Revenue Forecasting in Unforecastable Times
August 23, 2022
By Katherine Barrett and Richard Greene, senior advisors to GFRC and principals of Barrett and Greene, Inc.
It’s not news that budget forecasting in states, counties and cities in the United States is a perilous affair. When revenues come in higher than expected, people are happy, and elected officials don’t seem inclined to complain about the difficult work of the revenue officials responsible for predicting the future. But when things move in the opposite direction – and revenues come in below assumptions – that’s an entirely different story, and budgeters are called on to account for the reasons why they made mistakes.
Yet, both situations – revenues coming in above forecasts, and revenues coming in below forecasts – are just two sides of the same coin. Notwithstanding complicated models that states rely on to help predict the future, exogenous factors intervene on a regular basis. No forecaster can build a natural disaster into a formula. Similarly, an unexpected stock market boom can move revenues in the opposite direction, particularly in states like California and Connecticut that rely heavily on capital gains.
At no time in our memory have forecasts been so dramatically incorrect than over the last few years as the COVID-19 pandemic has run amuck through the United States and much of the rest of the world. When the pandemic first began, forecasters from coast to coast projected that revenues would be dropping like a stone. Even when the federal government began to send dollars to these entities, the assumptions were that rainy day funds would be drained and services would be slashed.
On March 19, 2021, the Council on Foreign Relations confidently pronounced that “Many U.S. state and local governments, on the front lines of the response to the coronavirus pandemic, are facing severe budget shortfalls. A distressing combination of dwindling tax revenues, record unemployment, and rising health costs have pushed them to cut back on spending for infrastructure and education—of which states and cities are by far the primary funders.”
As we all know now, such dire predictions didn’t materialize. In fact, less than a year later, according to the National Association of State Budget Officers, “These concerns did not come to pass, mainly because state revenues later performed considerably better than was expected early in the pandemic. This was driven by an influx of federal funds, higher-income workers being relatively insulated from the effects of the COVID-19 recession, the shifting of consumption from services to goods (more commonly taxed at the state level), and the expanded ability to tax online sales . . .”
Politically motivated politicians have taken advantage of this turn of events to complain that the federal government was far too generous to states and localities when it chose to send some $350 billion to them in the American Rescue Plan Act (ARPA). But we’d argue that such complaints miss out on the fact that these dollars were allocated at a time when forecasters were almost universally convinced that those dollars would be critical to keep services flowing.
Even when giant surprises don’t roll through state governments fiscal affairs, forecasting is a complicated matter. Fortunately, there are a handful of approaches that can improve the revenue estimating process, even if errors, large and small are inevitable.
Over the course of seven years, we worked on a series of reports under the rubric of Truth and Integrity in State Budgeting produced by the Volcker Alliance (founded by the famed chair of the United States Federal Reserve, Paul Volcker). That series of reports divided the world of budgeting into five major categories, one of which was budget forecasting.
Evaluations of the states’ skill in that area, rested on “four major indicators to evaluate budget forecasting: the use of consensus revenue estimates, multiyear revenue forecasts, and multiyear expenditure forecasts; and the inclusion of a detailed rationale for revenue growth projections at the time of the original budget,” according to the 2020 installment of the series.
We believe, based on 30 years studying state budgeting, that these four indicators stand up very well, and it’s worthwhile going into a little detail here about each of them:
- Consensus Revenue Estimates. When legislatures and governors’ offices each are working from a different set of revenue estimates the budget process is hobbled. Worse yet, there’s the risk that a governor’s office, say, may inflate revenue figures in an effort to create a budget that appears to be balanced at the beginning of the fiscal year, but falls apart as months pass and reality sets in.
As the Volcker Report states, “A consensus approach ensures that all budgeters base their work on the same projection. While it does not necessarily generate the most accurate forecasts, it can create a more efficient budgeting process that eliminates political squabbles over the amount of money available to support state expenditures.’
- The concept of multiyear revenue and expenditure forecasts is a critical one. It forces disclosure of the effects that a fiscal decision can make beyond the current year. Right now, for example, with surpluses aplenty, there can be a temptation to use one-time money for operating expenses, putting pressure on future years when the cash isn’t coming in. But expenditure and revenue forecasts that go out a few years can reveal what happens when the well runs dry; and perhaps inspire fiscal rectitude.
- The final Volcker criteria, the inclusion of a detailed rationale comes down to a matter of transparency. To quote the Volcker report again, “To help determine the validity of revenue estimates, it is important for states to disclose the methodology used in calculating the figures. For example, without knowing that estimates in energy-producing states such as Wyoming largely depend on severance taxes, the reasoning behind the forecasts is lost.”
The contents of this blog post reflect those of the author, and not necessarily those of the GFRC.