Abstract
This paper associates exchange rate crises and capital flight with the possibility of default on public debt resulting from fiscal rigidities. By including interest-bearing debt, both domestic and external, the model can generate the timing of an attack and can explain why domestic public bonds, even when perfectly indexed, cannot eliminate the possibility of a crisis. This fiscal framework provides explanations for the simultaneity of private capital flight and public foreign borrowing and wide observed fluctuations in real exchange rates, with recent Mexican experience as illustration.
One of the salient features of the recent debt crisis in several Latin American countries has been the coexistence of inverse capital flows. In particular, while public external borrowing was flowing in, private capital was flowing out, as domestic investors made massive switches from domestic financial assets into foreign assets.1 Since domestic public debt was generally quite substantial, a large proportion of these domestic assets were government obligations. Thus, white domestic creditors were reducing their exposure to public debt, foreign creditors were increasing theirs.
One explanation for this portfolio substitution is asymmetrical information available to economic agents: domestic creditors predicted a crisis well before foreign bankers. Although this argument is probably not completely devoid of empirical relevance, it hinges on the existence of substantial long-run irrationality, as this phenomenon occurred over several years and on a scale that should have eventually attracted the attention of the international banking community and fed it to revise its lending operations much earlier.
There exists, however, a fully rational way to justify these seemingly contradictory capital flows. It involves asymmetric risk, rather than asymmetric information. The debt crisis corresponded, in many cases, to a fiscal crisis in which governments were faced with the prospect of not being able to keep servicing their debt, domestic or foreign, as fiscal rigidities prevented a sufficiently rapid adjustment of the budget deficit, following, in particular, the occurrence of large external shocks and fast-rising external debt. In this situation, perceived risks of default on government obligations grew rapidly. However, while a declared default on foreign obligations is a major step which entails large potential costs to the borrowing country,2 a virtual default on domestic debt is much easier to obtain: all it requires is a discrete devaluation which, by raising the price level, erodes the real value of domestic debt. Faced with this asymmetry of exposure between domestic and foreign debt, domestic wealth holders are likely to be the first to pull the trigger on the debt crisis because they perceive themselves as “junior” creditors.3 Furthermore, if domestic assets can be taxed, private capital flight can be encouraged by public foreign borrowing because the government may be perceived as counting on revenue from domestic asset taxation to service its higher foreign debt in the future. Paradoxically, a country committed to servicing its external obligations can thus be more prone to capital flight as domestic creditors suspect that the government may be able to maintain external debt servicing only at the expense of taxing domestic assets.
The notion that investors shift away from domestic assets to avoid inflation taxes brought about by fiscal disequilibrium underlies the whole literature on speculative attacks.4 However, the origins and implications of the weak fiscal stance of the government have seldom been explored, since the analysis has been generally limited to the consideration of an “excessive” rate of credit creation, and money is the only asset explicitly considered. From the existing literature it is particularly difficult to understand why governments do not implement timely corrective measures to prevent a balance of payments crisis. Nor is it clear why the public does not acquire domestic bonds, which could serve as well as foreign bonds as an inflation hedge, and which may be closer substitutes for domestic money. Ex ante, exchange risk is not a satisfactory explanation since the interest premium on domestic bonds could fully adjust to discount expected movements in the exchange rate. Moreover, in some countries, domestic banks have been offering dollar-denominated deposits.
To justify capital flight and the preference for foreign over domestic instruments, it seems more appropriate, then, to generalize the risk factor associated with the overall financial solvency of the public sector to all domestic public debt—not only to money. The government’s weak fiscal stance, which may force it to impose an inflation tax, could also induce it to reduce domestic debt servicing on bonds.
If prices are not perfectly flexible, an explanation can also be found for the exchange rate overshooting that is often the product of balance of payments crises, such as the one experienced by Mexico in 1982. With price inertia, domestic debt servicing cannot be instantaneously reduced through a devaluation-induced jump in the price level. It can, however, be reduced through a fall in the domestic real interest rate, which is possible, even with interest rate parity, if, following an initial depreciation, the real exchange rate gradually appreciates over time, reducing the return of foreign bonds in terms of domestic currency. An expectational equilibrium is thus generated as the public’s fear of lower returns on domestic assets leads to a shift into foreign instruments, provoking an overshooting that enables the authorities to lower domestic interest rates without violating interest rate parity.
In this paper we develop simple models to clarify these issues, using the recent experience of Mexico as an illustration. Section I relates the occurrence of speculative attacks to the solvency of the public sector in an economy with interest-bearing debt, and perfect price flexibility but imperfect asset substitution. Unlike in the usual speculative attacks literature, endogenous tending ceilings are derived, which are fully consistent with the behavior of both domestic and foreign creditors. Section II explores exchange rate dynamics and overshooting phenomena in the opposite case of price inertia and perfect asset substitution. Section III illustrates the results of the analysis using Mexico’s recent experience, and the final section offers concluding comments.
