Content area
Full text
We use data envelopment analysis (DEA) to identify the large-cap mutual funds in the Morningstar 500 [1999] that are efficient or inefficient. We also want to identify the financial variables that differ significantly between efficient and inefficient funds, and determine the nature of these relationships.
That is, which variables differ significantly between efficient and inefficient funds, given the Sharpe index output variable measure of risk-return performance? And what is the nature of these relationships for efficient and inefficient funds?
RESEARCH METHODOLOGY
Data envelopment analysis has traditionally been used to analyze the relative efficiency of public organizations, such as schools, hospitals, prisons, and military operations. More recently, it has been applied to banks (Haslem, Scheraga, and Bedingfield [1999]) and mutual funds (see Basso and Funari [2001], McMullen and Strong [1998], and Murthi, Choi, and Desari [1997]).
DEA was originally developed for use in service organizations, where the form of the production function is unknown or perhaps not even considered. It has the advantage of being a flexible, non-parametric, technique that makes no assumptions about the form of the production function. Instead, it estimates an empirical best practice production frontier from the observed inputs/outputs of individual decision-making units (DMUs), which replicates individual DMU behavior rather than the average sample estimate of conventional production functions. A DMU is efficient when comparisons with other units indicate no inefficiency in the use of inputs or outputs, as measured by its position relative to the efficient frontier.
The DEA best practice frontier is generally piecewise-linear and approximates the true production function. DEA gets its name because the data from the best practice DMUs generate the production function, and thereby "envelop" the data from other DMUs. The term DEA has been constantly broadened, as additional models for enhancing its advantages in measuring input/output efficiency have been developed.
We use the IDEAS [2000] system to analyze mutual fund efficiency. The model allows for both variable and constant returns to scale. We also use an input-oriented determination of output slack or excess inputs with a wmls invariant measurement of efficiency and inefficiency (i.e., efficiency scores are independent of units of measurement). Our choice of model specifications has been guided by the managerial, institutional, and environmental characteristics of the sample mutual funds and their portfolios.