## _cr06284

## Source details

**Canonical URL:** [_cr06284](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2006/_cr06284.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2006/_cr06284.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2006/_cr06284.pdf.json)

---

### Potential growth and total factor productivity (TFP) — key findings
- Average trend growth was 2.2 percent during 2000–04, with TFP growth averaging 1.3 percent.
- This is down from 3.4 percent for trend growth and average TFP growth of 1.5 percent during 1990–99.
- The decline in trend growth started around 1999.
- In the period 1977–2004, average annual growth is about 0.2 of a percentage point higher in the most recent set of national accounts (as used in this study).
- Developments explaining the slowdown:
  - Deceleration in trend growth rate of the utilization-adjusted capital stock (peaked in 1998 after continuous acceleration following the 1991 trough).
  - Slowdown in trend growth rate of the labor force; demographic forces halved the labor force growth rate between 1999 and 2004.
  - Decline in hours worked associated with proliferation of temporary and part-time arrangements.
  - Decline in TFP growth began around 1999 and is statistically associated with labor market institutions (minimum-to-median wage ratio tapering after 1998; union density tapering after 1998).
  - Increases in replacement rate of unemployment benefits had a positive, albeit small, impact on TFP growth but increased the Nairu; the increase in replacement rate also had a statistically significant effect on adjusted capital stock and on the Nairu.

### Model specification, measurement of factor services, and data
- Production function: Cobb-Douglas in flows of services; Y = A L^α Z^β with α + β = 1 under constant returns to scale.
- Labor services: L = N H (N = number of employees; H = hours worked per employee).
- Capital services: Z = K U (K = physical capital stock; U = utilization proxied by capacity utilization rate in the business sector).
- Rationale for utilization adjustment:
  - Capacity utilization proxies intensive use of capital and bringing additional machines on stream; adjusting avoids bias where unobserved input movements would be interpreted as TFP changes.
- Data: Annual data for 1979–2004 on real output, total labor hours, and rate of capacity utilization (DNB); capital stock series from the Bureau of Economic Policy Analysis; capacity utilization available only for business sector while capital stock covers entire economy.

### Statistical properties and estimation approach
- Stationarity/unit-root evidence:
  - Levels of output and labor hours: stationary at 95 percent when a constant and time trend are included, nonstationary at 97.5 percent.
  - Rates of change of output and labor hours: stationary when alternative includes a constant.
  - Unadjusted capital stock: clearly stationary.
  - Adjusted capital stock: borderline nonstationary in level even at 97.5 percent; its rate of change clearly stationary.
  - Fractional integration test confirmed stationarity of adjusted capital stock in levels.
- Cointegration tests: trace statistic accepts absence of cointegration at 95 percent; λmax suggests one cointegrating vector; conflicting evidence led to estimating production function in first differences.
- Econometric strategy:
  - Estimate production function and TFP growth jointly in first differences.
  - Model ∆A (TFP growth) as AR(p) and estimate state-space form by maximum likelihood using Kalman filter.
  - Diagnostics: residuals (level and square) are white noise (Kolmogorov-Smirnov statistic); AR(1) adequate for ∆A.

### Estimated elasticities, TFP dynamics, and comparisons
- With adjusted capital stock and imposing constant returns to scale:
  - Output elasticity of labor (α) = 0.63.
  - Output elasticity of capital (β) = 0.37.
- TFP growth dynamics:
  - Autoregressive coefficient = 0.98 (highly significant), indicating very persistent TFP growth.
  - Selected estimated TFP growth points:
    - 1984: 2.5 percent.
    - 1993: 1.4 percent.
    - 1994–1996: about 1.6 percent.
    - 2004: 1.3 percent.
- Comparison with unadjusted Solow residual:
  - Over 1984–2004, estimated TFP growth using adjusted capital was 20 basis points higher on average per annum than the unadjusted Solow residual.
  - Volatility: full-sample volatility of unadjusted Solow residual ≈ 1.8; volatility of estimated (adjusted) TFP growth ≈ a bit over 0.3.
- Correlations:
  - Estimated TFP growth is not significantly correlated with output growth, employment, or hours worked.
  - Unadjusted/adjusted Solow residuals show strong correlation with output growth; the estimated TFP growth does not—supporting importance of accounting for capital utilization.

### Trend growth calculation and decomposition (results)
- Trend extraction uses ideal band-pass filter on capital stock services, estimated TFP, and labor force participation; Nairu provided by DNB.
- Dutch trend growth outcomes:
  - Average trend growth for 1984–2004 = 3.4 percent per annum.
  - Subperiods:
    - 1984–99: Average trend growth ≈ 3¾ percent; TFP contributed in second half of 1980s and was stable through most 1990s.
    - 2000–04: Trend growth decelerated to about 2¼ percent; TFP growth decelerated after peaking in 1999; slowdown also coincided with slowdown in trend capital stock growth and labor force growth.
  - Labor and capital services developments explain about 80 percent of deceleration of trend growth; TFP growth explains remainder.

### Decomposition of recent TFP deceleration (2000–2004) — quantitative contributions
- Net contributions (average annual percent change, 2000-2004):
  - TFP growth: 1.32
  - Technological change: 1.52
  - Efficiency change: 0.02
    - Of which: labor: -0.17
    - Of which: capital: 0.19
  - Change in utilization of trend inputs: -0.22
    - Of which: labor: -0.01
    - Of which: capital: -0.21
- Interpretation:
  - Deceleration of TFP growth mostly attributed to deceleration of labor efficiency (labor contributed nearly 20 basis points to TFP deceleration).
  - Technological change made the largest absolute contribution to TFP growth but was unchanged between 1990–99 and 2000–04 and thus did not explain the deceleration.

### Labor market institutions — empirical links to TFP
- Exploratory regression (dependent variable: TFP growth; sample 1984-2001) produced statistically significant coefficients (t-statistics in parentheses):
  - Constant: 1.443 (1.55 *)
  - Ratio of minimum to median wage: coefficient -0.25 (-4.47 *)
  - Union density (percentage): coefficient -0.22 (-3.50 *)
  - Gross replacement rates average: coefficient 0.08 (2.72 *)
  - Tax wedge, married, as percentage gross labor costs: coefficient 0.03 (1.05)
  - Payments, married, as percentage of gross wage earnings: coefficient -0.01 (-0.32)
  - Average effective tax wedge (percentage): coefficient 0.03 (1.58)
  - R2: 0.84
  - K-S 1/: 0.18
- Interpretations:
  - Higher minimum-to-median wage ratio and higher social benefits appear to affect TFP growth negatively (channels: reduced labor demand, poverty traps, hampered capital deepening).
  - Higher union density has negative effect (channels: reduced wage differentiation, reduced labor mobility, dampened human capital investment).
  - Gross replacement rates show positive coefficient—may reflect capital deepening/substitution of capital for labor; supporting regressions show replacement rate positively impacts adjusted capital stock and negatively impacts labor services.

### Market structure and sectoral patterns
- TFP growth depends on creative destruction and benefits from product market liberalization and increased competition.
- Sectors with relatively low recent TFP growth: nonmarket services, construction, and network industries (gas and electricity) with high state control.
- Simultaneous labor and product market reforms can yield sizable gains in employment and output and mitigate transition costs.

### Projections and scenarios if no further reforms
- Central scenario if no new reforms:
  - Annual trend growth likely to recover from low level as investment cycle rekindles but will likely be around 2 percent.
  - Aligns with projected 0.15 percent per annum fall in the labor force, an expected moderate increase in participation rate, rekindling of investment cycle, and TFP growth rates remaining at 1.3–1.4 percent per annum.
- Conclusion: without reforms to boost TFP, trend growth of around 2 percent is prudent for remaining years of the decade.

### Policy implications for raising productivity and trend growth
- Emphasize enhancing flexibility of labor and product markets.
- Simultaneous labor and product market reforms recommended to reduce markups, raise competition, and mitigate transition costs.
- Enhance knowledge and training levels of labor force and pursue further market liberalization.
- Reduce barriers to entrepreneurship and innovation and lower costs of doing business, especially in sheltered sectors and network industries.

---

### Developments in house prices and mortgage market — key facts
- House price growth decelerated sharply in Q4 2000 and remained subdued until recently; in 2005 prices accelerated again, with quarter-over-quarter increase in Q3 registering "3.1 percent".
- Late 1990s boom: housing price and asset increases added an average of "1–1½ percent" to real GDP growth between 1998 and 2000 (DNB estimate).
- Mortgages guaranteed by National Mortgage Guarantee (NHG) reached "€60 billion" in 2004.
- Proportion of mortgages subject to interest rate adjustment within two years increased from "21 percent in 2003" to "31 percent in 2005".
- Percentage of newly provided loans with LTV exceeding "100 percent" has been increasing.
- Ratio of mortgage debt-to-GDP is high at "more than 100 percent" and increasing.
- 2004 mortgage figures:
  - Dutch mortgage debt: "€518,115" (millions of euros).
  - Mortgage debt growth over 2003: "14.3" (percent).
  - Residential mortgage debt-to-GDP: "111.1" (percent).
  - Per capita mortgage debt: "€31,868".
- Cross-country comparisons (selected):
  - EU-15 mortgage debt-to-GDP: "46.4" (percent).
  - Denmark mortgage debt-to-GDP: "89.7" (percent); per capita: "€32,292".
  - United Kingdom mortgage debt-to-GDP: "72.5" (percent); per capita: "€20,835".
  - United States mortgage debt-to-GDP: "65" (percent).
- Netherlands population density: "478 people per square kilometer" — contributing to low housing supply elasticity.
- Cumulative real house price increase from trough to peak: "103 percent".

