## _cr14314 ## Source details **Canonical URL:** [_cr14314](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2014/_cr14314.pdf) ## Other formats - [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2014/_cr14314.pdf.md) - [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2014/_cr14314.pdf.json) --- ### The Evolution of Households’ Net Wealth - Household saving rate: - Over 2002–08 declined from a peak of 15.1 percent in 2003 to 6.7 percent in 2008; growth averaged 3.6 percent per year. - Rebounded sharply during 2009–11 as credit and housing bubbles burst and growth declined. - Since 2012 the saving rate fell rapidly; Cyprus’s decline appears deepest among referenced boom-bust countries (Ireland, Spain). - Private consumption constituted about 66 percent of aggregate demand over 2000–13. - Wealth and balance-sheet statistics: - Household net financial wealth increased to 147 percent of GDP by end-2012; remained at 138 percent of GDP by end-2013. - During 2005–08: liabilities increased by 9 billion (33 percent of GDP); non-financial (housing) wealth rose by an estimated 6 billion (14 percent of GDP); financial wealth increased by 19 billion (59 percent of GDP). - 2013 banking crisis impacts: household financial assets fell by about €5 billion in 2013; about €0.5 billion of deposit decline estimated due to the bank bail-in; decline in insurance and pension fund assets due to bail-in losses and valuation losses on stock portfolios; households used pension fund assets to finance consumption. - Household debt: 140 percent of GDP at end-2013; debt interest payments as a share of disposable income rose to around 13 percent in 2013, up from 9.5 percent in 2006. - Household non-performing loans reached 45 percent of total loans at end-June 2014. - Distributional note: net financial wealth large and positive but distribution less even than euro-area average; some households with limited wealth may need to raise saving to repair balance sheets. ### Theoretical determinants and empirical approach - Theoretical channels summarized: - Wealth channel (permanent income): higher real wealth → consume more, save less. - Real interest rate: substitution and income effects. - Uncertainty: higher uncertainty → precautionary saving; realized shocks may force drawing down savings. - Fiscal balances: loose fiscal policy may be associated with higher household saving (Ricardian effect), alternative interpretations link fiscal position and macro uncertainty. - Demographics: lifecycle effects from population age structure. - Empirical approach: - Two models estimated using quarterly data from 2004:Q1 to 2013:Q4. - Model 1 (change): explains year-on-year change in the saving rate using lagged explanatory variables (all explanatory variables enter with one lag). - Model 2 (level): explains the level of the saving rate, capturing wealth, precautionary, and financial channels with lags chosen by Akaike’s information criterion; alternative specification adds fiscal balance and household debt. ### Model 1 — Determinants of year-on-year change in household saving rate (2004Q1-2013Q4) - Regression fit: Adjusted R-squared = 0.817. Dependent variable: Year-on-year change of household saving rate. - Coefficients (lagged) and interpretations: - Net financial assets (change): coefficient = -0.099; t-statistic = -5.633; Probability = 0.000. - Interpretation: Every one percent of GDP increase in net financial assets (i.e., €165 million) is associated with a 0.1ppt decrease in saving rate. - Housing price (percent change): coefficient = -0.147; t-statistic = -2.463; Probability = 0.020. - Interpretation: Every one percent increase in housing price is estimated to lead to 0.15ppt decrease in saving rate. - Fiscal balance (change): coefficient = -0.681; t-statistic = -5.522; Probability = 0.000. - Interpretation: Every one percent of GDP fiscal consolidation likely associated with 0.7ppt decline in saving rate. - Unemployment (change): coefficient = -2.226; t-statistic = -8.189; Probability = 0.000. - Interpretation: Every one percentage point increase in unemployment rate is associated with a decline of saving rate by 2.2ppt. - Real deposit rate (change): coefficient = 0.453; t-statistic = 2.690; Probability = 0.012. - Interpretation: Positive coefficient suggests substitution effects dominate; higher interest rates encourage saving. - Constant = 2.587; t-statistic = 4.395; Probability = 0.000. - Decomposition (qualitative): - 2005–08 decline largely explained by increasing net financial wealth, rising housing prices, and some improvement in fiscal position. - 2009–11 rebound driven by worsening financial wealth and fiscal position. - Since 2012 rising unemployment dominant in inducing consumption smoothing and reducing saving rate. ### Model 2 — Determinants of the level of the saving rate (2004Q1-2013Q4) - Model channels: - Wealth channel: net financial wealth (level, percent of disposable income) and housing wealth (y-o-y percent change). - Precautionary channel: percent balance of consumers’ unemployment expectations over next 12 months (EC Business and Consumer Survey). - Financial channel: percent balance of households’ expectations about their financial situation over next 12 months (proxy for credit availability/asset-return expectations). - Alternative specification adds general government fiscal balance and household debt to disposable income ratio. - Estimated coefficients and fit: - Column (1) (baseline): Adjusted R-squared = 0.648. - Net financial assets (t-2): -0.066*** (t-statistic: -3.292). - Housing price (percent change, t-4): -0.153*** (t-statistic: -2.833). - Expected unemployment: 0.154** (t-statistic: 2.129). - Expected households financial condition: 0.327*** (t-statistic: 6.591). - Time: -0.252*** (t-statistic: -3.314). - Constant: 30.969*** (t-statistic: 5.316). - Column (2) (adds Fiscal Balance (t-6) and Household debt (t-4)): Adjusted R-squared = 0.755. - Net financial assets (t-2): -0.113*** (t-statistic: -4.298). - Housing price (percent change, t-4): -0.131** (t-statistic: -2.695). - Expected unemployment: 0.179** (t-statistic: 2.089). - Expected households financial condition: 0.271** (t-statistic: 2.477). - Fiscal Balance (t-6): -0.220 (t-statistic: -1.010) — not statistically significant. - Household debt (t-4): 0.325** (t-statistic: 2.736). - Time: -0.952*** (t-statistic: -3.984). - Constant: 4.161 (t-statistic: 0.365). - Significance notation: ***p<0.01, **p<0.05, *p<0.1. - Interpretation: - Wealth effects: net financial assets and housing prices negatively associated with saving rate (every one percent of GDP increase in net financial assets ≈ 0.1ppt decrease in saving rate; housing price accelerations ~0.1–0.15ppt decrease). - Precautionary effect: expected unemployment positive — a 10 percentage point increase in percent balance ≈ 1.8ppt increase in saving rate. - Financial channel: expected household financial condition positive; may capture expected asset returns, incentivizing saving. - Debt and fiscal balance: household debt positive (higher indebtedness motivates saving); fiscal balance loses significance once unemployment expectations included. ### Projections and household debt scenarios (out-of-sample forecasts) - Forecast path for saving rate: - Expected to decline further to 4½ percent in 2014 and then gradually increase to about 12 percent by 2020. - Assumptions underlying projection: - Further reduction in housing prices by an estimated 10 percent. - Reduction in the unemployment rate in line with the macroeconomic forecast by 5.6 percentage points. - Improvement in the fiscal balance by about 4.8 percent of GDP. - Gradual decline of real interest rate (assuming no change in nominal interest rate and convergence of inflation rate to around 2 percent). - Amount saved each period contributes to increase in household net financial assets (via financial assets increase or liabilities reduction). - Forecast constant dropped in out-of-sample forecasting. - Household debt projections: - Baseline (households use about half of their savings to retire debt): household debt projected to decline to 102 percent of GDP in 2020. - Alternative (entire amount of saving used to pay down debt): indebtedness could fall to 85 percent of GDP by 2020. - Caveats: forecasts subject to considerable uncertainty; data noisy and endogeneity concerns. ### Housing market: stylized facts, valuation, and drivers of boom-bust - Stylized facts: - Housing prices accelerated since 2002; nearly 30 percent y-o-y in 2004; prices more than doubled between 2003 and 2008. - By 2014:Q1 prices fell by 26 percent from their 2008 peak. - Mortgage interest rates declined from above 7 percent in 2004 to below 5 percent in 2010. - Average growth of housing loans: 6.5 percent in 2002–05 and 17 percent in 2006–10. - Stock of dwellings increased by 25 percent during 2005–10; building permits 2007–09 higher by 70 percent versus 2002–03. - Peak overvaluation around 30–50 percent in 2008. - Valuation assessments: - Common ratios (20-year averages) indicate average valuation gap across methods about 7 percent at 2014:Q1; price-to-income about 20 percent above long-term average (using 2012 incomes reduces gap to 8 percent). - Model-based valuation gap (simultaneous supply-demand housing model): close to 14 percent in 2014:Q1; confidence interval ±10 percent. - Model suggests current gap driven by excessive growth in the debt-service ratio; equilibrium debt-service ratio assumed at 15 percent (euro area average). - Housing model elasticities and contributions (selected): - ln(real inward FDI into real estate) coefficient: 0.67***(0.06). - ln(real inward FDI into construction) coefficient: 0.003***(0.001). - Long-term elasticity of real house prices to income: 0.29. - Long-term elasticity to interest rate: 0.08. - Contributions to equilibrium house price growth (2002–14): total contribution 2.3; valuation gap in Q1-2014: 14.3. - Drivers of the boom and bust (quantified): - Inward FDI into real estate pushed up prices by an average of 12 percent per year during 2004–08. - Mortgage credit growth added another 7 percent per year to prices during 2004–08. - Housing supply increased and lowered prices by 9 percent per year on average during the boom. - After 2009, supply continued to put downward pressure by 5 percent annually on average; FDI disappeared; declining employment also contributed to declines. - Policy implication: further household deleveraging needed to facilitate housing price adjustment; speed/magnitude depend on employment and income performance; declining prices affect household wealth and banks’ collateral values, creating feedback loops. ### Banking sector strategy, recapitalization, payment restrictions, and NPLs - Program and banking strategy (approved in May): - Objectives: complete recapitalization, restructure sector, develop strategy to lift payment restrictions, strengthen regulation and supervision. - Recapitalization and restructuring implemented: - Bank of Cyprus (BoC): recapitalized via bail-in of uninsured deposits; 47.5 percent of uninsured deposits converted into equity; BoC ownership on exit: uninsured