## _cr03245

## Source details

**Canonical URL:** [_cr03245](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2003/_cr03245.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2003/_cr03245.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2003/_cr03245.pdf.json)

---

### Real Estate Bull and Bear Markets; Household Wealth and Saving
- Household net worth fell from about 6¼ times personal disposable income to 4½ times personal disposable income since 2000.
- Fitted personal saving rate at end-2002: 4½ percent; actual personal saving rate at end-2002: 3½ percent (implying a 1 percentage point gap).
- Visual implication noted: personal saving rate would need to increase by around 2½-3 percentage points to match current levels of household wealth.
- Wealth composition changes (end-1995 to 2003Q1):
  - Net housing wealth: +$3¼ trillion.
  - Net equity wealth: +$2½ trillion.
  - Net non-equity financial wealth: +$6¼ trillion (reaching $21¾ trillion in 2003Q1).
- Asset-shift and liquidity:
  - Share of non-equity financial wealth held in demand deposits rose from around 25 percent to nearly 30 percent (2000-2003).
  - About one-quarter (about $5 trillion) of non-equity financial wealth is equity in noncorporate businesses and may be illiquid.
- Long-run consumption model (sample period 1970Q1-2002Q4):
  - Estimated wealth effect: 2-3½ cents per $1 increase in wealth, varying by aggregation.
  - Disaggregated specifications: largest marginal wealth effect from non-equity financial wealth.
  - Model diagnostics (columns (1)-(4)): Wald test statistics and p-values — 13.955 (p-value 0.050), 2.779 (p-value 0.068), 0.015 (p-value 0.985), 0.771 (p-value 0.514).
  - Akaike Info. Criterion reported as -6.448, -6.715, -6.879, -6.764 for columns (1)-(4).
- Interpretation:
  - Decline in equity wealth accounted for around a 2 percentage point rise in the personal saving rate during 2000-2003.
  - Strong income growth exerted upward pressure on the saving rate; growth in residential and non-equity financial wealth dampened it.
  - Consumption exhibits an exogenous upward trend, possibly linked to improved household access to credit.

### Changes in the Personal Saving Rate — Short-term Risks and Measurement
- Short-term GDP risk: a full 1 percentage point upward adjustment in the personal saving rate (to match fitted value) would reduce GDP growth by around ¾ percentage points.
- Downside scenario: a larger correction in saving is a risk if the economic recovery disappoints; net housing wealth vulnerable if house prices fall.
- Measurement and data construction notes:
  - Consumption services: consumer durables service flow imputed using BEA stock data and user-cost of capital across 13 categories.
  - Household net wealth classified into net residential wealth, net equity wealth, and net non-equity financial wealth (non-profits net wealth included in non-equity financial wealth).
  - Household labor income defined as personal disposable income less proprietors’ income, rental income, dividend income, and interest income.
  - Consumption, wealth, and income scaled by number of U.S. households.

### Demand, Housing Market Drivers, and Price Sustainability Risks
- Household formation and ownership:
  - Number of households rose by 9.7 million between 1996 and 2002; 9.2 million housing units added in same period.
  - Home ownership rate: 68 percent in 2002 (record).
- Mortgage markets and transaction costs:
  - Transaction costs for new mortgages declined from 250 basis points (1985) to below 50 basis points (2002).
  - Share of ARMs in new loans fell to below 20 percent (compared with 30 percent in the 1980s).
  - GSEs (Fannie Mae and Freddie Mac) issued bulk of securitized mortgages and standardized origination.
- Supply constraints and zoning:
  - House prices generally close to construction costs; divergences largely in a few metropolitan markets linked to zoning strictness.
- Four sources of concern on price sustainability:
  - Vulnerability to slower nominal income growth: example shows mortgage payments share of income under 4 percent nominal growth vs 2 percent scenarios (specific illustrative proportions given: 19.8 percent vs 22.2 percent by 2008; after 10 years 15.6 percent vs 19.7 percent).
  - Increase in interest rates could burden ARM holders; countervailing factors include lower long-term rates and longer average maturities from refinancing.
  - Divergence between house prices and rents: house price-rent ratio exceeds long-term average by 10-15 percent (Krainer, 2003).
  - Lower taxes on real estate capital gains (1997 Taxpayer Relief Act): capital gains exemption up to $500,000 for married homeowners; capital gains tax rate lowered to 20 percent; first-time home buyers allowed withdrawal up to $10,000 from IRAs without penalty.
- Empirical and structural model evidence:
  - Cross-study MSA assessments: no nation-wide bubble; some 20 MSAs repeatedly identified as excessively priced (out of 250 MSAs), concentrated in CA, CO, FL, MA, NJ, NY, WA, and DC (together ~25 percent of single-housing stock).
  - Structural three-stage least squares regional model (annual, 1978-2002) key coefficients:
    - Marginal income coefficient: 0.7 (new homes), 0.5 (existing homes).
    - Construction cost coefficients: 0.751 (new), 0.554 (existing).
    - Number of houses sold coefficients (supply): 0.224 (new), 0.194 (existing).
    - Regional gaps: West actual vs predicted gap 15–25 percent; Northeast somewhat smaller.

### Defined Benefit Pension Plans — Funding, Profits, PBGC, Accounting, and Policy
- Funding deterioration:
  - S&P 500 pension surplus in 1999 nearly $250 billion; surplus exhausted by end-2001; end-2002 DB plans had a deficit estimated at $216 billion (CSFB 2003).
  - Valuation losses roughly $400 billion between end-1999 and end-2002.
- Portfolio composition (selected percentages, end years reported in source table format):
  - Corporate Equities: 42 55 43 (for 1985 1999 2002 as presented).
  - Government Bonds: 19 11 15.
  - Corporate Bonds: 10 11 15.
  - Mutual Fund Shares: 16 5.
  - Other: 27 17 22.
- Representative sample of 19 underfunded plans:
  - These 19 firms account for $110 billion of total underfunding.
  - Nearly half of deterioration since 2000 from stock market losses; remainder from higher benefit obligations (lower discount rates) and net payouts.
  - Table excerpts: All 19 companies funding positions shifted from 1997–1999: 53.9 to 2002: -110.3 (figures presented in source).
- Pension funding rules and profit impacts:
  - Underfunded plans required to return to full funding over five- to 30-year periods; additional faster requirements if below 90 percent funded under specified conditions.
  - Legislation in 2002 raised required discounting interest rate to 120 percent of the 30-year Treasury yield.
  - S&P 500 firms tripled pension contributions in 2002 to $46 billion; contribution pressure reduced economic profit growth to 7½ percent in 2002 (5 percentage points lower than otherwise; CSFB, 2002).
- Fund staff macro simulation (impact on profit growth; S&P 500 aggregates):
  - Actual/Observed 2002: Corporate Contributions $46 billion; Underfunding $216 billion; Impact on Profit Growth -5.
  - Baseline: 2003 Contributions $60 billion; Underfunding $170 billion; Impact -2. 2004 Contributions $70 billion; Underfunding $110 billion; Impact -1.
  - Strong recovery: 2003 Contributions $46 billion; Underfunding $65 billion; Impact 0. 2004 same.
  - Weak recovery: 2003 Contributions $85 billion; Underfunding $220 billion; Impact -5. 2004 Contributions $100 billion; Underfunding $210 billion; Impact -2.
- PBGC position and risks:
  - PBGC assumed additional $9 billion in obligations over two years; its net actuarial position eroded to a $3½ billion deficit (Single-Employer Program net position: 2001 = 7.7; 2002 = -3.6 in billions as presented).
  - PBGC financials (2001/2002, billions): Total assets 21.8 / 25.4; Total liabilities 14.0 / 29.1; Net loss -2.0 / -11.4; Net position 7.7 / -3.6.
  - Current annual premiums: $19 per plan participant plus $9 per $l,000 of unfunded vested benefits.
  - Stochastic simulations imply a 30 percent probability PBGC would be in surplus by 2012 absent measures.
  - Tradeoff: raising premiums or risk-based pricing could induce adverse selection and plan terminations.
- Accounting and valuation concerns:
  - Firms use expected return on pension assets in earnings rather than actual returns; median expected return still above 8 percent despite negative market returns in 2001–2002.
  - For representative 19 firms, reported earnings exceeded actual returns by $130 billion over past two years (as presented).
  - Each 50 basis point increase in interest rates reduces PBO for S&P 500 firms by $60 billion (CSFB 2003).
- Policy implications recommended in source:
  - Restore PBGC position balancing premium changes with oversight to avoid moral hazard and adverse selection.
  - Improve accounting transparency: require explicit reporting of plan returns’ impact on reported earnings.
  - Tax policy adjustment: relax penalties against contributions to fully funded plans—proposal example allowing deduction for contributions to plans funded up to 110 percent.

### Fiscal Turnaround, FY 2004 Proposals, and Interest-Rate Effects
- Historical episodes: structural fiscal expansions and tax cuts associated with mixed medium-term output effects; tax cuts sometimes followed by short-term growth boosts but lower or unchanged growth over subsequent ten years.
- Model simulations of FY 2004 proposals:
  - MA and GI models: significant positive output effect over next two years, waning thereafter due to crowding out and higher real interest rates.
  - MA model: capital accumulation reduction lowers labor productivity by about ½ percentage point in 2017 relative to baseline.
- CBO model results (selected average changes in GDP from CBO baseline, in percent):
  - Textbook Growth Model: 2004-2008 = -0.2; 2009-2013 = -0.7.
  - Closed Economy Life-Cycle (Lower gov't consumption after 2013): 2004-2008 = -0.3; 2009-2013 = -1.5.
  - Closed Economy Life-Cycle (Higher lump-sum taxes after 2013): 2004-2008 = 0.5; 2009-2013 = 0.3.
  - Open Economy Life-Cycle (Lower gov't consumption after 2013): 2004-2008 = -0.6; 2009-2013 = -0.5.
  - Open Economy Life-Cycle (Higher lump-sum taxes after 2013): 2004-2008 = 0.3; 2009-2013 = 0.6.
  - Infinite Horizon Growth Model (Lower gov't consumption after 2013): 2004-2008 = 0.2; 2009-2013 = -0.6.
  - Infinite Horizon Growth Model (Higher lump-sum taxes after 2013): 2004-2008 = 0.9; 2009-2013 = 1.4.
- Empirical survey on deficits and interest rates:
  - Large-scale model average: 1 percentage point increase in primary deficit-to-GDP caused by tax cut → ~40 basis points increase in long-term interest rates after one year and ~60 basis points after ten years (higher if deficit caused by spending).
  - Recent reduced-form estimates: ~35 basis points per 1 percentage point increase in deficit-to-GDP on expected future interest rates.
  - Elmendorf and Mankiw (CEA) neoclassical estimate: ~2-3 basis points per 1 percentage point increase in debt-to-GDP.
  - Laubach (2003): ~15 basis points for a permanent 1 percentage point increase in deficit-to-GDP.
- International implications:
  - World real interest rates co-move; Ford and Laxton (1999) estimate a 1 percentage point increase in world net government debt raises real interest rates by around 20 basis points.
  - System estimates imply a 1 percentage point increase in world government debt-to-GDP induces ~10 basis points increase in national real interest rates over 1997-2002.
  - Using these estimates, a 15 percentage point increase in U.S. public debt ratio projected by CBO could lead to an average ½-1 percentage point increase in national real interest rates.

