MIT LIBRARIES
3 9080 02239 0253
hBNEi
HD28
.M414
i
SB
ALFRED
P.
WORKING PAPER
SLOAN SCHOOL OF MANAGEMENT
SOVEREIGN DEBT BUYBACKS CAN
BARGAINING COSTS
Julio
J.
Rotemberg
Working Paper #2082-88
October 1988
MASSACHUSETTS
INSTITUTE OF TECHNOLOGY
50 MEMORIAL DRIVE
CAMBRIDGE, MASSACHUSETTS 02139
SOVEREIGN DEBT BUYBACKS CAN
BARGAINING COSTS
Julio J. Rotemberg
Working Paper #2082-88
October 1988
Sovereign Debt Buybacks Can
*
Lower Bargaining Costs
Julio J.
Rotemberg
Sloan School of Management,
Massachusetts Institute of Technology
and
NBER
current version: October 24, 1988
Abstract
I
develop two models
in
which debt repurchases by highly indebted sovereign
nations are advantageous for
when
all
parties.
The models
are based on the idea that
sovereign debts are large, bargaining costs are large. Creditors spend
resources convincing the debtor that they are tough
stake. Also, the sanctions
when they have more
which are sometimes triggered when bargaining
produce an agreement are larger when debts are
more
larger.
at
fails to
For both these reasons
buybacks, which reduce the face value of the outstanding debt, can be beneficial.
The
resulting equilibria are constrained Pareto
dize buybacks increase overall welfare
I
also argue that
consistent with
*I wish to
Bulow and Rogoff
my
Optima. But, donors who subsi-
more than donors who make
direct gifts.
(1988) 's empirical evidence on buybacks
is
models.
thank David Scharfstein and Ken Froot for helpful conversations and the National Science and Sloan Foundations
for research support.
MAssACHusEffrMmDrT
OF TECHNOLOGY
LIBRARIES
Sovereign Debt Buybacks Can
Lower Bargaining Costs
Julio J.
Rotemberg
Sloan School of Management,
Massachusetts Institute of Technology
and
NBER
Highly indebted countries and their creditor banks have experimented with numerous
approaches to ameliorate the
the country repurchase
its
below par, countries pay
"LDC Debt
debt at
less
its
One
Crisis".
rather popular approach
is
to have
current market price. Since the debt currently trades
than they would
if
repurchase, or "buyback", recognizes that the debts are very unlikely to be paid in
Bulow and Rogoff
Several authors (Huizinga and Sachs (1987),
Krugman
The
they actually honored their debts.
full.
(1988), Froot (1988),
(1988)) have noted that other schemes are closely related to buybacks.
For
example, "debt/equity swaps" allow investors to purchase local currency with foreign debt
that they obtain in the secondary market. This
their local currency in
hard currency and use the hard currency to buy back the debt.
We would obviously
An
like to
know why
these schemes arise and
additional reason for studying these schemes
effect.
equivalent to letting investors pay for
is
is
that
These data can help disentangle the empirical
of the debt
crisis.
In a thought provoking paper,
who
benefits
from them.
we now have some data on
validity of various theoretical
Bulow and Rogoff
(1988)
their
models
show that the
March 1988 debt repurchase
of Bolivia raised the secondary market's price of Bolivia's
debt substantially.
much
It
rose so
that the secondary market's valuation of outstanding
debt was almost unchanged by the buyback.
Bulow and Rogoff (1988) 's explanation
Why
is
then did Bolivian
officials
do this?
that Bolivian officials were misguided or
unpatriotic. Stated differently, there exist models of rational optimizing governments which
are inconsistent with the facts.
To make
this point
with some generality, consider situations
where buybacks
affect only the flows of
real activity in either.
Ceish flows, these
The sum
payments between countries.^ They do not
borrowers and lenders apply the same discoimt factor to future
If
purely redistributive (or financial) buybacks cannot benefit
of the welfare effects on
lenders welfare improvement
is
all
parties
sums
rates
offset
presumably exceed those of
would imply that the
by borrower
to zero. So, these
matched by a reduction
Borrower countries are probably constrained
their discount rates
affect
lenders' benefit
in
in
the
all
concerned.
models imply that the
borrower welfare.
amount they can borrow. Thus,
Such a difference
their lenders.
from purely financial buybacks
is
in
discount
more than
This occurs because buybacks increase current payments from
losses.
the borrower while, at best, reducing future payments.
Bulow and Rogoff (1988) argue
that, in the case of sovereign debts,
only the lenders to benefit from purely financial buybacks.
whose repayment depends only on the
outstanding debt have a very small
By
Sovereign debts are those
debtor with reprisals and not
Therefore, buyback induced reductions in the face value of the
on borrower net worth.
deal for borrowers.
ability to threaten the
we should expect
effect
on future repayments. This makes them a bad
contrast, in the case of corporate debts the creditor can dissolve
the debtor and keep the net worth. Thus, buybacks are financed with cash which really
belongs to creditors. In this case, creditors lose from repurchases.
Both of these
are negotiated
and
stories
if,
assume that buybacks take place
as in
Bulow and Rogoff
in
the open market.
(1986), negotiation
is
If
buybacks
costless, purely financial
buybacks change nobody's welfare. The evidence that buybacks make lenders better
is
thus inconsistent with
Bulow and Rogoff
(1986).
they do, that countries are rational pursuers of their
It is
now
off
unrealistic to maintain, as
self interest
and that debts have only
redistributive effects.
For buybacks to improve everybody's welfare, they must have real consequences.
Acharya and Diwan (1988), Froot (1988), Helpman (1987), Krugman (1988) and Sachs
(1988) have formally studied one particular real effect of buybacks. Their effect relies on
'Dooley (1988) provides an example of such a model.
^See Eaton, Gersoviti and StigliU (1986) and the literature cited therein.
moral hazard.^ They argue that the
fall in
raises the country's incentive to invest.
This occurs because the country gets to keep a
larger fraction of the investment's return.
very large rates of return but there
it
The investments
available to the debtor have
no mechanism, other than debt reduction, to induce
Once they
the country to undertake them.
make
is
nominal debt outstanding induced by buybacks
possible for the country to pay
are undertaken however, their high returns
more per unit
of outstanding debt.
Therefore,
reductions in the face value of debt tend to increase payments on the debt that remains
outstanding. Lenders get more money, borrowers get some of the surplus from the valuable
investments and everybody benefits.
In this paper
I
model a second
real effect of
buybacks. As Sachs and Huizinga (1986)
argue informally, debt repurchases can lower bargaining costs. They do so in
because smaller debts lead to
are
many ways
information.
to
some
in
With
less
which bargaining
is
wasteful.
They stem from the imperfection
of
perfect information (as in Rubinstein (1982)) bargaining simply leads
division of the total pie.
would only be capital gains and
contrast,
models
wasteful bargaining between country and banks. There
There are no resources spent on the dividing
such bargaining were in fact typical, one would not expect protracted debt
By
my
we
itself.
crisis.
