## _wp0565

## Source details

**Canonical URL:** [_wp0565](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp0565.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp0565.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp0565.pdf.json)

---

### Introduction and context
- Slovakia ranked first among Central and Eastern European countries in an IMD (2004) competitiveness ranking.
- The World Bank classified Slovakia as the world’s top reformer of its investment climate over the past 12 months and as one of the 20 most attractive countries in the world for doing business (World Bank, 2004).
- Exchange rate and inflation context:
  - CPI-based real exchange rate appreciated by an average annual rate of 7½ percent during 2000–04.
  - CPI-based real exchange rate appreciated by almost 10 percent per year during 2002–04.
  - In nominal terms, the koruna appreciated by almost 7 percent against the euro in 2004, and by almost 2½ percent in the first two months of 2005.
- Policy trade-off:
  - Trade-off between limiting nominal appreciation and reducing inflation in the run-up to Slovakia’s planned euro adoption on January 1, 2009.
  - Risk of missing the Maastricht inflation criterion of 3 percent in 2007 if resisting nominal appreciation generates additional inflation.

### Four main findings (paper-level summary)
- Wages: Slovak wages remain relatively low compared with other new EU member states, even after adjusting for productivity differences.
- Prices and undervaluation: Until recently, Slovak prices remained relatively low compared to expected levels given relative income and productivity, implying real exchange rate undervaluation.
- Adjustment drivers: Administered price adjustments, indirect tax increases, and rapid nominal appreciation have eliminated most or all of the undervaluation, with a risk of overvaluation in 2005.
- Role of fiscal policy:
  - Estimated equilibrium rate of real appreciation during 2005−09 is close to 3 percent per year in the absence of fiscal consolidation.
  - If the share of government consumption in GDP declines in line with the authorities’ medium-term fiscal objectives, the estimated equilibrium real appreciation is significantly lower.

### Wages and unit labor costs (ULC)
- Cross-country wage comparisons (2003):
  - Latvia and Lithuania: 83 and 94 percent of Slovak wages, respectively.
  - Neighboring Visegrad countries (Poland, the Czech Republic, Hungary): wages as much as 50 percent higher than Slovakia.
  - Slovenia: wages as high as three times the Slovak level.
- Productivity-adjusted wages:
  - Slovakia’s data point lies slightly below the estimated log-linear relationship between Ln (Gross Wages) and Ln (GDP per Worker), indicating wages are low even given labor productivity.
- ULC dynamics:
  - Economywide ULC defined as nominal wage costs per koruna worth of value added.
  - Nominal wages grew broadly in line with nominal productivity between end-2001 and mid-2003; nominal ULC growth was close to zero in that period.
  - Between mid-2003 and mid-2004, nominal productivity growth exceeded nominal wage growth, implying nominal ULC fell and profitability increased.
- Manufacturing ULC:
  - Normalized to 100 in 2000, Slovakia’s ULC growth in manufacturing was initially slower than in the Czech Republic and Hungary (2000–2003) but caught up since 2003.
  - For 2000-04 as a whole, Slovakia, the Czech Republic, and Hungary remained more or less equally competitive, while all lost some competitiveness to Poland.
- Kotian estimates:
  - Slovak ULC estimated at about a quarter of the EU-15 average, and at about half of ULC levels in the Czech Republic or Hungary.
  - Kotian (2005) estimates that Slovak ULC would not be competitive if it were to reach 45 percent of the EU-15 average.

### Prices, PPP, and Balassa–Samuelson considerations
- Relative price levels:
  - Slovak prices about 47 percent of euro area prices in 2003 and around 50 percent in 2004.
- Nontradables and PPP:
  - PPP is unlikely to hold quickly because nontradables and non-market/government-subsidized services often have prices well below cost.
- Balassa–Samuelson mechanism and proxies:
  - Lower productivity in tradables relative to nontradables leads to lower wages and lower nontradable prices; tradables prices are internationally determined.
  - Empirical practice often uses overall productivity differential or income differential as proxies for the relative productivity differential.
  - Coudert and Couharde (2003) estimate elasticity of 0.25: every 1 percent increase in income per capita relative to euro area income per capita is associated with a real appreciation of 0.25 percent.
- Slovakia’s cross-country position:
  - Slovakia’s relative price level lies below the fitted cross-country relationship, indicating prices were low even given low relative income (2003).

