## _cr14174 ## Source details **Canonical URL:** [_cr14174](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2014/_cr14174.pdf) ## Other formats - [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2014/_cr14174.pdf.md) - [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2014/_cr14174.pdf.json) --- ### Global financial conditions and foreign participation in EMs government bond markets - Sample and variables: - Quarterly data for 17 emerging market economies over 2004:Q1–2013:Q2. - Dependent variables: level and volatility of the 5-year local currency yield and EMBI benchmark foreign currency yield. - Foreign participation (Fit): foreign holdings as percent of total government local-currency (foreign-currency) denominated government debt (Arslanalp and Tsuda (2014) dataset). - Investor-base concentration: normalized Herfindahl index (range 0–1) from six components (foreign: central bank, bank, non-bank; domestic: central bank, bank, non-bank). - Global shock indicators: US3mt (U.S. 3 month t-bill), US10Y (U.S. 10-year Treasury), VIX. - Controls: current account/GDP, real GDP growth (or volatility), inflation, forward exchange rate volatility; country fixed effects and time effects included. - Key empirical findings: - Higher foreign participation in local-currency sovereign bond markets: - Increases transmission of global financial shocks once foreign participation passes a threshold (above 30 percent). - Increases yield volatility and decreases yield levels for local-currency bonds. - Positive association with local-currency bond yield volatility becomes significant when instrumenting foreign holdings (lagged instruments or financial remoteness-based instrument). - Foreign holdings of foreign-currency denominated bonds show no significant impact on the level or volatility of foreign-currency yields. - Higher investor-base concentration amplifies sensitivity of local-currency sovereign yields to global shocks; diversification mitigates impact. - Strong macroeconomic fundamentals (low inflation, strong and stable output growth, moderate public debt) reduce level and volatility of EM sovereign bond yields. - Nonlinear/threshold results (local-currency yields): - Foreign participation thresholds: - Above 35 percent: a 100 bps increase in US10Y leads to an additional 100 bps amplification; overall rise ≈ 140 bps above threshold versus 40 bps below. - Above 30 percent: a 100 bps increase in US short-term rates increases EM yields by 140 bps (no significant impact below). - Around 32 percent foreign holdings: a two-standard deviation VIX shock translates to a 130 bps yield increase past threshold (no significant impact below). - Investor-base concentration threshold around median Herfindahl ≈ 0.2: - A 100 bp increase in US10Y raises local-currency yields by 70 bps when Herf>0.2 versus 30 bps when Herf≤0.2. - Selected coefficients (Panel Non-Linear Estimates, dependent variable: local-currency 5-year yields / benchmark yields): - U.S. 10-year yield coefficient: 0.418* (col 1), 0.309* (col 3). - U.S. 10-year yield * 1[FH>35] = 0.958*** (col 1). - VIX * 1[FH>32] = 0.0692** (col 2). - U.S. 3-month * 1[FH>30] = 1.332*** (col 3). - U.S. 10-year yield * 1[Herf>0.2] = 0.385** (col 4). - Threshold dummy coefficients: -2.654*** (col 1), -2.717*** (col 2), -1.587*** (col 3), -1.160** (col 4). - Real GDP growth: -0.196*** (col 1), -0.169** (col 2), -0.179*** (col 3), -0.234*** (col 4). - Forward exchange rate, log: 0.00199*** (col 1), 0.00198*** (col 2), 0.00173*** (col 3). - Inflation: 0.370*** (col 1), 0.389** (col 2), 0.330** (col 3), 0.0982 (col 4). - Observations: 367 (cols 1–3), 317 (col 4). R-squared: 0.331, 0.293, 0.306, 0.533. - P-values for θ3 + θ32 reported as: 1.4 [0.000] (col 1); 0.08 [0.000] (col 2); 1.4 [0.000] (col 3); 0.7 [0.000] (col 4). - Robustness and caveats: - Endogeneity addressed with two-quarter lags and financial remoteness instrument; results depend heavily on post-crisis period when many countries crossed thresholds. - Lack of investor-profile data (retail vs. institutional, hedging) limits generalization of threshold magnitudes. ### Poland-specific implications (sovereign bond markets) - Exposure and risks: - Substantial increase in foreign holdings of Polish sovereign bonds raises vulnerability to withdrawal of foreign capital with US monetary policy normalization. - The 2011 reduction in transfers to the second pillar and 2014 pension changes increased foreign investor share by reducing domestic institutional holdings. - Mitigating factors for Poland: - Relatively well-diversified investor base compared to typical EMs. - Strong macroeconomic fundamentals and sound policy frameworks. - Flexible Credit Line arrangement with IMF may dampen spillovers. ### Macroeconomic fundamentals and yield behavior (summary) - Strong fundamentals associated with lower yields and lower yield volatility. - Specific empirical relations: - Higher indebtedness increases yield volatility. - Higher output growth volatility increases yield volatility. - Higher real GDP growth decreases level and volatility of both local- and foreign-currency yields. - Higher inflation positively associated with yield levels. ### Corporate sector vulnerabilities in Poland - Stylized facts and stock levels: - Total nonconsolidated NFC debt rose by about 15 percentage points of GDP to 83 percent of GDP from 2005 to 2011; declined close to 5 percentage points to 78 percent of GDP in 2012. - Long-term loans reached 35 percent of debt in 2011; short-term loans ≈ 10 percent of debt. - Intercompany debt ≈ 25 percent of GDP in 2012 (~30% of total nonconsolidated debt). - External NFC debt ≈ 30 percent of GDP in 2012 (from 17½ percent in 2000); intercompany component of external NFC debt ≈ 17 percent of GDP at end-2012. - Excluding intercompany debt, nonfinancial sector external debt ≈ 12 percent of GDP. - NFC debt securities: total stock ≈ 10 percent of GDP; domestic-currency share rose to 38 percent in 2013 from 6 percent in 2000. - Vulnerabilities: - High external NFC debt relative to non-European EMs could be exposed if capital flows reverse. - Rising noninvestment grade issuance: outstanding non-investment grade debt securities ≈ 3 percent of GDP; non-investment grade issuance >80% of total NFC securities issuance in 2013. - Rollover needs: overall external NFC debt due within 12 months ≈ 12 percent of GDP (44 percent of total NFC debt); ≈ 60 percent of rollover needs due to intercompany debt. - Firm-level metrics: - Total liabilities to total assets stabilized just above 40 percent. - Short-term debt to total debt increased slightly in 2012. - RoA and RoE declined in 2012 toward 2008 levels. - Tobin’s Q in 2012 slightly above 1. - Default probabilities in 2012 ranged between 1.5 percent (BSM) and 8 percent (Z). - Debt-at-risk (ICR < 1): <5 percent in 2007; peaked at 25 percent in 2009; >10 percent by 2012. - Mitigating factors: - Large share of intercompany debt (close to 60 percent of external nonfinancial sector debt) and its long-term classification reduce sudden-stop vulnerability. - Increasing local-currency issuance lowers FX balance-sheet risk. ### Stress-test scenarios and results (firms) - Scenarios modeled on firm-level data (S&P Capital IQ sample; 372 firms in 2012; assets PLN 516,444 million; debt PLN 96,334 million): - Interest-rate shock: apply 500 basis point shock to baseline interest rate; assume return on financial assets increases by 300 bps (alternative: EBIT unaffected). - Profit shock: 25 percent decline in operating income (EBIT). - Combined shock: 500 bps interest increase + 25 percent EBIT decline; assume no improvement in returns on financial assets. - Exchange rate shock: 30 percent zloty depreciation; FX share of external NFC debt = 66 percent; hedging scenarios: no hedging, 50 percent, 80 percent hedge ratios. - Data cleaning: exclude outliers with baseline interest >50% or negative baseline ICR; missing financial assets set to zero. - Stress-test findings (summary): - Moderate vulnerabilities under baseline and moderate shocks. - Large combined interest and profit shocks could increase debt at risk substantially (example: combined shock could more than double debt-at-risk to ≈ 35 percent of debt). - FX valuation losses from a 30 percent depreciation: loss on maturing FX debt ≲ 5 percent of EBIT; total valuation effect (full outstanding FX debt) could be ≈ 20 percent of EBIT without hedges. - Intercompany debt reduces measured vulnerability; excluding intercompany debt materially lowers estimated losses. - Conclusions: - Small to moderate shocks: debt-at-risk contained. - Severe interest-rate shocks: could pose material risks, though prevalence of fixed-rate debt would mitigate. - FX shock impact highly dependent on hedge ratios and natural hedges. ### Pension system reform and fiscal implications (2014 changes) - Institutional changes: - 2014 reform moved about half of pension fund assets/liabilities to the first pillar; contributions redirected to first pillar with opt-in window (2.92 percent contribution to second pillar if actively chosen). - Pension funds banned from investing in government bonds; foreign securities limit to rise from 5 percent (2013) to 30 percent (by 2016). - Zipper: assets of persons 10 years before retirement gradually transferred to first pillar. - Baseline modelling and assumptions: - Static cohort model; returns in second pillar net of fees assumed equal to first pillar indexation; discount rate = GDP growth. - No behavioral responses or macro feedback included. - Baseline fiscal results: - 2014 changes deliver improvement in fiscal cash flows ≈ 30 percent of GDP (NPV) between 2014–60 versus no policy change. - Cumulative cash flows peak ≈ 31 percent of GDP by 2050; ≈ 29 percent of GDP by 2060. - One-off asset transfers in 2014 ≈ 9 percent of GDP; zipper transfers cumulative ≈ 18 percent of GDP by 2060. - Increased public-sector pension payments cumulative ≈ 28 percent of GDP by 2060 partially offset gains; one-off transfer mostly paid out by 2060 (≈ 82 percent). - If implicit and explicit liabilities treated equally, reform is fiscally neutral. - Sensitivity and scenarios: - Switching rates to first pillar: - If all contributors stay in second pillar: NPV fiscal gains = 23 percent of GDP. - If all contributors opt-out: NPV fiscal gains = 35 percent of GDP. - Four macro scenarios (selected stats preserved exactly from Table 1): - Scenario 1 (Historical Dynamics): real GDP growth averages 2.4 percent; wage bill growth averages 2.5 percent. - Scenario 2 (Severe Downturn): real GDP growth averages 1.7 percent; minimum -3.2 percent; 7 years of negative performance. - Scenario 3 (Sluggish Wages): wage bill growth averages 1.7 percent; wage bill to GDP drops from 26 ½ percent (2014) to 22 percent. - Scenario 4 (Protracted Crisis): protracted severe losses (cumulative 23% GDP growth and 20% wages drop over 2020–25); baseline GDP Growth stats: Actual 2001-13 Average 3.6 Min 1.2 Max 6.8 St. Dev. 1.9; Baseline Average 2.3 Min 1.2 Max 4.3 St. Dev. 0.9; Historical Dynamics Average 2.4 Min -0.7 Max 6.8 St. Dev. 1.6; Severe Downturn Average 1.7 Min -3.2 Max 6.8 St. Dev. 1.8; Sluggish wages Average 2.4 Min -0.3 Max 6.8 St. Dev. 1.5; Protracted Crisis Average 1.3 Min -7.1 Max 3.8 St. Dev. 2.5. - Baseline results robust: fiscal impact under distressed scenarios deviates by < 6 percent of GDP from baseline. - Indexation risks and legacy fiscal vulnerabilities: - Current zero-floor indexation of main notional accounts creates fiscal exposure; shock scenarios activate zero-floor with adverse consequences ranging from 3 percent to 25 percent of GDP. - Mitigation: allow negative indexation or base indexation on multi-year wage bill dynamics to reduce risk (may not fully offset Protracted Crisis scenario). - Replacement rates projected to decline from ≈ 60 percent (current) to ≈ 30 percent by 2060. - Special pension regimes (farmers, miners, uniformed services) create fiscal burdens: in 2010 special pensions covered 2.1 million beneficiaries (29% of regular system) and outlays = 2.7 percent of GDP (29½ percent of regular pensions). - Disability benefits misalignment may create incentives to claim disability; alignment with old-age pensions recommended. - Policy recommendations (pension chapter): - Maintain fiscal consolidation and permanent expenditure rule given long-run liabilities. - Consider allowing negative indexation of notional accounts or multi-year indexation to reduce fiscal risk. - Improve incentives for private savings (third pillar), financial literacy, and coverage of atypical contracts. - Reform special pension schemes (parametric adjustments) and align disability rules with core system. ### VAT cyclicality and asymmetries (EU sample including Poland) - Data and methods: - Sample: 21 EU countries, 1999:Q1–2013:Q4; quarterly VAT revenues and domestic demand in constant 2005 euros. - Estimators: PMG, MG, PMG-BO, CMG; control for cross-sectional dependence using CCE augmentation and Binders and Offermanns correction. - Asymmetric specifications split domestic demand shocks into negative and positive deviations. - Main empirical findings: - VAT revenues are strongly pro-cyclical and asymmetric: downturns reduce