## _wp0465

## Source details

**Canonical URL:** [_wp0465](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp0465.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp0465.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp0465.pdf.json)

---

### I. Introduction and summary
- Main driver of Croatia's current account deficit is the merchandise trade deficit due to stagnating exports and increasing imports over the last three years.
- Two approaches to changing competitiveness:
  - Internal: supply-side policies (curbing inflation; decreasing taxes such as producers' share in the social contribution tax or the profit tax; relaxing rigid labor market conditions).
  - External: devaluing/depreciating the currency.
- Paper objectives:
  - Establish whether a stable long-run relationship exists between the exchange rate and the merchandise trade balance.
  - Determine whether depreciation leads to a net long-run improvement of the trade balance and quantify the extent.
  - Estimate short-run dynamics, including evidence on the J-curve, its length and persistence.
  - Inform effects of expenditure-reducing policies and central bank intervention alternatives.
- Model and data overview:
  - Traditional two-country, imperfect-substitution model.
  - Aggregated flows of six major trading partners (on average, over 55 percent of total trade).
  - Real effective exchange rate (REER) constructed using CPI and PPI variants.
  - Domestic and foreign income proxied by GDP.
  - Observation period: 1994(1) to 2002(2).
  - Three modeling techniques and two competitiveness measures used for robustness.

### II. Data, variables, and econometric characteristics
- Key variables (log form, seasonally adjusted): TB (export-to-import ratio), Q_PPI, Q_CPI, YD, YF.
- REER construction (author):
  - q(t) = 100 * Πi (S_it * P_it / P_t)^(w_i), i = Euro (Austria, Germany, Italy, France), British sterling, Slovenian tolar; weights w fixed at average trade weight for 1996 to 2000.
- Foreign income: weighted average of real GDP indexes for six trading partners with fixed average export-share weights for 1996 to 2000.
- Sample and frequency:
  - Quarterly data for 1994(1) to 2002(1).
  - GDP data unavailable before 1994; fourth quarter 1993 structural break.
- Stationarity test results (levels and first differences, ADF and PP; MacKinnon CVs reported in source):
  - Levels (selected reported statistics, natural logs):
    - TB: ADF -0.601(4) CV -2.967 ; PP -1.839 CV -2.953
    - Q_cpi: ADF -1.800(0) CV -2.956 ; PP 5.355 CV -2.953
    - Q_ppi: ADF -2.956(5) CV -2.975 ; PP -1.919 CV -2.956
    - Yd: ADF -1.996(0) CV -2.953 ; PP -1.934 CV -2.953
    - Yf: ADF -3.491(3) CV -3.580 ; PP -0.914 CV -2.956
  - First differences (selected reported statistics):
    - d(TB): ADF -5.143(0) CV -2.956 ; PP -5.284 CV -2.956
    - d(Q_cpi): ADF -5.519(0) CV -2.959 ; PP -5.540 CV -2.959
    - d(Q_ppi): ADF -5.231(0) CV -2.959 ; PP -5.262 CV -2.959
    - d(Yd): ADF -4.220(0) CV -2.956 ; PP -4.202 CV -2.956
    - d(Yf): ADF -2.558(5) CV -2.980 ; PP -4.708 CV -2.959
  - Interpretation: TB, Q_PPI, YD, and YF generally non-stationary in levels; ambiguity for Q_CPI. First differences largely stationary. Given short sample and possible breaks, three techniques employed to ensure robustness.

### III. Econometric approaches
- Three complementary techniques used:
  1. ARDL bounds testing (Pesaran, Shin, and Smith (PSS)): avoids pre-classification of I(0)/I(1); uses F/Wald bounds for cointegration and then error-correction specification.
  2. Bewley-type ARDL (Bewley (1979) as per Wickens and Breusch (1988)): direct estimation of long-run coefficients via re-parameterization; TSLS used to address simultaneity.
  3. ARDL(IV) (Pesaran and Shin (1997)): instruments replace current and lagged differences to mitigate endogeneity and perform well in small samples.
- Shock experiment:
  - Impulse responses computed for a permanent 1 percent shock to the exchange rate, holding incomes constant.

### IV. Empirical cointegration results and long-run relationships
- PPI-version PSS cointegration test:
  - F-statistic = 12.17667, Probability = 0.000178
  - Chi-square = 48.70668, Probability = 0
  - Critical bounds (95 percent, intercept and no trend): Wald I(0) 17.69 and I(1) 23.34; F I(0) 2.85 and I(1) 4.049.
  - Conclusion: reject null of no cointegration — long-run relationship exists for PPI-version.
- Reported long-run PPI-version equation (as presented in source):
  - FDPPI = 1.396.478.87 + yyqtb (notation preserved exactly as in source: FDPPI yyqtb 87.478.396.1+−=)
  - Adjusted R2 = 0.83; F-statistic = 10.81; LM test = 1.66 (p = 0.23); RESET test = 2.32 (p = 0.15).
- CPI-version PSS cointegration test:
  - F-statistic = 5.314387, Probability = 0.005799
  - Chi-square = 21.25755, Probability = 0.000282
  - Critical bounds lead to inconclusive results for some statistics; null rejected per F-stat bounds.
- Reported long-run CPI equation (as in source):
  - FDCPI = 95.2 log y - 70.2 log q + 25.2 log (notation preserved)
  - Adjusted R^2 = 0.48; F-statistic = 3.18; LM test = 3.53 (p = 0.03); RESET test = 0.162 (p = 0.69).
- ARDL(IV) long-run reported estimates (notation preserved):
  - PPI version: FDPPI = 66.3 log y - 54.3 log q + 00.1 log (Adjusted R^2 = 0.98; LM test = 1.72 (p = 0.28); RESET = 0.76 (p = 0.41))
  - CPI version: FDCPI = 42.4 log y - 38.3 log q + 09.2 log (Adjusted R^2 = 0.96; LM test = 1.96 (p = 0.18); RESET = 0.33 (p = 0.57))

