DUE 4/18

7

195

foreCastIng governmental

revenues teChnIques and lImItatIons wIth the assIstanCe of drew s. barden*

Performance by another that entails consistency, competence, and reliability is likely to enhance greatly the likelihood of the development of a relationship of trust.

(Valerie Braithwaite 1998, 52)

State statutes and local ordinances define the sources and types of revenues potentially available to local governments, but to build a budget, governments must have some idea of the accessibility, amount, flow, and timing of those revenues over a fiscal year or biennium. Developing forecasts or best estimates of the expected revenues for the coming year is a critical step in the budget process. The forecast level of revenues, plus interfund transfers and any funds brought forward from previous years, define the resources available to fund expenditures. The amount of available resources defines the potential size, capacity, and mix of government services for the coming budget year. Revenue forecasts also help to define the political environment faced by elected officials, interest group advocates, nonprofit foundations, service providers, and other budget actors. An understanding of pending government finances and program potential allows community nonprof- its, foundations, and network providers to develop complementary and supplemental programs. Revenue forecasts also help to assure debt service payments and the potential for future capital investments. In sum, revenue forecasts form the foundation for the local government budget and for many polity-level governance decisions by community leaders.

Long-term revenue forecasts provide important information to support financial forecasts and strategic financial plans (chapter 10). These extended forecasts of three or more years reflect assumptions about the local and regional economies, employer behavior, community wealth, and state and federal support for intergovernmental revenues. Matched with extended forecasts of future program expenditures and service caseloads, these long-term revenue fore- casts help to ensure the ongoing structural balance of the local government budget (chapter 15). The long-term revenue forecasts also provide trends against which to context short-term budget-cycle forecasts.

Effective revenue forecasting is as much a professional art as it is a technical procedure. Un- derneath the rigor of mathematical and statistical techniques, revenue forecasts rest on a series of assumptions about demographic changes, future business behaviors, and evolving economic conditions. Regional, state, national, and even global economic conditions affect local businesses and their decisions to expand or to reduce capacity. For example, the decisions by local property owners to develop agricultural land into new housing and commercial facilities would typically

*Drew Barden, PhD, retired economist for the City of Portland, Oregon, provided the initial approach and content for this chapter.

196 REVENUES AND BUDGETING

reflect favorable economic conditions, available developers, other business plans, and often, per- sonal decisions of retirement or career change. In the same way, family decisions to build a new home, or to renovate and expand an existing home, reflect personal savings, family circumstances, and job security. Homebuilding and remodeling decisions ultimately affect government building permit revenues and property tax assessments. The cumulative effect of these small individual decisions across a community provides critical information to forecasters as they generate budget- cycle and long-term revenue estimates.

With so much uncertainty folded into forecasts, analysts and forecasters must strive for both technical accuracy and for credibility with forecast users. This uncertainty applies to both budget- cycle forecasts and long-term forecasts, but the need to understand the near-term revenue situation for budget development and adoption especially sharpens credibility concerns with the budget- cycle forecasts. Revenue forecasts must be sufficiently certain to support public statements and decisions by elected officials, executive administrators, nonprofit partners, and other community leaders. A forecaster’s ability to communicate and to explain is as important as the technical skill needed to model data into forecast estimates. True credibility is built year after year with accurate forecasts and effective explanations. We address the importance of confidence and organizational relationships in the first part of this chapter.

Effective forecasters and analysts bring a full toolbox of approaches and techniques to the revenue forecasting task. The techniques range in sophistication from educated estimation, to spreadsheet deterministic models, to averaging and trending techniques, to highly complex statisti- cal regression and systems models. Mathematical break-even analysis techniques are particularly important for forecasting revenues from fees, charges, and rates for services. Each technique brings its own assumptions, limitations, and strengths. The quality of a forecast reflects the forecaster’s skill at making valid assumptions, applying the appropriate models to the available data, and selecting a forecast from among a range of options. The second part of this chapter provides an overview of the major forecasting techniques and details their application.

Property taxes provide a major source of revenue for many local governments. The details of the property tax assessment and computation vary by state, but the basic system relies on the accurate assessment of all property in the jurisdiction, computation of tax owed, and collection of the tax. After explaining the property tax system in detail, the chapter summarizes a common technique used to forecast property tax revenues.

Revenue forecasts by local governments reflect the structures and requirements embedded in the parent state tax system. Statutory limitations on property tax revenues and government ex- penditures, also known as tax and expenditure limitations (TELs) affect the revenues available to local governments to support programs. Accurate property tax revenue forecasts must consider any tax limitations in the state tax code. This chapter closes with an overview of the main types of property tax limitations and a discussion of the effect of limitations on revenue and the viability of local governments over the medium and long terms.

REVENUE FORECASTING TO BUILD CREDIBILITy

Revenue forecasting forms the foundation of the public budget. The identification of the types, availability, and timing of revenues dictates the levels of spending possible in a budget. Over- estimating expected revenues results in midyear program cuts and dashed expectations. Under- estimating revenues leads to a scaled-back budget that leaves public needs underserved; it also leads to the embarrassment of excess revenues, charges of overtaxation, and the political task of explaining how a jurisdiction will constructively use the unexpected windfall. How to build accurate forecasts and gain the confidence of budget actors in those forecasts is the forecaster’s primary task.

FORECASTING GOVERNMENTAL REVENUES 197

Revenue Forecasts Support the Budget Process

Revenue forecasts contribute to the budget process at several critical points. Depending on the size of the local government and the budget process calendar, the activities of budget preparation adds a five- to ten-month lead time before a budget is actually adopted and executed. A budget- cycle revenue forecast must accurately cover the entire budget preparation time frame and carry through the final months of the fiscal year or biennium with the best possible estimates. Revenue forecast updates, often on a quarterly basis, provide the means to keep the initial round of forecasts updated over the months or years of budget preparation and execution. A review of the forecast accuracy from the previous fiscal year may often provide additional information on assumptions and analytic model effectiveness. The quarterly revenue forecast updates may trigger a revision and rework of agency program spending requests. Rather than a linear flow, the budget process becomes circular as new information from each subsequent forecast update is reviewed, consid- ered, and adjusted into the proposed budget.

On the technical level, the budget-cycle forecast estimates provide critical revenue levels to the department-level analysts beginning to schedule out potential programs for the budget year. Restrictions may limit the use of some funds in place of others, and certain funds may not be avail- able until partway through the budget year. The programming of revenues, purchases, and work activities during the budget preparation phase sorts out such timing conflicts. Periodic updates and refinements to revenue forecasts help the department analysts develop the most likely scenarios for potential programs and service levels. Analysts in the central budget office and executives refining budget allocations and program levels also benefit from the most recent revenue forecast updates. Budget-cycle revenue forecasts also help community issue advocates explain the chang- ing economic realities to their membership and clients.

Long-term revenue forecasts (three or more years’ time frame) prepared as part of the orga- nization’s financial forecasts and financial strategic plan provide a backdrop and context to the budget-cycle revenue forecasts and updates. The long-term forecasts reflect durable trends and assumptions about the local and regional demographics, business operations, and regional and state economic capacity, state and federal fiscal, monetary and revenue sharing policy, accumu- lated community wealth, and revenue generation potential. In contrast, the budget-cycle revenue forecasts help the budget actors understand the implications of recent and ongoing changes in economic conditions. The combination of strategic context and current detail helps executives (CEOs) and elected officials adjust their political agendas and budget decisions to both the limits and to the full potential of regional economic capacity and wealth.

Budget-cycle revenue forecasts become especially critical following release of a proposed budget by the CEO to the council, commission, or board. A decision by the CEO not to raise additional revenues sets a cap on program size and composition for the coming fiscal year. Any request for increased taxes must be presented to the council, commission, or board for hearing and adoption before any subsequent adoption of the full budget. Recent and accurate revenue forecasts form the basis for the political decisions of whether to raise additional revenues or to rely on current taxation levels. Once the budget is adopted and execution begun, revenue forecasters continue to supply refined estimates over the fiscal year. Program analysts can then adjust service delivery capacity as needed. Revised revenue forecasts also help community groups and advocates under- stand potential changes to government programs.

Thus, the main task of the forecast group is to generate a set of numerical estimates that work. These estimates norm budget process deliberations and get the budget and the agency through the immediate fiscal year or biennium without program cutbacks. Optimistic revenue estimates for the current years should not result in reduced revenues in future years. Overly aggressive front- loading of revenue estimates only leads to reduced service levels and dashed expectations in future

198 REVENUES AND BUDGETING

years. Long-term forecasts provide a larger context for making adjustments to program size and composition. The close of the fiscal year provides an opportunity for forecasters to review their approaches, techniques, and judgment against actual events and changes in the economy.

