## _wp0862

## Source details

**Canonical URL:** [_wp0862](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp0862.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp0862.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp0862.pdf.json)

---

### Introduction and motivation
- Many studies examine financial development and growth, but fewer assess whether financial development helps reduce poverty (exceptions: Dollar and Kraay, 2002; Honohan, 2004; Beck, Demirgüç-Kunt, and Levine, 2007).
- Financial development can affect poverty:
  - Indirectly by stimulating economic growth.
  - Directly by facilitating transactions and providing savings products that increase income and enable investment.
- Financial instability can undermine poverty reduction through direct disruptions to financial services and indirect macroeconomic volatility.

### Paper goals
- Explore whether better access to savings or credit opportunities (with a nonnegative real return) is the main direct channel through which financial development alleviates poverty.
- Assess whether financial instability is detrimental to the poor and thus weakens the beneficial impact of finance on the poor.

### Identification strategy and distinguishing features
- Distinguish direct effect of financial development on poverty from indirect effect through economic growth.
- Two distinguishing features; documented feature in the excerpt:
  - Focus on channels (credit or money) through which the poor benefit from formal financial intermediation, emphasizing McKinnon (1973) “conduit effect” (liquidity/savings channel).
- The analysis examines both a credit indicator and an indicator of liquidity (M3/GDP), not just credit.

### Theoretical channels: McKinnon conduit effect and alternatives
- Competing mechanisms:
  - Greenwood and Jovanovic (1990): early-stage financial development may exclude the poor due to high unit cost of small loans (possible inverted U-shape in inequality).
  - McKinnon conduit/complementarity view: higher real return on holding money increases self-financed investment by lowering opportunity cost of internal saving.
- Additional conduits:
  - Deposits (demand/savings with nonnegative real remuneration).
  - Credits (as systems mature, small-loan costs may be borne or pooled).

### Why financial instability disproportionately hurts the poor
- Direct effects:
  - Payment system disruptions and bank closures harm poor households lacking asset diversification.
  - Loan rationing disproportionately affects small borrowers.
  - Uncertain deposit liquidity dampens the conduit effect.
- Indirect effects via macro volatility:
  - Financial instability increases volatility of investment and relative prices, making growth more volatile and reducing average growth.
  - Growth volatility and asymmetric downturns raise poverty more than expansions reduce it.
  - Inflation volatility and unanticipated inflation can hurt the poor when incomes are not fully indexed.
  - Human capital investments (health, education) fall during income drops, harming long-term prospects.
- Heterogeneity: some countries show redistribution offsetting mean-income declines, but evidence is mixed.

### Empirical model specification
- Baseline: the average per capita income of the poorest 20 percent (log) is explained by:
  - Level of real GDP per capita (y),
  - Level of financial development (Fd),
  - Instability of financial development (Fi),
  - Inflation rate (Infl),
  - Country-specific effect u and error ε.
- Variables for financial development and instability are measured as averages over five years (the year of the poverty measure and the four previous years).
- Additional specifications:
  - Interactions of liquidity/credit with level of development or bank branch coverage.
  - Dynamic specification including initial poverty to test convergence.
  - Inclusion of growth, growth volatility, growth asymmetry, and inflation volatility to probe indirect channels.
  - Robustness tests with microeconomic determinants (education, government consumption, trade openness, legal/infrastructure variables, land distribution, climatic shocks).

### Data, sample, and estimation approach
- Sample focus: developing countries only.
- Poverty measures:
  - Average per capita income of the poorest 20 percent (log), constant 1985 USD (Dollar and Kraay database): selected sample of 75 developing countries and 187 observations between 1966 and 1999.
  - Headcount poverty: share earning less than $1 a day (1993 PPP): selected sample of 65 developing countries and 121 observations for 1980–2000.
  - Poverty gap (percent of poverty line) also used.
- Financial variables: two indicators—M3/GDP (liquidity) and Credit/GDP (private credit).
- Instability measures:
  - Preferred: average absolute residuals from OLS regression of xt on xt-1 and a time trend (five-year average of absolute residuals).
  - Alternative: standard deviation of the variable growth rate.
- Estimation: panel data with system GMM (Blundell and Bond 1998) and OLS for baseline; Hansen test and AR(2) serial correlation tests reported.

### Financial development indicators used (definitions preserved)
- M3/GDP: ratio to GDP of liquid assets of the financial system, M3 (currency plus demand and interest-bearing liabilities of banks and non banks).
- Credit/GDP: ratio to GDP of credits granted by financial intermediaries to the private sector (excluding public sector credit).

### Key empirical findings — summary
- Financial development is on average good for the poor; financial instability is harmful.
- Positive result for the poor holds primarily when financial development is measured by M3/GDP (liquidity ratio), supporting the McKinnon conduit effect.
- Credit/GDP is generally not significant for the income of the poor in system GMM estimations, suggesting limited direct access to credit for the poor in developing countries.
- Banking geographical coverage (number of bank branches per km2) reinforces the favorable impact of liquidity and can make credit more beneficial where branches are widespread.

