## _wp07138

## Source details

**Canonical URL:** [_wp07138](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07138.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07138.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07138.pdf.json)

---

### I. Introduction — scope and research questions
- Focus: characterize firms’ activity in world equity markets (1989–2000) and classify firms as:
  - “international firms”: those that cross-list directly, have equity traded as DRs, or raise equity capital in international stock markets;
  - “domestic firms”: all other firms.
- Key questions:
  - Do firms from all countries use international capital markets?
  - Which countries and which firms participate, and how extensively?
  - Which country- and firm-level characteristics predict internationalization?

### Data, sample, and definitions
- Sample coverage and key counts:
  - 39,517 firms from 111 countries covering 1989–2000.
  - 2,546 international firms; maximum 30,552 firm-year observations for international firms.
  - 36,971 domestic firms; maximum 223,740 firm-year observations for domestic firms.
  - Total countries in broader tables: 205; countries with active domestic stock markets: 151; countries with international activity: 117; share of countries with international activity: 63.1%.
  - Number of firms listed in domestic markets: 36,971; number of international firms: 2,546; share of international firms: 6.9%.
- Data sources for international activity:
  - Bank of New York DR Directory (2,259 listed programs as of March 2005; 1,990 firms with active DR programs in 80 countries).
  - Euromoney: 8,795 operations from 5,665 firms in 86 countries (January 1983–April 2001).
  - NASDAQ, NYSE, LSE, FSE records used; three financial centers considered: New York, London, Frankfurt.
- Definition scope and limitations:
  - International firms include cross-listing, DR trading, or capital raising abroad; excludes measures of foreign ownership in local markets and cross-listings in other centers.
  - U.S. and U.K. excluded as sample countries because they are classified as international financial centers.

### Country-level methodology
- Country variables primarily from World Bank’s World Development Indicators; additional indexes: financial openness, law and order, investor protection.
- Estimation techniques:
  - Descriptive median tests (Mann-Whitney).
  - Random-effects panel regressions; Tobit models (censoring at zero) for share of international firms; fixed-effects panels as robustness checks.
  - Separate specifications for developed and developing countries; Heckman selection attempts noted but not central.

### Country-level empirical findings
- Prevalence and concentration:
  - Only a relatively small fraction of countries and firms use international markets.
  - About 63 percent of countries have some international activity; developed countries: 75.9 percent share with international activity.
  - Roughly seven percent of firms overall are international.
- Key country correlates of greater internationalization (consistent across panel and Tobit):
  - Larger GDP (Log of GDP coefficients positive and significant).
  - Higher GDP per capita.
  - Better macroeconomic stability (lower inflation).
  - Greater trade openness (Exports+Imports)/GDP positive and highly significant.
  - Greater financial/stock market liberalization associated with more internationalization.
- Institutional measures:
  - Mixed evidence: better investor protection and law-and-order sometimes associated with less internationalization (interpreted as firms in weaker institutional environments using internationalization to bond); results vary with specification and sample.
- Illustrative magnitudes (from reported regressions):
  - A one percent increase in country GDP → increase in proportion of international firms of about three percentage points (panel) and about six percentage points (Tobit) in one specification.
  - A one percent increase in GDP per capita → about four percentage points (panel) and about two percentage points (Tobit) increase in proportion of international firms.
  - A one percent rise in inflation → ratio two percentage points lower (panel) and six percentage points lower (Tobit).
  - A one standard deviation increase in stock market liberalization → increase of 0.7 percentage points (panel) and 2.3 percentage points (Tobit) in the proportion of international firms.
  - In another specification: a hypothetical increase in GDP yields a 4.5 percentage point increase in the ratio of international firms; a one standard deviation increase in institutional indices yields a decrease of ten percentage points in panel estimations (and smaller effects in Tobit).

### Firm-level methodology
- Firm-level data from Worldscope; core variables:
  - Size: total assets (also tested employees and sales).
  - Growth: growth in total sales revenue (two-year geometric average).
  - Performance: return on assets.
  - International activity: foreign sales / total sales.
- Estimation techniques:
  - Median and distributional comparisons (Kolmogorov-Smirnov).
  - Probit model predicting going-abroad decisions using initial-period information.
  - Cox proportional hazards model using time-varying information up to year before internationalization.
  - Before/after regressions with before and after dummies, controlling for industry, country, and year.

### Firm-level descriptive statistics and distributions
- Cross-sectional medians (All Countries):
  - Total Assets (Million U.S. dollars): Domestic Firms 275.33; International Firms 1,654.1.
  - Log (1+Sales Growth): Domestic 0.048; International 0.081.
  - Return on Assets: Domestic 0.037; International 0.054.
  - Foreign Sales / Total Sales: Domestic 0.000; International 0.290.
- Kernel / K-S test statistics (distributional differences between domestic and international firms):
  - Total Assets: D=0.324, P-value=0.000
  - Sales Growth: D=0.154, P-value=0.000
  - Return on Assets: D=0.098, P-value=0.000
  - Foreign Sales to Total Sales: D=0.260, P-value=0.000
- Concentration metrics:
  - International firms represent 28 percent of market capitalization but only seven percent of firms.
  - Share of value traded abroad: 18 percent overall.
  - Regional examples:
    - Developed countries: Market capitalization of international firms/GDP 61.8%; Market capitalization of international firms/Total Market Capitalization 46.9%; Value Traded Abroad/Value Traded Domestically 28.5%.
    - Latin America & Caribbean: Value Traded Abroad/Value Traded Domestically 61.2%.
    - Africa: Market capitalization of internationalized firms 9.5 percent (regional note).

