Recent experiences have underlined the crucial need to budget for the long term. State budgets have faced significant uncertainty over the past two years. In early 2020, as the COVID-19 pandemic gripped the country, some experts and lawmakers worried about steep and sustained shortfalls for state budgets as sectors of the economy shuttered. The prospect of federal assistance allayed some concerns, but most states were compelled to make decisions and pass budgets while national leaders haggled over aid details. Even now, as the economy recovers, state leaders must consider how to rebuild their economies, allocate federal aid, and plan for the future while facing both short-term concerns such as the trajectory of the virus and variants, and longer-term trends including demographic changes, natural disasters, and technological changes.
Although many factors are beyond a state official’s control, lawmakers can take steps to ensure that they are prepared for action. Sound fiscal planning is the first step. In a memo sent on Nov. 16, 2021, Pew experts advised North Carolina leaders on ways to make more informed long-term budget decisions. The memo offers state lawmakers six fiscal principles to follow when constructing a long-term budget:
These principles are pertinent not only to North Carolina’s leaders, but to all state leaders. Text of the full memo follows.
###
To: Kristin Walker, North Carolina Deputy State Budget Director
From: Sheanna Gomes and Jennifer Janson, The Pew Charitable Trusts
Date: November 16, 2021
Subject: Long-Term Budgeting
In response to your request, this memo discusses long-term budget projections, revenue, and spending forecasts that look multiple years into the future. These projections can help inform key decisions, including whether the state can afford new costs (from increased spending or tax cuts), or whether it needs to adjust to maintain a balanced budget (by cutting spending or increasing revenue).
To serve that purpose, Pew’s research shows that states’ projections should follow several key principles, including:
Even states that don’t produce long-term budget projections often have the building blocks for these analyses in the form of revenue forecasts, caseload forecasts, and agency-specific planning documents. But it’s not enough for states to forecast revenue in one place, Medicaid costs in another, and education expenses in another. To assess overall fiscal health, states should bring these forecasts together to offer a more comprehensive picture.
Forecasts that focus only on a state’s general fund may be too narrow to adequately serve this purpose. Instead, states have gained valuable insights by examining a broad range of revenue sources and spending categories—including all major budget drivers.
A broad approach can highlight important interactions between different funds and spending categories. For example, a December 2020 forecast* written by the Maryland Department of Legislative Services for the state’s Spending Affordability Committee showed that in the short term a shortfall in one of the state’s dedicated education funds, the Education Trust Fund (ETF), would put pressure on the general fund (Page 39). The analysis assumed the ETF’s fiscal year 2020 shortfall would be backfilled by the general fund in the following fiscal year. However, by fiscal 2026 the state expected an unspent balance of nearly $5 billion in the ETF and the Blueprint Fund, another dedicated education account, largely because the governor had vetoed legislation to increase education spending.
Similarly, the most recent Fiscal Accountability Report published by Connecticut’s Office of Policy and Management included a long-term outlook for the state’s special transportation fund (Page 16). The projections showed that the fund’s $168.4 million fiscal 2020 surplus would flip to a $367.5 million deficit by the end of fiscal 2026—a shift the office attributed to the pandemic, reductions in energy prices and fuel consumption, and an overall decline in the economy. The problem was expected to grow, despite the state diverting general fund revenue to the transportation fund, including a portion of sales tax collections.
States’ budget forecasts should look far enough into the future to identify trends and risks. For example, if a major state tax is set to expire in six years, the projections should span at least that amount of time. Going out at least three years is a minimum, so that budget projections in both annual and biennial budget states look beyond the current budget cycle. (The Government Finance Officers Association recommends a minimum of five to 10 years.)
A benefit of looking several years into the future is that it can help policymakers make timely course corrections by bringing attention to potential issues before they become larger problems. For example, in 2012, Maryland’s six-year forecast showed that the growing price of teacher pensions was becoming a major expense for the state, which covered the entire cost. Having this information helped spur legislators to negotiate a cost-sharing arrangement with cities and counties, which were the governing bodies deciding teacher salaries and benefits.
