Liquidity, risk, and the collapse of the Mexican peso: a dynamic CGE interpretation.

• Author: Kildegaard, Arne
• Position: Computable general equilibrium

1. Introduction

   On December 19th, 1994, Mexico's central bank announced a "one-time" 13% devaluation of the narrow band within which the peso had been allowed to float. Investors reacted violently, and within 48 hours the attack on reserves forced Mexico to abandon exchange rate management entirely.(1) The free-floating peso began a slide leaving it 33% lower relative to the dollar by the new year, and 50% lower by the middle of February.

   This paper argues that the Bank of Mexico's December 19th regime change revealed new in- formation to the market, rationalizing the subsequent attack on reserves.(2) Specifically, the regime change announcement contained "news" regarding Mexico's stock of official foreign reserves. Insights borrowed from option pricing theory illustrate that such news leads investors to raise the risk premium assigned to peso-denominated loans. The general equilibrium implications of such a shock are shown to include a sharp and immediate devaluation.

   The shock is simulated using a multi-sector dynamic computable general equilibrium model of Mexico. Following the devaluation, the model predicts a gradual real revaluation extending over several years. In the short-run, resources shift into export and import-competing sectors (principally Mining and Manufacturing), while the current account moves into surplus and net foreign indebtedness is reduced. In the medium- to long-term, the short-run impact is reversed: non-traded goods production expands and export and import competing sectors contract, leaving Mexico with a smaller volume of outstanding debt and of international trade than would have been the case absent the shock.

   Section II draws on insights from option pricing theory to explain the nature of the shock which hit the Mexican economy. Section III describes the model used in the analysis. Section IV reports key results from simulating the shock described in section II with the model of section III. Section V explores the economic intuition behind the simulation results. Section VI discusses the sensitivity of the results to key parameter specifications, while section VII concludes.

2. The Shock

   Insights from option pricing theory have recently been brought to bear on the question of how risk premia should be calculated on loans to sovereign borrowers.(3) This provides a simple conceptual framework for understanding why the Bank of Mexico's regime change announcement had such a surprising and disruptive effect on the foreign exchange market.

   As with any loan, the ex-ante likelihood of repayment on loans to a sovereign borrower depends upon expected cash flows. With sovereign borrowers, it is only the dollar fraction of the foreign economy's overall cash-flow which matters, and this fraction depends directly on stochastic variables such as export commodity prices and world interest rates. In the event of a bad realization of the relevant price (vector), cash flows will be insufficient to meet payments, the borrower will default on some portion of the debt, and real losses will be forced on the lender.

   The problem of pricing this risk can be simplified by the insight that default is essentially the exercise of an option, implicit in the loan contract. Specifically, default occurs when the borrower exercises a "put" option on cash flows, or, perhaps more intuitively, on the commodity export(s) which drives cash flows. If, for example, a foreign currency crisis occurs when the price of a country's oil exports falls to $16/bbl., and in fact prices fall to $14/bbl., then in return for each barrel the defaulting country "receives" $14 cash and $2 in debt write-offs. It is precisely as if the lender had agreed to buy oil at $16/bbl. at the borrower's discretion. Put differently, it is as if the borrower had exercised an implicit put option on oil at a strike price of $16.

   The ex-ante value of this option to the borrower -- and hence the cost to the lender -- depends primarily on two factors: 1) the variance of the stochastic process driving cash flows (oil prices above); and 2) the proximity of the critical strike price to today's price ("the money"). The more variable the stochastic process, and the closer strike price to the money, the higher the value of the implicit option to the borrower, and therefore the higher the risk premium which rational lenders should attach to the loan in order to be compensated.

   One variable of critical interest to lenders, then, is a country's stock of foreign reserves. To the extent that reserves are low, the economy can sustain less bad news and continue to remain solvent. In other words, a lower level of reserves implies that the strike price on the implicit put option is closer to the money.

   In the Mexican case, data on foreign reserves have been caricatured as a state secret. The notoriously slow process by which the Bank of Mexico historically released these data relegated markets to inferring the current state of official foreign reserves (and other important variables) from the policy actions of the Bank.(4) The decision to devalue imposed two distinct capital losses on the Mexican government, each enormous in its own right: first, in light of the centrality of the government commitment on exchange rate policy in the 1988 stabilization and subsequently, devaluing represented a repudiation of the pacto framework and the loss of government reputational capital painstakingly acquired over the previous six years; second, in light of the rapidly growing exposure of the public sector to exchange rate risk emanating from reliance on dollar-denominated tesobonos,(5) devaluation imposed a direct capital loss on public account. Choosing this unattractive option could only be viewed as due to the lack of alternatives. The Bank of Mexico's decision to devalue the exchange rate band by 13% and its subsequent decision to abandon the band altogether can be understood to have delivered the information to the market that foreign reserves were unexpectedly low (as in fact they were), driving the strike price on the implicit put option closer to the money and therefore raising the risk premium required by lenders.(6)

3. A Dynamic CGE Model of Mexico

   The model employed in the analysis below is characterized by nine productive sectors, perfect capital mobility and foresight, a representative consumer, government, and a foreign account.(7) Representative competitive firms in each sector maximize the stock market value of sector-specific capital by choosing optimal inputs and investment (subject to convex adjustment costs), while the representative consumer maximizes the present value of a time-separable inter-temporal utility function.(8) The steady state is imposed after 85 years.

   The productive sectors correspond to the 1-digit level of disaggregation of the Mexican national accounts: 1) Agriculture, Forestry and Fishing; 2) Mining; 3) Manufacturing; 4) Construction; 5) Electricity, Gas and Water; 6) Wholesale and Retail Trade; 7) Transportation and Communications; 8) Financial Services and Real Estate; 9) Personal and Professional Services. Domestic commodities are produced by combining intermediate inputs, labor and capital. Figure 1 illustrates the nesting scheme. At the top level of the nesting tree, domestic production of commodity i ([Q.sub.i]) is a constant elasticity...