Building Budget Actor Trust in Revenue Forecasts

Elected officials, program administrators, community leaders, issue advocates, program clients, and citizens all bring different priorities and needs to the local government budget process (chapters 3 and 4). Executive administrators use the forecasts to help set priorities in a consolidated all- funds budget and to structure discussions with the legislative body (e.g., council, commission, or board) (see chapter 15). Elected officials use the forecasts to set political strategy and tactics and to set expectations with constituents, community leaders, and interest groups. Revenue forecasts help set the resource context for the community services network, for intermediary nonprofits, and for community foundations. The forecasts also help advocacy group leaders set expectations and define the limits of what is politically possible to their members and clients. Program manag- ers, contractors, and nonprofit partners may use the forecasts to program resources and to guide hiring decisions. Revenue forecasts set the framework for labor negotiations between unions and governments (see first teaching case in Part IV). Investors and financial rating agencies review revenue forecasts as part of risk assessment in the offer and purchase of public bonds and other debt instruments. In sum, revenue forecasts are applied both inside and outside of the budget process, in both the short and long term.

As described in chapter 4, the different budget actors bring different perceptions, values, and understandings of taxes and public finance to the budget process. These differences apply to revenue forecasting as well. A consensus reconciling all of these different needs and perspectives is unlikely, but it is important that the budget process actors enter into the debate with some agreement on the rules of the game. Agreement in advance, on the size of the estimated revenues and resources available for programs is one major rule that can help contain subsequent budget debates.

Revenue estimation and forecasting relies heavily on quantitative, statistical, and economic techniques, but sophisticated techniques are only one part of building trust in the forecasting process and its estimates. The techniques for assembling revenue estimates are partly science and partly artful judgment. This is true whether the forecast is developed by a clerk in a small city finance department, a central budget office staff analyst, a contracted expert consultant, or a small staff of economists and forecasters working for a large city or county government. The techniques that feel familiar and solid to the revenue forecaster may easily appear as a mysterious “black box” to the general public and to elected officials. Revenue forecasters must make a special effort to translate their approaches and processes into terms that are easily understood and trusted by the various actors and publics who get involved in the budget process. Along the way, program managers, peer reviewers, central budget office analysts, and interest group analysts will bring higher levels of sophistication to their critique of revenue forecasts. Forecasters must be ready to respond to these many different forms and levels of oversight.

The art of forecasting reflects the application of judgment in: (1) the selection of forecasting techniques and tools, (2) the evaluation and selection of assumptions, and (3) the selection of fore- cast results to bring political agreement. The selection of forecasting tools and techniques reflects the importance of the revenue source, the time available, the data available to support a forecast, and the variation in the data. We will summarize the data needs, application, and sophistication of the different techniques later on in this chapter. Each technique will yield a slightly different revenue estimate. An estimate is the numerical prediction of revenue generated under specific as- sumptions and by a given technique. Where a simpler, more easily understood technique provides acceptable results, building credibility may call for its selection and use.

FORECASTING GOVERNMENTAL REVENUES 199

The evaluation and selection of assumptions comes in the interpretation of law and regula- tion, and in the assumptions that structure an estimate analysis. Assumptions also flow from an understanding of the political and economic context in which the forecast is developed and used. The forecaster’s sense of context, understanding of the larger economy and revenue situation, and intuition support his or her selection of an estimate.

In the end, the final selection of a revenue forecast should come from a range of potential alternative estimates. Technically sound, unbiased analysis stands as the foundation of credibil- ity. Sound analysis develops a truthful picture of the revenue situation for better or for worse. Selecting the most conservative estimate may underestimate revenues. Selecting estimates that are more conservative also opens the estimates to political scrutiny and mistrust. Erring in the opposite direction of overly optimistic revenue may lead to uncomfortable belt-tightening later in the fiscal year. The most useful estimate may be the most likely estimate. The most likely, or baseline, revenue estimates are neither optimistic nor pessimistic. Ideally, the actual collected revenues come in at, or slightly above, the most likely estimate. Most likely estimates may, in fact, be slightly pessimistic.

Several additional principles help to inform the judgment of forecasters and analysts in the estimation and selection of revenue estimates.

• A forecast needs to be reproducible—that is, generated with a systematic and coherent set of underlying assumptions and conditions. The same initial conditions and assumptions should always produce the same results.

• The new forecast should always be compared to the old forecast so that elected officials and managers can see how and why resource estimates have changed. For all but the newest revenue sources, a previous forecast should always exist. The previous forecast might be the second-year estimate of last year’s long-term forecast.

• All revenue estimates need to be rationally developed out of history; this is particularly true for an estimate for a new revenue source.

• Initial conditions are extremely important. Next year’s forecast needs to be continuously conditioned by both the current year and by history. Any forecast should be accompanied by a history, estimates for the current year, and a multiyear resource forecast.

• Forecast results must be presented in a coherent document that clearly conveys the estimates to elected officials and managers. The forecaster needs to develop a core set of tables, text discussion, and graphics that appear in regular format, even as the content changes. Many jurisdictions establish a formal revenue manual to catalog the detail that supports each forecast.

This last point stresses the need for analysts and forecasters to effectively explain and com- municate their forecast results, assumptions, limitations, and processes. In the heat of debate, elected officials will often leave the assumptions and limitations behind as confusing details. Briefing materials should take this behavior into account and package assumptions and limita- tions in concise, easily presented concepts. In communicating the results and details of a revenue forecast, “less may be more.”

qualities of a Credible Forecast

Developing accurate revenue forecasts that closely match actual revenue collections is a technical challenge. As mentioned, forecasters must rely on data and cues from past and current conditions to forecast revenue collections up to a fiscal year or even 24 months (biennium) into the future. Hundreds of variables across numerous business sectors can influence forecasting assumptions,

200 REVENUES AND BUDGETING

decisions, and estimates. Local population growth, demographics and age profiles, employment rates, investment and employment decisions by major local employers, state government spend- ing and policy decisions, and local economic conditions provide the primary set of variables for forecasts. National-level economic and monetary decisions by the Federal Reserve, national and regional economic performance, and the performance of the major financial trading markets may add more variables to the forecast dataset. Forecasters often face the challenge of limited, non- existent, or inconsistent data, and data may need adjustment for seasonal factors or for changing economic conditions. In the end, the forecast estimates rarely match the actual revenue collections exactly. Consistent success in generating forecasts that nearly match collections enhances forecaster credibility with decision makers and users. Acceptable deviations reflect the particular techniques involved, as well as the importance and contribution of the revenue sources to the jurisdiction. Major deviations and repeated large deviations between forecasts and actual revenue collections are cause for a loss of forecaster credibility.

A credible forecast uses quality data and correctly applies the appropriate mathematical, statisti- cal, or economic techniques. Once an effective forecast model is established, analysts should try to continue to use the same technique as long as possible. The continued use of the same technique year after year builds familiarity and acceptance in elected officials and in the public. Widely used systems and widely accepted techniques add to their confidence. Any deviation from commonly used methods will need explanation and justification. Consistency with Federal Reserve forecasts and commentary, published statistics on regional labor, commerce and housing surveys, census and demographic statistics, and university research publications are also important.

At the state government level, revenue forecasts can have substantial impacts on state budgets and agency service levels. This tends to generate public concern and political interest in the forecast quality. To respond to this interest, governors and legislatures may convene a peer review process to review both quarterly and long-term forecasts. A peer review process typically involves a mix of academic economists, corporate and independent economists, and local and regional business analysts. One panel may review the selection of technical methods and mathematical models, while a second panel may review the recent economic situation and any assumptions underlying the forecasts. The findings of the peer review panel provide critique, explanation, and, ideally, ratification of the forecast. A full peer review process is applicable only to state governments and to the largest city and county local governments, but scaled-down reviews and the endorsement of expert consultants can help to add critical credibility to any forecast team and its work.