If
There
losses as, over time, bargaining positions change.
see very valuable resources spent as the countries
and the banks seek
agreements. First, individuals with important responsibilities such as running countries or
banks devote much time to
LDC
debt negotiations. Second, individuals engage in actions
which are costly to the coalition of country and banks as a whole to demonstrate their
bargaining power.
Third, the failure to reach agreement as the participants act tough
sometimes leads to costly sanctions.
These bargaining costs are closely related to those explored
(1983), Sobel
in
Fudenberg and Tirole
and Takahashi (1983) and Admati and Perry (1987). This theoretical
lit-
erature has principally focused on model with alternating offers where temporary failure
to reach agreement
is
costly because the future
is
discounted. Delay to reach agreement
^See Atkeson (1988) for an infinite horiron rather tan a finite horiion model of sovereign debt with moral hazard. Atkeis conducted at such a level of generality that ideas such as "buybacks" have no simple analytical counterpart.
Nonetheless, buybacks could well be going on behind the payment flows predicted by his model.
son's analysis
emerges
in equilibrium
and constitutes the only cost of bargaining.^ In practice, bargain-
employ tools other than delay to demonstrate
ing parties also
the cost of failing to reach agreement at any given time
is
their resolve.
Moreover,
more severe than the
often
simple postponement of agreement. Failure to reach agreement often triggers aggressive
added realism
retaliation. For
The
thus focus on these other costs of bargaining.
I
empirical relevance of these bargaining costs
the trade of Brazil was severely disrupted during
hard to demonstrate how expensive
certainly cost Brazil
and
its
1987 debt moratorium. Banks tried
its
debt increases. This
I
many
The
are splitting a smaller pie. Chest-beating loses
situations
where bargaining
Such negotiation and posturing (which
the level of debt
The
form of sanction
is
is
appeal.
I
have
in
mind
how committed they
are to a
good outcome.
then becomes privately
falls.
US
For instance,
laws. This
many debt
contracts with highly
means that US courts
level of these seizures could
depend on the
will enforce seizures
level of debt.
Another
the disruption of a country's access to trade credit. This disruption
actually not easy for banks to enforce.
is
Individual banks would like to deviate from this
sanction and offer trade credit unilaterally for a
fee.
Banks
keep each other in check when they have more at stake,
When
its
socially unproductive)
level of the debt.
indebted nations are subject to
assets.
of
are
that the sanctions which are often triggered by imperfect infor-
mation may depend on the
on foreign
some
how tough they
accompanied by costly negotiations. In these negotiations
is
the actors spend resources to demonstrate
is
that ministers and bank
first is
resources demonstrating
when they
The second reason
This disruption
argue that the costs of bargaining increase as the
true for two reasons.
is
presidents are less likely to spend as
less attractive as
not to repay.
for Brazil
bankers some resources.
In the models presented below
level of
would be
it
For example,
readily apparent.
is
i.e
will
presumably try harder to
when
the debt
is
the bargaining costs increase as the level of debt increases, there
bigger.
is
an obvious
(1985) to Admati and Perry (1987).
Saloner (1987) He shows that predatory actions
by firms can be used to demonstrate that one is willing to pay little for the purchase of one's competitor. This rationalises
predation in situations where the two parties are negotiating the terms of a horizontal merger.
*Not
*
all
models of
this type exhibit costly delay.
One paper which shows how bargaining
*See Sachs and Huiiinga (1987).
^New York Times, February 22, 1988.
Compare Gul and Sonnenschein
costs can take unexpected guises
is
.
Buybacks are a mechanism which allows these reduc-
benefit to reducing the extant debt.
Under
tions.
plausible conditions the social benefits from the debt reduction are shared
by both banks and debtor countries. Therefore both find some buybacks
The debtor
when they
gains because there
The
arise, are less costly.
posturing and because the disagreements,
less costly
is
creditor gains because posturing
some money up
well
and because he
it is
perfectly consistent for both to benefit even
receives
in their interest.
front.
The key
is
costly for
result of the paper
him
is
as
that
though buybacks increase the secondary
market price of debt substantially.
I
also
show
that, unless they are able to modify the institutions of bargaining, gov-
ernments cannot generally produce Pareto improvements. Nonetheless, governments who
wish to help the parties involved have a role
tearing
sum
it
transfers. This intervention
Of
tive
up produces a higher amount of
is
is
some
desirable because
it
reduces bargaining costs.
mechanism
of the debt. In this regard
it is
of debt reduction.
is
worthwhile as
well.
By
contrast, in
my
One
worth pointing out that,
hazard models, buybacks are often worthwhile for lenders only
ness
do lump
recipient welfare per unit spent than
course, buybacks are far from the only
to forgive
debt repurcheises. Buying up the debt and
in
if
alternain
moral
outright debt forgive-
models, lenders always lose from outright
forgiveness.
On
the
mechanism of choice. The key
require
is
the other hand, perhaps only certain special features of the model
little
feature of buybacks that
coordination and bargaining.
The
I
want
to
make buybacks
emphasize
is
that they
countries can enter a market where debt
traded anonymously. Therefore, the costs of bargaining and negotiation are avoided. In
other words, buybacks are attractive exactly where other methods for reaching agreement
are costly.
If it
were possible to bargain
The paper proceeds
posturing. In Section
as follows.
2, I
costlessly,
In Section
buybacks would be dominated.
1, I
extend this model while
present a model of negotiation and
in Section 3
I
study costly sanctions.
Section 4 concludes with a comparison of these models to the moral hazard models of
equilibriimi buybacks.
*See Krugman (1988) for the circumstancee under which this equhralence holds. When the buy back is financed, as in
Froot (1988) with funds not previously recoverable by creditors, the equivalence breaks down. Creditors can then benefit from
buybacks even where forgiveness would be detrimental to their interests.
1.
Costly Negotiation
I
consider a collection of lenders and a sovereign borrower.
borrower implies that, as
Stiglitz (1986),
much
in
at all
It is
and
punish the debtor.
ability to
on the borrowers' resources.
have debts exceeding even one year's GNP.
if
sovereignty of the
of the literature surveyed in Eaton, Gersovitz
repayments depend only on the creditors
Repayments do not depend
The
Problem debtors
thus reasonable to suppose as
I
rarely
do that,
they wished, they could easily repay their entire obligation.
The borrower has
value of D.
previously incurred a debt whose face value has a present discounted
This means that there
more than D. More
collecting
borrower and this
precisely, the lenders
have some
permit the lender to
if
inflict
D
the borrower has not paid
US
in full.
whether to engage
borrower the option of not engaging
this
courts condone
Otherwise they do not
relationship between lender and borrower has three stages.
buyback some debt,
on the
pain.
of period 1 the borrower decides
in
any buyback
in
if
buybacks
at
he so desires.
