When it rains, it pours: Mexico's bank nationalisation and the debt crisis of 1982

Abstract How are expropriations related to governments' debt defaults? The literature has shown that expropriation episodes and debt defaults have rarely coincided, suggesting that each event resulted from a different set of factors. The aim of this article is twofold. First, I analyse default–expropriation relationships in the years previous to the debt crisis of 1982. I show that while default and expropriation episodes did not always coincide, countries that expropriated at least once during the period were also those that defaulted more often. I observe that countries that expropriated had worse macroeconomic indicators than countries that did not. Second, I focus on the case of Mexico, when its announcement of a debt moratorium in August 1982 was followed, less than one month later, by the nationalisation of its banking system. Both events were outcomes of an acute economic crisis. The nationalisation announcement aggravated the crisis because an agreement with the International Monetary Fund (IMF) seemed increasingly uncertain. I provide evidence from the largely overlooked bond market (on which the government never defaulted) that shows that investors reacted negatively to the bank nationalisation. Finally, I present original, published and unpublished primary sources to demonstrate that commercial banks, as well as international organisations, expressed misgivings about the banks' nationalisation. This fact may have hindered the country's economic recovery through the deterioration of public confidence and a decline in foreign investment.


Introduction
World economic downturns are frequently accompanied by waves of government debt defaults and, in extreme cases, expropriations.Last year, The Economist estimated that fifty-three emerging economies were on the brink of default, while expropriation risk is estimated to be on the rise1 .The literature on sovereign debt has analysed the links between both types of events.In a nutshell, such government policies are grouped into the broad concept of «sovereign theft», where a government decides unilaterally to reduce investors' returns 2 .While scholars have developed theoretical models and established broad statistical correlates, we know less about how these kinds of events have been interlinked in the past.In this article, I revisit Mexico's debt crisis of 1982.I show that the loss in terms of reputation in financial markets due to the banks' expropriation was higher than previously assumed and affected the gravity of the crisis.Using archival material, contemporary press articles and government bond prices, I find that the government's negotiations with the International Monetary Fund (IMF) were delayed, resulting in an upsurge in default risk, which may have had an impact on investment and, therefore, on the country's economic recovery.
Mexico's debt crisis has played a prominent role in the history of financial crises.It marked the beginning of the so-called Latin American debt crisis, whereby many governments from low-and middle-income countries experienced difficulties repaying their debts 3 .The early 1980s also constituted a turning point in the volume of foreign capital flowing to developing countries.Since the mid-1970s, Latin American governments borrowed large sums of capital from syndicates of western commercial banks, via the Euroloan market.On 20 August 1982, Mexico's Finance Minister Jesus Silva-Herzog publicly announced a 3-month moratorium on all amortisation payments due to bank loans to the public sector.According to the literature, the fragile debt position of the government was largely unanticipated 4 .The announcement prompted the IMF, the Bank for International Settlements (BIS) and the U.S. government to adopt measures to avert a crisis that had the potential to severely affect the United States and, by extension, the world The reminder of this article is organised as follows.In the first section, I introduce the concept of sovereign theft, and argue that after defaulting, the probability of expropriation increased, given both the poor economic performance of the country and the costs in terms of external reputation.In section 2, I provide a brief overview of the literature on Mexico's debt crisis and the bank nationalisation.I show that previous research has not adequately analysed the consequences of nationalisation on Mexico's risk premium and on the debt crisis.In section 3, I analyse the relationships between sovereign defaults and expropriation episodes internationally.I pursue a set of statistical tests to show that the macroeconomic performance of expropriating and non-expropriating countries was dissimilar and demonstrate that expropriating countries performed worse in most variables.In section 4, I provide evidence of the impact of nationalisation on negotiations with the IMF and on default risk.Section 5 concludes.

