The Role of Selective Capital Account Openness in Resilience to Financial Shocks. Tong Hui (IMF) Shang-Jin Wei (Columbia Univ, NBER & CEPR)

Transcription

1 The Role of Selective Capital Account Openness in Resilience to Financial Shocks Tong Hui (IMF) Shang-Jin Wei (Columbia Univ, NBER & CEPR)

2 Motivations Innovation and industrial upgrading more dependent on external finance than average economic activities. One threat to successful innovation, upgrading, and economic growth is an economy s vulnerability to liquidity shocks and to financial sector crises. Role of selective capital account convertibility Resilience to finance shocks Example of government interventions meant to create a hospitable/safe environment for innovation and growth 2

3 The Goals How different are different types of capital flows Consequence during a crisis (some results) Consequence during tranquil periods (preliminary results) To do: How to combine them? Endogeneity? 3

4 A crisis as an angle to examine the composition effect in financial globalization A large literature on effects of international capital flows ( financial globalization ): Potential benefits: lower cost of capital, better risk sharing, knowledge spillover, better discipline Henry (2007); Stulz (2005), etc Stubborn lack of empirical corroboration Kose, Prasad, Rogoff and Wei (2003); Rodrik and Subramanian (2008) Composition matters: Frankel-Rose (1996), Wei (2001, 2006 and 2007) 4

5 What is the composition effect? Because bank loans (and maybe portfolio flows) are more volatile than FDI, the share of FDI in total capital inflows affects a country s experience with financial integration. Not all integration is equal. The composition and the likelihood of currency/bop crisis Frankel and Rose, 1996: Probit in a panel Classens, Dooley, and Warner, 1995 (skeptics): how reliable are the labels? Frankel and Wei, 2005: (semi-parametric) regression tree method Other types of compositions (maturity structure, share of foreign currency debt) The composition and the liquidity crunch Tong and Wei,

6 Agenda for the presentation Basic facts: Are different types of capital flows actually different in volatility? Preliminary evidence: The composition effect in the crisis: Is there evidence that it affects the severity of the liquidity crunch? To do: Effects of capital flows in tranquil times What lies behind the composition? 6

7 Results so far Are different types of capital flows actually different in volatility? Yes. FDI appears most stable; foreign loans appear most volatile The composition effect in the crisis: Is there evidence that it affects the spread of the liquidity crunch? Yes Countries that relied more on non-fdi before the crisis exhibit a more severe credit crunch during the crisis What lies behind the composition? Financial institutional quality is an important factor Property rights institutions can have different effects Consistent with recent theory 7

8 Part 1: Do the labels matter? 8

9 Figure 1: Capital Flow to Emerging Economies (in US$ Billions) Direct Investment Foreign Loans Portfolio Investment

10 Volatility of (FDI/GDP) and (Loan/GDP) ( , Measured by Standard Deviation) FDI/GDP Loan/GDP 10

11 Volatility of (FDI/GDP) and (Loan/GDP) ( , Measured by Coefficient of Variation) 11

12 Volatility measured by standard deviation 12

13 Volatility measured by standard deviation FDI/GDP Loan/GDP Portfolio/GDP 13

14 Part 2: Is there evidence supporting the composition effect from the latest crisis? 14

15 The current crisis as an opportunity Crisis = 危 (danger) 机 (opportunity) The global financial crisis as an opportunity to check the composition effect Whether/how a country s pre-crisis composition of capital inflows affects the severity of liquidity crunch experienced by its non-financial firms during a systemic crisis. 15

16 Methodology (Tong and Wei, 2009) Ask not how the aggregate stock prices have done; ask how individual stocks have done differently relative to each other. Main ideas: (1) If credit crunch worsens, this should be reflected in the relative stock price responses between those firms that rely heavily on external finance versus those that don t. (2) If pre-crisis composition matters for liquidity crunch, it should be reflected in crosscountry variations. 16

17 International lending -> support domestic lending E.g., the wholesale funding model of Korea FDI: Internal capital market of multin l firms Normal times: Desai, Foley and Forbes (2009) Crisis times? 17

18 Main Findings (1) Liquidity squeeze is wide-spread across countries. Firms that depend intrinsically more on external finance for working capital fare significantly worse 18

