## _wp07118

## Source details

**Canonical URL:** [_wp07118](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07118.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07118.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07118.pdf.json)

---

### Major findings and objectives
- Average private sector liabilities increased from 114 percent of GDP in 1995 to 143 percent of GDP in 2004 for a group of 18 advanced economies.
- Rate of buildup in private sector liabilities outpaced real GDP growth by more than 2.5 percentage points on average every year during the period.
- Between 1986 and 1995 the average private-credits-to-GDP ratio of these countries increased by a mere 0.8 percentage points per year.
- Cross-country standard deviation of changes in the credit-to-GDP ratio doubled between the pre- and post-1995 periods (from 0.13 to 0.27).
- Portugal’s private sector liabilities increased by 90 percentage points of GDP between 1995 and 2004; Japan and Canada experienced a decline in the ratio; Finland and France saw increases of 9 and 15 percentage points, respectively.
- Private sector financial liabilities are defined as all loans and securities liabilities (including derivatives but excluding shares) incurred by households and nonfinancial corporations; data are consolidated from OECD national accounts and Eurostat financial accounts.
- Purpose:
  - Document patterns of accumulation of private sector liabilities in advanced economies during 1995-2005.
  - Examine economic implications of recent buildups in liabilities at the industry level, focusing on financial intermediation effectiveness and conditions under which it could be enhanced.
- Contribution: exploits cross-country variation and analyzes impacts of changes—rather than levels—of credits on industry performance, controlling for unobserved country fixed factors.
- Equity finance: net equity finance drifted up by only about 1.5 percent of GDP over the last 10 years for the sample, while corporate borrowings increased by 20 percent of GDP during the same period.

### II. Recent developments in financial liabilities
- Households set the pace and cross-country variation:
  - Households accounted for more than half (53.7 percent) of the overall increase in private sector liabilities.
  - Household liabilities composed on average 41.7 percent of total private sector liabilities in 1995; by the period’s end household liabilities’ share jumped to 46.2 percent.
  - Average household liabilities increased from 48 percent of GDP in 1995 to 68 percent of GDP in 2005.
  - Households’ share increased in 12 of the 18 countries (DNK, US, AUS, NLD, ESP, SWE, FIN, PRT, ITA, GRC, JPN and CAN).
  - Countries with household liability increases >30 percent of GDP: Portugal, the Netherlands, Australia, Spain.
  - Countries with household liability increases <5 percent of GDP: Japan, Belgium, Canada.
  - Cross-country dispersion in household indebtedness in 2005: range from 110 percent of GDP in Denmark to 28 percent of GDP in Italy.
- Corporate sector:
  - Average corporate liabilities rose from 63 percent of GDP in 1995 to 75 percent of GDP in 2005.
  - Excluding Japan, corporate liabilities increased from 58 percent to 73 percent of GDP.
  - Portugal, the UK and Spain showed the largest corporate borrowings increases (>30 percentage points of GDP).
  - Loans accounted for 81 percent of total corporate liabilities in 2004, down from 85 percent in 1995.
  - Finland and Sweden saw loans’ share of corporate liabilities decline by 10 percentage points.
  - Securities account for >25 percent of total corporate borrowings in France and the Anglo-Saxon countries.
- Synchronization and variation:
  - Household and corporate liabilities expanded in sync in many countries; large variation across countries in magnitude.
  - Recent increases in corporate borrowings are associated with reductions in costs of external finance and improvements in resource allocation particularly in countries with:
    - a more arm’s length–based financial system, and
    - a less rigid labor market.
  - The association is stronger where the financial system has greater arm’s length content and the labor market is more flexible.
  - Paper abstracts from credit-growth effects on financial stability; prior literature links rapid credit growth to crises, but Moreno-Badia (2007) finds no direct adverse effects of credit growth on banking-sector stability across the euro area in recent years.

### Cross-country variation in private-sector liabilities (1995-2005 or latest)
- Average and dispersion:
  - Cross-country variation in changes in total private-sector liabilities: 3.7 percentage points.
  - Standard deviation: 3.3 percentage points.
- Mechanical contributors to variation:
  - Large differences in changes in household liabilities.
  - Large differences in changes in corporate liabilities.
  - Concurrence between changes in the two sectors contributes the rest of variance.
- Quantitative decomposition of variance:
  - Corporate liabilities account for 37 percent of the cross-country variation (variance).
  - Household liabilities account for 29 percent of the cross-country variation (variance).
  - Excluding Japan: corporate liabilities would have contributed 28 percent, and household liabilities 41 percent, of the overall variance.
- Observed sectoral patterns (1995-2005):
  - Increases in liabilities as a share of GDP generally occurred in both household and corporate sectors, with exceptions of Canada and Japan.
  - Countries with more rapid corporate expansion tend to have larger household increases (positive association).
  - Japan experienced a very sharp decline in corporate liabilities as a share of GDP.

### Macroeconomic relationships: expenditure composition and growth
- Private consumption and investment:
  - Buildups in total liabilities do not appear to raise the expenditure share of private consumption.
  - Increases in corporate liabilities seem associated with rises in the expenditure share of productive investment (non-housing component of gross fixed capital formation).
- Growth:
  - No relationship between increases in liabilities and economic growth appears in aggregate (simple regressions and convergence-adjusted regressions fail to find significant relationships).
  - Potential masking factors: unobserved time-varying country-level factors and reverse causality (GDP denominator effects, temporary cash-flow changes).