I. Public Debt and Capital Flight
Consider an economy in which prices are perfectly flexible and purchasing power parity (PPP) holds continuously, so that the real exchange rate equals one. The government finances its deficit through domestic and foreign debt. For greater simplicity, real domestic debt is taken as a single composite of money and bonds, bh with an average real interest cost rh. Since bh is an aggregate and rh is the real average interest rate, rh can be negative even when the real rate on the bond component is positive, if the inflation tax on money balances outweighs the real interest rate paid on bonds.5 With the domestic price index as a deflator, except for the real foreign debt and interest rate, bw and rf, which are expressed in terms of foreign prices, the real budget restriction of the public sector can be written as:
where dotted variables are time derivatives and g and t, real expenditures and taxes, are assumed to be fixed, reflecting fiscal weaknesses. Residents hold both domestic and foreign assets, but foreigners do not hold domestic assets. Consider the portfolio equilibrium condition:
where bf represents privately held foreign bonds and σ the semi-elasticity of substitution between home and foreign assets. Define
as real private financial wealth. Using equations (2) and (3), we can express the total interest bill on public sector debt for a given level of private wealth as: 6
Equation (4) may be rewritten:
where
As shown on the right-hand side of equation (5), the interest bill can be decomposed into two elements; rf(bw + bh) corresponds to the cost of servicing total public debt at the real world interest rate; the second term, − Φ, can be positive or negative depending on portfolio composition. When bh> λbf, the government must pay a premium on domestic debt and Φ is negative. On the other hand, when bh < λbh, Φ is positive and the government can extract a “tax revenue” on domestic assets.
The Φ contours shown in Figure 1 have the usual inflation tax shape. Tax revenues from private holdings of domestic debt first rise and then fall as the proportion of these holdings in total private wealth rises; Φ(0,σ) = 0, Φ then rises with bh, reaches a maximum for bh =
The shape of the iso-interest contours, R (σ), can then be diagrammed as in Figure 2. For bh = 0, all contours corresponding to the same R intersect the vertical axis at the same point, independently of σ For a given cr, the contours rise as bh becomes positive, reach a maximum for bh = bh, a point at which bhbh is minimized,7 then start falling, have a slope of minus one for bh = bh, and go toward -∞ as bh reaches W. As σ rises, their curvature decreases; when σ approaches +∞, they become straight lines of slope minus one, defined as R* in Figure 2.
The government can intervene by borrowing abroad and reducing domestic debt. In Figure 2, this operation moves the economy leftward on a line of slope equal to minus one. If assets are perfect substitutes, the total interest bill remains invariant, and the economy moves on the same R* contour. When σ is finite and assets are imperfect substitutes, a marginal intervention leaves the interest bill unchanged when bh = bh since the slope to the R contour at that point equals minus one. Elsewhere, a change in debt composition alters R. When bh > bh, the asset tax revenue rises when the composition of public debt shifts towards foreign obligations. When bh<=
Consider now the following experiment: suppose that a permanent negative fiscal shock, Δ(t − g) < 0, hits the economy, while the budget is initially in equilibrium with bh = bw = 0. If assets are perfect substitutes, an immediate default is unavoidable, since the total interest bill cannot be altered by intervention. A reduction of the domestic debt burden is more likely to occur than a default on foreign debt, because the former is much simpler to achieve. With perfect price flexibility, all it takes is a discrete devaluation, which, by instantaneously raising the domestic price level, reduces bh and moves the economy horizontally to the left in Figure 2, until an R* contour is reached that is low enough to close the fiscal gap.8
However, if assets are imperfect substitutes, the imposition of capital losses for domestic wealth owners can be avoided if a change in debt composition can by itself reduce R sufficiently. Suppose that the economy is initially on the R1 contour at a point such as H in Figure 2, where
Consider then what happens when a negative fiscal shock hits the budget. If Δ(g − t)<(R1 − R4) the government remains fully solvent and rational creditors, whether domestic or external, should be fully willing to finance the emerging fiscal gap through the acquisition of new debt. The authorities could then, for example, increase foreign borrowing along a vertical trajectory, from H up to Q, the point from which the minimum iso-interest contour that can be reached is R2(σ), such that Δ(g-t) = R1 − R2. To the extent that the government would rather devalue than default on its foreign obligations, and since the R2 contour could no longer be reached beyond Q through intervention, foreign borrowing beyond Q would eventually require a discrete devaluation to maintain foreign debt servicing. To avoid a certain capital loss, domestic creditors should then at that point shift from domestic into foreign assets, forcing the government to intervene. Capital flight, financed by a large burst of foreign borrowing, will occur as a result, moving the economy from Q to N.9
The new steady-state equilibrium is at N, and the fixed exchange rate rule should be abandoned at that point. To see this, let R3(σ) be the iso-interest contour whose tangent to the point of maximum domestic asset taxation, V in Figure 2, goes through N. Since R2-R3>0, a margin is stilt available to service additional borrowing, and foreign lending can, in principle, continue. However, if the government is committed to honoring its foreign obligations, any further external borrowing will require a default on domestic obligations, since the total interest bill can no longer be reduced by reshuffling debt through intervention. Even with a one-to-one conversion of new foreign borrowing into capital flight, the total interest bill rises and a jump devaluation remains unavoidable. Thus, any additional external borrowing should give rise to an increase of a larger magnitude in the private demand for foreign assets, as domestic creditors try to protect themselves from a certain capital loss. Since no net additional financing can therefore be obtained,10 a rational government should, at that point, abandon the fixed exchange rate rule and finance the deficit by a higher tax on domestic assets, made possible by the switch in private portfolio composition and a higher rate of depreciation, as documented in the usual speculative attacks literature. Here, however, the timing of the attack is determined within the model, instead of by an exogenous stock of foreign reserves.
