## _wp0402

## Source details

**Canonical URL:** [_wp0402](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp0402.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp0402.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp0402.pdf.json)

---

### I. Introduction and research question
- Focus: costs (or benefits) of giving up monetary independence and, in particular, the flexible nominal exchange rate as a stabilization tool for five central European countries (CECs): the Czech Republic, Hungary, Poland, the Slovak Republic, and Slovenia.
- Key conceptual trade-off: under high capital mobility and limited independent interest rate scope, the main question centers on the loss of the flexible exchange rate as a shock absorber.
- Usefulness of flexible exchange rates depends on shock type:
  - Real shocks: flexible exchange rates can act as useful absorbers by generating rapid adjustment in international relative prices when domestic prices adjust slowly (Mundell, 1964).
  - Monetary/financial shocks: exchange rate adjustment can be destabilizing—e.g., a negative financial shock that raises interest rates would appreciate the exchange rate and amplify output losses.
- New Open Economy Macroeconomics caveat: if pass-through from exchange rate to import prices is very small, exchange rates may be of little use even for real shocks (Engel (2002), Obstfeld (2001), Obstfeld (2002)), although empirical evidence remains supportive of exchange-rate effects on relative prices (Obstfeld (2001) and (2002)).

### II. Literature context and empirical approach
- Empirical tradition: structural vector auto-regression (SVAR) framework developed by Clarida and Gali (1994) with three variables—output, prices, and the real exchange rate—to identify supply (AS), demand (IS), and monetary/financial (LM) shocks.
- Alternative simpler approach: two-variable, two-shock model (Enders and Lee).
- This paper’s methodological choices:
  - Applies Clarida and Gali SVAR and the Enders and Lee two-variable model to the CECs.
  - Focuses on the nominal exchange rate because it is the policy instrument that would be given up after euro adoption.
  - Estimates SVARs only over periods with generally flexible exchange rates, using non-multiplicative dummies for regime changes and exceptional periods.

### III. Estimation period, data, and econometric design
- Observation periods used (Table 1):
  - Czech Republic: 1996:II – 2003:II
  - Hungary: 1995:III – 2003:II
  - Poland (1): 1995:II – 2003:II
  - Poland (2): 1998:II – 2003:II
  - Slovak Republic: 1997:I – 2003:II
  - Slovenia: 1993:I – 2003:II
- Regime and event dummies (selected):
  - D1: Asian and Russian Crises — 1 from 1997:V – 1998:VIII
  - D2: Float in Czech Republic — 1 from 1997:V onwards
  - D3: Widening of bands in Hungary — 1 from 2001:VI onwards
  - D4: Float in Poland — 1 from 2000:IV onwards
  - D5: Float in Slovak Republic — 1 from 1998:X onwards
  - D6: Inflation targeting in Czech Republic — 1 from 1998:I onwards
  - D7: Inflation targeting in Poland — 1 from 1998:IX onwards
- Data:
  - Monthly data from 1993 to 2003 for bilateral nominal exchange rate against the euro (domestic currency per euro; an increase is a depreciation), industrial production (IFS), and CPI (IFS).
  - Industrial production and CPI expressed relative to the euro area; real exchange rate constructed from the bilateral nominal exchange rate and relative CPI.
- Econometric specification:
  - Models estimated in first differences; lag length uniformly chosen as six periods.
  - VARs include a constant and monthly dummies (except Slovenia where output is seasonally adjusted).
  - Stationarity and cointegration tests performed; cointegration generally not found → VAR in first differences appropriate.
  - Outputs: impulse response functions (IRFs) and forecast error variance decompositions up to 48 months.

### IV. Box 1 — Identification of shocks in the Mundell-Fleming framework
- Long-run restrictions classify structural shocks:
  - Relative AS shocks: permanent effect on relative output.
  - Relative IS shocks: permanent effect on the real exchange rate but not on relative output.
  - Relative nominal (LM) shocks: no permanent effect on relative output and the real exchange rate.
- Expected IRF patterns (short- and medium-run) if identification holds:
  - AS shock: increases relative output; eventual fall in prices; ambiguous long-run real exchange rate effect.
  - IS shock: nominal and real exchange rate appreciate in the short run (sticky prices); relative output increases temporarily; real exchange rate appreciates if shock is permanent.
  - LM shock: lowers domestic interest rate relative to foreign rates; nominal and real exchange rates depreciate in the short run; relative output increases temporarily; long-run no effect on relative output.