At the beginning
all.
If
I
thus give the
he does decide to
becomes public knowledge at the beginning of period
a secondary market for the debtor's claims opens.
this,
ability to inflict pain
the reason the borrower pays back. However, the
is
this giving of pain only
The
an outside agent which prevents the lenders from
is
depends on whether buybacks
will take place.
By
The
1.
After
price q for the debt generally
using this timing convention,
consider the losses to debtors from buybacks considered by Dooley (1988) and
I
explicitly
Bulow and
Rogoff (1988). 1°
Bargaining starts
in the
second period. Bargaining involves negotiations between the
lenders as well as negotiations between the lenders and the borrower.
get under
way the
lenders discover
how tough they
this relatively late resolution of uncertainty.
Therefore, individual creditors cannot
start trying to coordinate their actions.
'We
know how tough
The second
There are several reasons
are.
One reason
As these negotiations
is
that there are
many
for
lenders.
lenders are collectively until they
re«ison the
toughness of each individual
would otherwise have a model of extortion. Pure extortionists are limited only by the available resources of the victim
or by the desire to induce the victim to keep producing resources.
'"If, by contrast, the debtor were allowed to make secret, unforeseen, buybacks he would obtain the debt at a lower price.
Such secret buybacks would lead to a redistribution of income from lenders to borrowers. Of course, for secret buybacks to
have this feature they must not always occur in equilibrium.
6
lender evolves randomly over time.
If
the date of bargaining
known with
bargaining strength of the lenders cannot be
is
set in advance, the actual
certainty before this date.
In the second period the lenders have the opportunity to convince the borrower that
they are tough. To act tough one must spend resources har2issing the borrower. In practice,
some trade and
the
US
credit flows
must be disrupted. This aggressive behavior
courts because, at this stage the borrower
is
not in
full
tolerated by
is
compliance with the terms
of the loan.
one of two types. Tough lenders need to spend
In period 2 lenders can be of
sources than soft lenders to
is
instill
less re-
a given degree of pain. Therefore the level of harassment
a useful signal of the lender's true
ability.
As a
tough lenders
result, in equilibrivmi, the
harass more than the soft lenders.
Bargaining continues
in the third period.
To keep the
at this point, lenders collectively issue a take-it-or-leave-
analysis simple
it
offer to the
I
assume
borrower.
that,
If
the
borrower does not pay the amount demanded the lenders unleash their collective punishment power. For this theory to make sense, the lenders must be willing to mete out
Subgame
these punishments ex post.
perfection of this type
and Rogoff (1986), these punishments are simple trade
worthwhile for the lenders even
if
is
guaranteed
seizures.
the seized goods are worth
if,
Bulow
as in
Such trade seizures are
much more
to the debtor
than to the creditor.
While the punishments
I
consider here can be interpreted as seizures,
that in practice other forms of punishment are employed as well.
are punishing a debtor to disrupt
There are two reasons
lenders. ^^ This
for this.
its
The
commitment tends
first is
that the
commitment
likely
much
who
as possible.
to sanctions helps the
to soften the position of this particular borrower. Such
who become
very angry
when
little.
The second reason
in
is
expect lenders
trade and financial relations as
a commitment can probably be ensured by hiring individuals
the debtor pays
I
it
is
that lenders often have other debtors.
There
is
some
benefit
having a reputation for toughness with these other borrowers. Penalizing a particular
"Just
like
commitment
to jail criminals helps the innocent.
7
may
borrower
help establish this reputation with others. Of course, lenders differ in their
exposures to different debtors. Also the conditions of debtors vary. As a result, reputation
not perfectly transferable. This
is
is
why
the reason
period two harassment
is
helpful to
the lenders.
Whether the punishments
are seizures of tradeable goods or whether they are
extensive, borrowers are unlikely to
inflict
To
much damage,
cripple a country,
lenders
it is
know
initially
how much damage
lenders can
must know a great deal about the country's
helpful to
impede
its
more
To
inflict.
vulnerabilities.
access to trade credit. This requires that
lenders be able to prevent other financial institutions from transacting with the debtor.
Tough
lenders have both the requisite knowledge and resolve.
It is for this
the disruption of trade and financial flows required by period two harassment
reason that
is
relatively
costless for these lenders.
The
which they make
in
period payments P.
fourth
is
borrower has four components. The
utility of the
the
The
first
period.
third
is
The second
is
the disutility from the pain
lenders offer. This loss equals Lt where r equals s or
Tough
utility
can be written
lenders find
it
B
t
N suffered
in
period
The
2.
the borrower does not accept the
if
depending on whether the lenders
easier to retaliate so Lt exceeds Lg. Total borrower
as:
U = -e[b + P + N +
where the expectation
I
the buy back payment
the present discounted value of third
the loss from the punishment after period 3
are soft or tough.
first is
E
is
L^
(1)
taken over the subjective distribution of types. In equation
(1),
assume that the borrower
is
indifferent
between paying one dollar
dollar of present discounted value in period 3. This
is
standard
if
in
period one and one
payments are discounted
at the borrower's opportunity cost of funds.
The pain
A^
is
a function of the effort
to define the functions
of pain
N
.
This
is
which give the
is
the type of the lenders.
level of effort
It is
convenient
needed by type r to obtain the
level
given by:
A=
The function /
A and
T
fr[N)
strictly increasing in the
=
S,t.
harassment N. For any given N, fr
8
(2)
is
larger
when
T
equal to s than
is
(which
I
lenders
who
write as /|)
is
when
it
equal to
is
when
strictly higher
t.
More importantly, the
derivative -4^
This captures the idea that those
r equals s.
are tough from period 3 on are likely to find harassment in period 2 relatively
inexpensive. These are the lenders
In the period
1
who
find
secondary market, q
it
the price of any debt instrument whose dis-
is
counted face value equals one. The buyback,
means that the outstanding nominal debt
easy to coordinate and police each other.
if it
after the
occurs, takes place at this price. This
buyback,
jD, is
given by:
b = D--.
(3)
9
Lenders act competitively
this price
to
it is
impossible to change
and keeping the debt.
selling his
If
q.
market. They perceive that
through their individual actions. As a result they must be indifferent between
selling at q
By
in this
claim on a single unit of present value, a lender gets a utility equals
he keeps the debt, he shares
harassment
costs.
I
assume that
in the
payments from the debtor
his share of
both
is
equal to his share of the claims
outstanding. Thiis, the claim on a unit of present value gives a lender
claim a period 2 utility equal
sell his
where
E
The
feature of (4)
is
who
W must equal
that lenders and borrowers use the
q.
same discount
future payments which equal one dollar of present discounted value
at the borrower's opportunity cost of funds also equal
lender's discount rate.
units,
is
My
chooses not to
to:
takes expectations over types. In equilibrium,
One important
as well as in the
if
when discounted
one dollar when discounted at the
notation, where everything
actually easiest to understand
rate.
is
denominated
one pretends that the future
isn't
in
discounted
discounted at
all.