### V. Short-run dynamics and impulse response findings (by approach and REER measure)
- PSS approach (PPI version) short-run dynamic (equation (21) as in source):
  - ∆tb_t = -0.53*EC_{t-1} - 0.82∆q^{PPI}_{t-1} - 1.06 ∆q^{PPI}_{t-2} + 4.00∆q^{PPI}_{t-3} - 0.25∆tb_{t-1} + 0.12∆tb_{t-2} - 0.09∆tb_{t-3} - 0.33∆tb_{t-4}
  - Impulse response to permanent depreciation:
    - Initial deterioration: total amount of 2.5 percent.
    - Length of adverse effect: two quarters.
    - New equilibrium after ten quarters.
    - New equilibrium level: 1.4 percent above the old equilibrium level.
- PSS approach (CPI version) short-run dynamic (equation (23) as in source):
  - ∆tb_t = -0.73 EC_{t-1} - 2.47∆q^{CPI}_{t-1} - 1.63∆q^{CPI}_{t-2} + 2.66∆q^{CPI}_{t-3} - 1.34∆y^{D}_{t-1} - 3.09∆y^{D}_{t-2} - 0.04∆tb_{t-1} + 0.21∆tb_{t-2} - 0.14∆tb_{t-3} - 0.34∆tb_{t-4}
  - CPI PSS impulse response:
    - Initial deterioration: 10.1 percent in total.
    - Length of adverse effect: four quarters.
    - New equilibrium after fifteen quarters.
    - New equilibrium level: 1.05 percent below the old equilibrium level.
- ARDL(IV) short-run impulse equations (notation preserved from source):
  - PPI (equation (26)) and CPI (equation (27)) reported with contemporaneous and lagged coefficients on q, yD, yF, and ∆tb terms.
  - ARDL(IV) impulse response findings (both variants):
    - Signs of a J-curve effect present.
    - Time to new equilibrium: six quarters.
    - Length of adverse effect: one quarter.
    - Magnitudes:
      - PPI: initial deterioration 0.5 percent; new equilibrium 1.29 percent above pre-depreciation equilibrium.
      - CPI: initial deterioration 1.1 percent; new equilibrium 0.75 percent above pre-depreciation equilibrium.
- Bewley ARDL approach (two-variable focus) impulse equations (notation preserved):
  - PPI (equation (28)): tb_t = 1.39 q^{PPI}_t - 6.84∆q^{PPI}_{t-1} - 1.03∆q^{PPI}_{t-2} + 3.12∆q^{PPI}_{t-3} - 3.03∆q^{PPI}_{t-4} - 0.13 ∆tb_{t-1}
    - Impulse outcomes:
      - Delayed J-curve lasting two quarters.
      - Adverse impact totaling 5.8 percent.
      - New equilibrium after eight quarters at 1.41 percent above old equilibrium.
  - CPI (equation (29)): tb_t = 1.81 q^{CPI}_t - 1.61∆q^{CPI}_{t-1} - 0.63∆q^{CPI}_{t-2} + 1.77∆q^{CPI}_{t-3} - 1.51∆q^{CPI}_{t-4} - 0.29∆tb_{t-1} - 0.26∆tb_{t-2} - 0.02∆tb_{t-3} - 0.13∆tb_{t-4}
    - Impulse outcomes:
      - No support for a J-curve.
      - New equilibrium after fifteen quarters at 1.9 percent above the old equilibrium.

### VI. Elasticities and aggregated quantitative results
- Elasticity summary (Table 4, entries preserved):
  - q (CPI based): Bewley 1.82, PSS 2.24, ARDL(IV) 2.09, average 2.05
  - q (PPI based): Bewley 1.39, PSS 1.96, ARDL(IV) 1.00, average 1.45
  - yD (CPI based): -2.70, -3.39, -3.05
  - yD (PPI based): -3.78, -3.54, -3.66
  - yF (CPI based): 2.94, 4.42, 3.68
  - yF (PPI based): 4.87, 3.66, 4.27
- Average long-run elasticity findings reported:
  - Long-run elasticity of the trade balance with respect to the exchange rate on average between 1.45 and 2.05.
  - Foreign income elasticity on average between 3.68 and 4.27.
  - Domestic income elasticity averaging between 3.05 and 3.66.
- Interpretation:
  - Elasticities reflect substitution effects and domestic substitute availability.
  - Domestic income elasticity lower than foreign income elasticity.
  - Higher growth in Croatia relative to six largest trading partners likely implies a net negative impact on the trade balance.

### VII. Aggregated permanent-depreciation quantitative summary (Table 5 and text)
- Selected Table 5 entries (preserving exact figures and signs):
  - BEW-CPI: Relative position of new long-run equilibrium level 1.9; Time for new equilibrium in quarters 15; Length of J-curve effect 0; Extent of J-curve effect 0.00
  - BEW-PPI: Relative position 1.41; Time for new equilibrium 8; Length of J-curve effect 2; Extent of J-curve effect -5.77
  - PSS-CPI: Relative position -1.05; Time for new equilibrium 15; Length of J-curve effect 4; Extent of J-curve effect -10.10
  - PSS-PPI: Relative position 1.35; Time for new equilibrium 10; Length of J-curve effect 2; Extent of J-curve effect -2.49
  - ARDL(IV)-CPI: Relative position 1.29; Time for new equilibrium 6; Length of J-curve effect 1; Extent of J-curve effect -0.52
  - ARDL(IV)-PPI: Relative position 0.75; Time for new equilibrium 6; Length of J-curve effect 1; Extent of J-curve effect -1.11
  - total average: Relative position 0.94; Time for new equilibrium in quarters 10; Length of J-curve effect 1.67; Extent of J-curve effect -3.33
  - average without PSS_CPI: Relative position 1.349; Time for new equilibrium in quarters 1.20; Length of J-curve effect -1.98
- Text summary measures:
  - On average, a permanent 1 percent depreciation results in a 0.94 percent improvement of the trade balance.
  - Ignoring the outlier (PSS with CPI), improvement amounts to 1.34 percent.
  - Average time to new equilibrium: approximately 2.5 years.
  - Average length of adverse effect of permanent depreciation: moderately above one quarter.
  - J-curve increase in the trade balance deficit estimated between 2.0 percent and 3.3 percent.
- Caveats:
  - Short time series and dynamic trade environment with many outliers recommend caution in interpretation.
  - Single-equation approach limitations: possible multiple cointegrating vectors and weak exogeneity issues may produce inefficiency or misleading estimates.

### VIII. Policy implications
- Main policy-relevant findings:
  - Improvement of the trade balance after permanent devaluation/depreciation is limited: a 1 percent permanent devaluation results in a new equilibrium level up to 1.34 percent above the old equilibrium (best case).
  - The initial adverse J-curve effect can be significant.
- Eight potential spillover/adverse effects of devaluation (enumerated in source; excerpt emphasizes six):
  1. Exchange rate pass-through on inflation can neutralize nominal depreciation and lead to real exchange rate appreciation.
  2. High dependence on imports for intermediate goods and raw materials can erode competitiveness gains from real depreciation.
  3. Increases in import and export prices may deteriorate domestic real income.
  4. Resource shift from non-tradable to tradable sector could create a wage gap, wage spiral, or net increase in unemployment given rigid labor markets.
  5. Indexed debt levels and high euroization could amplify adverse consequences; distribution and structure of such debt matter.
  6. Strong permanent depreciation could severely impact public finances by increasing domestic currency cost of debt servicing, risking measures to combat a general government deficit around 5.5 percent of GDP.
- Policy judgment from the study:
  - Given limited trade-balance gains and significant potential negative side effects, a permanent nominal depreciation in an asymmetric intervention policy or in a pegged exchange rate regime is unlikely to have a net favorable effect on the entire economy.

