Introduction1
Sun and Lynchâs (2008) Government Budget Forecasting: Theory and Practice provided one of the first systematic reviews of government budget forecasting more than a decade ago. More recently, Williams and Calabrese (2016) examined the recent literature on and techniques used in government budget forecasting. We received significant feedback inquiring about an even more thorough treatment of the topic, and the current volume is an outgrowth of these inquiries.
To understand budget forecasting, it is critical to understand what a budget is. While âbudgetâ is a common term used for a variety of purposes, it was introduced into American governmental practice to focus on the planning stage that precedes an appropriation (Cleveland 1913). With this link to planning, it is clear that forecasting is a critical component of public budgeting. Thus, the current volume is intended as a contribution to the understanding of an essential, yet still understudied, element of public budgeting systems. Further, governments have several budgets, but there are two major types: expense budgets and capital budgets. Expense budgets are for the acquisition of resources that are promptly consumed, usually in the day-to-day operations of the government. Capital budgets are for the acquisition of resources that are retained for long time periods. In both the literature and research, far more is known about expense budget forecasting than capital budgeting, even though capital budgets can be larger than operating budgets. While this collection mostly addresses the first sort, it includes one chapter that brings an aspect of capital budget forecasting into focus.
There are many facets to budget forecasting practice. These can include practices at the international, national, and subnational levels. In the United States, subnational refers to states, cities, counties, and special districts. Similar, but sometimes differently labeled, jurisdictions can be found in other countries. Because public service is also delivered by not-for-profit organizations, we have endeavored to include two chapters on their forecasting. This is a poorly studied domain, with virtually no literature examining how not-for-profit organizations forecast or assessing the quality of their forecasts. Budget forecasting includes the forecasting of revenue and expenditures. For expenditures, the forecast often focuses on factors that lead to or cause public expenditures, such as student enrollment, prison populations, or other similar factors that lead to the need for public funding.
For larger governments, budget forecasting generally reflects economic modelling of the jurisdictionâs domestic product. For smaller governments, it may reflect the use of simpler techniques. While there is wide reliance on stochastic modeling, some budget forecasting relies on judgment and some relies on deterministic models. The text includes material across many of these approaches. The chapters included in this volume were peer reviewed by experts on these topics.
This book has four parts. Part I contains chapters related to international and national budget forecasting. These practices have similarities and differences among the countries examined here: the United States, Germany, and a large sample of low-income and developing countries. This section also includes a discussion of macroeconomic modelling for forecasting, which commonly occurs with national forecasts as well as with larger states and potentially some larger local governments.
In Chap. 2, Gerald D. Cohen provides an overview of the economic theory and analysis used by U.S. federal government economists at the CBO and Troika to derive forecasts for productivity, labor force, inflation, and interest rates. These variables play a key role in the budgeting and policy formation process because exogenous changes in the economic outlook or policies, such as infrastructure or paid family leave that move the needle on these variables, can have significant impact on the budget outlook. Moreover, the interplay between these variables means that forecast errors can cascade.
In Chap. 3, Neil R. Ericsson and Andrew B. Martinez discuss the evaluation of budget forecasts using information from U.S. federal government agenciesâ forecasts. The authors review the extensive literature on forecast errors and demonstrate the use of various forecast methods with forecasts made by the Congressional Budget Office (CBO), the Office of Management and Budget, and the Analysis of the Presidentâs Budget over 30 years. The forecasts of each are examined for bias, efficiency, and other characteristics. They recommend a generalized approach for the study of forecast errors to obtain the best forecast.
In Chap. 4, Dörte Busch and Wolfgang Strehl discuss the legal framework of budgeting in Germany. They discuss the implications of the Maastricht treaty for budgeting and forecasts for budgets. The chapter shows a sharp rise in the debt-to-GDP ratio during the 2008â2012 recession and aftermath and a slow decline thereafter. They describe the links between forecasting and such matters as structural budget balance, the GDP, and the debt break.
In Chap. 5, Marco Cangiano and Rahul Pathak discuss the revenue forecasting landscape in middle and low-income countries with a focus on examining the existence of forecast bias and potential remedies. They construct a dataset of ex-ante revenue forecasts and ex-post realizations for 26 countries using the information from the Public Expenditure and Financial Accountability (PEFA ) reports, and find that most of these countries tend to overestimate their revenues. The forecast errors are significantly large and appear to correlate with the measures of income and administrative capacity. They review two institutional innovations for improving the budget process and forecasts: Semi-Autonomous Revenue Authorities (SARAs ) and Independent Fiscal Councils, although neither of these institutions has been explicitly tasked with providing independent revenue forecasts that could address the observed bias. Lastly, the chapter highlights the lack of research and data on revenue forecasting in low and middle-income countries and recommends future research.