Managing the Revenue Forecast Team

Revenue forecasters face the professional challenge of blending technical methods into an environ- ment of political pressures. Forecasters must be ready to fully display their approaches, methods, assumptions, and estimates to skeptical elected officials, community activists, and the public. In many jurisdictions, the budget process allows public hearing or formal work session agenda time for a formal recitation and oversight. A successful performance during oversight sessions sets the cornerstone for building credibility in the forecaster’s work. Executive administrators, central budget office analysts, and program managers also cast a critical eye on forecasts. Once these actors are satisfied and won over, they can act as trusted messengers to key community opinion leaders, stakeholders, advocacy groups, and interested citizens. Beyond the basic task of building credibility, forecasters often face increased political pressure to find additional revenue. Requests by executives and elected officials may cause forecasters to go back and review their assumptions of economic conditions and their selection of the most likely revenue estimates. Here, forecasters face the decision to support a marginally more optimistic revenue forecast, or to hold firm with a more pessimistic—but more likely—forecast. Chief executive officers, finance directors, chief

FORECASTING GOVERNMENTAL REVENUES 201

finance officers (CFOs), and staff directors need to provide the political support necessary to allow forecasters to present unbiased, truthful estimates of the revenue picture. Doing so will support the forecasters’ credibility and build support for the budget as a governance outcome.

During the budget preparation season, revenue forecasters live in a world of process deadlines, last-minute requests, and intense overwhelming workloads. Even in the smallest jurisdiction, there are too many revenue sources and too little time and money to be equally thorough and systematic in making an estimate for each revenue source. However, each revenue source rep- resents a different level of importance to the overall revenue picture. Revenue forecasters must concentrate their time and attention on those revenue sources with the greatest contribution and impact to the budget. Primary sources of revenue such as property taxes, income taxes, retail sales taxes, and major charges and fees require the greatest attention and the greatest rigor in forecast- ing techniques. If the collection and distribution of property taxes and retail sales tax revenue is centralized at the county or state level, the central authority may contract with expert consultants or professional organizations for forecasting services. Local governments may purchase forecasts on a fee subscription basis. More minor sources of revenue, especially those with little variation from year-to-year, deserve adequate but limited attention. Forecasters should not spend extensive time or effort developing complex techniques on very minor revenue sources with little historical data or annual variation. Time pressures reinforce the need to prioritize a forecaster’s attention. Limited time and data may force an analyst to make a “quick and dirty” forecast. In some cases, these quick but knowledgeable estimates will provide sufficient information for decision makers. Here, the relative or magnitude difference in estimates may be much more important than the absolute level of the estimates.

Analysts and economists work to develop accurate, defensible forecasts within the constraints of revenue importance, time, and money. Elected officials, especially in large cities, counties, and in state legislatures, face a much different reality as political actors, and their behaviors respond to a different set of needs and discomforts. An elected official may request a review of a forecast analysis to ensure its validity. Requesting a review may be a way of signaling the reality of excep- tionally low revenues to themselves, to disbelieving colleagues, and to constituency groups.

The availability of quality data is a critical requirement for effective forecasting. Ideally, data- sets contain sufficient points to support statistical analysis; oftentimes, however, this is not the case. In many instances, the collected dataset covers only a few recent years, has gaps, or reflects changing conditions that render early years inconsistent or incomparable with later years. Gath- ering the data necessary for statistical and trending analysis takes time and money. Organization effort and commitment are required to gather the cost information, performance measures, and output measures that fill out forecasting datasets. Support from program managers, staff directors, and other administrators helps assure that forecasters have the data they need to make the case to elected officials and to the public.

OVERVIEw OF REVENUE ESTIMATION TEChNIqUES

Revenue forecasting is the analytic process of using data from the past with contextual assump- tions to develop expected revenue estimates for future fiscal years, budget periods, or strategic planning time periods. A forecast is made up of a series of annual or periodic estimates. Estimates are the numerical result of an analysis process. For most forecasts, the levels of revenue generated in the past are seen as indicative of future revenue levels, and revenue forecasts are specifically designed to compute future estimates that exceed the time range of the dataset. This approach contrasts with most statistical procedures with assumptions that limit predictions to within the time range and variation of the dataset.

Forecasters gather the best data they can and make the best assumptions they can, but in the

202 REVENUES AND BUDGETING

end, forecasts are a best guess of future revenue levels. The future is uncertain, and unexpected events of varying intensity disrupt and challenge the familiarity of past experience. Changing economic conditions may quickly stretch beyond the experience and performance of the most complete dataset. For example, few analysts and observers could have foreseen the extreme fi- nancial conditions caused by the economic downturn of 2007–2009. By fall of 2008, real estate values became highly uncertain, consumer spending was colored with caution, and the returns from financial investments failed to produce expected returns. These extreme conditions resulted in data points that strongly deviated from historic data trends. Forecasting property, income, and sales tax revenue for 2009 and 2010 became especially difficult.

Commonly used revenue forecasting techniques include:

• Educated estimation • Deterministic spreadsheets • Break-even analysis • Different forms of averaging • Trending methods • Linear regression analysis • Systems and models of regression equations

We provide explanations and examples of several of these techniques on the textbook webpage. Each of these forecasting techniques responds to particular types of data, assumptions, and legal and economic situations. The above list, however, loosely orders the techniques by increasing statistical power and data requirements. The trending, regression, and other statistical techniques require the use of an electronic statistical analysis package. Graduate-level coursework in analytic methods typically provides the theory and background for using these types of forecasts. While using a high-powered technique may offer professional satisfaction, we caution analysts to keep things simple. A simpler model may give satisfactory results and be much easier to explain to elected officials, senior administrators, and the public.

Local governments typically rely on dozens of different revenue sources. State statutes and regulations, along with local ordinances, authorize many of these sources, including taxes, inter- governmental revenues, charges and fees for services, and fines and penalties. These laws define the revenue authorization, define which citizen or business group is subject to taxation, define the rates and rate caps that local governments may levy, and prescribe how the revenue may be used. It then becomes incumbent on each local government to identify those sources that it is authorized to use. The major forms of revenues are familiar and easy to identify—property taxes, retail sales taxes, or major state intergovernmental payments. Minor, less familiar authorizations for special purpose taxes may require detailed research for their identification and use. Each source of revenue deserves some level of forecast, and each forecast generates numerous assumptions, estimates, and supporting details. A revenue manual provides a central repository in which a local govern- ment can organize and document its revenue sources. All revenue sources should be described in a standard format that results in consistent information for each source. The entry for each revenue source should include the following information (e.g., Clark County, WA, Auditor’s Office 1999; Municipal Research and Services Center, WA 2009):

• a definition/description of the type of tax, charge, or revenue source; • the statutory authorization allowing collection or use with details to the budget fund level; • the mechanisms to assess and collect the revenue; • the revenue limitations that apply to the tax or source; • a description of the methodology for calculating revenue;

FORECASTING GOVERNMENTAL REVENUES 203

• an assessment of the stability and adequacy of the revenue; • a categorization of recurring or nonrecurring (one-time) revenue (GFOA 2013b); • the accounting details, including relationships to budget fund, budget line-item, or accounting

codes; • the model(s) used to forecast the revenue including assumptions, model description and

statistical or mathematical methods, key constants and parameters, and confidence intervals and sensitivity analyses;

• a list of demographic, economic, political, and policy factors that can affect model estimates;

• the limitations behind the forecast; • a revenue history for each source, with the actual annual collections for the past five or more

years; • a comparison of the actual annual collections with the forecast estimates for the past five or

more years; • any helpful notes and hints for analysis and explanation.

A well-maintained revenue manual provides a basis for public explanations and confidence building.

Educated Estimation

Educated estimation may provide the most appropriate estimation technique for minor revenue sources. The approach is especially applicable when revenues vary little from year to year, and when the forecaster is knowledgeable of the history and context for the source. In developing an educated estimate, a forecaster considers recent revenue levels and any changes that would af- fect the revenue stream. Changes affecting the future revenue stream might include demographic trends, revisions to the state or federal tax regulations, changes in tax or fee enforcement, or economic cycles. In another appropriate application, the forecaster may have knowledge of a one-time, isolated spike in a normally stable flow of revenues. An analyst may be forced into educated estimation when historical data are very limited or unavailable. Forecasters using edu- cated estimation should be extremely careful about documenting their assumptions. Accumulating estimates with comparable assumptions through time allows the development of a rudimentary dataset. This dataset can ultimately be used to support the forecasting of future estimates with more refined techniques.

Deterministic Spreadsheets

Property taxes and other taxes with defined regulatory structure often present clear relationships that forecasters can effectively model with deterministic forecasting. Deterministic models are simply a string of equations with known coefficient values that model the assessment and valuation of a particular tax or fee. These models lack the variable nature of stochastic (random), statistical models, but their step-by-step linked equations lend themselves to electronic spreadsheet model- ing. We explain how to apply a deterministic model to property tax revenue forecasting in the next section of this chapter.

