JEFFREY SACHS
Harvard University
HARRY HUIZINGA
Harvard University
U.S. CommercialBanks and the
Developing-Country
Debt Crisis
THE DEBTCRISISof the less developedcountriesbrokeout in August 1982,
with the announcementby Mexico that it would be unableto meet debt
obligationsthen falling due. Since then, more than forty developing
countrieshave been forced to reschedule debts with commercialbank
creditorsand to seek additionallendingand other forms of relief from
the internationalfinancialcommunity.' From the inception of the debt
crisis, the primaryU.S. concern has been the risks to the majorU.S.
commercialbanks, whose exposure in the developing countries has
significantlyexceeded theirtotal bankcapital.
Table 1 shows the exposure of the U.S. banks in the majordebtor
countriesat the end of 1986.The exposureis dividedby size of bank(the
nine largestbanks, as againstthe rest of the U.S. banks)and by type of
claim (on the public sector, as againstthe private sector). The concentrationof the claims is high: the exposure of the top nine banks in just
the top four countries, Argentina, Brazil, Mexico, and Venezuela,
accountsfor $41billion,or 45 percentof total U.S. bankexposureshown
in the table. The top nine banks account for a remarkable65 percent of
total exposureof U.S. banks in Latin America. Sovereignloans, those
to foreignpublicsector borrowers,accountfor abouttwo-thirdsof U.S.
banklendingto the less developed countries(LDCs).2
1. See JeffreySachs, "Managingthe LDC Debt Crisis,"BPEA,2:1986,pp. 397-431,
foranoverviewof thedebtcrisisandthemanagementof thecrisisby thecreditorcountries.
2. To give an idea of the global distributionof bank claims in the problemdebtor
countries,LatinAmericahas about$200billionof bankdebt outstanding,of whichabout
$75billionis owed to U.S. banks,$30billionto Japanesebanks(with$13billionin Mexico
555
556
Brookings Papers on Economic Activity, 2:1987
Table 1. Claims of U.S. Banks in the LDC Debtor Countries, End-1986
Millions of dollars except as noted
Claimsof
top nine banks
Claimsof
all other banks
Secondarymarket
Bid
Valueof
pricea
all public
debt
(dollars)
Public
Other
Public
Other
3,961
41
10,176
2,850
968
1,967
2
5,183
1,296
560
1,677
34
3,822
1,097
236
919
19
3,229
1,219
384
47.0
10.0
55.0
67.0
81.0
2,650
8
7,699
2,644
975
204
286
1,161
34
28
10
35
197
10
7
169
78
712
3
14
33
28
101
0
30
33.0
42.0
45.0
82.0
72.0
123
153
843
30
30
84
217
158
24
25
19
57
13
493
12
33
74
24
5
1
38
17
9
126
4
38.0
60.0
37.0
5.0
74.0
44
175
67
1
19
Mexico
Morocco
Nicaragua
Nigeria
Panama
8,960
405
17
404
261
4,393
282
8
263
1,117
5,571
65
41
144
114
4,732
140
0
92
701
53.0
65.5
5.0
28.0
64.0
7,701
308
3
153
240
Peru
Philippines
Poland
Romania
Senegal
511
2,611
290
93
20
307
1,092
73
22
2
383
942
89
14
6
145
462
17
11
0
11.0
67.0
43.0
87.0
61.0
98
2,381
163
93
16
Sudan
Uruguay
Venezuela
Yugoslavia
Zaire
31
653
4,206
965
8
6
45
2,301
350
4
1
162
1,355
413
1
1
69
1,251
337
0
2.0
68.0
67.0
70.0
24.5
1
554
3,726
965
2
18.0
Country
Argentina
Bolivia
Brazil
Chile
Colombia
Costa Rica
DominicanRepublic
Ecuador
Gabon
Guatemala
Honduras
Ivory Coast
Jamaica
Liberia
Malawi
Zambia
Total
Percentof capital
69
4
2
2
39,721
85
20,131
43
17,282
25
14,116
20
...
. . .
13
31,878
27
Sources: Federal Financial Institutions Examination Council, "Country Exposure Lending Survey: December
1986," Statistical release E-16 (126) (April 24, 1987); Salomon Brothers, Inc., Indicative Prices for Less DevelopedCountry Bank Loans (July 27, 1987).
a. Bid price for a $100 claim on the secondary market as of July 1987.
and $10 billionin Brazil), $40 billionto U.K. banks, and the remaining$55 billionor so
divided among German,French, Canadian,Swiss, and other banks. The estimate for
Japanis fromJapanEconomicInstitute,"Japan'sResponseto the LDCDebt Crisis,"JEI
Report29A(July31, 1987),andforthe UnitedKingdomfromMaxwellWatsonandothers,
InternationalCapitalMarkets:Developmentsand Prospects, 1986, OccasionalPaper43
(InternationalMonetaryFund, December1986).
Jeffrey Sachs and Harry Huizinga
557
The debt managementstrategypursuedby the United States and the
official financial community since 1982 has been geared toward the
protection of the large commercial banks, at least on a short-run
accounting basis.3 U.S. policy has been to maintaincurrent interest
servicing by the debtor countries to the U.S. banks and to avoid any
explicit debt forgiveness or even capitalizationof interest payments.4
U.S. regulatorshave appliedlax prudentialstandardsto bankswithlarge
LDC exposures, allowingthem to carryalmostall such exposure on the
books at face value, thoughits value on the secondarymarketis heavily
discounted. Banks have also been allowed to count as currentincome
all the interestpaymentsthey receive on the loans, even those payments
madepossible only by new "involuntary"loans to the debtorcountry.
By actingas if all is normal,the regulatorshave hoped to accomplish
threethings:to keep the debtorcountriesfromhaltinginterestpayments
or promotingalternativeproposals for debt forgiveness; to keep the
banksfromwithdrawingprecipitouslyfromthe debtorcountries;andto
keep depositors and other creditors of the banks from withdrawing
precipitouslyfromthe banks.In a limitedsense thatstrategyhasworked.
Worst-case scenarios of financial panic have been avoided, and the
banks have been given time to increase their capital ratios. U.S. bank
exposure in the problemdebtor countries as a percentageof the book
value of primarycapitalhas declined significantlysince 1982,as shown
in table 2. The regulatorylaxness may, however, have hindered the
adjustmentof the U.S. banks to the crisis by allowing them to move
slowly in rebuildingtheir capital base. Some banks have paid unduly
largedividendsat the expense of their capital in recent years, because
they have been allowed to overstate theireconomic incomes.
Howeverwell the regulatorytreatmenthas paperedover the crisis, it
has not solved it. Nor has it hiddenthatfact fromthe debtors,the banks,
or the marketplace.Despite the official optimismof the United States
and the creditor communityregardingthe debt crisis and despite the
seeminglyrelaxedattitudesof the U.S. regulators,most marketparticipants have conceded that much of the LDC debt will not be repaid. A
3. See Sachs, "Managingthe LDC Debt Crisis."
4. Debt forgivenessrefersto any restructuringof the debt that reducesthe expected
discountedvalueof futurepayments.Interestcapitalizationrefersto any schemeby which
a portionof interestdue is automaticallyrelentto the debtorfor later payment.Interest
capitalizationis distinguishedfrom "involuntarylending"packages,in which loans are
madeon an ad hoc basis so thatthe countrycan financesome of the interestdue.
Brookings Papers on Economic Activity, 2:1987
558
Table 2. Exposure in the Debtor Countries as a Percentage of Bank Capital, Various
Periods, 1982-86a
Region
End-1982 Mid-1984 End-1986
All U.S. banks
156.6
102.5
7.7
94.8
68.0
3.2
All LDCs
Latin America
Africa
186.5
118.8
10.2
All LDCs
Latin America
Africa
287.7
176.5
19.3
All LDCs
Latin America
Africa
116.0
78.6
3.8
96.1
65.2
3.3
55.0
39.7
1.3
70.6
29.0
41.6
84.7
34.1
50.6
116.1
46.7
69.4
Nine majorbanks
246.3
157.8
14.3
153.9
110.2
6.0
All other banks
Addendum
Total bank capital
(billions of dollars)
All U.S. banks
Nine major banks
All other banks
Source: Federal FinancialInstitutionsExaminationCouncil, "CountryExposure LendingSurvey," April 25,
1983,October15, 1984,and April24, 1987,issues.
a. Exposuresare total amountsowed to U.S. banks after adjustmentsfor guaranteesand externalborrowing.
Totalexposuresarecalculatedfor all LDCs(OPEC,nonoilLatinAmerica,nonoilAsia, nonoilAfrica);LatinAmerica
(nonoil Latin Americaplus Ecuadorand Venezuela);and Africa(nonoil Africa plus Algeria,Gabon, Libya, and
Nigeria).
good indicatorof long-termexpectations regardingLDC claims is the
price of those claims on the secondary market. Column 5 of table 1
records the secondary bid price for a $100 claim, as of July 1987. The
price for claims on Argentina, Brazil, and Mexico is in the range of
$45-$55. The averageprice for the entire U.S. bankportfolio, weighted
by exposure in the various countries, is $55.90 per $100 claim. The $57
billion of U.S. bank exposure to foreign governmentsin table 1 has a
secondarymarketvalue of $31.9 billion.5
5. Since the analysisin this paperwas completed,the secondarymarketprices have
fallenfurtherfor most of the debtorcountries.As of October6, 1987,SalomonBrothers
quotedthe followingprices for the largestcountries:Argentina,34; Brazil, 38; Mexico,
47; and the Philippines,55. Some of the reasons for the furtherdrop between July and
October 1987include:the tougheningof the Braziliannegotiatingposition in the fall of
1987(includingBrazil'scall for a majorconversionof debt to exit bondsat below-market
interestrates);the Peronistelectoralvictoryin Argentinain September1987;the political
instabilityin the Philippines;andthe sharprise in U.S.-denominatedinterestrates.
Jeffrey Sachs and Harry Huizinga
559
Althoughmany bankers and U.S. administrationspokesmen try to
arguethat the secondary marketprice of U.S. debt is a poor guide to
moregeneralmarketsentimentsconcerningthe LDC debt, stock market
prices of the commercial banks closely reflect the secondary market
valuationof the LDC exposure. As pessimismhas grownover the value
of the LDC claims in banks' portfolios, equity prices of banks have
dropped.6
The fact that stock market prices have been discounted helps to
explainthe currenteagerness of banks to sell their LDC exposures at a
discount,since they can accept a capitalloss in the books withoutfurther
depressingtheir market value. Citicorp's decision this past spring to
increase its loan loss reserves against Latin American exposure (an
action that was followed by the other majorbanks in the United States
and abroad)appearsto be a prelude to a policy of selling off the LDC
exposure at a significantdiscount. As we discuss later, this new policy
of sellingoff debt may have importantimplicationsfor public policy in
this area.
We organize our discussion of these developments in the following
manner.First, we brieflyconsiderthe underlyingcauses of the growing
marketdiscount on the LDC debt. Then we turnto an analysis of how
the banksand regulatorshave respondedto the crisis since 1982.Next,
we examinethe evidence that the stock marketis now valuingthe LDC
debt at the substantialdiscounts reflected in the secondary market.
Finally,we explorethe implicationsof the marketdiscountfor the future
of debt negotiationsandfor debt relief.
Why the LDC Debt Sells at a Discount
The shortcomingsof the currentU. S. debt managementstrategyhave
not gone unnoticed.In a 1986study Sachs pointed out that most of the
optimisticassessments of the debt crisis ignoredthe internaleconomic
dislocationscausedby the largedebtoverhang.7Most optimisticobserv6. An initialattemptto analyze the links of stock marketprices and LDC debt was
carriedout by StevenKyle andJeffreySachs, "DevelopingCountryDebtandthe Market
Valueof LargeCommercialBanks," WorkingPaper1470(NationalBureauof Economic
Research,September1984).That study, which used data through1983:3,also found a
significantnegativeeffect of LDC exposureon bankshareprices.
7. See Sachs, "Managingthe LDC Debt Crisis."
560
Brookings Papers on Economic Activity, 2:1987
ers have viewed the debtorcountries'problempurelyin termsof external
parameters such as OECD growth, world interest rates, and global
commoditiesprices. They have failed to take account of the economic
andpoliticaldisarraywithinthe debtorcountriesthathas resultedfrom,
or has at least been greatly aggravatedby, the debt crisis: low rates of
national saving and investment, large budget deficits, and recourse to
inflationaryfinance.8
Most of the LDC debtors have little real prospect of servicing the
interestdue on their externaldebt in the next few years. In the past five
years, they have madesignificantnet resourcetransfersto the creditors.
LatinAmerica,for example,has transferredabout5 percentof GNP per
year. But despite these transfers, the debt-exportratios of the major
debtorcountrieshave risen, not fallen. (See table 3.)
Recent increases in certainprimarycommodityprices, apparentlyin
a laggedresponse to the depreciationof the dollar,gave rise to hope that
the export prospects of the LDCs would improve. Ironically,however,
most of the price increaseshave been for nonfoodprimarycommodities
producedmainlyin the developed countriesor in the Asian developing
countries, most of which are not in crisis.9The prices for sugar, wheat,
beef, coffee, and cacao, the main Latin Americancommodityexports,
continue to be deeply depressed. Moreover,internationalinterestrates
have risen significantlyduring1987.
Many of the majordebtor countries are in fiscal turmoil,even after
sharpcuts in governmentspendingin recent years. The interest due on
the foreign debt constitutes such a large proportion of government
expenditures(around30 percentin manyof the debtorcountries)that it
standsin the way of budgetaryreform.'0 The voters in the new democracies in Latin America are not content to absorbfurtherfiscal austerity
8. Of course, manyof these problemspredatedthe debt crisis andindeedcontributed
to the onset of crisis, a point stressedin JeffreyD. Sachs, "ExternalDebt andMacroeconomicPerformancein LatinAmericaandEast Asia," BPEA,2:1985,pp. 523-64. In most
cases, the excessive budgetarydeficitsreflectdeep social andpoliticaldivisionswithinthe
debtorcountriesthat have been exacerbatedby the need to service a heavy foreigndebt
burden.
9. In LatinAmerica,Chilehas been the mainbeneficiaryof the rise in metalsprices,
since copperaccountsfor more than40 percentof Chile's merchandiseexports. Copper
prices have risen froman averageof 640 per poundin 1985to a price of 91? per poundin
October1987.