### Equilibrium real appreciation — Cross-section and time-series evidence
- Cross-section evidence (sample of 120 countries):
  - Implied “initial undervaluation” in 1993 of more than 40 percent for the koruna.
  - Distance to equilibrium declined slowly through late 1990s/early 2000s, then declined substantially beginning in 2003 due to nominal appreciation and CPI inflation.
  - On trends at the time of the study, remaining undervaluation expected to disappear in 2005 and may turn into overvaluation.
- Time-series measures used:
  - CPI-based (headline CPI differential).
  - Net-CPI-based (excludes administered prices, indirect taxes, and food for Slovakia; euro area core excludes energy, food, alcohol, tobacco).
  - PPI-based (manufacturing PPI) and ULC-based (unit labor cost differential).
- Administered prices and indirect taxes:
  - Administered prices grew by 10.5 percent in January 2004 and by 4.5 percent in January 2005 (Statistical Office of the Slovak Republic).
  - Correcting for these effects by using net inflation reduces overestimation of equilibrium appreciation.
- Empirical patterns:
  - Net-CPI-based real appreciation grew broadly in line with the productivity differential, except mid-1998 through mid-1999 when the nominal koruna depreciated after a regime change.
  - PPI-based and ULC-based real exchange rates also grew broadly in line with productivity differential.
  - PPI appreciation may reflect imperfect separation of tradables/nontradables, quality improvements, and marketing/reputation improvements associated with FDI.

### CPI composition and inflation contributions
- CPI component shares (selected ranges and example year 2004):
  - Administered prices accounted for approximately 14.0–20.8 percent of CPI in sample years (19.9 percent in 2004).
  - Net inflation accounted for approximately 57.9–64.5 percent of CPI in sample years (58.7 percent in 2004).
  - Tradables comprised approximately 49.7–39.5 percent of total CPI across years and constituted about 67.5–77.1 percent of net inflation.
  - Nontradables (market services) were approximately 14.8–19.1 percent of total CPI across years.
- Tradables vs. nontradables inflation:
  - Nontradables prices generally increased faster than tradables prices (consistent with Balassa–Samuelson).
  - Tradables prices also grew, but in 2003–2004 tradables prices declined in response to nominal koruna appreciation.
  - Because tradables make up a large share of CPI, CPI-based appreciation is driven by Balassa–Samuelson effects, administered price adjustments, indirect taxes, and increases in value added of tradables.

### Econometric methodology and cointegration results
- Methodology:
  - Behavioral Equilibrium Exchange Rate (BEER) model estimating long-run relationships between real exchange rate measures and fundamentals.
  - Fundamentals included productivity differential (Slovakia relative to euro area) and government consumption (share/growth), under the assumption government spending is biased toward nontradables and causes real appreciation.
- Unit root and cointegration:
  - All variables tested (real exchange rates deflated by headline CPI, net CPI, ULC, and PPI differentials; productivity differential; government consumption percent of GDP) are nonstationary; cointegration analysis is legitimate.
  - Johansen tests provide evidence that a unique cointegrating vector exists between the three variables for all RER measures.
  - Best estimates typically obtained with four lags; residuals for three- or four-lag regressions are well-behaved.
- Estimated cointegrating relationships (four-lag specification; real exchange rate coefficient normalized to one):
  - ln (CPI-based RER) = 1.77 ln (productivity differential) + 0.53 ln (government consumption)
    - Standard errors: (0.12) (0.21)
  - ln (net CPI-based RER) = 0.93 ln (productivity differential) + 0.45 ln (government consumption)
    - Standard errors: (0.08) (0.15)
  - ln (ULC-based RER) = 1.10 ln (productivity differential) + 1.40 ln (government consumption)
    - Standard errors: (0.17) (0.30)
  - ln (PPI-based RER) = 1.04 ln (productivity differential) + 1.18 ln (government consumption)
    - Standard errors: (0.10) (0.18)
- Interpretation:
  - Productivity differential coefficient much higher for headline CPI-based measure (1.77) than for other measures, indicating failure to correct for administered price and indirect tax increases would overestimate equilibrium appreciation.
  - For net CPI-based, ULC-based, and PPI-based measures, productivity coefficient not significantly different from unity (0.93, 1.10, 1.04), implying ~1 percent real appreciation per 1 percent productivity differential increase.
  - Government consumption elasticity varies by RER deflator:
    - CPI-based: ~0.5 elasticity (0.53 CPI-based; 0.45 net CPI-based).
    - ULC-based and PPI-based: elasticities not significantly different from one (1.40 and 1.18 respectively).
  - Estimates robust to lag and sample variations; recursive estimates show coefficient stability.

