Financial Integration, Monetary Unions, and Symmetry (Correlation of Cyclical Fluctuations) API-119 Lecture 8: Guest Lecture by Jeff Frankel Recap: Financial Integration Pros & Cons • Pros: In API-120 & API-119, we learned that functions of international financial markets include: – Consumption Smoothing • break the timing between production & consumption – Risk Diversification • insurance again shocks – Efficient Allocation of Capital • capital is at its most productive use; • investors earn a higher return. • Cons: – These come with costs, such as volatility & crises. – Capital does not always seem to flow the right direction! Evidence from International Data • In previous API119 lectures we saw many puzzles in the literature: • The theoretical benefits of financial integration do not match evidence from international data. – Feldstein-Horioka Puzzle • High saving-investment correlations – Risk Sharing Puzzle • Low consumption correlations – Lucas Paradox • Capital does not flow from rich to poor. The FeldsteinHorioka Puzzle Feldstein-Horioka regression: (I/GDP) = α + β (NS/GDP) + v. Feldstein (1980) argued that if capital were perfectly mobile, he would find β = 0. Instead, β was much closer to 1. The coefficient (“saving retention”) fell a bit subsequently, but still high. * * See Table 2, Appendix I in this powerpoint. There are many critiques of Feldstein-Horioka • (I) National Saving is endogenous • The “intertemporal optimization” critique -Private saving varies with the business cycle, – or with population or productivity growth. – Often: “A theoretical model can be constructed in which capital mobility is perfect and yet the saving-investment correlation is high.” – Obstfeld, Summers, Tesar… • The “maintained external balance” critique – – Fiscal policy is endogenous: governments react to trade imbalances – Tobin, Westphal, Caprio & Howard, Roubini, Bayoumi, Buiter. • (II) The world real interest rate is endogenous – = The “big-country” critique. – Murphy, Tobin, Obstfeld. – But that doesn’t help explain the cross-section findings. • The most common critique is that NS is endogenous, • which should call for an Instrumental Variable. • IV for Public Saving (BS): military spending • IV for Household Saving: dependency ratio. • Yet the IV estimates of the F-H coefficient (“saving retention”) are as high as the OLS estimates ! * See Table 3, Appendix I in this powerpoint. Evidence from Intra-national Data Studies of data among regions within a common currency do not show the puzzles: current account deficits & surpluses are big enough to allow saving and investment to go their separate ways • Feldstein-Horioka tests on regions within a common currency – Sub-regions within the UK: • Bayoumi, Tamim, & Andrew Rose (1992), "Domestic Saving and Intra-National Capital Flows," European Economic Review. – Provinces within Canada (for 1961-1990) : • Bayoumi, Tamim, & Gabriel Sterne (1992), "Regional Trading Blocs, Mobile Capital and Exchange Rate Coordination," IMF. – Prefectures of Japan: • Dekle, Robert (1996) "Saving-investment associations and capital mobility: On the evidence from Japanese regional data," JIE, Aug. • Iwamota & van Wincoop (2000): Estimated coefficient is 0.3 in cross section, 0.2 in panel. “Intranational versus International Saving and Investment co-movements, ” in Intranational Macroeconomics, Hess & van Wincoop, eds.. – States within the US (1950s data): • Sinn, Stefan (1992+), "Saving-Investment Correlations and Capital Mobility: On the Evidence from Annual Data," Economic Journal. • The same for nations under the gold standard (Bayoumi, 1990). • The finding is never a high positive correlation between NS & I – as is standard in international studies. The prediction of the full risk sharing model also holds up better on intra-national data: – Crucini & Hess (2000): cross-region consumption correlations > output correlations. • On data from US states, Canada provinces & Japan prefectures. » “International versus Intranational Risk Sharing,” in Intranational Macroeconomics, Hess & van Wincoop, eds. – Kalemli-Ozcan, Sorensen & Yosha (AER, 2003): intranational risk sharing >> international risk sharing. • On data from US states, Canada provinces, Japan prefectures, UK regions, Italian regions, & Spanish regions • Summary: – Regions that are known to share a common currency and to be highly integrated with respect to their goods markets pass the Feldstein-Horioka