depositors of BoC 80 percent, legacy Laiki uninsured depositors 18 percent, other creditors/shareholders 1 percent; restructuring included conversion/release composition of deposits and time-deposit conversions. - Hellenic Bank (HB): capital shortfall under PIMCO adverse scenario ~€300 million; capitalization completed via sale of shares (€100 million) and conversion of junior debt (€250 million). - Cooperative sector: state nationalized and recapitalized Central Cooperative Bank (CCB) with €1.5 billion; consolidation from 93 to 18 institutions; capital ratio target 4 percent CET1; restructuring through 2017. - Payment restrictions and deposit flows: - Milestone-based gradual lifting strategy during 2013–14. - System lost around 15 percent of total deposit base (excluding bailed-in amounts) during March-December 2013; deposits broadly stabilized in 2014. - Implemented measures included withdrawal limits, export-of-notes limits, restrictions on checks and card use abroad, wire-transfer thresholds, prohibition to open new bank accounts, and extension of term deposits. - NPLs and credit conditions: - NPLs at 54 percent of total loans or 143 percent of GDP at end-June (source provides 45 percent household NPL earlier; sector NPLs 54 percent). - High NPLs strain capital and liquidity, hinder credit and recovery. - Two broad approaches to manage NPLs: maintain on banks’ balance sheets (internal workout/APS) or transfer to centralized agencies (SPVs/AMCs) with trade-offs. - Cyprus opted for internal workout given limited fiscal space: banks set up internal workout units; sector-wide arrears management framework and code of conduct; CCB-supported recapitalization. - Preconditions for success: adequate incentives for restructuring, ongoing legal reform for foreclosure/insolvency, avoidance of delay in loss recognition. - Cross-country lesson: NPL reduction takes time—3–5 years on average; GDP growth is main driver of turnaround. ### Lending rates, funding, and normalization prospects - Drivers of high lending rates: - High deposit rates at all maturities drive lending rates; reflect risk premia and high credit risk from private indebtedness and large NPLs. - Net interest margins about 50–100 basis points above European average to build provisions and protect capital. - Liquidity available but held as cash/low-yield instruments; weak credit demand and limited collateral constrain lending. - Short-run prospects to lower rates limited: - Banks already reduced cost-to-income ratio below European average; ECB programs (TLTRO) expected to have limited impact given collateral constraints and weak lending opportunities. - BoC reliant on central bank funding; market access limited; rebuilding deposit base costly and may hurt profitability. - Path to normalization: - Restoring confidence, successful bank restructuring, reduction of NPLs, and completion of EU Comprehensive Assessment are key to reduce funding costs and allow gradual convergence of deposit and lending rates. - Reform of foreclosure and insolvency legal framework critical to NPL resolution and credit resumption. ### Public finances: consolidation, outcomes, and structural issues - 2012 fiscal consolidation package: - Measures about 2.6 percent of GDP; broadly balanced between revenue (VAT, social contributions) and expenditure (targeting social transfers, reduce public employment through attrition). - Outcome: modest primary balance improvement of slightly below 1 percent of GDP. - Spending and revenue structure (end-2012): - Compensation of employees relative to GDP among the highest in Europe (second only to Denmark in Eurostat database by end-2012). - Other spending categories close to or below median; intermediate consumption 0.8 percent below median. - Consumption tax base high; labour tax base low (reflecting tax evasion and enforcement weaknesses); capital tax base large due to large bank deposit stock. - Program objectives (May 2013) and measures: - Target primary balance of 4 percent of GDP by 2018 to put debt on downward path toward 100 percent of GDP by 2020. - About 7.5 percent of GDP in measures legislated at end-2012 and implemented in 2013–14; projected additional adjustment of 4.7 percent of GDP left for future years. - 2013–14 measures: nominal salary cuts of 9–15 percent; VAT and reduced VAT increased by 2 and 1 percentage points (2013 and 2014); corporate tax increased from 10 to 12.5 percent; excise increases; property tax increases; withholding tax on interest increased from 10 to 30 percent. - Structural reforms: COLA reform (December 2012), pension reforms (January 2013), Fiscal Responsibility and Budget Systems Law (early-2014), revenue administration reform (mid-2014), welfare reform (mid-2014). - Fiscal outcomes (2012–13): - Overall and primary deficits declined by about 1 percent of GDP in 2013. - Tax revenues stabilized around 26 percent of GDP in deep recession. - Primary expenditure reduced in nominal terms by about 5 percentage points but increased by 0.8 percent as share of GDP. - Selected indicators (% GDP): Revenue 2012 = 39.4; 2013 = 41.4. Expenditure 2012 = 45.8; 2013 = 46.8. Primary Balance 2012 = -3.2; 2013 = -2.0. Overall Balance 2012 = -6.4; 2013 = -5.4. Public Debt 2012 = 86.6; 2013 = 111.5. - Medium-term fiscal gap and policy guidance: - Estimated additional permanent consolidation needed: 3.4 percent of GDP to reach a primary surplus of 4 percent of GDP by 2018 and sustain it. - Policy guidance: focus on expenditure-side consolidation (compensation of employees and social transfers account for 70 percent of primary general government expenditure). - Options include targeted wage adjustments (a 2 percent flat cut ≈ 0.3 percent of GDP), taxing public pension gratuities (yield about 0.2 percent of GDP; fully phasing out could yield up to 0.5 percent), reducing employment in select sectors (e.g., 2,000 renewable-contract workers could save about 0.4 percent of GDP in 3 years), better targeting of social benefits and housing benefits (up to 0.5 percent of GDP), tuition and higher-education transfer reforms (up to 0.6 percent of GDP), and SOE operational improvements (transfers ~0.8 percent of 2012 GDP). - Emphasis on permanent, high-quality, spending-focused measures and institutional reforms in revenue administration and public financial management. - Fiscal risks and public financial management weaknesses: - Fragmented revenue administration, deficient commitment control, lack of medium-term orientation, limited use of cost-benefit for investment projects, and pensions sustainability concerns (2012 Ageing Report projects pensions public expenditures increase by 8.4 percentage points of GDP during 2010–60). ### Potential output, financial-cycle adjustments, and long-term growth projections - Potential growth trends (cross-method comparisons and financial-cycle augmentation): - Potential Growth — PF: 3.9 (1996-2001), 3.4 (2002-08), 0.7 (2009-11), -1.2 (2012-13). - Potential Growth — MV: 3.6 (1996-2001), 1.4 (2002-08), -1.5 (2012-13). - Potential Growth — HP filter: 3.4 (1996-2001), 0.2 (2009-11), -1.7 (2012-13). - Financially-neutral: 3.0 (2002-08), 1.3 (2009-11), -1.7 (2012-13). - Financially-neutral model implies boom period potential 2002–08 = 3.0 percent (lower than traditional 3.4–3.6), implying larger boom output gap. - Drivers of potential growth deterioration: - 1996–2001: capital accumulation and productivity gains strong. - 2002–08: employment growth contributed most; credit-financed domestic demand and property boom; positive output gap opened in 2006–07. - 2009–11: slowdown after global crisis; property market bust hit construction and financial services. - 2012–13: abrupt investment decline; housing construction fell by 62 percent between 2008 and end-2012; banking crisis in 2013; unemployment close to 17 percent at end-2013; long-term unemployment 42½ percent of unemployed; youth unemployment 39½ percent. - Long-term potential growth projection (2015–20) — production function assumptions: - Labor: employment projected to grow around 1 percent per year. - Capital: capital accumulation expected to slow to 2.5 percent per year. - Productivity (TFP): linked to education trends at 1.3 percent per year. - Projected potential output growth rate for 2015–20: 2.0 percent. - Projected contributions to potential growth (2015–20): Capital 20.0 (contribution 0.4ppt), Labor 40.0 (contribution 0.8ppt), TFP 35.0 (contribution 0.7ppt), Residual -0.6; sum = 2.0 percent potential growth. - Sensitivities and medium-term output loss estimates: - Every one percentage point increase in employment growth raises potential output by 0.8 percentage point. - Every one percentage point increase in capital accumulation raises potential output by close to 0.2 percentage point. - Estimated losses: Actual output in 2015 estimated to be 31 percent below its level in the absence of the crisis; potential output 22 percent below its “no-crisis” counterfactual. - Cyprus suffered one of the largest output losses due to the crisis, after Greece and Ireland. *Source: IMF staff analysis and estimates in content unit _cr14314.* ### 1. The Evolution of Households’ Net Wealth _______________________________________________ 9 ### 1. The Evolution of Households’ Net Wealth ### A. Introduction and context - Over 2002–08, the household saving rate declined from a peak of 15.1 percent in 2003 to 6.7 percent in 2008; growth averaged 3.6 percent per year. - During 2009–11 the saving rate rebounded sharply as credit and housing bubbles burst and growth declined. - Since 2012 the saving rate fell rapidly, even as the domestic recession deepened; the decline in Cyprus’s saving rate appears to have been the deepest among similar boom-bust countries referenced (Ireland, Spain). - Private consumption constituted about 66 percent of aggregate demand over 2000–13. ### B. Theoretical determinants of household saving (literature summary) - Net financial and non-financial wealth: Higher real wealth leads households to consume more and save less (permanent income hypothesis). - Real interest rate: Affects saving via substitution and income effects. - Uncertainty: Higher uncertainty (e.g., rising unemployment) tends to increase precautionary saving; if shocks materialize, households may dip into savings to smooth consumption. - Fiscal balances: Loose fiscal policies may be associated with higher household saving (Ricardian effect), though alternative interpretations exist linking fiscal position and macroeconomic uncertainty. - Demographics: Lifecycle income hypothesis implies changes in population age structure affect aggregate saving. ### C. Empirical approach — two models - Data: OLS regressions using quarterly data from 2004:Q1 to 2013:Q4. - Definition: Household saving rate defined consistent with the source (approximation for disposable income justified by compensation ≈ 60 percent of disposable income). - Models: - Model 1 (change): Explains year-on-year change in the saving rate using lagged explanatory variables. - Model 2 (level): Explores determinants of the level of the saving rate (details and results summarized in tables and figures in the