Point N will only be reached by a government that is willing to postpone an exchange rate adjustment, and to keep servicing its foreign debt, whatever the final cost of taxing its domestic obligations. The associated level of inflation or of financial repression can be so high, however, that it may force the government to abandon its exchange rate policy earlier, or to contemplate a partial default on its foreign obligations as a less costly way to close its financing gap. If this is anticipated by investors, capital flight will occur earlier on the HQ segment, say at Q′, when all the desired portfolio substitution from domestic into foreign assets can be carried out before intervention ceases or foreign lending is cut off. The new end-point on the R2 contour, M, is associated with a lower level of asset taxation. Depending, therefore, upon the perceived comparative willingness of the government to accept earlier exchange rate adjustment or higher eventual asset taxes, the economy may end up at any point of the KN segment. If the government’s intentions are not known, capital flight should gradually accelerate as foreign borrowing increases, since the rising level of asset taxation that will eventually be needed makes a collapse of a pegged exchange rate more and more likely over time, either because the government may stop intervening or because foreign creditors may decide on their own to stop lending.11 Furthermore, since in the context of uncertainty the collapse of the exchange rate regime cannot be perfectly anticipated, a jump devaluation will necessarily occur at the time of the collapse, as documented in the speculative attacks literature.12
II. Price Inertia and Overshooting
In the previous model, an immediate adjustment in the debt-servicing burden following a devaluation could be obtained as the result of an instantaneous adjustment of the price level, which reduced the real value of domestic debt. However, if prices are not perfectly flexible, adjustment can only be achieved through reducing the real interest rate. In this section, it will be shown that real interest rates can be substantially reduced even when a large share of domestic debt is composed of interest-bearing instruments that are close substitutes for foreign bonds, so that interest rate parity holds. Real interest rates will fall if a large overshooting of the real exchange rate is engineered at the time of the collapse of the nominal exchange rate.
Let Ph* be the equilibrium domestic price level obtained when PPP holds. Setting the foreign price equal to one, Ph* = E, where E is the nominal exchange rate. Suppose that Ph adjusts with some inertia toward its PPP equilibrium level: 13
With Ph as a deflator, equation (7) may be expressed in real terms:
where e is the real exchange rate. Since e can differ from 1, the budget restriction needs to be rewritten:
Suppose, finally, that interest rate parity is verified, so that:
then, replacing rh from equation (10) and ė/e from equation (8), equation (9) becomes:
Equations (8) and (11) form a differential system in e and bh.
Assume that price adjustment is fast enough so that the impact of additional external borrowing on the stock of foreign debt can be ignored. In the phase diagram of Figure 3, equation (8) is simply an inertial adjustment that gives rise to a horizontal equilibrium schedule, ee. On the other hand, the bond equilibrium schedule corresponding to equation (11), bb, is positively sloped around the equilibrium exchange rate if the following condition is satisfied:
Dynamics of Adjustment in the Price-Inertia Case
Citation: IMF Staff Papers 1987, 001; 10.5089/9781451946987.024.A004
Dynamics of Adjustment in the Price-Inertia Case
Citation: IMF Staff Papers 1987, 001; 10.5089/9781451946987.024.A004
Dynamics of Adjustment in the Price-Inertia Case
Citation: IMF Staff Papers 1987, 001; 10.5089/9781451946987.024.A004
In this case, a higher bh requires a compensating increase in e to maintain the budget in equilibrium. This is because a higher bh implies a higher domestic debt servicing, which can be neutralized in one of two ways: by the increase in the value of net public capital inflows that is caused by a devaluation when ·bw − rfbw is positive; or, by a reduction in ·rh which is only possible, with interest rate parity, if the exchange rate gradually appreciates after an initial devaluation. The latter effect is validated through price adjustment and applies to the value of domestic debt, hence the first term in equation (12). Suppose then that equation (12) is verified, in particular because price inertia is not too strong, so that the first term in equation (12) outweighs the second when there are net public capital outflows. Then it can be checked in Figure 3 that the dynamics of adjustment have the saddle-path property.
Suppose now that an unanticipated negative public finance shock—for example, a reduction in the flow of external lending—hits the budget. The bb schedule shifts leftward (Figure 4) and the real exchange rate overshoots before converging slowly back to equilibrium. The interpretation is straightforward: the shock reduces the ability of the government to service its domestic debt. The public then shifts toward foreign bonds and the nominal exchange rate depreciates, in turn inducing a depreciation of the real exchange rate, since the price level does not jump owing to inertia. The ensuing gradual appreciation of the exchange rate toward equilibrium reduces the return of foreign bonds in domestic currency and—through interest rate parity—the interest rate on home bonds, thus allowing for the collection of an interest tax and the adjustment of public finances. As the price level starts rising, real home bond balances fall and the domestic interest rate recovers gradually, until bond balances have been reduced to a level that can be serviced at the world interest rate.