Sovereign theft: A Theoretical Framework
Historical evidence shows that episodes of expropriation and sovereign defaults rarely coincide.However, the number of studies analysing these relationships is extremely limited.Tomz and Wright (2010) find that between 1929 and 2004, sovereign defaults and expropriations occurred in waves, even though they did not coincide in time.They also posit that at a country-level analysis, two groups of countries could be distinguished, representing 70 per cent of their sample.One group of countries had neither expropriated nor defaulted, whereas the other group was formed by countries that had defaulted and expropriated at least once, even if not simultaneously.Similarly, Eden et al. (2012) found that countries with a past record of expropriations have a higher likelihood of future expropriations and defaults.These authors claim that expropriations are prone to take place during «good times».Tomz and Wright (2007) showed that countries default during bad times, even though this relationship has been historically weak.
Tomz and Wright (2010) introduced a model in which a government's incentives to engage in sovereign theft depend on the state of the economy, the risk aversion of political leaders and the types of penalties stemming from defaulting and retaliation.The relevance of a retaliation in influencing a government's behaviour depends on the severity of the penalties from foreign creditors (in the case of a default), from foreign investors (in the case of an expropriation), and whether these penalties are shared across creditors and investors.The analysis contemplates a sequential model in which investments occur before a government receives a loan, whereas loans are repaid before profits are distributed.Thereafter, the government first decides whether to default or repay the loan.At a second stage, the government must decide whether to expropriate the earnings of the direct investors.The resulting prizes for the government hinge on its utility function, whose parameters depend on the prevailing costs and benefits of defaulting and expropriating.
First, we consider the benefits of default.The resources that a government saves from retained debt service can be used for other purposes.According to Bértola and Ocampo (2013), this was why many governments in Latin America decided to default at the onset of the Great Depression, as they could invest in domestic infrastructure and the provision of credit to boost economic activity.Generally, a decision to default depends on both economic and political factors.A government could be incentivised to divert the resources devoted to servicing its external debt to boost economic activity, thereby increasing the probability of its re-election (or increasing its social base).In this regard, the proximity of elections has been identified as a variable positively related to sovereign defaults6 .
Nevertheless, sovereign defaults are expensive 7 .A government that defaults is excluded from capital markets and will face higher borrowing costs in the future (Suter 1992;Tomz 2007;Catão and Mano 2017).Other types of costs include a decrease in international trade and, in extreme cases, a loss of sovereignty through the establishment of supervisory mechanisms or through the use of gunboat diplomacy8 .The capacity to sanction depends upon the identity of the lender and whether their coordination mechanisms can exclude a government from accessing new funds.Amador (2004) argues that unstable countries are less capable of evading the costs of default through increased savings, and thus, they face a persistent need to borrow externally.Therefore, and rather counterintuitively, these countries would be more prone to avoiding default.
Reputation plays a key role in the literature on sovereign defaults.A default may have a negative impact on a government's creditworthiness, which implies that borrowing could become more costly in the future (Catão and Mano 2017).One may ask how these elements differ from the decision to expropriate.Under the assumption of incomplete information, foreign investors do not know ex-ante the preferences of government.An expropriation may have an impact on their perception of the value that a government grants for future loans.Foreign investors interpret an expropriation as a government's least favourite preference for maintaining good relations with them.In turn, these perceptions may have a negative effect on the expected value of potential future projects, thereby triggering a decline in the levels of a country's foreign investment.
Overall, the cost-benefit calculation behind a government's decision to default does not entirely differ from the decision to expropriate.Two major benefits can be derived from expropriations.In equity contracts, expropriations can be directly related to business cycles.On the one hand, recessions can be propitious periods to expropriate.In these cases, a government may decide to expropriate out of «desperation» when public revenues decline and the need to increase spending mounts.On the other hand, an upward trend in the business cycle can also increase the temptation to expropriate.In those cases, «opportunism» may lead a government to expropriate, as the amount to be gained from expropriation is at its highest level.Finally, a government can gain control over operations, projects and transfer factors of production from foreign investors to the host country (Tomz and Wright 2010).Nevertheless, there may be risks if a firm is vertically integrated with other firms internationally, which may raise other difficulties.
A final question concerns the timing of expropriations and sovereign defaults, and whether they should coincide or be spaced out.Either action may trigger a negative perception of a government, which would then be qualified as «unreliable».Cole and Kehoe (1998) introduce spillover effects to show that debt defaults can affect other arenas.These authors illustrate their theoretical model with the 1985 decision of Peru's President Alan García to default and pursue a series of nationalisations during his stay in office.After defaulting, he expropriated foreign oil companies and, a couple of years later, banks and insurance companies.Peru's initial reputation was severely damaged and capital flight reacted very rapidly, so the cost of expropriation, along with foreign investment, declined sharply as investors relocated their assets abroad.The basic message of this model is that a government might minimise its costs if it decides to default and expropriate simultaneously.
Cole and Kehoe's (1998) analysis is motivated by the relevance of a government's reputation as a major incentive to repay its foreign debt, and places sovereign debt on the same foot as other arenas in which trust in a government is important.Contrary to Tomz and Wright's (2010) model, the sequence of events is not relevant.The basic assumption that both models share is the existence of spillover effects, even if the nature and intensity of the costs and benefits from either default or expropriation differ.In this regard, according to Cole and Kehoe (1998), the sequential narrative of events is secondary.A direct implication of the model is that an expropriation might have an impact on default risk, as creditors assume that a government is not trustworthy and may default in the future.
A major question that emerges is whether these models can be applied to Mexico's 1982 crisis.Previous works on Mexico's expropriation decision have emphasised the country's political context, including Mexico's authoritarian regime.While none of the models on sovereign theft mention the role of political institutions, a large body of literature shows that the level of a country's expropriation risk depends on political variables, including the political regime and the political orientation of the government in power9 .Therefore, rather than contradicting these works, the economic costs of defaulting and expropriation emphasised in the models of sovereign theft can be seen as complementary to the political context under which expropriation took place.

