## wpiea2019139

## Source details

**Canonical URL:** [wpiea2019139](https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019139.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2019/wpiea2019139.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2019/wpiea2019139.pdf.json)

---

### Overview and purpose
- Analyzes the evolution of the United States’ consolidated public sector balance sheet and projects it forward to analyze the fiscal position in a comprehensive manner.
- Leverages rich US data to go beyond government debt, covering the consolidated public sector between 1945 and 2016 and providing intertemporal projections that incorporate the government’s power to tax and promises of future expenses implied by current legislation.
- Expands scope to include state and local governments and public corporations (including the Federal Reserve System), consolidating cross-holdings across sectors.

### Methodology and data
- Descriptive analysis: summarize evolution of stocks of assets and liabilities since World War II (Financial Accounts of the United States – Z.1; IMF Fiscal Monitor (October 2018); IMF estimates on subsoil assets). Latest full-data year: 2016.
- Intertemporal analysis: follows Buiter (1983) to combine data with projections of macroeconomic dynamics and policies to estimate the government’s intertemporal balance sheet; present-value horizon T=50 (50 years).
- Discounting: all flows discounted using the projected effective nominal interest rate on general government debt.
- Medium-term forecasts (2018–23): April 2018 IMF WEO database; long-term (2024+) based on a neoclassical growth model informed by demographic evolution.
- Annexes detail assumptions, VAR model for loan losses, and asset valuation methods (oil and gas, coal/metals/other minerals).

### Key quantitative findings (2016 and selected years)
- General government gross debt: 107 percent of GDP in 2016.
- Public sector net financial worth: -101 percent of GDP in 2016.
- General government nonfinancial assets: 84.5 percent of GDP, including:
  - natural resource assets: 10.7 percent of GDP
  - fixed assets: 73.8 percent of GDP
- General government financial assets: about 26 percent of GDP, including:
  - student loans: 6 percent of GDP
- Pension liabilities (accumulated defined benefit obligations): 54 percent of GDP.
- Federal Reserve holdings of Treasury securities and GSE-issued debt and securities: 23 percent of GDP.
- State and local governments: spend about 14 percent of GDP and employ one-eighth of the workforce.
- Nominal GDP cited (2016 context): 18,624.5 ($bn).
- Consolidated public sector net worth over time:
  - Peaked at 35 percent of GDP in 1980.
  - Reached -27 percent of GDP in 2016.
- Static Public Sector Net Worth (percent of baseline GDP) – selected values:
  - 2017: -18.3
  - 2018 (baseline / adverse / severely adverse): -21.4 / -27.7 / -35.8
  - 2019 (baseline / adverse / severely adverse): -24.7 / -35.9 / -47.8
  - 2020 (baseline / adverse / severely adverse): -27.9 / -40.5 / -54.1
- Assets (percent of baseline GDP) – general government:
  - 2017: 109.4; 2018: 106.0; 2019: 102.5; 2020: 100.6
- Nonfinancial assets (percent of baseline GDP) – general government:
  - 2017: 83.5; 2018: 80.5; 2019: 77.4; 2020: 75.6
- Liabilities (percent of baseline GDP) – general government:
  - 2017: 127.7; 2018: 127.4; 2019: 127.2; 2020: 128.5
- Debt securities (percent of baseline GDP):
  - 2017: 96.9; 2018: 97.3; 2019: 97.7; 2020: 99.2

### Intertemporal sustainability and fiscal-adjustment measures
- Baseline intertemporal results (2016):
  - Present value of primary balances: -161 percent of GDP.
  - IFNW* (Intertemporal Financial Net Worth plus natural resource assets): -252 percent of GDP.
- Sustainability conclusion: under baseline assumptions, current fiscal policies in the US are not viable in the long term.
- Required adjustments (permanent from 2019):
  - Permanent 1 percent of GDP increase in the primary balance improves IFNW* by 49 percent of GDP.
  - Cutting the gap between baseline IFNW* and zero-IFNW* in half requires an adjustment of 2.6 percent of GDP.
  - Fully canceling the negative IFNW* requires a permanent adjustment of 5.1 percent of GDP.
- Timing and macro feedback:
  - An adjustment starting in 2024 would need to be 0.7 percent of GDP higher than one starting in 2019.
  - Sudden large adjustments risk recession and could further reduce public sector net worth; gradual adjustment preferred.
- Comparative debt-focused metrics:
  - IMF debt-sustainability framing: medium-term general government primary surplus of about 1¼ percent of GDP could lead to a durable decrease in the public debt ratio (≈ a fiscal adjustment of about 3 percent of GDP by 2027).
  - CBO estimates: permanent fiscal adjustment of 1.9 percent of GDP from 2019 to keep federal debt-to-GDP ratio in 2048 equal to current level; 3 percent of GDP to decrease it to its 50-year average by 2048.
  - Paper’s calculation: a permanent 3 percent of GDP adjustment starting in 2019 would bring the present value of primary balances close to zero but not fully resolve government employee pension gaps.

### Assumptions and caveat on interest-growth differential
- Long-term nominal growth: 3.8 %
- Long-term discount rate / effective long-term interest rate for general government debt: 4.4 %
- CBO projection for interest rates on federal government debt: 4.1%
- Assessment relies on positive future interest-growth differentials; if interest rates remain below GDP growth, solvency concerns may not apply, but analysis takes a cautious stance given uncertainty.

### Stress-testing the public sector balance sheet (2018–20 scenarios)
- Scenarios: Baseline, Adverse, Severely Adverse (aligned with Federal Reserve’s February 2018 supervisory scenarios).
- Adverse key shocks (2018): average Treasury yields cut in half; equity prices fall by 25 percent; real estate prices fall about 10 percent; economy picks up in 2019–20.
- Severely Adverse key shocks: real GDP decreases by 6.3 percent in 2018 and by 0.6 percent in 2019; unemployment climbs to 10 percent in 2019; short-term rates fall to zero while long-term rates stay unchanged; equity prices fall by more than 60 percent in 2018; real estate prices fall by 25 percent; recovery begins in 2020.
- Fiscal-flow impacts:
  - Revenue-to-GDP ratio temporarily decreases by 2.5 / 4.0 percentage points (Adverse / Severely Adverse) before slowly reverting to baseline.
  - Non-age-related spending held constant in nominal terms, implying higher expenditure-to-GDP ratios under shocks.
- Asset revaluation effects:
  - Nonfinancial assets fall by 2.5 / 5.6 percent of baseline GDP in 2020 (Adverse / Severely Adverse).
  - State and local pension fund assets drop by 3.4 / 7.3 percent of baseline GDP in 2020 (Adverse / Severely Adverse).
  - Static net worth projected decreases by 13 / 26 percent of baseline GDP by 2020 (Adverse / Severely Adverse); asset revaluation explains about 8 / 15 percent of GDP of the total decrease.
- Loan losses and GSEs:
  - Federally-held student loans: about 4.0 / 6.5 percent of the portfolio not paid back over 2018–2020, leading to a portfolio loss of 0.1 / 0.3 percent of baseline GDP (Adverse / Severely Adverse).
  - Mortgage portfolio held by GSEs: cumulative losses of 0.4 / 0.6 percent of baseline GDP over three years (Adverse / Severely Adverse).
  - FHFA estimate: potential incremental Treasury draws by Fannie Mae and Freddie Mac of about $100bn over two years under a severely adverse scenario (0.4 percent of baseline GDP).
  - Aggregate loan-loss contribution to public sector net worth decline: -0.4 / -0.9 percent of baseline GDP (Adverse / Severely Adverse).
- Debt accumulation by 2020 relative to baseline: additional debt of 2.4 / 8.7 percent of baseline GDP (Adverse / Severely Adverse).
- Memorandum: Nominal GDP ($bn)
  - Baseline: 2017 = 19391; 2018 = 20438; 2019 = 21577; 2020 = 22440
  - Adverse: 2018 = 19386; 2019 = 19908; 2020 = 20849
  - Severely adverse: 2018 = 18535; 2019 = 18725; 2020 = 19699