### Projections (2005–2009) and scenario implications
- Projected annual average rates (Slovakia, 2005-2009):
  - Real GDP growth: 4.7
  - Employment growth: 0.9
  - Productivity growth: 3.8
  - Government Consumption-to-GDP growth: -1.9
- Euro Area projections:
  - Real GDP growth: 2.1
  - Employment growth: 1.0
  - Productivity growth: 1.1
- Derived equilibrium appreciation projections:
  - Productivity differential growth (Slovakia minus euro area): 2.7
  - Productivity differential and government consumption growth (CPI-based real appreciation): 1.7
  - Productivity differential and government consumption growth (PPI-based and ULC-based real appreciation): 0.7
- Central projection and fiscal sensitivity:
  - Central projection: productivity-driven equilibrium real appreciation may approach 3 percent per year on average during 2005-09, driven by projected productivity differential growth of 2.7 percent per year and assuming administered prices and indirect taxes no longer play an important role.
  - If government consumption remains constant in percent of GDP, equilibrium real appreciation ≈ 3 percent per year.
  - If government consumption declines as projected (share declines to meet Maastricht fiscal deficit criterion of 3 percent of GDP in 2007), equilibrium real appreciation could be reduced to:
    - 1.7 percent per year using the 0.5 elasticity estimated for the CPI-based RER.
    - 0.7 percent per year using the unit elasticity estimated for the PPI-based and ULC-based RERs.
  - Conclusion: if government consumption growth is constrained per the authorities’ medium-term objectives, average rate of equilibrium real appreciation during 2005-09 could be only about one-fourth of what it would be without fiscal consolidation.

### Key conclusions and policy implications
- Main conclusions:
  - Slovak wages remain low compared to other new EU member states, even after productivity adjustments.
  - Slovak prices have remained low relative to income, implying past undervaluation of the real exchange rate until recently.
  - Administered price adjustments, indirect tax increases, and rapid nominal appreciation have eliminated most or all of this undervaluation; risk of overvaluation in 2005.
  - Rate of equilibrium real appreciation expected to be in line with productivity growth but can be reduced with fiscal consolidation.
- Policy implications:
  - Fiscal consolidation (reducing government consumption as a share of GDP) can help reduce equilibrium real appreciation and support monetary policy in achieving disinflation without losing competitiveness.
  - Resisting nominal appreciation via monetary intervention risks higher inflation and could jeopardize compliance with the Maastricht inflation criterion ahead of euro adoption on January 1, 2009.

*Source: _wp0565 — Appendix Tables, Appendix I (IMF staff analysis and projections).*

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

### _wp0565 - References

### Figures
- 1. Monetary Policy Developments ..........................................................................................4
- 2. New EU Member States: Relative Wages, 2003  ................................................................6
- 3. New EU Member States: Wages and Productivity, 2003 ....................................................6
- 4. Wages, Productivity, and Unit Labor Costs.........................................................................7
- 5. Unit Labor Costs in Manufacturing .....................................................................................7
- 6. Slovak Price Level Relative to Euro Area Price Level........................................................8
- 7. Relative Prices and Relative Income Levels, 2003..............................................................9
- 8. Slovakia: Equilibrium and Actual Relative Price Levels ..................................................11
- 9. Slovakia: Estimated Real Exchange Rate Undervaluation ................................................12
- 10. Contributions to Headline CPI Inflation ............................................................................13
- 11. Real Exchange Rate Indices and Productivity Differential ...............................................14
- 12. Real Exchange Rate Indices and Productivity Differential  ..............................................14
- 13. Tradables and Nontradables Prices ....................................................................................16

### Tables
- 1. Slovakia: Weights of CPI Components .............................................................................16
- 2. Projected Equilibrium Real Appreciation, 2005-2009.......................................................20

### Appendices
- 1. Unit Root Tests, Cointegration Tests, and Cointegration Estimates  ................................25
- 2. Recursive Coefficient Estimates of Real Exchange Rate Determinants............................28

*Source: _wp0565 - References..............................................................................................................*

### Appendix Tables

### _wp0565 - Appendix Tables

### Introduction and context
- Slovakia was recently placed first among Central and Eastern European countries in a competitiveness ranking conducted by the Swiss Institute for Management Development (IMD, 2004).
- The World Bank classified Slovakia as the world’s top reformer of its investment climate over the past 12 months and as one of the 20 most attractive countries in the world for doing business (World Bank, 2004).
- The Slovak koruna appreciated substantially against the euro in recent years:
  - CPI-based real exchange rate appreciated by an average annual rate of 7½ percent during 2000–04.
  - CPI-based real exchange rate appreciated by almost 10 percent per year during 2002–04.
  - In nominal terms, the koruna appreciated by almost 7 percent against the euro in 2004, and by almost 2½ percent in the first two months of 2005.
- Policy trade-off and policy-relevant dates:
  - Trade-off between limiting nominal appreciation and reducing inflation, relevant because Slovakia plans to adopt the euro on January 1, 2009.
  - Risk of missing the Maastricht inflation criterion of 3 percent in 2007 if resisting nominal appreciation generates additional inflation.