### Firm-level regression results — selection and hazard analyses
- Probit (predictive margins; using 1993 values):
  - Size (Log Total Assets): positive and highly significant; marginal effects examples: 0.014 ***, 0.010 ***, 0.039 ***, 0.030 *** across panels.
  - Growth (Log (1+Sales Growth)): positive and significant in many specifications (e.g., 0.008 **, 0.009 ***, 0.019 *).
  - Foreign Sales to Total Sales: positive and significant (e.g., 0.017 ***, 0.013 ***, 0.071 *).
  - Return on Assets: positive but significance varies.
  - Economic magnitudes (Probit, country-dummies specification):
    - A one percent increase in total assets → ~1.4 percentage points higher probability of becoming international for an average firm.
    - A one percent increase in sales growth → ~0.8 percentage points higher probability.
    - A one standard deviation increase in foreign sales → ~0.4 percentage points higher probability.
    - For developed-country firms: a one percent size increase → ~1 percentage point probability increase.
    - For developing-country firms: a one percent size increase → ~4 percentage points probability increase.
- Cox proportional hazards (time-to-event):
  - Size hazard ratios: around 1.716 *** to 1.833 *** (statistically significant).
  - Growth hazard ratios: >1 and often significant (e.g., 1.306 ***, 1.339 ***).
  - Foreign Sales to Total Sales: large hazard ratios (e.g., 3.198 ***, 2.677 ***, 5.650 ***).
  - Return on Assets: hazard ratios >1 in several specs, significance varies.
  - Economic magnitude example: standard deviation of log total assets = 1.61; exponentiated effect yields a 140 percent increase in baseline probability of becoming international for a one standard deviation increase in size.
  - Ranking of economic importance: size largest, then foreign sales, then sales growth; return on assets smaller.

### Before-and-after (difference-in-differences style) firm dynamics
- Pre-internationalization (median regressions / Table 7):
  - Prospective international firms are larger, grow faster, have higher return on assets, and higher foreign sales shares than domestic peers.
  - Size example: Before dummy effect on Total Assets = 3,666.7 *** [9.516]; After dummy effect = 5,858.1 *** [14.479]; difference (b)-(a) = 2,191 (P-value 0.000).
  - Sales Growth: before 0.060 ***; after 0.022 ***; difference -0.038 (P-value 0.002).
  - Return on Assets: before 0.021 ***; after 0.001 (not significant); difference -0.021 (P-value 0.006).
  - Foreign Sales to Total Sales: before 0.116 ***; after 0.146 ***; difference 0.030 (P-value 0.002).
- Interpretation:
  - Firms that become international are already advantaged on size, growth, returns, and foreign sales prior to internationalization.
  - After internationalization, firms become even larger and increase foreign sales share; growth and returns relative to pre-internationalization levels may slow.

### Conclusions and policy-relevant implications
- Overall assessment:
  - Direct use of international equity markets is concentrated in a small subset of firms and countries; internationalization limited to a relatively small group.
  - Country determinants: larger, higher-income, more open economies with better macro stability and greater financial liberalization see more firm internationalization; institutional quality shows mixed effects (possible “bonding” motive for firms from weaker-institution countries).
  - Firm determinants: larger firms, firms with more foreign sales, faster-growing firms, and higher-return firms are more likely to internationalize.
- Policy and research directions enumerated:
  - Distinguish listing/trading-only from capital-raising in international markets; differentiate cross-listing forms (direct vs ADR/GDR).
  - Collect firm-level governance variables (ownership, directors, auditors) to evaluate governance role relative to internationalization.
  - Study spillovers to domestic firms (liquidity, market integration versus adverse liquidity effects).
  - Identify country and firm characteristic thresholds enabling international capital issuance to guide policies expanding global market access.
  - For domestic exchanges: monitor risk of migration of sufficiently developed-country firms to international markets and implications for local market capitalization and liquidity.
  - Tailor local capital market development recognizing small firms are unlikely to access international equity markets directly.

*Source: _wp07138 - Appendix Table 4. The set of regression results for all countries (upper panel) confirms most, although not all of — PDF chapter/section content provided.*

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

### _wp07138 - References..............................................................................................................

### I. INTRODUCTION — scope and research questions
- Financial globalization increased significantly during the last decade, with greater cross-border capital flows, tighter links among financial markets, and greater presence of foreign financial firms.
- Standard aggregate measures (gross capital flows, stocks of foreign assets and liabilities, degree of co-movement of returns) suggest unprecedented international financial integration, but do not reveal:
  - how many firms from how many countries actively participate;
  - what proportion of the corporate sector internationalizes;
  - why firms seek to internationalize.
- This paper complements existing literature by analyzing firms’ activity in world equity markets, classifying firms as:
  - “international firms”: those that participate in international stock markets by raising capital, cross-listing, and/or issuing depositary receipts in global markets;
  - “domestic firms”: all other firms.
- Key questions addressed:
  - Does intensity of internationalization imply firms from all countries use international capital markets?
  - From which countries are firms more likely to participate?
  - How extensive is the process in countries that internationalize?
  - Do country characteristics matter for the degree of internationalization? If so, which ones?
  - Within internationalizing countries, are participating firms a specific subset, and are their ex-ante characteristics different?

### Literature context and competing views
- Two opposing views on macroeconomic determinants of firm internationalization:
  - View A ("escape"/bonding): worse domestic fundamentals increase use of international markets; international markets attractive to firms from countries with weak institutional environments because they offer bonding to systems that better protect investor rights.
  - View B (supply-driven): better domestic fundamentals make firms more attractive to global suppliers of capital; thus better domestic fundamentals increase access to international capital markets.
- Microeconomic firm-level determinants discussed:
  - Fixed costs to accessing international markets may make firm size important (e.g., compliance with international accounting standards, minimum market capitalization requirements).
  - Growth opportunities may increase likelihood of going abroad (firms with large unrealized growth opportunities).
  - Foreign sales can serve as international collateral, relaxing borrowing constraints (Caballero and Krishnamurthy, 2001, 2002).
  - High returns on capital and corporate governance measures may influence propensity to seek equity capital abroad.

### Data and sample
- Compiled sample: 39,517 firms from 111 countries covering the period 1989–2000.
  - 2,546 international firms, accounting for a maximum of 30,552 firm-year observations.
  - 36,971 domestic firms, accounting for a maximum of 223,740 firm-year observations.
- For each firm: collected firm-level information such as size, growth, performance, and foreign trade activity.
- Compiled aggregate country information related to domestic stock market development, internationalization, and other commonly used country attributes.
- Analysis conducted:
  - Country-level: to understand how widespread internationalization is and how country characteristics relate to internationalization.
  - Firm-level: description of firms accessing international markets, estimations of probability of firms becoming international using cross-sectional and panel estimations, and comparisons of firm characteristics over time.

