## _wp1057 - Bibliography ...........................................................................................................

## Source details

**Canonical URL:** [_wp1057 - Bibliography ...........................................................................................................](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1057.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1057.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1057.pdf.json)

---

### I. BACKGROUND
- Uses of structural fiscal balances:
  - provide an estimate of the underlying fiscal position and guidance on the health and direction of fiscal policy;
  - determine the size and direction of automatic stabilizers;
  - assess long-run fiscal sustainability by indicating trends as temporary factors dissipate.
- Stability and Growth Pact (SGP): all member states required to maintain a medium-term fiscal objective defined in terms of the structural fiscal balance.
- Measurement challenges:
  - estimation of potential output and output gaps;
  - adjustment of fiscal revenues for the effect of the business cycle using estimated revenue elasticities;
  - whether adjustments for asset price cycles, changes in the shares of various components of national income, or other factors are needed.
- Prominent approaches:
  - OECD approach (Girouard and Andre, 2005): adjusts for the business cycle but not for asset price cycles or changes in composition of national income.
  - European Commission variant (Larch and Turrini, 2009): basis for SGP structural balance calculations.
  - ECB disaggregated approach (Bouthevillain and others, 2001): takes changes in composition of national income into account.
- IMF’s traditional approach for Ireland (broadly following OECD approach):
  - Potential output estimated using a Cobb Douglas production function; output gaps then calculated.
  - Structural revenues calculated using an aggregate elasticity of revenue with respect to the output gap estimated to be 1.08, and the estimated output gap, to extract the cyclical component of revenue.
  - On expenditure side, only unemployment-related benefits considered cyclical; NAIRU estimated using a HP filter on unemployment rate data to estimate cyclical impact on unemployment benefits.
  - Structural balance = structural revenues − structural expenditures.
- Limitations of the traditional approach:
  - Does not account for asset price booms or sectoral changes in national income composition.
  - In Ireland’s case, not accounting for sectoral changes can create serious biases because different components of national income have different impacts on fiscal revenues.
  - Such biases are particularly dangerous during asset price booms because abrupt ends reveal large fiscal holes.

### II. IRISH EXPERIENCE
- Macroeconomic shocks and outcomes:
  - Ireland entered a deep recession in 2008 after years of strong growth, hit by three shocks: foreign trade, the financial sector, and the housing market.
  - The shocks to foreign trade and the financial sector were largely due to the global crisis; the housing market bust was primarily homegrown.
  - The economy contracted 3 percent in 2008.
  - The contraction intensified in 2009—GDP contracted 8.1 percent in Q1-Q3 2009.
- Asset price developments, 1995-2007:
  - Both house and equity prices more than quadrupled between 1995 and 2007.
  - Asset-based tax revenues increased substantially in importance, especially property-based taxes (stamp duty and capital taxes, and VAT on property).
  - Property-based taxes grew at a pace well above GDP growth during the boom.
  - When asset prices collapsed, asset-based taxes collapsed and a large hole in public finances emerged.
- Issues in estimating Irish structural balance:
  - Pre-crisis estimates based on the traditional OECD-based approach failed to warn of the scale of deterioration.
  - Small estimated output gaps (due to assumptions about labor market integration and migration effects keeping unemployment near NAIRU) led to treating buoyant property-based revenues as structural.
  - Asset prices and asset price cycles are not included in the standard production function approach, so the methodology lacked a mechanism to correct for asset bubbles.
  - The collapse in revenues in 2008 was well beyond what the standard methodology could explain, especially for asset-based taxes.
- Alternative output gap estimation and impact:
  - Athanasopoulou (2009) used a multivariate Kalman filtering approach with potential output following a stochastic trend with mean-reverting growth rate.
  - Revised output gap estimates produced a higher estimated output gap in 2007.
  - Correcting the output gap alone worsened the estimated structural fiscal balance in 2007 by 1¾ percentage points of GDP.
- Empirical discrepancy in 2008:
  - Given an elasticity for overall tax revenues with respect to nominal GDP estimated at slightly above unity, the collapse in tax revenues should have been only slightly higher than the collapse in nominal GDP.
  - In reality, revenue collapse in 2008 was 8½ percent—double the 4¼ percent contraction in nominal GDP.
  - Discrepancies were particularly striking for taxes sensitive to asset markets.
- Conclusion from Irish experience:
  - A more disaggregated approach explicitly considering asset prices and sectoral shifts is necessary to assess what portion of fiscal revenues is structural.
  - Expanding structural-balance calculation to account for asset price cycles and sectoral shifts indicates a large fiscal hole had emerged by 2007 even before the global crisis struck.