### Sensitivity of house prices and consumption to fundamentals
- Hofman (2005): "1 percent" increase in real interest rate reduces long-run equilibrium house price by about "10 percent".
- IMF estimate (Nadal de Simone, 2005): long-run elasticity of consumption w.r.t. per capita real net wealth = "0.21".
- CPB quarterly model SAFE (1995–2002): "40 percent" of consumption growth could be attributed to increase of households’ stock of dwellings.
- IMF staff regression (1970–2005): long-run coefficients — disposable income per household: "1.0222"; nominal interest rate: "-0.0451"; model fit R2 "0.97"; Max Gap "0.13".

### Risks and structural concerns
- High household indebtedness and new mortgage products allow heavy leverage and reduced principal repayment.
- NHG guarantee reduces banks’ incentives to scrutinize loans (mortgages essentially risk free for lenders).
- Rising interest rates could prompt downward house price adjustments affecting consumption and banking sector.
- Studies vary on overpricing as of Q2 2004: Hofman (2005) found prices consistent with fundamentals; others found overvaluation or scope for downward movement.

### Policy issues — tax deductibility and NHG
- Financial System Stability Assessment (2004) recommended phasing out tax deductibility of mortgage interest over medium term, preferably gradually.
- Reasons to eliminate mortgage interest tax deductibility:
  - Distorts housing markets by making mortgage debt attractive and reducing incentive to pay principal.
  - Favors wealthy households (greater tax advantage at higher tax brackets and larger assets).
  - Removing deductibility would help the budget (rising costs related to population aging) and could fund policies to make housing affordable for low-income households.
- NHG implicit cost is covered by government; NHG provided at no cost creates price distortion.

---

### Policy options for phasing out mortgage interest deductibility — three options analyzed
- Policy 1: Immediate and complete removal of tax deductibility of mortgage interest.
- Policy 2: Introduction of a nominal limit on mortgage loan size eligible for deduction; nominal limit set at €141,000 (average mortgage loan in 2005).
- Policy 3: Reduction of the deductible rate so mortgage interest can be deducted only at a rate of 30%.

### Household categories and illustrative household parameters (from Ministry of finance, March 20, 2006)
- Low-Income Household:
  - Average house price: 200,000
  - Average income of mortgage borrower: 27,000
  - Average loan size of borrower: 127,000
  - Mortgage interest rate (in percent): 5.3
  - Household tax category: 30%
  - Calculated deduction from income: 6,780
  - Tax saving (additional disposable income): 2,034
  - Leverage (Loan/income): 4.70
  - Loan/house price: 0.64
- Average-Income Household:
  - Average house price: 225,200
  - Average income of mortgage borrower: 49,000
  - Average loan size of borrower: 145,000
  - Mortgage interest rate (in percent): 5.3
  - Household tax category: 40%
  - Calculated deduction from income: 7,750
  - Tax saving (additional disposable income): 3,100
  - Leverage (Loan/income): 2.96
  - Loan/house price: 0.64
- High-Income Household:
  - Average house price: 325,000
  - Average income of mortgage borrower: 94,000
  - Average loan size of borrower: 209,000
  - Mortgage interest rate (in percent): 5.3
  - Household tax category: 50%
  - Calculated deduction from income: 11,170
  - Tax saving (additional disposable income): 5,585
  - Leverage (Loan/income): 2.22
  - Loan/house price: 0.64
- Note: Age: all households <65; income categories: low <€30,000; average €30-60,000; high >€60,000.

### Illustrative quantitative impacts (selected results)
- Under modest recession scenario, one year after policy change, price drop for average-income households:
  - Policy 1: -12.4
  - Policy 2: -3.0
  - Policy 3: -5.2
- Six years after policy change, no-policy-change scenario (real house price increases):
  - Normal economic growth: 29.1 percent
  - Exuberant economic growth: 48.8 percent
- Six years after policy change under Policy 2 (normal growth), real house price increases by household income:
  - Low-income households: 26.6 percent
  - Average-income households: 22.2 percent
  - High-income households: 14.4 percent
- Economic scenario definitions used:
  - Modest recession: Disposable income growth per household per year 0.00%; Change in interest rate, per year -0.10
  - Normal growth: Disposable income growth per household per year 3.50%; Change in interest rate, per year 0.15
  - Exuberant growth: Disposable income growth per household per year 5.00%; Change in interest rate, per year 0.20
- Illustrative impact under Policy 1 if removal coincides with modest recession:
  - Real house prices for high-income households decline by 20 percent in first year; reach 36 percent decline by end of fifth year if recession persists.

### Distributional and cyclical properties — comparison of Policy 2 and Policy 3
- Policy 3 (deductible rate reduced to 30%):
  - Progressive: affects higher tax-bracket households immediately; low-income households not affected.
  - Procyclical: foregone tax benefits larger as interest rates rise, compounding downward pressure on prices.
- Policy 2 (nominal limit at €141,000):
  - Progressive: affects high-income/high-loan households immediately.
  - Countercyclical: becomes binding gradually as prices rise; less binding when prices fall, providing cushion.
  - Considered attractive: progressive, countercyclical features, gradual impact on low- and medium-income households.

### Policy recommendations on mortgage-related risks
- Educate public about risks associated with different mortgage loans (part of code of conduct for mortgage lenders).
- Address price distortion from NHG guarantee being provided at no cost.
- Phase-out tax deductibility of mortgage interest payments to reduce incentives to leverage; direct additional revenue to well-targeted programs for low-income families and/or tax cuts to lessen downward impact on housing demand and stimulate labor participation.
- Prefer gradual phase-out to minimize disruptive effects; timing matters to avoid procyclicality.

---

### Tax revenue volatility — measurement and Netherlands evidence
- Primary volatility measure: coefficient of variation (CV) of tax revenues-to-GDP.
- Netherlands quarterly CVs (selected series):
  - Direct taxes and social contributions CV: 1975–84: 2.5; 1985–94: 3.0; 1995–2004: 4.4.
  - Corporate income tax CV: 1975–84: 8.2; 1985–94: 5.5; 1995–2004: 15.9.
  - Personal income tax and social contributions CV: 1975–84: 2.8; 1985–94: 3.1; 1995–2004: 4.4.
  - Indirect taxes CV: 1975–84: 4.4; 1985–94: 3.3; 1995–2004: 6.2.
  - Memo: Wage bill CV: 1975–84: 3.8; 1985–94: 1.7; 1995–2004: 1.2.
- Netherlands annual CVs (selected):
  - Direct taxes and social contributions CV: 1975–84: 2.2; 1985–94: 2.9; 1995–2004: 4.5.
  - Corporate income tax CV: 1975–84: 8.0; 1985–94: 5.3; 1995–2004: 16.3.
  - Personal income tax and social contributions CV: 1975–84: 2.5; 1985–94: 3.0; 1995–2004: 4.4.
  - Indirect taxes CV: 1975–84: 4.2; 1985–94: 3.2; 1995–2004: 6.4.
  - Memo: Wage bill CV: 1975–84: 3.6; 1985–94: 1.5; 1995–2004: 1.2.
- Key Netherlands findings:
  - Highest recent volatility (1995–2004, quarterly) in corporate income tax = 15.9; indirect taxes = 6.2; personal income tax and social contributions = 4.4.
  - Corporate income tax volatility more than doubled from 1975–84 to 1995–2004.
  - Personal income tax moved from low volatility in 1975–84 to near top in 1995–2004 (exceeded only by Ireland).

### Determinants and quantified role of business cycle and pensions
- Business cycle volatility (average absolute deviation of output gap) for the Netherlands: 1975–84: 1.8; 1985–94: 1.4; 1995–2005: 1.9; 2000–2005: 2.4.
- Short-term elasticities (OECD estimates):
  - Personal income taxes and social contributions elasticity to output gap: 1.25 percent (weighted avg).
  - Corporate income tax short-term elasticity: 1.5 percent.
- Quantified implications (Table 5 averages):
  - Output gap average absolute deviation: 1975–84: 1.8; 1985–94: 1.4; 1995–2005: 1.9; 2000–2005: 2.4.
  - Tax ratio — PIT and social contributions: 1975–84: 0.5; 1985–94: 0.4; 1995–2005: 0.5; 2000–2005: 0.6.
  - Tax ratio — CIT: 1975–84: 0.9; 1985–94: 0.7; 1995–2005: 1.0; 2000–2005: 1.2.
- Role of pension funds:
  - Dutch private pension system predominantly "defined benefit"; income tax deduction of pension contributions allowed.
  - Share of equities in pension funds rose from 10 percent in 1990 to more than 40 percent in 2005.
  - Correlations (Table 6) show strong negative relationships between tax ratios and pension premia, e.g. 1995–03: Direct taxes and social contributions -79.7; Personal income tax and social contributions -45.5; Total taxes -80.7.
  - ARCH/GARCH results: higher volatility in pension premia contributes to increased personal income tax volatility.