Mexico's Debt Crisis of 1982 and its Banking Nationalisation
Mexico's debt crisis has been analysed from a more general perspective in Latin America and in the context of increased financial integration10 .Broadly speaking, the literature has provided three major categories of reasons for Latin America's debt crises.The first group of reasons relates to the functioning of international financial markets.Scholars favouring this explanation posit that the years preceding the crisis had high international liquidity and increasing current account problems in developed countries, prompting Western commercial banks to increase their lending to developing countries through the Eurodollar market11 .This perspective suggests that the beliefs system was shared by regulators, governments in creditor countries and international organisations such as the IMF and the OECD.It has been echoed by various scholars, including Kahler (1985), Wellons (1985), Devlin (1989) and Altamura and Flores Zendejas (2020), who situate the role of politics and the existence of creditor countries' implicit support for their banks as the main factors boosting the 1970s lending boom.
A second group of explanations focuses on macroeconomic imbalances in Mexico and, more generally, in Latin America.Cline (1983) emphasises the general increase observed in debt levels, differentiating between non-oil and new oil exporters such as Mexico12 .This group of theories has been challenged by scholars such as Sachs (1985) or Bértola and Ocampo (2013) because these imbalances only increased in the year before the crisis.Furthermore, these authors claim that macroeconomic variables do not accurately differentiate countries that defaulted from those that did not.A final group of theories that analyses the relationships between Northern and Southern countries argues that changes in world economic conditions, including declining terms of trade and a general rise in interest rates, lie at the heart of Latin America's debt crisis (Diaz-Alejandro et al. 1984) 13 .How is the debt crisis related to the bank nationalisation in Mexico?For certain protagonists who have published their experiences, Mexico's bank nationalisation was a consequence of the economic and debt crises14 .However, this deteriorating macroeconomic context is not the only factor behind the decision, and scholars have provided other, personal and political reasons15 .The decision by President López Portillo to blame the banks for Mexico's economic crisis-summarised in the now infamous phrases «They have looted us.Mexico is not finished.They will not loot us again»-has been studied from different angles 16 .One concerns the roots of the government's decision.Turrent (2009) emphasised the role of an inner circle of economists who favoured the establishment of a planned economy, for which the banking sector played a key role.One economist advising the president was Carlos Tello, who has emerged in the literature as both a scholar and a major protagonist who became the head of the central bank in the aftermath of nationalisation.Tello (1984) describes the need for nationalisation given the necessity of ending the high concentration of the banking sector, deemed as being strategic from a developmentalist perspective.
A predominant perception of the bank nationalisation emphasises political strugglesalso influenced by ideological disputes-within the context of an economic crisis.Elizondo Mayer-Serra (2005) identifies three major dimensions behind the president's move.The first dimension was ideological and concerned the role of the state in the economy and the need to continue the direction established during the Mexican Revolution, when banks had been considered an enemy.The second dimension is financial, where capital flight could only be averted by the bank nationalisation and the imposition of exchange controls.Domestic banks were blamed for fostering capital outflows in the months prior to the crisis, thereby contributing to currency devaluation.Finally, Elizondo Mayer-Serra (2005) identified a political (and personal) dimension, whereby the president's intention was to show that he remained in power and, therefore, was still able to act against those who had originally speculated against the currency.
From a long-term perspective, Elizondo Mayer-Serra (2001) and Haber (2006) underscore the negative impact of an economy with poorly defined property rights on the banking sector's performance.In the same vein, Del Ángel and Martinelli Montoya (2009) argue that even if the decision to nationalise might have been a mistake, it was not irrational.These authors posit that relations between the government and banks had been tense since at least the early post-revolutionary period.As a result, the government was able to eventually expropriate without costs.The question is why, given the banks' vulnerability, the expropriation did not occur earlier.According to these authors, the banks' expropriation served to consolidate the power of an authoritarian regime such as the one prevailing in Mexico.One reason banks had not been expropriated before was because the government could extract rents from the private banks, but once this was no longer the case, such a policy could be implemented because other social groups also perceived it as necessary to confront the crisis.Maxfield (1990) also emphasises the conflict between the government (in particular, the national populists) and a bankers' alliance, which encompassed a financial-industrial conglomerate that had profited from increased integration with the world economy, thereby hindering the state's capacity to allocate bank credit.For this scholar, the president was «acorralado por los hechos» (cornered by events) and undertook a «defensive, last-resort measure» (Maxfield 1990, p. 142).She perceives the nationalisation and exchange control policies as failures as they did not break the power of the financialindustrial conglomerates and did not solve the state's financial situation.
One of the reasons for nationalisation is that it was the government's reaction to a looming banking crisis (Del Ángel 2005).Such a claim is not surprising given that sovereign defaults and banking crises have been interlinked in the literature on sovereign debt.Scholars such as Reinhart and Rogoff (2011) have argued that domestic banks may be particularly exposed to the risk of default by their own governments, as this would severely affect their portfolios.However, Haber (2006) posits that this was not the case in Mexico, as the central bank bought treasury-issued bonds.Banks may have been affected by the devaluation of the currency.In fact, currency crises could also have a negative effect on banks that have been exposed to currency risk through international transactions 17 .
Various scholars have supported the idea that bank nationalisation was an attempt by the Mexican government to bail out its banks.Elizondo Mayer-Serra (2001, p. 179) posits that international bankers received the news with «relief», as the government would be responsible for the banks' debts.According to Alvarez (2019), Mexico's banking system was fragile in the months before the crisis.To a large extent, this was not very different from the position of other banks in Latin America, including Brazil and Chile, which would experience a severe shock during the debt crisis.Mexico's government adopted a supportive attitude towards its banks, particularly those with foreign branches and who were active in the interbank market.Furthermore, these banks also had a large exposure to the country's public debt.Alvarez (2019, p. 185) reported that the largest commercial banks had foreign loans that represented from three to ten times their total capital.This condition weakened Mexico's bargaining position in its quest for a debt rescheduling.Alvarez (2019) suggests that foreign creditors did not necessarily see the Mexican nationalisation as a problem because the banks' liabilities would become part of the government's public debt.This new landscape reduced the risk of exclusion from interbank credit and facilitated the negotiations with the IMF.
While Mexican bankers complained that their international counterparts did not raise their voices to show solidarity with such an arbitrary policy, this perspective needs to be qualified, to say the least18 .First, Tello (1984, p. 126) admitted that he did not expect the decision to be popular among the international financial community or the U.S. government.Second, contrary to the supposed «relief» by international bankers, Boughton (2000, p. 301) reports that there was a panic in the interbank market on 7 September, 1 week after nationalisation, as international banks refused to roll over lines of credit to Mexican banks, severely undermining their position.Third, President López Portillo himself declared that the plan to nationalise the banks was not intended to rescue the bankers or to guarantee the repayment of Mexico's private debt, but, rather, to reduce the speculative bubble that started in the early months of 1982 (López Portillo 2008, p. 266).Finally, the government could have committed to guaranteeing the banks' debts, as it eventually did, without nationalizing the banks.As we will demonstrate, evidence from the press, the IMF and bond risk premia show a negative reaction to the nationalisation.

