DISCUSSION

• Date: 01 May 1971
• DOI: http://doi.org/10.1111/j.1540-6261.1971.tb00913.x
• Author: Stewart C. Myers
• Published date: 01 May 1971

DISCUSSION

STEWART C. MYERS*: I profited by reading Elton

and Gruber's paper. I had not

thought

very deeply about the

"debt management"

problem-i.e., the timing of

calls

and

the

timing and maturity of

debt issues. Nor had I paid

much attention to the

deci-

sion

rules proposed in the

literature. As Elton and Gruber point out, many of

these

are

clearly shaky, or at best

rough rules of thumb.

The authors have laid out the

sequential nature of the

debt management

problem

very

well

and have made good

progress towards a more

accurate formal analysis

of it.

Of

course, we are faced with

the usual difficulties in

immediate practical

application

of

dynamic programming. In the

case of debt

management, the essence of the

problem

can

rarely be captured except by

a stochastic model. Going

to this degree of

sophistica-

tion

at

the present state of the art

is very costly in terms of

required data and

computa-

tion. But

things will get simpler as further work is done

and we understand the

problem

better.

Now

to

criticism.

Although

I learned a good deal from the

paper,

I still

feel

only

partly

educated. There are two

unresolved general problems

that should be pointed

out

-not to

find fault with the paper,

but to put it in better

perspective. The first

problem

concerns

objectives and

assumptions. The second is to

decide whether a dynamic

pro-

gramming approach is worth the

trouble. Although lack of

space limits

me

to debt

management, the same two types

of problems will have to

be faced

in

other

potential

applications of

dynamic

programming

in

finance.

Elton and Gruber's assumed

objective is to minimize

the

present

value

of

interest

costs net

of "maneuvering

costs"-that is, net of

call

premiums

and

transaction

costs.

The interest

cost will

depend

on

the

degree

of

call

protection

given

when

new

bonds

are issued.

The

problem

is

to

determine, first,

when to

refinance; second,

the

maturity

of the

issue, and, third,

the

degree

of

call

protection.

Elton and Gruber

do not

stress

the

third

decision,

but

it is

important,

and

their models

can

be

adapted

to

consider

it.

If

the

goal

is to reduce

interest

cost,

then the

timing

of

debt issue and

refinancing is

essentially

a game played against

the market. Elton and Gruber have

described

the

right

way

to

play

the

game given

the

rules

they

have assumed. But

the

other

side

is

playing

too.

Moreover,

it

appears

to be a zero

sum

game;

one side

gains only

if

the

other loses.

One

would

think that there

is little

point

in

management's playing

the

game

unless

it

can

predict

the future course of interest rates better than bond

investors

can.

What

does it mean for

management

"not to

play

the

game?"

It does not mean

that

a firm

should

do

nothing

if it finds that

yields

have

dropped

enough to

make the call

option

valuable

on an

existing

issue. This is an

opportunity

that should not

be

thrown

away.

However,

if a new issue is contemplated,

and if

management

has no special

knowledge

about

future interest

rates,

then

there

would

seem to be little

point

in

"playing

the

game"

of

adjusting

call

protection

on

the new

bond or

of

calculating

when

it

in turn is

to

be

called.

If this is

an

accurate

picture,

the

problem

is simpler

than Elton

and Gruber

imply,

although

a dynamic

programming problem

still arises when call of an existing

bond

becomes

profitable.

But

multiple

refunding

does not have to be

brought

in to the call

decision.

This

line of

thinking

leads into a still more basic issue. If

managers

in

general

have

* Sloan

School

of

Management,

Massachusetts

Institute of

Technology.

538