## _wp1329 - 1. Summary Statistics

## Source details

**Canonical URL:** [_wp1329 - 1. Summary Statistics](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp1329.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp1329.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp1329.pdf.json)

---

### Introduction and research questions
- Investigates whether deeper financial systems dampen output, consumption, and investment volatility.
- Asks whether there are non-linearities (financial depth amplifying volatility at very high levels).
- Examines if financial depth reduces the effect of real external shocks on volatility.
- Assesses the role of structural characteristics related to trade and financial openness in mediating these effects.

### Data, sample, and measurement
- Sample: 110 countries over 1974–2008 (34 high-income, and 76 middle- and low-income countries).
- Panel frequency: five-year non-overlapping averages; panel is unbalanced.
- Primary data sources: IMF’s International Financial Statistics and World Economic Outlook; supplemented by World Bank and other sources.
- Financial depth (benchmark): private credit supplied by depository banks and other financial institutions as a share of GDP.
- Alternative financial depth measures (all as share of GDP): total liquid liabilities; depository banks’ assets; total deposits in financial institutions.
- Volatility measures:
  - Benchmark: standard deviation of the per-capita growth rate of the relevant variable (output, consumption, investment).
  - Robustness: deviation of per-capita growth rate from its Hodrick-Prescott filter trend.
- Real external shocks considered: country-specific terms-of-trade shocks (standard deviation); external demand shocks (standard deviation of partner country growth); commodity export price shocks (Aghion et al. (2010) approach).
- Banking crisis indicator: dummy for frequency of banking crisis within each five-year interval (Laeven and Valencia (2010)).
- Controls (averaged within each 5-year period unless otherwise noted): beginning-of-period real GDP per capita, standard deviation of real exchange rate changes, ratio of government balance to GDP, inflation rate, growth rates, trade openness, financial openness, Polity index.

### Key empirical findings
- Method: dynamic panel regressions estimated with system GMM (Arellano and Bond (1991); Arellano and Bover (1995); Blundell and Bond (1998)); robust two-step standard errors following Windmeijer (2004).
- Core result: financial depth (private credit/GDP) dampens volatility of output, consumption, and investment across countries, but only up to a threshold.
- Non-linearities:
  - Evidence of a U-shaped relationship: deeper financial systems reduce volatility up to a threshold, beyond which further deepening is associated with increased volatility.
  - Panel GMM estimated thresholds (Table 5, full sample):
    - Final consumption threshold: 132.9 (threshold SE 27.78; 95% CI (78.5,187.4))
    - Private consumption threshold: 118.6 (threshold SE 20.10; 95% CI (79.2,158.0))
    - Investment threshold: 106.0 (threshold SE 11.06; 95% CI (84.3,127.7))
  - HP-filter robustness threshold estimates: 120.6, 107.7, 100.1, 98.39 (with corresponding SEs and 95% CIs reported in Table 10).
  - Marginal significance: consumption-smoothing effect no longer statistically significant when private credit reaches 105 percent of GDP (over 18 percent of observations in the relevant regression are above this threshold); in 2008, 20 countries were above 105 percent of GDP.
  - Non-linear GDP-volatility effect significant only for a sub-sample of high- and middle-income countries (point estimate suggests marginal effect becomes negative when credit reaches 117 percent of GDP — reported as "available upon request").
- Shock absorption:
  - Deeper financial systems mitigate negative effects of real external shocks on macroeconomic volatility.
  - Interaction results:
    - Private credit × terms-of-trade volatility: negative and statistically significant — more financially developed economies are less adversely affected.
    - Private credit × partner growth volatility: negative and significant for private consumption volatility.
  - Triple interactions with structural characteristics:
    - Trade openness: financial depth reduces impact of terms-of-trade shocks on private consumption volatility in more open economies.
    - Financial openness: greater domestic financial depth attenuates impact of external demand and terms-of-trade shocks on consumption volatility in more financially open economies.
  - Commodity export price shocks: commodity export price coefficients positive and significant; interaction terms PC x TO x (sd) Commodity export price and PC x FO x (sd) Commodity export price are negative and significant in several specifications.
- Robustness:
  - Results hold across alternative control variables, income groups, time periods, alternative measures of financial depth and shocks, HP-filter volatility measure, and GMM specifications.
  - Alternative financial depth measures (total liquid liabilities; bank assets; deposits) confirm main results and non-linearities for consumption and investment; non-linear output-volatility relationship holds mainly for advanced and emerging market sub-samples.
  - Endogeneity addressed via system GMM; Hansen tests generally do not reject instrument validity (examples: Table 4 Hansen test p-value 0.272, 0.604, 0.305, 0.677); AR(2) test p-values reported across tables.

