## _031612b ## Source details **Canonical URL:** [_031612b](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2012/_031612b.pdf) ## Other formats - [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2012/_031612b.pdf.md) - [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2012/_031612b.pdf.json) --- ### II. BUSINESS CYCLE SYNCHRONICITY — THEORETICAL AND EMPIRICAL OVERVIEW - Theoretical findings and context: - Effects of trade and financial integration on output volatility and co-movement are model-dependent and hinge on: extent and nature of integration; types of shocks; levels of economic and financial development. - Trade integration generates demand and supply-side spillovers that can increase business cycle co-movement. - Specialization effects depend on intra- vs. inter-industry specialization and dominant shock types (Krugman, 1993; Frankel and Rose, 1998). - Financial integration has ambiguous effects: may reduce correlations via capital reallocation (specialization) or increase synchronization via demand-side equity/wealth effects and contagion (Claessens and Forbes, 2001). - Consumption risk-sharing theory: complete markets should smooth consumption and increase consumption growth co-movement; magnitudes depend on model features and are expected to be greater for EMs and ODEs (Prasad et al., 2003). - Empirical summary: - No systematic relationship found between intensity of trade/financial linkages and output volatility across a variety of studies. - Most empirical work suggests trade integration has a positive impact on synchronization; role of financial integration is debated and mixed. - Studies report divergent temporal patterns: some find rising synchronization among advanced economies since 1960 (Kose, Otrok, and Whiteman, 2008; Bordo and Helbling, 2011), others find declining correlations in recent decades (Heathcote and Perri, 2004; Stock and Watson, 2003). - Advanced vs. emerging markets: evidence on decoupling is mixed and inconclusive across methodologies and samples. ### II.B — CONSUMPTION RISK-SHARING AND VOLATILITY EVIDENCE - Empirical categories: 1. International correlation patterns: consumption correlations are relatively low; consumption correlations lower than output; consumption correlated more with domestic output than world output (Backus, Kehoe, and Kydland, 1992). 2. Tests of perfect risk sharing via regression models: often reject perfect risk sharing (Cochrane, 1991; Mace, 1991; Lewis, 1996). 3. Regression measures of risk-sharing and role of financial flows: evidence of rising international risk sharing in late 1990s while home bias declined in advanced countries (Sørenson et al., 2007). - Degree and distribution: - Modest degree of international risk sharing overall; only advanced countries improved risk sharing in recent globalization; developing countries largely excluded (Kose, Prasad, and Terrones, 2009). - Composition matters: portfolio debt—dominant inflow to EMs—appears poorly conducive to risk sharing. - Volatility patterns: - Kose, Prasad, and Terrones (2003b): ratio of consumption growth volatility to income growth volatility increased during recent globalization for EMs—consumption volatility rose by more than income volatility. - Evidence ambiguous: Bekaert, Harvey, and Lundblad (2006) find declines in consumption volatility after equity liberalizations; Kalemli-Ozcan et al. (2010) find foreign ownership associated with higher firm- and aggregate-level volatility. ### METHODOLOGY: GROWTH CORRELATIONS AND MULTIDIMENSIONAL SCALING (1995–2010) - Synchronicity measure: - Dissimilarity = "one minus the correlation coefficient of two series of real GDP growth" over 1995–2010. - Distance matrix and dimensionality reduction: - N-by-N distance matrix constructed for largest 100 and 140 economies; insignificant correlations included. - Multidimensional scaling (MDS) projects the synchronicity matrix into an N-by-2 coordinate matrix; stress statistic used to assess goodness of fit. - Alternative robustness check: t-Distributed Stochastic Neighbor Embedding yields robust results. ### MDS RESULTS — CLUSTERS AND REGIONAL PATTERNS (1995–2010) - Advanced economies (AE): - Core AE: EU, U.S., U.K., Canada form a tight cluster; Ireland and Portugal close to core; Greece further away from euro neighbors. - Interpretation: persistent asynchronous cycles among some euro members despite common exchange rate and monetary policy. - East Asia: - Supply-chain countries (Hong Kong SAR, Philippines, Malaysia, Taiwan Province of China, Korea, Singapore, Vietnam, Japan) sit between China and the AE core; not a tight cluster like AE core. - Implication: multiple potential sources of growth shocks; convergence among Asian EMs less pronounced than among advanced economies. - Latin America: - Most large Latin American EMs (except Mexico) located closer to the Asian EM cluster than to the AE core; possible role of commodity trade with China. - Commodity exporters / oil producers: - Majority scattered around AE and Asian/Latin American clusters despite similar terms-of-trade shocks and common pegs to the U.S. dollar. - Many oil producers situated closer to China and India than AE core. - Low income countries (LICs): - No observable clustering; idiosyncratic shocks (political cycles, weather) important drivers. ### APPLICATIONS TO CURRENCY AREAS AND UNIONS - Euro Area: - Post-euro: synchronization around a core (Germany, France, Italy, northern European countries); peripheral countries (Greece, Ireland, Portugal, Spain) remain outside EA core in business cycle synchronicity and unit labor cost convergence. - Structural fissures in unit labor costs and current account balances indicate deeper divergence. - West and Central African Monetary Union: - West African Monetary Union slightly more homogeneous; Central African Monetary Union split into two relatively distant sub-clusters—implies divergent outcomes under a common exchange rate policy. - Eastern Caribbean Union: - Correlations between members increased considerably in last 5 years of sample (presumably due to crisis); earlier maps show two sub-clusters. ### POLICY IMPLICATIONS — MDS-BASED SURVEILLANCE USES - Identify countries persistently asynchronous with peers despite common exchange rate/monetary policies (warning signals). - Detect gateways and clusters that may propagate shocks; consider multiple sources of growth shocks. - For currency unions/fixed-exchange-rate areas: examine structural fissures (unit labor costs, current account dynamics), sub-cluster splits, and anticipate divergent outcomes. - For commodity exporters: investigate why similar external shocks and common pegs do not produce similar cycle dynamics; monitor links to major demand centers (China, India). - For LICs: focus surveillance on idiosyncratic vulnerabilities (political cycles, weather) rather than expecting strong synchronicity. --- ### 6. GLOBAL TRADE AND FINANCIAL FLOWS NETWORK — A DENSE CENTER AND CLUSTER ANALYSIS - Network architecture: - High central density with edges concentrating at the center and petering out toward the periphery; some peripheral-peripheral direct links exist. - Core countries account for a large share of global trade and financial ties; core conditions materially impact the rest of the network. - Centrality measures and core composition: - Measures: degree centrality, closeness centrality, random walk betweenness centrality, eigenvector centrality. - Core = economies with centrality > 90th percentile for each measure: 22 economies in core. - Three core tiers (Table 1 composition preserved): - Tier 1: China, France, Germany, Italy, Japan, United Kingdom, United States, Thailand - Tier 2: Austria, Belgium, Canada, Korea, Netherlands, Spain, Switzerland - Tier 3: Australia, Denmark, India, Mexico, South Africa, Sweden, Taiwan, Province of China - Observations: - First two tiers dominated by Euro Area economies plus Canada, China, Japan, Korea, Switzerland, U.K., U.S. - Final tier largely EMs with Australia, Denmark, Sweden. - Surprises: absent G20 members in core include Argentina, Brazil, Indonesia, Saudi Arabia, Russia, Turkey; present smaller economies include Denmark, Sweden, Switzerland, Taiwan Province of China, Thailand. ### FROM GRAPHS TO OVERLAPPING CLUSTERS — CLIQUE PERCOLATION METHOD (CPM) - CPM advantages: - Identifies an unknown number of potentially overlapping clusters of varied size; allows multi-cluster membership (captures gatekeepers). - CPM empirical choices and results: - Best fit k range estimated between 5 and 8; percolation threshold around 31 percent; expected number of neighbors ≈ 10; weight pruning threshold ≈ 15 percent (bilateral relations where wij < 15 eliminated). - CPM yields 11 distinct clusters of varying sizes. - Graph separable into two core regions: one around the United States and China, another around Italy and Germany; some countries (France, Netherlands, U.K.) lie between both cores. - For k between 5 and 8: - Single cluster at 8-clique level: Australia, China, Indonesia, Japan, Malaysia, Singapore, Thailand, United States. - Three clusters at 7-clique level, two overlapping and one separated. - Overlap patterns: - Maximal overlap at 6-clique level, then 5-clique level. - 6-clique: 12 overlapping clusters with 72 countries. - 5-clique: 8 overlapping clusters with 133 countries; cluster 1 (5-clique) = giant cluster with 117 countries. ### GATEKEEPERS, GATEKEEPER CLUSTERS, AND IMPLICATIONS - Gatekeeper identification: - Gatekeepers link clusters and transmit/amplify/dampen shocks; can be individual countries or groups acting jointly. - Combined 7- and 8-clique analysis highlights China, Japan, Korea, United States as gatekeepers linking parts of Middle East and Pakistan to Asia. - Hong Kong SAR: nuanced gatekeeper role tied to Asian economies and the United States as an equity financing hub. - Gatekeeper cluster ranking: - Metric measures weighted contribution of members to cluster overlaps normalized for cluster size. - Cluster #11 under 6-clique confirmed as a top gatekeeper cluster; clusters #7 and #4 rank highly as well. - Policy implications: - Gatekeeper clusters map potential shock transmission paths and should inform surveillance of interconnectedness across trade and financial flows. - Identifying gatekeepers aids in anticipating which countries or clusters can propagate shocks and where monitoring and coordination may be most valuable. --- ### 22. CENTRALITY, GATEKEEPERS, AND SHOCK PROPAGATION — CLUSTER-LEVEL FINDINGS - Structural insights: - Majority membership in two of three clusters comprises very large economies. - U.S., Netherlands, China, Germany lie at the intersection of the three gatekeeper clusters. - Core economies act as gatekeepers through different core regions, creating overlap cascades from core to periphery. - Transmission findings: - Shocks typically pass through multiple "gates" to spread system-wide; potential for negative feedback within and between clusters substantial. - Strong transmission channels from Germany and Italy to Europe, CEESE, CIS, and parts of Middle East. - United States is a key conduit for spillovers elsewhere; U.S. policy responses can be particularly effective in stabilizing global impact from European stress. - Italy identified as a gatekeeper for core Europe and a conduit to Belgium, France, Germany, Netherlands, CEESE, CIS, emerging Middle East, and the United States. - U.K. acts as a shock absorber from Europe and a conduit for dollar-funded shocks to the U.S.; China’s density is more within Asia and is less receptive to incoming shocks but can produce material effects on its clusters if severely stressed. - Mechanical shock experiment (European shock propagation): - Uniform reduction of all link values by 15 percentage points used as a proxy for a global GDP shock; results map incoming share of envisioned shock and share of effects transmitted to others for each country. - Caveat: - Analysis maps potential routes and feedback loops mechanically and is not a quantitative gauge of relative intensity without further assessment of policy space and actions. --- ### VI. FINANCIAL INTERCONNECTIONS AND CROSS-BORDER POLICY DEPENDENCE — SUPERVISION, CLUSTERS, AND ARBITRAGE - Clusters from cross-border banking: - Cross-border bank operations create home-host clusters that can constrain domestic macro-financial policy via externalities and spillovers (examples: Nordic-Baltic, Austria-CEE, Singapore–South/South-East/East-Asia). - Parent funding can be material for host credit intermediation; shifts in parent funding risk host credit and solvency. - Centralized business models (centralized funding, treasury, risk management) reinforce clustering and intra-group exposures. - Policy space effects: - Parent funding can "clog the credit channel" of host monetary policy; alternative instruments may be needed (partial capital controls, tax treatment of inflows, subsidiarization limits). - Coordination and information sharing across supervisors are essential; absence raises vulnerability to contagion. - Supervisory coordination and crisis management: - Good supervision requires timely information, adequate resourcing, enforcement powers, and legal protection. - Cross-border resolution complexity: national frameworks can conflict; concerted coordination and equitable burden-sharing improve outcomes. - Regulatory and tax arbitrage: - Firms exploit policy differences to seek regulatory and tax advantages (example: Depfa relocation to Ireland; Depfa–HRE crisis and eventual SoFFin rescue > €100 billion). - Arbitrage can produce clusters whose continuation may be questioned post-crisis; eliminating arbitrage opportunities can reduce cluster-induced systemic risks. - Data gaps and surveillance constraints: - Significant information gaps exist: geographic distribution of banking sector holdings and portfolio assets often not publicly available; BIS data often net not gross; EPFR/CPIS/IIP limitations. - Data gaps hinder mapping exposures, early warning, and comprehensive Fund surveillance. --- ### CASE STUDIES AND REGIONAL APPLICATIONS - Saudi Arabia: monetary, commodity, and credit dynamics (1980–2010) - Backdrop: - U.S. Federal Reserve policy rate lowered from 5.25 percent to 4.75 percent on September 18, 2007; federal funds target rate reduced to 1 percent by end-October 2008. - Crude oil rose from over $65 per barrel in mid-2007 to over $130 per barrel by