### IX. Conclusion and suggested further research
- Key empirical conclusions:
  - On average, a permanent 1 percent depreciation yields an improvement of the trade balance between 0.94 percent and 1.3 percent.
  - New equilibrium established after approximately 2.5 years.
  - Average length of the adverse effect: moderately above one quarter.
  - Evidence of the J-curve effect in Croatia with an estimated increase in the trade balance deficit due to the J-curve between 2.0 percent and 3.3 percent.
  - Results sensitive to inclusion of dummy variables; one model (PSS-CPI) is an outlier.
  - Overall empirical consistency across methods with that exception.
- Suggested further research paths:
  - Estimate a VAR and/or VECM to account for feedback effects.
  - Model bilateral trade data to offset possible mismeasurement (e.g., shipbuilding trade with Italy).
  - Use disaggregated data by standard international trade classification; treat shipbuilding and trade in oil derivatives separately.

*Source: _wp0465 (excerpted sections II–IX and references).*

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

### _wp0465 - References

### I. Introduction and summary

- The main component of Croatia's considerable current account deficit is the merchandise trade deficit, which results from stagnating exports and increasing imports over the last three years.
- Two approaches to changing competitiveness are available:
  - Internal approach: supply-side policies such as curbing inflation, decreasing taxes (e.g., producers' share in the social contribution tax or the profit tax), or relaxing rigid labor market conditions.
  - External approach: devaluing/depreciating the currency.
- Public debate in Croatia is split between advocates of nominal exchange rate devaluation/depreciation and advocates of maintaining the managed floating exchange rate policy.
- A formal quantitative estimate of the long-run and short-run impacts of exchange rate changes on the merchandise trade balance has been lacking; this paper aims to close that gap.
- Motivations for quantifying responsiveness of the merchandise trade balance to exchange rate changes:
  - Establish whether a stable long-run relationship exists between the exchange rate and the merchandise trade balance.
  - If a long-run relationship exists, determine whether depreciation will lead to a net long-run improvement of the trade balance.
  - Quantify the extent of trade balance improvement to weigh benefits against costs of permanent depreciation.
  - Estimate short-run dynamics, including whether depreciation has an adverse short-run impact (the J-curve effect) and the persistence/extent of any adverse effect.
  - Contribute to understanding international transmission of changes in economic activity and prices.
  - Provide information on the impact of expenditure-reducing policies (monetary and fiscal) on the domestic trade balance.
  - Encourage further analysis of the impact of substantial government foreign lending on the trade balance and the current account.
  - Inform assessment of asymmetric central bank intervention policy versus alternatives (e.g., introduction of a crawling-peg exchange rate policy) to support permanent depreciation.

- Model and data setup:
  - The responsiveness is evaluated in a traditional two-country, imperfect-substitution model.
  - The trade balance is approximated by aggregated flows of Croatia's six major trading partners, which account, on average, for over 55 percent of its total trade.
  - Real effective exchange rate is modeled using the consumer price index (CPI) and the producer price index (PPI).
  - Domestic and foreign income are proxied by GDP.
  - Observation period: from the first quarter of 1994 to the second quarter of 2002.
  - Three modeling techniques and two competitiveness measures are employed to obtain more robust results, taking into account that Croatia is a transition country with only short time series and a dynamic trading environment characterized by many outliers.

- Aggregation and interpretation notes:
  - The estimated response of the trade balance to depreciation is a weighted average result because foreign income is aggregated across six partners and fluctuations are averaged in the real effective exchange rate.
  - The aggregate result incorporates asymmetric responses of trade flows across the six countries; nonexistence of a long-run relationship or J-curve effect at the aggregated level does not preclude validity for some individual countries.
  - Aggregated approach is used because policymakers prefer single averaged outcomes, disaggregated data tend to have larger measurement errors, and disaggregated functions are more likely to be misspecified (Grunfeld and Griliches, 1960; and Aigner and Goldfeld, 1974).
  - The definition of the aggregated foreign income variable also has problems.

- Empirical findings (as reported in the introduction):
  - Evidence is found of a long-run relationship between the trade balance and its arguments.
  - Long-run trade balance elasticity with respect to the exchange rate ranges between 1.5 and 2.1.
  - The long-run domestic income elasticity is lower than the foreign income elasticity.
  - On average, the J-curve effect does exist in Croatia.
  - The short-run adverse effect of depreciation on the trade balance lasts about one quarter, during which the trade balance deteriorates by around 2.0 percent.
  - It takes the trade balance about 2.5 years to reach a new equilibrium level, improving, on average, between 0.9 percent and 1.3 percent compared with the old equilibrium.

### Tables and figures (listed in source)

- Tables (titles as listed):
  - 1. Augmented Dickey-Fuller and Phillips-Perron Test Results for Nonstationarity of Variables
  - 2. Producer Price Index Version Model: Test for Cointegration
  - 3. Consumer Price Index Version Model: Test for Cointegration
  - 4. Elasticity of Trade Balance with Respect to Eeal Effective Exchange Rate (q), Domestic Income (yD), and Foreign Income (yF)
  - 5. Summary of Empirical Results

- Figures (titles as listed):
  - 1. Trade Balance and Real Effective Exchange Rate (PPI based)
  - 2. Trade Balance and Real Effective Exchange Rate (CPI based)
  - 3. Impulse Response: Pesaran, Shin, and Smith Model (PPI)
  - 4. Impulse Response: Pesaran, Shin, and Smith Model (CPI)
  - 5. Impulse Response: ARDL(IV) - PPI
  - 6. Impulse Response ARDL(IV) - CPI
  - 7. Impulse Response: Bewley Model (PPI)
  - 8. Impulse Response: Bewley Model (CPI)