VAT revenues more than upturns restore them. - Short-term elasticity: a 1 percentage point increase in domestic demand growth leads to a 1.0‒1.6 percentage point increase in VAT revenues (range across estimators). - Error-correction (adjustment) terms: -0.424*** to -0.627*** across estimators. - Asymmetric estimates (examples): - PMG-BO negative demand shock: 1.425*** (0.322); positive demand shock: 0.583* (0.306). - CMG negative demand shock: 0.827*** (0.217); positive demand shock: 0.727*** (0.257). - VAT elasticities higher in downturns (“bad times”) than in expansions. - Crisis-split results: sensitivity to domestic demand increased marginally in 2008–13. - Policy implications: - Use state-dependent (asymmetric) VAT elasticities for forecasting VAT revenues and fiscal balance. - Strengthen tax administration and VAT compliance, especially during downturns, to limit persistent revenue losses. - For Poland: adopt pro-cyclical VAT elasticities in forecasts; prioritize diagnosing VAT underperformance and strengthen VAT administration. ### Private saving accelerations (cross-country evidence and implications for Poland) - Identification and stylized facts: - Definition: three-year moving average over nine years increases by >4 percentage points relative to prior five-year average and saving rate after transition >10 percent of GDP. - Number of identified episodes: 86. - Average private saving rate acceleration: 13.6 percentage points of GDP; median 10.3 percentage points. - Unconditional decade probability: ≈ 25 percent that a country experiences a saving acceleration in a decade. - Accelerations concentrated in emerging/developing regions. - Determinants (baseline econometrics): - Positive correlates: higher real per capita GDP growth, higher public saving-to-GDP, natural resource discoveries (oil per capita, log). - Negative correlates: higher growth volatility, higher unemployment, greater financial openness; financial development positively correlated in EM/LIC subsample. - Selected marginal effects (examples): - Real per capita GDP growth: 0.00813*** (0.00214). - Public saving-to-GDP: 0.00589*** (0.000859). - Financial openness: -0.00929** (0.00451) in one specification. - Unemployment rate: -0.00233* (0.00134) in one specification. - Oil discoveries per capita, log: 0.217** (0.0989) in column (6). - Causality and outcomes: - Evidence indicates saving accelerations are mostly driven by stronger GDP growth (causality from growth → saving accelerations). - Average supplementary GDP growth during acceleration ≈ 1 percentage point per year (ATT results: 1.321***, 0.995***, 1.118***, 1.123*** across matching methods). - Saving accelerations associated with higher private investment post-episode (ATT ≈ 2.576–3.215 percentage points; significant). - Poland-specific diagnostic: - Poland's propensity to experience a private saving acceleration estimated ≈ 6 percent. - Constraints: elevated unemployment and fiscal imbalances reduce likelihood. - Policy recommendations for Poland: - Advance structural reforms (labor market) to raise potential growth and reduce unemployment. - Continue fiscal consolidation to improve public saving-to-GDP. - Preserve banking and financial stability to support voluntary private saving and capital accumulation. *Source: _cr14174 (IMF staff report excerpts).* ### 1. Global Financial Conditions and Foreign Participation in Ems Government Bond ### Global Financial Conditions and Foreign Participation in Ems Government Bond Markets ### Overview and motivation - Foreign holdings of Polish government bonds have increased substantially over the last decade, helping to lower government bond yields, particularly of those issued in local currency. - Foreign participation in emerging market (EM) local-currency sovereign bond markets increased in recent years, especially in the aftermath of the global financial crisis when supportive monetary policy by major central banks and ample global liquidity prevailed. - The increase in foreign participation raises concerns about increased sensitivity to shifts in market sentiment and the transmission of global financial shocks into EM sovereign bond markets. ### Data, sample, and key definitions - Sample: quarterly data for 17 emerging market economies over 2004:Q1–2013:Q2. - Dependent variables (rit): the level and volatility of the 5-year local currency yield and EMBI benchmark foreign currency yield. - 5-year bonds chosen to maximize sample size and capture the most widely traded paper in the secondary market. - Volatility is computed as the log of the standard deviation of weekly changes in the yield over each quarter (12 weeks), capturing within country-quarter volatility. - Foreign participation (Fit): foreign holdings of local-currency and foreign-currency denominated government debt in percent of total government local-currency (foreign-currency) denominated government debt. - Foreign holdings data source: Arslanalp and Tsuda (2014) dataset. - Investor base concentration: approximated using a normalized Herfindahl index computed from disaggregated data of institutional investor holdings of EM total government debt. - Instrumental strategies to address endogeneity: - Instrument with two quarter lags of foreign participation. - Use predicted values of foreign holdings ratios explained by a geographical measure of “financial remoteness” (time-invariant country measure with quarter-by-quarter time-varying estimates from auxiliary regressions). - Controls for country fundamentals (Xit): (i) current account balance to GDP; (ii) real GDP growth (or volatility of growth); (iii) inflation; and (iv) forward exchange rate volatility. - Model includes country fixed effects (iu) and time effects (τt). ### Main empirical findings - Foreign participation and investor base concentration matter for the transmission of global shocks: - Higher foreign participation in local-currency denominated sovereign bond markets increases the transmission of global financial shocks, especially once a threshold has been reached (foreign participation above 30 percent). - Higher foreign participation in local-currency denominated government bonds: - Increases the volatility of the yield. - Decreases the level of the yield. - The positive association with local-currency bond yield volatility becomes significant when instrumenting foreign holdings with lagged values or using the financial remoteness-based instrument. - Higher foreign holdings of foreign-currency denominated bonds do not appear to have an impact on the level or volatility of foreign-currency denominated bond yields. - Potential explanations: currency risk considerations, investor type (buy-and-hold vs. shorter-term), differing risk aversion, and differing ability to hedge foreign currency risk. - A higher concentration of the investor base (higher normalized Herfindahl index) makes EM local-currency sovereign yields more sensitive to global financial shocks; a diversified investor base can help ameliorate the impact of shocks. - Strong macroeconomic fundamentals—low inflation, strong and stable output growth, and moderate public debt levels—help reduce the level and the volatility of EM sovereign bond yields. ### Poland-specific implications - Given expected tightening in global financial conditions from eventual normalization of US monetary policy and the substantial increase in foreign holdings of Polish sovereign bonds, Poland could be adversely affected by a withdrawal of foreign capital. - The increase in foreign participation following modifications to Poland’s second pension pillar (which reduced private pension funds’ holdings of government bonds) contributed to a higher share of foreign investors in the sovereign bond market. - Mitigating factors for Poland include a relatively well-diversified investor base (compared to the typical EM), strong fundamentals, and sound policy frameworks that could help dampen the impact of adverse financial spillovers. ### Methodological notes and robustness - The specification extends earlier work by differentiating between EM foreign-currency and local-currency sovereign bonds and by covering a longer period that includes pre- and post-crisis dynamics. - Endogeneity concerns addressed via lagged instruments and a financial remoteness-based instrument; when foreign participation enters contemporaneously, instrumental variable techniques are used. - The disaggregated components of foreign holdings (foreign bank and non-bank participation) were tested individually but coefficients were not significantly different from one another; aggregate foreign holdings are used. _Prepared by Christian Ebeke (EUR) and Annette Kyobe (SPR); analysis draws on Arslanalp and Tsuda (2014) and follows the structure summarized in the source document._ ### 10. Stronger macroeconomic fundamentals are generally associated with lower yield levels ### 10. Stronger macroeconomic fundamentals are generally associated with lower yield levels ### Macroeconomic fundamentals and yield behavior - Stronger macroeconomic fundamentals are generally associated with lower yield levels and reduced yield volatility. - Higher indebtedness increases yield volatility. - Higher output growth volatility increases yield volatility. - Higher real GDP growth decreases the level and volatility of both local- and foreign-currency bond yields. - Higher inflation is positively associated with the level of bond yields. ### Empirical approach to foreign participation, investor concentration, and transmission of shocks - Objective: examine how foreign participation and investor-base concentration affect the transmission of global financial shocks to EM sovereign bond yields. - Global shock channels and indicators: - Liquidity channel: U.S. 3 month t-bill rate (US3mt). - Portfolio balance channel: U.S. 10-year Treasury bond rate (US10Y). - Confidence channel: VIX (VIX). - Dependent variable (r_it): yield of the 5-year government local-currency bond or the country-specific sovereign benchmark bond yield. - IF_it-1: foreign participation (or, alternatively, concentration of the investor base measured by a normalized Herfindahl index ranging between 0 and 1). - Herfindahl index computed from shares of six components (foreign: central bank, bank, non-bank; domestic: central bank, bank, non-bank). - Dummy D = 1 if lagged foreign participation ≥ D* (threshold). D ranges between 10 and 40 percent for foreign participation; between 0.1 and 0.25 for investor-base concentration. D is lagged by one quarter. - Global shocks S_t (US3mt, US10Y, VIX) are interacted with D to test nonlinear effects conditional on foreign exposure. - Threshold detection: foreign holdings cutoffs explored by 1 percentage point increments (0.01 unit increments for Herfindahl). Test of no nonlinear effect is H0: coefficient on interaction (θ3) = 0. Under OLS, optimal cutoff also minimizes residual sum of squares. ### Key empirical results and thresholds - Transmission of global financial shocks to local-currency yields is amplified at higher levels of foreign participation; no significant impact found on the level of foreign-currency yields. - Foreign participation threshold effects: - The results suggest that, above 30–35 percent, foreign participation increases the transmission of global financial shocks in a significant way. - When foreign holdings of local currency bonds lie above 35 percent: - The transmission of a shock to the U.S. 10 year yield is amplified by an additional 100 bps. - Past the 35 percent threshold, a 100 bps increase in the US yields results in a rise in EM yields of around 140 bps (compared with an increase of only 40 bps below the 35 percent threshold). - When foreign participation exceeds 30 percent: - Past this threshold, a 100 bps increase in U.S short-term interest rates increases EM yields by 140 bps (compared with no significant impact below the threshold). - In periods of global risk aversion (rising VIX), where foreign holdings reach around 32 percent: - A two-standard deviation shock in the VIX translates to a 130 bps increase in yields in EMs with foreign participation past the threshold (compared to no significant impact below the threshold). - Investor-base concentration threshold effects: - A threshold of investor-base concentration exists around the median Herfindahl index in the sample (around 0.2). - A high concentration of the investor base (above the 0.2 threshold) makes EM local-currency sovereign yields more sensitive to global financial shocks. - A 100 bp increase in the 10-year U.S. yield increases local-currency sovereign bond yields by 70 bps when the investor base is significantly concentrated (above the 0.2 threshold) compared to 30 bps when it is more diversified (below the threshold). - Findings from Table 5 (Panel Non-Linear Estimates; dependent variable: yield on local-currency 5-year government bonds and country-specific benchmark yields): - U.S. 10-year yield coefficient: 0.418* (column 1) and 0.309* (column 3) where applicable. - U.S. 10-year yield * 1[FH>35] = 0.958*** (column 1). - VIX * 1[FH>32] = 0.0692** (column 2). - U.S. 3-month * 1[FH>30] = 1.332*** (column 3). - U.S. 10-year yield * 1[Herf>0.2] = 0.385** (column 4). - Relevant