### Formal econometric ARCH findings on PIT volatility
- Semi-elasticities:
  - Pension premium volatility: a one percent increase in premium volatility leads to about 6 to 8 percent increase in personal income tax revenue volatility.
  - Output gap volatility: effect varies between 5 percent and 8 percent depending on specification.
- ARCH-specification results (selected table highlights preserve exact coefficients and statistics):
  - Table 2 (Equation (3) mean; Equation (5) variance):
    - Conditional mean: α1 = -1.30 [-66.93]; yt-1 α2 = 0.93 [23.48]
    - Conditional variance: β1 = 0.001 [7.58]; β2 = 0.50 [3.24]; β3 (PtV) = 0.07 [6.45]
    - Adj. R2 = 0.91; ARCH-LM = 4.58
  - Table 3 (Equation (4) mean; Equation (5) variance):
    - Conditional mean: α1 = -0.78 [-21.71]; wt α2 = 1.02 [14.28]
    - Conditional variance: β1 = 0.001 [7.93]; β2 = 0.95 [15.69]; β3 (PtV) = 0.06 [12.04]
    - Adj. R2 = 0.16; ARCH-LM = 4.35
  - Table 6 (Equation (3) mean; Equation (7) variance):
    - Conditional mean: α1 = -1.01 [-0.43]; yt-1 α2 = 0.99 [64.38]
    - Conditional variance: β1 = 0.01 [7.74]; β2 = 0.39 [4.19]; β3 (PtV) = 0.07 [3.82]; β4 (GtV) = 0.05 [7.90]
    - Adj. R2 = 0.97; ARCH-LM = 4.62
- Conclusion from ARCH evidence: pension premiums and business cycle volatility significantly explain increased PIT revenue volatility.

### Business cycle, fiscal implications, and recommended fiscal prudence
- Formal estimation: a one percent increase in output gap volatility leads to an increase in volatility of tax revenue that varies between 0.05 and 0.08 percent depending on specification.
- OECD estimate: overall budget elasticity ≈ 0.6 percent for the Netherlands.
- Historical output gap varied within ±4 percent of GDP over past 30 years and changed year-on-year by 1¼ percent on average; implication: fiscal balance can change by ¾ percentage points of GDP year-to-year and vary within ±2½ percent of GDP due to business cycle fluctuations.
- Policy implication: given higher tax revenue volatility, consideration could be given to a more conservative fiscal deficit signal value than the existing 2½ percent of GDP signal in the Dutch fiscal framework.

---

*Source: IMF staff analysis in _cr06284 (selected excerpts).*

### References..............................................................................................................

### References

### Key findings on potential growth and TFP in the Netherlands
- Average trend growth was 2.2 percent during 2000–04, with TFP growth averaging 1.3 percent.
- This is down from 3.4 percent for trend growth and average TFP growth of 1.5 percent during 1990–99.
- The decline in trend growth started around 1999.
- In the period 1977–2004, average annual growth is about 0.2 of a percentage point higher in the most recent set of national accounts (as used in this study).

### Determinants and dynamics of the slowdown
- The decline in trend growth was accompanied by:
  - A deceleration in the trend growth rate of the utilization-adjusted capital stock, which reached a peak in 1998 after continuous acceleration following the 1991 trough of the European recession.
  - A slowdown in the trend growth rate of the labor force; demographic forces halved the labor force growth rate between 1999 and 2004.
  - A contribution from a decline in hours worked associated with the proliferation of temporary and part-time arrangements.
- The decline in TFP growth started around 1999 and was statistically associated with institutional features of the labor market:
  - A tapering after 1998 in the secular fall in the ratio of the minimum wage to the median wage.
  - A tapering after 1998 in union density.
  - Increases in the replacement rate of unemployment benefits appear to have had a positive, albeit small, impact on TFP growth, but at the cost of increasing the Nairu.
- The paper finds a statistically significant effect of the increase in the replacement rate on the adjusted capital stock and on the Nairu.

### Methodology and measurement considerations
- The study estimates a Cobb-Douglas production function treating TFP growth as a latent variable.
- The capital stock is adjusted for its utilization rate using the capacity utilization rate in the business sector.
- Adjusting capital for utilization helps avoid biases in estimated labor shares and the spurious procyclicality of TFP that can arise when intensity of capital use is omitted.
- The estimation approach yields results consistent with:
  - Factor income shares computed from national accounts.
  - A constant returns to scale technology, supporting the use of the Cobb-Douglas production function.
  - A “statistically clean” TFP series that is not correlated with strongly cyclical variables (e.g., employment).

### Policy implications and considerations
- Trend growth estimates matter for fiscal policy design because real expenditure ceilings under the Dutch fiscal framework are set with regard to the trend growth rate.
- If trend growth is overestimated, the government may commit to a more expansionary policy stance than intended, complicating efforts to ensure structural fiscal consolidation during an upturn.
- Policies to raise productivity should emphasize enhancing the flexibility of both labor and product markets.
- Simultaneous labor and product market reforms have advantages: sizable gains in employment and output can result from reducing markups in both markets, and raising competition in goods markets can mitigate transition costs of labor and service market reforms.
- Measures to enhance the knowledge and training level of the labor force and further market liberalization are identified as necessary to address the fall in productivity growth.

### Structure of the broader document (selected chapter organization)
- Chapter I: POTENTIAL GROWTH AND TOTAL FACTOR PRODUCTIVITY IN THE NETHERLANDS
  - A. Overview and Introduction
  - B. Production Function Specification and Data Properties
  - C. (Estimation and results described in full chapter)
  - D. (Trend growth calculation)
  - E. (Determinants of deceleration explored)
  - F. Conclusions and policy implications

*Source: _cr06284 - References..............................................................................................................*

### 9.      This study estimates a production function specification that builds on the

### _cr06284 - 9.      This study estimates a production function specification that builds on the

### Model specification and measurement of factor services
- The study builds on Bils and Cho’s (1994) real business cycle model with changes in effective factor utilization and constant work effort.
- Labor services: L = N H, where N = number of employees and H = hours worked per employee; hours worked used to capture changes in number of workers, shifts, and duration of the workweek.
- Capital services: Z = K U, where K = physical capital stock and U = utilization, proxied by the rate of capacity utilization with hiring (business sector).
- Rationale for utilization adjustment:
  - Capacity utilization is highly correlated with the cycle and proxies both intensive use of existing capital (e.g., added shifts) and the bringing-on-stream of additional machines.
  - If the net physical capital stock understates excess capacity, input shares will be unrealistically high and TFP growth unrealistically low; unobserved input movements will be interpreted as TFP changes.
- Data:
  - Annual data for 1979–2004 on real output, total labor hours, and the rate of capacity utilization from the Netherlandsche Bank (DNB).
  - Capital stock series provided by the Bureau of Economic Policy Analysis.
  - Capacity utilization series available only for the business sector while capital stock covers the entire economy.

### Statistical properties and unit-root/cointegration testing
- Stationarity tests (augmented Dickey-Fuller with trend when appropriate):
  - Levels of output and labor hours: stationary at the 95 percent confidence level when a constant and time trend are included in the alternative hypothesis, nonstationary at the 97.5 percent level.
  - Rates of change of output and labor hours: stationary when the alternative includes a constant.
  - Unadjusted capital stock: clearly stationary.
  - Adjusted capital stock: borderline nonstationary in level even at the 97.5 percent confidence level; its rate of change clearly stationary when the alternative contains a constant.
- Alternative tests:
  - A unit root test with null of stationarity confirmed that levels of output, labor hours, and unadjusted capital stock are nonstationary, while adjusted capital stock is stationary.
  - A fractional integration test (Shimotsu and Phillips, 2005) confirmed stationarity of the adjusted capital stock series in levels.
- Cointegration tests (Cheung and Lai, 1993 correction for small sample bias):
  - Trace statistic: accepts absence of cointegration at the 95 percent level.
  - λmax statistic: suggests one cointegrating vector.
  - Given conflicting cointegration evidence and the requirement of two cointegrating vectors to estimate the model in levels (production relation and stationarity of adjusted capital stock), the production function was estimated in first differences.

### Econometric estimation approach
- Production function: Cobb-Douglas in flow of services:
  - Y = A L^α Z^β (formal expression presented).
  - In competitive equilibrium, α and β are output elasticities of labor and capital; α + β = 1 implies constant returns to scale and factor income shares.
- Estimation strategy:
  - Production function and TFP growth estimated jointly in first differences.
  - TFP growth modeled as a latent variable and specified as an autoregressive process:
    - ∆Y = ∆A + α ∆L + β ∆Z  (Equation (2) in first differences).
    - ∆A modeled as AR(p): ∆A_t = Σ_{i=1}^p ρ_i ∆A_{t−i} + ε_t (formal AR specification).
  - State-space form estimated by maximum likelihood using Kalman filter (prediction error decomposition).
- Model diagnostics and robustness:
  - Residuals (level and square) are white noise (Kolmogorov-Smirnov statistic).
  - TFP growth adequately modeled as AR(1); AR(2) and ARMA(1,1) variants yielded insignificant extra coefficients and did not affect factor share estimates or standard errors.

### Estimated elasticities and dynamics of TFP growth
- With adjusted capital stock and imposing constant returns to scale:
  - Output elasticity of labor (α) = 0.63.
  - Output elasticity of capital (β) = 0.37.
- TFP growth dynamics:
  - Autoregressive coefficient = 0.98 (highly significant), indicating very persistent TFP growth.
  - Estimated path of TFP growth (selected points):
    - 1984: 2.5 percent.
    - 1993: 1.4 percent.
    - 1994–1996: about 1.6 percent (for the following three years).
    - 2004: 1.3 percent.
  - A constant in the AR(1) was statistically insignificant and dropped.
- Comparison with unadjusted Solow residual:
  - Over 1984–2004, Dutch estimated TFP growth using adjusted capital was 20 basis points higher on average per annum than the unadjusted Solow residual.
  - Volatility:
    - Full-sample volatility of the unadjusted Solow residual ≈ 1.8.
    - Volatility of estimated (adjusted) TFP growth ≈ a bit over 0.3.
- Correlations and economic interpretation:
  - Estimated TFP growth is not significantly correlated with output growth, employment, or hours worked.
  - Unadjusted or adjusted Solow residuals show strong correlation with output growth; the estimated TFP growth does not.
  - These results support the importance of accounting for variations in capital utilization and imply that the cyclicality of Solow residuals can be explained without invoking increasing returns to scale.
- Additional robustness:
  - Tests of TFP growth exogeneity (Hall, 1989) using changes in real government military spending and oil prices found TFP growth uncorrelated with these variables; some weak correlation exists between the unadjusted Solow residual and military spending.