summer 2008. - Saudi Arabian Monetary Agency (SAMA) cut policy rate from 5 percent in October 2007 to 2 percent by mid-2008 to maintain the U.S. dollar peg. - Annual credit growth in Saudi Arabia rose from 6 percent in early 2007 to over 30 percent in July 2008; double digit inflation by mid-2008. - Main results: - Business cycle correlation with U.S. shifted: 1980s negative (supply-driven oil shocks), 2000s positive (demand-driven oil shocks from developing Asia). - Pass-through from oil prices to fiscal spending fell over past three decades; non-oil output volatility declined in 2000s. - Credit dynamics increasingly relevant for non-oil activity; importance of U.S. interest rate policy to Saudi non-oil sector likely rose in 2000s. - Regression (Saudi non-oil GDP on U.S. GDP, Developing Asia GDP, Oil Price; HP-filtered; samples 1980–2010, 1980–1995, 1996–2010) — coefficients and standard errors preserved exactly: - U.S. GDP: -0.66* (Std. error (0.24)) for 1980-2010; -1.02* (Std. error (0.48)) for 1980-1995; 0.20* (Std. error (0.10)) for 1996-2010. - Developing Asia GDP: -0.05 (Std. error (0.21)) for 1980-2010; 0.43 (Std. error (0.48)) for 1980-1995; 0.21* (Std. error (0.06)) for 1996-2010. - Oil Price: 0.18* (Std. error (0.04)) for 1980-2010; 0.25* (Std. error (0.07)) for 1980-1995; -0.01 (Std. error (0.02)) for 1996-2010. - R-squared (adj): 0.49 for 1980-2010; 0.69 for 1980-1995; 0.40 for 1996-2010. - Fiscal volatility statistics (exact values from Table 2): - Oil revenue growth volatility (std): 1980s 44.4 ; 1990s 21.1 ; 2000s 38.5 - Spending growth volatility (std): 1980s 19.8 ; 1990s 16.2 ; 2000s 6.4 - Correlation between oil revenue and spending growth: 1980s 0.9 ; 1990s 0.7 ; 2000s 0.2 - Non-oil growth volatility (std): 1980s 4.6 ; 1990s 1.6 ; 2000s 0.7 - Credit and monetary transmission: - Granger causality: full sample 1980–2010 — no Granger causality between real credit and non-oil GDP; subsample 1996–2010 — reject no Granger causality (greater role of credit). - VAR variance decomposition (forecast error variance of non-oil GDP explained by real credit after 5 years): 1996–2010: 58 percent; 1980–1995: 40 percent. - Monthly VAR (1997:1–2008:9; subsamples 1997:1–2003:12 and 2003:9–2008:9): selected coefficients for 2003:9–2008:9 preserved exactly: - 3-Month LIBOR on Credit: -4.01* (1.87) - 3-Month LIBOR on CPI: 0.54 (0.32) - Oil price on Credit: 0.14* (0.03) - Oil price on CPI: -0.01* (0.01) - International food price on Credit: -0.06 (0.09) - International food price on CPI: 0.04* (0.02) - NEER on Credit: 0.45* (0.22) - NEER on CPI: -0.04 (0.04) - R-squared (adj) for 2003:9–2008:9: Credit 0.27, CPI 0.41 - Interpretation: interest rate channel (proxied by LIBOR) became more effective in transmitting to credit in 2003:9–2008:9; oil demand from EMEs and financial deepening raise exposure. - Policy recommendations: - Strong fiscal management to stabilize non-oil economy. - Further refine macroprudential instruments to influence monetary conditions independent of interest rate policy. - Monitor tensions from simultaneous links to developing Asia and commitment to U.S. dollar peg. - Asian cluster and China centrality (output synchronicity, machinery and electronics) - Vertical production linkages (intermediate inputs) are main channel for trade-driven business cycle synchronicity. - Empirical patterns: - High correlations in industrial production across emerging Asia with respect to China, especially since mid-2005. - Policy rates more persistently correlated across cluster than output in some periods. - Rolling correlations with China rose and peaked during global crisis; divergence during 2004–2007/8 "peace" time. - Key numerical facts: - China exchange rate reform (July 2005): immediate 2.1 percent revaluation; daily fluctuation within +/- 0.3 percent band. - Cumulative appreciation of renminbi against U.S. dollar by end-2008: 20¼ percent. - China growth 2003–2008: grew at more than 10 percent per year. - Malaysia growth 2004–2007 average: 6 percent per year. - Malaysia central government deficit: 4.1 percent of GDP in 2003 to 3.2 percent in 2007. - Philippines peso appreciation 2005–2007: cumulative 25 percent; real effective appreciation 20 percent. - Policy implications: - Preservation of relative currency stability vis-à-vis China can be important for downstream/processing-trade economies. - Prudential measures become additional policy instruments when exchange rate instrument is constrained. - Scope exists for cluster-level policy dialogue and coordination. - Baltic–Nordic example (benchmarking, bank funding, and gatekeeper role) - Pre-crisis dynamics: - Baltics ran large current account deficits funded by Nordic bank consumer loans; convergence expectations raised consumption and credit. - Swedish banks acted as gatekeepers, funding Baltics (low wholesale funding rates, high Baltic growth). - Crisis dynamics: - Withdrawal of funding by Swedish banks amplified shock; collective bank pullback triggered destabilizing spiral, forcing crises in Latvia and severely impacting Estonia and Lithuania. - Swedish policy responses (liquidity, guarantees, support measures) dampened contagion; actions included expanded collateral policy, new loan facilities, swap lines ($30 billion), deposit guarantee increases, debt guarantee scheme up to SEK 1.5 trillion (later reduced to SEK 750 billion), recapitalization schemes, stabilization fund with SEK 15 billion initial contribution. - Gatekeeper VAR experiment: - VAR with VIX, G20 GDP-weighted growth, gatekeeper money market rate and GDP, destination money market rate and GDP (gatekeeper = Sweden; destinations = Estonia, Latvia, Lithuania). - Swedish monetary policy found to attenuate spillovers to Baltic money market rates and real GDP growth; conditional responses indicate material impact of gatekeeper policy choices. - Policy implications: - Gatekeeper policy space matters for cluster outcomes; coordinated measures can reduce amplification. - Defensive measures and ring-fencing may reduce cluster inclusion but protect domestic stability. - Harmonization of prudential standards benefits hosts and reduces externalities. --- ### 23. DATA GAPS, TRADE AND FINANCIAL CLUSTERS — SURVEILLANCE CONSIDERATIONS - Data and timing limitations: - Long lags exist between suspicion of problems and provision of data tracking interconnections; Data Gaps initiative aims to bridge some gaps. - By the time data are requested and received, it may be too late; larger policy space is essential given these limits. - Trade cluster surveillance considerations: - Export sector membership informs vulnerability to global factors (terms of trade); position in supply chains (upstream vs downstream) matters for reliance on imported inputs and vulnerability to upstream shocks. - Competition and similar export structures reduce policy instrument flexibility; trade credit links financial and trade clusters and transmits shocks. - Financial cluster surveillance considerations: - Funding models and parent balance sheet strength determine resilience; strategic importance and durability of financial centers matter for cluster persistence. - Integration degree affects transmission channels (FX swap market, interbank, maturity shortening). - Identifying funding and transmission channels supports targeted analysis. - Data limitations hampering surveillance: - Geographic distribution of a country’s banking sector holdings and fund managers’ portfolio assets often not publicly available. - BIS statistics often provide net not gross flows; EPFR and CPIS/IIP limitations complicate constructing relative importance metrics for regional centers. - These gaps impede mapping clusters and assessing systemic spillovers needed for Fund surveillance. *Source: ENHANCING SURVEILLANCE: INTERCONNECTEDNESS AND CLUSTERS (excerpted content unit _031612b).* ### References _______________________________________________________________________________________ 11 ### _031612b - References ### II. BUSINESS CYCLE SYNCHRONICITY — A MULTI DIMENSIONAL SCALING APPROACH: INTRODUCTION & OVERVIEW - The note provides an overview of theoretical and empirical literature on implications of greater trade and financial integration for volatility and synchronization of business cycles. - Theoretical predictions vary depending on: - extent and nature of integration (e.g., inter- vs. intra-industry trade), - types of shocks (e.g., industry-level vs. common shocks), - levels of economic and financial development (e.g., ability to diversify risk). - Empirical research reports: - no systematic relationship between intensity of trade and financial linkages and output volatility, - most empirical studies suggest trade integration has a positive impact on synchronization, - the role of financial integration in synchronization remains debated. ### II.A THEORETICAL IMPLICATIONS — VOLATILITY, SYNCHRONIZATION, AND RISK-SHARING - Varied effects: - Theoretical models yield different implications for how trade and financial linkages affect output volatility and co-movement (Hirata, Kose, and Otrok, 2012). - Context matters: - Effects depend on level of development, nature of shocks, and pattern of specialization. - Trade integration: - Generates demand and supply-side spillovers across countries, increasing business cycle co-movement. - Demand-side example: investment or consumption boom in one country increases demand for imports, boosting foreign economies. - Supply-side example: positive shock to tradable goods output lowers prices and reduces costs of imported inputs for other countries. - Impact of specialization: - Increased trade leads to increased specialization; impact on co-movement depends on intra- vs. inter-industry specialization and shock types. - If inter-industry specialization rises and industry-specific shocks dominate, co-movement may decrease (Krugman, 1993). - If common shocks dominate, co-movement may increase (Frankel and Rose, 1998). - Financial integration: - Ambiguous effects on cross-country output correlations. - Can reduce correlations by enabling capital reallocation consistent with comparative advantage (promoting specialization). - Can increase synchronization via demand-side effects (changes in equity prices affecting wealth dynamics) and contagion transmitted through financial linkages (Claessens and Forbes, 2001). - Diversification and risk sharing: - Financial integration allows capital-poor EMs and ODEs to diversify away from narrow production bases, potentially reducing output volatility. - At advanced stages, trade and financial flows may enhance specialization and increase vulnerability to industry-specific shocks (Imbs and Wacziarg, 2003; Kalemli-Ozcan, Sørensen, and Yosha, 2003). - Consumption risk-sharing: - Basic theory predicts financial integration should enable international insurance against income risk, smoothing consumption and increasing co-movement of consumption growth across countries. - With complete markets, trade in financial assets can delink national consumption from country-specific output fluctuations, making consumption growth less volatile relative to income growth. - Magnitudes of gains depend on model-specific features: volatility of output shocks, rate of relative risk aversion, risk-adjusted growth rate, and risk free interest rate. - Benefits expected to be greater for EMs and ODEs due to higher intrinsic volatility (Prasad et al., 2003). ### II.B EMPIRICAL EVIDENCE — BUSINESS CYCLE SYNCHRONIZATION - Impact of integration: - Empirical studies generally unable to provide a concrete explanation for how stronger global linkages affect co-movement of business cycles. - Cross-country or cross-region panel regressions used to assess role of global linkages on co-movement in advanced countries (Kose and Yi, 2006). - Imbs (2004): extent of financial linkages, sectoral similarity, and volume of intra-industry trade positively impact business cycle correlations. - Baxter and Kouparitsas (2005): international trade is the most important transmission channel for business cycle fluctuations. - Kose, Prasad, and Terrones (2003b): both trade and financial linkages have a positive impact on cross-country output and consumption correlations. - Diverse conclusions: - No definite answer on how co-movement varies with changes in trade and financial flows. - Findings on temporal evolution of synchronization with financial integration are mixed; differences arise from country coverage, sample periods, aggregation methods, and econometric approaches. - Some studies find declining output correlations among advanced economies over last three decades (Heathcote and Perri, 2004; Stock and Watson, 2003). - Evidence of strengthening links: - Other studies document stronger business cycle linkages over time. - Kose, Otrok, and Whiteman (2008): for advanced countries, business cycle correlations on average have risen since 1960 (Bayesian dynamic factor model). - Bordo and Helbling (2011): using annual data 1880–2008, synchronization among advanced countries increased over time. - European evidence mixed: Artis, Krolzig, and Toro (2004) find a European business cycle; Canova, Ciccarelli, and Ortega (2007) argue no evidence of a specific European cycle since the 1990s (Hirata, Kose, and Otrok, 2011). - Advanced vs. emerging markets (AMs and EMs): - Debate on temporal evolution of synchronization for AMs and EMs, focused on ability of EMs (especially Asia-Pacific) to decouple from U.S. slowdowns (Helbling et al., 2007; Kose and Prasad, 2010). - Many methodologies applied; results on decoupling are inconclusive. *Source: ENHANCING SURVEILLANCE: INTERCONNECTEDNESS AND CLUSTERS (excerpt from the References and Section I overview).