### State and Local Governments (SLGs): Finances, Reserves, and Fiscal Drag
- SLG trends and statistics:
  - SLG current expenditures account for nearly all of the 7 percent of GDP increase in general government spending since 1960.
  - Federal grants account for almost one quarter of SLG revenues in 2002.
  - At end-1990s SLGs ran current surpluses up to around ½ percent of GDP; current deficits reached ½ percent of GDP in 2002 and a deficit exceeding ¾ percent of GDP seemed likely in FY 2004.
  - Federal grants-in-aid to SLGs as percent of total receipts: 0.8 (1960), 2.6 (1980), 1.9 (1990), 2.9 (2002).
  - SLG share of current receipts as percent of GDP: 8.0 (1960), 11.3 (1980), 11.4 (1990), 12.5 (2002).
  - Composition of state/local receipts (2002, percent of total receipts): Personal income tax receipts 15.4; Corporate profits tax accruals 2.6; Sales taxes 25.6; Property taxes 20.5; Federal grants-in-aid 23.4.
- Reserve funds and adjustments:
  - State reserve funds: about 10 percent of state expenditures by end-FY 2000 (compared with less than 5 percent at end-1980s).
  - By end-FY 2003 reserves virtually exhausted after 16 states drew down reserves.
  - Some studies estimate reserves needed >18 percent of expenditures to accommodate severe macro shocks.
- FY 2002–FY 2003 state adjustments:
  - Spending-side: hiring freezes; cuts to prisons, education, childcare, local government support; Medicaid largely excluded from cuts.
  - Revenue-side: limited tax hikes; use of tobacco settlement funds totaling $32 billion between 1998-2002.
- Borrowing and credit:
  - Market debt of state and local governments rose from 12 percent of GDP in 2000 to 14 percent in early 2003.
  - Aggregate SLG adjustment estimated to produce a fiscal contraction of about ¼ percent of GDP in 2003, offsetting only a small part of a 1¾ percent of GDP federal stimulus.
- VAR results on fiscal shocks (dynamic effects):
  - One standard deviation shock to SLG consumption share: increases GDP by 0.4 percentage points immediately; effect declines to near zero by fifth quarter.
  - One standard deviation shock to SLG net taxes: negligible short-run effect; builds to almost 0.4 percentage points by fifth quarter; dissipates after 6 quarters.
  - One percentage point increase in net federal taxes: no significant impact on output gap.
  - One percentage point increase in federal spending: reduces output gap by ~0.2 percentage points in first quarter; dissipates after three quarters.

### Energy Policy, Taxation, and GEM Simulations
- U.S. energy facts:
  - Sector shares of Btu consumption (2001): Industrial 33 percent; Transportation 28 percent; Residential 21 percent; Commercial 18 percent.
  - Net petroleum imports projected to rise from 55 percent (2001) to 68 percent (2025) of U.S. petroleum consumption.
- Policy context:
  - NEP (May 2001) focused on boosting domestic supply, technology, and tax subsidies rather than consumption taxes.
  - Administration proposals and congressional bills include tax credits, loan guarantees, ANWR opening, ethanol mandates, and incentives for renewables and nuclear.
- Taxation rationale and elasticities:
  - Short-run price elasticities for energy and fuel demand: -0.13 to -0.26 (OECD, 2001a).
  - Long-run elasticities: -0.37 to -0.46 (OECD, 2001a).
  - Survey of 97 studies: short-run gasoline elasticity average -0.26; long-run average -0.86 (Dahl and Sterner, 1991).
  - CBO: a 15-cent gasoline tax hike could have raised $16 billion in 2003.
  - OECD: a 40-cent per gallon increase could be justified; a $100 per ton carbon tax would have yielded $110 billion in 1999.
- Staff GEM simulations (U.S. calibrated):
  - A 10 percentage point tax on petroleum products used as intermediate inputs: long-run U.S. GDP reduction 0.03 percent.
  - A 10 percentage point tax on final consumption of petroleum products: long-run U.S. GDP reduction 0.11 percent.
  - Detailed Table 1 (percent deviation from baseline) highlights short-run and long-run effects on Real GDP, Consumption, Investment, various prices, and Real exchange rate for three tax scopes (consumption, production, all energy) at 1 year, 5 years, and long run. Selected long-run outcomes:
    - 10 percent tax on energy used in consumption — Long Run: Real GDP -0.08; Consumption -0.04; Investment -0.22; Consumption price of energy 9.38.
    - 10 percent tax on energy used in production — Long Run: Real GDP -0.03; Consumption 0.00; Investment -0.13; Goods producers’ price of energy 7.62.
    - 10 percent tax on all energy — Long Run: Real GDP -0.11; Consumption -0.04; Investment -0.35; Consumption price of energy 7.02.
- Design and distributional notes:
  - Tax revenue recycling (to lower labor or investment taxes) can alleviate output losses.
  - Consideration needed for distributional impacts (e.g., rural vs urban households).
  - Efficiency comparisons: gasoline taxes more efficient than CAFE-like standards (CBO, 2002).
- Fiscal magnitudes of proposed incentives:
  - Administration proposals: tax incentives costing $8 billion over ten years for various clean-energy measures.
  - House: ~ $19 billion in tax breaks; Senate: ~ $16 billion over ten years.

### Globalization, Volatility of GDP Growth, and Country/Regional Findings
- Volatility periodization:
  - Three periods: Bretton Woods; Common Shocks (1973:1–86:2); Globalization period (1986:3–2001:4).
  - U.S. real GDP growth standard deviation declined significantly since the 1960s; similar declines for other G-7 except Germany and Japan.
- Dynamic latent factor model results (21 series across 7 countries):
  - World factor explains >25 percent of G-7 output variation and >15 percent of consumption and investment volatility.
  - Share of output variance explained by world factor rose from ~7 percent (Bretton Woods) to 19 percent (globalization period).
  - World factor explains ~60 percent of output variation in France and less than 13 percent in the United States.
  - Correlation between median world factor and U.S. output growth: 0.62.
- Implications:
  - Global factors have become more important in driving domestic business cycles; a U.S. rebound could have large spillovers, especially to Canada and the United Kingdom.
  - U.S. external linkages remain small relative to size of its economy, limiting pure export-driven recovery prospects.

### Mexico and Regional Integration (NAFTA) — Volatility and Synchronization
- NAFTA effects on U.S. trade/GDP (CBO, 2003):
  - Boost to U.S. exports ~0.12 percent of GDP in 2001; imports +0.1 percent of GDP.
  - Potential long-term U.S. GDP increase due to NAFTA estimated range 0.02 percent to 0.5 percent.
- Mexico macro volatility:
  - Volatility of macro aggregates declined markedly post-NAFTA.
  - Regional factor contribution to Mexican output volatility rose from <1 percent (1980-1993) to >19 percent (1994-2002).
  - Investment variance explained by regional factor increased almost tenfold.
- Model simulations of trade-friction reductions:
  - Post-NAFTA simulation: substantial increase in Mexican exports; Mexico’s output and investment respond more strongly to partner-country supply shocks in post-NAFTA period.
- Conclusions:
  - NAFTA increased trade openness, growth, and business-cycle synchronization among members; caution urged in generalizing results to other FTAA members.

### Regional Free Trade Arrangements (FTAs) Simulations — GTAP Static Results and Caveats
- Simulation framework: GTAP comparative static general equilibrium model; removes tariffs and quotas between partners; omits dynamic gains and non-goods provisions.
- United States-Chile FTA:
  - Modest welfare gains for both parties; Chile GDP small drop due to trade diversion; Chile exports of processed foods gain.
- United States-CAFTA FTA:
  - Central America GDP could increase by as much as 1.5 percent; CAFTA region sees large gains (CAFTA: 3858.9 and 1.491 percent in table presentation).
  - Trade creation in textiles and processed foods; trade diversion in machinery/equipment.
- United States-Australia FTA:
  - Small overall welfare/output effects given low existing tariffs; simulations suggest slight Australian GDP drop (reported as -1.2 and -0.034 in table format for Australia).
- Selected GTAP numeric table excerpts (percent change unless otherwise noted) presented in source with country-by-country values for U.S.-Chile, U.S.-CAFTA, and U.S.-Australia scenarios (table entries preserved in source).
- Policy implications:
  - FTAs generate modest direct gains for the United States; partner countries may receive larger welfare gains but can suffer trade diversion.
  - FTAs most beneficial when minimizing trade diversion, covering broad product/service areas, and sustaining reform momentum without undermining multilateral liberalization.

*Italicized source attribution: IMF staff analysis and estimates as presented in _cr03245._*

### 2. Real Estate Bull and Bear Markets in Large Metropolitan Areas ......................17

### 2. Real Estate Bull and Bear Markets in Large Metropolitan Areas

### Recent trends in household wealth and saving
- Household net worth fell from about 6¼ times personal disposable income to 4½ times personal disposable income since 2000.
- Figure-based implication: the personal saving rate would need to increase by around 2½-3 percentage points to match the current levels of household wealth (visual suggestion).
- Empirical finding: the saving rate is about 1 percentage point below its long-run trend (text estimate).
- Personal saving rate: fitted value 4½ percent at end-2002, actual rate 3½ percent (implying a 1 percentage point gap).

### Composition and shifts in household wealth (1995–2003Q1)
- Net housing wealth rose by $3¼ trillion between end-1995 and 2003Q1.
- Net equity wealth rose by $2½ trillion between end-1995 and 2003Q1.
- Net non-equity financial wealth rose by $6¼ trillion over the same period, reaching $21¾ trillion in 2003Q1 and exceeding the value of equities at the height of the stock market.
- During 2000-2003 the share of non-equity financial wealth held in demand deposits rose from around 25 percent to nearly 30 percent.
- Between 1969-1996 equity wealth accounted for just under 20 percent of household net wealth; during the late 1990s equity share rose to almost 45 percent at the bubble peak, and by 2003 had fallen to 25 percent, with other assets returning toward long-run averages.

### Liquidity and portfolio reallocation after the equity collapse
- The collapse of the equity bubble led households to redirect savings into less risky assets and lock in capital gains; much of the rise in non-equity financial wealth reflects increases in demand deposits and money market shares.
- Not all non-equity financial wealth is liquid: just under one-quarter (about $5 trillion) is equity in noncorporate businesses and could be difficult to sell quickly.

### Long-run consumption model and wealth effects
- A long-run consumption model based on the permanent income hypothesis (PIH) is estimated over the sample period 1970Q1-2002Q4 (see Table 1 for full estimates).
- Assets are decomposed into residential (ar), equity (ae), and non-equity financial (an) wealth in the cointegrating specification.
- Estimated wealth effect (impact on consumption of a $1 increase in wealth) ranges from 2-3½ cents depending on aggregation.
  - Smallest wealth effects occur when residential wealth is included separately rather than aggregated.
  - In disaggregated specifications, the largest wealth effect comes from non-equity financial wealth (interpreted as reflecting liquidity and lower transaction/precautionary costs).
- Combining residential and non-equity financial wealth into a composite term improves model fit relative to a single consolidated wealth term (Akaike Information Criterion comparisons reported).
- A Wald test: the hypothesis that income and wealth coefficients sum to one is rejected when all wealth is aggregated (column 1), but not rejected for disaggregated specifications (columns 2–4). Reported Wald test statistics and p-values include: 13.955 (p-value 0.050), 2.779 (p-value 0.068), 0.015 (p-value 0.985), 0.771 (p-value 0.514) across columns (1)-(4).
- Table 1 summary statistics include adjusted R-squared values and AIC: Adjusted R2 values and Akaike Info. Criterion reported as -6.448, -6.715, -6.879, -6.764 for columns (1)-(4). (Sample period: 1970Q1-2002Q4.)

### Interpreting consumption, saving, and asset composition effects
- The model suggests consumption has exhibited an exogenous upward trend, possibly related to improved household access to credit—consistent with Cerisola and De Masi (1999).
- The decline in equity wealth accounted for around a 2 percentage point rise in the personal saving rate during 2000-2003 (model parameters imply equity wealth decline pushed saving up by 2 percentage points).
- Strong income growth also exerted upward pressure on the saving rate in this period, while growth in residential and non-equity financial wealth exerted a dampening effect.
- Because the preferred (column 3) equation implies a fitted consumption path, the resulting fitted saving rate (4½ percent at end-2002) indicates the observed saving rate (3½ percent) was modestly below the level implied by fundamentals.

### Key quantitative figures and indicators cited
- Household net worth: fell from about 6¼ to 4½ times personal disposable income.
- Implied required increase in personal saving rate from visual inspection: around 2½-3 percentage points.
- Empirical saving-rate gap relative to fundamentals: about 1 percentage point.
- Wealth changes, end-1995 to 2003Q1: real housing +$3¼ trillion; equities +$2½ trillion; non-equity financial +$6¼ trillion (to $21¾ trillion).
- Share of non-equity financial wealth in demand deposits: ~25 percent → nearly 30 percent (2000-2003).
- Equity share of household net wealth: just under 20 percent (1969-1996) → almost 45 percent (bubble peak) → 25 percent (post-collapse).
- Estimated wealth effect: 2-3½ cents per $1 increase in wealth (varies by aggregation).
- Fitted vs actual personal saving rate at end-2002: 4½ percent (fitted) vs 3½ percent (actual).