threshold dummy coefficients: -2.654*** (col 1), -2.717*** (col 2), -1.587*** (col 3), -1.160** (col 4). - Real GDP growth coefficients: -0.196*** (col 1), -0.169** (col 2), -0.179*** (col 3), -0.234*** (col 4). - Forward exchange rate, log coefficient: 0.00199*** (col 1), 0.00198*** (col 2), 0.00173*** (col 3). - Inflation rate coefficients: 0.370*** (col 1), 0.389** (col 2), 0.330** (col 3), 0.0982 (col 4). - Model statistics: Observations 367 (cols 1–3), 317 (col 4). R-squared 0.331, 0.293, 0.306, 0.533 respectively. - P-values for θ3 + θ32 given as: 1.4 [0.000] (col 1), 0.08 [0.000] (col 2), 1.4 [0.000] (col 3), 0.7 [0.000] (col 4). ### Robustness and caveats - Results should be interpreted with caution: - Absence of data on the profile of foreign investors (retail vs. institutional) limits generalization of threshold values. - Most countries reached the identified thresholds after the global financial crisis, making threshold results heavily dependent on the post-crisis period. ### Policy implications and recommendations - Higher foreign participation in local-currency EM government bond markets: - Provides additional financing and decreases yield levels. - Increases yield volatility and amplifies transmission of global financial shocks once threshold levels (around 30 percent foreign investor share) are reached. - Mitigating factors that reduce vulnerability to spillovers: - Broaden and diversify the investor base to attenuate the impact of global shocks. - Preserve strong macroeconomic fundamentals—especially lower public debt and higher international reserves—to dampen destabilizing effects of reliance on foreign investors. - Country-specific note for Poland: - Foreign investors have a significant presence in Polish government bond markets. - Spillovers from global financial shocks may be dampened by Poland’s strong macroeconomic fundamentals, sound policy frameworks, relatively diversified investor base, and the Flexible Credit Line arrangement with the IMF. *Source: _cr14174 - 10. Stronger macroeconomic fundamentals are generally associated with lower yield levels (IMF).* ### References ### _cr14174 - References ### References cited - Arslanalp, Serkan, and Takahiro Tsuda, 2014, “Tracking Global Demand for Emerging Market Sovereign Debt”, IMF Working Paper No. 14/39, International Monetary Fund. - Chami, Ralph, Dalia Hakura, and Peter Montiel, 2009, “Remittances: An Automatic Output Stabilizer?”, IMF Working Paper No. 09/91, International Monetary Fund. - Combes, Jean-Louis, and Christian Ebeke, 2011, "Remittances and Household Consumption Instability in Developing Countries", World Development 39.7 (2011): 1076–1089. - Ebeke, Christian, and Yinqiu Lu, 2014, “Emerging Market Local Currency Bond Yields and Foreign Holdings in the Post-Lehman Period-a Fortune or Misfortune?”, IMF Working Paper No. 14/29, International Monetary Fund. - Lim, Jamus Jerome, Sanket Mohapatra, and Marc Stocker, 2014, “Tinker, Taper, QE, Bye? The Effect of Quantitative Easing on Financial Flows to Developing Countries”, Background Paper for Global Economic Prospects 2014, Washington, DC: World Bank. - Masten, Arjana, Coricelli, Fabrizio, and Masten, Igor, 2008, "Non-linear growth effects of financial development: Does financial integration matter?" Journal of International Money and Finance, 27.2 (2008): 295-313. - Peiris, Shanaka J., 2010, “Foreign Participation in Emerging Markets’ Local Currency Bond Markets (EPub)”, IMF Working Paper No. 10/88, International Monetary Fund. - Rose, Andrew K., and Mark M. Spiegel, 2009, "International Financial Remoteness and Macroeconomic Volatility", Journal of Development Economics 89.2 (2009): 250-257. ### Corporate sector vulnerabilities — overview and key findings - Chapter focus: nonfinancial corporate (NFC) sector vulnerabilities in Poland at the firm level and cross-country macro level. - Main observation: stock of NFC debt in Poland is high relative to peers, but mitigating factors exist (notably intercompany debt and increasing local currency denomination of debt securities). - Risk channels highlighted: - FX shocks may pose only a moderate risk where foreign currency debt is hedged. - Increase in noninvestment grade issuance of debt securities raises sensitivity to large interest rate shocks. - Potential macroeconomic implications through layoffs and increased unemployment if corporate financing deteriorates. ### Historical dynamics and recent developments - 2007: high fixed investment growth above 15 percent preceded deterioration in corporate sector balance. - 2008–09 crisis effects: - NPLs climbed, reaching around 12 percent of total loans by mid-2010 (and even higher for SMEs). - Credit to NFCs contracted sharply: year-on-year credit growth declined from more than 25 percent at end-2008 to close to -10 percent in spring 2010. - Bankruptcy cases in courts increased by about 60 percent during 2008–10. - Recovery signals: - Credit to NFCs slowly picking up after 2012 but remains low compared to other EMs. - Bankruptcies: 12-month sum down 6.8 percent year-on-year in March 2014 from growth of close to 20 percent a year earlier. - NPLs have started to edge down. ### Stylized facts — the stock - NFC debt trajectory: - Total nonconsolidated NFC debt increased by about 15 percentage points of GDP to 83 percent of GDP from 2005 to 2011. - Long-term loans increased by 10 percentage points of GDP, reaching 35 percent of debt in 2011. - Short-term loans remained around 10 percent of debt. - Moderate deleveraging: total nonconsolidated NFC debt declined by close to 5 percentage points of GDP from 2011 to 78 percent of GDP in 2012. - Intercompany debt: - Intercompany debt around 25 percent of GDP in 2012, accounting for just above 30 percent of total nonconsolidated debt. - Other accounts payable (including trade credits) amount to 27 percent of nonconsolidated debt. - External NFC debt: - External NFC debt rose to around 30 percent of GDP in 2012 from 17½ percent of GDP in 2000. - Intercompany component of external NFC debt rose from around 5 percent of GDP at end-2000 to around 17 percent of GDP at end-2012. - External intercompany debt (FDI debt instrument liabilities) accounted for close to 60 percent of total nonfinancial sector external debt in 2012. - Excluding intercompany debt, nonfinancial sector external debt about 12 percent of GDP (around the sample median). - Debt securities and currency composition: - Total stock (including domestic) of nonfinancial sector debt securities at 10 percent of GDP. - Share of NFC debt securities denominated in domestic currency increased to 38 percent in 2013 from 6 percent in 2000. ### Vulnerabilities and mitigating factors - Vulnerabilities: - High external NFC debt relative to many non-European EMs could erode financing quickly if capital flows reverse (e.g., Fed tapering or geopolitical turmoil). - Flow indicators hint at emerging vulnerabilities despite stock indicators showing relatively stable financing sources. - Rising noninvestment grade issuance increases sensitivity to large interest rate shocks. - Mitigants: - Large share of intercompany debt reduces vulnerability to sudden stops linked to global market turmoil because intercompany lending can be classified as long-term lending per IMF BPM6 (2009). - Increasing domestic currency denomination of debt securities reduces balance sheet risks from exchange rate depreciation. - External intercompany debt predominance (close to 60 percent of external nonfinancial sector debt) is a stabilizing factor relative to wholesale market borrowing. ### Data sources and measurement notes (Box 1) - Data sources used: - Standard & Poor’s Capital IQ: firm-level balance sheet data; sample in 2012 includes 372 nonfinancial companies; assets in 2012 sample PLN 516,444 million (32 percent of GDP); debt PLN 96,334 million (6 percent of GDP or 7½ percent of total NFC debt as per Eurostat). - Corporate Vulnerability Utility (IMF): default risk indicators based on Worldscope. - Moody’s KMV Credit Edge Plus: expected default frequencies (EDFs) for publicly listed companies; data extracted for 452 companies. - Dealogic Analytics: aggregates of NFC bonds and syndicated loan and equity issuance. - Eurostat: country-level financial balance sheet data for European NFC sectors; allows consolidated vs nonconsolidated decomposition and intercompany debt inference. - International Financial Statistics (IFS) / IIP: breakdown of aggregate external liabilities to compute nonfinancial sector external debt for “other sectors”. - Quarterly External Debt Statistics (QEDS): breakdown of nonfinancial sector external debt into short- and long-term and intercompany debt where available. - Limitations noted: - External debt statistics for “other sectors” do not separate households and NFCs; household share of nonfinancial sector external debt in 2013 was about 40 percent of total external debt but household external debt expected to be small. - Intercompany NFC debt with nonresident affiliates may not be fully captured in some datasets. - Cross-country comparable data on total NFC debt limited outside Europe. *Source: _cr14174 - References* ### 12. While the share of short-term external debt is high, the overall stock of short-term ### _cr14174 - 12. While the share of short-term external debt is high, the overall stock of short-term ### Short-term external debt: stock and composition - As of end-2013, short-term external debt accounted for about 33 percent of total nonfinancial sector external debt, excluding intercompany debt. - The share is close to unchanged when including intercompany debt. - The stock of short-term debt, excluding intercompany debt, is around 4 percent of GDP. - Poland’s overall external NFC debt (loans, trade credits, and other types of debt) accounted for about 12 percent of GDP as of end-2013. - Of the debt maturing within 12 months (rollover needs), close to 60 percent represented intercompany debt. ### Flow vulnerabilities and rollover risk - Rollover needs are high: overall external NFC debt due within the subsequent 12 months equals about 12 percent of GDP (44 percent of total NFC debt). - Close to 60 percent of rollover needs are due to intercompany debt, which would presumably be easier to roll over than wholesale financing. ### Quality and composition of corporate issuance - The outstanding stock of non-investment grade debt securities in Poland is around 3 percent of GDP. - Non-investment grade issuance increased to more than 80 percent of total NFC securities issuance in 2013 (50 percent in 2012) and is now above the median EM. - International corporate bond issuance in Poland has been modest, below that in emerging Europe and other EMs. - Currency and debt profiles of debt securities are described as sound: - The foreign currency share in overall NFC debt issuance has started to decline as more borrowers borrow in local currency. - Foreign currency issuance is below that of many other EMs. - Poland has lower debt to equity issuance than other EMs, resulting in declining debt to equity ratios. ### Firm-level vulnerabilities: accounting ratios and indicators - Leverage: - Total liabilities to total assets stabilized at just above 40 percent after a decade of decline. - Short-term debt to total debt picked up slightly in 2012. - Debt to equity and debt to assets are low relative to other emerging markets. - Liquidity: - The current ratio and quick ratio deteriorated in Poland in 2012, placing Poland among emerging markets that are relatively vulnerable on liquidity measures. - Profitability: - Return on assets (RoA) and return on equity (RoE) declined in 2012, approaching levels last seen in 2008. - Valuation: - Tobin’s Q in 2012 was only slightly above one, suggesting relatively efficient markets. - Default risk indicators: - Altman Z-Score, Ohlson O-Score (converted into probabilities), and Black-Scholes-Merton (BSM) Default Probability all showed a slight decline in 2012. - Default probabilities in 2012 ranged between 1.5 percent (BSM probability) and 8 percent (Z probability). - Distance to default increased in 2012, consistent with declining distress indicators. ### High-frequency default indicators and sectoral patterns - Moody’s KMV one-year-ahead probability of default: - The 90th percentile in early 2014 remained above 15 percent. - The median probability is low at less than 1 percent. - Improvements in 2013 were visible across a number of industries, notably construction and real estate development. - Durable consumer products: median company default probability recently recorded at around ½ percent (compared with >10 percent for more than 50 percent of companies in early 2009). ### Debt-at-risk history and cross-country context - Debt at risk (share of total corporate debt with ICR < 1): - Less than 5 percent in 2007. - Peaked at 25 percent in 2009. - Improved in 2010, worsened again in 2011–12; by 2012, debt at risk increased to above 10 percent of debt. - IMF (2014a) comparison: Poland’s debt at risk is on par with India and South Africa but above Colombia, Malaysia, and Mexico (as noted in the source). ### Stress-test scenarios and assumptions - Data: - Firm-level sample from Standard & Poor’s Capital IQ for public