Dynamics of Overshooting
Citation: IMF Staff Papers 1987, 001; 10.5089/9781451946987.024.A004
Dynamics of Overshooting
Citation: IMF Staff Papers 1987, 001; 10.5089/9781451946987.024.A004
Dynamics of Overshooting
Citation: IMF Staff Papers 1987, 001; 10.5089/9781451946987.024.A004
Several remarks can be made. First, the overshooting mechanism depicted in Figure 4 is the same as the one in Dornbusch’s well-known 1976 paper. However, while in Dornbusch’s model interest rate movements are essentially monetary phenomena arising from the monetary equilibrium condition, in our model they are purely fiscal manifestations produced by the need to equilibrate the budget restriction of the government. Second, there is a direct relationship between the magnitude of the overshooting and the amount of foreign public debt. As the burden of foreign debt servicing net of new foreign loans increases, the bb schedule becomes steeper and the size of the real exchange rate jump which is needed to accommodate a fiscal shock rises. When foreign debt is large enough to invert the sign of condition (12), adjustment through overshooting is no longer feasible with a single exchange rate regime.14 Third, the adjustment of domestic bond balances to the level that can be serviced by the state at the world interest rate provides a simple theory of private bond portfolio composition, if extended to a longer-term horizon that includes current account adjustments. In particular, if a certain level of real financial wealth is desired, the public will have to acquire abroad those bonds that cannot be serviced internally. Fiscally weaker countries can thus be expected to have higher shares of foreign bonds in private portfolios. Finally, given that short-term fiscal adjustment requires real exchange rate overshooting if prices are not fully flexible, and given that the more “dollarized” an economy the smaller its taxable base, it is clear that the existence of dollar-denominated domestic bonds tends to be destabilizing because it raises the size of the overshooting. Since the government may elect to default on its domestic dollar debt in order to avoid an excessive jump in the real exchange rate. investors may eventually lose confidence in this type of bond if their share in total domestic debt becomes substantial. This can explain why foreign dollar bonds may be preferred to their domestic counterpart.
III. The Mexican Experience
To fit the recent Mexican experience to the outlined models above, Table 1 gives data on the real public sector deficit for 1978–84, and on its sources of funds.15
Real Budget Restriction of the Mexican Public Sector, 1978–84
(In billions of 1978 pesos)
“Mexdollar” balances. For an explanation of how the “Mexdollar” system worked, see Ortiz (1983).
Computed as errors and omissions plus variation in short-term assets held abroad by residents.
Real Budget Restriction of the Mexican Public Sector, 1978–84
(In billions of 1978 pesos)
| Item | 1978 | 1979 | 1980 | 1981 | 1982 | 1983 | 1984 | ||
|---|---|---|---|---|---|---|---|---|---|
| Use of funds | |||||||||
| Pre-oil deficit | 109 | 142 | 236 | 416 | 429 | 83 | 78 | ||
| Domestic debt interest | 2 | -2 | 14 | 23 | 37 | 30 | 22 | ||
| Foreign debt interest | 46 | 56 | 61 | 70 | 158 | 143 | 141 | ||
| Source of funds | |||||||||
| Oil income | 27 | 38 | 110 | 122 | 150 | 190 | 171 | ||
| Inflation tax | 17 | 24 | 36 | 36 | 95 | 64 | 42 | ||
| Interest tax | 13 | 11 | 47 | 16 | 222 | 65 | 52 | ||
| Changes in real money balances | 8 | 18 | 1 | 2 | -15 | -33 | -5 | ||
| Changes in real peso bond balances | 18 | 36 | 38 | 51 | 105 | -60 | -30 | ||
| Changes in real dollar bond balancesa | — | 8 | 20 | 73 | -45 | -31 | -62 | ||
| External borrowing | 58 | 65 | 64 | 229 | 132 | 83 | 44 | ||
| Statistical discrepancy | 16 | -3 | -5 | -20 | -20 | -22 | 29 | ||
| Memorandum items | |||||||||
| Capital flightb | 14 | 21 | 68 | 160 | 179 | 105 | 48 | ||
| Real exchange rate | 22.8 | 19.3 | 15.4 | 12.8 | 24.0 | 24.4 | 18.5 | ||
| Nominal exchange | |||||||||
| rate | 22.8 | 22.8 | 23.2 | 26.2 | 148.5 | 161.3 | 190.0 | ||
“Mexdollar” balances. For an explanation of how the “Mexdollar” system worked, see Ortiz (1983).
Computed as errors and omissions plus variation in short-term assets held abroad by residents.
