TY - JOUR
AU1 - Fan,, Haichao
AU2 - Gao,, Xiang
AB - Abstract This article unbundles institutions protecting domestic and foreign creditors’ rights. We estimate a negative relation between the degree of domestic rights protection and the external stock of private non‐guaranteed debt in 85 developing countries. A supply‐side explanation is that strong domestic protection supports reliable outside financing options for potential external debt defaulters; foreign investors anticipating this would tighten credit constraints ex ante. Then we formalise the argument in a private borrowing and default model, and show that centralisation is no longer necessarily welfare superior. Creditor rights protection is a critical determinant of developing countries’ capacity to accumulate external debt. This is especially the case in the private‐sector debt market, where repayment enforcement relies more on third‐party institutions than self‐enforcing arrangements. However, the literature has overlooked the role played by institutions protecting domestic creditors’ rights. Instead, it had naturally concentrated on foreign creditor rights protection in debtor countries. Using variations in institutional quality, this article examines the effects of domestic creditor rights on external private debt and the consequent welfare implications. Domestic creditor rights were seldom contemplated in prior studies for two reasons. First, external debt of developing countries used to be comprised mostly of sovereign debt, where legal enforcement of rights is minimal. Hence theorists have focused on the economic incentives to repay.1 Broadly defined measures of institutions suffice to proxy for sovereign creditworthiness.2 Second, research on external private debt documented only the importance of foreign creditor rights. The domestic side has long been viewed as insignificant and, given that citizens were supposed to be well protected, the corresponding institutions had been assumed to function perfectly under government control. Let us first consider the composition of external debt. The period 1990–2009 has witnessed private sectors being actively promoted as independent participants in the global financial market. Capital flows to public sectors are declining, and the state’s role is transforming from financier to guarantor of funds. As a result, the private‐sector publicly non‐guaranteed (PNG) share of total external debt stocks on average increased from 5% to 17% for 85 base‐sample countries.3 Not only is PNG credit now a notable fraction but it also shows distinctive patterns. In each year during 2004–9, PNG debt is consistently more concentrated in countries with weaker domestic creditor rights, whereas countries that have built stronger foreign rights protection institutions accumulate larger stocks over time. Panels (a) and (b) of Figure 1 provide an illustration of this dichotomy. In contrast, as can be seen in (c) and (d), better quality in either institutions attracts more public and publicly guaranteed (PPG) debt. Fig. 1. Open in new tabDownload slide Cross‐country Ave. Stocks of PNG External Debt Per Capita, 2004–9. (a) PNG External Debt Stocks and Domestic Creditor Rights Protection. (b) PNG External Debt Stocks and Foreign Creditor Rights Protection. (c) PPG External Debt Stocks and Domestic Creditor Rights Protection. (d) PPG External Debt Stocks and Foreign Creditor Rights Protection Notes. Debt data are taken from the World Bank’s 2014 International Debt Statistics (IDS), which is a successor of the Global Development Finance since 2013, and the Institute of International Finance (IIF) 2014 Economic Database. PNG external debt comprises long‐term external obligations of private debtors that are not guaranteed for repayment by a public entity. PPG debt comprises long‐term external obligations of public debtors, including the national government, political subdivisions (or an agency of either), and autonomous public bodies, and external obligations of private debtors that are guaranteed for repayment by a public entity. The threshold for ‘weak’ versus ‘strong’ domestic (foreign) creditor rights protection is set at the median for its primary measure – debt recovery costs (investment profile) – in 79 countries (Gabon, Gambia, Guyana, Liberia, Montenegro, and Sudan were removed from the base sample because of missing creditor rights measures for certain years). We only plot debt level for relatively high‐income countries with GDP per capita above the sample median. Similar patterns, though the contrast is less strong, hold for relatively low‐income countries. Details on variable definitions and sources can be found in subsection 1.1. Fig. 1. Open in new tabDownload slide Cross‐country Ave. Stocks of PNG External Debt Per Capita, 2004–9. (a) PNG External Debt Stocks and Domestic Creditor Rights Protection. (b) PNG External Debt Stocks and Foreign Creditor Rights Protection. (c) PPG External Debt Stocks and Domestic Creditor Rights Protection. (d) PPG External Debt Stocks and Foreign Creditor Rights Protection Notes. Debt data are taken from the World Bank’s 2014 International Debt Statistics (IDS), which is a successor of the Global Development Finance since 2013, and the Institute of International Finance (IIF) 2014 Economic Database. PNG external debt comprises long‐term external obligations of private debtors that are not guaranteed for repayment by a public entity. PPG debt comprises long‐term external obligations of public debtors, including the national government, political subdivisions (or an agency of either), and autonomous public bodies, and external obligations of private debtors that are guaranteed for repayment by a public entity. The threshold for ‘weak’ versus ‘strong’ domestic (foreign) creditor rights protection is set at the median for its primary measure – debt recovery costs (investment profile) – in 79 countries (Gabon, Gambia, Guyana, Liberia, Montenegro, and Sudan were removed from the base sample because of missing creditor rights measures for certain years). We only plot debt level for relatively high‐income countries with GDP per capita above the sample median. Similar patterns, though the contrast is less strong, hold for relatively low‐income countries. Details on variable definitions and sources can be found in subsection 1.1. The second reason is even more unconvincing. As noted above, domestic rights matter in PNG external debt. With a lax judiciary operating at an arm’s length away from the government, their protection technology is just imperfect as foreign creditor rights. Despite potential overlaps, a wide disparity between domestic and foreign rights protection could emerge due to discriminatory clauses and biased enforcement. Figure 1(a) unveils a pattern that is inconsistent with straightforward intuitions. From a supply constrained point of view, we identify an outside‐option effect through which domestic rights may inversely affect external private debt. Stronger domestic protection supports richer sources of financing for privately‐owned firms,4 even after they are prohibited from international markets. Such an alternative raises borrowers’ post‐default value, and foreign investors anticipating this will tighten credit constraints. At first blush, this argument could apply equally to public borrowing. We assert, however, it is more relevant to private contracts. For one thing, the legal system plays a lesser role in public debt repayment. For another, given the limited liquidity of purely domestic funds, a large portion of funds available to individual defaulters is intermediated from overseas via non‐defaulted agents. Yet if the government defaults, the whole country is excluded, so it is much harder to mitigate the punishment similarly. Another explanation relates to the power of a demand‐side story: firms might be unwilling to borrow internationally when domestic institutions can guarantee an ample supply of local credit to draw on. We argue this substitution effect is small for several reasons. First, international markets often come to complement, rather than substitute for, the domestic capital pool. This is confirmed by a 0.55 correlation between PNG external debt and domestic credit to private sector in our sample. Second, not all firms have access to foreign funds, but for those who succeeded in inducing foreign investments or had the skill to issue bonds abroad, external debt is preferred because of limited monitoring and lower capital costs. Third, there is extensive evidence suggesting that creditors determine the terms of contracts and debtors are constrained.5 Hence, the observed debt stocks are likely to be the credit ceiling instead of optimal demand. On the econometric part, we follow Lane (2004) by maintaining the assumption that credit constraints bind for private sectors in developing countries. As in Alfaro et al. (2008), we perform cross‐country ordinary least squares (OLS) regressions. This treatment draws attention to the slow‐moving institutional variables. While they focused on how the strength of aggregate institutions affect private equity inflows; we attempt to estimate the separate effects of domestic and foreign rights protection on private debt stocks. To get a sense of the magnitude of the outside‐option effect, consider two countries with foreign creditor rights rankings below the lower quartile. If we move up from around 25th percentile (Nigeria) to around 75th percentile (Iran) in the distribution of domestic creditor rights, based on the results shown in column 1 of Table 2(a), we have $170 less debt per capita during the sample period on average. It represents a 35% decrease in the sample mean, which is $487. With a parallel pair of countries (Cameroon and Greece) that fall above the upper quartile in foreign creditor rights rankings, this negative effect turns positive ($937 more debt per capita as we move up from 25th to 75th percentile in domestic rights). It is because the outside‐option effect barely operates when foreign investments are fully guarded by quality protection of foreign creditors’ rights. Stronger domestic protection now indicates better governance and creditworthiness. These findings are robust to different measures of creditor rights, to controlling for a host of non‐legal determinants, and to a multiple instrumental variables (IV) strategy. We then use the limited commitment framework to formalise the supply‐constrained explanation.6 In the decentralised borrowing and default model, domestic and foreign rights protecting institutions are government‐remote enforcers of respective type of private contracts. A lender can file a lawsuit to seize a fraction of the borrower’s endowment in case of default events. This fraction varies across contracts, as courts treat domestic and foreign creditors differently, discriminating in one or the other direction. In the centralised borrowing and default benchmark, governments are the arbiter of public contracts. They trade internationally with no respect to foreigners’ rights, and domestically with their residents under perfect enforcement. Comparing the ex ante welfare of agents in the above two contexts, we find that centralisation is no longer necessarily superior. It is shown in a numerical example that the total welfare can be greater with decentralisation if the degree of protection of domestic creditors’ rights is substantially low. Our empirical strategy is similar to Acemoglu and Johnson (2005), who evaluate the importance of horizontal institutions supporting contracts amongst citizens and vertical institutions constraining expropriation by executives. We disregard where the executive role ends but introduce discriminatory rights protection. Erce (2012) adopts a case‐study approach to analyse patterns of discrimination during sovereign debt restructurings. Unlike his assumption that government chooses whether to respect intercreditor equity, we view creditor inequity as exogenous mistreatment. Esty (2004) investigates how domestic financial development and foreign creditor rights interact to affect the composition of loan syndicates. The author finds that banks provide a greater share of total funds to countries with less‐developed financial systems. Bae and Goyal (2009) separate the effects of creditor rights protection and property rights enforcement on foreign bank loans. While better protection of domestic creditors reduces spreads, it appears to have adverse impacts on size and maturity. Although disentangling domestic and foreign protection rights is a by‐product in their article, we show that their transaction‐level findings become more pronounced at the macro level, as firm‐specific disturbances cancel each other out. Our theory is closest to Jeske (2006) who develops a private borrowing model where resident default risk and sophisticated financial markets7 yield underborrowing due to externalities on domestic interest rates. Our work is also related to Uribe (2006), who finds no overborrowing in a risk‐free‐bond only economy with fixed debt limit imposed at the aggregate instead of the individual level. However, this finding is sensitive to assumptions such as identical agents and adjustable foreign interest rates.8 Turning to centralised default models, Kim and Zhang (2012) show in a simulation that equilibrium private debt levels could be higher or lower than centralisation, depending on the specification and severity of default penalties. By assuming the court must enforce all contracts or none, Broner and Ventura (2011) find that a large amount of foreign debt raises enforcement costs and thus hampers domestic trade. Erce (2012) concludes that decentralisation leads to underborrowing since sovereign default would devaluate entrepreneurs’ foreign collateral. The rest of the article is organised as follows. Section 1 presents reduced‐form evidence. Section 2 puts forward a theory. Section (3) estimates numerically the theory’s welfare, and Section 4 concludes. 