and private nonfinancial companies (including real estate companies) in Poland. - Private companies are a relatively small share of the sample. - Shock types considered: - Interest rate shock: - Baseline interest rate on debt = interest expense in year t relative to last year’s debt. - Apply a 500 basis point shock to the baseline rate. - Assume return on financial assets increases by 300 basis points (instead of 500 basis points) to reflect higher return on marketable securities, investment securities, and cash and bank balances. - Also explore abstracting from changes in the return on financial assets (EBIT unaffected). - Profit shock: - Operating income (EBIT) is shocked directly by assuming a 25 percent decline. - Combined interest rate and profit shock: - Adverse combined shock of the 500 basis point interest increase and 25 percent profit decline. - In this combined scenario, assume returns on financial assets do not improve as interest on debt increases. - Exchange rate shock: - Aggregate-level shock due to lack of firm-level currency composition data. - Close to 40 percent of total NFC debt (including intercompany debt) is external. - 66 percent of external debt of the NFC sector is in foreign currency (FX). - Consider a 30 percent exchange rate depreciation. - FX loss on debt = FX loss on debt principal + FX loss on foreign currency interest expense. - Hedge ratio assumptions explored: no hedging, 50 percent hedge ratio, and 80 percent hedge ratio. - Maturing FX debt relative to total FX debt outstanding proxied by the share of maturing FX denominated debt securities from Dealogic. - Data cleaning: - Outliers where the baseline interest rate was above 50 percent or the baseline ICR was negative were removed. - Missing observations for financial assets are set to zero. ### Stress-test findings (summary from text) - Stress tests indicate moderate vulnerabilities in the corporate sector under the scenarios considered. - Large shocks (for example combined interest and profit shocks, or large exchange rate depreciations without hedging) could cause strains given high rollover needs and segments of debt at risk. - A substantial share of vulnerabilities is likely related to intercompany debt; firm-level data limitations prevent a complete intercompany/non-intercompany breakdown in the stress tests. *Source: IMF staff analysis as presented in the provided content unit.* ### 24. The results suggest that Poland’s corporate sector is vulnerable to large interest rate ### _cr14174 - 24. The results suggest that Poland’s corporate sector is vulnerable to large interest rate shocks ### Corporate-sector vulnerability to interest-rate shocks - A significant increase in interest rates, quickly translated into higher corporate borrowing costs, combined with a 25 percent adverse profit shock could more than double debt at risk to reach around 35 percent of debt. - Baseline debt-at-risk is on par with that for the United Kingdom in IMF (2011), but the effect of shocks appears somewhat larger. - The analysis assumes debt is generally tied to variable interest rates; the potential prevalence of fixed-rate debt would mitigate the impact of the shock. - Intercompany debt may be less subject to interest-rate risk, depending on pricing and whether the shock is global vs. Poland-specific. ### FX exposure and hedging - The estimated loss on FX debt maturing in 2014 from 30 percent zloty depreciation is manageable (below 5 percent of EBIT). - In the absence of currency hedges, the total valuation effect (accounting for the full amount of outstanding FX debt) could be substantial (around 20 percent of EBIT). - Natural hedges, abstracted from the analysis, may mitigate FX valuation losses. - Intercompany debt could reduce FX shock risk; when excluding intercompany debt from NFC aggregates (extreme example: fully excluding intercompany debt), the estimated valuation loss related to 30 percent zloty depreciation could be substantially lower. ### Aggregate NFC debt, rollover needs, and sectoral conditions (Conclusions) - NFC debt and accompanying rollover needs in Poland are high. - Poland’s NFC debt does not stand out relative to other emerging European countries, but external NFC debt is high compared to non-European emerging markets. - The stock of nonperforming loans is high in the corporate sector (particularly among SMEs). - Noninvestment grade bond issuance has increased recently, and rollover needs are substantial. - Mitigating factors: - The share of intercompany debt in total NFC external debt is high; if less vulnerable to sudden stops than wholesale funding, this alleviates rollover risks. - Default risk has declined across most sectors. - Debt–to-equity ratios are declining. - More borrowers are able to issue debt in local currency, moderating potential FX risks. - Vulnerability summary: - Small to moderate shocks: debt at risk remains contained. - FX shocks: impact depends on hedging—little or no hedging could result in substantial valuation losses; significant hedging would lead to much lower losses. - Severe interest-rate shocks: could pose a risk, though not all debt may be subject to variable rates; intercompany lending likely mitigates default concerns. ### Stylized facts on VAT revenues and cyclicality (chapter continuation) - VAT collections in Poland have been lackluster since the 2008 financial crisis despite a VAT rate increase in 2011. - VAT share of GDP declined from a peak of 8.2 percent of GDP in 2007 to 7.0 percent in 2013. - The VAT compliance gap widened from 0.2 percent of GDP in 2007 to 1.5 percent of GDP in 2011; more recent data suggest the VAT gap may have widened further during 2012–13. - Cross-country sample and methodology: - Elasticity estimates use co-integrating panel models for 21 EU countries between 1999:Q1–2013:Q4. - The sample covers Austria, Belgium, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Luxembourg, Netherlands, Poland, Portugal, Slovak Republic, Slovenia, Spain, Sweden, and United Kingdom. - Main econometric results on VAT elasticities: - VAT elasticities are strongly pro-cyclical and asymmetric over the cycle. - Average point estimates range from 1.0 to 1.6 across estimation methodologies. - VAT elasticities tend to be substantially higher during cyclical downturns. - Implications and observed trends: - Negative domestic demand shocks may have significant effects on VAT collections, especially during crisis episodes. - The observed downward trend in VAT collection effectiveness since 2008 is not fully explained by cyclical dynamics; structural factors (e.g., tax compliance problems) appear relevant. - In 12 out of the 21 countries analyzed, VAT rates were hiked after 2008; unweighted average rates rose from 19.8 percent in 2007 to 21.5 percent in 2013. - A rough measure suggests VAT collection effectiveness declined by about ½ percentage point of GDP since 2008, broadly matching the increase in VAT rates. - Policy and forecasting implications: - Forecasting models for VAT revenues need to account for the cyclical position of the economy to reduce estimation bias; using an average static elasticity would tend to overestimate revenues during downturns and vice versa. - The weakening trend in VAT collection effectiveness calls for attention to tax administration issues. - Stylized dynamics around downturns: - Downturns are identified as episodes of at least 8 consecutive quarters with a negative output gap (HP-filter). - There were 29 downturn episodes in the sample, including 11 before the onset of the 2008 financial crisis. - Typical pattern: VAT revenues collapse at the onset of a downturn (between quarters T+0 and T+2) and rebound strongly in the second year (quarters T+4 to T+7). *Source: _cr14174 - 24. The results suggest that Poland’s corporate sector is vulnerable to large interest rate shocks (PDF chapter/section).* ### 14. In Poland, the cyclical behavior of VAT revenues followed a similar pattern. While ### 14. In Poland, the cyclical behavior of VAT revenues followed a similar pattern. ### C. Empirical Design - Model specification (baseline): year-on-year difference operator applied to logs; T = log of VAT tax revenues, y = log of domestic demand, v = standard VAT rate; sub-index i = countries, t = time. Parameters are country-specific, including the error correction term. - Estimation frameworks considered: - Dynamic panel GMM (Arellano and Bond, 1991; Blundell and Bond, 1998) with country dummies (homogeneous slopes, heterogeneous intercepts). - Mean Group (MG) estimator (Pesaran and Smith, 1995): allows full slope heterogeneity; consistent under heterogeneity. - Pooled Mean Group (PMG) estimator (Pesaran et al., 1999): assumes common long-run relationships and heterogeneous short-run dynamics. - Cross-sectional dependence addressed using Common Correlated Effects (CCE) estimators (Pesaran, 2006) and Binders and Offermanns (2007) correction: - Augment PMG and MG with cross-sectional averages of dependent and independent variables to proxy unobserved common factors. - Data: - Quarterly macroeconomic data from the European Commission. - Level values of VAT revenues and domestic demand in constant euro terms (2005 constant euro exchange rates). - Year-on-year changes of logs used to mitigate seasonality. - Sample: 21 European countries, 1999:Q1 to 2013:Q4. ### D. Results — Linear effects and crisis split - Core finding on cyclicality: - Strong pro-cyclicality of VAT revenues with respect to domestic demand. - A one percentage point increase in domestic demand growth leads to a 1.0‒1.6 percentage point increase of VAT revenues. - Adjustment mechanism: a departure from long-term equilibrium is reduced by about 42‒63 percent in the next quarter. - Key estimates from Table 1 (short-term coefficients; standard errors in parentheses): - Error correction term: -0.424*** (0.0475), -0.488*** (0.0484), -0.496*** (0.0511), -0.627*** (0.0528) across PMG, MG, PMG-BO, CMG. - d. log Domestic demand: 1.557*** (0.174), 1.644*** (0.200), 1.047*** (0.157), 0.998*** (0.212). - d. VAT rate: 1.565*** (0.443), 1.664*** (0.494), 1.126* (0.677), 0.225 (0.521). - Observations: 1,050; Number of countries: 21. - Crisis period sensitivity (Table 2; CMG estimates): - Before the crisis (CMG: Before the crisis): Error correction term -0.776*** (0.0790); d. log Domestic demand 0.862*** (0.327); d. VAT rate 0.749 (0.647); Observations 504. - Since the crisis (CMG: Since the crisis): Error correction term -0.660*** (0.0660); d. log Domestic demand 0.997*** (0.323); d. VAT rate 0.493 (1.076); Observations 546. - Interpretation: - Sensitivity of VAT revenues to domestic demand shocks increased marginally in 2008‒13. - Possible drivers: asymmetry in response (stronger declines than recoveries) and potential reduction in VAT efficiency and tax compliance post-crisis. ### E. Exploring Asymmetries and Non-Linear Effects - Asymmetric specification: - Domestic demand shocks split into negative and positive deviations relative to country-specific sample averages. - Equation [2] estimates separate elasticities for expansions (positive shocks) and contractions (negative shocks): coefficients capture VAT elasticity during domestic demand expansions and contractions. - Estimation approach to avoid bias: - Dynamic fixed-effect (DFE) panel with Least Square Dummy Variable Corrected Estimator (LSDVCE) (Bruno, 2005) used to mitigate Nickell bias and skewness in negative vs positive shock frequency. - Main asymmetric findings (Table 3; standard errors in parentheses): - PMG-BO: Negative demand shock 1.425*** (0.322); Positive demand shock 0.583* (0.306); Error correction term -0.364*** (0.0411); Observations 1,029; Number of countries 21. - CMG: Negative demand shock 0.827*** (0.217); Positive demand shock 0.727*** (0.257); Error correction term -0.631*** (0.0538); Observations 1,050; Number of countries 21. - DFE and LSDVC: Negative demand shock 1.599*** (0.279) / 1.599*** (0.243); Positive demand shock 0.978 (0.774) / 0.978*** (0.283); lagged negative demand shock significant and negative: -0.838*** (0.245) / -0.865*** (0.271). - Robustness: - Results robust across specifications (with/without cross-sectional dependence controls, added lags, homogeneity imposed). - Interpretation: - VAT elasticity is significantly higher in “bad times” than in “good times”: VAT revenues decline more during downturns than they recover during upturns. - Consistent with evidence of pro-cyclical VAT inefficiency and increases in VAT gaps during recessions. ### E (continued). Role of protracted shocks - When focusing on protracted episodes (sustained up to 4 quarters) of positive or negative domestic demand deviations: - VAT revenue cyclicality is not statistically different between protracted positive and negative shocks. - Key estimates from Table 4 (short-term coefficients; standard errors in parentheses): - Error correction term: -0.495*** (0.0516) PMG-BO; -0.530*** (0.0485) CMG. - Protracted negative demand shock: 1.318*** (0.264) PMG-BO; 0.900*** (0.318) CMG; 1.133*** (0.304) FE; 0.809*** (0.116) LSDVC. - Protracted positive demand shock: 0.898*** (0.327) PMG-BO; 0.778*** (0.271) CMG; 1.073** (0.428) FE; 0.846*** (0.231) LSDVC. - d. VAT rate estimates: 