### Public-sector structure, cross-holdings, QE, and propagation channels
- Two public-sector spheres distinguished by cross-holding relationships:
  - Federal sphere: federal government, GSEs (part under federal conservatorship), federal government employee pension funds, and the Federal Reserve.
  - State and local sphere: subnational governments and state and local government employee pension funds.
- Main cross-holdings (channels for fiscal risk propagation):
  - Treasury securities, GSE-backed securities and GSE-issued debt held by the central bank: 23 percent of GDP.
  - Claims by defined benefit pension funds on sponsoring government to cover unfunded pension liabilities: 20 percent of GDP.
  - Treasury securities held by federal government employee retirement funds: 10 percent of GDP.
  - Treasury and GSE-backed securities held by state and local governments: 6 percent of GDP.
- Quantitative easing (QE) 2008–14:
  - QE episodes: QE-1 (Nov'08-Mar '10); QE-2 (Nov '10 -Jun '12); QE-3 (Sep '12 -Oct '14).
  - From a consolidated public sector perspective, QE did not expand the public sector balance sheet size because Fed purchases were largely claims on other public sector entities; QE shifted interest recipients from the public to the Fed and shortened maturities at the public sector level.
- Federal Reserve:
  - QE expanded the Fed’s balance sheet materially, but consolidated public-sector asset-side impact was negligible; Fed assets overwhelmingly claims on other public sector entities.
  - Fed liabilities include reserve deposits issued in exchange for long-term securities purchased; baseline assumes QE “normalization plan” beginning late 2017 with specified unwinding amounts.

### GSEs, pension funds, and vulnerabilities
- GSEs and mortgage exposures:
  - Mortgage assets held by GSEs grew from 1 percent in 1960 to more than 30 percent in the early 2000s.
  - Pre-crisis, GSEs invested in riskier "non-agency MBS"; by mid-2008 Fannie Mae and Freddie Mac faced large losses and entered conservatorship; cumulative draws on the Treasury over 2008–2011: $187.5 billion.
  - Conservatorship led to reduction and recomposition of investment portfolios; private-label securities almost disappeared.
- Government employee retirement funds:
  - Aggregate pension fund assets significantly smaller than liabilities; general government covers shortfall.
  - Federal pension funds invest exclusively in Treasury securities; federal pension liabilities backed by federal promise.
  - State and local pension funding status volatile and heterogeneous; examples:
    - Wisconsin: surplus of 4.3 percent of state GDP.
    - Illinois: gap of 27 percent of state GDP.
  - Methodological sensitivity: using market-valuation techniques could materially increase reported unfunded liabilities (Rauh (2017) suggests tripling reported unfunded liabilities relative to current reporting).

### Long-term projections (beyond 2024) and demographic impacts
- Demographics:
  - Share of population over 60 increases from 20 percent in 2015 to 30 percent in 2075.
  - Aging erodes labor’s contribution to growth.
- Growth and price assumptions:
  - Real GDP growth: 1.9 percent in 2020 to 1.8 percent in 2075 under constant TFP growth.
  - Nominal GDP deflator: 2 percent.
  - Implicit interest rate assumed to gradually reach 4.4 percent within fifteen years.
- Long-term fiscal costs related to aging:
  - Public health expenditure projected to increase from about 9 percent of GDP in 2017 to 18 percent of GDP by 2070.
  - Pension costs projected to increase from 8.3 percent of GDP in 2017 to 10.3 percent of GDP by 2070.
  - Social Security expected to increase from 4.9 percent of GDP to 6.3 percent of GDP within 30 years.
- Revenue and non-age-related expenditure assumptions:
  - Federal revenue expected to increase by 2.6 percent of GDP from 2024 to 2048 and remain constant thereafter.
  - Primary non-age expenditure assumed to decline by 1.1 percent between 2018 to 2028 and remain constant thereafter.
  - Subnational flows implicitly held constant relative to GDP in the long term.

### Sensitivity to discount rate and key robustness findings
- Discount-rate sensitivity (impact on 2017 present values, percent of 2017 GDP; selected entries):
  - +10 basis points in 2018: impact on 2017 PV of primary balance +0.2; impact on 2017 PV of revenue -1.7.
  - +100 basis points in 2018: +1.6; -16.4.
  - +10 basis points to LT rate: +2.2; -15.6.
  - +100 basis points 2018–2067: +41.2; -353.2.
- Illustrative effects:
  - A one-shot, one-point increase in 2018 would diminish the 2017 present value of revenue by about 16 percent of 2017 GDP and increase the 2017 present value of primary balance by 1.6 percent of 2017 GDP.
  - A permanent extra point over the whole 50-year period would decrease the present value of revenue by some 350 percent of GDP, while improving the present value of primary balance by 41 percent of GDP.

### VAR model for loan losses (ANNEX 2) — model and outcomes
- Quarterly data from 1984 onward for "Net Loan Losses to Average Total Loans for Banks"; explanatory variables: change in unemployment rate, change in national house price index, change in real 30-year mortgage interest rate, plus lags and quarterly dummies.
- Key regression features:
  - Dependent-variable 1st lag coefficient: 0.89; p-value 0.00.
  - Change in unemployment contemporaneous coefficient: 0.11; p-value 0.04.
  - Pct. change in house prices 2nd lag coefficient: -6.29; p-value 0.00.
  - Number of observations: 132; R-squared: 0.97.
- Scenario calibration:
  - Applying coefficients to DFAST scenarios yields projected loan loss ratios.
  - Severely adverse scenario: peak loan losses as share of total loans slightly above those in the GFC.
  - Severely adverse scenario: total loan losses between 2018–20 would be USD 89 billion higher than under the baseline.
  - Historical comparator: USD 187.5 billion in total actual GSE drawings on the Treasury as of end-2016.