'^This simplifies the analysis without affecting the qualitative conclusions. Since borrowers have limited access to capita]
markets we expect them to discount the future more heavily. This would imply that future payments which give a disutility
of P to the borrower give a smaller utility to the lenders. This large discounting of future payments by the borrower reduces
the attractiveness of buybacks from his point of view. Nonetheless, there are still parameter configurations for which buybacks
benefit
all
parties.
Equal discounting by both parties rationalises my assumption that the debts are fully settled after one round of bargaining.
With equal discounting there is no reason to defer payments further. If borrowers discount the future more heavily, equilibrium
payments to lenders might well accrue more slowly over time. This would rationaliEe the protracted nature of debt crises.
9
To analyze
game,
this
borrower believes that the lender
expects to lose
any
offer
crZ/3
+
(l
—
Suppose that,
start with the last period.
I
<7)Lt
is
if
soft
in
period
3,
with probability o. This means that the borrower
he turns down the lenders'
offer.
As a
result he accepts
which involves a payment such that
P <oLs^{\-a)Lt.
The
lenders,
knowing
In period
this, will
there
2,
now
is
a
clcissic
is
soft.
will
pay at most ^lLs
+
(1
—
(5)
amount which makes
ask for the
signaling game.
rassment acts as a signal of the lenders type.
borrower
the
A*)I't
If
where
hold with equality.
(5)
In this
game
neither type harasses the borrower, the
/z is
the prior probability that the lender
There are now two cases to consider depending on whether
equal to \iLs
+
—
(1
//)!/(.
the level of ha-
D
is
smaller than or
In the former case, neither type gains by signaling since they
recoup the entire face value of the debt
any event. Thus,
in
in this case, the equilibrium
has no harcLssment and the period 3 payments equal D.
Suppose instead that
have been deleted,
D
exceeds /zLs
+
(l
—
fj.)Lt.
lack of harassment cannot be
Then, once dominated strategies
an equilibrium. Tough lenders would
deviate by inflicting an A^ such that:
mm{D, Lt) where
e is
=
fiLs
+
(1
-
fi)Lt
a small positive number. After deviating in this
payment equal
to
inm{D,Lt).
accede to their request.
is
fs{N)
As a
If
-
e
manner they would demand a
the borrower recognizes these deviators as tough, he will
result,
tough lenders gain from the deviation since ft{N)
smaller than fs{N). Moreover, borrowers would rationally recognize these deviators as
tough. After
all,
soft lenders
would
lose
from
this deviation
even
if it
led the
borrower to
subsequently view them as tough and pay them inm{D,Lt).
As shown by Cho and Kreps
unique equilibrium and that
inflict A^*
it
(1987) arguments of this type establish that there
is
a
separates the types. At this equilibrium the tough lenders
such that:
mm{D,Lt)-fs{N*)=L,.
"See Spence (1974).
'*See Cho and Kreps
(1987).
10
(6)
Equation
(6) implies
the borrower.
receiving
At
is
I
that soft lenders are indifferent between harassing and not harassing
assume they leave him alone. Tough lenders
strictly gain
by harassing and
m\n{D,Lt) instead of Lg.
changes in the extant face value of the debt matter as long as
this equilibrium,
D
smaller than or equal than Lt. In this case:
Equation
(7)
dN*
1
db
/;3
(7)
captures the basic reason for buybacks in this model.
A
reduction in the
outstanding debt lowers the amount the tough can collect without affecting the amount
the soft can collect. This makes imitation of the tough's behavior by the soft
The equilibrium
level of
harassment
is
such that
it
just deters the soft
less attractive.
from imitating the
tough. Therefore, reductions in the attractiveness of imitation reduce the equilibrium level
of harassment.
Equation
(7)
collect the face value of the
All that
is
necessary
is
derived under the strong assumption that tough lenders
is
debt
in full. In the
that there exist
some
next section,
I
show how
states of nature in
this
can be relaxed.
which tough lenders
collect
the entire face value of the debt.
The advantage
of debt reductions given in (7)
is
reflected in the secondary
price of debt. Using (4), the secondary value of the debt
market
is:
qb = ti{D-ft{N*)] + {l-,x)Ls
where
The
is
to
I
have substituted
simplest
way
D
for
payments to tough lenders and Ls
how
of understanding
compute the change
qD, the
in
(8)
for those to the soft.
the secondary market responds to changes in
total
D
market value of secondary debt:
m=,^D^=,i^.
dD
dD
(9)
f's
Since both
/x
and
f'-f
',,
*
are positive
value of debt increases although
it
and
increases
less
by
than one, the total secondary market
less
than the face value. The increase
value implies that lenders would lose from unilaterally destroying
some
in
of their claims on
the borrower. Such forgiveness amounts to a reduction in D.
*^This result distinguishes this model from Krugman (1988) where lenders benefit from buybacks only when they also benefit
from forgiveness. Forgiveness is potentially valuable in his model because it alleviates moral hazard. On the other hand,
buybacks are more valuable in my model because I assume that the debtor's ability to pay is not impaired by the buyback.
11
That the increase
market value
in
very surprising since the face value
more surprising
is
is
is
smaller than the increase in face value
only paid off with probability
that the total market value
This possibility arises because
may
harassment are almost equally costly to both types. As a
waste most of the increased collections
mechanism, the model can
in the
What
perhaps
is
well respond very little to changes in
can be arbitrarily close to
f'g
/x.
not
is
Then, marginal
f[.
result,
D.
increzises in
even tough lenders must
form of increased harassment. Through
this
easily account for the rather small reduction in the market's
valuation of Bolivian debt in the wake of the buyback.
Having determined how the
change
in q.
Using
and
(8)
total
market value of debt responds,
If
(1
_
D\
dD
two reasons
D
terms of
obtained from
when
D
reason
is
for q to fall
(10).
The
first
rises.
a reason for q to
rise.
thereby making them
dition for q to
fall is
and
this
(6)
(10)
f's
^
J
f's
and the properties
of the function /.
is
fall.
costly.
The second
On
is
two
the other hand, the third term in (10) gives
less
important.
on q
first
that the increase in debt increases
increased debt spreads the costs of harassment over
effect
^
There are
These two reasons are represented by the
The
While the overall
the
that soft lenders always collect the same, so payments to
the soft per unit of outstanding debt
equilibriimi harassment
^ ^ tA\
D
- f^)h
D\ D
is
now compute
(9):
^
where the inequality
I
is
ambiguous,
it
follows
from
more
units
(ll) that a sufficient con-
that:
.^
< ^.
