Domestic Financial Development in EMEs: Evidence and Implications Ettore Dorrucci, Alexis Meyer-Cirkel and Daniel Santabárbara III Foro de Investigadores de Bancos Centrales Miembros del Consejo Monetario Centroamericano Banco Central de Costa Rica, 27 August 2009 1
Outline 1. Motivation and objectives 2. DFD index methodology 3. Findings 4. Conclusions 2
Motivation and objectives Why an index of EME financial development, and why now? Where do EMEs stand in terms of domestic financial development compared with mature economies? Did we observe any financial catching-up of EMEs countries towards mature economies in the recent past? Does econometric analysis support the hypothesis that EMEs financial underdevelopment has been one structural factor contributing to the external imbalances? 3
Motivation and objectives What are the gaps left by the literature? IMF Financial Index (2006): Construct an index of financial development for a group of 17 OECD countries for 32 EMEs and industrialised nations Creane et al. (2007): Measure financial development for Middle East and North African countries. Use smaller set of dimensions Wide spectrum of literature proxy financial development with individual variables (e.g. La Porta et al 1998; Chinn and Ito 2005, 2007; Rajan and Zingales 2003; etc.) World Economic Forum Index (2008), provides the most comprehensive index possible, including up to 107 variables. However, it only focuses on the latest data available, with the aim to revise and update the database in years to come 4
Motivation and objectives Our contribution Developing a more precise and comprehensive definition of domestic financial development (DFD) Developing composite indexes of EME DFD financial development in 26 EMEs, using 6 mature economies as a benchmark Developing a database, consisting in 22 variables Assessing where EMEs stand in terms of financial development (broad indices) and how much progress they have made over time (narrow indices) Providing empirical evidence of the effect of DFD on global imbalances 5
Motivation and objectives Some preliminary stylized facts: FD taking place in EMEs? Funding possibilities have increased 6
DFD index methodology Step 1. Theoretical framework What was our benchmark when creating the index The OECD has put together a comprehensive and fairly detailed handbook on the construction of indices. We have used it as a roadmap in the process. Construction of an index: Step 1. Develop a theoretical framework What is badly defined is likely to be badly measured (OECD handbook, 2005) The literature on financial development in EMEs is often disappointing in terms of precision Chin and Ito (2005, p. 21): Financial development measured as activity of the stock market appears to depend upon capital account openness both individually and in interaction with the level of legal development Mendoza, Quadrini and Rios-Rull (2007, p.2): Financial development is characterised by the extent to which financial contracts are enforceable 7
DFD index methodology Step 1. Theoretical framework we try to take greater care regarding both precision and comprehensiveness Domestic financial development is the capability of one country to channel savings into investment efficiently and effectively within its own borders owing to (i) the quality of its institutional and regulatory framework, (ii) the size of its financial markets, the diversity of its financial instruments and private agents ease of access to them and (iii) the financial markets performance, e.g. in terms of efficiency and liquidity so as to have a guiding principle when setting the dimensions and choosing individual variables 8