### Policy-relevant observations, reform scenarios, and implications
- The public sector balance sheet provides a more comprehensive picture than debt alone: it incorporates nonfinancial assets, financial assets, public corporations, and implicit liabilities.
- QE is a public-sector-level debt-management operation that shifted interest-rate and maturity exposures among public entities rather than expanding consolidated public-sector assets.
- Fiscal risks have shifted since the crisis: GSEs’ balance sheets shrank and became less risky under conservatorship, while state and local pension funding shortfalls emerged as a major fiscal fragility.
- IMF-recommended policy directions and illustrative packages:
  - Reform social security and increase federal revenue-to-GDP ratio.
  - Illustrative federal tax package (2019): carbon tax, broad-based 5 percent VAT, higher fuel excises — could yield more than 70 percent of GDP of present-value primary balances over 50 years.
  - Illustrative social security reforms (2019): chained CPI for indexation, subject high earnings to payroll tax, accelerate increases in retirement age — could improve present value of primary balances by more than 50 percent of GDP.
- Stress-test implications:
  - In a severe shock, balance-sheet effects outside debt can reduce public sector net worth almost double the expected accumulation of debt.
  - Emerging fiscal risks include the fast-growing federal student-loan portfolio.
- Suggested further analysis:
  - Granular assessment of state and local pension funding status under alternative discount rates.
  - More detailed mortgage and student-loan loss modeling.
  - Regular repetition of balance-sheet stress tests to update baselines and track evolving fiscal environment.

### Data, valuation, and aggregation conventions (ANNEX 3 highlights)
- Natural-resource valuation:
  - Oil and gas: production from Rystad database (government-owned fields), price path from WEO, costs from Rystad; cash flows over 85-year horizon; discount rate for NPV: 4.5%.
  - Coal, metals, other minerals: World Bank estimates and USGS data used; interpolation for missing years; convert to current USD using WEO commodity price index; prorate by government ownership.
- Public corporations and central bank data drawn from Fed Financial Accounts (Z.1) and Central Bank Survey; consolidation removes intra-public-sector cross-holdings (Fed holdings of general government securities; federal deposits at the Fed; GSEs’ and pension funds’ holdings of government securities; government holdings of FPC equity; claims of pension funds on governments).
- Loan-loss projection approach: VAR model on banking-sector net loan losses, with GSE losses scaled by factor 1/3.

*Source: Introduction and subsequent sections (wpiea2019139).*

### Introduction ...........................................................................................................

### Introduction

### Overview and purpose
- The paper analyzes the evolution of the United States’ public sector balance sheet and projects it forward to analyze the fiscal position in a comprehensive manner.
- Balance sheets summarize all assets and liabilities and provide a measure of solvency; the analysis leverages rich US data to go beyond the usual focus on government debt.
- The analysis covers the consolidated public sector balance sheet between 1945 and 2016 and provides intertemporal projections incorporating the government’s power to tax and promises of future expenses implied by current legislation.

### Methodology and data
- Descriptive analysis: summarize evolution of stocks of assets and liabilities since World War II.
- Intertemporal analysis: follows Buiter (1983) to combine data with projections of macroeconomic dynamics and policies to estimate the government’s intertemporal balance sheet.
- Scope: expands prior work by including state and local governments and public corporations (including the Federal Reserve System), consolidating cross-holdings across sectors.
- Data sources: Financial Accounts of the United States – Z.1, IMF Fiscal Monitor (October 2018), IMF estimates on subsoil assets; latest year with full data availability is 2016. Annexes detail assumptions and data sources.

### Key quantitative findings (selected)
- General government gross debt: 107 percent of GDP in 2016.
- Public sector net financial worth: -101 percent of GDP in 2016.
- General government nonfinancial assets: 84.5 percent of GDP, including:
  - natural resource assets: 10.7 percent of GDP
  - fixed assets: 73.8 percent of GDP
- General government financial assets: about 26 percent of GDP, including:
  - student loans: 6 percent of GDP
- Pension liabilities (accumulated defined benefit obligations): 54 percent of GDP (asset position significantly inferior to liabilities).
- Federal Reserve holdings of Treasury securities and GSE-issued debt and securities: 23 percent of GDP.
- Subnational fiscal scale: state and local governments spend about 14 percent of GDP and employ one-eighth of the workforce.

### Intertemporal sustainability result
- Under baseline assumptions, current fiscal policies in the US are not viable in the long-term.
- To keep all explicit promises without overburdening future generations, the government needs either to raise an additional 2.6 percent of GDP in revenue per year or reduce some implicit promises to current generations.

### Assumptions and caveat on interest-growth differential
- Assessment of fiscal sustainability rests on the assumption of positive future interest-growth differentials.
- If interest rates remain below GDP growth rates in the long run, sovereign solvency concerns may not be warranted; however, given uncertainty and potential high costs of underestimating interest rates, the analysis takes a cautious approach.

### Risk exposures and stress-testing insights
- Historical analysis shows public sector net worth has been affected by shocks to asset prices, housing market conditions, and growth.
- Projected asset values under different shock scenarios quantify risk exposures across assets.
- Potential loss magnitudes:
  - Mortgage portfolios of major GSEs and federal student loans warrant vigilance.
  - Losses through equity portfolios of state and local government pension funds dwarf the above exposures.
- Cross-holdings within the public sector represent channels for fiscal risk propagation. Main cross-holdings include:
  - Treasury securities, GSE-backed securities and GSE-issued debt held by the central bank: 23 percent of GDP.
  - Claims by defined benefit pension funds on sponsoring government to cover unfunded pension liabilities: 20 percent of GDP.
  - Treasury securities held by federal government employee retirement funds: 10 percent of GDP.
  - Treasury and GSE-backed securities held by state and local governments: 6 percent of GDP.

### Policy-relevant observations and implications
- The public sector balance sheet provides a more comprehensive picture of fiscal position than general government debt alone; debt securities do not capture nonfinancial assets, financial assets, public corporations, or implicit liabilities.
- Quantitative easing (QE) increased the Fed’s balance sheet and interest rate exposure, and increased federal government average debt maturity; from a consolidated perspective, QE did not expand the public sector balance sheet and shifted interest rate risk from the federal government to the central bank.
- Aggregation can obscure limited risk-sharing mechanisms: risk sharing between states is limited, and federal insurance for municipalities against local/regional shocks is limited; granular analysis is therefore important for state and local governments.
- Accounting conventions matter: unfunded defined benefit pension obligations are significant and sensitive to discount rate methodology; social security payments, while not contractual, resemble debt service payments economically and can be changed over time.

