Economics Letters North-Holland
29 (1989) 237-241
CONTAGION
EFFECT ON SOVEREIGN
237
INTEREST
RATE SPREADS
John DOUKAS Concordta Uniuerstty, Montreal, Que., Canada H3G 2M8 Received Accepted
5 March 1988 25 April 1988
This paper analyzes the extent to which sovereign default risk in contagious. That is, contagion determine whether ‘news’ regarding the creditworthiness of a sovereign borrower affect the spreads empirical results support the view that a systematic risk element enters into the default risk perceptions existence of significant contagion effects in international lending.
tests are conducted to charged to others. The of lenders implying the
1. Introduction In recent years, much progress has been made in developing and understanding the complexities surrounding international debt. Several tests have been conducted ’ and intuitive economic interpretations have been provided in explaining risk premiums (or spreads over LIBOR margins) charged on sovereign loans in the Euro-credit market. 2 The emphasis on risk premiums stems from the notion that sovereign loans must be priced to reflect new information about the borrower that is made available to lenders in every period. In particular, it has been postulated that sovereign risk premium changes - or changes in spreads over LIBOR margins - can be explained by the arrival of ‘news’ about the borrower in the market place. This proposition has recently been empirically investigated by testing whether developing countries’ foreign debt has been efficiently priced in the sense that bank sovereign risk assessments were fully reflected in the pricing of foreign debt with different degrees of success. One of the main restrictive characteristics in most of these studies has been the assumption that new information about a borrower’s state of creditworthiness does not affect lenders’ subjective probabilities regarding the default risk of other borrowing countries. That is, absence of contagion effects in the international loan market. In this paper we examine the extent to which sovereign default risk is contagious. Thus, our paper is primarily motivated by the question, does new information about a borrower’s state of creditworthiness alter lenders’ subjective beliefs regarding the probability of default risk of other borrowing countries?
2. Empirical results Following Doukas and Jalilvand (1986), lenders’ assessment of sovereign risk, perceived probability of debt repudiation, is inversely associated with the burrower’s future (expected) productive ’ See Feder and Just (1977) Feder and Ross (1982) and Edwards (1984) among others. * See, for example, Cline (1983). Doukas and Jalilvand (1986). Eaton and Gersovitz (1980). and Saunders
0165-1765/89/$3.50
0 1989, Elsevier Science Publishers
B.V. (North-Holland)
(1986).
J. Doukas / Conragion effect on sovereign interest rate spreads
238 Table 1 Estimation
of output
equation
Independent variable
(1). a Country Argentina
Constant
dP,
Brazil
4.0134 (12.6810)
5.6294 (10.0001)
2.9130 (4.8304)
3.4291 (2.1015)
6.2080 (2.2432)
3.1980 (2.2045)
- 1.1150 (- 2.1827)
- 0.4427 ( ~ 2.1049)
-0.5107 (-2.3145)
dP,z
Mexico
d%
0.6432 (2.2814)
0.7103 (2.9082)
0.3219 (2.2014)
dq,mi
0.3118 (2.4416)
0.5425 (2.7320)
0.3027 (2.0045)
MSE
0.025
0.0304
0.0321
a Asymptotic
t-statistics
are shown in parentheses.
MSE is the mean square
error of the estimated
regression.
growth opportunities that the borrower may possess. This inverse relationship between lenders’ perceived probability of default and the borrowing country’s future growth opportunity set (implicit collateral in international lending) is obtained by treating default as a rational financial decision that determines the borrower’s default (or repayment) option on a syndicated Euro-loan. To assess the expected growth opportunities of a borrowing country we measure a borrower’s economic growth using Doukas’ (1987) output adjustment equation derived from a macroeconomic model following Bruno’s (1979) eclectic macroeconomic model which provides a general framework for analysis of stabilization policies in less developed countries. The output adjustment equation is 4, = PO+ PI dp, + PZ dp,, + P3 dm, + P4 dq,-t
+ e,,
(1)
where qr is the domestic production index 3, JI, is the domestic price level, p,, is the domestic price of imports, m, is the nominal money supply and C, is the error term. The output regression results reported in table 1 are quite satisfactory from both the point of view of the MSE and from the perspective of signs and the level of significance of the coefficients. To test the contagion effect hypothesis a straight forward test is performed by fitting the following equation: n=3 Ds:=P,+
c
P,d+~t>
(2)
/=I
where DS: measures the observed monthly change in the syndicated Eurocredit spread from country (before and after the announcement of q figures), q, represents the unexpected component of This measure on domestic productions is obtained based on an equally weighted average commodities produced in each country. The commodities included were as follows: (i) Argentina (meat, casu, wheat, hides and skins and wool), (ii) Brazil (coffee, soybeans and products, sugar and iron ore), (iii) Mexico (petroleum, cotton, sugar, coffee and shrimp).