DFD index methodology Step 1. Theoretical framework Why domestic financial development? Local actors perspective: Dimensions and variables are selected so as to reflect financial development from the perspective of an individual local economic agent. This would have to include his/her possibilities to freely access financial markets across borders Domestic markets perspective: Dimensions and variables are chosen in order to shed light on the development of the local financial market, its comprehensiveness, deepness, and safety. We basically ignore cross border financial access und usage, because we want to measure market completeness in relative terms (EMEs vs mature) 9
DFD index methodology Country coverage The group of EMEs within the G20: Argentina (ARG), Brazil (BRA), China (CHN), India (IND), Indonesia (IDN), South Korea (KOR), Mexico (MEX), Russia (RUS), Saudi Arabia (SAU), South Africa (ZAF), and Turkey (TUR). vigorous growth and economic development The group of oil and commodity exporting countries: Bahrain (BHR), Chile (CHL), Kuwait (KWT), Oman (OMP), Qatar (QAT), United Arab Emirates (UAE), and Venezuela (VEN). Role of OR and SWFs Other EMEs: Egypt (EGY), Hong Kong (HKG), Israel (ISR), Malaysia (MAL), Philippines (PHL), Singapore (SGP), Taiwan (TWN), Thailand (THA). Regional importance, development, and data availability The reference group: Some benchmark mature economies: United States, (US), Japan (JPN), United Kingdom (GBN), Euro area, including individual data on Germany (GER), France (FRA) and Italy (ITA) 10
DFD index methodology Step 2. General criterion to select variables Variables should be selected on the basis of their relevance, analytical soundness, accuracy, and accessibility. (OECD handbook, 2005) 1. INSTITUTIONS 2. SIZE OF AND ACCESS TO MARKETS 3. MARKET PERFORMANCE Quality of institutions Regulatory and Judicial framework Size of traditional private financial markets Financial innovation Possibility for residents to access finance Banks efficiency Liquidity (market turnover) Distribution of domestic asset base between the private and the official sector Level of corruption (-) Bureaucratic quality (+) Strength, impartiality and observance of the legal system (+) Investor protection (strength of minority shareholders) (+) Stock market value / GDP (+) Private bond market / GDP (+) Gross issuance of ABS and MBS / GDP (+) Number of bank branches per 100,000 inhabitants (+) Number of ATM machines per 100,000 inhabitants (+) Banks cost to income (-) Value of shares traded as a ratio of equity market capitalisation (threeyear moving average) (+) Central bank claims on the private sector over total claims on the private sector (-) Amount of public sector funding over total bank claims (-) Strength of collateral and bankruptcy laws in protecting the rights of borrowers and lenders (+) Degree of information available in lending operations (+) Total bank claims / GDP (+) Assets of non-bank financial institutions / GDP (+) Life insurance penetration (volume of life-insurance premia / GDP) (+) Non-life insurance penetration (volume of non-life-insurance premia / GDP) (+) Domestic private debt over domestic government debt (-) Efficiency in enforcing contracts and resolving commercial disputes (+) Note: Expected effect on DFD in parenthesis Cost of maintaining a savings account (annual fees) (-) 11
DFD index methodology Step 2. General criterion to select variables The dimensions to be covered Domestic Financial Development Index Institutional dimension Size and access to markets Market performance 12
DFD index methodology Step 2. General criterion to select variables Domestic Financial Development Index Institutional dimension Size and access to markets Market performance Regulatory and judicial framework Law and Order Investor protection Legal rights Enforcing contracts Credit information Quality of institutions Bureaucratic quality Corruption 13
DFD index methodology Step 2. General criterion to select variables Domestic Financial Development Index Institutional dimension Size and access to markets Market performance Size Traditional measures Stock market value/gdp Private bond market/gdp Total bank claims/gdp Assets of non-bank financial institutions/gdp Residents access to financial services Branch and ATM penetration Annual fees for savings account Life insurance penetration Non-life insurance penetration Size Financial innovation ABS and MBS, gross issuance/gdp 14