1.423* (0.768), 0.953* (0.500), 1.151** (0.446), 0.806* (0.416) across specifications. - Observations: 1,050 (PMG-BO, CMG, FE); 1,029 (LSDVC). Number of countries: 21. ### F. Policy Implications - Forecasting implications: - State-dependent (asymmetric) VAT elasticities should be used when forecasting VAT revenues and the fiscal balance. - The asymmetric specification provides better forecasting properties than a simple linear model and tracks VAT revenue dynamics in Poland and other EU countries well. - Compliance and administration: - Asymmetric VAT elasticities are consistent with worsening tax compliance during downturns. - When domestic demand is very weak relative to historical averages, risks of deeper revenue underperformance increase and tax compliance issues may intensify. - Such behaviors are difficult to reverse; tax-to-GDP ratios may remain well below pre-shock levels even after recovery. - Policy recommendation: pursue measures to strengthen revenue administration and safeguard VAT compliance, with particular focus on the VAT segment. ### G. Conclusion - Innovations and methodological contributions: - Use of cointegrating panel techniques, control for statutory VAT rates, and explicit treatment of cross-sectional dependence to account for common shocks and heterogeneous responses. - Main empirical conclusions: - Pro-cyclicality of VAT revenues is strong across EU countries. - Asymmetric effects: downturns reduce VAT revenues more than upturns restore them. - Increases in statutory VAT rates in EU countries have not translated into higher revenues; descriptive evidence shows a negative correlation between changes in VAT rates and VAT revenue performance (causality left for future work). - Practical implications summarized: - Forecasting exercises and assessments of structural fiscal stance must incorporate state-dependent VAT elasticities, especially during downturns. - VAT revenue losses during sharp slowdowns may be difficult to reverse; VAT revenues-to-GDP ratios in Poland and other EU countries could remain below pre-shock levels for a protracted period when VAT gaps or inefficiencies worsen. - For Poland specifically: use pro-cyclical VAT elasticities for forecasting, diagnose sources of VAT underperformance, and prioritize strengthening tax administration focused on VAT. *Source: IMF staff report chapter on VAT cyclicality and asymmetries (Poland and EU sample), 1999:Q1–2013:Q4.* ### References ### _cr14174 - References ### References - Aizenman, J. and Jinjarak, Y., 2008, “The Collection Efficiency of the Value Added Tax: Theory and International Evidence”, The Journal of International Trade & Economic Development, vol. 17(3), pages 391–410. - Arellano, M. and Bond, S. 1991, “Some Tests of Specification for Panel Data: Monte Carlo Evidence and An Application to Employment Equations”, Review of Economic Studies, vol. 58(2), pages 277–97, April. - Binder, M. and Offermanns, C., 2007, “International Investment Positions and Exchange Rate Dynamics: A Dynamic Panel Analysis”, CESifo Working Paper Series No. 2095. - Blundell, R. and Bond, S., 1998, “Initial Conditions and Moment Restrictions in Dynamic Panel Data Models”, Journal of Econometrics, vol. 87(1), pages 115–143, August. - Bruno, G., 2005, “Estimation and Inference in Dynamic Unbalanced Panel-Data Models with a Small Number of Individuals”, Stata Journal, vol. 5(4), pages 473–500, December. - Ebrill, L., Bodin, J.-P., Keen, M. and Summers, V., 2001, “The Modern VAT”, Washington: International Monetary Fund. - European Commission, 2013, “Study to Quantify and Analyze the VAT Gap in the EU-27 Member States: Final Report”, TAXUD/2012/DE/316, Warsaw, July 2013. - Hercowitz, Z. and Strawczynski, M., 2004, “Cyclical Ratcheting in Government Spending: Evidence from the OECD”, The Review of Economics and Statistics, vol. 86(1), pages 353–361, February. - Pesaran, M., 2004, “General Diagnostic Tests for Cross Section Dependence in Panels”, CESifo Working Paper Series No. 1229, CESifo Group Munich. - Pesaran, M., 2006, “Estimation and Inference in Large Heterogeneous Panels with a Multifactor Error Structure”, Econometrica, vol. 74(4), pages 967–1012. - Pesaran, M. and Smith, R., 1995, “Estimating Long-Run Relationships from Dynamic Heterogeneous Panels”, Journal of Econometrics, vol. 68(1), pages 79–113, July. - Pesaran, M., Y. Shin, and R. Smith, 1999, “Pooled Mean Group Estimation of Dynamic Heterogeneous Panels”, Journal of the American Statistical Association, vol. 94(446), pages 621–634. - Reckon, 2009, “Study to Quantify and Analyze the VAT Gap in the EU-25 Member States” Taxation Studies 0029, Directorate General Taxation and Customs Union, European Commission. - Westerlund, J., 2007, “Testing for Error Correction in Panel Data”, Oxford Bulletin of Economics and Statistics, vol. 69(6), pages 709–748, December. ### Key findings and scope (from chapter text) - Purpose: Assess the long-term fiscal impact of the 2014 pension changes in Poland and discuss key fiscal risks from pre-existing pension system flaws. - Projection horizon: 2014‒60. - Main baseline assumption: Half of the contributors switch to the first pillar (50 percent). - Baseline fiscal outcome: - The 2014 pension changes deliver an improvement in the fiscal accounts of about 30 percent of GDP in net present value terms between 2014‒60, compared with no policy change. - This improvement is matched by an increase in implicit pension liabilities; treated equally, the pension changes would be fiscally neutral. - Timing of cash flows: - Positive cash flows to the fiscal sector until around 2050, followed by increasing negative flows in the outer years of the projection (and beyond). ### Background and institutional features - System architecture: - Three pillars: first pillar = notional defined contribution (NDC), PAYGO, operated by social security administration; second pillar = individual accounts managed by private pension funds (OFEs); third pillar = voluntary individual accounts. - 2014 reform elements: - Transfer of about half of pension fund assets (and corresponding liabilities) to the first pillar. - Redirection of contributions to the first pillar and centralization of the payout phase in the first pillar. - Individuals could keep accounts in pension funds and contribute 2.92 percent of their salaries to these accounts by filing an administrative request within a four-month window; otherwise new contributions defaulted to the first pillar. - Pension funds banned from investing in government bonds; limits on holdings of foreign securities to rise from 5 percent in 2013 to 30 percent by 2016. - Payout centralization: all contributions of individuals 10 years before retirement directed to the first pillar; assets in the second pillar gradually transferred to the first pillar starting 10 years before retirement (the “zipper”). ### Methodology and assumptions - Model: Simplified static cohort model using age and gender profiles of balances in first- and second-pillar accounts as of end-2012, cohort-specific participation rates, age-earning profiles, social security contributions, and Eurostat demographic projections for 2015‒60. - Behavioral and macro feedback: Assumes no behavioral responses and no feedback effects from pension changes to macroeconomic and labor market conditions. - Retirement probabilities: All beneficiaries enter the system at the statutory retirement age. - Asset return assumption: Equal returns between NDC and second pillar assets; first pillar returns based on statutory indexation; second pillar (net of fees) assumed to evolve according to the average GDP growth of the previous five years. - Discounting: Net present value computed using a discount rate equal to GDP growth. ### Fiscal implications and liabilities - Transition costs: The partial unwinding of the funded second pillar recaptures a fraction of transition costs incurred after the 1999 reform. - Explicit versus implicit liabilities: - Asset transfers and redirections improve fiscal balance and reduce explicit public debt but increase pension liabilities recorded in NDC accounts. - If implicit and explicit liabilities are treated equally, the reform is fiscally neutral (reference to Mackenzie, 2003). - Legal debt thresholds referenced: - Polish Constitution caps public debt at 60 percent of GDP (national definition). - Public Finance Law requires mandatory fiscal consolidation if debt to GDP exceeds 55 percent of GDP. - Recent law amendment established two preventive debt thresholds at 43 and 48 percent of GDP, replacing a previous threshold of 50 percent of GDP. ### Legacy fiscal risks and vulnerabilities - Replacement rates: The 1999 reform led replacement rates to decline from around 60 percent currently to 30 percent in the future (projection to 2060), creating risks of old-age poverty. - No economy-wide savings gain: Accumulation of assets in OFEs between 1999‒2013 was almost perfectly matched by an increase in public sector debt. - Specific risks: - Non-negative indexation limits on NDC accounts could lead to fiscal imbalances in crisis scenarios (growth of pension liabilities could exceed growth of contributions). - Misalignment of disability benefits with the core pension system may create incentives to claim retirement as disabled. - Special pension schemes (farmers, miners, uniformed services) may have unaddressed imbalances posing long-term fiscal pressure. ### Policy recommendations (parametric and institutional options) - Addressing drop in replacement rates: - Expand pension coverage to workers under special temporary contracts (noting some temporary contracts fall under the civil code and do not require employers and employees to contribute to the social security system). - Assess alternatives to incentivize private savings in the voluntary third pillar. - Improve financial literacy. - Revising indexation of first-pillar NDC accounts: - Allow for negative coefficients in indexation. - Second-best: set indexation of the main account to the five-year average growth of the wage bill to reduce negative fiscal impact of temporary labor market shocks. - Adjusting disability formula: - Align disability benefits as much as possible with benefits of the core pension system to avoid adverse incentives and moral hazard. - Tackling special pension schemes: - Subject special pension schemes to parametric reform and align them with the core pension system to ensure actuarial solvency; expected to yield fiscal savings in the long run rather than substantial short-term fiscal impact. - Fiscal prudence: - Maintain fiscal consolidation and the new permanent expenditure rule, because reform yields larger payments in more distant future despite near-term debt reduction below the 55 percent of GDP legal threshold. ### Box 1 — The 1999 Pension Reform (summary of key parameters) - Reform converted system from defined-benefits to defined-contributions under a three-pillar structure. - Contribution structure: - Pension contributions maintained at 19½ percent of the wage bill. - Between 1999 and 2011, 7.3 percent of the wage bill transferred to the second pillar; 12.2 retained in the first pillar. - Coverage rules at reform: - Contributions to first and second pillars compulsory for individuals below age 30 at time of reform. - Ages 30–49 allowed voluntary contribution to second pillar. - Individuals aged 50 or more at reform not allowed to open private pension fund accounts. - Indexation and pension determination: - Individual notional accounts in first pillar indexed to the evolution of the nominal wage fund. - Second pillar returns based on fund portfolios (net of fees). - Upon retirement, pensions determined by dividing capitalized accounts in first and second pillars over life expectancy. - Over time, pensions indexed by CPI inflation plus 20 percent of the real wage growth. - Projected replacement rates under the system: drop from around 60 percent in 2014 to 30 percent by 2060. _Italic: Source — _cr14174 - References (IMF PDF)._ ### 18. The estimations leave outside some important risks that are difficult to model. Material ### 18. The estimations leave outside some important risks that are difficult to model. Material ### Model scope and accounting assumptions - Estimations exclude material risks (both positive and negative) from age-specific employment probabilities, productivity changes, and underestimated life expectancies; these risks are not assessed in this paper. - Periodic updates of the financial soundness of the pension system are recommended to track major trends and assess corrective measures. - Analysis focuses on cash flows and abstracts from differences in accounting criteria across methodologies (Box 2). - The one-off transfer of public debt and subsequent asset transfers in the context of the zipper are treated as positive fiscal cash flows (Annex 2). Corresponding pension liabilities are registered in notional accounts and indexed over time according to statutory formulas. - Associated pension payments are computed by dividing capitalized balances in first-pillar accounts at retirement (adjusted by pre-retirement asset transfers) by life expectancy; resulting annuities are indexed according to legal