1 Empirical Analysis 1.1 Data and Descriptive Statistics The authoritative source for external debt is the IDS published by the World Bank. The categories of debt stocks are long‐term debt, use of International Monetary Fund (IMF) credit, and short‐term debt. Long‐term debt is defined as having a maturity of more than one year. It is further divided into PPG and PNG debt. Use of IMF credit can be classified as PPG debt, for it denotes member governments’ drawings. We abstract from short‐term debt because the IDS permits no distinction between short‐term public and private debtors. One might fear that doing so will artificially reduce debt measured. We argue that short‐term debt only accounts for a small proportion – the average is below 10% for the developing world. In addition, short‐term lending tends to be motivated by speculation rather than fundamentals. We use two data sets that encompass all maturities and our results are robust. The first is the 2014 IIF Economic Database. Like the IDS, private debt backed by public entities is excluded. The difference lies in the IIF leaving out debt owed by private deposit banks. Since the core of the money market consists of interbank borrowing, its statistics are comparable to those from the IDS, thus acting as a good supplementary. The second is obtained from the Bank for International Settlements (BIS; Dembiermont et al., 2013), which further narrows down borrower identity to private non‐financial sectors. But the BIS series carry explicit guarantees, so large discrepancies exist between the BIS and the IDS estimates. Although other vendors do exist, their focus is on the asset side and the disaggregation is by the type of creditor.9 In the IDS, out of 124 countries with debt data available, only 76 are ranked by creditor rights proxies. For a total of 31 countries from the IIF, we have complete information on 29, including nine economies that are not covered by the IDS (Chile, Czech Republic, Greece, Israel, Poland, Russia, Slovak Republic, South Korea and Uruguay). These 85 countries combined (the IDS is used whenever it overlaps with the IIF) constitute the base sample.10 We choose the period 2004–9 to match with the length of domestic rights measures, to avoid sovereign debt crises,11 and to incorporate the whole cycle of the 2007 credit freeze. Moreover, the average maturity of new external private debt commitments is a little bit over six years. There is no gap during this period, but the beginning or ending year varies for a few countries.12 To proxy for the protection of domestic creditors’ rights, we construct five measures based on the World Bank’s Doing Business (WBDB) database in 2010. It gathers data from questionnaire responses to especially designed case scenarios involving 100% domestic‐owned parties. The primary and secondary measures are taken from the resolving insolvency scenario (Djankov et al., 2008): the ‘debt recovery costs’ spent as share of a debtor’s estate value and the ‘cents on the dollar recouped’ of a secured bank loan in insolvency proceedings. They are valid proxies given that insolvency and default events are governed by similar legal schemes. Our next two measures are based on WBDB’s enforcing contracts topic (Djankov et al., 2003): the ‘number of procedures’ to enforce a plaintiff’s claim and the ‘enforcement costs’ (calculated as share of the value of the claim) to complete all steps in a commercial dispute. The major flaw in these measures is that they are more informative with regard to protection rights affecting commercial transactions rather than financing activities. Following Gwartney et al. (2010), the above four measures are rescaled to a range of [0, 10], where a higher score implies better protection.13 The last measure is the 0–10 ‘strength of legal rights’ index from the getting credit scenario (Djankov et al., 2007). This index captures the extent to which rights of borrowers and lenders are protected by collateral and bankruptcy laws. Therefore, it is about the existence of laws that matter less than enforcement, and is less preferred.14 The protection of foreign creditors’ rights is proxied by two statistics from the ICRG’s Political Risk Rating project. The primary ‘investment profile’ component index assesses factors affecting the risk to foreign investment. The rating assigned is the sum of three subcomponents: contract viability/expropriation, profits repatriation, and payment delays, each has a maximum of 4 and a minimum of 0. We divide the sum by 12 and multiply by 10 to transform it into a [0, 10] range. The alternative is the average of contract ‘repudiation and expropriation’ risk indicators from the IRIS‐3 file (both range from 0 to 10). Contract repudiation addresses the likelihood that a country will modify or repudiate a contract with a foreign business. Expropriation evaluates the risk of outright confiscation and forced nationalisation of a foreign property. Since IRIS‐3 is a relatively old data set, we focus on a country’s average level during 1982–97 as opposed to the level in the last year of coverage, in order to minimise randomness and account for history dependence. For both foreign creditor rights proxies, a higher score equates to better protection. One disadvantage of the ICRG statistics is that they are by definition more relevant to FDI. Does this bias our results? While a definitive answer is hard, we regress FDI and debt stocks on the ICRG measures, respectively, and the estimates are similar in magnitude and equally significant. Another potential disadvantage is representativeness bias. The ICRG statistics are based on subjective evaluation by inside experts, unlike the WBDB statistics which are derived from surveys for investors, outside consultants and the general public. We believe this is not a serious issue, since setting debt limits is also a strategic decision made by insiders. Nevertheless, a group of IVs is used to correct for errors in measurement. Table 1 reports summary statistics. Indeed, a better rule of law (RoL) or a lower Gini ratio strengthens both sets of creditor rights institutions,15 reducing their disparities. But between‐institution variation still exists across our sample countries. The 1999 default of the commercial bank HBL illustrates discrimination against foreigners in Pakistan, which ranks at the 99th percentile of debt recovery costs but at the 19th percentile of investment profile. The bankruptcy of the lottery firm Sazka in 2011 illustrates discrimination against residents. Its home country Czech Republic scores the 49th on debt recovery costs and the highest on the investment profile. The 2003 Uruguayan debt restructuring adopts an equitable and collaborative approach, reflected in Uruguay’s high ratings on both debt recovery costs (93th) and investment profile (82th). Table 1 Summary Statistics for Debt and Creditor Protection Measures . Base sample . High‐RoL countries . Low‐RoL countries . Low‐Gini countries . High‐Gini countries . Average PNG external debt stocks per capita, 2004–9 (constant 2005 US$) 487 829 153 755 212 (1,103) (1,466) (325) (1,475) (336) Average PPG external debt stocks per capita, 2004–9 (constant 2005 US$) 979 1,383 585 1,221 732 (2,471) (3,401) (777) (3,363) (895) Measures of domestic creditor rights (adjusted to a range of 0–10) Average debt recovery costs, 2004–9 (primary measure) 5.19 5.29 5.10 5.65 4.72 (2.89) (2.70) (3.09) (2.88) (2.85) Average cents on the dollar recouped, 2004–9 (secondary measure) 4.82 5.63 4.04 5.49 4.14 (2.81) (2.81) (2.61) (2.68) (2.80) Average number of procedures, 2004–9 5.07 5.74 4.42 4.65 5.51 (2.85) (2.69) (2.89) (2.71) (2.96) Average enforcement costs, 2004–9 5.41 5.99 4.85 5.92 4.89 (2.12) (1.73) (2.32) (1.87) (2.55) Average strength of legal rights, 2005–9 4.92 5.81 4.04 5.18 4.65 (2.31) (2.30) (1.99) (2.34) (2.28) Measures of foreign creditor rights (adjusted to a range of 0–10) Average investment profile (IP), 2004–9 (primary measure) 6.72 7.53 5.92 6.88 6.54 (1.59) (1.22) (1.51) (1.41) (1.75) Average repudiation & expropriation (R&E), 1982–97 (secondary measure) 6.21 6.73 5.71 6.53 5.88 (1.28) (1.24) (1.12) (1.33) (1.15) . Base sample . High‐RoL countries . Low‐RoL countries . Low‐Gini countries . High‐Gini countries . Average PNG external debt stocks per capita, 2004–9 (constant 2005 US$) 487 829 153 755 212 (1,103) (1,466) (325) (1,475) (336) Average PPG external debt stocks per capita, 2004–9 (constant 2005 US$) 979 1,383 585 1,221 732 (2,471) (3,401) (777) (3,363) (895) Measures of domestic creditor rights (adjusted to a range of 0–10) Average debt recovery costs, 2004–9 (primary measure) 5.19 5.29 5.10 5.65 4.72 (2.89) (2.70) (3.09) (2.88) (2.85) Average cents on the dollar recouped, 2004–9 (secondary measure) 4.82 5.63 4.04 5.49 4.14 (2.81) (2.81) (2.61) (2.68) (2.80) Average number of procedures, 2004–9 5.07 5.74 4.42 4.65 5.51 (2.85) (2.69) (2.89) (2.71) (2.96) Average enforcement costs, 2004–9 5.41 5.99 4.85 5.92 4.89 (2.12) (1.73) (2.32) (1.87) (2.55) Average strength of legal rights, 2005–9 4.92 5.81 4.04 5.18 4.65 (2.31) (2.30) (1.99) (2.34) (2.28) Measures of foreign creditor rights (adjusted to a range of 0–10) Average investment profile (IP), 2004–9 (primary measure) 6.72 7.53 5.92 6.88 6.54 (1.59) (1.22) (1.51) (1.41) (1.75) Average repudiation & expropriation (R&E), 1982–97 (secondary measure) 6.21 6.73 5.71 6.53 5.88 (1.28) (1.24) (1.12) (1.33) (1.15) Notes. Means with standard deviations in parentheses. The RoL index (0–10), averaged over 2004–9, is taken from Kaufmann et al. (2010). The Gini ratio (0–100), averaged over 1978–2009 discontinuously, is sourced from the World Bank’s WDI and the Global Peace Index. High or low‐RoL (Gini) countries are defined as countries with the RoL index (Gini ratio) above or below the median of the base sample. The unconditional correlation coefficients between debt recovery costs and investment profile are 0.18, 0.12, 0.24, 0.12 and 0.21 for the base sample, high‐Rol, low‐Rol, low‐Gini, and high‐Gini countries, respectively. See the text for detailed variable definitions and sources. Open in new tab Table 1 Summary Statistics for Debt and Creditor Protection Measures . Base sample . High‐RoL countries . Low‐RoL countries . Low‐Gini countries . High‐Gini countries . Average PNG external debt stocks per capita, 2004–9 (constant 2005 US$) 487 829 153 755 212 (1,103) (1,466) (325) (1,475) (336) Average PPG external debt stocks per capita, 2004–9 (constant 2005 US$) 979 1,383 585 1,221 732 (2,471) (3,401) (777) (3,363) (895) Measures of domestic creditor rights (adjusted to a range of 0–10) Average debt recovery costs, 2004–9 (primary measure) 5.19 5.29 5.10 5.65 4.72 (2.89) (2.70) (3.09) (2.88) (2.85) Average cents on the dollar recouped, 2004–9 (secondary measure) 4.82 5.63 4.04 5.49 4.14 (2.81) (2.81) (2.61) (2.68) (2.80) Average number of procedures, 2004–9 5.07 5.74 4.42 4.65 5.51 (2.85) (2.69) (2.89) (2.71) (2.96) Average enforcement costs, 2004–9 5.41 5.99 4.85 5.92 4.89 (2.12) (1.73) (2.32) (1.87) (2.55) Average strength of legal rights, 2005–9 4.92 5.81 4.04 5.18 4.65 (2.31) (2.30) (1.99) (2.34) (2.28) Measures of foreign creditor rights (adjusted to a range of 0–10) Average investment profile (IP), 2004–9 (primary measure) 6.72 7.53 5.92 6.88 6.54 (1.59) (1.22) (1.51) (1.41) (1.75) Average repudiation & expropriation (R&E), 1982–97 (secondary measure) 6.21 6.73 5.71 6.53 5.88 (1.28) (1.24) (1.12) (1.33) (1.15) . Base sample . High‐RoL countries . Low‐RoL countries . Low‐Gini countries . High‐Gini countries . Average PNG external debt stocks per capita, 2004–9 (constant 2005 US$) 487 829 153 755 212 (1,103) (1,466) (325) (1,475) (336) Average PPG external debt stocks per capita, 2004–9 (constant 2005 US$) 979 1,383 585 1,221 732 (2,471) (3,401) (777) (3,363) (895) Measures of domestic creditor rights (adjusted to a range of 0–10) Average debt recovery costs, 2004–9 (primary measure) 5.19 5.29 5.10 5.65 4.72 (2.89) (2.70) (3.09) (2.88) (2.85) Average cents on the dollar recouped, 2004–9 (secondary measure) 4.82 5.63 4.04 5.49 4.14 (2.81) (2.81) (2.61) (2.68) (2.80) Average number of procedures, 2004–9 5.07 5.74 4.42 4.65 5.51 (2.85) (2.69) (2.89) (2.71) (2.96) Average enforcement costs, 2004–9 5.41 5.99 4.85 5.92 4.89 (2.12) (1.73) (2.32) (1.87) (2.55) Average strength of legal rights, 2005–9 4.92 5.81 4.04 5.18 4.65 (2.31) (2.30) (1.99) (2.34) (2.28) Measures of foreign creditor rights (adjusted to a range of 0–10) Average investment profile (IP), 2004–9 (primary measure) 6.72 7.53 5.92 6.88 6.54 (1.59) (1.22) (1.51) (1.41) (1.75) Average repudiation & expropriation (R&E), 1982–97 (secondary measure) 6.21 6.73 5.71 6.53 5.88 (1.28) (1.24) (1.12) (1.33) (1.15) Notes. Means with standard deviations in parentheses. The RoL index (0–10), averaged over 2004–9, is taken from Kaufmann et al. (2010). The Gini ratio (0–100), averaged over 1978–2009 discontinuously, is sourced from the World Bank’s WDI and the Global Peace Index. High or low‐RoL (Gini) countries are defined as countries with the RoL index (Gini ratio) above or below the median of the base sample. The unconditional correlation coefficients between debt recovery costs and investment profile are 0.18, 0.12, 0.24, 0.12 and 0.21 for the base sample, high‐Rol, low‐Rol, low‐Gini, and high‐Gini countries, respectively. See the text for detailed variable definitions and sources. Open in new tab Then there is some doubt as to whether Figure 1(a) is a result of inter‐creditor discrimination. Since foreign (domestic) creditor rights institutions support external (domestic) financing, if institutionalised discrimination prevails in the world (i.e. foreign and domestic rights protection are negatively correlated), domestic creditor rights and external private debt may as well be inversely related. A positive correlation between domestic and foreign creditor rights proxies should clear away the doubt. To further establish that this discrimination effect is not driving the results, we conduct regressions for high‐RoL and for low‐Gini countries, where agents are unlikely to experience unfair treatment. The negative relationship between domestic rights and external debt in absolute value is actually larger than in the base sample. 