Break-Even Analysis

Break-even analysis provides an analytic basis for determining fees and charges for service, and for rate-setting actions by local governments. Break-even analysis also applies to the determina-

204 REVENUES AND BUDGETING

tion of fees for licenses and permits, franchise access fees, tolls, and utility rates. Council, com- mission, or board actions on fees and rates serve not only to set consumer charges but also, in aggregate, to determine the level of total revenues generated by the program annually. Forecasts of aggregate revenue levels determine whether the program recovers its costs, or whether it will need a subsidy to meet program objectives.

Effective rate-setting actions and the accompanying cost analysis should reflect the full and true costs of providing the particular service, including all direct and indirect costs for operations and maintenance, administrative overhead, and charges for the use of capital facilities (GFOA 1996). For analysis purposes, analysts categorize costs as fixed, incremental/step, or variable. Fixed costs typi- cally cover program capital investments and the costs that occur at any service level—administrative facilities, basic staff, communications, and information systems. In general, the more customers the department or program serves, the lower the per client share of the fixed costs.

Incremental or step costs and variable costs reflect costs that increase or decrease as the level of output rises or falls. Incremental/step costs reflect a level of cost tied to a defined range of service output. They are often linked to facility or machinery capacity. Suppose, for example, that a small program uses a single 7-passenger van, then grows to a 15-passenger van, then three 15-passenger vans, but at some point grows to need one or more large buses. Incremental costs would reflect each cost level to purchase and operate the needed vehicles. Variable costs directly increase or decrease as client citizens or businesses purchase more or less of a good or service. Variable consumption can be closely metered, as for electricity from a public utility district or water consumption from a city water utility.

The relative contribution of the fixed, incremental, and variable cost components to a charge or fee rate schedule varies by policy decision. Governments and nonprofits may structure rates to include more cost in the fixed cost portion of the rate. This approach, in essence, designates a larger basic fee with relatively small increments of variable cost. Other contrasting fee structures place less weight on the fixed fee portion and proportionally more cost in a graduated schedule of incremental steps for higher consumption or higher service attention.

Categorizing costs as fixed, incremental/step, or variable allows analysts to compute the cost for each unit of service produced by a program. With an understanding of the unit cost of a ser- vice, and how the per unit cost rate changes over different levels of output, analysts can develop estimates of total program cost at varying levels of output. With estimates of total program cost, analysts can then estimate program expenditures, or the charge and fee revenues, needed to operate the program at a given level of service delivery over the coming fiscal year. Cost management and performance analysis techniques similar to those described in chapter 13 provide the techniques to develop the refined per unit cost information needed for a detailed break-even analysis.

Break-even analysis forecasting techniques have wide application in commercial businesses, nonprofits, and governments (Ullmann 1976; Dropkin, Halpin, and La Touche 2007; Gainer and Moyer 2005, 296–297; Chen, Forsythe, Weikart, and Williams 2009). In a very simple break-even analysis, all the costs necessary to produce the service for a single category of client are totaled into a total cost amount. Actual expenditures from the previous fiscal year provide the data to determine total program costs. To cover costs, the jurisdiction or nonprofit must collect an equal total amount of revenue—the total revenue. Governments often know the number of customers they serve. For example, a city water delivery system serves a known number of residential and commercial customers. Knowing the total cost (equal to the total revenue) and the number of output units, the analyst can compute a per customer charge. More advanced break-even analyses break customers into categories of users, break costs into different categories, and determine rate schedules that increase with increasing consumption of water or electricity.

For some services, local governments and especially nonprofits may be able to increase the number of customers as a means to increase revenues. This also allows a greater amortization

FORECASTING GOVERNMENTAL REVENUES 205

or sharing of the fixed and incremental costs. A detailed break-even analysis would also identify classes of customers or clients and would consider the relative costs of providing services to each class. Effective cost management would monetize, or place a value on, the extra program capac- ity and service attention required for high-cost users (e.g., Winthrop and Herr 2009; Cordes and Winthrop 2008). Once a detailed, unbiased analysis is completed, the jurisdiction’s legislative body (council, commission, or board of directors) can turn to the social justice issues of how to subsidize particular classes of clients. The council, commission, or board may decide to impose a surcharge for high-impact users, a full cost recovery rate, or a partial rate for subsidized cost recovery (GFOA 1996). Council or board members may decide to forego full cost recovery in order to ensure service to certain groups of low-income or disadvantaged citizens as part of meet- ing larger goals of fairness and broad public good. The issues of cost allocation, high-cost clients, full cost recovery, and subsidy allocation are also very relevant to nonprofit executives and board members. For many small social service nonprofits, a sliding scale fee structure allocates cost discounts based on a client’s ability to pay. To further explore these issues, we provide an exercise on break-even analysis and rate setting on the textbook webpage.

Different Forms of Averaging

When a dataset of historical revenue information shows a steady visible trend, or shows a constant explained variation, averaging may provide the most appropriate forecasting tool. As with educated estimation, averaging is appropriate for relatively minor revenue streams. In developing a forecast using averaging, an analyst should consider any changes in the legal, economic, demographic, or organizational environment that could produce changes in past trends or introduce new variation into the forecast estimate.

A forecaster using revenue averaging can follow several strategies. Where no trend appears in the data, the familiar arithmetic average of all available past per period revenue levels is normally used. If a simple trend appears in the historic revenue data, a prior moving or rolling average that uses only the most recent two, three, or four rolling data points may provide an effective estimate (Berman 2007). However, computing an estimate using two, three, or four rolling data points may result in a prediction above or below the trend in the data. The analyst needs to exercise judg- ment in selecting the number of data points to include in the rolling average estimate. Some local jurisdictions use averaging to forecast the catch-all category of Miscellaneous Revenues.

Trending Methods

Where historical data of past actual revenues and the forecast estimates for those revenues are available, a trending technique such as exponential smoothing may provide the most appropriate forecast estimate (cf. Ullman 1976; Chen, Forsythe, Weikart, and Williams 2009). If wide variation appears in the actual revenues, a trend becomes difficult to identify and the technique has limited effectiveness. The technique is appropriate for data with limited variation and without spikes.

The trending technique of exponential smoothing estimates the future revenue for the next year based on a relationship between the actual revenues for past years and their comparable forecast estimates. A weighting factor between 0 and 1 sets the relative balance between the actual and forecast estimates in the future prediction. While computer packages can perform the calculations to select the weighting factor and to make the estimations of future revenues, the analyst must carefully review the weighting factor selection. Too low a weighting factor dampens the effect of the actual revenues, while too high a factor limits the value of past estimates. Local jurisdictions may rely on trending techniques to forecast telephone utility license fees and state revenue sharing for cigarettes and alcoholic beverages.

206 REVENUES AND BUDGETING

Linear Regression Analysis

Linear regression analysis is commonly applied in a two-step procedure to develop revenue forecasts. This technique uses historical data to first build a mathematical relationship between past revenue levels and one or more economic, social, or demographic explanatory factors. The regression relationship is expressed as a mathematical equation. A regression dataset usually pairs the level of actual revenue collections with numerical data on each explanatory factor. Each explanatory factor serves as an independent variable in the regression equation. The data are often collected on an annual, quarterly, or monthly basis. In the second step of the procedure, forecasters use the computed regression relationship to estimate expected future revenue levels. Using linear regression techniques to forecast future revenues assumes that the conditions af- fecting the explanatory factors and their relationship to revenue generation remain unchanged from the past.

In contrast to the relatively simple data requirements of the previous techniques, linear regression has more comprehensive data needs. Gathering the wide variety of data measures and indicators from which to select out a final set of explanatory factors is often an expensive and a long-term process. Buying datasets from economic research firms, consulting firms, or universities some- times provides an alternative to the data collection process. Once the data are available, building a refined regression model to estimate revenues requires professional abilities and a computerized statistical package. As with other forecasting techniques, the amount of time and resources invested in a regression estimation effort should be proportional to the importance of the revenue source. Even with a long-term, comprehensive dataset and a high level of talent, regression models may generate a revenue estimate of only limited quality. Typically, when this occurs, a critical explana- tory factor has not been identified or is missing from the regression equation.

