Efficiency in the eurobond market: application of nonparametric techniques

Abstract

The aim of this article is to analyse the efficiency of eurobond issuers within the primary market, from 1995 to 2000. The study includes a reference to theoretical discussion and to the methodology used; detailed explanation of the variables considered which, in order to supplement those strictly financial, include others such as spread from Interest Rate Risk (IRR) and respective swap; rating, duration and size; and macroeconomic fundamentals of the issuer country. Results and conclusions obtained from the static and dynamic efficiency analyses are then illustrated and discussed.

Notes

¹ See Charnes et al . (1978).

² The article analysed efficiency with CRS and VRS to check whether there was an actual difference in the scores. As the difference was significant it was decided that the DEA Model with VRS would be applied.

³ i.e. the market yield or the swap for the issue currency, with the same duration and the same issue week.

⁴ The exponential function for rating quantification is f(x) = 100.125 − 100 e^{− x /2.84} where x = 18, 17, … , 0 depending on Aaa, Aa1, … , c category.

⁵ Duration, according to Macaulay's pioneer study (1938), is calculated as follows:

where: tj :	=	Maturity of the jth payment generated by the security.
C j :	=	Amount of the jth payment generated by the security.
R:	=	IRR of the equity corresponding to the time unit used to measure maturities.
Wj :	=	Estimation corresponding to the jth capital of the payment flow (where the sum of all wj 's equals one).
P:	=	Price of the security.
D:	=	Duration of the security.

⁶ Duration is equal to the repayment term only when the coupon is zero or when it has only one periodic voucher payment outstanding.

⁷ The Euromoney classification system is based on a combination of objective data and subjective information, the latter obtained through a selection of specialists. This system addresses the problem caused by the diversity of countries analysed, as well as the differences in quality and quantity of the information available in them.

⁸ Because of the size of the sample the efficiency levels of each issue have been omitted. The average data for each year is used instead to indicate the efficiency level the sector is at.

⁹ Although the euro has only been present at the 1999 and 2000 issues, it has eight totally efficient issues.

¹⁰ The global Model does not include year 2000 due to unavailability of macroeconomic data for that year.

.

¹² Fisher's index or Tornquist's index.

¹³ Vrts are variable returns to scale.

¹⁴ See Färe et al. (2001).

¹⁵ We define a eurobond portfolio as the set of issues offered by one issuer during a certain year.

¹⁶ A panel sample must be used in order to apply the Malmquist index.

¹⁷ The information is presented in the form of indexed numbers, so that the corresponding average annual increase rates result from subtracting one from the figures of the table.