1.2 Specification and Strategy To disentangle domestic and foreign creditor rights protection, we estimate the following equation: fm=αϕDm+βϕFm+γϕDm×ϕFm+δ′×zm+ϵm.(1) In country m, fm is the stock of PNG external debt per capita, ϕDm is a measure of domestic creditor rights, ϕFm is a measure of foreign creditor rights, and zm is a vector of controls including an intercept. Unless otherwise noted, all variables are time averaged over 2004–9. The interest centres around coefficients α, β, and γ, while δ captures effects of the controls in zm ⁠.16 The partial derivative of fm with respect to ϕDm is α+γϕFm ⁠. The interpretation of α is the effect of domestic creditor rights on the external level of PNG debt evaluated at ϕFm=0 ⁠. Our main hypothesis is that the estimated α^‐coefficient is significantly negative, which we identify as the outside‐option effect in the absence of foreign creditor rights protection. As ϕFm increases, α^+γ^ϕFm is still negative but its magnitude falls. This interactive effect is captured by a significant positive γ^ – stronger protection of foreign investors increases incentives to lend, crowding out concerns about better post‐default options. When ϕFm=ϕFmax ⁠, a threshold above which foreign investors in an average country are fully protected for debt disbursed, α^+γ^ϕFmax=0 and hence the outside‐option effect disappears. Let ϕFm continue to increase, α^+γ^ϕFm turns positive and becomes greater the higher is ϕFm ⁠. We do not model the positive link between ϕDm and fm at high degrees of protection of foreign creditors’ rights. We posit it as a signal of good governance inherent in domestic institutions, reinforced by stronger foreign creditor rights. Figure 2(a) depicts ∂fm/∂ϕDm with associated p‐value at different levels of ϕFm ⁠, including a 95% confidence interval and a histogram of investment profile. Fig. 2. Open in new tabDownload slide Estimation at Different Degrees of Creditor Rights Protection and Histograms. (a) Estimation at Different Degrees of Foreign Creditor Rights Protection. (b) Estimation at Different Degrees of Domestic Creditor Rights Protection Notes. The two subgraphs in panel (a) share the same horizontal axis. The lower subgraph illustrates the effect of domestic creditor rights protection (measured by debt recovery costs) depending on the level of foreign creditor rights protection (measured by investment profile); the grey area and the dashed curve, respectively, represent the 95% confidence interval and associated p‐values. The upper histogram is a graphical representation of the distribution of countries according to investment profile scores; the value on top of each bar is the number of countries that fall into the according bin and the grey curve is a normal density plot. Panel (b) is constructed in the same fashion, with strength of foreign creditor rights protection being the horizontal axis and the effect of foreign creditor rights protection on PNG external debt being the relationship of interest. Fig. 2. Open in new tabDownload slide Estimation at Different Degrees of Creditor Rights Protection and Histograms. (a) Estimation at Different Degrees of Foreign Creditor Rights Protection. (b) Estimation at Different Degrees of Domestic Creditor Rights Protection Notes. The two subgraphs in panel (a) share the same horizontal axis. The lower subgraph illustrates the effect of domestic creditor rights protection (measured by debt recovery costs) depending on the level of foreign creditor rights protection (measured by investment profile); the grey area and the dashed curve, respectively, represent the 95% confidence interval and associated p‐values. The upper histogram is a graphical representation of the distribution of countries according to investment profile scores; the value on top of each bar is the number of countries that fall into the according bin and the grey curve is a normal density plot. Panel (b) is constructed in the same fashion, with strength of foreign creditor rights protection being the horizontal axis and the effect of foreign creditor rights protection on PNG external debt being the relationship of interest. The estimated effect of ϕFm on fm ⁠, i.e., β^+γ^ϕDm ⁠, is positive over the range of values taken by ϕDm ⁠. According to our theoretical model, as ϕFm increases, on the one hand, foreign creditors’ incentives to lend will increase because the post‐foreign‐default value is reduced by further endowment loss (a positive component effect) but, on the other hand, such an incentive may decrease as well because stronger ϕFm also lowers the post‐domestic‐default‐value, which increases the benefits of having access to domestic markets after a foreign default, raising the post‐foreign‐default value (a negative component effect). It can be proved that the combined effect is, however, always positive. When ϕDm=0 ⁠, the post‐domestic‐default value is very sensitive to changes in ϕFm and hence the above negative component effect is maximised. The combined effect is positive but insignificant. As ϕDm improves, the negative component effect diminishes. Think of the polar case where domestic creditors are fully protected, ϕDm≥ϕDmax and β^+γ^ϕDmax is significant at the 1% level. The post‐domestic‐default value is so low such that no one will ever default domestically. The channel for the negative component effect is completely shut down. Our second hypothesis is that, given ϕDm=ϕD¯ (the mean value of ϕDm ⁠, greater than ϕDmax ⁠), the combined effect is significantly positive and becomes stronger the higher is ϕDm ⁠. Figure 2(b) is constructed in the same fashion as Figure 2(a), with ∂fm/∂ϕFm being the relationship of interest. Table 2(a) presents the results of estimating (1) without controlling for other debt determinants. In all regressions, the R 2s are fairly large, i.e., creditor rights have a strong explanatory power. The estimated outside‐option effect is always significantly negative for different combinations of various protection rights proxies.17 1.3 Robustness Checks 1.3.1 Omitted Variable: Adding A List of Controls Table 2(b) shows robustness to the inclusion of extra explanatory variables. Column (1) assesses implicit government guarantees using privately‐owned public debt as a percentage of GDP sourced from the Historical Public Debt Database, as the government would rescue private firms from the edge of default if they are big buyers of public bonds.18 Column (2) includes the log exchange rate to proxy involuntary private defaults under the pressure of currency devaluation by the central bank. Column (3) investigates the substitution effect by adding per capita liquid liability of the debt system (Beck and Demirgüç‐Kunt, 2009). Column (4) includes trade openness to examine the impact of direct sanctions. Column (5) captures economic fundamentals by adding the share of labour force who completed their secondary schooling (Barro and Lee, 2013). Column (6) concludes that external private debt is sensitive to government policies such as income tax. Table 2 Domestic and Foreign Creditor Rights Protection: OLS Regressions on Time‐series Averages (a) OLS estimation on a cross section of 85 countries . . (1) . (2) . (3) . (4) . (5) . (6) . (7) . (8) . (9) . (10) . Measure of ϕDm used: . Debt recovery costs . Cents on the $ recouped . Number of procedures . Enforcement costs . Strength of legal rights . Measure of ϕFm used: . IP . R&E . IP . R&E . IP . R&E . IP . R&E . IP . R&E . Domestic creditor rights when ϕFm=0 −0.34** −0.21* −0.47** −0.42** −0.14 −0.34*** −0.95** −0.61** −0.52** −0.65*** (0.13) (0.12) (0.20) (0.19) (0.15) (0.13) (0.42) (0.29) (0.21) (0.22) Foreign creditor rights when ϕDm=ϕD¯ 0.37*** 0.46*** 0.38*** 0.43*** 0.27** 0.37*** 0.27*** 0.38*** 0.29*** 0.38*** (0.11) (0.14) (0.12) (0.13) (0.09) (0.11) (0.08) (0.11) (0.10) (0.10) Interaction term 0.05** 0.03* 0.07** 0.07** 0.03 0.06*** 0.16** 0.12** 0.08*** 0.11*** (0.02) (0.02) (0.03) (0.03) (0.02) (0.02) (0.07) (0.05) (0.03) (0.04) R 2 0.27 0.26 0.33 0.31 0.27 0.30 0.41 0.35 0.34 0.36 (a) OLS estimation on a cross section of 85 countries . . (1) . (2) . (3) . (4) . (5) . (6) . (7) . (8) . (9) . (10) . Measure of ϕDm used: . Debt recovery costs . Cents on the $ recouped . Number of procedures . Enforcement costs . Strength of legal rights . Measure of ϕFm used: . IP . R&E . IP . R&E . IP . R&E . IP . R&E . IP . R&E . Domestic creditor rights when ϕFm=0 −0.34** −0.21* −0.47** −0.42** −0.14 −0.34*** −0.95** −0.61** −0.52** −0.65*** (0.13) (0.12) (0.20) (0.19) (0.15) (0.13) (0.42) (0.29) (0.21) (0.22) Foreign creditor rights when ϕDm=ϕD¯ 0.37*** 0.46*** 0.38*** 0.43*** 0.27** 0.37*** 0.27*** 0.38*** 0.29*** 0.38*** (0.11) (0.14) (0.12) (0.13) (0.09) (0.11) (0.08) (0.11) (0.10) (0.10) Interaction term 0.05** 0.03* 0.07** 0.07** 0.03 0.06*** 0.16** 0.12** 0.08*** 0.11*** (0.02) (0.02) (0.03) (0.03) (0.02) (0.02) (0.07) (0.05) (0.03) (0.04) R 2 0.27 0.26 0.33 0.31 0.27 0.30 0.41 0.35 0.34 0.36 (b) Robustness regressions: adding a list of controls . Control variable added: . Implicitly guar. . Public influence . Substitution eff. . Trade sanction . Fundamentals . Gov’t policies . Info. frictions . Incomplete mkt . Governance . Latitude . Debt recovery costs when ϕFm=0 −0.25** −0.30** −0.24* −0.29** −0.34** −0.27* −0.31** −0.25* −0.35** −0.26** (0.13) (0.14) (0.13) (0.14) (0.13) (0.15) (0.14) (0.14) (0.14) (0.12) Investment profile when ϕDm=ϕD¯ 0.24*** 0.26*** 0.19** 0.23*** 0.23*** 0.31*** 0.20*** 0.23*** 0.20*** 0.19*** (0.08) (0.10) (0.09) (0.08) (0.07) (0.10) (0.08) (0..08) (0.07) (0.07) Interaction term 0.04** 0.04** 0.03* 0.04** 0.05** 0.04** 0.04** 0.04* 0.05** 0.03* (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) Log real GDPpc in 2004 0.92*** 0.89*** 0.74*** 0.78*** 0.61** 0.97*** 0.89*** 0.96*** 0.63*** 0.56*** (0.26) (0.26) (0.23) (0.22) (0.25) (0.26) (0.23) (0.27) (0.20) (0.19) Control variable 0.31 0.09 0.06* 0.50 0.03** −0.02** −2.84** −0.04 0.29** 2.35*** (0.23) (0.08) (0.03) (0.75) (0.01) (0.01) (1.09) (0.03) (0.12) (0.66) Countries 85 85 85 85 77 69 85 84 85 85 R 2 0.39 0.39 0.42 0.39 0.47 0.45 0.44 0.40 0.41 0.45 (b) Robustness regressions: adding a list of controls . Control variable added: . Implicitly guar. . Public influence . Substitution eff. . Trade sanction . Fundamentals . Gov’t policies . Info. frictions . Incomplete mkt . Governance . Latitude . Debt recovery costs when ϕFm=0 −0.25** −0.30** −0.24* −0.29** −0.34** −0.27* −0.31** −0.25* −0.35** −0.26** (0.13) (0.14) (0.13) (0.14) (0.13) (0.15) (0.14) (0.14) (0.14) (0.12) Investment profile when ϕDm=ϕD¯ 0.24*** 0.26*** 0.19** 0.23*** 0.23*** 0.31*** 0.20*** 0.23*** 0.20*** 0.19*** (0.08) (0.10) (0.09) (0.08) (0.07) (0.10) (0.08) (0..08) (0.07) (0.07) Interaction term 0.04** 0.04** 0.03* 0.04** 0.05** 0.04** 0.04** 0.04* 0.05** 0.03* (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) Log real GDPpc in 2004 0.92*** 0.89*** 0.74*** 0.78*** 0.61** 0.97*** 0.89*** 0.96*** 0.63*** 0.56*** (0.26) (0.26) (0.23) (0.22) (0.25) (0.26) (0.23) (0.27) (0.20) (0.19) Control variable 0.31 0.09 0.06* 0.50 0.03** −0.02** −2.84** −0.04 0.29** 2.35*** (0.23) (0.08) (0.03) (0.75) (0.01) (0.01) (1.09) (0.03) (0.12) (0.66) Countries 85 85 85 85 77 69 85 84 85 85 R 2 0.39 0.39 0.42 0.39 0.47 0.45 0.44 0.40 0.41 0.45 Notes. The dependent variable is the PNG external debt stocks per capita, expressed in thousands of constant 2005 US$, and averaged over 2004–9. All regressions include an intercept and are estimated by OLS with robust standard errors in parentheses. *, ** and *** denote significant at 10%, 5% and 1%, respectively. Open in new tab Table 2 Domestic and Foreign Creditor Rights Protection: OLS Regressions on Time‐series Averages (a) OLS estimation on a cross section of 85 countries . . (1) . (2) . (3) . (4) . (5) . (6) . (7) . (8) . (9) . (10) . Measure of ϕDm used: . Debt recovery costs . Cents on the $ recouped . Number of procedures . Enforcement costs . Strength of legal rights . Measure of ϕFm used: . IP . R&E . IP . R&E . IP . R&E . IP . R&E . IP . R&E . Domestic creditor rights when ϕFm=0 −0.34** −0.21* −0.47** −0.42** −0.14 −0.34*** −0.95** −0.61** −0.52** −0.65*** (0.13) (0.12) (0.20) (0.19) (0.15) (0.13) (0.42) (0.29) (0.21) (0.22) Foreign creditor rights when ϕDm=ϕD¯ 0.37*** 0.46*** 0.38*** 0.43*** 0.27** 0.37*** 0.27*** 0.38*** 0.29*** 0.38*** (0.11) (0.14) (0.12) (0.13) (0.09) (0.11) (0.08) (0.11) (0.10) (0.10) Interaction term 0.05** 0.03* 0.07** 0.07** 0.03 0.06*** 0.16** 0.12** 0.08*** 