Real Budget Restriction of the Mexican Public Sector, 1978–84
(In billions of 1978 pesos)
| Item | 1978 | 1979 | 1980 | 1981 | 1982 | 1983 | 1984 | ||
|---|---|---|---|---|---|---|---|---|---|
| Use of funds | |||||||||
| Pre-oil deficit | 109 | 142 | 236 | 416 | 429 | 83 | 78 | ||
| Domestic debt interest | 2 | -2 | 14 | 23 | 37 | 30 | 22 | ||
| Foreign debt interest | 46 | 56 | 61 | 70 | 158 | 143 | 141 | ||
| Source of funds | |||||||||
| Oil income | 27 | 38 | 110 | 122 | 150 | 190 | 171 | ||
| Inflation tax | 17 | 24 | 36 | 36 | 95 | 64 | 42 | ||
| Interest tax | 13 | 11 | 47 | 16 | 222 | 65 | 52 | ||
| Changes in real money balances | 8 | 18 | 1 | 2 | -15 | -33 | -5 | ||
| Changes in real peso bond balances | 18 | 36 | 38 | 51 | 105 | -60 | -30 | ||
| Changes in real dollar bond balancesa | — | 8 | 20 | 73 | -45 | -31 | -62 | ||
| External borrowing | 58 | 65 | 64 | 229 | 132 | 83 | 44 | ||
| Statistical discrepancy | 16 | -3 | -5 | -20 | -20 | -22 | 29 | ||
| Memorandum items | |||||||||
| Capital flightb | 14 | 21 | 68 | 160 | 179 | 105 | 48 | ||
| Real exchange rate | 22.8 | 19.3 | 15.4 | 12.8 | 24.0 | 24.4 | 18.5 | ||
| Nominal exchange | |||||||||
| rate | 22.8 | 22.8 | 23.2 | 26.2 | 148.5 | 161.3 | 190.0 | ||
“Mexdollar” balances. For an explanation of how the “Mexdollar” system worked, see Ortiz (1983).
Computed as errors and omissions plus variation in short-term assets held abroad by residents.
The first obvious observation is that the pre-oil public deficit grew steadily from 1978 to 1982, the combined result of runaway public spending and lagging non-oil revenues, the latter mostly stemming from falling real prices of goods and services produced by state enterprises. Up to 1981, the deficit was financed by rapidly growing oil revenues derived from expanding exports and favorable world prices; by a steady increase in domestic credit, made possible by rapid economic growth; and by a significant tax on both money and bond balances, the latter facilitated by negative or low foreign real interest rates and by the continuous real appreciation of the peso. Foreign lending to the public sector from 1978 to 1980 was largely used to cover interest payments and did not, therefore, constitute a significant financing source over that period.
In 1980, signs of a lack of confidence in the path followed by the Government began to appear. Devaluation expectations, induced by a rapidly appreciating exchange rate, led to a growing “dollarization” of the economy, as reflected in the expansion of dollar-denominated domestic debt, which became the main source of domestic finance in 1981. Together with the increasing dollarization, capital flight accelerated sharply from 1980 on, reflecting growing doubts about the public sector’s ability to honor its domestic debt commitments, both in pesos and in dollars.
The year 1981 marked the turning point on the path leading to the crisis. The pre-oil budget deficit continued to explode while oil revenues reached a plateau, mainly because of falling oil prices and the continued appreciation of the peso. Despite the fact that domestic credit still grew substantially in real terms, owing to the “Mexdollar” system, a huge financial gap had to be covered by foreign borrowing, which, in that single year, rose in dollar terms by as much as it had grown in the previous five years. Simultaneously, as measured by the sum of errors and omissions and variations in short-term assets held abroad by residents,16 capital outflows reached more than $10 billion, most of it in the second half of 1981, when the lack of fiscal adjustment, in the face of adverse external shocks, became quite evident.17
In 1982, the peak of the crisis, the budget deficit grew even larger, a clear indication again of fiscal rigidities.18 The other dominant events were, of course, the interruption of foreign borrowing, following a last burst of intense borrowing to sustain the peso in the face of the large capital outflow observed from March to August; and the final collapse of the Mexdollar system, after investors massively lost confidence in that instrument. In August, Mexdollars were converted into pesos at 70 pesos a dollar, while the free exchange rate shot up to 130 pesos at the end of the month. This clearly amounted to a partial default by the state on its domestic dollar creditors.
The ongoing fiscal deficit and the foreign debt crisis created an enormous financial gap, which could not be financed by an increase in real domestic credit, given the exhaustion of the Mexdoltar system and the massive capital outflows.19 A portion of the deficit was financed by higher oil revenues associated with the depreciation of the peso. But even after accounting for oil income, there was still a shortfall of 317 billion pesos in 1982 which had to come, in effect, from taxes on domestic assets. As seen in Table 1, 30 percent of this shortfall was covered by an inflation tax; the rest by a huge interest tax.
The years 1983 and 1984 marked the aftermath of the crisis. The pre-oil deficit finally fell from its 429 billion pesos peak in 1982 to around 80 billion pesos, a quite impressive achievement. But, despite higher oil revenues associated with the devaluation, this reduction failed to absorb both the continued deterioration of net foreign capital inflows 20 (-60 billion pesos in 1983 and -97 billion pesos in 1984, down from -26 billion pesos in 1982 and 159 billion pesos in 1981) and the fall in real domestic credit associated with higher inflation, negative or low growth, and the portfolio reallocation resulting from the public’s preference for foreign bonds. Thus, high inflation and interest taxes were still needed; 106 billion pesos were obtained from money and 117 billion pesos from bonds over the two years 1983 and 1984.