*Source: _wp0465 - References*

### Section VI describes the technical and econometric characteristics of the data. Section VII

### _wp0465 - Section VI describes the technical and econometric characteristics of the data. Section VII

### Literature review
- Short-run and long-run relationships between the trade balance and exchange rate have mixed empirical evidence across countries and methodologies.
- Key country findings:
  - United States: Rose and Yellen (1989) (1960 to 1985) do not find support for the J-curve or a long-run relationship. Bahmani-Oskooee and Brooks (1999) (ARDL) do not find a J-curve but report a significant long-run relationship indicating real depreciation of the U.S. dollar has a favorable effect on the U.S. trade balance. Shirvani and Wilbratte (1997) (VECM) find a reversed L-curve. Demirden and Pastine (1995) show feedback effects may result in a J-curve.
  - Australia: Felmingham (1988) (1965 to 1985) finds no J-curve.
  - Canada and United States: Marwah and Klein (1996) (1977 to 1992, IV and OLS on disaggregated data) find an S-curve: initial deterioration after depreciation, followed by improvement, then later worsening.
  - Japan: Gupta-Kapoor and Ramakrishnan (1999) (VECM) and Jung and Doroodian (1998) find evidence of a J-curve.
  - Cross-country tests: Bahmani-Oskooee and Alse (1994) (Engle-Granger) find cointegration for 14 of 41 countries; evidence of J-curve in those cointegrated cases.
- Emerging and developing countries:
  - Thailand and Malaysia: Baharumshah (2001) (VAR, 1980 to 1996) finds a stable positive long-run relationship; short-run J-curve evidence mixed (delayed J-curve in Thailand, none in Malaysia).
  - Thailand bilateral (1973 to 1997): Bahmani-Oskooee and Kantipong (2001) find J-curve with U.S. and Japan only.
  - Eight developing countries (Upadhyaya and Dhakal, 1997): only Mexico shows long-run improvement from devaluation.
  - Middle East (Bahmani-Oskooee, 2001): favorable long-run effect of real depreciation on trade balance for Bahrain, Egypt, Jordan, Morocco, Syria, Tunisia, and Turkey.
  - Turkey: Akbostanci (2002) finds favorable long-run relationship and S-shaped short-run response; Kale (2001) finds negative long-run impact but delayed J-curve in short-run.
- Central and Eastern European Countries (CEEC): Hacker and Hatemi (2002) find positive long-run relationship for Czech Republic, Hungary, and Poland with respect to Germany; J-curve in Czech Republic and Poland, not Hungary.
- Overall synthesis:
  - The J-curve is not an empirical regularity—present in some countries, absent in others.
  - Evidence generally supports a favorable stable long-run relationship between trade balance and exchange rate, with notable exceptions such as the United States.

### Real exchange rate and merchandise trade development in Croatia
- Current account and merchandise trade:
  - Current account deficit peaked in 1997 at 12.5 percent of GDP.
  - By end-2002 preliminary data: current account deficit about 6.9 percent of GDP (five-year high as of 2002).
  - Main driver of the substantial deficit is the merchandise trade balance: stagnating exports and propulsive imports.
  - Net tourism income and workers' remittances cushion the negative balance but are vulnerable and often seasonal.
- Real effective exchange rate (REER) measures used:
  - Two REER variants computed: PPI-based and CPI-based (both normalized and seasonally adjusted).
  - Increase in exchange rate denotes a depreciation.
  - CPI vs PPI: CPI includes final traded and non-traded goods (includes imports), subject to distortions (administrative prices, excise duties); PPI excludes imports, includes intermediate goods, raw materials, capital goods, and is often proxy for domestic export prices.
  - Between 1998 and 2000 the REER(PPI) depreciated more than REER(CPI) due to strong growth of prices of services, tobacco products and beverages, partly related to excise duty increases.
- Trade balance dynamics and J-curve concept:
  - Trade balance measured as export-to-import ratio (TB).
  - Long-run: steady-state relationship between new exchange rate level and trade balance.
  - Short-run: J-curve — initial deterioration of trade balance after depreciation due to sticky domestic-currency export prices and lagged volume responses (recognition lag, decision lag, delivery lag, replacement lag, production lag).
  - J-curve represents the transition path from old equilibrium to new equilibrium following permanent exchange rate changes.

### Theoretical model
- Framework:
  - Standard two-country imperfect substitutes model (Goldstein and Kahn (1985), Rose and Yellen (1989)).
  - Imports and exports are not perfect substitutes for domestic goods; finite elasticities estimated.
- Demand and supply specifications (as in source equations):
  - Domestic import demand Md = f(Y, PM, P, ... ) with expected signs: ∂Md/∂Y > 0, ∂Md/∂PM < 0, ∂Md/∂P > 0.
  - Foreign import demand M*d = f(Y*, P*M, P*, e, ... ) with similar sign conventions.
  - Relative price of imports defined via foreign export price and real exchange rate q where q = P*e/P (increase in q refers to depreciation).
  - Export supply and foreign export supply functions Xs = f(Px, P) and X*s = f(P*x, P*).
  - Equilibrium conditions e X = Ms* and Xs = Md*.
- Trade balance definition and expectations:
  - Trade balance TB = (exports/price) − (imports/price) in reduced form TB = f(Y, Y*, Q) with sign expectations:
    - Real foreign income (Y*) positively related to TB.
    - Real exchange rate (Q) positively related to TB (depreciation improves TB in long run).
    - Domestic income (Y) negatively related to TB.

### Applied econometric models
- Rationale for multiple techniques: robustness given short sample and low-power/contradictory unit root tests; avoid simultaneity bias.
- Three techniques used:
  1. ARDL approach (Pesaran, Shin, and Smith (1995, 1996)) — avoids prior classification of I(0)/I(1); uses bounds testing (F-test/Wald) for cointegration and then estimates error-correction representation.
     - Null of no long-run relationship tested via F-statistic/Wald with non-standard distributions; critical bounds provided for I(0) and I(1) cases.
  2. Bewley-type (1979) ARDL approach (as discussed in Wickens and Breusch (1988)) — direct estimation of long-run coefficients by re-parameterizing distributed lag; uses TSLS to remove simultaneity; potential finite-sample bias and sensitivity to lagged dependent variable.
  3. ARDL with instrumental variables (Pesaran and Shin (1997)) — mixture of PSS and Bewley approaches; uses instruments to replace current and lagged differences in dependent variable by lagged levels; good small sample properties; robust to variable generating process.
- Objectives:
  - Determine cointegration among variables despite unit-root test ambiguity.
  - Obtain estimates robust to simultaneity bias.

### Data description — technical
- Sample and frequency:
  - Estimated period: 1994(1) to 2002(1) using quarterly data.
  - Earlier periods not used due to unreliable data: GDP data unavailable before 1994; stabilization program and currency change in fourth quarter 1993 represent structural break.
- Data sources and notes:
  - Merchandise exports, imports, trade direction: Croatian Bureau of Statistics (CBS) reports (including corrections for shipbuilding industry).
  - 2000 methodological break in customs data processing altered intra-quarter distribution but not significant quarterly spillovers.
  - Croatian PPI and CPI indices, and seven bilateral nominal exchange rates from CNB database.
  - PPI indices and real GDP for trading partners (Austria, France, Germany, Italy, Slovenia, United Kingdom) from IMF International Financial Statistics CD (February 2003).
  - Bosnia excluded due to data unavailability.
- Variable construction:
  - Merchandise trade balance defined as exports-to-imports ratio (avoids log of deficit problem).
  - Real effective exchange rate (q) constructed by author because official CNB REER includes invisibles (tourism) and thus is not appropriate here.
  - REER formula (author’s estimate):
    - q(t) = 100 * Πi (S_it * P_it / P_t)^(w_i)  (equation 18 in source)
    - i = Euro (Austria, Germany, Italy, France), British sterling, Slovenian tolar
    - t = 1994q1, 1994q2....2002q1
    - Weights w fixed at average trade weight for each trading partner from 1996 to 2000 (chosen as period after hostilities ended).
  - Foreign income variable: weighted average of real GDP indexes for six trading partners using variable export shares as weights, fixed average for 1996 to 2000.
- Seasonal adjustment and transformations:
  - All variables are seasonally adjusted and in log-form.
  - TB, domestic real GDP, and REER seasonally adjusted using TRAMO-SEATS ARIMA model.
  - Composite foreign demand index seasonally adjusted using X12 due to TRAMO-SEATS diagnostics problems (many outliers).