19 Main Findings (2) Pre-crisis volume of capital inflows does not predict severity of credit crunch in 2008 The (lack of a) volume effect hides an important composition effect The credit crunch is more severe in emerging countries with large exposure to non-fdi flows but less severe in countries with exposure to FDI. 19

20 Empirical Specification Key regressors Baseline findings and Robustness checks 20

21 Econometric Model The model (1) StockReturnij = α j + βfinancialdependenceij + Controlij + εij Pure cross-sectional Key regressors are all pre-determined by values in

22 Pre-crisis Financial Integration β = β + β (2) 1 2Financial Integration j The severity of an emerging economy s credit crunch is systematically linked to its pre-crisis capital inflows since the crisis may have triggered a reversal of global capital flows. Focus on 24 emerging markets. 22

23 Intrinsic Dependence on External Finance for Capital Investment Dependence on external finance for investment [capital expenditures - cash flow] =, capital expenditures We calculate the SIC 3-digit sector median for U.S. firms from 1990 to Then apply this same index to other countries, following Rajan and Zingales (1998). 23

24 Intrinsic Dependence on External Finance for Working Capital Intrinsic liquidity need for working capital Cash conversion cycle = 365* inventories-account payables cost of goods sold + account receivables total sales We calculate the SIC 3-digit sector median for U.S. firms from 1990 to 2006 Then apply the same index to other countries, similar to Raddatz (2006) and Kroszner, Laeven and Klingebiel (2007). 24

25 Examples DEP_INV: dependence on external finance for investment Low: Vegetable and animal oils; Textile goods; Sport and athletic goods High: Drugs and medicines; Office and computing mach. DEP_WK: dependence on external finance for working capital (liquidity needs) Low: Petroleum refineries, Soft drinks, Bakery products High: Radio, TV. and comm. eqp; Leather products; Scientific equipment 25

26 Control Variables Country fixed effects Firm-level features: four factor model (firm size, beta, market/book, momentum). Sector Sensitivity to Demand Shock Track stock return from 9/10/01 to 9/28/01 for each U.S. listed firm. Define the average stock return for each 3-digit SIC sector as the sector-level sensitivity to demand shock, as in Tong and Wei (2008). 26

27 Dependent Variable The difference in the log of stock price from July 31, 2007 to December 31, 2008 for 14,307 nonfinancial firms in 44 developed and emerging economies. 27

28 The Extent of Liquidity Crunch The fall in stock price is more severe for sectors with a greater dependence on external finance for working capital; In some cases, more severe for sectors with dependence on external finance for investment as well (Table 3) 28

29 Stock Returns from 7/31/07 to 12/31/08 Dependence for Investment (DEF_INV) Dependence Working Capital (DEF_WK) * *** *** Beta*Market Return 0.270*** Firm Size 3.748*** Market/Book *** Momentum Demand Sensitivity *** 29

30 Pre-crisis Exposure to Capital Inflows (Averaged from , % of GDP, 24 emerging economies) Country Total inflow FDI FPI Foreign Loans Chile China Hungary India Korea

31 Capital flow volume is insignificant We start with a de facto measure of financial integration: the country s annual inflow over GDP averaged from 2002 to Capital flow volume is not significantly associated with the severity of credit crunch. (Table 5) 31

32 Table 2. Role of Pre-crisis Financial Integration (Emerging Markets; Volume Effect) (Dependent Var: Stock return from 7/31/07 12/31/08) DEF_INV DEF_INV*Inflow DEF_WK * DEF_WK*Inflow Firm level controls Yes Yes Industry fixed effects No Yes Observations

33 But composition matters a great deal Separating FDI, portfolio investment, and foreign loan (over GDP). FDI reduces the liquidity crunch while Non-FDI aggravates it. (Table 6) Results hold even after we control firmlevel factors and industry fixed effects. 33

34 Table 3. Role of Pre-crisis Financial Integration in Emerging Markets (Dep Var: Stock return from 7/31/07 12/31/08) DEF_INV DEF_INV*FDI ** DEF_INV*FPI ** ** DEF_INV*Foreign Loan * DEF_WK ** DEF_WK*FDI ** DEF_WK*FPI DEF_WK*Foreign Loan * * Firm-level controls No Yes Sector fixed effects No Yes 34

