Journal Issue
Share
Article

Republic of Estonia: Selected Issues

Author(s):
International Monetary Fund
Published Date:
March 2009
Share
  • ShareShare
Show Summary Details

II. Assessment of Current Account Stability1

A. Overview and Key Conclusions

1. This chapter assesses current account stability in Estonia. It focuses on measuring current account disequilibrium, defined as the gap between the underlying current account (the current account stripped of temporary factors) and the equilibrium current account (a balance that leads net foreign assets to evolve in a manner consistent with the economy’s fundamentals). It also draws implications for real exchange rate misalignment.

2. Econometric estimates based on CGER methodologies, suggest that as of October 2008 there was a current account gap of 1½-4¼ percent of GDP, which corresponds to a real exchange rate overvaluation of 3-9 percent. Direct estimates of overvaluation point to a somewhat higher overvaluation, in the order of 16-19 percent.

B. Equilibrium and Underlying Current Account Measures

3. Since estimates of the equilibrium and underlying current accounts are subject to significant uncertainty, we use several complementary approaches to establish a range of values.

Equilibrium Current Account

Macrobalance Approach

4. The macrobalance approach uses coefficients from cross-country panel regressions to calculate the current account balance that is consistent with medium-term fundamentals—captured by the independent variables in the regression. These variables include demographic features, growth, fiscal positions and net foreign liabilities—all chosen for their likely influence on the savings-investment balance.2Table 1 shows four alternative estimates of Estonia’s equilibrium current account deficit based on this approach. These estimates cover a wide range—from 2.3 to 11.6 percent of GDP.3

Table 1.Equilibrium Medium-Term Current Account Balance Macrobalance Approach
VariablesRegressorsCoefficientsContributions to equilibrium current account
Pooled estimationHybrid pooled estimationFixed effects estimationPooled estimationHybrid pooled estimationFixed effects estimation
(IMF, 2008)(Rahman, 2008)(IMF, 2008)(Rahman, 2008)
Fiscal balance (% of GDP)1/−3.90.200.220.190.32−0.8−0.8−0.7−1.2
Old−age dependency (ratio)1/−0.3−0.140.04−0.12−0.230.00.00.00.1
Population growth2/−0.5−1.21−0.63−1.03−0.470.60.30.50.2
Initial net foreign assets (% of GDP)3/−77.00.020.03−1.5−2.2
Lagged CA balance to GDP (%GDP)4/−12.40.37−4.5
Oil Balance (%GDP)−3.80.230.440.170.31−0.9−1.7−0.6−1.2
Output growth2/2.8−0.21−0.18−0.16−0.27−0.6−0.5−0.4−0.7
Relative Income5/45.90.02−0.010.020.9−0.30.90.0
Investment climate400.0−0.01−4.0
Fixed effect6/100.0−0.09−8.8
Constant100.00.000.020.000.02.0−0.30.0
Equilibrium Current Account Balance (percent of GDP)−2.3−7.2−5.2−11.6
Source: IMF staff calculations. Coefficients taken from IMF (2008) and Rahman (2008).

End-term projections (i.e., for 2014) of all variables are used as regressors with the exception of those flow variables thatenter as lagged variables (for which the projected average for the period 2006-09 is used) and those variables that enterin the form of initial stocks (for which the end -September 2008 value is used)

Relative to trading partners

End-September 2008 ratio.

Average for the period 2006-09.

Relative to the US (end-2007)

Calculated such that the average prediction error for the period 1996-2007 is zero.

Source: IMF staff calculations. Coefficients taken from IMF (2008) and Rahman (2008).

End-term projections (i.e., for 2014) of all variables are used as regressors with the exception of those flow variables thatenter as lagged variables (for which the projected average for the period 2006-09 is used) and those variables that enterin the form of initial stocks (for which the end -September 2008 value is used)

Relative to trading partners

End-September 2008 ratio.

Average for the period 2006-09.

Relative to the US (end-2007)

Calculated such that the average prediction error for the period 1996-2007 is zero.