### Key statistics and descriptive figures (five-year panel; Table 1)
- SD of real per-capita growth of GDP: Obs 880, Mean 3.47, Std. Dev. 3.37, Min 0.06, Max 34.39
- Final consumption volatility: Obs 641, Mean 15.00, Std. Dev. 8.85, Min 0.04, Max 176.10
- Private consumption volatility: Obs 663, Mean 5.90, Std. Dev. 10.31, Min 0.01, Max 178.62
- Investment volatility: Obs 635, Mean 15.14, Std. Dev. 14.43, Min 0.00, Max 124.86
- Private credit (financial depth): Obs 728, Mean 39.74, Std. Dev. 37.44, Min 1.39, Max 195.77
- Liquid liabilities: Obs 703, Mean 42.31, Std. Dev. 28.85, Min 0.65, Max 223.09
- Bank assets: Obs 717, Mean 44.04, Std. Dev. 38.85, Min 1.99, Max 241.28
- Deposits: Obs 718, Mean 37.28, Std. Dev. 30.52, Min 1.74, Max 211.73
- Partner growth volatility: Obs 875, Mean 1.15, Std. Dev. 1.23, Min -2.48, Max 11.93
- Terms-of-trade volatility: Obs 850, Mean 0.0, Std. Dev. 36.85, Min -57.6, Max 128.79
- Commodity export price volatility: Obs 818, Mean 0.04, Std. Dev. 0.23, Min -1.18, Max 1.52
- Banking crisis: Obs 880, Mean 0.31, Std. Dev. 0.83, Min 0.00, Max 5.00

### Main regression magnitudes and determinants
- Panel GMM result: "a one-standard-deviation increase in private credit to GDP (37.4 percent of GDP in the sample) reduces the volatility of growth, consumption, and investment by over 15 percent."
- Polity index: stronger political institutions associated with reduced macroeconomic volatility.
- Trade openness associated with higher consumption variability.
- Financial openness associated with greater GDP volatility.
- Exchange rate volatility associated with higher GDP volatility.
- Sub-sample (low- and middle-income countries): financial depth has a more pronounced smoothing effect on consumption and investment volatility; coefficient estimates larger for consumption volatility than for full sample.

### Robustness and endogeneity diagnostics (selected)
- Hansen test p-values reported in tables; example: Table 4 Hansen p-values 0.272, 0.604, 0.305, 0.677 across Columns 1-4.
- AR(2) test p-values reported to assess serial correlation across specifications.
- Alternative instrument lags (third vs second lag) yield very similar estimates (results available upon request).
- Three-year, five-year, seven-year panels produce broadly consistent results (Table 13); three-year panel increases number of instruments and yields very high Hansen p-values (close to 1) but AR-2 test supports specification.

### Policy-relevant conclusions and recommendations
- Core policy conclusions:
  - Deeper financial systems generally dampen volatility of output, consumption, and investment and help cushion against adverse real external shocks.
  - The consumption-smoothing role of financial depth is especially important in countries with high trade and financial openness.
  - There exists a non-linear (U-shaped) relationship: financial depth reduces volatility up to estimated thresholds, beyond which further deepening can amplify volatility.
- Quantitative policy-relevant magnitudes:
  - One-standard-deviation increase in private credit to GDP (37.4 percent of GDP) reduces volatility of growth, consumption, and investment by over 15 percent.
  - Estimated thresholds (private credit to GDP) where benefits reverse: 132.9 (final consumption), 118.6 (private consumption), 106.0 (investment).
- Policy considerations:
  - Low-income and emerging market economies can benefit from further financial deepening to reduce welfare costs of volatility, given higher vulnerability to terms-of-trade, commodity price, and partner volatility.
  - Advanced economies with very high financial depth should be aware of threshold risks where excessive financial depth may amplify volatility; potential channels include misallocation of funds, increased leverage, and risk-taking induced by implicit safety nets.
  - Policies should be calibrated to country context (stage of development, trade and financial openness) to maximize benefits of financial deepening while containing risks beyond identified thresholds.
- Research implications:
  - Further empirical work needed to identify factors driving threshold effects and channels through which high levels of financial depth propagate and amplify instability, to better inform targeted regulation and macroprudential policies.