10. For a discussion of the budgetaryburdenresultingfrom the foreign debt, see
Helmut Reisen and Axel Van Trotsenburg,"The Budgetaryand TransferProblemof
Jeffrey Sachs and Harry Huizinga
561
Table 3. Ratio of External Debt to Exports, 1982, 1984, 1986, and 1987a
Percent
Country
1982
1984
1986
1987b
Argentina
Brazil
Chile
Colombia
Ecuador
405
339
333
191
239
461
322
402
254
259
536
425
402
198
333
554
471
370
235
464
Mexico
Nigeria
Peru
Philippines
Venezuela
299
84
269
269
84
292
158
356
309
158
413
300
497
308
322
366
310
551
309
278
Total
264
290
385
385
Source: MorganGuarantyTrust Companyof New York, WorldFinanicial Markets (June-July 1987), p. 4,
table6.
a. The debt-exportratio is the averagegross externaldebt as a percentageof exports of goods, services, and
privatetransfers.
b. Projections.
for the sake of foreign creditors. The recent rise in interest rates will
intensify the fiscal pressures. The large fiscal deficits are now being
financed in large part through an expansion of credit by the central
banks, a process that will result in high inflation. For several years,
inflationhas been at triple-digitannualrates in Argentina,Brazil, and
Peru. It topped a 150 percent annualrate in the springand summerof
1987inMexico. The 20,000percenthyperinflationin Boliviawas brought
under control only after Bolivia stopped all interest payments on the
externalbankdebt.
Oneresultof the internaleconomic disarrayhas been a burgeoningof
unilateralactions on the debt, particularlyin the democraticcountries
in LatinAmerica.Bolivia, Brazil, Costa Rica, the DominicanRepublic,
Ecuador,Honduras,and Peruhave all unilaterallysuspendedpartor all
of the interest servicingon their foreign debt in the past two years. In
Argentinaand Mexico, the two majordebtor countries that have not
suspended,the banks found it necessary in 1987to relend much of the
interest due in order to forestall a unilateralsuspension of payments.
MajorDebtorCountries"(Paris:OECDDevelopmentCentre, 1987),and JeffreySachs,
"TradeandExchangeRatePoliciesin Growth-Oriented
AdjustmentPrograms,"Working
Paper2552(NationalBureauof EconomicResearch,April1987).
562
Brookings Papers on Economic Activity, 2:1987
Mexico received approximately$6 billionin bankcredits, andArgentina
recently signedan agreementfor $2 billionin new bankcredits."I
Three other large debtors, Chile, the Philippines, and Venezuela,
have been servicingtheirdebts recently without substantialrefinancing
of interest. Chile, of course, is not so mucha model debtorcountryas a
modelauthoritariancountrywhose governmentcan imposethe requisite
domestic austerity to make it possible to service the debt.'2 In the
Philippines,internalinstabilityatfirstpreventedthe Aquinogovernment
from taking a tough stand with creditors. The government therefore
signeda reschedulingagreementin 1987with no concertedlendingfrom
the banks.13But there is now a good chance that a unilateralpartial
suspensionof debt servicingwill be declaredby the PhilippineCongress.
In Venezuela, as well, the governmentis underfierce politicalpressure
to abandonits recent debt reschedulingagreement. Even the government's own political party has called for reopening negotiations to
achieve debt relief.
In debt agreementsnegotiatedin the past year, the banks have lost
ground.Inthe firstroundofreschedulings,in 1983,debtwas recontracted
with an interest rate spreadof about 2 percentagepoints over LIBOR
(the London InterbankOfferedRate). In the second roundof reschedulings, in 1984-85, the spreadfell to about 1.2 percentagepoints. In the
recent round, the spread has fallen further, to less than 1 percentage
point. Similarly, commissions have declined, and the maturitiesand
grace periodson the rescheduleddebts have also increased.14
11. These amounts correspondto approximatelyone year's interest payments on
medium-andlong-termsovereigndebt.
12. Moreover,the banksmorethanfully refinancedChile'sinterestpaymentsduring
1983-84(that is, the "concertedlending"to Chile in 1983-84exceeded Chile's interest
payments),so that Chile actuallyreceived a net resourcetransferfrom the commercial
banksas late as 1984.Interestingly,concertedlendingto Chilehas been moregenerouson
average(whenthe new lendingis measuredas a ratioto the existingdebt)during1983-86
thanhas concertedlendingto any otherproblemdebtorcountry.
13. In her speech upon the opening of the PhilippineCongress, PresidentAquino
declared,"We cannothelpbutfeel ourforeigncreditorstook undueandunfairadvantage
of the internaldifferenceswe hadwithfactionsintenton subvertingthis Governmentand
destroyingour democracy."(As quotedin the Financial Times,July 28, 1987.)The first
act of both houses of the new PhilippineCongresswas to call for an investigationinto the
foreigndebt.
14. See Watson, International Capital Markets, 1986, for details on spreads and
maturitiesin debt reschedulingsthrough 1986. See "The Risk Game: A Survey of
InternationalBanking,"Economist(March21-27, 1987),p. 18, for an update.
Jeffrey Sachs and Harry Huizinga
563
It is thus not difficultto understandthe growing discount on LDC
paper in the secondary market. The economies in most cases are not
getting better, and the countries are increasingly demanding more
concessions in reflection of that reality. Moreover, the international
macroeconomicenvironment,particularlyregardinginterest rates and
commodityprices, remainsunsatisfactory.Detailedprice quotationson
the secondary markethave been availableonly for the past year, with
severalinvestmentbanksnow circulatingpricesheets, butallindications
are that the discount has been growing and the prices falling over the
past few years, as shown in table 4.
On a cross-countrybasis, the discount on the LDC debt, denoted
Price, can be explained as a function of four variables:the debt-GNP
ratio,D/GNP (the debt-exportratioworksaboutas well); the rateof real
GNPgrowthbetween 1980and 1985,GNPGROWTH;a dummyvariable,
SUSP, indicatingwhether the country has unilaterallysuspendeddebt
service payments; and a dummy variable, ATRR, indicatingwhether
U.S. bank regulatorshave mandatedan allocated reserve, that is, a
write-down,for the country's assets on the books of the U.S. banks.
The followingsimple regressionmodel accounts well for the secondary
marketprices as of July 1987.Numbersin parenthesesare t-statistics.
Price = 77.2 - 9.6 ATRR - 17.2 SUSP
(16.3) (1.2)
(6.3)
-
0.15 D/GNP
(2.7)
+ 2.2 GNPGROWTH,
(2.2)
R2=
0.84; 28 observations.
The dummyvariableSUSP equals 1 if the country suspendedinterest
payments in 1987, and 2 if the country suspended interest payments
before 1987and is still in suspension. Accordingto the equation, a $100
claimon a debt-freeLDC with 6 percentannualgrowthwould command
a secondarymarketprice of $90.40 [$77.20 + (6 x 2.2)]. On the other
hand, a country like Bolivia, with a debt-GNP ratio of 136.8, real
GNP growthof - 4.5 percenta year, a requiredwrite-offfor U.S. banks
(ATRR= 1), and morethan two years in debt suspension(SUSP = 2),
has a predicted price of $2.78 [$77.20 - 9.6 - (17.2 x 2) - (0.15 x
136.8) - (2.2 x 4.5)], comparedwith an actualprice of $10.
A key problemwithinterpretingthe secondarymarketpricesinvolves
564
Brookings Papers on Economic Activity, 2:1987
Table 4. Secondary Market Bid Prices for LDC Debt, Various Periods, November 1985October 1987a
Dollars
Coluntry
November
1985
August
1986
April
1987
July
1987
October
1987
Argentina
Brazil
Mexico
Peru
Ecuador
n.a.
75-83
78-82
32-36
n.a.
66
76
56
n.a.
65
60
63
59
17
56
47
55
53
11
45
34
38
47
5
30
Sources: November 1985, from Economist (16 November 1985); August 1986, from Euromoney (August 1986), p.
71; 1987 figures from Salomon Brothers, Inc., Indicative Prices for Less Developed Country Bank Loans (April 20,
1987; July 20, 1987; and October 6, 1987).
n.a. Not available.
a. Figures are the bid price for a $100 claim on the secondary market.
the role of SUSP. Onits face, the pricingequationsuggeststhata country
can manipulatethe secondary marketprice of its debt by suspending
debt servicing. To the extent that debt service relief is then tied to the
secondary market price of debt, as in some of the relief proposals
discussed later, there mightbe the moral hazardproblemof countries
unilaterallysuspendingdebt paymentas a strategicmaneuverto benefit
from relief. The moralhazardargumentis overdrawn,however, to the
extent thatS USP is proxyingfor othercountrycharacteristicsthatmake
debt servicingparticularlydifficultfor that country:politicalinstability,
adverse export structure,financialcollapse, and so forth. In that case,
SUSP is simply another indicator of "ability to pay," rather than a
manipulablestrategicvariable.
Patterns of Debt Management by the Banks and Bank
Regulators
In response to the debt crisis, U.S. banks have virtually stopped
makingnew loans to the problemdebtor countries, with the little new
lendingthat remainsbeing confinedto specific bailout packages. Bank
earnings, for the most part, did not suffer until 1987, when banks set
aside reserves to cover possible losses on LDC claims. Under pressure
from regulators, the banks increased their primarycapital base and
therebyreducedthe ratioof LDC exposure to capital.
Jeffrey Sachs and Harry Huizinga
BANK
EXPOSURE
IN THE
565
LDCS
The change in bank lending is illustratedin table 5. Although the
widelypublicizednegotiatedloan agreementsaretermed"new money'"
packages, U.S. bank exposure to the problemdebtor countries fell in
absolutedollaramountduring1982-86,afterrisingrapidlyduring197982. The absolute decline in lendingbelies the myth that the banks have
continued throughoutthe crisis to provide net "new money" to the
debtor countries, though at a reduced rate of increase. The widely
publicizedconcertedlendingagreementsin recentyears have been loans
to governments.As table 5 shows, claims on the public sector rose 53
percent during 1982-86. But claims on the private sector declined 48
percent. At the same time that the banks have been providing "new
money" to governments, they have been withdrawingloans from the
privatesector. Three other factors can also account for the differential
growth in claims on the public and private sectors. To some extent,
private sector debts have become public sector debts as governments
have taken over some of the foreign obligations of the private sector
since the beginningof the debt crisis. Secondly, the decline in exposures
to the private sector represents, in part, a write-off of claims on the
privatesector, ratherthan an amortizationof loans. Third, declines in
exposurealso reflectsales by the banksof theirLDC claims, or declines
due to debt-equityswaps. Given the publisheddata it is impossible to
distinguishchangesin exposure due to new loans, amortizations,writeoffs, sales, swaps, or public sector assumptionsof privatesector debt.
The notion of "new money" is also misleadingbecause most "new
money" packages after 1982 have involved considerably less in new
loans than is due to the same creditors in interest. Thus, even when
Mexico or Argentinagets a so-called new loan after months of hairraisingnegotiations,the check is still writtenby the debtorgovernment
to the commercialbank. Technically,the net resourcetransfer(equalto
new lendingnet of amortizationsand interest payments)to the debtors
is negative. These negative net resource transferspoint up one of the
fallaciesin a popularargumentagainst LDC default-that if a country
defaults,it will be not be able to attractnew bank money. Losing new
money will be of little concern to a debtor country if the reductionin
566
Brookings Papers on Economic Activity, 2:1987
Table5. Changesin Bank Loan Exposure,1979-86
Percentage change in
exposure, 1979-82
Country
Argentina
Bolivia
Total
Public
71
165
- 31
-8
Percentage change in
exposure, 1982-86
Private
Total
Public
Private
41
4
84
- 54
- 75
-- 70
- 84
-44
Brazil
Chile
Colombia
50
147
47
78
17
83
38
226
35
10
6
- 33
92
267
19
- 36
- 50
- 57
Costa Rica
- 12
27
- 35
- 16
42
- 81
Dominican Republic
Ecuador
Gabon
33
29
-33
10
22
-35
65
33
2
- 15
7
-72
49
147
-76
-75
- 77
-30
Guatemala
-47
57
- 54
- 60
27
- 75
Honduras
- 34
30
- 57
-9
17
- 38
46
11
42
8
63
19
-43
- 22
-41
-05
-50
- 68
Liberia
- 16
- 43
- 15
- 67
- 55
- 67
Malawi
- 20
-41
46
- 54
-49
-61
113
15
- 2
149
131
-23
70
54
102
121
-76
501
-3
18
- 84
- 51
50
27
- 84
- 39
-38
9
- 84
- 63
31
485
24
- 61
- 3
- 65
82
43
- 18
- 31
- 1
27
99
13
- 28
139
18
- 33
- 34
-47
- 11
- 69
- 50
- 2
45
-44
- 15
-72
-53
- 89
- 79
- 35
251
- 62
- 38
-94
Ivory Coast
Jamaica
Mexico
Morocco
Nicaragua
Nigeria
Panama
Peru
Philippines
Poland
Romania
Senegal
Sudan
8
28
-56
- 82
- 83
- 67
230
34
--71
-39
492
28
- 85
-37
65
38
- 64
-73
1
-21
- 11
-91
28
15
250
-94
- 59
-47
- 64
21
Zambia
25
- 11
231
- 60
- 39
-92
Overall exposure
42
52
36
- 12
53
-48
Uruguay
Venezuela
Yugoslavia
Zaire
Sources: Federal Financial Institutions Examination Council, "Country Exposure Lending Survey," various
issues.
interestpaymentsachieved by defaultexceeds the new money that the
countryis able to borrowby not defaulting.
The patternof concertedlendingpackagesamongthe debtorgovernments also illustratesthe venerableeconomic adage, "If you owe your
bank?100, you're in trouble;if you owe yourbank?1,000,000,then he's
Jeffrey Sachs and Harry Huizinga
567
in trouble."Very systematically,it is the countrieswith largedebts that
have been ableto bargainfor new lendingfromthe banks.This is evident
from the data in table 6. For each country, we measure the size of
concertedloans in year t, CLI,as a proportionof disburseddebt at the
end of year t - 1, D_ 1. On average, the ratio CL,/D,-1 is far higherfor
the large debtors, Argentina, Brazil, Chile, and Mexico, than for the
rest. Venezuela is a significantexception to this rule, since the current
Venezuelanadministrationhas, curiously,nevertriedto bargainfor new
money. To summarize the data in table 6, the fifteen small debtor
countriesin the table had 3.4 percentof the debt at the end of 1983,but
received only 0.3 of the concertedloans during1986.