(12)
_
It
is
worth reiterating that
Otherwise, small increases
in
D
this analysis only applies
do not
raise the
amount
means that the equilibrium harassment does not change.
when
D
is
smaller than L^.
collected by the tough.
This
In turn, this implies that the
secondary market price of the existing debt remains the same.
We now come to the
than or equal to L^.
If
first
stage of the game.
I
assume
for the
moment
the country decides to abstain from buybacks, (8)
12
D
with D
that
is
smaller
equal to
D
gives the secondary
market value of the debt.
denote this value by
I
q.
Equation
(6)
gives the equilibrium level of harassment.
Suppose the country does decide
value of debt
off.
I
now
falls.
If
(12)
is
some
to repurchcise
debt.
met, the secondary price of debt
One
focus on the borrower.
issue
monopsonistically in the market for his
which
own
arises
debt.
is
himself to the amount of resources
B
the lenders are better
whether the borrower can act
the secondary price of debt as independent of the buyback's
stitutional arrangements of the secondary market.
rises so
he acts competitively, he perceives
If
Whether competitive or monopsonistic behavior
Then, the outstanding
is
size.
more
realistic
depends on the
Suppose the borrower cannot cormnit
he spends on repurchasing debt. Suppose further
that lenders simultaneously announce a price at which they are willing to
on the secondary market. In
(4).
Since they
all
announce
in-
equilibriimi, all lenders
this price, the
he wants at this price. He thus acts as
announce a q which
sell their
is
claims
equal to
W in
borrower can indeed purchase as much debt as
if:
db
\
(^^)
lB=-qSuppose instead that, before the lenders announce the price
for their claims, the
borrower
can commit himself to an expenditure B. This commitment induces monopsonistic behavior.
Now
when
B
the borrower realizes that the change in the outstanding face value of the debt
changes
is
given by differentiating
db
dB
For very small buybacks,
(13)
and
(14).
D
is
-1
(14)
q.^D-b)^
essentially equal to
D so there
Whether small buybacks are worthwhile reduces
the borrower gains by buying a small
answer to
(3):
this question
I
first
amount
is
no difference between
to the question of
of his debt at the price q.
whether
To compute the
rewrite (1) as:
U = -B - ii{b ^ N"^ Differentiating this expression with respect to
13
B
a.t
[\
-
D—
^i)Ls.
D:
(15)
The expression
increjise their effort
large
-i gives the
of pain the borrower suffers
when
by one unit. This amount can be arbitrarily large.
enough that the expression
One can
amount
(16)
is
Here the costs of harassment are so large
bargaining occurs under
As a
full
when
for the soft that essentially
needed by the tough to separate themselves. In
^i.
can easily be
It
positive.
interpret bargaining under full information as occurring
zero so, from (8), q exceeds
soft lenders
N*
is
information. This simply reproduces the result of
is
essentially
Bulow and
my
results.
small buybacks are worthwhile, borrowers often choose interior solutions where
the debt outstanding continues to trade at a discount.
because higher levels of
B
The
interior
optima tend to
exist
correspond to higher secondary market prices. Whether (13) or
(14) applies, higher values of q
make
additional debt repurchases less attractive.
an interior solution, the best buyback
borrower
for the
db
The LHS
infinite.
always negative when
is
Rogoff (1988) and helps highlight the importance of imperfect information for
When
is
no harassment
this full information case
result the expression in (16)
f'g
is
Assuming
such that
dN*\
(
of (17) gives the borrower's benefit from an additional unit spent on the out-
standing debt. That benefit must, at an optimum equal the cost. The benefit consists of
two parts.
in
D
LHS
First, the
reduces harassment.
of (17)
fall
when
^
borrower pays
B
less
with probability
The second order conditions
increases. This condition
is
as in (14) than
is
when he
—35
J5,
is
smaller
for this
.
Second, the reduction
optimum
met because the
q lowers the absolute value of both the expressions (13)
For any positive level of
/x,
when
and
require that the
resulting increase in
(14).
the borrower acts monopsonistically
acts competitively. This implies that the equilibrium level of
bigger in the latter case.
When
B
the borrower acts monopsonistically, he reduces his
secondary market purchases so that he benefits from a lower price.
Equation
(17) applies
lutions are also possible.
when
These
debt that the outstanding debt
D
< Ls
the borrower pays
D
is
the borrower chooses an interior solution. Extremal soarise
when
the borrower chooses to buyback so
paid with probability one.
to both types in period 3. So,
14
If
much
buybacks are so large that
no signaling
is
necessary and
there
is
no equilibrium harassment. As a
result, q
one. Reduction in buybacks from this
is
On
point increzise equilibrium harassment and so might be strictly undesirable.
hand, further
To gain
a borrower
buybacks have no
increzises in
effect
on harassment.
further insight into these extremal points,
who
acts competitively
(13) in (17) extremal values of
B
I
study the conditions under which
would choose to lower
are worthwhile
the other
D
until q
is
one. Using (7) and
if:
or
f'siO)
If (18) is
I
have studied buybacks assuming that
D
is
smaller than or equal to
For larger values of the debt, small buybacks do not change the amount paid in the
third period
and so leave harassment unaffected. Thus, small buybacks do not
borrower. Nonetheless, even for such large debts,
and jumping
The
Lg,
(18)
not satisfied, borrowers shy away from these extrema.
Until now,
Li.
< YZT-
is
to a point
level of
B
which
which
is
is
satisfies (17) or, in the
desirable.
may be worth having
a large buyback
either extremal or satisfies (17).
an equilibrium when borrowers buy debt
this equilibrium
it
extremal case, that which equates
freely.
The remaining question
At the monopsonistic equilibrium the borrower
to purchasing an additional unit of debt while the lenders are strictly better
monopsonistic equilibrium a very small increase
for
in
repurchases
even slightly larger buybacks, simply mandating a change
in
is
the borrower. Thus the equilibriima
whether
By
agents.
all
could be
made
is
we do not
is
whether
indifferent
off.
So, at the
Pareto improving. Yet,
repurchases does not lead
and hurting
consider changes in institutions and ask simply
better off by using the existing institutions differently.
lump sum
such redistributions are possible, optimality reigns
The reasons
and
constrained Pareto optimal in a certain sense. This
contrast, unconstrained Pareto optimality allows
If
is
D
is
to Pareto improvements. Further buybacks raise q thereby benefitting lenders
optimality obtains as long as
benefit the
when harassment
that both lenders and borrowers care equally about
15
redistributions across
disappears.
money and they both
lose
from harassment.
down
written
until
Unconstrained Pareto optimality thus requires that the debt be
harassment ceases being worthwhile. Yet, unless (18) holds, the equi-
librium level of harassment
is
positive.
Hence, buybacks are usually smaller than those
dictated by unconstrained Pareto optimality.
While
while,
this
argument suggests that buybacks beyond the equilibrium
transfers.