*Source: _wp1329 - 1. Summary Statistics (IMF PDF content provided).*

### 1. Summary Statistics...................................................................................................

### _wp1329 - 1. Summary Statistics...................................................................................................

### Introduction and research questions
- Investigates whether deeper financial systems dampen output, consumption, and investment volatility.
- Asks whether there are non-linearities (financial depth amplifying volatility at very high levels).
- Examines if financial depth reduces the effect of real external shocks on volatility.
- Assesses the role of structural characteristics related to trade and financial openness in mediating these effects.

### Key empirical findings
- Uses dynamic panel analysis for 110 advanced and developing countries over the years 1974-2008.
- Finds strong support that financial depth (proxied by private credit as a share of GDP) dampens volatility of output, consumption, and investment across countries, but only up to a certain point.
- Identifies that at very high levels (over 100 percent of GDP) financial depth magnifies consumption and investment volatility.
- Finds robust evidence that deeper financial systems serve as shock absorbers, mitigating the negative effects of real external shocks on macroeconomic volatility.
- When country-specific external shocks are interacted with financial depth, results are robust and highly significant, with a more pronounced smoothing effect on consumption and investment volatility.
- Shows that the volatility-dampening role of finance is related to the degree of openness: financial depth reduces the impact of terms of trade, external demand, and commodity price shocks on consumption volatility particularly when trade and financial openness are high.
- Results hold against robustness tests including alternative control variables, countries’ income levels, different time periods, alternative measures of financial depth and shocks, and the use of GMM methods to address endogeneity.

### Theoretical and empirical context
- Theoretical channels cited:
  - Financial frictions and agency/information asymmetries propagate shocks via the credit channel (Bernanke and Gertler, 1990; Kiyotaki and Moore, 1997; Greenwald and Stiglitz, 1991).
  - Deeper financial systems relax borrowing constraints, promote risk sharing, alleviate cash constraints, and facilitate diversification (Caballero and Krishnamurty, 2001; Aghion et al., 1999; Acemoglu and Zilibotti, 1997).
  - Financial development can also increase risk-taking and leverage, potentially raising volatility (Shliefer and Vishny, 2010; Wagner, 2010; Brunnermeier et al., 2012; Quadrini, 2011).
- Empirical literature highlights mixed findings:
  - U-shaped relationship reported by Easterly et al. (2000) with output volatility increasing when private credit reaches 100 percent of GDP.
  - Denizer et al. (2002) generally support a negative correlation but find private sector credit/GDP not always significant.
  - Acemoglu et al. (2003) and Beck et al. (2006) do not find a robust relation between financial intermediary development and growth volatility.
  - Evidence that trade and financial openness correlate with higher output volatility (Kose et al., 2003) and that exposure to external risk increases income and consumption volatility (Rodrik, 1998).

### Data, sample, and measurement
- Panel comprised of 110 countries over 1974–2008.
  - Country composition: 34 high-income, and 76 middle- and low-income countries.
- Time-series data transformed into five-year averages to construct the panel.
- Excludes the recent global financial crisis from the sample period.
- Primary data sources: IMF’s International Financial Statistics and World Economic Outlook; supplemented by World Bank and other sources.
- Panel of country and time-period observations is unbalanced.
- Financial depth proxied by private credit extended by banks and other institutions as a share of GDP.
- Macroeconomic volatility measured for output, consumption, and investment (specific volatility measures referenced in the paper’s methodology).

### Econometric methodology and addressing endogeneity
- Uses a dynamic panel regression framework estimated with system GMM:
  - References: Arellano and Bond (1991); Arellano and Bover (1995); Blundell and Bond (1998).
  - Robust two-step standard errors computed following Windmeijer (2004).
- Baseline regression specification (notation preserved from source):
  - V_{i,t} = α + Σ_{j=1}^{p} ρ_j V_{i,t−j} + β FD_{i,t} + γ X_{i,t} + τ_t + μ_i + ε_{i,t}
  - Where V is macroeconomic volatility, FD is financial depth, X is a set of control variables, τ_t is the time-specific effect (for each non-overlapping five-year period), μ_i is the country-specific effect, and ε_{i,t} is the error term.
- The approach addresses joint endogeneity of explanatory variables and potential biases from country-specific effects.