Expropriations and Debt Defaults, 1966-1982
What is the empirical evidence of the default and expropriation episodes for the year before the crisis?We divided the sample between countries that expropriated at least once and those that did not.A first statistical overview shows that the first group (expropriators) defaulted 7 per cent of the time, whereas the second group (non-expropriators) defaulted 2 per cent.A Welch two-sample t-test shows that the means' difference is significant 19 .A more precise perspective of this relationship is provided in Figure 1, where we depict the correlation between the frequency of defaults and the frequency of expropriations between 1966 and 1982 20 .We divided the full set of countries into two groups using the K-means clustering technique (James et al., 2013, p. 523).We utilised an algorithm to determine the point at which the distance between the points was minimised to a centroid.The algorithm then implements successive iterations until the optimal number of clusters is defined.We repeated this procedure with bootstrap and using the Within Sum of Square (WSS) specification and found that the optimal number was two clusters 21 .
Figure 1a shows the cluster of countries with a higher number of expropriations and defaults, including Algeria, Peru and Mexico.It is noteworthy that the relationship between both variables is negative even if the slope is not pronounced.Figure 1b shows that at lower levels of expropriations and defaults, the correlation between both variables is positive.We added 95 per cent confidence intervals as shaded areas to account for the low number of observations.Figure 1 suggests that above a certain number of sovereign theft episodes, the drivers of defaults and expropriations might converge, even if we only computed three cases of expropriation and defaults during the same year (Chile in 1972and 1973, and Ghana in 1974).
Mexico's case suggests that the state of its economy could have been highly relevant in explaining its default and expropriation events in 1982.They occurred at the onset of a downturn in the business cycle.The country had been experiencing a decline in growth rates and a squeeze in public finance.The loans contracted by the government had variable rates and were adjustable every 6 months.Therefore, in a period of rising interest rates in the world economy, the debt burden increased, and the incentive to default escalated 22 .This condition also coincided with a decline in commodity prices, particularly oil prices, which reduced public revenues sharply.The crisis accelerated after an initial devaluation in February 1982, with a continuous fall in international reserves.Therefore, given the sequence of events, the default should have increased the probability of expropriation as a «desperation» reaction. 19The value of the t test is 4.22, and the p-value is 0.00 (the difference in the means' confidence interval at 95% is [0.02 0.06].We repeated the same analysis with subsamples (1970-1982) and 1976-1982).Results do not differ greatly (available upon request). 20Frequencies are represented as the number of years where there is a default or an expropriation event over the total number of years reported for each country.Data on expropriations are from Tomz and Wright (2010), who adopt a definition of expropriation that includes nationalization of foreign firms, coerced sales, interventions or requisitions and renegotiations, on which a government compels direct investors to «accept substantial changes in a contract or negotiations» (p.98).Data on (external) defaults are from the Online Appendix of Catão and Mano (2017).They define a default as «a unilateral interruption of repayment of interest and/or principal on contractual debt obligations by a sovereign government» (p.94).This data was complemented with Reinhart and Rogoff (2011), who use the same definition of default.We begin from the year 1966, because expropriation data is almost complete after that year.The total number of country-year observations is 1,078.
21 This is also an optimal result given the low number of observations.A correlation analysis between both variables for the whole sample displays a correlation coefficient (ρ value) of 0.39 and a t-test value of 13.84 with a confidence interval between [0.33 0.44] at 95% confidence interval. 22See Devlin (1989) and Rockerbie (1993).
Revista de Historia Económica / Journal of Iberian and Latin American Economic History What is the role of macroeconomic indicators in the frequency of sovereign theft episodes?We now focus on the difference between countries that expropriated at least once and those that did not during the years before the crisis (1977)(1978)(1979)(1980)(1981)(1982)  23 .We compared the means of a group of representative indicators and pursued a mean equality test (t-test) between both groups of countries for the following variables: economic growth (real per capita GDP growth), level of public debt (debt to GDP ratio), budget balance (as a ratio to GDP), current account balance (to GDP), level of public investment (to GDP), inflation, reserve to money supply ratio, average nominal exchange rate depreciation (percentage changes of domestic currency to the U.S. dollar) and risk premia (spreads) of syndicate bank loans 24 .
Table 1 presents the results.On average, expropriating countries display lower economic growth, higher debt, lower public investment, higher inflation rates and larger fiscal deficits.Their currencies also exhibit higher levels of depreciation to the U.S. dollar.The means' differences in reserves, exports and current account balances were not statistically significant.Figure 2 presents the evolution of each indicator divided between the expropriating countries (group 1) and the rest (group 0).Notably, the differences in most indicators are persistent and even widen in certain cases between both groups of countries (in particular, inflation, exchange rate depreciation and public investment levels).
Are there any spillover effects from expropriations on default risk?An expropriation might lead to worsening government borrowing terms if we assume that the government's reputation for foreign investors also affects the debt market.While an international panel analysis of sovereign risk premia goes beyond the scope of this article, Figure 3 provides a preliminary overview.We tested whether there is a significant correlation between the expropriation ratio and risk premia on government loans in 1981 for Source: See text. 23The sample of countries is reduced due to data availability.Countries included in this analysis are the following: Algeria, Argentina, Bolivia, Brazil, Chile, China, Colombia, Costa Rica, Cyprus, Ecuador, Gabon, Greece, Hungary, India, Ivory Coast, Jamaica, South Korea, Malaysia, Mauritius, Mexico, Morocco, Niger, Nigeria, Pakistan, Panama, Papua, Peru, Philippines, Poland, Portugal, Romania, South Africa, Spain, Thailand, Tunisia, Uruguay, Venezuela, and Yugoslavia. 24Indicators included in the analysis were based on data availability.Our data comes mainly from the World Bank (1987) which published the data for developing countries.Missing data was completed with (International Financial Statistics Yearbook, 19861986).Spreads are from Rockerbie (1993), who provides a weighted, yearly (average) spread measure that considers interest rate, amount, and maturity of each loan.Original source is the Euromoney Magazine.For a discussion on spreads in this period, see Negrete Cárdenas (2000).
countries that expropriated at least once in the period considered.We first implemented the same cluster analysis as in Figure 1 and show the results for two clusters.Figure 3a depicts cluster 1, which includes countries with more frequent expropriation events.The relationship is negative, with a correlation coefficient of −0.28, even though the t-statistic is not significant, suggesting that at a certain level of expropriations' frequency, spreads do not increase.On the contrary, Figure 3b shows that for countries that did not expropriate very frequently, this relationship is positive (even if the correlation coefficient is not significant either).
Figure 3 situates Mexico's case in a broader, international context.Mexico is included in cluster 1, even though its level of spread is situated in the lower half of the sample (both clusters considered).In this regard, and in line with previous findings, the country was considered safe as a sovereign borrower until 1981 25 .Somewhat paradoxically, given its past records of expropriations, the penalty on the debt market from an additional expropriation might not have been as high as countries with fewer expropriation events (those in cluster 2).
As we show below, neither the press nor the markets seem to have anticipated Mexico's banks' expropriation.One might wonder how reputational costs were considered in the decision to default and to expropriate.As previous literature has shown, Mexico's government representatives intended to minimise the costs of the moratorium through the design of a strategy aimed at obtaining the support of the U.S. Treasury, the IMF and commercial banks 26 .While the moratorium mainly affected the government's (*) Significant at the 10 per cent; (**) significant at the 5 per cent; (***) significant at the 1 per cent and (no) not significant. 25The Euromoney Magazine published a ranking based on conditions under which each borrower contracted a loan in dollars or deutsche marks, using the London Interbank Offered Rate (LIBOR) as a reference rate.In the 1981 ranking, Mexico was placed 27 (out of 69 countries), better than countries such as Italy and South Korea (both non-defaulters).The Euromoney Magazine, «The Country Risk League Table», 47-51 February 1982.The BIS reported that creditor banks had maintained their Mexican exposure due to the country's oil wealth and its importance «to the lending banks as a source of balance-sheet growth and of profits».(Bank for International Settlements 1983, p. 128). 26According to Kraft (1984, 4), some advisors to Silva-Herzog were favorable to a declaration of unilateral default; this was rapidly ruled out as it was considered as a challenge to creditors with effects similar to an atom bomb.
Revista de Historia Económica / Journal of Iberian and Latin American Economic History   banking loans, the government decided not to interrupt payments on its bonds or on its trade-related credits.Regarding the expropriation, the government repeatedly insisted that it only affected domestic banks.This differentiated treatment (averting a negative effect on other sectors) may have been motivated by the relevance of foreign investment in the economy.Even then, as we shall see in the next section, we posit that the expropriation affected the country's reputation beyond the damage caused by the default.

