Evaluating the Financial Systems of Emerging Market Economies: Evaluating the Financial Systems of Emerging Market Economies Applied General Equilibrium Development Economics Robert M. Townsend The Marschak Lecture The Econometric Society Bogota, Colombia October 5th, 2007


Outline of the Lecture: Outline of the Lecture The importance of measurement, and actually doing it: from individuals to the national economy A Research Algorithm – Application: Thailand General Equilibrium CME Benchmark Standards: anomalies, but with some successes Policy Impact: financial institutions help move toward standard Salient Patterns in the Data: national, regional village, household Micro and Macro Decompositions: quasi-analytic, what is important Applied General Equilibrium Models: estimation, simulation Policy Analysis: welfare gains Anomalies: next steps Applied General Equilibrium Development Economics


Slide3: Research Algorithm MEASUREMENT: COMMON MACRO/MACRO PLATFORM AND ACTUALLY DOING IT


Slide4: [The Circular Flow. Source: Colander's Economics, Microeconomics, and Macroeconomics, 5th Edition by David C. Colander, ©2004 McGraw Hill/Irwin The national income accounts and the associated "circular flow" diagram envision little production in the household sector.


Slide5: Even as constructed, non farm proprietary income has been large relative to other factor payments in the data. Nonfarm proprietary income still dominates corporate profits, for example.


Slide6: . TOWNSEND THAI SURVEYS


Slide7: [Source: Samphantharak and Townsend (2006)] The national income accounts are based on corporate financial accounts. These distinguish stocks in the balance sheets from cash flow, which is distinguished in turn from (accrued) income.


.: . Village Level Work with Archawa Paweenawat


.: . . Thai Research Project Platform: Data Base Archive for Research


.: . National Accounts Underlie Computable General Equilibrium Model But be careful about aggregated production function Households as Firms- Heaton, Lucas, Vissing-Jorgensen, DeNardi, Basseto Be careful about presumed household economies Caveats: the difficulty of measurement Sectors Services Quality Prices Consumption versus investment Gifts


.: . Also on the wish list: Better Measurement Combo Surveys of Firms and Owners


Research Algorithm continued… Neoclassical Anomalies: Research Algorithm continued… Neoclassical Anomalies If markets and institutions were perfect and there were no policy distortions, then certain benchmark standards would be implied. From GE models of entire economy, but could be kinship group, local village, region for example. May work for consumption but not investment; some shocks and not others Looking at micro evidence for appropriate assumptions about market/institutional structure These diagnostics tests are relatively easy to do and can be part of a Financial Sector Evaluation


Slide13: Relative to these benchmarks there are many anomalies in the Thai economy, even for those using formal credit and savings instruments Many households and businesses appear to be constrained in occupation choice Rates of Return Decline With Wealth for Constrained Business and are low for wealthy unconstrained. May hold cash and low return assets. Poor households and SME enterprise are particularly vulnerable in consumption and investment to variation in income and cash flow Evidence of Incomplete markets – rubber price shocks not covered


The Risk Sharing Equation : The Risk Sharing Equation


IMPORTANT EXCEPTIONS: IMPORTANT EXCEPTIONS


Slide16: Aggregation applied at subunits networks within village and related industrial conglomerates [Family Networks in Villages. Source: Townsend Thai Data research with Samphantharak] [Examples of Simple Group Structures. Source: Samphantharak (2003)]


RAINFALL IS COVERED: RAINFALL IS COVERED [Average Rainfall in Thailand. Source: Data from Thai Meteorological Department]


Research Algorithm continuedECONOMY IS NOT NEOCLASSICAL- POLICY IMPACT : Research Algorithm continued ECONOMY IS NOT NEOCLASSICAL- POLICY IMPACT Government program innovations and plausibly exogenous variation in access to intermediation have had nontrivial impacts on households and businesses. Often not simply an effect of wealth transfer Good financial institutions bring us back toward neoclassical standard Exploiting exogenous variation or likely instruments


Slide19: (8.1.5) (8.1.3) The new one million baht village funds program seems to have increased consumption, profits from businesses, labor income, agricultural investment, and total borrowing above and beyond village fund credit, while raising default rates and lowering assets/savings. Instrument is inverse number of households per village (with Kaboski)