0.11*** (0.02) (0.02) (0.03) (0.03) (0.02) (0.02) (0.07) (0.05) (0.03) (0.04) R 2 0.27 0.26 0.33 0.31 0.27 0.30 0.41 0.35 0.34 0.36 (a) OLS estimation on a cross section of 85 countries . . (1) . (2) . (3) . (4) . (5) . (6) . (7) . (8) . (9) . (10) . Measure of ϕDm used: . Debt recovery costs . Cents on the $ recouped . Number of procedures . Enforcement costs . Strength of legal rights . Measure of ϕFm used: . IP . R&E . IP . R&E . IP . R&E . IP . R&E . IP . R&E . Domestic creditor rights when ϕFm=0 −0.34** −0.21* −0.47** −0.42** −0.14 −0.34*** −0.95** −0.61** −0.52** −0.65*** (0.13) (0.12) (0.20) (0.19) (0.15) (0.13) (0.42) (0.29) (0.21) (0.22) Foreign creditor rights when ϕDm=ϕD¯ 0.37*** 0.46*** 0.38*** 0.43*** 0.27** 0.37*** 0.27*** 0.38*** 0.29*** 0.38*** (0.11) (0.14) (0.12) (0.13) (0.09) (0.11) (0.08) (0.11) (0.10) (0.10) Interaction term 0.05** 0.03* 0.07** 0.07** 0.03 0.06*** 0.16** 0.12** 0.08*** 0.11*** (0.02) (0.02) (0.03) (0.03) (0.02) (0.02) (0.07) (0.05) (0.03) (0.04) R 2 0.27 0.26 0.33 0.31 0.27 0.30 0.41 0.35 0.34 0.36 (b) Robustness regressions: adding a list of controls . Control variable added: . Implicitly guar. . Public influence . Substitution eff. . Trade sanction . Fundamentals . Gov’t policies . Info. frictions . Incomplete mkt . Governance . Latitude . Debt recovery costs when ϕFm=0 −0.25** −0.30** −0.24* −0.29** −0.34** −0.27* −0.31** −0.25* −0.35** −0.26** (0.13) (0.14) (0.13) (0.14) (0.13) (0.15) (0.14) (0.14) (0.14) (0.12) Investment profile when ϕDm=ϕD¯ 0.24*** 0.26*** 0.19** 0.23*** 0.23*** 0.31*** 0.20*** 0.23*** 0.20*** 0.19*** (0.08) (0.10) (0.09) (0.08) (0.07) (0.10) (0.08) (0..08) (0.07) (0.07) Interaction term 0.04** 0.04** 0.03* 0.04** 0.05** 0.04** 0.04** 0.04* 0.05** 0.03* (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) Log real GDPpc in 2004 0.92*** 0.89*** 0.74*** 0.78*** 0.61** 0.97*** 0.89*** 0.96*** 0.63*** 0.56*** (0.26) (0.26) (0.23) (0.22) (0.25) (0.26) (0.23) (0.27) (0.20) (0.19) Control variable 0.31 0.09 0.06* 0.50 0.03** −0.02** −2.84** −0.04 0.29** 2.35*** (0.23) (0.08) (0.03) (0.75) (0.01) (0.01) (1.09) (0.03) (0.12) (0.66) Countries 85 85 85 85 77 69 85 84 85 85 R 2 0.39 0.39 0.42 0.39 0.47 0.45 0.44 0.40 0.41 0.45 (b) Robustness regressions: adding a list of controls . Control variable added: . Implicitly guar. . Public influence . Substitution eff. . Trade sanction . Fundamentals . Gov’t policies . Info. frictions . Incomplete mkt . Governance . Latitude . Debt recovery costs when ϕFm=0 −0.25** −0.30** −0.24* −0.29** −0.34** −0.27* −0.31** −0.25* −0.35** −0.26** (0.13) (0.14) (0.13) (0.14) (0.13) (0.15) (0.14) (0.14) (0.14) (0.12) Investment profile when ϕDm=ϕD¯ 0.24*** 0.26*** 0.19** 0.23*** 0.23*** 0.31*** 0.20*** 0.23*** 0.20*** 0.19*** (0.08) (0.10) (0.09) (0.08) (0.07) (0.10) (0.08) (0..08) (0.07) (0.07) Interaction term 0.04** 0.04** 0.03* 0.04** 0.05** 0.04** 0.04** 0.04* 0.05** 0.03* (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) Log real GDPpc in 2004 0.92*** 0.89*** 0.74*** 0.78*** 0.61** 0.97*** 0.89*** 0.96*** 0.63*** 0.56*** (0.26) (0.26) (0.23) (0.22) (0.25) (0.26) (0.23) (0.27) (0.20) (0.19) Control variable 0.31 0.09 0.06* 0.50 0.03** −0.02** −2.84** −0.04 0.29** 2.35*** (0.23) (0.08) (0.03) (0.75) (0.01) (0.01) (1.09) (0.03) (0.12) (0.66) Countries 85 85 85 85 77 69 85 84 85 85 R 2 0.39 0.39 0.42 0.39 0.47 0.45 0.44 0.40 0.41 0.45 Notes. The dependent variable is the PNG external debt stocks per capita, expressed in thousands of constant 2005 US$, and averaged over 2004–9. All regressions include an intercept and are estimated by OLS with robust standard errors in parentheses. *, ** and *** denote significant at 10%, 5% and 1%, respectively. Open in new tab Columns (7)–(8) control for other reasons of market imperfections besides limited commitment: asymmetric information and market incompleteness. The former is measured by ‘distantness’ (Alfaro et al., 2008), a GDP‐weighted average of distances from the capital city of a country to the capital cities of rest countries. The latter is measured by M2/M1. Column (9) inspects other aspects of politics besides institutions: governance and democracy. We use the bureaucracy quality score from the ICRG as a proxy for decision‐making within the government and use the DEMOC index from the Polity IV to account for access to government offices. The last column emphasises the latitude since countries closer to the equator are argued to be less developed economically and institutionally. We repeat all exercises for every year during the sample period. The coefficients are, in general, smaller but they remain statistically significant except for 2009. We suspect this is due to the cut‐off of foreign lending following the global contagion of the 2007 credit crisis. Our results are also robust to a panel data analysis, particular subsamples, and different normalisation methods of the dependent variable. 1.3.2 Multicollinearity: Partitioning Protection Rights One might worry about multicollinearity because of high correlations between creditor rights and the interaction term. To show that we are capturing each variable’s independent effect, we define whether protection rights are ‘weak’ or ‘strong’ in country m along the domestic and foreign dimension. Weak protection is associated with a 0, and strong protection is associated with a 1. The investors’ protection rights set is thus divided into four states: {(0, 0), (1, 0),(0, 1), (1, 1)}. The pair (0, 0) denotes weak protection of both domestic and foreign creditors, the pair (1, 0) denotes strong protection of domestic creditors but weak protection of foreign creditors, the pair (0, 1) denotes weak protection of domestic creditors but strong protection of foreign creditors and the pair (1, 1) denotes strong protection of both domestic and foreign creditors. Then it is possible to create dummy variables, ϕ0,0m ⁠, ϕ1,0m ⁠, ϕ0,1m and ϕ1,1m ⁠, for each of the four states and estimate specification (2) (below) instead. To define the threshold for the ‘weak’ versus ‘strong’ protection, we take the first principal component of the primary and secondary measures of domestic creditor rights and the first principal component of both measures of foreign creditor rights. The threshold is set at the median for the respective principal component. Our hypotheses could be cast in terms of restrictions on the coefficients from: fm=κ+β0,0ϕ0,0m+β1,0ϕ1,0m+β0,1ϕ0,1m+δzm+ϵm,(2) where κ is an intercept, β0,0 ⁠, β1,0 and β0,1 are the differences in the effect of each state vis‐à‐vis the omitted state (1, 1), and zm=1 if the log of real GDP per capita in 2004 is above its sample mean; zm=0 otherwise. In particular, our main hypothesis would be β^0,0>β^1,0 and a negative β^0,1 ⁠. Our second hypothesis would be β^0,1>β^0,0 and a negative β^1,0 ⁠. Although we can not reject some of these hypotheses’ null counterparts (because partition ignores the interaction effect), the sorting order of these estimated coefficients in Table 3 is consistent with our conjectures. Table 3 Robustness Regressions: Partitioning Protection Rights Dummy: weak protection of both domestic and foreign creditors −0.61** (0.28) Dummy: strong protection of domestic creditors; weak protection of foreign creditors −0.66** (0.29) Dummy: weak protection of domestic creditors; strong protection of foreign creditors −0.41 (0.37) Dummy: log real GDPpc in 2004 0.67*** (0.16) Countries 85 R 2 0.22 One‐sided F‐test for β^0,0>β^1,0 (p‐value) 0.28 One‐sided F‐test for β^0,1>β^0,0 (p‐value) 0.17 Dummy: weak protection of both domestic and foreign creditors −0.61** (0.28) Dummy: strong protection of domestic creditors; weak protection of foreign creditors −0.66** (0.29) Dummy: weak protection of domestic creditors; strong protection of foreign creditors −0.41 (0.37) Dummy: log real GDPpc in 2004 0.67*** (0.16) Countries 85 R 2 0.22 One‐sided F‐test for β^0,0>β^1,0 (p‐value) 0.28 One‐sided F‐test for β^0,1>β^0,0 (p‐value) 0.17 Notes. The dependent variable is the PNG external debt stocks per capita, expressed in thousands of constant 2005 US$, and averaged over 2004–9. The regression includes an intercept and is estimated by OLS with robust standard errors in parentheses. *, ** and *** denote significant at 10%, 5% and 1%, respectively. Open in new tab Table 3 Robustness Regressions: Partitioning Protection Rights Dummy: weak protection of both domestic and foreign creditors −0.61** (0.28) Dummy: strong protection of domestic creditors; weak protection of foreign creditors −0.66** (0.29) Dummy: weak protection of domestic creditors; strong protection of foreign creditors −0.41 (0.37) Dummy: log real GDPpc in 2004 0.67*** (0.16) Countries 85 R 2 0.22 One‐sided F‐test for β^0,0>β^1,0 (p‐value) 0.28 One‐sided F‐test for β^0,1>β^0,0 (p‐value) 0.17 Dummy: weak protection of both domestic and foreign creditors −0.61** (0.28) Dummy: strong protection of domestic creditors; weak protection of foreign creditors −0.66** (0.29) Dummy: weak protection of domestic creditors; strong protection of foreign creditors −0.41 (0.37) Dummy: log real GDPpc in 2004 0.67*** (0.16) Countries 85 R 2 0.22 One‐sided F‐test for β^0,0>β^1,0 (p‐value) 0.28 One‐sided F‐test for β^0,1>β^0,0 (p‐value) 0.17 Notes. The dependent variable is the PNG external debt stocks per capita, expressed in thousands of constant 2005 US$, and averaged over 2004–9. The regression includes an intercept and is estimated by OLS with robust standard errors in parentheses. *, ** and *** denote significant at 10%, 5% and 1%, respectively. Open in new tab 1.3.3 Simultaneity: Using Historical Measures of Creditor Rights It is plausible to argue that more capital inflows incentivise creditor rights institutions building,19 or that analysts downgrade protection rights ratings of countries with large external debt burden. Table 4 checks into the effect of historical creditor rights protection (La Porta et al., 1998) on the current level of PNG external debt in column (2). Though insignificant due to a small sample, the coefficient reported for ϕDm is more negative than its modern counterpart in column (1), after the positive reverse causality is removed. However, a relatively small difference between coefficients from the two columns implies that double causality is not critical among developing countries.20 Table 4 Robustness Regressions: Using Historical Creditor Rights Measures . (1) . (2) . Average strength of legal rights, 2005–9 when ϕFm=0 −0.12 (0.18) Average investment profile, 2004–9 when ϕDm=ϕD¯ 0.18** (0.08) Domestic creditor rights index in 1998 when ϕFm=0 −0.34 (0.31) Average investment profile, 1984–99 when ϕDm=ϕD¯ 0.44* (0.22) Interaction term 0.02 0.07 (0.04) (0.07) Countries 26 26 R 2 0.20 0.10 . (1) . (2) . Average strength of legal rights, 2005–9 when ϕFm=0 −0.12 (0.18) Average investment profile, 2004–9 when ϕDm=ϕD¯ 0.18** (0.08) Domestic creditor rights index in 1998 when ϕFm=0 −0.34 (0.31) Average investment profile, 1984–99 when ϕDm=ϕD¯ 0.44* (0.22) Interaction term 0.02 0.07 (0.04) (0.07) Countries 26 26 R 2 0.20 0.10 Notes. The dependent variable is the PNG external debt stocks per capita, expressed in thousands of constant 2005 US$, and averaged over 2004–9. Both regressions include an intercept and are estimated by OLS with robust standard errors in parentheses. Amongst our five proxies for domestic creditor rights, only the strength of legal rights proxy has historical values. *, ** and *** denote significant at 10%, 5% and 1%, respectively. Open in new tab Table 4 Robustness Regressions: Using Historical Creditor Rights Measures . (1) . (2) . Average strength of legal rights, 2005–9 when ϕFm=0 −0.12 (0.18) Average investment profile, 2004–9 when ϕDm=ϕD¯ 0.18** (0.08) Domestic creditor rights index in 1998 when ϕFm=0 −0.34 (0.31) Average investment profile, 1984–99 when ϕDm=ϕD¯ 0.44* (0.22) Interaction term 0.02 0.07 (0.04) (0.07) Countries 26 26 R 2 0.20 0.10 . (1) . (2) . Average strength of legal rights, 2005–9 when ϕFm=0 −0.12 (0.18) Average investment profile, 2004–9 when ϕDm=ϕD¯ 0.18** (0.08) Domestic creditor rights index in 1998 when ϕFm=0 −0.34 (0.31) Average investment profile, 1984–99 when ϕDm=ϕD¯ 0.44* (0.22) Interaction term 0.02 0.07 (0.04) (0.07) Countries 26 26 R 2 0.20 0.10 Notes. The dependent variable is the PNG external debt stocks per capita, expressed in thousands of constant 2005 US$, and averaged over 2004–9. Both regressions include an intercept and are estimated by OLS with robust standard errors in parentheses. Amongst our five proxies for domestic creditor rights, only the strength of legal rights proxy has historical values. *, ** and *** denote significant at 10%, 5% and 1%, respectively. Open in new tab 1.3.4 Measurement Error: Instrumenting Creditor Rights By Multiple Ivs Attenuation bias may occur pervasively since our proxies could correspond poorly to the concepts we purport to measure. Besides, p‐value for the Hausman test is far below 1%, so the null hypothesis that protection rights are exogenous is rejected. We address this issue with a multiple IV approach; if successful, we should see a large increase in the absolute value of coefficients on domestic and foreign creditor rights measures. Following the line of research led by La Porta et al. (1998), the legal origin is an important determinant of institutional variations. Thus, our first instrument is a dummy for ‘German civil law’. It equals 1 if a country can be traced to a German legal system; 0 if a country’s legal rules originate in the ‘common law’ or the ‘French civil law’ tradition. We expect it to be significantly and positively related to ϕFm ⁠, as German civil‐law countries offer the strongest law enforcement and the greatest protection for foreign investors. The second instrument is Roeder’s (2001) ethnolinguistic fractionalisation index computed for the year of 1985. It is defined as the probability that two randomly selected persons from a given country will not belong to the same ethnic or linguistic group. We expect it to be significantly and negatively related to ϕDm ⁠. The reason is that the presence of many different ethnolinguistic groups causes domestic tensions, as bureaucrats will favour members in the same group (Mauro, 1995). Table 5 summarises the results of the two‐stage least squares (2SLS) regression in column (2) of panel (a), the corresponding first‐stage regression in columns (1) and (2) of (b), and the OLS counterpart in column (1) of (a). To test for IV validity, we resort to linkages between religious beliefs and legal institutions (Stulz and Williamson, 2003). Instrumenting protection rights with the percentage of population belonging to three widely spread religions (La Porta et al., 1999), we show in columns (3) and (4) that our instruments are excludable from the main regression. Large over‐identification test statistics also support validity. Table 5 Robustness Regressions: Instrumenting Creditor Rights with Multiple IVs . (1) . (2) . (3) . (4) . . OLS . 