Investment Financing and Financial Development: Evidence from Viet Nam Conference on Understanding Banks in Emerging Markets (CEPR, EBRD, EBC, RoF) Conor M. O Toole 1 Carol Newman 2 1 Economic Analysis Division Economic and Social Research Institute 2 Department of Economics Trinity College Dublin
Introduction Modelling Investment Data and Econometrics So many banks in Ha Noi... Results Conclusions and Policy Implications Annex
1 Introduction Why Viet Nam? 2 Modelling Investment Investment Framework Empirical Model with Financial Development 3 Data and Econometrics Data Econometric Methodology 4 Results Main Results Additional Findings and Robustness Checks 5 Conclusions and Policy Implications 6 Annex
Extensive literature concerning Does finance cause growth? King & Levine (1993), Honohan and Beck (2007), Beck (2013), Beck and Levine (2004) Policy prescriptions for developing economies include measures to facilitate financial reform and development Benefits of financial development (Levine, 2005) Increased and more efficient investment Better monitoring and corporate governance Improved risk management Trade facilitation
RQ :Does financial development reduce firm financing constraints? Build on Rajan and Zingales (1998), Love (2003), Love and Zicchino (2006), Guariglia and Poncet (2008), Beck et al. (2008) Complementary to research on financial reform and access to finance Galindo et al. (2007), Abiad et al. (2008), Haramillo et al. (1996), Barajas et al. (2000), Gelos and Werner (2002) Complement work on SMEs and financing constraints in development Beck and Demirguc-Kunt (2006a,b), Ayyagari et al (2007), Beck et al. (2008)
Innovations and contributions of this paper 1 Firm level data on non-listed SMEs in industry and services; 2 Direct data on how firms financed investment; 3 Within country variation in financial development; and 4 Definition of financial development (1) Financial depth (2) State involvement (3) Market financing
Why Viet Nam? Fast growing, realigning economy with considerable product and capital market liberalisation (WTO accession 2007) Economic Growth Investment Source: General Statistics Office, Vietnam
Why Viet Nam? Major changes to financial operating environment evident at macro level Financial liberalisation Domestic credit to private sector (as % of GDP) Banking sector concentration Expansion of monetary base Source: Beck et al. (2000, 2009); Cihak et al. (2013) World Bank WDI; Abiad et al (2010).
Investment Framework Modelling investment Neoclassical Q model of finance: ( ) I K it = α 0 + 1 a Q i, + c i + λ t + θ jt + ε it Accelerator model: ( ) I = β 0 + β 2 s it + β 3 s i + c i + λ t + θ jt + ε it K it Financing constraints: Cash flow? Cash stock? ( ) Direct measure: IF it = IF IF +ExF it ( ) I K it = α + β Q Q i, + β IF IF i, + c i + η t + λ j + ɛ it A-priori expectations: β Q > 0 β IF > 0