### Results — income of the poorest 20 percent (selected exact coefficients and findings)
- Table 1 (M3/GDP specifications): reported M3/GDP coefficients across models include: 0.331; 0.525; 0.069; 0.447; 0.350; 0.380; 0.359; 0.450; 0.350; 0.388; 0.358; 5.994.
- Instability of M3/GDP coefficients include: -3.746; -4.627; -4.137; -4.101; -4.028; -4.680; -3.931; -3.388; -3.222; -104.988.
- Table 2 (Credit/GDP specifications): Credit/GDP coefficients include: 0.296; 0.169; 0.074; -0.162; 0.120; -0.102; -0.085; -0.161; -0.138; 0.108; -0.327; 9.502.
- Instability of Credit/GDP reported values include: 2.825; 3.798; -0.762; 4.164; 3.585; 3.792; 5.906; 4.343; 3.940; -5.510.
- Evidence of convergence when including lagged income: example coefficient of convergence in column 5 of Table 1 is –0.6 with a t statistic equal to –5.30.
- Elasticity of income of the poor with respect to mean income is not significantly different from 1 in almost all estimations (consistent with Dollar and Kraay, 2002).

### Headcount poverty index (selected findings)
- Financial development is negatively associated with the poverty headcount (Table 3).
- Financial instability increases the headcount (Table 3).
- Credit level and instability remain insignificant in system GMM estimates for headcount (Table 4), though OLS shows some significance at 10 percent.
- Inflation instability increases poverty incidence (column 7 of Tables 3 and 4).
- Some counter-intuitive coefficients for growth instability and asymmetry may reflect pro-poor policies, social safety nets, or informal coping mechanisms.

### Poverty gap and robustness
- Poverty gap results (Tables 5 and 6) corroborate other poverty measures:
  - Financial development associated with reductions in the poverty gap.
  - Financial instability associated with increases in the poverty gap in some specifications.
- Robustness tests:
  - Controls added include M3/GDP square, instability of agricultural value-added, Gini of land distribution, government consumption, civil liberties, primary enrolment, trade openness, road density.
  - Financial development and financial instability generally retain expected signs and significance in most regressions.
  - Significant control variables in some specifications:
    - Trade openness and higher government consumption tend to worsen poverty.
    - Infrastructure availability and human capital accumulation tend to alleviate poverty.
  - Outlier removal confirms main findings.
  - Alternative instability measure (std dev of growth rate of M3/GDP):
    - Financial instability is negatively associated with log income of poorest 20% (at the 0.05 level).
    - Financial instability is positively correlated with headcount index (at the 0.10 level).
    - No effect on the poverty gap.
    - The impact of financial development on the poor remains broadly similar.

### Economic relevance and channels (preserved exact summary Table 11 and quantitative impacts)
- Table 11: selected estimated coefficients preserved exactly as reported
  - Impact of Financial development (M3/GDP)
    - On Economic growth: 0.250*** (Column 1, Table 7)
    - On Financial instability: 4.897*** (Guillaumont and Kpodar (2006))
    - On Income of the poor: 0.447* (Column 4, Table 1)
    - On Headcount poverty: -0.326*** (Column 4, Table 3)
    - On Poverty gap: -0.148** (Column 4, Table 5)
  - Impact of Financial instability (Instability of M3/GDP)
    - On Economic growth: -1.557** (Column 1, Table 7)
    - On Income of the poor: -4.627** (Column 4, Table 1)
    - On Headcount poverty: 1.590* (Column 4, Table 3)
    - On Poverty gap: 1.011* (Column 4, Table 5)
  - Impact of Economic growth (GDP per capita growth)
    - On Income of the poor: 1.023*** (Column 4, Table 1)
    - On Headcount poverty: -0.175*** (Column 4, Table 3)
    - On Poverty gap: -0.086*** (Column 4, Table 5)
  - Significance notation: * significant at 10%; ** significant at 5%; *** significant at 1%
- Reported quantitative scenario (as reported):
  - An exogenous 10 percentage point increase in M3/GDP would:
    - Directly produce a 4.5 percent increase in the average income of the poorest 20 percent.
    - Increase economic growth by 2.5 percent, with an equivalent rise in the average income of the poor.
    - Generate additional financial instability of about 0.49, which translates into:
      - A 0.76 percent drop in both economic growth and the average income of the poor.
      - Additional effects causing the income of the poor to decrease by 2.26 percent.
    - Net effect: a 10 percentage point increase in M3/GDP would net out to about a 4 percent increase in the average income of the poor.
    - For the same increase in M3/GDP:
      - Headcount poverty would have decreased by 2.8 percentage points.
      - Poverty gap would have been narrowed by 1.1 percentage points.

### Mechanism and interpretation
- Dominant channel: McKinnon conduit effect — savings mobilization and liquidity (M3/GDP) provide direct benefits to the poor by raising real returns to savers and enabling self-financed investment.
- Credit tends to benefit the poor later in financial development; credit access remains limited in many developing countries.
- Financial instability undermines the conduit effect and offsets poverty-reducing benefits.

### Policy implications and recommendations (preserved phrasing and priorities)
- Financial liberalization and reforms should explicitly account for the risk of financial instability because instability can dampen or cancel poverty-reducing impacts.
- Reforms to liberalize the financial sector and foster intermediation should be accompanied by:
  - sound macroeconomic policies;
  - gradual external openness;
  - firm banking supervision.
- Reform sequencing:
  - First-generation reforms to eliminate financial repression by liberalizing interest rates should take priority over second-generation reforms to improve access to affordable finance via institutional lending-system improvements.
  - At early stages, improving savings mobilization is effective for poverty alleviation when banks are reluctant to lend to the poor; saving accumulation will later improve access to credit.
- Encourage microfinance lending to the poor, noting credit growth does not necessarily benefit them directly, especially with geographically concentrated bank branches.
- Advocate for a “distributed architecture, with larger institutions such as banks, MFI-network umbrella organizations, or the post office taking the contract and subcontracting parts of it to rural agencies, including MFIs.” (Honohan and Beck (2007) paraphrase).