### V. Main empirical findings — Two-variable SVAR (nominal exchange rate and relative output)
- Conceptual point: neutral shocks have no long-run effect on relative output; non-neutral shocks have a long-run impact on relative output.
- Stationarity/cointegration: variables I(1); no cointegration → VAR in first differences appropriate.
- Variance decomposition (selected outcomes at 12 months):
  - At least three quarters of variability in the nominal exchange rate is explained by the neutral shock in most countries.
  - Exception: Poland (1998:II – 2003:II), where the neutral shock explains 53 percent of the exchange rate.
  - Variability in relative output is mostly determined by non-neutral shocks: contribution of the non-neutral shock ranges from 60 percent in Slovenia to over 90 percent in the Slovak Republic. Poland (short period) shows contribution of 52 percent.
- Representative detailed entries (from Table A2) — nominal exchange rate neutral/non-neutral at selected horizons:
  - Czech Republic nominal exchange rate at 12 months: neutral 92 non-neutral 8 (S.E. 0.022).
  - Hungary nominal exchange rate at 12 months: neutral 18 non-neutral 82 (S.E. 0.048).
  - Poland PO(1) nominal exchange rate at 12 months: neutral 79 non-neutral 21 (S.E. 0.024).
  - Poland PO(2) nominal exchange rate at 12 months: neutral 27 non-neutral 73 (S.E. 0.041).
  - Slovak Republic nominal exchange rate at 12 months: neutral 75 non-neutral 25 (S.E. 0.018).
  - Slovenia nominal exchange rate at 12 months: neutral 6 non-neutral 94 (S.E. 0.038).
- Interpretation: the nominal exchange rate does not generally respond to the shocks that cause most output fluctuations → evidence that the exchange rate has not served as an effective output absorber in these cases.
- IRFs: ambiguous for distinguishing LM vs. IS within neutral shocks when the aggregate effect of the neutral shock on output is very small.

### VI. Main empirical findings — Three-variable SVAR (LM, IS, AS)
- Model adds real exchange rate and imposes long-run restrictions to separate LM, IS, and AS shocks.
- Stationarity/cointegration: real exchange rate generally I(1) (Hungary exception); cointegration tests do not reject unit root → VAR in first differences used.
- IRF patterns (consistent with theoretical priors except for Hungary):
  - IS shock: nominal exchange rate appreciates; prices rise so real exchange rate appreciates even more; output temporarily increases. Exchange rate movement dampens IS shock impact on output. Hungary exception: positive IS shock leads to a depreciation.
  - LM shock: nominal exchange rate depreciates permanently; output increases temporarily; real exchange rate depreciates in the short run, then returns as prices adjust. Hungary: LM shock difficult to quantify. Slovenia: output temporarily decreases after LM shock (identification difficulty).
  - AS shock: increases output; ambiguous effects on exchange rates.
- Variance decomposition (selected outcomes at 12 months, Table A4):
  - Relative output: mainly driven by AS shocks — contribution over 60 percent in the Czech Republic, Hungary, and Slovenia; more than 90 percent in the Slovak Republic. Poland (long period) shows greater LM/IS contributions; Poland (short period) stands out with a relatively small AS contribution.
  - Nominal exchange rate: predominantly explained by a combination of LM and IS shocks. Except Hungary, the contribution of LM shocks is 50 percent or higher in all countries and reaches 80 percent in Slovenia.
  - Real exchange rate: also predominantly driven by LM and IS shocks. For all countries except Hungary, the relative contribution of LM shocks to the real exchange rate is smaller than to the nominal exchange rate — implying studies using real exchange rates may overestimate the importance of IS shocks and underestimate LM shocks.
- Representative numeric entries (selected):
  - Czech Republic real exchange rate at 12 months: money 59 demand 32 supply 9 (S.E. 0.023).
  - Czech Republic nominal exchange rate (selected formatting in source): nominal decomposition shows substantial supply percentages in the table (see source).
  - Hungary real exchange rate at 12 months: money 60 demand 29 supply 12 (S.E. 0.014).
  - Poland PO(1) real exchange rate at 48 months: money 35 demand 57 supply 8 (S.E. 0.027).
  - Poland PO(2) real exchange rate at 48 months: money 21 demand 47 supply 32 (S.E. 0.031).
  - Slovak Republic real exchange rate at 48 months: money 37 demand 48 supply 15 (S.E. 0.022).
  - Slovenia real exchange rate at 48 months: money 45 demand 43 supply 12 (S.E. 0.008).
  - Nominal exchange rate LM contribution: 50 percent or higher in all countries except Hungary; reaches 80 percent in Slovenia (aggregate statement from text).

### VII. Key numeric and procedural facts (preserved exactly)
- Data frequency and sample: monthly data from 1993 to 2003.
- VAR lag length: six periods.
- Forecast horizons for variance decompositions and IRFs: up to 48 months.
- Regime change dates and magnitudes:
  - Hungary: ± 2.25 percent bands (March 1995); widening to ±15 percent (mid-2001).
  - Poland: intervention band widened to ±10% (February 1998); inflation targeting introduced September 1998; full float April 2000.
  - Slovak Republic: band widened to ± 7 percent by early 1997; float October 1998.
- Representative variance-decomposition outcomes:
  - Poland (1998:II – 2003:II): neutral shock explains 53 percent of exchange rate variability.
  - Contribution of non-neutral shocks to relative output: ranges from 60 percent (Slovenia) to over 90 percent (Slovak Republic); Poland (short period) 52 percent.
  - LM shocks contribution to nominal exchange rate: 50 percent or higher in all countries except Hungary; reaches 80 percent in Slovenia.
  - Aggregate reported LM contribution in smaller, more open CECs: between 58 and 80 percent.