### Main empirical findings (as reported)
- Only a relatively small fraction of countries and firms use international markets.
- Country-level correlates of firms going abroad:
  - Firms more likely to go abroad are located in countries with:
    - bigger economies;
    - higher income levels;
    - better macroeconomic environments;
    - worse institutional environments.
- Firm-level correlates of international firms:
  - International firms tend to be:
    - larger;
    - growing faster;
    - having higher rates of return;
    - having more foreign sales.
- Interpretation:
  - Firms that internationalize are drawn from particular groups of countries and appear different from other firms.
  - Certain country- and firm-level characteristics are important factors in the internationalization process, suggesting barriers for many firms to participate directly in international capital markets.

### Contributions and improvements over previous work
- Dataset is large and diverse across countries, firms, and years, allowing characterization of both country- and firm-level factors related to likelihood of using international equity markets.
- Captures broader internationalization process by including:
  - access to more than one international stock exchange;
  - different forms of internationalization (capital raising, issuing, and trading).
- Period studied (1989–2000) spans 12 years during which many developing countries introduced reforms and experienced high internationalization up to the burst of the dotcom bubble.

### Paper organization and remaining sections
- Sections 2 and 3: data and methodology.
- Sections 4 and 5: country- and firm-level summary statistics and results on the extent of internationalization.
- Section 6: conclusion.
- Appendix and tables (as listed in the document):
  - Tables: Stock Market Internationalization by Region; Stock Market Internationalization and Country Characteristics; Country-Level Regressions; Differences Across Firms and Countries; Probit Regressions; Cox Regressions; Differences in Firm Characteristics Between International and Domestic Firms.
  - Figures: Internationalization Process; Distribution of Countries According to International Activity; Distribution of Firm Characteristics for Domestic and International Firms.
  - Appendix Tables: Series Description and Data Sources; Number of International and Domestic Firms by Country; Industry Classification; Fixed-Effect Country-Level Regressions.

*Source: _wp07138 - References (extract of the working paper content provided).*

### Appendix Table 1, while the list of countries covered and the groupings by income level are

### _wp07138 - Appendix Table 1, while the list of countries covered and the groupings by income level are

### Data sources for international activity
- U.S. sources:
  - Bank of New York DR Directory: covers depositary receipt (DR) programs in NYSE, NASDAQ, and AMEX; includes all currently active programs dating back to January 1956, with most after 1980. As of March 2005, there were a total of 2,259 listed programs. The resulting database accounts for 1,990 firms with active DR programs in 80 countries.
  - NASDAQ database: covers data on foreign companies listed on NASDAQ since 1989.
  - NYSE dataset: data on foreign companies cross-listed on NYSE.
  - Euromoney database: covers all operations of capital raised in international markets by firms; reports 8,795 operations from 5,665 firms in 86 countries during January 1983–April 2001.
- European sources:
  - London Stock Exchange (LSE) dataset: contains information on international activity for firms from 63 countries from January 1997 onward.
  - Frankfurt Stock Exchange (FSE) dataset: covers international firms for 68 countries for 1999 and 2000. Note: the FSE time series are much shorter, and firms are not classified as international if they internalize only through FSE because the exact timing of internationalization cannot be pinpointed. Instead, FSE data are used to remove firms from the pool of domestic firms.

### Definition and scope of "international firms"
- International firms are defined as those that:
  - cross-list directly, or
  - have their equity traded in the form of DRs, or
  - raise capital in international stock markets.
- The definition captures capital raising without listing and is more general than listing-only definitions.
- Exclusions and limitations:
  - The degree to which foreign investors hold shares or trade in local markets is not considered (most countries do not distinguish local vs. foreign investors).
  - Only three financial centers are considered (New York, London, Frankfurt); firms cross-listing in other developed-country centers are not treated as international in this study.
  - Firms that exit international status (e.g., end ADR programs or delist) are not studied; exits are noted as fairly limited over the period.

### Country-level data and macro variables
- Country variables mainly from World Bank’s World Development Indicators, except indexes on financial liberalization and institutional environment (see Appendix Table 1 for details).
- Country-level variables used in analysis:
  - GDP in U.S. dollars
  - GDP per capita
  - Inflation
  - Fiscal surplus
  - Trade openness (exports and imports relative to GDP)
  - Indexes on: financial openness, law and order, and degree of investor protection
- Additional country-level variables considered but not always used due to potential endogeneity: development of local financial markets and degree of cross-border equity flows.

### Firm-level data and micro variables
- Firm characteristics from Worldscope for all listed firms in local markets, including balance sheet and income statement information.
- Core firm-level variables employed (four headings: size, growth, performance, international activity):
  - Size proxy: total assets (also tested number of employees and total sales revenue with similar results)
  - Growth proxy: growth in total sales revenue
  - Performance proxy: return on assets
  - International real activity proxy: share of foreign sales over total sales
  - Sector/industry dummy variables to control for sectoral effects
- Other firm variables tested but not reported: growth in number of employees, assets per employee, research per employee, research over revenue, property, plant, and equipment growth, price-to-earnings ratio, price-to-book value, leverage, short-term debt to total debt, retained earnings, capital expenses over total assets. Corporate governance data at the individual-firm level were not available.

### Methodology — country-level and panel estimations
- Country-level analysis:
  - Descriptive statistics (median equality tests) to compare countries with more/less or any/no internationalization.
  - Random-effects (linear) panel and Tobit models to investigate determinants of the share of international firms out of total listed firms. Tobit accounts for dependent variable censoring at zero.
  - Separate estimations for developed and developing countries to test income-group effects.
  - Fixed-effects panel regressions estimated as an alternative technique. Fixed-effects Tobit models not estimated due to lack of consistent estimators.
  - Heckman selection models were attempted but difficult to implement; the specifications tried were generally supportive of main results though statistical significance in the main equation was typically less prevalent.

### Methodology — firm-level estimations
- Summary comparisons:
  - Compare medians between international and domestic firms for whole sample and for developed vs. developing country groups.
  - Also compare whether international (and domestic) firms differ across developed and developing countries.
- Predictive estimators:
  - Probit model: predicts going-abroad decision over a future time period using only information as of an initial date (conservative predictor; not affected by preparatory changes).
  - Cox proportional hazards model: uses all available information up to the year before internationalization. Hazard specification:
    - h(t) = h0(t) exp(β'X), where h0(t) is baseline hazard and β is vector of coefficients.
    - Semi-parametric: assumes hazard ratio h(t)/h0(t) is constant over time; no assumption on baseline hazard.
- Firm explanatory variables used (initial or previous year values):
  - logarithm of total assets
  - logarithm of sales growth
  - return on assets
  - proportion of sales abroad
  - sector dummies and time dummies (Cox model) to account for industry- and time-specific effects
- Country factors controlled in two ways:
  - include individual country dummies, or
  - include country characteristics (GDP, GDP per capita, inflation, stock market liberalization, trade, investor protection)
- Additional regressions:
  - Regressions with key firm characteristics as dependent variables and before/after internationalization dummies as independent variables, controlling for industry, country, and year effects. Coefficients on before dummy indicate differences prior to internationalization; after dummy indicates differences after internationalization.