*Source: IMF staff analysis as presented in the chapter "U.S. Household Saving: Long-Run Determinants and Short-Term Risks," sample period 1970Q1-2002Q4.*

### 5. Changes in the Personal Saving Rate

### 5. Changes in the Personal Saving Rate

### Short-term outlook risk from saving adjustments
- Despite a moderate shortfall in personal saving relative to levels determined by income and wealth, any upward adjustment in saving could weigh on the short-term outlook.
- If saving were to adjust by the full 1 percentage point of disposable income necessary to return the actual rate to its fitted value, the effect would be to reduce GDP growth by around ¾ percentage points.
- A larger correction in the saving rate remains a risk if the economic recovery disappoints:
  - A slower-than-expected recovery could lead to a continued decline in household wealth and weaker labor market conditions.
  - Net housing wealth is particularly vulnerable because a slowing economy could depress house prices; households have borrowed extensively against the value of their homes.

### Structural and longer-term considerations
- The trend decline in the saving rate may not continue into the future.
- Financial innovation has likely allowed households to save less while achieving the same level of wealth and consumption (Cerisola and De Masi, 1999). If the pace of innovation has slowed or has been nearly completed, the saving rate is unlikely to continue its trend decline.

### Data construction and measurement relevant to saving analysis
- Consumption services
  - The chapter provides a direct estimate of household consumption services to estimate long-run cointegration equations.
  - The service flow of consumer durables is imputed from the consumer durables stock (from annual BEA data) following OECD (2001, pp. 54-69).
  - The service flow is taken as the user cost of capital times the stock of durables for each of the 13 major categories of consumer durables.
  - Depreciation is from the BEA; interest rates for user cost are calculated assuming correspondence to rates on auto loans, home mortgages, or personal loans (the 13 cost of capital terms use one of these three rates).
  - Nominal and real stock data are interpolated to a quarterly frequency using the quarter’s share of nominal spending on consumer durables in the yearly total.
  - An implicit price deflator is derived from the nominal and real service flow data; the 13 categories are chain-weighted together, then chain-weighted with nondurables and services consumption to form a chain aggregate.
- Household net wealth (three broad classifications used)
  - Net residential wealth: difference between owner-occupied real estate assets and household mortgages.
  - Net equity wealth: difference between broad equity assets and security credits. Broad equity assets are defined as the sum of specified household and institutional equity holdings (per Davis and Palumbo, 2001, p. 46) with a particular adjustment for life insurance company reserves.
  - Net non-equity financial wealth: residual of net financial wealth less net equity wealth; includes demand deposits, Treasury securities, corporate securities, municipal bonds, imputed equity value of noncorporate businesses, and non-profits wealth. Non-equity financial liabilities include consumer credit, bank loans, and nonprofit liabilities.
  - Net wealth of non-profits is included in non-equity financial wealth (despite significant real estate holdings) to avoid biasing the net housing wealth coefficient.
  - Consumer durable assets are excluded from net household wealth because their service flow is included in consumption services.
- Household labor income
  - Labor income is taken as personal disposable income less: proprietors’ income (with inventory valuation and capital consumption adjustments); rental income (with capital consumption adjustments); dividend income; and interest income.
- Household normalization
  - Consumption, wealth, and income terms are scaled by the number of U.S. households to gauge the effect of household formation on long-run equilibrium relationships.
  - Households are used rather than per capita measures because household formation rates affect residential wealth accumulation and many services are consumed at the household level (e.g., utilities, homeowner’s rent).

*Source: _cr03245 - 5. Changes in the Personal Saving Rate*

### 9.      Demand also has been reinforced by demographic trends that boosted the number

### 9.      Demand also has been reinforced by demographic trends that boosted the number

### Demand drivers and housing market context
- The number of households rose by 9.7 million between 1996 and 2002, compared to 9.2 million units added to the housing stock during that period.
- The home ownership rate reached a record 68 percent in 2002.
- Demand drivers cited:
  - Coming of age of the last cohorts of the baby-boom generation (Mankiw and Weil 1989).
  - Decline in average household size.
  - Pickup in immigration (note: “The number of legal immigrants to the United States averaged 900,000 per year in the 1990s—compared to an average of 730,000 in the 1980s—and census estimates also indicate a rising inflow of illegal immigrants.”).

### Role of GSEs, mortgage markets, and transaction costs
- The growing use of mortgage-backed securities (MBS) facilitated separation of mortgage origination and investment in mortgages.
- Government-sponsored enterprises (GSEs) such as Fannie Mae and Freddie Mac:
  - Issued the bulk of securitized mortgages.
  - Standardized the mortgage application process and introduced greater levels of competition in origination.
  - Their on-balance-sheet holdings (financed using the GSEs’ triple-A rating) may have partially reduced costs to borrowers (illustrated by a positive spread between conforming and noneligible “jumbo” loans).
- Transaction costs for new mortgages have on average declined from 250 basis points of the loan amount in 1985 to below 50 basis points in 2002.
- The share of adjustable-rate mortgages (ARMs) in newly closed mortgage loans fell to below 20 percent (compared with 30 percent in the 1980s), and a refinancing wave increased average maturity of outstanding mortgage debt.

### Supply constraints, zoning, and local price divergences
- Glaeser and Gyourko (2003) found that, in general, house prices were fairly close to physical construction costs across most of the United States.
- Divergences between house prices and construction costs were largely limited to a few metropolitan markets and were largely related to measures of zoning strictness (restrictions on the size and characteristics of houses).
- Variables related to the size of housing lots or population density were not found to have a significant influence on house prices, suggesting supply factors (e.g., zoning) can play a major role in explaining excessive real estate prices in some major urban centers.

### Reasons for concern (risks to price sustainability)
- Four sources of concern over price sustainability:
  - Vulnerability to slower nominal income growth:
    - If present low inflation persists and nominal household income growth fell below the 4 percent average achieved during the 1990s, the value of both mortgage debt and debt service relative to income would decline relatively slowly, leaving households vulnerable to income shocks or unemployment (Baker 2002a).
    - Illustrative example: assuming a household allocates 25 percent of disposable income to payments on a new 30-year fixed mortgage at 6 percent interest, with 4 percent nominal income growth mortgage payments account for 19.8 percent of income by 2008 (compared to 22.2 percent under 2 percent nominal growth); after 10 years the proportions are 15.6 percent and 19.7 percent, respectively.
  - Increase in interest rates:
    - Higher interest rates could burden holders of ARMs and reduce overall housing affordability.
    - Countervailing factors: drop in long-term mortgage rates accompanied by decline in ARMs share to below 20 percent and refinancing that increased average maturities.
  - Divergence between house prices and rents:
    - The recent divergence is viewed by some as signaling the need for market correction.
    - Krainer (2003) finds the house price-rent ratio currently exceeds its long-term average by 10-15 percent, but shows the ratio would return to this average if rents continued to grow in line with their long-term trend and house prices were flat for a period of two to three years.
  - Lower taxes on real estate capital gains:
    - The 1997 Taxpayer Relief Act exempted capital gains of up to $500,000 for married homeowners.
    - The tax rate on capital gains was lowered to 20 percent.
    - First-time home buyers were allowed to withdraw up to $10,000 from individual retirement accounts without penalty.
    - These changes could have contributed to upward price shifts and, by facilitating realization and withdrawal of capital gains, could contribute to greater price volatility in a downturn (Knight and Eakin, 1998).

### Empirical evidence and model-based results
- Cross-study assessments:
  - Recent analyses using data for metropolitan statistical areas (MSAs) (Burns, 2002; JCHS, 2003; Youngblood, 2003; Zandi, 2002) found no signs of a nation-wide housing bubble.
  - These studies concluded price levels in most areas are broadly consistent with increases in personal income, though each identified a different group of MSAs with prices excessive relative to fundamentals.
  - Only some 20 MSAs (out of a total of 250 MSAs) were identified as being excessively priced in more than one of the four papers.
  - These 20 markets include the largest metropolitan regions and are located in the states of California, Colorado, Florida, Massachusetts, New Jersey, New York, Washington, and the District of Columbia. Together these states account for about 25 percent of the single-housing stock in the United States.
  - Deteriorating employment in some high-priced markets has had limited impact on housing price inflation; signs of an actual price decline exist in only one MSA so far.
- Structural model estimates (three-stage least squares, annual data by major census regions, 1978-2002):
  - Model jointly estimated demand and supply for new and existing house prices (endogenous variables: real house prices and number of homes sold).
  - Demand-side explanatory variables initially included real disposable income, real mortgage rate, unemployment rate, financial wealth, and a variable representing age structure; financial wealth and age-structure variables were dropped due to insignificance.
  - Supply-side explanatory variables initially included construction costs, housing stock, and home ownership levels; housing stock and home ownership proved insignificant and were replaced by a time trend for new house price model and by average household size for existing home price model.
  - Key quantitative findings:
    - Marginal income coefficient: 0.7 for new homes and 0.5 for existing homes.
    - Construction costs coefficients: 0.751 (New House Prices supply equation) and 0.554 (Existing House Prices supply equation).
    - Number of houses sold coefficients (supply equations): 0.224 (New) and 0.194 (Existing).
    - Regional gaps: house prices are mostly in line with model predictions in the Midwest and South; in the West and Northeast actual prices are somewhat above model predictions, with a gap between actual and predicted values of 15–25 percent in the West and somewhat smaller in the Northeast.
  - Interpretation notes:
    - The income coefficients are broadly consistent with existing literature and suggest new houses are viewed as more luxurious relative to older houses.
    - Statistical caveats: slow price adjustments reduce power of cointegration tests; a cointegrating vector was found in the model for new house prices, but none for existing home prices, complicating interpretation of estimated coefficients and their significance.

*Source: IMF staff estimates and cited studies in the provided content.*

### 2. Defined Benefit Pension Plans:

### _cr03245 - 2. Defined Benefit Pension Plans:

### A. Factors Underlying the Erosion of DB Plans
- In 1999, pension plans of firms in the S&P 500 had been overfunded, in net present value terms, by nearly $250 billion; by end-2001 this surplus was exhausted, and by end-2002, DB plans had a deficit estimated at $216 billion (CSFB 2003).
- Rating agencies have begun to scrutinize pension obligations and have downgraded many companies with large liabilities, causing stock prices to fall and credit spreads to widen.
- Valuation and market developments:
  - Pension plans sold significant net amounts of equities during the latter half of the 1990s, yet equity price increases pushed the share of pension plan assets held in equities to around 55 percent by the end of the decade.
  - Valuation losses totaled roughly $400 billion between end-1999 and end-2002.
  - Lower long-term interest rates used to discount future pension payouts have significantly increased the net present value of pension obligations.
- Tax and contribution incentives:
  - Contributions are deductible for income tax purposes only if the plan is underfunded; contributions that take the plan above this point are subject to corporate income tax and an excise tax, discouraging contributions when plans are in surplus.
- Longer-term perspective:
  - Federal Reserve Flow of Funds data suggest that if the share of DB plan assets in equities had been strictly limited to 40 percent from the 1980s, plan assets would have been $250 billion below their actual end-2002 level.
  - Capital gains averaged nearly $200 billion per year during the latter half of the 1990s.
  - Cumulative stock market gains since 1985 fell from a peak of $1.4 trillion in 1999 to around $1 trillion at the end of 2002.
- Table 1 (as presented in source) — Defined Benefit Corporate Pension Plans: Portfolio Composition (In percent of total financial assets, end of year):
  - 1985 1999 2002
  - Corporate Equities425543
  - Government Bonds191115
  - Corporate Bonds101115
  - Mutual Fund Shares165
  - Other271722