### Robustness and sensitivity
- Results tested for robustness to:
  - Alternative control variables.
  - Countries’ income levels.
  - Different time periods.
  - Alternative measures of financial depth and shocks.
  - Endogeneity via GMM instrumentation.
- Findings reported as robust across these tests.

### Implications emphasized in the text
- Financial deepening generally stabilizes macroeconomic aggregates up to moderate levels of financial depth, implying welfare gains through reduced volatility.
- At very high levels of financial depth (over 100 percent of GDP), the stabilizing effects can reverse for consumption and investment, suggesting non-linearities and potential instability from excessive financial expansion.
- Deepening financial systems enhances shock absorption, especially in open economies where trade and financial openness increase exposure to external shocks.

*Source: _wp1329 - 1. Summary Statistics (IMF PDF content provided).*

### Appendix I presents the list of countries included in the sample.

### _wp1329 - Appendix I presents the list of countries included in the sample

### Measurement and data
- Financial depth (benchmark): aggregate private credit provided by deposit money banks and other financial institutions as a share of GDP (following King and Levine (1993)).
- Alternative financial depth measures considered (all measured as a share of GDP): total liquid liabilities; depository banks’ assets; total deposits in financial institutions.
- Volatility measures:
  - Benchmark: standard deviation of the per-capita growth rate of the relevant variable (output, consumption, investment).
  - Robustness: deviation of per-capita growth rate from its Hodrick-Prescott filter trend.
- Real external shocks considered:
  - Country-specific terms-of-trade shocks (standard deviation).
  - External demand shocks measured as the standard deviation of partner country growth.
  - Robustness: country-specific commodity price shocks using the approach in Aghion et al. (2010).
- Banking crisis indicator: dummy variable indicating the frequency of a banking crisis within each five-year interval (Laeven and Valencia (2010)).
- Controls (averaged within each 5-year period unless otherwise noted): beginning-of-period real GDP per capita, standard deviation of real exchange rate changes, ratio of government balance to GDP, inflation rate, growth rates of real per-capita GDP/consumption/investment, trade openness (exports+imports to GDP), financial openness (gross inflows as in Kose et al. (2009)), Polity index.

Key sample descriptors and summary statistics (five-year panel; from Table 1)
- Observations and descriptive stats (selected):
  - SD of real per-capita growth of GDP: Obs 880, Mean 3.47, Std. Dev. 3.37, Min 0.06, Max 34.39
  - Final consumption volatility: Obs 641, Mean 15.00, Std. Dev. 8.85, Min 0.04, Max 176.10
  - Private consumption volatility: Obs 663, Mean 5.90, Std. Dev. 10.31, Min 0.01, Max 178.62
  - Investment volatility: Obs 635, Mean 15.14, Std. Dev. 14.43, Min 0.00, Max 124.86
  - Private credit (financial depth): Obs 728, Mean 39.74, Std. Dev. 37.44, Min 1.39, Max 195.77
  - Liquid liabilities: Obs 703, Mean 42.31, Std. Dev. 28.85, Min 0.65, Max 223.09
  - Bank assets: Obs 717, Mean 44.04, Std. Dev. 38.85, Min 1.99, Max 241.28
  - Deposits: Obs 718, Mean 37.28, Std. Dev. 30.52, Min 1.74, Max 211.73
  - Partner growth volatility: Obs 875, Mean 1.15, Std. Dev. 1.23, Min -2.48, Max 11.93
  - Terms-of-trade volatility: Obs 850, Mean 0.0, Std. Dev. 36.85, Min -57.6, Max 128.79
  - Commodity export price volatility: Obs 818, Mean 0.04, Std. Dev. 0.23, Min -1.18, Max 1.52
  - Banking crisis: Obs 880, Mean 0.31, Std. Dev. 0.83, Min 0.00, Max 5.00

- Time horizon covered in figures and analysis: 1974-2008 (Figure 1 caption).