### III. THE MODEL
- Purpose: estimate fiscal revenues accounting for asset prices and sectoral changes; the equation is a reduced form linking revenue and its base to explanatory macro variables.
- Two formulations depending on stationarity.

- A. Model with Stationary Data
  - Log-linear model with lags (Rt as function of Mit and elasticities ali).
  - Structural revenue computed using estimated coefficients and structural levels of explanatory variables.

- B. Model with Nonstationary Data
  - Stationarity-inducing transformation: divide variables by nominal GDP (mt = Mit / Yt) to form ratios.
  - Structural value of the revenue/GDP ratio expressed as product of scaled variables raised to estimated elasticities.
  - Alternatives if scaling fails:
    - estimate in first differences; or
    - use an error correction model allowing cointegration, which includes lagged structural revenue and requires an identification assumption for level in at least one year.
- Note: paper abstracts from discretionary measures effects on tax elasticities; European Commission (2009) adjusts revenue data for discretionary measures (excluding Ireland) and finds relatively small overall impact on tax elasticity estimates though substantial divergences in certain years.

### IV. ESTIMATION RESULTS (introductory material)
- Log-linear regressions estimated for all revenue categories on annual data for 1987- (dataset end not shown in provided content).
- Estimation choices framed by stationarity properties; multiple specifications (levels, ratios to GDP, first differences, error-correction) used depending on data properties.
- Methodological points:
  - Use of lags of explanatory variables and elasticities to capture dynamic responses.
  - Growth-rate (error-correction) specifications include lagged structural revenue and require an identification assumption for level in at least one year.

### Unit-root testing and data treatment
- All variables are expressed in logs and in percent of GDP.
- Ng-Perron (MZa) unit-root test results reject the unit-root hypothesis for all variables using Ng-Perron (2001) test.
  - Critical values: 1 percent level, -23.8; 5 percent level, -17.3.
- Selected MZa statistics:
  - Personal income tax: -42.19 ***
  - Corporate tax: -39.14 ***
  - VAT: -17.84 **
  - Excise duty: -21.19 **
  - Capital taxes: -28.12 ***
  - Stamp duty: -73.38 ***
  - Other taxes: -22.04 **
  - Social insurance contributions: -30.63 ***
  - Miscellaneous current revenue: -18.58 **
  - Capital revenue: -22.06 **
  - Personal consumption: -68.37 ***
  - Residential investment: -99.30 ***
  - Negative of net factor income from abroad: -65.15 ***
  - Exports of goods and services: -39.50 ***
  - House price index: -27.50 ***
  - Equity price index: -33.06 ***
- Division by GDP implies regressions explain movements in revenue/GDP ratios rather than levels; GDP does not enter as a stand-alone explanatory variable.

### Regression specification and estimated determinants of revenues
- Final significant explanatory variables (all variables in percent of GDP): nominal consumption, nominal residential investment, house price index, equity price index, exports, and net factor income from abroad.
- OLS used; justification: explanatory variables unlikely to be significantly influenced by movements in any single revenue type.
- Coefficient interpretation (variables in percent of GDP):
  - positive coefficient → faster-than-GDP increases in explanatory variable associated with faster-than-GDP increases in revenue type;
  - negative coefficient → faster-than-GDP increases in explanatory variable associated with slower-than-GDP increases in revenue type.
- Regression fit (selected R-squares):
  - PIT: 0.96
  - Other taxes: 0.79
  - Social contributions: 0.71
  - Misc. rev.: 0.96
  - Capital rev.: 0.95
  - CIT: 0.87
  - VAT: 0.94
  - Excise: 0.86
  - Capital tax: 0.93
  - Stamp duty: 0.33

### Summary of key regression findings (direction and significance)
- Personal consumption/GDP ratio:
  - Significant positive impact on PIT, VAT, excise tax, other taxes, and social contributions (to GDP).
- Residential investment/GDP ratio:
  - Significant positive impact on VAT, capital tax, stamp duty, and social contributions.
  - Positive but just insignificant at the 10 percent level on other taxes.
  - Significant negative impact on miscellaneous current revenue.
- Negative of net factor income from abroad/GDP:
  - Significant negative impact on PIT, CIT, and miscellaneous current revenue.
- Exports/GDP ratio:
  - Significant positive impact on CIT/GDP.
- House price index/GDP ratio:
  - Significant positive impact on stamp duty/GDP.
  - Significant negative impact on capital revenues/GDP.
- Equity price index/GDP ratio:
  - Significant positive impact on PIT and excise tax/GDP.
  - Significant negative impact on other taxes, social contributions, and miscellaneous current revenues/GDP.