### VIII. Conclusions and policy implications
- Short-run macroeconomic responses identified by the Clarida and Gali–type SVAR generally align with Mundell-Fleming theoretical priors, supporting shock identification despite short samples and structural changes.
- Exchange rate role during periods of (relatively) flexible exchange rates in these five CECs:
  - (i) the exchange rate has responded little to the shocks that affect output; and
  - (ii) LM shocks have contributed significantly to nominal exchange rate variability, with contributions particularly high (between 58 and 80 percent) in the smaller, more open CECs.
- Overall interpretation: the nominal exchange rate appears on average to have served as much or more as an unhelpful propagator of LM shocks than as a useful absorber of IS shocks.
- Policy inference: these results cast doubt on the usefulness of exchange rate flexibility as a shock absorber in the CECs studied; the costs of losing exchange rate flexibility in the CECs are limited, if even positive.

*Source document: _wp0402 - REFERENCES (PDF)*

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

### _wp0402 - References..............................................................................................................

### I. Introduction and research question
- Focus: costs (or benefits) of giving up monetary independence and, in particular, the flexible nominal exchange rate as a stabilization tool for five central European countries (CECs): the Czech Republic, Hungary, Poland, the Slovak Republic, and Slovenia.
- Key conceptual trade-off: under high capital mobility and limited independent interest rate scope, the main question centers on the loss of the flexible exchange rate as a shock absorber.
- The usefulness of flexible exchange rates depends on the types of shocks:
  - Real shocks: flexible exchange rates can act as useful absorbers by generating rapid adjustment in international relative prices when domestic prices adjust slowly (Mundell, 1964).
  - Monetary/financial shocks: exchange rate adjustment can be destabilizing—e.g., a negative financial shock that raises interest rates would appreciate the exchange rate and amplify output losses.
- New Open Economy Macroeconomics caveat: if pass-through from exchange rate to import prices is very small, exchange rates may be of little use even for real shocks (Engel (2002), Obstfeld (2001), Obstfeld (2002)), although empirical evidence remains supportive of exchange-rate effects on relative prices (Obstfeld (2001) and (2002)).

### II. Literature context and empirical approach
- Empirical tradition: structural vector auto-regression (SVAR) framework developed by Clarida and Gali (1994) with three variables (often relative to trading partners)—output, prices, and the real exchange rate—to identify supply (AS), demand (IS), and monetary/financial (LM) shocks (see Box 1).
- Alternative simpler approach: two-variable, two-shock model (Enders and Lee).
- This paper’s methodological choices:
  - Applies Clarida and Gali SVAR and the Enders and Lee two-variable model to the CECs.
  - Focuses on the nominal exchange rate (rather than just the real exchange rate) because the nominal rate’s flexibility is what would be given up after euro adoption.
  - Estimates SVARs only over periods with generally flexible exchange rates (to avoid distortions from systematic policy intervention), using non-multiplicative dummies for regime changes and exceptional periods.

### Box 1 — Classification and identification of shocks in the Mundell-Fleming (MF) model
- Framework: Clarida and Gali (1994) derive a stochastic Obstfeld (1985) open economy macro model with endogenous output, prices, and the real exchange rate; long-run restrictions (Blanchard and Quah (1989)) identify three structural shocks: relative supply (AS), demand (IS), and monetary/financial market (LM).
- Long-run effects used for identification:
  - Relative AS shocks: permanent effect on relative output (productivity and labor market shocks).
  - Relative IS shocks: permanent effect on the real exchange rate but not on relative output.
  - Relative nominal (LM) shocks: no permanent effect on relative output and the real exchange rate.
- Short- and medium-run expected IRF patterns if identification is correct:
  - AS shock: increases relative output; short-run impact on real and nominal exchange rates ambiguous; eventual fall in prices; long-run ambiguous effect on real exchange rate.
  - IS shock: increases relative demand; short-run nominal and (due to sticky prices) real exchange rate appreciate and relative output increases; eventually relative output returns to prior level; the real exchange rate appreciates if shock is permanent.
  - LM shock: lowers domestic interest rate relative to foreign rates; short-run real and nominal exchange rates depreciate and relative output increases; long-run no effect on real exchange rate as relative output returns to prior level.
- Note: IS and LM shocks can be classified together as neutral shocks.

### III. Estimation period, data, and econometric design
- Estimation rationale: restrict samples to periods of generally flexible exchange rates to avoid bias from periods of systematic policy intervention (e.g., exchange market intervention).
- Dummies: include (non-multiplicative) dummies to account for changes in monetary and exchange rate regimes and exceptional periods (following Creel and Levasseur (2003) and Süppel (2003)); these are described in Table 2 (not reproduced here).
- Observation periods used (Table 1):
  - Czech Republic: 1996:II – 2003:II
  - Hungary: 1995:III – 2003:II
  - Poland (1): 1995:II – 2003:II
  - Poland (2): 1998:II – 2003:II
  - Slovak Republic: 1997:I – 2003:II
  - Slovenia: 1993:I – 2003:II
- Country-specific note: Czech Republic introduced exchange rate flexibility in February 1996 when a peg was replaced by a ± 7.5 percent band; dummies proposed for adoption in early 1997 of a (managed) flexible exchange rate regime and other events.

