## _cr08375

## Source details

**Canonical URL:** [_cr08375](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2008/_cr08375.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2008/_cr08375.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2008/_cr08375.pdf.json)

---

### Executive summary and scope
- Document: REPUBLIC OF ARMENIA — Selected Issues (Selected Issues Paper supporting the Staff Report).
- Timeframes and samples:
  - Fiscal VAR analysis period: 1998:1–2007:4.
  - Foreign exchange intervention empirical sample: January 3, 2001–May 30, 2008.
- High-level conclusions:
  - Building up fiscal space during expansions is key for effective stimulatory policies during downturns.
  - Armenia’s underlying fiscal balance started to deteriorate in 2004 during rapid economic growth.
  - CBA foreign exchange market interventions (2001–2008) show limited effectiveness in changing the exchange rate level and only tentative effects in reducing volatility.
  - Dram appreciation has removed prior undervaluation; real exchange rate is near or possibly slightly above equilibrium.
  - Pass-through estimates show significant downward rigidity in import prices in response to exchange rate movements, consistent with concerns about limited competition in key import sectors.

### Chapter I — Enhancing fiscal policy: framework, measurement, and evidence
- Recommended fiscal stance measure:
  - Underlying fiscal balance = overall balance excluding grants and external interest payments and including all subsidies (including the gas subsidy that existed until May 2008).
  - Coverage ideally: central government, state fund for social insurance (consolidated), local governments, state-owned enterprises, and off-budgetary accounts. Government spending directed to imports should be excluded where feasible.
- Institutional framework and risks:
  - Current framework: three-year medium-term expenditure framework (MTEF) plus a debt management law with elements of a fiscal responsibility law.
  - Explicit rules: ceiling on public debt at 60 percent of GDP; constraints on the overall balance when debt is above 50 percent of GDP.
  - Public debt judged low with no short-term risk to fiscal sustainability; medium-term fiscal risks should be routinely assessed and quantified.
- Fiscal stance diagnostics (1998–2007):
  - Using overall balance vs. underlying balance produces different diagnostics; off-budget gas subsidy introduced in 2006 distorts the overall balance.
  - Key finding: fiscal policy has been pro-cyclical in recent years — underlying and cyclically-adjusted balances deteriorated since 2004 while actual output grew above potential.
- VAR-based fiscal impulse results (spending shocks):
  - Identification: focus on spending shocks (tax shocks left unidentified); government spending treated as predetermined at quarterly frequency.
  - Main findings:
    - Spending shocks appear expansionary and inflationary.
    - Prices take about two quarters to react (response not statistically significant).
    - A 1 percent increase in real per capita government spending can lead to up to 0.5 percent increase in per capita GDP (short-run estimate).
    - Expansionary spending is associated with increased employment (short-term, small magnitude), reduced unemployment (insignificant), higher private consumption and private investment (complementarity), higher manufacturing activity, and a widening trade deficit (large, persistent, statistically significant for at least one year).
  - Caveats: high estimation uncertainty and data quality issues (informality, under-reporting).

### Chapter I — Cyclically-adjusted balance: methodology, parameters, and results
- Potential output estimation:
  - Method: Hodrick-Prescott (HP) filter on log GDP, annual data 1994–2007; HP smoothing parameter λ = 100.
  - Finding: Actual output below potential before 2003; growth exceeded potential since 2001 (except 2004); actual output above potential since 2005.
- Elasticity assumptions used in cyclically-adjustment:
  - Tax revenue elasticity (η): empirical point estimate 0.47 (from regression of quarterly changes in real tax receipts on lead and lags of changes in real output).
    - Feasible range considered: minimum 0, maximum around 2.
  - Government current spending elasticity (ξ): baseline chosen ξ = -0.1 given weak automatic stabilizers; range considered ξ ∈ (-0.1, -0.8). Regression returned -0.49 but deemed unrealistically high.
- Illustrative implications:
  - If tax elasticity close to 2, negative revenue gap when actual GDP below potential could reach 2.5 percentage points of GDP.
  - Cyclical temporary component of revenues in 2007 could be as large as 1.5 percentage points of GDP.
  - With weak automatic stabilizers, cyclically-adjusted expenditure should not deviate much from actual expenditure unless ξ relatively large.
- Cyclically-adjusted balance results:
  - Baseline (η = 0.47, ξ = -0.1): cyclical factors exacerbated overall fiscal deficit during 1999–2004; by 2004 cyclically-adjusted deficit close to 2 percent of GDP; since 2004 cyclically-adjusted balance deteriorated faster than overall balance while output above potential; by 2006–2007 cyclical factors helped reduce the underlying deficit.
  - Sensitivity: gap magnitude mostly determined by revenue elasticity variability. Alternative combinations (η = 0.25 or 1, ξ = -0.1 or -0.2) examined for robustness.

### Chapter I — Institutional recommendation: macro-fiscal unit
- Rationale: strengthen fiscal framework, forecasting, and monitoring; provide strategic multi-year coordination.
- Suggested core functions:
  - Develop fiscal framework: forecasting, analysis, monitoring short- and medium-term fiscal developments.
  - Monitor implementation: assess budget execution and links to growth and inflation.
  - Protect the fiscal framework: advise on emerging near- and medium-term risks; flag corrective actions; identify policy alternatives.
- Value-added tasks:
  - Ensure internally-consistent macroeconomic framework.
  - Perform fiscal modeling, forecasting, impact analysis, and scenario/stress testing.
  - Align fiscal policy with government objectives (macroeconomic stability, medium-term growth, poverty reduction, fiscal and debt sustainability).
  - Monitor in-year outcomes and coordinate corrective policy design.

### Chapter I — Policy guidance
- Short-run guidance:
  - Fiscal stance should become more neutral going forward.
  - If signs of overheating persist, fiscal stance needs tightening to allow active stabilization.
  - If the international downturn significantly impacts Armenia, tighter fiscal stance may not be desirable.
- Overarching recommendation: build up fiscal space during expansions to enable stimulatory policies during downturns.

### Chapter II — Foreign exchange market intervention: objectives, structure, and empirical evaluation
- CBA objectives and regime:
  - CBA maintained a flexible exchange rate while intervening to “moderate the rate of change and prevent undue fluctuations” rather than establishing a level.
  - Exchange rate regime reclassification in September 2006: from “independently floating” to “managed float with no predetermined path or level for the exchange rate.”
- Transmission channels:
  - Monetary policy channel main path (interventions largely unsterilized and directly impact reserve money).
  - Signaling channel may operate; microstructure (order flow) channel possibly important but intraday effects not testable with available data.
- Market structure and ARMEX volumes:
  - Average daily trade volume: around $3.2 million in 2007/08.
  - Trading days with volume not exceeding $1 million: almost half of trading days.
  - Trading days exceeding $5 million: about 15 percent of trading days.
  - On active trading days: 20–25 trades; average contract size close to $100,000.
  - Wholesale market averaged approximately 20 percent of total turnover (range 0–50 percent).
- Intervention magnitudes and liquidity management:
  - Net issuance of CBA securities for liquidity management: around AMD28 billion in 2006 and AMD10 billion in 2007.
  - Net liquidity injections from foreign exchange market intervention: AMD95 billion in 2006 and AMD117 billion in 2007.
  - 2001–2003: CBA net foreign exchange sales amounted to between 15 and 20 percent of beginning-of-year reserve money.
  - From 2004, interventions turned to net purchases and increased significantly until 2007, when they exceeded 90 percent of end-2006 reserve money.
  - In 2008 intervention activity moderated and returned to net foreign exchange sales.

