Long-Term Interest Rate Projections and the Federal Debt
Congressional Budget Office (CBO) projections of federal debt, already dire, depend on interest rate forecasts which are likely too low
The Congressional Budget Office (CBO) is a nonpartisan federal agency responsible for providing economic and budgetary analysis to Congress. Its projections are critical for shaping federal fiscal policy, as they serve as the foundation for decisions on government spending, taxation, and debt management. Accurate projections are essential because they influence how policymakers plan for long-term economic stability and respond to fiscal challenges.
In its latest long-term budget outlook, the CBO projects a gradual increase in the budget deficit reaching 8.5 percent of GDP in 2054, while the public debt ratio is forecast to reach 166 percent.
Key to CBO’s long-term budget projections are its long-term projections of interest rates that affect the budget, including rates on various debt instruments issued by the Treasury Department. While the CBO’s interest rate projections for the coming decade show higher rates than over the past 30 years, rates on government securities are only forecast to be about a third of a percentage point higher over the entire 2024–2054 period of the projection compared to the 1994-2023 average (4.2% vs 3.8%).
In this article I explain how the CBO’s underlying assumptions for factors that determine the trajectory of interest rates are likely overly optimistic. These assumptions are often based on outdated economic trends, structural factors with high degrees of uncertainty, or a limited scope of empirical literature.
Such an approach underemphasizes the role of a growing public debt ratio on interest rates, while overemphasizing the role of other structural factors. With a stock of public debt that is set to exceed total GDP in the near term and a debt maturity structure that heightens the risk of upward trending interest rates, the CBO's projections may not accurately reflect the true costs of our current fiscal trajectory.
Factors that Determine CBO’s Projection of Interest Rates
In CBO’s assessment, structural factors—demographics, attitudes toward saving and investment, and the amount of federal debt—largely determine interest rates in the long term. In its latest long-term budget outlook, the agency expects trends in structural factors that have prevailed since the early 1980’s to continue putting downward pressure on interest rates. Specifically, the report cites (1) slower growth of the labor force, (2) more private foreign and domestic savings available for investment, and (3) slower growth of TFP to depress interest rates through 2054.
The agency’s interest rate model lists six underlying factors with categories of research literature for each of the underlying factors (see below). The latest long-term budget forecast lists two factors related to the empirical literature on demographics: (1) labor force growth and (2) private domestic and foreign saving. Given this, the current interest rate model seems to place a lot of weight on the empirical demographics literature when determining future interest rate trends.
Interest Rate Assumptions Underestimate Impact of Debt
While in the CBO model the structural factors listed above put downward pressure on long-term rates, a growing debt ratio (public debt to GDP) puts upward pressure on rates. The empirical literature finds that each percentage point increase in the debt ratio reduces private investment by about 0.23 to 0.24 percentage points (Vanlaer, 2021; Kostarakos, 2022). The literature faces challenges because the U.S. hasn't historically experienced extreme debt-to-GDP levels, limiting the direct applicability of past studies. Other countries with higher debt ratios differ by lacking reserve currency status and safe-haven bonds, meaning the U.S. may encounter different effects. Additionally, debt's impact on private investment could be nonlinear, with stronger effects beyond certain thresholds, making linear models potentially inadequate (Salmon, 2021).
The decline in private investment resulting from debt crowd out reduces the amount of capital per worker and further increases interest rates and the return on capital over time. The CBOs production function typically uses an impact parameter of 2.5 basis points (bps) on the debt to GDP ratio—in other words, for every percentage point (ppt) increase in the debt ratio, the CBO model assumes upward pressure on interest rates of 2.5 bps. This figure is based on the middle estimate of one 2019 CBO working paper which estimated the impact of debt on interest rates to be between 2 and 3 bps (Gamber and Seliski, 2019). In its most recent long-term budget outlook, the CBO estimates an even smaller impact of debt on interest rates, expecting each ppt increase in the debt ratio to raise rates by just 2 bps.
However, a review of the empirical literature, especially newer studies with updated datasets and more comprehensive approaches, suggests that a 2 or 2.5 bps assumption is likely underestimating the impact of debt on long-term rates. Table 1 below highlights some of the findings in the related literature. Early empirical analysis of the impact of public debt on interest rates typically finds estimates ranging from 3 to 4 bps (Laubach, 2003; Engen and Hubbard, 2004). In a 2004 economic analysis, Brookings Institution economists, William Gale and Peter Orszag estimated the impact of budgetary dynamics and savings on interest rates (Gale and Orszag, 2006). The authors find that every 1 ppt increase in the debt ratio raises long-term interest rates by 4.9 bps. Using panel data for 19 developed economies, a 2006 IMF working paper found the simulated and estimated interest rate effects of government debt to be between 2 and 5 bps when using a Fixed Effects model (Kinoshita, 2006).