### Annex I — selected descriptive statistics (figures preserved exactly)
- Sample coverage:
  - Sample of 75 developing countries for income of the poorest 20% (1966-2000).
  - Sample of 65 developing countries for headcount index (1980-2000).
- Selected descriptive figures (rows preserved exactly as in source; first number = observations):
  - Mean income of the poorest 20%  
    - 187 670 502 42 2882
  - GDP per capita (constant 1985 USD at PPP)  
    - 187 2461 1871 329 11738
  - Headcount index  
    - 121 0.258 0.214 0.001 0.823
  - M3/GDP  
    - 187 0.365 0.214 0.073 1.419  
    - 121 0.355 0.190 0.115 1.294
  - Instability of M3/GDP  
    - 187 0.023 0.018 0.004 0.126  
    - 120 0.027 0.027 0.004 0.183
  - Credit/GDP  
    - 166 0.210 0.164 0.012 0.993  
    - 116 0.203 0.151 0.014 0.745
  - Instability of Credit/GDP  
    - 153 0.016 0.012 0.004 0.076  
    - 113 0.017 0.012 0.000 0.059
  - Inflation rate  
    - 155 0.207 0.258 -0.040 1.727  
    - 116 0.398 1.337 0.006 12.642
  - Growth  
    - 187 4.705 4.752 -12.169 21.152  
    - 120 3.922 3.409 -6.516 18.829
  - Gini index of land distribution  
    - 144 66.52 16.59 31.21 93.31  
    - 91 66.71 16.98 31.77 93.31
  - Primary school enrolment rate (Education)  
    - 170 0.927 0.231 0.245 1.376  
    - 120 0.915 0.251 0.245 1.341
  - Number of bank branches per km2  
    - 138 9.635 13.383 0.109 71.921  
    - 85 7.107 10.007 0.109 47.461
  - Trade openness (%)  
    - 184 57.67 32.23 8.17 181.82  
    - 120 58.22 28.93 9.50 137.91

*Source: _wp0862 - References (excerpt provided).*

### References .............................................................................................................

### _wp0862 - References .............................................................................................................

### Introduction
- Many studies consider the relationship between financial development and growth, but the question of whether financial development helps reduce poverty has received limited empirical attention (exceptions include Dollar and Kraay, 2002; Honohan, 2004; and Beck, Demirgüç-Kunt, and Levine, 2007).
- Growth is a powerful way to reduce poverty (Bruno, Ravallion, and Squire, 1998), but increases in inequality can undermine the benefits of growth for the poor (Kanbur, 2001).
- Financial development can affect poverty through multiple channels:
  - Indirectly by stimulating economic growth.
  - Directly by facilitating transactions and allowing the poor to benefit from financial services (primarily savings products) that increase their income (through interest earned) and enhance their ability to undertake profitable investments and other activities.
- Financial instability can undermine poverty reduction because the poor are generally more vulnerable than the rich to unstable and malfunctioning financial institutions and indirectly through negative macroeconomic impacts (e.g., volatility of growth, inflation).

### Paper goals
- The paper aims to:
  - Explore whether better access to savings or credit opportunities (with a nonnegative real return) is the main direct channel through which financial development works to alleviate poverty.
  - Assess whether financial instability is detrimental to the poor and thus weakens the beneficial impact of finance on the poor.

### Identification strategy and distinguishing features
- The authors seek to distinguish the direct effect of financial development on poverty reduction from its indirect positive effect through economic growth.
- Two features set this analysis apart from earlier work. The text supplied documents the first feature:
  1. Focus on the channels (credit or money) through which poor people benefit from formal financial intermediation. The analysis emphasizes the “motive of finance” for money demand suggested by Keynes (1937) and rehabilitated by McKinnon (1973) when he presented the “conduit effect.” This conduit-effect perspective assumes that even if financial institutions do not provide credit to the poor who must self-finance investment, they are useful because they offer profitable financial opportunities for savings.
- Consequently, the paper examines both a credit indicator and an indicator of liquidity, not just credit as in Dollar and Kraay (2002), Honohan (2004) and Beck, Demirgüç-Kunt, and Levine (2007).

*Source: _wp0862 - References*

### 2. We also recognize that financial development is accompanied by crises that are likely to

### _wp0862 - 2. We also recognize that financial development is accompanied by crises that are likely to

### Overview and research question
- Purpose: Reconcile two schools of thought — (a) financial development has a positive effect on growth and may benefit the poor, and (b) credit growth predicts banking and currency crises (Loayza and Ranciere, 2006).
- Strategy: Include both an indicator of financial development and an indicator of financial instability in regressions explaining poverty.