### Chapter II — Empirical strategy and data
- Two-stage identification to address endogeneity:
  - Stage 1: TOBIT dynamic censored regression to estimate CBA intervention reaction function for purchases and sales separately.
  - Stage 2: Use predicted interventions (P^*_t, S^*_t) as instruments in GARCH(1,1) models for exchange rate returns (Δ ln s_t) and volatility.
- Controls and proxies:
  - Δ ln e_t: daily Euro/dollar nominal exchange rate return.
  - Δ(i - r^*)_{t-1}: first-differenced interest differential (domestic interbank rate minus LIBOR).
  - ΔOIL_t: first differenced monthly moving average of one-year lagged annual oil price changes (proxy for FX inflows).
- Summary statistics (selected):
  - AMD/$ exchange rate: Mean 483.315; Median 513.735; Maximum 591.763; Minimum 301.593; Std. Dev. 3.816.
  - AMD daily exchange rate return (Δ(ln s_t)): Mean 0.000; Std. Dev. 0.003; Maximum 0.034; Minimum -0.027.
  - Foreign exchange purchases (I_P_t) (in million $): Mean 0.493; Median 0.000; Maximum 25.690; Std. Dev. 1.806.
  - Foreign exchange sales (I_S_t) (in million $): Mean 0.182; Median 0.000; Maximum 13.425; Std. Dev. 0.884.
  - Sample period: January 3, 2001 to May 30, 2008.

### Chapter II — Empirical findings and interpretation
- First stage (policy reaction function) key findings:
  - Exchange rate appreciation (depreciation) triggers foreign exchange purchases (sales).
  - Significant dependence on deviations from a 20-day moving average for purchases.
  - Strong intervention clustering: one-day persistence in interventions.
- Second stage (GARCH) key findings:
  - Limited effectiveness of interventions on exchange rate returns (level): intervention coefficients in mean equation small and often not in theoretically expected direction; purchases associated with appreciation (negative I^P coefficient) consistent with simultaneity—interventions tend to follow movements.
  - Limited evidence that interventions reduce volatility:
    - Foreign exchange sales associated with lower exchange rate volatility (variance equation negative and significant for I^S in some specifications).
    - Purchases show little or no volatility-reducing effect and in the recent subperiod may be associated with slightly higher volatility.
  - Control variables behave as expected: autoregressive term positive and significant; relative dollar strength depreciates domestic currency; oil proxy associated with appreciation; interest differentials not significant.
- Robustness:
  - Results robust to alternative samples, GARCH variants, control variables, and lag lengths.
- Caveats and data limitations:
  - Interventions only observed at daily frequency; intraday stabilizing effects in a thin market may not be captured.
  - Simultaneity can bias coefficients downward despite two-stage approach; interpretation requires caution.

### Chapter II — Policy implications and consistency with monetary objectives
- Identified inconsistency (2005–2007):
  - Large unsterilized foreign exchange purchases injected domestic liquidity, blunting the interest rate channel and fueling money and credit growth.
  - Surplus liquidity delayed commercial bank lending rate increases until the third quarter of 2008 despite earlier CBA repo rate hikes.
  - Unsterilized interventions likely contributed to credit growth and domestic demand; transmission lag from money growth to inflation about one year suggests additional inflation pressures could arise.
- Resolution in 2008:
  - FX sales and interest rate increases aligned, resulting in a tightening of the monetary policy stance.
- Trade-offs:
  - Intervening to smooth exchange rate movements can conflict with inflation targeting if interventions are unsterilized and large relative to reserve money or market depth.

### Chapter III — Exchange rate appreciation, pass-through, competitiveness, and external sustainability
- Exchange rate assessment:
  - Dram appreciation removed previous undervaluation; real exchange rate near or possibly slightly above equilibrium in 2007 across methods (PPP, BEER, External Sustainability, Macroeconomic Balance).
- External Sustainability (ES) approach assumptions and outcomes:
  - NFA/GDP benchmark: NFA/GDP as of end-2006 = -24.2 percent.
  - Inflation rate of NFA: 2.5 percent.
  - Benchmark GDP growth rate: 6 percent; alternative scenarios: low-growth 3 percent; high-growth 9 percent.
  - Current account deficits required to stabilize NFA/GDP:
    - benchmark growth: -1.9 percent
    - low growth: -1.3 percent
    - high growth: -2.5 percent
  - ES results broadly consistent with PPP and BEER: no undervaluation; possible slight overvaluation in 2007.
- Macroeconomic Balance (MB) approach:
  - Six estimates of current account “norm” range from about a deficit of 2 percent of GDP to a deficit of 14 percent of GDP; underlying current account estimated at a deficit of 5 percent of GDP in 2006. Actual current account lies within estimated range, suggesting exchange rate near equilibrium.
- Exchange rate pass-through and domestic competition:
  - VAR-based pass-through (May 2005–August 2007) findings:
    - Import price responses are much larger for exchange rate depreciations than for appreciations; only depreciation responses statistically significant.
    - Pass-through about three times larger for depreciations than for appreciations.
    - Speed of adjustment: import prices adjust over about 3 months.
    - Interpretation: significant downward rigidity in import prices to appreciation; transportation costs unlikely to explain asymmetry; evidence points to monopolistic practices in key import sectors.
- Structural competitiveness and policy priorities:
  - Continued appreciation implies competitiveness preservation requires productivity gains via business environment improvements.
  - Key reform priorities:
    - Improve the business environment to raise productivity.
    - Improve tax and customs administration.
    - Deepen financial intermediation.
    - Reduce corruption.
    - Discontinue monopolistic practices in the import sector.
  - Competitiveness indicators: Armenia ranked 93 in the Global Competitiveness Index 2007-2008; weaknesses in infrastructure, institutions, and financial market sophistication.

### Overall policy conclusions
- Fiscal policy:
  - Build fiscal space during expansions to enable counter-cyclical fiscal responses in downturns.
  - Move fiscal stance toward neutrality; tighten if overheating signs persist; provide stimulus if international downturn significantly affects Armenia.
  - Institutionalize macro-fiscal analysis (dedicated macro-fiscal unit) and incorporate cyclically-adjusted indicators and fiscal impulse measures into budget processes.
- Monetary and exchange rate policy:
  - Unsterilized large-scale FX interventions can conflict with inflation-targeting; sterilization capability and consistent liquidity management are critical.
  - FX sales may modestly reduce exchange rate volatility, but interventions generally do not shift exchange rate level materially given endogeneity and market structure.
- Competitiveness and external sustainability:
  - Dram appreciation has reduced undervaluation concerns but raises the need for structural reforms to sustain competitiveness and mitigate medium-term vulnerabilities.
  - Address import-sector market structure to reduce pricing rigidities and improve pass-through dynamics.