### Data description — econometric characteristics
- Variables considered: TB, Q_PPI, Q_CPI, YD, YF.
- Stationarity tests employed: Augmented Dickey-Fuller (ADF) and Phillips-Perron (PP).
  - ADF follows Dolado, Jenkinson and Sosvilla-Rivero (1990) implementation (Enders (1995)).
  - PP relaxes autocorrelation and heteroskedasticity assumptions via non-parametric correction.
- Table 1 test results (as reported in source; variables in natural logarithms):
  - Levels:
    - Trade balance (TB): ADF -0.601(4) CV -2.967 ; PP -1.839 CV -2.953
    - Real effective exchange rate CPI (Q_cpi): ADF -1.800(0) CV -2.956 ; PP 5.355 CV -2.953
    - Real effective exchange rate PPI (Q_ppi): ADF -2.956(5) CV -2.975 ; PP -1.919 CV -2.956
    - Domestic GDP (Yd): ADF -1.996(0) CV -2.953 ; PP -1.934 CV -2.953
    - Foreign GDP (Yf): ADF -3.491(3) CV -3.580 ; PP -0.914 CV -2.956
  - First differences:
    - d(TB): ADF -5.143(0) CV -2.956 ; PP -5.284 CV -2.956
    - d(Q_cpi): ADF -5.519(0) CV -2.959 ; PP -5.540 CV -2.959
    - d(Q_ppi): ADF -5.231(0) CV -2.959 ; PP -5.262 CV -2.959
    - d(Yd): ADF -4.220(0) CV -2.956 ; PP -4.202 CV -2.956
    - d(Yf): ADF -2.558(5) CV -2.980 ; PP -4.708 CV -2.959
  - Notes: CV denotes MacKinnon critical values for rejecting unit root hypothesis.
- Interpretation of tests:
  - Both tests consistently suggest TB, Q_PPI, domestic GDP, and foreign GDP are non-stationary in levels (borderline cases noted).
  - Q_CPI: non-stationary per ADF, stationary per PP.
  - In first differences, four variables are stationary according to both tests; foreign demand is an exception (not passing ADF but stationary per PP).
  - Given short series and plausible structural breaks, preferred strategy is to employ multiple techniques: PSS bounds testing (does not require classification I(0)/I(1)), ARDL(IV), and Bewley approach for robustness.

### Empirical results — overview and specific PPI-version cointegration test
- Estimation strategy:
  - Use two REER measures (CPI and PPI based) and three modeling techniques (PSS, ARDL(IV), BEW).
  - Report long-run relationships and short-run dynamics; compute impulse responses for a permanent 1 percent shock to the exchange rate (shock applied to exchange rate, other incomes held constant).
- PPI-version model — cointegration test (Pesaran, Shin, and Smith method):
  - Reported test statistics:
    - F-statistic = 12.17667, Probability = 0.000178
    - Chi-square = 48.70668, Probability = 0
  - Reported critical value bounds (95 percent, intercept and no trend):
    - Wald statistic: I(0) 17.69 and I(1) 23.34. Report notes: test results are above upper-bound, hence null rejected.
    - F statistic: I(0) 2.85 and I(1) 4.049. Report notes: test results are above upper-bound, hence null rejected.
  - Conclusion: reject null of no cointegration — long-run relationship exists for the PPI-version model.
- Estimated long-run PPI-version equation (as reported):
  - FDPPI yyqtb 87.478.396.1+−=      (equation numbering from source)
    - (As presented in source: FDPPI yyqtb87.478.396.1+−=)
  - Goodness-of-fit and diagnostics:
    - Adjusted R2 = 0.83
    - F-statistic = 10.81
    - LM test = 1.66 (p = 0.23)
    - Reset test = 2.32 (p = 0.15)
- Short-run dynamics:
  - After establishing cointegration, Engle and Granger (1987) style error-correction term specified to obtain short-run dynamics.
  - Short-run response to a permanent 1 percent depreciation is analyzed via impulse response diagrams (description of diagram provided; full diagram and numerical impulse responses appear later in the source beyond the provided excerpt).

*Source: Excerpt from _wp0465 (sections II–VII).*

### references to a certain time period since the impulse should relate to a reaction of the

### _wp0465 - references to a certain time period since the impulse should relate to a reaction of the

### Impulse response: PSS approach (PPI version)
- Estimated short-run dynamic model (equation (21)):
  - ∆tb_t = -0.53*EC_{t-1} - 0.82∆q^{PPI}_{t-1} - 1.06 ∆q^{PPI}_{t-2} + 4.00∆q^{PPI}_{t-3} - 0.25∆tb_{t-1} + 0.12∆tb_{t-2} - 0.09∆tb_{t-3} - 0.33∆tb_{t-4}
  - EC represents the error correction term.
- Impulse response outcome after a permanent depreciation:
  - Initial deterioration of the trade balance by a total amount of 2.5 percent.
  - Length of the adverse effect: two quarters.
  - New equilibrium established ten quarters after the shock.
  - New equilibrium level: 1.4 percent above the old equilibrium level.

### Cointegration and PSS approach (CPI version)
- Tests (Table 3) for no cointegration:
  - F-statistic = 5.314387, Probability = 0.005799.
  - Chi-square = 21.25755, Probability = 0.000282.
  - Notes: Critical value bounds of the Wald statistic with intercept and no trend at 95 percent I(0) 17.69 and I(1) 23.34. Hence, inconclusive result.
  - Critical value bounds of the F statistic with intercept and no trend at 95 percent I(0) 2.85 and I(1) 4.049. Hence, the null hypothesis is rejected.
- Long-run CPI version (equation (22)):
  - FDCPI = 95.2 log y - 70.2 log q + 25.2 log (notation preserved as in source)
  - Adjusted R^2 = 0.48, F-statistic = 3.18, LM test = 3.53 (p = 0.03), Reset test = 0.162 (p = 0.69)
- Short-run dynamic (equation (23)):
  - ∆tb_t = -0.73 EC_{t-1} - 2.47∆q^{CPI}_{t-1} - 1.63∆q^{CPI}_{t-2} + 2.66∆q^{CPI}_{t-3} - 1.34∆y^{D}_{t-1} - 3.09∆y^{D}_{t-2} - 0.04∆tb_{t-1} + 0.21∆tb_{t-2} - 0.14∆tb_{t-3} - 0.34∆tb_{t-4}
- CPI PSS impulse response after a permanent depreciation:
  - Initial deterioration by 10.1 percent in total.
  - Length of the adverse effect: four quarters.
  - New equilibrium established fifteen quarters after the shock.
  - New equilibrium level: 1.05 percent below the old equilibrium level.
- Summary: PPI and CPI PSS versions give opposite-signed long-run impulse outcomes despite evidence of cointegration.