5. In the rest of this note, we focus on the results obtained in the hybrid pooled estimation and in Rahman (2008), which give us an average equilibrium current account deficit of approximately 6.2 percent of GDP. The other two approaches have features that are problematic. The pooled estimation in Lee et al (2008) omits factors specific to Eastern European countries, such as investment climate (as captured by Rahman, 2008) or the initially low capital stock (which creates persistence in the current account deficit and may be captured by the lagged current account deficit). And the fixed effects estimation is dominated by the fixed-effect term, making it difficult to interpret.4

External Sustainability Approach

6. The external sustainability approach calculates the current account balance that would be consistent with a specified evolution of the net foreign assets position. A natural benchmark is the deficit that stabilizes net foreign assets (NFA) as a share of GDP.5

7. The NFA-stabilizing balance (cabS) can be calculated from equation (1), where g is real GDP growth, π is the inflation rate (GDP deflator), and nfa is NFA as a share of GDP.

8. Assuming real GDP growth of 5 percent (staff’s medium-term growth projection), and a rate of change in prices of 2.25 percent (in line with projected medium-term increase in the rate of change of Euro-denominated foreign prices), implies that sustaining Estonia’s net foreign assets position (-77 percent of GDP, as of end-September 2008) requires running a current account deficit of approximately 5.2 percent of GDP.6

Underlying Current Account

9. We use two alternative methodologies to estimate the underlying current account, the projections-based method and the elasticities-based method. The first method equates the underlying current account with staff’s medium-term current account projections (which are part of staff’s macroframework). The second method uses elasticities estimated in earlier CGER empirical work (Isard and Faruqee, 1998, Chapter V) to calculate the underlying current account by stripping out cyclical influences on the actual current account and adding in the lagged impact of the past appreciation of the real exchange rate.

10. The projections-based method suggests an underlying current account balance of minus 9.1 percent of GDP while the elasticities-based method suggests a lower underlying balance (-10.9 percent of GDP). We will come back to the differences in the estimates in section D. At this point, however, it is convenient to highlight that the figure obtained in the elasticities-based method is the results of two large and opposite effects (Table 2): On the one hand, the closing of the output gap is estimated to reduce the current account deficit by around 4 percentage points.7 On the other hand, the impact of past real exchange rate appreciation is estimated to increase the deficit by approximately the same magnitude.8

Table 2.Deriving the Underlying Current Account Balance Under the Elasticities-Based Method
ElasticityDataAdjustment (in percent of GDP)
Current account balance (12 months ending in October 2008) (A)−9.9
Temporary factors (B)−1.0
One−off factors0.0
Estonian business cycle (output gap) 1/1.53.03.7
Export partners' business cycle1.50.6−0.7
Real exchange rate movements−4.1
20080.60−0.50.2
20070.256.7−3.4
20060.154.6−0.9
Underlying current account balance (A+B)−10.9
Memorandum items:
Exports elasticity to the real exchange rate0.71
Imports elasticity to the real exchangerate0.92
Exports to GDP72.8
Imports to GDP82.4

Methodology: Isard and Faruqee (1998), OP167, Chapter VSource: IMF staff calculations.

Output gap measured as the average gap obtained from the Hodrick-Prescottfilter and the production function approach.

Methodology: Isard and Faruqee (1998), OP167, Chapter VSource: IMF staff calculations.

Output gap measured as the average gap obtained from the Hodrick-Prescottfilter and the production function approach.

Current Account Gap and the Assessment of the Real Exchange Rate

11. Table 3 derives the estimated current account gap for each of the approaches described above and shows the corresponding real exchange rate misalignment, based on a real exchange rate elasticity of the current account calculated using the coefficients in Isard and Faruquee (1998). The calculation of the current account gap takes into account the projected medium-term capital transfers from the European Union, which are a mitigating factor of the current account imbalance.

Table 3.Current Account Stability. Summary Table(percent of GDP unless otherwise indicated)
MB1/ES2/
Equilibrium CA (A)−6.2−5.2
Underlying current account (B)
Projections-based method−9.1−9.1
Elasticities-based method3/−10.9−10.9
CA gap (C=A-B)
Projections-based method2.93.9
Elasticities-based method3/4.75.7
Mitigating factor: Capital Transfers (I)4/1.51.5
CA gap net of mitigating factor (J=HI)
Projections−based method1.42.4
Elasticities-based method3/3.24.2
Overvaluation of the real exchange rate5/
Projections-based method3.15.3
Elasticities-based method3/7.29.4
Source: IMF staff calculations.

Average of the results obtained in the hybrid pooled estimation and in Rahman (2008)

Stabilizes net foreign assets at the September-2008 ratio (-77 percent of GDP)

Calculated using coefficients from Isard and Faruquee (1998, chapter V)

Medium-term projected transfers.