source). ### D. Explanatory variables used in the change model - Household net financial wealth: year-on-year change in ratio to GDP (net financial assets = household financial assets minus household financial liabilities). - Housing wealth proxy: year-on-year percent change in housing prices (housing price index). - Unemployment rate: year-on-year level change in harmonized unemployment rate. - General government fiscal balance: year-on-year change in ratio of four-quarter rolling average fiscal balance to four-quarter rolling average GDP. - Real deposit rate: year-on-year level change in rate of deposits redeemable at notice up to 3 months minus HICP inflation rate. - All explanatory variables enter with one lag to minimize endogeneity. ### E. Main regression findings (Determinants of the change of household saving rate) - Regression period: 2004Q1-2013Q4. Dependent variable: Year-on-year change of household saving rate. Adjusted R-squared = 0.817. - Coefficients and significance: - Net financial assets (change, lagged): coefficient = -0.099; t-statistic = -5.633; Probability = 0.000. - Interpretation given: Every one percent of GDP increase in net financial assets (i.e., €165 million) is associated with a 0.1ppt decrease in saving rate. - Housing price (percent change, lagged): coefficient = -0.147; t-statistic = -2.463; Probability = 0.020. - Interpretation given: Every one percent increase in housing price is estimated to lead to 0.15ppt decrease in saving rate. - Fiscal balance (change, lagged): coefficient = -0.681; t-statistic = -5.522; Probability = 0.000. - Interpretation given: Every one percent of GDP fiscal consolidation likely associated with 0.7ppt decline in saving rate. - Unemployment (change, lagged): coefficient = -2.226; t-statistic = -8.189; Probability = 0.000. - Interpretation given: Every one percentage point increase in unemployment rate is associated with a decline of saving rate by 2.2ppt. - Real deposit rate (change, lagged): coefficient = 0.453; t-statistic = 2.690; Probability = 0.012. - Interpretation given: Positive coefficient suggests substitution effects dominate; higher interest rates encourage saving. - Constant = 2.587; t-statistic = 4.395; Probability = 0.000. ### F. Decomposition of saving-rate dynamics (qualitative findings) - 2005–08 decline in saving rate largely explained by: - Increasing net financial wealth. - Rising housing prices. - Improvement in the fiscal position (to some extent). - 2009–11 rebound in saving rate mainly driven by: - Worsening of financial wealth. - Worsening of the fiscal position. - Since 2012 (deepening recession) the dominant factor inducing consumption smoothing and reducing saving rate is: - Rising unemployment. ### G. Box 1 — The Evolution of Households’ Net Wealth (key statistics and events) - Household net financial wealth increased to 147 percent of GDP by end-2012. - During 2005–08: - Households accumulated significant liabilities, which increased by 9 billion (33 percent of GDP). - Non-financial (housing) wealth rose by an estimated 6 billion (14 percent of GDP). - Financial wealth increased by 19 billion (59 percent of GDP). - 2013 banking crisis impacts: - Household financial assets fell by about €5 billion in 2013. - About €0.5 billion of deposit decline estimated due to the bank bail-in; remainder likely deposit drawdown to support consumption. - Decline in insurance and pension fund assets due to bail-in losses and valuation losses on stock portfolios. - Households used pension fund assets to finance consumption. - By end-2013 net financial wealth remained at 138 percent of GDP. - Household debt: - At 140 percent of GDP at end-2013, Cyprus’s household debt is among the highest in the euro area. - Debt interest payments as a share of disposable income rose to around 13 percent in 2013, up from 9.5 percent in 2006. - Household non-performing loans reached 45 percent of total loans at end-June 2014. - Distributional note: - Net financial wealth remains large and positive at end-2013 but distribution is less even than the euro-area average; some households with limited wealth may need to raise saving to repair balance sheets. *Source: IMF staff (Yinqiu Lu); chapter "The Cypriot Household Saving Rate" in the provided PDF content.* ### 8. The second model attempts to explain the level of the saving rate. The model is based ### 8. The second model attempts to explain the level of the saving rate. The model is based ### Model specification and channels - The model follows Carroll et al. (2012) and captures three channels: - Wealth channel: - Household net financial wealth (level in percent of disposable income). - Housing wealth (year-on-year percent change in housing prices). - Precautionary channel: - Percent balance of consumers’ unemployment expectations over next 12 months (percent balance) to reflect economic uncertainty and precautionary saving motives. - Financial channel: - Percent balance of households’ expectations about their financial situation over next 12 months (percent balance) as a proxy for credit availability (lending survey data for Cyprus not available until 2009). This indicator may also reflect rising income prospects and should be interpreted with caution. - Main equation (notation preserved as in source): ܵ௧ܺߙ൅ܿൌ௜,௧ି௝ߝ൅ݐߛ൅௧, where ܵ௧ is the household saving rate and ܺ௜,௧ి௝ represents the explanatory variables listed above with lags chosen based on Akaike’s information criterion. A time trend is included. - Alternative specification adds two variables (represented by ܤ௜,௧ి௣): general government fiscal balance, and household debt to disposable income ratio. ### Data notes and variable construction - Percent balance definitions (from EC Business and Consumer Survey): - Unemployment expectations percent balance = percent expecting increase in unemployment over next 12 months minus percent expecting decrease. Quarterly series constructed by averaging monthly data. - Financial situation percent balance = percent reporting an improvement in households’ financial situation over next 12 months minus percent reporting deterioration. Quarterly series constructed by averaging monthly data. - Other tested variables (loan deposit ratios, interest spread) were insignificant. ### Estimated coefficients and model fit (Table 2 results; Dependent variable: Level of Household Saving Rate; Period: 2004Q1-2013Q4) - Column (1): - Net financial assets (t-2): -0.066*** (t-statistic: -3.292) - Housing price (percent change, t-4): -0.153*** (t-statistic: -2.833) - Expected unemployment: 0.154** (t-statistic: 2.129) - Expected households financial condition: 0.327*** (t-statistic: 6.591) - Time: -0.252*** (t-statistic: -3.314) - Constant: 30.969*** (t-statistic: 5.316) - Adjusted R-squared: 0.648 - Column (2) (adds Fiscal Balance (t-6) and Household debt (t-4)): - Net financial assets (t-2): -0.113*** (t-statistic: -4.298) - Housing price (percent change, t-4): -0.131** (t-statistic: -2.695) - Expected unemployment: 0.179** (t-statistic: 2.089) - Expected households financial condition: 0.271** (t-statistic: 2.477) - Fiscal Balance (t-6): -0.220 (t-statistic: -1.010) — not statistically significant - Household debt (t-4): 0.325** (t-statistic: 2.736) - Time: -0.952*** (t-statistic: -3.984) - Constant: 4.161 (t-statistic: 0.365) - Adjusted R-squared: 0.755 - Note: T-statistics in brackets. ***p<0.01, **p<0.05, *p<0.1. ### Interpretation of key coefficients and signs - Wealth effects: - Net financial assets: every one percent of GDP increase in net financial assets is associated with a 0.1ppt decrease in saving rate (coefficients: -0.066 and -0.113 reported). - Housing prices: a one percentage point acceleration in the change of housing price is associated with a 0.1ppt decrease in the saving rate (coefficients: -0.153 and -0.131). - Precautionary effect: - Expected unemployment: positive coefficient; every 10 percentage point increase in the percent balance is associated with a 1.8ppt increase in the saving rate (coefficients 0.154 and 0.179 imply this mapping). - This positive sign contrasts with the negative coefficient of actual unemployment in the previous model and reflects precautionary saving when households expect higher future unemployment. - Financial channel: - Expected household financial situation: positive coefficient (0.327*** and 0.271**) contrary to expectation that higher credit availability would reduce saving. The chosen index may capture an expectation of higher return on assets, which incentivizes saving. - Debt and fiscal balance: - Household debt (t-4): positive coefficient (0.325**) — higher indebtedness motivates households to save. - Fiscal balance: not statistically significant in column (2); once macroeconomic uncertainty is accounted for (via unemployment expectations), Ricardian effects appear to disappear. ### Decomposition of historical contributions - 2007–08 decline in saving rate: - Largely explained by increase in net financial wealth and to some extent decline in unemployment expectations. - 2009–11 saving rate dynamics: - Mainly driven by worsening of both financial and housing wealth. - Since 2012: - Worsening expected financial conditions appear to be the leading factor for the decline in the saving rate. - Over the whole period: - Rising household debt acts to curtail the decline in the saving rate. ### Projections (Out-of-sample forecasts and assumptions) - Forecast path: - Saving rate expected to decline further to 4½ percent in 2014 and then gradually increase to about 12 percent by 2020. - Assumptions underlying the projection: - Further reduction in housing prices by an estimated 10 percent. - Reduction in the unemployment rate in line with the macroeconomic forecast by 5.6 percentage points. - Improvement in the fiscal balance by about 4.8 percent of GDP (in line with authorities’ medium-term fiscal targets). - Gradual decline of real interest rate (assuming no change in nominal interest rate and convergence of inflation rate to around 2 percent). - Assumption that the amount saved by households in each period contributes to an increase in household net financial assets in each period (through either an increase in financial assets or a reduction in liabilities). - Forecast methodology note: - The constant is dropped in the out-of-sample forecasting, as no dynamic feedback is built in the model, and, in an equilibrium state, the constant should be zero. - 10 percent housing price reduction: noted as close to the mid-point of the estimated price gap of 7–14 percent relative to equilibrium. ### Household debt projections and scenarios - Baseline projection (assuming households use about half of their savings to retire debt and accounting for feedback between saving rate and net financial assets): - Household debt projected to decline gradually to 102 percent of GDP in 2020, consistent with the upper range of estimated deleveraging needs toward a more sustainable debt level. - Alternative projection (if the entire amount of saving is used to pay down debt): - Household indebtedness could fall to 85 percent of GDP by 2020. ### Conclusion and policy-relevant findings - Main findings: - Household net wealth and unemployment expectations are key determinants of the saving rate in Cyprus. - Pre-crisis period: Cypriot households dissaved as wealth increased and credit financed consumption. - Post-2008: Households initially increased saving as a precautionary measure as macroeconomic uncertainty and unemployment rose, then reverted to a declining saving rate to smooth consumption once shocks materialized. - Forward-looking: household saving rate expected to increase to 12 percent in 2020, accompanied by a decline in household debt. - Policy implications (implicit in results): - Monitoring household net financial wealth, housing prices, unemployment expectations, household debt, and expected household financial conditions is important for understanding saving behavior and designing policies to support deleveraging and sustainable consumption smoothing. ### Caveats, data limitations, and estimation concerns - Data issues: - Data is noisy; some series required generating quarterly data from annual trends. - Data uncertainty due to multiple data sources and relatively short time-period may affect regression results. - In compensation of employees, both annual and quarterly data are used; quarterly figures do not precisely sum to annual figures but differences are positive and consistent over time. - Endogeneity and identification: - Endogeneity between household wealth and the saving rate, and between unemployment and the saving rate, may not have been fully controlled through lags. - Distributional considerations: - Lack of micro-level data prevents exploration of distributional aspects of wealth. Wealth is likely unevenly distributed; high-indebted households with limited wealth may reduce saving more than average to support consumption under stress. - Forecast uncertainty: - Forward-looking projections are subject to considerable uncertainty and should be interpreted with care. - Statistical robustness: - Other potential credit-availability variables were tested (loan deposit ratios, interest spread) and found insignificant. --- *Source: Excerpt from IMF staff analysis in _cr14314 - 8. The second model attempts to explain the level of the saving rate. The model is based (PDF chapter).* ### 8. Finally, we also use an alternate methodology that takes into account the financial ### _cr14314 - 8. Finally, we also use an alternate methodology that takes into account the financial ### Methodology: incorporating the financial cycle - The model follows Borio et al. (2013), augmenting the output gap equation in the HP filter with financial cycle variables, such as credit growth and property prices. - Credit growth and housing price are included to estimate the “financially-neutral” output gap for Cyprus, reflecting the large pre-crisis credit expansion and property market boom and bust. ### Results from standard and financial-cycle-augmented models - Standard methods indicate potential growth in Cyprus has been on a declining trend since the mid 1990s. - Augmented production-function (PF) estimates: - Potential growth was rapid during 1996–2001, slightly declined up to 2008, and slowed considerably after 2008. - Potential growth turned negative in 2012 and further deteriorated with the 2013 banking crisis. - Cross-method comparisons (selected period averages and end periods): - Potential Growth — PF: 3.9 (1996-2001), 3.4 (2002-08), 0.7 (2009-11), -1.2 (2012-13) - Potential Growth — MV: 3.6 (1996-2001), 1.4 (2002-08), -1.5 (2012-13) - Potential Growth — HP filter: 3.4 (1996-2001), 0.2 (2009-11), -1.7 (2012-13) - Potential Growth — Financially-neutral: 3.0 (2002-08), 1.3 (2009-11), -1.7 (2012-13) - The financially-neutral model estimates: - Boom period (2002–08) potential: 3.0 percent (lower than traditional estimates of 3.4–3.6 percent), implying a larger boom output gap. - Bust period: 1.3 percent (2009–11) and -1.7 percent (2012–13), broadly in line with other methods. ### Analysis of potential output developments through end-2013 - 1996–2001: - Potential growth driven by capital accumulation (infrastructure and housing) and productivity gains (TFP), with strong growth in high productivity sectors (financial services). - A negative output gap opened in 1996–99 due to a strong earthquake and Green line disruptions; gap gradually closed over 2000–11. - 2002–08: - Employment growth contributed most to potential growth; capital accumulation and productivity contributions slowed relative to the previous period. - Rapid expansion of credit financed domestic demand and a property boom; construction sector expanded using immigrant labor. - A positive output gap opened in 2006–07 as activity outpaced potential. - 2009–11: - Potential growth slowed following the global crisis; property market bust hit construction and financial services, reducing employment growth. - Actual output fell in 2009, recovered moderately in 2010, then fell below potential in 2011 after power plant explosion. - 2012–13: - Investment declined abruptly in 2012; housing construction fell by 62 percent between 2008 and end-2012. - Banking crisis in 2013 led to broad contraction across sectors. - Unemployment rose to close to 17 percent at end-2013 (almost double end-2011 and four times end-2008); long-term unemployment was 42½ percent of the unemployed at end-2013; youth unemployment rose to 39½ percent at end-2013. - Potential output growth (PF approach) estimated at around -1.2 percent, driven by falling employment and declining productivity. ### Projecting long-term potential growth (2015–20) — production function assumptions and projections - Assumptions: - Labor: Employment projected to grow at around 1 percent per year (in line with Eurostat’s population growth projections); lower than pre-crisis due to likely loss of human capital, hysteresis effects, and absence of large immigrant inflows. - Capital: Capital accumulation expected to slow to 2.5 percent per year (versus average 5 percent per year pre-crisis). - Productivity (TFP): TFP growth linked with long-term education trends at 1.3 percent per year. - Projected contributions to potential growth (Sum = 100): - 1996-2008: Capital 30.9; Labor 47.7; TPF 21.3 - 2015-20: Capital 20.0; Labor 40.0; TPF 35.0 - Projected potential output growth rate for 2015–20: 2.0 percent (lower than pre-crisis average). - Future Potential Growth (2015-2020) components: - Capital: 2.5 (average), contribution 0.4ppt (0.2ppt in sensitivity table) - Labor: 1.0 (average), contribution 0.8ppt (0.8ppt in sensitivity table) - TFP: 0.7 (with education 1.6 / 1.3 shown in table) - Residual: -0.6 - Potential Growth: 2.0 ### Sensitivity and medium-term output loss estimates - Sensitivities: - Every one percentage point increase in employment growth raises potential output by 0.8 percentage point. - Every one percentage point increase in the rate of capital accumulation raises potential output by close to 0.2 percentage point. - Medium-term output loss methodology: - Actual output (Y) uses actual data and latest macro projections. - Potential output (Y*) estimated from the production function. - Counterfactual potential output without the crisis (Y**) taken from IMF (2010) production function projections through 2010. - Estimated losses (comparative and absolute): - Actual output in 2015 is estimated to be 31 percent below its level in the absence of the crisis. - Potential output is 22 percent below its “no-crisis” counterfactual. - Cyprus is estimated to have suffered one of the largest output losses due to the crisis, after Greece and Ireland. - Pashardes and Pashourtidou (2013) estimated output loss relative to projected GDP for 2012–20 around 17–48 percent under different potential growth assumptions. ### Concluding remarks and policy recommendations - Main findings: - Potential growth was strong but gradually declining over 1996–2008, with an accelerated decline in 2009–11 and negative potential growth since 2012 due to the domestic downturn and 2013 banking crisis. - The crisis has likely caused long-lasting damage to both physical and human capital, reducing long-term potential growth to about 2 percent. - Policy recommendations to boost long-term potential growth: - Improve labor market outcomes and labor productivity; ensure wages and productivity remain aligned. - Active labor market policies to reactivate the unemployed, especially vocational training and educational programs for workers. - Structural reforms where Cyprus lags best practice, including: - Opening up closed professions by streamlining licensing restrictions. - Removing barriers to competition, including protection of firms and price controls. - Reducing red tape. - Strengthening the legal system. - Fostering innovation. - Estimated impact of structural reforms in these areas: 1.5–3 percentage points (Cheptea and Velculescu, 2014; and IMF, forthcoming). *Source: IMF staff estimates and analysis contained in the provided content unit.* ### Appendix I. Methodology ### Appendix I. Methodology ### Multivariate Model - Framework: Model of Benes et al. (2010) with three gap equations evolving: - Output gap (y), unemployment gap (u), capacity utilization gap (c). - Inflation equation links level and change of output gap to annual inflation, 4ߨ, and includes inflation expectations that follow a random walk. - Okun’s law links unemployment gap u to the output gap y. - Capacity utilization gap c is modeled as related to the output gap to help improve potential output and output gap estimates. - Equilibrium variables evolve dynamically linking potential output (ܻത௧), labor share (ߠ), an output growth trend (ܩ௧௒ത), NAIRU (ܷഥ௧), steady-state NAIRU (ܷௌௌ), growth in equilibrium unemployment (ܩ௧௎ഥ), and growth in trend capacity utilization (ܩ௧஼̅), with specified autoregressive forms for trends (e.g., ܩ௧௒തܩ߬ and ܩ௧௎ഥ involving terms like (߬1െ) and (ߙ1െ)). - Estimation: full system estimated by regularized maximum likelihood (Ljung, 1999), a Bayesian methodology requiring priors. - Assumed steady-state levels and labor share (exact values): - Trend growth ܩௌௌ௒ത: 1.8 - Long-run equilibrium unemployment ܷௌௌ: 6.1 - Labor share ߠ: 0.83 ### Financially-neutral Estimate - Basis: Borio et al. (2013) approach embedding financial variables into a state-space HP filter to capture output gap movements driven by financial conditions without directly changing potential output. - State-space HP filter specification expresses observed log real GDP (ݕ௧ = ln(ܻ௧)) and a latent potential component, with normally and independently distributed errors ߝ. - Financial-augmented equation: output gap equation augmented by vector of financial variables ݔ௧, which affects the output gap but not potential output. - Financial variables included (as used in this paper): - Real credit growth (credit growth adjusted for HICP inflation) - Real growth rate of residential housing price - Real growth of general index of Cyprus stock exchange - HICP inflation rate - Industry capacity utilization - Data sources and sample: - Credit: Central Bank of Cyprus - Housing price, HICP, industry capacity utilization: Eurostat - Stock exchange index: Cyprus Stock Exchange - Frequency and period: Quarterly data covering 1995: Q1 to 2013: Q4 (to the extent data available) - Estimation: Bayesian estimation of the state-space model. ### Cyprus Housing Market — Stylized Facts and Model Inputs - Boom and bust chronology and magnitudes: - Housing prices increasing trend since 2002; acceleration in 2004 after EU entry with residential property prices increasing by nearly 30 percent y-o-y in 2004. - Prices more than doubled between 2003 and 2008. - Housing bubble burst with global crisis; by 2014:Q1 prices fell by 26 percent from their 2008 peak. - During boom: mortgage interest rates declined from above 7 percent in 2004 to below 5 percent in 2010. - Average growth of housing loans: 6.5 percent in 2002–05 and 17 percent in 2006–10. - Stock of dwellings increased by 25 percent during 2005–10. - Issuance of building permits in 2007–09 higher by 70 percent compared to 2002–03. - Construction costs (excluding land) increased at an average yearly rate close to 5 percent. - Cyprus peak overvaluation around 30–50 percent in 2008; Estonia experienced 45–65 percent overvaluation in 2007. - Cross-country timing: Cyprus boom similar to Ireland and Spain on the upside but price adjustment on the downside slower (Cyprus 25 percent decline over five years versus 60–70 percent declines in Ireland and Spain over six years). - Historical comparisons: - In the two years prior to peak, real house prices in Cyprus increased by 33 percent (Helbling (2005) cross-country average 32 percent). - During the bust, real house prices declined by 26 percent from peak (average historical decline 27 percent over 4–5 years). ### Housing Market Valuation Methods and Findings - Step 1 — Common ratios (deviations from 20-year long-term averages): - Ratios used: - Real house prices (house prices relative to overall price level) - Price-to-income ratio - Price-to-cost ratio (house prices including land, construction costs, developers’ markup divided by construction costs) - Long-term average computed over last 20 years for consistency and availability. - Findings at 2014:Q1: - Average valuation gap across methods about 7 percent. - Real house prices and price-to-cost ratio approximately at long-term levels. - Price-to-income ratio about 20 percent above its long-term average (partly due to sharp income declines in the recession). - Using 2012 income level for price-to-income ratio reduces the valuation gap to 8 percent. - Alternative long-term benchmarks (1995–2005 or 1995–2008) could increase valuation gaps by 10/15/20 percent under the price-to-income/price-to-cost/real house price ratios respectively. - Peak overvaluation and cross-country status: - Cyprus peak overvaluation around 30–50 percent in 2008 (second largest in euro area after Estonia). - Current euro-area average house prices adjusted to long-term values, but cross-country heterogeneity remains: Cyprus, France, Belgium, Austria still show some overvaluation; Estonia, Ireland, Spain, Greece undershot long-term values by 10–30 percent. - Step 2 — Simultaneous supply-demand housing model (log specification, building on Kasparova and White (2001), Tumbarello and Wang (2010), extended for FDI): - Demand equation: Real house prices (P) depend on housing stock (S), employment relative to population (E/pop), inward FDI into real estate (FDIre), and interaction between average income (Y) and debt-service ratio (w). - Debt-service ratio w computed from net present value of average income with assumed loan maturity: - Average remaining maturity assumed 15 years, so number of quarterly outstanding repayments n = 60. - Supply equation: Housing stock (S) is function of housing investment and lagged stock; investment depends on interest rate (r), price of dwellings (P), and inward FDI into construction (FDIc). Time to build assumed about 2–3 quarters. - Empirical diagnostics: - Correlation between change in housing stock and building permits strongest (95 percent) when using two to three quarterly lags. ### Housing Market Model Results (selected) - Estimation sample: 2002:Q1–2014:Q1. - Estimated elasticities and signs (all parameters reported as significant and with expected sign): - A 1 percentage point increase in real FDI into real estate raises real house prices by 0.67 percentage points. - A 1 percentage point increase in the share of income servicing mortgage raises real house prices by 0.54 percentage points. - A 1 percentage point increase in employment relative to population raises real house prices by 0.75 percentage points. - A 1 percentage point increase in the stock of dwellings lowers real house prices by 3 percent. - Housing supply increases by 0.003 (model text truncated here; full supply coefficient context in source). *Appendix I. Methodology — _cr14314* ### 0.012 percentage points with a percentage increase in real FDI into construction and house price ### _cr14314 - 0.012 percentage points with a percentage increase in real FDI into construction and house price ### Key model findings on drivers of the boom and bust - Inward FDI into the real estate pushed up prices by an average of 12 percent per year during 2004–08. - Mortgage credit growth added another 7 percent per year to prices during 2004–08. - Housing supply increased and lowered prices by 9 percent per year on average during the boom period. - As supply continued to rise after the downturn started in 2009, it continued to put downward pressure on house prices by 5 percent annually on average. - Declining employment contributed to price declines; FDI disappeared after the crisis onset. - Bank mortgage lending supported prices until the trend reversed in 2013. ### Estimated elasticities and coefficients (selected) - Long-term elasticity of real house prices to income: 0.29. - Long-term elasticity of real house prices to interest rate: 0.08. - Coefficient on ln(real inward FDI into real estate) in the model: 0.67***(0.06). - Coefficient on ln(real inward FDI into construction) in the model: 0.003***(0.001). - Coefficient on interest rate (housing supply equation): -0.002***(0.001). - Autoregressive coefficient ln(real house prices): 0.012***(0.004). - Lagged housing stock coefficients: ln(stock of dwellings, lagged two quarters) = 2.1***(0.17); ln(stock of dwellings, lagged three quarters) = -1.12***(0.17). - Intercepts reported: 39.8***(4.7) and 0.53***(0.08) for respective equations. - Observations used: 4747. - R2 values reported: 0.96 and 0.99 for model equations shown. ### Valuation gap and equilibrium assessment - The model suggests a valuation gap relative to equilibrium of close to 14 percent in 2014:Q1. - Valuation gap derived as difference between actual real house price growth rate during 2002–2014:Q1 and the equilibrium cumulative growth rate. - The current gap of 14 percent is driven by the excessive growth in the debt-service ratio. - House prices had already declined by 26 percent since their peak in 2008. - Common house price ratios indicate an overvaluation gap around 0–20 percent, with an average across methodologies of 7 percent. - Model-based valuation gap: 14 percent. - The valuation gap carries a confidence interval of ±10 percent. ### Contributions to equilibrium house price growth (long-term growth 2002–14, y-o-y percent) and valuation gap components (Table 2) - dln(real income): Coefficient 0.29; Long-term growth 1.2; Contribution to equilibrium house price growth 0.3. - dln(debt service ratio) 1/: Coefficient 0.29; Long-term growth 1.0; Contribution 0.3. Note: Sustainable debt service ratio assumed at 15 percent - the euro area average (ECB, 2013). - dln(employed/population): Coefficient 0.4; Long-term growth -0.6; Contribution -0.3. - dln(real inward FDI into real estate): Coefficient 0.36; Long-term growth 13.7; Contribution 5.0. - dln(real inward FDI into construction): Coefficient -0.12; Long-term growth 26.3; Contribution -3.0. - d(interest rate): Coefficient 0.08; Long-term growth -0.1; Contribution 0.0. - Total contribution to equilibrium house price growth: 2.3; Valuation gap in Q1-2014: 14.3. ### Mechanics and interpretation from the reduced-form exercise - The reduced-form demand equation was derived by substituting supply into the demand function using estimated parameters from 3SLS. - Rapid credit growth and FDI were primary demand-side drivers of the boom; rising supply in a declining economy with falling employment contributed to the bust. - The contribution of real income to price increases was small, suggesting banks relaxed lending standards by relying more on collateral values than on borrowers’ capacity to repay. - The equilibrium debt-service ratio used in calculations: 15 percent. ### Policy implications and adjustment dynamics - Current gap driven by excessive debt-service ratio implies further household deleveraging is needed to facilitate housing price adjustment toward equilibrium. - The speed and magnitude of further house price adjustment will depend on the performance of the economy (employment and income growth could buffer price falls). - House prices could undershoot equilibrium if income and employment continue to decline. - Declining house prices affect household wealth and consumption, and lower collateral values impair banks’ balance sheets and lending capacity, creating feedback loops that can exacerbate price falls in the short run. ### Data, estimation method, and caveats - Price of dwellings derived from Central Bank of Cyprus (CBC) quarterly residential property price index (2006–2014) and earlier CBC index (2002–05) which averages Pashardes-Savva, BuySell, and internal CBC indices; earlier series for 1995–01 interpolated from 1983 and 2002 price measures. - Other data sources: Cystat, Haver, and CBC (housing stock, construction costs, building permits, earnings, housing loans by MFIs, FDI, population, employed persons, interest rates on housing loans). - The model estimated using 3SLS following Greene (2003): first-step OLS to obtain fitted values, second-step structural estimation, third-step to introduce non-zero covariances and improve efficiency. - Point estimates sensitive to choice of long-run benchmarks; alternative plausible long-run values affect equilibrium growth and valuation gap. - Reported tests: White's heteroskedasticity test 1/χ2(34) =45; (0.11) and χ2(34) =37; (0.31); Durbin-Watson autocorrelation test 1/d(8,46)=0.8 and d(8,46)=1.5. - Standard errors shown in parentheses; significance denoted: *** p<0.01 and ** p<0.05. Real values obtained using HICP. *International Monetary Fund — Selected excerpts from the Cyprus housing market analysis and related sections in the provided content* ### 8. A program supported with international financial assistance was approved in May. ### _cr14314 - 8. A program supported with international financial assistance was approved in May. ### Program Objectives and Banking Sector Strategy - A program supported with international financial assistance was approved in May. - The comprehensive banking sector strategy aimed at: - completing the recapitalization of the banking sector; - restructuring the sector; - developing a strategy to gradually lift payment restrictions; and - strengthening