procedure
of the major exportable
J. Doukas / Contagion effect on sovereign interest rate spreads Table 2 Reaction
of Euro-syndicate
interest
rate spreads
to new information.
Borrowing countries
Estimated oo
a1
1978-1983
Brazil
0.214 (1.002)
- 0.083 (- 1.125)
1978-1983
Mexico
0.820 (1.256)
1978-1982
Argentina
0.325 (1.157)
Period
a qe is the expected parentheses.
and
qu is the unexpected
a
coefficients
component
239
Summary
statistics
R2
DW
- 0.618 (- 3.452)
0.092
2.14
- 0.149 ( - 1.074)
- 0.485 (- 3.540)
0.035
2.11
- 0.101 (-1.110)
- 0.652 (3.137)
0.044
1.97
of ‘news’
for the borrowing
country:
t-statistics
are shown
in
‘news’ for country j obtained from the residuals of the estimation of eq. (1) reported in table 1, and cr is the error term. Eq. (2) permits to examine, for instance, how ‘news’ for Brazil and Mexico affects the Argentinian interest rate spreads charged in the Euro-credit market. To the extent that the spread over the LIBOR charged on syndicated Euro-loans reflects the true risk associated with lending to LDCs [see, for example, Feder and Just (1977), Eaton and Gersovitz (1980), Feder and Ross (1982) Edwards (1984) and Saunders (1986), among others] estimation of the above equation is expected to show whether a systematic (undiversifiable) risk element enters into the default risk premia in international lending are perceptions of lending institution. 4 That is, whether default-risk independently determined by country-specific events or in conjunction with events originating in other borrowing countries. Therefore, discovery of a high degree of contagion would imply a high level of systematic, or undiversifiable, risk in international lending. Conversely, a low level of contagion would imply that sovereign risks are basically independent or unsystematic events that can be minimized by holding a geographically diversified portfolio of sovereign loans. Under the assumption that the international lending market is efficient it is hypothesized then that only the unexpected movements of economic growth (or ‘news’), qi, impact on the sovereign credit market causing risk premia changes, DS’. Monthly data on spread merging for public and publicly guaranteed Euro-syndicated loans granted to Brazil, Mexico and Argentina were obtained from various issues of the Euromoney magazine’s Market Commentary: Syndicated Loans. 5 Because Brazil, Mexico and Argentina were the only countries that borrowed funds from the Euro-syndicated credit market almost every month during our sample period, we did not include other sovereign borrowers. 6 The spread series spanned the period from January 1978 to December 1983 for Brazil and Mexico and from January 1978 to May 1982 for Argentina. The sample considered in this study contains only LIBOR priced syndicated medium-term loans denominated in U.S. dollars. ’ Table 2 presents the results obtained from the independent estimation of eq. (2) for Brazil, Mexico and Argentina. In general, the empirical evidence confirms the view that international interest Folkerts-Landau (1985), however, has argued that yields on LDCs bonds do reflect the true risk associated with lending to the LDCs. Recently, Edwards (1986) found no substantive differences between bond yields and bank spreads on a country-risk determination study. The data were originally obtained from the Euromoney Syndication Guide. For the few months that these countries did not borrow from the Euro-syndicated credit market we used the previous month’s spread. Concerning the currency denomination problem on international lending see McDonald (1979, p. 630).
240 Table 3 Estimation
J. Doukas / Contagion effect on sovereign interest rate spreads
of spread
equation
(2) for contagion
effects in Euro-syndicate
interest
rate spreads.
a
n=3 W=Po+
c ,=I
P,q:+cr.