### Theoretical channels: McKinnon conduit effect and alternative views
- Main hypothesis: Financial development can directly improve the well-being of the poor beyond its effect through economic growth.
- Two competing mechanisms:
  - Regressive/early-stage view (Greenwood and Jovanovic, 1990): high unit cost of small loans and setup costs can exclude the poor initially, potentially producing an inverted U-shape in income inequality as financial development proceeds.
  - Conduit/complementarity view (McKinnon, 1973): money and capital are complements; higher real return on holding money increases self-financed investment by lowering the opportunity cost of internal saving, enlarging the financial “conduit” for capital accumulation.
- Additional conduits:
  - Deposits: banks offer demand or savings deposits with nonnegative real remuneration for poor households.
  - Credits: as financial systems mature, banks and formal institutions may bear costs of small loans or enable pooled loans to the poor (Rajan and Zingales, 2003; Mosley, 1999).
- Empirical literature supports the complementarity hypothesis that demand for real money balances depends positively on real income and real returns and that the investment ratio is positively related to the real return on money balances (multiple cited studies).

### Why financial instability disproportionately hurts the poor
- Direct effects:
  - Disruptions of the payment system and bank closures harm poor households who cannot diversify assets or access foreign banks.
  - Banks in difficulty ration small loans; small borrowers are less profitable and have little negotiating power.
  - The McKinnon conduit effect is dampened or cancelled when deposit liquidity is uncertain.
- Indirect effects via macroeconomic volatility:
  - Financial instability exacerbates volatility in the investment rate and in relative prices (real exchange rate), making growth more volatile.
  - Growth volatility reduces average growth and thereby impedes durable poverty reduction (Ramey and Ramey, 1995).
  - Poor are more vulnerable to asymmetric effects of downturns (de Janvry and Sadoulet, 2000); regressions raise poverty more than expansions reduce it.
  - Inflation volatility and unanticipated inflation can hurt the poor when state-determined incomes are not fully indexed (Easterly and Fischer, 2001).
  - Drops in real income can lower poor households’ investment in health and education, reducing human capital.
- Heterogeneity: effects may differ across countries; evidence from several African countries indicates redistribution effects that sometimes mitigate mean-income declines (Christiaensen, Demery, and Paternostra, 2003), but Ravallion (2001) finds “no sign that distributional change helps protect the poor during contractions in average living standards” (p. 1806).

### Empirical model specification
- Baseline logic:
  - The average per capita income of the poorest 20 percent is explained by:
    - Level of real GDP per capita,
    - Level of financial development,
    - Instability of financial development,
    - Inflation rate (control).
- Basic equation (described in text):
  - Poverty indicator Pv is a function of y (GDP per capita), Fd (level of financial development), Fi (level of financial instability), Infl (inflation rate), country-specific effect u, and error term ε. Variables for financial development and instability are measured as averages over five years (the year of the poverty measure and the four previous years).
- Additional specifications considered:
  - Interaction of liquidity/credit levels with level of economic development or access to banking (geographical coverage of bank branches) to identify where McKinnon conduit is most relevant.
  - Dynamic specification including initial poverty level to test convergence effects (as in Beck, Demirgüç-Kunt, and Levine, 2007).
  - Inclusion of economic growth, growth volatility, growth asymmetry, and inflation volatility to test whether financial instability’s impact on the poor operates through macroeconomic volatility.
  - Robustness tests adding microeconomic determinants of poverty: primary education, government consumption, trade openness, legal environment, infrastructure, land distribution, climatic shocks.

### Data, sample, and estimation approach
- Sample focus: developing countries only (to reduce heterogeneity and because McKinnon conduit plays out in low financially developed economies).
- Poverty measures used:
  - Average per capita income of the poorest 20 percent (logarithm), in 1985 constant dollars (Dollar and Kraay database).
    - Dollar and Kraay database: at least two spaced observations of mean income of the poor for 92 countries (observations within countries are separated by at least five years over the period 1950–1999, the median interval being six years).
    - Selected sample: 75 developing countries and 187 observations between 1966 and 1999 (sample size varies by specification).
  - Headcount poverty: share earning less than $1 a day (1993 PPP).
    - Database coverage: 84 developing and transition countries; 21 have only one observation for 1980–2002.
    - Selected sample: 65 developing countries and 121 observations for the period 1980–2000.
  - Poverty gap as percent of poverty line (third indicator available).
- Income controls:
  - For poorest 20 percent mean income: GDP per capita in constant 1985 US$ at PPP (as in Dollar and Kraay, 2002).
  - For headcount and poverty gap: GDP per capita in constant 1993 US$ at PPP (World Bank) to ensure consistency.
- Financial variables:
  - Financial development and financial instability are key variables (two common indicators of financial development are used; further variable definitions and sources are provided in Annex II as noted in the source).
- Estimation method:
  - Panel data over the period 1966–2000 (as applicable) estimated with the system GMM estimator to address dynamics and potential endogeneity.
  - System GMM chosen in part because it is less sensitive to known data problems in poverty series.

### Key empirical finding summaries (as reported)
- Financial development is on average good for the poor, unlike financial instability.
- This positive result for the poor holds only when financial development is measured by the ratio of money to GDP (liquidity ratio), providing evidence that the conduit effect assumption is relevant.