For a better appreciation of how these tax revenues were collected, the dynamics of the free exchange rate are shown in Figure 5, which uses the monthly closing date of the peso spot price at the New York Stock Exchange. The figure shows that overshooting reached its peak in August 1982. The free exchange rate then appreciated fairly smoothly until the end of 1984.21 This gradual appreciation allowed domestic real interest rates to fall below the world rate, at very negative levels in 1982 and early 1983 and 1984, as shown in Figure 6, which compares real bond returns in Mexico and in the U.S. over the period 1978–84. As it can be seen in the figure, the foreign real return in dollars was much higher than its domestic peso counterpart in 1982 and early 1983 and 1984, while due to the expected real appreciation of the peso, foreign real returns in pesos, computed on the basis of the forward exchange market, were broadly comparable to domestic returns.
Real Exchange Rates, 1982–84
Citation: IMF Staff Papers 1987, 001; 10.5089/9781451946987.024.A004
Source: Banco de México, Indicadores Económicos. Real exchange rates were obtained by deflating nominal exchange rates with the ratio of Mexican to U.S. consumer price indices.Real Exchange Rates, 1982–84
Citation: IMF Staff Papers 1987, 001; 10.5089/9781451946987.024.A004
Source: Banco de México, Indicadores Económicos. Real exchange rates were obtained by deflating nominal exchange rates with the ratio of Mexican to U.S. consumer price indices.Real Exchange Rates, 1982–84
Citation: IMF Staff Papers 1987, 001; 10.5089/9781451946987.024.A004
Source: Banco de México, Indicadores Económicos. Real exchange rates were obtained by deflating nominal exchange rates with the ratio of Mexican to U.S. consumer price indices.
Real Domestic and Foreign Returns, 1978–84
Citation: IMF Staff Papers 1987, 001; 10.5089/9781451946987.024.A004
Real Domestic and Foreign Returns, 1978–84
Citation: IMF Staff Papers 1987, 001; 10.5089/9781451946987.024.A004
Real Domestic and Foreign Returns, 1978–84
Citation: IMF Staff Papers 1987, 001; 10.5089/9781451946987.024.A004
IV. Conclusions
This paper has analyzed the linkages between fiscal rigidities, public debt, and exchange market adjustments. It is based on the concept that asset demand shifts between domestic and foreign bonds respond to the perceived ability of the state to service its domestic debt. In this context, internal or external shocks with an adverse impact on public finance create fiscal gaps that have to be filled, in the absence of fiscal adjustment, by taxing domestic holders of public debt. Asset substitution is then likely to give rise to capital flight or to sharp exchange rate movements. It was demonstrated that, with imperfect asset substitutability, capital flight could be financed by foreign borrowing. This was a rational outcome if foreign creditors expected the state to tax its domestic creditors as a means of maintaining foreign debt servicing. In turn, it was rational for domestic creditors to attack the local currency and massively convert their domestic assets into foreign bonds so as to avoid a devaluation-induced default on these assets. In contrast to methods used in the usual speculative attacks literature, debt ceilings fully consistent with rational behavior on the part of both domestic and foreign creditors of the public sector were derived endogenously.
We also showed that exchange rate overshootings are likely to occur in response to unanticipated public finance shocks if prices are not fully flexible. Price inflexibility implies that the real value of domestic debt cannot fall instantaneously, so that domestic debt servicing must be reduced through a fall in domestic real interest rates. In turn, with interest rate parity, this reduction can only result from an expected real exchange rate appreciation following an initial discrete devaluation.
Although claims of rational behavior based on underlying fundamentals cannot be pushed too far, our models suggest ways in which to interpret the recent Mexican experience that do not rely on sheer speculation and disequilibrium dynamics. Available evidence from the Mexican crisis seems to provide reasonable support for our analysis.
This approach can be extended in a number of directions. Among them, two lines of analysis in particular merit investigation: the introduction of a fully stochastic framework such as that recently used by Penati and Pennachi (1986) for the monetary model; and the analysis of fiscal rigidities and asset substitution in a richer intertemporal framework, which could integrate issues of credibility and temporal inconsistencies.22
APPENDIX
Notes on Table 1: Budget Restriction of the Mexican Public Sector
The data for Table 1 were obtained from Indicadores Económicos from the Bank of Mexico. The public sector deficit figures correspond to the concept of “financial deficit,” less oil tax income and interest payments. Domestic public debt was obtained as the sum of total domestic credit given by the financial sector to the public sector, plus bonds sold directly to the public. Foreign credit was obtained from balance of payments data.23 Finally, the figure for domestic dollar credit was computed as the flow of dollar credit granted by the financial sector, less foreign credit obtained by development banks, since this portion is already accounted for as external credit. From the comparison of the nominal figures, a statistical discrepancy term was obtained for each year. A further separation between money and bonds was obtained by dividing domestic peso credit from banks to the public sector, into interest and non-interest bearing assets, according to the proportion of Ml into peso liabilities of the financial system.
Figures were then deflated by the yearly average consumer price index. Real interest payments on domestic debt were computed by applying the U.S. real interest rate, obtained on the basis of the Treasury bill rate deflated by the consumer price index, to the stock of domestic bonds. For greater clarity of interpretation, interest payments on foreign debt were kept in nominal terms and taken directly from the balance of payments. In the case of peso credit flows, real increases were obtained by deflating end-of-period nominal balances by the end-of-period price index. The inflation tax on money balances was computed to be consistent with this deflation procedure, and the peso bonds tax was obtained as residual.