### Four main findings (paper-level summary)
- Wages: Slovak wages remain relatively low compared with other new EU member states, even after adjusting for productivity differences.
- Prices and undervaluation: Until recently, Slovak prices remained relatively low compared to expected levels given relative income and productivity, implying real exchange rate undervaluation.
- Adjustment drivers: Administered price adjustments, indirect tax increases, and rapid nominal appreciation have eliminated most or all of the undervaluation, with a risk of overvaluation in 2005.
- Role of fiscal policy: The estimated equilibrium exchange rate will continue to appreciate in line with productivity growth, but fiscal consolidation could mitigate this real appreciation:
  - In the absence of fiscal consolidation, the estimated equilibrium rate of real appreciation during 2005−09 is close to 3 percent per year.
  - If the share of government consumption in GDP declines in line with the authorities’ medium-term fiscal objectives, the estimated equilibrium real appreciation is significantly lower.

### Wages and unit labor costs (ULC)
- Cross-country wage comparisons (2003):
  - Latvia and Lithuania: 83 and 94 percent of Slovak wages, respectively.
  - Neighboring Visegrad countries (Poland, the Czech Republic, Hungary): wages as much as 50 percent higher than Slovakia.
  - Slovenia: wages as high as three times the Slovak level.
- Productivity-adjusted wages:
  - Slovakia’s data point lies slightly below the estimated log-linear relationship between Ln (Gross Wages) and Ln (GDP per Worker), suggesting wages are low even given labor productivity.
- Unit labor costs and trends:
  - Economywide ULC defined as nominal wage costs per koruna worth of value added.
  - Nominal wages grew broadly in line with nominal productivity between end-2001 and mid-2003; nominal ULC growth was close to zero in that period.
  - Between mid-2003 and mid-2004, nominal productivity growth exceeded nominal wage growth, implying nominal ULC fell and profitability increased.
  - ULC in manufacturing:
    - Normalized to 100 in 2000, Slovakia’s ULC growth in manufacturing was initially slower than in the Czech Republic and Hungary (2000–2003), implying increased competitiveness, but since 2003 Slovak ULC growth caught up with the others.
    - For the period 2000-04 as a whole, Slovakia, the Czech Republic, and Hungary remained more or less equally competitive, while all lost some competitiveness to Poland.
  - Kotian estimates (SLovak ULC context):
    - Slovak ULC estimated at about a quarter of the EU-15 average, and at about half of ULC levels in the Czech Republic or Hungary.
    - Kotian (2005) estimates that Slovak ULC would not be competitive if it were to reach 45 percent of the EU-15 average.

### Prices, PPP, and Balassa–Samuelson considerations
- Relative price level (Slovakia vs. euro area):
  - Slovak prices about 47 percent of euro area prices in 2003 and around 50 percent in 2004.
- Law of one price and nontradables:
  - PPP is unlikely to hold quickly because nontradables (e.g., haircuts, restaurant meals) do not equalize in price internationally.
  - Non-market/govt-subsidized services (education, health care, public housing, utilities) often have prices well below cost, affecting PPP measures.
- Balassa–Samuelson hypothesis:
  - Lower productivity in tradables relative to nontradables leads to lower wages and hence lower nontradable prices; tradables prices are determined internationally and thus less affected.
  - Empirical practice often uses overall productivity differential or income differential as proxies for the relative productivity differential.
- Cross-country relationship and Slovakia’s position:
  - Coudert and Couharde (2003) estimate an elasticity of 0.25: every 1 percent increase in income per capita relative to euro area income per capita is associated with a real appreciation of 0.25 percent.
  - Slovakia’s relative price level lies below the fitted cross-country relationship, indicating relative price level was low even given low relative income (2003).