### Calculating structural revenues
- Structural revenue/GDP ratio for each revenue type requires estimating how far each explanatory variable is from its structural (fundamental) value.
- Structural values obtained using HP filter (trend extraction), with projections up to 2014 to minimize end-point problems.
- HP-filter extraction satisfactory for most variables except residential investment, where the housing bubble was largely ascribed to trend by the HP filter.
- Residential investment/GDP decomposition used:
  - Residential investment/GDP = (price index for residential construction / GDP per capita) × (real residential construction / population) × population / GDP.
- Assumption: residential investment has a stable long-run relationship with nominal GDP; deviations reflect cyclical or bubble factors that return to fundamentals over the long run.
- Alternative treatment for residential investment:
  - Use projected path where nominal residential investment declines about 42 percent in 2009 (noted as slightly optimistic versus first three quarters of 2009 decline of 48 percent), a further contraction of about 30 percent in 2010, and positive growth thereafter similar to overall nominal GDP growth.
  - Projected 2014 residential investment/GDP taken as the structural value (implied fundamental value similar to early 1990s).
- For all other explanatory variables the HP-filtered trends are used as structural values.
- Overall structural revenue is the sum of structural levels of different revenue types.
- Largest divergences between actual and structural revenues (percent of GDP) occur for VAT, capital taxes, stamp duty, and social contributions.

### Structural expenditures and structural balance
- Assumption: unemployment benefits are the only significant cyclical component of fiscal expenditure.
  - Total expenditure = expenditure excluding unemployment benefits + unemployment benefits.
  - Unemployment benefits = unemployment rate × labor force × average benefits.
  - Structural unemployment benefits = NAIRU × labor force × average benefits.
  - Structural expenditure = expenditure excluding unemployment benefits + structural unemployment benefits.
- Structural expenditure divided by potential GDP (potential GDP obtained using Athanasopoulou approach) to compute structural balance: structural revenue/GDP ratio minus structural expenditure/potential GDP ratio.

- Table 3: Structural Fiscal Position (in percent of potential GDP unless otherwise stated)
  - Structural revenue (2000–2008): 33.5, 31.5, 31.3, 30.0, 31.6, 31.2, 30.9, 31.6, 29.7
  - Structural expenditure (2000–2008): 32.7, 33.8, 33.8, 32.9, 33.0, 33.5, 34.8, 38.6, 42.7
  - Structural balance (2000–2008): 0.8, -2.3, -2.5, -2.9, -1.4, -2.3, -3.9, -7.0, -13.0
- Memo items:
  - Structural balance under alternative approaches (production function approach): 4.3, 0.9, -0.4, 0.5, 1.6, 1.5, 2.4, -1.3, -6.7
  - Using Athanasopoulou output gap (alternative): 2/2.5, -0.5, -1.5, -0.1, 1.2, 1.1, 1.5, -3.0, -8.7
  - Headline revenue (%GDP): 35.4, 33.4, 32.4, 32.8, 34.2, 34.6, 36.4, 35.8, 34.1
  - Headline expenditure (%GDP): 30.6, 32.5, 32.7, 32.4, 32.8, 33.0, 33.5, 35.7, 41.2
  - Headline balance (%GDP): 4.8, 0.9, -0.3, 0.4, 1.4, 1.6, 2.9, 0.1, -7.2
- Key interpretation:
  - The structural fiscal deficit was significantly higher than previously estimated: increased to 7 percent of potential GDP in 2007 and to 13 percent of potential GDP in 2008.
  - Most structural deterioration occurred on the expenditure side rather than the revenue side: between 2000 and 2008 structural expenditures increased by 10 percentage points of potential GDP, compared with a 3.8 percentage point decline in structural revenues.
  - Structural expenditures rose by almost 8 percentage points of potential GDP between 2006 and 2008.