*Source: IMF staff analysis (text and tables provided in the content unit).*

### 2008. The views expressed in this document are those of the staff team and do not necessarily reflect

### REPUBLIC OF ARMENIA — Selected Issues

### Executive Summary
- This Selected Issues Paper provides additional background information and analysis to support the discussion in the Staff Report.
- Chapter I: Enhancing Fiscal Policy in Armenia
  - Building up fiscal space during economic expansions is key for the effective implementation of stimulatory policies during downturns.
  - Following a period of significant fiscal consolidation, Armenia’s underlying fiscal balance started to deteriorate in 2004—at a time of rapid economic growth.
  - The chapter analyzes the role of fiscal policy as an effective demand management tool, covering:
    - impact of fiscal policy on aggregate demand (focusing on calculations of the fiscal impulse), and
    - impact on aggregate supply (focusing on calculations of the cyclically-adjusted fiscal balance).
- Chapter II: Smoothing Exchange Rate or Fuelling Inflation: What are the Effects of Foreign Exchange Market Intervention in Armenia?
  - The Armenian dram has appreciated considerably over the past six years, driven by strong foreign exchange inflows.
  - Foreign exchange market interventions by the central bank increased substantially in response to persistent appreciation pressures and heightened market volatility.
  - The chapter analyzes central bank interventions from 2001 to 2008 with emphasis on:
    - effectiveness of intervention in moderating the extent and volatility of exchange rate changes,
    - trade-offs between inflation targeting and exchange rate policy objectives, and
    - possible inconsistencies between monetary and intervention policies followed in the recent past.
- Chapter III: Exchange Rate Appreciation and External Competitiveness in Armenia
  - Recent dramatic appreciation of the dram has removed any previous undervaluation, bringing the real exchange rate near or possibly slightly above equilibrium.
  - Pass-through estimates show significant downward rigidity in import prices in response to exchange rate movements.
  - The current account deficit has worsened sharply during the past couple of years, raising concerns about external sustainability.
  - Although risks of a disruptive external correction remain modest, there is need to strengthen competitiveness through accelerated structural reforms to reduce medium term vulnerabilities.

### Chapter I — Enhancing Fiscal Policy in Armenia: Background and Framework
- Paper prepared by Borja Gracia (FAD), Holger Floerkemeier (MCD), Reginald Darius (SPR); approved by Middle East and Central Asia Department; October 31, 2008.
- Publication and availability details:
  - Price: $18.00 a copy
  - International Monetary Fund ● Publication Services, 700 19th Street, N.W. ● Washington, D.C. 20431
  - Telephone: (202) 623-7430 ● Telefax: (202) 623-7201 ● E-mail: publications@imf.org ● Internet: http://www.imf.org
- Context and key background findings:
  - Armenia’s chronic low tax revenue has dominated the fiscal policy debate during the last decade; tax revenue remains one of the lowest in the region and significantly below potential.
  - IMF, WB, and USAID technical assistance concluded that, unless significant structural change takes place, further improvements in tax collection are unlikely and recent ones are not sustainable.
  - Identified needs in tax reform:
    - upgrade the structure and business processes of the tax and customs administrations,
    - address tax policy deficiencies (absence of a VAT threshold, a “simplified” regime difficult to administer and comply with, and presumptive regimes that are, in reality, privileged regimes).
  - Recent legislative changes and a comprehensive modernization plan (VAT threshold, simplified regime for businesses below the threshold) are described as promising steps.
- Policy focus shift:
  - Emphasis should move to the role of fiscal policy as an effective demand management tool given rising income levels, increasing global market integration, and limited effectiveness of monetary policy.
  - Policy discussions should concentrate on fiscal effects on:
    - aggregate demand (price level, exchange rate, external current account), and
    - aggregate supply (cyclical effects of revenues and expenditures).

### Chapter I — Importance of a Consistent Fiscal Framework
- Observations on fiscal policy implementation:
  - In many transition countries, implementing consistent fiscal policies with a medium-term perspective has proven difficult.
  - Lack of formal analysis linking short-term decisions with medium-term cost implications leads to accommodation of pressures to relax fiscal positions and ignores contingent or implicit public liabilities.
- Desirable policy framework characteristics:
  - well-established, transparent, clear frameworks;
  - policies that are sound, based on solid analysis, and predictable;
  - consistency between monetary and fiscal frameworks aligned with common, publicized long-term objectives to increase credibility and anchor expectations.
  - For Armenia, fiscal policy needs alignment with the existing inflation-targeting regime and support monetary policy in a challenging international economic environment.
- Current Armenian fiscal framework and limitations:
  - Current framework: a three-year medium-term expenditure framework (MTEF) complemented by a debt management law with elements of a fiscal responsibility law.
  - Explicit rules in the framework:
    - ceiling on public debt at 60 percent of GDP,
    - constraints on the overall balance when debt is above 50 percent of GDP.
  - The framework has been weakened by major policy decisions taken outside the MTEF with permanent fiscal implications (example: recent large increase in average pensions).
- Fiscal sustainability and risks:
  - Public debt is low, posing no short-term risk to fiscal sustainability; this makes anchoring short-term fiscal policy to long-term sustainability difficult.
  - Medium-term fiscal risks should be routinely assessed and discussed in budget documents, and their impact on long-term fiscal sustainability quantified.
  - Recent hikes in energy and food prices, and the potential serious economic downturn from the international financial crisis, highlighted the need for more pro-active fiscal policy and clearer understanding of fiscal impacts on growth and inflation.

### Chapter I — Measuring the Fiscal Stance
- Definition and recommended measure:
  - An accurate measure of the fiscal stance is key for policy design.
  - Recommended measure for Armenia: the underlying fiscal balance defined as the overall balance excluding grants and external interest payments and including all subsidies (i.e., the gas subsidy that existed until May 2008).
  - Coverage ideally includes central government, state fund for social insurance (recently consolidated into the central government budget), local governments, state-owned enterprises, and off-budgetary accounts.
  - Government spending directed to imports should also be excluded, though this is technically difficult.
- Implications of different measures:
  - Different measures lead to different policy diagnostics and prescriptions (Figure I.1 indicates variation among overall balance, underlying balance, grants, and gas subsidy measured in percent of GDP for 1998–2007).
  - Using the overall balance (as defined in the 2005–08 PRGF-supported program) suggests:
    - a drastic fiscal adjustment during 1999–2002,
    - a moderate deterioration until 2005,
    - a slight improvement thereafter.
  - Using the underlying balance suggests:
    - the fiscal adjustment was less drastic and spread over 1999–2004,
    - steady deterioration of the fiscal stance during the last three years (relative to the report).
  - Large grant disbursements in 2002 and 2003 explain initial discrepancy; the off-budget gas subsidy introduced in 2006 explains the more recent gap.
  - The gas subsidy, though ultimately financing imports, acted as a direct transfer to residents and businesses, affecting disposable income and domestic aggregate demand; therefore any meaningful measure of the fiscal stance must include the gas subsidy.
  - A third measure—the cyclically adjusted balance—will be discussed later in the chapter.