### Key descriptive findings and statistics
- Bank of New York DR Directory: 2,259 listed programs as of March 2005; 1,990 firms with active DR programs in 80 countries.
- Euromoney database: reports 8,795 operations from 5,665 firms in 86 countries during January 1983–April 2001.
- LSE dataset: international activity for firms from 63 countries from January 1997 onward.
- FSE dataset: international firms for 68 countries for 1999 and 2000.
- Aggregate time-series patterns (Figure 1 descriptions):
  - Number of new international firms relative to listed firms:
    - Strong upward trend for developed countries.
    - Moderate increase then tapering for developing countries.
  - For developing countries: internationalization peaks in 1994 at 170 firms annually, then tapers off to below 80 new firms annually (partly reflecting privatization patterns).
  - For developed countries: internationalization almost doubles between 1998 and 2000 from 108 new international firms to 190 firms annually.
  - Market capitalization ratio (international firms to total market capitalization):
    - Developed economies: increases from less than 20 percent in 1989 to 47 percent in 2000.
    - Developing countries: increases to 19 percent by end of period (from a much lower base).
  - Trading abroad relative to domestic trading:
    - Developing countries: average trading ratio rises from a few percentage points to 52 percent in 1996; at the end of the period, 29 percent of trading takes place abroad on average.
    - Developed countries: average ratio of trading abroad to home trading rises from 15 to 28 percent over the same period.
- Cross-sectional and regional statistics for year 2000 (Table 1 descriptions):
  - Share of countries with some degree of internationalization: about 63 percent overall.
  - Share highest for developed countries at 76 percent.
  - About 2,500 international firms out of around 37,000 firms in the sample (about seven percent of firms overall).
  - Regional/income variation in share of international firms (share of international firms out of domestic listed firms):
    - Latin America: 15 percent
    - Developed countries: 12.7 percent
    - Central and Eastern Europe: 2 percent
  - Dataset coverage notes:
    - Ratios may overstate internationalization because domestic firm coverage is not complete (only domestic firms listed on main stock market are used).
    - Ratios may underestimate internationalization because internationalization into all financial centers is not covered.
- Vehicles for internationalization in dataset:
  - 45 percent of international firms cross-list or raise capital in a public exchange
  - 10 percent use ADRs/GDRs

*Source: _wp07138 - Appendix Table 1, while the list of countries covered and the groupings by income level are*

### Appendix Table 2 provides the total number of firms, distinguished by domestic and international firms, covered

### _wp07138 - Appendix Table 2 provides the total number of firms, distinguished by domestic and international firms, covered

### Data and sample coverage
- Appendix Table 2 provides the total number of firms, distinguished by domestic and international firms, covered for each country.
- The sample of domestically listed firms does not cover all firms within the country because several firm-specific indicators were needed that are not available for the whole array of domestic firms and because multiple listing outlets and multiple data sources restrict the sample.
- For ratios, the table provides the average of the within group ratios, not the ratio of the averages.

### Sectoral composition of international firms
- Most international firms are from the manufacturing, services, and finance sectors (see Appendix Table 3).
- Firms active in transportation and public utilities, public administration, and services see more peers migrate.
- Many international firms in developed countries come from the manufacturing and services sector.
- In developing countries many international firms come from the manufacturing, finance, and utilities sectors.
- Otherwise, no strong sectoral trends are identified.

### Market capitalization and value traded: concentration of international firms
- Internationalizing firms tend to be larger:
  - Share of market capitalization represented by international firms: 28 percent.
  - Share in numbers (international firms as percent of firms): seven percent.
- In terms of value traded:
  - Share of value traded abroad: 18 percent.
  - International firms are more heavily traded abroad than domestic firms.
- Regional variations:
  - For developed countries, share of market capitalization represented by internationalized firms: 47 percent.
  - For Africa, share of market capitalization represented by internationalized firms: 9.5 percent.
  - For Latin America, trading abroad is 61 percent of trading domestically.
  - For developed countries, trading abroad is 28 percent of trading domestically.
  - For most other regions, share of trading abroad is less than two percent.
- Country-level market capitalization examples:
  - Total market capitalization of Eastern European firms amounts to 11 percent of these countries’ GDP.
  - Total market capitalization for developed countries amounts to 109 percent of GDP.
- Implication: internationalization of a few firms can have greater impact where domestic stock markets are less developed.

### Trends in internationalization over time
- Proportion of countries with some international firms rises steadily:
  - About 70 percent of countries have at least one international firm by end of the sample period.
- Countries where fraction of international firms exceeds ten percent:
  - Rise over the period but reaches only some 32 percent at the end of the sample period.
- Proportion of firms internationalizing (global sample):
  - Rises from three to seven percent over the period.
  - These proportions are very similar when averaged by country observations.
- Conclusion: internationalization remains limited to a small group of firms and countries.

### Cross-country distributions and comparisons
- In 2000, many countries have little or no internationalization:
  - Fewer than ten percent of firms are international in almost 66 percent of the countries.
- Quintile comparisons (examples cited):
  - For the 4th quintile: on average ten percent of firms are international.
  - For the 5th quintile: on average 30 percent of firms are international.
  - Share of market capitalization of international firms for the 4th quintile: on average 44 percent.
  - Share of market capitalization of international firms for the 5th quintile: on average 71 percent.
  - For the 5th quintile, international trading amounts on average to 51 percent of domestic trading.
  - Some countries have trading abroad relative to domestic trading exceeding 100 percent.
- General pattern: average international firm is much larger than average domestic firm; trading abroad is less prevalent than listing/capital-raising abroad for most countries.