* ### 13. The case against decoupling. Some studies employ simple correlations to make a case ### 13. The case against decoupling. Some studies employ simple correlations to make a case ### The case against decoupling - Flood and Rose (2009) use GDP data for 64 countries over the period 1974–2007. - After de-trending these series using various filters, they analyze rolling-window correlations across advanced countries and developing economies and conclude that, while the average level of cross-correlations changes over time, there is no strong evidence that these correlations have become statistically significantly lower in the later parts of their sample. - Walti (2009) reports that the extent of co-movement of cycles across advanced countries and EMs has not changed much since the early 1980s. ### The case for decoupling - Kose, Otrok, and Prasad (2012) examine sources of macroeconomic fluctuations in advanced and emerging market countries using dynamic factor models and series of output, consumption, and investment for the 1960–2008 period. - Their findings indicate substantial convergence of business cycles among advanced economies and among emerging markets over time, and a concomitant divergence of business cycles between these two groups. - Mumtaz, Simonelli, and Surico (2011) employ a dynamic factor model and report findings confirming decoupling of business cycles using data for a group of 36 countries over a 75-year period. - Additional studies supporting the decoupling hypothesis include Dooley and Hutchison (2009); Rossi (2009); Fidrmuc and Korhonen (2010). ### Volatility of output and consumption: empirical evidence - No systematic empirical relationship: - Existing evidence, using a variety of regression models with different country samples and time periods, reports no systematic empirical relationship between the intensity of trade and financial linkages and output volatility. - Kose, Prasad, and Terrones (2003b) report that the ratio of consumption growth volatility to income growth volatility actually increased during the recent period of globalization for emerging market economies. - They find that the volatility of consumption rose (perhaps because of crises experienced by some of these economies) by more than income volatility did. - This result runs counter to the theoretical prediction that financial integration allows countries to share income risk and smooth consumption. - They also report that increasing financial integration is associated with rising relative volatility of consumption, albeit only up to a threshold; beyond a certain level of financial integration, an increase in integration actually reduces the relative volatility of consumption. - Ambiguous evidence: - Bekaert, Harvey, and Lundblad (2006) find that, following equity market liberalizations, there is an outright decline in consumption volatility. - Kalemli-Ozcan, Sorenson, and Volosovych (2010) report a positive association between foreign ownership and the volatility of a firm’s various outcomes, a result that extends to aggregate data as well. - Differences across studies could arise from variations in definitions of financial integration, measures of consumption volatility, data samples, and methodologies; overall evidence is ambiguous. ### Emerging markets (EMs) and procyclicality of capital flows - Observed phenomenon: - Positive productivity and output growth shocks in EMs led to consumption booms financed by international investors. - Domestic financial liberalization at the time of opening to international financial flows loosened financing constraints, accentuating consumption booms. - When negative shocks hit, EMs rapidly lost access to international financial markets, depressing consumption. - Procyclicality explanation: - A growing literature suggests the procyclical nature of capital flows explains the adverse impact of international financial integration on consumption volatility in EMs. - One manifestation is “sudden stops” of capital inflows (Calvo and Reinhart, 1999). ### Consumption and income risk-sharing: categories of empirical literature - Three categories: 1. Studies focusing on patterns of international correlations to determine degree of consumption risk sharing (Kose, Prasad, and Terrones, 2003a, 2009; Ambler, Cardia, and Zimmermann, 2004). - Findings indicate theoretical predictions of perfect risk sharing lack empirical support for three reasons (Backus, Kehoe, and Kydland, 1992): - Correlations between consumption paths of various countries are relatively low. - Consumption correlations are lower than those of output. - Correlations between consumption and domestic output are generally higher than those between consumption and world output. 2. Studies testing perfect risk sharing via regression models using reduced form solutions or first order conditions (Cochrane, 1991; Mace, 1991). - Lewis (1996) finds the hypothesis of risk sharing is rejected for countries with few or limited capital controls; correlations between domestic consumption and output are higher for countries with more restrictions, suggesting less risk sharing. 3. Studies employing regression models to measure extent of risk-sharing and impact of financial flows on risk sharing. - Sørenson, Yosha, Wu, and Zhu (2007) analyze relationship between home bias and international risk sharing, documenting that international risk sharing rose during the late 1990s while home bias in debt and equity holdings declined in advanced countries. ### Degree and distribution of international risk sharing - Modest degree of international risk sharing: - Kose, Prasad, and Terrones (2009), using a variety of empirical techniques, conclude there is at best a modest degree of international risk sharing, and certainly nowhere near theoretical levels. - Only advanced countries have attained better risk sharing outcomes during the recent period of globalization; developing countries have been largely shut out. - Even EMs, despite large increases in cross-border capital flows, have seen little change in their ability to share risk (Giannone and Reichlin, 2006; Moser, Pointner, and Scharler, 2004). - Composition of flows matters: portfolio debt—the dominant form of capital inflows to EMs—does not seem conducive to risk sharing. ### Overall assessment and open questions - Inconclusive findings: - Theoretical studies and empirical evidence are inconclusive regarding implications of financial integration for business cycle volatility and co-movement. - Financial integration does not always reduce amplitude of business and financial cycles and may increase it. - Risk-sharing benefits of integration are apparent in theory but hard to show conclusively in empirical work. - Possible reason for difficulty in obtaining sharp results: - Changes in the nature of shocks as cross-border linkages become more intensive. ### Changing nature of shocks and linkages - Integration can affect co-movement by changing nature and frequency of shocks: - Stronger trade and financial linkages may necessitate higher policy coordination, raising correlations between shocks associated with nation-specific fiscal and/or monetary policies, potentially increasing business cycle synchronization (Darvas, Rose, and Szapary, 2005; Flood and Rose, 2009). - Productivity shocks could become more correlated if rising trade and financial integration leads to increased knowledge and productivity spillovers across countries (Kose, Prasad, and Terrones, 2009). - Increased financial integration and developments in communication technologies lead to faster dissemination of news shocks in financial markets, which could increase business cycle synchronization via wealth and consumption channels. *Source: _031612b - 13. The case against decoupling. Some studies employ simple correlations to make a case* ### 3. Growth correlations. We use the correlation of annual GDP growth rates as a measure of ### _031612b - 3. Growth correlations. We use the correlation of annual GDP growth rates as a measure of ### Methodology: measuring business cycle synchronicity - Synchronicity measure: dissimilarity defined as "one minus the correlation coefficient of two series of real GDP growth" over the sample period 1995–2010. - Distance matrix construction: - An N by N distance matrix is constructed where N is the number of countries in the analysis. - Analysis focuses on two country groups: the largest 100 and 140 economies. - Insignificant correlations are included in the distance matrix; the standard error of a correlation coefficient is a function of the number of observations and the magnitude of the coefficient. - Dimensionality reduction: - Multidimensional scaling (MDS) projects the N-by-N synchronicity matrix into an N-by-2 coordinate matrix. - The algorithm regresses actual dissimilarities on distances in the two-dimensional map and minimizes the sum of squared differences (stress). - Two-dimensional maps chosen for visual interpretability; goodness of fit characterized by the stress number. - Alternative robustness check: t-Distributed Stochastic Neighbor Embedding yields robust results. ### Results: clusters and regional patterns (1995–2010) - Advanced economies (AE): - The core AE, EU countries, the U.S., the U.K., and Canada form a tight cluster. - Ireland and Portugal are situated close to the advanced economy core; Greece is further away from its euro area neighbors than some non-euro area countries. - Interpretation: persistent asynchronous business cycles among certain euro area members despite common exchange rate and monetary policy. - East Asia: - East Asian supply chain countries (Hong Kong SAR, Philippines, Malaysia, Taiwan, Province of China, Korea, Singapore, Vietnam, and Japan) are situated between China and the AE core. - These East Asian countries do not form a tight cluster like the AE core. - Interpretation: surveillance should account for multiple countries as potential sources of growth shocks; convergence among Asian emerging markets is not nearly as significant as among advanced economies. - Latin American emerging markets (EMs): - Most large Latin American EMs (except Mexico) are situated closer to the Asian EM cluster than to the AE core. - Possible explanation: increased importance of commodity trade with China. - Stylized fact: AE business cycle synchronization has been close; EMs’ cycles have been somewhat less synchronized than AEs, although more synchronized with each other. - Commodity exporters / oil producers: - Some weak clustering around Saudi Arabia, but majority of oil producers are scattered around the AE and Asian/Latin American clusters. - Observations: - Despite similar terms of trade shocks and common exchange rate policies pegged to the U.S. dollar, oil producers’ business cycles are not that similar, indicating tension between policy and fundamentals. - Most oil producers are situated closer to the large Asian EMs (China and India) than the AE core, suggesting increasing importance of Asian demand. - Low income countries (LICs): - No observable clustering among LICs. - Interpretation: idiosyncratic shocks (political cycle, weather patterns, etc.) are important drivers of LICs’ business cycles. ### Applications to common currency areas and unions - Euro Area (EA): - Since euro adoption, business cycles within the EA have become largely synchronized around a core comprising Germany, France, Italy, and some northern European countries. - Peripheral countries such as Greece, Ireland, Portugal, and Spain remain located well outside that EA core in terms of business cycle synchronicity and unit labor cost convergence. - Recent EU entrants have moved closer to the Euro Area, reflecting increased trading ties. - Structural fissures through the EA core: - Unit labor costs: France and Italy are far removed from Germany on some measures. - Current account balances: Germany groups with Austria and the Netherlands; Belgium, Italy, and France join Greece, Ireland, Portugal, and Spain to form another group. - Interpretation: split in current account dynamics reflects divergence in national saving-investment balances and deeper structural differences among core EA countries. - West and Central African Monetary Union: - Synchronicity map centered on the World real GDP series to show relationships to the world business cycle. - West African Monetary Union is slightly more homogeneous. - Central African Monetary Union splits into two sub-clusters that are relatively distant from each other. - Implication: divergent macroeconomic outcomes may arise against a backdrop of a common exchange rate policy; merits further surveillance exploration. - Eastern Caribbean Union: - Correlations between member countries increased considerably in the last 5 years (relative to the sample), presumably due to the economic crisis. - Earlier maps show two sub-clusters rather than one. - Surveillance should consider roles of natural disasters, fiscal space, and differences in income. ### Policy implications and surveillance recommendations - Use MDS-based synchronicity maps as a surveillance tool to identify: - Countries whose business cycles are persistently asynchronous with peers despite common exchange rate or monetary policies (warning signals). - Gateways and clusters that may propagate shocks across regions (importance of considering multiple sources of growth shocks). - For currency unions and fixed-exchange-rate areas: - Examine structural fissures (e.g., unit labor costs, current account dynamics) that may indicate deeper divergence beneath convergent macroeconomic policies. - Explore sub-cluster splits within unions (e.g., Central African Monetary Union) to anticipate divergent macroeconomic outcomes under common policies. - For commodity exporters: - Investigate why similar external shocks and common peg policies do not translate into similar business cycle dynamics. - Monitor links to major demand centers (e.g., China and India) given positioning closer to Asian EM clusters. - For LICs: - Focus surveillance on idiosyncratic shock vulnerabilities (political cycles, weather) rather than expecting strong cross-country synchronicity. *Prepared by Philip Liu and Sergejs Saksonovs (SPR); figures and maps based on MDS of GDP growth correlations (1995–2010) and staff calculations.