### Industry-level approach: rationale and empirical strategy
- Motivation:
  - Reduce bias from omitted variables and reverse causality by exploiting cross-industry variation.
  - Disaggregated analysis examines differential effects of finance on industries with different dependence on external finance.
- Theoretical basis:
  - Agency problems raise external finance costs; industries differ in dependence on external finance due to technological factors.
  - Rajan and Zingales (1998) logic: if buildups in liabilities reduce costs of external finance, industries more reliant on external finance should grow relatively faster in countries with faster liability buildups.
- Econometric specification:
  - Dependent variable: change in industry i in country c = (average annual real growth in value-added 1995-2003) − (average annual real growth 1987-1995).
  - Key regressors:
    - iE: country-invariant measure of industry dependence on external finance.
    - cLΔ: average annual change in liabilities as a share of GDP during 1995-2003 for country c.
    - cX: country characteristics (financial system type, labor market flexibility, changes in real short-term and long-term interest rates between 2003-2005 and 1995-1997).
    - Full sets of country and industry dummies included.
  - Hypothesis: positive coefficient on interaction (dependence * change in liabilities) (β̂1) if buildups reduce costs of external finance.
  - Panel approach uses cross-industry variation and drops time-series dimension to reduce endogeneity.

### Measures and controls
- Dependence on external finance:
  - From de Serres, Kobayakawa, Slok and Vartia (2006), constructed in spirit of Rajan and Zingales (1998).
  - Defined from US listed firms as (capital expenditures − cash flows from operations) / capital expenditures.
  - 2-digit coverage including services; data covering 1990-2003.
- Financial system measure:
  - 1995 financial index from IMF (2006) World Economic Outlook (September 2006).
  - Aggregate of: degree of traditional banking intermediation, level of development of new financial intermediation, importance of capital markets.
  - Higher score = more arm’s length-based financial system.
- Labor market flexibility:
  - Defined as −ln(1+x), where x is the employment rigidity index from “Doing Business” (2003).
  - Higher score = more flexible labor market.
- Expected roles: more arm’s length financial systems and more flexible labor markets should augment impact of increased credit availability on industries dependent on external finance.

### Regression results (industry-level summary)
- General finding:
  - Increased availability of credit is associated with reductions in costs of external finance primarily in countries with:
    - more arm’s length–based financial systems, and
    - more flexible labor markets.
- Selected coefficient patterns (from Table 1 and related tables; t-statistics based on robust SE clustered by country; *, **, *** denote 10%, 5%, 1% significance):
  - Dependence on External Finance * Change in Liab to GDP: Coef est. 0.04; t-stat. 0.98.
  - Dependence on External Finance * Change in Liab to GDP (alternate column): Coef est. -0.49; t-stat. -3.42***.
  - Dependence on External Finance * Financial Index: Coef est. 0.02; t-stat. 1.60; Coef est. -0.03; t-stat. -2.36**.
  - Dependence on External Finance * Financial Index * Change in Liab to GDP: Coef est. 1.35; t-stat. 4.04***; Coef est. 2.24; t-stat. 3.93***.
  - Dependence on External Finance * Change in Real Short-term Interest Rate: Coef est. -0.29; t-stat. -2.07*; similar negative estimates across columns.
  - Dependence on External Finance * Change in Real Long-term Interest Rate: Coef est. 0.32; t-stat. 2.74**; similar positive estimates across columns.
- Sample and specification notes:
  - Japan excluded to improve precision.
  - Observations with industry growth >20% or < -20% excluded in both periods.
  - # Observations: 355 per column; R-sq.: 0.63 per column.
  - Country fixed effects and industry fixed effects included.

### Economic magnitudes and interpretations
- Median average annual change in total liabilities to GDP ratio during 1995-2003: 3.8 percentage points.
- Median average annual growth of all industries in the sample: 2 percent.
- Using corporate liabilities improves measurement of credit conditions; empirical finding: changes in corporate liabilities are more strongly associated with dependent industries’ relative growth than changes in total liabilities.
- Example magnitude (corporate liabilities interaction with financial system):
  - In a financial system at the 75th percentile on the financial index, every percentage point by which corporate liabilities outgrow GDP per year on average during 1995-2003 is associated with an increase in average annual growth of an industry at the 75th percentile in dependence (vs. an industry at the 25th percentile) of 0.12 percentage points higher.
- Labor market flexibility interaction:
  - In countries with more flexible labor markets, increases in total private sector borrowings are strongly associated with faster relative growth of industries reliant on external funding (significant at 1 percent level).
  - Association between increases in liabilities and dependent-industry relative growth is much stronger for corporate liabilities than total liabilities when allowed to vary with labor market flexibility.
- Collinearity:
  - Correlation between financial index and labor market flexibility: 0.78, which complicates simultaneous inference.
  - Negative correlation between recent increases in corporate liabilities and the financial index: -0.17 (reverse-causality concern considered unlikely).