However, the actors
essence of
my
analysis
private self interest.
problem
Of
This argument
practical implications are unclear.
its
It
is
will
relies
on
level are
costless
worth-
lump-sum
presumably bargain over these redistributions. The
that, as they bargain, they find socially costly actions in their
may simply
be impossible to orchestrate a solution to the bargaining
at negligible cost.
course,
if
outsiders wish to transfer resources to the lenders and the borrower, they
can improve the welfare of both. Moreover, by subsidizing buybacks outsiders can improve
borrower and lender welfare by more than by simply giving them resources. To take an
extreme example, suppose a donor pays the face value of the debt
The donor then annuls the claims on the borrower. The
Lenders gain D{1
costs.
Using
—
q).
(8) this
in full to the lenders.
cost of this intervention
is
D.
Borrowers gain their expected payments as well as the punishment
sum
equals:
Dq +
+
,,{N*
ft{N*)).
Therefore, this intervention nets borrowers and lenders put together:
D + n{N* + ft{N*)).
By
contrast,
if
the donor had simply distributed D, and this didn't affect bargaining ability
So borrower and lenders put
neither side would deviate from the original equilibrium.
together would only gain D. Giving the donation through intervention in the secondary
n{N* +
loan market thus produces an extra gain of
One
rically.
difficulty
Many
with this analysis
donors would
like to
is
that
it
ft{N*)].
treats borrower
and lender welfare symmet-
help borrowing nations for political reasons.
donors do not care about lender welfare,
.
it
is
16-
hard to see
why
If
these
they should intervene in
the secondary debt market. Perhaps, such interventions are warranted
mechanisms
Such an analysis
for taxing the lenders.
While the model of
model of the
this section
size of sovereign debts.
is
D
is
Lg
given that both sides have the
same discount
it
this paper.
can also be interpreted as a
then the amount that banks loan
to sovereign borrowers. This interpretation raises a puzzle.
exceeds
there are other
beyond the scope of
is
a model of buybacks,
The equilibrium
if
Why
rates
would the equilibrium
D
and that such debts create
bargaining costs?
The answer
once again, that the equilibrium
is,
Even when the borrowers take
up debts. They
to pile
On
soft.
q as given, they
not unconstrained Pareto optimal.
is
may have
a socially inefficient incentive
realize that increased indebtedness does not raise
payments
to the
average, the payments on an additional unit of debt tend to be smaller than
From a purely
financial viewpoint, additional debt
is
q.
thus attractive. This must of course
be weighted against the additional bargaining costs. However, since the lenders bear some
of the bargaining costs, the borrower
my
In any event,
may accumulate
assumption on borrowers' opportunities for changing
appropriate for buybacks than for debt accumulation.
market
socially excessive debts.
I
assume that there
is
are
more
an anonymous
debt and that borrowers are allowed to participate. This seems
for this
D
realistic for
buybacks. In the case of loan origination, on the other hand, the institutions are rather
different.
Loans to nations tends to originate through large consortia. These consortia
have numerous mechanisms
for preventing
new
borrowers from excessive increases in debt.
They can pressure
potential
Loan origination
best understood as a different bargaining
is
lenders, they can pressure the borrowers themselves.
game which occurs
before
the one studied here.
One
peculiar feature of the loans to nations
optimism surroimding
default
is
loomed
their origination.
in the horizon.
who
are highly indebted today
and high
the
Lenders seemed unaware that the possibility of
Perhaps the lenders were
irrational.
A
different possibility
that the debts would have been manageable without the world recession, low
prices
is
interest rates of the early 1980's.
'See Guttentag and Herring (1985).
17
commodity
In other words, the level of debt
was
When
such that no bargaining was anticipated.
rose,
it
became obvious that lenders known
payments by the countries
resistance to
to be soft
would
collect little.
The bargaining
considered here ensued.
2.
An
Random Repayments
Extension:
The model
of section 1 has the advantage of simplicity.
The
correct in this section.
first
weakness
is
It
has two weaknesses which
that third period payments depend only on the
coordinating capabilities of the lender. They do not depend on the
of the borrower.
The second weakness
random
characteristics
that the range of initial debts
is
buybacks are desirable appears somewhat
I
restricted.
D
for
which
For debts below Lg buybacks are
unnecessary. For debts above Lt they are not locally useful.
In this section
reprisals
autarky
unknown
is
is
initially
beginning of period
leave-it offer.
offer entails
assume that the
I
period
in
2.
ability of the
For instance, the borrower's ability to do well under
random. This bargaining position of the borrower
3.
if
P
the borrower would suffer a loss of
L
gives the probability that
L from
revealed at the
their take-it-or-
the reprisals, the
such that:
P=
2, this level
is
becomes known just before the lenders make
It
Therefore,
a payment
As of period
borrower to tolerate third period
is
L
m\n{L,D)
unknown. What
is
below L
if
is
known
is
that the density function Ht{L)
the lender's type
support \L^,Lt]- To capture the idea that
is r.
soft lenders are less
This density function has
capable of punishment,
I
assume that:
Ls
and
<Lt
that, wherever the supports overlap:
Hs{L)>Ht{L).
As a
result of these assumptions the lenders in period 2,
do not know the payment they
will receive.
18
who know
their
own
type,
Their welfare continues to be given by the
expected value of payments. Similarly the borrowers,
by
L
in equilibrium, care
With these
who
never incur the penalties given
about the expected value of their payments.
provisos, equations (l)-(4) remain valid in this modified model.
principal difference
is
that pooling of types without harassment
is
now
ruled out whenever:
D>L,.
If
the debt exceeds the smallest
amount that a
The
(19)
soft lender
can ever expect to get then
tough lenders find signaling worthwhile.
If
the debt supports an equilibrium with signaling, small buybacks are worthwhile
as long as
two additional conditions are met. The
harassment very
costly.
The second
the payments to the tough.
soft find imitation of the
applied only
tough
Lt exceeded
if
Only
Z),
less
they
first is
that borrowers find increased
is
that reductions in the face value of the debt reduce
in
the presence of such payment reductions will the
worthwhile. Whereas before such payment reductions
now apply whenever:
D<Lt.
In other words,
payments
than the value of their
more
3.
fall if
reprisal.
there
is
(20)
any state of nature
in
which the tough
collect less
Conditions (19) and (20) appear substantially weaker and
plausible than the corresponding conditions of the original model.
Costly Reprisals
In sections 1
and
2
I
have assumed that the lenders can discover the borrower's
relevant characteristics costlessly.
As a
result, their take-it- or-leave-it offers extracted all
the surplus at no further social cost. In practical situations,
can learn so much so
easily.
it is
unlikely that the borrowers
Ascertaining the borrower's ability to sustain punishment
probably no easier than ascertaining the lenders ability to mete
One way
of capturing these difficulties
is
it
is
out.
to consider borrower posturing as well.