### A (continued). Representative sample analysis
- A representative sample of 19 underfunded pension plans:
  - These 19 firms account for $110 billion of the total underfunding (slightly more than half of the S&P 500 total).
  - For these companies, nearly half of deterioration in funding since 2000 resulted from stock market losses; the balance arose from increased benefit obligations (due to lower discount rates) and net payouts.
  - Over 1997–2002, equity market holdings made a positive contribution to funding levels, in excess of $100 billion.
- Table 2 (selected illustrative figures from company reports, year-end; in billions of dollars):
  - All 19 companies: 1997–1999 53.9; 2002 -110.3 (presented in source as part of a larger table)
- Table 3. Change in Funding Status of Defined Benefit Pension Plans (Selected companies, billions of dollars):
  - 2001–2002 and 1997–2002:
    - Overall change in funding status-147-129
    - Capital gains/loss(-)-64104
    - Change in pension benefit obligation-67-113
    - Net payout in benefits-33-101
    - Other17-19

### B. Pension Funding and Profits
- Funding rules and required contributions:
  - Underfunded plans are required to return to full funding over a five- to 30-year period.
  - If a plan is less than 90 percent funded, additional contributions are required that would take the plan back to 90 percent within a three- to five-year period; this higher requirement applies only if the plan has been under the 90 percent threshold for two of the last three years, or if it is less than 80 percent funded.
  - Companies can apply for a three-year waiver; the Labor Department can extend amortization periods by as long as ten years.
  - Legislation in 2002 raised the interest rate plans are required to use to calculate the present value of pension liabilities to 120 percent of the 30-year Treasury bond yield; legislation is being considered to allow discounting using corporate bond rates.
- Recent and prospective profit impacts:
  - S&P 500 firms tripled their pension contributions in 2002 over the previous year to $46 billion.
  - As a result, the growth of economic profits was kept to 7½ percent in 2002, 5 percentage points lower than would have been the case if contributions had been unchanged (CSFB, 2002).
  - For the 19 firms examined, benefit payments currently exceed contributions by a factor of three.
- Macroeconomic simulation of pension fund finances (Fund staff estimates; corporate contributions and underfunding are S&P 500 aggregate figures; impact on profit growth in percent):
  - Actual/Observed:
    - 2002 Corporate Contributions $46 billion; Underfunding of S&P 500 Companies $216 billion; Impact on Profit Growth -5
  - Baseline scenario:
    - 2003 Corporate Contributions $60 billion; Underfunding $170 billion; Impact on Profit Growth -2
    - 2004 Corporate Contributions $70 billion; Underfunding $110 billion; Impact on Profit Growth -1
  - Strong recovery scenario:
    - 2003 Corporate Contributions $46 billion; Underfunding $65 billion; Impact on Profit Growth 0
    - 2004 Corporate Contributions $46 billion; Underfunding $65 billion; Impact on Profit Growth 0
  - Weak recovery scenario:
    - 2003 Corporate Contributions $85 billion; Underfunding $220 billion; Impact on Profit Growth -5
    - 2004 Corporate Contributions $100 billion; Underfunding $210 billion; Impact on Profit Growth -2

### C. Pension Funding and the PBGC
- PBGC exposures and recent developments:
  - The PBGC assumed additional $9 billion in benefit obligations over the past two years from failed plans; valuation losses on its own assets contributed to erosion of its net actuarial position into a $3½ billion deficit.
- PBGC financials (Single-Employer Program, billions of dollars; source: PBGC, Annual Report, 2002):
  - Year 2001 2002
  - Financial Position
    - Total assets21.825.4
    - Total liabilities14.029.1
    - Net loss-2.0-11.4
    - Net position7.7-3.6
  - Summary of Operations
    - Premium income0.80.8
    - Losses from terminations0.79.3
    - Investment income (loss)-0.70.3
    - Benefits paid1.01.5
- Risk assessment and policy tradeoffs:
  - Stochastic simulations in the PBGC’s 2002 annual report suggest a 30 percent probability that the PBGC would be in a surplus position by 2012 in the absence of measures.
  - Current annual premiums are $19 per plan participant, plus an additional charge of $9 per $l,000 of unfunded vested benefits.
  - Raising premiums or shifting to risk-based premiums could create adverse selection, encouraging healthier firms to terminate DB plans and switch to defined-contribution or 401(k)-type plans.
  - Moral hazard exists because financially weak firms can continue to promise generous pension benefits that the PBGC would largely cover in bankruptcy.
  - Restoring PBGC’s financial position may require balancing premium structure changes with proactive oversight to avoid imposing additional large obligations on the Corporation.

### D. Pension Accounting, Valuation, and Earnings
- Accounting treatment problems:
  - Current accounting rules allow firms to calculate pension plan earnings using an expected return on pension assets rather than actual returns, which avoids short-term asset price volatility in reported earnings but can obscure firms’ underlying financial positions during prolonged market downturns.
- Magnitude of misreporting:
  - Many firms reduced expected return assumptions by a percentage point or more over the past two years, but the median expected return is still above 8 percent, well above the sharply negative returns actually achieved in 2001 and 2002.
  - For the representative sample of 19 firms, reported earnings exceeded actual returns by $130 billion during the past two years.
  - Income booked on pension plan returns between 1997 and 1999 was $65 billion less than actual gains for those firms (reflecting assumed returns of 8 to 10 percent versus actual S&P 500 returns over 1995–1999 averaging over 25 percent).
- Market valuation distortions:
  - Research by Coronado and Sharpe (2003) suggests investors placed “an unjustifiably high valuation” on firms with substantial pension earnings because valuations did not fully differentiate between core earnings and pension earnings.

### E. Concluding Observations and Policy Implications
- Key conclusions:
  - The recent financial problems of DB plans are not wholly the result of excessive equity investment; valuation gains in the 1990s, insufficient contributions during the 1990s, long-standing demographic problems, and the decline in bond yields have all been important contributors.
  - Pension shortfalls may adversely affect corporate profits in the near term, but pressures could abate with an economic recovery that boosts stock valuations and raises interest rates. Baseline and weaker scenarios would still leave significant underfunding.
  - Restoring the PBGC’s financial position is likely to require additional measures; care is needed to ensure insured firms do not impose additional burdens on the Corporation.
  - Improving accounting transparency for pension fund results—requiring more explicit reporting of the impact of plan returns on reported earnings and presenting this information more prominently in profit and loss statements—would reduce misrepresentation.
  - Tax policy adjustment: relaxing tax penalties against contributions to fully funded pension plans could strengthen plan finances. One proposal is allowing firms to deduct contributions to plans with higher funding levels—perhaps 110 percent rather than the current 100 percent limit—to encourage full funding over the cycle while limiting tax avoidance opportunities.
- Additional observations on sensitivity:
  - Projected Benefit Obligations (PBOs) are extremely sensitive to interest rates; one estimate cited indicates each 50 basis point increase in interest rates reduces the PBO for S&P 500 firms by $60 billion (CSFB 2003).
  - Funding problems concentrated in sectors with significant excess capacity—airlines and autos—could mute marginal impacts on investment.

*Source: _cr03245 - 2. Defined Benefit Pension Plans*

### 2.      The turnaround in the fiscal situation—and the Administration’s call for further

### 2.      The turnaround in the fiscal situation—and the Administration’s call for further

### A. Historical experience and simulations of the FY 2004 budget proposal
- Context and debate
  - The Administration’s FY 2004 budget proposals revived debate on whether tax cuts generate supply-side benefits sufficient to offset adverse effects of larger fiscal deficits on interest rates and private investment (CEA, 2003).
  - Alternate view: higher deficits would ultimately lower output.

- Historical episodes of large fiscal expansion
  - Three recent cases when the federal fiscal balance fell by at least 1½ percent of GDP, in cyclically adjusted terms, over a two-year period (Figure 1).
  - During 1965-67, structural outlays rose by 1.8 percent of GDP (military spending); tax cuts caused structural revenues to fall by 1.1 percent of GDP between 1974-76 and by 2.3 percent of GDP during 1981-83.
  - Although tax cuts were associated with some acceleration in real GDP growth over each of the subsequent three years, real GDP growth declined, or remained essentially unchanged in each of the subsequent ten years (Table 1).

- Model simulations of FY 2004 proposals
  - Macro-econometric models (Macroeconomic Advisors (MA) and Global Insight (GI)) predict a significant positive effect on output growth over the next two years, but the boost to aggregate demand wanes thereafter owing to crowding-out of private investment from higher real interest rates.
  - In the MA model, lower capital accumulation driven by the proposals lowers labor productivity by about ½ percentage point in 2017, relative to the baseline (Figure 2).
  - CBO study: in a “textbook” neoclassical growth model the budget would lower GDP ¾ percent below baseline during 2009-2013. Using life-cycle and infinite-horizon forward-looking models, the budget proposals would only increase long-run output if tax cuts are anticipated to be reversed in the future (Table 3).
    - Table 3 (selected average changes in GDP from CBO baseline, in percent):
      - Textbook Growth Model: 2004-2008 = -0.2; 2009-2013 = -0.7
      - Closed Economy Life-Cycle Growth Model (Lower government consumption after 2013): 2004-2008 = -0.3; 2009-2013 = -1.5
      - Closed Economy Life-Cycle Growth Model (Higher lump-sum taxes after 2013): 2004-2008 = 0.5; 2009-2013 = 0.3
      - Open Economy Life-Cycle Growth Model (Lower government consumption after 2013): 2004-2008 = -0.6; 2009-2013 = -0.5
      - Open Economy Life-Cycle Growth Model (Higher lump-sum taxes after 2013): 2004-2008 = 0.3; 2009-2013 = 0.6
      - Infinite Horizon Growth Model (Lower government consumption after 2013): 2004-2008 = 0.2; 2009-2013 = -0.6
      - Infinite Horizon Growth Model (Higher lump-sum taxes after 2013): 2004-2008 = 0.9; 2009-2013 = 1.4
  - CBO also notes that in an open economy context, net inflows of foreign capital can help offset the decline in national savings and alleviate crowding out.

### B. Fiscal deficits and real interest rates
- Empirical focus
  - A key indicator of crowding-out risk is the impact of budget policies on interest rates; most empirical work focuses on the connection between fiscal deficits and interest rates.

- Results from large-scale macro-econometric models
  - These models generally indicate fiscal deficits have a sizeable effect on interest rates. The size of crowding-out depends on: the monetary policy reaction function, the interest-rate sensitivity of investment, openness of the economy, and how expectations of future policies are modeled.

- Findings from surveys and studies
  - Gale and Orszag (2002): average prediction from these models is that a 1 percentage point increase in the primary deficit-to-GDP ratio, caused by a tax cut, is followed by a 40 basis point increase in long-term interest rates after one year, and a 60 basis point increase after ten years. If the same increase in the primary deficit is induced by higher government spending, the increases are 60 basis points after one year and 130 basis points after ten years.
  - Reduced-form econometric studies have provided mixed results partly because of difficulty capturing responses to expected future fiscal policies. More recent papers (surveyed in Table 4) that address expectations find a positive and significant impact of expected budget deficits on expected future interest rates—averaging 35 basis points for a one percentage point increase in the deficit-to-GDP ratio (roughly in line with large-scale model estimates).
  - Elmendorf and Mankiw (1999) neoclassical framework (used in Administration estimates): assuming a one dollar increase in public debt reduces the long-run stock of capital by 60 cents, a 1 percentage point increase in the debt-to-GDP ratio leads to an increase of real interest rates of only around 2-3 basis points (CEA, 2003).
  - Laubach (2003) argues the above could be consistent with an increase of interest rates of approximately 15 basis points following a permanent one percentage point increase in the deficit-to-GDP ratio.
  - The CEA’s relatively modest interest-rate effect is predicated on substantial crowding out: in their example, a 5 percent of GDP increase in government debt would lower the capital stock by around 3 percent of GDP; given estimates of the gross marginal productivity of capital of around 10 percent, this implies a permanent reduction in output of roughly one third of a percent.

- Selected quantitative survey results (representative findings reported in source)
  - Reduced-form / model averages and selected study findings summarized:
    - Large-scale model average: roughly 40 bps (1 year) and 60 bps (10 years) for a 1 percentage point primary deficit-to-GDP increase due to tax cuts; larger for spending-induced deficits (60 bps and 130 bps).
    - Recent reduced-form estimates: average ~35 bps increase in expected future interest rates per 1 percentage point increase in deficit-to-GDP.
    - Elmendorf/Mankiw CEA estimate: ~2-3 basis points per 1 percentage point increase in debt-to-GDP (steady-state).
    - Laubach (2003): approximately 15 basis points for a permanent 1 percentage point increase in deficit-to-GDP.