### Conceptual caveats about using net worth
- Net worth is a measure of policy space, not a policy objective per se.
- Using policy space to respond to shocks can be welfare enhancing even if it reduces public sector net worth (e.g., countercyclical stimulus during the global financial crisis reduced public sector net worth while supporting private sector net worth).
- Policy analysis should account for the government’s ability to affect prices, exchange rates, interest rates, and economic activity; changes in balance sheet size/composition may affect these variables.
- Cyclicality of asset prices should be considered to avoid procyclical expenditures.

### Structure of the paper
- Section II: builds the static public sector balance sheet for 2016, discusses data sources and limitations.
- Section III: analyzes the evolution of the public sector balance sheet since the end of World War II, consolidated and by subsector, with focus on the past fifteen years and the global financial crisis.
- Section IV: estimates and analyzes the public sector intertemporal net worth.
- Section V: presents results of a fiscal stress test applied to the public sector balance sheet.

*Source: Introduction (wpiea2019139).*

### 2016. While Table 1.a. adheres to the GFS classification of subsectors, Table 1.b. breaks

### wpiea2019139 - 2016. While Table 1.a. adheres to the GFS classification of subsectors, Table 1.b. breaks

### Overview
- The analysis distinguishes two public-sector "spheres" based on cross-holding relationships:
  - Federal sphere: federal government, GSEs (part under federal conservatorship), federal government employee pension funds (asset/liability mismatches guaranteed by the federal government), and the Federal reserve (holds a significant portfolio of Treasury and GSE-backed debt securities).
  - State and local sphere: subnational governments and state and local government employee pension funds (asset/liability mismatches guaranteed by their respective sponsoring governments).
- Public sector cross-holdings are not consolidated in Figure 1.

### Evolution of the consolidated public sector (1945–2016)
- Public sector net worth:
  - Peaked at 35 percent of GDP in 1980.
  - Reached -27 percent of GDP in 2016.
  - Since the 1980s the decline mainly reflects the federal government’s position; state and local government net worth has fluctuated around a level of 40 percent of GDP.
- General government net savings have been persistently negative since the 1960s; only the late 1990s consolidation improved net worth through fiscal policy.
- Valuation effects explain a large share of changes in net worth:
  - Nonfinancial asset prices increased during the high-inflation 1970s.
  - Corporate equity valuation changes affected the balance sheet positively except after the dot-com bubble and during the global financial crisis.
- Balance sheet size and composition:
  - Financial assets have grown in importance since the early 1980s; nonfinancial assets declined slightly as a share of GDP.
  - Financial assets now largely consist of loans, debt securities, equities, and other assets, totaling about 90 percent of GDP (as of 2016).
- Nominal GDP cited: 18,624.5 ($bn).

### Cross-holdings and quantitative easing (QE)
- Cross-holdings:
  - Have increased over time, particularly during financial turmoil.
  - Cross-holdings of Treasury securities and GSE-issued debt have more than tripled over fifty years, held by GSEs, pension funds, state and local governments, and the Federal Reserve.
- Federal Reserve intervention and QE:
  - Fed purchases of GSE debt, GSE-backed securities, and Treasury securities occurred over 2008–14.
  - QE episodes labeled:
    - QE-1 (Nov'08-Mar '10)
    - QE-2 (Nov '10 -Jun '12)
    - QE-3 (Sep '12 -Oct '14)
  - The Fed initiated an unwinding ("normalization plan") in late 2017.
  - At the consolidated public sector level, QE did not increase the consolidated public sector balance sheet size because Fed assets bought under QE were primarily claims on other public sector entities.
  - QE instead shifted interest recipients from the public to the Fed and shortened maturities at the public sector level.

### General government: nonfinancial assets
- Stock of general government nonfinancial assets has averaged around 75 percent of GDP since 1945, with ownership shifting from federal to state and local governments.
- Federal fixed assets ratio to GDP has continuously declined since the mid-1970s, with prolonged low or negative net capital formation.
  - Average age of total federal government assets increased from 14 years in the 1960s to 24 years in 2016.
- State and local government fixed assets ratio to GDP has almost doubled since 1945.
- Structures and buildings form the bulk of general government nonfinancial assets, most owned by state and local governments.
- Subsoil (natural resource) assets account for much short-term fluctuation in the nonfinancial asset portfolio.

### General government: financial assets and liabilities
- Federal government financial assets declined as a share of GDP until 2008.
- Student loans:
  - Federal government holdings of student loans expanded substantially after the financial crisis.
  - Student loans reached 5.9 percent of GDP in 2017.
  - More than three quarters of all student loan debt is now held by the federal government.
- TARP and the bailout of General Motors had relatively small and temporary balance-sheet impacts.
- State and local governments’ financial assets have grown substantially since the 1970s and include debt securities, equities, and mortgages; a large share are claims on other public sector entities.
- General government liabilities are mostly debt securities, loans, and guarantees to defined benefit pension funds. Treasury debt securities have more than doubled since 2000.
- State and local liabilities have been volatile due to guarantees to retirement funds; strong fiscal rules constrain deficits and debt issuance but not pension benefits.

### Government employee retirement funds
- Aggregate government employee pension fund assets are significantly smaller than liabilities; general government covers the shortfall.
- Federal pension funds:
  - Guarantees to federal government pension funds have been shrinking since the early 1980s following the closing of CSRS to new entrants and the replacement with FERS (legally obliged to remain fully funded, with federal government guarantee if necessary).
  - Both CSRS and FERS invest exclusively in Treasury securities; all federal pension benefit liabilities are backed by a federal government promise.
- State and local pension funds:
  - Funding status has been volatile; liabilities have consistently increased while assets (mainly equities) have experienced large swings.
  - During the late 1990s stock-price boom, pension funds became overfunded, yielding negative claims on governments; the global financial crisis reversed this.
  - Cross-state heterogeneity in funding status (funding status measured as percent of state GDP):
    - Wisconsin: surplus of 4.3 percent of state GDP.
    - Illinois: gap of 27 percent of state GDP.
  - Variation in returns across funds contributes to heterogeneity; Shoag (2014) reports a within-year standard deviation of returns across state pension plans of 2 to 3 percentage points and a cross-sectional standard deviation in cumulative 20-year returns of nearly 100 percentage points.
- Methodological note:
  - Pension liability numbers use Federal Reserve financial accounts and discount-rate assumptions guided by US government accounting standards (long-term expected rates of return on plan assets); Rauh (2017) suggests market-valuation techniques would triple reported unfunded liabilities relative to current reporting.

### The Federal Reserve System
- Pre-2008: Fed balance sheet shrank as a percentage of GDP until the early 1990s and stayed roughly constant thereafter.
- Global financial crisis effects:
  - 2008: Increase in Fed lending as banks accessed the discount window; Fed intervened in mortgage-backed securities markets.
  - QE over 2008–2014 expanded the Fed’s balance sheet materially, but the consolidated public sector balance sheet saw negligible asset-side impact because QE purchases were mainly claims on other public sector entities.
  - QE is interpreted at the public-sector level as a debt management operation and a transfer of interest recipients from the public to the Fed.
- Fed asset composition overwhelmingly consists of claims on other public sector entities; Fed liabilities include reserve deposits issued in exchange for long-term securities purchased.