provisions. - Note on rates and comparisons: - Average real returns 1999–2012: second pillar about 3.4 percent; first pillar 3.9 percent; compared with 3.9 percent GDP growth. - In theory, discount rate should measure opportunity cost of fiscal resources or a risk-free return; in absence of such assets, GDP growth is used as a technically manageable discount rate. - Alternatively, the cash flows associated with the one-off debt transfer could be measured by the drop in associated debt service over time; financial result will be equivalent under the assumption of equality between the interest rate on public debt and the rate of return used to compute the net present value. ### Box 2. Accounting for the Fiscal Impact of Pension Changes - Channels affecting government accounts: - One-off asset transfer: in early 2014, pension funds transferred to the first pillar holdings of T-bonds (written-off immediately), government-guaranteed bonds, municipal bonds, and cash totaling at 51½ percent of their total assets. - Opt-out from second pillar: insured can opt-out and redirect all new contributions to the first pillar (assets remain in private funds until gradually transferred to the first pillar starting 10 years before retirement). - Lower contribution rate to second pillar: at 2.92 percent instead of 3.5 percent of income. - Redirecting contributions of the elderly: contributions of workers close to retirement will go to the first pillar starting 10 years before retirement. - Gradual transfer of assets: assets of persons 10 years before retirement will be gradually transferred to the first pillar. - Treatment differences by methodology: - Polish statistics: one-off debt transfer and gradual pre-retirement transfers registered above-the-line as first pillar revenue. - ESA95: classifies both one-off and gradual pre-retirement transfers as revenues. - ESA2010: will classify the one-off debt transfer as a below-the-line financing transaction; accounting of subsequent asset transfers in the zipper may be below-the-line financing, gradually registered as revenue as pensions are paid out. - IMF GFS: classifies the one-off debt transfer as a below-the-line transaction; zipper asset transfers registered as financing below-the-line and gradually reversed when corresponding pensions are paid out. - Implication: Methodological differences lead to discrepancies in fiscal accounts. Classifying asset transfers as financing (below-the-line) will put downward pressure on public debt without matching entries on the fiscal balance; asset transfers increase implicit liabilities, materializing in higher pension payments with matching fiscal revenues at a later stage. ### C. Assessing the Impact of the 2014 Pension Changes — Baseline Results - Baseline uses long-term official projections for comparability; key macro variables (real GDP and real wage growth) follow MoF long-term projections (Text Figure 1). - Model behavior noted: - Left panel: wage dynamics stronger than GDP growth, compensating for decreasing labor supply, leading to stable labor share in GDP. - Right panel: shows contributors to notional accounts in the first pillar, contributors under the notional defined contribution system, and contributors to OFEs. Sharp drop in 2014 from assumption that half of contributors opt out from the second pillar. - Cash flows computed using these projections and a static cohort model, enabling comparison with official results. - Main baseline fiscal-impact findings: - Impact on fiscal cash flows is positive over the projection horizon and broadly in line with official estimates. - Cumulative cash flows build up in the short- and medium-term and peak at about 31 percent of GDP by 2050. - By 2060 cumulative gains decline to around 29 percent of GDP. - Factors affecting net cash flows: - Higher flow of contributions to the first pillar: - Additional contributions relative to no policy change cause substantial drop in transition costs. - Gains originate from: (i) individuals assumed to opt-out of the second pillar; (ii) reduction in contributions to OFEs for individuals who maintain second-pillar accounts; (iii) individuals approaching retirement, whose contributions are entirely redirected to first pillar starting 10 years before retirement. - Using rate of return equal to GDP growth, these additional contributions generate an improvement in fiscal accounts by some 30 percent of GDP in net present value terms between 2014–60. - Asset transfers: - Pension fund holdings of government debt and government-guaranteed debt transferred in 2014 represent 9 percent of GDP, resulting in lower debt service during the projected horizon. - Assets in the second pillar transferred to the first pillar starting 10 years before retirement provide a cumulative 18 percent of GDP in cash flows to the fiscal by 2060. - Higher pension payments: - Pension benefits remain unaffected by the 2014 changes, but increased contributions and asset transfers to the first pillar lead to higher public-sector pension payments by a cumulative 28 percent of GDP by 2060, partially offsetting previous cash flow gains. - Specifically, effect of one-off transfer will be largely wiped out by 2060, as 82 percent of the initial cash flow gain will have been paid out in pension benefits. - Infinite-horizon note: - Positive effect on fiscal aggregates likely smaller under an infinite horizon; improvements in cash flows are matched by increases in implicit pension liabilities that must be repaid beyond the projected horizon. - Net fiscal gains are on a downward trajectory at the end of the projection, suggesting equilibrium is not yet reached by 2060; assessment over an infinite horizon likely produces lower cash flow gains. ### Sensitivity Analysis on the Impact of the 2014 Pension Changes - Switching rates to the first pillar: - Voluntary participation in the funded second pillar is not crucial for fiscal impact. - If all contributors decide to stay in the second pillar, net present value of fiscal gains = 23 percent of GDP under the baseline. - If all contributors opt-out of the second pillar, net present value of fiscal gains = 35 percent of GDP. - Conclusion: voluntary participation not the key component for long-term fiscal impact; gradual transfer of assets prior to retirement plays a more significant role. - Four macroeconomic scenarios (relative dynamics based on historic correlations calibrated with VAR models): - Scenario 1: Historical Dynamics — real GDP growth averages 2.4 percent; wage bill growth averages 2.5 percent; wage bill to GDP stabilizes around 25 percent. - Scenario 2: Severe Downturn — 3.8 percentage point drop in GDP growth relative to baseline in years 2020‒23, followed by permanent drop in potential growth equivalent to one standard deviation; real GDP growth averages 1.7 percent, with a minimum of ‒3.2 percent and 7 years of negative performance. - Scenario 3: Sluggish Wages — labor productivity slows, wage bill growth averages 1.7 percent during projection; ratio of wage bill to GDP drops from 26 ½ percent in 2014 to 22 percent at end of projection. - Scenario 4: Protracted Crisis — severe protracted crisis implying cumulative 23 percent and 20 percent drop in GDP growth and wages over 2020‒25 relative to baseline. - Table 1 summary statistics (selected variables under alternative scenarios) — preserved exactly as presented: - GDP Growth: Actual 2001-13 Average 3.6 Min 1.2 Max 6.8 St. Dev. 1.9; Baseline Average 2.3 Min 1.2 Max 4.3 St. Dev. 0.9; Historical Dynamics Average 2.4 Min -0.7 Max 6.8 St. Dev. 1.6; Severe Downturn Average 1.7 Min -3.2 Max 6.8 St. Dev. 1.8; Sluggish wages Average 2.4 Min -0.3 Max 6.8 St. Dev. 1.5; Protracted Crisis Average 1.3 Min -7.1 Max 3.8 St. Dev. 2.5. - No. Years with Negative Dynamics: Actual 2001-13 0; Baseline 0; Historical Dynamics 0; Severe Downturn 6; Sluggish wages 3; Protracted Crisis 6. - Wage Bill Growth: Actual 2001-13 Average 3.1 Min -1.4 Max 10.8 St. Dev. 3.0; Baseline Average 2.4 Min 0.8 Max 5.5 St. Dev. 1.2; Historical Dynamics Average 2.5 Min -0.8 Max 4.4 St. Dev. 1.2; Severe Downturn Average 1.8 Min -2.1 Max 4.4 St. Dev. 1.4; Sluggish wages Average 1.7 Min -0.6 Max 4.4 St. Dev. 1.2; Protracted Crisis Average 1.7 Min -10.3 Max 5.5 St. Dev. 2.8. - No. Years with Negative Dynamics (wage bill): Actual 2001-13 3; Baseline 0; Historical Dynamics 8; Severe Downturn 12; Sluggish wages 5; Protracted Crisis 3. - Robustness finding: - Baseline results fairly robust to sensitivity analysis; fiscal impact under distressed scenarios deviates by less than 6 percent of GDP from baseline (Text Figure 3). - Robustness depends on stability of share of wage bill to GDP; indexation of sub-accounts linked to average GDP growth means downturns slow growth of pension liabilities, helping counteract weaker contributions. - Sluggish wage growth scenario has the largest adverse impact because slowdown of wage bill growth relative to GDP erodes net fiscal inflows. ### Impact of the 2014 Changes on the Second Pillar - Scaling back outcome: - One-off debt transfer leads to sharp decline in size of second pillar from around 18 percent of GDP in 2013 to around 9 percent of GDP in 2014. - Subsequent drop in flow of contributions to second pillar plus zipper transfers reduces second pillar assets over time. - Projections: - Under no policy change, second pillar assets would have reached 36 percent of GDP by 2060 (Text Figure 4). - Under 2014 pension changes, second pillar assets projected to stabilize at around 7 percent of GDP in 2060. - The difference in second pillar asset size matches estimated increase in cash flows to fiscal and increase in implicit pension liabilities for the government. - Viability depends on opt-in share: - If all contributors shift contributions to the first pillar, second pillar will cease to exist before 2050 as remaining assets are transferred to the first pillar. - If all current and future contributors opt in, second pillar assets increase to about 17 percent of GDP by 2030 and stabilize at around 14 percent of GDP by 2060. ### D. Assessing Legacy Fiscal Risks - Despite positive fiscal effects, pension changes do not eliminate fiscal risks from legacy flaws; larger first pillar will still require further reforms to mitigate contingencies. - Section assesses fiscal risks from indexation of notional pension accounts, inadequate replacement rates, special pension regimes, and disability benefits, and provides policy recommendations. ### First-Pillar Indexation - Indexation rules and risks: - Indexation of notional accounts does not guarantee long-term financial viability. A defined-contribution pension scheme can be viable if indexation of notional pension capital does not exceed dynamics of pension contributions (Samuelson, 1958). - Original 1999 reform set indexation below dynamics of pension contribution base and allowed negative indexation coefficients. Rules were later revised to fully reflect growth of covered wage bill and to impose a zero-floor limit on indexation. - Indexation of notional sub-accounts created in 2011 set to 5-year average nominal GDP growth and also restrained by a zero-floor limit. - Risk assessment approach: - Fiscal risks of asymmetric indexation assessed by computing projections under current rules for distressed macroeconomic scenarios and comparing to a counterfactual policy that allows for negative indexation. - This isolates risks related to asymmetric indexation and quantifies them in the long term. *Italic: IMF staff estimates based on ZUS, GUS, KNF and Eurostat.