### C. International implications of higher U.S. public debt
- Global linkages and motivation
  - Integration of capital markets over the last three decades raises the possibility of important spillovers from U.S. fiscal policy to the rest of the world: higher fiscal deficits and public debt in one country tend to absorb global savings and might cause higher world interest rates.

- Evidence that real interest rates co-move
  - Country-specific real interest rates have tended to move together over the last three decades. Table 5 shows positive and generally quite high real interest rate correlations for industrialized countries, suggesting a “world” real interest rate.
  - Figure 3 presents proxies for a world real interest rate (unweighted average, GDP-weighted average, and first principal component) and shows significant increases during most of the 1980s (a period of rapid growth of world public debt), followed by declines over most of the 1990s despite still-high levels of world public debt.

- Studies on world fiscal variables and interest rates
  - Helbling and Wescott (1995), Orr and Conway (2002), and Ford and Laxton (1999) find that net public debt (world or national) is a significant determinant of the “world” real interest rate or national real interest rates.
    - Ford and Laxton (1999): a 1 percentage point increase in world net government debt raises real interest rates by around 20 basis points.
  - A re-examination using 11 industrialized countries over 1977-2002 (G-7 plus Belgium, Netherlands, Switzerland, Denmark) regressed national real interest rates on world fiscal variables (world net public debt-to-GDP ratio and share of real GDP absorbed by government consumption and investment). System estimates using GMM and instrumental variables were also performed to address potential endogeneity.

- Empirical results summary
  - Regression results generally confirm that an increase in world public debt affects national real interest rates, but cannot rule out a break in the relationship over the 1990s.
  - OLS coefficients on world fiscal variables generally have the expected signs, but are significant at the 5 percent level in only a few cases.
  - Augmenting OLS regressions with country-specific public debt-to-GDP ratios generally does little to improve results; the country-specific coefficient is rarely both significant and positive.

*Prepared by Roberto Cardarelli and Ayhan Kose; content as presented in the source chapter.*

### 18.      The system estimates show that both world public debt and government absorption

### _cr03245 - 18.      The system estimates show that both world public debt and government absorption

### World fiscal variables and national real interest rates — main findings
- A 1 percentage point increase in the world government debt to GDP ratio induces an increase in national real interest rates of around 10 basis points over the 1997-2002 period.
- The estimated effect is robust to addition of other variables (business cycle, monetary policy, inflation changes), but stability of coefficients over the period is difficult to test given the relatively scarce number of observations.
- Most desirable properties of GMM estimators are only valid asymptotically; point estimates should be taken with caution.
- Using these estimates, the 15 percentage point increase in the U.S. public debt ratio projected over the next decade by the CBO would lead to an average ½-1 percentage point increase in national real interest rates.

### Empirical approach and diagnostic notes
- Regressors include world GDP-weighted net government debt to GDP (WGND) and the GDP-weighted average of national real government absorption (WGA), plus the first difference of WGA (DWGA) and other world-weighted controls.
- The dependent variable: national 12-month Euromarket interest rates deflated by same-period CPI inflation.
- Data are semiannual from 1977:1 to 2002:2 (system observations: 566). Newey-West heteroskedasticity and autocorrelation-consistent standard errors reported; p-values are shown in square brackets in tables.
- Instrument set: second and third lags of WGND and WGA; Wu-Hausman tests reported support exogeneity of WGND in OLS regressions but could not rule out exogeneity of government consumption.
- A Chow test on two equally sized sub-samples rejects the null of stability of coefficients.

### Key regression coefficients and diagnostics (selected)
- WGND (world net government debt to GDP): coefficient ~0.12 (p-values [0.00] across reported specifications).
- WGA (GDP-weighted real government absorption): coefficients around 1.34 to 2.79 (p-values [0.00]).
- DWGA (first difference of WGA): mixed signs; selected reported value 1.24 with p-value [0.00] in some specifications and -0.04 with p-value [0.88] in another.
- Euro Dummy (1999:1-2002:2): negative coefficients (e.g., -1.47) with p-values [0.00] in several specifications.
- J-statistics (model specification test) reported with p-values indicating the null that the model is well specified is not rejected (examples: J-statistic 2.75 [0.99]; 10.80 [0.46]; others with p-values [0.99], [0.78], [0.89]).

### State and local governments (SLGs) in the United States — context and trends
- SLG sector: current expenditures by SLGs have grown strongly in recent decades and account for nearly all of the 7 percent of GDP increase in general government spending since 1960.
- SLG investment has remained essentially constant in relation to GDP; given the decline in federal investment, SLGs became the principal source of public investment.
- Growing SLG expenditures financed by tax and other revenue increases amounting to 4½ percent of GDP since 1960, and an increase in federal grants of 2 percent of GDP.
- Federal grants account for almost one quarter of total SLG revenues in 2002.

### Recent fiscal developments for SLGs (selected statistics and dynamics)
- At the end of the 1990s, SLGs ran current surpluses up to around ½ percent of GDP; with the economy weakening, current deficits reached a post-war peak of ½ percent of GDP in 2002.
- A deficit exceeding ¾ percent of GDP seems likely in FY 2004, with almost 90 percent of states projecting revenue shortfalls that will exceed 5 percent of their general funds.
- Health-related spending grew by ½ percent of GDP during the downturn period, driven mostly by Medicaid.
- SLG spending on income-support and welfare programs rose by 0.1 percent of GDP between 1999 and 2002.
- Corporate and personal income tax revenues (roughly one fifth of total state receipts) declined by roughly ¼ percent of GDP during 2001-2002.
- Federal grants-in-aid to state and local governments as percent of total receipts: 0.8 (1960), 2.6 (1980), 1.9 (1990), 2.9 (2002).
- State and local governments’ share of current receipts as percent of GDP: 8.0 (1960), 11.3 (1980), 11.4 (1990), 12.5 (2002).
- Composition of state and local receipts (in percent of total receipts, 2002): Personal income tax receipts 15.4; Corporate profits tax accruals 2.6; Sales taxes 25.6; Property taxes 20.5; Federal grants-in-aid 23.4.

### Drivers of SLG revenue and expenditure changes
- Revenue decline drivers:
  - Economic slowdown reduced labor incomes; collapse of stock market eroded capital gains (large effect in California and East Coast).
  - Prior state tax rate cuts left states more dependent on cyclically-sensitive income tax revenue.
  - Federal tax cuts (2001 and 2003) estimated to lower state tax revenues by about $5 billion (with specific 2003 provisions estimated to cost states $3 billion absent state countermeasures).
- Expenditure pressures:
  - Increased Medicaid generosity and broader U.S. health-care cost pressures.
  - Federal mandates that require states to provide specified services (e.g., Medicaid eligibility and mandatory categories), shifting costs to states.

### Fiscal rules and adjustment capacity at the state level
- All states but one have balanced-budget requirements, determined by state constitutions or law; typically apply only to current budgets.
- States may borrow for capital spending; many can balance budgets on an ex ante basis and can delay payments or build arrears to shift spending across years.
- Until recently, states could draw on significant reserve funds accumulated during the 1990s surpluses.

*Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2003/_cr03245.pdf*

### 10. However, the depletion of

### _cr03245 - 10. However, the depletion of

### Reserve funds and recent use
- By the end of FY 2000, state reserve funds stood at about 10 percent of state expenditures, compared to less than 5 percent at the end of the 1980s (Table 4).
- In recent years, some 16 states have had to cover their deficits by drawing down these reserves, leaving overall reserve balances virtually exhausted by end-FY 2003.
- Some studies estimate that states would need reserves of more than 18 percent of expenditures to accommodate a macroeconomic shock of the magnitude of the 1990-91 recession (Lav and Berube, 1999).
- Note in-source Table 4 line as presented: Total reserves  2/ 48.841.022.06.3
- Footnote: 2/ Sum of general fund balances and rainy-day funds.

### Adjustments undertaken by states (FY 2002–FY 2003)
- Spending-side measures implemented:
  - Hiring freezes.
  - Cuts in spending for prisons, education, childcare, and support for local governments (NASBO and NGA, 2003).
  - Medicaid largely excluded from cuts because of cost-sharing with the federal government, though states tightened eligibility for optional participants and adopted several cost-saving measures (including preferred drug lists).
- Revenue-side measures and sources:
  - Little emphasis so far on tax hikes.
  - Tobacco settlement funds amounted to $32 billion between 1998-2002 (Lindblom, 2003) and may be used to cover revenue shortfalls.

### Borrowing, debt, and credit effects
- States increased borrowing; some reclassified operating expenses as capital expenditures.
- Market debt owed by state and local governments increased from 12 percent of GDP in 2000 to 14 percent in early 2003, compared to an 18 percent peak during the 1990-1991 recession (Figure 3).
- State credit ratings and risk premiums have not been significantly affected overall, except for several states facing more severe financial difficulties (Figure 4).

### Outlook for FY 2004 and near-term fiscal stance
- Budget difficulties expected to worsen in FY 2004.
- National Governors Association surveys suggest further cuts in program expenditures, including education, human and health services, and aid to local governments.
- State spending is expected to fall by around ¼ percent in real terms in FY 2004.
- Governors in 29 states recommended tax and fee increases for FY 2004 with an expected yield of $17.5 billion (or 0.2 percent of GDP)—the largest since 1979.

### Aggregate fiscal drag and interactions with federal policy
- Estimated adjustment by SLGs to satisfy balanced-budget requirements would result in a fiscal contraction of about ¼ percent of GDP in 2003, offsetting only a small part of a 1¾ percent of GDP fiscal stimulus at the federal level.
- SLG policies are not expected to add to the slight withdrawal of federal stimulus in 2004 (Table 5).
- Table 5 (calendar year data; in percent of GDP) excerpts as presented:
  - Change in actual balances (NIPA basis)
    - General government 0.7-1.9-3.0-1.80.6
    - Federal government 0.5-1.6-2.7-1.90.4
  - Change in structural balances
    - General government 0.5-1.1-2.7-1.50.4
    - Federal government (budget basis) 0.5-1.2-2.4-1.70.4
- Footnotes: Sources: Budget of the U.S. Government, various issues; and Fund staff estimates. 1/ Estimated.

### Fiscal multipliers and uncertainty
- Literature-based ranges for fiscal multipliers (primarily for general government):
  - Spending increases: 0.3-1.4
  - Tax cuts: 0.2-1.3 (Hemming, et al., 2002)
- Low-end estimates compatible with Ricardian offset; some studies suggest fiscal multipliers can turn negative if fiscal policy increases uncertainty or crowds out private investment.
- VAR model approach employed to account for feedbacks between federal and SLG fiscal variables; fiscal variables used: tax revenues net of transfers to persons; public consumption expenditure; federal grants to SLGs. Four lags were used.

### VAR results: dynamic effects of SLG vs. federal fiscal shocks
- A one standard deviation shock to the share of SLG consumption spending in GDP:
  - Would reduce the output gap—hence increase GDP—by 0.4 percentage points immediately, with the effect slowly decreasing to almost zero by the fifth quarter.
- A one standard deviation shock to SLG net taxes:
  - Has negligible effect on GDP in the first quarter, with the impact slowly building and reaching almost 0.4 percentage points in the fifth quarter; the effect dissipates completely after 6 quarters.
- Comparative effects:
  - A one percentage point increase in net federal taxes as a share of GDP would have no significant impact on the output gap.
  - A one percentage point increase in federal spending would reduce the output gap by about 0.2 percentage points in the first quarter, with the effect dissipating entirely after three quarters.
- Caution: VAR estimation limitations noted—many SLG budget components in NIPA are available only with a two-year lag, most recent data are estimated, and quarterly data are interpolated from annual data.