* ### 6. A dense center. As expected, the network representing trade and financial flows displays a ### _031612b - 6. A dense center. As expected, the network representing trade and financial flows displays a ### A dense center - The network representing trade and financial flows displays a high density of interactions. - The distribution of edges between vertices is not uniform; there is a very dense set of relationships at the center of the network that peter-out toward the periphery. - Some peripheral vertices are linked directly to other peripheral vertices. - Figure 1 illustrates the Global Trade and Financial Flows Graph (not reproduced here). ### Architecture and function - Core countries (red vertices in Figure 1) account for a large part of global trade and financial relationships. - Economic and financial conditions in the core are likely to have material impact on the remainder of the network. - Variation in density and differing functional roles of vertices imply complexity that requires systematic deconstruction of the graph into building blocks (vertices and their inter-relations) for further understanding. - The text explores techniques and metrics to unpack this complexity. ### Quantifying and Rank-Ordering Centrality - Purpose: identify which vertices belong to the core versus the periphery and rank-order their importance. - Centrality measures used: - Degree centrality: quantifies the number of connections any given vertex has to all others. Degree can be computed as incoming links (in degree), outgoing links (out degree), or all links (degree). - Closeness centrality: measures the mean of the shortest path in terms of the number of paths between one vertex and all others. - Random walk betweenness centrality: approximated by the expected number of times that a random walk between any starting and ending vertex will pass through an intermediate set of vertices averaged over all starting and ending vertices. - Eigenvector centrality: defines both the number and the quality of the connections any given vertex has within the network. - Most central economies: - Economies with centrality higher than the 90th percentile for each measure were identified. - The core of the trade and financial flows graph consists of 22 economies. - Three distinct tiers appear at the core. Tier composition (Table 1): - Tier 1: China, France, Germany, Italy, Japan, United Kingdom, United States, Thailand - Tier 2: Austria, Belgium, Canada, Korea, Netherlands, Spain, Switzerland - Tier 3: Australia, Denmark, India, Mexico, South Africa, Sweden, Taiwan, Province of China - The first two tiers are dominated by Euro Area economies as well as Canada, China, Japan, Korea, Switzerland, the United Kingdom, and the United States. - The final tier is largely made up of EMs along with Australia, Denmark, and Sweden. - Interpretation: - Presence of China and key trading partners at the core indicates China’s role in Asian and global supply chains. - Presence of key global financial intermediation centers (United Kingdom, United States, Switzerland) reflects importance in global financial flows, portfolio allocation, and management. - Presence of key EMs confirms that developing economies are becoming regional magnets for financial and trade flows. - Overall, the core largely reflects many Group of Twenty (G20) constituent economies. - Surprises: - Some G20 members are absent from the core: Argentina, Brazil, Indonesia, Saudi Arabia, Russia, and Turkey. - Some non-G20 or smaller economies are present in the core: Denmark, Sweden, Switzerland, Taiwan Province of China, and Thailand. - This motivates exploring the structural and functional properties of the graph via uncovering community (cluster) organization and the role of specific vertices within it. ### From Networks (Graph) to Clusters - Cluster analysis aims to uncover natural organization of vertices (countries) and cohesive communities. - Limitations of traditional partitioning algorithms: - They produce separable non-overlapping, non-nested communities and often require ex-ante definition of number of clusters. - Single-vertex membership only (no overlaps) can miss economically meaningful adjacencies and multi-cluster membership that facilitate cross-cluster interactions. - Clique Percolation Method (CPM): - Allows identification of an unknown number of potentially overlapping clusters with varied membership size. - Models interactions using global trade and financial flows to map potential shock transmission paths. - Countries that share very dense connections (based on trade, BIS, and portfolio flow data) are considered “clustered.” - Clusters share overlap regions facilitated by countries belonging to multiple clusters; such countries are categorized as “gatekeepers.” - Theoretical underpinnings are detailed in Annex 1; Box 1 provides a step-by-step demonstration. - CPM results (preliminary): - CPM yields 11 distinct clusters of varying sizes. - The graph is separable into two core regions: one surrounding the United States and China, and one surrounding Italy and Germany. - A group of countries (e.g., France, The Netherlands, United Kingdom) appears between both core regions. - Figure 3 presents preliminary clustering (not reproduced here). - Box 1 (Cluster identification primer) key definitions and steps: - k-clique: cluster order defined by number of member vertices (e.g., 3-clique). - Clusters are adjacent if they share k-1 vertices; cluster chain is union of adjacent clusters of a given order; maximally connected clusters are found via template rolling. - Four-step maximally connected cluster identification: identify critical regime threshold / edge weight threshold / k range; enumerate clusters starting at largest k and remove identified clustered vertices and edges iteratively; construct clique overlap matrix for each k; reconstruct original graph with identified overlaps and divisions. ### Cluster Analysis Results and Implications - Gatekeepers and gatekeeper clusters: - Some economies serve as gatekeepers—linking different clusters and transmitting (passing through, amplifying, or dampening) shocks. - Groups of economies can together act as gatekeepers; method distinguishes individual gatekeepers and clusters of gatekeepers. - Cluster-to-cluster relationships: - There is a threshold parameter k for defining cluster size (see Annex 1). - For k between 5 and 8 (previously identified as the critical percolation region): - A single cluster arises at the 8-clique level, which includes: Australia, China, Indonesia, Japan, Malaysia, Singapore, Thailand, and the United States. - Three clusters are identified at the 7-clique level, two of which are overlapping, while one stands separated. - Gatekeeper economies (from combining 7 and 8 clique levels): - European economies cluster together, along with Tunisia and Algeria (inclusion due to trade relations and remittances). - China, Japan, Korea, and the United States act as gatekeeper countries linking key Middle Eastern economies and Pakistan to Asia. - Korea, India, and Brunei are members of adjacent clusters and are clustered together despite having no direct connecting edges. - Hong Kong SAR is identified with a nuanced gatekeeper role tied to Asian economies and the United States; the analysis contextualizes Hong Kong SAR’s role as an equity financing hub driven by regional dynamics. - Figure 4 provides a country-level view of first clusters (not reproduced here). - Gatekeeper clusters and overlap structure: - Maximal overlap between clusters takes place at the 6-clique level followed by the 5-clique level. - At the 6-clique level: 12 overlapping clusters containing 72 countries. - At the 5-clique level: 8 overlapping clusters containing 133 countries. - At the 5-clique level, cluster 1 is a giant cluster containing 117 countries. - Clusters are themselves clustered into different yet overlapping groups; overlaps highlight particular clusters (red circles in Figure 5) that act as gatekeeper clusters. - Gatekeeper cluster ranking: - A metric gauges relative involvement of gatekeeper countries in each cluster by measuring weighted contributions of cluster members to each cluster’s overlap relative to average overlap, normalized for cluster size. - Using this metric, cluster #11 under the 6-clique level is confirmed as a top gatekeeper cluster. - Cluster #7 and cluster #4 rank higher than nearly half the other clusters in terms of gatekeeper involvement (see Figure 6). - Characteristics and implications of gatekeepers: - Gatekeeper clusters are groupings of countries through which shock transmission can be mapped. - Membership of top gatekeeper clusters (notably cluster #s 4, 7, and 11 under the 6-clique regime) likely contains countries or groups with characteristics that facilitate transmission roles. - Identifying such gatekeepers helps map potential shock transmission paths and informs surveillance of interconnectedness across trade and financial flows. *Source: _031612b - 6. A dense center. As expected, the network representing trade and financial flows displays a* ### 22. Centrality and gatekeepers. It is clear that the majority of the country membership, in two ### 22. Centrality and gatekeepers. It is clear that the majority of the country membership, in two ### Centrality and gatekeepers - The majority of the country membership in two of the three clusters consists of very large and important economies. - The United States, The Netherlands, China, and Germany lie at the intersection of the three gatekeeper clusters. - Core economies act (via different core regions) as gatekeepers for multiple other important economies which in turn provide a gatekeeping function throughout the entire graph. - The graph of global trade and financial flows is rebuilt based on the gatekeeping score of each cluster. ### Shock propagation (cluster-level implications) - Cluster #1 (the giant cluster under the 5-clique level) is large but its major gatekeeper countries’ functions are already accounted for in other clusters. - The giant cluster helps identify: - Weakly gatekeeping countries that may link strongly peripheral countries to the rest of the network. - How peripheral countries may be linked to one another, thereby linking smaller but more important clusters indirectly. - Without this identification, it is unclear which parts of the periphery rely directly on the core and which reside in cluster chains. ### Illustration and key transmission findings (Table 2 and accompanying text) - Gatekeepers are coherently linked to other gatekeepers, creating an overlap cascade from the core of the system to the periphery. - A shock transmitted from any one point must go through multiple “gates” to spread throughout the system. - There is substantial potential for negative feedback effects within and between clusters. - Potential transmission channels are strongest from Germany and Italy to Europe, CEESE, CIS, and parts of the Middle East. - The United States is a key conduit for spillovers elsewhere. - European stress in Italy or Germany can have first-round direct effects on the U.S.; if not stemmed, additional feedback effects are likely to be transmitted from other European countries to both core economies such as the U.S. and peripheral regions. ### Putting it all together — Cluster-Based Surveillance: analyzing shock propagation and vulnerabilities - Italy is identified as a gatekeeper for core Europe and as a conduit of significant shocks to Belgium, France, Germany, and the Netherlands. - Italy acts as a gatekeeper for five other areas: (i) the CEESE region; (ii) CIS economies; (iii) emerging Middle Eastern economies such as Tunisia and Jordan; and (iv) the United States. - Additional results: - Unlike core Europe and the CEESE economies, the CIS economies may not be as susceptible to secondary reverberation effects. - Certain African economies (Burkina Faso, Cameroon, Gabon, and Senegal) are susceptible to two distinct channels of reverberation aside from the direct impact from Italy: their interconnectedness with France, and their interconnectedness amongst each other. - The U.S. (rather than China) is the larger source of shock transmission from Italy to both Latin America and to Asia. - Implication: U.S. policies to counteract a shock from Italy are likely to have a greater positive impact on global stability at the margin than isolated actions from China alone; concerted action by both will bring the greatest gains, as China is a “gatekeeper” to emerging Asia. - The U.K. appears as a shock absorber from Europe; owing to its unique role as a global intermediator of U.S. dollar assets, it acts as a conduit for shocks to the U.S., with real effects manifested in economies that issued or whose financial sectors utilized those funding vehicles. - China does not receive shocks from others; China’s density lies more within Asia and stickiness in trade relationships is likely to mute the impact of certain shocks on China. However, severe stress in China would produce material effects on countries in its clusters including the United States. ### Mechanical shock experiment (European shock propagation, Figure 10) - A shock is mechanically applied where the value of all links are uniformly reduced by 15 percentage points to proxy for a global GDP shock. - The figure reports, for each country, two numbers: the share of the envisioned shock transmitted to it (incoming from others) on the left, and the share of its effects on others on the right. ### Caveat - The analysis mechanically maps potential transmission routes between clusters and potential feedback loops within clusters. - It should not be interpreted as a quantitative gauge of the relative intensity of shocks between countries without subsequent assessment of policy space and action. - The analysis is a first step toward understanding how a shock could impact the system. ### Annex 1 — Understanding Clique Percolation Method (CPM): method and parameterization - CPM introduces a parameter k defining the size threshold for clustering and identifies Maximal Country Clusters (MCC) as maximally adjacent k-cliques. - Definitions and procedures: - k-clique: a fully connected sub-graph containing k vertices. - k-clique adjacency: two k-cliques connected by k−1 vertices. - k-clique template rolling: begin at a vertex, identify its k-clique, roll the template through adjacent k-cliques to explore a k-clique percolation cluster. - Maximal k-clique cluster: a k-clique that is not a subset of a larger k-clique. - Heuristic to choose k and weight threshold t: - Let G = (V,E) be the connected weighted graph; wij denotes edge weight between i and j. - N = average number of neighbors over all V in G. - t = threshold for edge weights to construct G' with wij ≥ t. - p = probability for an edge to connect any two vertices; pc = critical value threshold for p at which a giant k-clique percolation cluster emerges. - At percolation threshold pc the expected number of adjacent k-cliques reachable by rolling is 1, leading to: - 1 (1)( )1 k c kVkp   = - equivalently: 1 1 () [( 1)() ] k c pkkV k   = −·− - Express N as a function of pc and k: - 1 ()() 1 c V Nkpk k  = · − - Weight threshold t estimation: - maxmin () 0.5 ()() cc