### Chapter I — Determining the Impact of Fiscal Policy: the “Fiscal Impulse”
- Methodological approach:
  - The chapter analyzes the effect of government spending on the Armenian economy using a VAR framework to estimate the impact of changes in government spending on economic activity for the period 1998:1–2007:4.
  - Definitions used:
    - Net taxes = sum of tax revenues and nontax revenues, less transfer payments and net interest paid by the government.
    - Government spending = purchases of goods and services (current and capital).
  - All macro-fiscal variables are transformed in real per capita terms.
- Purpose:
  - The VAR-based approach is used instead of relying solely on the cyclically-adjusted underlying balance to estimate the fiscal stance and quantify the fiscal impulse.

*Prepared by Borja Gracia (FAD), Holger Floerkemeier (MCD), Reginald Darius (SPR); Approved by Middle East and Central Asia Department; October 31, 2008.*

### 10.      The structural transformation of the Armenian economy since independence makes

### 10.      The structural transformation of the Armenian economy since independence makes

### Identification and methodology
- Tax shocks are left unidentified because frequent tax policy changes, inadequate tax administration, low compliance, and a large informal economy make the relationship between tax revenues and economic activity weak and unpredictable.
- The analysis focuses on spending shocks; government spending variables are treated as predetermined with respect to macroeconomic shocks and unanticipated tax changes.
- Identification assumption: changes in government spending are undertaken for reasons other than immediate reaction to macroeconomic conditions. The quarterly frequency of the data makes this assumption plausible.
- VAR specification notes:
  - Impulse responses reported to a one percent shock to real (per capita) government spending.
  - Horizon: two years; one-standard deviation error bands via Monte Carlo with 1000 replications.
  - Given evidence of non-stationarity, the VAR is estimated in first differences with no linear trend. Two lags used.
  - Results virtually unchanged if interest rates are added to the endogenous vector.

### Impact of spending shocks
- Main empirical findings:
  - Spending shocks appear expansionary and inflationary.
  - The expansionary effect could finance at most half of the spending increase, resulting in a significant deficit.
  - Prices take two quarters to react to government spending shocks (response not statistically significant).
  - A 1 percent increase in real per capita government spending can lead to up to 0.5 percent increase in per capita GDP (short-run estimate).
- Caveats:
  - High uncertainty in estimations and data quality issues constrain interpretation.

### Impact on the labor market
- Responses to a government spending shock:
  - Employment: expected sign, statistically significant only in the short-term, magnitude small, relatively persistent.
  - Unemployment rate: opposite (expected) sign, relatively large but insignificant response, relatively persistent.
- Data quality caveat: high informality and systematic under-reporting of employment and wages in Armenia warrant caution in interpreting labor market results.

### Impact on economic activity
- Key responses to a government spending shock:
  - Private per capita consumption: relatively large and persistent impact (not statistically significant after three quarters).
  - Private investment: reacts strongly and positively (government spending appears complementary to private investment).
  - Manufacturing activity: responds positively.
  - Trade balance: government spending results in a widening of the trade deficit; effect is large, persistent, and statistically significant for at least one year.
- Policy implication: evidence suggests scope for fiscal policy as a short-term counter-cyclical tool given positive and statistically significant short-run impact on activity, but inflationary pressures and larger fiscal and trade deficits follow increased spending.

### Estimating a cyclically-adjusted fiscal balance
- Purpose: separate cyclical factors from trends to assess structural fiscal position; cyclically-adjusted underlying fiscal balance strips away revenue and expenditure components attributable to output deviations from potential (the output gap).

#### Estimating potential output and the output gap
- Method: Hodrick-Prescott (HP) filter applied to the log of GDP; annual data 1994–2007 used, analysis focused on 1998–2007; value of 100=λ (generally used for annual data).
- Findings (annual and annualized quarterly):
  - Actual output remained below potential before 2003 (negative output gap).
  - Growth of actual output exceeded potential since 2001 (except 2004).
  - Actual output appears above potential since 2005.
  - Annualized quarterly data produce qualitatively similar results but smaller cycles; negative output gap until 2005, increasingly positive thereafter.

#### Calculating the cyclically-adjusted fiscal balance (elasticity framework)
- Assumption: each budget component adjusted proportionally to ratio of potential to actual output according to its output elasticity.
- Notation (as in source):
  - T = actual tax revenue (could be disaggregated by tax i)
  - G = actual current government expenditures
  - η_i = output elasticity of tax i
  - ξ = output elasticity of current government expenditure (including the gas subsidy)
  - capital G = government spending on capital projects
- Cyclical adjustment uses assumed elasticities to scale each component to potential output.

#### Tax revenues
- Empirical estimate: simple regression of quarterly changes in real tax receipts on lead and lags of changes in real output yields a point estimate of 0.47 (average elasticity over the period).
- Given uncertainty, analyses focus on a range of feasible values for tax elasticity: minimum 0, maximum around 2 (as observed for other countries).
- Illustrative consequences:
  - Negative revenue gap when actual GDP below potential could reach 2.5 percentage points of GDP if elasticity close to 2.
  - The cyclical temporary component of revenues in 2007 could be as large as 1.5 percentage points of GDP.
  - Cyclical factors contributed marginally to revenue performance in 2003/04.
  - Example note from figure: In 2000, assuming η = 1, the business cycle resulted in tax revenues around 1 percentage points of GDP lower than if GDP had been at potential.

#### Government spending
- Automatic stabilizers are very weak in Armenia:
  - Weak unemployment benefits and weak countercyclical expenditure policies.
  - Nominal progressivity of tax system effectively eliminated by tax administration practices.
- Regression estimate for spending elasticity returned -0.49, but deemed unrealistically high given weak stabilizers; baseline chosen ξ = -0.1.
- Range considered for ξ: (-0.1, -0.8).
- Implications:
  - With weak automatic stabilizers, cyclically-adjusted expenditure should not deviate much from actual expenditure unless elasticity relatively large.
  - Example from figure: In 2000, assuming ξ = -0.4, business cycle resulted in government spending around 0.4 percentage points of GDP higher than if GDP had been at potential.

#### Cyclically-adjusted underlying balance: results and sensitivity
- Baseline elasticities: tax elasticity η = 0.47; spending elasticity ξ = -0.1.
- Findings (1998–2007):
  - Gap (difference between cyclically-adjusted balance and underlying balance) shows cyclical factors exacerbated overall fiscal deficit during 1999–2004.
  - By 2004, cyclically-adjusted deficit close to 2 percent of GDP (during period of fiscal adjustment).
  - Since 2004, cyclically-adjusted balance deteriorated faster than overall balance while actual output grew above potential.
  - By 2006 and 2007, cyclical factors helped reduce the size of the underlying deficit (i.e., positive cyclical revenue contribution).
  - Using quarterly data yields qualitatively similar results.
- Sensitivity to elasticity assumptions:
  - Given assumed low variability of government spending to the business cycle, the gap magnitude is mostly determined by revenue elasticity variability.
  - Figure comparisons include combinations: η = 0.25 or 1 with ξ = -0.1 or -0.2; baseline η = 0.47, ξ = -0.49 also shown for contrast.
  - Example from figure: In 2000, assuming η = 0.25 and ξ = -0.1, the business cycle resulted in an underlying fiscal balance around 0.4 percentage points of GDP higher than if GDP had been at potential.