*Source: _wp0862 - 2. We also recognize that financial development is accompanied by crises that are likely to*

### 1. The ratio to GDP of the liquid assets of the financial system, M3 (currency plus demand

### _wp0862 - 1. The ratio to GDP of the liquid assets of the financial system, M3 (currency plus demand 

### Financial development indicators used
- Two indicators:
  - The ratio to GDP of the liquid assets of the financial system, M3 (currency plus demand and interest-bearing liabilities of banks and non banks).
  - The ratio to GDP of the value of credits granted by financial intermediaries to the private sector.
- Motivation:
  - M3/GDP relates to the ability of financial systems to provide transaction services and saving opportunities and is relevant for testing the McKinnon conduit effect.
  - Credit/GDP (excluding credit to the public sector) measures the role of financial intermediaries in channeling funds to productive agents and possibly to the poor.

### Measuring instability of financial indicators
- Two approaches to measure instability:
  - Standard deviation of the variable growth rate (common but relies on stochastic-trend assumptions).
  - Average absolute value of residuals from OLS regression of xt on xt-1 and a time trend (preferred, does not assume purely stochastic or deterministic trend).
- Implementation details:
  - Financial instability for each country: regress the indicator on its lagged value and a linear trend for 1966–2000; compute the average absolute residual over five years (the year of the poverty measurement and the four preceding years).
  - Rationale: absolute changes in M3/GDP (or credit ratio) assumed to measure instability more accurately than relative changes.

### Econometric methodology
- Estimators used:
  - OLS for baseline regressions.
  - System GMM (dynamic panel generalized method-of-moment) to account for country-specific effects and address endogeneity, measurement errors, and omitted variables.
    - System GMM combines first-differenced equations with lagged levels as instruments and levels equations with lagged differences as instruments.
    - Blundell and Bond (1998) motivations: system GMM performs better than first-differenced GMM in small samples with weak instruments.
- Specification and tests:
  - Hansen test of over-identifying restrictions used to test validity of instruments (null: instruments not correlated with residuals).
  - Serial correlation test for no second-order serial correlation (null).
  - In reported regressions, both tests suggest inability to reject the null hypotheses.
  - First-order serial correlation tests indicate one cannot accept the null of no first-order serial correlation (results not reported to save space).

### Results — income of the poorest 20 percent
- General:
  - Tables 1 and 2 present results for M3/GDP and credit/GDP respectively; first two columns OLS, remaining columns system GMM.
- Findings for M3/GDP (Table 1):
  - M3/GDP level and instability are significantly correlated with the mean income of the poor.
  - Evidence does not reject the hypothesis of a positive direct effect of financial development on the standard of living of the poor.
  - Financial instability significantly reduces the income of the poor.
  - Controlling for financial instability improves the magnitude and significance of the liquidity ratio coefficient.
  - There is a positive correlation between instability and the level of financial development; on average, M3/GDP is more beneficial to the poor in countries with stable financial systems.
  - Interaction effects:
    - Liquidity ratio (M3/GDP) × number of bank branches per km2: positive and significant (column 11) — banking geographical coverage reinforces the favorable impact of liquidity on poor incomes.
    - Liquidity ratio has a stronger effect on poverty alleviation when country income is low; the negative effect of instability weakens when the level of development is higher (column 12) — consistent with McKinnon conduit effect being particularly relevant in poor countries.
  - Alternate indicator:
    - Ratio of saving deposits (M3-M1)/GDP yields similar results; positive coefficient slightly higher and negative coefficient of its instability is smaller.
- Findings for credit/GDP (Table 2):
  - Credit indicators (level and instability) are generally not significant except in OLS where private credit ratio is significant at 10 percent (column 1).
  - Suggests that in developing countries access to credit for the poor remains a challenge; the main channel for benefiting the poor is the McKinnon conduit effect captured by liquidity (M3/GDP) rather than credit/GDP.
  - Interaction effect:
    - Credit ratio × number of bank branches per km2: positive and significant (column 11) — banking coverage boosts benefits from more available credit by improving delivery of financial services.
  - Relationship between credit/GDP and income of the poor does not appear to depend on level of development (column 12).
- Robustness and sample issues:
  - Introducing the initial level (lagged income of the poor) to test convergence (column 5):
    - Financial development measured by M3/GDP remains favorable; financial instability remains detrimental.
    - Evidence of convergence: coefficient of convergence (initial level coefficient minus one) is negative and significant (example: coefficient of convergence in column 5 of Table 1 is –0.6 with a t statistic equal to –5.30).
    - Smaller sample size when lagged dependent variable is included; lagged variable dropped in subsequent regressions.
  - Consistent with Dollar and Kraay (2002): elasticity of income of the poor with respect to mean income is not significantly different from 1 in almost all estimations.
  - Sample composition may explain differences with other studies (e.g., Beck, Demirgüç-Kunt, and Levine, 2007):
    - For private credit ratio, sample average in Beck, Demirgüç-Kunt, and Levine (2007) is 40 percent of GDP, compared with 21 percent for the present sample.

### Other covariates, channels, and interpretation
- Inflation and growth:
  - Inflation rate generally has the expected negative sign in most regressions, but coefficient often lacks conventional significance.
  - GDP growth rate has a positive impact at 10% level of confidence on income of the poor in some specifications, without changing financial instability coefficients substantially.
  - Growth instability and inflation instability are not significant in either table.
  - Therefore, cannot conclude that the detrimental effect of financial instability on the poor operates through growth instability or inflation instability.
- Asymmetric income periods:
  - Interaction between log GDP per capita and a dummy equal to 1 when average growth is negative during the five previous years (as in Dollar and Kraay, 2002) appears negatively correlated with mean income of the poor (column 9 of Tables 1 and 2).
- Caveats:
  - Standard error of coefficient for financial instability may be biased because the instability measure is derived from a regression; bootstrap estimates indicate the bias is negligible.
  - Estimating instability over 1966–2000 may not capture structural breaks, though inclusion of lagged dependent variable partially mitigates this risk.

