Objectives and Methodology
This study had two major objectives:
(1) to develop a system to evaluate (from a financial viewpoint) the internal processes involved in the management of the share-market portfolios of managed investment funds such as mutual funds and trusts, superannuation funds, life funds of life insurance offices, and other funds whose assets are managed by professional money managers; and
(2) to apply the system in evaluating the internal research and management activities within a major Australian institutional investment fund.
Very little appears to have been written in the academic or professional literatures about internal performance evaluation of managed investment funds, or evaluation of the internal investment activities which make up the total performance of any fund. The methodology employed here involved identifying the activities within the fund
management structure at which securities were subject to implicit or explicit decisions, and also identifying the times at which various decisions were made at each of those activities. The internal decision processes were then evaluated by comparing actual with expected returns on securities as they were "passed" or "not passed" from one activity to the next within the fund management structure.
A unique feature of the study was the data used. Data comprising the full population of the internal research, decisions, and activities concerning the share-market portfolio were collected from a major Australian institutional investor. Institution X, for the sample period. Data on investment research also were collected from a second major institutional investor. The research reports prepared within these institutions were at no time made public, but were prepared for the exclusive use of the institutions themselves. A large
sample of share brokers' earnings forecasts also were collected for comparison with those of the institutional analysts. Preliminary analysis showed that the earnings forecasts of each group of analysts were more accurate than forecasts produced by naive mechanical models.
A matched portfolio variation of the residual analysis technique pioneered by Fama, Fisher, Jensen and Roll was used to investigate actual versus expected returns on securities as they passed through the Identified decision points within the fund management structure. Residual analysis using the familiar market model procedure was rejected for use here because of problems of bias. Problems in assessing the significance of observed differences between actual and expected returns, or excess returns, also were addressed.
The results suggest that, during the. period studied here, certain categories of the security
analysts' research reports produced within Institution X contained private information, that is, information not incorporated in share-market prices prior to or at the time the reports were prepared. Revisions of earnings forecasts made after a company visit showed relatively large potential excess returns over the subsequent 12 months, and these excess returns were of the correct sign and of the same order of magnitude for both positive and negative forecast revisions. Recommendations to act in the analysts' research reports also showed potential excess returns over subsequent periods, especially negative action recommendations.
Total purchases made by Institution X showed slightly negative excess returns over the 12 months subsequent to purchase. Further analysis identified categories of purchased securities which detracted from the performance of the total purchases. Also, time lags within the internal management structure which detracted from the
performance of the fund's purchase program were identified and evaluated. One major factor associated with these time lags appeared to be the purchasing authorization process as followed within Institution X.
Very few sales of shares were made by Institution X during the sample period. Analysis of the analysts' research reports had indicated certain categories of securities which showed negative excess returns subsequent to the research reports. By acting on the implicit or explicit sell signals for those categories of securities. Institution X would have avoided the associated negative excess returns during the period studied here.