Explaining Interest Rates in Local Government Borrowing

ABSTRACT

This article analyzes the usefulness of municipal accounting information in the interest rates agreed on by financial institutions in the European context. Most of the literature has focused on municipal bonds, mainly in the United States. Our core contribution is to study which factors determine the interest rates on bank loans to municipalities. This framework is different, since there is neither a secondary bond market nor credit ratings, and therefore information asymmetry is higher. Specifically, we investigate if municipal financial indicators and financial reports’ quality have an influence on interest rates paid by Spanish municipalities in the period 2001–2008. Our empirical results indicate that, in general, the municipal financial situation exerts an influence on the credit policy of lenders, who charge a risk premium to those municipalities with less current surplus, which is an indicator of the municipality's saving ability. Furthermore, there is also some inertia in the municipal annual interest rate (year t − 1 interest determines around 27% of year t interest). Other findings indicate that high levels of expenditures and debt per capita increase the interest cost of municipal borrowing. Finally, municipalities that disclose full detailed financial reports pay less interest because information asymmetry and transaction costs are lower.

Notes

^a Some specific municipalities paid an annual interest rate below 12-month euribor. However, Figure 2 shows that, on average, this difference is positive in each and every year. At the Spanish central level, this negative spread existed in all the period 2002–2008 (comparing average annual interest rate of 12-month sovereign national bonds with average 12-month euribor). Therefore, the interest paid by municipalities is higher than the interest paid by the central government in Spain. In our view, the reason for this is two-fold: (1) Although art. 49.3 of the Spanish Municipal Finances Act grants municipal borrowing the same conditions and benefits than national sovereign debt, it is reasonable to think that lenders deem municipalities as a riskier investment, for example, because of payment delays. (2) Spanish municipal borrowing takes the form of bank loans instead of a security market with credit ratings; therefore, lenders require a higher interest rate as a compensation for the transaction costs.

^b The term “debt” refers to total interest-bearing debt; i.e., both short-term and long-term debt. There is no information available on the specific level of short- or long-term debt. Accordingly, it does not include accounts payable.

Notes: All estimations have been carried out using the two-step GMM estimator. All variables are treated as endogenous and the lagged independent variables are used as instrument. Z statistic in brackets. m₂ is test statistic for second-order autocorrelations in residuals, distributed as standard normal N (0,1) under the null hypothesis of no serial correlation. Hansen test is a test of overidentifying restrictions, distributed as chi-square under the null of instrument validity. Degrees of freedom in brackets.

^a According to Feroz and Wilson (1992), the impact of the financial situation and the reporting quality on the interest rate depend on the size of the municipality. Thus, we have split the sample into two groups to account for the impact of size. The criterion has been ≤20.000/ > 20.000 inhabitants (as of December 31, 2008). This population, for example, is the median of municipalities issuing debt in Simonsen, Robbins, and Helgerson (2001).

^b We have rerun the regressions replacing the variable revgrowth with another variable representing revenues per capita. Results don't change, which confirms the robustness of our model.

^c Unreported regressions with population in natural logarithm confirm the same results for coefficients, which again reinforces the robustness of the model.

^***significant at 1%;.

^**significant at 5%;.

^*significant at 10% level.

Notes: All estimations have been carried out using the two-step GMM estimator. All variables are treated as endogenous and the lagged independent variables are used as instrument. Z statistic in brackets. m₂ is test statistic for second-order autocorrelations in residuals, distributed as standard normal N(0,1) under the null hypothesis of no serial correlation. Hansen test is a test of overidentifying restrictions, distributed as chi-square under the null of instrument validity. Degrees of freedom in brackets.

^a We have split the sample in two groups to assess separately the impact of size. The criterion has been ≤20.000/>20.000 inhabitants (as of 31.12.2008).

^b We have rerun the regressions replacing the variable revgrowth with another variable representing revenues per capita. Results don't change, which confirms the robustness of our model.

^c Unreported regressions with population in natural logarithm confirm the same results for coefficients, which again reinforces the robustness of the model.

^***significant at 1%;.

^**significant at 5%;.

^*significant at 10% level.

aSome specific municipalities paid an annual interest rate below 12-month euribor. However, Figure 2 shows that, on average, this difference is positive in each and every year. At the Spanish central level, this negative spread existed in all the period 2002–2008 (comparing average annual interest rate of 12-month sovereign national bonds with average 12-month euribor). Therefore, the interest paid by municipalities is higher than the interest paid by the central government in Spain. In our view, the reason for this is twofold: (1) Although art. 49.3 of the Spanish Municipal Finances Act grants municipal borrowing the same conditions and benefits than national sovereign debt, it is reasonable to think that lenders deem municipalities as a riskier investment, for example, because of payment delays. (2) Spanish municipal borrowing takes the form of bank loans instead of a security market with credit ratings, therefore lenders require a higher interest rate as a compensation for the transaction costs.