BANK
EARNINGS
Ironically, during 1982-86 the debt crisis did not have a serious
adverseeffect on the reportedcurrentearningsof the banks,even though
it called into questiontheirvery solvency. Whiledoubtsgrew aboutthe
long-termwillingnessof the debtorcountriesto service their debts and
whileprincipalrepaymentswere postponedfor manyyearsin the course
of reschedulings, most LDCs continued to service the interest due,
thoughsometimesonly afterthe banksloaned them some of the money
to do so. Even when interest paymentswere clearly tied to new loans,
the bankregulatorsallowed the banks to reportthe interest received in
full as currentincome, ratherthan, for example, requiringthat part of
the interest be allocated to loan loss reserves, and therefore not be
countedas currentincome.
As shownin table 7, reportednet income rose between 1980and 1986
for all of the nine major banks, with the conspicuous exception of
BankAmerica,whichsufferedmajorlosses on its domesticloanportfolio.
In some cases the measuredincome was even enhanced by the crisis,
becausein 1983and 1984manyof the reschedulingagreementsinvolved
significantfront-endfees and an increase in the interest rate spreads
builtinto the loan agreements.As table8 shows, the shareof LDC assets
on a nonaccrualbasis at the end of 1986is only slightlyhigherthan the
ratioof domesticloans on a nonaccrualbasis."5
15. Nonaccrualloans are loans in which interest servicing is behind schedule or
sufficientlysporadicso that interestis creditedto the bankonly as it is received (thatis,
on a cashbasis),ratherthanthe moretypicalprocedureof creditinginterestas it accrues.
568
Brookings Papers on Economic Activity, 2:1987
Table 6. Medium-Term Concerted Lending as a Percentage of Debt Outstanding
from Private Financial Institutions, 1983-86
Percent
1983
1984
1985
1986
Average,
1983-86
Argentinaa
Bolivia
Brazil
Chile
Colombia
12
0
11
35
0
18
0
14
16
0
0
0
0
9
29
0
0
0
0
0
8
0
6
15
7
Congo
Costa Rica
Dominican Republic
Ecuador
Gabon
0
0
0
20
0
0
0
0
0
0
0
0
0
0
0
9
0
0
0
0
2
0
0
5
0
Guatemala
Honduras
Ivory Coast
Jamaica
Liberia
0
0
0
0
0
0
0
0
0
0
0
0
4
0
0
0
0
0
0
0
0
0
1
0
0
Madagascar
Malawi
Mexico
Morocco
Nicaragua
0
0
11
0
0
0
0
6
0
0
0
0
0
0
0
0
0
8
0
0
0
0
6
0
0
Nigeria
Panama
Peru
Philippines
Senegal
0
0
16
0
0
0
0
0
18
0
0
3
0
0
0
4
0
0
0
0
1
1
4
5
0
Sudan
Togo
Uruguay
Venezuela
Yugoslavia
0
0
18
0
41
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
5
0
10
0
0
0
0
0
0
0
0
0
0
Country
Zaire
Zambia
Sources: Authors' calculations with data from World Bank, World Debt Tables: External Debt of Developitng
Couintries, 1986-1987 (World Bank, 1987); World Bank, World Debt Tables, Second Siupplement (World Bank, 1987);
and Maxwell Watson and others Internzational Capital Markets: Developments and Prospects, 1986, Occasional
Paper 43 (IMF, December 1986). For each year, we calculate the ratio of the concerted loan CL, to the disbursed
debt at time t- 1, D,t 1.
a. In 1987 Argentina received a concerted loan amounting to 5 percent of its 1986 outstanding loans.
569
Jeffrey Sachs and Harty Huizinga
Table 7. Bank Reported Net Income, 1980-87
Millions of dollars
Bank
1980
1981
1982
1983
1984
1985
1986
1987a
Citicorp
BankAmerica
Chase Manhattan
Manufacturer's Hanover
J. P. Morgan
449
643
354
229
342
531
445
412
252
348
723
390
308
295
394
860
391
430
337
460
890
346
406
353
538
998
337
565
408
705
1058
-518
585
411
873
-999
-929
-832
- 1103
952
Chemical
Security Pacific
First Interstate
Bankers Trust
First Chicago
174
181
225
214
63
205
206
236
188
119
241
234
221
223
137
301
264
247
260
184
341
291
276
307
86
390
323
313
371
169
402
386
338
428
276
-703
112
- 165
-151
-438
Sources: Compustat data base and Keefe, Bruyette, and Woods, Inc., Keefe NatiotznvideBankscatn (July 17, 1987).
a. Projected.
Table 8. Percentage of Bank Exposure to Latin America on Nonaccrual and Percentage
of Other Bank Assets on Nonaccrual, End-1986a
Latin
debt
Other
assets
Citicorp
BankAmerica
Chase Manhattan
Manufacturer's Hanover
J. P. Morgan
3.8
6.1
3.0
0.8
1.8
1.6
3.6
2.0
3.0
0.8
Chemical
Security Pacific
First Interstate
Bankers Trust
First Chicago
1.3
1.6
4.4
3.5
2.4
2.3
1.9
1.7
1.5
2.1
Average
2.9
2.0
Bank
Source: Based on data from Salomon Brothers, Review of Bankperformiance, 1986 (Salomon Brothers, 1987).
a. Nonaccrual loans are loans in which interest is credited by the bank on a cash basis rather than as it accrues.
Latin exposure includes loans to Argentina, Chile, Mexico, and Venezuela.
In assessing the effects of the debt crisis on measuredearnings,one
must draw a distinctionbetween the bank claims on the public sector
and those on the private sector. For the sovereign, or public sector,
loans, the vast bulk of interest due has been paid on a timely basis.
Among the majordebtors before 1987, only Argentinafell behind on
interestpaymentson sovereign debt, in 1984and early 1985. Brazilian
sovereign debt has been in suspension since February 20, 1987. By
570
Brookings Papers on Economic Activity, 2:1987
contrast,privatedebtorsin Argentina,Mexico, andVenezuelahave had
periods of fairly significantarrearageson their debt, thoughby the end
of 1986most of those arrearageshadbeen eliminated.Also, an unknown
proportionof the privatedebt has been lost forever in the form of firmlevel bankruptciesor in debt workouts with the creditors at slightly
concessionaryrates.
Onlyin 1987have the income statementsof the banksbegunto suffer,
as some of the largerdebtors,especially Brazil,have suspendedinterest
paymentsand, moreimportant,as bankshave madesignificantadditions
to loan loss reserves. Because of loan loss provisions, the large U.S.
banksposted losses of about $10 billionin the second quarterof 1987.
It is useful here to make clear the meaningof the recent additionsto
loan loss reservesby Citicorpandthe otherleadingbanks.Table9 shows
the size of the additionsand the shareof Latin Americanexposure that
is now covered by the reserves. That shareis calculatedby subtracting
all domestic nonperformingassets from the banks' total loan loss
reserves. The net reservesare then comparedwith the exposurein Latin
America. Citicorp's stated goal was to cover 25 percent of its Latin
Americanexposure.
Since the loan loss reserves are "unallocated," that is, not tied to
particularloans, or even to particularcountries, they do not involve a
write-downin value of particularassets. More obviously, they do not
involve any forgiveness by the banks of any part of the debts owed by
the developingcountries. The increase in unallocatedreserves reduces
reportedincomeof the banks, but it does not reducetaxableincome. On
the balancesheet, the increaseis a transferfrom shareholders'equity to
loan loss reserves. It does not affect measuredprimarycapital of the
bank because U.S. bank regulatorscount loan loss reserves as part of
primarycapital.
The additionto reserves does not affect the cash flow of the banks. In
that sense it is a cosmetic move only. In the future,if the bankswrite off
some portion of their LDC exposure, either by selling the assets at a
discount or by settling with the countries at below-marketterms, they
will be able to chargethe losses to the loan reserves withoutany effect
on reportedincome. At thatpoint, however, the capitalbase of the bank
would shrink,andthe taxableearningsof the bankwouldfall in line with
the write-off. Thus, by acceptinglarge reportedlosses now, the banks
will be betterplaced to reportpositive earningsin the future,even if the
LDC loans go sour.
571
Jeffrey Sachs and Harry Huizinga
Table 9. Bank Loan Loss Reserves, Net and as a Percentage of Latin Exposurea
Millions of dollars except where noted
Bank
Citicorp
BankAmerica
Chase Manhattan
Manufacturer's
Hanover
J. P. Morgan
Chemical
SecurityPacific
FirstInterstate
BankersTrust
Average
Loan loss
reserve,
end-1986
+
Loan loss
reserve
additioni,
1987
-
Domestic
nonperforminig
assets
Net loan
loss
reserve
Net reser ve
as percentage of exposure to
Latin fouir
1,698
2,172
1,065
1,008
910
3,000
1,100
1,600
1,700
0
2,022
3,148
980
1,761
316
2,676
124
1,685
947
594
27
2
26
14
14
669
729
536
591
1,100
500
750
700
1,015
1,132
1,238
526
754
97
48
765
18
7
4
28
1,042
1,161
1,349
854
16
Sources: Authors' calculations with data from New York Times, July 2, 1987; Salomon Brothers, Reviewv of
Bankperformance, 1986.
a. Data on loan loss reserves are updated through July 2, 1987; all other data are for end-1986. Latin exposure
includes loans to Argentina, Brazil, Mexico, and Venezuela.
CAPITAL ADEQUACY
Even before the debt crisis hit, U.S. bank regulatorshadjudged that
the capital-assetratiosof U.S. bankswere insufficient.New regulations
promulgatedin the early 1980s called for a rise in the ratio of primary
capitalto total assets, from the prevailinglow levels of about4 percent
to levels of 5.5 percent. Total capital(primarycapitalplus certaintypes
of qualifyingsubordinateddebt)was requiredto rise to 6 percentof total
bankassets.
A vast literatureon bankingregulationhas stressed the need for such
prudentiallimits.16 Banks are highly leveraged institutions, subject to
the possibility of large fluctuationsin net worth and also to various
incentiveproblems.A small decrease in the average value of a bank's
assets can dramaticallyreduce the bank's net worth and even drive the
bank into bankruptcy.17Moreover, because banks are operatingwith
borrowedfundsand because most of those funds are insuredby federal
deposit insurance, bank managers may have the incentive to take
16. Foragood discussionof the issues, see GeorgeJ. Benstonandothers,Perspectives
on Safe & Sound Banking: Past, Present, and Future (MIT Press, 1986).
17. Considera bankthathas depositsof $95, loansof $100,andprimarycapital(in this
case equalto shareholder'sequity)of $5. A 5 percentreductionin the value of the assets
wipesout 100percentof the bankcapital.
572
Brookings Papers on Economic Activity, 2:1987
Table 10. Bank Primary Capital as a Percentage of Total Assets, 1980-86
Bank
1980
1981
1982
1983
1984
1985
1986
Citicorp
BankAmerica
Chase Manhattan
Manufacturer's Hanover
J. P. Morgan
3.8
4.0
3.8
3.6
4.7
4.1
3.9
4.2
3.8
5.1
4.2
4.3
4.7
4.6
5.6
4.9
5.1
5.4
5.0
6.9
5.9
5.8
6.4
5.7
7.0
6.2
6.1
6.9
6.3
8.0
6.8
6.9
7.0
7.2
8.3
Chemical
Security Pacific
First Interstate
Bankers Trust
First Chicago
3.7
4.9
5.1
3.5
4.7
3.9
4.7
5.0
4.0
4.3
5.0
4.9
5.0
4.5
5.0
5.5
5.3
5.8
5.6
5.6
6.3
5.8
6.1
6.2
6.1
7.0
6.4
6.2
6.4
7.2
7.2
6.4
6.1
6.5
8.3
Average
4.2
4.3
4.8
5.5
6.1
6.7
7.1
Source: SalomonBrothers,Review of Banzkperfor,nance, variouseditions.
excessive gamblesif bankcapitalis too low a shareof total assets. If the
gamble goes well, the shareholdersenjoy an enormous proportional
returnto their claims. If the gamblegoes poorly, the shareholderslose
only the small amount of the net worth, and the deposit insurance
institutionmust makeup the differenceto the depositors.
Another aspect of prudentialsupervision, one that was obviously
overlookedin the 1970sandearly 1980s,is the requirementthatthe bank
not commitmore than 15percentof its capitalin loans to any borrower.
In fact, the loans to the Brazilian government and to the Mexican
governmentgreatlyexceeded 15percentof capitalfor manyof the large
U.S. banks,but the rulewas not invokedbecause the regulatorsallowed
the banks to treat the various official borrowers, such as parastatals,
central government, and developmentbanks, in Mexico and Brazil as
distinct borrowers even though they were all backed by the same
governmentguarantee.In the event, allof the loansto allof the borrowers
went bad at the same time. The multipleborrowersindeed reflected a
single risk, as mighthave been expected.
On paper, the capital adequacy rules have been enforced, and the
capitalbase of the U.S. bankshas been strengthened.But at least some
of the improvementreflectsaccountingconventionsratherthananactual
strengtheningof bank balance sheets. For bank capital to protect the
bank from bankruptcyand to forestall adverse incentive problems, it
shouldconsist mostly of shareholders'equity, and it shouldbe properly
valued. But the measure of primarycapital used for capital adequacy
JeJfrey Sachs and Harry Huizinga
573
Table 11. Bank Shareholders' Equity as a Percentage of Total Assets, 1981-June 1987
1981
1982
1983
1984
1985
1986
June
1987
Citicorp
BankAmerica
Chase Manhattan
Manufacturer'sHanover
J. P. Morgan
3.6
3.4
3.9
3.2
4.5
3.7
3.7
3.9
3.9
4.6
4.3
4.2
4.3
4.2
5.7
4.2
4.3
4.5
4.3
5.7
4.4
3.8
5.0
4.6
6.3
4.6
3.8
5.1
5.0
6.6
2.7
3.0
3.2
2.7
6.2
Chemical
SecurityPacific
First Interstate
BankersTrust
First Chicago
3.5
4.0
4.3
3.9
3.7
4.1
3.9
4.4
3.7
3.9
4.5
4.4
4.7
4.4
4.8
4.9
4.2
4.9
4.6
4.8
4.9
4.5
5.1
4.9
5.3
5.1
4.5
4.9
4.7
5.9
3.0
3.3
3.3
3.4
n.a.