### Country characteristics related to internationalization (summary statistics)
- Countries with international activity vs. those without:
  - Countries with international activity are generally larger and have higher income per capita.
  - They appear to show lower inflation rates but also tend to show lower fiscal surpluses.
  - They are more open financially and have better law and order and investor protection (differences not always statistically significant).
  - They engage less in trade.
  - Suggestion: more developed countries see more internationalization.
- Top 20 percent vs. bottom 20 percent of countries ranked by share of market capitalization abroad:
  - Highest-degree internationalization countries tend to be larger, richer, have better inflation performance, somewhat worse fiscal management, more open to financial flows, less trade in goods and services, and better law and order and investor rights.
  - All differences are statistically significant at least at the five percent level.
- Sorting by share of value traded abroad or by share of number listed abroad yields largely similar patterns; small differences in trade-to-GDP ratios (example: 0.74 versus 0.73) and changes in statistical significance of some institutional indexes.

### Multivariate regression evidence (country-level analysis)
- Estimation methods used:
  - Random-effects panel regressions.
  - Tobit regressions to account for countries with zero international firms.
  - Fixed effects panel estimations used as robustness checks (reported elsewhere).
- Main consistent findings (Tables 3a and 3b):
  - Larger country size and higher income are associated with higher degrees of internationalization.
  - Better macroeconomic stability (e.g., lower inflation) is associated with higher internationalization.
  - Greater trade openness and greater financial openness are associated with more internationalization.
  - For the smaller country sample, countries with better investor protection and better law and order have less internationalization (coefficients not always statistically significant, possibly due to high correlation with other macro variables).
  - Tobit regressions show larger and a few more statistically significant coefficients than panel regressions; no qualitative differences otherwise.
- Results for developing countries (Table 3b):
  - Results are not different from the full sample except law and order becomes consistently statistically significant in both panel and Tobit models.
  - Some coefficients are larger in magnitude for developing countries, suggesting internationalization there is more driven by country characteristics.
  - No particular group of countries drives the overall relationships.

### Interpretation and overarching conclusions
- More developed countries with better macroeconomic fundamentals, higher trade activity, and greater financial openness tend to see more internationalization.
- Countries with better institutional environments (particularly legal quality) tend to see less internationalization, potentially because firms in countries with weaker legal development use internationalization to bond to higher standards.
- These findings are consistent with related research (Claessens and others, 2006) and support the hypothesis that international investors’ demand factors play an important role in firms’ ability to go international.

### Economic significance and illustrative magnitudes from regressions
- Using regression results from the first specification of Tables 3a and 3b:
  - A one percent increase in the country’s GDP would result in an increase in the proportion of international firms to total firms of about three percentage points in the panel estimations and about six percentage points in the Tobit estimations.
  - A one percent increase in GDP per capita would result in an increase in the proportion of international firms to total firms of about four percentage points in the panel estimations and about two percentage points in the Tobit estimations.
  - A one percent rise in inflation would result in a ratio two percentage points lower (panel) and six percentage points lower (Tobit).
  - For non-log variables, effects are interpreted via standard deviation changes:
    - A one standard deviation increase in the stock market liberalization index results in an increase of 0.7 percentage points (panel) and 2.3 percentage points (Tobit) in the proportion of international firms.
- Tobit coefficients tend to be larger and sometimes more statistically significant than panel estimates, but censoring at zero does not alter overall conclusions.

*Source: _wp07138 - Appendix Table 2 provides the total number of firms, distinguished by domestic and international firms, covered*

### Appendix Table 4. The set of regression results for all countries (upper panel) confirms most, although not all of

### _wp07138 - Appendix Table 4. The set of regression results for all countries (upper panel) confirms most, although not all of

### Appendix Table 4 — country-level regressions: main findings
- The set of regression results for all countries (upper panel) confirms most, although not all, of the findings of the panel and Tobit regressions.
- Major exception: coefficients for GDP per capita are negative (different in sign). Interpretation provided:
  - Using fixed effects at the country level already controls for country differences, of which GDP per capita is the most important indicator.
  - The negative sign for GDP per capita then refers to the effect of time-variation in GDP per capita on internationalization, with growing economies observing less internationalization.
- Some coefficients lose statistical significance in the all-countries regressions (mostly those for fiscal surplus and stock markets liberalization); sometimes coefficients become significant when they were not before.
- Results using the set of developing countries only (lower panel) confirm the results for all countries.
- Tobit regression interpretation: valid when the proportion of international firms in a country is different from zero (i.e., part of the uncensored part of the data).

### Magnitudes and illustrative effects (reported comparisons)
- A hypothetical increase in GDP would result in a 4.5 percentage point increase for both estimation methods in the ratio of international firms.
- Institutional variables: in panel estimations, a one standard deviation increase in any of the indices yields a decrease of ten percentage points in the ratio of international firms.
- For Tobit estimations, the institutional effect is reduced to roughly two percentage points in the case of the investor protection Index.
- For developing countries, these magnitudes are very similar to those obtained using the whole sample of countries.
- Sample statistics: mean fraction of international firms in the sample is 13 percent, with a standard deviation of 16 percent.

### V. Internationalization: firm-level perspective — summary statistics and distributions
- Median tests (Table 4): international firms are larger, grow faster, have higher returns on assets, and carry on more international business than domestic firms.
- Size difference: almost a factor of six — 275 million U.S. dollars (domestic median) versus 1,654 million U.S. dollars (international median).
- Kernel distributions and Kolmogorov-Smirnov tests: distributions differ between international and domestic firms for assets, growth, returns on assets, and international business.
  - Greatest differences: assets and foreign sales distributions.
  - Smaller differences: sales growth distributions.
  - Least differences: return on assets distribution.
- Differences persist when separating developing and developed countries, though differences in statistical significance are smaller for developing countries (possibly due to sample size).
- Median comparisons between developed and developing countries:
  - Average asset sizes:
    - Developing countries: international 916 million U.S. dollars; domestic 165 million U.S. dollars.
    - Developed countries: international 2,454 million U.S. dollars; domestic 316 million U.S. dollars.
  - Relative sizes:
    - International firm vs domestic firm in developing countries: five to six times larger.
    - International firm vs domestic firm in developed countries: seven to eight times larger.
  - Interpretation: international firms are much larger than domestic firms in both groups; patterns of differences are similar across country groups.