### Trend growth calculation and decomposition
- Trend output growth formula used (all variables in annual growth rates):
  - Ȳ_t = Nairu_t + P̄_t + K̄_t + Ā_t multiplied and adjusted as in production-function consistent expression (formal expression provided in text).
- Trend extraction:
  - Ideal band-pass filter (Ouliaris, 2001) applied to series on capital stock services, estimated TFP, and labor force participation rate.
  - Nairu provided by the DNB; potential employment contribution defined consistent with Nairu.
- Dutch trend growth outcomes:
  - Average trend growth for 1984–2004 = 3.4 percent per annum.
  - Two distinct subperiods:
    - 1984–99:
      - Average trend growth ≈ 3¾ percent.
      - High trend growth coincided with increases in labor force participation and employment growth; trends in hours worked similar.
      - TFP growth contributed during the second half of the 1980s and was quite stable during most of the 1990s.
    - 2000–04:
      - Trend growth decelerated to about 2¼ percent.
      - TFP growth decelerated after peaking in 1999 and contributed somewhat to the slowdown.
      - Deceleration also coincided with slowdown in trend capital stock growth and subsequently in labor force growth.
      - Trend growth in utilization-adjusted capital stock peaked in 1998 after prior capital deepening.
      - Labor services growth declined after 1999 due to deceleration of trend labor force growth and shifts toward temporary and part-time arrangements.
      - Overall, developments in labor and capital services explain about 80 percent of the deceleration of trend growth; TFP growth explains the remainder.

*Italic: Source: _cr06284 - 9.      This study estimates a production function specification that builds on the*

### 22.      The recent deceleration of TFP growth can mostly be attributed to a

### _cr06284 - 22.      The recent deceleration of TFP growth can mostly be attributed to a

### Deceleration of TFP: decomposition and proximate cause
- The recent deceleration of TFP growth can mostly be attributed to a deceleration of labor efficiency.
- Decomposition approach:
  - TFP growth split into: (i) technological change; (ii) changes in the efficiency with which factors are used (deviations of labor and capital from their trend growth rates); and (iii) changes in the utilization of trend inputs.
  - Technological change is the residual after accounting for changes in trend inputs utilization.
- Net contributions (Table 5, average annual percent change, 2000-2004):
  - TFP growth: 1.32
  - Technological change: 1.52
  - Efficiency change: 0.02
    - Of which: labor: -0.17
    - Of which: capital: 0.19
  - Change in utilization of trend inputs: -0.22
    - Of which: labor: -0.01
    - Of which: capital: -0.21
- The deceleration in labor services trend growth contributed nearly 20 basis points to the deceleration of TFP growth, mostly due to a decline in the efficiency in its use.
- Technological change made the largest contribution to TFP growth but was unchanged between 1990–99 and 2000–04, so it did not explain the deceleration.

### Quantitative context for trend growth and TFP
- Average annual growth and related series (Table 4):
  - 1984-2004: Average Annual Growth 1.38; Unadjusted Solow Residual 1.59; TFP Growth 3.36
  - 1984-1999: Average Annual Growth 1.77; Unadjusted Solow Residual 1.67; TFP Growth 3.73
  - 2000-2004: Average Annual Growth 0.15; Unadjusted Solow Residual 1.32; TFP Growth 2.17
  - Standard deviation (1984-2004): 1.84; Unadjusted Solow Residual 0.32; TFP 1.31

### Labor market institutions: empirical findings linking institutions to TFP
- Exploratory econometric exercise using Dutch labor market institutional indicators (Nickel and Nunziatta (2001), extended by Giuseppe Nicoletti) yielded statistically significant relationships (Table 6; dependent variable: TFP growth; sample 1984-2001):
  - Constant: 1.443 (t-statistic 1.55 *)
  - Ratio of minimum to median wage: coefficient -0.25 (t-statistic -4.47 *)
  - Union density (percentage): coefficient -0.22 (t-statistic -3.50 *)
  - Gross replacement rates average: coefficient 0.08 (t-statistic 2.72 *)
  - Tax wedge, married, as percentage gross labor costs: coefficient 0.03 (t-statistic 1.05)
  - Payments, married, as percentage of gross wage earnings: coefficient -0.01 (t-statistic -0.32)
  - Average effective tax wedge (percentage): coefficient 0.03 (t-statistic 1.58)
  - R2: 0.84
  - K-S 1/: 0.18
- Interpretation of significant coefficients:
  - Higher ratio of minimum to median wage and higher social benefits appear to have affected TFP growth negatively (statistically significant).
    - Suggested channels: reduced labor demand, poverty traps, hampered capital deepening, pulling the economy below the production possibility frontier.
  - Higher union density (proxy for coverage of collective agreements) has a statistically significant negative effect on TFP growth.
    - Suggested channels: reduced wage differentiation, reduced labor mobility, dampened human capital investment, increased costs of firm entry and exit.
  - Gross replacement rates showed a statistically significant positive impact on TFP growth in the regression, but this may reflect substitution of capital for labor (capital deepening with labor-saving technology).
    - Supporting evidence cited: regressions show replacement rate positively and significantly impacts adjusted capital stock and negatively and significantly impacts labor services.

### Market structure, creative destruction, and sectoral patterns
- TFP growth depends also on "creative destruction" and benefits from further liberalized goods and services markets.
- Sectors with relatively low recent TFP growth in the Netherlands:
  - Nonmarket services and construction (sheltered sectors).
  - Network industries such as gas and electricity (areas with high state control).
- Evidence (cited): product market deregulation increases competition and reduces barriers to entry and exit, maximizing positive output and employment effects of labor market reforms that result in "wage moderation" (Estevão, 2005).

### Projections, scenarios, and implications if no further reforms
- Without further reforms, TFP growth and thus Dutch trend growth are likely to be significantly below their long-run averages.
- Demographic headwinds: aging of the population expected to weigh negatively on TFP growth and trend growth.
- Study’s central scenario if no new reforms:
  - Annual trend growth likely to recover from recent low level as the investment cycle rekindles but will likely be around 2 percent.
  - This aligns with:
    - Projected 0.15 percent per annum fall in the labor force,
    - An expected moderate increase in the participation rate,
    - Rekindling of the investment cycle as offsetting factors,
    - TFP growth rates remaining at 1.3–1.4 percent per annum.
- Conclusion: in the absence of further reforms that boost TFP growth, trend growth of around 2 percent would seem a prudent estimate for the remaining years of this decade.

### Policy recommendations and reform priorities
- To raise TFP and trend growth the Netherlands should:
  - Enhance labor market flexibility to facilitate labor reallocation and scaling down of relatively inefficient firms.
  - Reduce barriers to entrepreneurship and innovation.
  - Continue opening up goods and services markets (product market liberalization) to increase competition, lower entry and exit costs, and foster creative destruction.
  - Reduce the cost of doing business and increase competition, especially in sheltered sectors and network industries.
- Rationale: Past bold structural reforms in the 1980s and early 1990s strongly improved Dutch growth performance; similar ambition is needed to avoid the trend growth rate being stuck at historically low levels and further weakened by population aging.

### Concluding synthesis
- Drivers of recent slowdown in trend GDP growth:
  - Deceleration of labor force trend growth.
  - Moderation of capital deepening.
  - Deceleration of TFP growth (labor efficiency deterioration was a key proximate factor).
  - Net decline in hours worked due to proliferation of temporary and part-time work arrangements.
- Estimated TFP growth in this analysis is free of cyclical components and driven by underlying structural factors; factor elasticities align closely with national accounts factor shares.
- Policy emphasis: boosting TFP growth—through labor market flexibility, entrepreneurship and innovation support, and product market opening—is the primary avenue for raising trend growth given demographic constraints.

*Source: IMF staff analysis in _cr06284 - 22.      The recent deceleration of TFP growth can mostly be attributed to a (chapter/section).*

### 1.      Developments in house prices are of considerable interest in the Netherlands.

### Developments in house prices are of considerable interest in the Netherlands

### Recent developments in the Dutch housing market
- House price growth decelerated sharply in Q4 2000 and remained subdued until recently; in 2005 prices accelerated again, with the quarter-over-quarter increase in the third quarter registering "3.1 percent".
- During the late 1990s boom, housing price and asset increases added an average of "1–1½ percent" to real GDP growth between 1998 and 2000 (DNB estimate).
- Reports indicate deteriorating credit quality associated with continued mortgage expansion.
- Mortgages guaranteed by the National Mortgage Guarantee reached "€60 billion" in 2004.
- The proportion of mortgages subject to an interest rate adjustment within two years increased from "21 percent in 2003" to "31 percent in 2005".
- The percentage of newly provided loans with loan-to-value ratios (LTV) exceeding "100 percent" has been increasing.
- The ratio of mortgage debt-to-GDP is high at "more than 100 percent" and has been increasing.