### Effects of energy price shocks (IMF GEM model)
- Model calibration and assumptions:
  - Two-country GEM (United States and rest of world).
  - Energy (oil and natural gas) consumption set at 2.4 percent of GDP (real producer prices), of which 1.1 percentage points are produced domestically and the remainder are imported.
  - Roughly half of energy consumed is used in production of intermediate goods and half in the final consumption good.
- Shock experiments:
  - A 50 percent increase in the price of oil and gas is simulated.
  - Three durations considered:
    - Increase lasts for one quarter (solid).
    - Increase with one-year half life (dashed).
    - Increase with a five-year half life (dotted).
- Key quantitative results:
  - One-quarter shock: real GDP reduced by roughly ¾ percent relative to baseline in the first quarter; return to baseline almost instantaneous; long-term impact negligible.
  - Long-lived shock (five-year half life): maximum GDP drop roughly 1 percent occurring after three quarters; GDP 0.4 percent below baseline after ten years.
  - Under long-lived increase, consumers bear most adjustment: initial increase in current account deficit raises foreign debt; servicing/repayment depresses consumer spending relative to baseline for a sustained period.
  - Investment initially falls due to higher user cost of capital (tighter monetary policy) and lower return to capital from increased costs and limited ability to raise output prices; subsequently investment may rise above baseline as capital stock re-equilibrates.
- Monetary policy and inflation response:
  - Monetary policy modeled with an inflation-targeting reaction function that anchors inflation expectations.
  - CPI inflation increases for a short period initially, then returns to baseline relatively quickly, aided by nominal wage stickiness assumption (workers do not attempt to maintain real wages under persistent energy price increases).

*Source: IMF staff report content (sections 10–VI excerpt).*

### 9. However, there are several reasons why these simulations may understate the

### _cr03245 - 9. However, there are several reasons why these simulations may understate the

### Limitations of the simulations and sources of understatement
- The model assumes the deterioration in the U.S. current account would be financed by increased borrowing from the rest of the world; if the ability to borrow is constrained (including by shifts in confidence or portfolio preferences), U.S. consumption and investment may decline more than the simulations suggest.
- If the monetary authority were more accommodative of the inflationary impact of the shock and workers bargained aggressively to maintain their real consumption wage, the persistent shock could lead to more persistent CPI inflation and longer-lived real output effects.
- Simulations assume energy-exporting countries (included in the rest-of-the-world block) increase consumption in proportion to higher energy revenues; if energy exporters’ saving rates temporarily increased, demand for U.S. exports could be weaker than simulated.
- Historical large energy price shocks often coincided with significant geopolitical events that affected investor and consumer confidence; the shocks considered here do not include such effects.
- Footnote findings preserved:
  - Simulations assuming that real energy prices are expected to return to baseline more gradually do not show an appreciably larger first-quarter impact on real GDP or headline CPI inflation. Even if the shock is expected to be permanent, the first quarter impact on real GDP is only 1 percent.
  - The two-country setup implies the rest of the world (which includes energy exporters) experiences a positive terms of trade shock; initially output in the rest of the world declines slightly more than in the United States since production is assumed to be more energy-intensive. Under more persistent shocks positive income effects eventually lead to a much smaller decline in absorption than in the United States; positive effects would essentially be confined to oil exporters.

### Conclusion on impact of temporary energy price shocks
- Simulation result: impact on U.S. growth of temporary energy price shocks should be mild.
- A short-lived spike in energy prices (such as occurred during the past year) would have a modest and short-lived impact on growth.
- Even with expectations that a shock would be longer-lived, the impact on growth would be only marginally larger.
- Caution: results must be interpreted with caution because the analysis does not incorporate confidence effects that have accompanied past oil price shocks.

*References cited in the source: Hunt, B., 2003; Pesenti, P., 2003.*

### VII. ENERGY POLICY IN THE UNITED STATES: THE ROLE OF TAXATION — overview
- Context: Post-2001 Administration’s National Energy Policy (NEP) debate driven by geopolitical/economic consequences of U.S. dependence on oil imports and environmental consequences (including greenhouse gas emissions) of U.S. energy intensity.
- Observation: None of the initiatives emphasized taxes to discourage energy consumption; focus has been on boosting domestic supply and technology and on tax subsidies for production and efficiency at substantial fiscal cost.
- Proposition: The chapter suggests considering consumption-based energy taxes to meet energy and fiscal policy objectives.

### A. Energy use in the United States — key facts and statistics
- U.S. energy intensity of GDP has fallen steadily over the last half-century but remains well above most other industrial countries (Figure 1).
- U.S. consumption remains 30–50 percent higher than in Europe.
- Cross-country price differences (1996 example): average U.S. gasoline prices in 1996 were more than 50 percent below European prices and 10-15 percent lower than Canadian prices.
- Sector shares of Btu consumption in 2001:
  - Industrial: 33 percent
  - Transportation: 28 percent
  - Residential: 21 percent
  - Commercial: 18 percent
- Fuel shares and trends:
  - Petroleum share fell from nearly 50 percent in the mid-1970s to around 40 percent by the end of the 1990s.
  - Natural gas share: peaked at 32 percent in 1970 and now stands at around 25 percent.
  - Coal share: roughly similar to natural gas as main fuel for electricity generation.
  - Electricity from nuclear, hydroelectric, and other non-fossil sources: just under 15 percent of total energy use.
- Emissions:
  - U.S. greenhouse gas (GHG) emissions per unit of GDP are among the highest of major industrialized countries (Figure 5), broadly consistent with energy intensity rankings.
- Petroleum imports:
  - Net imports of petroleum projected by the U.S. Department of Energy to continue to grow strongly.
  - Share of net imports in total U.S. petroleum consumption expected to increase from 55 percent in 2001 to 68 percent in 2025.

### B. Energy policy — NEP and related measures
- NEP (May 2001) principal focus: address “fundamental imbalance between supply and demand” and projected increase in U.S. dependence on energy imports; promote “dependable, affordable, and environmentally sound production and distribution of energy.”
- NEP proposals included:
  - subsidies to promote conservation by households;
  - funding for research and development into alternative energy sources;
  - establishment of a new regulatory structure for the electricity sector, including extension of the tradable emissions permit system on sulphur dioxide and introduction of similar systems for nitrogen oxides and mercury;
  - revisions to emissions standards for autos and household appliances;
  - tax credits for fuel efficient vehicles, landfill methane projects, electricity from wind and biomass, residential solar energy property, purchase of new hybrid or fuel-cell vehicles;
  - opening the Arctic National Wildlife Reserve (ANWR) for oil exploration and pipelines and earmarking associated royalties for conservation.
- Congressional bills: House and Senate versions both include loan guarantees and tax credits for pipeline development; tax incentives for natural gas production; increased R&D funding; incentives for nuclear energy; tax credits for renewable energy sources including biomass and waste. House version would open ANWR and mandate increased use of ethanol in gasoline.
- Environmental policy stance:
  - Administration rejected the Kyoto protocol in 2001; Kyoto would bind countries to a 7 percent cut by 2008-2012 compared with 1990 levels; estimates placed the cost of achieving this reduction as high as 2 percent of GDP.
  - Administration proposed Clear Skies Initiative (2002): commitment to reducing U.S. emission intensity (GHG emissions as a share of real GDP) by 18 percent by 2012, to be met primarily through combined measures including extension of existing cap-and-trade programs.
  - Goulder (2002) cited: Administration’s target would leave emissions roughly 10 percent higher than at the beginning of the decade and nearly 30 percent above the Kyoto Protocol target.

### C. Reducing energy consumption: taxation and other instruments
- Taxes are widely viewed as an effective instrument to restrain demand, encourage efficient resource use, and align private and social costs in presence of externalities (Sandmo, 1976).
- Simulation evidence (Goulder and Schneider, 1999): achieving a 10 percent reduction in CO2 emissions would be ten times more costly if technology subsidies were used as a stand-alone measure compared with a broader approach combining technology subsidies with policies to raise the cost of carbon (tradable permits or carbon taxes).
- Studies (Parry, 2002; Parry and Small, 2002) using representative agent models identify key factors determining optimal fuel tax (in decreasing order of importance):
  - social cost of automotive congestion,
  - revenue-raising capacity of the tax,
  - extent fuel consumption imposes environmental externalities.
- These studies suggest U.S. and Canadian gasoline taxes may be only half their optimal level after adjusting for low population density and congestion externalities.
- Alternative approaches (regulation and tax incentives) have drawbacks:
  - CBO (2002) comparison: gasoline taxes are considerably less costly from economic efficiency perspective than CAFE-like standards; CAFE standards do not directly target fuel-saving activities by the consumer.

### Box 1. Energy excise taxes in the United States — federal and state key figures
- Federal level:
  - Fuel taxes average $0.184 per gallon.
  - Joint Committee on Taxation and IRS estimates federal fuel taxes yielded $29.6 billion (0.3 percent of GDP) in FY 2003.
  - Excise tax on coal yielded $550 million.
  - Excise tax on sale of automobiles with low fuel economy ratings yielded $78 million.
  - Specific excises include:
    - Tax of $0.43 per gallon rail diesel fuel and inland waterways fuel; $0.068 per gallon motorboat fuel, small engine gasoline, and special fuels.
  - Excises dedicated to designated funds include:
    - Abandoned Mine Reclamation Fund: Tax of $0.35 per ton of surface coal, $0.15 per ton of coal mined underground, $0.10 per ton of lignite (average tax estimated about $0.26 per ton in 1999).
    - Highway Trust Fund: Tax of $0.043 per gallon motor fuel.
    - Leaking Underground Storage Tank Trust Funds: Tax of $0.001 per gallon motor fuel.
    - Nuclear Waste Fund: Tax estimated to impose a 1.45 percent cost increment for power provided from nuclear energy in 1999.
    - Other dedicated funds and user fees noted for pipeline safety and uranium enrichment decommissioning.
  - Excise taxes dedicated to health-related trust funds include:
    - Black Lung Disability Trust Fund: Minimum of $0.55 per ton of coal or 4.4 percent of sales revenue if selling price is less than $25 per ton from surface mines or $12.50 per ton for surface coal.
  - Excise tax on sale of automobiles with low fuel economy ratings:
    - Tax ranging from $1,000 for an automobile rated between 21.5 and 22.5 miles per gallon (mpg) to $7,700 for an automobile rated at less than 12.5 mpg.
- State and local level:
  - In 2002, excise taxes on motor fuel represented 6 percent of total taxes collected by states.
  - Total state and local taxes on fuel varied from $0.08 per gallon in Alaska to $0.35 per gallon in New York.
  - Many states levy severance taxes on oil, gas and coal production; state energy severance taxes accounted for less than (text truncated in source).

*Italicized source attribution: IMF staff paper content from the supplied PDF excerpt.*

### 0.8 percent of total state tax revenue in 2002.

### _cr03245 - 0.8 percent of total state tax revenue in 2002.

### Macroeconomic effects of energy taxation: demand elasticities and revenue potential
- Short-run price elasticities for energy and fuel demand are estimated in the range of -0.13 to -0.26 (OECD, 2001a).
- Long-run elasticities for energy and fuel demand are estimated in the range of -0.37 to -0.46 (OECD, 2001a).
- A detailed survey of 97 econometric studies of the elasticity of demand for gasoline found:
  - Short-run elasticity averaged -0.26 (Dahl and Sterner, 1991).
  - Long-run elasticity averaged -0.86 (Dahl and Sterner, 1991).
- Fiscal revenue examples:
  - CBO estimates that a 15-cent hike in gasoline taxes could have raised $16 billion in additional budget revenue in 2003, more than doubling existing revenues (CBO, 2002).
  - OECD suggests an increase in fuel taxes of 40 cents per gallon could be justified given externalities associated with road use, but notes that roughly three quarters of U.S. carbon emissions are not taxed at all (OECD, 2001b).
  - A carbon tax of $100 per ton would have yielded $110 billion in 1999 (OECD, 2001b).

### Efficiency, policy design, and distributional considerations
- Subsidies for new fuel-efficient technologies are questioned given uncertainty in selecting technologies likely to yield significant payoffs (Sutherland, 1999).
- Market-oriented regulatory approaches:
  - Cap-and-trade emissions permit system for SO2 has generally been viewed as a success (CBO, 2000).
  - Drawback: no upper limit to costs polluters may incur to achieve quantitative targets.
  - Possible remedies: issue additional permits if permit prices exceed a ceiling; grant a percentage of free permits instead of auctioning them (Goulder, 2002).
- Use of additional tax revenue affects macro outcomes:
  - Adverse effects on output can be alleviated if revenue is used to lower taxes on labor or investment (Nordhaus, 1993; Bovenberg and Goulder, 1996).
  - Revenues can be used to compensate population segments vulnerable to tax increases (e.g., rural versus urban households; CBO, 2002).