DFD index methodology Step 2. General criterion to select variables Domestic Financial Development Index Institutional dimension Size and access to markets Market performance Technical efficiency Cost/Income Liquidity Depth: Stock market turnover ratio Distribution of domestic asset base Domestic private debt/ domestic government debt Central bank claims on private sector/ total claims on private sector Bank claims on public sector/ total bank claims 15
DFD index methodology Step 3. Multivariate analysis How to normalise the data in order to construct the index? Avoiding adding apples and oranges (OECD handbook, 2005) A variety of normalisation methods exist: 1. Ranking 2. Standardisation (or z-scores) 3. Re-scaling 4. Distance to a reference max( x 5. Categorical scale Indicators (above or below the mean are assigned -1, 0, or 1) I x, c, t x c, t min( x c, t all c, t all ) ) min( x c, t all ) chooses the largest value of variable x over all time periods from the pool of benchmark countries 16
DFD index methodology Step 3. Multivariate analysis Weighting and aggregation: No matter which method is used, weights are essentially value judgments. While some analysts might choose weights based only on statistical methods, others might reward (punish) the components that are deemed more (less) influential depending on expert opinion to better reflect the policy priorities or theoretical factors. (OECD handbook, 2005) Majority of prominent indices (e.g. Human Development Index, IMF Financial Index) use an equal weighting policy Equal weighting has indeed been our approach 1. Within each of the three dimensions, individual variables also received equal weighting. 2. We assigned each of the three dimensions of our index an equal weight 17
Findings Venezuela Indonesia Russia Philippines Argentina Brazil Oman Egypt India Mexico UAE Thailand Turkey Saudi Arabia Kuwait Chile China South Africa Qatar Israel Bahrain Malaysia Euro G3 Taiwan Korea Singapore Japan UK Hong Kong US Index of domestic financial development (2006) 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 18
Finding 1: Comparison EMEs-G7 Index of domestic financial development: Ranking and scores 2006 Composite index of domestic financial development (DFD) Country Rank Score Scale 1-100 1 st dimension: Institutions and rules supporting DFD 2 nd dimension: Financial market size and access to finance Economies with high level of financial development 3 rd Dimension: Selected proxies of financial market performance Rank Rank Rank United States 1 77.3 1 1 1 Hong Kong 2 69.8 3 7 4 UK 2 69.8 4 3 11 Japan 4 66.2 5 2 22 Singapore 4 66.2 2 9 16 Korea 6 64.6 8 5 2 Taiwan 7 61.7 12 4 6 EA-G3 8 58.6 10 6 5 19
Finding 1: Comparison EMEs-G7 Index of domestic financial development: Ranking and scores 2006 Composite index of domestic financial development (DFD) Country Rank Score Scale 1-100 1 st Dimension: Institutions and rules supporting DFD 2 nd Dimension: Financial market size and access to finance 3 rd Dimension: Selected proxies of financial market performance Rank Rank Rank Malaysia 9 57.9 7 11 8 Bahrain 10 55.4 13 12 7 Israel 11 54.4 6 10 18 Qatar 12 51.8 9 20 17 South Africa 13 49.8 18 8 13 China 14 49.5 21 16 3 Chile 15 48.4 11 13 25 Kuwait 16 48.1 15 17 15 Economies with intermediate level of financial development 20
Finding 1: Comparison EMEs-G7 Index of domestic financial development: Ranking and scores 2006 Composite index of domestic financial development (DFD) Country Rank Score Scale 1-100 1 st Dimension: Institutions and rules supporting DFD 2 nd Dimension: Financial market size and access to finance 3 rd Dimension: Selected proxies of financial market performance Rank Rank Rank Saudi Arabia 17 45.9 19 26 20 Turkey 18 45.5 16 21 12 Thailand 19 45.0 20 18 9 UAE 20 44.0 26 15 14 Mexico 21 43.2 14 23 21 India 22 42.4 22 19 24 Egypt 23 42.2 29 22 23 Oman 24 41.1 23 28 10 Brazil 25 40.8 24 14 26 Argentina 26 39.6 17 29 29 Philippines 27 36.9 27 24 28 Russia 27 36.9 28 27 19 Indonesia 29 34.1 25 30 30 Venezuela 30 29.4 30 25 27 Economies with low level of financial development 21