### Household leverage and mortgage debt
- The Netherlands has among the highest mortgage debt levels in industrialized countries.
- In 2004:
  - Dutch mortgage debt was "€518,115" (millions of euros).
  - Mortgage debt growth over 2003 in the Netherlands was "14.3" (percent).
  - Residential mortgage debt-to-GDP in the Netherlands was "111.1" (percent).
  - Per capita mortgage debt in the Netherlands was "€31,868".
- By comparison (selected figures from Table 2):
  - EU-15 mortgage debt-to-GDP: "46.4" (percent).
  - Denmark mortgage debt-to-GDP: "89.7" (percent); per capita: "€32,292".
  - United Kingdom mortgage debt-to-GDP: "72.5" (percent); per capita: "€20,835".
  - United States mortgage debt-to-GDP: "65" (percent).
- In 2004, mortgage debt in the Netherlands increased by "14 percent", compared with increases of "11" and "6 percent" in the United Kingdom and Denmark, respectively.
- The Netherlands is densely populated with "478 people per square kilometer", contributing to low housing supply elasticity due to limited undeveloped land and strict zoning.

### Cross-country comparisons and historical context
- Cumulative real house price increase from trough to peak in the Netherlands was "103 percent" (surpassing neighbors).
- Table 1 (accumulated inflation-adjusted house price increases as of end-2004) highlights:
  - Netherlands: Peak-to-peak "52", Trough-to-peak "103", Duration of increase "19", Price boom last peak "1978", last trough "1985".
  - Belgium cumulative increases and durations reported similarly (table retained in source).
- Housing price cycles historically tend to last "3–4 years" with cumulative price increases averaging around "30 percent" during boom phases; the current expansion in several industrial countries has surpassed this record by far.
- Netherlands experienced a boom in the 1970s followed by a complete reversal in the early 1980s; current increases are exceptional both cross-country and historically for the Netherlands.

### Sensitivity of house prices and consumption to fundamentals
- Multiple studies find housing prices highly sensitive to interest rate movements:
  - Hofman (2005): a "1 percent" increase in the real interest rate reduces the long-run equilibrium house price by about "10 percent".
  - IMF estimate (Nadal de Simone, 2005): long-run elasticity of consumption with respect to per capita real net wealth is "0.21".
  - CPB quarterly model SAFE (1995–2002): "40 percent" of consumption growth could be attributed to the increase of households’ stock of dwellings.
- Price developments are asymmetric; downward adjustments are slower (Verbruggen and others (2005)), implying prolonged adjustment could keep consumption lower for longer.
- IMF staff regression results (1970–2005) indicate strong explanatory power and statistical significance:
  - Long-run coefficients:
    - Disposable income per household: "1.0222" (coefficient), significant at 5 percent.
    - Nominal interest rate: "-0.0451" (coefficient), significant at 5 percent.
  - Short-run coefficients and lags also significant (see Table 3 in source).
  - Model fit: R2 "0.97"; Max Gap "0.13"; Test on Residuals critical value and result reported in Table 3.

### Risks and structural concerns
- Current indebtedness levels in industrial countries, particularly the Netherlands, are unprecedented.
- New mortgage products have allowed households to:
  - Leverage heavily,
  - Not pay back principal,
  - Take on interest rate and other risks associated with such products.
- Mortgages guaranteed by the National Mortgage Guarantee reduce banks’ incentives to scrutinize loans (loans are essentially risk free for lenders).
- Given the high sensitivity of prices to interest rates and the high household leverage, rising interest rates could prompt downward house price adjustments with negative effects on consumption and the banking sector.
- Estimations of overpricing vary across studies due to sample period, methodology, and variable inclusion; some studies (Hofman (2005)) found prices consistent with fundamentals as of Q2 2004, while others (Verbruggen et al. (2005); Van den End and Kakes (2002)) found evidence of overvaluation or scope for downward movement.

### Policy issues highlighted (tax deductibility and broader policy review)
- Fiscal incentives have made mortgage debt attractive; the paper pays special attention to tax deductibility of mortgage interest payments.
- The Financial System Stability Assessment of 2004 recommended phasing out tax deductibility of mortgage interest over the medium term, preferably gradually to avoid disruptive effects.
- Reasons for eliminating mortgage interest tax deductibility noted in the source:
  - It introduces a distortion in housing markets—making holding mortgage debt attractive and reducing incentive to pay back principal.
  - It favors wealthy households because the tax advantage increases with higher tax brackets and with larger real estate assets.
  - Removing tax deductibility would help the budget in the context of rising costs related to population aging.
  - Revenue from removing the incentive could be used for policies that more directly make housing affordable for low-income households who currently cannot afford a house due to high house prices.
- The paper attempts to quantify and shed light on the expected price impact of phasing out this incentive, drawing on experiences of other countries that removed tax deductibility.
- The paper recommends reviewing policies that encourage excessive household leverage and those affecting banks’ lending practices, given untested systemic risks.

*Source: IMF staff chapter "_cr06284 - 1.      Developments in house prices are of considerable interest in the Netherlands."*

### 18.      Various steps could help reduce the incentives for the risks that households and

### Various steps could help reduce the incentives for the risks that households and banks are taking

### Incentives, NHG guarantee, and mortgage solvency treatment
- The NHG guarantee encourages banks to extend mortgage loans and shifts the risk of default to the government.
- Mortgages guaranteed by the NHG carry no credit risk for banks and have a risk weighting of zero.
- Current Dutch solvency requirements distinguish mortgages by perceived riskiness:
  - Mortgages ≤ 75 percent of the liquidation value of the property: risk weighting of 50 percent.
  - Mortgages > 75 percent of the liquidation value ("top mortgages"): the "loan surplus" above liquidation value has a risk weighting of 100 percent.
- The implicit cost of the NHG guarantee is covered by the government; because it comes at no cost to banks and households, it introduces a price distortion.

### Tax deductibility of mortgage interest payments and household leverage
- Tax deductibility of mortgage interest payments is a fiscal advantage that strongly encourages households to leverage themselves.
- Empirical studies cited show household leverage is highly sensitive to the tax advantage created by interest rate deductibility (Dunsky and Follain, 2000; Follain and Dunsky, 1997; Follain and Ling, 1991; Ling and McGill, 1998; and Hendershott, Pryce, and White 2003).
- Removing interest deductibility:
  - Provides a strong incentive for homeowners with existing loans to pay off their loans.
  - Causes new homeowners to choose loans with less leverage.
  - Particularly affects households that are not income-constrained and are in higher tax brackets, who tend to choose higher leverage because of tax advantages.
- Social goal assessment:
  - Unclear whether interest deductibility makes housing more affordable because housing prices adjust to tax incentives.
  - Interest deductibility favors households in higher tax brackets, using foregone government resources in ways less advantageous to lower-income households.
  - Encourages households to leverage while reducing the incentive to pay down principal.

### Country experiences phasing out mortgage interest deductibility — overview
- Timing and speed of phase-out are important variables for policy design.
- The experiences of the United Kingdom and Sweden are examined as selective case studies.

### United Kingdom: gradual full elimination (1974–99)
- Phase-out timeline and measures:
  - 1974: Ceiling introduced on mortgage size eligible for deductibility; limit set at ₤25,000.
  - 1983: Ceiling raised to ₤30,000; limit never raised again.
  - Maximum deductible tax rate reductions:
    - From a 40 percent maximum income tax rate to 25 percent in 1992.
    - To 20 percent in 1994.
    - To 10 percent in 1995.
    - Down to zero in 1999.
- Binding ceiling development:
  - 1974: median house price ₤10,800; mortgagors above limit: 0 percent.
  - 1983: median house price ₤29,400; mortgagors above limit: 5.4 percent.
  - 1988 to 1991: median house price ₤63,000; mortgagors above limit: 48.4 percent.
  - 1995 to 1998: median house price ₤73,800; mortgagors above limit: 67.4 percent.
- Observed outcomes:
  - The nominal ceiling became progressively binding as house prices rose.
  - The ceiling was introduced toward the end of a recession and just before an upturn, which meant immediate upward pressure on house prices after policy introduction.
  - With the reform taking place over 25 years, the phasing-out appeared to have a limited impact on house prices.
- Econometric estimate (full sample 1970–2005):
  - R2 = 0.8
  - Coefficients and significance:
    - ∆ Real disposable income: Coefficient 1.8162, t-statistic *3.85, Significance 0.00
    - Nominal interest rate (detrended): Coefficient -0.01567, t-statistic *-4.00, Significance 0.00
    - ∆ Representative HH tax deductibility: Coefficient 0.000102, t-statistic **1.66, Significance 0.10
    - ∆ Real house price (lag 1 period): Coefficient 0.30942, t-statistic *2.30, Significance 0.02
  - Note: * Denotes significance at the 5 % critical level; ** Denotes significance at the 10% critical level.
- Interpretation:
  - The coefficient on tax deductibility is small but significant at the 10 percent level.
  - The moderate estimated impact may reflect the long time period over which deductibility was phased out.

### Sweden: faster partial reduction (1985–91) and confounding factors
- Sweden reduced maximum deductible rate:
  - Pre-reform maximum deductible rate: 80 percent.
  - 1985: reduced to 50 percent.
  - 1991: reduced to 30 percent.
- Concurrent reforms and macro conditions:
  - Late 1980s–early 1990s: significant tax reforms (including labor income taxes in 1991: basic marginal tax rate of 30 percent and top marginal tax rate of 50 percent; previous top marginal tax rate up to 75 percent).
  - Various changes to taxation of owner-occupied housing and other financial assets aimed at reducing inter-asset distortions and overinvestment in housing and durables.
  - Recession in the early 1990s.
- Analytical limitations:
  - Data limitations and simultaneous tax changes make it impractical to empirically isolate the impact of changes to mortgage interest deductibility on house prices.
  - Example limitation: Swedish disposable income national series starts only in 1993, after key policy changes.
- Graphical observations:
  - The 1991 reduction to 30 percent occurred during a recession when house prices had stagnated and started to fall; the reduction likely compounded the house price decline.
  - The 1985 reduction (80 to 50 percent) occurred amid strong economic growth, affected mainly high-income households, and real house prices grew strongly despite the partial reduction.