### ARDL (IV) approach: long-run and impulse responses
- Long-run relationships:
  - PPI version (equation (24)): FDPPI = 66.3 log y - 54.3 log q + 00.1 log (notation preserved)
    - Adjusted R^2: 0.98, LM test: 1.72 (p = 0.28), Reset test: 0.76 (p = 0.41)
  - CPI version (equation (25)): FDCPI = 42.4 log y - 38.3 log q + 09.2 log (notation preserved)
    - Adjusted R^2: 0.96, LM test: 1.96 (p = 0.18), Reset test: 0.33 (p = 0.57)
- ARDL(IV) short-run impulse equations:
  - PPI (equation (26)):
    - ∆tb_t = -0.99 q^{PPI}_t - 0.41∆q^{PPI}_{t-1} - 2.00∆q^{PPI}_{t-1} - 0.83∆q^{PPI}_{t-2} + 2.13∆q^{PPI}_{t-3} - 3.53 y^{D}_t + 0.32∆y^{D}_t - 1.47∆y^{D}_{t-1} + 3.64 y^{F}_t - 1.76∆y^{F}_t - 11.63∆y^{F}_{t-1} + 0.34∆tb_t + 0.13∆tb_{t-1}
  - CPI (equation (27)):
    - ∆tb_t = 2.09 q^{CPI}_t - 2.61∆q^{CPI}_t - 1.21∆q^{CPI}_{t-1} - 3.39 y^{D}_t + 1.06∆y^{D}_t + 1.45∆y^{D}_{t-1} + 4.42 y^{F}_t - 6.73∆y^{F}_t - 9.63∆y^{F}_{t-1} + 0.62∆tb_t + 0.26∆tb_{t-1}
- ARDL(IV) impulse response findings:
  - Both PPI and CPI ARDL(IV) indicate signs of a J-curve effect.
  - Time to new trade balance equilibrium in both ARDL(IV) cases: six quarters.
  - Length of the adverse effect in both cases: one quarter.
  - Initial deterioration and new equilibrium magnitudes:
    - PPI: initial deterioration of 0.5 percent; new equilibrium 1.29 percent above the pre-depreciation equilibrium.
    - CPI: initial deterioration of 1.1 percent; new equilibrium 0.75 percent above the pre-depreciation equilibrium.
- Interpretation: Differences attributed to different trends in price indices; PPI inflation lower than CPI inflation leading to more pronounced PPI-based REER depreciation and lower trade flow response.

### Bewley ARDL approach: two-variable focus (trade balance and exchange rate)
- Bewley impulse equations:
  - PPI (equation (28)): tb_t = 1.39 q^{PPI}_t - 6.84∆q^{PPI}_{t-1} - 1.03∆q^{PPI}_{t-2} + 3.12∆q^{PPI}_{t-3} - 3.03∆q^{PPI}_{t-4} - 0.13 ∆tb_{t-1}
  - CPI (equation (29)): tb_t = 1.81 q^{CPI}_t - 1.61∆q^{CPI}_{t-1} - 0.63∆q^{CPI}_{t-2} + 1.77∆q^{CPI}_{t-3} - 1.51∆q^{CPI}_{t-4} - 0.29∆tb_{t-1} - 0.26∆tb_{t-2} - 0.02∆tb_{t-3} - 0.13∆tb_{t-4}
- Bewley model impulse outcomes:
  - PPI:
    - Delayed J-curve effect lasting two quarters.
    - Adverse impact on trade balance totaling 5.8 percent.
    - New equilibrium after eight quarters at 1.41 percent above old equilibrium.
  - CPI:
    - No support for a J-curve effect.
    - New equilibrium reached after 15 quarters at 1.9 percent above the old equilibrium.

### Elasticities: summary and averages (Table 4)
- Elasticity of the trade balance with respect to the Real Effective Exchange Rate (q), Domestic Income (yD) and Foreign Income (yF):
  - q (CPI based): Bewley 1.82, PSS 2.24, ARDL(IV) 2.09, average 2.05
  - q (PPI based): Bewley 1.39, PSS 1.96, ARDL(IV) 1.00, average 1.45
  - yD (CPI based): -2.70, -3.39, -3.05
  - yD (PPI based): -3.78, -3.54, -3.66
  - yF (CPI based): 2.94, 4.42, 3.68
  - yF (PPI based): 4.87, 3.66, 4.27
- Average long-run elasticity findings reported in text:
  - Long-run elasticity of the trade balance with respect to the exchange rate on average between 1.45 and 2.05.
  - Foreign income elasticity on average between 3.68 and 4.27.
  - Domestic income elasticity averaging between 3.05 and 3.66.
- Interpretation notes:
  - Elasticities reflect substitution effects and availability of domestic substitutes.
  - Elasticity of domestic income is lower than elasticity of foreign income.
  - Higher growth rates in Croatia compared to six largest trading partners likely imply a net negative impact on the trade balance.
  - Potential structural break from Croatia's full membership in CEFTA not directly captured.

### Summary of empirical results regarding permanent depreciation (Table 5 and text)
- Table 5 excerpts (preserving entries):
  - BEW-CPI: Relative position of new long-run equilibrium level 1.9, Time for new equilibrium in quarters 15, Length of J-curve effect 0, Extent of J-curve effect 0.00
  - BEW-PPI: Relative position 1.41, Time for new equilibrium 8, Length of J-curve effect 2, Extent of J-curve effect -5.77
  - PSS-CPI: Relative position -1.05, Time for new equilibrium 15, Length of J-curve effect 4, Extent of J-curve effect -10.10
  - PSS-PPI: Relative position 1.35, Time for new equilibrium 10, Length of J-curve effect 2, Extent of J-curve effect -2.49
  - ARDL(IV)-CPI: Relative position 1.29, Time for new equilibrium 6, Length of J-curve effect 1, Extent of J-curve effect -0.52
  - ARDL(IV)-PPI: Relative position 0.75, Time for new equilibrium 6, Length of J-curve effect 1, Extent of J-curve effect -1.11
  - total average: Relative position 0.94, Time for new equilibrium in quarters 10, Length of J-curve effect 1.67, Extent of J-curve effect -3.33
  - average without PSS_CPI: Relative position 1.349, Time for new equilibrium in quarters 1.20, Length of J-curve effect -1.98
- Text summary:
  - On average, a permanent 1 percent depreciation of the domestic currency results in a 0.94 percent improvement of the trade balance.
  - Ignoring the outlier (PSS with CPI), improvement amounts to 1.34 percent.
  - Average time to new equilibrium: approximately 2.5 years.
  - Average length of the adverse effect of permanent depreciation: moderately above one quarter.
  - Increase of the trade balance as a consequence of the negative J-curve effect estimated between 2.0 and 3.3 percent.
- Caveats:
  - Short time series and dynamic trade environment with many outliers suggest caution.
  - Single-equation approach limitations: potential multiple cointegrating vectors and weak exogeneity issues may lead to inefficiency or misleading estimates.