### Main regression findings
- Cross-section (Table 3): coefficients on private credit are negative and statistically significant in all regressions for the full sample, suggesting deeper financial systems are associated with less macroeconomic volatility.
- Panel GMM (Table 4; five-year non-overlapping averages):
  - Private credit coefficient significantly negative in all regressions across country groups.
  - Economic magnitude: "a one-standard-deviation increase in private credit to GDP (37.4 percent of GDP in the sample) reduces the volatility of growth, consumption, and investment by over 15 percent."
  - Polity index: stronger political institutions associated with reduced macroeconomic volatility.
  - Trade openness associated with higher consumption variability; financial openness associated with greater GDP volatility; exchange rate volatility associated with higher GDP volatility.
- Sub-sample (low- and middle-income countries): beneficial effect of financial depth on smoothing consumption and investment volatility is more pronounced; coefficient estimates larger for consumption volatility than for the full sample.

### Non-linear (U-shaped) relationship between financial depth and volatility
- Quadratic specifications (Table 5 and Table 10):
  - Both linear and quadratic terms on private credit are statistically significant in almost all regressions.
  - Evidence of a U-shaped effect: deeper financial systems reduce volatility up to a threshold, beyond which further deepening is associated with increased volatility.
- Estimated threshold values (Table 5; Panel GMM, full sample):
  - Final consumption threshold: 132.9 (threshold SE 27.78; 95% CI (78.5,187.4))
  - Private consumption threshold: 118.6 (threshold SE 20.10; 95% CI (79.2,158.0))
  - Investment threshold: 106.0 (threshold SE 11.06; 95% CI (84.3,127.7))
- HP-filter robustness (Table 10) threshold estimates:
  - Threshold values reported: 120.6, 107.7, 100.1, 98.39 with corresponding SEs and 95% CIs shown in the table.
- Marginal significance noted:
  - Figure 3 (described in text): positive effect of financial depth in reducing consumption volatility is no longer statistically significant when private credit reaches 105 percent of GDP (over 18 percent of the observations in the regression of Column 2 are above this threshold).
  - In 2008, there were 20 countries above this 105 percent level; examples noted: United States, Ireland, the United Kingdom, Spain, Portugal, and Cyprus.
- Non-linear effect on GDP volatility is only significant for a sub-sample of high- and middle-income countries (point estimate suggesting marginal effect becomes negative when credit reaches 117 percent of GDP — reported as "available upon request").

### Role of shocks and interactions
- Controlling for shocks (Table 6):
  - Terms-of-trade fluctuations increase volatility of output, consumption, and investment.
  - External demand (partner growth) volatility associated only with higher output volatility; insignificant for consumption and investment volatility.
  - Banking crises are important determinants of output and investment volatility.
  - Financial depth effect remains negative and statistically significant when controlling for banking crises, terms-of-trade fluctuations, and external demand one-by-one and jointly (Columns 4, 8, 12, 16).
- Interaction results (Table 7):
  - Interaction of private credit and terms-of-trade volatility: negative and statistically significant — more financially developed economies are less adversely affected by terms-of-trade volatility ( Columns 1-4).
  - Interaction of private credit and partner growth volatility: negative and significant for private consumption volatility (Columns 5-8); effects less significant for output and investment.
- Triple interactions and structural characteristics (Tables 8-9):
  - Trade openness: while trade openness increases macroeconomic volatility, financial depth reduces the impact of terms-of-trade shocks on private consumption volatility in more open economies (Table 8, Column 3).
  - Financial openness: financial openness associated with higher macroeconomic volatility, but greater domestic financial depth attenuates the impact of external demand and terms-of-trade shocks on consumption volatility in more financially open economies (Table 9, Columns 2-3 and 6-7).
- Commodity export price shocks (Table 12):
  - Commodity export price shocks increase volatility (sd) Commodity export price coefficients positive and significant.
  - Interaction terms: PC x TO x (sd) Commodity export price and PC x FO x (sd) Commodity export price — negative and significant in several specifications, indicating deeper financial systems dampen the volatility effects of commodity export price shocks, particularly for countries with greater trade and financial openness.