In percent. The elasticity of net exports to the real exchange rate is set at -0.45 based on the coefficients in Isard and Faruquee (1998, chapter V)

Source: IMF staff calculations.

Average of the results obtained in the hybrid pooled estimation and in Rahman (2008)

Stabilizes net foreign assets at the September-2008 ratio (-77 percent of GDP)

Calculated using coefficients from Isard and Faruquee (1998, chapter V)

Medium-term projected transfers.

In percent. The elasticity of net exports to the real exchange rate is set at -0.45 based on the coefficients in Isard and Faruquee (1998, chapter V)

C. Direct Estimates of the Misalignment of the Real Exchange Rate

12. The results described in the previous section can be complemented with a direct estimate of the equilibrium real exchange rate (ERER). We follow two complementary methods:

  • The first method estimates the ERER using coefficients estimated in cross-country regressions. The application of this approach was problematic since the time series is short (see footnote 19) and since Estonia is out of the sample—thus it was necessary to infer the key country “fixed effect” from the data and the other estimated coefficients. The coefficients are discussed in Lee et al (2008), where corrections have been made to account for effects specific to countries in Central and Eastern Europe.9
  • The second method estimates the ERER as an average of historical observations of the real exchange rate. Observations from the early years of Estonia’s transition may not very relevant given the substantive transformations the economy has undergone. Therefore, we use the average of the last five years to estimate the ERER (we look both at the CPI-based real exchange rate and at the unit labor costs-based real exchange rate).

13. The results (Table 4) show an overvaluation of Estonia’s real exchange rate of approximately 16-19 percent. The first method suggests that the main drivers of the ERER are: the fixed-effects coefficient, the terms of trade index, and the productivity gap (the latter does not have a large contribution in the result shown in the table, but has an important role in explaining the appreciation during 1996–2003).

Table 4.Equilibrium Real Exchange Rate Approach
First Method (estimation method)1/
Second Method (averages method)2/
RegressorsCEE coefficients3/Contribution to equilibrium real exchange rate
Productivity gap4/0.011.420.01
NFA (as share of average exports and imports)−1.340.04−0.05
Government consumption (as share of GDP)0.170.000.00
Trade restrictions index (Sachs and Warner, 1995)0.000.140.00
Share of administered prices5/2.00−0.02−0.04
Commodity terms of trade index (logarithm)6/4.610.391.80
Country-specific constant parameter (fixed effect)7/1.002.952.95
Logarithm of equilibrium real exchange rate (A)4.67
Logarithm of actual real exchange rate (B)4.86
Estimated overvaluation (B-A) (in percent)18.615.5
Source: IMF staff calculations. CPI-based REER series is from IMF REER database. ULC-based REER series is from the European Commission (Economic and Financial Affairs).

End-term projections (i.e., for 2014) of all variables are used as regressors. Projections of the productivity gap equal to the value observed in 2006.

Average deviation (of CPI-based REER and ULC-based REER) from their past five-year historical averages.

CEE countries are Poland, Slovak Republic, Czech Republic, Hungary, and Slovenia

Logarithm of productivity of tradables minus logarithm of productivity of nontradables

Measure of regulated prices, EBRD

Base year is 2005 (value in 2005 is 4.60)

Calculated such that the average prediction error (i.e., the average misalignment) for the period 1996-2007 is zero

Source: IMF staff calculations. CPI-based REER series is from IMF REER database. ULC-based REER series is from the European Commission (Economic and Financial Affairs).

End-term projections (i.e., for 2014) of all variables are used as regressors. Projections of the productivity gap equal to the value observed in 2006.

Average deviation (of CPI-based REER and ULC-based REER) from their past five-year historical averages.

CEE countries are Poland, Slovak Republic, Czech Republic, Hungary, and Slovenia

Logarithm of productivity of tradables minus logarithm of productivity of nontradables

Measure of regulated prices, EBRD

Base year is 2005 (value in 2005 is 4.60)

Calculated such that the average prediction error (i.e., the average misalignment) for the period 1996-2007 is zero

D. Assessment of the Results

14. The results reported above provide a benchmark for assessing Estonia’s current account stability. However, it is important to highlight some caveats that suggest that some methods could be overstating the size of the current account gaps and the degree of overvaluation in a rapidly converging economy with overheated demand.