Average
3.8
4.0
4.6
4.6
4.9
5.0
3.4
Bank
Source: New York Times, July 2, 1987; and Salomon Brothers, Reviewvof Batnkperformatnce,various editions.
n.a. Not available.
requirementsincludes both equity and loan loss reserves. Thus, even
whenthe banksmakeloan loss provisionsbecause they anticipatefuture
losses on assets, measured primarycapital is unaffected, because the
loanloss provisioninvolves a transferbetween shareholders'equityand
loan loss reserves, both of which are fully counted in primarycapital.
Moreover,because the LDC claims are carriedin the books at full face
value, and until recently were not covered by loan loss provisions, the
book values of shareholders'equity clearlyoverstatedthe marketvalue
of shareholders'equity.
Thus, U.S. banks enjoyed risingcapital-assetratios during1982-86,
as shown in table 10, but suffered a significantdecline in the ratio of
shareholders'equity to assets as of mid-1987(table 11),when the banks
madea substantialincreasein loan loss reserves. The conclusion seems
to be thatthe regulatorshave raisedthe ratioof shareholders'equity to
total assets but little in the 1980s.Because the loan loss reserves on the
LatinAmericanclaims still cover no more than 25 percent of the Latin
exposureand because the marketsare signalinga discount on the debt
of perhaps45 to 50 percent, it seems clear that shareholders'equity is
still overstatedon account of the LDC debt, even afterthe additionsto
loan loss reserves.
Regulatorylaxness, a "businessas usual" attitude,certainlycontributed to the failureof the banks to make a greateradvance in rebuilding
574
Brookings Papers on Economic Activity, 2:1987
their equity base. It was clear from the beginningof the debt crisis that
at least some of the interestearningson the LDC debt shouldhave been
regardedas fictitious, particularlywhen leading debtors requirednew
involuntaryloans to meet interestpaymentson existing debts. Prudent
regulatorsmighthave requiredthat the banksbuildup capitalin partby
reducingdividendpayouts. But the majorbanks have maintaineddividendpayout ratios since 1982as if the debt crisis had not occurred,as is
evident in table 12. BankAmericawas particularlyflagrant.Even when
its earningswere fallingbecause of bad domestic loans, not to mention
badforeignloans, it continuedto pay significiantdividends,leadingto a
sharprise in the ratio of dividendsto income. Now the bank is fighting
for survival.
Our conclusion that banks have rebuiltcapital slowly must be temperedto the extent thatotherassets of the banksare undervaluedon the
books relative to true market values. One reason to think that other
assets are indeed undervaluedis that, as we show in the next section,
the marketvalues of manyof the largebankswere at or above theirbook
values as of the summer of 1987, despite the clear evidence that the
marketvalues of theirLDC claimswere far below theirbook values.
We attemptedto create an equity-assetratio based solely on market
values ratherthanbook values, by calculatingthe marketvalueof overall
bankassets as the sum of the marketvalue of bankequity and the book
value of bank liabilities. We assumed that the banks' liabilities, which
are mostly short-termfixed-incomeliabilities,have a marketvalue equal
to book value. We then took the ratio of the marketvalue of equity to
the constructed marketvalue of assets. We found that on average for
the ten large banks, the ratio of equity at marketvalue to assets rose
from 3.2 percent in 1983and 3.6 percent in 1984to 5.5 percent in June
1987, suggesting some real increase in capital adequacy. The sharp
decline in the stock market in October 1987 has probablypushed the
market-basedratio of equity to assets back down sharply, close to the
levels of 1984.
Therewould be one practicalimplicationfor LDC debt management
if the banks' non-LDCclaims are carriedon the books at below-market
value. As the losses on the LDC assets are realized, for example, by
sales of debtin the secondarymarket,the bankswouldbe ableto cushion
the effect on their overall capital by selling off other assets that are
undervaluedon the books andtakingthe capitalgains. Citicorpbeganto
adopt this strategy in the fall of 1987by selling a part of its real estate
Jeffrey Sachs and Harry Huizinga
575
Table 12. Dividend Payout Ratios for Ten Banks with Large LDC Exposure, 1980-86
Bank
Citicorp
BankAmerica
Chase Manhattan
Manufacturer's Hanover
1980
1981
1982
35
37
31
29
33
50
59
70
28
44
38
37
32
37
36
34
27
34
27
1983
J. P. Morgan
35
27
37
37
Chemical
BankersTrust
31
20
29
28
Wells Fargo
Marine Midland
Irving Bank
36
36
33
23
28
26
28
28
37
33
29
36
Average
31
33
37
36
37
1984
1985
1986
32
32
35
86
- 43a
0
41
45
34
30
38
29
31
37
27
36
27
32
34
26
30
38
36
29
34
26
28
28
32
31
41
3lb
28
Source: Salomon Brothers, Review of Bantkperformanice,various editions.
a. BankAmerica paid a dividend of $1.16 per common share despite losses of $2.68 per share.
b. Excluding BankAmerica.
equity at a significantcapital gain to offset the reported losses on its
LDC portfolio.
PENDING REGULATORY
DECISIONS
Two importantregulatorymatters are now pending. The first, and
specific, matteris the accountingtreatmentof the Braziliandebt. The
federalbank regulatorscan requirethe banks to make allocated provisions for loans to foreign governmentsunder the system of Allocated
TransferReserve Risks (ATRR) established in the 1983 International
LendingSupervisionAct. In this system, an interagencycommittee of
the FederalReserve Board,the FederalDeposit InsuranceCorporation,
andthe Comptrollerof the Currencycan declarethe loan to a countryto
be value-impairedandcompel a write-downof the assets. Generally,for
loans to be declared value-impairedthey must meet more than one of
fourconditions:interestis morethan 180days overdue;the countryhas
no InternationalMonetaryFundprogramandno prospectof negotiating
one; the countryhas not met its reschedulingterms for a year; and the
country has no definite prospect for an orderly restoration of debt
servicingin the nearfuture.18The finaldecision is at the discretionof the
18. Forafurtherdiscussion,see MaxwellWatson,PeterKeller,andDonaldMathieson,
International Capital Markets: Developments and Prospects, 1984, Occasional Paper 31
(IMF, August1984),pp. 17-18, and later issues; and "U.S. BankRegulatorsare Called
Likelyto RequireWrite-Downof BrazilLoans," WallStreetJournal,September1, 1987.
576
Brookings Papers on Economic Activity, 2:1987
bank regulators.Typically, the requiredwrite-off is 10 percent in the
first year and 15 percent in the second year and each succeeding year
that the loans are deemed to be value-impaired.The ATRR has so far
been appliedonly to a few smallerdebtorcountries:Bolivia, Nicaragua,
Peru, Poland, Sudan,and Zaire.
Brazil's suspension of interest payments on its sovereign debt was
announcedon February20, 1987.On April2, several banks announced
that they were placing Brazilianloans on a nonaccrualbasis. As of
August 20, 1987,the loans were in suspension of interest for 180 days.
The interagencytask force met in Octoberof 1987to decide whetherthe
Braziliandebt should be declaredvalue-impaired.An interimarrangement that, if carriedout, will provide for a partialpayment of interest
due in 1987will forestalla declarationthatthe Brazilianloans are valueimpaireduntilthe task force meets next, in the springof 1988.
The second, much more general, matteris capital adequacy regulations. The FederalReserveBoardandthe Bankof Englandhave recently
agreed to harmonizetheir accountingtreatmentfor the supervisionof
capital adequacy. The details of the agreement have not been fully
workedout andare in any event not yet public. It is expected, however,
that the new system will weight assets by quality to provide a more
refinedmeasureof bankcapital.
Existing accountingpractices differ markedlyin the two countries.
Now, for instance,the Bankof Englandrequiresthatloanloss provisions
of U.K. banksagainstLDCdebtsbe allocatedby country.Theprovisions
are charged against the capital base of the bank and may be charged
againstcurrentincome for tax purposes. Recently, moreover,the Bank
of England has instituted a scoring system by which the U.K. banks
must evaluate their risks on all LDC loans, and thereby decide upon
reserve levels.
Latin American Exposure and the Market Valuation of
Commercial Banks
The regulatorsand banks have so far operated as though claims on
the LDCs are worththeirfullface value, despiteoverwhelmingevidence
to the contrary. The stock markets, however, have seen throughthe
Jeffrey Sachs and Harry Huizinga
577
accountingveil and written down the value of banks with heavy exposures in the problemdebtorcountries.
A precise estimate of the stock marketvaluationof the LDC claims
is difficult because data are limited: banks are required to report
exposures in the LDCs only when total loans to a country exceed 1
percentof total assets. Therefore,while muchis knownaboutthe crossbank exposures of individualbanks in Argentina,Brazil, Mexico, and
Venezuela, little is known about the claims by individualbanks on the
otherproblemcountries,whichaccountforabout30percentof exposure,
as shown in table 1.
An initiallook at bank share prices confirmsthat the markets have
reacted to the bad news of recent years. Table 13 compares prices,
earnings, and dividends of nine banks with the heaviest recorded
exposure in Argentina, Brazil, Venezuela, and Mexico with those of
nine banks with no exposure. For the heavily exposed banks, with an
averageexposureof 130percentof book value, the averageratioof stock
marketvalue to book value at the end of 1986was 1.0, comparedwith
1.5 for the banks with zero exposure. Similarly, the heavily exposed
bankshada price-earningsratioof 6.6, comparedwith 10.3for the banks
with zero exposure. The differencein marketvalue is not a function of
the differencein currentearnings,but ratherthe price thatthe marketis
assigningto those earnings.Putanotherway, the marketis castingdoubt
on the futureearningsof the heavily exposed banksby capitalizingthose
banksat a lower price-earningsratio.
The last two columns highlighttwo considerationsdiscussed earlier.
Withthe exception of BankAmerica,with its extremelyweak domestic
portfolio, the heavily exposed banks had a rate of earningsrelative to
book value thatis comparableto thatof the bankswith zero exposureyet anotherindicationthat throughthe end of 1986(beforethe loan loss
reserves in 1987and the Brazilianmoratorium)the debt crisis posed a
problem of future earnings, not current earnings. The last column
highlightsthe fact thatthe dividendpayoutratiosof the heavily exposed
banks have not been systematically lower than those of the lightly
exposedbanks.The two exceptionsareBankAmerica,which suspended
its dividendin 1986, and First Security-Utah, which paid dividendsin
excess of currentearningsin 1986.
More generally, bank analysts concur that the currentmarket discounts are in line with, or even greaterthan, the quoted prices on the
578
Brookings Papers on Economic Activity, 2:1987
Table 13. Comparing Banks with Large Exposure and No Exposure in Latin America
Bank
Exposurebook
value
Stock
pricebook
value
Priceearnings
Earningsbook
value
Dividendearnings
ratioa
ratiob
ratioc
ratiod
ratioe
Citicorp
BankAmerica
Chase Manhattan
Manufacturer's Hanover
J. P. Morgan
1.2
1.7
1.4
1.8
0.9
Banks with large exposure
1.1
6.6
0.12
0.5
5.4
-0.17
5.1
0.12
0.8
0.12
0.6
4.7
1.8
9.6
0.17
Chemical
Wells Fargo
Marine Midland
Irving Bank
1.4
0.7
1.1
1.4
0.8
1.6
0.8
0.8
5.4
9.3
6.8
6.1
0.13
0.14
0.11
0.12
0.37
0.31
0.28
0.33
Average
1.3
1.0
6.6
0.10
0.30
Midlantic Banks Inc.
Michigan National
Meridian Bancorp.
BayBanks
First Security-Utah
0.0
0.0
0.0
0.0
0.0
Banks with no exposure'
1.6
9.5
0.17
1.3
8.5
0.11
0.14
1.2
10.0
1.4
9.0
0.13
13.0
0.01
0.9
0.27
0.34
0.43
0.38
2.68
State Street Boston
Commerce Bankshares
Dominion Bankshares
Amsouth Bankcorp.
0.0
0.0
0.0
0.0
2.7
1.1
1.5
1.6
15.1
9.2
9.3
9.2
0.16
0.11
0.15
0.16
0.22
0.29
0.36
0.37
Average
0.0
1.5
10.3
0.13
0.59
0.38
0.00
0.33
0.37
0.29
Sources: Keefe, Bruyette,and Woods,Inc., Keefe Nationwide Bankscan, variousissues; and SalomonBrothers,
Review of Bankperformance, 1986.
a. Exposureto Argentina,Brazil,Mexico, and Venezuelaover bankbook value for 1986.
b. Stock priceover per sharebook value as of mid-1987.
c. Price-earningsratioexpectedfor 1987priorto recentmajoradditionsto loan loss reserves.
d. Per shareearningsover book valuefor 1986.
e. Currentannualdividendratefor mid-1987over 1986earnings.
f. No recordedexposure(banksmustreportLDCexposureonly whenexposureexceeds 1 percentof totalassets).
secondary market.19Such a view helps to explain why, when Citicorp
unexpectedly announceda $3 billion increase in loan loss reserves in
mid-May,the marketsreacted by raisingCiticorpprices more than 10
percentthe week of the announcement.Clearly,the announcementwas
received not as bad news of losses, but as good news of a management
strategyto confrontthe losses aggressively.Oneinvestmentanalystwas
19. ThomasHanley of SalomonBrothersis quotedin FortuneMagazine(March30,
1987,p. 104)as declaring:"The discount [in the stock market]is even greaterthan the
priceconcessionsaccordedThirdWorlddebtcurrentlytradingin the secondarymarket."