and risk sharing tests, while standard international data fail the tests. – Also, recall that tests by price-based criteria such Covered Interest Parity, financial markets are highly integrated. – An implication: exchange rate variability or other sources of imperfect integration of goods markets may be the source of real interest differentials and quantity-based findings of “low capital mobility” – although this is not necessarily the authors’ interpretations of their own results. Optimal Currency Areas • Symmetry of GDP fluctuations is one of the main criteria for Optimal Currency Areas (OCA). • It goes back to Mundell (1961). • OCA theory says that the benefits of a common currency outweigh the costs if: – – – – the shocks and cycles are similar the countries are open to trade with each other the degree of labor mobility is high a system of risk-sharing is in place • through stabilizing fiscal transfers • or through stabilizing private capital flows. Endogeneity – “Lucas Critique” • OCA theory talks as if trade patterns and other parameters are exogenous and unchanging. • But the original motivation for currency unions such as EMU was to promote trade within the region! – Rose (2000) showed that countries with a common currency do indeed trade more, as much as x2 or x3. • An application of Lucas Critique: you cannot rely on ex ante statistical estimates to analyze the outcome of a change in regime (joining), because the “parameters” will change after the new regime is in effect! – You do not know in advance if it is optimal or not. – One would have to derive everything from deeper parameters that don’t change. If intra-regional trade is endogenous with respect to MU decision, then cyclical correlations are likely to be as well. (I) Eichengreen-Krugman hypothesis on the sign of the endogenous effect of on correlations. • Krugman (1993): when MU boosts intra-regional trade, – countries specialize according to comparative advantage. – So trade shocks become more idiosyncratic (asymmetric). – Countries share production risk via integrated capital markets. • More specialization in production induces a higher degree of asymmetry (lower correlation of cycles) – => Even if countries appear to satisfy OCA criterion ex ante, • they may fail it ex post. (II) Frankel-Rose hypothesis on the sign of the endogenous effect of on correlations. • Once countries are in EMU and trade more – it leads to a lower degree of asymmetry, – more highly correlated business cycles. – => Even if countries appear to fail OCA criterion ex ante, • they may satisfy it ex post. • FR: empirically, more trade leads to more symmetry • Frankel-Rose (1998), “The Endogeneity of the Optimum Currency Area Criteria,” Econ.J. Frankel-Rose Regression • Corr (GDPi, GDPj) = a + b Tradeij + controls + error – i-j are countries (pair-wise regression); – Cross-section estimated over 5 year window • b is estimated to be positive with strong statistical significance, – even when endogeneity of trade is handled by IV • from the gravity model (proximity of pair, size, etc.). – Supports F-R hypothesis over Eichengreen-Krugman. Frankel-Rose: Both OCA criteria, not just intra-union trade but also symmetry, are more likely to hold ex post than ex ante. Trade OCA criterion line Countries should adopt common currency Countries should float Correlation of Business Cycles Across Countries “symmetry” of shocks Evidence on the Competing Channel: Building Block I • Kalemli-Ozcan, Sorensen & Yosha (2003): more capital market integration => more specialization, using data from US states – Regression: Spec i = a + b fin.integration i + error – Spec i is a state/region within a country – Spec i measures how much i’s production differs from the rest of states within the country – Integration i measures how financially integrated is i with the rest of the states within the country. – Estimate of b is positive and significant. Evidence on Competing Channel: Building Block II • Kalemli-Ozcan, Sorensen, Yosha (2001) (KSY): more specialization => less asymmetry. • Recall FR Regression: – Corr (GDPi, GDPj) = a + b Tradeij + controls + error. • KSY Regression: – Corr (GDPi, GDPagg) = a + b Fin.Integrationi + controls + error – Dependent Variable ≡ correlation of i with the aggregate – (country i is in) To Summarize Trade Finance Policy Knowledge MORE MORE Frankel & Rose Intra-Industry LESS Inter-Industry Kalemli-Ozcan, Sorenson & Yosha MORE