Borrowing countries
Period
Estimated
coefficients
Summary
PO
PI
P2
statistics
Px
R2
DW
1978-1983
Brazil
0.097 (1.312)
- 0.574 B (- 2.260)
- 0.162 A ( - 2.010)
-0.240 M ( ~ 1.967)
0.101
2.18
1978-1983
Mexico
0.349 (0.871)
-0.426 M (- 2.873)
- 0.185 A (- 1.983)
-0.227 ” (2.159)
0.072
2.07
1978-1982
Argentina
0.185 (1.206)
- 0.541 * (- 2.027)
B -0.217 ( - 2.270)
-0.144 M ( - 2.003)
0.095
2.16
a &statistics
are shown in parentheses.
B = (qpnl),
A = ( qfrge”tl”d), M = (4,yex’co).
spreads, across major sovereign borrowers, are strongly associated with new information characterizing these countries’ state of creditworthiness. As may be seen, in all regressions only the unexpected component of borrower’s growth opportunities aggregate, q,, is significant and with the expected sign. Consistent with the market efficiency hypothesis, these results suggest that sovereign risk considerations are systematically incorporated in the process of Euro-syndicate credit pricing. Table 3 presents the results obtained from the estimation of equation (2) for the three sovereign borrowers simultaneously. These results are quite interesting. First, the coefficient of the countryspecific (&) ‘news’ variable remains significant at conventional levels. It is interesting to note that the point estimate of this coefficient appears to be, in general, slightly lower but significant. Another more interesting finding is that a systematic risk element enters into the default risk perceptions of lending institutions indicating the existence of significant contagion effects in international lending. As table 3 shows, the non country-specific coefficients, & and &, are always negative, as expected, and in all cases significant. These results imply that individual country interest rate spreads are influenced not only by the creditworthiness of the borrowing country itself but also by the creditworthiness of other borrowing countries. This finding, in fact, suggests that, as discussed above and elsewhere [see Carron (1982) and Saunders (1986)], sovereign loans are not independently priced. 3. Conclusions This paper has examined the interaction between country interest rate (over the LIBOR) spreads and the arrival of ‘news’ about sovereign borrowers’ creditworthiness in the market place. That is, contagion tests were conducted to determine the extent to which interest rate spreads are systematically affected by a common risk element. The results provide evidence supporting the contagion effect hypothesis which asserts that a considerable systematic risk is embedded into the default risk perceptions of lenders, and therefore, this risk may not be minimized by holding a geographically diversified portfolio of sovereign loans. References Carron, A., 1982, Financial Activity 2, 395-453.
crises:
Recent
experience
in U.S. and
international
markets,
Brookings
Papers
in Economic
J. Doukas / Contagion effect on sovereign interest rate spreads
241
Cline, W., 1983, International debt and the stability of the world economy (Institute for International Economics, Washington, DC). Donaldson, T.H., 1979, Lending in international commercial banking (Halsted Press, New York). Doukas, J., 1987, LDC stabilization policies, price and output adjustment: Theory and evidence, 1947783, Applied Economics 19, March, 447-458. Doukas, J. and A. Jalilvand, 1986, Sovereign risk and international lending: Theoretical and empirical analysis, Studies in Banking and Finance 3, 131-145. Eaton, J. and M. Gersovitz, 1980, Debt with potential repudiation: An empirical analysis, Review of Economic Studies 7, March, 3-21. Edwards, S., 1984, LDC foreign borrowing and default risk: An empirical investigation, 1976-80, American Economic Review 74, Sept., 726-734. Feder, G., and R.E. Just, 1977, An analysis of credit terms in the Eurodollar market, European Economic Review 9, May, 221-243. Feder, G. and K. Ross, 1982, Risk assessments and risk premiums in the Eurodollar market, Journal of Finance 37, June, 679-691. McDonald, D., 1982 Debt capacity and developing country borrowing: A survey of the literature, IMF Staff Papers, Aug., 603-646. Saunders, A., 1986a, The determinants of country risk: A selective survey of the literature, Studies in Banking and Finance 3, l-38. Saunders, A., 1986b, An examination of the contagion effect in the international loan market, Studies in Banking and Finance 3, 219-247.