* ### 31. Indexation of the main accounts may pose a large fiscal risk. Each of the shock scenarios ### _cr14174 - 31. Indexation of the main accounts may pose a large fiscal risk. Each of the shock scenarios ### Indexation and fiscal risk - Each of the shock scenarios would activate zero-floor indexation limit in the main account, with adverse fiscal consequences compared to a scenario allowing negative indexation. - In the projections, this impact would range from 3 percent to 25 percent of GDP (Text Figure 5). - Shocks heavily affecting the wage bill (as under the Sluggish Wages and the Protracted Crisis scenarios) could be particularly costly for public finances under the current indexation mechanism of the main account. - Footnote: The calculations differ from those presented in the sensitivity analysis in two respects. First, the shocks are applied to all the notional accounts in the first pillar (including the main accounts and the full balances of the sub-accounts). Second, the calculations are compared with a counterfactual policy that allows for negative indexation in the first pillar. ### Mitigation options for indexation risk - The largest fiscal risk emanates from the indexation of the main account; the sub-account could cause imbalances only in extreme cases. - Most transparent mitigation: allow negative indexation of notional accounts, as envisaged by the original pension reform. - Alternative mitigation: make indexation of the main account dependent on multi-year instead of annual dynamics of covered wage bill. - This change would suffice to eliminate the risks under the Historical scenario. - It would significantly mitigate risks under Severe Downturn and Sluggish Wages scenarios (all modeled as randomized scenarios based on VAR model). - However, it would not suffice to reduce the substantially adverse fiscal impact of the Protracted Crisis scenario. ### Replacement rates and long-term pension adequacy - The pre-reform system featured low effective retirement age and high replacement rates; 1999 reform moved system toward defined-contribution. - Model assumption: returns in the second pillar (net of fees) are equal to first pillar indexation. - Implication: market returns follow GDP and wage bill dynamics, making pensions indifferent to split of contributions between pillars; 2014 reforms neutral for replacement rates. - Projection: replacement rate projected to decline from around 60 percent currently to some 30 percent by 2060 (Text Figure 7). - EU Commission Ageing Report (EC, 2012) example: suggests a 27 percentage point drop in replacement rates by 2060. - The 2012 increase in statutory retirement ages (from 65 to 67 for men and from 60 to 67 for women) prevented a 7 percentage point deeper drop in replacement rates (Text Figure 8). - Gender equalization of retirement age reduced pension disparities between men and women. ### Old-age poverty risk and minimum pension guarantee - The risk of poverty among Polish retirees is currently below EU average, but will likely become more pronounced along the projected decline in replacement rates. - Estimates suggest the share of pensioners at risk of poverty could increase substantially in the long-run (Text Figure 9). - The threat of old age poverty appears particularly high assuming poverty thresholds remain unchanged relative to wages (social minimum and equivalised income), which may overstate long-term risks. - Legal minimum pension indexation accounts for only 20 percent of real wage growth (Text Figure 10). - By 2060, almost no retirees are projected to fall below the minimum pension, practically eliminating fiscal risk of topping-up benefits. - However, the minimum pension’s declining level relative to wages provides little comfort in alleviating old-age poverty risk. - Policy implication: develop efficient social assistance mechanisms for poor pensioners as important long-term challenge. ### Special pension regimes: coverage and fiscal burden - In 2010, occupational schemes for miners, “uniformed services”, and farmers: - Covered 2.1 million beneficiaries, equivalent to 29 percent of those in the regular system. - Outlays on special pensions were 2.7 percent of GDP or 29½ percent of regular pensions (Text Figure 11). - Ratio of contributions to pension outlays in occupational schemes tends to be low; uniformed personnel scheme fully paid by the state budget. - In 2010, gap between contributions paid by farmers and miners and pension spending under their schemes was around 1½ percent of GDP. - This gap would increase to 2.4 percent of GDP if uniformed personnel pensions are also included. - Special schemes are more generous and have lower effective retirement ages than the regular system (Text Figure 12). - Farmer benefits well below average, but contributions cover only 10 percent of outlays. - Possible reform alternatives: - Uniformed services: rationalize with parametric changes; 2013 changes increased required work period for pension rights from 15 to 25 years and set minimum retirement age at 55 years; consider further changes to link pensions to nature of work. - Miners: keep privileges only for those truly working under hardship conditions, or merge miners into main system and fund privileges via explicit budget subsidy or higher employer contributions. - Farmers: plans to cover them with personal income tax and make contributions and benefits dependent on income; revise enrollment criteria tied to land ownership to reduce misuse. ### Disability benefits and incentives - Maintaining current disability formula implies growing incentives for misuse. - Historical context: incidence of disability very high in early 1990s; subsequent tightening reduced abuse. - Under current rules, disability benefits could exceed projected pensions some 10‒15 years before statutory retirement age for lower income groups and among workers with shorter contribution periods (Text Figure 13). - Recommendation: retain insurance component in disability formula but better align disability benefits with old-age pensions to reduce adverse incentives as replacement rates drop. ### Conclusion and key policy recommendations - 2014 pension changes improve fiscal accounts by partial unwinding of second pillar, increasing contributions to first pillar and reducing (explicit) public debt and associated interest payments; public sector will take on larger future pension payments. - Statutory indexation clauses in notional accounts are linked to nominal wage bill and GDP, providing a normal link between implicit liabilities and contribution flows; NDC capitalization supports actuarial balance. - Legacy issues remain and merit policy action: - Excessively low replacement rates may increase old-age poverty incidence; promote awareness of insufficiency of projected pensions and explore alternatives to stimulate additional private savings. - Revise indexation mechanism of the main account: non-negative indexation may jeopardize financial soundness and translate into fiscal contingencies; mitigate by relaxing non-negative indexation clauses or linking indexation to multi-year wage bill dynamics. - Align relatively generous disability formula with old-age pension benefits to reduce adverse incentives as replacement rates decline. - Tackle longstanding reforms to special pension schemes for farmers, miners, and uniformed services. *Source: IMF staff estimates and analysis as presented in the chapter.* ### 3. Transfers to the second pillar were reduced in 2011. As part of a fiscal consolidation ### 3. Transfers to the second pillar were reduced in 2011. As part of a fiscal consolidation ### Pension contribution reallocation and indexing - Transfers to the second pillar were cut to 2.3 percent of the wage bill starting in May 2011. - The remaining 5 percent was redirected to subaccounts in the first pillar, boosting fiscal revenues. - These sub-accounts would be indexed at the average nominal GDP growth of the previous five years. - Authorities planned a partial reversion of pension contributions back to the second pillar, from 2.8 percent of the wage bill in 2013 to 3.5 percent of the wage bill by 2017. - The zipper mechanism: asset transfers associated with the zipper reduce public debt and the associated interest bill; transfers under the zipper, Tt, represent a positive cash flow affecting explicit debt. ### Statutory retirement age and uniformed services - Statutory retirement ages were increased starting in 2013. - A decision to gradually increase the statutory retirement age to 67 and equalize for men and women is expected to increase activity rates among the older age groups and improve replacement rates. - An increase in the retirement age for new entrants to uniformed services implemented in 2013 also contributes to improving the financial soundness of the pension system. ### Long-term demographic pressures - Declining fertility rates and increases in life expectancy imply adverse population dynamics. - The old-age dependency ratio (persons aged 65 and older relative to those between 15 and 64) is projected to increase from about 20 percent in 2012 to 70 percent by 2060 (Jabłonowski, and Müller, 2014). - These population dynamics are likely to exert substantial pressures on the PAYG system, with adverse fiscal implications. ### Stylized accounting and fiscal dynamics (Annex II) - Notional and capitalized accounts described: - Individual notional accounts in the first pillar (A1) indexed to the evolution of the nominal contribution base, subject to a zero-floor. - Individual sub-accounts created in 2011 (A2) indexed at the average GDP growth of the previous five years, subject to a zero-floor. - Individual sub-accounts after the pension changes (AR2), reflecting additional balances from redirection and zipper (AR2 > A2). - Individual accounts in the second pillar (A3), growing in line with market returns net of management fees. - Cash-flow effect: status quo fiscal flows equal pension contributions minus pension payments minus interest payments on pension-related public debt. Pension changes increase first-pillar contributions and reduce interest expenditure initially, while larger pension payments from the first pillar emerge over time as contributors retire. - Public debt implications: - Explicit government debt associated with the status quo is Dt; under the 2014 pension changes it is DR,t. - The transfer of public bonds from pension funds to the first pillar in 2014 causes a one-off drop in public debt of about 9 percent of GDP. - Public debt under the pension changes falls relative to status quo for several years, though long-term dynamics include growing capitalized pension payments from notional account A2. - Implicit liabilities: - Evolution of A1 is common to both policy alternatives and indexed at the nominal growth of the contribution base ωt. - Incremental impact on notional sub-accounts reflects indexation yt (average GDP growth of the previous five years). - Net effect on overall fiscal liabilities: - Net impact equals the net increase in explicit public debt plus the net increase in implicit public liabilities. - If the interest rate on public debt equals average GDP growth (the notional indexation), the net effect on overall public debt is zero. - If interest on public debt is higher than the indexation rate, the active policy increases public liabilities because savings on public debt service exceed notional account indexation. ### Private saving accelerations: predictability and implications for Poland - Cross-country findings: - Identified 86 episodes of rapid and sustained accelerations in domestic private saving ratios. - The unconditional probability that a country will experience a private saving acceleration sometime during a decade is around 25 percent. - Saving transitions are typically preceded by high and stable GDP per capita growth, low unemployment, and sustained strong fiscal positions. - Large natural resource discoveries are strong predictors of saving accelerations. - Implications for Poland: - Poland’s private saving-to-GDP ratio has been on a continuous declining trend; this raises concerns about financing investment and adequate retirement income given converging low pension replacement rates. - Factors limiting Poland’s likelihood of a private saving acceleration include low public savings and elevated unemployment. - Predicted propensity for Poland to experience a saving acceleration is relatively low at around 6 percent. - Policy implications: preserve macroeconomic stability gains, address elevated levels of unemployment, and further reduce the fiscal deficit to support sustained increases in private sector savings. *Source: REPUBLIC OF POLAND — INTERNATIONAL MONETARY FUND (excerpts)._ ### 8. The paper then asks whether saving accelerations are accompanied by increases in ### _cr14174 - 8. The paper then asks whether saving accelerations are accompanied by increases in ### Research question and main conclusion - Question: Are saving accelerations accompanied by increases in domestic investment and commensurate real GDP growth? - Main conclusion: Increases in private saving are fueled by higher growth realizations but are not necessarily a cause of stronger GDP growth. Episodes of sustained acceleration of private sector savings are mostly the result of stronger GDP growth performance and not necessarily their cause. ### Empirical design: identifying private saving transitions - Definition of a private saving transition (country undergoes a transition in year T if): - The three-year moving average of its saving rate over a nine-year period starting at T exceeds by more than 4 percentage points the five-year average of its private saving rate prior to T. - The private saving rate after the transition is higher than 10 percent of GDP. - Formal filter logic (parameters and horizon): - Threshold increase x = 4 percent of GDP. - Nine-year horizon starting at year 0 implies n = 6. - If conditions are satisfied for multiple consecutive years within a country and fewer than 10 years separate candidate dates, designate the earliest year as the transition year. - Practical notes on sample construction: - Sample begins in 1960; sample is unbalanced due to data availability. - For identification, drop first four and last two years of data per country; drop years t+2 ... t+10 of an episode to ensure minimum 10-year distance between accelerations. ### Sample outcomes and descriptive statistics - Number of identified saving acceleration episodes: 86. - Average private saving rate acceleration (before vs. after): 13.6 percentage points of GDP. - Median private saving rate acceleration: 10.3 percentage points of GDP. - Unconditional probability calculation: - Possible occasions in which an episode could have occurred: 3517. - Episodes / possible occasions = 86 / 3517 = 2.5 percent (probability of a growth transition taking place