Fluctuations Asymmetry LESS LESS Which one dominates? • Empirical papers above show both are important. • No study yet runs a horse race between the two channels – which would require pair-wise data both on trade linkages and on financial linkages • But we have an experiment to evaluate – European Crisis: 2010-….. Joining Euro Zone • One of the arguments made in favor of Euro zone in the past is that even when member countries are hit by asymmetric shocks, they still do not need independent monetary policies • Even if output shocks are asymmetric, consumption will not be, thanks to integrated capital markets! • Sure enough, the periphery countries ran huge CA deficits after joining: NS << I . After the formation of the Euro, current account imbalances rose sharply Current Crisis • But the current euro crisis is a clear indication that such insurance has not been achieved among Euro zone countries! Smoothing Fluctuations: Evidence • US Smoothing (1999-2005): – Capital Markets: 55% – Federal Government: 15% – Credit Markets: 30% • Euro Zone Smoothing (1999-2005): – Capital Markets: 10% – Euro Zone Government: 0% (there is no such gov) – Credit Markets: 35% Euro Zone Fluctuations and Smoothing • Clearly capital markets did not do the job, whether because: – Not integrated enough – Not enough time passed since common currency – Markets still segmented – with different laws and jurisdictions – Or maybe capital flows are procyclical – contrary to theory ! • And of course this is not a fiscal union – so there are supposed to be no fiscal transfers. 1300 1200 1100 1000 900 800 700 0 Jan-95 Apr-95 Jul-95 Oct-95 Jan-96 Apr-96 Jul-96 Oct-96 Jan-97 Apr-97 Jul-97 Oct-97 Jan-98 Apr-98 Jul-98 Oct-98 Jan-99 Apr-99 Jul-99 Oct-99 Jan-00 Apr-00 Jul-00 Oct-00 Jan-01 Apr-01 Jul-01 Oct-01 Jan-02 Apr-02 Jul-02 Oct-02 Jan-03 Apr-03 Jul-03 Oct-03 Jan-04 Apr-04 Jul-04 Oct-04 Jan-05 Apr-05 Jul-05 Oct-05 Jan-06 Apr-06 Jul-06 Oct-06 Jan-07 Apr-07 Jul-07 Oct-07 Jan-08 Apr-08 Jul-08 Oct-08 Jan-09 Apr-09 Jul-09 Oct-09 Jan-10 Apr-10 Jul-10 Oct-10 Jan-11 Apr-11 Jul-11 Spreads over German Bunds show integrated markets 1400 Big question: why did markets think these periphery countries were as safe as Germany upon joining euro? Greece Portugal Ireland 600 500 400 300 Spain 200 Italy 100 Maastricht Criteria • Many viewed that convergence according to Maastricht criteria will prepare the countries for OCA. Most important criteria: – Debt levels < 60% of GDP – All countries violated European Debt Levels Why it is a political problem in a currency area with no fiscal union: Exposure to Greek Debt • • • • • • • • • • • • • • • • ECB (bought in open market) Greek banks (held as collateral by the ECB) Greek pension funds and insurance comp. French banks German banks UK banks Portuguese banks US banks Dutch banks Italian banks Austrian banks Swiss banks Belgian banks Japanese banks Spanish banks Others (insurance, hedge funds) 55.0 40.0 30.0 56.9 28.3 14.7 10.2 8.7 5.2 4.5 3.3 3.0 2.0 1.3 1.1 20.0 Frankel Appendices on Measuring International Capital Mobility 1. Feldstein-Horioka for developing countries. 2. Interest Rate Parity: Country premium • • vs. currency premium for Latin America in 1994. Periphery euro countries versus emerging markets 2006-10. Appendix 1: The FeldsteinHorioka coefficieint (“saving retention”) appears no higher for developing countries than for industrialized countries – the opposite of what one would expect if measured barriers to capital mobility. • IV for Public Saving (BS): military spending • IV for Household Saving: dependency ratio. ˡ • Yet the IV estimates of the F-H coefficient (“saving retention”) are as high as the OLS estimates ! ˡ Appendix 2: Measuring factors in interest differentials • Sometimes the effect of capital controls can be isolated by offshore-onshore interest differentials – including by the covered interest differential to take out currencies difference (for countries with forward markets), – or differential in local $-linked bonds vs. US T-bills. • Sometimes currency premia can be decomposed. • The effect of default risk can be isolated by the sovereign spread on bank loans or bonds (EMBI). Total interest differential (Local – US ) = (Currency premium) + (country premium) = (Δse + exchange risk premium) + (country premium) On the eve of the Mexican peso crisis Sovereign interest rates, in 3 crises Source: IMF Worst mis-pricing: Greece’s sovereign spreads 2003-08