Empirical Model with Financial Development Financial Development - Decomposed 1 Financial depth 1 Business credit to private sector as percent of industrial output, FinDepth pt 2 State involvement 1 SOE share of total loans, (L S ) pt 2 SOE share of loans to SOE share of output, ( LS )pt 3 Market financing of investment ( ) I K it 1 % of inv lending by commercial banks to % of inv lending by state, ) ( CL GL pt = α + β Q Q i, + β IF IF i, + β OFD OFD p, + φ IFOFD (IF OFD) p, + c i + η t + λ j + ρ p + ψ pj + ɛ it A-priori expectations Financial depth φ IFOFD < 0 State involvement φ IFOFD > 0 Market financing of investment φ IFOFD < 0
Data Data Vietnamese Enterprise Survey (VES) over the period 2002-2008 Data prepared and cleaned under DANIDA funded project Surveys all firms > 30 employees with a representative sample of firms < 30 employees Covers all sectors of the economy Including industry and market services Data includes all 64 rural and urban provinces Provincial measures of FD aggregated from firm data
Data Summary Statistics VARIABLE N Mean St Dev I K 43,138.927 1.489 IF 44,153.691.404 ( ) CF K 37,867.116.312 ( ) S K 44,153 4.716 4.651 ( P ) K 44,153.076.405 FinDepth 44,153.269.136 ( ) LS L T 44,153.416.188 ( ) LS 44,153 1.245.447 ( ) CL GL 42,817 134.804 228.215 Private 44,153.784.411 State 44,153.128.334 Joint Venture 44,153.023.150 Solely Foreign 44,153.065.247 Services 44,153.388.487 SME 44,153.872.334
Data Figure: Mean of Provincial Financial Development Indicators Private business credit to industrial output SOE share of loans SOE loan share to SOE output share Commercial to state loans Source: Authors calculation using Vietnamese Enterprise Survey data
Econometric Methodology A number of important econometric issues are raised Measuring Q Panel VAR on firm fundamentals x it = Ax i, + κ i + γ t + u it q it = ( c [I δa] 1 δa ) x it VAR includes mvpk and CF/K ratio Estimate using System GMM (Holtz-Eakin et al. (1988)) Further considerations Errors-in-Variables and endogeneity Firm level heterogeneity Spatial and temporal serial correlation in errors Choosen methodology Difference GMM (Arellano & Bond (1991)) with Helmert transformation (Arellano & Bover (1995)) Cluster robust standard errors Exogeneity condition for instruments: E(ɛ it X i,t s ) = 0 s > 1 (1)
Main Results Table: GMM Investment Equation Estimates - All Firms Dep Var Constraint Financial Depth State Involvement Market Financing Overall I it K i, (1) (2) (3) (4) (5) (6) (7) (8) (9) 10 Q i, 0.110*** 0.110*** 0.109*** 0.109*** 0.109*** 0.110*** 0.110*** 0.111*** 0.111*** 0.111*** (0.011) (0.011) (0.011) (0.011) (0.011) (0.011) (0.011) (0.012) (0.012) (0.012) IF i, 0.048*** 0.050*** 0.013 0.046*** 0.025 0.048*** 0.056*** 0.062*** 0.034* 0.026 (0.015) (0.015) (0.017) (0.015) (0.016) (0.015) (0.016) (0.017) (0.018) (0.019) FinDepth p, -0.350** -0.364** -0.593*** (0.147) (0.147) (0.158) IF i, FinDepth p, -0.827*** -0.645*** (L S ) p, 0.546*** 0.563*** (0.182) (0.212) (0.072) (0.072) IF i, (L S ) p, 0.611*** p, IF i, ( CL GL ) p, IF i, ( CL GL ) p, p, (0.148) 0.064** 0.069** 0.053 (0.031) (0.031) (0.033) 0.147** 0.223*** (0.062) (0.066) -0.000-0.000 0.000 (0.000) (0.000) (0.000) -0.001*** -0.000** (0.000) (0.000) Sargan/Hansen J (p-value) 0.446/0.462 0.457/0.469 0.436/0.449 0.392/0.384 0.364/0.352 0.449/0.463 0.438/0.450 0.804/0.771 0.820/0.791 0.798/0.762 Res AR(2) (p-value) 0.410 0.419 0.428 0.450 0.507 0.399 0.409 0.573 0.553 0.588 n 38,912 38,912 38,912 38,912 38,912 38,912 38,912 35,214 35,214 35,214 * p<0.10, ** p<0.05, *** p<0.01