### Government-Sponsored Enterprises (GSEs)
- GSEs (notably Freddie Mac and Fannie Mae) and GSE-backed mortgage pools have grown considerably; represent one of the largest post-war government interventions in the U.S. economy.
- Mortgage assets held by GSEs:
  - Grew from 1 percent in 1960 to more than 30 percent in the early 2000s.
- Pre-crisis investment behavior:
  - In the 2000s GSEs invested in riskier "non-agency MBS" (securities issued by private investment banks backed by subprime mortgage pools).
  - Fannie Mae and Freddie Mac owned a portfolio of over $300 billion of corporate mortgage-backed securities by the end of (text ends).

*Source: IMF staff calculation and related figures and tables from the provided content.*

### 2007. As for their core credit guarantee business, while GSEs can only guarantee mortgages

### wpiea2019139 - 2007. As for their core credit guarantee business, while GSEs can only guarantee mortgages

### GSE credit exposure, losses, and conservatorship
- Between 2003 and 2007, loans with loan-to-property value ratios above 90 percent grew from 7 to 16 percent of total newly purchased loans for Fannie Mae.
- When the subprime crisis hit and commercial banks froze mortgage investments and securitization, GSEs increased guarantee activities to preserve access to affordable housing and liquidity on the secondary housing market.
- In the second half of 2007 and first half of 2008, Fannie Mae and Freddie Mac posted very significant losses due to credit losses on guaranteed mortgages and mark-to-market losses on the risky investment portfolio.
- By mid-2008, though technically solvent on regulatory book values, Fannie Mae and Freddie Mac were arguably insolvent on an economic basis.
- September 2008: Federal Housing Finance Agency (FHFA) appointed conservator; senior preferred stock purchase agreements signed with the US Treasury, permitting draws on Treasury cash in exchange for senior preferred stock and, since 2012, remitting all profits to the federal government ("full income sweep").
- The agreements planned wind-down of the firms’ investment portfolios; investment portfolios were reduced and recomposed, with private-label securities almost fully disappearing from the balance sheets.
- Cost to federal government over 2008-2011: Fannie Mae and Freddie Mac drew up a cumulative $187.5 billion on the Treasury.
- Since the conservatorship period both firms returned to posting positive profits, remitted to the Treasury under the agreements.

### Impact of the 2008–09 global financial crisis on the public sector balance sheet
- Crisis led to rapid and lasting accumulation of debt liabilities while public sector asset stock slightly decreased.
- Issuance of treasury and municipal securities represented more than 30 percent of GDP over four years (2008–11).
- Other public sector entities’ holdings increased by 7 percentage points of GDP over 2008–11, mostly due to the first waves of quantitative easing.
- Fast reduction in stock of mortgage assets: -7 percentage points of GDP between 2008 and 2011.
- Federal government student portfolio "tripled" over the same period.
- Federal government purchases under TARP: more than $400 billion of toxic equities and debt securities purchased; all holdings ultimately sold and proceeds by end of program in 2014 overcame total purchases; TARP impact on balance sheet became hardly visible within three years.
- Crisis-related valuation effects:
  - 2008 revaluation effects led to a 4 percent of GDP drop in total corporate equity assets of state and local pension funds; losses were quickly recovered when stock market prices picked up.
- Disposal of federal nonfinancial assets over past twenty years reached a cumulative ½ percent of GDP.

### Intertemporal public sector balance sheet: methodology and baseline results
- Intertemporal balance sheet incorporates present values of future revenue and primary expenditure flows computed over a 50-year horizon (T=50).
- All flows discounted using the projected effective nominal interest rate on general government debt.
- Medium-term forecasts (2018–23) taken from April 2018 IMF WEO database and incorporate Tax Cuts and Jobs Act (2017) effects.
- Long-term (2024 and beyond) output projections based on a neoclassical growth model informed by demographic evolution.
- Long-term fiscal forecasts driven by demographic impact on expenditure trends (lowering the primary balance by 10 points of GDP by 2067), partly offset by expected long-term evolution of federal revenue (+2.6 points of GDP by 2067) and non-age-related expenditure (-1.3 percent of GDP by 2067).
- Assumed long-run effective nominal interest rate for general government debt larger than nominal GDP growth:
  - Long-term nominal growth: 3.8 %
  - Long-term discount rate / effective long-term interest rate for general government debt: 4.4 %
  - CBO projection for interest rates on federal government debt: 4.1%
- Excluding fixed nonfinancial assets and including natural resource assets, define IFNW* = IFNW + natural resource assets.
- Baseline (no-policy-change) intertemporal results (2016):
  - Present value of primary balances: -161 percent of GDP.
  - IFNW*: -252 percent of GDP.
- Interpretation: under current policies in the baseline, the US fiscal position is unsustainable.

### Assessing fiscal adjustment needs (baseline-based estimates)
- A permanent 1 percent of GDP increase in the primary balance from 2019 onwards would improve IFNW* by 49 percent of GDP.
- Cutting the gap between baseline IFNW* and zero-IFNW* in half requires an adjustment of 2.6 percent of GDP (permanent from 2019).
- Fully canceling the negative IFNW* would require a permanent fiscal adjustment of 5.1 percent of GDP from 2019 onwards.
- Timing and macro feedback:
  - Postponing adjustment increases cost: an adjustment starting in 2024 would need to be 0.7 percent of GDP higher to ensure fiscal sustainability.
  - Sudden large adjustments risk causing recession and could further reduce public sector net worth; macro-fiscal feedback implies adjustment should be gradual.
- Comparison with debt-focused metrics:
  - IMF’s debt sustainability analysis: medium-term general government primary surplus of about 1¼ percent of GDP could lead to a durable decrease in public debt ratio; this corresponds to a fiscal adjustment of about 3 percent of GDP by 2027.
  - CBO estimates:
    - To keep federal debt-to-GDP ratio in 2048 equal to its current level, a permanent fiscal adjustment of 1.9 percent of GDP from 2019 onwards would be necessary.
    - For federal debt-to-GDP ratio to decrease to its 50-year average by 2048, a permanent fiscal adjustment of 3 percent of GDP would be necessary.
  - According to the paper’s calculations, a permanent 3 percent of GDP fiscal adjustment starting in 2019 would bring the present value of primary balances close to zero but would not fully resolve gaps related to existing government employee pensions.