#### Overall assessment of the fiscal stance (1998–2007)
- Key conclusion: fiscal policy has been pro-cyclical in recent years.
  - Both underlying and cyclically-adjusted balances show deterioration of fiscal position since 2004—at a time when actual output was growing above potential.
  - This conclusion is not evident when looking only at the overall balance according to budget definitions.
  - After large grants inflows moderated post-2003/04, overall, cyclically-adjusted, and underlying balances provided a similar diagnostic; the introduction of off-budget gas subsidy in 2006 again distorted the overall balance.

### Macro-fiscal unit: institutional recommendations
- Recent Armenian measures:
  - Strengthening public financial management systems.
  - Further institutionalizing the MTEF.
  - Adopting performance-based budgeting over the next few years.
  - Strengthening macroeconomic projections and analysis with fiscal impulse indicators during budget preparation.
- Recommendation: establish a strong macro-fiscal policy department to provide a strategic, coordinated, multi-year approach to macro-fiscal policy.
- Suggested core functions of an effective macro-fiscal policy department:
  - Develop the fiscal framework: forecasting, analysis, monitoring short- and medium-term fiscal developments.
  - Monitor implementation: assess budget execution and link to growth and inflation developments.
  - Protect the fiscal framework: advise on emerging near- and medium-term risks; flag need for corrective actions; identify alternatives.
- Value-added tasks (top-down, big-picture perspective):
  - Ensure internally-consistent macroeconomic framework via macroeconomic analysis.
  - Perform fiscal modeling, forecasting and impact analysis to develop multi-year fiscal projections under baseline unchanged policies and simulate policy changes.
  - Align fiscal policy with government economic objectives; choose appropriate fiscal paths to attain objectives such as macroeconomic stability, medium-term growth, poverty reduction, and fiscal and debt sustainability.
  - Assess consistency via macroeconomic analysis and debt/fiscal sustainability analysis; refine revenue, expenditure and debt targets.
  - Identify fiscal risks through alternative scenarios and stress tests.
  - Monitor in-year budget outcomes—financing, revenues, expenditures; explain deviations from plans; assess implications for end-year forecasts; design corrective policies when required; coordinate closely with other units to avoid duplication.

_Copyright and figure sources as presented in the chapter: Source: IMF staff estimates._

### 30.      Expansionary fiscal policy seems to have a positive impact on economic activity and

### _cr08375 - 30.      Expansionary fiscal policy seems to have a positive impact on economic activity and

### Fiscal policy effects and recommendations
- Expansionary fiscal policy:
  - Positive impact on economic activity and employment.
  - Also increases inflationary pressures and worsens the trade balance and the fiscal position of the central government.
  - Sustained expansionary fiscal policy may lead to increased macroeconomic imbalances and threaten fiscal sustainability.
- Policy recommendation:
  - Building up fiscal space during expansions is key for the effective implementation of stimulatory policies during downturns.
- Recent fiscal stance in Armenia:
  - After an initial phase of fiscal consolidation until 2004, the underlying fiscal balance has steadily deteriorated during a period of rapid economic growth.
  - Going forward, the fiscal stance should become more neutral.
  - Short-run guidance:
    - If signs of overheating persist, the fiscal stance needs to be tightened to allow fiscal policy to play a more active stabilization role.
    - If the international economic downturn impacts the Armenian economy significantly, a tighter fiscal stance may not be desirable.

### Foreign exchange market intervention: objectives and macro links
- CBA objectives and regime:
  - The Central Bank of Armenia (CBA) maintained a flexible exchange rate regime while increasingly engaging in foreign exchange market interventions aimed at “moderating the rate of change and preventing undue fluctuations in the exchange rate, rather than establishing a level of it.”
  - Exchange rate regime reclassification (September 2006): from “independently floating” to a “managed float with no predetermined path or level for the exchange rate.”
- Transmission channels from intervention to exchange rate:
  - Monetary policy channel is the main transmission path in Armenia because interventions have largely been unsterilized and directly impact reserve money.
  - Signaling channel may operate because CBA market activity is observable by market participants despite interventions not being announced.
  - Microstructure (order flow) channel may be important due to volatile and lumpy foreign exchange flows, but intraday effects cannot be empirically tested with available data.
- Sterilization and liquidity management:
  - Apart from issuing its own short-term securities, the CBA lacked instruments to sterilize intervention-induced liquidity changes.
  - Net issuance of CBA securities for liquidity management: around AMD28 billion in 2006 and AMD10 billion in 2007.
  - Net liquidity injections from foreign exchange market intervention: AMD95 billion in 2006 and AMD117 billion in 2007.

### Foreign exchange market structure and trends
- Market characteristics:
  - Small number of market participants, comparatively low but volatile turnover.
  - Retail market dominated by commercial banks; foreign exchange bureaus are small and household-focused.
  - Wholesale market averaged approximately 20 percent of total turnover, with variations between 0 and 50 percent.
- ARMEX trading patterns and volumes:
  - Trading activity increased significantly since June 2006, but average daily trade volumes remained low:
    - Average daily trade volume: around $3.2 million in 2007/08.
    - Trading days with volume not exceeding $1 million: almost half of trading days.
    - Trading days exceeding $5 million: about 15 percent of trading days.
    - On active trading days, number of trades: between 20–25 trades.
    - Average contract size on active days: close to $100,000.
- Seasonality of flows and interventions:
  - First quarter: lower economic activity, exports, and remittances → subdued turnover and intervention activity; CBA interventions mainly foreign exchange sales for liquidity management.
  - Fourth quarter: remittances and capital inflows strongest → CBA foreign exchange purchases mainly during last three to four months of the year.
- Intervention magnitude and evolution:
  - 2001–2003: CBA net foreign exchange sales amounted to between 15 and 20 percent of beginning-of-year reserve money.
  - From 2004, interventions turned to net purchases and increased significantly until 2007, when they exceeded 90 percent of end-2006 reserve money.
  - In 2008, with diminished appreciation pressures and other shocks, intervention activity moderated and returned to net foreign exchange sales.