Main Results Overall Marginal Effects ( ) ( ) I IF = 0+( 0.645) FinDepth+(0.223) LS CL +( 0.0001) GL Year Overall Marginal Effect 2002.051*** 2004.028*** 2006 -.008 2008 -.066*** Standard errors calculated using bootstrap methods on MF distribution * p<0.10, ** p<0.05, *** p<0.01
Additional Findings and Robustness Checks Test if effects are heterogeneous across firms Define an indicator variable for: 1. SMEs 2.Foreign firms 3. Market services Interact with all main variables Test for robustness using cash flow - investment sensitivities and alternative clustering
Additional Findings and Robustness Checks Table: GMM Investment Equation Estimates - Domestic Firms by Size Category ( IF TF ) ( ) IF TF ( ) IF TF (1) (2) (3) (4) 0.023 0.001 0.012 0.016 (0.023) (0.027) (0.024) (0.025) M 0.053* 0.095*** 0.062* 0.056* (0.031) (0.036) (0.032) (0.034) S 0.019-0.035-0.013 0.038 (0.031) (0.037) (0.034) (0.033) ( ) IF FinDepth TF -0.073 (0.271) ( ) IF M FinDepth TF 0.541 (0.385) ( ) IF S FinDepth TF -1.589*** ( ) IF TF L T ( ) IF M TF ( ) IF S TF ( ) IF TF ( ) IF M TF ( ) IF S TF L T L T (0.401) 0.091 (0.258) -0.132 (0.360) 0.896*** (0.341) -0.077 (0.103) 0.062 (0.140) 0.353*** (0.136) Sargan test (p-value) 0.450 0.441 0.358 0.427 Hansens J (p-value) 0.465 0.455 0.346 0.437 Res AR(2) (p-value) 0.415 0.431 0.509 0.418 n 38,912 38,912 38,912 38,912 * p<0.10, ** p<0.05, *** p<0.01
Additional Findings and Robustness Checks Table: GMM Investment Equation Estimates - Domestic Firms by Size Category ( IF TF ) (1) (2) (3) (4) (5) (6) 0.023 0.001 0.012 0.016 0.024 0.036 (0.023) (0.027) (0.024) (0.025) (0.024) (0.029) ( ) IF TF M 0.053* 0.095*** 0.062* 0.056* 0.055* 0.066* (0.031) (0.036) (0.032) (0.034) (0.031) (0.039) ( ) IF TF S 0.019-0.035-0.013 0.038 0.015-0.038 (0.031) (0.037) (0.034) (0.033) (0.031) (0.040) ( ) ( ) IF CL TF 0.000 GL ( ) ( ) IF M CL TF GL ( ) ( ) IF S CL TF GL (0.000) 0.000 (0.000) -0.001*** (0.000) Sargan test (p-value) 0.594 0.505 0.435 0.473 0.325 0.836 Hansens J (p-value) 0.691 0.532 0.445 0.496 0.390 0.814 Res AR(2) (p-value) 0.401 0.476 0.451 0.372 0.294 0.629 n 38,912 38,912 38,912 38,912 38,283 35,214 * p<0.10, ** p<0.05, *** p<0.01
Additional Findings and Robustness Checks Effects heterogeneous across firms SMEs FD financing constraints for small firms No effect for medium firms Services Less constrained FD stronger financing constraints Main beneficiaries Foreign firms No effect of FD, use IF Evidence that FOR and SOEs don t compete for finance