Mexico's risk perception
In this section, we provide evidence showing that the decision to nationalise worsened the levels of country risk, as demonstrated by the sharp increase in the yields of a Mexican bond being quoted on the London Stock Market27 .We complement the results with an overview from the international press to show that the nationalisation was unexpected and generated a negative perception.We then present archival evidence on the debt negotiation process that shows that the nationalisation interrupted and delayed an agreement to obtain an IMF programme and a debt rescheduling.This interruption was not reported in the press.We can only speculate on a counterfactual case in which the effects of the nationalisation on the negotiations were known and whether this would have had a stronger effect on risk premia.

Risk Assessment
During those years, rating agencies did not publish sovereign ratings (Gaillard 2012;Altamura and Flores Zendejas 2020).Nevertheless, certain publications disclosed relevant information on the economic and financial conditions of a large group of countries.
Moody's annual reports on Mexico provided a description of the financial system and the bonds issued by the Mexican government.Its 1982 annual report disclosed the number of banks nationalised: forty-seven deposit banks, thirty multiservice banks, twenty- seven financial banks, three credit mortgage institutions and seven capitalisation credit banks.Moody's (1983) report lists bonds quoting in different markets.This information is presented in Table 2.It displays the bonds listed at the end of 1982 and their outstanding amounts in early 1983.While most of these bonds were denominated in U.S. dollars, others included marks, yen, pounds, francs and Swiss francs.The maturities of these bonds were variable and, in some cases, could be extended.It is interesting to note that it was unclear whether these bonds would be excluded from the moratorium.Kraft (1984) posits that there was considerable disagreement between European and U.S. banks regarding their treatment, as Europeans aimed to exclude them.As these bonds had different currency denominations and had been issued in different countries, it might have been more difficult to find a common ground for equal treatment in the event of a moratorium or a default.
We use this information to analyse the yields of a long-term Mexican government bond issued in pound sterling.I measure country risk as the spread between the yield of a long- term government bond denominated in pound sterling and a similar bond issued by the U.K. government28 .This risk indicator shows how the new policies were perceived by the market.If investors and banks believed that nationalisation could have been beneficial for Mexico's debt capacity, we should observe a decrease in the yield premia.On the contrary, an increase in this variable could reflect a loss in confidence in the capacity of the government to honour its debts and, more generally, a deteriorated investors' perception of the country's protection of property rights and the rule of law.Bonds were secondary to bank loans' volume.Alvarez (2019) reports that around 81 per cent of the government's external debt was owed to commercial banks, while bonds were less than 5 per cent29 .For this paper's purposes, it is interesting to note that the yields of such bonds allow us to assess the impact of Mexico's policies on country risk.
Figure 4 shows the daily evolution between July and the end of October.While the spread shows a slight increase since July, a first relevant upward movement can be observed around 8-10 August, the moment when Silva-Herzog began his communications with the U.S. government.In the first 2 days after the nationalisation announcement, the yield increased from 879 basis points to 947, an increase of approximately 8 per cent.While this value stabilised in the following weeks, by the end of the month it had climbed again to more than 1000 basis points.Clearly, the state of the negotiations and the economic policies implemented did not contribute to a decline in the risk perception of the government.
A structural break test allows us to identify shocks that might have affected investors' risk perception.The risk premia series has a unit root for the sample (Table 3), permitting us to test for the existence of a structural break using the Zivot-Andrews and Lee-Strazicich LM tests (Table 4) 30 .In accordance with the tests, we cannot reject the null hypothesis that the series has a unit root with a structural change in the trend for 9/03/1982 and 8/18/1982 and in the intercept for 8/18/1982 and 8/20/1982.The date for the structural breaks on 18 August corresponds to the first day trading after the news went public on the negotiations between Mexico's government and the U.S. government31 .The second break corresponds to the second day of the bank nationalisation.The structural changes identified correspond to increases in the spreads.Using a breakpoint regression (Bai and Perron 2003), I estimated the difference in the parameters that arise from a structural change.Table 5 present the parameters associated with the trend and intercept for two key periods of the sample, showing an increase in both cases.
Table 5 shows that, while the results from the Bai-Perron test are significant for the first date, those for 3 September are not.Overall, the increase in the spread after the nationalisation is less pronounced than that observed as Mexico's debt repayment difficulties became known.This narrative suggests that the loss of the government's reputation in the debt market could have influenced Mexico's risk perception in other sectors.Moreover, the sharp increases in the spreads in August demonstrate that while the Notes: H 0 : The variable has a unit root.We reject the null hypothesis with a significance of 10 per cent (*); 5 per cent (**) and 1 per cent (***).