### Empirical investigation: scope, data, and methodology
- Objective:
  - Investigate the effectiveness of CBA foreign exchange market intervention during January 2001 to May 2008, focusing on the recent period of strong dram appreciation pressures.
  - Policy effectiveness measured by impact of interventions on level and volatility of exchange rate returns (daily percentage changes).
- Data and sample:
  - Data sources: CBA (daily dram/dollar spot exchange rates, interventions, interbank interest rates), Bloomberg (euro/dollar exchange rates, overnight and two-week LIBOR rates, daily oil price).
  - Sample period: January 3, 2001 to May 30, 2008.
- Key summary statistics (from Table II.1):
  - AMD/$ exchange rate: Mean 483.315, Median 513.735, Maximum 591.763, Minimum 301.593, Std. Dev. 3.816, Skewness -0.570, Kurtosis 1.892, Jarque-Bera 203.66, Stationarity: no
  - AMD daily exchange rate return (Δ(ln s_t)): Mean 0.000, Median 0.000, Maximum 0.034, Minimum -0.027, Std. Dev. 0.003, Skewness 0.150, Kurtosis 33.477, Jarque-Bera 74,893.39, Stationarity: yes
  - AMD deviation from HP trend (HPΔ s_t): Mean 0.000, Median 0.684, Maximum 26.006, Minimum -19.641, Std. Dev. 6.087, Skewness -0.320, Kurtosis 4.111, Jarque-Bera 132.46, Stationarity: yes
  - Foreign exchange purchases (I_P_t) (in million $): Mean 0.493, Median 0.000, Maximum 25.690, Minimum 0.000, Std. Dev. 1.806, Skewness 6.269, Kurtosis 57.595, Jarque-Bera 252,984.30
  - Foreign exchange sales (I_S_t) (in million $): Mean 0.182, Median 0.000, Maximum 13.425, Minimum 0.000, Std. Dev. 0.884, Skewness 7.077, Kurtosis 65.230, Jarque-Bera 328,372.50
  - Euro daily exchange rate return (Δ ln e_t): Mean 0.000, Median 0.000, Maximum 0.024, Minimum -0.024, Std. Dev. 0.006, Skewness 0.056, Kurtosis 4.040, Jarque-Bera 88.22, Stationarity: yes
  - 1st Diff. AMD/LIBOR interest differential (Δ(i-r*)_t): Mean -0.010, Median 0.001, Maximum 6.720, Minimum -6.440, Std. Dev. 0.968, Skewness -0.111, Kurtosis 12.442, Jarque-Bera 7,191.76, Stationarity: yes
  - 1st Diff. One-year-lagged MA oil price change (Δ(oil)_t): Mean 0.000, Median 0.000, Maximum 0.021, Minimum -0.024, Std. Dev. 0.007, Skewness -0.131, Kurtosis 3.040, Jarque-Bera 5.67, Stationarity: yes
- Methodology:
  - Base model: standard GARCH(1,1).
  - Several model variants estimated with different explanatory variables and alternative GARCH specifications for robustness.
  - Endogeneity/simultaneity concern:
    - Interventions may be endogenous because significant exchange rate movements can trigger interventions.
    - To address identification, a two-stage instrumental variable approach was used:
      - Step 1: Estimate CBA policy reaction function using a TOBIT dynamic censored regression model.
      - Step 2: Use predicted values from the TOBIT as instruments in the GARCH specification.

### Empirical findings on effectiveness of intervention
- Main empirical conclusions:
  - Limited effectiveness of foreign exchange market interventions:
    - Intervention does not seem to have had a substantial impact on dram exchange rate returns (level).
    - GARCH models lend only tentative support to the claim that interventions have contributed to reducing exchange rate volatility.
  - Interpretational caveats:
    - Intervention data are only available at daily frequency; interventions may have moderated intraday exchange rate volatility in a thin market, which cannot be captured with available data.
    - Simultaneity may bias coefficients downward, potentially underestimating intervention effectiveness; the two-stage approach was used to mitigate this.
- Consistency with literature:
  - Results align with broader literature finding generally no long-term effects of interventions on exchange rate levels, but some evidence of small short-term effects in advanced economies.
  - Evidence for effectiveness in developing and transition countries is mixed; interventions can sometimes increase exchange rate volatility.
  - Theoretical considerations suggesting potentially greater effectiveness in developing/transition countries:
    - Less sterilization (so monetary channel operates).
    - Exchange rate regulations can increase central bank information advantage.
    - Relative size of interventions often large compared to market turnover.

*Prepared by Holger Floerkemeier; empirical analysis covers January 3, 2001–May 30, 2008.*

### 20. In the first stage, the CBA’s foreign exchange intervention policy reaction function is

### 20. In the first stage, the CBA’s foreign exchange intervention policy reaction function is

### Methodology — First stage (intervention reaction function)
- Specification: TOBIT estimated separately for foreign exchange purchases (x = p) and sales (x = s).
- Equation form (as specified): I_t = max{0, γ_0 + γ_1 I_{t-1} + γ_2 Δs_{t-1} + γ_3 (ln s_{t-1} - ln s^T_{t-1}) + ε_t}
  - Dependent variable I_t: actual amount of dollar purchases (sales).
  - s_t: nominal dram/dollar exchange rate.
  - s^T_t: 20-day moving average exchange rate.
- Hypothesis: CBA “leans against the wind” to avoid excessive short-term exchange rate swings; intervention depends on one-day lagged intervention, immediate exchange rate changes, and deviation from a short-term moving average.
- Model variations tested: alternative specifications including LIBOR/dram interest rate differentials and/or different lag lengths; GARCH models estimated with original intervention data for comparison.

### Methodology — Second stage (impact on returns and volatility)
- Mean and variance modeled with GARCH(1,1).
- Mean equation (Equation (2) structure): Δln s_t = β_0 + β_1 Δln s_{t-1} + β_2 P^*_t + β_3 S^*_t + β_4 Δln e_t + β_5 Δ(i - r^*)_{t-1} + β_6 ΔOIL_t + ε_t
  - P^*_t and S^*_t: instrumented amounts of CBA exchange rate purchases and sales from first stage.
  - Δln e_t: daily Euro/dollar nominal exchange rate return.
  - Δ(i - r^*)_{t-1}: first-differenced interest differential between domestic interbank rate and LIBOR in percent per year.
  - ΔOIL_t: first differenced monthly moving average of one-year lagged annual oil price changes (proxy for FX inflows).
- Variance equation (Equation (4) structure): includes ARCH and GARCH terms, instrumented P^*_t and S^*_t, control variables, day-of-week, holiday, and post-holiday dummies.
- Distributional notes: residuals exhibit fat tails; Bollerslev-Wooldridge robust standard errors and covariances used; Student’s t tested for robustness.

### Determinants of foreign exchange intervention — First stage results (Table II.2)
- Key findings:
  - Exchange rate appreciation (depreciation) triggers foreign exchange purchases (sales).
  - Only the coefficient for deviations from the moving average is significant for foreign exchange purchases.
  - Strong intervention clustering: lagged intervention coefficients indicate one-day persistence.
- Selected reported statistics (Tobit results):
  - C (constant): -6.91 *** (p-value significance denoted by ***), -5.81 ***
    - Standard errors: (0.424), (0.442)
  - I_{t-1}: 1.07 ***, -1.00 ***
    - Standard errors: (0.079), (0.121)
  - Δs_{t-1}: -19.44, 185.58 **
    - Standard errors: (74.433), (77.901)
  - (ln s_{t-1} - ln s^T_{t-1}): -143.39 ***, 56.39 ***
    - Standard errors: (21.618), (21.825)
  - SCALE:C(5): 5.14 ***, 4.21 ***
    - Standard errors: (0.249), (0.269)
  - Mean dep var: 0.494, -0.182
  - S.E. of reg: 1.564, 0.844
  - Prob > LR: 0.000, 0.000
  - Left cens obs: 16560, 0
  - Right cens obs: 0, 1761
  - Uncensored obs: 278173
  - Total obs: 19341934
- Note: Tobit coefficient interpretation caveat—coefficients do not equal marginal effects directly.