Jeffrey Sachs and Harry Huizinga
579
quotedas explaining,"Therewas a huge sigh of relief thatthe bad news
was out. "20
The remainderof this section presentsregressionresultsthatprovide
a somewhatmoreprecise estimateof the marketvaluationof LDC loans
to the banks, as implicit in the banks' stock prices. Four parallel
approachesyield largelyconsistentresults. The firstapproachestimates
the marketvalue of LDC claims held by variousbanks. The second and
thirdapproachesrelate the banks' price-earningsratio and the returns
to holdingbankstocks to measuresof LDCexposure.The finalapproach
studies the movement of bank share prices in response to important
LDC exposure-relatednews.
VALUING
BANK ASSETS
The marketvalues of the securitieson the two sides of bankbalance
sheets shouldbe equal. Thus the marketvalue of a bank's assets should
equalthe marketvalue of its combinedshareholders'equity and liabilities. The marketvalue of a bank'sequityis observedin the stock market.
A bank'sliabilitiesareprimarilyshort-termliabilitiessuch as customers'
bankdeposits and short-termCDs and can be assumedto have a market
valueclose to book value. The marketvalueof a bank'spreferredequity,
whichfor most banksis less than 10percentof shareholders'equity, can
also be assumedequalto book value.2'
Using 01 to denote the market value of one dollar of claim on the
LDCs, with 1 - 01 the market discount on the LDC claim, and 02 to
denotethe marketvalueof one dollarof otherassets, we use the following
relationship:
(1)
MVc + BVp + BVl = O1Aldc+ 02Aother,
where
MVc =
BVp =
BV =
Aldc =
Aother =
marketvalue of outstandingcommonequity
book value of preferredequity
book value of liabilities
book value of LDC exposure
book value of other assets.
20. Michael Metz, portfolio strategist of Oppenheimerand Co., quoted in Ellen
Freilich,"StockPricesFall FifthSession in Row," WashingtonPost, May21, 1987.
21. Thislatterassumptionis madeto ease problemsof datacollection.
Brookings Papers on Economic Activity, 2:1987
580
Using Atotal to denote total assets, we can substituteAtotal - Aldc
for Aother, and divide by Atotal to get the relationship:
MVc + BVp + BVl
(2)
wherea
Atotal
=
02andb
= 01 -
Aldc
=-a?+bb Atotal
02.
The above equationis estimatedfor a cross section of banksfor each
of the years 1982 through 1986 and for June 1987. The regression for
June 1987uses the end-of-1986data for exposure and asset values, but
uses the June 1987stock prices to compute the marketvalue of assets.
Because the banksare requiredto disclose LDC exposure to individual
countries only if exposure is in excess of 1 percent of assets, comprehensive exposure data are availableonly for the majorborrowercountries. In particular,Aldc is limitedto includethe exposure to Argentina,
Brazil, Chile, Mexico, and Venezuela. Thatlimitationintroducesa bias
in the point estimateof b thatwe discuss below.
Table 14 reports the estimationresults. For each of the years 1983
throughJune 1987the estimatedcoefficientof the exposure-assetsratio
is negative and is statisticallysignificant.As the coefficient is equal to
the differencebetweenthe marketvalues of otherassets andLDC loans,
it is clearthat a dollaron the books to the LDCs contributesless to bank
market value than a dollar lent elsewhere. As noted, the exposure
variablecovers only aboutthree-fourthsof LDCexposure.The omission
of other LDC loans biases the coefficient upwardin absolute value to
the extent that banks that are heavily exposed in Argentina, Brazil,
Venezuela, and Mexico are also heavily exposed elsewhere, and to the
extent that other LDC assets are also selling at a discount. Thus we
shouldadjustthe estimatedcoefficientdownward.
If a bank's exposure to the four majordebtors were perfectly correlatedwith otherLDC exposure,if the marketsknew aboutthe remaining
exposure, and if the rest of the LDC debt sold at the same discount as
thatof the fourmajordebtors,then an unbiasedestimateof b for all LDC
debt would be approximatelythree-fourthsof the actualestimate, since
the four majordebtorsaccountfor about three-fourthsof the total bank
exposure. In the absence of perfect correlationand perfect knowledge
of the rest of the banks' LDC portfolios, an adjustment factor of
somethinggreater than three-fourthsis appropriate.We decrease the
581
Jeffrey Sachs and Harry Huizinga
Table 14. Market Valuation of Bank Assets and Latin Exposure, 1982-June 1987a
Exposure-assets
Year
Constant
1987c
1.026
1986
1.020
1985
1.008
1984
0.994
1983
0.992
1982
0.980
ratiob
-0.576
(-2.6)
-0.610
(-2.7)
-0.456
(-4.4)
-0.223
(-3.1)
-0.174
(-3.0)
0.049
(0.8)
Summarystatistic
Numberof
R2
observations
0.18
33
0.18
33
0.30
48
0.17
50
0.16
50
0.01
49
Source: Authors'calculations.
a. The dependentvariableis the sum of the marketvalue of commonstock plus the book values of preferred
stock and liabilities.Numbersin parenthesesare t-statistics.
b. The exposure-assetsratio denotes exposureto Argentina,Brazil, Chile, Mexico, and Venezuelaover book
valueof bankassets.
c. DatathroughJune.
point estimate by a factor of 0.8 to get our preferredpoint estimate of
the value of the LDC debt.
We use the estimates of table 14to findthe implicitmarketprices per
$100of face value of claim, as shown below.
1982
1983
1984
1985
1986
1987 (June)
$102
$85
$82
$64
$53
$57
The series shows that the marketstarteddiscountingthe LDC debt not
in 1982,when Mexico firstannouncedits inabilityto service its foreign
debt, butin 1983.Ever since 1983therehas been a trendtowardsgreater
discounts,a findingconsistentwith the trendin secondarymarketprices
observedin table4.
BANK STOCK EXCESS-RETURN
EQUATIONS
These findingsare supportedby the results of a set of bank stock
excess-returnregressions reported in table 15. Excess returns for a
particularbank in a particularperiod are measured as the difference
betweenthe holding-periodyieldforthe bank(capitalgainsplusdividend
yield) and the holding-periodyield of the Standardand Poor 500 stock
582
Brookings Papers on Economic Activity, 2:1987
Table 15. Bank Stock Excess Returns and Latin Exposure, June 1982-June 1987a
Summnarystatistic
Year
Constant
Exposurebook value
ratiob
1982-87e
0.156
(0.64)
-0.035
(-0.80)
-0.650
(- 2.54)
-0.034
(-0.69)
1982
Dummyc
-1.851
(-3.47)
. . .
CUMd
R2
Number
of
observations
. . .
0.41
27
-0.035
0.01
38
1983
0.266
-0.180
. . .
-0.214
0.23
38
1984
(5.79)
0.387
(2.91)
(- 3.26)
-0.158
(- 0.94)
. . .
-0.372
0.02
39
0.129
-0.049
. . .
- 0.421
0.01
38
(2.56)
(-0.68)
0.03
26
-0.390
0.20
26
...
0.02
26
-0.461
0.27
26
1985
1986
1987f
-0.451
0.070
(- 6.92)
-0.410
(- 6.50)
(0.83)
0.031
(0.39)
-0.165
-0.020
(- 2.37)
-0.111
(1.74)
(- 0.23)
-0.071
(-0.89)
.
-0.484
(-2.23)
. . .
-0.628
(-2.88)
. ..
Source: Authors'calculations.
a. The dependentvariableis the stock holdingrate of return(computedfrom stock price changeand dividends)
minusthe StandardandPoor500holdingratetimesthe bankbetacoefficient.Numbersin parenthesesare t-statistics.
b. Exposureto Argentina,Brazil,Chile,Mexico,andVenezuelaoverbankbookvalue.Forthe five-yearregression,
exposurefor 1984was chosen.
c. Dummyequalto 1.0 for a Texas bank(FirstCity Bank).
d. Cumulativenegativeexcess return.
e. Data fromJune 1982throughJune 1987.
f. Data throughJune.
marketindex multipliedby the individualbank's beta coefficient. The
regressionsrelate excess returnto the ratio of exposure to Argentina,
Brazil, Chile, Mexico, and Venezuela to bank book value. The table
reports an excess-returnequation spanningthe entire 1982-June 1987
period and also a set of yearly regressions.The five-yearexcess-return
regressionand alternativeregressionsfor 1986and January-June1987
include a dummy variable equal to 1 for banks located in Texas. The
dummycapturesthe effects of the oil slumpon the profitabilityof Texas
banks.
The five-year regression indicates that a bank with an exposure-tobook-valueratioof 1 would have suffereda negativeexcess returnof 65
percent. From the yearly regressions, we find a statisticallysignificant
effect of exposure only for 1983. However, the estimated coefficients
are negative for all years except 1986. By summing the coefficient
Jeffrey Sachs and Harry Huizinga
583
Table 16. Bank Price-Earnings Ratios and Latin Exposure, 1982-June 1987a
Suimmary
Exposurebook value
Year
Constant
ratiob
1987c
0.628
(13.74)
9.703
(10.34)
10.264
(20.31)
14.674
(4.62)
7.662
(13.85)
5.839
(28.81)
-3.771
(-3.71)
-3.632
(-2.95)
-4.138
(-5.63)
-5.242
(-1.31)
-1.443
(-2.10)
-0.210
(-0.91)
1986
1985
1984
1983
1982
statistic
K2
Number of
observations
0.29
36
0.20
36
0.41
48
0.04
48
0.09
49
0.02
49
Source: Authors'calculations.See text description.
a. The dependentvariableis the price-earnings
ratio.Numbersin parenthesesare t-statistics.
b. Exposureto Argentina,Brazil,Chile, Mexico, and Venezuelaover bankbook value.
c. DatathroughJune.
estimatesfor each of the years, we can get an alternativeestimateof the
cumulativenegative excess return(CUM) associated with LDC exposure, as shown in the fourthcolumn of table 15. On this basis, by June
1987the cumulativeexcess returnwas - 46 percent for a bank with an
exposure-to-book-valueratioof 1.
PRICE-EARNINGS
RATIOS
As a thirdway to test the relationshipbetween stock prices and LDC
exposure, we regress the banks' price-earningsratios on the ratio of
exposureto the fourmajorLDC debtorsto book value. As the sovereign
borrowershave been currentin their interestpayments, with the major
exception of Argentina during 1984, earnings associated with Latin
exposurehave not sufferedmuch. Low expectations about futuredebt
servicing, however, should be expected to depress the price-earnings
ratios. Table 16 shows that the estimated coefficient on the exposure
variableis indeed negative for all six years and that it is statistically
significantafter 1984.
The relative value of the constant term and the coefficient on the
exposurevariableprovideanindicationof the discounton LDCexposure
584
Brookings Papers on Economic Activity, 2:1987
relative to other assets. A bank with no exposure has a price-earnings
ratio given by the constantterm. A bank with the same book value but
with assets that are only LDC claims (with an exposure-capitalratio of
1) has a price-earningsratio equal to the constant minus the coefficient
on the exposure variable. Assumingthat currentearningsare proportional to book value, regardlessof the distributionbetween LDC claims
and other assets, we can divide the two price-earningsratios to get the
marketprice of the LDC claims relativeto the price of other assets (for
the same size book value of each type of asset). Assuming that other
assets have a price of 1 andthat the coefficienton the exposure variable
is overstatedby a factor of (1/0.8)for reasons describedearlier, we get
the alternativeestimatesof the LDC prices shown below.
1982
$97
1983
$85
1984
$71
1985
$68
1986
$70
1987(June)
$72
These estimates are broadly consistent with those from the assetvalue approach, though the implied discounts are somewhat smaller.
Note thatif the measuredcurrentearningson the LDC assets are smaller
thanthe earningson the alternativeassets per dollarof book value, then
ourproceduresin this section would understatethe discounton the LDC
claims.
PRICES
AND
LDC DEBT
NEWS
Movements of bank prices a day or a few days following important
news about LDC claims also help to bolster the views that the markets
are sensitive to the valueof the LDCclaims. We have alreadymentioned
the market'spositivereactionto Citicorp'sannouncementof the increase
in loan loss reserves. Three other examples-the announcementof the
Austral plan in Argentina,the announcementof the Cruzado plan in
Brazil,andthe announcementof Brazil'sunilateralsuspensionof interest
servicing-reinforce the point.
On Friday night of June 18, 1985, President Alfonsin of Argentina
announced an accord with the InternationalMonetary Fund on an
imaginative stabilization program and monetary reform. Simultaneously, the U.S. Treasuryannouncedin Washingtonthatit hadsucceeded
in assemblinga multilateral$480 million short-termloan for Argentina
to assist with its immediate loan obligations. Even though on the
Jeffrey Sachs and Harry Huizinga
585
followingMondaythe WallStreetJournalheadlinedan article, "Argentina's Latest Austerity Programis Greeted with Skepticism by Analysts," bank stock prices did well that day. The results of a regression
of bankstock returnson thatMondayon the ratioof Argentineexposure
to bankbook value is shown in the firstline of table 17. Accordingto the
equation,Argentineassets rose in marketvalue approximately$12 per
$100 of claims (the coefficient 0.097 is scaled up by the ratio of market
value to book value to get 0.12).
On Monday, March 3, 1986, Brazil announced a similar austerity
programthatincludedan agreementreachedthe previousSaturdaywith
foreignprivatecreditors.The agreementcalledfor a reductionin interest
paymentsof $150 millionin 1985and 1986on $1.5 billion of debt and a
refinancingof $6 billionthatmaturedin 1985.The second line of table 17
shows thatthe combinationof the reschedulingnegotiationsandthe new
programwas disappointingto bank stock investors, with bank stock
returnson thatday significantlynegativelyrelatedto the banks'Brazilian
claims.Each$100of Brazilianpublicclaimsis estimatedto have declined
in value by $8.10, and each $100of privateclaimsby $2.70.
Almost a year later, on Friday, February 20, 1987, the Brazilian
Ministerof Finance, Mr. Dilson Funaro, sent a telex to Brazil's 700
creditorbanks announcinga moratoriumof interest payments on medium-termandlong-termcommercialbankdebt. The WallStreetJournal
commentedthat internationalbankers had grown used to debt alarms
andthatthey were takingBrazil's action in stride. Indeed, the thirdline
of table 17 shows that on February20 bank stock returns were only
weakly negatively related to the ratio of exposure to Brazil to book
value. However, duringthe following week, bank stocks tumbled as
Braziltookfurthersteps thatindicatedits resolve. OnMonday,February
23, ChaseManhattanBank, ChemicalBank,andCiticorpeach lost more
than5 percentof theirstock values. OnWednesday,February25, Brazil
tightenedits policy by telling its banks not to repay foreign creditors
seeking to recall short-termcredit. The fourth line in table 17, which
relatesthe returnon bank stocks between February20 and February26
to the exposure-book-value ratio, shows that the cumulative effect
duringthe week of Brazil'sinterestmoratoriumon bank stock prices is
significantandhighlynegative.