REFERENCES
Blanco, Herminio, and Peter Garber, “Recurrent Devaluation and Speculative Attacks on the Mexican Peso,” Journal of Political Economy (Chicago), Vol. 94 (February 1986), pp. 148–66.
Connolly, Michael, and Dean Taylor, “The Exact Timing of the Collapse of an Exchange Rate Regime and Its Impact on the Relative Price of Traded Goods,” Journal of Money, Credit and Banking (Columbus, Ohio), Vol. 16 (May 1984), pp. 194–207.
Diaz Alejandro, Carlos, “Latin American Debt: I Don’t Think We Are in Kansas Anymore,” Brookings Papers on Economic Activity: 2 (1984), The Brookings Institution (Washington), pp. 335–403.
Dooley, Michael, “Country-Specific Risk Premiums, Capital Flight and Net Investment Income Payments in Selected Developing Economies” (unpublished; Washington: International Monetary Fund, 1986).
Dooley, Michael, and Peter Isard, “Tax Avoidance and Exchange Rate Determination” (unpublished; Washington: International Monetary Fund, 1986).
Dornbusch, Rudiger, “Expectations and Exchange Rate Dynamics,” Journal of Political Economy (Chicago), Vol. 84 (December 1976), pp. 1161–76.
Eaton, Jonathan, Mark Gersovitz, and Joseph Stiglitz, “The Pure Theory of Country Risk,” NBER Working Paper 1984 (Cambridge, Massachusetts: National Bureau of Economic Research, April 1986).
Flood, Robert, and Peter Garber, “Collapsing Exchange Rate Regimes: Some Linear Examples,” Journal of International Economics (Amsterdam), Vol. 17 (August 1984), pp. 1–13.
Ize, Alain, “Investment, Capital Flight and Political Risk: The Case of Mexico,” (unpublished; Mexico City: Colegio de México, January 1985).
Ize, Alain, and Guillermo Ortiz, “Political Risk, Asset Substitution and Exchange Rate Dynamics: The Mexican Financial Crisis of 1982,” Documentos de Trabajo No. 5 (Mexico City: Colegio de México, October 1984).
Khan, Mohsin S., and Nadeem Ul Haque, “Foreign Borrowing and Capita! Flight: A Formal Analysis,” Staff Papers, International Monetary Fund (Washington), Vol. 32 (December 1985), pp. 606–28.
Krugman, Paul, “A Model of Balance of Payments Crisis,” Journal of Money, Credit and Banking (Columbus, Ohio), Vol. 11 (August 1979), pp. 311–25.
Lizondo, Saul, “Foreign Exchange Futures Prices Under Fixed Exchange Rates,” Journal of International Economics (Amsterdam), Vol. 14 (February 1983), pp. 69–84.
Obstfeld, Maurice (1984a), “Rational and Self-Fulfilling Balance of Payments Crisis,” NBER Working Paper 1486 (Cambridge, Massachusetts: National Bureau of Economic Research, November).
Obstfeld, Maurice (1984b), “Speculative Attack and the External Constraint in a Maximizing Model of the Balance of Payments,” NBER Working Paper 1437 (Cambridge, Massachusetts: National Bureau of Economic Research, August 1984).
Ortiz, Guillermo, “Currency Substitution in Mexico: The Dollarization Problem,” Journal of Money, Credit and Banking (Columbus, Ohio), Vol. 15 (May 1983), pp. 174–85.
Ortiz, Guillermo, “Economic Expansion, Crisis and Adjustment in Mexico (1977–1983),” The Economics of the Caribbean Basin, ed. by Michael Connolly and John McDermott (New York: Praeger, 1985), pp. 68–98.
Penati, Alessandro, and George Pennachi, “Optimal Portfolio Choice and the Collapse of a Fixed Exchange Rate Regime” (unpublished; Philadelphia: University of Pennsylvania, August 1986).
Sachs, Jeffrey (1984a), “Latin American Debt: I Don’t Think We Are in Kansas Anymore—Comment,” Brookings Papers on Economic Activity: 2 (1984), The Brookings Institution (Washington), pp. 405–7.
Sachs, Jeffrey (1984b), “Theoretical Issues in International Borrowing,” NBER Working Paper 1189 (Cambridge, Massachusetts: National Bureau of Economic Research, August 1983).
Salant, Stephen, and Dale Henderson, “Market Anticipation of Government Policies and the Price of Gold,” Journal of Political Economy (Chicago), Vol. 86 (August 1978), pp. 627–48.
Van Wijnbergen, Sweder, “Fiscal Deficits, Exchange Rate Crises and Inflation” (unpublished; Washington: The World Bank, February 1986).
Mr. Ize, an economist in the Fiscal Affairs Department, is a graduate of Columbia and Stanford Universities. This paper was initiated while he was with the Economics Department of the Colegio de México and an economist at the Bank of México. Mr. Ortiz is an Executive Director of the Fund and a graduate of the Univer-sidad Nacional Autónoma de México and Stanford University.
This phenomenon has puzzled many observers of the Latin American scene. See, in particular, Diaz Alejandro (1984), and Sachs (1984a). Another type of asymmetry in capital flows, private capital flowing out at the same time as private external borrowing was flowing in, is analyzed in Khan and U1 Haque (1985).