### Policy options and quantified reform scenarios
- Fiscal adjustment options recommended by IMF (2018a) include reforming social security and increasing the federal revenue-to-GDP ratio.
- Illustrative long-term policy packages (based on CBO estimates expanded into the long term):
  - A federal tax package adopted in 2019 that includes the creation of a carbon tax, the creation of a broad-based 5 percent value-added tax, and an increase of excise taxes on fuels could yield more than 70 percent of GDP of present-value primary balances over 50 years.
  - A social security reform package including:
    - use of chained inflation for indexation of social security benefits from 2019,
    - submission of high earnings to social security payroll tax,
    - acceleration of planned increases in retirement age,
    could improve the present value of primary balances by more than 50 percent of GDP.

### Stress-testing the US public sector balance sheet
- Fiscal stress test methodology projects the static balance sheet three years forward under baseline and shock scenarios; assumes no countercyclical fiscal policy (no policy change).
- Two adverse shock scenarios aligned with the Federal Reserve’s February 2018 supervisory scenarios for 2018–20: "Adverse" and "Severely adverse".
- Example macro outcome under the Adverse scenario:
  - Real GDP declines by 2.1 percent in 2018.
  - Unemployment rate rises to 7 percent in the scenario (full set of scenario outcomes continue beyond excerpt).

*Source: IMF staff calculations; excerpt from wpiea2019139 (PDF chapter/section).*

### 2019. Average yields on Treasury securities are cut in half in 2018, while equity

### wpiea2019139 - 2019. Average yields on Treasury securities are cut in half in 2018, while equity prices fall by 25 percent, and real estate prices by about 10 percent. The economy starts to pick up in 2019 and 2020, with a return to positive growth, a slow decline in unemployment, and the reversal of most of the drop in Treasury yields and equity prices.

### Stress scenarios and macroeconomic trajectories
- Baseline / Adverse / Severely Adverse scenario descriptions and key shocks:
  - Baseline: return to positive growth in 2019 and 2020, slow decline in unemployment, reversal of most of the drop in Treasury yields and equity prices.
  - Adverse scenario (summary of major shocks in 2018):
    - Average yields on Treasury securities are cut in half in 2018.
    - Equity prices fall by 25 percent.
    - Real estate prices fall by about 10 percent.
    - Economy starts to pick up in 2019 and 2020.
  - Severely Adverse scenario:
    - Real GDP decreases by 6.3 percent in 2018 and by 0.6 percent in 2019.
    - Unemployment climbs to 10 percent in 2019.
    - Short-term rates fall to zero while long-term rates stay unchanged (steeper yield curve).
    - Equity prices fall by more than 60 percent in 2018.
    - Real estate prices fall by 25 percent.
    - Economy only starts to pick up in 2020.

### Fiscal flows and deficit dynamics
- Under both shock scenarios, the deficit-to-GDP ratio is severely hit due to:
  - Rapid drop in tax revenue.
  - Expenditure rigidities.
- Revenue-to-GDP ratio assumptions under shock scenarios:
  - Revenue-to-GDP ratio temporarily decreases by 2.5 / 4.0 percentage points, before slowly reverting to the baseline ratio. (Adverse / Severely Adverse)
- Expenditure-to-GDP ratio behavior:
  - Non-age-related spending is assumed to remain equal to the baseline scenario in nominal terms, implying an increase in expenditure-to-GDP ratio under both shocks.
  - Age-related expenditure-to-GDP ratios are kept stable (health costs assumed not to increase as much as under the baseline).

### Asset revaluation and static balance sheet effects
- Nonfinancial assets (NFA) and real estate:
  - Assume that 20 percent of the previous year’s price change is translated into NFA revaluation effects.
  - Revaluation plays negatively in 2019 and 2020.
  - Overall, relative to the baseline scenario, nonfinancial assets fall by 2.5 / 5.6 percent of baseline GDP in 2020. (Adverse / Severely Adverse)
- Corporate equity assets (three quarters held by state and local pension funds):
  - Most significant revaluation happens in 2018.
  - In 2020, relative to the baseline scenario, state and local pension fund assets are expected to drop by 3.4 / 7.3 percent of baseline GDP. (Adverse / Severely Adverse)
  - Asset/liability mismatch deepens; mismatch ultimately sponsored by state and local governments (neutral for pension funds’ balance sheets), but can pose liquidity issues for some state or local governments.
- Overall static net worth impacts by 2020:
  - Projected decrease in US public sector static net worth of 13 (respectively 26) percent of baseline GDP by 2020 under the two shocks. (Adverse / Severely Adverse)
  - Asset revaluation effects explain about 8 / 15 percent of GDP of the total decrease in public sector net worth in 2020 relative to the baseline. (Adverse / Severely Adverse)

### Loan losses, GSEs, and student loans
- Federally-held student loans:
  - About 4.0 / 6.5 percent of federally-held student loans would not be paid back over the period 2018-2020, leading to a portfolio loss of 0.1 / 0.3 percent of baseline GDP. (Adverse / Severely Adverse)
  - Impact relatively limited due to federal government recovery power, rarity of discharge cases, and availability of temporary relief (deferment, forbearance, grace periods). Impact uncertain given limited historical data and potential policy change.
- Mortgage loan portfolio held by GSEs:
  - Cumulative losses of 0.4 percent / 0.6 percent of baseline GDP over three years. (Adverse / Severely Adverse)
  - FHFA estimate under a severely adverse scenario: potential incremental treasury draws by Fannie Mae and Freddie Mac of about $100bn over two years (0.4 percent of baseline GDP).
  - Any mismatch on GSE balance sheets will be offset by the federal government; loan losses for GSEs would ultimately lead to higher federal government debt.
- Aggregate loan-loss contribution to public sector net worth decline:
  - Expected student and mortgage loan losses are relatively small at -0.4 / -0.9 percent of baseline GDP. (Adverse / Severely Adverse)

### Debt accumulation and intertemporal fiscal resources
- Relative to baseline, deepened fiscal deficits lead to additional debt by 2020 of:
  - 2.4 / 8.7 percent of baseline GDP. (Adverse / Severely Adverse)
- Persistent loss in the level of real GDP reduces the intertemporal amount of resources available for discretionary fiscal policy under both shock scenarios relative to the baseline, making fiscal adjustment more difficult.