The fifth line of the table focuses on the announcementby Citicorp
on May 19, 1987,that it would add $3 billion to its loan loss reserves in
cl~~~~~~~~~~
0
40.
0.
0
C;
C;~~~~~~~
0
0
WI~
~ ~ ~~ ~ ~ ~ ~ ~ ~~~~~~~~~~~~.
.0
>0
0
0
0.~~~~~~~~~0
MU
rA
On
'
*
~~~~~~~~~:3
.0
0
c-
~~~~
~~~~~~c..
00
0
II
~ ~ ~ ~ ~ ~ ~
'IC~~~~0
0
I
0 Q
rA
WI
0 >
0
0 -0
>
40
~~M0-0>
0
0~~~~~~~~r
c
0
0 0
*
.
*
~~~~~~~~~~~~~~0
.
0.
.0
.0
587
Jeffrey Sachs and Harry Huizinga
anticipationof future write-downsof Latin loans. Line 5 relates Latin
exposure over book value to stock price movements between the day
before and the day after Citicorp's announcement.The added dummy
variableis for Citicorpitself. The regression shows that.Citicorpstock
went up 4 percent, while other banks' stocks went down slightly (and
withoutstatisticalsignificance).
Stock Market Values and Debt Renegotiation
Theevidenceon the marketvalueof LDCdebthasa crucialimplication
for futurenegotiationsbetween debtorsand creditors,as well as for the
policy options of the official community. Since banks' stock prices
alreadyreflect significantlosses on the debt, banks shouldbe willingto
tradetheir LDC debt of a given face value for a safer asset with a lower
face value. Inthe simplestcase of sucha debtconversion,forthe moment
ignoringtax and accountingcomplications, if the stock marketvalues
the debt at $60 per $100 of face value, then a bank's shareholderswill
benefitif the bank sells each $100of debt for cash at any price in excess
of $60.Ofcourse, the swapneednotbe for cash;any marketablesecurity,
such as a bond or an equity claim, with a marketvalue in excess of $60
will do. Such a debt conversion could result either from direct negotiations between creditorsand debtorsor throughthe policy actions of the
officialcommunity,as illustratedbelow. Several benefits are likely to
resultfromdebt conversion schemes that convert the currentdebt, now
pricedat a discount,into cash or into new claims at a reducedface value
thatare thenpricednearthe new lower face value. We shall suggestthat
thereis a strongcase for policymakersto takepositive actionsto support
such debt conversions.
DEBT
CONVERSIONS
THROUGH
BILATERAL
DEBTOR-CREDITOR
ARRANGEMENTS
Debt conversions may be arrangeddirectly between debtors and
creditorsin many ways. The simplest, with a long tradition,is for the
debtorto enterthe secondarymarketfor its debtandto repurchasesome
or all of the debtfor cash at a deep discount.Such repurchasesof heavily
discountedbonds took place in the 1930s. A widespread use of debt
repurchases,however, carrieswith it severalproblems.
588
Brookings Papers on Economic Activity, 2:1987
A first problem is contractual. Most of the existing debt contracts
with the commercialbanks have a "sharingprovision," which requires
that all paymentsby the debtorto the creditorsmust be equally shared
by the participatingbanks. Technically,a debt repurchaseviolates this
clause, since the bankthat sells its claim gets a lump-sumpaymentthat
is not receivedby the otherbanks.The creditorsanddebtorcan negotiate
a waiver to eliminate the sharing provision, though such a waiver
generallyrequiresthe nearlyunanimousconsent of the bank creditors.
A waiver was negotiatedduring1987in the case of Bolivia, under the
restrictiveconditionthatBoliviawill repurchaseits debtonly withfunds
that have been donatedto Bolivia by foreigngovernmentsexpressly for
the purposeof debt repurchases.
A second problem involves the regulatoryenvironmentfacing the
banks. The maindifficultyis that when a bank sells its claimfor cash, it
must recorda capitalloss equalto the differenceof the face value of the
claim and the purchaseprice. This capitalloss reduces the book value
of bank capital and may triggerregulatoryproblems by reducing the
ratio of primarycapital(measuredat book value) to assets. If the bank
sells a $100 claim, valued in the secondary market at $60, for $65, it
wouldenjoy a $5 gainat marketprices. In book value, however, it would
have to recorda $35loss. Althoughthe stock marketgenerallyresponds
to the change in market valuation, and not in book valuation, if the
decline in book value were largeenoughto cause the bankto come close
to or fall below regulatorylimits on book-value capital-to-assetratios,
then the freedomof maneuverof the bankmightbe jeopardized.
Clearly,the regulatoryenvironmentimposes a bias againstdebt sales,
since it now allows an asset worth $60 to be held on the books at $100
untilthatasset is actuallysold at its reducedvalue. Onepossibleresponse
of the regulatorscould be to ease the regulationsto allow the capitalloss
from debt sales to be amortized over a period of several years, an
approachrecently introducedfor some kinds of bad farmloans. Other
"tricks" are also availableto disguisethe debt repurchaseand avoid an
immediatewrite-down.22
22. One such trickis for the debtorcountryto put the $60cash in a custodialaccount
managedby the creditors,to guaranteea bondwith below-marketinterestratesthathas a
presentvalue equal to $60 (for example, a $100face value consol, with a coupon rate 40
percent below marketinterest rates, that gives the bond a marketvalue of $60). The
creditorsthen swap their$100of debt for the guaranteed("defeased")bond, also with a
Jeffrey Sachs and Harry Huizinga
589
A relatedregulatoryproblemis that if a bank sells some of its claim
on a country at a discount, then the regulators and the bank's own
auditorsmightforce it to write down the rest of its claims on the debtor.
Uncertaintiesaboutthe regulatoryresponse have apparentlyprevented
manybanksfromsellingoff smallportionsof theirexposureina particular
country.
A thirdproblemwith repurchasesis thatthe debtorcountrygenerally
does not have the cash availableto makea majorrepurchaseof its debt.
If the requiredcash were available,the debt itself would likely not sell
at a deep discount. To the extent that debt conversions are desirable,it
mightmake sense for the official communityto give or lend money to
debtorsto makepossible a large-scalerepurchase.This is essentiallythe
experiment now under way with Bolivia. We return to the possible
involvementof the official communityto supportdebt conversions in
the next section.
Otherforms of debt conversion have a similareffect as debt repurchases, though they may be differentin appearance.In a debt-equity
swap, for example, a potentialforeign direct investor purchases some
debt fromthe banks on the secondarymarketand bringsthe debt to the
debtor country's central bank. The central bank purchases the debt
using local currency, under the restrictionthat the domestic currency
then be used to make a foreigndirectinvestment. As long as the central
bank repurchasesthe debt from the investor at close to the secondary
marketprice, thatis, at the price, convertedinto local currency,thatthe
investor paid to the banks, then the transactionis essentially a cash
repurchaseof debt when viewed from the perspective of the central
bank. From the debtor country's point of view, the key, and perhaps
only, advantageof such a mechanismover a directrepurchaseof debt is
that it allows the debtor to get aroundthe sharingprovision discussed
earlier.23Otherwise,debt-equityswaps have the same advantagesand
disadvantagesof directdebt repurchases.
face value of $100, but a safe marketvalue of $60. Under standardaccountingrules, a
swap of $100of debt for $100 of bonds of the same debtorgenerallydoes not requirea
write-downin book valuesof the claim.
23. Untilrecently,most centralbankswere redeemingthe debt at close to par,rather
thanclose to the secondarymarketprice, givingthe foreigninvestorthe spreadbetween
the secondarymarketpriceandthe face valueof the debt. In principle,a repurchaseat par
590
Brookings Papers on Economic Activity, 2:1987
A new form of debt conversion that may be similarto a repurchase,
thoughless costly to the debtorin terms of currentcash flow, is an exit
bond. Withan exit bond, the creditorswaps his bankdebt (say, of $100)
for a bond of the debtor country with the same face value, that is, the
same eventual principalrepayment, but with a below-marketcoupon
rate. The bondthereforehas a contractualpresentvalue (a presentvalue
assumingno default)that is below the face value of the existing bank
debt.24
Why shoulda creditorswap a bankloan for an exit bond of the same
debtorthathas a lower contractualpresentvalue?The bondis supposed
to be superiorto the existingbankdebtfor severalreasons. First, holders
of the bondsareexplicitlyrelievedof the obligationto makecontributions
to concerted-lendingpackagesthatmaybe negotiatedbetweenthe banks
and the country in the future. Second, the debtor undertakes, either
explicitly or implicitly,to give the bonds a senior status relative to the
remainingbank debt, that is, to service the bonds before servicingany
of the bankdebt. If the senior status is credible, the bond may be a very
safe claim.25
Third,as discussed in the next section, the creditorsas a groupmay
benefitfromthe introductionof exit bonds, since certainefficiencygains
may act as an investmentincentive, thoughas with manyincentive schemes, the largest
effect was to give a large lump-sumtransferto inframarginalinvestors that would have
invested in the country anyway. Recently, central banks have been findingways to
recapturemost of the discounton the debt, eitherby repurchasingthe debt at close to the
secondarymarketprice or by auctioningthe rightto participatein a debt-equityswap
amongpotentialforeigninvestors,therebyrecapturingthe surpluspreviouslyaccruingto
the foreigninvestors.
24. In contractualterms, the currentbankclaimhas a presentvalue equalto its face
value, since it carriesa marketrateof interest.
25. Considerthe case of Brazil,for example,with about$70billionof bankdebt, and
recent annualnet resource transfersto the banks of about $6 billion. If Brazil were to
convert$10billionof its debt into exit bondswith a long maturityanda fixedinterestrate
of 6 percent,given a safe marketrate of 10 percent,the interestdue on the bonds would
be $600millionper year. If the $600millionis crediblyseniorto the remainingbankdebt,
there would be little doubt about Brazil's capacity to service the bonds, since the $600
millionis farless thanBrazil'srevealedannualcapacityandwillingnessto pay. The bonds
wouldthereforebe a relativelysafe asset, andwouldthereforebe pricedat about$60 per
$100of principal(thatis, 0.6 marketinterestrate x $100,assuminga long maturityon the
bond). Assuming that bank claims on Brazil are now selling at below $60 per $100,
individualbanksshouldbe willingto swaptheirbankdebtfor the exit bonds.
591
Jeffrey Sachs and Harry Huizinga
may arise from the fact that the exit bonds reduce the debtorcountry's
contractual debt service obligations. We shall see that the debtor
country'sabilityandwillingnessto service its debts may well rise as the
contractualobligationsto do so fall.
SOME
TAX
AND
REGULATORY
ASPECTS
OF DEBT
CONVERSIONS
Two details are importantin understandingthe full regulatoryand
financialramificationsof debt conversions. The first involves book
accounting. As indicated earlier, debt conversions may or may not
require immediatewrite-downs of book values of the banks' claims,
dependingon how the transactionis carriedout. In general, if an asset
is swappedfor cash or some nondebtclaim, the bankmustbook the new
asset at its currentmarketvalue andwrite down any differencebetween
the face value of the debt and the marketvalue of the asset received in
return.Thus, if debt with face value of $100is pricedat $60and is traded
for cash or equity worth $65, the bankrecordsa loss of $35.
This loss is importantfor two reasons. First, it reduces the bank's
bookcapital,whichis the measureusedfor regulatorypurposes.Second,
it can be chargedagainsttaxes. If the bank pays the corporatetax rate
on the margin(34 percentin 1988),then the tax savingswould be worth
$40 x 0.34, or $13.60. Thus, the full value to the bank of selling a debt
for $60 would be $73.60. Put differently,if the stock marketis valuing
the debtat $60, the bankshouldbe willingto sell the debtfor $39in cash,
since $39 plus the tax saving of $21-0.34 x ($100 - $39)-equals the
marketvalue.
If the bankdebt is swappedfor a new formof debt, however, then the
accountingandtax rules may be different.If $100of bankdebt at market
interest rates is swapped for a $100 exit bond with a below-market
interestrate, the bank does not in generalhave to write down the value
of its claimunless and untilthe exit bond is sold on the market,at which
time the write-offwould be the differencebetween $100 and the price
receivedfor the bond. This accountingrule (known as FASB 15) gives
the banksgreatflexibilitywith respect to exit bonds. They can choose
the timeto take the capitalloss on the debt even afterthey swap the debt
for exit bonds.
Brookings Papers on Economic Activity, 2:1987
592
WHY
DEBT
CONVERSIONS
ARE ATTRACTIVE
Debt conversions can benefit the creditorsas a group as well as the
debtors.26Most debtorcountries'currentdebtfar exceeds the expected
discountedvalue of net debt servicing,that is, debt servicingnet of new
concerted lending. Keeping on the books debt that is far in excess of
what can reasonablybe expected to be repaidentails efficiency losses
that are often ignored, but are a central aspect of the case for debt
conversions.
Theefficiencylosses arewidelyrecognizedin the context of corporate
or personalbankruptcy,butnot yet in the context of excessive sovereign
debt. There are good efficiencyreasons why a corporationor individual
with excessive debts is allowed to dischargethose debts in the context
of bankruptcy.In a corporateChapter11 proceeding, it is recognized
thateconomic efficiencymay dictatethatan overly indebtedfirmshould
continue to operate, but that efficient operations require an explicit
conversion of the debt into claims with a reduced contractual debt
service obligation.Otherwise,the firmscannotoperateexcept in a crisis
mode: they are denied suppliers'credits; they are subjectedto creditor
lawsuits; they have a hardtime collecting on accounts receivable;they
cannot get new financingfor investmentprojects; and they cannot get
workersto invest injob-specifictraining.Bankruptcycourts do not tell
the corporationto continueto operatewith all of its debt intact, simply
to roll over the debt and pretendthat it can service all of the debt in the
future. It is recognized that the overhang of debt itself prevents the
smooth operationof the firm,to the ultimatedetrimentof the creditors.