### Policy implications (Section VIII)
- Main policy-relevant findings:
  - Improvement of the trade balance after a permanent devaluation/depreciation is limited: a 1 percent permanent devaluation results in a new equilibrium level up to 1.34 percent above the old equilibrium (best case).
  - The initial adverse J-curve effect can be significant.
- Eight potential spillover/adverse effects of devaluation (as listed in source):
  1. Pass-through of the exchange rate on inflation can neutralize nominal depreciation and lead to real exchange rate appreciation.
  2. High dependence on imports for intermediate goods and raw materials can erode competitiveness gains from real depreciation.
  3. Increases in import and export prices may deteriorate domestic real income.
  4. Resource shift from non-tradable to tradable sector could create a wage gap, wage spiral, or net increase in unemployment given rigid labor markets.
  5. Level of indexed debt held by enterprises and households (high euroization) could amplify adverse consequences; distribution and structure of such debt matter.
  6. Strong, permanent depreciation could seriously impact public finances by increasing domestic currency cost of debt servicing, risking measures to combat a general government deficit around 5.5 percent of GDP.
  - (Source enumerates eight potential spillovers; the text elaborates primarily on six distinct adverse points above in the provided excerpt.)
- Policy judgment:
  - Given limited trade-balance gains and significant potential negative side effects, a permanent nominal depreciation in an asymmetric intervention policy or pegged exchange rate regime is unlikely to have a net favorable effect on the entire economy.

### Conclusion and suggestions for further research (Section IX)
- Key empirical conclusions:
  - On average, a permanent 1 percent depreciation yields an improvement of the trade balance between 0.94 percent and 1.3 percent.
  - New equilibrium established after approximately 2.5 years.
  - Average length of the adverse effect: moderately above one quarter.
  - Evidence of the J-curve effect in Croatia with an estimated increase in the trade balance deficit due to the J-curve between 2.0 percent and 3.3 percent.
  - Results are somewhat sensitive to inclusion of dummy variables and one model is an outlier; overall empirical consistency across methods with that exception.
  - Intuition suggests a permanent nominal depreciation is unlikely to produce net favorable economy-wide effects once side effects are considered.
- Suggested further research paths:
  - Estimate a VAR and/or VECM to account for feedback effects.
  - Model bilateral trade data to offset possible mismeasurement (e.g., shipbuilding trade data with Italy).
  - Use disaggregated data by standard international trade classification; separately treat shipbuilding and trade in oil derivatives.

*Source: author’s calculations (content unit: _wp0465 - references to a certain time period since the impulse should relate to a reaction of the)*

### REFERENCES

### _wp0465 - REFERENCES

### J-curve and trade balance studies
- Abeysinghe,  Tilak,  and  Tan  Lin  Yeok,  1998,  “Exchange  Rate  Appreciation  and  Export  Competitiveness. The Case of Singapore”, Applied Economics, Vol. 30, pp. 51 - 55
- Akbostanci,  Elif,  2002,  “Dynamics  of  the  Trade  Balance:  The  Turkish  J-curve”,  ERC Working Papers, 02/05
- Baharumshah,  Ahmad  Z.,  2001,  “The  Effect  of  Exchange  Rate  on  Bilateral  Trade  Balance:  New Evidence from Malaysia and Thailand”, Asian Economic Journal, Vol. 15, No. 3, pp. 291 - 311
- Bahmani-Oskoee,  Mohsen,  and  Alse,  Janardhanan,  1994,  “Short-Run  Versus  Long-Run  Effects  of  Devaluation:  Error-Correction  Modeling  and  Cointegration”,  Eastern Economic Journal, Vol. 20, No. 4, Fall 1994, pp. 453 - 464
- Bahmani-Oskoee,  Mohsen,  and  Taggert  J.  Brooks,  1999,  “Bilateral  J-curve  Between  U.S.  and Her Trading Partners”, Weltwirtschaftliches Archiv, Vol. 135(1), pp. 156 - 165
- Bahmani-Oskoee,  Mohsen,  2001,  “Nominal  and  Real  Effective  Exchange  Rates  of  Middle  Eastern  Countries  and  Their  Trade  Performance”,  Applied  Economics,  Vol.  33,  pp. 103 - 111
- Bahmani-Oskoee,  Mohsen,  and  Kantiapong,  Tatchawan,  2001,  “Bilateral  J-Curve  Between  Thailand and Her Trading Partners”, Journal of Economic Development, Vol. 26, No. 2, December, pp. 107 - 117
- Briguglio,  Lino,  1989,  “The  Impact  of  a  Devaluation  on  the  Maltese  Trade  Balance  With  Special  Reference  to  the  Price  And  Income  Reversal  Effects”,  Applied  Economics, Vol. 21, pp. 325 - 337
- Buluswar, Murli D., Henry Thomson, and Kamal P. Upadhyaya, 1996, “Devaluation and the Trade  Blance  in  India:  Stationarity  and  Cointegration”,  Applied  Economics, Vol. 28, pp. 429 - 432
- Doroodian,  Khosrow,  Jung,  Chulho,  and  Roy,  Boyd,  1999,  “The  J-Curve  Effect  and  US  Agricultural and Industrial Trade”, Applied Economics, Vol. 31, pp. 687 - 695
- Jung,  Chulho,  and  Khosrow,  Doroodian,  1998,  “The  Persistence  of  Japan's  Trade  Surplus”,  International Economic Journal, Vol. 12, Number 1, Spring 1998
- Kale,  Pelin,  2001,  “Turkey's  Trade  Balance  in  the  Short  And  The  Long  Run:  Error  Correction  Modeling  and  Cointegration”,  The  International  Trade  Journal,  Vol.  1,  no. 1, Spring, pp. 27-55
- Marwah, Kanta, and Lawrence R., Klein, 1996, “Estimation of J - Curves: United States and Canada”, Canadian Journal of Economics, Vol. 29, No. 3, August, pp. 523 - 539
- Meade,  E.  Ellen,  1988,  “Exchange  Rates,  Adjustment,  and  The  J-Curve”,  Federal  Reserve  Bulletin, October
- Rose,  Andrew,  and  Janet  L.,  Yellen,  1989,  “Is  There  a  J-curve?”,  Journal  of  Monetary  economics, 24, pp. 53-68
- Upadhyaya, Kamal P., and Dhakal, Dharmendra, 1997, “Devaluation and the Trade Balance: Estimating the Long Run Effect”, Applied Economics Letters, Vol. 4, pp. 343 - 345
- Wilson, Peter, 2001, “Exchange Rates and the Trade Balance for Dynamic Asian Economies -  Does  the  J-Curve  Exist  for  Singapore,  Malaysia,  and  Korea?”,  Open  Economies  Review, Vol. 12, pp. 389 - 413
- Wilson,  William  T.,  1993,  “J-curve  Effect  and  Exchange  Rate  Pass-Through:  An  Empirical  Investigation  of  the  United  States”,  The  International  Trade  Journal,  Vol.  7,  No.  4,  Summer, pp. 463 - 483