### Phasing out mortgage interest deductibility in the Netherlands: timing, speed, and illustrative scenarios
- Policy design considerations:
  - Timing of phase-out relative to economic conjuncture and speed (time period over which policies bind) matter for outcomes.
  - Analysis treats a loss in household income when tax deductibility is reduced as equivalent to a drop in disposable income; estimates of determinants of real house prices in the Netherlands (Table 3) are used to quantify illustrative price impacts.
  - Numbers are illustrative, not forecasts.
- Three policy options analyzed:
  - Policy 1: Immediate and complete removal of tax deductibility.
  - Policy 2: Introduction of a nominal limit on mortgage loan size eligible for deduction; nominal limit set at the average size of a mortgage loan in 2005, making the option gradually more binding as house prices increase.
  - Policy 3: Reduction of the deductible rate so that the rate applying to mortgage interest deduction is below the marginal income tax rate.
- Economic scenarios:
  - Figure 7 summarizes impacts under the "normal economic growth" scenario; assumptions for this and two other scenarios are presented in Table 6 (not reproduced here).
- Heterogeneous household impacts by income category (Table 6 figures preserved exactly):
  - Low-Income Household:
    - Average house price for category: 200,000
    - Average income of mortgage borrower: 27,000
    - Average loan size of borrower: 127,000
    - Mortgage interest rate (in percent): 5.3
    - Household tax category: 30%
    - Calculated deduction from income: 6,780
    - Tax saving (additional disposable income): 2,034
    - Leverage of average household (Loan/income): 4.70
    - Loan/house price: 0.64
  - Average-Income Household:
    - Average house price for category: 225,200
    - Average income of mortgage borrower: 49,000
    - Average loan size of borrower: 145,000
    - Mortgage interest rate (in percent): 5.3
    - Household tax category: 40%
    - Calculated deduction from income: 7,750
    - Tax saving (additional disposable income): 3,100
    - Leverage of average household (Loan/income): 2.96
    - Loan/house price: 0.64
  - High-Income Household:
    - Average house price for category: 325,000
    - Average income of mortgage borrower: 94,000
    - Average loan size of borrower: 209,000
    - Mortgage interest rate (in percent): 5.3
    - Household tax category: 50%
    - Calculated deduction from income: 11,170
    - Tax saving (additional disposable income): 5,585
    - Leverage of average household (Loan/income): 2.22
    - Loan/house price: 0.64
  - Source: Ministry of finance, March 20, 2006 ("Actualisatie varianten fiscale behandeling eigenwoningbezit").
  - Note: Age: all households <65; income categories: low <€30,000; average €30-60,000; high >€60,000.
- Illustrative quantitative impact under Policy 1 (immediate and complete removal):
  - If removal coincides with the modest recession scenario:
    - Real house prices for high-income households would decline by 20 percent in the first year after the policy change.
    - If the modest recession were to continue, the real house price decline for high-income households would reach 36 percent at the end of the fifth year.
  - A more severe recession, or higher interest rates, would lead to a more severe drop in real prices.

*Source: IMF staff chapter on mortgage interest deductibility and household leverage (content unit _cr06284 - 18).*

### 34.      Under Policies 2 and 3—which show the tax deductibility of mortgage interest

### _cr06284 - 34.      Under Policies 2 and 3—which show the tax deductibility of mortgage interest

### Effects of Policies 2 and 3 on house prices (summary of quantitative results)
- One year after a policy change, price drop for average-income households during a modest recession:
  - Policy 1: -12.4
  - Policy 2: -3.0
  - Policy 3: -5.2
- Six years after policy change, no-policy-change scenario (real house price increases):
  - Normal economic growth: 29.1 percent
  - Exuberant economic growth: 48.8 percent
- Six years after policy change, under Policy 2 (nominal limit on mortgage loans that qualify for interest deduction), assuming normal economic growth, real house price increases by household income:
  - Low-income households: 26.6 percent
  - Average-income households: 22.2 percent
  - High-income households: 14.4 percent
- Results under Policy 3 are broadly similar to Policy 2 for implications on real house prices.
- Definitions of economic scenarios (as used in the quantitative analysis):
  - Modest recession: Disposable income growth per household per year 0.00%; Change in interest rate, per year -0.10
  - Normal growth: Disposable income growth per household per year 3.50%; Change in interest rate, per year 0.15
  - Exuberant growth: Disposable income growth per household per year 5.00%; Change in interest rate, per year 0.20
- Policy scenarios:
  - Policy 1: Immediate abolition of tax deductibility of mortgage interest rates.
  - Policy 2: Mortgage loans that qualify for mortgage interest rate deduction are capped by a nominal limit. The nominal limit is set at €141,000, which is the average mortgage loan in the Netherlands in 2005.
  - Policy 3: The marginal tax rate at which mortgage interest payments can be deducted from income is reduced to 30%.
- Household types:
  - HH1: Low-income household
  - HH2: Average-income household
  - HH3: High-income household

### Distributional and cyclical properties of Policy 2 and Policy 3
- Policy 3 (reduced deductible rate to 30%):
  - Progressive: affects tax-paying households in sufficiently high tax brackets immediately; low-income households are not affected.
  - Immediate large impact on average- and high-income households.
  - Procyclical element: foregone tax benefits are greater as interest rates rise; increase in interest rates also puts downward pressure on house prices, so the change compounds downward pressure on house prices.
- Policy 2 (nominal limit on deductible mortgage loan amount):
  - Progressive: households with high incomes and high loan values are affected immediately; larger loans entail larger foregone tax benefits.
  - Countercyclical element: low- and medium-income households become affected more gradually as the nominal limit only becomes binding over time when house prices increase.
  - If rising interest rates put downward pressure on house prices, the nominal limit becomes less binding as prices fall, providing a countercyclical cushion.
  - When interest rates rise, households could benefit from higher deductions offsetting somewhat higher interest payments.

### Concluding findings and policy recommendations (paragraphs 36–41)
- Observations on recent Dutch housing and mortgage developments:
  - Low interest rates, a shift of bank lending toward the mortgage market, and new mortgage products contributed to buoyant house price developments in the Netherlands in the last two decades.
  - Price increases have been considerable and raise the possibility of a slowdown or temporary reversal if interest rates were to rise, with potential to weaken consumption and growth.
  - Mortgage debt in the Netherlands is higher than in any other industrialized country.
  - New mortgage products and increased loan accessibility present efficiency gains but associated risks need to be well managed.
  - Government-financed insurance schemes for mortgage debt (NHG) have shifted some risk from banks to the government, reducing banks’ incentive to scrutinize mortgage contracts under NHG guarantee.
- Measures to reduce vulnerability of the economy and financial system:
  - Adequately educate the public about risks associated with different mortgage loans (part of the code of conduct for mortgage lenders).
  - Address price distortion from NHG guarantee being provided at no cost to households and banks.
  - Phase-out of tax deductibility of mortgage interest payments would reduce incentives for households to leverage; additional tax revenues could be directed to well-targeted programs for low-income families and/or allow tax cuts to lessen downward impact on housing demand and stimulate labor participation.
- Key considerations when phasing out tax deductibility:
  - (i) Have a progressive impact, with a larger share of the burden on the rich.
  - (ii) Avoid procyclicality whenever possible.
  - (iii) Minimize sudden disruptive effects—favor a gradual phase-out to reduce disruptive impacts.
- Policy preference in light of the above:
  - Introducing a nominal limit on the tax-deductible component of mortgage loans (Policy 2) is identified as an attractive option because it is progressive, has countercyclical features, and tends to have a gradual impact on low- and medium-income households as house prices increase.
  - Reducing the deductible rate (Policy 3) has some positive features but procyclical elements make it less attractive.
  - A gradual phase-out is likely to produce a more gradual decline in house prices when implemented over time and during periods of stronger economic growth; timing of policy matters to avoid disruptive effects.

### Additional analytical findings (appendix highlights)
- Stationarity and cointegration testing (United Kingdom and Netherlands, 1970–2005) informed model specification for estimating price impacts and dynamics:
  - Econometric tests reported include Augmented Dickey Fuller (ADF), Elliot, Rothenberg, and Stock (∆FGLS), and Johansen-Juselius Maximum Likelihood Test for cointegration.
  - For the Netherlands, the nominal interest rate is a borderline case but deemed stationary after detrending using the Elliot, Rothenberg test.
  - For the Netherlands, cointegration tests indicate two cointegration relationships when a stationary variable is included, implying the system is correctly identified for the sample period.

*Source: IMF staff analysis in _cr06284 (selected excerpts).*

### 50.      There are a variety of indicators that have been used to assess tax revenue volatility in

### _cr06284 - 50.      There are a variety of indicators that have been used to assess tax revenue volatility in

### Measurement and methods
- Primary volatility measure: coefficient of variation (CV) defined as the standard deviation relative to the mean of the ratio of tax revenues to GDP, applied to quarterly and annual frequencies.
- Robustness check: econometric evidence using regression analysis, including autoregressive estimation of conditional volatility and regressions of personal income tax (including social contributions) on wage income.
- Heteroskedastic-consistent covariance method used for regression inference (Bollerslev-Wooldridge robust standard error and covariance method).