Empirical support for that hypothesis can be found in Dooley (1986) who finds, on the basis of a cross-section sample of eight countries, that capital flight is inversely related to the risk premium on external debt and therefore to the differential risk faced by residents and nonresidents.
See the seminal contributions in Salant and Henderson (1978) and in Krugman (1979). More recent contributions include Flood and Garber (1984), Obstfeld (1984a and b), Connolly and Taylor (1984), and, in the case of Mexico, Blanco and Garber (1986). Van Wijnbergen (1986) independently developed a model relatinp fiscal rigidities to asset taxes and capital flows, somewhat similar to ours, while Dooley and Isard (1986) present a general conceptual framework that includes risks of asset taxation into asset demands, and analyzes their macroeconomic impact in a two-country framework.
Conversely, rh could become negative even with no inflation, if domestic bonds were sufficiently imperfect substitutes for foreign bonds. In practice, however, the margin for imposing a pure interest tax on bonds without additional inflation is likely to be limited, first, because nominal rates must be positive and this constraint becomes rapidly binding unless the initial inflation rate is already high, and second, because the degree of substitutability between foreign and domestic bonds may be high. Hence, higher inflation can be expected to be the main vehicle of asset taxation, both because it directly erodes money balances and because it allows for negative real interest rates on bonds.
Real wealth is assumed constant. Although a discrete devaluation does in fact reduce wealth, nothing of substance would change by incorporating changes in wealth into the analysis.
To see this, note that dR = rfdbw+ d(rhbh) = 0 on an iso-interest contour. The slope is horizontal when db = 0, which corresponds to the point where d(rhbh = 0, which maximizes the asset tax.
Although it does not have the same legal implications, a discrete unexpected devaluation can be perceived by the public as a partial default on domestic debt since it amounts to changing unilaterally the terms of an implicit debt contract between the government and private agents.
Similar dynamics would be obtained if the deficit had initially been financed by internal borrowing. In that case, domestic borrowing first rises, and then collapses, as the economy moves back to N following a burst of capital flight.
This can explain the reluctance of countries with already heavy levels of asset taxation to engage in further international borrowing, since the additional foreign debt may be immediately exchanged for domestic debt and converted into deposits abroad.
The default risk on foreign loans that is implied here could be assimilated in the sovereign risk analyzed in the debt repudiation literature (see Sachs (1984b)). The defaulting benefits emphasized here are less inflation and less financial repression, which are not usually considered in the literature.
See, in particular, Lizondo (1983) and Flood and Garber (1984).
A more usual adjustment equation would be of the type: ·Ph = ·Ph* + ν(·Ph* − ·Ph). However, when converted into real terms, the inflation rate becomes a determinant of the speed of adjustment. Although this would not alter the analysis in any significant way, a simpler formulation in real terms was chosen.
Overshooting would still work however with a dual exchange system, since in that case the burden of foreign debt servicing does not rise with a devaluation.
Details on the elaboration of this table can be found in the Appendix.
There has been a wide debate on the issue of capital flight measurement. See for example Dooley (1986). While the definition used here has clear shortcomings, it has the advantage of simplicity.
See Ortiz (1985) for an account of the economic events of this period. Apart from the lack of control over the deficit, another costly mistake of the López Portillo administration was maintaining the nominal value of the peso. In the context of our model, devaluing earlier would have been greatly desirable because it would have increased the peso revenues obtained from foreign borrowing and oil earnings. A real 50 percent devaluation implemented at the end of 1980, for example, would have raised oil income in 1981 by 61 (1978) billion pesos and reduced the need for foreign debt by more than two thirds in that year, even considering that the burden of servicing that debt would have risen, as a result of the higher exchange rate. By improving government finance, the reduction in foreign indebtedness would, in turn, have helped stabilize the exchange market, as seen earlier in the model.
Public spending grew, although the Government had announced spending cuts on at least three occasions, in July 1981 and in February and April 1982. Its credibility was understandably damaged by those repeated failed attempts. Besides the downward rigidity of expenditures, two other factors help to explain the increase in the deficit: a fall in non-oil tax income owing to the recession, and exchange losses suffered when the Government converted Mexdollar deposits into pesos at a rate of 70 pesos a dollar. The loss was compounded by the fact that the Government nationalized a banking system that had been forced to convert dollar loans at 50 pesos a dollar.
Although real peso bond balances grew by twice as much as in the previous year, this was essentially the result of the forced conversion of dollar bonds into pesos. Altogether, money and domestic bonds grew by only 45 billion of 1978 pesos, versus 126 billion pesos for 1981.
Foreign borrowing less foreign debt interest.
September and October 1982 are out of line owing to an expectational error motivated by a wrong setting for the controlled exchange rate during those months at an unrealistically low level, given the rates of inflation that had prevailed since January. On the other hand, a small speculative bubble occurred in mid-1984 provoked by a combination of “bad news” concerning the inertia of inflation and the behavior of oil prices and external interest rates.
The concept of fiscal weaknesses or fiscal rigidities could perhaps be endogenized. In Ize and Ortiz (1984) and Ize (1985), spending rigidities are related to the set of political pressures that the government faces.
Average yearly controlled rates were used to convert balance of payments figures since it was at that rate that the state sold its foreign currency surplus to the private sector.