Slide20: Arguably, exogenous variation in villages funds by: Policy (emergency services training, monitoring, pledged saving, collateral) Type (rice bank, buffalo bank, production credit group, women’s groups) Implies variation in impact (asset accumulation, risk sharing, occupation choice, and reliance on money lenders). Variation is coming from ministries promoting different institutions and not tracking failures


Slide21: WITH MAURO ALEM Instrumented variation includes whether headman says village has access, distance to district center, surprise local variation, for the impact on consumption and investment smoothing.


Slide22: Caveats General Equilibrium Effects - wages (treatment on non-treated) Instruments without Modeling - (with Sergio Urzua)


Salient Facts:Macro: Growth, Inequality Poverty, Financial Deepening: Salient Facts: Macro: Growth, Inequality Poverty, Financial Deepening


Slide24: Growth has been relatively high for the past 50 years, but with a sharp drop in 1997 and the recession in the years of following the financial crisis. But the trend of long term industrialization dominates the data. Caveat: this is not Latin America or Africa


Slide25: Inequality by almost all measures has been increasing since at least 1976, along with income, but unlike the growth of income, inequality peaks in 1992, with some backtracking for the crisis


Slide26: There has been a steady decrease in the faction of poor and distance of the poor from the poverty line, with only a slight wobble in the crisis. However, in panel data poverty is shown to be a transient rather than chronic phenomenon, especially if income data are used. [Summary Statistics of Income in Thai SES. Source: Jeong (2000)]


Slide27: Financial deepening displays astounding trends relative to the U.S. There was repression. Then part of the increase starting in 1986 can be attributable to financial liberalization


Slide28: Variation in Policy comes from this history Repression Liberalization Village funds of different ministries By the 1990's commercial bank regulation appears deficient and government transfers masked the distortion. Crisis and recession Post crisis, the government involvement in the financial sector has increased (million baht fund). BAAC debt moratorium


Salient Facts continued:From Macro Back Toward Micro Relative to Cross Country Comparisons : Salient Facts continued: From Macro Back Toward Micro Relative to Cross Country Comparisons


Decompositions Quasi Analytic- Both Micro and Macro: Decompositions Quasi Analytic- Both Micro and Macro List of common factors/variables; financial intermediation is among Others are occupation choice, education, urbanization Caution: these are like correlations, but some causality was established earlier


Slide35: Micro Kuznets decompositions Increasing access/use of the formal sector along with high and increasing income differentials account for a nontrivial part of growth of per capita income and increasing inequality, albeit with other factors (Jeong thesis)


Slide36: [Decomposition of Inequality Change. Source: Jeong (2001)] Chapter 4 – Equations (4.2.2) (4.2.4) (4.2.5) (4.2.6) (4.2.6) (4.2.6)


.: . Macro, total factor productivity is largely explained, It is NOT an unmeasured residual aggregate shock Access-no access dichotomy is used- (with Hyeok Jeong)


Importance: Caveat: unobserved heterogeneity, talent GOAL: do all this in one economy Importance Causes of growth and favorite sons/daughters: FINANCIAL INTERMEDIATION OCCUPATION CHOICE/INDUSTRIAIZATION EDUCATION URBANIZATION TECHNOLOGY, OPENNESS? INFRASTRUCTURE/INSTITUTIONS? CORRUPTION , POLITICAL ECONOMY?


Applied General Equilibrium Models : Applied General Equilibrium Models Thai economy as an integrated micro/macro system, with the choices of diverse individual agents aggregated up to explain macro variables. Start with dual models (perfectly intermediated sector vs. financial autarky) Algorithm continued MODEL BUILDING BASED ON SALIENT PATTERNS, FACTORS


.: . With Xavier Giné Parameters Estimated with Micro data WEALTH COSTS


Slide41: A model of occupation choice WITH EXOGENOUS FINANCIAL DRIVER and occupation choice explains well the upturn in the Thai economy at the time of a financial liberalization


Slide42: The general equilibrium effect of price changes from financial liberalization can cause losses for existing firms that use unskilled labor.