### Netherlands — Magnitude and trends (quarterly data, Table 1)
- Direct taxes and social contributions CV:
  - 1975–84: 2.5
  - 1985–94: 3.0
  - 1995–2004: 4.4
- Corporate income tax CV:
  - 1975–84: 8.2
  - 1985–94: 5.5
  - 1995–2004: 15.9
- Personal income tax and social contributions CV:
  - 1975–84: 2.8
  - 1985–94: 3.1
  - 1995–2004: 4.4
- Indirect taxes CV:
  - 1975–84: 4.4
  - 1985–94: 3.3
  - 1995–2004: 6.2
- Memo: Wage bill CV:
  - 1975–84: 3.8
  - 1985–94: 1.7
  - 1995–2004: 1.2
- Source: Quarterly data from OECD Economic Outlook 78.

### Netherlands — Magnitude and trends (annual data, Table 2)
- Direct taxes and social contributions CV:
  - 1975–84: 2.2
  - 1985–94: 2.9
  - 1995–2004: 4.5
- Corporate income tax CV:
  - 1975–84: 8.0
  - 1985–94: 5.3
  - 1995–2004: 16.3
- Personal income tax and social contributions CV:
  - 1975–84: 2.5
  - 1985–94: 3.0
  - 1995–2004: 4.4
- Indirect taxes CV:
  - 1975–84: 4.2
  - 1985–94: 3.2
  - 1995–2004: 6.4
- Memo: Wage bill CV:
  - 1975–84: 3.6
  - 1985–94: 1.5
  - 1995–2004: 1.2
- Source: Annual data from OECD Economic Outlook 78.

### Key findings for the Netherlands
- The highest recent volatility (1995–2004, quarterly) is observed in corporate income tax at 15.9 percent, followed by indirect taxes (6.2) and personal income tax and social contributions (4.4).
- Over time (direct taxes and social contributions CV): 2.5 (1975–84) → 3.0 (1985–94) → 4.4 (1995–2004).
- Corporate income tax volatility more than doubled from 1975–84 to 1995–2004 despite a mid-period decline.
- Corporate taxes account for around 8 percent of total tax receipts; personal income tax and social security contributions account for almost 60 percent of total revenues.
- Increased volatility in personal income tax and social contributions: 3.1 percent (1985–94) → 4.4 percent (1995–2004).
- Results are robust to data frequency (annual analysis yields similar results).

### Econometric evidence
- Autoregressive analysis of personal income tax revenues (including social contributions) shows a noticeable increase in conditional volatility over the past decade (1970–2005 series examined).
- Regression of personal income tax on wage income indicates increased volatility in PIT relative to the tax base in 1995–2004 (deviation from the conditional mean increased substantially).
- The increase in conditional volatility implies a larger forecasting error margin for personal income tax revenues and social contributions.

### European context — EU-15 (Table 3)
- Direct taxes and social contributions CV (annual, EU-15 average):
  - 1975–84: 6.8
  - 1985–94: 4.9
  - 1995–2004: 3.2
- Corporate income tax CV (EU-15 average):
  - 1975–84: 16.9
  - 1985–94: 19.9
  - 1995–2004: 18.3
- Personal income tax and social security contributions CV (EU-15 average):
  - 1975–84: 7.2
  - 1985–94: 4.8
  - 1995–2004: 3.4
- Indirect taxes CV (EU-15 average):
  - 1975–84: 5.8
  - 1985–94: 4.0
  - 1995–2004: 4.0
- Memo: Wage bill CV (EU-15 average):
  - 1975–84: 2.6
  - 1985–94: 2.2
  - 1995–2004: 2.0
- Source: Annual data from OECD Economic Outlook 78.
- Cross-country pattern: average EU-15 volatility declined over the decades, unlike the rising volatility in the Netherlands—especially pronounced for personal income tax.

### Cross-country ranking insights
- Personal income tax: Netherlands moved from almost the lowest volatility in 1975–84 to near the top in 1995–2004 (exceeded only by Ireland).
- Corporate income tax: Netherlands had low relative volatility in earlier periods and remained in the lower half of the distribution in 1995–2004 despite a rebound.
- Indirect taxes: Netherlands moved from the middle of the range in the first two decades to near the top in 1995–2004.

### Determinants of tax revenue volatility in the Netherlands
- Investigated factors: amplitude of business cycles, volatility of the wage bill, volatility in pension premiums, corporate location decisions, fluctuations in house prices, conduct of fiscal policy.
- Business cycle volatility (output gap average absolute deviation, Table 4):
  - France: 1.2 (1975–84), 1.5 (1985–94), 0.9 (1995–2005), 0.9 (2000–05)
  - Ireland: 2.8, 2.2, 2.6, 1.8
  - Italy: 1.9, 1.4, 0.9, 0.9
  - Sweden: 2.0, 2.9, 1.4, 0.9
  - United Kingdom: 1.9, 2.5, 0.5, 0.4
  - Germany: 2.3, 2.0, 1.1, 1.5
  - Average: 2.0, 2.0, 1.3, 1.3
  - The Netherlands: 1.8, 1.4, 1.9, 2.4
- Netherlands’ output gap volatility widened in 1995–2005 relative to 1985–94 and appears larger in 2000–05 than in prior decades.
- Short-term elasticities (OECD estimates):
  - Personal income taxes and social contributions elasticity to the output gap: 1.25 percent (weighted average of 1.7 and 0.6).
  - Corporate income tax short-term elasticity: 1.5 percent.
- Interpretation:
  - Short-term elasticity > 1 implies tax revenues amplify cyclical swings (e.g., for PIT: revenues increase 25 percent faster than the economy in upturns and decrease roughly 25 percent more in downturns).
  - Corporate tax elasticity > 1 reflects non-symmetrical treatment of profits and losses and provisions for loss carry-forward.

### Quantified impact of business cycle volatility (Table 5)
- Method: volatility of nominal revenues = average absolute deviation of output gap × short-term elasticity. Volatility of tax ratios = average absolute deviation of output gap × (short-term elasticity − 1).
- Table 5 average absolute deviations (percentages):
  - Output gap:
    - 1975–84: 1.8
    - 1985–94: 1.4
    - 1995–2005: 1.9
    - 2000–2005: 2.4
  - Nominal revenue — PIT and social contributions:
    - 1975–84: 2.3
    - 1985–94: 1.8
    - 1995–2005: 2.4
    - 2000–2005: 3.0
  - Nominal revenue — CIT:
    - 1975–84: 2.8
    - 1985–94: 2.2
    - 1995–2005: 2.9
    - 2000–2005: 3.6
  - Tax ratio — PIT and social contributions:
    - 1975–84: 0.5
    - 1985–94: 0.4
    - 1995–2005: 0.5
    - 2000–2005: 0.6
  - Tax ratio — CIT:
    - 1975–84: 0.9
    - 1985–94: 0.7
    - 1995–2005: 1.0
    - 2000–2005: 1.2
- Conclusion: higher business cycle volatility explains part of the observed increase in tax revenue volatility.

### Implications for forecasting and sensitivity illustration
- Increased standard deviation of PIT residuals (from AR and PIT-on-wage regressions) implies larger forecasting error margins for personal income tax revenues and social contributions.
- Example sensitivity: a 2 percent increase in wages, with PIT elasticity ≈ 1.25, implies expected PIT and social contributions growth of 2½ percent.
  - Using recent increased volatility (Table 5 deviation), actual growth would vary between 1.3 and 3.7 percent — a margin of 2.4 percentage points.
  - The same wage increase 20 years earlier would have implied growth between 1.8 and 3.2 percent — a smaller margin.

*Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2006/_cr06284.pdf*

### 1.4 percentage points.

### _cr06284 - 1.4 percentage points.

### Business cycle and tax revenue volatility
- Formal econometric analysis attributes a significant part of the increase in tax revenue volatility to business cycle fluctuations.
- Estimated magnitude: effect varies between 0.05 and 0.08 depending on specification. Example interpretation in text: "a one percent increase in the output gap volatility would lead to an increase in the volatility of tax revenue that varies between 0.05 and 0.08 percent."
- OECD estimate: overall budget elasticity at around 0.6 percent for the Netherlands.
- Historical output gap behavior cited:
  - output gap varied within an interval of around ±4 percent of GDP over the past 30 years;
  - changed year-on-year by 1¼ percent, on average, in absolute terms.
  - implication: fiscal balance can be expected to change by ¾ percentage points of GDP from year to year and vary within an interval of around ±2½ percent of GDP due to business cycle fluctuations.

### The role of pension funds
- Two defining characteristics of the Dutch private pension system with implications for tax receipts:
  - Predominantly "defined benefit" structure (unlike several other European countries with defined contribution schemes).
  - Netherlands (with sole exception of Denmark) allows income tax deduction of pension contributions.
- Mechanism: changes in pension contributions (which may follow changes in funds’ asset holdings) have a direct impact on tax receipts through deductibility.
- Asset allocation and timing:
  - share of equities in pension funds’ investment portfolios increased from 10 percent in 1990 to more than 40 percent in 2005;
  - large increases in pension assets during the late 1990s and significant downward pressure in the early 2000s following stock price declines.
- Empirical relationship: negative relationship between tax ratios and pension premia; Table 6 correlations (periods and series preserved exactly as in source):
  - 1975–84: Direct taxes and social contributions -47.6; Corporate income tax 19.9; Personal income tax and social contributions -52.5; Total taxes -40.1.
  - 1985–94: Direct taxes and social contributions -23.9; Corporate income tax -38.0; Personal income tax and social contributions -17.9; Total taxes -17.9.
  - 1995–03: Direct taxes and social contributions -79.7; Corporate income tax -78.4; Personal income tax and social contributions -45.5; Total taxes -80.7.
  - 1975–03: Direct taxes and social contributions -42.7; Corporate income tax -15.4; Personal income tax and social contributions -33.1; Total taxes -45.3.
- ARCH/GARCH empirical finding: higher volatility in pension premia is important in explaining increased personal income tax volatility; positive parameter value for pension premia volatility in the conditional variance function implies increasing pension premia have contributed to increased tax revenue volatility.