### Quantitative snapshots (selected table and series values preserved as presented)
- Static Public Sector Net Worth (percent of baseline GDP) – selected values from Table 2:
  - 2017: -18.3
  - 2018 (baseline / adverse / severely adverse): -21.4 / -27.7 / -35.8
  - 2019 (baseline / adverse / severely adverse): -24.7 / -35.9 / -47.8
  - 2020 (baseline / adverse / severely adverse): -27.9 / -40.5 / -54.1
- Assets (percent of baseline GDP) – general government:
  - 2017: 109.4
  - 2018: 106.0
  - 2019: 102.5
  - 2020: 100.6
- Nonfinancial assets (percent of baseline GDP) – general government:
  - 2017: 83.5
  - 2018: 80.5
  - 2019: 77.4
  - 2020: 75.6
- Liabilities (percent of baseline GDP) – general government:
  - 2017: 127.7
  - 2018: 127.4
  - 2019: 127.2
  - 2020: 128.5
- Debt securities (percent of baseline GDP):
  - 2017: 96.9
  - 2018: 97.3
  - 2019: 97.7
  - 2020: 99.2
- Memorandum items — Nominal GDP ($bn):
  - Baseline scenario: 2017 = 19391; 2018 = 20438; 2019 = 21577; 2020 = 22440
  - Adverse scenario: 2018 = 19386; 2019 = 19908; 2020 = 20849
  - Severely adverse scenario: 2018 = 18535; 2019 = 18725; 2020 = 19699

### Medium-term assumptions and projection methodology (2018–2023)
- Baseline scenario basis:
  - Based on April 2018 WEO macro-fiscal projections.
  - Takes into account the 2017 Tax Cuts and Jobs Act:
    - Peak static fiscal cost of 1.3 percent of GDP in 2019.
    - Peak dynamic positive impact on real GDP levels of about 1.2 percent in 2020.
  - Implicit interest rate rises from 2.5 percent in 2017 to 3.5 percent in 2023.
- Asset and liability transaction assumptions:
  - General government: rapid accumulation of financial assets (especially federal student loan portfolio) extended into forward years; liabilities incurrence computed as residual from WEO’s net lending path minus transactions in financial assets; net capital formation based on WEO projections; no transactions on natural resource asset portfolio assumed.
  - Public corporations: average transactions of financial assets over the last five years carried forward (in particular continued increase of mortgage loan portfolio); existing pension liabilities assumed to grow at same pace as present value of pension expenditure; other liabilities are residual assuming operating balance of public corporations equals its five-year average (zero).
- Price/revaluation assumptions:
  - Equity assets: revaluation proxied by Dow Jones Total Stock Market Index, forecast at 5 percent every year by the Federal Reserve.
  - Natural resource assets: revalued at 3 percent every year (assumed evolution of global Brent crude oil prices).
  - Revaluation of other nonfinancial assets depends on one-year-lagged evolution of house and commercial real estate price indexes (Fed’s scenario).
  - Empirical rule: about 20% of price changes are reflected in the following year’s revaluation of nonfinancial assets other than natural resources.
- Loan-loss projection approach:
  - Use historical relationship between banking sector quarterly net loan losses and macroeconomic variables via a VAR model (house prices, unemployment rate, mortgage interest rate).
  - For GSE-held loans, projected loan losses are scaled by a factor of 1/3.
- Quantitative easing unwinding assumptions:
  - Baseline consistent with Federal Reserve announcements: $420 billion worth of debt assets to be dropped in 2018, followed by $600 billion in 2019, 2020 and 2021. This is offset by concurrent decrease of the Fed’s currency and deposit liabilities by the exact same amount.

### Conclusions and implications for policy analysis
- By including public sector assets and other liabilities, intertemporal balance sheet analysis indicates:
  - The fiscal adjustment need is more significant than suggested by debt-only analysis.
  - Shift in major fiscal risks since the crisis: GSEs’ balance sheets have shrunk and become less risky under federal conservatorship, while funding shortfalls for state and local pension funds have emerged as a source of fiscal fragility.
  - In a severe shock, balance sheet effects outside the realm of debt expose public sector net worth to a negative impact almost double in size to the expected accumulation of debt.
  - Emerging fiscal risks include the fast-growing federal portfolio of student loans.
- Suggested further analysis:
  - More granular analysis of funding status of state and local pension funds, including different discount rate assumptions, to better assess mismatches and inform pension reform scenarios.
  - Further analysis of mortgage loan and student loan data to refine loss-assessment models.
  - Repeating the stress test exercise regularly to update baseline assumptions and track changes in the fiscal environment.

*Source: IMF staff calculations and accompanying material from the referenced chapter (wpiea2019139).*

### 2. Long-term (beyond 2024)

### 2. Long-term (beyond 2024)

### Macro-fiscal projections and demographic assumptions
- Methodology
  - Growth accounting approach developed by FAD.
  - Population projections: United Nations’ World Population Prospects, medium-fertility rate scenario, broken down by five-year cohorts, by age and gender.
- Demographic trajectory and impact
  - Share of population over 60 years old increases from 20 percent in 2015 to 30 percent in 2075.
  - Aging leads to a slow erosion of the contribution of labor to growth.
- Growth and price assumptions
  - Under assumed constant total factor productivity growth, real GDP growth: 1.9 percent in 2020 to 1.8 percent in 2075 (with some minor oscillations).
  - Nominal GDP deflator set at 2 percent.
  - Implicit interest rate assumed to gradually reach a long-term level within fifteen years: 4.4 percent.

### Long-term fiscal costs related to aging
- Health and pension expenditure projections (FAD / CBO sources)
  - Public health expenditure projected to double within 60 years as a share of GDP: from about 9 percent of GDP in 2017 to 18 percent of GDP by 2070.
  - Pension cost projected to increase from 8.3 percent of GDP in 2017 to 10.3 percent of GDP by 2070.
  - Social Security (CBO long-term outlook): expected to increase from 4.9 percent of GDP to 6.3 percent of GDP within 30 years.
- Drivers of health expenditure increase
  - Aging population (expenditure higher for older age groups).
  - Observed trend of health care costs per capita growing on average faster than GDP per capita.

### Revenue and non-age-related expenditure assumptions
- Revenue
  - Federal government revenue flows expected to increase by 2.6 percent of GDP from 2024 to 2048 and remain constant as a share of GDP beyond 2048.
  - Contributors: expiration of some provisions of the Tax Cuts and Jobs Act (0.7 percent) and structural features of the individual income tax including real bracket creep (about 1.2 percent of GDP).
- Primary (non-age) expenditure
  - Primary expenditure other than age-related spending assumed to decline by 1.1 percent between 2018 to 2028 (based on historical downward trend for federal discretionary spending) and remain constant as a share of GDP beyond 2028.
- Subnational government flows
  - Scenario implicitly keeps subnational revenue and expenditure constant relative to GDP over the long-term.

### Sensitivity of present values to discount rate assumptions
- Discounting policy
  - Discount rate used to calculate present values is the implicit interest rate.
  - Higher discount rates lower present values of future revenues and expenditures.
- Key sensitivity results (impact on 2017 present values, in percentage point of 2017 GDP) — Table A1 highlights:
  - +10 basis points in 2018: Impact on 2017 PV of primary balance +0.2; Impact on 2017 PV of revenue -1.7.
  - +100 basis points in 2018: +1.6; -16.4.
  - +10 basis points in 2028: +0.1; -1.3.
  - +100 basis points in 2028: +1.4; -13.0.
  - +10 basis points 2018–2027: +1.5; -15.0.
  - +100 basis points 2018–2027: +14.1; -143.0.
  - +10 basis points to LT rate: +2.2; -15.6.
  - +100 basis points 2018–2067: +41.2; -353.2.
- Illustrative effects reported in text
  - A one-shot, one-point increase in 2018 would diminish the 2017 present value of revenue by about 16 percent of 2017 GDP and increase the 2017 present value of primary balance by 1.6 percent of 2017 GDP.
  - A permanent extra point over the whole 50-year period would decrease the present value of revenue by some 350 percent of GDP, while improving the present value of primary balance by 41 percent of GDP.