tkpk ww= ··− - Empirical parameter choices for the trade and financial flows network: - Best fit k range estimated as falling between 5 and 8. - Within that range, the percolation threshold is around 31 percent. - The expected number of neighbors for each vertex is around 10. - The estimated weight pruning threshold is 15 percent. - Bilateral relations where wij < 15 are eliminated from the analysis. *Source: _031612b - 22. Centrality and gatekeepers. It is clear that the majority of the country membership, in two* ### 1. Backdrop. On September 18, 2007, the U.S. Federal Reserve lowered its policy rate from ### _031612b - 1. Backdrop. On September 18, 2007, the U.S. Federal Reserve lowered its policy rate from ### Backdrop: monetary and commodity shocks (2007–2008) - U.S. Federal Reserve policy rate lowered from 5.25 percent to 4.75 percent on September 18, 2007; federal funds target rate reduced to 1 percent by the end of October 2008. - Crude oil prices surged from over $65 per barrel in mid-2007 to over $130 per barrel by the summer of 2008. - Saudi Arabian Monetary Agency (SAMA) cut its policy rate from 5 percent in October 2007 to 2 percent by mid-2008 to prevent speculative capital inflows and maintain the exchange rate peg to the U.S. dollar. - Annual credit growth in Saudi Arabia increased from 6 percent in early 2007 to over 30 percent in July 2008. - Resulting inflationary pressure culminated in double digit inflation by mid-2008. ### Purpose and motivation - Focus: evolution of interconnectedness and its impact on policy constraints in Saudi Arabia. - Three external factors analyzed: (i) global oil prices, (ii) U.S. business cycle, (iii) developing Asia’s business cycle. - Key structural characteristics of Saudi Arabia (as stated): - Oil sector accounts for over half of GDP. - Oil exports account for over 80 percent of export receipts. - Oil revenues primarily accrue to the government, making the public sector central and dominant in the non-oil economy. - Exchange rate pegged to the U.S. dollar with a relatively open financial account implies interest rate policy closely follows the U.S. Federal Reserve. ### Summary of main results - Business cycle correlation with the U.S. shifted over three decades: - 1980s: supply driven oil shocks caused divergence (negative correlation). - 2000s: demand driven oil shocks (reflecting high growth in developing Asia) resulted in convergence (positive correlation). - Pass-through from global oil prices to fiscal spending has fallen over the past three decades; this may explain reduced output volatility. - Credit dynamics are increasingly relevant for non-oil economic activity. - Importance of U.S. interest rate policy for Saudi Arabia has likely risen in the 2000s. ### Policy implications - With the fixed exchange rate and ongoing financial deepening, synchronization of Saudi and U.S. business cycles is increasingly relevant for monetary policy stabilization. - Structural interconnectedness with developing Asia has increased through growing trade flows and Asia’s rising influence in the global oil market. - Tensions between policy objectives are more likely when global oil prices and the Asian business cycle move counter-cyclically with the U.S. business cycle. - Policy priorities highlighted: - Strong fiscal management. - Further refinement of macro prudential instruments to influence monetary conditions independently of interest rate policy. ### Business cycles and global oil prices: observations (1980–2010) - Data: annually de-trended real GDP for the U.S., developing Asia, and Saudi non-oil GDP; average global oil price; sample covers 1980 to 2010. - Four observations: - Clear negative relationship between U.S. and Saudi economic fluctuations in the 1980s; reversed to positive in the mid-1990s. - Developing Asia’s correlation with Saudi non-oil GDP weak in the first half of the sample, becoming positively correlated at the end of the 1990s and through the 2000s. - Oil prices tend to be positively correlated with Saudi non-oil GDP throughout the whole sample period. - Volatility in Saudi non-oil GDP falls significantly in the 2000s (also relative to the U.S. and developing Asia). ### Simple correlation analysis (regression summary) - Method: Saudi non-oil GDP as dependent variable; U.S. real GDP, developing Asia real GDP, and average oil price as independent variables; all variables detrended using the HP-filter; regressions run for full sample and subsamples 1980–1995 and 1996–2010. - Key regression results shown in Table 1 (coefficients and standard errors listed exactly as presented): - U.S. GDP: -0.66* (Std. error (0.24)) for 1980-2010; -1.02* (Std. error (0.48)) for 1980-1995; 0.20* (Std. error (0.10)) for 1996-2010. - Developing Asia GDP: -0.05 (Std. error (0.21)) for 1980-2010; 0.43 (Std. error (0.48)) for 1980-1995; 0.21* (Std. error (0.06)) for 1996-2010. - Oil Price: 0.18* (Std. error (0.04)) for 1980-2010; 0.25* (Std. error (0.07)) for 1980-1995; -0.01 (Std. error (0.02)) for 1996-2010. - R-squared (adj): 0.49 for 1980-2010; 0.69 for 1980-1995; 0.40 for 1996-2010. - Interpretation: - Full sample: U.S. real GDP negatively and statistically significantly related to Saudi non-oil GDP. - Subsamples: negative relationship holds in 1981–1995; turns positive in 1996–2010. - Developing Asia GDP: not significant over full sample; positive and significant in 1996–2010. - Oil price: positive impact on non-oil GDP in 1980–1995; neither economically nor statistically significant in 1996–2010—possibly reflecting demand-driven oil prices and reduced pass-through from oil revenues to fiscal spending. ### Demand vs. supply driven oil shocks (explanation for changing correlations) - Mechanism: - For Saudi Arabia (net oil exporter): both demand- and supply-driven rises in oil prices cause oil export revenue increases. - For the U.S. (net oil importer): supply-driven price rises increase production costs, contract output, and raise prices—adverse for U.S. output. - Historical narrative: - 1980s and early 1990s: supply disruptions (Iranian Revolution 1978–1979; Iraq–Iran war 1980–1981; first Gulf war 1990) led to large windfalls for Saudi Arabia but recessions in the U.S. and Europe. - Late 1990s and 2000s: consistent upward trend in oil prices since 1998 attributed to strong growth in developing Asia and rising oil demand in China. - China crude consumption increased from 3 mbd to above 8 mbd in 2010; China’s share of global consumption increased from 4 percent to 10 percent. - Developing Asia’s share of global crude oil consumption increased from 11 percent in 1995 to over 19 percent in 2010. - Period 2004–2008: oil price climbed from an average of $40 per barrel to over $130 per barrel; developing Asia accounted for over 43 percent of the global increase in crude oil consumption while North America and Europe combined accounted for 21 percent. - Global economy grew by over 19 percent from 2004 to 2008, while total oil production only rose by 1.8 percent (from 80.6 mbd to 82.0 mbd). - Consequence: oil price cycle in the 2000s became increasingly pro-cyclical, leading Saudi non-oil GDP to co-move positively with both the U.S. and developing Asia; annual non-oil growth rates of 4 and 5 percent in the late 2000s—the highest since the early 1980s. ### Fiscal policy and oil revenue volatility - Role of fiscal policy: - Government spending amounts to over 80 percent of non-oil GDP. - Fiscal policy is the main driver of non-oil growth and has been counter-cyclical with respect to the oil price cycle: draw on reserves or issue debt when oil prices are low; use surpluses to retire debt/build reserves when oil prices are high. - Outcome and volatility statistics (Table 2 values preserved exactly): - Oil revenue growth volatility (std), Spending growth volatility (std), Correlation between oil revenue and spending growth, Non-oil growth volatility (std): - 1980s: 44.4 ; 19.8 ; 0.9 ; 4.6 - 1990s: 21.1 ; 16.2 ; 0.7 ; 1.6 - 2000s: 38.5 ; 6.4 ; 0.2 ; 0.7 - Observations: - Positive relationship between oil revenue and spending growth across decades. - Oil revenue volatility: high in 1980s, declined in 1990s, rose again in 2000s. - Spending volatility: fell consistently across the three decades. - Correlation between revenue and spending: high in 1980s and 1990s; fell significantly in 2000s. - Fiscal authorities appear progressively more successful in smoothing spending despite continued oil revenue volatility. - Possible asymmetry: different responses to increases versus decreases in oil revenues may explain lower non-oil GDP volatility in the 2000s. ### Financial deepening and monetary policy - Financial structure: - Financial system dominated by commercial banks; monetary policy influences activity primarily through exchange rate and bank lending. - With U.S. dollar peg and open financial account, SAMA’s ability to affect the economy via exchange rate and short-term interest rate is limited. - Al-Jasser and Banafe (1999) noted weak interest and credit channels due to presence of government controlled Specialized Credit Institutions (SCIs), lack of financial leverage, and imperfect pass-through of the policy rate to lending rates due to imperfect competition. - Financial development since 1999: - Banking system has grown significantly in size; relative importance of SCIs has declined. - SAMA has undertaken liberalization to allow increased competition (SAMA, 2003). - Increased presence of foreign banks following Saudi Arabia’s accession to the WTO (Ramady, 2010). - As financial deepening reduces frictions, the effectiveness of monetary policy is assumed to have increased over time. - Credit and SCIs (figures referenced): - Figure 4: Credit by commercial banks (Share of non-oil GDP) — time series from 1969 to 2009 shown. - Figure 5: Credit by Specialized Credit Institutions (SCIs) (Share of non-oil GDP) — time series 1987–2010 shown. *Prepared by Niklas Westelius (MCD); ENHANCING SURVEILLANCE: INTERCONNECTEDNESS AND CLUSTERS — INTERNATIONAL MONETARY FUND* ### 12. Credit and non-oil GDP. The first step in assessing the evolving relevance of monetary ### _031612b - 12. Credit and non-oil GDP. The first step in assessing the evolving relevance of monetary ### Granger causality: role of real credit and non-oil GDP - Question addressed: Can real credit help forecast fluctuations in non-oil GDP (and vice versa) and has this ability changed over time? - Data and method: annual de-trended data of real credit and non-oil GDP, 1980–2010. - Findings: - Full sample (1980–2010): real credit does not Granger cause non-oil GDP and vice versa. - Subsample 1996–2010: the null of no Granger causality is rejected, supporting greater importance of real credit to the non-oil economy in this later period. ### Impulse responses: bi-variate VAR analysis of credit shocks - Model specification: - Simple bi-variate vector autoregressive model (VAR) with two lags (AIC lag order selection criterion). - Identification assumes non-oil GDP reacts contemporaneously to a shock to credit, but credit does not react contemporaneously to GDP (results robust to reverse ordering). - Estimated for full sample and subsamples 1980–1995 and 1996–2010. - Findings: - Full sample: response of non-oil GDP to a one standard deviation shock to real credit is positive in the first two years, but insignificant. - Subsamples: - Both sub-periods: non-oil GDP responds positively initially to real credit shocks. - 1996–2010: the positive effect is more prolonged and statistically significant, though the magnitude is smaller than in earlier subsample. ### Variance decomposition: credit’s contribution to non-oil GDP variance - Result: real credit explains a larger share of the forecast error variance of non-oil GDP in 1996–2010 compared to 1980–1995. - 1996–2010: 58 percent after 5 years. - 1980–1995: 40 percent after 5 years. - Interpretation: increased relevance of credit for the non-oil economy in the 1996–2010 period. ### Interest rates and real credit: monthly VAR to test imported interest rate effects - Question addressed: How has the imported interest rate policy affected real credit over time? - Model specification: - Monthly VAR with credit and the CPI as endogenous variables. - Exogenous variables: 3-month LIBOR, oil price, international food price index (and nominal effective exchange rate, NEER, included in results). - Variables log-differenced except LIBOR. - Full sample: 1997:1 to 2008:9 (end date chosen to exclude global financial crisis). - Subsamples: 1997:1–2003:12 and 2003:9–2008:9 to evaluate evolution of the interest rate channel. - Main empirical findings (text and Table 3 results): - Full sample: LIBOR is negatively correlated with credit growth and positively related to inflation; net effect of a rise in LIBOR is a decline in real credit growth. However, exogenous variables are not statistically significant in the full sample. - Subsample results reveal evolving significance: - 1997:1–2003:12: no statistically significant relationship between LIBOR and credit growth; LIBOR has a negative and statistically significant effect on inflation in this period. - 2003:9–2008:9: LIBOR has a negative and statistically significant impact on credit growth; the LIBOR–inflation relationship breaks down in this later period. - Other exogenous variable effects in 2003:9–2008:9: - Oil price has a positive impact on credit growth (coefficient reported as 0.14* with (0.03)). - International food prices have a positive impact on inflation (coefficient reported as 0.04* with (0.02)). - Nominal effective exchange rate (NEER) is positively and significantly correlated with credit growth (coefficient reported as 0.45* with (0.22)). - Selected coefficients from Table 3 for 2003:9–2008:9 (as reported): - 3-Month LIBOR on Credit: -4.01* (1.87) - 3-Month LIBOR on CPI: 0.54 (0.32) - Oil price on Credit: 0.14* (0.03) - Oil price on CPI: -0.01* (0.01) - International food price on Credit: -0.06 (0.09) - International food price on CPI: 0.04* (0.02) - NEER on Credit: 0.45* (0.22) - NEER on CPI: -0.04 (0.04) - R-squared (adj) reported for 2003:9–2008:9: Credit 0.27, CPI 0.41 - Interpretation: the interest rate channel, as proxied by LIBOR, appears to have become more effective in transmitting to credit in the later sub-period (2003:9–2008:9), implying greater sensitivity of domestic credit to the imported interest rate. ### Conclusion: implications