.: . . FINANCIAL DEEPENING With Kenichi Ueda


Slide44: A model with endogenous financial access delivers observed long term trends but not that upturn - not a stationary time series Need to take into account financial policy intervention


.: . . [Use of savings. Source: Townsend and Ueda (2007)]


Slide46: Add Kenichi slide on finacial liberalization [Welfare gains from reduction of marginal costs (left) and from reduction in entry costs (right). Source: Townsend and Ueda (2007)] Be careful about difference and differences Growth and welfare are not synonymous


From Macro to Regional: From Macro to Regional


Slide48: PREDICTION ERROS - GJ PLUS SLIDE OF TRANSACTIONS COSTS BY TYPE OF BANK [1996 GJ Access Index Simulation Differences. Source: Felkner and Townsend (2004)]


.: . . [Financial Deepening Simulation - k^ Defined by Actual Wealth Distribution and Participation Rate. Source: Felkner and Townsend (2004)]


Slide50: . [ Source: Jeong and Townsend (2005)] Down to Household-Level Anomalies MODEL ECONOMY THAI ECONOMY


Slide51: Moving beyond Dual Sector Models Endogenous Wealth Creation Alleviates Constraints (Karaivanov, Buera and Shin) Simple borrowing limits


Slide52: With Joe Kaboski Village fund with particular obstacle


Choices are shown to be constrained by real obstacles to trade. There seems to be MORAL HAZARD in entrepreneurial effort and project choice. A further example is MONITORING by joint liability borrowers. There seems to ADVERSE SELECTION, the exclusion of safer customers from the loan market. There seem to be LIMITED COMMITMENT problems, with loan size limited by collateral or wealth, and a tendency for strategic default limited by unofficial sanctions.: Choices are shown to be constrained by real obstacles to trade. There seems to be MORAL HAZARD in entrepreneurial effort and project choice. A further example is MONITORING by joint liability borrowers. There seems to ADVERSE SELECTION, the exclusion of safer customers from the loan market. There seem to be LIMITED COMMITMENT problems, with loan size limited by collateral or wealth, and a tendency for strategic default limited by unofficial sanctions. Adding Impediments to Trade and Distinguishing Work with Paulson, Karaivanov, Ahlin,


.: . Dynamic Mechanism Design The Pareto frontier With Alex Karaivanov


Slide55: Vuong tests distinguishing regimes


Slide56: With Sergey Mityakov and Juliano Assuncao More Work on the Supply Side - Industrial Organization


Slide57: Other Related Efforts There are relatively few contributions of this kind, and practically none in developing countries. Banerjee and Duflo: cross-country growth dynamics and TFP pioneered by Lucas among others are hard to reconcile with an aggregated production function, that is, as if the neoclassical framework were assumed to cover the micro data. Build toward a new micro-founded model with a small number of alternative technologies and varying fixed costs. They view their contribution as a preliminary attempt but of interest precisely because there are few other studies and almost none which combine micro estimates with endogenous growth and inequality dynamics. Clearly progress can be made: Heckman, Cameron, and Taber study wage dynamics and inequality in dynamic general equilibrium models estimated with US data Cagetti and DeNardi study entrepreneurial wealth in inequality in the U.S. with structural g.e. models. Some of the asset pricing literature is solidly in this tradition (Hansen, Cochrane, Singleton, Lucas). Real business cycle literature International Trade


Slide58: Applied General Equilibrium Development Economics The whole may be greater than the sum of the parts There is relatively little work in development that combines micro economics and macro economics Relatively little work that combines both theory and data Both the micro and macro data are put into a common framework for measurement Various theories can be rejected in the data (fathering further rounds of iterative research agenda) Modified and new theories which link growth, inequality, poverty, and financial deepening. Research to assess and quantity the heterogeneous impact of financial policy change at the level of households and firms while being consistent with the facts of growth, inequality, and poverty.


Slide59: Green BAAC: All Regions Both Commercial: All regions Green Baac: One Region With Sergey Mityakov and Juliano Assunçao