Borrower governments would then show how tough they are by ruining their countries
with resolve. Their electorate would show how tough they are by choosing obstinate and
19
otherwise undesirable rulers.
to those of sections
little
Instead,
ter.
I
and
1
do not pursue such niodels here because they would add
2.
look at a different consequence of uncertainty about the borrower's charac-
I
This uncertainty leads to costly reprisals when the borrower turns out to be tougher
than expected.
Formally,
borrowers decline this
are not
meant
I
offer.
Such declines trigger sanctions. Unlike harassment, sanctions
to extract concessions
As discussed
from
this particular
in Section 1, these sanctions
borrower.
have three
seizures of tradeables, the sanctions help the lenders recoup
ment
and
to further disruption of trade
Finally, these additional disruptions
Whether punishments
follow
some
Commit-
of their losses.
financial flows also serves to soften the borrower.
enhance the lenders reputation with other borrowers.
from commitment
ability,
from the desire to
when
collect
some-
the outstanding debt
as fair. Also, gaining a reputation for punishing large debtors probably
cutback
may
they consist of
Tough punishments probably convey a stronger reputation when people regard
higher.
them
When
roles.
thing or from reputational concerns they are likely to increase
is
Tough
allow the lenders to issue a take-it-or-leave-it offer.
in large
borrowing. This
well allow lenders
who
is
precisely
what the banks
promotes a
desire. Finally, the
US Courts
hold larger unpaid debts to seize more assets and punish the
borrower more severely.
My
model of costly punishments
both sections the
utility of
is
the borrower
rule out the harassment costs
very similar to that of sections
is
given by
N. This harassment
is
(1).
specifies a
borrower
L
is
payment P.
is
q.
In period 2, lenders
In period 3, the offer
is
and
make a
take-itnar-leave
either accepted or
it
it
is
offer.
now no
buyback
This
rejected
in
simply
I
is
as before, borrowers can
1,
As
2.
Unlike earlier sections
unnecessary because there
uncertainty about lender's ability to punish. In period
debt at the market price
1
offer
and the
punished.
I
assume that, from the perspective of
a
random
all
agents in periods
1
and
2,
the punishment
variable related to the level of debt. In period three, however, the borrower
knows L. The assumption of symmetric information about L
game considerably
simplifies the analysis.
Without such
20
in the early stages of the
late revelation of the true
L one
would have to worry about borrowers signaling their L via the buybacks they engage
As of periods
1
and
2,
let
h{L,D). The support of
p.d.f.
^
tiff
punishments,
When
is
H{L, D) be the density function
this distribution
Their optimal
P
[L,
L
with associated
Because increases
I/].
exceeds L.
Any
offer thus majcimizes
offer
P
is
know
that borrowers will turn
down with
thus turned
-R-
change
in
consider changes in D.
R
offers
1
— H.
(21)
interior solution, the equilibrium offer satisfies:
Expected borrower welfare from period 2 on
now
down
expected period 3 revenue R:
l-H{P,b)-
I
raise
probability
R = P\1-H{P,D)].
Assuming an
D
in
negative.
crafting their period 2 offers, lenders
whose payment
is
of
in.^^
from an increase
in
D
From
Ph{P, b)
is
0.
(22)
is:
Lh{L,D)dL.
(23)
application of the envelope theorem to (21) the
is:
dR_
db~
Since this expression
=
dH
db'
positive, revenues increase
when
creditors acquire a tool which
raises penalties.
Differentiation of (22) gives the change in P:
Ih
V
+
P—
dP +
^^)
The term multiplying dP must be
condition
is
met
^
\dD
+ P-^
dDj
d£>
=
24
0.
^
'
positive for the second order conditions to hold. This
as long as the density does not increase dramatically with L. If
one ignores
*^Such signaling could never be sufficient to induce full separation of types. If lenders could perfectly infer L from the
repurchase of debt, they would extract payments equal to L. There would be no punishments in equilibrium. The trouble is
that all borrowers would then pretend to be that type whose total payments are lowest. For a rather different model where the
extent of buybacks does produce complete separjition of types see Acharya and
model where the extent of buybacks can distinguish borrowers with
21
Diwan
(1988).
different discount rates.
They focus on a moral haeard
dD
the derivative of the density h with respect to D, the term multiplying
seems plausible, the offered payment
this case, as
Ignoring
^
dD
for
any given P. This
D
raises the benefit
contrast, increases in h induced by increeises in
more concerned about the
large increases \n
h
as
D
rises are
negative. In
rises as the threat increases.
the only consequence of an increase in
agreement goes up
lenders
P
is
is
that
the probability of
it
from a small increase
D have the opposite effect.
in
P.
By
They make the
decline in agreements induced by rises in P. Reasonable
needed
for this
second
effect to
dominate.
Is
a positive
—^ even reasonable?
dD
D
Increases in
h
for big L's.
equilibriimi
P
lower
H. This means
The question
is
that they tend to lower h for small L's and raise
of whether h tends to rise or
high or small.
to
It is difficult
make an
fall
with
D thiis
depends whether
a priori judgment on this question.
Given Bulow and Rogoff (1988) 's evidence, a relevant question
is
how
q responds.
Since lenders must be indifferent between keeping and selling their claims at the price
qD must
the market value of the debt
q,
equal expected revenues R. Therefore the change
in q is given by:
The
expression in (25)
first
price of the claims tends to
claims.
On
dD'
first effect
now come
welfare
q
dD
D
negative while the second
fall
is
positive.
On
the one
hand the
dominates because the increase
in
D
increases revenues. For
revenues
is
minor.
In the presence of buybacks period
to first period buybacks.
1
lender
simply:
is
1
'
because a given revenue must be distributed over more
B+
Period
PdH
DdD
the other hand the increased threat from the larger
small =^, the
I
is
dq
borrower welfare
is
/
J
other words
it
(26)
Lh{L,D)dL.
(27)
instead:
-B-RThe expression under
R.
Li
the integral sign captures the cost of equilibrium punishments. In
captures the social cost of disagreement. Ignoring this expression, (26) and
22
(27) are
symmetric and buybacks have only redistributive consequences.
make
agents better off only by reducing the social cost of disagreement.
all
A
small buyback raises
B
from zero to dB. Equations
^.
small buybacks reduce the outstanding debt by
—^
d(B + R)
-5-—
=
dB
Expression (28)
is
+
1
B+R
is
therefore:
that
(28
dD
More
D
generally, the only effect of
Lenders therefore gain, and
falls.
B + R rises, q rises. As discussed above, these increases
not too pronoimced.
of D on the distribution of L
I
if
in
demonstrate that
^.
proportional to expression (25).
is
The change
(14)
PdH
q
buybacks from the perspective of the lenders
and
(13)
Buybacks can
in q
occur as long as the effect
is
The
derivative of (27) with respect to
include two terms.