2SLS . 2SLS . 2SLS . (a) OLS and second stage IV estimates Debt recovery costs when ϕFm=0 (⁠ ϕDm^ when ϕFm^=0 if 2SLS) −0.29** −3.04*** −1.73*** −1.98*** (0.14) (1.09) (0.48) (0.47) Investment profile when ϕDm=ϕD¯ (⁠ ϕFm^ when ϕDm^=mean value of ϕDm^ if 2SLS) 0.24*** 0.75* 0.26 0.78** (0.08) (0.43) (0.33) (0.39) Interaction term (⁠ ϕDm^×ϕFm^ if 2SLS) 0.04** 0.49*** 0.27*** 0.28*** (0.02) (0.17) (0.07) (0.06) Log real GDPpc in 2004 0.84*** −0.31 0.52 −0.21 (0.23) (0.71) (0.55) (0.55) Dummy: German civil law legal origin 0.89 (0.83) Ethnolinguistic fractionalisation in 1985 −1.20 (0.78) R 2 in OLS 0.38 0.61 0.52 0.54 . (1) . (2) . (3) . (4) . . OLS . 2SLS . 2SLS . 2SLS . (a) OLS and second stage IV estimates Debt recovery costs when ϕFm=0 (⁠ ϕDm^ when ϕFm^=0 if 2SLS) −0.29** −3.04*** −1.73*** −1.98*** (0.14) (1.09) (0.48) (0.47) Investment profile when ϕDm=ϕD¯ (⁠ ϕFm^ when ϕDm^=mean value of ϕDm^ if 2SLS) 0.24*** 0.75* 0.26 0.78** (0.08) (0.43) (0.33) (0.39) Interaction term (⁠ ϕDm^×ϕFm^ if 2SLS) 0.04** 0.49*** 0.27*** 0.28*** (0.02) (0.17) (0.07) (0.06) Log real GDPpc in 2004 0.84*** −0.31 0.52 −0.21 (0.23) (0.71) (0.55) (0.55) Dummy: German civil law legal origin 0.89 (0.83) Ethnolinguistic fractionalisation in 1985 −1.20 (0.78) R 2 in OLS 0.38 0.61 0.52 0.54 (b) First stage for dependent variable . ϕDm . ϕFm . ϕDm . ϕFm . Dummy: German civil law legal origin −0.49 1.05** −0.43 0.86* (0.96) (0.50) (0.96) (0.51) Ethnolinguistic fractionalisation in 1985 −3.93*** 0.28 −2.95** 0.18 (1.24) (0.56) (1.26) (0.65) Percentage of Protestant population in 1980 −0.05* 0.00 (0.03) (0.01) Percentage of Catholic population in 1980 −0.01 −0.01 (0.01) (0.01) Percentage of Muslim population in 1980 0.01 −0.00 (0.01) (0.01) Log real GDPpc in 2004 0.35 1.41*** 0.76 1.51*** (0.76) (0.38) (0.76) (0.35) Countries 85 85 85 85 R2 in first stage 0.17 0.29 0.24 0.29 (b) First stage for dependent variable . ϕDm . ϕFm . ϕDm . ϕFm . Dummy: German civil law legal origin −0.49 1.05** −0.43 0.86* (0.96) (0.50) (0.96) (0.51) Ethnolinguistic fractionalisation in 1985 −3.93*** 0.28 −2.95** 0.18 (1.24) (0.56) (1.26) (0.65) Percentage of Protestant population in 1980 −0.05* 0.00 (0.03) (0.01) Percentage of Catholic population in 1980 −0.01 −0.01 (0.01) (0.01) Percentage of Muslim population in 1980 0.01 −0.00 (0.01) (0.01) Log real GDPpc in 2004 0.35 1.41*** 0.76 1.51*** (0.76) (0.38) (0.76) (0.35) Countries 85 85 85 85 R2 in first stage 0.17 0.29 0.24 0.29 Notes. The dependent variable is the PNG external debt stocks per capita, expressed in thousands of constant 2005 US$, and averaged over 2004–9. All regressions include an intercept and are estimated by OLS with robust standard errors in parentheses. *, ** and *** denote significant at 10%, 5% and 1%, respectively. Open in new tab Table 5 Robustness Regressions: Instrumenting Creditor Rights with Multiple IVs . (1) . (2) . (3) . (4) . . OLS . 2SLS . 2SLS . 2SLS . (a) OLS and second stage IV estimates Debt recovery costs when ϕFm=0 (⁠ ϕDm^ when ϕFm^=0 if 2SLS) −0.29** −3.04*** −1.73*** −1.98*** (0.14) (1.09) (0.48) (0.47) Investment profile when ϕDm=ϕD¯ (⁠ ϕFm^ when ϕDm^=mean value of ϕDm^ if 2SLS) 0.24*** 0.75* 0.26 0.78** (0.08) (0.43) (0.33) (0.39) Interaction term (⁠ ϕDm^×ϕFm^ if 2SLS) 0.04** 0.49*** 0.27*** 0.28*** (0.02) (0.17) (0.07) (0.06) Log real GDPpc in 2004 0.84*** −0.31 0.52 −0.21 (0.23) (0.71) (0.55) (0.55) Dummy: German civil law legal origin 0.89 (0.83) Ethnolinguistic fractionalisation in 1985 −1.20 (0.78) R 2 in OLS 0.38 0.61 0.52 0.54 . (1) . (2) . (3) . (4) . . OLS . 2SLS . 2SLS . 2SLS . (a) OLS and second stage IV estimates Debt recovery costs when ϕFm=0 (⁠ ϕDm^ when ϕFm^=0 if 2SLS) −0.29** −3.04*** −1.73*** −1.98*** (0.14) (1.09) (0.48) (0.47) Investment profile when ϕDm=ϕD¯ (⁠ ϕFm^ when ϕDm^=mean value of ϕDm^ if 2SLS) 0.24*** 0.75* 0.26 0.78** (0.08) (0.43) (0.33) (0.39) Interaction term (⁠ ϕDm^×ϕFm^ if 2SLS) 0.04** 0.49*** 0.27*** 0.28*** (0.02) (0.17) (0.07) (0.06) Log real GDPpc in 2004 0.84*** −0.31 0.52 −0.21 (0.23) (0.71) (0.55) (0.55) Dummy: German civil law legal origin 0.89 (0.83) Ethnolinguistic fractionalisation in 1985 −1.20 (0.78) R 2 in OLS 0.38 0.61 0.52 0.54 (b) First stage for dependent variable . ϕDm . ϕFm . ϕDm . ϕFm . Dummy: German civil law legal origin −0.49 1.05** −0.43 0.86* (0.96) (0.50) (0.96) (0.51) Ethnolinguistic fractionalisation in 1985 −3.93*** 0.28 −2.95** 0.18 (1.24) (0.56) (1.26) (0.65) Percentage of Protestant population in 1980 −0.05* 0.00 (0.03) (0.01) Percentage of Catholic population in 1980 −0.01 −0.01 (0.01) (0.01) Percentage of Muslim population in 1980 0.01 −0.00 (0.01) (0.01) Log real GDPpc in 2004 0.35 1.41*** 0.76 1.51*** (0.76) (0.38) (0.76) (0.35) Countries 85 85 85 85 R2 in first stage 0.17 0.29 0.24 0.29 (b) First stage for dependent variable . ϕDm . ϕFm . ϕDm . ϕFm . Dummy: German civil law legal origin −0.49 1.05** −0.43 0.86* (0.96) (0.50) (0.96) (0.51) Ethnolinguistic fractionalisation in 1985 −3.93*** 0.28 −2.95** 0.18 (1.24) (0.56) (1.26) (0.65) Percentage of Protestant population in 1980 −0.05* 0.00 (0.03) (0.01) Percentage of Catholic population in 1980 −0.01 −0.01 (0.01) (0.01) Percentage of Muslim population in 1980 0.01 −0.00 (0.01) (0.01) Log real GDPpc in 2004 0.35 1.41*** 0.76 1.51*** (0.76) (0.38) (0.76) (0.35) Countries 85 85 85 85 R2 in first stage 0.17 0.29 0.24 0.29 Notes. The dependent variable is the PNG external debt stocks per capita, expressed in thousands of constant 2005 US$, and averaged over 2004–9. All regressions include an intercept and are estimated by OLS with robust standard errors in parentheses. *, ** and *** denote significant at 10%, 5% and 1%, respectively. Open in new tab 2 Model Economy 2.1 A Decentralised Borrowing and Default Setup The world consists of M countries, each populated by N private sectors with a unit mass of identical agents in each sector. Time is discrete and infinite.21 Information at date t is represented by state θt ⁠, drawn from a finite space. A history until t is summarised in θt ⁠. Let π(θt) be the probability of θt as of date 0, with π(θ0)=1 ⁠, and π(θr|θt) be the conditional probability of θr given θt ⁠.22 There is one non‐storable good denoted by enm(θt) ⁠, the endowment received by type n in country m at θt ⁠. cnm(θt) denotes the corresponding consumption. Agents have lifetime preferences of the form: ∑t=0∞∑θtπ(θt)βtu[cnm(θt)]=E0∑t=0∞βtu[cnm(θt)],(3) where β is the discount factor and u(·) is the period utility function satisfying the Inada conditions. Agents trade state‐contingent contracts in home market exclusively, and in a single international market universally. Denote the prices of a full set of domestic and external contracts, purchased at θt in exchange for 1 unit of good next period given θt+1 occurs by pm(θt,θt+1) and q(θt,θt+1) ⁠. dnm(θt,θt+1) and fnm(θt,θt+1) are the quantities, with negative values denoting debts. The budget constraint for each history is given by: enm(θt)+dnm(θt)+fnm(θt)≥cnm(θt)+∑θt+1[pm(θt,θt+1)dnm(θt,θt+1)+q(θt,θt+1)fnm(θt,θt+1)].(BC) When two agents enter into a contract, the borrower gets the funds but the lender must rely on a promise. Such a commitment is not credible, since at maturity the borrower may default. To give borrowers incentives to repay, assume that contracts are enforced by the threat of output loss and market exclusion. First, two sets of institutions, indexed by i ∈ {D, F}, exist to protect domestic and foreign creditors’ rights separately. Creditors seize an exogenous fraction ϕim∈0,ϕimax of endowment in periods after the default event under i’s jurisdiction.23 A gap between independent ϕDm and ϕFm stands for discrimination. The second form of threat is permanent reversion to financial autarchy. After a domestic default an individual is prohibited from trading in both markets. The value received in the agent’s autarchy (AA) is: Anm(θt,ϕim)=∑r=t∞∑θr|θtπ(θr|θt)βr−tu[(1−ϕDm−ϕFm)enm(θr)]. One can think of this as a setup where a financial sector arranges all transfers.24 Since it cares more about the home market, it will refuse to serve domestic defaulters, compelling them to repudiate internationally. Suppose a foreign default occurs first, an individual can still trade domestic contracts at original prices. The maximised utility obtained from the agent’s international autarchy (AIA) is as follows: Vnm[θt,ϕim,{pm},dnm(θt)]=max{cnm,dnm}∑r=t∞∑θr|θtπ(θr|θt)βr−tu[cnm(θr)]s.t. (1−ϕFm)enm(θr)+dnm(θr)≥cnm(θr)+∑θr+1pm(θr,θr+1)dnm(θr,θr+1)∀θr(4) ∑s=r∞∑θs|θrπ(θs|θr)βs−ru[cnm(θs)]≥Anm(θr,ϕim)∀θr(5) dnm(θr,θr+1)≥−B∀(θr,θr+1)and{pm},dnm(θt)given, where (4) is the budget constraint, (5) is the domestic participation constraint and the last is a no‐Ponzi condition with B > 0 taken to be large enough to not bind. Whether a foreign default can trigger a domestic one depends on dnm carried into the (AIA) problem. Let dnm,* be the cutoff making type n in country m indifferent between living in (AA) and (AIA). As Vnm is strictly increasing in dnm ⁠, if dnm≤dnm,* ⁠, which implies Vnm≤Anm ⁠, the agent would immediately default on domestic debts and we are back to (AA). Otherwise, Vnm>Anm ⁠, simultaneous defaults are undesirable. The following participation constraint captures this limited enforcement friction, and require that the utility from repayment must be no less than that from strategic domestic or foreign default, whichever is higher, ∑r=t∞∑θr|θtπ(θr|θt)βr−tu[cnm(θr)]≥Anm(θt,ϕim)ifdnm(θt)≤dnm,*(θt)Vnm[θt,ϕim,{pm},dnm(θt)]otherwise∀θt.(PC) 2.2 Competitive Equilibrium The agent’s problem (AP) is to choose sequences for consumption and domestic and external contracts to maximise (3) subject to (BC), (PC) and no‐Ponzi conditions, taking dnm(θ0) and fnm(θ0) as given. The necessary and sufficient first‐order conditions (FOCs) are in Appendix A. Let us define a default‐free competitive equilibrium in this economy. Definition 1. A Trade Equilibrium is an allocation {cnm,T,dnm,T,fnm,T} and a price sequence {pm,T,qT} for all countries and sectors such that: (i) each agent solves its (AP) given prices, institutional quality, and initial contract holdings; (ii) resource feasibility is satisfied at all θt ⁠, ∑m∑ncnm,T(θt)≤∑m∑nenm(θt) ⁠; and (iii) for all (θt,θt+1) ⁠, the international market is clear ∑m∑nfnm,T(θt,θt+1)=0 and all M domestic markets are clear ∑ndnm,T(θt,θt+1)=0,∀m ⁠. Proposition 1 characterises constrainedness and equilibrium prices. To conserve space, we denote the inter‐temporal marginal rate of substitution (IMRS) as: IMRSnm,T(θt,θt+1)=π(θt+1|θt)βu′cnm,T(θt,θt+1)u′[cnm,T(θt)]∀n,m,(θt,θt+1). Proposition 1 In equilibrium, for any (θt,θt+1) ⁠, the M countries can be classified into two groups: (i) quasi‐unconstrained countries, where agents are either participation unconstrained in both markets, i.e., (PC) is slack, or dually participation constrained, i.e., Vnm≤Anm and (PC) binds, qT(θt,θt+1)=pm,T(θt,θt+1)≥IMRSnm,T(θt,θt+1); (ii) internationally constrained countries, where all agents are participation constrained in the international market, i.e., Vnm>Anm and (PC) binds, qT(θt,θt+1)>max{pm,T(θt,θt+1),IMRSnm,T(θt,θt+1)}. Proof. See Appendix B. The trade equilibrium has two features that act as effective deterrents to foreign defaults (note that a domestic default never happens before a foreign one). One novel feature is that IMRSs are unequalised across sectors within a country and, hence, agents face partially uninsurable idiosyncratic risk. Another feature is that the price of the external contract is the maximum of all domestic prices. A defaulting borrower on external debt will only have access to a domestic market where constraints (5)s are tighter and interest rates are higher in later periods when the individual wants to borrow. 