### Staff GEM simulations: scenarios and quantitative results
- Simulation context: a version of the staff’s Global Economy Model (GEM) calibrated to the U.S. economy.
- A 10 percentage point increase in taxes on petroleum products used as intermediate production inputs:
  - Long-run U.S. GDP reduction: 0.03 percent (Table 1).
- A 10 percentage point tax levied on final consumption of petroleum products:
  - Long-run U.S. GDP reduction: 0.11 percent (Table 1).
- Mechanisms and international effects:
  - U.S. petroleum imports represent almost 20 percent of the world market, so part of the burden of higher U.S. taxes is shifted to the rest of the world through lower prices and an appreciated U.S. dollar.
  - Short-run effects of a U.S. tax on energy used in production could be positive due to different speeds of adjustment for producer and consumer prices.
  - Higher elasticity of substitution between petroleum products and other goods/services increases likelihood that domestic tax causes world prices to fall and mitigates U.S. output declines.
  - Simulations do not account for possible responses by world energy producers to changed market conditions.

### Table 1 — Simulated impact on the U.S. economy of energy taxes (Percent deviation from baseline)
- 10 percent tax on energy used in consumption:
  - After 1 Year: Real GDP -0.03; Consumption -0.03; Investment -0.30; Consumption price of energy 7.69; Goods producers’ price of energy -2.09; Oil producers’ price of energy -2.59; Real exchange rate 0.16.
  - After 5 Years: Real GDP -0.02; Consumption 0.01; Investment -0.30; Consumption price of energy 8.33; Goods producers’ price of energy -1.51; Oil producers’ price of energy -1.87; Real exchange rate 0.23.
  - Long Run: Real GDP -0.08; Consumption -0.04; Investment -0.22; Consumption price of energy 9.38; Goods producers’ price of energy -0.56; Oil producers’ price of energy -0.67; Real exchange rate 0.22.
- 10 percent tax on energy used in production:
  - After 1 Year: Real GDP 0.02; Consumption 0.03; Investment 0.00; Consumption price of energy -0.30; Goods producers’ price of energy 9.67; Oil producers’ price of energy -0.38; Real exchange rate 0.12.
  - After 5 Years: Real GDP 0.02; Consumption 0.05; Investment -0.01; Consumption price of energy -1.00; Goods producers’ price of energy 8.91; Oil producers’ price of energy -1.26; Real exchange rate 0.11.
  - Long Run: Real GDP -0.03; Consumption 0.00; Investment -0.13; Consumption price of energy -2.23; Goods producers’ price of energy 7.62; Oil producers’ price of energy -2.79; Real exchange rate 0.26.
- 10 percent tax on all energy:
  - After 1 Year: Real GDP -0.01; Consumption 0.04; Investment -0.30; Consumption price of energy 7.39; Goods producers’ price of energy 7.39; Oil producers’ price of energy -2.95; Real exchange rate 0.37.
  - After 5 Years: Real GDP -0.01; Consumption 0.06; Investment -0.32; Consumption price of energy 7.27; Goods producers’ price of energy 7.27; Oil producers’ price of energy -3.09; Real exchange rate 0.34.
  - Long Run: Real GDP -0.11; Consumption -0.04; Investment -0.35; Consumption price of energy 7.02; Goods producers’ price of energy 7.02; Oil producers’ price of energy -3.45; Real exchange rate 0.47.
- Source for table: Fund staff calculations.

### Recent energy policy initiatives (selected measures and fiscal magnitudes)
- Administration proposals:
  - Tax incentives costing $8 billion over ten years for development of clean coal technology; research and development of renewable energy resources; purchase of nuclear power plants; electricity produced using wind and biomass; purchases of solar panels for homeowners; purchases of hybrid gas-electric vehicles; and co-generation plants.
  - Implements restructuring of electricity sector to promote competition, and enhance reliability and efficiency.
  - Extends the Price-Anderson Act and provides incentives for new nuclear plants.
  - Opens Alaska’s Arctic National Wildlife Refuge (ANWR) for oil and gas exploration.
  - Supports construction of a pipeline from the Alaska North Slope.
  - Supports ethanol mandate.
  - Expedites study of impediments to oil and gas exploration on federal lands.
  - Provides tax breaks and funds for fuel efficient technologies.
  - Proposes to cut greenhouse gas intensity by 18 percent over ten years through a “voluntary” program and funding for research on climate change.
- House and Senate measures (comparative highlights):
  - House calls for approximately $19 billion in tax breaks to promote energy production and conservation.
  - Senate calls for approximately $16 billion in tax breaks over ten years for renewable energy and conservation programs and traditional fossil fuel energy producers.
  - House passed H.R. 6, Energy Tax Policy Act of 2003 (April 11, 2003) and Senate considered S. 517, Energy Tax Policy Act of 2002 (April 25, 2002).
  - House: Repeals Public Utility Holding Company Act (PUHCA) and reforms PURPA.
  - Senate: Allows FERC to issue rules creating a national wholesale electricity market (“Standard Market Design”) while permitting states to oversee retail markets; repeals PUHCA; repeals PURPA conditional on FERC findings.
  - House and Senate include provisions for loan guarantees and tax incentives for an Alaskan natural gas pipeline (80 percent loan guarantee for up to $18 billion noted in House version).
  - Renewable and alternative fuels: House establishes a blending requirement of 2.7 billion gallons of renewable fuel with gasoline in 2005; Senate mandates increase to 5 billion gallons by 2012 and requires 10 percent production from select renewable energy resources by 2020 in one provision.
  - Hydropower: incentives and licensing reforms appear across proposals.

### Globalization and business cycles: rising linkages and implications for synchronization
- Trade:
  - U.S. trade (sum of exports and imports) has grown roughly three times faster than output since 1960, reaching about 20 percent of GDP in 2002.
  - Trade growth has been even more rapid among other G-7 economies.
  - Vertical specialization has increased; processed imports as a share of merchandise exports rose for both the United States and other G-7 countries.
  - Vertical specialization explains 30 percent of the growth in industrialized countries’ exports from the 1970s to the 1990s (Hummels, Ishii, and Yi, 2001).
- Capital flows:
  - For the United States, the sum of inflows and outflows of foreign direct investment and other private capital flows surged from less than 3 percent in the early 1970s to more than 18 percent of GDP in 2000.
  - For other G-7 countries, gross capital flows have on average reached nearly 45 percent of output.
- Ambiguity of globalization’s effect on business-cycle synchronization:
  - Increased trade can heighten demand- and supply-side spillovers, but greater inter-industry specialization may insulate countries if industry-specific shocks dominate.
  - Increased capital mobility can increase synchronization via shock transmission, but financial globalization can also enable specialization that insulates countries (Kalemli-Ozcan, Sorenson, and Yosha, 2003).
  - Empirical studies present mixed evidence on changes in synchronization since 1960 or 1970s (Heathcoate and Perri, 2003; Helbling and Bayoumi, 2003; Doyle and Faust, 2002; Kose, Prasad, and Terrones, 2003; Kose, Otrok, Whiteman, 2003a).
- Analytical approach:
  - Three sub-periods considered to isolate common shocks from globalization effects: Bretton Woods (1960:1–72:4); common shocks period (oil price shocks and subsequent disinflation); globalization period.

*Sources: EIA (1999), Lazzari (2003), U.S. Census Bureau (2003), CBO (2002), and Fund staff calculations.*

### 4. Volatility of GDP Growth in Different Periods

### 4. Volatility of GDP Growth in Different Periods

### Overview and periodization
- Analysis divides the postwar experience into three periods: Bretton Woods; Common Shocks (1973:1–86:2); and the globalization period (1986:3-2001:4), which "coincided with dramatic increases in the volume of international trade and financial flows."
- Comparison across these periods highlights changes in both the volatility of national aggregates and the co-movement of business cycles across the G-7.

### Evidence from volatility and correlations
- The standard deviation of U.S. real GDP growth "has declined significantly since the 1960s"; this decline is "true for the rest of the G-7, except for Germany and Japan, where output volatility has increased in the past decade."
- The correlation between U.S. output and output in:
  - Canada, France, and the United Kingdom has increased.
  - Germany and Japan has declined.
- Casual inspection of factor estimates: the world factor aligns with major business cycle episodes of the past 40 years (expansion of the 1960s/early 1970s; recessions of mid-1970s, early 1980s, early 1990s; expansion of late 1980s; synchronized severe downturn in 2000).
- Note: "Both the world factor and the estimated U.S. country factor capture some of the NBER reference cycle dates" but the U.S. country factor and world factor diverge in the 1990s (U.S. country factor captures prolonged U.S. expansion while the world factor displays downturns); the world factor "appears to have acted as a significant drag on U.S. growth in the latter half of 2001."

### Dynamic latent factor model: structure and key quantitative results
- Model decomposition (per Kose, Otrok, and Whiteman (2003b)):
  - 1 single world factor (fworld)
  - 7 country-specific factors (ficountry, one per country)
  - 21 idiosyncratic factors specific to each observable (εi,t)
- Observables: output, consumption and investment for each of seven countries (21 time series; 21 regression equations).
- Variance-decomposition findings:
  - The world factor is responsible for "more than 25 percent of G-7 output variation, and for more than 15 percent of the volatility of consumption and investment."
  - Importance of world factor differs across countries:
    - "roughly 60 percent of output variation in France"
    - "less than 13 percent in the United States"
  - Over time, the share of output variance explained by the world factor increased "from roughly 7 percent in the Bretton Woods period to 19 percent in the globalization period."
  - The share of investment variance explained by the world factor "tripled" between those periods.
  - There was "only a marginal increase in the average variance of consumption explained by the world factor," implying remaining consumption-smoothing opportunities across countries.
- Supplementary correlation note: Kose, Otrok, and Whiteman (2003a) report that "the correlation between the median world factor and U.S. output growth is 0.62."

### Country-specific observations
- France: world factor explains roughly "60 percent of output variation."
- United States: world factor explains "less than 13 percent of output variation."
- Germany and Japan: world factor "appears to have been less important in explaining output volatility," reflecting large domestic shocks (Japanese asset-price collapse and banking crisis; German aftershocks of unification); however, "given the current export dependence of both countries’ economies, global developments have again gained in importance in recent years."

### Key implications and policy-relevant conclusions
- "Global factors have played a larger role in explaining the dynamics of business cycles in the G-7 countries during the period of globalization." Exceptions: Germany and Japan where country-specific and idiosyncratic factors temporarily overshadowed trade and financial linkages.
- Specific implications highlighted in the source:
  - "The U.S. role in the global economy has increased. Global factors have become more important in driving domestic business cycles during the past decade, and related research shows that the U.S. economy has been a major force for global growth."
  - "A rebound in economic activity in the United States could have large spillover effects for other G-7 countries. This would be especially true for Canada and the United Kingdom, where global factors have become more important in explaining business cycle fluctuations in recent years."
  - "However, it is less likely that an increase in output in other G-7 countries would have as large an impact on the U.S. economy. Trade and financial linkages between the United States and other countries, although much larger now than in the past, are still too small to generate a “pure” export driven recovery in the United States. Despite the rapid growth in trade and financial flows, the volume of U.S. international trade and financial flows is still quite small relative to the size of its economy."

*Source: _cr03245 - 4. Volatility of GDP Growth in Different Periods*

### 7. Mexico: Volatility of Macroeconomic Aggregates

### 7. Mexico: Volatility of Macroeconomic Aggregates

### NAFTA’s effect on U.S. trade and GDP
- CBO (2003) estimates:
  - Boost to U.S. exports was only around 0.12 percent of GDP in 2001.
  - NAFTA raised imports by about 0.1 percent of GDP.
  - Potential long-term increase in the level of U.S. GDP due to NAFTA estimated in the range of 0.02 percent to 0.5 percent.
- ITC (1997) and Gould (1998) provide estimates broadly in line with the CBO finding on imports.