### Conclusions and policy implications
- The paper extends standard structural-balance estimation for Ireland by explicitly accounting for asset price cycles, the housing bubble, and changes in national income composition.
- Asset price cycles and the housing bubble had significant impacts on Irish fiscal revenues, especially on indirect taxes like VAT, capital gains, and stamp duty; equity prices significantly impact personal income taxes.
- Changes in the sectoral composition of GDP materially affected the evolution of structural revenues.
- Not accounting for asset price cycles, bubbles, or sectoral changes creates a large bias in structural-balance calculations that cannot generally be corrected by altering the output gap estimate alone; in Ireland’s case this bias made revenue prospects appear rosier and helped prompt an expenditure relaxation that created a large fiscal gap.
- Recommendation/implication:
  - expand standard OECD-based methodology to incorporate asset price cycles and sectoral composition effects (the paper advocates the approach used here).
  - ECB is noted as already moving to refine the treatment of asset-based taxes in a disaggregated approach.

*Source: _wp1057 - Bibliography ...........................................................................................................*

### Bibliography ...........................................................................................................

### _wp1057 - Bibliography ...........................................................................................................

### I. BACKGROUND
- Structural fiscal balances are used to:
  - provide an estimate of the underlying fiscal position and guidance on the health and direction of fiscal policy;
  - determine the size and direction of automatic stabilizers;
  - assess long-run fiscal sustainability by indicating trends as temporary factors dissipate.
- Under the Stability and Growth Pact (SGP) all member states are required to maintain a medium-term fiscal objective defined in terms of the structural fiscal balance.
- Measurement challenges in calculating the structural balance include:
  - estimation of potential output and output gaps;
  - adjustment of fiscal revenues for the effect of the business cycle using estimated revenue elasticities;
  - whether adjustments for asset price cycles, changes in the shares of various components of national income, or other factors are needed.
- Prominent approaches:
  - OECD approach (Girouard and Andre, 2005): adjusts for the business cycle but not for asset price cycles or changes in composition of national income.
  - European Commission variant (Larch and Turrini, 2009): basis for SGP structural balance calculations.
  - ECB disaggregated approach (Bouthevillain and others, 2001): takes changes in composition of national income into account.
- IMF’s traditional approach for Ireland (broadly following OECD approach):
  - Potential output estimated using a Cobb Douglas production function; output gaps then calculated.
  - Structural revenues calculated using an aggregate elasticity of revenue with respect to the output gap estimated to be 1.08, and the estimated output gap, to extract the cyclical component of revenue.
  - On expenditure side, only unemployment-related benefits considered cyclical; NAIRU estimated using a HP filter on unemployment rate data to estimate cyclical impact on unemployment benefits.
  - Structural balance = structural revenues − structural expenditures.
- Limitations of the traditional approach:
  - Does not account for asset price booms or sectoral changes in national income composition.
  - In Ireland’s case, not accounting for sectoral changes can create serious biases because different components of national income have different impacts on fiscal revenues.
  - Such biases are particularly dangerous during asset price booms because abrupt ends reveal large fiscal holes.