2.3 On the Link Between Theory and Empirics First, we have assumed trade in state‐contingent securities. In domestic markets, agents typically have a vast set of assets available. In the international market, although the majority of asset offerings are non‐contingent, debt stocks in developing countries are frequently rescheduled upon changes in the borrowers’ situation, let alone bundled contingencies like excusable default and convertibility provisions. At the beginning of θt ⁠, ∑θt+1q(θt,θt+1)fnm(θt,θt+1) can be interpreted as net foreign asset offerings, while the same summation over only negative quantities represents gross external debt positions. After θt+1 is realised, as there is only one active contract, net and gross foreign asset stocks are identical. The per capita external debt stock of an internationally constrained country at node (θt,θt+1) is given by: fm(θt,θt+1)=1N∑n=1Nfnm(θt,θt+1). Normalising one period to the average maturity of PNG external contract and pricing the numéraire good at market par value, it seems reasonable to relate the absolute value of fm to our dependent variable in the empirical section. Thus, one could map a cross section of internationally constrained countries at θt+1 into the 85 base‐sample developing countries over 2004–9. Second, in the model, default is not allowed and ϕim is the degree of creditors’ rights protection off the equilibrium path. In contrast, empirical proxies measure how default events are resolved in practice. But if we look at the underlying proxy construction methodology, these proxies are derived from opinions based on hypothetical cases involving contract breaches, though the original purpose is to make data comparable across countries. In addition, models with equilibrium default require market incompleteness and are often used to calibrate business cycle indicators during sovereign default episodes. We, however, would like to capture foreign investor decisions, given payments are made as scheduled. 2.4 Comparative‐Statics Exercises In quasi‐unconstrained countries, consider dually participation constrained agents at θt+1 ⁠. Higher ϕDm reduces Anm and slackens (PC). The implicit ceiling on debt borrowed at θt with payment contingent on the occurrence of θt+1 can be raised. Since external contract is priced the same as domestic contract, agents may cut external debt holdings, at the same time borrow more than the debt limit increase domestically. This is the so called substitution effect. An alike argument applies to a rise in ϕFm ⁠. Our focus is on internationally constrained countries. Holding ϕFm=0 ⁠, a lower ϕDm increases Anm at all θr with r ≥ t + 1. In θt+1’s (AIA) problem, larger Anms tighten a series of constraint (5)s, curtail domestic credit and lead to consumption dispersion. As argued by De Santis (2007), even a small increase in dispersion risk will cause a big fall in Vnm and hence a large relaxation of (PC).25 The implicit θt+1‐contingent debt ceiling can again be raised. Since external contract is cheaper under this circumstance, it is preferable to borrow more internationally. This outside‐option effect is estimated by α^ in the empirical part. Now allow ϕFm to rise until near its ceiling ϕFmax ⁠. As a result of further endowment loss, (PC) binds at a lower level of Vnm ⁠. Moreover, the (5) series become less stringent, for Vnm decreases less than Anm due to concavity. So the change in ϕDm has to be larger to make an impact that matters. The outside‐option effect is undermined. This pattern of interaction is captured by γ^ ⁠. Next hold ϕDm constant. A higher ϕFm ⁠, on the one hand, tightens (4) and decreases Vnm (a large positive impact on fm ⁠) but, on the other hand, it also relaxes the (5) series and offsets the reduction in Vnm (a small negative impact on fm ⁠). When ϕDm=0 ⁠, the combined impact is positive but insignificant, as is confirmed empirically. As ϕDm improves, the negative component impact diminishes and eventually disappears at ϕDmax ⁠. Starting from there, foreign creditor rights protection exerts a strongly positive effect on external debt holdings. If we evaluate ϕDm at ϕD¯>ϕDmax ⁠, this pure positive effect is estimated by β^ in the empirical section. A Centralised Borrowing and Default Benchmark In this subsection, we define a global planner’s constrained efficient allocations in the above‐mentioned economy. The solution can be found in a simpler setup, where each government trades internationally with other governments and allocates endowments plus sovereign borrowings domestically as a utilitarian planner. Subject to the same type of limited enforcement problem, sovereign debts are made available only to the extent that payments can be enforced by the threat of reversion to autarchy. Let φn be the welfare weight assigned to type n sector for any m; the autarchy value is given by: Vm(θt)=max{cnm}∑n=1Nφn∑r=t∞∑θr|θtπ(θr|θt)βr−tu[cnm(θr)]s.t. ∑n=1Nenm(θr)≥∑n=1Ncnm(θr)∀θr. Before a default, the country‐specific planner’s problem (PP) is: max{cnm,Fm}E0∑n=1Nφn∑t=0∞βtu[cnm(θt)]s.t. ∑n=1Nenm(θt)+Fm(θt)≥∑n=1Ncnm(θt)+∑θt+1q(θt,θt+1)Fm(θt,θt+1)∀θt ∑n=1Nφn∑r=t∞∑θr|θtπ(θr|θt)βr−tu[cnm(θr)]≥Vm(θt)∀θt(6) Fm(θt,θt+1)≥−B∀(θt,θt+1)andFm(θ0)given. Definition 2. An equilibrium with country‐specific planners is an allocation {cnm,G,Fm,G} and a price sequence {qG} for all countries such that: (i) each planner solves its (PP) given prices, a subservient judiciary, and initial contract holdings; (ii) resource feasibility is satisfied at all θt ⁠, ∑m∑ncnm,G(θt)≤∑m∑nenm(θt) ⁠; and (iii) for all (θt,θt+1) ⁠, the sovereign debt market is clear ∑mFm,G(θt,θt+1)=0 ⁠. The weighted sum of agents’ utilities could be higher or lower under decentralisation than the one under centralisation. For comparison, we introduce an auxiliary planner’s problem (APP), which is (PP) modified in that the right hand side of (6) in each θt is replaced by: Vm,T(θt,ϕim)=∑{n:dnm,T≤dnm,*}φnAnm(θt,ϕim)+∑{n:dnm,T>dnm,*}φnVnm[θt,ϕim,{pm,T},dnm,T(θt)], and a series of (5)s is added. With appropriate transfers of initial contracts, the solutions to all agents’ (AP)s also solve the country’s (APP). If we assume (APP)’s constraint set is a superset of (PP)’s constraint set, i.e. Vm(θt)>Vm,T(θt,ϕim)and∑r=t∞∑θr|θtπ(θr|θt)βr−tu[cnm,G(θr)]>Anm(θt,ϕim)∀m,θt, then centralisation becomes welfare inferior. This is because the resulting perfection of the domestic creditor rights institution can overall be detrimental. Extending this result worldwide, the trade equilibrium specified might be more efficient than the second‐best allocations from Definition 2 in the sense of Kehoe and Levine (1993). While it is challenging to find analytically the exact range of ϕim that leads to such outcomes, the next Section provides an illustrative example to help build boundary conditions numerically. (3) Efficiency: an Expository Example Consider a toy economy of two countries, m ∈ {1,2}, two agents, n ∈ {a, b}, u(·) =ln(·) and four equiprobable histories. In Table 6, we interpret 1 as the average endowment, ±y as the systematic shock and ±ɛ as the idiosyncratic shock. It is assumed that y, ε∈0,12 so that all endowments are positive. In each history, the value of either shock alternates between positive and negative deterministically. However, the sum of systematic shocks in two countries and the sum of idiosyncratic shocks of two agents in a given country are always 0. Table 6 Sequences of Endowments . History 1 . History 2 . History 3 . History 4 . History 1 . History 2 . History 3 . History 4 . . At even‐numbered periods t = 0, 2, 4, … . At odd‐numbered periods t = 1, 3, 5 … . ea1 1+y+ε 1+y−ε 1−y−ε 1−y+ε 1−y−ε 1−y+ε 1+y+ε 1+y−ε eb1 1+y−ε 1+y+ε 1−y+ε 1−y−ε 1−y+ε 1−y−ε 1+y−ε 1+y+ε ea2 1−y−ε 1−y+ε 1+y+ε 1+y−ε 1+y+ε 1+y−ε 1−y−ε 1−y+ε eb2 1−y+ε 1−y−ε 1+y−ε 1+y+ε 1+y−ε 1+y+ε 1−y−ε 1−y−ε . History 1 . History 2 . History 3 . History 4 . History 1 . History 2 . History 3 . History 4 . . At even‐numbered periods t = 0, 2, 4, … . At odd‐numbered periods t = 1, 3, 5 … . ea1 1+y+ε 1+y−ε 1−y−ε 1−y+ε 1−y−ε 1−y+ε 1+y+ε 1+y−ε eb1 1+y−ε 1+y+ε 1−y+ε 1−y−ε 1−y+ε 1−y−ε 1+y−ε 1+y+ε ea2 1−y−ε 1−y+ε 1+y+ε 1+y−ε 1+y+ε 1+y−ε 1−y−ε 1−y+ε eb2 1−y+ε 1−y−ε 1+y−ε 1+y+ε 1+y−ε 1+y+ε 1−y−ε 1−y−ε Open in new tab Table 6 Sequences of Endowments . History 1 . History 2 . History 3 . History 4 . History 1 . History 2 . History 3 . History 4 . . At even‐numbered periods t = 0, 2, 4, … . At odd‐numbered periods t = 1, 3, 5 … . ea1 1+y+ε 1+y−ε 1−y−ε 1−y+ε 1−y−ε 1−y+ε 1+y+ε 1+y−ε eb1 1+y−ε 1+y+ε 1−y+ε 1−y−ε 1−y+ε 1−y−ε 1+y−ε 1+y+ε ea2 1−y−ε 1−y+ε 1+y+ε 1+y−ε 1+y+ε 1+y−ε 1−y−ε 1−y+ε eb2 1−y+ε 1−y−ε 1+y−ε 1+y+ε 1+y−ε 1+y+ε 1−y−ε 1−y−ε . History 1 . History 2 . History 3 . History 4 . History 1 . History 2 . History 3 . History 4 . . At even‐numbered periods t = 0, 2, 4, … . At odd‐numbered periods t = 1, 3, 5 … . ea1 1+y+ε 1+y−ε 1−y−ε 1−y+ε 1−y−ε 1−y+ε 1+y+ε 1+y−ε eb1 1+y−ε 1+y+ε 1−y+ε 1−y−ε 1−y+ε 1−y−ε 1+y−ε 1+y+ε ea2 1−y−ε 1−y+ε 1+y+ε 1+y−ε 1+y+ε 1+y−ε 1−y−ε 1−y+ε eb2 1−y+ε 1−y−ε 1+y−ε 1+y+ε 1+y−ε 1+y+ε 1−y−ε 1−y−ε Open in new tab As before, agents trade domestic and external contracts, only this time the contracts are lifelong agreements signed ex ante, i.e. before the realisation of the history. Once the history has been revealed, agents decide whether to default depending on the value of outside options. Let xnm denote the consumption deviation for agent n in country m. Ex post, the lifetime preference by sticking to agreements can be rescaled as: U(xnm)=ln(1+xnm)+βln(1−xnm). Institutions play an important part in the decentralised borrowing and default setup. To simplify the exposition, assume that foreign payments are unenforced, ϕFm=0,∀m ⁠, and domestic payments are imperfectly enforced, ϕDm=ϕD∈0,ϕDmax,∀m ⁠. If we can show that decentralisation allows higher welfare even in the absence of foreign institutions, this result would be stronger for any positive ϕFm ⁠. Without loss of generality, suppose history 1 occurs. By symmetry, the optimal consumption deviations are: xa1=xaT=z+ζ(ϕD),xb1=xbT=z−ζ(ϕD);xa2=−xaT=−z−ζ(ϕD),xb2=−xbT=−z+ζ(ϕD), where z ≥ 0 and ζ(ϕD)≥0 are the systematic and idiosyncratic component of consumption deviation, respectively. Note ζ(ϕD) is a decreasing function and ζ(ϕDmax)=0 ⁠. Denote p=β[1−z+ζ(ϕD)]/[1+z−ζ(ϕD)] as the one‐period domestic contract price in an internationally constrained country, while agents in a quasi‐constrained country confront a higher domestic contract price: q=β1+z+ζ(ϕD)1−z−ζ(ϕD).(7) The external contract price remains at a constant level of q. The domestic contract price flips back and forth between p in odd‐numbered periods and q in even‐numbered periods. At t = 0, both agents in country 1 are participation constrained in the international market. In an equivalent Arrow–Debreu setup, the period‐0 dated domestic price and the net capital outflow (NCO) are: P1(t)=(pq)t2q(pq)t−12;NCOn1(t)=y−zz−y∀n∈{a,b},att=0,2,4,…att=1,3,5,…. According to (A.3) in Appendix A, the summation of present values of future NCOs should be equal to 0: ∑t=0∞P1(t)NCOn1(t)=(y−z)+q(z−y)1−pq=0.(8) In country 2, agent a is participation unconstrained in both markets; and b is dually participation constrained: U(xb2)=ln[(1−ϕD)(1−y+ε)]+βln[(1−ϕD)(1+y−ε)].(9) There are two solutions to (8). The first is (AIA), or z = y, while the second requires q = 1, which further implies z+ζ(ϕD)=1−β/1+β using (7). Together with (9), we have two equations to solve for two unknowns: z and ζ(ϕD) ⁠. Assuming the same welfare weight is assigned to the two agents, the weighted sum of their ex ante utilities in either country is of the form: E0U(xaT)+U(xbT)2=U(xaT)+U(−xaT)+U(xbT)+U(−xbT)4.(10) Turning to the centralised borrowing and default benchmark, idiosyncratic risks are eliminated. At history 1, each country can be aggregated into a planner with the following consumption deviation: xa1=xb1=xG;xa2=xb2=−xG, where xG is the smallest deviation satisfying country 1’s participation constraint at even‐numbered t:26 xG=minx≥0{x:U(x)≥U(y)}. The ex ante utility of the planner in either country is of the form: E0[U(xG)]=U(xG)+U(−xG)2.(11) When ϕD is small, the weighted sum of agents’ ex ante utilities under decentralisation may exceed the planner’s ex ante utility under centralisation. To see this graphically, set the parameters, β = 0.85, y = 0.1 and ɛ = 0.25. In the extreme case when ϕD=0 ⁠, all ex post utilities are ranked in panels (a) and (b) of Figure 3 and, because a larger size of international capital flows can be supported under decentralisation, (10) is strictly higher than (11). As shown in (c) and (d), as ϕD increases, (10) falls below (11) eventually. We can decompose the difference between (10) and (11) into three parts: E0U(xaT)+U(xbT)2−E0[U(z)]⏟Part 1: lossdueto lessdomestic borrowing+E0U1−β1+β−E0[U(xG)]⏟Part 2: loss due to theexternality of private anddefault borrowing+E0[U(z)]−E0U1−β1+β⏟Part 3: gain frommore internationalborrowing. Fig. 3. Open in new tabDownload slide Ex Post and Ex Ante Utilities in the Numerical Example. (a) Ex Post Lifetime Utilities Given History 1 Occurs. (b) Ex Post Lifetime Utilities Given History 1 Occurs. (c) Comparison of Net Capital Outflows in Two Setups. (d) Comparison of Ex Ante Utilities in Two Setups Notes. β is set to 0.85 so that some government debts can be supported. Let y = 0.1 and ɛ = 0.25 to capture the notion that aggregate risk is an order of magnitude smaller than idiosyncratic risk. The size of NCO in the decentralised setup will exceed that in the centralised setup when ϕD<ϕD† ⁠. As ϕD continues to deteriorate and falls below ϕD‡ ⁠, the weighted sum of agents’ utility is strictly higher than the planner’s utility ex ante. Fig. 3. Open in new tabDownload slide Ex Post and Ex Ante Utilities in the Numerical Example. (a) Ex Post Lifetime Utilities Given History 1 Occurs. (b) Ex Post Lifetime Utilities Given History 1 Occurs. (c) Comparison of Net Capital Outflows in Two Setups. (d) Comparison of Ex Ante Utilities in Two Setups Notes. β is set to 0.85 so that some government debts can be supported. Let y = 0.1 and ɛ = 0.25 to capture the notion that aggregate risk is an order of magnitude smaller than idiosyncratic risk. The size of NCO in the decentralised setup will exceed that in the centralised setup when ϕD<ϕD† ⁠. As ϕD continues to deteriorate and falls below ϕD‡ ⁠, the weighted sum of agents’ utility is strictly higher than the planner’s utility ex ante. The concavity of U(·) ensures Part 1 is a pure loss. The monotonicity of E0[U(·)] ensures Part 2 and 3 are indeed loss and gain, respectively. Lower post‐foreign‐default utility improves international borrowing, creating gains that outweigh the total losses as long as ϕD is below ϕD‡=0.004 (in this numerical example, ϕDmax=0.011 ⁠). 