regulation and supervision. ### Recapitalization and Restructuring Steps (Implemented) - Bank of Cyprus (BoC) recapitalization: - Recapitalization was completed through additional bail in of uninsured deposits. - BoC (including assets transferred from Laiki) was subject to a separate fair-value assessment to reassess capital needs required to maintain a CT1 capital ratio above the regulatory minimum of 9 percent through end-2015. - Total share of BoC uninsured deposits converted into equity amounted to 47.5 percent. - Upon exit from resolution at end-July, ownership structure of BoC: - uninsured depositors of BoC owned 80 percent of the bank; - legacy Laiki uninsured depositors received an 18 percent stake in BoC in exchange for assets transferred to BoC; - the rest (1 percent) owned by other BoC creditors and shareholders. - Remaining frozen uninsured deposits were restructured and converted into time deposits with 6, 9, and 12 month maturities, extendable by another 6, 9, and 12 months as needed. - Conversion/release composition noted: - 12.5% converted into 12 month time dep. - 12.5% converted into 9 month time dep. - 12.5% converted into 6 month time dep. - 10% released - 15% released - 22.5% blocked as buffer for final bail-in - 37.5% converted into equity - 47.5% converted into equity - 30% blocked - Hellenic Bank (HB) recapitalization: - After sale of its Greek operations, HB’s bank capital shortfall under PIMCO’s adverse scenario amounted to about €300 million, although the bank remained solvent, with positive equity. - Authorities launched an offer participation via equity participation and conversion of junior debt into Tier I convertible instruments. - Successful completion of the offer at end-October resulted in full capitalization through: - sale of shares (€100 million) to domestic and foreign investors; and - conversion of junior debt (€250 million). - Cooperative credit sector recapitalization: - The cooperative credit sector was fully recapitalized by the state. - In March 2014 the state nationalized and recapitalized the Central Cooperative Bank (CCB) by injecting €1.5 billion. - Consolidation of the sector from 93 into 18 institutions. - Institutions received capital from the CCB to ensure a capital ratio of 4 percent CET1. - The CCB put in place a restructuring plan for the sector to be implemented through 2017. ### Payment Restrictions, Deposit Flows, and Stabilization - Milestone-based strategy implemented during 2013–14 to gradually lift domestic payment restrictions as key banking sector steps were achieved. - Deposit outflows during 2013: - System lost around 15 percent of the total deposit base (excluding bailed-in amounts) during March-December 20–13. - Deposits broadly stabilized in 2014. - Payment and transfer measures (as implemented): - Cash: - Withdrawal limit: €9,000 per month - Export of bank notes limit: €2,000 per journey - Checks, credit, debit: - Cashing of checks prohibited - Use of credit card abroad limited to €5,000 per person per month - Wire-transfers: - Business transfers: > €300,000 domestic; > €20,000 cross-border — subject to approval - Individuals may transfer up to €3,000 to another domestic bank - Legal persons may transfer up to €50,000 to another domestic bank - Prohibition to open new bank accounts - Extension of term deposits ### Current Challenges and Key Statistics - Non-performing loans (NPLs): - NPLs have risen rapidly, standing at 54 percent of total loans or 143 percent of GDP at end-June. - Growing NPLs strain capital and liquidity positions, hinder credit and economic recovery. - Credit risk and lending prospects: - Credit risk remains high; prospects for bank lending stay weak, reflected in high lending rates and strained bank-funding conditions. - Public finances constraint: - Public sector debt exceeding 110 percent of GDP and still rising, limiting fiscal space to provide direct support such as via an AMC. ### Approaches to Managing NPLs (Cross-Country Options and Trade-offs) - Two broad approaches used internationally: - Maintain NPLs on banks’ balance sheets (decentralized/internal approach). - Transfer NPLs partially or fully to centralized agencies (centralized approach). - Decentralized / Internal options: - Bank internal-restructuring units: - Banks manage NPLs via specialized units; examples: Iceland, Portugal, Denmark, Greece; used selectively in Spain and Ireland. - Retains upside to banks but keeps risks on balance sheets for years; can distract from core operations. - Asset-protection schemes (APSs): - State provides insurance/guarantee while NPLs remain on balance sheets. - Shifts downside risk to public sector, may create contingent liabilities and recurring fees for banks; does little to facilitate workout. - Centralized options: - Bank Special-Purpose Vehicles (SPVs): - NPLs transferred to a separate legal entity, often subsidiary of the bank; risks often remain consolidated with the parent bank initially. - Examples: Sweden (1992), Germany (2009). - Asset-Management Companies (AMCs): - NPLs sold to a separate entity majority-owned by third-party investors; banks crystallize losses upfront. - Variants: - Private-sector AMCs: require willing buyers; often feasible only for narrowly defined portfolios. - Private-Public AMCs: state shares risk via equity, funding, or guarantees (examples: Ireland NAMA, Spain SAREB); potential political interference and EUROSTAT consolidation issues. - Public AMCs: fully state-owned and funded (examples: U.S. Resolution Trust Corporation (1989), Finland “Arsenal” (1993), Sweden “Securum” (1992)); require strong governance and oversight. - Cross-country lessons: - AMCs effective for clearly defined portfolios; need strong legal framework, political independence, liquefiable assets, professional management, skilled staffing, effective information systems. - Internal workout success depends on an effective NPL resolution framework, proper incentives, sufficient capital, and provisioning rules that prohibit forbearance. - Reducing NPLs takes time: 3–5 years on average for a downward path in many cases; GDP growth is a main driver of turnaround in NPLs. ### Cyprus Strategy and Preconditions for Success - Cyprus opted for an internal workout strategy given limited fiscal space. - Banks set up internal workout units and procedures. - Policies guided by a sector-wide arrears management framework and code of conduct. - Banks required time to build operational capacity to deal with NPLs. - Critical dependencies for success: - Adequate incentives for borrowers and lenders to reach restructuring solutions. - Ongoing reform of foreclosure and insolvency legal framework to: - Provide the threat of foreclosure to incentivize borrower payment or negotiation; - Provide incentives for banks to proactively restructure NPLs; - Ensure borrowers can meet renegotiated obligations while lenders receive cash-flow and avoid additional capital charges (after provisions are utilized). - Foreclosure should be a last resort solution. - Risk of delay: - Bank shareholders—largely uninsured depositors and junior bond holders who already suffered large losses—may be reluctant to recognize additional losses and instead wait for potential upside, which could create a vicious cycle of weak balance sheets, weak credit, weak economy, and worsening NPLs. *Source: IMF staff summary of the provided text.* ### 23. Cyprus’s lending rates remain very high in cross-country comparison. High deposit rates ### 23. Cyprus’s lending rates remain very high in cross-country comparison. High deposit rates ### Drivers of high lending rates - High deposit rates appear to be the key driver of lending rates; they are high at all maturities, reflecting the risk premia demanded by investors to maintain funds in the Cyprus banking system. - Banks face high credit risk given high private sector indebtedness and large NPLs. - Net interest margins (NIM) are currently about 50–100 basis points above the European average, and are needed to allow banks to build provisions and protect their capital base. - Liquidity appears to be less of a constraint: even banks with ample funding maintain it as cash or low-yielding instruments rather than use it for lending, likely due to weak credit demand. ### Short-run prospects for lowering lending rates - Prospects for lowering lending rates in the short-run are limited. - Banks have little room to further improve costs, having already reduced their cost to income ratio to below the European average by closing branches, reducing staff, and cutting wages and other operational expenses. - New ECB programs such as the TLTRO are expected to have only a limited impact on Cyprus. To make use of these programs—which allow banks to borrow commensurate with the amount of net lending to the economy, excluding mortgage loans—banks need ECB-eligible collateral and opportunities to lend; some Cyprus banks have limited collateral, while demand for credit is weak. ### Funding structure, market access, and deposit base - BoC remains highly reliant on central bank funding, which—while comparatively inexpensive—comes with a perceived stigma. - Shifting toward market funding would help boost confidence, but access to wholesale markets is limited, in part due to limited collateral buffers. - Junior funding could be an option but is relatively more expensive. - Banks need to rebuild their deposit base; to attract cash outside the system and ensure that the gradual removal of remaining external payment restrictions is not disruptive, they need to offer competitive rates and longer maturities, which will be relatively expensive and hurt profitability. ### Restoring confidence, NPLs, and normalization of rates - Restoring confidence in the system is needed to allow for an eventual reduction in both deposit and lending rates. - Authorities need to continue implementing their banking strategy to strengthen investor and depositor confidence. The recent recapitalization of BoC is a key step toward normalization of its operations. - Successfully meeting the requirements of the ongoing pan-European comprehensive bank assessment will be another important milestone. - Progressing with bank restructuring and the reduction of NPLs and an eventual gradual lifting of remaining payment restrictions are essential to strengthen confidence and eventually allow for a normalization of deposit and lending rates and the resumption of credit growth. ### Broader banking-sector context and conclusion - Cyprus has taken unprecedented measures to stabilize its banking system: banks have been recapitalized using private and public sector resources, the sector has consolidated and deleveraged significantly, and liquidity and deposit situation has improved despite certain constraints on capital movements remaining in place. - The recovery of the Cypriot banking sector hinges primarily on the successful and sustainable resolution of NPLs. High NPLs remain the key vulnerability, exposing balance sheets to significant downside risks and binding capital and operational capacity, thereby limiting new lending while the economy remains in recession and recovery is expected to be modest. - Uncertainties related to NPLs contribute to high funding costs, translating into lending rates that further constrain credit demand. - A successful completion of the ongoing EU Comprehensive Assessment can help boost confidence, reduce liquidity pressures, and allow for a gradual convergence of deposit rates with the EUR average along the yield curve. - Progress on normalizing liquidity and funding conditions will shape the future strategy for exiting from capital controls; while core banks’ need to access market funding would partly offset funding cost savings, it can help increase the safety and stability of the banking sector. - Addressing NPLs resolutely, including reform of the legal regime for foreclosures and insolvency, is critical to banks’ long-term viability and the resumption of credit and growth; cross-country experience suggests this takes time and resolve. *Source: IMF staff report excerpt (section 23 and related concluding passages).