The same kinds of efficiency losses apply to sovereign debts. The
overhangof the debt itself can hampereconomic performance,even if
the reality is being concealed by concerted loans or by arrears. As a
simple illustration, suppose that a country owes $10 billion of debt.
Assume that principalrepaymentsare always rescheduled(thatis, that
the debthas infinitematurity).The safe marketinterestrateis 10percent.
26. The argumentin this section, thatdebtreliefcan provideefficiencygains, was first
set forthin JeffreySachs, "The Debt Overhangof DevelopingCountries,"forthcoming
inthe memorialvolumeforCarlosF. Diaz-Alejandroto be publishedby theWiderInstitute,
Helsinki, Finland, 1988. Another recent paper that explores a similartheme is Paul
Krugman,"BootstrapDebt Relief" (MIT, 1987).
Jeffrey Sachs and Harry Hiuizinga
593
To begin, suppose that the country's annualdebt servicingcapacity, its
capacity to make a net resource transfer,is $600 million, assumingno
changein policies. The countryrepaysall thatit can each period, though
its repayment capacity may depend on the kinds of policies that it
follows. The marketvalue of the debt would simply be the discounted
valueof $600million,or $6billionat marketinterestrates.The secondary
marketprice would thereforebe $60per $100of debt.
In the present debt-managementarrangements,assumingno breakdown in negotiations,and withoutdebt conversions, the countrywould
pay the full $1 billionof interestin the firstyear, andget a concertedloan
of $400million, so that its net resource transferwould be $600 million,
which is its ability to pay. Next year, the debt would be $10.4 billion,
with interest due of $1.04 billion. The country would again make a net
resourcetransferof $600million,requiringa new concertedloan of $440
million.Each year the countrywould get a new concertedloan in order
to keep the net transferat $600million.The debt would eventuallygrow
at a rateapproaching10percentper year, the rateof interest.Obviously,
the debtorand creditorare engagedin a simple "Ponzi scheme" to hide
the fact of partialdefault.
Supposethat, instead, all of the debt were converted into exit bonds
with 6 percent interest. The bonds would be perfectly safe, since the
countrywould and could pay $600millionper year on the bonds. There
would be no defaults and no concerted lending, because the debt
conversionwouldobviatethe need for concertedlending.It wouldleave
the position of the debtorsand creditorsunchangedwith respect to net
resource transferseach year. The debt conversion would impose no
losses to the creditorsregardingdebt service receipts, and would allow
bothdebtorsandcreditorsto avoidthe costs of negotiatingthe concerted
loans each year.
Perhapsmoreimportant,the debt conversionwould avoid the risk of
an actual breakdownin debt negotiationsat some point in the future,
leadingto an outrightdefault. The risk of a negotiatingbreakdownis
present in most kinds of negotations, but is particularlyacute in the
bargainingbetween banks and debtor countries, because both the
creditorsand debtorsare negotiatingon a wide varietyof fronts, and so
have the incentiveto stake out toughpositions to avoid the appearance
of weaknessin othersettings.Actualbreakdownsof negotiations,which
have occurred with Nigeria, Peru, and several other countries, are
594
Brookings Papers on Economic Activity, 2:1987
costly. The debtor'sinternationaltradecan be hamperedby a dryingup
of internationaltradecreditsorotherformsof disruption.Thisdisruption
representsa dead-weightloss to both the debtorandthe creditors,a loss
that can be avoidedby the debt conversion.
Note that the marketprice of debt will in general reflectthe fears of
sucha futurebreakdown.In ourexample,therefore,inwhichthe country
can service 60 percent of the debt, the marketprice of the debt would
generally sell at below 60 percent, say $55 per $100. Then, a switch to
exit bonds would involve a capital gain from $55 to $60 that would be
sharedby the creditorsand debtors.
Next, suppose that by undertakinga structuralreformprogramthat
would requireit to forgo $100millionin currentconsumption,the debtor
country could raise its output and thereforeits debt servicingcapacity
by $20 million this year and every year in the future. Debt servicing
capacity, and therefore debt servicing (underthe assumptionthat the
country pays all that it can), would rise to $620 million per year. The
secondarymarketvalue of the debt would rise to $62 per $100.
Such a reformwould requirethe debtorto sacrificeconsumptionfor
the sake of the foreign creditors, since the returns to the structural
reformeffortwouldbe appropriatedby futuredebt servicing.The debtor
would have little incentive to make such a structuralchange, which
could also be politicallysuicidalif oppositionpartiesattackthe governmentfor sacrificingcurrentconsumptionfor the sakeof foreigncreditors.
The assumptionbehind such an attack, that the creditorswould appropriatea largeportion, if not all, of the improvementsin the economy, is
reasonable. Concertedlending and reschedulingsare grantedto countries in dire economic difficulties. Once an economy improves, the
debtoris expected to service its debts in full, and its power to resist debt
servicingis diminished.No player, not the banks, the IMF, the World
Bank, or the creditorgovernments,will excuse a countrywith a healthy
economy from debt servicingon the groundsthat it was once in trouble
and chose to undertakeneeded reforms.
Couldthe reformsbe financedby the foreigncreditorsto increasethe
country'sdebt servicingcapacity,ratherthanreducingits consumption?
Probably not. Supposing that $100 million is lent to the country in
addition to the $400 million of interest refinancing,with the overall
concerted loan totaling $500 million. There is no guarantee that the
structuralreforms would actually be undertaken.The country might
Jeffrey Sachs and Harry Huizinga
595
promiseto undertakethe reforms, receive the loan, use the money for
consumptionpurposes instead, and continue to pay $600 million in net
resourcetransfersin the future. It would benefitfromthis policy choice
by raisingcurrentconsumptionby $100 million at no real future cost.
Because promises to reform are notoriously difficult to regulate and
because most IMF and WorldBankconditionalityrequirementsare not
met by borrowingcountries, it is a good bet that such lending would
simplynot be undertaken.
Now supposethatthe debt-conversionexercise is undertakeninstead.
The contractualdebt burdenis reducedto $600millionper year through
the use of exit bonds. If the country undertakesthe reform effort, it
would reduce currentconsumptionby $100million, but increase future
consumptionpotential by $20 million per year. The foreign creditors
would no longer appropriatethe benefit, since their claims have been
reducedto a fixed $600 million by the debt conversion. As long as the
consumers'rateof time discountis sufficientlylow, the reformwill now
look attractive.The creditor'swelfareis unchangedby the debt conversion, andthe debtor'sis raised.
As a result of the efficiency gains, it would generallybe possible to
structurethe debtconversionto benefitboththe debtorandthe creditors.
For example, the debt worth $60 in this examplecould be swappedinto
exit bonds worth $61 (for example, paying 6.1 percent interest), still
leavingthe debtorwith enoughincentive to carryout the reforms.
The key point of this extended example is that an overhangof debt
creates various inefficiencies. First, it requires continuous and costly
renegotiationof the debt. Second, it raises the specter of a costly
breakdownof negotiation,whichwoulddisruptthe tradingarrangements
of the debtor, and thereby impose costs on both the debtor and the
creditors. Third, and perhaps most important,the debt can act as a
heavy tax on economic reform. Under the current arrangements,the
returnsto reformare appropriatedheavily, if not entirely,by the foreign
creditors. The result is twofold: no individual government has an
incentiveto undertakeadjustments,andpoliticalpartiesthatareopposed
to reformhave an attractivecase to take to the electorate. This latter
concernis farfromabstract.In recentlegislativeelections, the reformist
government of President Alfonsin in Argentina lost heavily to the
Peronists, who have been arguingagainst reforms to increase foreign
debtrepayments,urginga debt moratoriuminstead.
Brookings Papers on Economic Activity, 2:1987
596
A CRUCIAL
CAVEAT
ON EXIT
BONDS
The previous discussion has highlightedthe potential usefulness of
exit bonds by showing how the conversion of all bank debt into exit
bonds could reduce the contractualobligations of the debtor country
while at the same time maintaining,or even raising,the marketvalue of
the resultingclaims held by the creditors.But in a more generalsetting,
the conversion of bank debt into exit bonds may result in a fall in the
value of the creditors' claims ratherthan a rise, thereby undercutting
the case for exit bonds.
As an illustration, suppose that a debtor country will be able and
willingto repay its debt fully if world commodityprices recover (probability 0.6), and will defaultentirely if commodityprices stay the same
or fall further(probability0.4). In the secondary market, $100 of debt
would sell for $60. Now supposethatthe debtis convertedto exit bonds,
carryinga coupon interestrate at 60 percentof the marketinterestrate.
Thecontractualpresentvalueof a $100long-termbondwouldbe reduced
to $60. What would be the new marketvalue of the creditors' claims?
Evidently, the new exit bonds would also have a 60 percent chance of
being fully serviced and a 40 percent chance of being fully defaulted.
Thus, the exit bonds would sell at a 60 percent discount relative to the
contractualobligationsof the bonds. In other words, the marketprice
wouldbe 60 percentof $60, or $36. Obviously,in this case, the creditors
lose substantiallyby giving up their bank claims worth $60 for an exit
bond worth$36.
Therefore,a discounton the bankdebt in the secondarymarketdoes
not mean that creditorscan automaticallybenefit, or at least stay even,
from a conversion of debt into exit bonds. In the examplejust cited,
there is no efficiency gain to makingthe debt conversion, since payoffs
depend purely on the exogenous world commodity price, not on the
policies of the debtor. There is, however, a loss to the creditors, since
with the conversion to exit bonds, the banks give up the chance of
receivingthe full $100 repaymentof their debt in the event of favorable
worldcommodityprices. Technically,partof the value of the creditors'
claims on the country is the option value of sharingin high commodity
prices. When the debt is converted into exit bonds with a lower
contractualvalue, partof that option value is lost.
Jeffrey Sachs and Hariy Hiuizinga
597
Exit bonds thereforehave the followingminuses and pluses fromthe
perspectiveof the creditorsas a group. On the negative side, since exit
bonds reduce the contractualpresent value of futuredebt repayments,
the bankslose the option value of gettingfully or substantiallyrepaidon
theirbankdebtif exogenous events are highlyfavorable.Onthe positive
side, the exit bondsoffer variousefficiencybenefits,which can raise the
marketvalue of the resultingclaims. To reiterate,these efficiencygains
include:avoidingthe costs of continuousdebt negotiations,avoidingthe
chanceof a costly breakdowninfuturedebtnegotiations,andstimulating
economicreformin the debtorcountry.Thebalanceof costs andbenefits
dependson the relativeimportanceof these considerations.
We are inclined to believe that the benefits of debt conversion for
manydebtorcountriesoutweighthe possible costs, thoughour reasons
are necessarily impressionisticand must be considered on a case-bycase basis. The 50 percent discount on bank debt for most countries
reflects the fact that the debt is twice too large to be serviced with
regularity,ratherthanthe fact that exogenous events will resultin all or
no repaymentwith probability0.5. In other words, the option value of
waitingfor full or nearlyfull repaymentis of little value. Moreover, as
we have pointed out, it is likely to be seriously self-defeating,since if
the creditorswaitfor full repayment,the debt overhangwill stiflereform
andtend to bringto power less reformistand more radicalregimesthat
indeedwill choose to suspendall debt repayments.
Put yet another way, to the extent that the creditors really face a
probabilitydistributioninvolving complete, as against no, repayments
on the debt, it is a probabilitydistributionthat they themselves can
influence.A reductionin the contractualdebtburdenthroughsome form
of debtconversionwill bolsterthe politicalstandingof those who would
repaythe debt.
Debt Conversionsand Public Policy
We believe thatthe benefitsof debt conversionwarrantpublicpolicy
action.At the minimum,the regulatoryenvironmentcan be modifiedso
thatarbitrarybook-valuecalculationsdo not standinthe way of desirable
exchanges. But many commentatorshave proposed going further by
using public money to smooth the debt-conversionoperations. In one
598
Brookings Papers on Economic Activity, 2:1987
popularproposal,the debt conversionwould be intermediatedby a new
internationaldebt facility. The facility, which could, for example, be
part of the World Bank, would accept the exit bonds of the debtor
countryand issue its own bonds to the commercialbanks in returnfor
the existing bankdebt, which then would be extinguished.The country
would owe money to the facility in the form of exit bonds. The facility
would owe money to the banks in the form of guaranteedbonds. The
bankswould get a safe claim, the bond of the debt facility, ratherthana
risky exit bond of the debtorcountry.
This proposalis nearlyidenticalto two others. In the first,the official
creditorcommunity,again, perhaps,the WorldBank, would provide a
guaranteeon the exit bonds issued by the debtorcountry.In the second,
the official creditor community would lend the debtor countries the
money necessary to make cash repurchasesof debt in the secondary
market.In all three cases, the contractualburdensof the debtorcountry
would be reducedin line with the discount on its debt in the secondary
market,andthe claimson the debtorcountrywouldeffectivelybe shifted
to the officialcreditorcommunityandaway fromthe commercialbanks.
The cost to the officialcommunitywould be the differencein value of
its own bonds, which are a safe asset, and the exit bonds of the debtor
country.27Suppose, for example, that the country's debt now sells at
$60 per $100. If the facility issues $60 of guaranteedbonds to the banks
and accepts exit bonds from the debtor country with contractualobligations also worth $60, that is, the same coupons and principleas the
safe bonds, the facility's net worth will be reducedif the exit bonds sell
at a discountbecause of defaultrisk. We have given examples in which
the $60 of exit bonds will indeed be worth the full $60, with no costs to
the facility, andcases in which the bonds would be worthonly $36, with
the facility losing $24 in present-valueterms.
Why is such a facility needed at all? The main reason is that a largescale debt conversion poses significantcollective action problemsthat
can best be overcome with official intervention. Bankers' fears about
regulatoryproblems, the legal status of exit bonds, the problem of
contagioneffects whereby terms to one country influencenegotiations
27. The facilitymighteven sell off some or all of the exit bonds in the open marketto
realizethe capitalloss andreducefuturerisks.
Jeffrey Sachs and Harry Huizinga
599
with other countries, and the difficulty of collective decisionmaking
among many banks make it difficult to carry out a large-scale debt
conversionwithoutconsiderableofficialsupport.