### Exchange rate pass-through, competitiveness, and prices
- Altermann,  Bill,  1997,  “Are  Producer  Prices  Good  Proxies  for  Export  Prices?”,  Monthly  Labor Review, October
- Junz,  Helen  B.,  and  Rudolf  R.,  Rhomberg,  1973,  “Prices  and  Export  Performance  of  Industrial Countries, 1953 – 63”, American Economic Review, Vol. 63, pp. 412 - 418
- Hooper,  Peter,  Johnson  Karen,  and  Jaime,  Marquez,  1998,  “Trade  Elasticities  for  G-7  Countries”, International   Finance   Discussion   Papers   Number   609,   Board   of   Governors of the Federal Reserve System, April
- Menon, Jayant, 1995, “Exchange Rate Pass-Through”, Journal of Economic Survey, Vol. 9, No. 2, pp. 197 - 231
- Marsh, Ian W., and Stephen P., Tokarick, 1994, “Competitiveness Indicators: A Theoretical and Empirical Assessment", IMF Working Paper, WP/94/29
- Sinha,  Dipendra,  2001,  “A  Note  on  Trade  Elasticities  in  Asian  Countries”,  International Trade Journal, Vol. 15, No. 2, Summer, pp. 221 - 237
- Marquez,  Jaime,  1994,  “The  Econometrics  of  Elasticities  or  the  Elasticity  of  Econometrics:  an  Empirical  Analysis  of  the  Behavior  of  U.S.  Imports”,  The  Review  of  Economics  and Statistics, pp. 471-481
- Marquez,  Jaime,  1999,  “Long-Period  Trade  Elasticities  for  Canada,  Japan  and  the  United States”, Review of International Economics, Vol. 7(1), pp. 102 - 116
- Hacker,  R.  Scott,  and  Abudlnasser,  Hatemi-J,  2002,  “The  Effect  of  Real  Exchange  Rate  Changes  on  Trade  Balances  in  the  Short  and  Long  Run:  Evidence  From  German  Trade  With  Transitional  European  Economies”,  www.ihh.hj.se/ersasise2002/papers/Hacker_and%20Hatermi_J_Emerging_Markets.pdf

### Econometric methods, cointegration, unit roots, and exogeneity
- Aigner,   Dennis   J.,   and   Stephen   M.   Goldfeld,   1974,   Estimation   and   Prediction   from   Aggregate   Data   When   Agregates   Are   Measured   More   Accurately   Then   Their   Components, Econometrica, Vol. 42, pp. 113 - 134
- Bewley,  Ronald,  1979,  “The  Direct  Estimation  of  The  Equilibrium  Response  in  a  Linear  Dynamic Model”, Economics Letters 3, pp. 357 - 361
- Bewley,  Ronald,  and  Denzil,  Fiebig,  ,  1990,  “Why  Are  Long-Run  Parameter  Estimates  So  Disparate?”, The Review of Economics and Statistics, pp. 345 - 349
- Enders, Walter, 1995, “Applied Econometric Time Series”, Wiley, New York
- Engle,   Robert   F.,   David   F.   Hendry   and   Jean-Francois   Richard,   1983,   “Exogeneity”,   Econometrica, 51, 2, pp. 277 - 304
- Ericsson, R. Neil, and John S. Irons, 1994, “Testing Exogeneity”, Oxford University Press
- Grunfeld,  Y.,  and  Zvi,  Griliches,  1960,  “Is  Aggregation  Necessarily  Bad?”,  Review  of  Economics and Statistics, Vol. 42, pp. 1-13
- Kremers, Jeroen J., Neil R. Ericsson, and Juan J. Dolado, 1992, “The Power of Cointegration Tests”, Oxford Bulletin of Economics and Statistics, 54, 3
- Pesaran,   Hashem,   and   Yongcheol,   Shin,   1997,   “An   Autoregressive   Distributed   Lag   Modelling Approach to Cointegration Analysis”, DAE Working Papers Amalgamated Series, University of Cambridge
- Pesaran, Hashem, Yongcheol, Shin, and Ron J. Smith, 1996, “Testing for the Existence of a Long-Run Relationship”, DAE Working Papers Amalgamated Series, no. 9622, University of Cambridge
- Schwert, G. William, 1989, “Tests for Unit Roots: A Monte Carlo Investigation”, Journal of Business and Economic Statistics, April 1989, Vol. 7 Issue 2, pp. 147 - 160
- Urbain,  Jean-Pierre,  1992,  “On  Weak  Exogeneity  in  Error  Correction  Models”,  Oxford Bulletin of Economics and Statistics, 54, 2, pp. 187 - 207
- Ericsson, R. Neil, and John S. Irons, 1994, “Testing Exogeneity”, Oxford University Press

### Miscellaneous empirical and sectoral studies
- Arize, Augustine, 1987, “The Supply and Demand for Imports and Exports in a Simultaneous Model”, Applied Economics, Vol. 19, pp. 1233 - 1247
- Gupta-Kapoor, Anju, and Ramakrishnan Uma, 1999, “Is There a J-curve? A New Estimation for Japan”, International Economic Journal, Vol. 13, No. 4, Winter, pp. 71 - 79
- Hooper,  Peter,  Johnson  Karen,  and  Jaime,  Marquez,  1998,  “Trade  Elasticities  for  G-7  Countries”, International Finance Discussion Papers Number 609, Board of Governors of the Federal Reserve System, April
- Wilson,  William  T.,  1993,  “J-curve  Effect  and  Exchange  Rate  Pass-Through:  An  Empirical  Investigation  of  the  United  States”,  The  International  Trade  Journal,  Vol.  7,  No.  4,  Summer, pp. 463 - 483

*References list as provided in _wp0465 - REFERENCES*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp0465.pdf_