### Impact of intervention on the level and volatility of exchange rate returns — GARCH results (Table II.3 and narrative)
- Samples estimated:
  - Full sample: Jan. 2001 - May 2008 (Observations: 1,934)
  - First period: Jan. 2001 - Oct. 2005 (Observations: 1,259)
  - Second period: Nov. 2005 - May 2008 (Observations: 674)
- Mean equation highlights (impact per $1 million intervention):
  - Δ(ln s_{t-1}): 0.304 *** (full), 0.271 *** (first), 0.499 *** (second)
    - Standard errors: (0.036), (0.054), (0.053)
  - I^P_{t-1} (purchases):
    - Full sample: -0.226 ** (0.112)
    - First period: -0.056 (0.181)
    - Second period: -0.193 *** (0.0684)
  - I^S_{t-1} (sales):
    - Full sample: -0.028 (0.066)
    - First period: -1.287 (1.042)
    - Second period: 0.067 (0.116)
  - Δ(ln e_t): 0.018 * (full), 0.004 (first), 0.063 *** (second)
    - Standard errors: (0.009), (0.007), (0.017)
  - Δ(i - r^*)_{t-1}: 0.000 across samples (standard errors (0.000))
  - Δ(oil)_{t-1}: -0.010 * (full), -0.012 * (first), -0.007 (second) with standard errors (0.006), (0.007), (0.018)
- Variance equation highlights:
  - RESID(-1)^2: 0.150 ** (full) (0.062); 0.150 * (first) (0.090); 0.150 *** (second) (0.037)
  - GARCH(-1): 0.600 *** across samples (standard errors: (0.035), (0.134), (0.063))
  - I^P_{t-1} in variance: 0.000 (full), 0.000 (first), 0.000 * (second) (standard errors (0.000), (0.002), (0.000))
  - I^S_{t-1} in variance: -0.001 *** (full) (0.000), -0.002 (first) (0.003), -0.001 *** (second) (0.000)
  - Δ(i - r^*)_{t-1} in variance: 0.000 *** (full) (0.000)
- Model fit:
  - R^2: 0.111 (full), 0.061 (first), 0.269 (second)
  - Adj. R^2: 0.100, 0.045, 0.244
  - Log-L: 9,000; 5,823; 3,267
- Interpretation and findings:
  - Foreign exchange purchases are associated with an appreciation of the domestic currency (negative I^P coefficient), possibly reflecting simultaneity: interventions tend to follow exchange rate movements rather than drive them.
  - Foreign exchange sales do not show a consistent significant impact on exchange rate returns; sales may have been used mainly for liquidity management in calm periods.
  - Foreign exchange sales are associated with lower exchange rate volatility; purchases show little or no volatility-reducing effect and in the second subperiod may be associated with slightly higher volatility.
  - Control variables behave as expected: autoregressive term positive and significant; relative dollar strength depreciates domestic currency; rising capital inflow proxy (oil) associated with appreciation; exchange rate not sensitive to interest differentials.

### Alternative specification — Deviations from trend (Hodrick-Prescott filter) (Table II.4)
- Dependent variable: HPΔs_t (deviation of dram/dollar from trend).
- Key coefficients (impact per $1 million intervention):
  - I^P_{t-1}:
    - Full sample: -75.313 *** (21.687)
    - First period: -24.017 (72.755)
    - Second period: -107.003 *** (31.183)
  - I^S_{t-1}:
    - Full sample: 87.681 *** (20.077)
    - First period: -33.316 (354.055)
    - Second period: 98.616 (67.993)
- Variance equation highlights:
  - RESID(-1)^2: 0.259 *** (full) (0.073); 0.191 * (first) (0.118); 0.094 ** (second) (0.073)
  - GARCH(-1): 0.532 ***(full) (0.038); 0.542 ***(first) (0.169); 0.484 ***(second) (0.154)
  - I^P_{t-1} in variance: -0.019 (full) (0.078); -0.081 (first) (0.117); -0.004 (second) (0.060)
  - I^S_{t-1} in variance: -0.234 *** (full) (0.023); -0.479 (first) (0.984); -0.073 (second) (0.078)
- Interpretation:
  - Results broadly similar to original model: foreign exchange purchases unexpectedly associated with appreciation; foreign exchange sales associated with depreciation over the full sample.
  - Interventions appear to follow exchange rate movements rather than inducing them.
  - Foreign exchange sales associated with slightly lower volatility; purchases negative in variance equation but not significant.

### Robustness and overall inference
- Robustness checks: choice of sample periods, alternative GARCH variants, inclusion/exclusion of control variables, and varying lag lengths leave main results unchanged.
- Overall conclusion from estimations:
  - Interventions appear not to result in exchange rate movements in the intended direction.
  - Interventions have only a very limited impact on the volatility of the exchange rate.

### Consistency of exchange rate and monetary policies — empirical observations and implications
- Policy shift: CBA gradually shifted from a monetary anchor to an inflation targeting regime since 2006; forecast level of inflation as intermediate target and short-term interest rate as operational target.
- Constraints: unstable money demand, shifts in monetization and dollarization, limited depth of financial markets, substantial excess liquidity.
- Operations and effects:
  - Between 2005 and 2007, CBA issued significant short-term securities to mop up excess liquidity.
  - Foreign exchange sales supplemented sterilization during times of relative FX market calm (usually first quarter).
  - From second half of 2007, a credit boom emerged, driven by imports, consumption, real-estate; private sector credit surge reduced excess reserves and increased domestic demand and inflationary pressures.
  - Rapid remittance and capital inflows maintained upward pressure on the exchange rate, triggering sharp appreciation rallies and conversions of dollar holdings into dram.
- Liquidity and inflation links:
  - Unsterilized foreign exchange purchases amounted to about 90 percent of end-2006 reserve money, driving rapid reserve money growth.
  - Surplus liquidity blunted the interest rate channel; commercial banks delayed raising lending rates until the third quarter of 2008, more than a year after CBA started raising the repo rate.
  - Unsterilized interventions likely fuelled credit growth and domestic demand; with an approximate transmission lag from money growth to inflation of about a year, additional inflation pressures could arise.
- Resolution in 2008:
  - Temporary inconsistency between inflation targeting and FX intervention resolved in 2008: foreign exchange market interventions (FX sales) and interest rate increases aligned, resulting in a tightening of the monetary policy stance.

*Source: IMF staff analysis (text and tables provided in the content unit).*

### 39. This paper analyzed the impact of CBA interventions in the foreign exchange market

### This paper analyzed the impact of CBA interventions in the foreign exchange market from 2001 to 2008

### Study scope and objectives
- Analyzed the impact of CBA interventions in the foreign exchange market from 2001 to 2008.
- Main focus: investigate the effectiveness of foreign exchange purchases and sales in moderating the extent and volatility of exchange rate changes.
- Discussed trade-offs between inflation targeting and exchange rate policy objectives and possible inconsistencies between monetary and intervention policies followed in the recent past.