The likely costs to the official communityof intermediatinga largescale conversionwould be modest. Suppose that each creditorcountry
participatesin the internationaldebtfacilityinproportionto the exposure
of its banks in the problemdebtor countries. As table 1 showed, U.S.
banks hold approximately$57 billion in claims on governmentsof the
problemdebtorcountries. The secondarymarketvalue of those claims
was some $32billionin July 1987.If the debtfacilitygave the U.S. banks
guaranteedbonds worth $32 billion in returnfor the debt and accepted
exit bondsfromthe debtorcountrieswith contractualobligationsworth
$32 billion, the capital cost to the United States would be the market
discountfrom the $32 billion contractualvalue. At best, the claims on
the LDCs would be worththe full $32billion:therewould be no residual
cost to the United States. At the very worst, the bonds would sell at the
same discount as the originalbank debt, at 55 percent of contractual
value. The claimson the LDCs in thatcase wouldbe worth$17.6billion,
and the transaction would cost the U.S. government $14.4 billion.
Presumably,this capital loss could be amortizedover many years, so
that U.S. taxpayers would end up paying $1 billion to $2 billion each
year for several years. Such costs could be reduced furtherby conditioning the debt relief on economic reform measures in each debtor
country. (We have stressed that the debt relief itself should strengthen
the incentivesfor actuallycarryingout the reforms.)
Even this upper limit of $14 billion seems a modest cost to reduce
LDC debt to the secondarymarketlevels for all thirtyproblemdebtor
countries in table 1. The achievement, from the U.S. perspective, is
considerable:the debt is reduced to levels that the market deems
manageable;the U.S. banks are taken out of the game, and out of risk,
without imposingfurtherlosses; the eliminationof the debt overhang
enhancesthe possibilityof efficiency gains in the debtorcountries;the
political positions of moderates in the LDCs is bolstered; and new
democraticregimesin muchof LatinAmericaandthe Philippineswould
likely be strengthenedby a reductionof theircontractualdebt servicing
obligations.Finally,this kind of relief is an efficientformof foreignaid,
sincethe U.S. contributionswouldalso be matchedby the othercreditor
governments.The Germans, Japanese, British, and other countries
600
Brookings Papers on Economic Activity, 2:1987
Table 18. SecondaryMarketValueof Claimsof FinancialInstitutionson the Problem
DebtorNations
Millionsof dollarsexcept where noted
Secondarymarket
Country
Argentina
Bolivia
Brazil
Chile
Colombia
Debt to
financial
institutionsa
Bid
priceb
(dollars)
Total value
20,395.3
126.3
49,624.7
12,084.8
4,144.2
47
10
55
67
81
9,585.8
12.6
27,293.6
8,096.8
3,356.8
Costa Rica
DominicanRepublic
Ecuador
Gabon
Guatemala
1,530.4
328.4
4,972.5
532
101.1
33
42
45
82
72
505.0
137.9
2,237.6
436.2
72.8
Honduras
Ivory Coast
Jamaica
Liberia
Malawi
164.8
2,486.6
406.5
41.4
53.7
38
60
37
5
74
62.6
1,492.0
150.4
2.1
39.7
Mexico
Morocco
Nicaragua
Nigeria
Panama
58,757.3
2568
1,144.9
6,515.2
1,877.6
53
65.5
5
28
64
31,141.4
1,682.0
57.2
1,824.3
1,201.7
Peru
Philippines
Romania
Senegal
Sudan
3,224.6
4,206.6
2,261.4
233.5
553.6
11
67
87
61
2
354.7
2,818.4
1,967.4
142.4
11.1
Uruguay
Venezuela
Yugoslavia
Zaire
Zambia
1,300.5
9,968.2
4,510.3
402.9
226.5
68
67
70
24.5
18
884.3
6,678.7
3,157.2
98.7
40.8
Total
194,743.8
...
105,542.3
Sources: WorldBank;and SalomonBrothers,Indicative Prices for Less Developed Country Batnk Loatns (July
27, 1987).
a. End of 1986.
b. Bid pricefor a $100claimon the secondarymarketas of July 1987.
Jeffrey Sachs and Harty Huizinga
601
would also be contributingtheir share, so that the United States would
avoid carryingan unduepartof the burden.
The costs, from the point of view of the entire creditorcommunity,
are shown in table 18. As of the end of 1986, the world's financial
institutions, almost exclusively banks, had medium- and long-term
claims on the governments of the problem debtor countries of $195
billion, with a secondarymarketvalue of $105billion. Thus a complete
swap of debts into exit bonds for the thirtycountries in table 18 would
requireofficial guaranteesof $105 billion. At best, the capital cost of
these guaranteeswill be zero: the debtorcountrieswill fully service the
reduced burdenof the exit bonds. At worst, the exit bonds would be
valuedat essentially the same discount as the currentbank debt, about
$54per $100of face value.28The internationalcapitalloss would thus be
on the orderof (1 - 0.54) x $105, or $48 billionfor the entire creditor
community,of whichthe U.S. sharewouldbe approximately$14billion.
28. The discountis slightlydifferentfromthe discountfor the U.S. banksbecause of
the compositionof the globalportfolio.
Comments
and Discussion
John B. Shoven: I liked this paper a lot, possibly because I am not
nearlyas knowledgeableas the authorsaboutthis subject,andtherefore
I learned a lot. The basic story is not a new one. The policy of bank
regulatorshas been to allow banks to carrytheir LDC loans at full face
value and treat the interest flow from them as income, even if that
interest is largely financedby additionalloans. It looks now as though
the object has been to hide the facts from Americandepositors. But I
have been on this BrookingsPanelfor a long time, and a review of our
previous insights on this subjectputs things in a differentlight. Robert
Solomon'sconclusionin the fall of 1981,less thana yearbefore the onset
of the debt crises, that Argentina, Brazil, Chile, Mexico, Peru, the
Philippines,South Korea, and Thailandwere "creditworthy"met with
generalagreementin the discussion. Jeff Sachs noted that only Ghana
and North Korea had repudiatedtheir debt in the postwar period and
that Ghanahad laterrescheduled.Some sympathymay be due the bank
accountantswho were carryingthis debt at full value for so long. Now
thatdefaultand severe concessions have become commonplace,we say
that the regulatorswere trying to hide something. But, at least for a
while, we too could not see the true value of these assets.
Thepaperreviews the exposureof U.S. banks,particularlythe largest
ones, to LatinAmericandebt,gives us the valueof thatdebton secondary
markets,andfindsthatthe implicitvalue of these loans in the U.S. stock
market (found by analyzing the market value of the bank stocks) is
consistent with the quotes fromthe secondarymarket.
For the lay audience, includingmyself, some interestingfacts are
uncovered along the way. First, both shareholderequity and loan loss
reserves count as primarycapitalfor the bank. Therefore,a bank that
recognizes that it has inadequateloan loss reserves and increases them
602
Jeffrey Sachs and Harty Huizinga
603
does not lower its primarycapitalby doing so. I have to agree with the
authorsthatthis seems a littlecrazyandthatregulatorsshouldbe looking
at stockholders' equity in determininga bank's primarycapital. For
manybankstoday, the majorityof primarycapitalis loan loss reserves.
Second, the additionalreserves that Citibankand others have been
setting aside are not allocated to particularloans or even countries.
Therefore, while they reduce reported income, they do not reduce
taxable income. And, from the best that I can tell, almost all of these
banksare taxpayers.This behavioris somewhatcurious.Most of us like
to reportas smallan income as possible to the IRS, andcompaniesoften
reportlower earningson tax account than they reportto stockholders.
Here,just the opposite is happening.The banksare reportinglargeloss
set-asides to their shareholders,without takingthe steps necessary to
reducetaxableincome.
While I learned much from the Sachs and Huizingapaper, I would
like to have learned more. First, I would like to know more about the
secondary market for LDC debt. What is the volume, who are the
participants,what is the bid-ask spread, and so forth? Are the major
buyers (as well as sellers) of the troubledloans the commercialbanks?
Canthe third-worldcountriesbuy back theirown obligationsat 56 cents
on the dollar?I will say more aboutthis in a minute.
Second, after what the authors revealed about bank accounting
practices, I was surprisedto find them using (apparently)book-value
figuresfor shareholders'equity in table 11, where they are examining
the adequacyand composition of primarycapital. Their own table 13
shows that ratios of marketto book value range widely across banks
(from0.5 to 2.7 for the bankslisted). My impressionis that the ratio has
also changedthroughtimeandthattable 11understatesthe improvement
of the last few years in shareholderequity at marketvalue. The problem
of using book figures is illustratedby their June 1987 number, which
shows that shareholders'equity had dropped sharply from year-end
1986.This reflects the loan loss reserves, which arejust a paper entry,
andnot the stock marketvaluationof the equity claimson banks.
Theauthorsfindthatstock investorswere notfooled by the overstated
income statements and balance sheets, but suggest that management
mayhave been, at least in the settingof dividendpolicy. I thinkthey are
a bit naive here. I would be the first to admitthat, althoughthere is no
shortageof theories, we don't know why firmspay dividends. But the
604
Brookin2s Papers on Economic Activity, 2:1987
paymentof dividendsdoes seem to set up expectationsfor theircontinuation. The banks have been aggressively trying to sell new equity.
Several banks are now preparingnew equity issues, and almost all of
them have actively used dividendreinvestmentplansto increaseequity.
Cuttingor eliminatingdividendsin this situationmay appropriatelybe
viewed as a costly strategyto be followed only if absolutelynecessary.
The authorsdiscuss brieflythe idea of an internationaldebt facility
that would shift claims on the debtor countries from the commercial
banks to the official creditorcommunity. The idea seems like a good
one. But if the creditorcountriesparticipatingin the facility offer only
market prices to the banks and borrowers, they are not following an
aggressivepolicy. The governmentsare simplymakingofficialwhat the
marketshave alreadyrecognized.I fail to see why this price is a natural
one for the termsof the debt swap, andfeel that moregeneroustermsto
both lendersandborrowerscould be consideredby the creditorgovernments, obviously at substantialcost to those governments.
GeneralDiscussion
Discussion about the relevance of secondary marketprices on LDC
debt was spirited. Robert Lawrence noted that the use of secondary
marketprices createsa moralhazardby providingdebtorcountrieswith
an incentive to adoptpolicies thatreduce the marketprice of theirdebt.
He suggestedthatrelief shouldbe based on some concept of sustainable
debt, ratherthanon the secondarymarketvalue of LDC loans. Charles
Schultze addedthat, so far as the debtorcountrieshave alreadyanticipated debt relief of the Sachs variety, they may alreadyhave acted so as
to reduce the price of their debt. Thus there is no way to know what
combinationof economic fundamentalsand bargainingstrategy is reflected in present prices. Schultze was also concerned that the use of
secondary marketprices would produce perverse political rewardsby
providingmore debt relief to those countriesthat had done a relatively
poorjob of managingtheir affairs.To the extent that the U.S. governmentwantedto providesome subsidyto the debtorcountries,he argued,
it would be politicallydifficultto do so in a way thatappearedto bail out
the lending banks or that helped debtor countries in relation to those
secondarymarketprices.
Jeffrey Sachs and Harry Huizinga
605
James Duesenberrysupportedthe main policy argumentthat some
form of debt relief is desirable, both for its economic benefits to the
debtornationsandforits foreignpolicy importanceforthe UnitedStates.
But he reasoned that the market value of debt and the value of bank
stocks bear no necessary relation either to the needs of individual
countriesfor aid or to the extent thatbanksshouldbe helpedout of their
problemby the government.He suggestedthat a betterplan for solving
the debt problemmight guaranteea cross-section of debt and provide
relief to debtor nations on a basis that is more closely related to their
individualneeds andto the scope for economicimprovementin response
to debt relief in individualcountries.
Robert Hall extended Sachs's point that the debt relief plan would
entailboth efficiencygains andlosses. At present, bankloans represent
a call option on the LDCs with adverse incentive effects on their
performance:good economic performanceresults in largerpaymentsto
the U.S. banks, while poor performancereduces those payments. He
agreedwith Sachs that a write-downof the LDC debt could reduce this
adverse incentive effect, thus providinga present efficiency gain. But
Hall noted that future lending would be less efficient because lenders
wouldtakeaccountof the fact thatpreviouscontractshadbeen rewritten
ex post. Dwight Jaffee added that debt relief involvingthe sale of bank
loans to thirdpartiesmightreducethe qualityandquantityof LDC loans
in the futurebecause if banksgot rid of manyexistingloans, they would
lose their incentive to continue lending in order to maintainpayments
on outstandingloans.
PaulKrugmancounteredthatwe could conceivablybe on the far side
of the "debt-reliefLaffer curve," in the sense that debt relief reduces
the probabilityof default in the future so that the expected value of
payments on future contracts becomes larger and more secure. He
emphasizedthat that outcome would be far more likely if debt relief
were made conditionalon economic reforms, as Sachs suggests. However, Georgevon Furstenbergquestionedwhether any new facility for
purchasingdebt in Sachs's scheme could enforce such economic reforms, observing that the IMF and World Bank are unable to do so.
Sachsreemphasizedthatremovalof the presentdisincentivesto institute
economic reforms-the banks' call options to which Hall referredwould make the future different from the present, and that further
efficiency gains would come from reducingthe dead-weightloss now
associatedwith bargainingbetween debtorsand lenders.
606
Brookings Papers on Economic Activity, 2:1987
Several participantsquestioned the regression results relatingdebt
exposureto the marketvalue of banks. BenjaminFriedmanwarnedthat
regressionsexplainingthe excess returnson bank stocks by their LDC
debt exposure would be biased if that exposure itself affected the beta
coefficient that was used in forming the excess-returns variable. The
banks' beta could be affected in this way because improvinggeneral
economic conditions would improve the prospects of debt repayment
andthusbenefitbankstocks disproportionately.JamesPoterbareasoned
that the relationbetween bankexposure to LDC debt and marketvalue
or excess returnswould be exaggeratedby otherportfoliodecisions that
banks may have been induced to take because of their large LDC debt
exposures. Similarly, von Furstenbergreasoned that the relation of
marketvalue to book value of bankswith and withoutlargeexposure to
LDC debt could reflect long-standingdifferences in lending behavior
between majormoney-centerbanksand others.