### Equilibrium real appreciation — Cross-section estimates
- Cross-section evidence (sample of 120 countries) suggests:
  - An “initial undervaluation” in 1993 of more than 40 percent for the koruna.
  - Distance to equilibrium declined slowly through late 1990s and early 2000s, then declined substantially beginning in 2003 due to nominal appreciation and CPI inflation.
  - On current trends at the time of the study, remaining undervaluation was expected to disappear in 2005 and may turn into overvaluation.

### Equilibrium real appreciation — Time-series estimates and measures
- Three real exchange rate measures used in time-series estimation:
  - CPI-based (headline CPI differential).
  - Net-CPI-based (excludes administered prices, indirect taxes, and food for Slovakia; core inflation for euro area excludes energy, food, alcohol, tobacco).
  - PPI-based (manufacturing PPI) and ULC-based (unit labor cost differential).
- Correcting for initial undervaluation:
  - A substantial part of CPI-based real appreciation reflected catch-up from initial undervaluation driven by administered price and indirect tax increases.
  - Adjusting for administered price and indirect tax increases by using net inflation reduces overestimation of equilibrium appreciation.
  - Administered prices growth: example figures—administered prices grew by 10.5 percent in January 2004 and by 4.5 percent in January 2005 (Statistical Office of the Slovak Republic).
- Empirical patterns:
  - Net-CPI-based real appreciation grew broadly in line with the productivity differential, except for mid-1998 through mid-1999 when nominal koruna depreciated following regime change.
  - PPI-manufacturing and ULC-based real exchange rates also grew broadly in line with the productivity differential.
  - PPI-based appreciation cannot be explained exclusively by Balassa–Samuelson (which affects nontradables), but may reflect:
    - Imperfect separation of tradables/nontradables in the PPI.
    - Quality improvements in domestically produced tradables not fully captured by PPI basket revisions.
    - Improvements in marketing/reputation of tradables (often associated with FDI and productivity growth).
- CPI composition and contributions:
  - Between 1993 and 2004 total CPI = 100 percent.
  - Administered prices accounted for approximately 14.0–20.8 percent of CPI in sample years (e.g., 19.9 percent in 2004).
  - Net inflation accounted for approximately 57.9–64.5 percent of CPI in sample years (e.g., 58.7 percent in 2004).
  - Tradables comprised approximately 49.7–39.5 percent of total CPI across years and constituted about 67.5–77.1 percent of net inflation.
  - Nontradables (market services) were approximately 14.8–19.1 percent of total CPI across years.
- Tradables versus nontradables inflation:
  - Nontradables prices generally increased faster than tradables prices (consistent with Balassa–Samuelson), but tradables prices also grew, and in 2003–2004 tradables prices declined in response to nominal koruna appreciation.
  - Because tradables make up a large share of CPI, CPI-based appreciation is driven by multiple factors: Balassa–Samuelson effect, administered price adjustments, indirect taxes, and increases in value added of tradables.

### Econometric methodology and core estimation result
- Methodology:
  - Adopted a Behavioral Equilibrium Exchange Rate (BEER) model: estimate the long-run statistical relationship between the real exchange rate and fundamentals.
  - Fundamentals included:
    - Productivity differential (Slovakia relative to euro area).
    - Government consumption (share/growth) as a determinant, under the assumption government spending is biased toward nontradables and thus tends to cause real appreciation.
- Key econometric result reported in the text:
  - All measures of real appreciation in Slovakia have been cointegrated with both productivity growth and the growth in government consumption.

### Policy implications and conclusions (extracted from findings)
- Fiscal policy can support monetary policy in limiting real appreciation:
  - Fiscal consolidation (decline in share of government consumption in GDP in line with medium-term objectives) could significantly reduce equilibrium real appreciation compared with a no-consolidation scenario.
- Exchange rate management considerations:
  - Resisting nominal appreciation via monetary intervention risks higher inflation, which could jeopardize Maastricht inflation criterion compliance ahead of euro adoption on January 1, 2009.
  - Given projected equilibrium appreciation close to 3 percent per year without fiscal consolidation over 2005−09, policy coordination (fiscal consolidation + monetary stance) can ease the trade-off between nominal appreciation and inflation control.