Additional Findings and Robustness Checks Table: GMM Estimates - All Firms - Q Model ( ) IK Constraint Financial Depth State Interventionism Market Financing Overall t Q 0.083*** 0.032*** 0.023** 0.028*** 0.036*** 0.041*** ( CF K )t (0.023) (0.010) (0.010) (0.009) (0.013) (0.011) 0.368*** 0.007 0.180* 0.168 0.185** 0.086 (0.078) (0.103) (0.104) (0.119) (0.093) (0.122) FinDepth -0.712*** -1.041*** (0.169) (0.196) ( ) CF K t -2.099* -3.499*** L T ( ) CF K t ( ) CF K t ( CL GL ) L T ( ) ( CF CL K t GL ) (1.209) (1.076) -0.595*** (0.081) 0.145 (1.210) -0.097*** -0.108*** (0.030) (0.034) -0.557 1.209*** (0.713) (0.211) 0.001*** 0.001*** (0.000) (0.000) -0.001*** -0.001* (0.000) (0.000) Sargan test (p-value) 0.22 0.10 0.06 0.18 0.41 0.83 Hansens J (p-value) 0.10 0.08 0.20 0.12 0.18 0.76 Res AR(1) (p-value) 0.0 0.0 0.0 0.0 0.0 0.0 Res AR(2) (p-value) 0.73 0.74 0.85 0.99 0.96 0.61 Time/Province/Sector Dummies Yes Yes Yes Yes Yes Yes Sector-Province Dummies Yes Yes Yes Yes Yes Yes n 51,871 42,517 52,482 52,482 45,356 38,121 * p<0.10, ** p<0.05, *** p<0.01 All estimates are robust to heteroskedasticity and clustered at the firm level Instruments are lagged CF/K, sales to capital dated t 3 and deeper, as well as lags of variables in main equation where correct
Conclusions Clear evidence that financial development reduces financing constraints in Vietnam Constraints are: 1 Decreasing in financial depth 2 Increasing in the use of finance by SOEs 3 Decreasing in the degree of loans allocated on market terms Distributional impacts are evident with small firms and service firms benefiting Policy Implications Vietnam had gradual but progressive opening of capital markets Policy makers must ensure financial development effects those starved of finance Mix of more credit and better allocation criteria required Micro-finance institutions not used in Viet Nam - network of former state lenders Future Research Does financial development improve the marginal productivity of capital? Household decisions, access to credit and financial development
Many thanks for your time. Questions, comments, suggestions... Email: conor.otoole@esri.ie
Definitions Financial Development Ease by which firms with positive NPV projects can finance investment Financing Constraint Wedge between internal and external cost of capital
Table: GMM Investment Equation Estimates - Solely Foreign Firms FOR ==1 ( IF TF ) ( ) IF TF (1) (2) (3) (4) (5) (6) 0.046*** 0.008 0.022 0.053*** 0.044*** 0.030 (0.016) (0.018) (0.017) (0.016) (0.016) (0.019) FOR 0.040 0.072* 0.055 0.020 0.056 0.089* (0.037) (0.039) (0.037) (0.045) (0.037) (0.046) ( ) IF FinDepth TF -0.852*** (0.191) ( ) IF FOR FinDepth TF 0.318 ( ) IF TF L T ( ) IF FOR TF ( ) IF TF ( ) IF FOR TF ( ) ( IF CL TF GL ) L T ( ) ( IF FOR CL TF GL ) (0.451) 0.625*** (0.152) -0.139 (0.517) 0.170*** (0.065) -0.352** (0.168) -0.001*** (0.000) 0.001* (0.000) Sargan test (p-value) 0.450 0.441 0.358 0.427 0.298 0.820 Hansens J (p-value) 0.465 0.455 0.346 0.437 0.353 0.792 Res AR(2) (p-value) 0.415 0.431 0.509 0.418 0.313 0.545 n 38,912 38,912 38,912 38,912 38,283 35,214 * p<0.10, ** p<0.05, *** p<0.01