### Resource allocation, productivity, and heterogeneity
- Main resource-allocation finding:
  - Expansion in private-sector liabilities is associated with improved resource allocation: industries with higher underlying growth trends (1987-1995) exhibited greater subsequent relative growth in countries with faster recent buildup in total liabilities.
  - Stronger results when liabilities restricted to corporate liabilities: coefficient statistically significant at the 1 percent level and implies for every percentage point by which corporate liabilities outgrow GDP each year on average during 1995-2003, the relative underlying growth trend of industries is enhanced by about 11 percent.
  - Quantified heterogeneity by financial system:
    - In a financial system at the 75th percentile on the financial index, for every percentage point by which corporate liabilities outgrew GDP each year on average during 1995-2003, the relative underlying growth trend of industries is enhanced by about 17 percent; at the 25th percentile the corresponding figure is 3 percent.
- Productivity effects:
  - Increases in corporate liabilities appear to improve labor productivity of growing industries.
  - Key statistics:
    - Observed 0.08 percentage points difference in the increase in productivity growth (Table 3, regression 6).
    - Median average annual change in the corporate liabilities to GDP ratio is 1.4 percentage point.
    - Median difference in underlying growth trend between an industry at the 75th percentile and one at the 25th percentile is 3.1 percentage points.
    - Median annual productivity growth of all the sample industries is 1.91 percent during 1995-2003.
    - In countries with a more arm’s length-based financial system (75th percentile), for every percentage point rise in the corporate liabilities to GDP ratio each year on average during 1995-2003, one percentage point difference in average annual growth during 1987-1995 would translate into 0.16 percentage points difference in the increase in productivity growth.
- Selected regression coefficients related to resource allocation and productivity (t-statistics shown):
  - Growth in 1987-1995 * Change in Liab to GDP: Coef est. 11.90; t-stat. 3.13***.
  - Growth in 1987-1995 * Change in Liab to GDP: Coef est. 45.09; t-stat. 5.60***.
  - Growth in 1987-1995 * Labor Mkt Flexibility * Change in Liab to GDP: Coef est. 10.57; t-stat. 4.07***; Coef est. 8.55; t-stat. 2.06*.
  - Growth in 1987-1995 * Financial Index * Change in Liab to GDP: Coef est. 107.52; t-stat. 2.20**.
  - Growth in 1987-1995 * Change in Real Short-term Interest Rate: Coef est. -13.76; t-stat. -2.73**; other negative estimates also reported.
  - Productivity growth baseline coefficients: Coef est. 0.10; t-stat. ~1.02 across columns.

### Role of financial system structure and labor market flexibility
- Financial system:
  - Positive association between increases in corporate liabilities and productivity of growing industries is significantly stronger in countries with more arm’s length-based financial systems (Table 3, regression 8).
  - Correlations (excluding Japan) of increases in arm’s length content with increases in total private sector liabilities and corporate liabilities are -0.32 and -0.43, respectively.
- Labor market flexibility:
  - Labor market flexibility facilitates faster relative growth of growing industries, likely via more fluid redeployment of labor.
  - Relative growth of growing industries is strongest in countries with both large increases in corporate borrowing and high labor market flexibility.
  - Labor market flexibility strengthens the association between corporate-liability increases and growing industries’ relative productivity growth.
- Joint inference hampered by high correlation (0.78) between financial index and labor market flexibility, but both appear important.

### Aggregate vs. industry-level contrasts
- Aggregate facts:
  - Since around 1995, private-sector financial liabilities as a share of GDP among 18 advanced economies jumped on average by over 40 percentage points.
  - Large cross-country variation: Portugal more than doubled its liabilities-to-GDP ratio; Japan, Canada and Finland showed very small increases or declines.
  - Aggregate macro analysis shows no clear relationship between liability accumulation and economic growth, potentially due to omitted-variable confounding.
- Industry-level evidence:
  - Identifies channels through which financial development benefits the real economy: reductions in costs of external finance and improved resource allocation that strengthen value-added growth and productivity growth of growing industries — particularly in countries with arm’s length financial systems and flexible labor markets.

### Policy implications and recommendations
- Financial deepening alone is unlikely to deliver full benefits associated with financial development.
- How financial capital is intermediated and ease of labor redeployment matter for the economic value of newly created credit.
- Broad-based policies likely to enhance economic contributions of the financial sector:
  - deepen firms’ borrowing capacity,
  - hasten transition toward a more arm’s length financial system, and
  - relax labor market regulations.
- Caution: transition to a more arm’s length-based financial system may be associated with higher private-sector vulnerability to cyclical changes; accompany transitions with appropriate prudential measures and reforms in labor and product markets to mitigate adverse cyclicality effects.

### Supporting industry- and appendix-level facts
- Selected entries from Table A1 (industry dependence on external finance):
  - Chemicals and chemical products: 246.2.
  - Real Estate, Renting and business activities, computer, R&D services 70-74: 3.35.
  - Post and telecommunications 64: 1.67.
  - Electrical and optimal equipment 30-33: 1.62.
  - Wood and products of wood and cork 20: -0.45.
  - Fabricated metal products except machinery and equipment 28: -0.25.
  - Construction 45: -0.19.
- Table A2 lists fastest- and slowest-growing industries, 1987-1995, by country (detailed ISIC entries provided in the source).

*Source: Excerpt from IMF working paper _wp07118 (I. INTRODUCTION; II. RECENT DEVELOPMENTS IN FINANCIAL LIABILITIES; selected tables and regression results as provided).*

### References ................................................................................................29

### I. INTRODUCTION

### Major findings and objectives
- The average level of private sector liabilities increased from 114 percent of GDP in 1995 to 143 percent of GDP in 2004 for a group of 18 advanced economies.
- The rate of buildup in private sector liabilities outpaced real GDP growth by more than 2.5 percentage points on average every year during the period.
- Between 1986 and 1995 the average private-credits-to-GDP ratio of these countries increased by a mere 0.8 percentage points per year.
- Cross-country heterogeneity intensified after 1995: the cross-country standard deviation of changes in the credit-to-GDP ratio doubled between the pre- and post-1995 periods (from 0.13 to 0.27).
- Portugal’s private sector liabilities increased by 90 percentage points of GDP between 1995 and 2004; Japan and Canada experienced a decline in the ratio; Finland and France saw increases of 9 and 15 percentage points, respectively.
- Private sector financial liabilities are defined as all loans and securities liabilities (including derivatives but excluding shares) incurred by households and nonfinancial corporations; data are consolidated from OECD national accounts and Eurostat financial accounts.