In Chap. 6, Rudolph Penner examines long-term projections of U.S. federal budget totals. He finds that the predictions are largely driven by a growing elderly population with spending on Social Security, Medicare, and Medicaid growing more rapidly than tax revenues. This is predicted to lead to an explosion in the debt-to-GDP ratio. Recent unusually low interest rates and the dot-com boom of the late 1990s moderated this ratio. However, the Great Recession caused the debt-to-GDP ratio to briefly rise much faster than expected. Despite these surprises the rapid growth of programs serving the elderly has been forecasted fairly accurately. Penner concludes by discussing program designs that adjust to surprises by changing indexing or using trigger mechanisms.
In Chap. 7, James W. Douglas and Ringa Raudla assess the Congressional Budget Officeâs (CBO ) ability to use information effectively to make quality projections by examining whether one-year ahead and five-year cumulative projections updated in the summer are more accurate than its initial winter projections for fiscal years 1978 through 2017. They find that the updated projections are generally more accurate, suggesting that the CBO is effectively using the new information it collects over a short period of time (generally 6â7 months) to improve the quality of its forecasts.
Part II reviews state and local government budget forecasting within the United States. There has been extensive research on these practices extending back at least as far as the 1950s. While it is well known that these governments tend to underforecast their revenue for various reasons, much else remains unknown. This section provides new empirical evidence regarding some of these matters. This part addresses various matters of forecast accuracy, transparency, and bias. It also addresses different types of bias and their likely consequence. Finally, it includes a discussion of the forecasting of especially smallâand poorly resourcedâlocal governments.
In Chap. 8, Emily Franklin, Carolyn Bourdeaux, and Alex Hathaway find that significant variation in forecasting practices, particularly consensus forecasting processes, makes it difficult to assess the accuracy and transparency of government revenue forecasting. They look at the diversity of the revenue forecasting processes across the 50 states between FY2015 and FY2017 and assess the extent to which state forecasts have proven to be accurate and transparent. Their results show that these state forecasts were about as accurate as previous research would lead one to expect. More detailed analyses of several states demonstrate that reported revenue forecasts do not always reflect what the state expects to receive in revenue. They conclude that forecasts exist within institutional and political frameworks that can influence the accuracy and transparency of the forecast.
In Chap. 9, Melissa McShea and Joseph Cordes ask whether states can improve the accuracy of revenue forecasts by using more advanced time series and Bayesian vector autoregression (BVAR ) forecasting methods. Using state revenue data from Virginia, they first estimate baseline forecasts using autoregression (AR ) and vector autoregression (VAR ). They then present a theoretical case for estimating a BVAR model, and present and compare forecasts based on AR, VAR, and BVAR. They conclude that there are gains in forecast accuracy in using the BVAR, but that BVAR is not a panacea for forecasting the extremely volatile corporate income tax revenue series.
In Chap. 10, Thad Calabrese and Daniel Williams examine whether the well-established risk-averse behavior associated with revenue underforecasting also extends to overforecasting expenditures. They argue that appropriated funds go to agencies, so overestimated expenditures may generate unintended agency-level discretion rather than buffer against forecast errors. However, when central budget offices, rather than agencies, control expenditure categories, those categories may be overfunded without creating such discretion. They find initial evidence for this sort of behavior, with governments overfunding centralized accounts when they use program budgeting and more general overfunding when line-item control is used.
In Chap. 11, Geoffrey Propheter examines the link between underforecasting and consequential tax increases through the examination of New York City property taxes. This link is labeled fiscal obfuscation, a form of fiscal illusion, because voters cannot easily see the link between underforecasting and tax increases. He argues that underforecasting leads to ratcheting tax revenue because the underforecast suggests a need for additional revenue, and the revenue actually received becomes the basis for planning for the next cycle. He examines whether the tax is excessively and intentionally underforecasted by the mayorâs budgeting agency.
In Chap. 12, Vincent Reitano examines the practice of small local governmentâs use of judgment-based forecasting rather than using mathematical modeling. He finds that the literature attributes the use of judgmental forecasts in small local governments to limited resources, which may constrain forecast-related employment and software, and to political official preferences. These factors may lead to large forecast errors and may inhibit the use of long-term forecasts for strategic planning. He recommends empirical research into the relationship between forecast errors and government size and qualitative research into government forecasting practices.
In Chap. 13, Daniel Williams and Thad Calabrese investigate current year budget forecast performance for municipal governments in the United States. They find that there is ample research...