### II. IRISH EXPERIENCE
- Ireland entered a deep recession in 2008 after years of strong growth, hit by three shocks: foreign trade, the financial sector, and the housing market.
- The shocks to foreign trade and the financial sector were largely due to the global crisis; the housing market bust was primarily homegrown.
- Key macro outcomes:
  - The economy contracted 3 percent in 2008.
  - The contraction intensified in 2009—GDP contracted 8.1 percent in Q1-Q3 2009.
- Asset price developments, 1995-2007:
  - Both house and equity prices more than quadrupled between 1995 and 2007.
  - Asset-based tax revenues increased substantially in importance, especially property-based taxes (stamp duty and capital taxes, and VAT on property).
  - With residential investment and house prices soaring, property-based taxes grew at a pace well above GDP growth.
  - The change in tax structure increased exposure of fiscal revenues to asset market corrections; when asset prices collapsed, asset-based taxes collapsed and a large hole in public finances emerged.
- Issues in estimating Irish structural balance:
  - Pre-crisis estimates based on the traditional OECD-based approach failed to warn of the scale of deterioration.
  - Small estimated output gaps (due to assumptions about labor market integration and migration effects keeping unemployment near NAIRU) led to treating buoyant property-based revenues as structural.
  - Asset prices and asset price cycles are not included in the standard production function approach, so the methodology lacked a mechanism to correct for asset bubbles.
  - The collapse in revenues in 2008 was well beyond what the standard methodology could explain, especially for asset-based taxes.
- Alternative output gap estimation:
  - Athanasopoulou (2009) used a multivariate Kalman filtering approach with potential output following a stochastic trend with mean-reverting growth rate.
  - The revised output gap estimates differ significantly from production function estimates and produced a higher estimated output gap in 2007.
  - Correcting the output gap alone worsened the estimated structural fiscal balance in 2007 by 1¾ percentage points of GDP.
- Empirical observations on revenue vs nominal GDP:
  - Given an elasticity for overall tax revenues with respect to nominal GDP estimated at slightly above unity, the collapse in tax revenues should have been only slightly higher than the collapse in nominal GDP.
  - In reality, revenue collapse in 2008 was 8½ percent—double the 4¼ percent contraction in nominal GDP.
  - Discrepancies were particularly striking for taxes sensitive to asset markets.
  - Prior literature (Eschenbach and Schuknecht (2002), Girouard and Price (2004), Morris and Schuknecht (2007)) finds that excluding asset prices can lead to serious biases in estimation of structural balances.
- Conclusion from Irish experience:
  - A more disaggregated approach explicitly considering asset prices and sectoral shifts is necessary to assess what portion of fiscal revenues is structural.
  - The paper expands structural-balance calculation for Ireland to take explicit account of asset price cycles and sectoral shifts and finds these make a substantial difference—indicating a large fiscal hole had emerged by 2007 even before the global crisis struck.

### III. THE MODEL
- Purpose: estimate fiscal revenues accounting for asset prices and sectoral changes; the estimated equation can be seen as a reduced form linking revenue and its base to explanatory macro variables.
- Model depends on stationarity of data. Two formulations presented.

- A. Model with Stationary Data
  - Equation (1) (log-linear model with lags):
    - Rt = product over i and lags of (M_{i,t-l} ^ a_{i,l}) times error term (expressed in logged form in source).
    - Definitions:
      - Rt represents revenue in time t.
      - Mit represents the value of the ith explanatory variable in time t.
      - ali represents the elasticity of revenue to the lth lag of the ith explanatory variable.
      - εt is the error term.
  - Structural revenue formula (equation (2)):
    - Estimated structural revenue is the product over explanatory variables and lags of (M_{i,t-l}^â_{i,l}) using estimated coefficients (hats) and structural levels (stars) as given.

- B. Model with Nonstationary Data
  - Stationarity-inducing transformation: division by nominal GDP (Yt) to form ratios (mt = Mit / Yt), since ratios of macro variables to GDP are likely bounded and more likely stationary; this was found to work for variables used in the paper.
  - Structural value of the revenue/GDP ratio (equation (3)):
    - rt (ratio of revenue to GDP) = product over i and lags of (m_{i,t-l} ^ â_{i,l}) using scaled variables.
  - If scaling does not work, alternatives:
    - Estimate in first differences; or
    - Use an error correction model allowing cointegration (equation (4) in source), which relates growth rates and includes an equilibrium term.
  - Error-correction growth formulation (equation (5)):
    - Structural revenue determined by a product formula that includes lagged structural revenue—pinning down levels requires additional assumptions to identify structural revenue in at least one year.
- Note: The paper abstracts from consideration of the effects of discretionary measures on tax elasticities; European Commission (2009) attempts to adjust revenue data for discretionary measures in a number of countries (excluding Ireland) and finds overall relatively small impact on tax elasticity estimates though substantial divergences observed in certain years.

### IV. ESTIMATION RESULTS (introductory material)
- Log-linear regressions were estimated for all revenue categories on annual data for 1987- (dataset end not shown in provided content).
- The paper frames estimation choices by stationarity properties and offers multiple model specifications (levels, ratios to GDP, first differences or error-correction models) depending on data properties.
- Key methodological points preserved from the model description relevant to estimation:
  - Use of lags of explanatory variables and elasticities to capture dynamic responses.
  - When estimating in growth rates (error-correction form), structural revenue determination involves the lagged value of structural revenue, requiring an identification assumption for level in at least one year.