Table 1: Impact Estimates of One Percentage Point Higher Debt-to-GDP Ratio on Interest Rates
With the onset of the Great Financial Crisis, some economists updated their prior estimates with new datasets and methodology. For example, in a journal article published in 2009, Thomas Laubach presented new evidence on the interest rate effects of budget deficits and debt (Laubach, 2009). Laubach finds that a 1 ppt increase in the projected debt ratio raises interest rates by about 4 bps—slightly higher than his 2003 estimate of about 3.5 bps. A 2010 IMF working paper considering a wide range of country-specific factors for a panel of 31 advanced and emerging market economies found notably large effects of debt on bond yields, with the impact on rates being larger than 5 bps (Kumar and Baldacci, 2010).
In more recent years, economists have been updating prior empirical analysis with new datasets, methodologies, and additional variables. In a long blog post from 2019, budgetary economist Ernie Tedeschi updated and expanded the empirical framework on Francis and Veronica Warnock (Warnock and Warnock, 2009; Tedeschi, 2019). Observing the period 1984-2018, the results of Tedeschi’s framework reveal that each ppt increase in the debt ratio raises the 10-year yield by 4.21 bps, all else equal. In 2022, American Enterprise Institute economists further expanded (1981-2022) and updated Tedeschi’s framework to consider the impact of quantitative easing (Mantus and Warshawsky, 2022). The authors find that a ppt increase in publicly held debt raises long-term rates by 4.5 bps, slightly higher than Tedeschi’s 2019 estimate. Finally, a recently published journal article provides a comprehensive high-frequency identification approach to the relationship between the debt ratio and interest rates, finding an impact of 4.3 bps. These new approaches use updated timeframes—older studies use data covering late 1970’s through mid-noughties, while these studies run from about 1980 through 2019 or 2022. Some of these new approaches measure how interest rates change in a high-frequency window around release (i.e. 24-hour window), while others run regressions on a recursive sample to see how the coefficient changes over time. Newer studies also add additional variables to their models, including growth in the dollar, the elderly population share, and Federal Reserve holdings of treasury securities.
In reviewing the findings of the empirical literature, the CBO’s 2 bps (or 2.5 bps) assumption is likely underestimating the impact of debt on long-term rates. This estimate seems well below the range of 3 to 5 bps commonly found in the literature and well below the 4 bps or higher estimates in the latest empirical studies. Perhaps the 2019 CBO analysis omits the second-round effect that the debt ratio is effectively increasing by more than 1 ppt. In other words, while the CBO accounts for the initial impact of a ppt increase in the debt ratio, the higher interest leads to higher borrowing costs, which in turn contribute to faster debt accumulation. These compounding effects (where debt and interest rates mutually reinforce each other's growth) are not fully captured by models that only measure the immediate impact. In this case, a ppt increase in the debt ratio projected several years out should be associated with an increase in interest rates larger than the one implied by 1 ppt increase in the steady state debt ratio predicted by the model.
Another possible issue with the 2019 CBO paper is its methodology. As Cotton (2024) points out: “Structural approaches, such as those involving vector autoregressions (VAR), rely on a strict formulation of the relationship between deficits/debt and interest rates, which may not be empirically accurate. Running regressions over wide time periods, such as by quarter or year, risks the possibility that confounding factors will drive the results.”
To better align its underlying assumptions with the empirical literature, the CBO production function should use a debt impact parameter of at least 4 bps (not 2 or 2.5 bps).
Other Structural Factors are Overly Dependent on Past Trends and Involve Substantial Uncertainty
As noted above, the latest CBO report on the long-term budget outlook places a lot of weight on demographic factors depressing interest rates in the coming 3 decades. Changes in labor force growth are largely dependent on the demographics research literature, while demographics are also a key factor in private domestic and foreign saving. The cited research places a lot of weight on past demographic trends, largely overlooking the literature on future expectations in demographic changes, and subsequent changes in savings, and labor force participation. CBO acknowledges these challenges and how this contributes to uncertainty in its 2020 working paper (Gamber, 2020): “Forecasting the independent influences of those channels is challenging and contributes to the uncertainty in interest rate forecasts.”