*Source: _wp0565 - Appendix Tables (excerpts provided from the IMF working paper content).*

### Appendix I reports evidence of stable cointegration relationships between

### _wp0565 - Appendix I reports evidence of stable cointegration relationships between

### Cointegration evidence and data properties
- All variables (the real exchange rate deflated by headline CPI, net CPI, ULC, and PPI differentials; the productivity differential; and government consumption in percent of GDP) are nonstationary, implying it is legitimate to search for a cointegration relationship.
- Johansen cointegration tests (Table A2) provide evidence that a unique cointegrating vector exists between the three variables for all different measures of the real exchange rate.
- Detailed estimates of the cointegrating vectors are reported in Table A3, with the coefficient for the real exchange rate normalized to one.
- Best estimates—in terms of minimizing the information criteria—are generally obtained when four lags are included; residuals for regressions with three or four lags are well-behaved. Two-lag specifications show weaker reliability due to heteroskedasticity test outcomes.

### Estimated cointegrating relationships (four-lag specification)
- ln (CPI-based RER) = 1.77 ln (productivity differential) + 0.53 ln (government consumption)
  - Standard errors: (0.12) (0.21)
- ln (net CPI-based RER) = 0.93 ln (productivity differential) + 0.45 ln (government consumption)
  - Standard errors: (0.08) (0.15)
- ln (ULC-based RER) = 1.10 ln (productivity differential) + 1.40 ln (government consumption)
  - Standard errors: (0.17) (0.30)
- ln (PPI-based RER) = 1.04 ln (productivity differential) + 1.18 ln (government consumption)
  - Standard errors: (0.10) (0.18)

### Interpretation of coefficients and robustness
- Productivity differential:
  - The estimated coefficient for the productivity differential is much higher for the (headline) CPI-based measure (1.77) than for the other three measures, indicating that failure to correct for administered price and indirect tax increases would lead to an overestimate of equilibrium real appreciation.
  - For the net CPI-based, ULC-based, and PPI-based measures, the estimated coefficient for the productivity differential is not significantly different from unity (0.93, 1.10, 1.04 respectively), implying that for every 1 percent increase in the productivity differential, the real exchange rate appreciates by approximately 1 percent.
- Government consumption elasticity:
  - Elasticity with respect to government consumption depends on the RER deflator.
  - ULC-based and PPI-based measures: elasticity not significantly different from one (1.40 and 1.18 respectively).
  - CPI-based measures: elasticity significantly smaller—approximately 0.5 (0.53 for CPI-based; 0.45 for net CPI-based).
  - Possible explanations: government consumption includes civil servant salaries that affect economywide unit labor costs (ULC effect). No clear explanation provided for the strong PPI-based effect.
- Robustness checks:
  - Estimates robust to variations in the number of lags and in the sample period.
  - Recursive estimation (shrinking the sample one observation at a time) shows coefficients are very stable over time, with confidence intervals narrowing as sample size increases.

### Projections (2005–2009) and scenario implications
- Projected annual average rates (Slovakia, 2005-2009):
  - Real GDP growth: 4.7
  - Employment growth: 0.9
  - Productivity growth: 3.8
  - Government Consumption-to-GDP growth: -1.9
- Euro Area projections:
  - Real GDP growth: 2.1
  - Employment growth: 1.0
  - Productivity growth: 1.1
- Derived equilibrium real appreciation projections:
  - Productivity differential growth (Slovakia minus euro area): 2.7
  - Productivity differential and government consumption growth (CPI-based real appreciation): 1.7
  - Productivity differential and government consumption growth (PPI-based and ULC-based real appreciation): 0.7
- Central projection: productivity-driven equilibrium real appreciation may approach 3 percent per year on average during the next five years, driven by projected productivity differential growth of 2.7 percent per year during 2005-09 and assuming administered prices and indirect taxes no longer play an important role.
- Fiscal scenario sensitivity:
  - If government consumption remains constant in percent of GDP, equilibrium real appreciation ≈ 3 percent per year.
  - If government consumption declines as projected (share of government consumption gradually declines to meet Maastricht fiscal deficit criterion of 3 percent of GDP in 2007), equilibrium real appreciation could be reduced to:
    - 1.7 percent per year using the 0.5 elasticity estimated for the CPI-based RER.
    - 0.7 percent per year using the unit elasticity estimated for the PPI-based and ULC-based RERs.
  - Conclusion from scenarios: if government consumption growth is constrained per the authorities’ medium-term objectives, the average rate of equilibrium real appreciation during 2005-09 could be only about one-fourth of what it would be without fiscal consolidation.