As an example, a large city government uses a regression equation model to forecast annual revenues from business licenses. This revenue source is too large and important to rely on trending or averaging techniques. It is also highly affected by the business cycle and the national economy. After an analysis and selection of explanatory variables, the final regression equation contains three variables to explain revenues: the annual unemployment rate for the metropolitan area, national corporate profits, and the license revenues from the preceding year. In another example, regression models have been developed to forecast lottery ticket sales and the future revenues from a new lottery game. Important explanatory factors that affect lottery ticket sales and revenue include variables that a state lottery board can control and other factors related to seasonality. Variables controlled by the lottery board include promotions, advertising campaigns, game changes, and accounting changes. Seasonal variables can include week of the month, month or season of the year, and holidays.

Linear regression techniques offer analysts a set of highly flexible tools for computing revenue estimates. For difficult forecasting situations, advance regression techniques may prove helpful. These advanced techniques include nonlinear regression and variable transformations. Obtain professional guidance when considering the use of these advanced techniques.

Systems and Models of Regression Equations

As revenue sources grow in terms of absolute size and contribution, the application of more com- plex forecasting techniques may become justified. In an increasing step in forecasting complexity, several linear regression equations may be linked together into a system to forecast a revenue stream. State government forecasters, forecasting teams in the very largest county and city gov- ernments, and expert consultants use these complex techniques to forecast personal and corporate income taxes. At this advanced level, complex models of linked regression equations are used

FORECASTING GOVERNMENTAL REVENUES 207

to describe a state’s entire economy. This type of forecasting incorporates key information from regional and national economic conditions as well as from recent state-level tax collections. Us- ing these advanced techniques requires credibility with elected officials and the public. Without clear presentations, constant communication, and an open, peer-reviewed process, public trust in revenue forecasts may bottom out.

Effectively interpreting the estimates from a complex system requires considering both the estimates generated and the variation or error that surrounds the estimates. Typically, the error that surrounds an estimate is the uncertainty of the statistical, mathematical, and random sam- pling procedures used in the estimation process. It provides a range of probability and a quality context for the estimate. The error may originate from several sources: (1) a variety of structural assumptions built into the system of equations; (2) the interaction of several statistical estimates and regression equations; and (3) the variation in the data used to run the model. In the case of estimates generated from a single linear regression model, statistics from the regression calculation will provide error and quality measurements. But in systems of several regression equations and in larger complex systems, sensitivity analysis procedures are used to develop a range of possible revenue estimates under different data conditions and structural assumptions. For example, one state presents a “low,” a “likely,” and a “high” set of estimates for its quarterly update revenue forecasts. Professional review of both forecast estimates and any sensitivity or error analysis is critical to ensuring public trust in the forecasting effort. Analysts should always exercise special caution when using large and complex forecasting systems. Statistical models can capture only a portion of the real world and its many variables. Changes in the economy or in external condi- tions outside the range of the data or the models in the system may render computed estimates inappropriate and unusable. The forecasting system should be used as a tool, not as an absolute predictor of future revenues.

Selecting a Revenue Estimate

We have just outlined the major methods for forecasting current year and future year revenues. Each technique has data needs and brings strengths and weaknesses for a forecast estimate. Se- lecting the appropriate technique, making the appropriate assumptions, and selecting the most appropriate estimate are left to the forecaster’s professional judgment. The use of judgment is really the art of the revenue forecast.

The wide range of techniques from the simple to the complex reminds us that a remote rural town clerk can make effective revenue forecast estimates that are fully adequate to the setting. To develop a high-quality calculation, forecasters may apply several techniques to de- velop a set of comparable estimates. For example, averaging techniques, trending, and simple linear regression may all be used to develop estimates of future retail sales tax revenues. The forecasting feature in commercial accounting software offers another option for generating revenue estimates. Review of the estimates developed by the different models—but using the same data—points out their relative differences. One technique may generate a higher level of future annual revenues, while another technique may generate a lower level of revenues. One technique may generate a gentler, smoother model of the historical data and of future estimates. Another model may generate a more volatile model that closely tracks historical variation. One regression model may have a very good statistical fit or match to the historical data, but then grow wildly to unrealistic future estimates beyond the first year or two. Another regression model may fit well, but then sink to unrealistically low levels. An effective estimate will have both good statistical fit to the data and a realistic fit to expected future economic and business circumstances for the budget year.

Selecting a forecast model with both good statistical and realistic fit to economic capacity and

208 REVENUES AND BUDGETING

conditions is especially important for long-term forecasts of three or more years. This is because unrealistic increases or decreases in the mathematical and statistical estimates often appear more strongly in the out-years of the forecast. Additionally, the statistical error or the statistical quality of the estimates becomes poorer in the out-years of a long-term forecast. One approach to manag- ing the statistical error of long-term revenue forecasts is to develop and select a baseline forecast set on a preferred combination of statistical and realistic fits. Analysts then apply the sensitivity analysis procedure described above to learn how the baseline forecast behaves under varying assumptions and data inputs (Swanson 2008). This approach would typically generate a range of “low,” “likely,” and “high” revenue forecasts. With the context of a range of forecast scenarios, analysts and planners can increase the tolerance of financial forecasts and financial plans to changes in economic conditions.

PROPERTy TAX MEChANICS AND REVENUE FORECASTS

As we described in the previous chapter, property (or ad valorem, [value-based]) taxes are the most important form of local government revenue. While some jurisdictions rely heavily on income taxes or other forms of revenue, some form of property tax almost always contributes to the mix of local government revenues. A full understanding of the property tax system and its mechanics is critical for revenue forecasting and for budget development.

Valuation Forms the Basis of the Property Tax

In many states, a popularly elected county assessor holds responsibility for implementing the property tax system and for collecting property taxes. In other states, an appointed administrator serves in the role. A staff of professional assessors and analysts works under the direction of the assessor. The task of the staff is to: regularly visit every piece of property in the county; assess and estimate the value of the property; apply the appropriate tax rate to the assessed property value; produce the tax bills and notify the property owner; collect any tax due; and follow up on delinquent payments. As mentioned in the previous chapter, the county may seize and then sell properties with delinquent taxes.

Establishing a timely, accurate, and fair assessed value for each property is the most difficult aspect of the property tax system. While most property owners accept their assessments, concerned owners may want clarification, and skeptical owners may want to challenge the assessment. State law and county governments establish an administrative appeals process under which property owners may appeal their assessments (cf. Washington County, OR 2011; and Washington State Department of Revenue 2010). The process typically begins with a hearing before a county board of property tax appeals or board of equalization. The initial appeals procedures are often informal in nature to support citizen initiative. Should the county appeals process fail to bring a property owner satisfaction, state-level review procedures or judicial review are available. The varied background and varied skill levels of assessors and the decentralized location of assessors in remote county offices have sometimes led to inconsistent and inaccurate appraisals. Profes- sional organizations (IAAO 2011) and strong training have helped to improve assessment quality. However, state regulations may reserve high value and unique industrial properties for skilled state revenue agency assessors.

Typically, the county assessor provides tax assessment and revenue collection services for the county government and for all of the other governments and special service districts within the county. These related governments and programs include: cities, towns, and townships; school districts; regional transportation and port districts; special service districts for fire, insect control, flood control, and hospitals; and state government levies. A local government or district may

FORECASTING GOVERNMENTAL REVENUES 209

impose one or several tax levies. Thus, any particular piece of property in the county falls into a number of overlapping taxing levies as established by several of these governments.

Visualize the overlapping taxing districts as a stack of layers or pancakes. Geographic infor- mation systems (GIS) effectively capture this stacking and overlapping arrangement. A tax code area is a portion of the county where the combination of overlapping tax districts is the same. A large county may have more than 100 tax code areas, and a smaller jurisdiction or district likely has several, as demonstrated in Exhibit 7.1. The left column in Exhibit 7.1 demonstrates a hypo- thetical array of levies imposing taxes on a single piece of property in tax code area 110002. The levies fall into several types:

• general or current services fund for operating and maintenance expenses (e.g., City of Upper Cascadia and Upper Cascadia County general funds, Upper Cascadia school district M&O general fund, port general fund);

• operations and maintenance funds for specific programs (e.g., school equalization, mental health, mosquito control, EMS);

• levies for specific equipment, capital purchases, or construction (schools capital project fund, schools vehicle transportation [buses]);

• bonded debt service levies for construction and capital purchases (schools debt service, port bonds, and library, parks, and recreation).