The
first is
the
fall in
B gives the benefits to borrowers.
B+R
first
made by
term
is
Jl
Q
q
An
positive in the normaJ case.
it
—\^
depends on
'
dD
.
seems reasonable to suppose that they lower h
only over these small values of L,
is
it is
is:
(29)
dD
increase in
B
lenders and thereby reduce the costs of disagreements.
ambiguous sign because
this
The second
fP
Ldh{L,D)
rLdHL^^^
Ph{P,D)dP
dD
q
The
given by (28).
These benefits
tends to lower the offer
The second term has an
However, since increases
in
D
lower
H
,
it
for small values of L. Since the integral
is
likely to fall as well.
The economic
interpretation of
the following. Reductions in outstanding debt reduce the typical punishment. This
means that the
probability associated with punishments below the cutoff
P
tends to
rise.
and
also
This raises the costs of potential disagreement.
Only
if
the expression (29)
larger than (28), are small
buybacks are
It is
is
positive,
which remains a distinct
buybacks worthwhile. Even when
socially worthwhile, they
tempting to interpret
may
(29)
is
possibility,
positive so that small
not occur in equilibrium.
this result as well as the related
one
in Section 1 as sug-
gesting that outside agents should promote buybacks. This does not follow
agents are purely self interested.
transfers which are necessary to
gaining costs.
On
As
is
before,
it
is
make everybody
if
these outside
probably impossible to implement the
better off without also incurring bar-
the other hand buyback subsidies remain a tool of choice for donors.
23
A
donor who destroys the whole debt
to the integral in (27) over
4.
after purchasing
it
at face value
produces a gain equal
and above the cost of the program.
Conclusions
Economists find
interest.
It
tempting to attribute himian and institutional behavior to
it
self-
therefore important to understand whether the currently popular debt
is
repurchases can be attributed to rational motives. This paper has emphasized that
purchases can help lower bargaining cost. This benefit
is
re-
complementary to the saving
in
agency costs emphasized by Acharya and Diwan (1988), Froot (1988), Helpman (1987),
Krugmaji (1988) and Sachs (1988). Both these savings accrue because there
information. Perhaps
it
would be more accurate
is
imperfect
to say that they accrue because contracts
are imperfect.
In the moral hazard context, for instance,
observe the investments of the borrowers.
What
it
is
doesn't really matter whether lenders
essential
that borrower and lender be
is
unable to write enforceable contracts which stipulate payments as a function of investment.
Similarly, in
my model
of section
1, it is
essential that
borrower and lenders are unable to
write contracts with payments as a function of the level of harassment.
One
difficulty
with the agency-theoretic approach
macroeconomic variables such
as investment
demands on the macroeconomic
do seem
is
that contracts contingent on
feasible.
policies of borrowers.
The IMF makes very
probably curtail investment since tight monetary policies and cutbacks
are standard. This presents a challenge to agency based models
trying to improve the debt
suffer
The main
in the role
in public
investment
one believes the
IMF
is
from the same problem. Perhaps
contracts contingent on harassment are feasible. However,
what
if
requirements
crisis.
Perhaps the bargaining models presented here
contractually
IMF
Overall, the
specific
it
appears
difficult to specify
sorts of behavior constitute costly harassment.
difference
between the moral hazard models and the bargaining models
they leave for outside donors. In
my
bargaining models,
it
makes sense
is
for the
donors to spend their resources pushing buybacks. This rationalizes the actual behavior
of governments.
By
contrast, with costless bargaining
24
it
seems
difficult to
understand why
governments should interfere
5.
in
what
are basically agency relationships.^^
References
Acharya, Sankarshan and Ishac Diwan: "Debt Conversion Schemes as a Signal of
Creditworthiness: Theory and Evidence", mimeo, June 1988
Atkeson, Andrew: "International Lending with moral Hazard and Risk of Repudiation", Ph.D. Thesis, Stanford Business School, May 1988
Bulow, Jeremy and Kenneth Rogoff: "A Constant Recontracting Model of Sovereign
Debt", Journal of Political Economy, forthcoming, 1986
"Sovereign Debt Restructurings: Panacea or Pangloss", mimeo, June 1988
Cho, In-Koo and David M. Kreps: "Signaling Games and Stable Equilibria", Quarterly
Journal of Economics, 102, May 1987, 179-222
P.: "Buy-Backs and Market Valuation of External Debt", InternaMonetary Fund Staff Papers, 35, June 1988, 215-29
Dooley, Michael
tional
MarK Gersovitz and Joseph Stiglitz:"The Pure Theory of Country
Risk", European Economic Review, 30, 1986, 465-67
Eaton, Jonathan,
Kenneth A.:"Buybacks, Exit Bonds and the Optimality of Debt and Liquidity
RelieP forthcoming in International Economic Review, 1988
Froot,
Fudenberg, Drew and Jean Tirole: "Sequential Bargaining with Incomplete Information", Review of Economic Studies, 50, 1983, 221-47
Gul, Faruk and Hugo Sonnenschein: "Bargaining with One-Sided Uncertainty", Working Paper, 1985
Guttentag, Jack M. and Richard Herring: "Commercial Bank Lending to Developing
Countries: From Overlending to Underlending to Structural Reform" in Smith
and Cuddington eds.: International Debt and the Developing Countries, The
World bank, Washington DC, 1985
Helpman, Elhanan:"The Simple Analytics of Debt-Equity Swaps and Debt Forgiveness" Working Paper 1987
,
Krugman, Paul
R.:
Staff Papers,
"Market-Based Debt Reduction Schemes" forthcoming
May
in the
1988
'*See Prescott and Townsend (1984) for a proof that moral hatard alone does not warrant government intervention.
25
IMF
Prescott, Edward C. and Robert M. Townsend:"Pareto Optimal and Competitive
Equilibria with Adverse Selection and Moral Hazard", Econometrica, 52, January
1984, 21-46
Rubinstein, Ariel: "Perfect Equilibrium in a Bargaining Model", Econometrica, 50,
1982, 97-110
Sachs, Jeffrey and Harry Huizinga: "US Commercial Banks and the Developing Country Debt Crisis", Brookings Papers on Economic Activity 2, 1987, 555-606
"The Debt Overhang of Developing Countries" in De Macedo and
Findlay eds. Diaz Memorial Volume, Wider Institute, Helsinki, 1988
Sachs, Jeffrey:
Saloner, Garth: "Predation, Mergers and Incomplete Information",
Economics, 18, Summer 1987, 165-86
Rand Journal
of
and I. Takahashi:"A Multi-Stage Model of Bargaining", Review of Economic Studies, 50, 1983, 411-26
Sobel, Joel
Spence, Michael: Market Signaling, Harvard University Press, 1974.
/
•-
'-'
_.
26
Date Due
MIT LIBRARIES
DUPL
3 9080 02239 0253