Finding 1: Comparison EMEs-G7 Still a lot of room for EMEs financial catching up 22
.2.4.6.8 Finding 1: Comparison EMEs-G7 0 Performance sub-index.4.5.6.7.8.9 Comparison among index dimensions 2006 USA USA KOR VEN ITA ZAF BHR BRA UAE CHNKWT IND THA EGY PHL TUR MEX RUS SAU OMNARG IDN TWNFRA KOR EUR ISR MYS CHL QAT JPN GBN DEU HKG SGP VEN EGY RUS PHL CHN TWN EUR BHR FRA ITA OMN UAE THA ZAF TUR KWT IND BRA SAU MEX ARG QAT CHL MYS ISR DEU HKG GBN SGP JPN IDN.2.4.6.8 1 Institutions sub-index.2.4.6.8 1 Institutions sub-index 23
Finding 2: EMEs financial development over time How to see the evolution over time? Reduce the amount of variables taking into account data availability and our definition of development. Hence, create a narrow index of DFD Size sub-index Institutions sub-index G6 80 70 60 50 40 30 20 10 Performance sub-index EMEs 80 70 60 50 40 30 20 10 0 1991 1994 1997 2000 2003 2006 0 24
Finding 2: EMEs financial catching up Size sub-index EMEs are catching up mainly in terms of size EMEs G6 75 50 25 0 1991 1994 1997 2000 2003 2006 EMEs financial market size compared with benchmark mature economies (G7 except Canada) 25
Finding 2: EMEs financial catching up Size dimension EMEs are catching up mainly in terms of size G6 Brazil Mexico Saudi Arabia United Arab Emirates India Korea Russia China 75 50 25 0 1992 1994 1996 1998 2000 2002 2004 2006 Selected EMEs financial market size compared with benchmark mature economies (G7 except Canada) 26
Finding 3: Relevance of domestic financial development for global imbalances Weighted Average of per Capita Income Net capital flowing from EMEs to advanced countries Weighted Average of Income in Surplus and Deficit Countries 1 0.9 0.8 0.7 0.6 Group of Countries with Current Account Deficits High Income 0.5 0.4 0.3 0.2 0.1 0 Group of Countries with Current Account Surpluses 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 Low Income Surplus Countries Deficit Countries Weighted average of income in the two groups of countries with current account deficits and surpluses (1981-2008) 27
Finding 3: Relevance of domestic financial development for global imbalances How DFD could contribute to explain it Borrowing constraints Y EMEs Precautionary S S > I (X-M) Official accumulation of internationally traded assets OR SWFs Net capital inflows from EMEs HFDs I and/or C (X-M) NO borrowing constraints 28
Finding 3: Relevance of domestic financial development for global imbalances Two applications of the DFD index In our analysis the current account is driven by savings and investment: y i, t 1DFDi, t 2FOi, t 3 DFD i, t FOi, t ) jcontrol variables j ( u j i, t i, t Dependent variables (y): Current account balance, national savings, gross capital formation, public and private savings and investment (expressed in GDP ratios) Objective variable (DFD): Domestic financial development, financial openness (FO), and their interaction Control variables: Net foreign assets to GDP; income per capita relative to the United States; output growth rates; dependency ratio; trade openness; and regional dummies and time fixed effects 29
Finding 3: Relevance of domestic financial development for global imbalances Two applications of the DFD index Database: Annual data from 1985-2006 26 EMEs considered + 6 benchmark G7 countries Unbalanced panel 3-year non-overlapping averages Estimation technique: GMM estimator Robustness checks with the OLS (preferred to fixed effect estimator) 30
Finding 3: Relevance of domestic financial development for global imbalances 1. How do DFD and FO affect the current account? We analyze current account dynamics for both lenders (L) and borrowers (B): CA i, t L 1 DFD L i, t B 1 DFD B i, t L 2 FO L i, t B 2 FO B i, t L 3 ( DFD i, t FO i, t ) L B 3 ( DFD i, t FO i, t ) B j j Control variables j i, t u i, t Variables with superscript L (B) take their actual value if the country is a lender (borrower), otherwise 0 Coefficients should be interpreted as the actual effect of changes in DFD, FO and their interaction on the current account to GDP of lending (borrowing) countries Sample: EMEs + G7 31