Exhibit 7.1

Property Tax Rate Detail for Tax Code Area 110002 (rate on $1,000 assessed valuation)

Governmental entity/levy Total

Schools Schools General Fund Maint. and Ops. 2.38025 Schools Capital Projects Fund (Levy) 0.21873 Schools Debt Service (Bonds) 2.19842 Schools Vehicle Transportation (Levy) 0.08863

State and County State School Equalization (state) 2.47115 Upper Cascadia County General Fund 1.34063 Mental Health 0.01250 Developmental Disabilities 0.01250 Mosquito Control 0.00842 Soldiers/Sailors Fund (state) 0.00874 Conservation Futures Fund (state) 0.06250

City (City of Upper Cascadia) City of Upper Cascadia General Fund 3.36833 Emergency Medical Services (Levy) 0.23368 Park/Recreation Bond 0.14296 Library Bond 0.15281

Port District (Port of Upper Cascadia) Port General 0.26700 Port Bonds 0.16826 Cemetery (District #1) 0.02106 Fire District n/a Library n/a County Road Fund n/a

Total Rate 13.15657 Total Assessed Property Value $76,010,000

210 REVENUES AND BUDGETING

Tax code area 110002 lies within an incorporated city boundary. The services provided by the City of Upper Cascadia include fire and police, planning and development, road maintenance, library operations, and park operations. Library and park facilities are constructed under separate levies to cover capital bonds. In contrast, county residents living outside the city in an unincorpo- rated area do not receive this enhanced level of services. Residents in unincorporated areas would contribute to the county road fund under a countywide levy, and they would pay separate levies for a fire district and a library district if they had voted to tax themselves for these services (bottom of Exhibit 7.1). Where cities, towns, and townships do not provide a particular service, residents may vote to establish a special service district funded by a tax levy. Such a levy would appear as a tax district listed in Exhibit 7.1. Notice also that the school district imposes four different levies for operations, for capital facilities projects, for debt service, and for school bus purchases.

Exhibit 7.1 includes several state property tax levies. In this state, the county serves as the assessment and collection agent for the state government. The state school equalization levy is substantial, at $2.47115 per $1,000 assessed value. The state collects the levy and uses the revenue to help equalize school funding across rich and poor districts. The state has also imposed several small statewide initiatives for veterans’ support and conservation futures. Some states may not have state-level assessments.

The county assessor establishes an assessed value for each piece of eligible property in the county. Residential property is typically assessed on the value of the land and the permanent structures. Commercial property assessment may include land, facilities, equipment, inventory, and furnishings. In many states, the assessed value directly reflects the full fair market value of the property. In other states, the value may be reduced by a fixed dollar exemption, by a percent- age reduction, or by a rollback to the values of a starting year. The assessed value or its adjusted substitute determines the value variable of the property tax computation. The imposed burden of the property tax is expressed as the millage, or dollar amount levied on $1,000 of assessed property value. This is the property tax millage rate.

To compute the tax for an individual residential property owner, apply the following formula:

Property Tax = (Property Assessed Value) × (Total Dollar Rate/1,000) (7.1)

Suppose our City of Upper Cascadia homeowner owns a house and property assessed at $200,000. Using the total millage rate from the bottom of Exhibit 7.1, the total annual property tax owed would be calculated as:

Property Tax = ($200,000) × (13.15657/1,000) (7.2) Property Tax = $2,631 (7.3)

While a homeowner is most concerned about his or her individual tax burden, governmental entities are most interested in estimating the total potential tax levy or revenue generated from their tax base. To develop this total potential revenue, the county assessor must compute the tax revenue from each tax code area within the jurisdiction and sum these components into a total revenue figure. Exhibit 7.2 lists all the property tax code areas that fall within the Upper Cascadia city limits.

The assessed value column typically includes the values of taxable real property—multifamily residential, single-family residential, commercial, industrial—and business equipment, inventory, and furnishings. Tax code area 115000 likely contains major industrial and commercial facilities generating an assessed value exceeding $2 billion. The assessor omits any exempted properties such as churches, most government buildings and facilities, and public schools. Note that the

FORECASTING GOVERNMENTAL REVENUES 211

uniform tax millage rate of $3.36833 per $1,000 valuation applies in the four different tax areas that fall within the city boundaries. The combined revenue generated from each of the tax areas within a jurisdiction’s boundary is the total levy:

Total Tax Levy = (Total Assessed Property Value) × (Dollar Rate/1,000) (7.4)

Following through on our example, for the City of Upper Cascadia, the total tax levy from all tax areas in the city is:

Total Tax Levy = ($3,047,320,127) × (3.36833/1,000) (7.5) Total Tax Levy = $10,264,380 (7.6)

Forecasts of the total tax levy provide a base for budget construction for a jurisdiction. How- ever, a few taxpayers fail to make their payments in a timely manner. Taxpayer delinquencies are an inevitable occurrence, and analysts routinely apply a 2 to 5 percent reduction to the estimated total revenues (e.g., Waldhart and Reschovsky 2012; Johnson 2011). The delinquency-adjusted total levy becomes the revenue estimate used in budget construction.

Property Tax System Dynamics

The basic property tax relationship described in Equation 7.1 has implications for the property owners in a community and its local governments. The assessed values of properties can change for several reasons, but local governments need a continuing flow of revenues. The property tax formula responds to these changes (Franklin 2010).

The property tax formula applies to all eligible land and property in a tax code area. The as- sessed value of each property is multiplied by the total rate, and each property contributes its share of the total revenues collected by the county. In our homeowner example (see Equations 7.1, 7.2, and 7.3), the homeowner paid $2,631 annually. If all the homes in the tax area were identical, each homeowner would pay the same $2,631. Each home would be assessed at $200,000, and the tax rate for each property would total $13.15657 per thousand dollars of value. Suppose also that Upper Cascadia Library District proposes an annual operating budget of $1 million. Each of the 380 homes and properties in the community would contribute $2,631 for an equal share of the levy needed to meet the $1 million proposed budget.

In the real world, however, the local real estate market fluctuates with economic growth or contraction. The $200,000 market value of each house and property will fall with a sustained con- traction in the local economy. Yet, local government continues to need $1 million annually to cover

Exhibit 7.2

City of Upper Cascadia Potential Property Tax Revenues, 2008, by Source Tax Code Areas (general fund levy for regular maintenance and operations)

Area Assessed value Millage Potential tax revenue

110002 $76,010,000 3.36833 $256,027 110031 $17,402,283 3.36833 $58,617 112032 $351,673,333 3.36833 $1,184,552 115000 $2,602,234,511 3.36833 $8,765,185

Total Levy $3,047,320,127 3.36833 $10,264,380

212 REVENUES AND BUDGETING

service costs. Thus, if the assessed values drop, the tax rate must be increased to compensate and to assure delivery of $1 million. Similarly, if property valuations increase in a growing economy and the assessed value of a home increases, the tax rate will decrease slightly to deliver only $1 million to local governments. It is important to recognize that value and rate, as a mathematical product function, compensate in changing conditions to assure $1 million to local government. If the assessed values of all of the houses in the community are equal, and the tax rate is equal, each homeowner’s share of the total tax levy will be equal.

More realistically, some homeowners make investments to their homes—remodel kitchens or bathrooms, add landscaping, and construct major additions. These investments increase the value of those particular homes and their assessed value relative to other homes in the community— say to a home value of $220,000. Similarly, other homeowners perform minimal maintenance, and their houses decay. The assessed value of these homes will fall relative to the others—say to $180,000. But, many homes will continue near the average assessed value of $200,000. The tax rate applied to all properties is the same—13.15657. Thus, houses with above average assessed values will pay a larger tax, and those of lesser value will pay a lesser tax. In the end, however, the averaged property values of all 380 properties and the standard tax rate must produce the $1 million needed by local government. New construction and improvements to homes and property add new value to the tax code area. This increased value serves to spread the tax burden and to reduce the tax rate on the existing property owners to generate our hypothetical $1 million of revenues, though not at a one-for-one dollar value, since new construction requires additional schools, public safety, and other governmental services and facilities.

Grasping these dynamics aids our understanding of how property tax limitations have their effect. Controls on any part of the property tax equation affect the flow of revenue to local governments. We summarize these effects and property tax limitations in the last part of the chapter below.

Forecasting Property Tax Revenues

Detailed property assessments provide a rich database for forecasting annual property tax revenues. Property taxes are also grounded in a highly structured regulatory procedure. A deterministic spreadsheet model can effectively integrate these features into an effective revenue forecasting model. Forecasters develop a model for each code tax area within a jurisdiction (Exhibit 7.2). The combined forecasts from all tax code areas would provide a revenue forecast for the city, county, or special district.