### Robustness checks and endogeneity
- Alternative volatility measure (HP-filter): main negative relationship between financial depth and volatility and the non-monotonic (U-shaped) relationship remain (Table 10).
- Alternative financial depth measures (Table 11): results confirmed using (i) total liquid liabilities, (ii) depository banks’ assets, and (iii) total deposits in financial institutions; non-linear relationship persists for consumption and investment volatility; non-linear output-volatility relationship holds mainly for advanced and emerging market sub-samples.
- Endogeneity concerns addressed:
  - Dynamic panel GMM estimator used to control for potential endogeneity and for including initial level of volatility.
  - Hansen tests of over-identifying restrictions reported across tables generally do not reject instrument validity (p-values reported in tables; examples: Table 4 Hansen test p-value 0.272, 0.604, 0.305, 0.677 for Columns 1-4).
  - AR(2) test p-values reported to assess serial correlation (examples across tables).
  - Robustness: alternative instrument lags (third vs second lag) yield very similar estimates (results available upon request).
  - Interaction-based identification reduces endogeneity concerns relative to single-variable effects.

- Other robustness dimensions:
  - Different time horizons: three-year, five-year, seven-year panels show broadly consistent results (Table 13). Note: three-year panel increases number of instruments and yields very high Hansen p-values (close to 1) but AR-2 test supports specification.
  - Omission of continents and inclusion of alternative controls (e.g., replacing Polity index or including banking crises) do not materially change main results (available upon request).

### Policy-relevant conclusions and implications
- Core findings:
  - Deeper financial systems dampen volatility of output, consumption, and investment growth and help cushion against adverse real external shocks.
  - The consumption-smoothing role of financial depth is especially important in countries with high trade and financial openness.
  - There exists a non-linear (U-shaped) relationship: financial depth reduces volatility up to country-specific thresholds, beyond which further deepening can amplify volatility.
- Quantitative policy-relevant magnitudes:
  - One-standard-deviation increase in private credit to GDP (37.4 percent of GDP in the sample) reduces volatility of growth, consumption, and investment by over 15 percent (Panel GMM result).
  - Estimated thresholds (approximate levels of private credit to GDP where benefits reverse): 132.9 (final consumption), 118.6 (private consumption), 106.0 (investment) — see Table 5.
- Policy considerations:
  - Low-income and emerging market economies, given their higher vulnerability to terms-of-trade, commodity price, and partner volatility, can benefit from further financial deepening to reduce welfare costs of volatility.
  - For advanced economies with very high financial depth, policymakers should be aware of threshold risks where excessive financial depth may amplify volatility — possible channels include misallocation of funds, increased leverage, and risk-taking induced by implicit safety nets.
  - Policies should be calibrated to country context (stage of development, trade and financial openness) to maximize benefits of financial deepening while containing risks beyond the identified thresholds.
- Research implications:
  - Further empirical work needed to identify factors driving the threshold effects and the channels through which high levels of financial depth propagate and amplify instability, to better inform targeted regulation and macroprudential policies.

*Source: _wp1329 - Appendix I presents the list of countries included in the sample.*

### Appendix 1. List of Countries

### _wp1329 - Appendix 1. List of Countries

### High Income
- Australia
- Austria
- Bahrain
- Canada
- Cyprus
- Czech Republic
- Denmark
- Estonia
- Finland
- France
- Germany
- Greece
- Hungary
- Ireland
- Israel
- Italy
- Japan
- Korea, Rep.
- Netherlands
- New Zealand
- Norway
- Oman
- Portugal
- Singapore
- Slovak Republic
- Slovenia
- Spain
- Switzerland
- United Kingdom
- United States
- Mexico
- Morocco
- Mauritius
- Malaysia
- Latvia
- Lithuania
- Libya
- Gabon
- Kazakhstan
- Iran, Islamic Rep.
- India
- Guatemala
- Honduras
- Honduras
- Guatemala

### Mi ddl e  I ncome
- Albania
- Algeria
- Angola
- Argentina
- Armenia
- Botswana
- Brazil
- Chile
- China
- Colombia
- Costa Rica
- Croatia
- Dominican Republic
- Ecuador
- Egypt, Arab Rep.
- El Salvador
- Equatorial Guinea
- Gabon
- Guatemala
- Honduras
- India
- Indonesia
- Iran, Islamic Rep.
- Kazakhstan
- Kyrgyz Republic
- Lao PDR
- Libya
- Madagascar
- Malawi
- Mali
- Mauritania
- Mexico
- Morocco
- Mozambique
- Nepal
- Niger
- Nigeria
- Pakistan
- Papua New Guinea
- Philippines
- Poland
- Russian Federation
- South Africa
- Thailand
- Turkey
- Uruguay
- Venezuela, RB
- Zambia