Finding 3: Relevance of domestic financial development for global imbalances 1. How do DFD and FO affect the current account? Current account/gdp Coefficient p-value Narrow DFD Index (L) -0.155*** (0.008) Narrow DFD Index (B) -0.256*** (0.000) Financial openness (L) -0.0183* (0.100) Financial openness (B) -0.0342*** (0.001) Narrow DFD*Financial openness (L) 0.020 (0.464) Narrow DFD*Financial openness (B) 0.0662*** (0.005) Observations Number of countries Hansen test Robust p-values in parentheses, *** p<0.01, ** p<0.05, * p<0.1 Variables in italics are instrumented through the GMM procedure with their first lags Time dummies not reported 93 23 (1.000) Increasing DFD within lending countries seems to have helped channel savings from abroad to their domestic market. In borrowing countries, growing DFD also tends to attract foreign capital, thus contributing to even higher current account deficits. Higher financial openness is related to lower lending by surplus countries and higher borrowing by deficit countries. We also find some evidence of substitutability between DFD and FO in borrowing countries but not in lending ones 32
Finding 3: Relevance of domestic financial development for global imbalances 2. How do DFD and FO affect the determinants of S-I? Savings/GDP Investment /GDP Public savings/gdp Private savings/gdp Public investment /GDP Private investment /GDP Narrow DFD Index 0.023 0.223** -0.005 0.027-0.068 0.247*** (0.844) (0.048) (0.956) (0.667) (0.219) (0.009) Financial openness 0.0433** 0.047-0.001 0.0443*** 0.003 0.037 (0.045) (0.192) (0.940) (0.005) (0.790) (0.210) Narrow DFD*Financial openness -0.115* -0.088 0.006-0.121*** 0.003-0.072 (0.063) (0.365) (0.860) (0.008) (0.905) (0.376) Observations 63 63 63 63 52 52 Number of countries 18 18 18 18 14 14 Hansen test (1.000) (1.000) (1.000) (1.000) (1.000) (1.000) Growing DFD is related to increases in investment, mainly driven by private investment. Financial openness tends to increase private savings and total savings significantly, possibly because households and firms tend to save more to protect against external shocks. DFD has also the expected negative impact on private savings, but requires to be accompanied by financial openness. 33
Conclusions We construct, on the basis of an new database, composite indices to measure domestic financial development that indicates that: The scope for EMEs financial catching up is still substantial, although this scope varies significantly from country to country In several countries, we observe that successful institutional and regulatory environments of certain EMEs have not yet translated into good-sized and high-performing financial intermediaries and markets There is some indication that the financial catching up process has already started in some EMEs The econometric analysis points out that a relative increase in financial sector development is associated with: a reduction of current account balances, an increase in gross capital formation and a decrease in private savings 34
Thank you! 35
DFD index methodology Why our DFD index does not include Causal variables versus components of the Index Financial openness and capital restrictions Necessary to differentiate elements part of domestic financial market from factors relevant for their development. We accept that financial openness and development may be mutually reinforcing, nevertheless they should remain separate concepts. Macroeconomic variables (e.g. inflation, budget balance to GDP, FDI/GDP). 36