### IV. Main empirical findings and interpretation
- Two-variable (two-shock) model result:
  - If the nominal exchange rate does not respond to the shocks affecting output, it suggests the exchange rate is not useful as an output stabilizer.
  - Empirical result: output is predominantly influenced by one type of shock—the non-neutral—while the nominal exchange rate is largely determined by the other type—the neutral—raising doubts about the exchange rate’s usefulness as a shock absorber.
  - Caveat: this result is not conclusive by itself because a very effective exchange rate stabilizer could in principle shield output from neutral shocks (Canzoneri et al).
- Three-variable SVAR (IS, AS, LM) result:
  - Some exchange rate variability is attributable to IS shocks (potentially useful).
  - A larger share of exchange rate variability is attributable to LM shocks (unhelpful), with the share of LM shocks particularly high in the smaller, more open CECs.
  - Overall interpretation: judging from observed exchange rate movements, the nominal exchange rate appears on average to have served as much or more as an unhelpful propagator of LM shocks than as a useful absorber of IS shocks.
- Policy implication drawn:
  - The cost of losing the flexible nominal exchange rate as a stabilization tool in the CECs is modest, if at all positive, particularly in the smaller countries.

### V. Paper structure (as presented)
- Section II: estimation period, econometric methodology, and data.
- Section III: empirical results.
- Section IV: concluding remarks.

*Source: _wp0402 - References..............................................................................................................*

### introduction in January 1998 of inflation targeting (IT).

### _wp0402 - introduction in January 1998 of inflation targeting (IT).

### Overview and institutional/regime changes
- Hungary: narrow bands of ± 2.25 percent introduced in March 1995. Full exchange rate flexibility introduced only in mid-2001 with the widening of the exchange rate band to ±15 percent; estimation carried out from mid-1995 onwards with a dummy for the widening in mid-2001.  
- Poland: moved from a crawling peg to a crawling band regime in May 1995; estimation also performed from February 1998 onwards when the intervention band was widened to ±10% (following Süppel (2003)). Dummies proposed for the introduction of IT in September 1998 and the full float in April 2000.  
- Slovak Republic: band gradually widened during 1996 to ± 7 percent by early 1997; a dummy is proposed for the float of the koruna in October 1998.  
- Slovenia: official guidance of the exchange rate has been considerable, no significant change since 1993, and estimation begins in 1993.

### Econometric methodology
- Approach: Structural VAR (SVAR) methodology following Blanchard and Quah (1989); long-run restrictions used to identify structural shocks (neutral vs. non-neutral in two-variable model; LM, IS, and supply (AS) shocks in three-variable model). Short-run responses left unrestricted.  
- Specification:
  - Models estimated in first differences to impose long-run restrictions on levels.  
  - Stationarity and cointegration tests performed to verify appropriateness of first-differences specification.  
  - VARs include a constant term and monthly dummies to capture seasonality (monthly dummies not included for Slovenia since output data were only available on a seasonally adjusted basis).  
  - Lag length: uniformly chosen as six periods in both models based on lag length tests.  
  - Period- and regime-specific dummies tested; a dummy is maintained if significant in at least one of the three equations.
- Defined dummies (Table 2):
  - D1: Asian and Russian Crises — 1 from 1997:V – 1998:VIII
  - D2: Float in Czech Republic — 1 from 1997:V onwards
  - D3: Widening of bands in Hungary — 1 from 2001:VI onwards
  - D4: Float in Poland — 1 from 2000:IV onwards
  - D5: Float in Slovak Republic — 1 from 1998:X onwards
  - D6: Inflation targeting in Czech Republic — 1 from 1998:I onwards
  - D7: Inflation targeting in Poland — 1 from 1998:IX onwards
- Outputs retrieved after estimation and restrictions: impulse response functions (IRFs) and forecast error variance decompositions up to 48 months.

### Data
- Monthly data from 1993 to 2003 for:
  - bilateral nominal exchange rate against the euro (from Eurostat) — units: domestic currency per euro (an increase is a depreciation of the exchange rate);
  - industrial production (from IMF’s International Financial Statistics (IFS) database);
  - CPI index (from IMF’s IFS database).
- Industrial production and CPI expressed relative to the euro area to capture asymmetric shocks.  
- Real exchange rate constructed from the bilateral nominal exchange rate and relative price level vis-à-vis the Euro area (CPI).  
- Tradeoff noted: industrial production covers only part of economic output; GDP available only quarterly would reduce observations too much.