### Low Income
- Paraguay
- Peru
- Philippines
- Burkina Faso
- Burundi
- Cote d'Ivoire
- Cambodia
- Chad
- Cameroon
- Ethiopia
- Cape Verde
- Gambia, The
- Ghana
- Guinea-Bissau
- Haiti
- Congo, Rep.
- Kenya
- Kyrgyz Republic
- Lao PDR
- Lesotho
- Libya
- Madagascar
- Malawi
- Mali
- Mauritania
- Mozambique
- Nepal
- Niger
- Nigeria
- Papua New Guinea
- Rwanda
- Senegal
- Sierra Leone
- Solomon Islands
- Tanzania
- Togo
- Uganda
- Zambi a

### Appendix 2. Variables Used in Regression Analysis
- Variables: GDP growth
  - Description: Annual percentage growth of per-capita real GDP.
  - Source: IMF WEO
- Variables: Final Cons umption growth
  - Description: Annual percentage growth of per-capita real final cons umption.
  - Source: IMF WEO
- Variables: Private Cons umption growth
  - Description: Annual percentage growth of per-capita real private cons umption.
  - Source: IMF WEO
- Variables: Inves tment growth
  - Description: Annual percentage growth of per-capita real inves tment.
  - Source: IMF WEO
- Variables: GDP volatility
  - Description: Standard deviation of real GDP growth.
  - Source: IMF WEO
- Variables: Final cons umption volatility
  - Description: Standard deviation of real final cons umption growth.
  - Source: IMF WEO
- Variables: Private cons umption volatility
  - Description: Standard deviation of real private cons umption growth.
  - Source: IMF WEO
- Variables: Investment volatility
  - Description: Standard deviation of real inves tment growth.
  - Source: IMF WEO
- Variables: Liquid liabilities
  - Description: M3 as  percentage of GDP.
  - Source: World Bank
- Variables: Bank assets
  - Description: Depos itory banks 's  as s ets as  percentage of GDP
  - Source: World Bank
- Variables: Private credit
  - Description: Private credit s upplied by depos itory banks  and other financial institutions  as  percentage of GDP.
  - Source: World Bank
- Variables: Depos its
  - Description: Depos it to depos itory banks  and other financial institutions  as  percentage of GDP.
  - Source: World Bank
- Variables: External demand shocks
  - Description: Standard deviation of the partner's  percentage growth weighted by a country's export ratio.
  - Source: IMF WEO
- Variables: Terms  of trade s hocks
  - Description: Standard deviation of percentage change in term of trade of goods.
  - Source: IMF WEO
- Variables: Commodity price s hocks
  - Description: Standard deviation of percentage change in commodity export prices   weighted by s hare of commodity exports in total exports.
  - Source: IMF WEO
- Variables: Banking Crisis
  - Description: Dummy variable for banking cris es .
  - Source: Laeven and Valencia (2010)
- Variables: Trade openness
  - Description: Sum of exports  and imports  as  a s hare of GDP.
  - Source: IMF WEO
- Variables: Financial opennes s
  - Description: Sum of gros s  foreign direct inves tment, portfolio and equity inves tment liabilities  as percentage of GDP.
  - Source: IMF WEO
- Variables: Inflation
  - Description: Inflation meas ured by cons umer price index.
  - Source: IMF WEO
- Variables: Government balance
  - Description: Central government balance as  percentage of GDP.
  - Source: IMF WEO
- Variables: Polity index
  - Description: Index ranging from -10 (autocratic regimes ) to +10 (democratic regimes).
  - Source: POLITY IV
- Variables: Real exchange rate volatility
  - Description: Standard deviation of real exchange rate in unit of national currency per US dollar.
  - Source: IMF WEO
- Variables: Initial real GDP per capita
  - Description: Natural log of real GDP per capita at the beginning of each period.
  - Source: IMF WEO

*Content derived from _wp1329 - Appendix 1. List of Countries (PDF).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp1329.pdf_