reputational costs of the moratorium announcement were high, the marginal costs of nationalisation were low.In this regard, these findings suggest that the timing of defaulting and expropriation was a rational decision from a reputational perspective.

Reactions in the Press
It has been argued that the international press regarded the announcement positively (Basáñez and Camp 1984, p. 203;Elizondo Mayer-Serra 2001, p. 179).Nevertheless, this characterisation is not entirely accurate.An overview of the press shows that the general view was one of uncertainty and surprise.I first used the Fativa Dow Jones database, an international news database produced by Dow Jones, a major global provider of economic and financial information32 .I focused on all articles published in English between 2 September and 30 September 1982.My joint search on «Mexico» and «banks» released 266 entries33 .From this figure, seventeen articles were published on 2 September, the Note: H 0 : ∃ unit root with structural change on the intercept, trend or both.***indicates there is no statistical evidence to reject H 0 at 99 per cent confidence, **at 95 per cent and *at 90 per cent.
Note: In parentheses it is the critical value of the test.The Crash and Break at the Lee-Strazicich refer to the intercept and the trend, respectively; one and two refer to the number of structural breaks that the test allows.Source: Own estimations.day after the nationalisation.Ten articles were neutral or unrelated, five reported negative opinions and two reported positive opinions.However, one of these two articles, titled «Takeover pleases US Banks», recognised that banks had been shocked and that, in the long run, nationalisation could reduce confidence in the Mexican banking system.Positive articles thereafter remained ambiguous regarding the final effects of the nationalisation.One example is «Mexico backs Banks' Debts». 34It reproduced a telex signed by Silva-Herzog, where the government announced that it would guarantee the debts of the banks that had been nationalised, something that remained unclear.In fact, from the total number of articles found in Factiva, the word «confidence» was repeated seventy-four times.It was mostly related with negative sentiments including «undermine», «decline», «lack» and «decrease» while I only found the words «bolster» and «confidence» three times.
While a proper text analysis goes beyond the scope of this article, the qualitative evidence suggests that market sentiment remained negative.An illustrative example can be found in an article published by the Canadian Globe and Mail newspaper that qualified the nationalisation as «an abrupt move that stunned economic analysts» 35 .It claimed that the announcement would have a negative effect on foreign confidence in Mexico's ability to face its economic crisis.The same article quoted an interview with a U.S. business representative who predicted panic withdrawals from Mexican banks.The article mentioned that the move cast some doubt on the likelihood that the government would reach an agreement with the IMF, with which it was negotiating a 3-year $4 billion loan.It speculated on the kind of conditionality that the IMF was requesting, which included policies such as the reduction of public expenses through denationalisation and reduced restrictions on currency trading 36 .
Other articles emphasised the relevance of the event.The United Press International reported that the bank nationalisation was seen as a threatening catastrophe by the financial community 37 .The New York Times (NYT) compared the event to Mexico's expropriation of foreign oil companies in 1938.Even so, the NYT continued, central banks from ten Western nations had approved a $1.85 billion loan to the country 38 .The Financial Times reported that it was unclear whether Citibank would be affected by the nationalisation move.It quoted international bankers who predicted that the move would have only a minimal impact on the banks' operations, as the sector had always been tightly controlled by the central bank.It was also reported that the U.S. Treasury Secretary had not been notified in advance of Mexico's nationalisation plans.The article reported that the Mexican peso had depreciated in the New York foreign exchange market 39 .In a similar vein, The Associated Press reported that stock, bond and precious metals prices soared as investors worried about the stability of foreign governments and currencies sought «safe harbors for their money».It stated that there could be more nationalisations of banks in Latin or South America 40 .
Uncertainty also concerned the assets held by Mexican banks, as many of them were shareholders in several companies.I complemented my analysis with an overview of The Economist Historical Archives database and the October issue of The Euromoney Magazine, then the main magazine reporting on commercial banks' lending.The magazine Euromoney reported that the political party in power had always opposed the nationalisation of Mexican private banks, something that had been proposed by the Unified Socialist Mexican Party.Even so, the magazine reported that the measure had received high levels of popular support, despite the President's poor image as a «discredited politician who had brought his country to the verge of ruin» 41 .The article indicated that the government planned to sell the companies owned by the nationalised banks.However, the article quoted a banker asking «Who will want to buy these companies?Who can say the government won't step in and nationalise them?»The Economist also reported that leftist political parties and trade unions had been delighted, but emphasised the uncertainty created by the President's speech 42 .