Studies on the impact of demographic changes find that these dynamics have put downward pressure on interest rates since 1980 ranging from a cumulative 90 bps to 130 bps (Rachel & Smith, 2015; Gagnon et al., 2016; Lisack et al., 2017). However, determining whether these trends will continue into the foreseeable future is not an easy task. In fact, the range of estimates forecasting the future impact of demographics on interest rates does not suggest any conclusive consensus in the literature. Lisack et al. (2017) forecasts modest downward pressure on interest rates of 37 bps through 2050, while Carvalho et al. (2016) forecast slightly larger downward effects of 50 bps. On the other hand, other studies forecast demographic changes putting upward pressure on interest rates in the coming decades (Rachel and Smith, 2015). Goodhart and Pradhan (2017) argue that future demographic changes are more likely to raise interest rates as retirees from the Boomer generation spend down their savings.
Aside from demographic structural changes, CBO’s long-term report also cites slower growth of total factor productivity (TFP) as a key factor depressing long-term interest rates. Similar to the literature on demographics, there is a lot of uncertainty in forecasting the impact of changes in TFP on long-term interest rates. Past movements in trend output cannot explain the decline in real interest rates over the entire 1980–2019 period (Rachel and Smith, 2017). Lunsford and West (2019) find that the correlation between real output growth and real interest rates on government debt is variable, weak, and sometimes of the wrong sign. Rachel and Summers (2019) attempt to explain the past decline in interest rates and conclude: “the model suggests that a bulk of the decline in real interest rates is due to factors other than trend GDP growth’.
Short-Term Debt Maturity Structure Amplifies the Cost of Rising Interest Rates
Debt held by the public in the U.S. has a relatively short average maturity of just 71 months (just under 6 years). More notably, 54% of this debt matures and must be rolled over within the next three years. This exposes the federal government to heightened interest rate risks in the short term, as a large portion of outstanding debt will be subject to new borrowing rates in the near future. This makes the U.S. particularly vulnerable to sudden increases in interest rates. When short-term rates spike, a significant portion of the national debt must be refinanced at these elevated rates, which can quickly drive up the cost of servicing the debt.
In September 2024 alone, the Treasury had to roll over $2.56 trillion in maturing securities, with the average interest rate on total interest-bearing debt now exceeding 3.3%. High interest rates thus not only affect new debt issuance and current deficits, but also the cost of rolling over existing debt as it matures, compounding the fiscal impact in both the short and long term. As a result, the government's short debt maturity structure exacerbates the financial strain of rising interest rates. If rates continue to climb, more debt will need to be rolled over at higher costs, further pressuring the federal budget and increasing overall debt service obligations.
Reevaluating Long-Term Risks and Fiscal Assumptions
The CBO's long-term interest rate projections significantly understate the potential risks posed by rising public debt and structural uncertainties. In particular, the 2 bps (or 2.5 bps) assumption used to estimate the impact of rising debt on interest rates appears to be conservative when compared to more recent empirical literature, which suggests the effect is likely closer to 4 bps. This underestimation could lead to an overly optimistic view of future interest rate environments and the corresponding cost of borrowing. Moreover, the CBO’s reliance on demographic trends and total factor productivity growth as drivers of lower long-term rates is fraught with uncertainty. While these factors have historically exerted downward pressure, future projections are far less predictable.
Finally, the short-term maturity structure of the U.S. debt amplifies the risks associated with rising interest rates. This dynamic is not sufficiently reflected in the CBO’s budget models, which may underestimate how quickly higher rates could translate into escalating government costs. A reassessment of the assumptions underpinning these projections is necessary to more accurately reflect the challenges posed by the growing national debt and the heightened sensitivity of the federal budget and broader economy to interest rate fluctuations.
Applying the CBO’s current debt impact parameter of 2 bps, the CBO forecasts that the nominal rate on 10-year treasury securities will be 4.13% in 2034 and 4.44% in 2054. If we instead apply an impact parameter of 4 bps, then the forecast nominal rate on 10-year treasuries rises to 4.47% in 2034 and 5.78% in 2054. Even modest adjustments to the debt impact parameter can lead to significantly higher long-term interest rates, magnifying the fiscal burden and underscoring the need for more cautious assumptions in long-term debt projections.