^b The term “debt” refers to total interest-bearing debt; i.e., both short-term and long-term debt. There is not information available on the specific level of short- or long-term debt. Accordingly, it does not include accounts payable.

Notes: All estimations have been carried out using the two-step GMM estimator. All variables are treated as endogenous. Two sets of instruments work in the model. On the one hand, the lagged independent variables. On the other hand, pop<5, pop>65 and meanage (Capeci 1991). Z statistic in brackets. m₂ is test statistic for second-order autocorrelations in residuals, distributed as standard normal N(0,1) under the null hypothesis of no serial correlation. Hansen test is a test of overidentifying restrictions, distributed as chi-square under the null of instrument validity. Degrees of freedom in brackets.

^a According to Feroz and Wilson (1992), the impact of the financial situation and the reporting quality on the interest rate depend on the size of the municipality. Thus, we have split the sample into two groups to account for the impact of size. The criterion has been≤20.000/>20.000 inhabitants (as of Dec. 31, 2008). This population, for example, is the median of municipalities issuing debt in Simonsen, Robbins, and Helgerson (2001).

^b Unreported regressions with population in natural logarithm confirm the same results for coefficients.

^***significant at 1%;.

^**significant at 5%;.

^*significant at 10% level.

Notes: All estimations have been carried out using the two-step GMM estimator. All variables are treated as endogenous. Two sets of instruments work in the model. On the one hand, the lagged independent variables. On the other hand, pop <5, pop >65 and meanage (Capeci 1991). Z statistic in brackets. m₂ is test statistic for second-order autocorrelations in residuals, distributed as standard normal N(0,1) under the null hypothesis of no serial correlation. Hansen test is a test of overidentifying restrictions, distributed as chi-square under the null of instrument validity. Degrees of freedom in brackets.

^a We have split the sample into two groups to assess separately the impact of size. The criterion has been ≤20.000/ > 20.000 inhabitants (as of 31.12.2008).

^b Unreported regressions with population in natural logarithm show the same results.

^***significant at 1%;.

^**significant at 5%;.

^*significant at 10% level.

1. Even in these two cities, the percentage of securities out of the total financial debt is less than 50%: 19.77% and 47.47% for Madrid and Barcelona, respectively, as of 31 December 2009.

2. Wall Street Journal, February 26, 2010, or BBC news: http://news.bbc.co.uk/2/hi/ 7584097.stm.

3. Wall Street Journal, April 18, 2011.

4. Year 2001 drops due to interest rate calculation. Variables interest and grossav are averages from our sample. Variables newhouses and spread are actual numbers published by the Spanish National Statistics Institute and the Central Bank of Spain, respectively.

5. We have eliminated errors or misleading data by rejecting cases with annual interest rates below 0% and above 50%. To check the reliability of our refined sample, we have also run non-reported regressions with 0%–10% interest thresholds, and results remain unchanged.

6. Other studies with a similar approach are Beaver (1966), Wallace (1981), Copeland and Ingram (1982), Ingram and Copeland (1982), Wilson and Howard (1984), Engen and Hubbard (2004), Johnson and Kriz (2005), Christiaens et al. (2010), and Lassen (2010).

7. The highest correlation between explanatory variables is .42. This figure is far below the .76 correlation reported by Leonard (1983) to confirm multicollinearity between two variables in his empirical analysis.

8. In this regard, see Hastie (1972), Wallace (1981), Leonard (1983), Wilson and Howard (1984), Apostolou, Reeve, and Giroux (1984), Rivers and Yates (1997), Simonsen, Robbins, and Helgerson (2001), Lewis (2003), Gore (2004), and Butler, Fauver, and Mortal (2009).

9. Basically, the law requires less than 3.000.000? of budgeted revenues for the LG to report in the simplified way.

Additional information

Notes on contributors

FRANCISCO BASTIDA

Francisco Bastida ([email protected]) is Professor of Accounting and Finance at the University of Murcia in Spain. His current research focuses on the impact of transparency in government financial reporting from an international perspective, and public administration financial analysis.

MARíA-DOLORES GUILLAMÓN

María-Dolores Guillamón ([email protected]) is Assistant Professor of Accounting and Finance at the University of Murcia in Spain. Her research focuses on the analysis of the public sector financial activity, both national and local.

BERNARDINO BENITO

Bernardino Benito([email protected]) is Professor of Accounting and Finance at the University of Murcia in Spain. He regularly serves as a consultant for public administrations. He has more than 70 publications in different national and international journals and is co-author of different books published in international circles. He is an editorial board member of Spanish Accounting Review, International Journal of Economics and Accounting, Journal of Business Administration Research, Economics and Business Letters, and Lex Localis.