### Key conclusions and policy implications
- Main conclusions:
  - Slovak wages remain low compared to other new EU member states, even after productivity adjustments.
  - Slovak prices have remained low relative to income, implying past undervaluation of the real exchange rate until recently.
  - Administered price adjustments, indirect tax increases, and rapid nominal appreciation have eliminated most or all of this undervaluation; there is even a risk of overvaluation in 2005.
  - The rate of equilibrium real appreciation is expected to be in line with productivity growth but can be reduced with fiscal consolidation.
- Policy implication:
  - By reducing government consumption as a share of GDP (fiscal consolidation), fiscal policy can help reduce equilibrium real appreciation and thereby support monetary policy in achieving disinflation without losing competitiveness.

*Source: Appendix I, _wp0565 (IMF staff analysis and projections).*

### Appendix I. Unit Root Tests, Cointegration Tests, and Cointegration Estimates

### Appendix I. Unit Root Tests, Cointegration Tests, and Cointegration Estimates

### ADF Unit Root Tests (Table A1)
- Variables tested (periods indicated in source): 1996:Q1-2004:Q2 and 1996:Q2-2004:Q2.
- Regression specifications:
  - Levels tests: "The regression includes a constant, a trend, and seasonal dummies. The critical values of the Augmented Dickey-Fuller (ADF) t-statistics are -3.55 for the 5 percent level and -4.25 for the 1 percent level. The null hypothesis is that of a unit root in levels, i.e., rejection of the null means that the variable is stationary in levels. The symbols * and ** denote significance at the 5 percent and 1 percent level, respectively."
  - Differences tests: "The regression includes a constant and seasonal dummies. The critical values of the ADF t-statistics are -2.95 for the 5 percent level and -3.64 for the 1 percent level. The null hypothesis is that of a unit root in differences, i.e., rejection of the null means that the variable is stationary in differences."
- Reported ADF results (as presented in the source; rows correspond to lags 3, 2, 1, 0 for each variable in sequence):
  - Header text: lagt-adfbetalagt-adfbeta
  - Rows (raw values as in source):
    - 3 -2.1380.6453 -3.768** -0.224
    - 2 -2.9370.5942 -3.815** 0.001
    - 1 -2.5790.6851 -3.155* 0.266
    - 0 -1.7430.7910 -4.058** 0.271
    - 3 -0.6070.9563 -3.845** -0.288
    - 2 -0.9000.9352 -4.017** -0.097
    - 1 -0.9610.9331 -3.777** 0.132
    - 0 -0.7010.9520 -4.758** 0.180
    - 3 -2.0290.7403 -2.971* -0.010
    - 2 -2.0470.7582 -2.896 0.137
    - 1 -1.8390.7921 -3.241* 0.186
    - 0 -1.4750.8390 -4.700** 0.130
    - 3 -1.7520.7483 -3.617* -0.185
    - 2 -2.4150.6992 -3.604* 0.029
    - 1 -2.0350.7681 -3.078* 0.261
    - 0 -1.4570.8380 -4.222** 0.218
    - 3 -2.5490.1853 -3.461* -1.141
    - 2 -2.9150.1892 -4.677** -1.066
    - 1 -3.1880.2561 -5.355** -0.671
    - 0 -4.128* 0.2490 -7.962** -0.378
    - 3 -3.412 -0.0143 -5.355** -1.223
    - 2 -3.2810.0812 -5.991** -0.944
    - 1 -4.249* -0.0091 -9.707** -0.891
    - 0 -6.004** -0.0460 -14.41** -0.497
  - Variables listed in source (in order shown): ln (CPI-based RER); ∆ln (CPI-based RER); ln (PPI-based RER); ∆ln (PPI-based RER); ln (government consumption); ∆ln (government consumption); ln (productivity differential); ∆ln (productivity differential); ln (net CPI-based RER); ∆ln (net CPI-based RER); ln (ULC-based RER); ∆ln (ULC-based RER).