### Purpose and contribution
- Two-fold purpose: (1) document basic facts about the pattern of accumulation of private sector liabilities in advanced economies during 1995-2005; (2) examine economic implications of recent buildups in liabilities at the industry level, focusing on financial intermediation effectiveness and conditions under which it could be enhanced.
- Contribution: exploits cross-country variation and analyzes impacts of changes—rather than levels—of credits on industry performance, thus controlling for unobserved country fixed factors and yielding focused policy implications.
- Equity finance is not explicitly examined; net equity finance drifted up by only about 1.5 percent of GDP over the last 10 years for the sample, while corporate borrowings increased by 20 percent of GDP during the same period.

### II. RECENT DEVELOPMENTS IN FINANCIAL LIABILITIES

### A. Households Set the Pace and Continue to Display Large Cross-Country Variation
- Households accounted for more than half (53.7 percent) of the overall increase in private sector liabilities.
- In 1995, household liabilities composed on average 41.7 percent of total private sector liabilities; by the period’s end household liabilities’ share jumped to 46.2 percent.
- For 12 of the 18 countries, the percentage of total liabilities going to the household sector increased.
- Average household liabilities increased from 48 percent of GDP in 1995 to 68 percent of GDP in 2005.
- All sample countries experienced a rise in household liabilities as a share of GDP, including Japan and Canada where total liabilities grew slower than GDP.
- Countries with the sharpest household liability increases (more than 30 percent of GDP) include Portugal, the Netherlands, Australia, and Spain.
- Japan, Belgium, and Canada: household liabilities rose by less than 5 percent of GDP.
- Cross-country dispersion: household indebtedness levels range from 110 percent of GDP in Denmark to 28 percent of GDP in Italy in 2005; disparity neither significantly widened nor narrowed over the last 10 years.
- The 12 countries whose household liabilities show an increase in share are DNK, US, AUS, NLD, ESP, SWE, FIN, PRT, ITA, GRC, JPN and CAN.

### B. Corporate Sector Debt Expands Strongly From a High Base and Loans Remain Dominant
- Average corporate liabilities as a share of GDP increased from 63 percent in 1995 to 75 percent in 2005.
- Excluding Japan, corporate liabilities increased from 58 percent to 73 percent of GDP.
- Portugal, the UK and Spain showed the largest increases in corporate borrowings as a share of GDP, by more than 30 percentage points of GDP.
- Japan and Canada: corporate liabilities grew slower than GDP.
- Australia and Finland: corporate indebtedness rose by less than 5 percentage points of GDP.
- Cross-country variation (excluding Japan): standard deviation in corporate liabilities was 11 percentage points in 1995 and 15 percentage points in 2004; corresponding household figures are 20 percentage points in 1995 and 25 percentage points in 2004.
- Loans remain the core source of corporate borrowed funds: loans accounted for 81 percent of total corporate liabilities in 2004, down from 85 percent in 1995.
- Finland and Sweden experienced the largest shift away from loans (loans’ share of corporate liabilities declined by 10 percentage points).
- Securities account for more than 25 percent of total corporate borrowings in France and the Anglo-Saxon countries.

### C. Synchronization and Cross-Country Variation
- Household and corporate liabilities expanded in sync in many countries; there is a large variation across countries in the magnitude of increases.
- The paper highlights that recent increases in corporate borrowings are associated with reductions in costs of external finance and improvements in resource allocation, particularly in countries with:
  - a more arm’s length–based financial system (transactions based on publicly available information rather than long-standing relationships), and
  - a less rigid labor market.
- The association is stronger where the financial system has a greater arm's length content and the labor market is more flexible.
- The paper abstracts from effects of credit growth on financial stability; prior literature finds rapid credit growth often precedes balance-of-payments and/or banking crises, but Moreno-Badia (2007) finds no direct adverse effects of credit growth on banking-sector stability across the euro area in recent years.

*Excerpt from the IMF working paper (I. INTRODUCTION; II. RECENT DEVELOPMENTS IN FINANCIAL LIABILITIES).*

### 3.7 percentage points, which is almost matched by the standard deviation (3.3 percentage

### _wp07118 - 3.7 percentage points, which is almost matched by the standard deviation (3.3 percentage

### Cross-country variation in private-sector liabilities (1995-2005 or latest)
- Average and dispersion:
  - Cross-country variation in changes in total private-sector liabilities: 3.7 percentage points.
  - Standard deviation: 3.3 percentage points.
- Mechanical contributors to large cross-country variation:
  - Large differences in changes in liabilities in the household sector.
  - Large differences in changes in liabilities in the corporate sector.
  - Cross-country concurrence between changes in liabilities in the two sectors.
- Quantitative decomposition of variance in changes in total liabilities:
  - Corporate liabilities account for 37 percent of the cross-country variation (variance).
  - Household liabilities account for 29 percent of the cross-country variation (variance).
  - The concurrence between changes in liabilities in the two sectors contributes the rest.
  - Footnote: Excluding Japan (which experienced a very sharp decline in corporate liabilities as a share of GDP), corporate liabilities would have contributed 28 percent, and household liabilities 41 percent, of the overall variance.

