A Model of the Trust Investment Process

Geoffrey P. E. Clarkson

Summerized at 99.5.17

by Kim Kyoung-Soo

Contents

• Introduction

• Postulates and Data for the Model

• Selection of the Current List of Stocks-The “A” List

• The Formulation of an Investment Policy

• The Selection of a Portfolio

• Testing the Model

Introduction

• Trust funds held by banks in the United States currently amount to nearly $60 billions

• The purpose of this model: to simulate the process employed in the investment of trust fund in common stocks.

Introduction(Cont’d)

• An investor is processing information

• The decision-making in investment process is under uncertainty

• The basis of the analysis– the operations at a medium-sized national bank– the decision maker is the trust investment

officer

Postulates and Data for the Model• The main postulates for the analysis

– A memory

– Search and selection procedures

– A set of rules or criteria

• Interviewing departmental officers and observing behavior at committee meetings

• “Protocols” recorded the trust officer’s decision processes for accounts that arose in the course of his work

Three parts of the Trust Investment Process• The selection of a list of stocks suitable for

current investment - the “A” List

• The formulation of an investment policy

• The selection of a portfolio

Structure of Decision Process

Selection of the Current List of Stocks-The “A” List• 1A. Processing the Raw Information

• 1B. Evaluation the Data

• 1C. The Interrelations

Processing the Raw Information

• The same processes for three main categories(general economy, industry, company)

• 1A1. Determination of attributes and their values

• 1A2. The Formation of expectations

Determination of Attributes and Their Values• Current Value

– The historical values of each attribute in the system for the last ten years are entered in the form of lists

• For each attribute, the model generate 3 additional lists– Ten-year average– Recent changes– Average rate of change

The Formation of Expectations

• Economy and industry expectation lists– Nonnumerical data - values converted into ”abov

e”, “below”, “equal to”• the forecasted value

• the first value on the Recent Change List

• Company expectation lists– contain 1 or 2 entries in numerical form– obtained by converting the 12 month or 3- to 5-ye

ar forecast into an expected rate of change

Evaluating the Data

• The model evaluates the data by creating two main lists– 1B1. The Relative Performance List– 1B2. The Relative Value List

• cf. List of info. for each attr. of each co.– A list of the last 10 values of the attribute– The mean of these tem values– A list of the rates of change of these values– The mean of these rates of change– an Expectation List contains the expected rate of change for th

e coming year/the next 3 to 5 years

The Relative Performance List

(1) A list made for each of the basic lists for each attribute of the mean for that attribute for each company

(2) The vector of the means of the distribution of means for each attribute

(3) A percentage deviation of each mean from the distribution mean (2)

(4) Converting (3) into 3-valued scale

The Relative Performance List(cont’d)

• aij represent the class of all company attribute mea

ns where– i : the number of attribute for each company

– j : the number of companies for each industry

The Relative Performance List(cont’d)• The Mean of the Distribution of Means for

attribute i

• The list of all such means forms the vector

The Relative Performance List - The Percentage Deviations• The array of percentage deviations of each a

ij from its respective mean ãi

The Relative Value List

• The first entry: the difference between the mean price earnings ratio and the price earnings ratio of the forecasted earnings

• The second entry: the difference between the historical mean and the current price earnings ratio

The Relative Value List(cont’d)

– P = current market price– E = expected earnings per share for the current year– E* = forecasted earnings per share 3 to 5 years from now– P/E = ten year average of price earnings ratio

• The results are recorded as 3-valued scale.

The Interrelations

• Selecting the “A” List– select the “A” list under simplified conditions

• Relaxing the Conditions-A, B– examine the complexities by relaxing the condit

ions– assign names to the values which appear on the

Relative Value List.

Selecting the “A” List

• The scanner– examines each Economy and Industry Expectati

on List that have received new information

• The Selector – searches through the Relative Value Lists of all

companies and placing on the “A” list those companies whose Relative Values are

• (x) = “above” or “equal to”• (y) = “above”, “equal to”, or “below”

Relaxing the Conditions-A

• The effects of adding new information

• after a change the information is processed as 1A and 1B, and new values are placed on the Relative Value List

• The Scanner proceeds and initiates the selection procedure.

• The Selector examins the “A” List first, and removes any companies (x) = “below”

Relaxing the Conditions-A(cont’d)• The Selector then proceeds to the remaining list of c

ompanies and places on the “A” list those companies whose Relative Values are– (x) = “above” or “equal to”

• When more than one category receives new information,– the Scanner always go to most general category first, e.g.

economy or industry, and then proceeds down.– At this point changes in the Economy and Industry Expec

tation Lists are translated into “hold” or “delete hold”, and companies on the “A” list are not taken off.

Relaxing the Conditions-B

• 3 categories of changes in the Economy and Industry Expectation Lists– Forecated Value Falls below Recent Change Va

lue– Recent Change Value Falls below Forecasted V

alue– Forecasted Values and Recent Change Values b

oth Turn Down

The Formulation of an Investment Policy• Trust investors formulate investment policie

s for two types of funds– large trust funds e.g. Common Trust Funds– individual trust accounts

cf. The model does not consider the problem of investing the funds of Common Trust Funds

Individual Trust Accounts

• The Scanner– tests for the value of the list of information on th

e client, and converts it into a 2-valued scale-”Low” or “Not Low”(correspond to being below or not below a median value)

• The Selector– take the pattern of answers generated by the Scan

ner and convert it into the appropriate investment policy

Selector for Investment Policy

The Selection of a Portfolio

• A process of mapping the set of industries and companies on the “A” List onto the investment policy– Selection of industries appropriate to the invest

ment policy– Selection of companies– Diversification– Size of participation

Selection of Industries Appropriate to the Investment Policy• The lists were derived by direct questioning

and investment officer’s behavior.

• The model takes these lists as given and by searching through the “A” List derives, for each goal, a list of those industries and companies that are on the “A” List.

Selection of Companies

• Eliminate all companies that do not meet the two tests

• The Scanner orders the companies in each industry on the basis of the dominant attribute of the investment policy

• The Selector takes the first company from the industry that is at the head of the list and applies a set of tests-Growth-portfolio discrimination net to it

Growth-portfolio discrimination net

Diversification

• All accounts participate in at least five industries

• Participation in stocks be limited, in general to one per industry

Size of Participation

• The “Share Selector”– determines the number of shares to be purchased

of each company– the rule of process

• The total funds to be invested in common stocks are divided by the number of participations desired.

• To determine the number of shares to be purchased, the average number of dollars to be invested in each company is divided by the price of the particular company’s stock.

Testing the Model

• The results of these tests indicate that the trust investment program selected a greater proportion of correct securites that did any one of the alternative models.