### Other contributing factors
- Global integration and EU enlargement:
  - likely affected corporate location decisions to book profits and minimize tax burdens, increasing tax revenue volatility.
  - larger and more volatile transaction flows and asset price swings can amplify tax revenue fluctuations.
  - Netherlands relies heavily on trade (Figure 6 cited on trade openness).
- Mortgage interest payment deductibility:
  - mortgage interest payments are (for the most part) deductible in the Netherlands.
  - volatility in mortgage interest payments, mortgage debt, house prices, and mortgage rates can affect volatility of foregone income tax revenue.
- Housing wealth and consumption channels:
  - CPB (2006) study: housing wealth has been growing steadily since 1995 except 2000–2003 (growth rate declined slightly and turned slightly negative in 2003).
  - CPB estimates marginal propensity to consume out of housing wealth for durable goods to be 0.1 percent.
  - Asymmetry: impact of house price gains on durables consumption not significant, while impact of losses is significant and twice as large—households cut durables spending more strongly on wealth losses than they increase spending on gains; this asymmetry may have increased volatility in indirect taxes.
- Fiscal policy and tax-regime changes:
  - shifts in tax regime (e.g., shift from direct to more indirect taxation in 2001) can cause sudden changes in tax revenues as percent of GDP followed by adjustment and increased volatility.
  - past fiscal framework practice: government used ex post deviations of tax revenues from pre-established nominal reference levels for discretionary tax changes—such practice may have increased volatility of tax revenues as a share of GDP, though measurement is difficult without a tax policy change series.

### Empirical framework and data (Appendix summary)
- Objective: determine whether fluctuations in pension premiums and the business cycle explain increased volatility of personal income tax revenues in the Netherlands.
- Methodology: linear autoregressive conditional heteroskedastic (ARCH) framework (Engle (1982) and Bollerslev (1986) extensions) to estimate volatility of tax revenues and pension premiums; allows simultaneous estimation of parameters for explanatory variables in conditional variance.
- Key modeling choices:
  - conditional mean of personal income tax revenue modeled alternately as an AR(1) or as a lagged function of wage income (wage income as share of GDP).
  - conditional variance of personal income tax revenue modeled to include lagged squared residuals and additional explanatory variables: conditional variance of pension premiums and output gap volatility (absolute deviation from its conditional mean).
  - pension premiums modeled as an autoregressive process to capture their conditional variance; that conditional variance incorporated into the ARCH specification for personal income tax revenues.
- Sample and data:
  - sample period 1970—2005, quarterly data on GDP, output gap, personal income tax revenues, pension premiums, and wage income from OECD and Statistics Netherlands.
  - variables show structural events around 1980 and 1990; first differencing renders series stationary per ADF-statistics.
- Diagnostic tests:
  - LM-tests for conditional heteroskedasticity indicate presence of ARCH(1) in personal income tax revenues series; hypothesis of conditional homoskedasticity for pension premiums is not rejected.
  - sign-and-size-biased test showed no asymmetric impact of positive/negative shocks on conditional variance.
  - outliers detected in 1995 and 2001 reflecting tax-regime changes; outlier-robust GM estimation used in AR(1) context.

### Conclusions and policy implications
- Observed changes:
  - tax revenue volatility increased substantially over the last decade in the Netherlands, particularly for corporate taxes; volatility in personal income taxes and indirect taxes also increased substantially.
  - on average, other EU countries did not exhibit similar increases; volatility was more or less constant or slightly declining.
- Factors identified as contributing to increased volatility:
  - larger business cycle fluctuations (global activity, trade, asset prices);
  - tax deductibility of mortgage interest payments;
  - corporate location and profit-booking decisions amid tax competition;
  - tax deductibility of pension premia (empirical evidence supports this as a key factor).
- Policy implication highlighted:
  - the Dutch fiscal framework includes a fiscal deficit signal value of 2½ percent of GDP (triggers corrective measures to avoid breaching the 3 percent Maastricht limit).
  - this signal value was adopted when budgetary uncertainties were somewhat less; higher tax revenue volatility suggests consideration could be given to a somewhat more conservative signal value.

*Source: _cr06284 - 1.4 percentage points.*

### 87.      The formal estimation results support the hypothesis that the volatility of personal

### _cr06284 - 87.      The formal estimation results support the hypothesis that the volatility of personal

### Main findings on personal income tax revenue volatility
- The volatility of personal income tax revenue is affected by developments in pension premiums and the business cycle.
- The conditional mean of personal income tax revenues can be modeled both as an autoregressive process and as a function of wage income, with elasticities:
  - autoregressive specification: 0.9—1.0
  - wage-income specification: 1.0—1.10
- The larger-than-unity tax revenue elasticity for personal income tax reflects the progressive nature of the tax system in the Netherlands.
- Pension premiums, the business cycle, and their combination play a significant role in determining fluctuations in personal income tax revenues around the conditional mean.
- All estimated coefficients in the ARCH-specifications have the correct signs; positive signs support the hypothesis that an increase in pension premiums or business cycle increases the volatility of personal income tax revenues.
- The estimate for the constant term is significantly different from zero.
- The parameter estimates for the lagged residual term, the pension premium volatility, and the volatility of the output gap are significantly different from zero.
- Semi-elasticities:
  - Pension premium volatility: a one percent increase in premium volatility leads to about 6 to 8 percent increase in personal income tax revenue volatility.
  - Output gap volatility: the effect varies between 5 percent and 8 percent, depending on the functional form of the conditional mean and the conditional variance.
- Direct regression of squared residuals from both conditional-expectation models on the conditional variance of pension contribution rates and the output gap volatility produced t-values of 2.2 and 3.2, respectively, indicating pension contributions and business cycle significantly explain increased personal income tax revenue volatility.
- The Akaike criterion discriminates in favor of:
  - the autoregressive model for the conditional mean
  - the conditional variance being better explained by a combination of pension contribution and output gap volatility

### Estimation approach and diagnostics
- ARCH-specification was used for personal income tax revenue volatility as specified in Equations (3)–(6).
- Numerical algorithm: Berndt-Hall-Hall-Hausman (BHHH, 1974) used to find quasi maximum likelihood parameter estimates for the ARCH-specification.
- Robustness checks: different starting values used to ensure convergence to the global maximum.
- Standard errors and covariance: Bollerslev-Wooldridge’s procedure used to obtain robust standard errors and covariance.
- ARCH LM-test statistics reported in tables; specific ARCH-LM values include: 4.58, 4.35, 4.65, 4.40, 4.62, 4.37 across specifications.

### Key quantitative estimation results (selected from Tables 2–7)
- Table 2 (Equation (3) mean; Equation (5) variance):
  - Conditional mean: Constant α1 = -1.30 [-66.93]; yt-1 α2 = 0.93 [23.48]
  - Conditional variance: Constant β1 = 0.001 [7.58]; β2 (lagged squared residual) = 0.50 [3.24]; β3 (PtV) = 0.07 [6.45]
  - Adj. R2 = 0.91; ARCH-LM = 4.58
- Table 3 (Equation (4) mean; Equation (5) variance):
  - Conditional mean: Constant α1 = -0. 78 [-21.71]; wt α2 = 1.02 [14.28]
  - Conditional variance: Constant β1 = 0.001 [7.93]; β2 = 0.95 [15.69]; β3 (PtV) = 0.06 [12.04]
  - Adj. R2 = 0.16; ARCH-LM = 4.35
- Table 4 (Equation (3) mean; Equation (6) variance):
  - Conditional mean: Constant α1 = -1.29 [-10.46]; yt-1 α2 = 0.99 [76.35]
  - Conditional variance: Constant β1 = 0.01 [5.19]; β2 = 0.60 [1.99]; Β3 (GtV) = 0.04 [3.89]
  - Adj. R2 = 0.97; ARCH-LM = 4.65
- Table 5 (Equation (4) mean; Equation (6) variance):
  - Conditional mean: Constant α1 = -0. 44 [-1.95]; wt α2 = 1.10 [4.00]
  - Conditional variance: Constant β1 = 0.02 [1.97]; β2 = 0.74 [4.81]; Β3 (GtV) = 0.08 [3.12]
  - Adj. R2 = 0.19; ARCH-LM = 4.40
- Table 6 (Equation (3) mean; Equation (7) variance):
  - Conditional mean: Constant α1 = -1.01 [-0.43]; yt-1 α2 = 0.99 [64.38]
  - Conditional variance: Constant β1 = 0.01 [7.74]; β2 = 0.39 [4.19]; β3 (PtV) = 0.07 [3.82]; Β4 (GtV) = 0.05 [7.90]
  - Adj. R2 = 0.97; ARCH-LM = 4.62
- Table 7 (Equation (4) mean; Equation (7) variance):
  - Conditional mean: Constant α1 = -0. 89 [-21.71]; wt α2 = 0.73 [3.43]
  - Conditional variance: Constant β1 = 0.01 [9.16]; β2 = 0.22 [4.53]; β3 (PtV) = 0.06 [10.66]; Β4 (GtV) = 0.04 [1.88]
  - Adj. R2 = 0.16; ARCH-LM = 4.37

### Variable definitions used in estimations
- yt-1 = personal income tax revenue as percent of GDP
- wt-1 = wage income as percent of GDP
- 2 1−t ε = the lagged squared residuals of Equation (5)
- PtV = the conditional variance of the pension premiums
- GtV = the “volatility” of the output gap

*Source: _cr06284 (excerpts from ARCH-specification estimation results and tables).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2006/_cr06284.pdf_