for surveillance and domestic policy - Summary of main results: - Two important developments emphasized: - Growing oil demand from EMEs has become increasingly important for oil market dynamics. - Financial sector development in Saudi Arabia has strengthened the monetary transmission mechanism. - Implications: - Greater influence of EMEs’ economic fluctuations on Saudi oil export revenues. - Greater influence of U.S. monetary policy on the Saudi non-oil sector. - Fiscal policy has been increasingly successful at smoothing spending despite oil revenue volatility, possibly explaining lower output volatility in the 2000s. - Going forward — policy implications and risks: - Continued growing links with developing Asia (China and India) and commitment to the U.S. dollar peg create potential tensions when global oil prices move counter-cyclically with the U.S. business cycle. - Example: 2011 — collapse of Libya’s oil exports boosted Saudi oil revenues and fiscal spending, while the Euro area debt crisis raised global downturn risks; combination could produce inflationary pressures from expansionary fiscal policy and accommodative U.S. monetary policy. - Policy recommendations: - Effectively use fiscal policy as a stabilizing tool. - Further refine macroprudential instruments to influence monetary conditions independent of interest rate policy. *Source: IMF — _031612b - 12. Credit and non-oil GDP. The first step in assessing the evolving relevance of monetary* ### 10. Output synchronicity. Given vertically integrated production structures, particularly for ### 10. Output synchronicity. Given vertically integrated production structures, particularly for ### Output synchronicity and cluster dynamics - The Asian cluster, particularly machinery and electronics, is closely linked to the global electronics business cycle, producing largely synchronized business cycles with China. - GDP changes are quite synchronized across the cluster, although the depth of the contraction during the recent crisis varies across countries. - Vertical production linkages (use of intermediate inputs across bilateral sectors) are identified as the main channel through which trade generates business cycle synchronicity. ### Correlation analysis (industrial production, policy rates, exchange rates) - Output variables: high correlations in industrial production across the cluster, especially since mid-2005, and particularly for emerging Asia with respect to China. - Policy rates: reasonably synchronized across the cluster, though correlation with China is less obvious compared to industrial production. - Exchange rates: relatively higher correlations across emerging Asian countries in the cluster, but on the lower side for China. - Empirical interpretation caveat: simultaneity implies caution in reading unconditional correlation matrices, but regional factors have contributed substantially to rising business cycle correlations. ### Rolling correlations with China (2002–2010) - GDP growth and industrial production correlations: - Show divergence during “peace” time 2004-2007/8. - Rise and peak during the global crisis (sharp increase in co-movement during periods of financial stress). - Policy rate correlations: - Appear more persistent over the entire period, including times of lower output correlation. - Example pattern: drop in correlation of GDP growth and industrial production from 2002 to 2006, while policy rate correlations remain relatively high — indicating possible tradeoffs/stresses in policy making for cluster members. ### Historical decomposition of industrial production cycles - Decomposition into domestic and external/cluster-specific factors shows: - Cycles largely synchronized between 2000 and 2004, and during the crisis. - Divergence between country-specific and cluster-wide cycles between 2004 and 2008. - For emerging Asian countries such as Malaysia, Thailand and the Philippines, the importance of cluster-wide factors in explaining variance in industrial production has increased since 2002. ### Relative cyclical positions and implications for policy mix - Output gaps across the cluster were less synchronized with China between 2004 and 2008, allowing scope for divergent policy mixes. - Output gaps turned positive starting 2004 for Malaysia, Singapore and Thailand, suggesting a need for policy tightening. - China had negative output gaps until mid-2006, allowing relatively accommodative policies. - From 2007 onwards, output gaps became more correlated, declining rapidly through early 2009 and recovering thereafter. - Implication: differences in cyclical positions can justify divergent national policy actions, but rising synchronization reduces that scope. ### Interlinkages, China’s centrality, and policy space - China’s role: - Acts as a “gatekeeper” with rising financial linkages and vertically integrated trade structures. - Preservation of place in production structures relative to China can incentivize maintaining relative currency stability vis-à-vis each other. - Can serve as a “circuit breaker” during global demand shocks because of significant policy buffers; its countercyclical response mitigated regional impact during the crisis. - Policy space constraints for downstream/processing-trade economies (e.g., Malaysia, Thailand, Philippines, Vietnam): - The exchange rate instrument may become limited in managing macroeconomic conditions and capital inflows. - Prudential measures become an additional policy instrument when traditional policy space is effectively constrained by other cluster members’ policies. - Scope for policy coordination: - Evidence suggests policymakers paid attention to regional currency developments as China strengthened exchange rate flexibility. - A cluster-level policy dialogue and coordination appear feasible and potentially beneficial. ### Key policy-relevant events and exact numerical facts - China exchange rate reform (July 2005): immediate 2.1 percent revaluation; daily fluctuation within +/- 0.3 percent band; link to an undisclosed basket of currencies. - Cumulative appreciation of the renminbi against the U.S. dollar by end-2008: 20¼ percent. - China growth 2003–2008: grew at more than 10 percent per year. - Malaysia: - Grew at an average of 6 percent per year between 2004 and 2007. - Central government deficit: 4.1 percent of GDP in 2003 to 3.2 percent in 2007. - Abandoned peg on July 22, 2005 (one day after China). - Philippines: - Peso appreciated by a cumulative 25 percent against the U.S. dollar during 2005–2007 period discussed. - Real effective appreciation of 20 percent. - Thailand: - Imposed capital controls in end-2006; some controls lifted only in mid-2008. - National currencies (index 2005:1=100) showed relative stability against the RMB for Hong Kong, Malaysia, Philippines, Singapore, Thailand, Vietnam after China’s mid-2005 reforms. - Box 1 notable findings: - Renminbi appreciation likely has positive effects on firms exporting to China and negative effects on firms providing inputs for China’s processing exports or importing Chinese products with large imported content. - Singapore, Malaysia and Thailand are more impacted by exchange rate volatility with respect to China given their position in the supply chain and reliance on imported inputs. - Policy responses during 2003–2008 varied across countries reflecting initial conditions and policy buffers; responses included fiscal consolidation, exchange rate appreciation, administrative measures, capital controls, and reliance on structural reforms. *Source: IMF, “10. Output synchronicity. Given vertically integrated production structures, particularly for” (excerpts).* ### References ### _031612b - References ### References - Aiyar, Shekhar, (2011), “What impact would a Deleveraging by Euro Area Banks have on Asia?” Mimeo, IMF, December 6, 2011. - Di Giovanni, Julian and Andrei Levchenko, (2009), “Putting the Parts Together: Trade, Vertical Linkages, and Business Cycle Comovement,” IMF Working Paper #181. - Eichengreen, Barry and Hui Tong, (2011), “The External Impact of China’s Exchange Rate Policy: Evidence from Firm Level Data” IMF Working Paper #155. - Frankel, Jeffrey A. and Andrew K. Rose, (1996), “The Endogeneity of the Optimum Currency Area Criteria,” NBER Working Paper 5700. - Fratzscher, Marcel and Arnaud Mehl, (2011), “China’s Dominance Hypothesis and the Emergence of a Tri-Polar Currency System,” Center for Economic Policy Research, International Macroeconomics Discussion Papers DP8671. - He, Dong and Wei Liao, (2011), “Asian Business Cycle Synchronization,” Hong Kong Institute for Monetary Research Working Paper No. 06/2011. - Imbs , Jean, (2004), “Trade, Finance, Specialization, and Synchronization,” The Review of Economics and Statistics, Vol. 86, No. 3, pp. 723–734. - Ito, Takatoshi, (2010), “China as Number One: How About the Renminbi?” Asian Economic Policy Review, Vol. 5, Issue 2, pp. 249–276. - Ng, Eric C.Y., (2010), “Production Fragmentation and Business-cycle Comovement,” Journal of International Economics, Vol. 82, pp. 1–14. - Riad, Nagwa, Luca Errico, Christian Henn, Christian Saborowski, Mika Saito and Jarkko Turunen, (2012), “Changing Patterns of Global Trade,” IMF Departmental Paper Series, Washington, DC: International Monetary Fund. ### Appendix Table A1: Unconditional Correlation Matrices — key notes - Data is monthly and covers the period 1998:1–2011:8. - Unconditional Correlation Industrial Production 1996:1-2011:08 (T-stats in parenthesis). (Matrix entries and t-stats presented in source.) - Unconditional Correlation Policy Rates 1996:1-2011:08 (Matrix entries and t-stats presented in source.) - Unconditional Correlation Exchange Rates 1996:1-2011:08 (Matrix entries and t-stats presented in source.) - Source: Haver and IMF IFS as indicated in the matrices. ### VI. Financial Interconnections and Cross-Border Policy Dependence — overview - Objective: assess the nature of cross-country financial stability and policy linkages arising from cross-border operations of financial institutions. - Key constraint: IMF surveillance hampered by lack of access to adequate information on financial institutions’ exposures to countries, markets, and sectors. - Trade-off: membership in financially integrated clusters increases contagion risk but yields benefits from coordination and information sharing in macro-prudential policy, regulation, supervision, and resolution. - Regulatory arbitrage: clusters generated by financial firms’ attempts at regulatory arbitrage may require policy action to eliminate those opportunities. ### A. Background — core points - Financial interlinkages: cross-border operations of banks establish cross-country clusters that can constrain domestic macro-financial policy effectiveness via increased policy externalities. - Spillovers and clusters: banks operating abroad create clusters (home + host jurisdictions); when such operations are systemically important for hosts they create spillovers of home country policies to hosts and can impose externalities on the home when host operations are material to bank revenues. Examples: Nordic-Baltic, Austria-CEE, and potentially South East Asia with Singapore. - Arbitrage drivers: policy initiatives without harmonized prudential regulation and cross-border cooperation encourage regulatory arbitrage; tax arbitrage also drives cross-border decisions and is difficult to eliminate. - Cooperation requirement: countries in clusters may need to share supervisory information, coordinate prudential standards, and cooperate in supervision and resolution. Defensive policies may be used if coordination is unachievable. - IMF implication: Fund surveillance would benefit from comprehensive monitoring of clusters and requires access to information on G-SIFIs’ exposures to countries, markets, and sectors. ### B. Cross-Border Banking Linkages I: Credit Intermediation — architecture and examples - Home-host linkages: - Parent funding is an important source of capital for domestic credit intermediation in host countries. - Host country exposures can be material to BHC liquidity and solvency; downturns in hosts can threaten refinancing capacity and solvency of parents. - Nordic-Baltic example: five large banks chartered in Sweden (Nordea, SEB, and Swedbank), Denmark (Danske), and Norway (DnB NOR) dominate deposit and credit markets throughout the region. - Swedbank example: deposits and lending by the parent bank, on a gross basis, to the Baltic affiliates increased its share of non-equity liabilities from 17½ percent at end-2005 to 46 percent by end-2008. Parent funding rose from 28 percent of local deposits to over 107 percent over this 3 year period. - Austria-CEE example: Austrian banks Erste and Raiffeisen active in the Czech Republic, Hungary, Poland, Romania, and Slovakia; in all but Poland the affiliates are systemically important and are material to the parent banks. - Singapore example: Singaporean banks generate and conduit regional capital flows into South-, South-East-, and East-Asia as measured by proportion of bank-sourced external capital (numbers in Table 4 drawn from BIS locational banking statistics and confidential BIS reporting). - Centralized business models: - Centralization of credit, funding, treasury, and risk management across borders reinforces clustering and direct financial interlinkages. - Corporate credit often managed centrally; example: Finnish corporate credit in DnB NOR group, Danish corporate credit managed by DnB NOR or Nordea Finland. - Austrian banks used affiliates in some CEE countries to fund and extend credit in Swiss franc (CHF), eliminating market risk but exposing borrowers and banks to currency appreciation (CHF vis-à-vis the Forint). - Group-level funding often raised from global wholesale capital markets in foreign currencies and distributed to regional operations; enterprise risk management centralizes capital buffers with group-level treasury hedging. ### How does this Affect Domestic Policy Space within the Cluster? - Monetary policy constraints: - Direct parent funding can constrain the effectiveness of host country monetary policy in containing credit and asset booms because funding costs are driven by global markets rather than host policy. - In open capital accounts, this can "clog the credit channel" of monetary policy. - Alternative instruments may be necessary, including partial capital controls (example: India), tax treatment of capital inflows sensitive to instrument/maturity/currency (example: Brazil), and subsidiarization with strict limits on intra-group exposures (examples: Mexico and Spain). - Need for cooperation: - Sole reliance on domestic prudential regulation and supervision of local affiliates may be difficult and inefficient for cross-border BHCs with substantial intra-group funding exposures. - Example: lowering household leverage in Norway via loan-to-value and loan-to-income limits and higher mortgage risk weights may have limited impact without corresponding measures in other Nordic countries due to passporting under EEA rules and branches of foreign banks. - Suggested coordinated actions (Nordic context): (i) agree to raise risk weights on Norwegian residential mortgage exposures; or (ii) agree on uniformly higher risk weights on their banks’ residential mortgage exposures independent of location of the borrower. ### Positive and Negative Policy Spillovers — observations and examples - Countercyclical macro-prudential strengthening in one country can increase efficiency of countercyclical policies elsewhere in the cluster. - Weaknesses in countercyclical macroeconomic management can yield high corporate, household, and public indebtedness alongside asset bubbles, limiting scope to deploy targeted fiscal, monetary, and financial instruments during downturns. - Baltic housing market overheating and high household mortgage indebtedness pre-global crisis were not adequately addressed by domestic macro-prudential tools; Latvia showed procyclicality in fiscal and monetary policy after EU accession in 2005. - Nordic monetary and financial stability policies did not adequately address risks from rising wholesale funding and intra-group credit exposures funding the Baltic boom. - Confluence of exogenous shocks starting in Q3 of 2008—the liquidity tightening after the Lehman bust and the recession in Germany—exacerbated the Nordic-Baltic downturn; threats to SEB and Swedbank liquidity and solvency arose from Baltic exposures. - Consequences in H1 2009: Riksbank liquidity swap lines had to cover all Baltic central banks despite differing fundamentals (e.g., Estonia vs. Latvia). Swedish authorities adopted general creditor guarantees and interventions subscribed to by SEB and Swedbank in 2009, substantially increasing contingent fiscal liabilities. - Defensive measures: a country may raise liquidity and capital buffers and ring-fence domestic banking operations to reduce vulnerability to contagion, which reduces degree of inclusion in the cluster. - Harmonization benefit: host authorities also benefit from harmonization to higher prudential standards or consistent implementation across the cluster, as illustrated by experiences with Icelandic parent banks and affected U.K., Dutch, and Isle of Man creditors and authorities. *Source: _031612b - References (IMF PDF content provided).* ### 11. Information sharing and coordinated supervision. Absence of supervisory information ### 11. Information sharing and coordinated supervision. Absence of supervisory information ### Key findings on information sharing and supervisory coordination - Absence of supervisory information sharing or of coordinated supervisory action may increase vulnerability to exogenous shocks and risk of financial contagion for individual countries or the entire cluster. - Good supervision requires taking appropriate action on the basis of accurate, timely, and comprehensive information to stem excessive risk taking by regulated institutions; it also requires endowing supervisors with an adequate budget and sufficient enforcement powers and legal protection. - Executing good supervision within a cluster can become problematic, particularly in a setting of high intra-group exposures, in the absence of supervisory information sharing and adequate scope for coordinated supervisory action. - Poor resourcing or empowerment of supervision, or capture by national banking champions, can exert a significant negative externality on other countries within the cluster, especially where information sharing is suboptimal. - The inability of national supervisory action to affect the behavior of business units operating outside national boundaries represents a problem for domestic financial stability and may force measures to ring-fence local operations, thereby limiting participation in the cluster. ### Illustrative supervisory cooperation examples - The pre-crisis Nordic supervisory MoU embedded an institution-specific component and paved the way for supervisory information sharing and joint inspections in cases such as Nordea and Sampo. - In the wider EU context, joint inspections were not mandated but bilateral cooperation arrangements could and did exist (example: Belgium–Netherlands cooperation for Fortis), and such arrangements did not preclude a national approach to resolution. ### Crisis management and cross-border resolution - Coordinated management of BHC failure or systemic financial distress can reduce the economic costs of a financial crisis. - The cross-border nature of BHC operations complicates resolution where intra-group exposures are high because of centralized liquidity management (examples: Dexia and Lehman) or parent funding importance for affiliates (example: Baltic operations of Swedish banks). - National resolution frameworks and crisis management responsibilities can conflict with efficient and equitable resolution of cross-border banking groups. - Home countries may favor subsidiary structures to limit parent bank liability—and hence contingent fiscal costs—in distress; home legislation enabling subordination of deposits in overseas branches during restructuring could shift fiscal costs to host jurisdictions. - Concerted coordination on joint restructuring solutions, even absent ex-ante burden sharing agreements, can be successful and welfare improving. Equitable distribution of resolution costs in proportion to the share of the enterprise held by each jurisdiction, and clear communication of the adopted strategy, can allow distressed institutions to obtain finance and buy time to sell or retrench operations. ### Clusters and regulatory/tax arbitrage - Policy initiatives that dissolve barriers to movement of financial capital and promote cross-border market access can lead to highly integrated financial systems (example: EU). - Banks may utilize different jurisdictions to raise market funding, manage risk, engage in credit intermediation, and extend brokerage services. - In some cases (Nordic-Baltic, Austria-CEE clusters) these choices reflect economies of scale and scope; in other cases they reflect banks’ desire to exploit tax and regulatory arbitrage or ease of cross-border penetration without corresponding convergence of prudential standards or supervisory coordination. ### Case study: Deutsche Pfandbrief Bank (Depfa) and Hypo Real Estate (HRE) - In 2001 corporate restructuring of Deutsche Pfandbriefebank, AG resulted in a new parent bank, Depfa Bank plc, incorporated in Ireland. Motivation included potential for substantial tax savings and high asset-liability maturity mismatches (ALMMs) permitted in Ireland but strongly regulated in Germany after 1999. Responsibility for consolidated supervision moved from Germany to Ireland. - Depfa group combined a thin spread business with assumption of high liquidity and interest rate risk, making it especially vulnerable to increases in global risk aversion and liquidity squeezes such as those in 2007–08. - Depfa bank’s primary funding came from jumbo covered bonds. The group used regulatory space under Irish regulations to swap down overall funding duration substantially. Market commentary suggests Depfa continued to attract funding using this strategy well into 2008, helped by the solid asset quality of its cover pool of German public sector loans. - Depfa bank remained a mid-sized issuer of Pfandbrief; liquidity or solvency problems at the group level could disrupt confidence in the Pfandbrief market and affect public sector refinancing needs. - Depfa group’s takeover by German mortgage lender HRE in 2007 combined Depfa’s ALMM with HRE’s portfolio of legacy bad loans. Because HRE was larger in asset size, the merged institution had a riskier funding profile and poor asset quality. - Depfa group’s ALMM proved costly to HRE and German authorities, being a major source of HRE’s vulnerability to funding squeezes throughout 2008 and ultimately leading to a takeover by SoFFin as part of a rescue package amounting to over €100 billion, a substantial proportion of which was funded by the state. ### Policy spillovers and supervisory lessons from Depfa–HRE - Ability of banks to engage in corporate restructuring and relocation of home country without corresponding changes in business operations can be problematic absent harmonized prudential standards; regulatory arbitrage often leads to excessive risk taking and high eventual costs to concerned countries. - Pre-crisis legal gateways for supervisory information sharing were either not sufficiently precise or not appropriately utilized for German authorities to access necessary prudential financial information on Depfa on an ongoing basis from Irish authorities. - Maintenance of sound prudential policy by German authorities for domestically licensed Pfandbrief banks did not prevent excessive risk accumulation by group affiliates located outside Germany. BaFin did not automatically obtain regular information on this risk accumulation and was not alerted in advance to the need for defensive action. - Subsequent to HRE’s takeover, BaFin assumed the role of consolidated supervisor; BaFin requested its Irish counterpart for permission to review Depfa’s books in March 2008 and on the basis of this review, issued an order to HRE to close out Depfa’s unhedged positions. - A more general problem arises from misaligned regulatory incentives in clusters: the home agency may have less incentive to allocate scarce prudential resources than the host agency if the institution’s primary activity takes place in the host jurisdiction. An example: Bank DnB Nord, the Danish subsidiary of DnB NOR, does not have substantial group-level business in Denmark but its Baltic branches are systemically important in that region. ### Arbitrage, OFCs, and cluster sustainability - Use of offshore financial centers (OFCs) by large global BHCs for booking transactions and hedging can reflect tax breaks and regulatory advantages sought by OFCs; while OFCs gain business, the practice may increase global (unhedged) risk for BHCs and pose stability and fiscal risks for home authorities in the absence of supervisory information sharing. - If clustering primarily reflects firms’ exploitation of regulatory arbitrage, authorities may question cluster continuation—especially where boom-period benefits accrue to financial enterprises while downturn costs are widely distributed among taxpayers across jurisdictions. Appropriate policies may include eliminating arbitrage opportunities and, thereby, the clusters they create. ### Challenges in identifying clusters: information gaps - Mapping financial interlinkages for cluster identification is constrained by significant information gaps; these limit ongoing assessments necessary for Fund surveillance and advice. - Unavailability of information in the public domain: - Geographic distribution of a country’s banking sector holdings in other countries is not publicly available through BIS banking statistics. - Geographic distribution of portfolio assets held by fund managers domiciled in a given country are not publicly available and typically need private sources such as EPFR and Lipper. - Unavailability of comprehensive cross-country information: - Confidential BIS data does not carry geographic distribution of risk exposures for a majority of countries including large emerging markets such as Brazil and Mexico. - Corresponding information is not supplied by EPFR on bonds and equities funds’ holdings, complicating identification where such countries are important originators or conduits of capital flows. - Gaps in information content: - BIS data provides net—and not gross—flows; reliance on net flows blunts early warning capacity because gross banking flows can provide useful information. - Data limitations make it difficult to determine whether a country (e.g., Singapore) is an originator of bank lending or merely a booking location for tax purposes—this affects assessment of the materiality of policy decisions in regional financial centers. - Constructing metrics of relative importance of potential regional financial centers for driving external capital requires reconciling external capital stocks/flows (e.g., from CPIS or IIP databases) with banking booking data, which is generally difficult beyond banks. *Source: IMF — "11. Information sharing and coordinated supervision. Absence of supervisory information" (extracted content).