### Comparison of macroeconomic projections with other institutions (Table A2)
- Real GDP growth
  - 2018–28: FAD 1.9; CBO 1.9; GAO 1.9
  - 2029–68: FAD 1.9; CBO 1.9; GAO 2.1
- Inflation
  - 2018–28: FAD 2.2; CBO 2.1; GAO 2.1
  - 2029–68: FAD 2.0; CBO 2.0; GAO 2.1
- Nominal interest rates (note: CBO and GAO project federal government debt rates only)
  - 2018–28: FAD 3.5; CBO 3.1; GAO 3.1
  - 2029–68: FAD 4.3; CBO 3.9; GAO 3.7 / 3.8

### VAR model for loan losses (ANNEX 2)
- Data and scope
  - Quarterly series used: “Net Loan Losses to Average Total Loans for Banks” (Federal Reserve Bank of St. Louis) available from 1984.
  - Recognized caveats: mortgage collateral reduces losses relative to unsecured credit; student loans respond differently to shocks; borrower distress resolution differs across loan categories.
  - GSE-specific adjustment: projected loan losses scaled by factor 1/3 because historical GSE losses in 2010–2011 were considerably lower than banking sector averages.
- Explanatory variables (quarterly, contemporaneous + three lags)
  - Change in unemployment rate (BLS, monthly).
  - Change in national house price index (all transactions).
  - Change in real 30-year mortgage interest rate (Freddie Mac, weekly).
  - Quarterly dummies included.
- Model performance and findings
  - Out-of-sample prediction up to 2007q4 captures about half of the increase in loan losses during the GFC; including the GFC in estimation sample increases responsiveness.
  - Regression highlights: few statistically significant variables — second lag of changes in house prices and changes in unemployment rate (contemporaneous and second lag).
  - Loan losses are highly persistent.
- VAR regression results (Table A2: coefficients and p-values)
  - Dependent variable: Net Loan Losses
  - Dep. Var. 1st lag: coefficient 0.89; p-value 0.00
  - Dep. Var. 2nd lag: coefficient 0.10; p-value 0.38
  - Dep. Var. 3rd lag: coefficient -0.14; p-value 0.07
  - Change in unemployment rate
    - contemporaneous: coefficient 0.11; p-value 0.04
    - 1st lag: coefficient 0.08; p-value 0.20
    - 2nd lag: coefficient 0.10; p-value 0.10
    - 3rd lag: coefficient 0.08; p-value 0.16
  - Pct. change in house prices
    - contemporaneous: coefficient 1.73; p-value 0.27
    - 1st lag: coefficient 1.92; p-value 0.27
    - 2nd lag: coefficient -6.29; p-value 0.00
    - 3rd lag: coefficient -0.70; p-value 0.67
  - Change in real interest rate
    - contemporaneous: coefficient 0.02; p-value 0.51
    - 1st lag: coefficient -0.01; p-value 0.71
    - 2nd lag: coefficient 0.02; p-value 0.45
    - 3rd lag: coefficient -0.02; p-value 0.43
  - Number of observations: 132
  - R-squared: 0.97
- Scenario projections and calibrated outcomes
  - Applying coefficients to DFAST scenarios yields projected loan loss ratios.
  - Severely adverse scenario: peak loan losses as share of total loans slightly above those in the GFC.
  - Severely adverse scenario: total loan losses between 2018–20 would be USD 89 billion higher than under the baseline.
  - Historical comparator: USD 187.5 billion in total actual GSE drawings on the Treasury as of end-2016.
  - Note: exercise does not take into account provisioning rules or impacts on tax assets.

### Data sources and assumptions for the static public sector balance sheet (ANNEX 3)
- General approach
  - Central/General Government: authorities’ submission to STA for dissemination in GFS Yearbook database.
  - Estimates for stocks of mineral and energy resources equal the present value of expected pre-tax cash flows from commercial exploitation; sources and methods differ by commodity.
- Oil and gas valuation
  - Production: Rystad database (only government owned fields).
  - Prices (USD): WEO forecasts available at the end of the reference year.
  - Costs of production (USD): Rystad database.
  - Cash flows calculated over an 85-year horizon.
  - Discount rate for net present value: 4.5% (equal to average (2000–22) long-term (10-year) government bond yields in WEO plus one percent).
- Coal, metals and other minerals valuation
  - Sources:
    - Estimates (constant 2014 USD) from World Bank “The Changing Wealth of Nations 2018” for years 2000, 2005, 2010, 2014.
    - United States Geological Survey data on 2016 reserves and 2014–16 production where available.
    - Prices (USD): WEO actual commodity prices for 2000–16.
    - US Department of Interior Natural Resources Revenue Data, including production for federal lands and water.
  - Missing years: linear interpolation of available observations; 2015 and 2016 follow evolution of reserves where data exist; otherwise assume value unchanged from 2014 onward.
  - Estimates converted to current USD using WEO commodity price index and pro-rated using government ownership information.
- Central Bank and Public Corporations
  - Central Bank: authorities’ submission to STA of Central Bank Survey through Standardized Report Format.
  - Nonfinancial Public Corporations: none; NFPCs consolidated in general government data.
  - Financial Public Corporations (other than Central Bank): Fed’s Financial Accounts of the United States - Z.1 (vintage March 8, 2018) — tables F.x and L.x for specified entities (Federal Government Employee Retirement Funds, State and Local Government Employee Retirement Funds, GSE, Agency- and GSE-Backed Mortgage Pools).
  - No data adjustments other than consolidation of Agency- and GSE-backed securities held by Pension Funds and GSEs.
  - Total FPCs (“NPCT Time Series”): aggregation of CB and other FPC, less Fed’s holdings of agency- and GSE-backed securities and GSEs’ checkable deposits at the Fed.
- Public sector aggregation
  - Public Sector = aggregation of General Government and Total FPC, less:
    - Fed’s holdings of general government securities;
    - Federal government deposits at the Fed;
    - GSEs and Government Employee Retirement Funds’ holdings of government (federal and municipal) securities;
    - Government holding of financial public corporations’ equity;
    - Claims of Government Employee Retirement Funds on the government units as pension managers.

*Source: IMF staff (content unit: 2. Long-term (beyond 2024)).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019139.pdf_