4 Conclusion There is a consensus on the role of institutions in finance. Hypothetically, if we can sort out the complex and contingent relations of entangled rights protecting institutions, it is intuitive that its domestic side supports home market, and its foreign side boosts external intermediation. Using data of 85 developing countries, we document a negative relationship between the domestic arm of creditor rights protection and external private debt. Two offhand explanations emerge in this regard. First, one can attribute the negative effect to ex ante substitutability between domestic and external debts. But this is only true at late stage of financial development, where the distinction between home and foreign transactions blurs. Second, both sets of creditor rights are likely to be the result of a common base, so we may get the negative relation because of discrimination. But we observe no persistent intercreditor discrimination across our sample countries. We provide a new perspective: the curtailment of credit can be a strategic decision by foreign investors in response to domestic borrowers’ better post‐default scenario. To attain efficient borrowing, domestic rights building should be combined with eliminating foreign discrimination; otherwise, there is an incentive to default internationally first, causing foreign creditors to tighten constraints. The empirical result is evident in the international round tripping in relation to financial transactions in Africa, India and China. It also underpins the ‘revived Bretton Woods’ interpretation of the capital flow structure vis‐à‐vis emerging Asia (Dooley et al., 2004). We contribute to the theoretical literature by modelling creditor rights continuously in the traditional debt repudiation models. During capital account liberalisation, imperfect domestic protection rights have conflicting effects. Worse domestic rights protection generates a loss by hindering local assets trade but it also induces foreign creditors to raise debt ceiling. Under certain, not unusual conditions, the latter outside‐option effect can create gains that outweigh domestic costs. Appendix A: Optimality Conditions for (Ap) Drop n, m. Let λ and ν be the Lagrange multipliers on (4) and (5). Then FOCs of the (AIA) problem yield: λ(θr)=π(θr|θt)βr−tu′[cθtA(θr)]1+∑s=tr∑θr|θsν(θs)βt−sπ(θr|θs)π(θr|θt), where cθtA(θr) is the optimal consumption after a foreign default at θt ⁠. Apply the envelope theorem, ∂V[θt,ϕi,{p},d(θt)]∂d(θt)=λ(θt)=[1+ν(θt)]u′[cθtA(θt)]. Let μ be the Lagrange multiplier on (PC). Using the (AP)’s FOCs at node (θt,θt+1) ⁠, pT=IMRST1+μπβt+1(1+L)ifdT≤d*1+μπβt+1(1+L)u′(cT)−(1+ν)u′(cθtA)u′(cT)otherwise;(A.1) qT=IMRST1+μπβt+1(1+L),(A.2) where L is defined as the weighted sum of all μs until θt ⁠, L=∑r=0t∑θr≤θtπ(θt|θr)π(θt)β−rμ(θr). Notice that (A.1) and (A.2) may be insufficient, for the (AP) has a non‐convex constraint set. This follows from the observation that V[θt,ϕi,{p},d(θt)] is concave in d(θt) ⁠, ∀θt ⁠, ϕi ⁠, and {p}. To surmount the problem, replace (PC) when d>d* by the following constraints that are weaker but necessary. The first constraint states that the stream of future NCOs, discounted at domestic prices, must be no larger than the present value of legal penalties on a foreign default, ∑r=t∞∑θr|θtP(θr)∑θr+1q(θr,θr+1)f(θr,θr+1)−f(θr)≤ϕF∑r=t∞∑θr|θtP(θr)e(θr)∀θt,(A.3) where P(θr)=∏r=0t−1p(θr,θr+1) is the date 0 price of a r‐period domestic contingent contract. The second constraint is designed to prevent domestic default given a foreign default has occurred, ∑r=t∞∑θr|θtP^(θr)∑θr+1p(θr,θr+1)d(θr,θr+1)−d(θr)≤ϕD∑r=t∞∑θr|θtP^(θr)e(θr)∀θt,(A.4) where P^(θr)=P(θt)π(θr|θt)βr−t{u′[(1−ϕD−ϕF)e(θr)]}/{u′[(1−ϕD−ϕF)e(θt)]} is the IMRS between (AA) endowments in θt and θr ⁠. The last constraint is imposed on agents with μ(θt)>0 and (PC) binding, Domestic credit unconstrainedif(1+ν)u′(cθtA)=u′(cT)d(θt)≥dT(θt)if(1+ν)u′(cθtA)<u′(cT)d(θt)≤dT(θt)if(1+ν)u′(cθtA)>u′(cT).(A.5) Intuitively, under (1+ν)u′(cθtA)<u′(cT) ⁠, more domestic borrowing than dT causes the right hand side of (PC) to decrease at a slower rate than the left hand side, violating the binding (PC), ∂Vnm[θt,ϕi,{p},d(θt)]∂d(θt)|d=dT=(1+ν)u′(cθtA)<u′(cT)=∂∑r=t∞∑θr|θtπ(θr|θt)βr-tu[c(θr)]∂d(θt)|c=cT. Otherwise, the above relationship reverses. Those agents are lending constrained domestically. A slack (PC) implies that (A.3)–(A.4) are slack. A binding (PC) implies that (A.3) and (A.5) bind. Let μ~ ⁠, μ^ ⁠, and ξ be the Lagrange multipliers on (A.3), (A.4), and (A.5). The FOCs of the alternative (AP) at node (θt,θt+1) are: pT=IMRST1+μπβt+1(1+L^)ifdT≤d*1+μ^πβt+1(1+L^)0if(1+ν)u′(cθtA)=u′(cT)P^−ξ/μ^u′(cT)if(1+ν)u′(cθtA)<u′(cT)P^+ξ/μ^u′(cT)if(1+ν)u′(cθtA)>u′(cT)otherwise;(A.6) qT=IMRST1+μπβt+1(1+L~)ifdT≤d*1+μ~Pπβt+1(1+L~)otherwise,(A.7) where L^(θt)=∑r=0t∑{θr:dT≤d*}π(θt|θr)π(θt)β-rμ(θr)+∑{θr:dT>d*}P^(θr)π(θt)u′[cT(θt)]β-rμ^(θr),L~(θt)=∑r=0t∑{θr:dT≤d*}π(θt|θr)π(θt)β-rμ(θr)+∑{θr:dT>d*}P(θr)π(θt)u′[cT(θt)]β-rμ~(θr). After rescaling μ~ ⁠, μ^ ⁠, and ξ, (A.6)–(A.7) are identical to (A.1) and (A.2). The new constraint set is convex by construction. Together with a transversality condition, (A.1) and (A.2) are indeed sufficient. Appendix B: Proof of Proposition 1 Note that, for all n, m, and (θt,θt+1) ⁠, (A.1) and (A.2) imply qT≥pT and qT≥IMRST ⁠. Then qT=pT in a country means that, for all of its agents, either they are dually participation unconstrained with μ = 0 or they face a binding (PC) with μ > 0 given V ≤ A. qT>pT in a country means that all agents face a binding (PC) with μ > 0 given V > A; the IMRS of a agent in such a country could be lower, equal to, or higher than the domestic price, depending on whether the individual is borrowing constrained, lending constrained, or credit unconstrained domestically according to (A.5). Footnotes 1 " Panizza et al. (2009) provide a literature review on newly‐written theoretical contributions on this topic. 2 " Faria and Mauro (2009) construct an index from the World Bank’s Worldwide Governance Indicators (WGI) to show that better institutions tilt countries’ external capital structures towards equity and away from debt. Lane (2004) and Qian (2012) construct another broad assessment by averaging International Country Risk Guide’s (ICRG) IRIS‐3 variables from the Political Risk Services group. Lane (2004) emphasises the role of institutions in determining the stock of external public debt and Qian (2012) examines how institutions affect the likelihood of sovereign defaults. 3 " See subsection 1.1 for the construction of the base sample. The cross‐country average stock of PNG external debt per capita also grew from $60 to $663 over these two decades. 4 " Using bank‐level data from 148 countries over 2000–6, Cole and Ariss (2013) present evidence that better legal protection, especially in developing countries, lead to higher levels of private credit and, consequently, to better financial‐sector development. 5 " See Rajan and Zingales (1998), Galindo and Schiantarelli (2003), Harrison and McMillan (2003), Rand (2007) and Héricourt and Poncet (2009). Collectively, they show that private sectors face severe external debt constraints. 6 " In this literature, lending is sustained by sanction and/or output loss. Early models assume that defaulters are completely and permanently excluded from their only way of insuring against shocks – capital markets, see Eaton and Gersovitz (1981), Kehoe and Levine (1993) and Alvarez and Jermann (2000). One recent strand of research replaces complete exclusion with partial exclusion, see Jeske (2006), Wright (2006) and Hellwig and Lorenzoni (2007). Using a recursive method, another quantitative strand of research looks at temporary exclusion with default allowed in equilibrium (see Aguiar and Gopinath, 2006; Arellano, 2008; Kim and Zhang, 2012). The third strand of literature criticises what if debtors have other opportunities to smooth consumption (Bulow and Rogoff, 1989; Kehoe and Perri, 2002). 7 " Capital markets are ‘sophisticated’, in the sense that lenders are able to know the nationality and financial conditions of borrowers, and can freely impose tailored credit constraints on different individuals. 8 " Fernández‐Arias and Lombardo (2002) study the same issue and find overborrowing. The key difference between the two studies is that in Uribe (2006), when the constraint is binding, the interest rate for foreign borrowing adjusts upward and induces agents to internalise the credit limit, while Fernández‐Arias and Lombardo (2002) assume that the world interest rate is fixed and agents fail to internalise the debt ceiling in a deterministic environment. 9 " The IMF’s International Financial Statistics (IFS) include both public and private issuers and holders of debt securities. PNG debtors are not clearly identified for most countries for long periods. Recent efforts have been made to achieve higher accuracy and wider coverage but not in the direction of distinguishing PPG and PNG debt. To improve accuracy, the literature draws attention to the importance of valuation effects. Lane and Milesi‐Ferretti (2007) measure debt liabilities by cumulating flows and adjust for the effects of exchange rate changes. Kraay et al. (2005) construct estimates of foreigners’ position in domestic loans using the price of investment goods in local currency. To widen coverage, the IMF merges the IFS with the World Bank’s IDS in its External Debt Statistics. The Joint External Debt Hub combines data developed individually in the BIS, the IMF, the Organisation for Economic Cooperation and Development (OECD), and the World Bank. 10 " Most base‐sample countries are emerging and developing economies, though 10 of them (Chile, Czech Republic, Greece, Hungary, Israel, Mexico, Poland, Slovak Republic, South Korea and Turkey) are member of the OECD. Chile is the only OECD member which also belongs to the organisation of developing countries, the Group of 77. 11 " Arteta and Hale (2008) present evidence of a decline in foreign credit to domestic private firms during periods of sovereign crises (Moody’s Investors Service, 2011; Reinhart and Rogoff, 2011; Beers and Nadueau, 2014). In the base sample, Dominican Republic carried out pre‐emptive restructurings between 2004–5. Zimbabwe restructured over 98.5% of domestic short‐term debts. Ecuador defaulted on $3.9 billion worth of external debts in late 2008. 12 " Montenegro has no data before 2006 (Serbia and Montenegro ceases to exist as one country when the two republics declared independence on June 2006). Data on the Slovak Republic and Israel are blank for 2008–9 and 2009 respectively. 13 " For each year, the formula used for rescaling is (Vmax−V−m)/(Vmax−Vmin) or (V+m−Vmin)/(Vmax−Vmin) multiplied by 10. V−m represents debt recovery costs, procedural complexity, or enforcement costs; V+m represents cents on the dollar recouped. The values for Vmax and Vmin are set at 1.5 standard deviations above and below the sample average respectively. Countries with values outside of the [Vmax,Vmin] range receive ratings of either 0 or 10 accordingly. 14 " See Bhattacharya and Daouk (2009). 15 " Chong and Gradstein (2007) establish double causality between better institutional quality and a more equal income distribution. 16 " We find no evidence of significance when adding the square terms in ϕDm and ϕFm ⁠, so they are not included. 17 " One exception is when we use a number of procedures as the measure of domestic rights and investment profile as the measure of foreign rights. The reason may be that foreign individuals are subject to similar procedural steps as residents in explicit laws. 18 " Chile’s debt nationalisation in 1982 is one example. Alternatively, we follow Schich and Lindh (2012) in using credit rating agencies‘ rating uplifts to proxy the value of implicit guarantees and the results remain unchanged. 19 " Governments might reform actively through weakening incumbents’ opposition to financial development (Rajan and Zingales, 2003) and/or respond positively to investors’ wishes of quality legal services (Ahlquist and Prakash, 2010). 20 " This result coincides with the insignificant causal relation discovered by Gourinchas and Jeanne (2006) in terms of capital flows affecting GDP per capita, and by Klein and Olivei (2008) in relation to capital account liberalisation affecting financial depth. 21 " A finite horizon can lead to persistent indeterminacy in dynamic competitive equilibria (Azariadis and Lambertini, 2002). 22 " It is not necessary to assume a particular evolution process, e.g., a Markov process. 23 " We assume ϕDm+ϕFm<1 for all m to ensure both domestic and foreign creditors’ rights can be preserved if they simultaneously file claims. By doing so, like Reinhart and Rogoff (2011), we need not to deal with debt seniority. 24 " See Ábrahám and Cárceles‐Poveda (2006) for a limited commitment model with competitive financial intermediaries. 