Furthermore, states should acknowledge risks that may develop or grow beyond the forecast period, even if such risks are too speculative to quantify or too distant to have substantial budget effects within the next few years. Long-term demographic trends, natural disasters, technological changes, or climate change could fit in this category. For example, although Florida’s Legislative Budget Commission’s annual Long Range Financial Outlook doesn’t project hurricane costs, it includes a section on “significant risks to the forecast” (Page 41) that discusses various ways hurricanes could cause financial challenges.
One weakness of some projections is that states treat the exercise more like an executive budget proposal—conveying what the governor thinks tax and spending policy should be—rather than beginning by estimating costs and revenue collections absent policy changes. It’s easy for state budgets to appear structurally balanced if, for example, the analysis presumes spending will only grow by 2% a year because that is the governor’s preference. However, a good long-term projection would examine whether it’s realistic that spending will only grow by 2% a year based on the state’s programs and commitments.
To do that, a current services baseline is a helpful analytical framework. Under the current services model, states estimate what they would need to spend to keep a program at a consistent level of service—if their schools continue to have the same teacher-to-pupil ratio, for example. A key reason to use the current services approach is that a budget that is flat in nominal terms doesn’t guarantee that the state will be able to maintain a consistent level of services because costs could rise (or shrink) from factors such as inflation, population growth, or demographic changes. Current service estimates account for these factors to generate realistic spending forecasts.
An exercise completed by Pennsylvania’s Independent Fiscal Office (IFO) as a part of its five-year outlook shows how a current services baseline can substantially affect states’ conclusions about spending demands. The IFO projected expenditure totals using both a current services baseline and a “cost-to-carry” baseline (Page 30). The cost-to-carry method used in the outlook increased projected spending only to the extent it was needed to comply with federal or state mandates, fulfill debt service or pension commitments, or provide care for those under state agency jurisdiction. All other spending captured in the cost-to-carry baseline was kept at the same level as in the first year of the forecast, unlike the current services model.
Starting in 2021-2022, the IFO shows spending growing by 3.1% per year under the current services method and 2.6% per year under the cost-to-carry method. But this small difference compounds over time, to the point that the aggregate difference between the two baselines is nearly $1.2 billion by fiscal 2026. IFO identified this gap despite projecting the state’s population to decline slightly over the study period—in a growing state, the difference would likely be larger.
Although states should start with baseline estimates of revenue and spending absent policy changes, the projections are often most valuable for informing state leaders’ actions if they also include analysis of how policy decisions under consideration would affect the numbers. This analysis can show to what degree proposals to increase revenue or cut spending would improve the state’s structural budget position or whether new initiatives are affordable without risking long-term deficits.
For example, Maryland’s most recent six-year forecast was completed in December 2020 at a time when key fiscal decisions were in flux because of gubernatorial vetoes. The governor had vetoed legislation to significantly increase education funding and to increase taxes on tobacco, digital goods, and digital advertising. To provide useful information despite this uncertainty, the forecasters analyzed the size of the state’s structural deficit (Page 31) and the balance of dedicated education accounts (Page 40) both with and without the legislature overriding the vetoes.
The analysis directly informed lawmakers’ debate on the veto overrides. Supporters of the education funding law pointed out that the analysis showed the dedicated education accounts would (nearly) have enough money to pay for the education funding increase through fiscal 2026, while opponents questioned taking on new spending commitments at a time when the state had an ongoing structural deficit in the general fund.
Maryland’s approach of analyzing multiple scenarios can also help states avoid a question that sometimes complicates long-term budget projections: whether to use a “current law” or a “current policy” baseline. If, for instance, a tax credit is scheduled to expire next year, a current law projection would assume that it does in fact expire (because the scheduled expiration is currently the law), while a current policy projection would assume the tax credit will continue (because offering the tax credit is the state’s current policy until the expiration actually happens). But because analysts can’t be sure what policymakers will ultimately decide—and because policymakers benefit from seeing how different options would affect the state’s bottom line—analyzing both current law and current policy is often preferable to picking between them.
Temporary or one-time revenue and spending can obscure a state’s underlying financial situation. Today, for example, states are enjoying surpluses supported by federal relief, but this temporary revenue doesn’t necessarily mean ongoing revenue can pay for ongoing spending. To estimate states’ structural surplus or deficit, long-term budget projections should isolate and compare ongoing revenue to ongoing spending.