* ### 4. Benchmarking policies. At the same time, the Baltic countries were very conscious of the ### _031612b - 4. Benchmarking policies. At the same time, the Baltic countries were very conscious of the ### Benchmarking and policy constraints - Politicians in all three Baltic countries emphasized the need to keep taxes low, especially corporate rates. - Competitiveness concerns: with nominal exchange rates fixed, real adjustments had to come from wage cuts. - As part of the euro accession process, policymakers were constrained by the desire to demonstrate credibility; Estonia committed to “balance or better” in their budget. - Policymakers were conscious of benchmarking against each other. ### Severe adjustments and macrofinancial impact - Pre-crisis, all three Baltics ran large current account deficits, supported by consumer loans from Nordic banks. - When the crisis hit, deficits became surpluses; in Latvia this was partly an accounting issue (central bank recorded losses written off by parent banks as positive entries in the income account, amounting to a few percent of GDP). - Fiscal adjustment: as revenues dropped with GDP, fiscal deficits were closed through combinations of expenditure cuts and new revenue sources. - Deleveraging in the medium-term was recognized as necessary and has been happening gradually. - Financial-sector movements transmitted to the real economy through drops in consumption and investment, notably in construction. ### Pre-crisis dynamics and policy limits - Co-movement: nominal GDP levels, industrial production indices, and per capita GDP show relatively high co-movement among the Baltics and with others, particularly during the financial crisis. - Convergence vs over-heating: - EU accession expected to produce convergence—higher-than-average growth supported by lower real prices, improved competitiveness, more investment and exports. - Access to credit from the EU implied consumption would rise during convergence, making sustainability hard to judge. - Attempts to rein in lending had limited effect because many cross-border mechanisms were not legally binding and did not specify actions in distress. - Identified weaknesses: uncertainty about deposit insurance, burden-sharing in insolvency, and constraints on small countries with restricted policy space under fixed exchange rates. - Sweden emerged as a gatekeeper: Swedish banks had incentives to fund Baltics (low wholesale funding rates, high Baltic growth), quickly built large market shares, and persisted in that role. ### Crisis: amplification, feedback, and dampening - Amplification: - Withdrawal of funding by Swedish banks amplified the global shock; short-term wholesale funding transformed into longer-term loans became unviable as global liquidity dried up. - Collective bank pullback created a destabilizing spiral, forcing crisis in Latvia and severely impacting Estonia and Lithuania. - Feedback: - Baltic loan losses contributed a large proportion of Nordic banks’ losses, though share of overall lending to Baltics was relatively low compared to Nordic lending. - Sweden quickly supported banks and tightened supervision to prevent future concentration of risk. - Loss reversals have occurred, but low growth and rebalancing will likely weigh on future profits and tax revenues; risks to Swedish authorities from insolvency are likely lower. - Dampening: - Swedish authorities provided liquidity in krona and foreign currencies and directly supported troubled banks, stabilizing markets and easing deleveraging pace. - Swedish actions allowed Baltic policymakers to focus more on macroeconomic adjustments rather than large financial-sector bailouts. ### Box 1 — Key Crisis Intervention Measures (measures and status) - Measures to increase liquidity in Krona and foreign currencies (mostly introduced in September–December 2008) - The Riksbank expanded a collateral policy framework by fully accepting covered bonds and lowering the minimum credit rating requirements for long-term securities pledged as collateral. — Active. - The Riksbank set up new 3, 6, and 12-month loan facilities to facilitate banks’ access to longer-term funds. — Replaced by 28-day loan facility with a penalty rate, April 2010. - The Riksbank established a new temporary credit facility using commercial paper as collateral to facilitate the supply of credit for non-financial companies. — Discontinued, September 2009. - The Riksbank granted emergency liquidity assistance to Kaupthing Bank Sverige AB and Carnegie Investment Bank AB (both SEK 5 billion). Later outcomes: former bank sold to Alandsbanken by its Icelandic owners; latter taken over by the state and later sold to private investors. — All repaid. - The Riksbank and U.S. Fed set up temporary reciprocal swap facilities ($30 billion). A separate swap facility was also established with the ECB. — Active, but currently no use. - New dollar tem loan facilities (with the maturity of 28 and 84 days) were introduced. — Discontinued, November 2009. - The SNDO held extraordinary auctions of treasury bills to provide markets with securities generally accepted as collateral; funds raised invested (via repos) in covered bonds to support covered (mortgage) bond markets. — Discontinued. - Measures to support banks’ capital and assure market confidence (all measures introduced in October 2008) - The deposit guarantee was raised from SEK 250,000 to SEK 500,000, and the coverage was broadened to include all types of deposit. — To be raised to EUR 100,000 (about SEK 895,000) from 2011. - The parliament enacted the “Government Support to Credit Institutions Act,” giving the government unlimited fiscal powers to finance measures needed to ensure financial system stability; support can be provided through guarantees, capital injections, or other means and provided based on contracts between the government and a bank. — Active. - National Debt Office given power to take over a troubled bank if there is a serious systemic risk and bank capital falls below 25 percent of the regulatory requirement. — Active. - A debt guarantee scheme for the medium-term borrowing of banks and mortgage institutions was established; total amount set at SEK 1.5 trillion (later reduced to SEK 750 billion). An institution applying to the guarantee would pay fees and be subject to restrictions on remuneration for senior management. — Active, but currently no use. - A recapitalization scheme for solvent banks and other credit institutions was established; government capital injection takes the form of shares or hybrid capital (Tier 1 capital). Participating institutions are subject to restrictions on remuneration for senior management. — Active, but currently no use. - A stabilization fund was set up to finance government measures to support the financial system; sources are annual fees from banks and other credit institutions. The government contributed SEK 15 billion for establishment of this fund. — Active. ### Gatekeeper role: counterfactual VAR analysis - Model and sample: - VAR model with six variables: VIX; G20 GDP weighted real GDP growth (proxy for global shocks); money market rate and real GDP growth in the gatekeeper country; money market rate and real GDP growth in the destination country. - Destination countries: Estonia, Latvia, Lithuania; gatekeeper: Sweden. - Estimation period: typically from the first quarter of 1995 to the third quarter of 2011 (with some variation across countries). - Identification: - Structural shocks identified by Cholesky ordering with contemporaneous chain: VIX, G20 real GDP growth, gatekeeper money market rate, gatekeeper real GDP growth, destination money market rate, destination real GDP growth. - Assumption: shock to gatekeeper money market rate reflects gatekeeper’s monetary policy (caveat: during financial stress, link between policy and money market rate may weaken). - Comparison approach: - Compare two impulse responses: unrestricted model vs model with gatekeeper money market rate restricted to a constant (implemented by adding shocks needed to keep the rate constant). - Results: - Swedish monetary policy attenuates spillovers from negative global shocks to money market rates and real GDP growth in the Baltic countries, especially for money market rates in Estonia and Latvia. - For Estonia and Latvia, the conditional median expected path of the money market rate (conditional on Sweden holding policy unchanged) lies outside the 90 percent confidence interval of the unrestricted response after three to four quarters (confidence intervals computed via bootstrap with bias correction per Kilian, 1998). - A negative global shock lowers Swedish money market rates (presumably due to more accommodative monetary policy); if Sweden’s policy were unchanged, this would imply positive shocks to money market rates and higher rates in destination countries. - Differences in Swedish policy effects on real GDP: strongest for Latvia and Lithuania (in response to a VIX shock); for Estonia the conditional median impulse response of real GDP growth is within the 90 percent confidence interval of the unconditional one. - Caveats: - Uncertainty of conditional impulse response estimates, time variation, and low number of observations for some countries. - Nonetheless, policy in gatekeepers can materially impact other countries in the cluster; cluster members should be aware. ### Core and cluster connections; policy implications - Benefits and limits: - Even with perfect coordination between a gatekeeper and cluster members, conditions in the global core will affect the cluster (easy access to wholesale funding driven by global core conditions). - With open capital flows, it is difficult for a cluster or gatekeeper to lean against large flows originating from the core. - If peripheral clusters’ conditions are correlated (e.g., similar business models), the robustness of clusters matters to the core. - Gatekeepers’ policy space matters: - Nordic gatekeepers (Sweden, Austria) had policy space to act as shock absorbers: gross government debt below 60 percent of GDP, current account surpluses, and comfortable fiscal accounts in 2008 (small deficit for Austria, small surplus for Sweden). - Sweden and Austria quickly bolstered banks operating in their clusters; such responses would not have been possible if gatekeepers lacked market access. - Contrast: Bulgaria and Romania with Greece as gatekeeper—there the cluster members could not rely on the gatekeeper to be a shock absorber. - Flexibility: - Gatekeeper flexibility is important given uncertainty about future shocks. - Alternative policy choices by gatekeepers could have produced different amplification/dampening outcomes; absence of a backstop by gatekeepers could have forced more severe deleveraging in recipient countries, potentially amplifying shocks and threatening bank health more broadly. *Source: EXCERPT — ENHANCING SURVEILLANCE: INTERCONNECTEDNESS AND CLUSTERS (content unit _031612b).* ### 23. Data. Better data would help. There have been long lags between the initial suspicion of ### 23. Data. Better data would help. ### A. Background - Long lags have existed "between the initial suspicion of something wrong and the provision of data to track certain kinds of interconnections." - The ongoing Data Gaps initiative is attempting to bridge some of these gaps. - By the time data are requested and received, it may already be too late; as flows become larger, the role of gatekeepers as potential shock absorbers could also grow. - Larger policy space is likely to be essential given these data and timing limitations. ### B. Trade Clusters — considerations for surveillance - Export sectors - Knowing the trade sectors to which a country belongs informs the extent to which macroeconomic conditions (domestically and in the cluster) are impacted by global factors/cycles (e.g., terms of trade). - Divergent macroeconomic policies and outcomes across cluster members can yield lessons for response options. - Position in supply chains - A country’s upstream or downstream position matters; analysis of value-added trade is required to determine reliance on imported inputs for export production. - Advanced countries tend to be upstream (important sources of inputs); Emerging markets (EMs) tend to be further downstream due to heavier reliance on imported inputs. - A supply shock in an upstream country (for example, the earthquake in Japan in 2011) can reverberate through the cluster if reliance on inputs from that country is heavy and short-term substitution is difficult; impacts also depend on conditions and policies in gatekeepers. - Competition - Countries with similar export structures are more likely to compete in third markets and may have lesser flexibility in some policy instruments (e.g., nominal exchange rates). - If output gaps do not co-move, this can create policy trade-offs (for example, a country may find it harder to appreciate nominally for competitiveness reasons even if output is above potential). - Trade credit - Trade credit is a link between financial and trade clusters and a channel for financial shock spillovers. - Example: Euro Area banks have provided a large proportion of total trade credit in Asia; deleveraging by Euro Area banks could adversely impact Asian trade and finance. - Susceptibility varies with trading partners’ funding models. ### C. Financial Clusters — considerations for surveillance - Funding model - The funding model underpinning cross-border lending matters; during the crisis, foreign banks that relied on retail deposits behaved similarly to domestic banks. - The parent - The balance sheet strength of the parent bank matters in both stressful and "normal" times. - Prior to the crisis, abundant liquidity in advanced markets was channeled to EMs, contributing in some cases to rapid growth and high leverage. - During the crisis, subsidiaries of parents that were weathering the crisis well received headquarters’ financial support; subsidiaries of parents more severely affected were cut off from the head office and had to reduce exposures to the corporate sector. - Strategic importance - Whether an economy has markets of strategic importance to others matters for durability of presence by foreign institutions. - EU-headquartered banks considering CEE part of their core market created strong incentives to maintain presence even under stress. - Economies serving primarily as booking centers for financial transactions (e.g., some off-shore centers) are likely substitutable within or across clusters. - Countries hosting platforms/infrastructure/markets vital to a cluster are likely durable elements and could affect other cluster members and other clusters. - Integration - Degree of integration across clusters affects transmission. - In the crisis, difficulties in cross-border financing affected domestic liquidity in EMs through several channels: - funding costs in domestic currency increased in several economies as dislocation of the cross-border funding market prompted institutions to fund U.S. dollar borrowing through the FX swap market; - heightened counterparty risk affected borrowing in the interbank market; - there was a shortening in the maturity structure of funding. - Other channels - Identifying key channels allows more targeted analysis of transmission across clusters. - Domestic interbank money markets were relatively less affected in the crisis compared to the FX market. - In most EMs, effects on domestic bond markets were relatively limited; in EMs with comparatively deep bond markets, concerns about foreign investor sales led to sharp widening in bid-ask spreads that effectively stopped market activity. *ENHANCING SURVEILLANCE: INTERCONNECTEDNESS AND CLUSTERS — INTERNATIONAL MONETARY FUND* --- _Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2012/_031612b.pdf_