Negotiations with the IMF
While Silva-Herzog had been having frequent contact with the IMF since early 1982, he asked the IMF to send a mission to Mexico that started on 23 July.On 13 August, he entered into formal negotiations to obtain an IMF loan (Kraft 1984, p. 10;Boughton 2000, p. 289).Before an agreement could be reached, the Mexican government needed to find the resources necessary to continue servicing its debt.As part of the emergency support received, there was a US$700 billion swap with the Federal Reserve, a US$1 billion advance payment for purchases of Mexican petroleum, and US$1 billion in loans from the Commodity Creditor Corporation.Finally, the BIS granted a US$1.5 billion credit line to Mexico's central bank.According to Anthony Salomon, President of the Federal Reserve Bank of New York, these loans would allow the government the necessary time to reach an agreement with creditor banks and the IMF for an extended arrangement 43 .
The success of the negotiations of Mexico's government with the IMF was dependent upon the willingness of commercial banks to maintain their credits to Mexico during the period requested (90 days) and on the government's capacity to convince bankers that the measures to be undertaken would be sufficient to redress the economic crisis.The bankers had been informed about the IMF mission and the possibility of obtaining a 3-year Extended Fund Facility (EFF) programme 44 .At the end of August, the IMF Executive Directors waited to learn the position of the banks before deciding on the nature of the programme and the size of the loan, even though they remained sceptical about whether the conditions initially announced by the Mexican government would be met.A positive assessment was provided by the fact that commercial banks had initially reacted positively to the meeting with Silva-Herzog.Overall, optimism regarding the finding of an optimal solution was still in the minds of the IMF officials 45 .
Therefore, the announcement of Mexico's bank nationalisation came as a dampener to the negotiations.Silva-Herzog was uncertain about whether he would continue the negotiations, and the IMF decided to send a new mission after 20 September to evaluate the policies announced by the Mexican President.The IMF also became pessimistic about whether an agreement could have been reached before the new administration took office.It was not only that it had been difficult to include members of the new administration in the negotiations, but also that «changes in the direction of certain policies that followed announcements made by the President on September 1 obviously complicated the process» 46 .
The announcement had an immediate impact on negotiations for the EFF programme.According to IMF documents, not only did the Director General of Mexico's central bank resign, but officials with whom the IMF had been negotiating «expressed shock and surprise about the measures announced by the President».The author of the memorandum also seemed to confirm that the President did not appear to accept any blame for what had happened.However, the breakdown of the negotiations was not supposed to be made public.Furthermore, the new Director General of the Bank of Mexico, Carlos Tello, apparently needed time to implement the new policies.IMF officials had been well aware that he had been hostile to the IMF in the past, and therefore, «may not have any interest in continuing the negotiations».It seemed nevertheless, that while the outgoing administration would not continue, the finance minister would push to reach an agreement if he remained in his post 47 .
In a Memorandum on the annual meeting between the IMF and the World Bank in Toronto, Nigel Carter-then Personal Assistant to the Managing Director-described how Ted Beza (IMF Managing Director) recalled that Mexico's nationalisation and exchange controls took place while the Fund's mission was in Mexico.According to Carter, Beza forecasted that such measures would delay negotiations since the IMF needed to assess the impact of those measures and understand how they could affect the programme that the IMF was designing.He also outlined the programme, as discussed with Silva-Herzog, including the need for adjustment 48 .
Banks from different countries had been following Mexico's negotiations with the IMF and also reacted negatively to the nationalisation and to its potential effects.Nevertheless, in a few cases, banks felt less concerned about the crisis and the outcome of the negotiations.This was the case of the Swiss banks.As the Swiss National Banking Association (SNB) reported, it did not see the need to initiate an investigation, as foreign participation in the Mexican banking sector had been forbidden since 1932.At that time, only City Bank was allowed to stay 49 .For the SNB, the only factor that needed to be monitored in Mexico was Swiss banks' exposure to Mexico's public debt.However, the report expressed some misgivings about the political and macroeconomic evolution of the country.
A very different case was that of the British banks.At the Bank of England (BoE), the Sovereign Risk Committee closely followed the position of Mexico's banks.It is interesting to note that British commercial banks did not necessarily share the views of their U.S. counterparts, which they considered as too optimistic.One of the reports of the same meeting in Toronto, a note addressed to Anthony Loehnis-then Executive Director of External Finance-and members of the «Mexican Task Force», reported that David Ardron (in charge of the Sovereign Risk Committee) said that «US banks seemed very relaxed» and maintained their existing levels of deposits with Mexican banks.However, the author of the report seemed to favour a more cautious attitude 50 .Some days later, the position of the BoE seems to have become more positive.In an unsigned Memorandum addressed to the BoE governors, the author praised the Mexicans for having successfully achieved their first and overriding objective, namely, to gain prompt and close control over their net external payments.The report then continued: «Whatever misgivings may be felt about the nationalization of Mexican banks and the change of personnel at the top of the central bank, the whole situation would by now have been a lot worse if the Mexicans had simply allowed things to drift out of control, with a spiraling depreciation of the peso and growing internal disorder.The example they have in practice set is quite a good one» 51 .
Regarding the interbank liabilities of the Mexican banks, nevertheless, the report stressed that «the Mexicans themselves did not fully appreciate the special nature of these liabilities, and their special importance for Mexico's future credit standing and for the credit standing of other major creditors in the international markets.However, they were soon made aware of this special importance and agreed to exclude these liabilities from the moratorium.Once again, the action was a correct response to a developing emergency» 52 .
The report then stressed the different options ahead to avert major withdrawals of interbank deposits from Mexican banks-still an unsolved issue-to provide further liquidity to Mexico's central bank and make sure it was able to provide support to its banks.In other words, nationalisation did not seem to have offered any kind of solution-although exchange rate controls helped-but remained an international coordination challenge between creditor central banks, the IMF and the BIS.If anything, it increased the uncertainty of the outcome of the crisis.