### Observed patterns in sectoral liability changes
- General patterns 1995-2005:
  - Increases in liabilities as a share of GDP generally occurred in both household and corporate sectors, with exceptions of Canada and Japan.
  - Countries with more rapid expansion in corporate liabilities tend to have larger increases in household liabilities (positive association shown in Figure 6).
- Exceptions and notable country behavior:
  - Japan experienced a very sharp decline in corporate liabilities as a share of GDP.

### Macroeconomic relationships: expenditure composition and growth
- Private consumption and investment:
  - Buildups in total liabilities do not appear to raise the expenditure share of private consumption (Figure 7A).
  - Increases in corporate liabilities seem associated with rises in the expenditure share of productive investment (the non-housing component of gross fixed capital formation) (Figure 7B).
- Growth:
  - Overall, no relationship between increases in liabilities and economic growth appears (Figure 8).
  - Simple growth regressions (not reported here) explicitly taking care of effects of economic convergence also fail to find any significant relationships between buildups in liabilities and economic growth. (Footnote)
- Potential masking factors:
  - Unobserved time-varying country-level factors may affect growth and be correlated with liabilities buildups.
  - Reverse causality: macroeconomic developments might affect private sector indebtedness (e.g., faster growth might lower measures of indebtedness since GDP enters as denominator; temporary increases in cash flows might reduce need for borrowings).

### Industry-level approach: rationale and empirical strategy
- Motivation:
  - To reduce bias from omitted variables and reverse causality, analysis focuses on mechanisms through which financial development affects growth using industry-level data.
  - Disaggregated analysis examines differential effects of finance on industries with different characteristics (e.g., dependence on external finance).
- Theoretical basis:
  - Agency problems (information asymmetry, moral hazard) make external finance costlier than internal funds, potentially inhibiting firm growth.
  - Industries differ in "dependence" on external finance due to technological factors (capital intensity, cash flow generation).
  - Rajan and Zingales (1998) logic: if buildups in liabilities reduce costs of external finance, industries more reliant on external finance should exhibit relatively higher growth in countries with faster liability buildups.

### Basic econometric specification (industry-level test)
- Regression form (variant of Rajan and Zingales, 1998):
  - Dependent variable: change in industry i in country c: (average annual real growth in value-added 1995-2003) minus (average annual real growth 1987-1995).
  - Key regressors and definitions:
    - iE: country-invariant measure of dependence on external finance of industry i.
    - cLΔ: average annual change in liabilities as a share of GDP during 1995-2003 of country c.
    - cX: country characteristics (benchmark includes measure of the nature of a country’s financial system, labor market flexibility, and changes in real short-term and long-term interest rates between 2003-2005 and 1995-1997).
    - ccD}{ and iiD}{: full sets of country and industry dummies.
  - Hypothesis: if buildups in liabilities reduce costs of external finance, estimated coefficient on interaction (dependence * change in liabilities) (β̂1) should be positive.
- Panel approach:
  - Uses cross-industry dimension instead of time-series to reduce bias from time-varying country-specific factors.
  - Eliminates time-series dimension to reduce endogeneity.

### Measure of dependence on external finance
- Source and construction:
  - Measure taken from de Serres, Kobayakawa, Slok and Vartia (2006), constructed in the spirit of Rajan and Zingales (1998).
  - Defined from US listed firms as (capital expenditures minus cash flows from operations) divided by capital expenditures.
  - de Serres et al.’s measure:
    - Based on 2-digit level data and covers services industries as well as manufacturing.
    - Constructed with data covering 1990-2003 (more recent than Rajan and Zingales (1980s) measure).
  - Assumption: technological factors governing industries’ reliance on external finance persist across countries; US listed firms face relatively small costs of external finance.

### Controls: financial system type and labor market flexibility
- Financial system measure:
  - 1995 financial index from IMF (2006) World Economic Outlook (September 2006).
  - Aggregate of three subindices: degree of traditional banking intermediation, level of development of new financial intermediation, and importance of capital markets.
  - Higher score implies more arm’s length-based financial system (more competitive banking sector, more financial innovations, more developed capital markets).
- Labor market flexibility measure:
  - Negative of the log of the employment rigidity index from “Doing Business” (2003) survey by the World Bank: –ln(1+x), where x is the employment rigidity index.
  - Index averages difficulties of hiring, firing, and adjusting work hours.
  - A higher score refers to a more flexible labor market.
- Expected roles:
  - More arm’s length-based financial systems and more flexible labor markets should augment the impact of increased availability of credit on industries that depend on external finance.

