Comments on “Financial Development_ Structure and …

Comments on“Financial Development: Structure and Dynamics”by Augusto de la Torre, ErikFeyenand AlainIze
Conference on Financial Structure and Economic DevelopmentWorld Bank, Washington, DCJune 16, 2011MartinCihakLead Economist, WB/FPDCE
A very relevant paper
Outlines a comprehensive frameworkfour fundamental types of frictionsassociated market failures / paradigmsShows some interesting empirical resultsincluding a useful regional study on LAC countriesParticularly relevant for work on WBG’s newGlobal Financial Development Reportshould include a part on “benchmarking” financial sector development in countries around the globe
The findings are plausible
The notion that financial sector development is nonlinear, bumpy, and that there is a both a “bright” and “dark” side to financeseems veryplausible.It is hard to argue againstideas of path-dependenceand leapfrogging (e.g.: checks, credit cards, mobile banking,…).A number of observations that sound realistic (and consistent with recent evidence):much of financial development isexplosivebut it has decreasing impacts (returns) on real developmentfinance resembles “luxury good”(its use explodes as income rises, yet its benefits exhibit declining returns).finance may become excessive at some pointthispoint is influenced by quality of policies to reduce the risks
The paper has ambitious goals
It attempts to isolate the policycomponentoffinancial development(“enablingenvironment”) fromthe (exogenous component of the) outcome financial indicator.This is a difficult task; there is little consensus on the appropriate methodology
Main comments
The link between the theoretical and the empirical parts could be strengthenedThe empirical part, as the authors say, is at this point more an illustration than in-depth econometric analysis.Need some more clarity on the hypotheses that are being tested, what is the test, rejected/notrejected,at what level.More clarity on the caveats/weaknesses of data.
Patterns of financial development
Financial development described as a largely logical progression“Gradual grinding down/easing of frictions”Butnew frictionscan emerge.Focus on “leapfrogging”, but there may also be “backpaddaling” and dead endsResultslargely driven by cross-country variation yet interpreted as developments over time“government borrowing appears early in the game”, “bank deposits emerge before credit”, etc.yet data don’t go before 1980s
Choice of variables, data
Overall, the choice of variables is reasonable, given the data constraintsIndicators of reach, efficiency and liquidity, globalization, and soundnessSomeindicators could be added,e.g., on access to financeThe indicators are analyzed one by one. How to come up with a comprehensive assessment?But need more caveats on underlying dataImprecise proxies (net margins vs. efficiency)The data are basic aggregate ratios (no differentiation within systems; do not capture evolution in financial technology--e.g. checks/cards/mobile payments)Missing observations, issues of comparability (accounting standards etc)Data adjusted for a number of controls, including GDP/capitaThe inclusion of GDP/capita is problematicIt one of the “real” development variables that are presumably affected by financial development (=>causality/endogeneityissues)
Econometrics: panel vs. cross-section?
Instead of panel regressions, the authors opt to run cross-sectional regressions (on medians), and to show graphically a set of paths around that crosssection.Thisapproach allows for an interesting graphicalpresentation.Butsome info about the dynamics of the systems may be lost by not usingthe panel datafor a panel regression (# of observationscouldincrease from~100toa few thousand).Thiscould potentially better explain the impact of changes in ininstitutional/legal/regulatoryframework,innovation in globalfinance, andother typesshocks.
Additional econometric comments
Core of the empirical analysis: mapping gaps into forcing variables (Tab 6):Howdoes the regression technique deal with possible measurement errors in the dependent variable (=estimated “gap”)?Are the “gaps” comparable across countries? (The gap in finance for each country is given by the distance of the indicator from the “norm”, which comes from adjusting country medians on economic development and other country-specific characteristics. )Towhat extent does the lack of significance on some of the Enabling Environment Indicatorsreflect thatthey are correlated with realGDP growth or that they may also explain the incidence of credit crunches?Credit crunches can also be approximated as inClaessens,KoseandTerrones(2011):CC= bottom 25% of world sample of credit declines; severe CC= those in bottom 12.5 %.Given the high concordance between housing prices and credit (Claessenset al. 2011), housing price busts could also be defined and included in the regression.Analogously, for the regressions on stock markets, the authors could also include equity price busts (based on real stock prices) or real currency busts (based on REERs).To what extent is the distance from the norm explained by the set of forcing variables?For example,in Table 9, it would be useful to show not only the explained gaps but also the actual gap.
Things toadd in the explanation of gaps in financial development…
Influenceof explicit policy actions (e.g. financial liberalization) on reducing gaps.Externalshocks (e.g., swings in risk appetite, changes in world interest rates) that typically affect the inflow of foreign capital and, hence, the depth of domestic financialmarketFeaturesof the broader policy environment (e.g., exchange rate regime, capital market openness—this may have an impact on financial dollarization, FX markets, FXhedinginstruments, and ability to manage FXrisks)Featuresof the market structure (e.g., market concentration – may have an impact on the speed of setting up credit info sharing systems, degree of stateownership)Ownership(state, foreign, domestic private– may have an impact on speed of introduction of newtechnologies)Whatabout appearance of derivates and sophisticated products (CDO, CDO-sq, CDS, etc.)
Summing up…
A very useful paper, both the theoretical part and the empirical part.Thelink between thetwo partscould be strengthenedAlso, clearer caveats on data weaknesses.And moreclarity on the hypotheses that are beingtested.