* ### 13. A fiscal consolidation package was implemented in 2012 (Box 2). With exception of a ### 13. A fiscal consolidation package was implemented in 2012 (Box 2) ### Fiscal consolidation package (2012) - A fiscal consolidation package was legislated in 2011 and had its full impact in 2012, including measures of about 2.6 percent of GDP. - The package was broadly balanced between revenue and expenditure: - Revenue: increases in VAT and social contributions. - Expenditure: measures to better target social transfers and reduce public sector employment through attrition, among others. - Outcome: the package helped to contain the deterioration in the fiscal balance due to the economic downturn, resulting in a modest primary balance improvement of slightly below 1 percent of GDP. - Prior fiscal measures: a pension reform in 2009 introduced gradual increases of 1.3 percent in the contribution rate every 5 years and increased the vesting period for pension entitlements. ### Expenditure aggregates and public expenditure structure (end-2012) - Compensation of employees relative to GDP: - Stands out in cross-country comparison; was among the highest in Europe, second only to Denmark in the Eurostat database by the end of 2012. - Other spending categories: - Total expenditure and most spending categories are either close to the median or below it. - Intermediate consumption was 0.8 percent below the median spending, with Cyprus ranking 24th out of 31 countries in the sample. - Gross capital formation and capital transfers were close to the median. - Employment and sectoral imbalances: - General government employment relative to total employment is slightly below average compared to other European countries (2011). - Evidence of excess employment in education: - Low and declining student-teacher ratio for basic education. - Secondary education: average of only 8 students per teacher in 2011 versus European average of 14 in 2010 for basic education. - Spending composition (difference with median in sample, % of GDP, 2007–2012 shown in source table): - Intermediate Consumption: -0.9, -1.0, -0.9, -0.5, -0.6, -0.8 - Gross Capital formation: -0.3, -0.2, 0.7, 0.4, 0.6, -0.1 - Capital Transfers: 0.1, -0.2, 0.1, 0.3, -0.2, 0.3 - Category-specific observations: - Compensation costs exceeded 65 percent of spending in general public services, defense, public order and safety, and education by end-2012. - Education spending as a share of GDP in 2012 was 1.1 percent above the median of comparator countries. - Cyprus spent more on social-exclusion and housing benefits relative to GDP than any other country in the sample. ### Revenue aggregates, tax bases, and implicit tax rates - Real estate cycle impact: - Total revenues-to-GDP ratio was above median in 2007 at the peak of the housing boom; by 2012 total revenues-to-GDP declined below the cross-country median as real-estate related revenues dissipated. - Revenue category differences (difference with median in sample, % of GDP, 2007–2012 from source table): - Total Revenues: 3.7, 0.9, -0.7, 0.0, -1.9, -2.6 - Taxes on Production and Imports: 5.5, 5.3, 2.4, 2.3, 1.1, 1.6 - of which Taxes on Land, Buildings, and Other Structures: 0.9, 0.5, 0.0, 0.1, 0.0, -0.1 - Current Taxes on Income and Wealth: 1.4, 1.3, 0.5, 0.6, 0.2, 0.0 - of which Taxes on Holding gains: 2.8, 1.6, 0.3, 0.4, 0.3, 0.1 - Social Security Contributions: -3.7, -3.9, -4.0, -3.3, -3.3, -3.1 - Tax base estimates (difference with median in sample, % of GDP, 2007–2012 from source table): - Consumption Tax Base: 18.9, 18.9, 12.9, 14.0, 16.4, 17.4 - Labour Tax Base: -3.9, -5.9, -5.5, -4.1, -3.7, -5.1 - Capital Tax Base: 3.8, 9.8, 5.0, 1.5, 5.1, 7.3 - Key interpretations: - Consumption tax base consistently exceeded the median by more than 10 percentage points of GDP, reflecting the importance of tourism and relatively good enforcement of consumption taxes. - Labour tax base was below the median by at least around 4 percentage points of GDP, reflecting tax evasion and relatively weaker enforcement of labor taxes. - Capital tax base was relatively large, reflecting an unusually large bank deposit stock (subject to withholding tax on interest); potential understatement if foreign companies paying taxes in Cyprus were included. - Implicit tax rates (difference with median in sample, percentage points, 2007–2012 from source table): - Implicit Tax Rate on Consumption: -1.1, -0.8, -1.6, -2.1, -3.5, -3.6 - Implicit Tax Rate on Labour: -10.1, -8.7, -7.5, -6.8, -7.0, -5.9 - Implicit Tax Rate on Capital: 10.9, 5.7, 6.9, 7.7, 6.1, 3.7 - Observations: - Implicit tax rates for consumption and labor-income taxation were considerably below the median: consumption implicit rate 3.6 percentage points below the median (2012), labor implicit rate almost 6 percentage points below the median (2012). - Large positive differential on implicit tax rate on capital and business income is notable despite a corporate tax rate of 10 percent in 2012. ### Selected structural fiscal issues: revenue administration and public financial management - Revenue administration: - Inefficient split into two agencies organized by tax type (VAT services and Inland Revenue) rather than taxpayer segments. - Consequences: fragmentation of collection processes; reduced probability of detection of non-compliance; higher compliance costs for taxpayers; increased administrative cost of collecting revenues. - Different enforcement powers and procedures (stronger for VAT services than Inland Revenue) likely contributed to a large consumption tax base and weaker labor income tax base. - Public financial management weaknesses: - Deficient commitment measurement and control and expenditure overruns: - Not all commitments tracked by government accounting system, leading to risk of overruns and arrears accumulation; practice of approving within-year supplementary budgets undermined fiscal credibility. - Absence of a system to monitor and manage fiscal risks: - Detailed data on government guarantees was not available or sufficient to assess risks, exposing the budget to uncertainty and ad-hoc adjustments. - Lack of medium-term orientation: - Budgeting largely annual and incremental, limiting capacity to adapt to changing circumstances (e.g., education allocations not adjusted for declining student population; housing benefits not adjusted). - Limited use of cost-benefit analysis for investment projects: - Budget-financed projects lacked proper appraisal, raising risk of poor allocation and reduced growth-friendliness of capital budgets. - Difficulties in managing resources to achieve specific policy objectives: - Focus on inputs and absence of program/activity classification with performance indicators made it difficult to assess cost-effectiveness of activities. - Public sector pay and promotion: - Pay and promotion were de-linked from performance; seniority-based automatic increases led to promoted individuals receiving similar salary increases to non-promoted near 100 percent during 2002–12 (World Bank preliminary findings). - COLA mechanism automatically indexed wages to the consumer price index, contributing to increasing compensation costs and a public-wage premium that eroded competitiveness. - Social protection and pensions: - Social protection fragmented with weak targeting; multiple benefits across ministries, differing eligibility criteria, and benefit overlaps led to inefficiencies and leakage. - Pension system unsustainable long-run: - The 2012 Ageing Report projected annual public expenditures on pensions to increase by 8.4 percentage points of GDP during 2010–60, compared to an average of 1.4 percent of GDP. ### Program fiscal response and outcomes to date (post-2012 program) - Program objectives (May 2013): - Target a primary balance of 4 percent of GDP by 2018 to put debt on a downward path toward 100 percent of GDP by 2020. - Consolidation phased over 6 years with substantial upfront adjustment to address credibility and financing constraints. - Measures legislated and projected: - About 7.5 percent of GDP in measures were legislated at end-2012 and implemented in 2013–14. - Projected additional adjustment of 4.7 percent of GDP left for future years. - 2013–14 consolidation composition: - Broadly balanced between revenue and expenditure. - Expenditure measures targeted compensation of employees and social transfers: - Nominal salary cuts of 9–15 percent (depending on the salary level) were implemented in 2013 and 2014; allowances were reduced. - Social transfer measures aimed to save while improving targeting via eligibility criteria on income and assets and rationalization of overlapping benefits. - Revenue measures raised tax rates and broadened tax bases: - VAT and reduced VAT increased by 2 and 1 percentage points in 2013 and 2014, respectively. - Corporate tax rate increased from 10 to 12.5 percent. - Excise rates for energy, tobacco, and alcohol increased in both 2013 and 2014. - Property tax rates were raised and thresholds for exemptions lowered. - Withholding tax rate on interest increased from 10 to 30 percent. - Structural reforms supporting consolidation (legislated/implemented 2012–mid-2014): - COLA reform (December 2012): reduced wage indexation to half the increase in the consumer price index and suspended it automatically after two consecutive quarters of negative growth; reformed COLA starts operating after current COLA suspension expires at end-2016. - Pension reforms (January 2013) for GSIS and GEPS: increased statutory retirement ages, introduced early retirement penalties, allowed automatic adjustment of statutory retirement age by life expectancy at retirement; GEPS closed to new entrants in 2011. - Fiscal Responsibility and Budget Systems Law (FRBSL) adopted in early-2014: - Introduces medium-term budgeting with binding expenditure ceilings supported by fiscal rules; framework for monitoring/managing fiscal risks (including from PPPs); emphasis shift from inputs to outputs and performance; strengthens evaluation and management of public-investment projects. - Revenue administration reform (legislated mid-2014): - Law allowing integration of VAT services and IRD into a single agency organized by taxpayer segment. - Legislation amended to allow freezing and seizure of assets in cases of undisputed tax debt. - Welfare reform (legislated mid-2014): - Introduced a single Guaranteed Minimum Income (GMI) scheme replacing many public assistance benefits; centralized administration of most social benefits in the Ministry of Labor; tightened eligibility criteria and enhanced oversight via a single beneficiary database cross-checked with other databases. *Source: IMF Staff Report content unit _cr14314 (excerpts).