### Regression results (summary)
- General finding:
  - Increased availability of credit is associated with a general reduction in costs of external finance, but mostly only in countries with:
    - A more arm’s length-based financial system.
    - A more flexible labor market.
- Specifics from Table 1 (industry growth in 1995-2003; t-statistics based on robust standard errors clustered by country; *, **, *** denote statistical significance at 10%, 5% and 1%):
  - Dependence on External Finance * Change in Liab to GDP:
    - Coef est. 0.04; t-stat. 0.98 (Corporate Liabilities column).
    - Coef est. -0.49; t-stat. -3.42*** (Corporate Liabilities? table shows multiple columns).
    - Coef est. 0.05; t-stat. 0.67; Coef est. -0.88; t-stat. -3.45*** (Total Priv Sector Liabilities columns).
  - Dependence on External Finance * Financial Index:
    - Coef est. 0.02; t-stat. 1.60.
    - Coef est. -0.03; t-stat. -2.36**.
    - Coef est. 0.02; t-stat. 1.95*.
    - Coef est. -0.01; t-stat. -0.98.
  - Dependence on External Finance * Financial Index * Change in Liab to GDP:
    - Coef est. 1.35; t-stat. 4.04***.
    - Coef est. 2.24; t-stat. 3.93***.
  - Dependence on External Finance * Change in Real Short-term Interest Rate:
    - Coef est. -0.29; t-stat. -2.07*.
    - Coef est. -0.27; t-stat. -2.20**.
    - Coef est. -0.29; t-stat. -2.07*.
    - Coef est. -0.28; t-stat. -2.58**.
  - Dependence on External Finance * Change in Real Long-term Interest Rate:
    - Coef est. 0.32; t-stat. 2.74**.
    - Coef est. 0.23; t-stat. 2.13**.
    - Coef est. 0.29; t-stat. 2.90**.
    - Coef est. 0.23; t-stat. 2.72**.
- Sample and specification notes:
  - All industry-level data (except measure of industry dependence) from Groningen Growth and Development Center sectoral database.
  - Japan excluded in regressions to gain precision.
  - Observations excluded where industry growth in country was >20% or < -20% during 1987-1995 or 1995-2003.
  - # Observations: 355 (per column).
  - R-sq.: 0.63 (per column).
  - Country fixed effects and Industry fixed effects included.

### Economic interpretation and magnitudes
- Interaction with financial system:
  - In more arm’s length-based financial systems, increases in total liabilities are more strongly associated with general reductions in the costs of external finance (Table 1, regression 2).
  - The magnitude of this association is described as considerable in economic terms.
- Illustrative example (text truncated in source; available partial statement):
  - “Let’s take as example a financial system that would rank at the 75th percentile among the sample countries on the financial index. For every percentage point by which total liabilities outgrow GDP per year on average during 1995-2003, the increase in average annual growth of an industry ranking at the 75th percentile in terms of dependence on external finance would be” — (source text cuts off; no further numeric completion provided).

*Italic: Source: Excerpt from IMF working paper content (figures, tables and notes as provided in the supplied PDF text).*

### 0.1 percentage points higher than an industry ranking at the 25

### 0.1 percentage points higher than an industry ranking at the 25

### Liabilities, costs of external finance, and industry growth
- Median average annual change in total liabilities to GDP ratio during 1995-2003 was 3.8 percentage points; median average annual growth of all industries in the sample was 2 percent.
- Using corporate rather than total liabilities is expected to improve accuracy in measuring credit conditions faced by firms, even if total liabilities raise precision.
- Empirical finding: changes in corporate liabilities are more strongly associated with dependent industries’ relative growth than changes in total liabilities (Table 1, regression 3).
- Financial system interaction:
  - In a financial system ranked at the 75th percentile on the financial index, every percentage point by which corporate liabilities outgrow GDP per year on average during 1995-2003 is associated with an increase in average annual growth of an industry at the 75th percentile in dependence (vs. an industry at the 25th percentile) of 0.12 percentage points higher.
  - In this specification, the type of financial system itself does not significantly affect changes in industries’ growth.
- Labor market flexibility interaction:
  - Neither labor market flexibility nor changes in total liabilities alone are related to relative growth of dependent industries, but their interaction is significant (Table 2, regression 2).
  - In countries with more flexible labor markets, more total private sector borrowings are strongly (at 1 percent significance level) associated with faster relative growth of industries reliant on external funding.
  - The association between increases in liabilities and dependent-industry relative growth is much stronger for corporate liabilities than for total liabilities when allowed to vary with labor market flexibility (Table 2, regression 4).
- Joint inclusion of financial system type and labor market flexibility:
  - Effects of each variable on the association between increased liabilities and relative growth are similar but smaller when both are included (Table 2, regressions 5 and 6).
  - Collinearity concern: correlation between the financial index and labor market flexibility is 0.78, which may preclude meaningful simultaneous inference.

### Heterogeneity across financial systems and labor markets
- Summary conclusion:
  - In countries with more arm’s length-based financial systems and more flexible labor markets, recent increases in corporate liabilities partly reflect general reductions in costs of external finance, disproportionately benefiting more dependent industries.
  - In countries with more relationship-based financial systems and more rigid labor markets, there is no evidence of this effect; increases in corporate liabilities might have even led to slower relative growth of dependent industries.
- Note on endogeneity concern:
  - The negative correlation between recent increases in corporate liabilities and the financial index is -0.17, suggesting the reverse-causality concern (that fast-growing dependent industries cause higher corporate liabilities and are easier to finance in arm’s length systems) is not plausible.