### Firm-level regressions: Probit and Cox estimates — determinants
- Probit regressions (using firm information as of 1993 to predict internationalization over the next seven years):
  - Larger firms and firms growing faster are more likely to go abroad.
  - More foreign sales increase chances of internationalization (collateral value to having international activities).
  - Return on assets: coefficient positive but not always statistically significant; may be noisy or redundant with growth measures.
  - Using country dummies vs country variables: results consistent; specifications employ firm characteristics and either country dummies or country variables to control country characteristics.
  - Probit results split by country group:
    - No major differences in sign; significance generally maintained.
    - Firms from developing countries appear more sensitive to firm characteristics (size, sales growth, foreign sales).
- Economic significance estimates (Probit, regressions with country dummies):
  - For an average firm:
    - A one percent increase in total assets raises the probability of becoming international by approximately 1.4 percentage points.
    - A one percent increase in sales growth raises the probability of becoming international by 0.8 percentage points.
    - A one standard deviation increase in foreign sales raises the probability by 0.4 percentage points.
  - For firms from developed countries:
    - A one percent increase in size relates to a one percentage point increase in probability of becoming international.
  - For firms from developing countries:
    - A one percent increase in size relates to about a four percentage point increase in probability of becoming international.
  - Overall interpretation: firm characteristics matter more for firms in developing countries; internationalization is more sensitive to firm attributes among developing country firms.

- Cox regressions (time-varying firm information up to the year before listing abroad):
  - A coefficient greater than one indicates increases in the variable enhance probability of going abroad; less than one indicates decreases.
  - Coefficients indicate the percentage change in the probability of observing a firm becoming international, relative to the base probability, due to a one standard deviation change in each explanatory variable.
  - Cox results confirm Probit findings: firm size, sales growth, and foreign sales share positively related to probability of internationalizing; return on assets positive and statistically significant in Cox (in contrast to some Probit specifications).
  - Differentiation by country groups:
    - Most firm characteristics statistically significant for both developed and developing countries.
    - Sales growth: not statistically significant for developing countries in Cox.
    - Return on assets: not statistically significant in all Cox cases.
    - Foreign sales variable: more important for developing countries in some specifications; coefficient on foreign sales larger for developed countries in other specifications.
  - Economic magnitude example (hazard ratio interpretation):
    - Standard deviation of the logarithm of total assets = 1.61.
    - Logarithm of the estimated coefficient = 1.716.
    - Multiplying and exponentiating yields a 140 percent increase in the baseline probability of becoming international with a one standard deviation increase in size for an average firm.
    - Relative ordering of economic importance: size largest, followed by foreign sales, then sales growth; return on assets effect relatively small.

### Before-and-after internationalization analysis (difference-in-differences style)
- Method: regress firm characteristics on before-internationalization and after-internationalization dummies, including country, industry, and year dummies.
  - Before dummy = 1 for years before firm becomes international; after dummy = 1 in and after year firm becomes international.
- Table 7 results (four firm characteristic regressions):
  - Prior to internationalization, prospective international firms are larger, have higher sales growth, higher return on assets, and higher foreign sales to assets than domestic firms — all statistically significant.
  - Subsequent to internationalization, relative to domestic firms, international firms are larger, grow faster, and have a higher share of foreign sales. Return on assets coefficient post-internationalization is not statistically significant.
  - Comparing after vs before coefficients:
    - All differences are statistically significant but signs differ:
      - Upon becoming international, firms become even larger and further increase share of foreign sales.
      - Post-internationalization firms grow at a somewhat slower pace and lower return on assets relative to their pre-internationalization levels.
  - Interpretation: before-and-after results confirm Probit and Cox findings; internationalization associated with increased size and foreign sales share, and possible growth pattern changes (growth slows, returns fall) after listing abroad.

### VI. Conclusions — synthesis and policy-relevant implications
- Extent of international financial integration via firms using international equity markets appears more limited than commonly thought:
  - Many countries have few participating firms; far fewer countries have a non-negligible proportion of internationally active firms.
  - Only certain firms and countries participate in international equity markets.
- Determinants:
  - Country characteristics: more developed countries with better macroeconomic (but worse institutional) conditions and more open economies have more international firms.
  - Firm characteristics: larger firms and firms with more foreign sales are significantly more likely to internationalize; firms that grow faster and have higher rates of returns are also more prone to going abroad.
- Suggested extensions and policy issues (enumerated research/policy directions in the source):
  - Distinguish firms that only list or trade internationally from those that raise capital in those markets; distinguish forms of listing (cross-listing versus ADRs/GDRs).
  - Collect and analyze firm-specific governance variables (ownership structures, independent directors, accountants) to assess their role relative to internationalization as corporate governance tools.
  - Study spillovers to domestic firms from internationally active firms (positive spillovers: freed domestic financing, integrated markets; negative spillovers: adverse effects on domestic market liquidity).
  - Understand country and firm characteristic thresholds that enable firms to issue capital internationally to inform policies that increase firms’ access to global capital markets.
  - Implications for domestic stock exchanges: sufficiently developed countries risk migration of firms to international markets, with consequences for local market capitalization, liquidity, and market viability.
  - Recognize which firms are unlikely to access international equity markets (e.g., small firms) and tailor local capital market development accordingly.
- Broader conclusion: designing preferred forms of financial market development hinges on understanding the nature and determinants of international financial integration.

Italic source attribution:
*Source: _wp07138 - Appendix Table 4. The set of regression results for all countries (upper panel) confirms most, although not all of — PDF chapter/section content provided.*

### 2000. International firms are those identified as having at least one active depositary receipt program, having

### _wp07138 - 2000. International firms are those identified as having at least one active depositary receipt program, having

### Definition and sample
- International firms are those identified as having at least one active depositary receipt program, having raised equity capital in international markets, or being listed on the London Stock Exchange, NASDAQ, or NYSE.
- The United States and the United Kingdom are not included in the sample due to their classification as international financial centers.
- Sample period referenced: 1989–2000.

### Distribution of firm characteristics (Figure 4)
- Kernel distributions estimated for total assets, sales growth, return on assets, and foreign sales to total sales for domestic and international firms.
- Kolmogorov-Smirnov combined test statistics and p-values:
  - Total Assets: D=0.324, P-value=0.000
  - Sales Growth: D=0.154, P-value=0.000
  - Return on Assets: D=0.098, P-value=0.000
  - Foreign Sales to Total Sales: D=0.260, P-value=0.000
- Visual density scales in figure (examples from axes):
  - Total Assets density axis up to 5,000
  - Sales Growth axis from -0.5 to 0.5
  - Return on Assets axis from -0.2 to 0.3
  - Foreign Sales to Total Sales axis from 0.0 to 1.0
- Legend: Domestic Firms vs International Firms.