Conclusions
In this paper, I have placed Mexico's debt crisis in a broader, international context.While the literature on the 1982 debt crisis has focused on sovereign debt problems of defaulting countries, my analysis has encompassed the role of expropriations.In the late 1970s and the early 1980s, countries prone to default were also those that had experienced an expropriation episode at least once during the same period.Countries that expropriated also had worse macroeconomic fundamentals, thereby increasing their likelihood of defaulting.In this regard, Mexico was not unique.Mexico's debt crisis and bank nationalisation corresponded to the government's reaction to an adverse macroeconomic context, as was the case in other countries.Mexico was particular to the extent that the government defaulted and expropriated within such a short time frame.
Second, I have shown that financial markets did not welcome the decision to nationalise.An overview of the international press, and the reactions of the bond market show that the decision was perceived as negative.My analysis of the risk premia on Mexico's government pound sterling bond shows that markets reacted negatively to nationalisation, a result that contradicts the argument on the supposedly positive assessment of the nationalisation on Mexico's debt situation.However, this negative reaction was modest compared to the risk premia's increases in the aftermath of the moratorium announcement.Therefore, our findings do not support the theoretical arguments of reputational spillover effects originated by an expropriation and affecting government bond prices.Further research could provide international evidence of the bilateral relationships between expropriation events and sovereign debt markets.These relationships require further analysis for the most recent years.
Finally, the Mexican government's attempts to obtain an IMF loan were affected by the expropriation, thereby delaying the agreement and increasing the uncertainties in the political management of the economy.The archival evidence of this article confirms Boughton's (2000) narrative on the effects of the expropriation on those negotiations.Fortunately, the nationalisation did not fully affect the final outcome, which was largely dependent upon actions of other actors, including foreign central banks, foreign governments and international organisations, all of which were willing to avert a major international crisis.While foreign investment fell considerably in the years following the crisis, this rapid, multilateral and coordinated response, mitigated the effects of an expropriation on the government of Mexico's reputation, as would have been certainly the case in a pure market-driven framework.

Figure 2 .
Figure 2. Macroeconomic indicators.Inflation is reported as average percentage increases.Exchange rate is average annual change, in percentage, of the nominal exchange rate between a currency and the US dollar.Source: See text.Figures reported as percentages. .
doi.org/10.1017/S0212610923000174Published online by Cambridge University Press

Figure 3 .
Figure 3. Default risk and expropriation ratios, 1981.Sources: See text.Spreads and expropriation ratios as percentages.

Figure 4 .
Figure 4. Yield spreads of a Mexican government bond in 1982 (pound sterling denominated).Source: Own computation from the Financial Times.

Table 1 .
Mean equality tests, macroeconomic indicators Notes: The P-value in parentheses.

Table 2 .
Bonds reported in the Moody's International Manual, March 1983

Table 3 .
Unit root test table

Table 4 .
Crash and break tests

Table 5 .
Bai-Perron tests Note: the P-value is presented in parentheses.Source: Own estimations.