### Johansen Cointegration Tests (Table A2)
- Test period and controls: "The tests are conducted for the period 1996:Q2-2004:Q2, and include four lags, a constant, three seasonal quarterly dummies, a dummy for the change in exchange rate regime (1998:Q3-1999:Q3), and a dummy for the summer 2002 depreciation (2002:Q2-Q3), which was related to political concerns about the outcome of the September 2002 elections."
- Notes on statistics: "The statistics λtrace and λmax are Johansen's trace eigenvalue and maximal eigenvalue statistics. The statistics λ'trace and λ'max incorporate a degrees-of-freedom correction, and a small sample adjustment. The null hypotheses are whether the cointegration rank r equals zero (no cointegration), is less than or equal to one (at most one cointegrating vector), or is less than or equal to two (at most two cointegrating vectors). The symbols * and ** denote significance at the 5 percent and 1 percent level, respectively."
- Reported cointegration results (raw blocks as in source):
  - Block 1:
    - rank r=0 45.90 [0.000]** 37.28 [0.000]** 29.21 [0.059] 23.73 [0.019]*
    - r ≤ 1 8.61 [0.409] 8.43 [0.344] 5.48 [0.756] 5.37 [0.698]
    - r ≤ 2 0 20.18 [0.672] 0.18 [0.672] 0.11 [0.736] 0.11 [0.736]
  - Block 2:
    - rank r=0 46.73 [0.000]** 40.51 [0.000]** 29.74 [0.051] 25.78 [0.008]**
    - r ≤ 1 16.22 [0.674] 6.12 [0.604] 3.96 [0.900] 3.89 [0.863]
    - r ≤ 2 20.10 [0.751] 0.10 [0.751] 0.06 [0.800] 0.06 [0.800]
  - Block 3:
    - rank r=0 33.27 [0.018]* 26.87 [0.005]** 21.17 [0.358] 17.10 [0.173]
    - r ≤ 1 16.40 [0.653] 5.92 [0.629] 4.07 [0.892] 3.77 [0.875]
    - r ≤ 2 20.48 [0.490] 0.48 [0.490] 0.30 [0.582] 0.30 [0.582]
  - Block 4:
    - rank r=0 52.89 [0.000]** 45.08 [0.000]** 34.23 [0.014]* 29.17 [0.002]**
    - r ≤ 1 17.81 [0.493] 7.62 [0.427] 5.05 [0.801] 4.93 [0.750]
    - r ≤ 2 20.19 [0.663] 0.19 [0.663] 0.12 [0.726] 0.12 [0.726]
- CPI-based RER, Net CPI-based RER, ULC-based RER, and PPI-based RER are the cointegration sets reported.

### Estimated Cointegration Equations (Table A3)
- Sample and controls: "PPI-based RER (1996:Q2-2004:Q2). All regressions include a constant, three seasonal quarterly dummies, a dummy for the change in exchange rate regime (1998:Q3-1999:Q3), and a dummy for the summer 2002 depreciation (2002:Q2-Q3), which was related to political concerns about the outcome of the September 2002 elections. The symbols * and ** denote significance at the 5 percent and 1 percent level, respectively."
- Reported coefficient sequences and associated statistics (as presented in the source):
  - Dependent variable: (columns headed by) Number of lags 4 3 2 4 3 2 4 3 2 4 3 2
  - Ln (prod diff):
    - Coefficients: 1.77 1.74 1.78 0.93 0.86 0.94 1.10 1.06 1.33 1.04 0.93 1.08
    - Standard errors row (aligned under above): (0.12) (0.11) (0.12) (0.08) (0.09) (0.10) (0.17) (0.20) (0.25) (0.10) (0.10) (0.13)
  - Ln (govt cons):
    - Coefficients: 0.53 0.50 0.51 0.45 0.29 0.43 1.40 1.35 1.92 1.18 0.89 1.20
    - Standard errors row (aligned under above): (0.21) (0.18) (0.19) (0.15) (0.14) (0.16) (0.30) (0.33) (0.37) (0.18) (0.17) (0.19)
  - Diagnostic statistics (vectors reported across the lags):
    - Vector normality: 0.66 0.62 0.06 0.10 0.19 0.19 0.51 0.76 0.17 0.47 0.43 0.06
    - Vector heteroskedasticity: 0.95 0.79 0.03* 0.91 0.83 0.05 0.92 0.78 0.24 0.96 0.82 0.06
  - Log likelihoods (per column): 294 285 266 305 294 273 294 281 265 307 291 272
  - Information criteria 2/ (AIC, HQ, SC) reported (per column):
    - AIC: -14.8 -14.8 -14.2 -15.5 -15.3 -14.6 -14.8 -14.5 -14.1 -15.6 -15.2 -14.5
    - HQ: -14.0 -14.2 -13.7 -14.7 -14.7 -14.1 -14.0 -13.9 -13.6 -14.8 -14.5 -14.1
    - SC: -12.5 -12.9 -12.7 -13.2 -13.5 -13.1 -12.5 -12.7 -12.6 -13.3 -13.3 -13.1
- Note on information criteria: "The three information criteria reported are the Akaike information criterion (AIC), the Hannan-Quinn criterion (HQ), and the Schwartz criterion (SC). In all cases, the log-likelihood constant is included."

*Source: Appendix I. Unit Root Tests, Cointegration Tests, and Cointegration Estimates.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp0565.pdf_