* ### 28. As a result of these policies, the 2013 ### As a result of these policies, the 2013 ### Fiscal outcomes and key indicators (2012–13) - Overall and primary deficits declined by about 1 percent of GDP in 2013. - Tax revenues were stabilized at around 26 percent of GDP despite the deep recession. - Taxes on current income and wealth increased as a share of GDP due to significant measures. - Primary expenditure was reduced in nominal terms by about 5 percentage points, but increased by 0.8 percent as a share of GDP. - In July 2014, the cumulative primary balance on a cash basis was a surplus of 2.1 percent of GDP, relative to an expected balance in the 4th review. - Selected fiscal indicators 2012–13 (% GDP) (Source: Eurostat): - Revenue: 2012 = 39.4; 2013 = 41.4 - Taxes on production and Imports: 2012 = 14.9; 2013 = 14.5 - Current Taxes on Income and Wealth: 2012 = 11.1; 2013 = 11.6 - Social Security Contributions: 2012 = 8.5; 2013 = 8.9 - Other: 2012 = 4.9; 2013 = 6.3 - Expenditure: 2012 = 45.8; 2013 = 46.8 - Primary Expenditure: 2012 = 42.6; 2013 = 43.4 - Interest Bill: 2012 = 3.2; 2013 = 3.4 - Primary Balance: 2012 = -3.2; 2013 = -2.0 - Overall Balance: 2012 = -6.4; 2013 = -5.4 - Public Debt: 2012 = 86.6; 2013 = 111.5 - Footnote: 174.8 million Euros in sign in gas bonuses was added to the Eurostat other revenue and primary expenditure data in 2013 to prevent understating primary expenditure in 2013. ### Unwinding of 2008–12 spending increases (2013 dynamics) - Close to a half of the nominal increase in spending during 2008–12 was unwound in 2013. - Changes in major spending categories (percent of 2012 GDP): - Change in Primary Expenditure 2008–12 = 8.3; 2012–13 = -3.6 - Recurrent Expenditure 2008–12 = 8.1; 2012–13 = -2.7 - Social Transfers 2008–12 = 4.2; 2012–13 = -0.1 - Housing Benefits 2008–12 = 0.0; 2012–13 = -0.3 - Compensation of Employees 2008–12 = 2.9; 2012–13 = -1.5 - Intermediate Consumption 2008–12 = 0.6; 2012–13 = -0.7 - Subsidies 2008–12 = 0.2; 2012–13 = 0.0 - Other Current Expenditure 2008–12 = 0.2; 2012–13 = -0.1 - Capital Expenditure 2008–12 = 0.2; 2012–13 = -0.9 - Drivers: - Compensation of employees fell by 1.5 percentage points of 2012 GDP in nominal terms during 2012–13, representing about half the increase in 2008–12. - Discretionary spending fell, particularly intermediate consumption and capital expenditure, as completion of large investment projects and tight control more than unwound earlier increases. - Social transfers and housing benefits contributed modestly to the decline given increased retirements and higher unemployment benefit demand. ### Compensation of employees and public sector pay - Despite reductions, compensation of employees remains among the highest in the EU. - Even after wage cuts of 9–15 percent and a 3 percent decline in central government employment, compensation of employees as a share of GDP is the fourth largest in the EU (calculation uses 2012 GDP). - Nominal salary reductions accounted for most of the decline in compensation of employees in 2013 (% of 2012 GDP) across components: - Central Government Pensions and Gratuities; Central Government Wages and Salaries; Central Government Social Contributions; Semi-Government and Local government compensation; Allowances; Overtime Fees (chart referenced in source). - A distortion note: The 2013 compensation of employee figure is distorted by about 0.5 percent of GDP due to payment of gratuities given increases in early retirements. ### Social transfers and housing benefits - Social transfers remained broadly unchanged in nominal terms despite rationalization measures, largely reflecting automatic stabilizers. - Measures in 2013 rationalized a number of benefits (social-exclusion, family, disability), but these were largely offset by: - Increase in unemployment benefits (including redundancy payments) as unemployment rose to close to 16 percent in 2013 from 12 percent in 2012 and under 4 percent in 2008. - Surge in retirements from labor market weakness and early retirements, which more than offset introduction of penalties for early retirement. - Composition of benefit changes (percent of 2012 GDP) showed increases in Unemployment Benefits and Old Age Benefits among others (chart referenced in source). ### Medium-term fiscal gap and targets - Macroeconomic outlook: gradual and moderate recovery with high private sector deleveraging needs and protracted high unemployment, limiting revenue buoyancy and sustaining unemployment benefit spending. - Unfreezing of pensions and public wages starting in 2017 will likely boost spending. - Estimated additional permanent consolidation needed: 3.4 percent of GDP to reach a primary surplus of 4 percent of GDP by 2018 and sustain it thereafter. - This estimate is lower than initial program estimate of 4.7 percent of GDP reflecting over-performance relative to program targets. - Primary Balance and fiscal gap estimates (percent of GDP, cash basis): - Primary Balance with Unchanged Policies 2015 = -1.3; 2016 = 0.2; 2017 = 0.1; 2018 = 0.6 - Primary Balance 4th Review Targets 2015 = -1.0; 2016 = 1.8; 2017 = 3.0; 2018 = 4.0 - Cumulative Fiscal Gap 2015 = 0.3; 2016 = 1.6; 2017 = 2.9; 2018 = 3.4 - Source: IMF Staff Estimates. ### Policy guidance: focus on expenditure-side consolidation - Rationale: - Significant revenue measures already implemented stabilized revenues; spending remains elevated by historical standards. - Bulk of additional measures should target largest spending items: compensation of employees and social transfers (account for 70 percent of primary general government expenditure) to avoid compressing smaller, growth-friendly budget areas. - Limit further adjustment in discretionary spending (gross capital formation and intermediate consumption) to keep the budget growth friendly and ensure durability. - Specific options to rationalize compensation of employees (examples and estimated fiscal yields): - Reducing public sector wages: - A 2 percent flat cut is estimated to generate savings of 0.3 percent of GDP. - Targeted cuts to salary distribution levels where gaps with private sector are largest suggested. - Subjecting public pension gratuities to income tax: - Gratuities average about €104,637 per person in 2013 and are fully exempt from income tax. - Taxing them is estimated to yield about 0.2 percent of GDP. - Fully phasing them out over time could yield up to 0.5 percent of GDP. - Such measures may need to be accompanied by raising eligibility age for gratuity and/or unreduced pension to limit early retirement incentives. - Reducing employment in select sectors: - Address excess employment in education via increasing working hours, increasing class size, extending school schedules, rationalizing non-teaching staff. - Reducing 2,000 workers on renewable contracts could save about 0.4 percent of GDP in 3 years. - Accompanying public administration reform to address fiscal consequences of unfreezing wages from 2017 (automatic salary increase of 3.6 percent on average and other increases). Key reform measures to consider: - Carrying out a compensation survey to benchmark government salaries with appropriate private sector comparators and keep it updated. - Aligning salaries with private sector informed by compensation survey results. - Reforming pay scales to eliminate automatic increases unrelated to performance. - Revamping the performance appraisal system. - Facilitating mobility across the public sector. - Apply reform to broader public sector including SOEs to maximize benefits. ### Other spending rationalization opportunities - Social benefits: - Review benefits introduced or expanded in 2008–12 to phase out or better target to the vulnerable. - Given favorable housing indicators and the new welfare reform including a housing rental allowance through the GMI, housing benefits could be better targeted and gradually phased out, yielding up to 0.5 percent of GDP. - Education and other current expenditure: - Introduce/increase gradually tuition fees for tertiary education and reduce transfers to universities; elimination of transfers could yield up to 0.6 percent of GDP. - Retain and target existing education grants to vulnerable students to preserve access to tertiary education. - Basic education subsidies could be means tested/gradually phased out; full phase-out could yield up to 0.4 percent of GDP. - Semi-government institutions and SOEs: - Operational improvements could allow reduction in transfers which represented about 0.8 percent of 2012 GDP in 2013 (figure excludes transfers to universities and includes transfers to several semi-government organizations and SOEs). ### Conclusion and policy implications - Root cause: fiscal imbalances built up primarily from permanent spending increases financed by temporary real-estate revenues. When real-estate related revenues dwindled, the permanent increases in compensation of employees and social transfers widened the fiscal deficit. - Actions taken: authorities implemented fiscal measures of 7.5 percent of GDP in 2013–14, leading to an improvement in the primary balance of about 1 percent of GDP in 2013 despite the deep recession; revenues remained broadly stable; about half of the nominal spending increase during 2008–12 was unwound; and structural reforms were advanced in pensions, wage indexation, revenue administration, budget systems, and welfare. - Remaining need: an additional 3.4 percent of GDP in fiscal measures is estimated to be needed in the medium term to reach the primary surplus target of 4 percent of GDP by 2018 to put debt on a sustained downward path toward 100 percent of GDP by 2020. - To ensure durable consolidation: - Measures should be permanent, high quality, and spending-focused. - Options include revising wage levels and pay scales, eliminating automatic increases, reducing employment in overstaffed areas (education), and rationalizing public-pension lump-sum payments. - Review social benefits expanded during 2008–12 and improve targeting of education subsidies while increasing tertiary-education fees could be considered. - Even after reaching a 4 percent primary surplus, maintaining it for some time (10 years or more) is necessary to return debt to pre-crisis levels and toward the European target of 60 percent of GDP. - Underpin consolidation with durable fiscal structural reforms in revenue administration, public financial management, and public administration. *Source: IMF staff report excerpt (Eurostat and IMF staff estimates cited within source).* --- _Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2014/_cr14314.pdf_