For this chapter's analysis, two dependent variables are constructed to measure the extent and likelihood of project finance deals using bank loan syndication for PPP projects in the sample of the seven Asian countries. The first variable, debt concentration, is measured as the average loan share provided by each bank in the loan transaction. This approach was used by Esty and Megginson (2003) to analyze debt concentration. When no information on each arranger's participation in a loan tranche is available, which applies to our case, an equal sharing of the loan amount is assumed.4 Table 5.1 shows the average debt concentration in the sample is $134.4 million, with a minimum debt concentration of $580,000 and a maximum of $2.1 billion, which indicates a high concentration of debt among project finance transactions.
Table 5.1: Summary Statistics of Key Variables
| Variable | Mean | Standard Deviation | Minimum | Maximum |
| Debt concentration | 134.40 | 238.50 | 0.58 | 2048.48 |
| Loan syndication dummy | 0.42 | 0.49 | 0.00 | 1.00 |
| Rule of law | -0.05 | 0.37 | -0.61 | 1.02 |
| S&P rating | 0.03 | 0.16 | 0.00 | 1.00 |
| Tier 1 ratio | 12.89 | 8.20 | 7.75 | 92.15 |
| Liquid-to-total assets | 11.41 | 6.33 | 2.94 | 53.94 |
| Bank regulation variable | 0.31 | 32.56 | -39.61 | 113.28 |
| Cost over income | 53.77 | 13.99 | 16.01 | 102.17 |
| Nonperforming loans | 2.94 | 1.57 | 0.01 | 7.28 |
| Loan maturity (EW) | 12.94 | 4.31 | 2.43 | 26.53 |
| Loan security | 0.34 | 0.47 | 0.00 | 1.00 |
| Tranche size | 160.17 | 266.79 | 0.29 | 2048.48 |
| Repo rate | 5.52 | 2.39 | 1.50 | 9.00 |
| Inflation rate | 5.35 | 3.48 | -0.90 | 18.68 |
| Credit default swap | 117.68 | 40.32 | 43.82 | 469.78 |
EW = equally weighted, S&P = Standard & Poor's.
Sources: Author's estimates, based on Thomson One Banker; Thomson Datastream; World Bank, Worldwide Governance Indicators and World Development Indicators; Bloomberg L.P.; and CEIC Data Company.
The second dependent variable, the loan syndication dummy, is a dummy variable equal to 1 if the loan is provided by more than one bank, and 0 otherwise. Forty-two percent of the projects in the sample are funded via loan syndication. All project deals in Malaysia and Viet Nam were funded by loan syndicates, but only 22% in India.
The main independent variable, legal risk, is proxied by the rule of law index from the World Bank's Worldwide Governance Indicators Database.5 The rule of law composite index ranges from -2.5 to 2.5, with higher values corresponding to better rule of law. It is important to note that this variable is an inverse scale in proxying for legal risk, which means legal risk decreases as the rule of law index increases. In the seven-country sample, the index average is -0.05, with a minimum score of -0.61 and a maximum score of 1.02, which denotes a higher legal risk.
The second independent variable, information asymmetry, is captured by the dummy variable equal to 1 if the project borrower has a Standard & Poor's senior unsecured debt rating, 0 otherwise. For Lee and Mullineaux (2001) and Godlewski (2008), the logic of using this variable is that higher quality information is available on firms with a credit rating. In the sample, only 3% of the project borrowers had a credit rating, indicating a high level of information asymmetry in the seven countries.
For bank-specific characteristics, the typical arranger in the sample shows a solid tier 1 level of 12.9%, which confirms that the banks in the sample are well-capitalized. Liquidity accounts for 11.4% of total assets, and the loan portfolio shows a low degree of riskiness (2.9%). The nonperforming-loan variable is used to proxy both asset quality and a bank's supervisory mechanism.6
Bank efficiency in terms of the cost-to-income ratio is about 54%. The ratio is also interacted with the regulatory quality index of the Worldwide Governance Indicators to capture banking regulation.7 While this approach is slightly different from the one used by Godlewski (2008), the interaction of bank efficiency and the regulatory environment should provide a good proxy for banking regulation, as it directly captures the efficiency aspect of the banking structure and the quality of regulations in promoting private sector development, including banks.8
Loan characteristics are also used in the empirical analysis to test whether debt-related variables affect the extent and propensity of banks to participate in loan syndicates. Dennis and Mullineaux (2000) note that "certain characteristics of the loan itself may affect the agent bank's capacity to syndicate either because the characteristic serves to attenuate agency costs or because it influences the perceived value to the buyer for non-agency- related reasons." Some of these characteristics include loan maturity and loan collateral, which were discussed in the literature review. In the seven-country sample, the average maturity of a PPP project finance deal is 13 years, spanning 2.4 years-26.5 years. The loan maturity variable is also used to proxy for recontracting adjustments to PPP infrastructure projects.
Loan security is a dummy variable equal to 1 if the loan is collateralized, 0 otherwise. Thirty-four percent of the sample are projects secured through collateral. The loan's tranche size is also used as an independent variable, because loan size affects a bank's decision to syndicate a loan. The average tranche size in the sample is $160 million, with a minimum of $300,000 and a maximum of $2 billion.
The dataset is further complemented by country variables that proxy for macroeconomic and financial conditions. The inflation rate is used to proxy for macroeconomic stability, the repo rate for monetary policy, and the credit default swap spread for a country's credit risk profile. The average inflation rate in the sample is 5.4%, and the average repo rate is 5.5%. The average credit default swap spread is 117.7.