### Stock market internationalization by region (Table 1)
- Country coverage and counts:
  - Total countries in sample: 205
  - Number of countries with active domestic stock markets: 151
  - Number of countries with international activity: 117
  - Share of countries with international activity: 63.1%
  - Number of firms listed in the domestic market: 36,971
  - Number of international firms: 2,546
  - Share of international firms: 6.9%
- Regional breakdown (selected rows):
  - Developing Countries: Number of Countries 161; Number of Countries with Active Stock Markets 128; Number of Countries with International Activity 85; Share of Countries with International Activity 58.5%; Number of Firms Listed in the Domestic Market 23,910; Number of International Firms 889; Share of International Firms 3.7%
  - Developed Countries: Number of Countries 44; Number of Countries with Active Stock Markets 29; Number of Countries with International Activity 22; Share of Countries with International Activity 75.9%; Number of Firms Listed in the Domestic Market 13,061; Number of International Firms 1,657; Share of International Firms 12.7%
- Market-cap and trading aggregates (regional averages):
  - Total (all countries): Total Market Capitalization/GDP 50.6%; Market Capitalization of International Firms/GDP 19.7%; Market Capitalization of International Firms/Total Market Capitalization 28.2%; Value Traded Domestically/GDP 34.3%; Value Traded Abroad/GDP 3.7%; Value Traded Abroad/Value Traded Domestically 17.7%
  - Developing Countries: Total Market Capitalization/GDP 27.4%; Market Capitalization of International Firms/GDP 5.3%; Market Capitalization of International Firms/Total Market Capitalization 19.9%; Value Traded Domestically/GDP 14.4%; Value Traded Abroad/GDP 0.8%; Value Traded Abroad/Value Traded Domestically 13.5%
  - Latin America & Caribbean: Total Market Capitalization/GDP 29.0%; Market Capitalization of International Firms/GDP 4.7%; Market Capitalization of International Firms/Total Market Capitalization 23.6%; Value Traded Domestically/GDP 3.1%; Value Traded Abroad/GDP 2.0%; Value Traded Abroad/Value Traded Domestically 61.2%
  - Developed Countries (market aggregates): Total Market Capitalization/GDP 109.1%; Market Capitalization of International Firms/GDP 61.8%; Market Capitalization of International Firms/Total Market Capitalization 46.9%; Value Traded Domestically/GDP 86.1%; Value Traded Abroad/GDP 12.8%; Value Traded Abroad/Value Traded Domestically 28.5%

### Country characteristics and internationalization (Tables 2a and 2b)
- Median comparisons and Mann-Whitney U-tests show statistically significant differences between countries with and without activity in international equity markets on many dimensions.
- All countries (medians, All Countries panel):
  - Log of GDP (Million U.S. dollars): Countries without Activity 8.73; Countries with Activity 11.20; Z-statistic -21.70; P-value 0.00
  - Log of GDP per capita (U.S. dollars): Countries without Activity 7.32; Countries with Activity 8.19; Z-statistic -10.69; P-value 0.00
  - Log of (1+Inflation): Countries without Activity 0.08; Countries with Activity 0.06; Z-statistic 0.41; P-value 0.68
  - Fiscal Surplus/GDP: Countries without Activity 0.03; Countries with Activity 0.02; Z-statistic 2.47; P-value 0.01
  - Stock Market Liberalization median: Countries without Activity 0.00; Countries with Activity 1.00; Z-statistic -17.15; P-value 0.00
  - Trade (Exports+Imports)/GDP: Countries without Activity 0.80; Countries with Activity 0.67; Z-statistic 25.13; P-value 0.00
- Comparisons by extremes (bottom 20 percent vs top 20 percent of internationalization) highlight larger gaps:
  - Least Internationalized vs Most Internationalized (by market capitalization of international firms): Log of GDP medians 9.04 vs 12.37; Z-statistic -14.46; P-value 0.00
  - Log of GDP per capita medians 7.53 vs 9.89; Z-statistic -13.41; P-value 0.00
  - Stock Market Liberalization medians 1.00 vs 1.00 but Z-statistic -5.29; P-value 0.00 (stock market liberalization measure varies across subpanels)
- Table 2b similarly reports medians and significance when sorting countries by value traded abroad/value traded domestically and by number of international firms; consistently larger economies, higher GDP per capita, lower inflation, higher trade openness, and stronger investor protection/law-and-order measures associate with greater internationalization.

### Country-level regressions (Tables 3a and 3b)
- Dependent variable: annual share of international firms over total firms in each country.
- Key coefficient magnitudes (All Countries, Table 3a, Panel and Tobit specifications):
  - Log of GDP: coefficients range from 0.027 *** to 0.096 *** across specifications (Z-statistics reported, e.g., [5.212], [26.984]).
  - Log of GDP per Capita: coefficients around 0.010 *** to 0.041 *** (e.g., [5.933], [7.025]).
  - Log (1+Inflation): negative and significant in multiple specifications (e.g., -0.019 ***, -0.065 ***, [3.578], [7.103]).
  - Fiscal Surplus/GDP: positive and often significant (e.g., 0.226 ***, 0.470 ***, [4.084], [7.397]).
  - Trade (Exports+Imports)/GDP: positive and highly significant (e.g., 0.157 ***, 0.199 ***, [12.321], [28.296]).
  - Stock Market Liberalization: positive and significant when included (e.g., 0.017 ***, 0.053 ***, [3.493], [9.250]).
  - Investor Protection Index and Law and Order Index show mixed significance (some negative coefficients reported).
- Developing countries (Table 3b) show similar qualitative patterns:
  - Log of GDP: positive and significant (e.g., 0.022 *** to 0.061 ***).
  - Log of GDP per Capita: positive and significant in many specs.
  - Log (1+Inflation): negative and significant.
  - Trade (Exports+Imports)/GDP: positive and significant (e.g., 0.082 *** to 0.117 ***).
  - Stock Market Liberalization: positive and significant (e.g., 0.018 ***, 0.070 ***).