To do so, analysts should separate one-time revenue and spending—items such as money from a legal settlement—from the calculation. But even categories of revenue and spending that are typically considered ongoing can temporarily be higher or lower than normal. For example, state Medicaid costs may be temporarily inflated because of federal requirements to maintain eligibility during the COVID-19 public health emergency. Or, during an economic downturn, revenue falls below the long-term trend. While it’s potentially more complicated than separating obvious one-time items, ideally states’ structural balance estimates are based on an analysis of normal, ongoing revenue and spending growth, removing factors that temporarily may cause the numbers to be higher or lower than the long-term trend.
Pennsylvania, for example, uses this analytical framework for its structural balance estimates, stating (on Page 10) “By definition, the determination of a structural budget surplus or deficit assumes an economy that operates at full employment and excludes all one-time measures such as temporary borrowing, shifting of funds or payment delays.” In other words, their estimates (Page 41) are not based on an economy in recession because in recessions revenue is lower than the long-term trend. They also remove one-time revenue from the estimates, noting, for instance, the $431 million in special fund transfers the state used to balance its fiscal 2021 budget.
As Pennsylvania’s analysis acknowledges, by assuming the absence of a recession in the forecast period, its structural balance estimation “leans optimistic.” With that in mind, an alternative approach is to forecast plausible ranges for ongoing revenues and ongoing spending. For example, the main four-year outlook published by the California Legislative Analyst’s Office in November 2020 showed that major tax revenues were projected to increase by an average of less than 1% each year, while general fund expenditures were expected to average a 4.4% increase each year. But to supplement this primary forecast, they also forecast revenue and spending in more likely and less likely ranges (Page 11). This approach helped the analysts avoid the potential false precision of a single forecast, while providing clear findings about structural balance. Their conclusion: “It is quite unlikely revenues will end up growing fast enough to cover the growth in costs necessary to maintain current levels of government services.”
Policymakers don’t just need to know whether their state’s budget is structurally balanced—they also need to know why or why not. Identifying the factors causing a structural deficit, for example, is a starting point for the state to develop policies to return to long-term balance. Some long-term projections fall short by simply presenting numbers with limited explanation and context, but others include detailed narratives explaining key trends.
In Colorado, the fiscal 2021 long-range financial plan showed that the state’s spending growth was unsustainable without policy changes—budget officials projected that spending would increase by 4.9% a year over the next five years, but that the state’s taxpayer’s bill of rights would allow for increases of only 3%. The plan included a detailed discussion of “key budget drivers,” including factors driving the unsustainable increases.
In the plan, the budget situations of each of the state’s “big six” agencies—departments that account for 84.4% of state spending—were discussed in turn. For example, the discussion for the Department of Health Care Policy & Financing noted that enrollment in the state’s Medicaid program and children’s health plan was expected to increase by as much as 20% by fiscal 2023. The section on the Colorado Department of Transportation pointed out that by leaving the gas tax at 22 cents per gallon for the past 28 years, the state had allowed the tax to lose two-thirds of its purchasing power. And, in addition to the agency-specific analysis, the plan also discussed cross-cutting factors affecting the state’s budget situation, including weak economic growth and an aging population.
As states describe the factors influencing the state’s financial outlook, they should document key assumptions and methodological decisions—acknowledging when and why the eventual numbers could differ from the projections. California’s Legislative Analyst’s Office November 2020 outlook included an appendix describing key assumptions (Page 16). These assumptions included that the state wouldn’t receive additional federal COVID-19 relief funding, that the enhanced federal Medicaid match would expire at the end of the 2021 calendar year, that the state’s COVID-19 response costs in fiscal 2022 would be half of the preceding year, and that lawmakers would allow a tax on managed care organizations to expire as scheduled.
As it turned out, not all the office’s assumptions were correct. The federal government provided significant additional relief in March 2021 in the American Rescue Plan Act and extended the expanded Medicaid match at least through the first three months of 2022. But, by highlighting upfront the importance of these assumptions and the possibility that they would be wrong, the office signaled to policymakers the inherent uncertainty in long-term forecasting and the need to use the numbers with caution.