As we summarized above, a deterministic model is simply a set of linked equations. Typically, the total assessed value for the tax code area is entered into the model. Estimates of the value of new construction are added to the total assessed value. If the assessed value is assumed equal to the fair market value, the model might adjust for broad changes in real estate valuation. If the model is applied in a budget-driven state that allows the jurisdiction to set a budget level and then raise funds, the model would input the needed level of total revenue. The model would then compute the tax rate for the coming fiscal year. Alternatively, in rate-driven states, the defined tax rate would be applied against the assessed values to compute the total available revenue. Forecasters can add tax limitations restrictions on assessed values, tax rates, or total levies as necessary. Once the model computes values, rates, and levies for the coming year, the same procedures can be used to forecast the successive year. The model can be cycled as needed to compute forecasts for the out-years.

PROPERTy TAX LIMITATIONS AND LIMITATION EFFECTS

Tax limitations and related expenditure limitations (tax and expenditure limitations, or TELs) to control tax burdens and the size of government have recently been imposed in most states (U.S.

FORECASTING GOVERNMENTAL REVENUES 213

Advisory Commission on Intergovernmental Relations [ACIR] 1995). Of these TELs, limitations on property taxes have been especially common. But property tax limitations have been imposed almost as long as the property tax has been used. The first round of limitations appeared in the 1880s and coincided with local home rule efforts. More recently, the current round of limitations began in the late 1970s and 1980s. Many of these limitations remain in statute and in force. In some states, multiple citizen initiatives have resulted in an accumulation of statutes and a confusing combination of controls. The actual impacts of the combined limitations often produce unexpected consequences as the economic conditions change.

Restricting the flow of property tax revenues requires placing restrictions on one or several components of the property tax equation (Equation 7.4). As a basis for this discussion, we repro- duce the jurisdiction-level version of the basic property tax equation here:

Total Tax Levy = (Total Assessed Property Value) × (Total Dollar Rate/1,000) (7.4)

In this equation, the total tax levy represents the combined revenues from all tax code areas within the jurisdiction, while the total assessed property value covers all the eligible property in the jurisdiction. The dollar rate divided by 1,000 is the millage rate adopted by the voters and adjusted or capped in conformance with state law.

A catalogue of tax and expenditure limitations (TELs) reveals six methods for controlling property tax revenues (ACIR 1995; Mullins and Joyce 1996):

1. Overall Property Tax Rate Limitation

This limitation mechanism covers aggregate total tax rate faced by a property owner in a particular tax code area. The dollar rate of $13.15657 per thousand dollars of assessed valuation for our City of Upper Cascadia homeowner is such an aggregated (Exhibit 7.1) tax rate. An overall rate limitation would cap or reduce this aggregate total rate.

2. Specific Property Tax Rate Limitations

This type of limitation caps or reduces specific rates individually, or by a class of rates. For ex- ample, a state may place a category rate cap on the school operations and maintenance rate, or on the general government rate, which combines all city, county, and special district governments. Some states allow the voters to lift a state-set rate cap through a ballot referendum (cf. Maher and Skidmore 2009). Super-majority requirements for measure acceptance may be even more restrictive than a simple majority vote.

3. Property Tax Levy Limit

This limitation caps the total tax levy generated (left side of Equation 7.4). Levy limits set a cap on the rate the levy may grow from year to year. The cap is sometimes linked to the rate of infla- tion and population growth. For example, Initiative 747 in Washington State in 2007 capped the total levy limit at a 1 percent per year increase for most local jurisdictions. Voters may lift the levy cap by ballot referendum.

4. Limits on Assessment Increases

This restriction controls increases in the assessed value of property. Many states embrace the assessed value as fair market value and place no restriction on the total assessed property value

214 REVENUES AND BUDGETING

component in Equation 7.4. Limitation controls in these states use rate and total levy limits to achieve revenue control. But many other states place controls on the assessed value of property. These controls include exempting a portion of the assessed value from tax, rolling back the value of property to a prescribed starting year, or limiting the growth of assessed value by a prescribed annual percentage. The exemption of a portion of value is used by many states. Some states use taxpayer age-based categories to provide tax relief to senior property owners (NCSL 2002). In contrast, California’s Proposition 13, adopted in 1978, set values at 1975 levels and placed a maxi- mum 2 percent annual growth rate on values for property held continuously since the imposition of the initiative (McCaffery and Bowman 1978).

5. General Revenue or General Expenditure Limit

This restriction steps outside the property tax equation and places limits on all revenue received by a jurisdiction. Similarly, a general expenditure limit places a cap on total spending by the jurisdiction. This type of restriction is often proposed at the state level to constrain state general fund spending (e.g., State of Colorado, Taxpayer Bill of Rights of 1992 [TABOR]; Martell and Teske 2007).

6. Full Disclosure or Truth in Taxation Requirements

This approach may focus on changes to the property tax, but also may step outside the property tax equation to include other types of taxes. This type of control requires governments to fully explain why tax increases are occurring, and may require public discussion and a formal vote before a tax increase. However, a public vote to raise taxes supersedes any information or discussion. This is a relatively weak tax limitation measure, but it may help to increase the quality and uniformity of assessments (Cornia and Walters 2005). Utah (Cornia and Walters 2005) and Florida have used this type of control (Riley and Colby 1991, 95).

The TEL statutes, regulations, and voter initiatives in a state may use one or several of these restrictions. Some features may be especially binding, and others not especially effective. A limitation on only one part of the property tax equation (see Equation 7.4) leaves room for minor adjustments in other factors. For example, a rollback restriction on appraised value may be cir- cumvented by a marginal tax rate increase. While a limitation on one factor in Equation 7.4 may reduce a resident’s tax bill and the revenue flow to local government, effective constraints on two of the three factors in the equation are required to ensure tax relief. State law may allow specific exemptions from the limitations, and in the end, administrators must dig through and thoroughly understand the details and exemptions of their state’s property tax system. Oftentimes, the vari- ous caps and limitations adopted by a state are “stacked” on each other. The entire tax process then becomes arcane and hard for the average citizen to follow, thereby lowering the trust of the citizenry in the system.

Changing economic conditions and changes in real estate assessed values may make the TEL more or less binding. A boom of economic growth, along with new building and facility construc- tion, may inflate assessed property values in the short term. The increase in value may allow a burst of revenues that hides the long-term effect of a limitation. A burgeoning growing economy in the 1990s obscured the ultimate effect of the Colorado Taxpayer Bill of Rights (TABOR) limitation until the early 2000s, when the economy changed. At that point, the TABOR generated a severe reduction in public services (Center for Budget and Policy Priorities [CBPP] 2013; Martell and Teske 2007). However, in Oregon, a property tax limitation from the late 1990s had aggressively adjusted assessed property values to below market rates. During the recent economic downturn, these adjusted property values remained relatively constant even as real market values declined

FORECASTING GOVERNMENTAL REVENUES 215

steeply, and taxpayers grumbled that the adjusted values should also decline. The divergence and dampened fluctuations in adjusted assessed values helped to hold property tax revenues to a stable level over the downturn. Public education helped to remind taxpayers of the benefits, and burdens, of the tax limitation over a full economic cycle.

Imposing a tax rate cap as a property tax limitation may bring unintended consequences. A tax rate limitation imposes a cap on the tax millage rate, but also brings up the issue of how the community should prioritize the component array of services funded under the cap. Exhibit 7.1 demonstrates the many services funded out of the aggregate $13.15657 per thousand dollars as- sessed value tax rate. As the combined rate is capped and reduced, the question becomes, Which services should take the largest reduction? Each state resolves this issue differently. Some states apply a uniform reduction that compresses all rates equally. Other states reserve priority funding to senior districts, and then let junior districts take the suppression or prorated rate reductions to bring the aggregate junior district rate under the prescribed limit. Junior districts might include single service special districts for flood control, cemetery, library, parks and recreation, or mos- quito control. If all junior district rates are reduced to zero, senior district rates are then prorated. Such a division makes sense on paper until circumstances demand full services from a junior flood or fire district. Compression, versus suppression and proration, demonstrate two different approaches for reducing district tax rates to meet statutory limits with a final goal of ensuring compliance with a rate cap.