### Results — Two-variable SVAR: exchange rate and output
- Model: first differences of nominal exchange rate (∆r_t) and relative industrial production (∆y_t); distinguishes neutral shocks (no long-run effect on relative output) and non-neutral shocks (long-run impact on output).
- Stationarity/cointegration:
  - ADF and PP tests: levels generally not stationary; first differences are stationary → variables I(1).  
  - Cointegration tests: residuals from regressing nominal exchange rate on relative output and a constant indicate that the null of a unit root cannot be rejected for all countries → variables are not cointegrated → VAR in first differences appropriate.
- Variance decomposition (Table 3) — key findings after 12 months:
  - At least three quarters of variability in the nominal exchange rate is explained by the neutral shock in most countries. Exception: Poland over 1998:II – 2003:II, where the neutral shock explains 53 percent of the exchange rate.  
  - Variability in relative output is mostly determined by non-neutral shocks: contribution of the non-neutral shock ranges from 60 percent in Slovenia to over 90 percent in the Slovak Republic. Poland (short period) exception with contribution of 52 percent.
- Interpretation: nominal exchange rate does not respond to the shocks that cause most output fluctuations — evidence that the exchange rate has not served as an absorber in these cases.
- IRFs: ambiguous for distinguishing LM vs. IS within neutral shocks because the aggregate effect of the neutral shock on output is very small in most countries; cannot determine whether depreciations in response to neutral shocks reflect positive LM shocks or negative real demand (IS) shocks.

### Results — Three-variable SVAR: LM, IS, and AS shocks
- Model extends two-variable VAR by adding the real exchange rate ∆r_t and imposes long-run restrictions to distinguish:
  - LM shock (ε^m_t): no long-run impact on real exchange rate and no long-run impact on relative output.
  - IS shock (ε^d_t): no long-run impact on relative output.
- Stationarity/cointegration:
  - Real exchange rate levels generally I(1) (except Hungary where ADF and PP reject unit root); differences stationary.  
  - Cointegration tests: ADF on residuals from regression of real exchange rate on nominal exchange rate, relative output, and a constant — null of a unit root cannot be rejected → variables not cointegrated → VAR in first differences appropriate.
- Estimation specifics: six lags, seasonal dummies, and testing of period/regime dummies. Monetary policy (D2) and inflation targeting (D6) dummies accepted for the Czech Republic and Poland in the long period.
- IRF patterns (consistent with theoretical priors except for Hungary):
  - IS shock: nominal exchange rate appreciates; prices rise so real exchange rate appreciates even more; output temporarily increases. The exchange rate movement dampens the impact of the IS shock on output. Hungary exception: positive IS shock leads to a depreciation, consistent with policy choice to offset competitiveness effects.
  - LM shock: nominal exchange rate depreciates permanently; output increases temporarily; real exchange rate depreciates in short run as prices rise slowly, then returns as prices catch up. Hungary: LM shock difficult to quantify. Slovenia: output temporarily decreases after LM shock, indicating identification difficulties.
  - AS shock: increases output and has ambiguous effects on exchange rates.
- Variance decomposition (Table 4) — after 12 months:
  - Relative output: mainly driven by AS shocks — contribution over 60 percent in the Czech Republic, Hungary, and Slovenia; more than 90 percent in the Slovak Republic. Poland (long period) shows greater LM/IS contributions; Poland in the short period stands out with a relatively small AS contribution.
  - Nominal exchange rate: predominantly explained by a combination of LM and IS shocks. Except Hungary, the contribution of LM shocks is 50 percent or higher in all countries and reaches 80 percent in Slovenia. Hungary’s results less reliable due to policy intervention affecting identification.
  - Polish nominal exchange rate: around 40 percent driven by IS shocks in the short period; supply shocks more important in Poland than in other CECs.
  - Real exchange rate: also predominantly driven by LM and IS shocks. For all countries except Hungary, the relative contribution of LM shocks to the real exchange rate is smaller than to the nominal exchange rate — implying studies using real exchange rates may overestimate the importance of IS shocks and underestimate LM shocks.

### Key numeric and procedural facts
- Data frequency and sample: monthly data from 1993 to 2003.  
- VAR lag length: six periods.  
- Forecast horizons for variance decompositions and IRFs: up to 48 months.  
- Regime change dates and magnitudes:
  - Hungary: ± 2.25 percent bands (March 1995); widening to ±15 percent (mid-2001).
  - Poland: intervention band widened to ±10% (February 1998); IT introduced September 1998; full float April 2000.
  - Slovak Republic: band widened to ± 7 percent by early 1997; float October 1998.
- Representative variance-decomposition outcomes cited:
  - Poland (1998:II – 2003:II): neutral shock explains 53 percent of exchange rate variability.
  - Contribution of non-neutral shocks to relative output: ranges from 60 percent (Slovenia) to over 90 percent (Slovak Republic); Poland (short period) 52 percent.
  - LM shocks contribution to nominal exchange rate: 50 percent or higher in all countries except Hungary; reaches 80 percent in Slovenia.
  - Overall reported LM contribution in smaller, more open CECs: between 58 and 80 percent (conclusion summary).