### NAFTA and the North American business cycle — comovement and volatility
- Evidence of increased synchronicity of business cycles among NAFTA members:
  - Marked increase in cross-country correlations of major macroeconomic aggregates, including output, consumption, and investment.
  - In most cases, the increases in correlations are statistically significant.
- Macroeconomic volatility in Mexico declined markedly after NAFTA:
  - Uniform and sizeable decline in the variance of several macroeconomic aggregates between the 1980-1993 and 1996-2002 periods.
  - Decreased volatility may reflect:
    - Vertical specialization in the NAFTA period.
    - Increased importance of more stable regional factors driving the Mexican business cycle.
    - Imported stability of domestic macroeconomic policies.

### Dynamic factor model results (Kose, Otrok, and Whiteman methodology)
- Macroeconomic fluctuations decomposed into:
  1. A regional factor common across all variables/countries.
  2. Country-specific factors common across main aggregates within a country.
  3. Factors specific to each variable.
- Key quantitative findings:
  - Proportion of output volatility explained by the regional factor rose from less than 1 percent in the period 1980-1993 to more than 19 percent in 1994-2002.
  - Variance of investment accounted for by the regional factor increased almost tenfold between 1980-1993 and 1994-2002.
  - The regional factor has increasingly explained volatility of manufacturing and industrial production over time.

### Dynamic stochastic multi-country business cycle model (simulation of trade-friction reduction)
- Model features:
  - Extends Backus, Kehoe, Kydland (1994) to three countries, trading frictions, and international financial autarky (see Kose and Yi (2003) for details).
  - In the model, imports from Mexico are used as intermediate inputs to produce final consumption and investment goods in the United States and Canada.
  - NAFTA impact simulated by changing the level of trading frictions (proxying tariffs and non-tariff barriers).
- Simulation outcomes:
  - Pre- and post-NAFTA simulations show a substantial increase in Mexican exports resulting from the lowering of tariffs after NAFTA.
  - Mexico’s output and investment respond much more strongly to temporary supply shocks in partner countries during the post-NAFTA period than in the pre-NAFTA period.
- Impulse response illustration:
  - Figure 8a: GDP response to a positive supply shock in U.S. and Canada — post-NAFTA responses exceed pre-NAFTA responses (percent deviation from the steady state shown across quarters).
  - Figure 8b: Investment response to a positive supply shock in U.S. and Canada — post-NAFTA responses exceed pre-NAFTA responses (percent deviation from the steady state shown across quarters).

### Conclusions and policy implications
- Quantification caveats:
  - Difficult to quantify NAFTA’s impact precisely given concurrent shocks (U.S. boom and 2000 stock market collapse/recession; Mexico’s tequila crisis and mid-1990s recession; peso devaluation; strength of the U.S. economy).
  - Mexico had joined GATT in 1986 and undertook trade liberalization and privatization/deregulation during 1988-1994, which also influenced outcomes.
- Evidence-based conclusions:
  - NAFTA had an important effect on growth and business cycle dynamics among its members.
  - Mexico benefited from a substantial increase in the volume of international trade and financial flows, as well as stronger growth.
  - Business cycles among NAFTA partners became considerably more synchronized.
  - Substantial increase in the degree to which Mexican output volatility was driven by regional versus domestic factors.
- Implications for FTAA:
  - The Free Trade Area of the Americas (FTAA) could have potentially significant effects on its developing country members.
  - Caution in generalizing from Mexico’s experience is warranted given Mexico’s specific circumstances (depreciated peso, strong U.S. economy, common border).
  - Analysis suggests that, in addition to boosting economic efficiency, increased foreign investment flows and deeper trade and financial linkages under an FTAA could help promote greater macroeconomic stability in the region.

*Source: _cr03245 - 7. Mexico: Volatility of Macroeconomic Aggregates*

### 8.      Nonetheless, regional agreements appear to provide helpful opportunities to

### _cr03245 - 8.      Nonetheless, regional agreements appear to provide helpful opportunities to 

### Regional agreements: role and key design characteristics
- Regional agreements can provide helpful opportunities to promote trade liberalization, especially when political and other factors impede unilateral or multilateral approaches.
- The key to ensuring favorable effects is to strive toward maintaining relatively low external barriers—i.e., “open regionalism”—to minimize trade diversion.
- Characteristics likely to maximize benefits and minimize trade diversion:
  - Regional diversity. Export diversity may be associated with greater complementarity of product ranges across countries, and greater trade with advanced countries may bring advantages to developing countries through increased investment flows and technology transfers. This suggests, for example, that the benefits of North-South arrangements exceed those of South-South arrangements.
  - Comprehensive coverage of products. FTAs are likely to bear greater fruit if they are extended beyond manufactured trade, and include agricultural products and services. Even more benefits can occur under comprehensive approaches that liberalize foreign direct investment, strengthen competition policy and improve regulatory frameworks.
  - Reform momentum. FTAs may help lock in broader reform agendas among participating countries (for example, encouraging reforms in investment protection and customs administration). Care is needed to ensure reforms are consistent and appropriate for countries’ stage of development.

### U.S. model for bilateral and regional trade arrangements
- The U.S. model meets many of these criteria. As part of the negotiation of the FTAA, timetables are to be established for removal of all trade restrictions on manufactured goods, agriculture, and services.
- Hemisphere-wide rules would be established for intellectual property rights, subsidies, antidumping, countervailing duties, government procurement, investment, competition policy, and dispute settlement.
- The challenge is to ensure these efforts do not undermine momentum for multilateral liberalization, which is still the first-best alternative.

### Simulations of Free Trade Arrangements with the United States — framework and caveats
- Simulations cover United States-Chile, United States-Central America (CAFTA), and United States-Australia.
- Framework: Global Trade Analysis Project (GTAP) comparative static general equilibrium model; assumed removal of all tariffs on goods as well as textile and clothing quotas between partners.
- Important shortcomings of the analytical framework (which may understate FTA benefits):
  - GTAP does not fully capture dynamic gains from trade liberalization (some studies suggest dynamic gains could be twice as large as static gains).
  - The model cannot consider effects of non-goods-related provisions of FTAs (services, investment, intellectual property), which could have larger effects.
  - The analysis does not take into account potential spillovers between FTAs or effects of multilateral liberalization.

### United States-Chile FTA — context and simulation summary
- Context:
  - Nearly 20 percent of Chilean exports are destined for the United States, and 20 percent of Chilean imports come from the United States.
  - Chile has a low and mostly uniform MFN tariff of 6 percent, and is an active participant in other regional and bilateral arrangements.
- Simulation highlights (effects focus solely on liberalizing goods trade):
  - Modest welfare gains for both Chile and the United States.
  - Chilean exports of processed foods, and to a smaller extent basic crops and textiles and clothing, receive a particular boost.
  - Modest welfare gain and small drop in Chile’s GDP reflect trade diversion as imports of U.S. machinery and equipment replace lower cost imports from the EU, Japan, and the rest of Asia.
  - Results similar to those prepared by the USITC (2003).
  - Chile’s strategy of establishing FTAs with other major trading partners helps reduce potential trade diversion.

### United States-CAFTA FTA — context and simulation summary
- Context:
  - Around 70 percent of Honduras’ exports are destined for the United States.
  - Many products already enter the United States under preferential arrangements, but barriers are relatively high in textile products and agriculture.
- Simulation highlights:
  - An FTA would have important welfare benefits for Central America.
  - GDP would increase by as much as 1.5 percent in Central America, with smaller gains for the United States.
  - Benefits arise mainly from boosts in sales of textiles and clothing and processed food, more than offsetting trade diversion from Japan, the rest of Asia, and Europe in machinery and equipment and textiles.
  - Trade creation effects: imports of basic manufactured imports from the United States would supplant higher cost CAFTA production, implying lower intra-CAFTA (duty-free) trade in these products.
  - Because of the region’s size and higher initial trade barriers, a U.S.-CAFTA agreement would have a greater impact on the rest of Latin America than a U.S.-Chile FTA.
  - Simulations do not account for the scheduled global liberalization of textile and clothing quotas (expiration of the Multifibre Agreement in 2005) or the planned FTAA; sustained benefits to CAFTA may be lower than estimated, highlighting the importance of using FTA preferences to spur efficiency enhancements.

### United States-Australia FTA — context and simulation summary
- Context:
  - About 10 percent of Australian exports are destined for the United States; APEC countries account for 72 percent of Australia’s exports.
  - Applied MFN tariffs currently average 4.3 percent, although tariffs on textile items are closer to 15 percent.
- Simulation highlights:
  - Reflecting already low existing tariff rates, a U.S.-Australia agreement would have relatively small overall welfare and output effects.
  - Staff simulations suggest Australia’s GDP would drop slightly owing to diversion of imports of machinery and equipment, basic manufactured products, and textiles from Japan, Asia, and the EU.
  - Impact on other countries would be small.
  - Australian producers of textiles and processed crops and animal (meat and dairy) products would reap significant gains.
  - Static model results are qualitatively similar to ACIL (2003), which show full bilateral liberalization would reduce Australian GDP by 0.09 percent by 2010.
  - The Center for International Economics (2001) estimated Australia’s GDP would rise by 0.33 percent by 2006 and 0.4 percent by 2010, assuming a 0.35 percent boost to services sector productivity as a result of a U.S.-Australia FTA.

### Simulation numeric results (GTAP estimates) — Table 3 (Percent change unless otherwise noted)
- United States-Chile
  - United States: 218.2 0.000
  - European Union: -96.3 0.000
  - Japan: -31.6 0.000
  - Canada: -24.3 0.000
  - Australia: -3.7 0.000
  - Brazil: -59.8 -0.004
  - Mexico: -23.1 0.000
  - Argentina: -15.5 -0.002
  - Chile: 4.1 -0.071
  - CAFTA: -13.5 -0.005
  - Rest of Latin America: -20.9 -0.003
  - Asia (excl. Japan): -63.3 -0.001
  - Mid-East/North Africa: -13.7 -0.001
  - Sub-Saharan Africa: -4.4 0.000
  - Rest of the World: -6.6 0.000
  - World: -154.4 0.000
- United States-CAFTA
  - United States: 964.1 0.007
  - European Union: -990.6 -0.004
  - Japan: -511.6 -0.002
  - Canada: -270.0 -0.003
  - Australia: -36.5 -0.001
  - Brazil: -123.4 -0.008
  - Mexico: -267.7 -0.006
  - Argentina: -32.0 -0.002
  - Chile: -26.8 -0.007
  - CAFTA: 3858.9 1.491
  - Rest of Latin America: -231.0 -0.028
  - Asia (excl. Japan): -1672.2 -0.029
  - Mid-East/North Africa: -180.4 -0.005
  - Sub-Saharan Africa: -27.6 0.000
  - Rest of the World: -232.9 -0.004
  - World: 220.2 -0.004
- United States-Australia
  - United States: 404.0 0.001
  - European Union: -155.2 0.000
  - Japan: -137.9 -0.001
  - Canada: -50.9 0.000
  - Australia: -1.2 -0.034
  - Brazil: -15.4 -0.001
  - Mexico: -15.1 0.001
  - Argentina: -4.2 0.000
  - Chile: -4.1 -0.001
  - CAFTA: -19.9 -0.008
  - Rest of Latin America: -11.3 -0.002
  - Asia (excl. Japan): -209.5 -0.002
  - Mid-East/North Africa: -22.0 -0.001
  - Sub-Saharan Africa: -3.4 -0.001
  - Rest of the World: -13.5 0.000
  - World: -259.6 0.000

### Conclusion and policy implications
- Estimates may understate gains from FTAs because they omit dynamic benefits and non-merchandise liberalization effects.
- Simulations suggest welfare gains for the United States are small but positive and that partner countries could suffer losses related to trade diversion.
- When initial trade barriers are low (as in Chile and Australia), gains from further liberalization in goods are limited.
- Nonmembers are adversely affected, including countries with prior FTAs with the United States (e.g., Mexico and Canada).
- Gains for countries could be eroded as more FTAs come into force.
- Policy implications:
  - FTAs are most beneficial if designed to minimize trade diversion and not to dilute momentum toward multilateral trade liberalization, spur competing regional trade blocs, or impose excessive administrative burdens.
  - Additional FTA elements—labor, the environment, intellectual property rights, and capital flows—should be designed to support broader liberalization.

*Source: Fund staff estimates from GTAP simulations.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2003/_cr03245.pdf_