*Italic: Source: _wp1057 - Bibliography ...........................................................................................................*

### 2008. Data used were all divided by GDP. For all variables (in percent of GDP), the unit root

### _wp1057 - 2008. Data used were all divided by GDP. For all variables (in percent of GDP), the unit root

### Unit-root testing and data treatment
- All variables are expressed in logs and in percent of GDP.
- Ng-Perron (MZa) unit-root test results (Table 1) reject the unit-root hypothesis for all variables using Ng-Perron (2001) test. Critical values: 1 percent level, -23.8; 5 percent level, -17.3.
- MZa statistics (Table 1, selected):
  - Personal income tax: -42.19 ***
  - Corporate tax: -39.14 ***
  - VAT: -17.84 **
  - Excise duty: -21.19 **
  - Capital taxes: -28.12 ***
  - Stamp duty: -73.38 ***
  - Other taxes: -22.04 **
  - Social insurance contributions: -30.63 ***
  - Miscellaneous current revenue: -18.58 **
  - Capital revenue: -22.06 **
  - Personal consumption: -68.37 ***
  - Residential investment: -99.30 ***
  - Negative of net factor income from abroad: -65.15 ***
  - Exports of goods and services: -39.50 ***
  - House price index: -27.50 ***
  - Equity price index: -33.06 ***
- Division by GDP implies regressions explain movements in revenue/GDP ratios rather than levels; GDP does not enter as a stand-alone explanatory variable.

### Regression specification and estimated determinants of revenues
- Variables considered and final significant explanatory variables: nominal consumption, nominal residential investment, house price index, equity price index, exports, and net factor income from abroad.
- Ordinary Least Squares (OLS) used; authors argue OLS yields consistent estimates because explanatory variables are unlikely to be significantly influenced by movements in any single revenue type.
- Coefficient interpretation (all variables in percent of GDP): a positive coefficient implies faster-than-GDP increases in the explanatory variable are associated with faster-than-GDP increases in the revenue type; a negative coefficient implies faster-than-GDP increases in the explanatory variable are associated with slower-than-GDP increases in the revenue type.
- Regression fit (Table 2, selected R-squares for dependent variables):
  - PIT: 0.96
  - Other taxes: 0.79
  - Social contributions: 0.71
  - Misc. rev.: 0.96
  - Capital rev.: 0.95
  - CIT: 0.87
  - VAT: 0.94
  - Excise: 0.86
  - Capital tax: 0.93
  - Stamp duty: 0.33

### Summary of key regression findings (direction and significance)
- Personal consumption/GDP ratio:
  - Significant positive impact on PIT, VAT, excise tax, other taxes, and social contributions (to GDP).
- Residential investment/GDP ratio:
  - Significant positive impact on VAT, capital tax, stamp duty, and social contributions.
  - Positive but just insignificant at the 10 percent level on other taxes.
  - Significant negative impact on miscellaneous current revenue (possible explanation: residential investment raises GDP without stimulating the base for miscellaneous current revenues, lowering the ratio).
- Negative of net factor income from abroad/GDP (magnitude of net factor income outflows):
  - Significant negative impact on PIT, CIT, and miscellaneous current revenue. Interpretation: growth in multinationals’ activities raised GDP faster than the bases for these revenues.
- Exports/GDP ratio:
  - Significant positive impact on CIT/GDP.
- House price index/GDP ratio:
  - Significant positive impact on stamp duty/GDP.
  - Significant negative impact on capital revenues/GDP (increases in house prices stimulate GDP more than the base for capital revenues).
- Equity price index/GDP ratio:
  - Significant positive impact on PIT and excise tax/GDP.
  - Significant negative impact on other taxes, social contributions, and miscellaneous current revenues/GDP (increases in equity prices stimulate GDP more than the bases for these revenue types).