### Methodology
- Employed a two-stage approach:
  - Stage 1: modeled the intervention policy reaction function of the central bank.
  - Stage 2: used the intervention forecasts from Stage 1 as instruments in a GARCH model of the level and volatility of dram/dollar exchange rate returns.

### Empirical findings
- Overall evidence: only weak support that foreign exchange interventions have been effective in Armenia.
- Predictability of interventions:
  - The incidence and size of foreign exchange interventions can be fairly well predicted by short- and medium-term exchange rate movements and prior intervention activity.
- Impact on exchange rate returns:
  - Evidence regarding the impact of intervention on exchange rate returns is rather weak.
  - Where intervention coefficients were statistically significant in the mean equation, the effects were small and not in the theoretically expected direction.
- Impact on volatility:
  - In the variance equation, the model indicates that foreign exchange sales may have some limited impact on lowering the volatility of exchange rate returns.

### Caveats and limitations
- Methodological challenges:
  - Difficulty disentangling cause and effect due to the identification problem resulting from the simultaneity of exchange rate changes and intervention activity.
- Data limitations:
  - Potential intraday stabilizing effects of interventions could not be investigated due to lack of ultra-high frequency data.
  - Intraday effects—particularly relevant in the context of very volatile market activity and lumpy foreign exchange flows—might exist and could create additional interday exchange rate dynamics, but were not observable in the study.

### Monetary and exchange rate policy consistency
- Large-scale unsterilized foreign exchange interventions have, at times, conflicted with monetary tightening pursued to fight rising inflation pressures in 2006 and particularly the second half of 2007.
- Policy transmission and macro effects noted:
  - Injection of large amounts of domestic currency liquidity (while partly accommodating cash dollarization) may have blunted the interest rate channel of monetary policy.
  - Such interventions have fuelled money and credit growth.
  - These developments may eventually result in higher inflation from domestic demand pressures.

*Source: IMF staff paper, paragraphs 39–43.*

### 15.      A third approach for estimating the equilibrium exchange rate is the external

### _cr08375 - 15.      A third approach for estimating the equilibrium exchange rate is the external

### External Sustainability (ES) approach
- Purpose: Relates sustainability of external stock position, current account position, and the real exchange rate via a simple calibration requiring assumptions about potential growth, inflation, and rates of return on external assets/liabilities.
- Key macroeconomic assumptions:
  - NFA/GDP benchmark used: NFA/GDP position as of end-2006 (-24.2 percent).
  - Inflation rate of NFA: 2.5 percent.
  - Benchmark GDP growth rate: 6 percent.
  - Alternative growth scenarios considered: low-growth scenario (3 percent), high-growth scenario (9 percent).
- Current account deficits required to stabilize the NFA/GDP ratio for each growth scenario:
  - benchmark growth: -1.9 percent
  - low growth: -1.3 percent
  - high growth: -2.5 percent
- Result: ES approach, like PPP and BEER approaches, indicates the real exchange rate is no longer undervalued and may have become somewhat overvalued in 2007. The required REER adjustment is broadly similar across the three scenarios. The magnitude of the required exchange rate adjustment is obtained by computing the elasticity of the current account balance with respect to the real exchange rate.

### Macroeconomic Balance (MB) approach
- Specification: Adapts the Consultative Group on Exchange Rate Issues (CGER) formulation to Armenia’s data and economy; current account is related to fiscal balance, demographics, initial net foreign asset position, oil prices/terms of trade, output growth, and relative income (compared to the U.S.). For non-oil Armenia, the specification adds the real exchange rate and FDI as explanatory variables.
- Estimation and results:
  - Six estimates of the current account “norm” obtained, ranging from a deficit of about 2 percent of GDP to a deficit of 14 percent of GDP.
  - The actual current account level lies within this range, suggesting the exchange rate was near the equilibrium level.
  - The underlying current account was estimated at a deficit of 5 percent of GDP in 2006.
  - Each specification was estimated using pooled OLS, fixed effects, and GMM estimators; each was also estimated with variables expressed as deviations from partner-country averages as well as in (log) levels.

### Exchange Rate Pass-Through and Domestic Competition
- Context and methodology:
  - Motivated by anecdotal evidence of monopolistic practices in key import sectors (sugar, fuel) and their potential impact on pass-through from exchange rate changes to domestic imported goods prices.
  - Empirical method: VAR model loosely following McCarthy (2007), using monthly data for the period May 2005 – August 2007 (excluding the pre-stabilization period).
  - VAR endogenous variables: price of oil in domestic currency, output gap, nominal effective exchange rate, and import prices. Variables expressed in first differences and natural logs (output gap already stationary).
  - Identification: oil price identifies supply shocks; output gap identifies demand shocks; exchange rate responds contemporaneously to supply and demand shocks; import price index responds contemporaneously to supply, demand, and exchange rate shocks.
- Main findings:
  - Import price responses are much larger for exchange rate depreciations than for appreciations; only responses to exchange rate depreciations are statistically significant.
  - Exchange rate pass-through is much larger (about three times) for depreciations than for appreciations.
  - Speed of adjustment: import prices adjust to both appreciations and depreciations over about 3 months.
  - Interpretation: Asymmetric pass-through indicates significant downward rigidity in import prices in response to exchange rate movements; transportation costs alone are unlikely to explain this asymmetry, pointing instead to monopolistic practices in the import sector that have limited the role of dram appreciation in containing inflationary pressures.

### Structural Competitiveness
- Broad message: Continued exchange rate appreciation over the medium term implies competitiveness preservation requires sustained productivity improvements, largely via improving the business environment.
- Business environment observations:
  - Recent trends show moderate improvements in ease of doing business, but Armenia’s performance remains weaker than several regional comparators.
  - Progress has been made on getting credit, but Armenia continues to rank poorly on cross-border trade and paying taxes, highlighting the need to accelerate tax administration reforms.
- Competitiveness rankings:
  - Armenia is ranked at 93 in the Global Competitiveness Index (GCI) 2007-2008.
  - Armenia scores poorly on infrastructure, institutions, and financial market sophistication.
  - Given a weak and highly concentrated export base, significant structural reforms are needed to boost competitiveness.

### Conclusions and policy recommendations
- Exchange rate assessment:
  - The recent dramatic appreciation of the dram removed the previously existing undervaluation, bringing the real exchange rate near or possibly slightly above equilibrium; this result is consistent across differing methods used.
  - Pass-through estimates point to significant downward rigidity in import prices in response to exchange rate movements.
- Policy recommendations to safeguard competitiveness amid dram appreciation:
  - Improve the business environment to raise productivity.
  - Improve tax and customs administration.
  - Deepen financial intermediation.
  - Reduce corruption.
  - Discontinue monopolistic practices in the import sector.

*Source: _cr08375 - 15.      A third approach for estimating the equilibrium exchange rate is the external*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2008/_cr08375.pdf_