*Italic source: Content unit from the IMF PDF chapter _wp0862 (excerpt provided).*

### introduction of this variable reduces the marginal impact and the significance of M3/GDP

### _wp0862 - introduction of this variable reduces the marginal impact and the significance of M3/GDP

### Headcount poverty index
- Model estimated using the share of the population living on less than US$1 a day as the poverty measure.
- Expected coefficient signs are opposite those using income of the poorest 20 percent.
- Findings:
  - Financial development is negatively associated with the poverty headcount (Table 3).
  - Financial instability increases the headcount (Table 3).
  - The significance level of the ratio of private credit to GDP improves in OLS estimates (columns 1 and 2 of Table 4) compared to Table 2, but credit level and instability remain insignificant in the system GMM estimates (Table 4).
  - Growth instability and asymmetry show counter-intuitive coefficients in some specifications (column 6 and 8 of Table 3), possibly reflecting:
    - Propoor policies in countries vulnerable to exogenous shocks or volatile growth.
    - Social safety nets being relatively well-targeted to the very poorest (average income of the poorest 20 percent is about $2.2 a day, more than double the $1 a day poverty line).
    - Reliance on informal coping mechanisms by the poor.
  - Inflation instability increases poverty incidence (column 7 of Tables 3 and 4).

### Poverty gap
- Poverty gap expressed as a percentage of the poverty line; sensitive to changes in incomes of the poor even when below the poverty line and estimates resources needed to bring all poor out of poverty.
- Results (Tables 5 and 6) corroborate findings from other poverty measures:
  - Financial development associated with reductions in the poverty gap.
  - Financial instability associated with increases in the poverty gap in some specifications.

### Robustness tests
- Additional controls introduced include:
  - M3/GDP square, instability of agricultural value-added, inequality of land distribution, government consumption, civil liberty index, primary school enrolment, trade openness, and road density.
- Summary of robustness outcomes (Tables 8, 9, and 10):
  - Financial development and financial instability generally keep the expected signs and remain significant in most regressions.
  - Only a few control variables are significant:
    - More openness to trade (Tables 8, 9, and 10) and higher government consumption (Table 8) seem to worsen poverty.
    - Availability of infrastructure (Tables 8 and 10) and human capital accumulation (Tables 9 and 10) seem to alleviate poverty.
- Outlier analysis:
  - Removing observations with residuals larger than two standard deviations of the dependent variable confirms the main findings.
- Alternative instability measure:
  - Using the standard deviation of the growth rate of M3/GDP:
    - Financial instability is negatively associated with the log of income of the poorest 20 percent (at the 0.05 level).
    - Financial instability is positively correlated with the headcount index (at the 0.10 level).
    - No effect on the poverty gap.
    - The size and significance of financial development’s impact on the poor do not change dramatically.

### Economic relevance and channels
- Conceptual framework: financial development affects poverty directly and indirectly through economic growth and financial instability.
- Growth regression (Table 7) results:
  - Financial development stimulates economic growth.
  - Financial instability reduces economic growth.
  - High human capital, low black market premium, and political stability are beneficial to economic growth.
- Table 11: Summary of estimated coefficients (preserved exactly as reported)
  - Impact of Financial development (M3/GDP)
    - On Economic growth: 0.250*** (Column 1, Table 7)
    - On Financial instability: 4.897*** (Guillaumont and Kpodar (2006))
    - On Income of the poor: 0.447* (Column 4, Table 1)
    - On Headcount poverty: -0.326*** (Column 4, Table 3)
    - On Poverty gap: -0.148** (Column 4, Table 5)
  - Impact of Financial instability (Instability of M3/GDP)
    - On Economic growth: -1.557** (Column 1, Table 7)
    - On Income of the poor: -4.627** (Column 4, Table 1)
    - On Headcount poverty: 1.590* (Column 4, Table 3)
    - On Poverty gap: 1.011* (Column 4, Table 5)
  - Impact of Economic growth (GDP per capita growth)
    - On Income of the poor: 1.023*** (Column 4, Table 1)
    - On Headcount poverty: -0.175*** (Column 4, Table 3)
    - On Poverty gap: -0.086*** (Column 4, Table 5)
  - Significance notation: * significant at 10%; ** significant at 5%; *** significant at 1%

### Quantitative impacts (as reported)
- An exogenous 10 percentage point increase in M3/GDP (about half the sample standard deviation) would:
  - Directly produce a 4.5 percent increase in the average income of the poorest 20 percent.
  - Increase economic growth by 2.5 percent, with an equivalent rise in the average income of the poor.
  - Generate additional financial instability of about 0.49, which translates into:
    - A 0.76 percent drop in both economic growth and the average income of the poor.
    - Additional effects causing the income of the poor to decrease by 2.26 percent.
  - Net effect: a 10 percentage point increase in M3/GDP would net out to about a 4 percent increase in the average income of the poor.
  - For the same increase in M3/GDP:
    - Headcount poverty would have decreased by 2.8 percentage points.
    - Poverty gap would have been narrowed by 1.1 percentage points.