in any given year). - Interpreted: a typical country would have about a 25 percent chance of experiencing a growth transition at some point in any given decade. - Distributional patterns: - Saving accelerations are predominantly observed in emerging and developing countries. - Most accelerations concentrated in four regions: Sub-Saharan Africa; Latin America; MENA; Asia. - Of the 86 private saving transitions, 34 percent experienced only one transition, and 12 percent experienced two accelerations. - Typical saving-rate dynamics around the transition: - Typical jump in the private saving rate around year 0 is about 11 percentage points of GDP. - Median saving rate goes from 10 percent (pre-transition) to about 21 percent of GDP in the five to ten years following the transition. ### Econometric model and covariates - Dependent variable: dummy d = 1 for the 3 years centered on the first year of the private saving acceleration episode (t −1, t, t +1); 0 otherwise. - Control variables X are measured as averages over the 5 years before the saving acceleration window (to avoid reverse causality). - Groups of potential explanatory variables tested: - GDP growth and volatility (income growth volatility = standard deviation of annual real per capita GDP growth rate over past five years). Data: IMF WEO. - Unemployment rate. Data: IMF WEO. - Public sector saving rate. Data: IMF WEO. - Economic globalization: trade openness (exports + imports normalized by GDP) and financial openness (Chinn and Ito measure). - Financial development: M2-to-GDP (primary proxy) and bank deposits-to-GDP (tested). Data: World Bank Financial Development and Structure Database. - Natural resource discoveries: values of oil discoveries (Cotet and Tsui (2013)). ### Key econometric findings (baseline) - A good macroeconomic environment is a key factor behind private saving accelerations. - Main correlates (direction and interpretation): - Output growth volatility: negative effect on the probability of saving accelerations; persistent macroeconomic instability reduces the likelihood of sustained precautionary saving. - Public saving-to-GDP: positively correlated with the probability of private saving accelerations (contrary to simple Ricardian equivalence predictions). - Unemployment rate: higher and protracted unemployment reduces the likelihood of private saving accelerations. - Financial openness: negatively correlated with private saving accelerations (access to foreign capital reduces propensity for large sustained increases in domestic private savings). - Natural resource discoveries (oil discoveries per capita, log): positively associated with subsequent private saving accelerations. - Selected estimated marginal effects and standard errors reported in Table 2 (coefficients evaluated at the average of control variables; standard errors in parentheses): - Real per capita GDP growth: 0.00813*** (0.00214) - Per capita growth volatility, log: -0.00312 (0.00771) in column (1); -0.0177* (0.00957) in column (2); -0.0129* (0.00754) in column (3); -0.0192** (0.00752) in column (4); -0.00385 (0.00645) in column (5); -0.0244*** (0.00864) in column (6); -0.0256*** (0.00866) in column (7) - Public saving-to-GDP: 0.00589*** (0.000859) in column (1); 0.00631*** (0.00131) in column (2); 0.00676*** (0.000964) in column (3); 0.00682*** (0.00102) in column (4); 0.00459*** (0.000881) in column (5); 0.00582*** (0.00128) in column (6); 0.00596*** (0.00122) in column (7) - Trade openness: coefficients small and close to zero (examples: -6.92e-05 (0.000140); 0.000152 (0.000128)) - Financial openness: -0.00929** (0.00451) in column (1); -0.0154*** (0.00476) in column (2); other columns show mixed significance - M2-to-GDP: 3.25e-05 (0.000176) in column (1); -6.37e-05 (0.000176) in column (2); 0.000354* (0.000183) in column (3) - Bank deposits-to-GDP: 0.000163 (0.000196) in one specification; -0.000138 (0.000134) in another - Unemployment rate: -0.00233* (0.00134) in one specification; -0.00266** (0.00120) in another - Oil discoveries per capita, log: 0.217** (0.0989) in column (6); 0.213** (0.0965) in column (7) - Real per capita GDP, log: -0.0740* (0.0389) in one specification; -0.0989*** (0.0384); -0.103* (0.0596); and other magnitudes in additional columns - Model sample and fit notes (from Table 2 footnotes): - Year-specific effects included in many specifications. - Number of accelerations across columns: 51, 19, 50, 45, 18, 44, 44. - Countries included across columns: 126, 80, 125, 113, 78, 99, 99. - Observations across columns: 1,901; 821; 1,901; 1,733; 1,198; 1,618; 1,568. - Significance legend: *** p<0.01, ** p<0.05, * p<0.1. ### Robustness checks - Modified logit framework (King and Zeng (2001)) to correct for rare-occurrence bias yields broadly similar results to the probit specifications. - Robust findings across alternative specifications: - Saving accelerations tend to be preceded by high economic growth and low growth volatility. - Higher public savings and lower unemployment are associated with higher probability of private saving accelerations. - Large discoveries of natural resources are strongly correlated with subsequent private saving accelerations. *Source: IMF staff calculations and estimates (as presented in the supplied chapter excerpt).* ### 18. Using the sub-sample of EMs and low-income countries (LICs) only does not materially ### _cr14174 - 18. Using the sub-sample of EMs and low-income countries (LICs) only does not materially ### Robustness: Emerging Markets (EMs) and Low-Income Countries (LICs) sub-sample - Using the sub-sample of developing countries produces broadly similar results to the full sample; key exception: - Financial development variable becomes statistically (and positively) correlated with the probability of saving accelerations in the EMs and LICs sub-sample. - Interpretation offered: the marginal benefit of further financial deepening is stronger in the EMs and LICs sub-sample. - Estimation notes: - Simple regression coefficients reported instead of marginal effects. - Relogit is a logit model corrected for rare occurrence bias as suggested by King and Zeng (2001). - Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. ### Empirical correlates (selected coefficients reported exactly as in source) - From the EMs sample Relogit table (columns shown with coefficients and standard errors in parentheses): - Real per capita GDP growth: 0.0987*** (0.0292); 0.121*** (0.0307) - Per capita growth volatility, log: -0.00352 (0.114); -0.216 (0.201); -0.112 (0.223); -0.401*** (0.127); -0.356*** (0.124) - Public saving-to-GDP: 0.0761*** (0.0121); 0.130*** (0.0216); 0.0835*** (0.0160); 0.138*** (0.0220); 0.0937*** (0.0203); 0.0815*** (0.0200) - Financial openness: -0.120 (0.0838); -0.304*** (0.106); 0.0341 (0.0998); -0.274** (0.113); 0.0352 (0.106) - Bank deposits-to-GDP: -0.00164 (0.00332); -0.00396 (0.00364); 0.00511 (0.00330); -0.00225 (0.00429) - Unemployment rate: -0.0979** (0.0433); -0.0890* (0.0483) - Real per capita GDP, log: -1.552** (0.654); -3.327 (2.071); -0.552 (0.791); -0.657 (0.754) - (Real per capita GDP, log)2: 0.0748* (0.0437); 0.185 (0.121); 0.0115 (0.0509); 0.0169 (0.0477) - Oil discoveries per capita, log: 2.608* (1.556); 3.387** (1.547) - Sample sizes and counts (EMs table): - Number of accelerations: 46; 18; 45; 18; 42; 44 (corresponding to columns) - Countries: 122; 79; 121; 78; 97; 99 - Observations: 2,106; 1,200; 2,101; 1,198; 1,521; 1,618 ### Is Poland ready to experience an acceleration in its private saving ratio? - Main conclusion: - Poland faces challenges to experiencing a significant acceleration of private savings; key constraints are elevated unemployment and fiscal imbalances. - Diagnostic exercises: - Unconditional average performance of countries that experienced private saving accelerations compared with Poland (Table 5). - Propensity score estimates indicate Poland has around 6 percent chance of undergoing a private saving acceleration (Text Figure 2). - Poland’s propensity scores have increased recently but remain below critical thresholds due to elevated unemployment rate and fiscal imbalances. - Selected comparisons (Table 5: medians among accelerations post 1995; non-resource rich sample; Poland average over 2000–10): - Real GDP growth volatility: 3.0; 2.3; 1.7 - Unemployment rate: 9.3; 9.5; 12.2 - Real per capita GDP growth: 3.3; 2.03; 4.7 - Public saving-to-GDP: 2.4; 1.6; -0.5 - Financial development (M2-to-GDP): 33.0; 30.7; 48.6 - Financial development (bank deposits-to-GDP): 28.9; 28.0; 40.2 - Trade openness (Exports and Imports-to-GDP): 68.7; 59.5; 81.1 - Financial openness (KAOPEN index): -0.54; -0.6; 0.1 - Foreign direct investment-to-GDP: 2.1; 2.3; 4.4 ### Causality between saving accelerations and GDP growth (method and findings) - Research design (three steps): - Measure association between the acceleration dummy and average growth during the following nine years to reveal supplementary real per capita growth needed to fuel saving accelerations. - Test whether growth and private investment ratios are significantly higher approximately 10 years after an acceleration. - Use propensity score matching to reduce endogeneity: each acceleration country-observation matched to a counterfactual non-acceleration observation with similar predicted probability; compare macro outcomes. - Main findings: - Supplementary increase in GDP during the acceleration phase is around 1 percentage point per year. - Saving accelerations do not seem to significantly increase GDP growth afterwards, nor do they reduce it. - Direction of causality runs from higher and stable growth to private saving accelerations. - Saving accelerations are associated with higher private investment in the long run. ### Quantitative treatment effects (Table 6: average treatment effect on the treated, ATT; bootstrapped z-statistics in parentheses) - Change to growth during the acceleration episode: - Nearest-neighbor matching: 1.321*** (3.058) - 3 Nearest-neighbor matching: 0.995*** (2.783) - Radius matching: 1.118*** (3.814) - Kernel matching: 1.123*** (3.823) - Observations: 1,136 (for all four methods) - Change to growth post-acceleration episode: - Nearest-neighbor matching: 0.523 (1.054) - 3 Nearest-neighbor matching: 0.520 (1.174) - Radius matching: 0.464 (1.296) - Kernel matching: 0.457 (1.280) - Observations: 1,059 (for all four methods) - Change to private investment ratio post-acceleration episode: - Nearest-neighbor matching: 2.689** (2.414) - 3 Nearest-neighbor matching: 3.215*** (3.511) - Radius matching: 2.576*** (4.130) - Kernel matching: 2.592*** (4.138) - Observations: 810 (for all four methods) - Estimation notes: - Bootstrapped z-statistics based on 100 replications. - *, **, and *** indicate significance levels of 10 percent, 5 percent, and 1 percent, respectively. - Probit-selection equations include full set of determinants of saving transitions except the unemployment rate to maximize sample size. Estimations based on the common-support sample. ### Aggregate empirical findings - Frequency and distribution: - Of the 135 countries studied, 69 have had, at least, one private saving acceleration between 1960 and 2012. - Emerging and developing countries have experienced 78 percent of world’s overall saving accelerations. - Determinants associated with private saving accelerations: - Strong economic performance precedes surges in private saving: higher rates of GDP growth, lower GDP growth volatility, lower unemployment, and healthy public finances. - Financial development matters in the sub-sample of developing countries. - Strong fiscal positions are one of the most robust determinants of private saving transitions. - Large discoveries of natural resources lead to a surge in private saving rate. ### Policy implications and recommendations (for Poland) - High-level implications: - The direction of causality suggests medium-term policies that promote a sustainable macroeconomic environment and address structural issues (long-term unemployment, job quality) generate growth and protracted increases in private saving. - Saving accelerations contribute to capital accumulation through higher private investment in the long run. - Policy recommendations explicitly listed: - Advance structural reforms including labor market reforms to increase potential growth and reduce persistent high unemployment. - Continue with fiscal consolidation to reduce public debt. - Preserve banking and financial stability to safeguard confidence in the domestic financial sector as a means to increase formal and voluntary saving. - Rationale for Poland: - Efforts to promote voluntary private savings will help reinforce external sustainability and contribute to a decent retirement income for older members of society. - The most germane policies for Poland relate to the fiscal situation and the labor market; addressing persistent high unemployment and labor market duality and re-equilibrating public sector finances are the most important challenges. *Source: IMF staff estimates (extracted from _cr14174 - 18. Using the sub-sample of EMs and low-income countries (LICs) only does not materially).* --- _Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2014/_cr14174.pdf_