### Conclusions and policy implications
- Short-run macroeconomic responses identified by the Clarida and Gali–type SVAR generally align with Mundell-Fleming theoretical priors, supporting shock identification despite short samples and structural changes.  
- Exchange rate role:
  - On average during the period with (relatively) flexible exchange rates in the five CECs, (i) the exchange rate has responded little to the shocks that affect output; and (ii) LM shocks have contributed significantly to nominal exchange rate variability, with contributions particularly high (between 58 and 80 percent) in the smaller, more open CECs.  
  - The exchange rate appears on average to have served as much or more as an unhelpful propagator of LM shocks than as a useful absorber of IS shocks.
- Policy inference: these results cast doubt on the usefulness of exchange rate flexibility as a shock absorber in the CECs studied; the costs of losing exchange rate flexibility in the CECs are limited, if even positive.

*Source: _wp0402 - introduction in January 1998 of inflation targeting (IT).*

### REFERENCES

### _wp0402 - REFERENCES

### References (selected)
- Artis, M.J., and M. Ehrmann (2000), “The exchange rate – a shock-absorber or source of shocks? A study of four open economies”, EUI Working Papers, RSC No. 2000/38; CEPR Discussion Papers no. 2550.
- Blanchard, O.J., and D. Quah (1989), “The Dynamic Effects of Aggregate Demand and Supply Disturbances”, American Economic Review 79(4), 655-673.
- Buiter, W., 1995 “Macroeconomc Policy During a Transition to Monetary Union”, Centre for Economic Performance Discussion Paper N. 261.
- Canzoneri, M., J. Valles and J. Vinals (1996), “Do Exchange Rates Move to Address International Macroeconomic Imbalances?”, CEPR Discussion Papers no. 1498, London.
- Clarida R., and J. Gali (1994), “Sources of Real Exchange Rate Fluctuations: How Important Are Nominal Shocks?”, Carnegie-Rochester Conference Series on Public Policy 41, 1-56.
- Creel, J., and S. Levasseur (2003), “How would a fixed exchange rate regime fit the transition economies? The case of the Czech Republic, Hungary and Poland”, mimeo, OFCE, Paris
- Dibooglu, S. and A. Kutan (2001), “Sources of Real and Nominal Exchange Rate Fluctuations in Transition Economies: The Case of Poland and Hungary”, Journal of Comparative Economics 29, 257-275.
- Enders, W. and B. Lee (1997), “Accounting for Real and Nominal Exchange Rate Movements in the Post-Bretton Woods Period”, Journal of International Money and Finance 16(2), 223-254.
- Engel, C., 2002, “The Responsiveness of Consumer Prices to Exchange Rates and the Implications for Exchange-Rate Policy: A Survey of a Few Recent New Open-Economy Macro Models,” NBER Working Paper No. 8725.
- Gros, D., and Hobza (2003), “Exchange Rate Variability as an OCA Criterion: Are the Candidates Ripe for the Euro?”, International Center for Economic Growth, Working paper 23.
- Hoffmaister, A., and C.A. Végh, 1995, “Disinflation and the Recession-Now-Versus-Recession-Later Hypothesis: Evidence from Uruguay,” IMF Working Paper 95/99.
- International Monetary Fund, 1997, Finland, Selected Issues (Washington).
- Kontolemis, Z., and K. Ross (forthcoming), “Exchange Rate Fluctuations in Transition Economies”.
- McKinnon, J. (1991), “Critical Values for Cointegration Tests”, in Engle R.F. and C.W.J. Granger (eds.), Long Run Economic Relationships, Oxford University Press.
- Mundell, R., 1961, “A Theory of Optimum Currency Areas,” The American Economic Review, Vol. 51, No. 4, pp. 657-65.
- Obstfeld, M., 1985, “Floating Exchange Rates: Experience and Prospects,” Brookings Papers on Economic Activity, No. 2, pp. 369–450.
- Obstfeld, M., 2002, “Exchange Rates and Adjustment: Perspectives from the New Open Economy Macroeconomics,” NBER Working Paper No. 9118.
- Obstfeld, M., 2001, “International Macroeconomics: Beyond the Mundell-Fleming Model,” IMF Staff Papers, Vol. 47, pp. 1–39.
- Süppel, R. (2003), “Economic Dynamics in EU Accession Countries: A Case for Exchange Rate Flexibility?”, mimeo, European Central Bank, Frankfurt

### Figures: Impulse Response Functions (qualitative summary)
- Figures report impulse responses (levels) of nominal exchange rates, real exchange rates, prices, and relative output to 1 standard deviation structural shocks over horizons up to 48 months for: Czech Republic, Hungary, Poland (two specifications: PO(1), PO(2)), Slovak Republic, and Slovenia.
- Labels and series shown include distinctions between neutral shock and non-neutral shock (Figures 2) and between money shock, real demand shock, and supply shock (Figures 3).
- Time axis units: Time (in months) with horizons including 0, 3, 6, 9, 12, 15, 18, 21, 24, 30, 36, 42, 48 depending on panel.
- Magnitude ranges illustrated (examples from panels):
  - Czech Republic nominal exchange rate panel spans approximately -0.005 to 0.03 (percent).
  - Hungary nominal exchange rate panel spans approximately -0.02 to 0.05 (percent).
  - Poland panels span approximately -0.03 to 0.03 (percent) depending on specification.
  - Slovak Republic nominal exchange rate panel spans approximately -0.04 to 0.03 (percent).
  - Slovenia nominal exchange rate panel spans approximately 0 to 0.014 (percent).
- Real exchange rate, prices, and relative output panels similarly show small magnitude responses (e.g., Czech Republic: Real Exchange Rate panel spans -0.025 to 0.020; Czech Republic: Relative Output panel spans 0.000 to 0.030).

