguished in the studies are residential and commercial/industrial. The agricultural class is often omitted from the studies, since it is assessed in many states based on use value rather than market value. Property classes whose assessed values are set by the state, especially railroads and utility property, are also often omitted from ratio studies since the purpose of the studies is to estimate the level of assessment of locally assessed property (the level for state-assessed property being already known).
A few of the states that calculate ratios by class but do not allow classified assessment use tests of statistical significance such as the binomial, Mann-Whitney or Kolmogorov-Smirnov, to determine if significant differences appear between the ratios found in different property classes, a finding that would be an indication of discriminatory assessing. These tests are particularly useful in that they can be used when ratios are not normally distributed. The binomial and chi-square tests are also used in at least one state, Idaho, for determining whether the sample may be assumed to be from a normal population.
As mentioned earlier, virtually all states have a clear, unambiguous valuation standard and approximately half have some statutory requirement for reassessment at a fixed interval. In addition to these standard regulations, a variety of actions may be taken by the state government when an assessing jurisdiction is found to be out of compliance with state requirements, based upon the results of a ratio study.
If the statistics produced from the study are not satisfactory, many states order a reassessment to be done earlier than would occur under the normal cycle. Occasionally, the state performs the needed reassessment and bills the costs to the assessing unit. Some states refuse to certify the assessment roll or prevent assessment notices from being mailed to property owners until the roll is corrected. In other instances, state financial aid is withheld or reduced. A few states remove the assessor from office, require training, and/or withhold wages until the roll is corrected. About one-third of the states use the ratio study results to require an adjustment of the aggregate assessing unit value when the actual assessment level is found to be significantly different from the mandatory level. Another fifth require an immediate change in individual parcel values when the statistic indicates that the municipality is out of compliance.
A small number of states take little or no action as a result of the ratio study findings. Some only advise the assessors of the findings, or others offer financial or technical assistance incentives to reassess. In New Mexico, for example, the state does not have authority to force assessing units to comply with statutory standards. However, tax consultants are now beginning to use the ratio study results as evidence in court challenges to assessments. In Michigan, state equalization factors cannot be fully implemented, because of legislative restrictions on the amount of equalization adjustment that can be made as a result of any single ratio study.
New York's program of segment equalization rates -- which produces separate estimates of the level of assessment in school district portions of certain assessing units -- appears to be unique. In most of the other states studied, a finding of differential assessing within the same assessing unit would result in enforcement of statutory standards. Since New York takes no remedial action to correct the assessment anomalies that give rise to different levels of assessment in segments within the same assessing unit, its program appears to actually facilitate the continuance of assessing practices that have been demonstrated to be contrary to statute.
Several state officials said that they believe the equalization and state oversight process is inadequate in their states, either because of insufficient staff to verify sales data and to augment sales with appraisals, or because the existing state mandates governing the assessment process are not strong enough or sufficiently enforceable to protect taxpayers.
Assessing jurisdictions in most states have recourse to an appeal process if they dispute their state's ratio findings. Many initial disagreements are involved with determining which sales are not arm's-length, and should therefore be excluded from the study. These disagreements are often settled at an informal hearing at the state agency, before the ratio studies are completed. Formal appeals in most states occur through an administrative hearing at the same agency. Further appeal can be continued in the state court system. This typical appeals process is essentially the same one that is used in New York.
In some instances, where ratios are computed for state aid purposes, school districts and other taxing districts receiving aid may appeal ratio results. In North Carolina, railroads and utilities are given a reduction in their tax levies in the counties whose ratios are more than a pre-determined percentage below market value. The utility companies in that state have hired consultants to do their own ratio studies in order to monitor the state's findings.
In most states officials are of the opinion that the appeal process works well. Appeal rates are typically below ten percent, and state decisions and practices are generally upheld by the courts in states where ratio standards are enforced. In about a quarter of the states studied, however, appeal mechanisms exist but are not often used due to weak enforcement of state assessment mandates.
Information concerning the cost of conducting ratio studies was not always available and was sometimes too incomplete to allow for interstate comparisons or firm conclusions. Among the problems encountered in comparing costs were: (1) in many instances, employees working on ratio studies are also responsible for other tasks, making the cost of the study difficult to isolate; (2) in most cases, state officials could only cite the direct costs of the program, and were unable to determine indirect costs or overhead such as rent, computer costs, legal fees, etc. that were frequently shared with functions other than the ratio study, (3) the tasks and budgets in some states are divided among more than one agency; and (4) in states with low budgets, some of the tasks, such as editing of sales data, are performed by the assessing units.