### Firm-level selection and hazard analyses (Tables 5 and 6)
- Probit regressions (Table 5) — marginal probabilities of becoming international (1993 values; robust clustered standard errors):
  - Size (Log of Total Assets): positive and significant; marginal probabilities reported include 0.014 ***, 0.010 ***, 0.039 ***, 0.030 *** across panels (Z-statistics e.g., [14.285], [11.799]).
  - Growth (Log (1+Sales Growth)): positive and significant in many specifications (e.g., 0.008 **, 0.009 ***, 0.019 *, Z-statistics [2.425], [3.727]).
  - Return on Assets: small positive coefficients, significance varies (e.g., 0.011 * with bracketed statistics).
  - Foreign Sales to Total Sales: positive and significant (e.g., 0.017 ***, 0.013 ***, 0.071 *).
  - Models include country dummies or country variables depending on panel; industry dummies included.
  - Number of observations and international firm counts vary across panels (examples: Observations 5,254 with Number of International Firms 303; Observations 4,620 with Number of International Firms 254).
- Cox regressions (Table 6) — hazard ratios for becoming an international firm (exponentiated coefficients; clustered standard errors):
  - Size (Log of Total Assets): hazard ratios around 1.716 *** to 1.833 *** across panels (e.g., [17.720], [8.551]).
  - Growth (Log (1+Sales Growth)): hazard ratios >1 and often significant (e.g., 1.306 ***, 1.339 ***, [4.898], [3.859]).
  - Return on Assets: hazard ratios >1 in several specs (e.g., 1.375 **, 1.750 **, with z-statistics).
  - Foreign Sales to Total Sales: large positive hazard ratios (e.g., 3.198 ***, 2.677 ***, 5.650 ***; [5.654], [4.286], [3.664]).
  - Panels reported with country dummies and with country variables; sample sizes vary (e.g., Number of Observations 64,172; Number of International Firms 2,145 in one panel).

### Differences in firm characteristics and dynamics (Tables 4 and 7)
- Cross-sectional medians (Table 4, All Countries medians):
  - Total Assets (Million U.S. dollars): Domestic Firms 275.33; International Firms 1,654.1 (observations: Domestic 64,480; International 10,323)
  - Log (1+Sales Growth): Domestic 0.048; International 0.081
  - Return on Assets: Domestic 0.037; International 0.054
  - Foreign Sales / Total Sales: Domestic 0.000; International 0.290 (observations used vary)
- Mann-Whitney U-tests show statistically significant differences between domestic and international firms across size, growth, performance, and foreign sales measures (Z-statistics and P-values reported; examples: Total Assets Z-statistic 82.78, P-value 0.00; Foreign Sales / Total Sales Z-statistic 55.50, P-value 0.00).
- Panel regressions comparing before- and after-internationalization (Table 7):
  - Before Internationalization Dummy (a):
    - Total Assets: 3,666.7 *** [9.516]
    - Sales Growth: 0.060 *** [5.064]
    - Return on Assets: 0.021 *** [2.933]
    - Foreign Sales to Total Sales: 0.116 *** [10.143]
  - After Internationalization Dummy (b):
    - Total Assets: 5,858.1 *** [14.479]
    - Sales Growth: 0.022 *** [4.387]
    - Return on Assets: 0.001 [0.296]
    - Foreign Sales to Total Sales: 0.146 *** [17.442]
  - Difference (b)-(a):
    - Total Assets: 2,191 (Test (b)-(a)=0; P-value 0.000)
    - Sales Growth: -0.038 (P-value 0.002)
    - Return on Assets: -0.021 (P-value 0.006)
    - Foreign Sales to Total Sales: 0.030 (P-value 0.002)
  - Number of observations: Total Assets/Sales Growth/Return on Assets regressions 74,803; Foreign Sales to Total Sales 50,034. Number of firms: 10,154 for most regressions; 8,048 for foreign sales regression.

*Source: _wp07138 - 2000. International firms are those identified as having at least one active depositary receipt program, having (IMF PDF content provided).*

### Appendix Table 1. Series Desc

### Appendix Table 1. Series Desc

### Description and Data Sources
- The data cover the period 1989–2000.

- Series Names, Description, Source:
  - Countr y-Level
    - GDP at market prices (current U.S. dollars)
      - Gross domestic product (GDP) at purchaser prices. GDP data is converted from domestic currencies using yearly averageofficial exchange rates. For a few countries where the official exchange rate does not reflect the rate effectively applied toactual foreign exchange transactions, an alternative conversion factor is used.
      - World Bank: World Development Indicators
    - GDP per capita at market prices (current U.S. dollars)
      - Gross domestic product divided by mid-year population.
      - World Bank: World Development Indicators
    - Inflation, consumer prices (percent per year)
      - Inflation as measured by the consumer price index.
      - World Bank: World Development Indicators
    - Fiscal surplus
      - Overall  budget  balance,  defined  as  current  and  capital  revenue  and  official  grants  received,  less  total  expenditure  and lending minus repayments. Data are for central government only.
      - World Bank: World Development Indicators
    - Stock market liberalization
      - Takes the value one after the year of full stock market liberalization as indicated by combining liberalization dates of Bekaert et al. (2005), Kaminsky and Schmukler (2003), and Vinhas de Souza (2005).
      - Bekaert et al. (2005), Kaminsky and Schmukler (2003), and Vinhas de Souza (2005).
    - Trade (ex ports plus im ports)
      - Sum of ex ports and im ports of g oods and services, ex pressed in current U.S. dollars.
      - World Bank: World Development Indicators
    - Law and order Index
      - Law and order index.
      - International Countr y Risk Guide (ICRG)

  - Firm-Level
    - Investor p rotection index
      - Investor p rotection index.
      - World Bank: Doin g Business
    - Total assets
      - Total assets at the end of the most recent fiscal y ear, ex pressed in U.S. dollars.
      - Worldscope
    - Two-year average sales growth
      - Geometric average annual sales growth over the last two years. Sales are expressed in U.S. dollars.
      - Worldscope
    - Return on assets
      - Sum of net income (before preferred dividends) and interest expenses on capitalized debt (after tax), over last year's total assets.
      - Worldscope
    - Forei g n sales to total sales
      - Annual international sales over ne t sales or revenues.
      - Worldscope

*Source: _wp07138 - Appendix Table 1. Series Desc*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07138.pdf_