### Table A1. Test statistics and model setup — 2 VAR model (selected entries)
- Countries and sample periods:
  - CZ: 1996:02 – 2003:02
  - HU: 1995:03 – 2003:02
  - PO(1): 1995:05 – 2003:02
  - PO(2): 1998:02 – 2003:02
  - SR: 1997:01 – 2003:02
  - SA: 1993:01 – 2003:02
- Unit root test on level (ADF test statistic) — example entries:
  - CZ NER: -1.06 (Intercept)
  - HU NER: -1.74
  - PO(1) Relative output: -4.54***
- Unit root test on difference (ADF test statistic) — example entries:
  - CZ NER (difference): -9.74***
  - HU NER (difference): -7.02***
  - SA Relative output (difference): -9.62***
- Cointegration test (ADF) entries: values reported per country (e.g., CZ: -2.10; HU: -2.17).
- VAR modeling:
  - Lag length: 6 for all countries.
  - Dummies: month (included where significant); SA indicates output is seasonally adjusted.

- Significance legend:
  - *: Null hypothesis can be rejected at the 10% level
  - **: Null hypothesis can be rejected at the 5% level
  - ***: Null hypothesis can be rejected at the 1% level

### Table A2. CECs — Forecast Error Variance Decomposition (2 VAR model, all variables in logarithmic first differences)
- Structure: For each country and variable (nominal exchange rate; relative output) the table reports Shock S.E., and percent decomposition into neutral and non-neutral shocks at horizons 1, 2, 3, 6, 9, 12, 18, 24, 36, 48 (months).
- Czech Republic (example rows):
  - Variable: nominal exchange rate — Shock S.E. and decomposition:
    - 1: S.E. 0.021 neutral 95 non-neutral 5
    - 2: S.E. 0.021 neutral 95 non-neutral 5
    - 3: S.E. 0.022 neutral 95 non-neutral 5
    - 6: S.E. 0.022 neutral 95 non-neutral 5
    - 9: S.E. 0.022 neutral 92 non-neutral 8
    - 12: S.E. 0.022 neutral 92 non-neutral 8
    - 18: S.E. 0.022 neutral 92 non-neutral 8
    - 24: S.E. 0.022 neutral 92 non-neutral 8
    - 36: S.E. 0.022 neutral 92 non-neutral 8
    - 48: S.E. 0.022 neutral 92 non-neutral 8
  - Variable: relative output — Shock S.E. and decomposition:
    - 1: S.E. 0.013 neutral 83 non-neutral 17
    - 2: S.E. 0.014 neutral 83 non-neutral 17
    - 3: S.E. 0.015 neutral 84 non-neutral 16
    - 6: S.E. 0.016 neutral 83 non-neutral 17
    - 9: S.E. 0.016 neutral 84 non-neutral 16
    - 12: S.E. 0.016 neutral 84 non-neutral 16
    - 18: S.E. 0.016 neutral 84 non-neutral 16
    - 24: S.E. 0.016 neutral 84 non-neutral 16
    - 36: S.E. 0.016 neutral 84 non-neutral 16
    - 48: S.E. 0.016 neutral 84 non-neutral 16
- Hungary (selected):
  - Nominal exchange rate:
    - 1: S.E. 0.033 neutral 14 non-neutral 86
    - 12: S.E. 0.048 neutral 18 non-neutral 82
    - 48: S.E. 0.049 neutral 18 non-neutral 82
  - Relative output:
    - 1: S.E. 0.032 neutral 10 non-neutral 90
    - 12: S.E. 0.054 neutral 9 non-neutral 91
    - 48: S.E. 0.054 neutral 9 non-neutral 91
- Poland (PO(1) and PO(2)) (selected):
  - PO(1) nominal exchange rate:
    - 1: S.E. 0.021 neutral 83 non-neutral 17
    - 12: S.E. 0.024 neutral 79 non-neutral 21
    - 48: S.E. 0.024 neutral 78 non-neutral 22
  - PO(1) relative output:
    - 1: S.E. 0.024 neutral 62 non-neutral 38
    - 12: S.E. 0.029 neutral 53 non-neutral 47
    - 48: S.E. 0.029 neutral 52 non-neutral 48
  - PO(2) nominal exchange rate:
    - 1: S.E. 0.027 neutral 13 non-neutral 87
    - 12: S.E. 0.041 neutral 27 non-neutral 73
    - 48: S.E. 0.042 neutral 27 non-neutral 73
  - PO(2) relative output:
    - 1: S.E. 0.026 neutral 23 non-neutral 77
    - 12: S.E. 0.041 neutral 48 non-neutral 52
    - 48: S.E. 0.042 neutral 49 non-neutral 51
- Slovak Republic:
  - Nominal exchange rate:
    - 1: S.E. 0.015 neutral 96 non-neutral 4
    - 12: S.E. 0.018 neutral 75 non-neutral 25
    - 48: S.E. 0.018 neutral 74 non-neutral 26
  - Relative output:
    - 1: S.E. 0.006 neutral 73 non-neutral 27
    - 12: S.E. 0.008 neutral 79 non-neutral 21
    - 48: S.E. 0.008 neutral 79 non-neutral 21
- Slovenia:
  - Nominal exchange rate:
    - 1: S.E. 0.029 neutral 1 non-neutral 99
    - 12: S.E. 0.038 neutral 6 non-neutral 94
    - 48: S.E. 0.039 neutral 6 non-neutral 94
  - Relative output:
    - 1: S.E. 0.009 neutral 38 non-neutral 62
    - 12: S.E. 0.009 neutral 40 non-neutral 60
    - 48: S.E. 0.009 neutral 40 non-neutral 60