DFD index methodology Why our DFD index does not include Variables with ambiguous effects Debt securities issuance/gdp: This variable has strong non-linear effect on financial development as it includes government debt. No issuance of government debt is negative: lack of safe assets, no reliable yield curve, etc (Herring and Chatusripitak, 2000). But too much debt is negative too: risk of default, crowding out of private sector, stability concerns (Calvo 1995). Concentration and profitability of banking sector: The sign of this variable might be country specific. High concentration could imply a lack of competition, while on the other hand it could just reflect economies of scale or scope (see Demsetz 1973, 1974; Peltzman 1977) 37
DFD index methodology Step 3. Multivariate analysis Dealing with missing data: The idea of imputation of data can be both seductive and dangerous (OECD handbook, 2005) In the database, the lack of data on specific variables was evidenced in two different ways: Data were available but missing at some point in a country we used the last available data for that country. Data were not available for a country a. If there was evidence of inexistence of a financial market segment (e.g. ABS issuance), we assigned the value of zero. b. Otherwise, we looked other existent variables in the same subdimension and took the simple average of the rank achieved in those index values. We then considered the value achieved in the variable of interest for a country with the same rank, and used that value. 38
Finding 3: Relevance of domestic financial development for global imbalances An application of the DFD index Current account/gdp Coefficient p-value Narrow DFD Index (L) -0.155*** (0.008) Narrow DFD Index (B) -0.256*** (0.000) Financial openness (L) -0.0183* (0.100) Financial openness (B) -0.0342*** (0.001) Narrow DFD*Financial openness (L) 0.020 (0.464) Narrow DFD*Financial openness (B) 0.0662*** (0.005) Net foreign assets/gdp 0.028 (0.211) Government balance/gdp -0.037 (0.782) Relative per capita income to US 0.037 (0.311) Age dependency ratio -0.104** (0.010) Real GDP growth -0.022 (0.797) Trade openness 0.0358*** (0.000) Gulf Cooperation Countries 0.399*** (0.000) Commodity exporters 0.0257** (0.016) Asian financial centres -0.036 (0.361) Constant 0.134*** (0.001) Observations Number of countries Hansen test 93 23 (1.000) 39
Finding 3: Relevance of domestic financial development for global imbalances An application of the DFD index Savings/GDP Investment /GDP Public savings/gdp Private savings/gdp Public investment /GDP Private investment /GDP Narrow DFD Index 0.023 0.223** -0.005 0.027-0.068 0.247*** (0.844) (0.048) (0.956) (0.667) (0.219) (0.009) Financial openness 0.0433** 0.047-0.001 0.0443*** 0.003 0.037 (0.045) (0.192) (0.940) (0.005) (0.790) (0.210) Narrow DFD*Financial openness -0.115* -0.088 0.006-0.121*** 0.003-0.072 (0.063) (0.365) (0.860) (0.008) (0.905) (0.376) Net foreign assets/gdp 0.055 0.013 0.003 0.052 0.110*** -0.027 (0.328) (0.684) (0.916) (0.194) (0.001) (0.216) Government balance/gdp -0.504* -0.213 0.239-0.744*** -0.409** 0.251 (0.058) (0.405) (0.172) (0.005) (0.027) (0.355) Relative per capita income to US -0.141-0.191** -0.057-0.084-0.254*** -0.010 (0.102) (0.017) (0.441) (0.448) (0.000) (0.903) Age dependency ratio -0.606*** -0.450*** -0.199** -0.406** -0.356*** -0.095 (0.001) (0.002) (0.026) (0.028) (0.005) (0.312) Real GDP growth 0.706*** 0.960*** 0.280** 0.427** 0.140 0.804*** (0.001) (0.000) (0.019) (0.028) (0.243) (0.001) Trade openness 0.0854*** 0.025 0.0630*** 0.022 0.0556** -0.035 (0.000) (0.368) (0.007) (0.122) (0.029) (0.125) Gulf Cooperation Countries 0.461*** -0.006 0.203*** 0.258*** 0.123* -0.136 (0.000) (0.937) (0.004) (0.003) (0.070) (0.192) Commodity exporters 0.0716* 0.030 0.0584* 0.013 0.0398** -0.016 (0.053) (0.273) (0.077) (0.365) (0.011) (0.321) Asian financial centres -0.135-0.067-0.194* 0.059 (0.356) (0.603) (0.065) (0.507) Constant 0.510*** 0.349*** 0.099 0.411*** 0.280*** 0.114* (0.000) (0.000) (0.106) (0.001) (0.001) (0.066) Observations 63 63 63 63 52 52 Number of countries 18 18 18 18 14 14 Hansen test (1.000) (1.000) (1.000) (1.000) (1.000) (1.000) 40