The available data on total annual program costs (25 states) show wide variability, ranging from $30,000 in North Dakota to New York's approximately $18.5 million in the 1994-95 fiscal year and a projected $15 million in the 1995-96 fiscal year. These figures represent both direct costs and indirect costs. The next largest expenditure to New York's was found in Florida, which spends $8.3 million annually (includes the cost of certain other functions besides the ratio study). The third most expensive program is found in Wisconsin, where $4.8 million is expended annually in direct plus indirect costs.
Obviously, states differ markedly in size, so it is necessary to adjust these figures to make the comparisons more meaningful. When costs are divided by state population, the following pattern emerges. The range (from $0.05 per capita to $1.03 per capita) is still quite wide, but only a fraction as wide as the total dollar range cited above. New York and Wisconsin led the nation, at $1.03 and $0.98 per capita, more than 50 percent higher than third-place Florida ($0.64 per capita). The great majority of states for which any type of cost data were available cited figures of less than $0.40 per capita.
It is noteworthy that two of the states with the highest costs, New York and Wisconsin, are among the minority of states having municipal-level assessment administration. Wisconsin has an even greater level of local government fragmentation than New York, with over 1,800 assessing units. Undoubtedly, this factor contributes to the cost of the ratio study, as it does in New York. Florida, with both county level assessment and relatively high costs, is the exception among the high cost states. Were its full indirect costs known, they could potentially place Florida considerably closer to New York and Wisconsin in terms of total expenditure per capita. However, since the figure given, based on the budget for the Department of Revenue's Property Tax Assessment Program Section, also includes expenditures for such things as aerial photography and administration of tax refunds -- functions that would not normally be associated with a ratio study -- it is difficult to compare the Florida figures to those of New York and Wisconsin.
One item of particular interest was appraisal costs. Most states were unable to provide figures on the average cost of an appraisal. Officials from some states said they would like to do more appraisals to augment sales data but are unable to do so because of insufficient funding. Missouri, one of the few states that relies exclusively on appraisals, cited appraisal costs of $55 per parcel. South Carolina and Maine, states that use a mixture of sales and appraisals, cited costs of $25 per parcel and $40 to $60 per parcel, respectively. However, an official from Kansas, a state that also supplements its sales with appraisals, cited a cost of $200 per parcel for developing appraisals that complied with highest standards of professional appraisal practice as promulgated by the Appraisal Foundation. Although these figures are inadequate to support any firm conclusions, they demonstrate the potential range in costs and perhaps also the fact that an appraisal that may be considered adequate in one state may be considered lacking in another.
Based on the review of practices found elsewhere in the nation in relation to New York's system, several significant points have emerged and they are summarized below. The intent of these ideas and suggestions is to indicate those areas where New York's system for equalization of local assessments and related functions might benefit from techniques and approaches developed in the other states. Also pointed out are key features of New York's system which appear to have particular merit and thus should be retained.
All the suggestions made could be implemented under the current structure of local government and assessment administration in New York; none is dependent on county-level assessing, full-value assessing or other such arrangements that are common elsewhere. Some would require statutory amendments but others could be accomplished through changing rules or procedures. Overall, the ideas presented are directed toward the objectives of maintaining statistical reliability, reducing resource requirements, streamlining operations, avoiding duplication, and making better use of existing data.
Conduct a single ratio study
The current practice of conducting two separate ratio studies, based on two entirely separate data sets -- a sales-based study and an appraisal-based study -- is duplicative and confusing. It results in two different estimates of the overall residential ratio in a given community, with the technical and conceptual differences between the estimates virtually impossible to explain to the average local government official or citizen. Only one study should be necessary to determine the residential ratio, and the same one should establish the ratios for the other major property classes and the overall ratio for the assessing unit. The two existing studies should therefore be combined, with the result that sales would be used extensively in the equalization study (see No. 3, below).