Limitations that modify or roll back and adjust assessed valuations of property may also generate unwanted consequences. Application of an assessed value modification begins at a date specified in the ballot referendum, statute, or regulation. Property assessments on existing property are then constrained into the future, often with a limited percentage annual value increase. In actuality, real estate and property values for entire neighborhoods may escalate steeply after the implementa- tion date specified in the limitation, but this value increase may not be recognized in a modified assessment. In contrast, new growth may be valued at its full market value and carries a higher tax burden. Limitations on assessed value can quickly set up two classes of property owners and a shift in tax burden to new growth and recently arrived residents and businesses (Martens 2011; McCaffery and Bowman 1978, 534). Efforts to limit property taxes can quickly challenge taxpay- ers’ perceptions of fairness and horizontal equity.

The implications of TELS extend beyond the mechanics of the property tax equation. Property tax limitations can generate unintended shifts in funding, along with shifts in political power and control. Mullins and Joyce (1996) in a nationwide analysis of TELs reported numerous unex- pected consequences resulted from limitations. In general, TELs increased the centralization of government authority to the state level, lessened local responsiveness, increased the use of nontax revenue (charges and fees) by local governments, and lessened government’s ability to respond to the needs of dependent populations (Mullins and Joyce 76). Mullins and Joyce note that after tax limitations, the overall size of government doesn’t increase, but state government expenditure decisions become much more important. State government funds have had a tendency to backfill the loss of revenues caused by local level limitations. With greater funding authority shifted to the state level, local governments may become less responsive to local needs and preferences. Mullins and Joyce conclude that the centralization caused by TELs conflicts with free market public choice advocates’ preference for local control of government and taxation. With state governments under fiscal pressure following the economic downturn of 2007–2009, little revenue often remained to backfill local government property tax revenue losses.

To the positive, the capping of local property taxes and the centralization of funding to the state government can have an effect of equalizing school funding across the rich and poor school districts (Oregon Legislative Revenue Office 1999).1 To the negative, McCaffery and Bowman (1978, 535) describe how California’s Proposition 13 threatened to undo the California legisla-

216 REVENUES AND BUDGETING

ture’s response to the Serrano v. Priest (1971, 1976, 1977) California Supreme Court decision that requires school finance restructuring in the interest of greater equalization of expenditures across school districts.

Voters and legislatures enact TELs at the state level as amendments to the state constitution or as statutes. In either case, a TEL represents a major part of the institutional structure that de- fines local government finance and revenue collection. Adoption of a TEL is both a mechanical manifestation of tax containment and a symbolic statement of a desire to limit a particular form of tax or of government in general. Public opinion polling and campaign literature may provide explanations behind citizen adoption of a TEL, but once enacted into law, local government ad- ministrators must implement the TEL structure.

To enact a TEL, local administrators and elected officials must fully follow state regulations generated by the limitation statute, interpret any previous limitations for a collective effect, and design an organization and policy to best implement the limitation. The limitation singly, or in combination with earlier restrictions, may severely reduce local property tax revenues. As the teaching case at the start of this section demonstrated, administrators and the community must truthfully adjust staffing, policy, program delivery mechanisms, and program service levels to match revenue reductions. If the impact of the TEL is delayed and subtle, administrators must take on the difficult task of communicating its potential impacts to elected leaders and citizens. Long-term revenue forecasts and strategic financial plans are the principal tools for describing an impending budget structural imbalance and its implications on service delivery and on the jurisdiction’s creditworthiness.

Administrators and elected officials must know their community and understand how the TEL should fit against local policy and values. Once a TEL is imposed, administrators must ask them- selves: What is the TEL limitation trying to tell me, my government, and my elected officials? A review of the community culture may reveal that a small minority of taxpayers are explicitly concerned with excessive or unfair taxes, or with an unconstrained government presence. On the other hand, a major portion of community taxpayers may be messaging a fear that government is too big relative to community resources, is too complex, is espousing and supporting values dif- ferent from those of the community, or is not providing comparable worth of services. Outreach and an adjustment of services may be in order. But, where political support and community values message otherwise, administrators should recognize and honor community calls for government response and services. In many states, the taxing district residents may challenge a tax rate cap or levy limit by a ballot initiative (Maher and Skidmore 2009). In a nearly offensive response, local governments and special districts in Washington State have offered voters the option of lifting a tax levy cap, and then in a second ballot question establishing a permanent tax rate to fund a specified service. Where tax limits cannot be lifted, administrators may need to identify alternate sources of revenue to respond to community needs. Nontax charges and fees for service may provide such a locally controlled revenue source. Alternatively, administrators may need to look to the nonprofit sector as a source of resources and program support.

REVENUE FORECASTING SUMMARy

Forecast revenues set the foundation for the public budget and its allocation of resources to pri- oritized needs and programs. Generating revenue forecasts is a blend of science and art. Forecast- ers must have strong skills in statistical and mathematical methods as well as a good handle on economic conditions and their effects on local businesses. Forecasters also need a strong sense of professional judgment in the selection of assumptions and in the selection of forecast estimates from an array of choices. Most important, forecasters must not only generate forecasts but also

FORECASTING GOVERNMENTAL REVENUES 217

effectively communicate their methods and professional choices to elected officials, community leaders, interest advocates, and the public. Every forecast becomes an opportunity for building confidence in professional workmanship and in the larger budget process. Effective forecasting should also demonstrate the effects of tax limitations on local governments. A truthful picture of the short- and long-term effects of limitations builds credibility with taxpayers. A truthful revenue picture allows government leaders to return to voters with a request for additional funds to meet increased community needs.

STUDy qUESTIONS

1. Take the role of a fiscal analyst in a large port or airport special district government. The executive director and finance officer have asked you to take charge of all revenue forecasting to support the annual budget process. List the primary sources of revenue for your district. What technique would you use to forecast future revenues from each source? What data would you need to support a forecast, and how long would it take to collect?

2. Again, as a special district fiscal analyst, what priority would you give to each revenue source? How would you prioritize time and resources in developing revenue forecasts? Which sources would require substantial effort with regression modeling? Which other sources could use educated estimation?

3. What qualities and standards would you want in a revenue forecast? How would these qualities build public confidence in your work and in the budget process? Should your forecast assumptions necessarily match those used by the adjacent county in their fore- casts? Explain?

4. Most forecasting approaches use historical data adjusted by current and near-term condi- tions to forecast future revenues. The economic downturn of 2007–2009 injected such uncertainty into the financial and real estate markets that the trends in historical data provided little help for the 2009 and 2010 revenue forecasts. For example, once robust county revenues from building permit fees and development service charges fell to near zero in hard-hit states. What approaches and data would you use to develop forecasts for building permit fees and development charges estimates under conditions of drastic economic and employment uncertainty?

5. Review your local government, special district, or nonprofit organization and identify a major fee charged for services. Which user or customer group in the community car- ries the burden of this fee? Should this fee be reauthorized as a tax broadly borne by all citizens? Is the fee structured for partial or full cost recovery, or for profit genera- tion? Analyze the cost structure of the fee and identify the base charge, incremental charges, fixed costs, step costs, and variable costs and explain their integration into a total fee.

5. On the textbook website at www.pdx.edu/cps/budget-book, exercise 7.1 Revenue Fore- casting: Property Tax provides an opportunity to forecast property tax revenues using a simplified deterministic spreadsheet model. To practice forecasting retail sales tax rev- enues with a simplified trendline model, work exercise 7.2 Revenue Forecasting: Retail Sales Tax.

6. Research property tax limitations in your state or in an adjacent state. How do the limita- tions restrict or affect the assessed value, tax rate, or total levy of residential homeowners? How has the limitation shifted tax burden among groups of taxpayers? Has the limitation system created favored treatment among certain groups of homeowners?

218 REVENUES AND BUDGETING

7. How have property tax limitations affected the revenue available to local governments in your state? Consider both immediate and long-term revenue consequences. Has major business, real estate, or residential development delayed the impact of property tax limitations in your area? Have depressed economic conditions delayed the effect of limitations? What larger political implications have the limitations caused?

NOTE

1. For additional information on state-level school equalization of funds, see Melissa Beard, Heather Moss, Isabel Munoz-Colon, and Sarah Reyneveld, A Reference Guide of Six States: K–12 Funding Formulas in Colorado, Kentucky, Maryland, Massachusetts, North Carolina, and Oregon (Olympia, WA: Washington Learns, Office of Financial Management, and Office of Superintendent of Public Instruction, August 2006), www.washingtonlearns.wa.gov/materials/SixStateStudyREALFinal.pdf (accessed May 16, 2011).