### Calculating structural revenues
- Structural revenue/GDP ratio for each revenue type requires estimating how far each explanatory variable is from its structural (fundamental) value.
- Structural values for explanatory variables obtained using HP filter (trend extraction), with projections up to 2014 to minimize end-point problems.
- HP-filter extraction satisfactory for most variables except residential investment, where the housing bubble was largely ascribed to trend by the HP filter.
- For residential investment/GDP the paper decomposes:
  - Residential investment/GDP = (price index for residential construction / GDP per capita) × (real residential construction / population) × population / GDP (equation (6) and text description).
- Authors assume residential investment has a stable long-run relationship with nominal GDP; deviations reflect cyclical or bubble factors that return to fundamentals over the long run.
- Alternative treatment for residential investment:
  - Use projected path where nominal residential investment declines about 42 percent in 2009 (author notes this is slightly optimistic versus first three quarters of 2009 decline of 48 percent), a further contraction of about 30 percent in 2010, and positive growth thereafter similar to overall nominal GDP growth.
  - The projected 2014 residential investment/GDP is taken as the structural value (implied fundamental value similar to early 1990s).
- For all other explanatory variables the HP-filtered trends are used as structural values.
- Overall structural revenue is the sum of structural levels of different revenue types.
- Figure 5 (described) indicates largest divergences between actual and structural revenues (percent of GDP) are for VAT, capital taxes, stamp duty, and social contributions—revenue types most affected by residential investment effects and the property bubble.

### Structural expenditures and structural balance
- Assumption: unemployment benefits are the only significant cyclical component of fiscal expenditure.
  - Total expenditure = expenditure excluding unemployment benefits + unemployment benefits (equation (7)).
  - Unemployment benefits = unemployment rate × labor force × average benefits (equation (8)).
  - Structural unemployment benefits = NAIRU × labor force × average benefits (equation (9)).
  - Structural expenditure = expenditure excluding unemployment benefits + structural unemployment benefits (equation (10)).
- Structural expenditure divided by potential GDP (potential GDP obtained using Athanasopoulou approach) to compute structural balance: structural revenue/GDP ratio minus structural expenditure/potential GDP ratio.

- Table 3: Structural Fiscal Position (in percent of potential GDP unless otherwise stated)
  - Structural revenue (2000–2008): 33.5, 31.5, 31.3, 30.0, 31.6, 31.2, 30.9, 31.6, 29.7
  - Structural expenditure (2000–2008): 32.7, 33.8, 33.8, 32.9, 33.0, 33.5, 34.8, 38.6, 42.7
  - Structural balance (2000–2008): 0.8, -2.3, -2.5, -2.9, -1.4, -2.3, -3.9, -7.0, -13.0
- Memo items (Table 3):
  - Structural balance under alternative approaches (production function approach): 4.3, 0.9, -0.4, 0.5, 1.6, 1.5, 2.4, -1.3, -6.7
  - Using Athanasopoulou output gap (alternative): 2/2.5, -0.5, -1.5, -0.1, 1.2, 1.1, 1.5, -3.0, -8.7
  - Headline revenue (%GDP): 35.4, 33.4, 32.4, 32.8, 34.2, 34.6, 36.4, 35.8, 34.1
  - Headline expenditure (%GDP): 30.6, 32.5, 32.7, 32.4, 32.8, 33.0, 33.5, 35.7, 41.2
  - Headline balance (%GDP): 4.8, 0.9, -0.3, 0.4, 1.4, 1.6, 2.9, 0.1, -7.2

- Key interpretation:
  - The structural fiscal deficit was significantly higher than previously estimated: increased to 7 percent of potential GDP in 2007 and to 13 percent of potential GDP in 2008.
  - Most structural deterioration occurred on the expenditure side rather than the revenue side: between 2000 and 2008 structural expenditures increased by 10 percentage points of potential GDP, compared with a 3.8 percentage point decline in structural revenues.
  - Structural expenditures rose by almost 8 percentage points of potential GDP between 2006 and 2008.

### Conclusions and policy implications
- The paper extends standard structural-balance estimation for Ireland by explicitly accounting for asset price cycles, the housing bubble, and changes in national income composition.
- Asset price cycles and the housing bubble had significant impacts on Irish fiscal revenues, especially on indirect taxes like VAT, capital gains, and stamp duty; equity prices significantly impact personal income taxes.
- Changes in the sectoral composition of GDP materially affected the evolution of structural revenues.
- Not accounting for asset price cycles, bubbles, or sectoral changes creates a large bias in structural-balance calculations that cannot generally be corrected by altering the output gap estimate alone; in Ireland’s case this bias made revenue prospects appear rosier and helped prompt an expenditure relaxation that created a large fiscal gap.
- Recommendation/implication: expand standard OECD-based methodology to incorporate asset price cycles and sectoral composition effects (the paper advocates the approach used here). The ECB is noted as already moving to refine the treatment of asset-based taxes in a disaggregated approach.

*Source: Author's calculations and content from the provided IMF working paper excerpt.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1057.pdf_