### Conclusion (authors’ stated main takeaways)
- Three main conclusions:
  1. Financial development is propoor, with the direct effect stronger than the effect through economic growth.
  2. Financial instability hurts the poor and partially offsets the benefit of financial development.

*Source: _wp0862 - introduction of this variable reduces the marginal impact and the significance of M3/GDP*

### 3. The McKinnon conduit effect is most likely the main channel through which the poor

### _wp0862 - 3. The McKinnon conduit effect is most likely the main channel through which the poor

### Mechanism: how financial development affects the poor
- The poor face liquidity constraints on physical and human capital investments; improvement in financial intermediation offers them a higher return on their savings.
- Credit becomes more available to the poor later in the process of financial development, after savings accumulation.
- Bank crises are particularly detrimental to the poor, who have very few opportunities to diversify their assets.
- Dominance of the McKinnon conduit effect: first-stage benefits accrue primarily through savings mobilization (higher returns to savers) rather than immediate direct credit access.

### Policy implications and recommendations
- Financial liberalization and reforms should explicitly account for the risk of financial instability because instability can dampen or cancel the poverty-reducing impact of financial development.
- Policies to liberalize the financial sector and foster financial intermediation should be accompanied by:
  - sound macroeconomic policies;
  - gradual external openness;
  - firm banking supervision.
- Prioritization of reform sequencing:
  - First-generation reforms to eliminate financial repression by liberalizing interest rates should take priority over second-generation reforms to improve access to affordable finance via institutional lending-system improvements.
  - At early stages of financial sector development, improving savings mobilization is effective in alleviating poverty, particularly when banks are reluctant to provide credit to the poor.
  - Saving accumulation will later improve access to credit.
- Encourage microfinance lending to the poor, recognizing that credit growth does not necessarily benefit them directly, especially when bank branches are geographically concentrated.
- Financial system design recommendation (Honohan and Beck (2007) paraphrase): promote “a distributed architecture, with larger institutions such as banks, MFI-network umbrella organizations, or the post office taking the contract and subcontracting parts of it to rural agencies, including MFIs.”

### Empirical findings (selected quantitative results preserved exactly as reported)
- Table 1 (Dependent variable: the log of average income of the poorest 20%):
  - Log of GDP per capita (y) coefficients reported across models include: 0.949; 0.947; 1.123; 1.023; 0.682; 1.064; 1.019; 1.023; 1.074; 1.008; 0.886; 1.004.
  - M3/GDP coefficients reported across models include: 0.331; 0.525; 0.069; 0.447; 0.350; 0.380; 0.359; 0.450; 0.350; 0.388; 0.358; 5.994.
  - Instability of M3/GDP coefficients reported include: -3.746; -4.627; -4.137; -4.101; -4.028; -4.680; -3.931; -3.388; -3.222; -104.988.
- Table 2 (Dependent variable: the log of average income of the poorest 20%, using Credit/GDP):
  - Log of GDP per capita (y) coefficients include: 0.960; 0.977; 1.170; 1.182; 0.729; 1.081; 1.128; 1.182; 1.157; 1.095; 0.971; 1.459.
  - Credit/GDP coefficients include: 0.296; 0.169; 0.074; -0.162; 0.120; -0.102; -0.085; -0.161; -0.138; 0.108; -0.327; 9.502.
  - Instability of Credit/GDP reported values include: 2.825; 3.798; -0.762; 4.164; 3.585; 3.792; 5.906; 4.343; 3.940; -5.510.
- Tables 3–6 (incidence of poverty and poverty gap, M3/GDP and Credit/GDP specifications) show consistent patterns:
  - M3/GDP and Credit/GDP often have negative coefficients with poverty measures (e.g., M3/GDP coefficients such as -0.296; -0.412; -0.103; -0.326 in Table 3).
  - Instability measures often have positive coefficients with poverty measures (e.g., Instability of M3/GDP = 2.022 in Table 3; Instability of M3/GDP = 0.994 in Table 5; Instability of M3/GDP = 1.336 in Table 10).
- Table 7 (System GMM, dependent variable: the growth rate of GDP per capita):
  - M3/GDP = 0.250 (t statistic reported as (4.49)***).
  - Instability of M3/GDP = -1.557 (t statistic reported as (2.01)**).
  - Education = 0.183 (t statistic reported as (2.85)***).
  - Other reported coefficients include: Log of initial GDP per capita = 0.0002; Inflation (a) = 0.002; Government consumption/GDP = -0.004; Trade openness (Log) = -0.012; Black market premium (a) = -0.044 (t statistic (2.07)**); Civil liberties index = -0.010; Political instability = -0.058 (t statistic (3.31)***).
  - Observations = 304; Number of countries = 69; Hansen Test (prob) = 1.00; AR(2) (prob) = 0.40.

### Interpretations drawn in the source
- The magnitude of the direct effect of financial development on poverty reduction underscores the relevance of active financial sector reforms.
- Financial instability undermines the poverty-reducing benefits of financial development; thus stability-enhancing policies are required alongside liberalization and intermediation-promoting reforms.
- Early-stage reforms that increase savings mobilization (the McKinnon conduit) are particularly effective for poverty alleviation when bank credit to the poor is limited; later, savings help improve access to credit.