### Resource allocation: expanding liabilities and allocative efficiency
- Theoretical channels:
  - Financial development can lower the costs of risks and encourage financing of projects with higher mean returns (even if riskier).
  - Banking competition can reduce vested interests in incumbents and promote funding of more productive entrants.
- Empirical approach:
  - Basic test: whether increases in liabilities strengthen developments of growing industries (industries’ underlying growth trend measured by value-added growth in 1987-1995).
  - If increased borrowing is related to better resource allocation, the coefficient on the interaction of underlying growth trend and change in liabilities should be positive.
- Main results:
  - Expansion in private-sector liabilities is associated with improved resource allocation: increases in relative growth of industries with higher underlying growth trends are greater in countries with faster recent buildup in total liabilities (Table 3, regression 1).
  - Results are stronger when liabilities are restricted to corporate liabilities: the coefficient is statistically significant at the 1 percent level and implies that for every percentage point by which corporate liabilities outgrow GDP each year on average during 1995-2003, the relative underlying growth trend of industries is enhanced by about 11 percent (Table 3, regression 2).
    - Example interpretation preserved from source: the increase in growth of an industry that grew by 2x% in the earlier period would be 0.11x percentage points higher than that of an industry that previously grew by x%.
  - Interaction with financial system:
    - The interaction of the underlying growth trend and the financial index enters significantly: growing industries are more likely to subsequently grow faster in countries with more arm’s length-based financial systems.
    - Quantified heterogeneity: in a financial system at the 75th percentile on the financial index, for every percentage point by which corporate liabilities outgrew GDP each year on average during 1995-2003, the relative underlying growth trend of industries is enhanced by about 17 percent; at the 25th percentile on the financial index the corresponding figure is 3 percent.
  - Productivity effects:
    - Increases in corporate liabilities appear to improve labor productivity of growing industries: for every percentage point by which corporate liabilities outgrew GDP each year on average during 1995-2003, one percentage point higher average annual growth during 1987-1995 would translate into (result continued in source).
- Sample and robustness notes:
  - Outlying Japan is excluded from estimations.
  - Observations associated with abnormal real value-added or productivity growth (more than 20% or less than -20% per year on average) are excluded.
  - Financial sector, public sector, industries heavily reliant on natural resources, private household businesses are dropped.
  - Any industry whose average value-added accounts for less than 0.5% of the country’s total value-added is excluded.
  - T-statistics are based on robust standard errors clustered by country.
  - *, **, *** denote statistical significance at 10%, 5% and 1% level, respectively.

*Source: _wp07118 - 0.1 percentage points higher than an industry ranking at the 25 (PDF chapter/section).*

### 0.08 percentage points difference in the increase in productivity growth (Table 3, regression

### _wp07118 - 0.08 percentage points difference in the increase in productivity growth (Table 3, regression

### Key empirical findings
- An observed 0.08 percentage points difference in the increase in productivity growth (Table 3, regression 6).
- Median average annual change in the corporate liabilities to GDP ratio is 1.4 percentage point.
- Median difference in underlying growth trend between an industry at the 75th percentile and one at the 25th percentile is 3.1 percentage points.
- Median annual productivity growth of all the sample industries is 1.91 percent during 1995-2003.
- In countries with a more arm’s length-based financial system (75th percentile in arm’s length content), for every percentage point rise in the corporate liabilities to GDP ratio each year on average during 1995-2003, one percentage point difference in average annual growth during 1987-1995 would translate into 0.16 percentage points difference in the increase in productivity growth.

### Regression evidence (selected coefficients from Table 4 and related regressions)
- Growth in 1987-1995 * Change in Liab to GDP: Coef est. 11.90, t-stat. 3.13***
- Growth in 1987-1995 * Change in Liab to GDP: Coef est. 45.09, t-stat. 5.60***
- Growth in 1987-1995 * Change in Liab to GDP: Coef est. 27.74, t-stat. 0.79
- Growth in 1987-1995 * Change in Liab to GDP: Coef est. 8.47, t-stat. 2.16**
- Growth in 1987-1995 * Change in Liab to GDP: Coef est. 39.78, t-stat. 5.36***
- Growth in 1987-1995 * Change in Liab to GDP: Coef est. -25.00, t-stat. -0.88

- Growth in 1987-1995 * Labor Mkt Flexibility: Coef est. 0.14, t-stat. 2.60**
- Growth in 1987-1995 * Labor Mkt Flexibility: Coef est. 0.03, t-stat. 0.51
- Growth in 1987-1995 * Labor Mkt Flexibility: Coef est. 0.02, t-stat. 0.25
- Growth in 1987-1995 * Labor Mkt Flexibility: Coef est. 0.14, t-stat. 2.73**
- Growth in 1987-1995 * Labor Mkt Flexibility: Coef est. 0.03, t-stat. 0.58
- Growth in 1987-1995 * Labor Mkt Flexibility: Coef est. 0.04, t-stat. 0.52

- Growth in 1987-1995 * Labor Mkt Flexiblity * Change in Liab to GDP: Coef est. 10.57, t-stat. 4.07***
- Growth in 1987-1995 * Labor Mkt Flexiblity * Change in Liab to GDP: Coef est. 8.55, t-stat. 2.06*
- Growth in 1987-1995 * Labor Mkt Flexiblity * Change in Liab to GDP: Coef est. 9.98, t-stat. 3.81***
- Growth in 1987-1995 * Labor Mkt Flexiblity * Change in Liab to GDP: Coef est. 3.49, t-stat. 1.07

- Growth in 1987-1995 * Financial Index: Coef est. 0.21, t-stat. 0.24
- Growth in 1987-1995 * Financial Index: Coef est. -0.32, t-stat. -0.34

- Growth in 1987-1995 * Financial Index * Change in Liab to GDP: Coef est. 26.57, t-stat. 0.47
- Growth in 1987-1995 * Financial Index * Change in Liab to GDP: Coef est. 107.52, t-stat. 2.20**