Table: GMM Investment Equation Estimates - Services Firms - SERV = 1 ( IF TF ) ( ) IF TF (1) (2) (3) (4) Overall 0.043* 0.057** 0.073*** 0.076*** 0.064** (0.025) (0.024) (0.022) (0.029) (0.030) SERV -0.138*** -0.106** -0.051-0.163*** -0.159*** (0.047) (0.042) (0.038) (0.049) (0.054) ( ) IF FinDepth TF -0.694** -0.534* (0.291) (0.329) ( ) IF TF -1.047** -0.520 ( ) IF TF L T ( ) IF SERV TF ( ) IF TF ( ) IF SERV TF ( ) ( IF CL TF GL ) L T ( ) ( IF SERV CL TF GL ) (0.510) (0.615) 0.299 (0.226) 0.938** (0.390) 0.038 0.099 (0.098) (0.106) 0.300* 0.462*** (0.156) (0.176) -0.000 0.000 (0.000) (0.000) -0.002*** -0.002*** (0.001) (0.001) Sargan test (p-value) 0.314 0.264 0.306 0.752 0.617 Hansens J (p-value) 0.346 0.284 0.345 0.733 0.720 Res AR(2) (p-value) 0.484 0.565 0.417 0.562 0.617 n 29,330 29,330 29,330 29,330 25,759 * p<0.10, ** p<0.05, *** p<0.01
Attrition Year No of Firms % of Total 2002 22,050 8% 2003 28,588 10% 2004 37,192 13% 2005 45,024 15% 2006 61,560 21% 2007 41,048 14% 2008 57,398 20% Source: VES
Summary Statistics VARIABLE N Mean St Dev I K 43,138.927 1.489 IF 44,153.691.404 ( ) CF K 37,867.116.312 ( ) S K 44,153 4.716 4.651 ( P ) K 44,153.076.405 FinDepth 44,153.269.136 ( ) LS L T 44,153.416.188 ( ) LS 44,153 1.245.447 ( ) CL GL 42,817 134.804 228.215 Private 44,153.784.411 State 44,153.128.334 Joint Venture 44,153.023.150 Solely Foreign 44,153.065.247 Services 44,153.388.487 SME 44,153.872.334
Variable Definitions Table: Overview of variables in empirical model Variable Description Source I K Investment to beginning period capital stock VES S K Total sales to beginning period capital stock VES CF K Net income plus depreciation VES HHI j The Herfindahl index of revenue concentration (4 digit sector) VES FinDepth Credit to the private sector as a percentage of output VES (L S ) SOE Share of Outstanding Loans VES ( ) LS GDP ( ) S CL GL SOE Loans Share relative to SOE Output Share VES % of Loans by Commercial Banks to % Loans by Gov Banks VES FOR Firms with 100% Foreign Ownership VES JV Joint ventures with foreign companies VES SERV Market Services Sectors VES SME Firms less than 250 employees Eurostat
Table: GMM Estimates - All Firms - Simple Accelerator Model ( ) IK Constraint Financial Depth State Interventionism Market Financing Overall t ( y t ) 1.796*** 0.243** 0.192* 0.273** 0.458*** 0.400*** ( CF K )t (0.525) (0.096) (0.100) (0.123) (0.171) (0.133) 0.421*** -0.116 0.245** 0.162 0.230** 0.141 (0.131) (0.163) (0.098) (0.119) (0.089) (0.116) FinDepth -0.743*** -0.985*** (0.171) (0.206) ( ) CF K t -6.331*** -3.484*** L T ( ) CF K t ( ) CF K t ( CL GL ) L T ( ) ( CF CL K t GL ) (1.192) (1.043) -0.552*** (0.099) 0.302 (1.219) -0.083** -0.089** (0.033) (0.036) -0.909 1.147*** (0.679) (0.203) 0.001*** 0.001*** (0.000) (0.000) -0.001*** -0.001** (0.000) (0.000) Sargan test (p-value) 0.31 0.25 0.05 0.19 0.364 0.449 Hansens J (p-value) 0.22 0.27 0.11 0.14 0.352 Res AR(1) (p-value) 0.0 0.0 0.0 0.0 0.0 0.0 Res AR(2) (p-value) 0.05 0.68 0.94 0.89 0.507 0.399 Time/Province/Sector Dummies Yes Yes Yes Yes Yes Yes Sector-Province Dummies Yes Yes Yes Yes Yes Yes n 92,281 42,441 52,356 52,356 45,250 38,055 * p<0.10, ** p<0.05, *** p<0.01 All estimates are robust to heteroskedasticity and clustered at the firm level Instruments are lagged CF/K, sales to capital dated t 3 and deeper, as well as lags of variables in main equation where correct