*Source: _wp0862 - 3. The McKinnon conduit effect is most likely the main channel through which the poor (PDF chapter/section content provided).*

### Annex I. Descriptive Statistics

### Annex I. Descriptive Statistics

### Sample coverage
- Sample of 75 developing countries for which the income of the poorest 20% is available (1966-2000)
- Sample of 65 developing countries for which the headcount index is available (1980-2000)

### Descriptive statistics (selected variables; figures preserved exactly as in source)
- Mean income of the poorest 20%  
  - 187 670 502 42 2882
- GDP per capita (constant 1985 USD at PPP)  
  - 187 2461 1871 329 11738
- Headcount index  
  - 121 0.258 0.214 0.001 0.823
- Poverty gap (as a percent of the poverty line) / GDP per capita (constant 1993 USD at PPP)  
  - 119 120 0.101 2881 0.113 2043 0.000 440 0.522 9920
- M3/GDP  
  - 187 0.365 0.214 0.073 1.419  
  - 121 0.355 0.190 0.115 1.294
- Instability of M3/GDP  
  - 187 0.023 0.018 0.004 0.126  
  - 120 0.027 0.027 0.004 0.183
- Credit/GDP  
  - 166 0.210 0.164 0.012 0.993  
  - 116 0.203 0.151 0.014 0.745
- Instability of Credit/GDP  
  - 153 0.016 0.012 0.004 0.076  
  - 113 0.017 0.012 0.000 0.059
- Inflation rate  
  - 155 0.207 0.258 -0.040 1.727  
  - 116 0.398 1.337 0.006 12.642
- Instability of the agricultural value added  
  - 183 1.249 1.044 0.135 6.110  
  - 119 1.280 1.150 0.041 6.823
- Instability of growth  
  - 187 0.028 0.017 0.005 0.096  
  - 120 0.025 0.014 0.003 0.084
- Growth  
  - 187 4.705 4.752 -12.169 21.152  
  - 120 3.922 3.409 -6.516 18.829
- Gini index of land distribution  
  - 144 66.52 16.59 31.21 93.31  
  - 91 66.71 16.98 31.77 93.31
- Primary school enrolment rate (Education)  
  - 170 0.927 0.231 0.245 1.376  
  - 120 0.915 0.251 0.245 1.341
- Civil liberties index  
  - 162 2.872 1.305 0 6  
  - 121 2.864 1.294 0 6
- Road density  
  - 73 0.240 0.346 0.007 1.581  
  - 85 0.269 0.398 0.007 1.660
- Trade openness (%)  
  - 184 57.67 32.23 8.17 181.82  
  - 120 58.22 28.93 9.50 137.91
- Number of bank branches per km2  
  - 138 9.635 13.383 0.109 71.921  
  - 85 7.107 10.007 0.109 47.461
- Government consumption/GDP  
  - 184 0.130 0.051 0.044 0.318  
  - 120 0.131 0.049 0.046 0.295

### Variables and sources (definitions preserved)
- Log of income of the poorest 20%: Log of average incomes in bottom quintile, constant 1985 USD at PPP
- Log of GDP per capita (1985 PPP): Log of average per capita income, 1985 USD at PPP
- Headcount index: The percentage of the population living below the $1/day international poverty line
- Poverty gap: The average shortfall of the poor with respect to the poverty line, multiplied by the headcount ratio
- Log of GDP per capita (1993 PPP): Log of GDP per capita based on purchasing power parity (PPP)
- Growth: Growth of real GDP
- M3/GDP: Liquid liabilities as a percentage of GDP
- Inflation rate: Growth of consumer price index
- Agriculture value added: Agriculture value added as a share of GDP
- GDP per capita growth: Growth of real GDP per capita
- Education: Primary school enrolment rate
- Road density: The ratio of total road network (km) to country's total area (square km)
- Government consumption/GDP: Government expenditures as share of GDP
- Credit/GDP: Private Credit by Deposit Money Banks to GDP — Financial Structure Database 2001; The World Bank
- Gini of land distribution: Land distribution inequality (Data are average over 1950 to 1990) — Lundberg and Squire (2003)
- Civil liberties index: Civil liberties are measured on a one-to-seven scale, with one representing the highest degree of freedom and seven the lowest. For easing interpretation of this variable, we use in the regressions 7 minus Index of civil liberties — Freedom House Database 1999
- Financial openness: Sum of short and long term private debts, publicly or not guaranteed, divided by GDP — Global Development Finance 2002
- Political instability: Number of riots, attacks, strikes and coup d'état — CERDI database (2000)
- Number of bank branches per km2: Number of bank branches divided by the area of the country — Claessens (2006)
- Black market premium: The percentage difference between the black market rate and the official exchange rate — Global Development Network Database (1999)
- Trade openness: Sum of real exports and imports as share of GDP — Penn World Table 6.1; Dollar and Kraay (2002)
- World Bank Global Poverty Index Database; International Financial Statistics and World Development Indicators are cited as sources in the table

### Annex III. Correlations: Poverty and Financial Variables
- M3/GDP correlation lines as presented: M3/GDP                                    0.37-0.45-0.39..
- Private credit/GDP correlation line as presented: 0.33-0.38-0.350.78

*Source: _wp0862 - Annex I. Descriptive Statistics*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp0862.pdf_