25 " In other words, a worse outside option enhances international borrowing capacity. This idea has been used by Krueger and Perri (2006) to show that higher income volatility results in more hedging among households in a closed economy. 26 " To support some sovereign debt, the systematic shock y must satisfy two restrictions given an exogenous discount factor. The first is U(y) > 0, or − ln(1 + y)/ ln(1 − y) > β, otherwise consumption is fully smoothed, xG=0 ⁠; the second restriction is argmaxx≥0U(x)<y ⁠, or y > (1 − β)/(1 + β), otherwise the planner’s autarchy is preferred, xG=y ⁠. References Ábrahám , Á. and Cárceles‐Poveda , E. ( 2006 ). ‘ Endogenous incomplete markets, enforcement constraints, and intermediation ’, Theoretical Economics , vol. 1 ( 4 ), pp. 439 – 59 . OpenURL Placeholder Text WorldCat Acemoglu , D. and Johnson , S. ( 2005 ). ‘ Unbundling institutions ’, Journal of Political Economy , vol. 113 ( 5 ), pp. 949 – 95 . Google Scholar Crossref Search ADS WorldCat Aguiar , M. and Gopinath , G. ( 2006 ). ‘ Defaultable debt, interest rates, and the current account ’, Journal of International Economics , vol. 69 ( 1 ), pp. 64 – 83 . Google Scholar Crossref Search ADS WorldCat Ahlquist , J. and Prakash , A. ( 2010 ). ‘ FDI and the costs of contract enforcement in developing countries ’, Policy Sciences , vol. 43 ( 2 ), pp. 181 – 200 . Google Scholar Crossref Search ADS WorldCat Alfaro , L. , Kalemli‐Ozcan , S. and Volosovych , V. ( 2008 ). ‘ Why doesn’t capital flow from rich to poor countries? An empirical investigation ’, Review of Economics and Statistics , vol. 90 ( 2 ), pp. 347 – 68 . Google Scholar Crossref Search ADS WorldCat Alvarez , F. and Jermann , U. ( 2000 ). ‘ Efficiency, equilibrium, and asset pricing with risk of default ’, Econometrica , vol. 68 ( 4 ), pp. 775 – 98 . Google Scholar Crossref Search ADS WorldCat Arellano , C. ( 2008 ). ‘ Default risk and income fluctuations in emerging economies ’, American Economic Review , vol. 98 ( 3 ), pp. 690 – 712 . Google Scholar Crossref Search ADS WorldCat Arteta , C. and Hale , G. ( 2008 ). ‘ Sovereign debt crises and credit to the private sector ’, Journal of International Economics , vol. 74 ( 1 ), pp. 53 – 69 . Google Scholar Crossref Search ADS WorldCat Azariadis , C. and Lambertini , L. ( 2002 ). ‘ Endogenous debt constraints in lifecycle economies ’, Review of Economic Studies , vol. 70 ( 3 ), pp. 461 – 87 . Google Scholar Crossref Search ADS WorldCat Bae , K. and Goyal , V. ( 2009 ). ‘ Creditor rights, enforcement, and bank loans ’, Journal of Finance , vol. 64 ( 2 ), pp. 823 – 60 . Google Scholar Crossref Search ADS WorldCat Barro , R. and Lee , J.W. ( 2013 ). ‘ A new data set of educational attainment in the world, 1950–2010 ’, Journal of Development Economics , vol. 104 ( 3 ), pp. 184 – 98 . Google Scholar Crossref Search ADS WorldCat Beck , T. and Demirgüç‐Kunt , A. ( 2009 ). ‘ Financial institutions and markets across countries and over time – data and analysis ’, Working Paper, World Bank Policy Research. Beers , D. and Nadueau J.‐S. ( 2014 ), ‘ Introducing a new database of sovereign defaults ’, Working Paper, Bank of Canada. Bhattacharya , U. and Daouk , H. ( 2009 ). ‘ When no law is better than a good law ’, Review of Finance , vol. 13 ( 4 ), pp. 577 – 627 . Google Scholar Crossref Search ADS WorldCat Broner , F. and Ventura , J. ( 2011 ). ‘ Globalization and risk sharing ’, Review of Economic Studies , vol. 78 ( 1 ), pp. 49 – 82 . Google Scholar Crossref Search ADS WorldCat Bulow , J. and Rogoff , K. ( 1989 ). ‘ Sovereign debt: is to forgive or forget? ’ American Economic Review , vol. 79 ( 1 ), pp. 43 – 50 . OpenURL Placeholder Text WorldCat Chong , A. and Gradstein , M. ( 2007 ). ‘ Inequality and institutions ’, Review of Economics and Statistics , vol. 89 ( 3 ), pp. 454 – 65 . Google Scholar Crossref Search ADS WorldCat Cole , R. and Ariss , R. ( 2013 ). ‘ Legal origin, creditor protection and bank lending around the world ’, available at: http://ssrn.com/abstract=997582 (last accessed: 17 February 2016). De Santis , M. ( 2007 ). ‘ Individual consumption risk and the welfare cost of business cycles ’, American Economic Review , vol. 97 ( 4 ), pp. 1488 – 506 . Google Scholar Crossref Search ADS WorldCat Dembiermont , C. , Drehmann , M. and Muksakunratana , S. ( 2013 ), ‘ How much does the private sector really borrow? A new database for total credit to the private non‐finanancial sector ’, BIS Quarterly Review , (March): pp. 65 – 81 . OpenURL Placeholder Text WorldCat Djankov , S. , Hart , O., McLiesh , C. and Shleifer , A. ( 2008 ). ? Debt enforcement around the world ’, Journal of Political Economy , vol. 116 ( 6 ), pp. 1105 ‐ 49 . Google Scholar Crossref Search ADS WorldCat Djankov , S. , La Porta , R., López‐de‐Silanes , F. and Shleifer , A. ( 2003 ). ‘ Courts ’, Quarterly Journal of Economics , vol. 118 ( 2 ), pp. 342 – 87 . Google Scholar Crossref Search ADS WorldCat Djankov , S. , McLiesh , C. and Shleifer , A. ( 2007 ). ‘ Private credit in 129 countries ’, Journal of Financial Economics , vol. 84 ( 2 ), pp. 299 – 329 . Google Scholar Crossref Search ADS WorldCat Dooley , M. , Folkerts‐Landau , D. and Garber , P. ( 2004 ). ‘ The revived Bretton Woods system ’, International Journal of Finance and Economics , vol. 9 ( 4 ), pp. 307 – 13 . Google Scholar Crossref Search ADS WorldCat Eaton , J. and Gersovitz , M. ( 1981 ). ‘ Debt with potential repudiation: theoretical and empirical analysis ’, Review of Economic Studies , vol. 48 ( 2 ), pp. 289 – 309 . Google Scholar Crossref Search ADS WorldCat Erce , A. ( 2012 ). ‘ Selective sovereign defaults ’, Working Paper, Federal Reserve Bank of Dallas Globalization and Monetary Policy Institute. Esty , B. ( 2004 ). ‘ When do foreign banks finance domestic projects? New evidence on the importance of legal and financial systems ’, available at: http://ssrn.com/abstract=594526 (last accessed: 17 February 2016). Faria , A. and Mauro , P. ( 2009 ). ‘ Institutions and the external capital structure of countries ’, Journal of International Money and Finance , vol. 28 ( 3 ), pp. 367 – 91 . Google Scholar Crossref Search ADS WorldCat Fernández‐Arias , E. and Lombardo , D. ( 2002 ). ‘Market discipline and exuberant foreign borrowing’, in ( L. Hernández and K. Schmidt‐Hebbel, eds.), Banking, Financial Integration, and International Crises , pp. 333 – 60 , Santiago, Chile : Central Bank of Chile . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Galindo , A. and Schiantarelli , F. ( 2003 ). Credit Constraints and Investment in Latin America , Washington, DC : Inter‐American Development Bank . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Gourinchas , P.O. and Jeanne , O. ( 2006 ), ‘ The elusive gains from international financial integration ’, Review of Economic Studies , vol. 73 ( 3 ), pp. 715 – 41 . Google Scholar Crossref Search ADS WorldCat Gwartney , D. , Hall , C. and Lawson , R. ( 2010 ). Economic Freedom of the World: 2010 Annual Report , Vancouver : The Fraser Institute . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Harrison , A. and McMillan , M. ( 2003 ). ‘ Does direct foreign investment affect domestic credit constraints? ’, Journal of International Economics , vol. 61 ( 1 ), pp. 73 – 100 . Google Scholar Crossref Search ADS WorldCat Hellwig , C. and Lorenzoni , G. ( 2007 ). ‘ Bubbles and self‐enforcing debt ’, Econometrica , vol. 77 ( 4 ), pp. 1137 – 64 . OpenURL Placeholder Text WorldCat Héricourt , J. and Poncet , S. ( 2009 ). ‘ FDI and credit constraints: firm‐level evidence from China ’, Economic Systems , vol. 33 ( 1 ), pp. 1 – 21 . Google Scholar Crossref Search ADS WorldCat Jeske , K. ( 2006 ). ‘ Private international debt with risk of repudiation ’, Journal of Political Economy , vol. 114 ( 3 ), pp. 576 – 93 . Google Scholar Crossref Search ADS WorldCat Kaufmann , D. , Kraay , A. and Mastruzzi , M. ( 2010 ). ‘ The worldwide governance indicators methodology and analytical issues ’, Working Paper, World Bank Policy Research. Kehoe , T. and Levine , D. ( 1993 ). ‘ Debt‐constrained asset markets ’, Review of Economic Studies , vol. 60 ( 4 ), pp. 865 – 88 . Google Scholar Crossref Search ADS WorldCat Kehoe , P. and Perri , F. ( 2002 ). ‘ International business cycles with endogenous incomplete markets ’, Econometrica , vol. 70 ( 3 ), pp. 907 – 28 . Google Scholar Crossref Search ADS WorldCat Kim , Y.J. and Zhang , J. ( 2012 ). ‘ Decentralized borrowing and centralized default ’, Journal of International Economics , vol. 88 ( 1 ), pp. 121 – 33 . Google Scholar Crossref Search ADS WorldCat Klein , M. and Olivei , G. ( 2008 ). ‘ Capital account liberalization, financial depth, and economic growth ’, Journal of International Money and Finance , vol. 27 ( 6 ), pp. 861 – 75 . Google Scholar Crossref Search ADS WorldCat Kraay , A. , Loayze , N., Servén , L. and Ventura , J. ( 2005 ). ‘ Country portfolios ’, Journal of the European Economic Association , vol. 3 ( 4 ), pp. 915 – 45 . Google Scholar Crossref Search ADS WorldCat Krueger , D. and Perri , F. ( 2006 ). ‘ Does income inequality lead to consumption inequality? evidence and theory ’, Review of Economic Studies , vol. 73 ( 1 ), pp. 163 – 93 . Google Scholar Crossref Search ADS WorldCat La Porta , R. , López‐de‐Silanes , F., Shleifer , A. and Vishny , R. ( 1998 ). ‘ Law and finance ’, Journal of Political Economy , vol. 106 ( 6 ), pp. 1113 – 55 . Google Scholar Crossref Search ADS WorldCat La Porta , R. , López‐de‐Silanes , F., Shleifer , A. and Vishny , R. ( 1999 ). ‘ The quality of government ’, Journal of Law, Economics and Organization , vol. 15 ( 1 ), pp. 222 – 79 . Google Scholar Crossref Search ADS WorldCat Lane , P. ( 2004 ). ‘ Empirical perspectives on long‐term external debt ’, B.E. Journal of Macroeconomics , vol. 4 ( 1 ), pp. 1 – 23 . OpenURL Placeholder Text WorldCat Lane , P. and Milesi‐Ferretti , G.M. ( 2007 ). ‘ The external wealth of nations mark II: revised and extended estimates of foreign assets and liabilities, 1970–2004 ’, Journal of International Economics , vol. 73 ( 2 ), pp. 223 – 50 . Google Scholar Crossref Search ADS WorldCat Mauro , P. ( 1995 ). ‘ Corruption and growth ’, Quarterly Journal of Economics , vol. 110 ( 3 ), pp. 681 – 712 . Google Scholar Crossref Search ADS WorldCat Moody’s Investors Service ( 2011 ), Sovereign Default and Recovery Rates, 1983–2010 , New York : Moody’s Global Credit Research . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Panizza , U. , Sturzenegger , F. and Zettelmeyer , J. ( 2009 ). ‘ The economics and law of sovereign debt and default ’, Journal of Economic Literature , vol. 47 ( 3 ), pp. 1 – 47 . Google Scholar Crossref Search ADS WorldCat Qian , R. ( 2012 ). ‘ Why do some countries default more often than others? The role of institutions ’, Working Paper, World Bank Policy Research. Rajan , R. and Zingales , L. ( 1998 ). ‘ Financial dependence and growth ’, American Economic Review , vol. 88 ( 3 ), pp. 559 – 86 . OpenURL Placeholder Text WorldCat Rajan , R. and Zingales , L. ( 2003 ). ‘ The great reversals: the politics of financial development in the twentieth century ’, Journal of Financial Economics , vol. 69 ( 1 ), pp. 5 – 50 . Google Scholar Crossref Search ADS WorldCat Rand , J. ( 2007 ). ‘ Credit constraints and determinants of the cost of capital in Vietnamese manufacturing ’, Small Business Economics , vol. 29 ( 1 ), pp. 1 – 13 . Google Scholar Crossref Search ADS WorldCat Reinhart , C. and Rogoff , K. ( 2011 ). ‘ The forgotten history of domestic debt ’, Economic Journal , vol. 121 ( 552 ), pp. 319 – 50 . Google Scholar Crossref Search ADS WorldCat Roeder , P. ( 2001 ). ‘ Ethnolinguistic fractionalization (ELF) indices, 1961 and 1985 ’, available at: http://pages.ucsd.edu/~proeder/elf.htm (last accessed: 17 February 2016). Schich , S. and Lindh , S. ( 2012 ). ‘ Implicit guarantees for bank debt: what do we stand? ’, OECD Journal: Financial Market Trends , vol. 2012 ( 1 ), pp. 1 – 22 . Google Scholar Crossref Search ADS WorldCat Stulz , R. and Williamson , R. ( 2003 ). ‘ Culture, openness, and finance ’, Journal of Financial Economics , vol. 70 ( 3 ), pp. 313 – 49 . Google Scholar Crossref Search ADS WorldCat Uribe , M. ( 2006 ). ‘ On overborrowing ’, American Economic Review , vol. 96 ( 2 ), pp. 417 – 21 . Google Scholar Crossref Search ADS WorldCat Wright , M. ( 2006 ). ‘ Private capital flows, capital controls, and default risk ’, Journal of International Economics , vol. 69 ( 1 ), pp. 120 – 49 . Google Scholar Crossref Search ADS WorldCat Author notes " The authors are grateful to the editor Morten Ravn and two anonymous referees. We also thank Daron Acemoglu, John Beghin, Joydeep Bhattachaya, Yao Amber Li, Gray Calhoun, Gene Grossman, Kasten Jeske, Jiandong Ju, Harvey Lapan, John Schroeter, Rajesh Singh, Mark Wright, Jing Zhang, Oleksandr Zhylyevskyy, Zhiwei Xu and participants at various seminars for helpful comments and suggestions, as well as Guangyuan Guo for excellent research assistance. Haichao Fan acknowledges the financial support from the Shanghai Pujiang Program (15PJC041) and the fundamental research funds from Shanghai University of Finance and Economics (2013110715). Xiang Gao acknowledges the financial support from the Shanghai Pujiang Program (12PJC048) and the National Natural Science Foundation of China (71501117). © 2016 Royal Economic Society
TI - Domestic Creditor Rights and External Private Debt
JF - The Economic Journal
DO - 10.1111/ecoj.12380
DA - 2017-11-01
UR - https://www.deepdyve.com/lp/oxford-university-press/domestic-creditor-rights-and-external-private-debt-oA26BiX3Sf
SP - 2410
VL - 127
IS - 606
DP - DeepDyve
ER -