The overwhelming majority of states have succeeded in conducting an annual study, using very recent assessment rolls as a basis, at lower costs than are incurred in New York for a biennial study using outdated rolls. However, few face the level of complexity in New York's local government structure and not all of them make as much effort as New York to insure that the ratio data are representative of all property types. An annual study, based on the most recent assessment rolls, would be very desirable in New York, and it would insure currency of ratio estimates and market values. It would also avoid the volatility that can result from introducing multi-year ratio changes in a single year.
Despite these advantages, data quality and representativeness should not be reduced to achieve an annual study. It is critical that the study statistics be accurate and reliable, and these objectives should not be sacrificed to achieve annual periodicity.
At present, sales ratios are not used directly in determining overall ratios for equalization of local assessments. Given the cost of appraisals, it is difficult to defend this practice in the case of residential property in urbanized areas. In these areas, sales are sufficiently numerous to insure adequate representation of the property type in question without any appreciable sacrifice of randomness. Some sales of non-residential property might also be utilized, as might sales of residential property in more rural assessing units. However, adequate representation of all major property types in these instances will require that sales be supplemented with appraisals.
It is very important that all the significant property classes on an assessment roll be represented in the data in order to insure that estimates of the overall ratio and total market value are not biased and volatile from year to year. Without full representation, serious bias could occur in assessing units where certain property classes are over-assessed or under-assessed, and annual fluctuations could be excessive. Representativeness can be assured by random selection in the case of appraisal parcels. Sales can be classified according to major characteristics such as property use and assessed value, and selected or weighted according to the incidence on the roll of parcels with like characteristics.
Increased use of sales data can only be accomplished if the quality of the data is improved. At present, erroneous data is tolerated because the errors usually do not materially affect the median-based measure of residential ratio for which the sales are used and there is no impact on major programs such as state education aid and tax apportionment. Use of sales ratios in weighted-mean calculations (see No. 5, below) that affect these key state and local programs requires a high standard of data accuracy and reliability.
Better screening of sales is needed at both the state and local levels. Buyers or sellers may need to be contacted in some cases, particularly those involving commercial properties, to insure correctness of the data.
The weighted mean is the best measure of central tendency for purposes of estimating the total market value of taxable real property in an assessing unit or in a property class within an assessing unit. Its characteristics are especially important where there is directional bias in assessing, i.e., where high-value properties tend to be over-assessed or under-assessed relative to low-value properties. Median ratios should also be computed for the major property classes and for the entire roll for purposes of verifying inter-class uniformity and for use in assessment appeals.
The COD is the most widely used statistic for this purpose and is endorsed by the IAAO.
The predominant use of ratio studies in the nation is verification that local levels of assessment meet statutory standards. Despite New York's lack of a statewide assessment level requirement, a large number of adjoining assessing units have voluntarily adopted a standard (usually 100%). Where such a common standard exists within a school district or county taxing unit, and where values are both reasonably current and reflect the same valuation year, the municipalities in question should be certified for apportionment of school and county taxes based directly on the assessments.
A consistent statewide value standard is used in determining property wealth for distributing state education aid. When local assessments are based on a market level and period consistent with this standard, and can be verified and certified as such by the state, the total local assessed value could be used as a measure of full value for aid purposes.
Objective standards should be developed by the state for use in verifying and certifying local assessment levels. In the initial year of a revaluation, a procedure audit should be sufficient for verification purposes. In subsequent years, objective statistical tests of the level of assessment (such as the binomial test), as recommended by the IAAO, should be performed as part of the certification process.
This practice is now required by law for some localities, including certain school districts, communities using homestead/non-homestead classified tax systems, and municipalities within county assessing units. It may also be done at the state's option in cases where differential assessment is demonstrated in two or more school district portions of a given assessing unit. Significant state resources are expended annually in the process. At present, a finding of differential assessment levels within an assessing unit usually does not lead to correction of the underlying problem through a revaluation.
Although separate estimates for assessing unit portions may be required for a year or two to prevent short-term inequities in tax apportionment or state aid allocation, a revaluation would accomplish this more effectively in subsequent years. However such revaluations are less likely to occur if the state tries to correct local deficiencies on a long-term basis through the equalization process.
1In addition to ratio studies, which quantify de facto deviations from the standard, some states also use a "procedure audit" to track local compliance with statutory requirements and generally accepted standards of valuation and tax administration.
See: Peter L. Davis, Building a More Powerful Ratio Study with Fewer Resources, Kansas Division of Property Valuation, 1995.