### Table A3. Test statistics and model setup — 3 VAR model (selected entries)
- Unit root test on level (ADF test statistic) for RER (trend and intercept) — example entries:
  - CZ RER: -2.78
  - HU RER: -3.81**
  - SA RER: -2.94
- Unit root test on difference (ADF test statistic) for RER:
  - CZ: -10.60***
  - HU: -6.73***
  - PO(1): -6.68***
  - PO(2): -5.06***
  - SR: -6.58***
  - SA: -6.37***
- Cointegration test (ADF) — example:
  - SA: -4.28** (none)
- VAR modeling:
  - Lag length: 6.00 for all countries.
  - Optimal lag length criteria reported (LR, FPE, AIC, SC, HQ) with numeric codes in the table.
  - Dummies included vary by country (examples: D2, D6, month for CZ; D8, month for PO(1)); SA indicates output is seasonally adjusted.

### Table A4. CECs — Forecast Error Variance Decomposition (3 VAR model, all variables in logarithmic first differences) (selected entries)
- Format: For each country and variable (real exchange rate, nominal exchange rate, relative output) the table reports Shock S.E. and percent decomposition into money, demand, and supply shocks at horizons 1, 2, 3, 6, 9, 12, 18, 24, 36, 48.
- Czech Republic (selected):
  - Real exchange rate (S.E. and decomposition):
    - 1: S.E. 0.019 money 80 demand 20 supply 0
    - 12: S.E. 0.023 money 59 demand 32 supply 9
    - 48: S.E. 0.024 money 58 demand 32 supply 10
  - Nominal exchange rate:
    - 1: S.E. 0.019 money 0 demand 32 supply 68 (note formatting in source shows concatenated numbers)
    - 12: S.E. 0.023 ... (see full table for detailed series)
  - Relative output:
    - 1: S.E. 0.034 money 0 demand 32 supply 68 (formatting in source indicates percentages reported)
- Hungary (selected):
  - Real exchange rate:
    - 1: S.E. 0.012 money 67 demand 31 supply 2
    - 12: S.E. 0.014 money 60 demand 29 supply 12
    - 48: S.E. 0.014 money 59 demand 29 supply 12
  - Nominal exchange rate:
    - 1: S.E. 0.0123... (source formatting shows 0.01232599) money 0 demand 33 supply 67 (see table for exact cell formatting)
  - Relative output:
    - 1: S.E. 0.033 money 6 demand 19 supply 75 (table contains exact sequences)
- Poland (PO(1) and PO(2)) (selected):
  - PO(1) real exchange rate:
    - 1: S.E. 0.020 money 41 demand 56 supply 3
    - 48: S.E. 0.027 money 35 demand 57 supply 8
  - PO(2) real exchange rate:
    - 1: S.E. 0.023 money 25 demand 52 supply 22
    - 48: S.E. 0.031 money 21 demand 47 supply 32
- Slovak Republic (selected):
  - Real exchange rate:
    - 1: S.E. 0.017 money 28 demand 61 supply 10
    - 48: S.E. 0.022 money 37 demand 48 supply 15
  - Nominal exchange rate:
    - 1: S.E. 0.015 money 70? (source shows 0.0157017130.030 7 193 — full numeric layout in table)
- Slovenia (selected):
  - Real exchange rate:
    - 1: S.E. 0.006 money 31 demand 57 supply 13
    - 48: S.E. 0.008 money 45 demand 43 supply 12
  - Nominal exchange rate:
    - 1: S.E. 0.006780220 money 27 demand 76? (see full table for exact formatting)

(Note: Table cells contain tightly formatted numeric sequences in the source; the entries above preserve the numeric values and decomposition categories as they appear.)

*Italic: Source document: _wp0402 - REFERENCES (PDF) *

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp0402.pdf_