35 Robustness Tests Add domestic financial development (x dependence on external finance) no effect Add an indicator of market vs bank system (and x DEF) Add firm-level leverage ratio and (x DEF) Add a dummy for a stock that also has an ADR (and x DEF) Add measure of cyclicality (X FD) no effect Directly measure demand-sensitivity (source: Tong and Wei, 2008). 35

36 Role of Pre-crisis Level of Domestic Financial Development (Emerging Markets, Robustness check) (Dep Var: Stock return from 7/31/07 12/31/08) DEF_INV*FDI 2.647** 3.213** DEF_INV*FPI ** ** DEF_INV*Foreign Loan DEF_WK*FDI * DEF_WK*FPI DEF_WK*Foreign Loan (Domestic Private Credit/GDP)*DEP_INV (Domestic Private Credit/GDP)*DEP_WK (Credit+ Market Capitalization/GDP)*DEP_INV (Credit +Market Capitalization/GDP) *DEP_WK -6E-05 Sector fixed effect and firm-level controls Yes Yes 36

37 Could realized volume and composition of capital inflows be endogenous? Try de jure measures of capital account convertibility with attention on different restrictions on different types of capital flows 37

38 De Jure Measure of Capital Openness A de jure measure on capital openness, based on the IMF s Annual Report on Exchange Arrangements and Exchange Restrictions for year Openness on inward capital flows: FDI, FPI (stocks and bonds), and foreign loan (commercial and financial credits) respectively. Confirm the findings for de facto classification, particularly for DEP_INV. (Table 9) 38

39 Role of Pre-crisis Financial Integration De Jure Classification (emerging economies) (Dep Var: Stock return from 7/31/07 12/31/08) DEF_INV DEF_INV*FDI *** DEF_INV*FPI * ** DEF_INV*Foreign Loan DEF_WK DEF_WK*FDI DEF_WK*FPI * DEF_WK*Foreign Loan Industry fixed effects No Yes 39

40 Case study: Lehman Brothers Bankruptcy Examine stock returns from Sept 12, 2008 to Sept 16, 2008 for manufacturing firms in 24 emerging economies. Same qualitative result: pre-crisis FDI flows alleviate the credit constraints, while Non-FDI flows make it worse. 40

41 Table 11. Stock Returns around Lehman Brother Bankruptcy Case 6 DEF_INV*FDI 0.330*** DEF_INV*FPI DEF_INV*ForeignLoans DEF_WK*FDI DEF_WK*FPI ** DEF_WK*ForeignLoans * Leverage ** Leverage*FDI Leverage*FPI Leverage*ForeignLoans Sector dummies and firm controls Yes 41

42 Part 3: Overall assessment Tranquil periods (early results, ) Some evidence that capital inflows alleviate liquidity constraints Question: How to put the results from tranquil and crisis periods together? 42

43 Part 4: Endogeneity of composition Is it institutional quality or level of financial development lurking behind the composition effect? 43

44 Explanation? Caballero, Farhi, and Gourichas (2008) Mendoza, Enrique, Quadrini, and Rios-Rull (2009), "Financial integration, financial deepness and global imbalances," NBER Working Paper Ju and Wei (AEJ-EP 2010), Domestic institutions and the bypass effect of financial globalization Good property rights protection (low corruption) helps to attract more FDI, maybe less loans Better financial system would retain more domestic savings at home More domestic investment -> lower MPK Less FDI 44

45 Ju and Wei (AEJ 2010): Domestic Institutions and the Bypass Effect of Financial Globalization Embed Holmstrom and Tirole (1997) into a 2-country general equilibrium model When c=b=0, MPK=financial interest rate. Inefficient financial system (a high c or a higher b) reduces the financial interest rate for a given MPK Two-way capital flows Different effect of financial institutions/property rights on K flows 45

46 Linking composition to institutions and financial development 46

47 Linking composition to institutions and financial development OLS Measurement errors + endogeneity 2SLS: Using history as instruments Settler mortality: Acemoglu, Johnson and Robinson Legal origin: La Porta, Lopez-di-Silanes, Shleifer and Vishny 47

48 48

49 49

50 50

51 51

52 Policy lessons Looking at the average effect of capital inflows may mistakenly conclude that capital flows don t matter Since the composition matters, reforms on capital account need to be sequenced Future work: need to quantify benefits of non-fdi in good times and costs in bad times Recognize determinants of the composition How innovation and growth are affected by selective capital account openness 52