- Growth in 1987-1995 * Change in Real Short-term Interest Rate: Coef est. -7.37, t-stat. -1.22
- Growth in 1987-1995 * Change in Real Short-term Interest Rate: Coef est. -6.72, t-stat. -1.59
- Growth in 1987-1995 * Change in Real Short-term Interest Rate: Coef est. -7.36, t-stat. -1.61
- Growth in 1987-1995 * Change in Real Short-term Interest Rate: Coef est. -13.76, t-stat. -2.73**
- Growth in 1987-1995 * Change in Real Short-term Interest Rate: Coef est. -0.85, t-stat. -2.04*
- Growth in 1987-1995 * Change in Real Short-term Interest Rate: Coef est. -9.07, t-stat. -2.39**

- Growth in 1987-1995 * Change in Real Long-term Interest Rate: Coef est. 3.78, t-stat. 0.92
- Growth in 1987-1995 * Change in Real Long-term Interest Rate: Coef est. 0.40, t-stat. 0.12
- Growth in 1987-1995 * Change in Real Long-term Interest Rate: Coef est. 0.20, t-stat. 0.05
- Growth in 1987-1995 * Change in Real Long-term Interest Rate: Coef est. 1.20, t-stat. 0.28
- Growth in 1987-1995 * Change in Real Long-term Interest Rate: Coef est. -1.97, t-stat. -0.54
- Growth in 1987-1995 * Change in Real Long-term Interest Rate: Coef est. -3.77, t-stat. -1.17

- Growth in 1987-1995 (constant terms shown): Coef est. 0.21, t-stat. 1.49; Coef est. -0.16, t-stat. -1.26; Coef est. -0.31, t-stat. -0.59; Coef est. 0.03, t-stat. 0.20; Coef est. -0.32, t-stat. -2.89**; Coef est. -0.22, t-stat. -0.36

- Productivity Growth in 1987-1995: Coef est. 0.10, t-stat. 1.02; Coef est. 0.10, t-stat. 1.01; Coef est. 0.10, t-stat. 1.03

Notes on the regressions:
- Labor Market Flexibility is defined as -ln(rigidity+1), where rigidity is the employment rigidity index from "Doing Business," 2003 edition.
- The t-statistics are based on robust standard errors clustered by country.
- *, **, *** denote statistical significance at 10%, 5% and 1% level, respectively.
- Results on resource allocation are robust to exclusion of industry fixed effects.

### Role of financial system structure
- The positive association between increases in corporate liabilities and the productivity of growing industries is significantly stronger in countries with a more arm’s length-based financial system (Table 3, regression 8).
- The effect of an arm’s length financial system on augmenting the liabilities–allocative-efficiency link is analogous to the effect of labor market flexibility.
- Correlations (excluding outlying Japan) of increases in the arm’s length content of a financial system with increases in total private sector liabilities and increases in corporate liabilities are -0.32 and -0.43, respectively.

### Role of labor market flexibility
- Labor market flexibility facilitates growing industries to grow faster (Table 4, regression 1), likely via more fluid redeployment of labor from waning to booming industries.
- The relative growth of growing industries is strongest in countries with both large increases in corporate borrowing and high labor market flexibility (Table 4, regression 2).
- Labor market flexibility strengthens the association between increases in corporate liabilities and growing industries’ relative productivity growth (Table 4, regressions 5).
- High correlation between financial system measures and labor market rigidity prevents meaningful simultaneous inference of both effects, but both variables appear to play separate, important roles.
- When both financial-system arm’s length content and labor market flexibility enter regressions together (Table 4, regressions 3 and 6), each enhancement effect on the liabilities–performance association is similar to, but weaker than, when the other is excluded.

### Aggregate and industry-level contrasts
- Since around 1995, private-sector financial liabilities as a share of GDP among a group of 18 advanced economies jumped on average by over 40 percentage points, with large cross-country variation (example: Portugal more than doubled its liabilities to GDP ratio; Japan, Canada and Finland showed very small increases or declines).
- Aggregate macroeconomic analysis shows no clear relationship between accumulation of private sector liabilities and economic growth in the last 10 years, potentially due to omitted-variable confounding.
- Industry-level evidence identifies significant associations between recent increases in corporate borrowings and channels through which financial development benefits the real economy: reductions in costs of external finance and improved resource allocation that strengthen value-added growth and productivity growth of growing industries — particularly in countries with arm’s length financial systems and flexible labor markets.

### Policy implications and recommendations
- Financial deepening (increases in private-sector financial liabilities) alone is unlikely to be sufficient to deliver full benefits traditionally associated with financial development.
- How financial capital is intermediated (arm’s length vs. relationship-based) and the ease of labor redeployment have important implications for the economic value of newly created credits.
- Broad-based policies that:
  - deepen firms’ borrowing capacity,
  - hasten transition toward a more arm’s length financial system, and
  - relax labor market regulations,
  are likely to generate greater economic contributions from the financial sector.
- The transition to a more arm’s length-based financial system may be associated with higher private-sector vulnerability to cyclical changes; such transitions should be accompanied by appropriate prudential measures and reforms in labor and product markets to mitigate adverse cyclicality effects.

### Supporting industry- and appendix-level facts
- Table A1 lists industry dependence on external finance (selected entries): Chemicals and chemical products 246.2; Real Estate, Renting and business activities, computer, R&D services 70-74 3.35; Post and telecommunications 64 1.67; Electrical and optimal equipment 30-33 1.62; Wood and products of wood and cork 20-0.45; Fabricated metal products except machinery and equipment 28-0.25; Construction 45-0.19.
- Table A2 lists fastest- and slowest-growing industries, 1987-1995, by country (detailed ISIC entries provided in the source).

*Source: _wp07118 (selected excerpts as provided).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07118.pdf_