15. The degree of overvaluation calculated using the equilibrium exchange rate approach could be overestimated. First, the terms of trade series does not capture the improvements in terms of trade that Estonia has had in the last two years. If a terms of trade series calculated from export and import price deflators is used instead, the estimated overvaluation using the first method falls from 18.6 percent to 15 percent. Second, firms’ profitability remains high, suggesting that increases in labor productivity and the capacity to charge higher prices on account of quality improvements have sustained firms’ competitiveness in spite of higher wages (Figure 8 of the staff report).

16. The underlying current account deficit estimated using the elasticities-based approach could be overestimated. This is illustrated by a comparison between the actual current account and the current account forecasted by the elasticities-based approach.

  • Figure 1 illustrates that the elasticities-based approach tends to overestimate the size of the current account deficit and the overestimation seems to be related to the contribution of the component capturing the impact of past exchange rate changes. This is perhaps due to the coefficients used for the adjustment for past changes of the real exchange rate being estimated in a sample of countries whose real exchange rate was stationary (so past changes in the real exchange rate were likely measuring deviations from equilibrium). Estonia, however, has had a secular appreciation of the real exchange rate, as expected along an equilibrium path of a converging economy. Therefore, past changes of the real exchange rate also reflect changes in productivity and were not likely to lead to a future deterioration of the current account.10

Figure 1.Estonia: Forecasting the Current Account Deficit Using the Elasticities Method (percent of GDP) 1/

(Year-on-year percent change)

Source: Bank of Estonia (actual data) and IMF staff estimates (forecast)

1/ Data for 2008 corresponds to the 12 months ending in October 2008.

  • Figure 2 illustrates that the elasticities-based approach performs better (i.e., with no bias) in Estonia when we restrict our attention to the cyclical component (i.e., we exclude the impact of the past changes of the REER).

Figure 2.Estonia: Actual and Estimated Current Account Deficit (percent of GDP) 1/

(Year-on-year percent change)

Source: Bank of Estonia (actual data) and IMF staff estimates.

1/ Data for 2008 corresponds to the 12 months ending in October 2008.

E. Conclusions

17. Econometric estimates based on CGER methodologies, suggest that as of October 2008 there was a current account gap of 1½-4¼ percent of GDP, which corresponds to a real exchange rate overvaluation of 3–9 percent. Direct estimates of overvaluation point to a somewhat higher overvaluation, in the order of 16–19 percent.

18. The range estimated for the current account gap is primarily driven by the range of estimations of the underlying current account. Estimations of the latter variable using elasticities obtained from previous CGER work seem to overestimate the magnitude of the underlying current account deficit.

REFERENCES

    LeeJaewooGian MariaMilesi-FerrettiJonathanOstryAlessandroPrati and Luca AntonioRicci.(2008). Exchange Rate Assessments: CGER Methodologies IMF Occasional Paper 261

    IsardPeter and HamidFaruqee(1998). Exchange Rate Assessment: Extension of the Macrobalance Approach IMF Occasional Paper 167 chapter V.

    RahmanJesmin(2008). Current Account Developments in New Member States of the European Union: Equilibrium Excess and Euphoria.IMF WP/08/92.

    • Search Google Scholar
    • Export Citation
1

Prepared by Pedro Rodriguez.

2

Lee et al. (2008) and Rahman (2008) describe the methodology and coefficients.

3

The results associated with Rahman’s (2008) differ somewhat from those reported in her paper because of differences in the value of the regressors, particularly of the NFA position (which is lower in the paper).

4

In short samples (such as those available for Estonia), the fixed effect tends to be significantly influenced by the historical realizations of the dependent variable (see Lee et al [2008], page 5).

5

While this could have been a poor benchmark for Estonia fifteen years ago when the country was initiating its convergence, it is a more useful benchmark now given the large and negative NFA position.

6

The assumption on the rate of change in prices could be seen as conservative given Estonia’s currently high inflation. However, assuming a higher inflation rate (e.g. 2.7 percent as in staff’s baseline scenario), would barely change the result.

7

This would be achieved through lower imports associated with lower consumption and investment, and possibly also a shift of resources from the nontradable sector (e.g., construction) to the tradable sector.

8

Isard and Faruquee (1998) obtain that changes in the real exchange rate tend to be passed to the current account over a three year period.

9

The adjustments made in Lee et al (2008, chapter 3) imply a higher coefficient for the productivity differential and a lower (actually zero) coefficient for the impact of government consumption.

10

This analytical issue was pointed out by the Estonian authorities in the technical discussions.

Other Resources Citing This Publication