Prospective Deficits and the Asian Currency Crisis

The paper rejects both the “self-fulfilling prophecy” account favored by Asian policymakers and the standard first-generation view that ongoing fiscal deficits are monetized until reserves are exhausted, arguing instead that the crisis reflected fundamentals: large prospective deficits associated with implicit bailout guarantees to failing banking systems (Introduction). The authors note that “the explanation preferred by policy makers in Asia is that the currency crisis was a self fulfilling prophecy,” while “the most natural alternative explanation… is that it reflected profligate fiscal policy” of the kind featured in first-generation models; against both, they argue that “the Asian currency crises were caused by fundamentals: large prospective deficits associated with implicit bailout guarantees to failing banking systems,” where “the expectation that these future deficits would (at least in part) be financed by seignorage revenues led to a collapse of the fixed exchange rate regimes in Asia” (Introduction).

Because the governments of the crisis countries were not running the large, growing deficits that first-generation models require – “governments of the crisis countries… were running either surpluses or small deficits” (Introduction). Table 5 reports fiscal surpluses for Indonesia, Korea, Malaysia, the Philippines, and Thailand alongside Japan and the U.S. for 1995-1997, and the paper concludes that “the crises certainly could not have been predicted on the basis of large and/or growing pre-crisis fiscal deficits in the crisis countries.”

The model is a continuous-time, perfect-foresight small open economy populated by an infinitely lived representative agent and a government, both with access to international capital markets, in which purchasing power parity holds and the representative agent faces a cash-in-advance constraint on consumption (Section 2). The household maximizes a CRRA lifetime utility function subject to a flow budget constraint that allows for real domestic government debt and net foreign assets, and its first-order conditions imply a consumption-Euler-type relation, equation (2.6), that “will play a central role in our derivations” (Section 2.1). The government purchases goods, makes lump-sum transfers, levies lump-sum taxes, borrows at the real interest rate, and can print money, subject to its own intertemporal budget constraint (2.8), under which “the present value of future surpluses including the value of seignorage revenues must equal the value of the government’s net initial debt” (Section 2.2).

At time zero agents learn that transfers will increase permanently after some future date T’, raising the present value of the deficit by an amount Psi; since the government’s budget constraint under a sustainable fixed rate collects no seignorage, satisfying the new, higher present-value requirement is only possible if the government eventually raises seignorage revenues, which requires abandoning the peg (Section 3). The paper shows formally that “since the government must collect seignorage revenues it must, at some point, abandon the fixed exchange regime,” because under a sustained peg “no seignorage revenue has been raised by the government” no matter how much money is printed, since private agents simply trade away money they do not want to hold (Section 3, footnote 11).

The paper distinguishes the date of the speculative attack (t, when the fixed rate collapses) from the later date (T) at which the government actually implements the new money-growth policy needed to raise seignorage, arguing that “disentangling these two events considerably enriches the dynamic implications of the model”* (Introduction). Solving the model’s differential equation for the price level under the government’s threshold rule, the authors find that in their calibrated benchmark cases the attack date is strictly earlier than the policy-change date (t* less than T), because the anticipated future increase in money supply raises expected inflation and reduces money demand well before the increase in money itself occurs (Section 3.2).

In both calibrations the speculative attack (t=2.27 for Korea, t=2.42 for Thailand) occurs after agents learn of the higher prospective deficit but before the government implements its new monetary policy at T=3, so that “an econometrician looking at the data would see an exchange rate crisis – but he would not see large deficits, high growth rates of money or high rates of inflation prior to the speculative attack”** (Introduction; Section 4.2). The paper explicitly draws out the interpretive consequence: “the econometrician could well conclude that the attack was a multiple equilibrium phenomenon. In fact it reflects fundamentals: high prospective deficits” (Introduction).

No – changes in the debt threshold only affect the timing of the attack and how much monetary policy must adjust to balance the budget, not whether an attack occurs, and delaying via borrowing raises future inflation (Section 4.4-4.5). The authors state that “changes in the threshold rule… do not affect the inevitability of the speculative attack, only its timing and how much monetary policy must be adjusted to balance the government’s budget,” and that “a government can substantially delay the collapse of a fixed exchange rate regime by borrowing but only at the cost of higher future inflation” (Section 4.4). In the extreme “present value rule” case with no debt ceiling at all, the government “could delay the attack for a very long (but finite) period of time” only “at the cost of hyperinflation,” an outcome the authors attribute to two special assumptions of their model (cash-in-advance applies only to consumption, and output is exogenous) rather than treating it as a realistic policy option (Section 4.5).

The optimal policy is to abandon the fixed exchange rate as soon as information about the higher prospective deficit arrives, financing the required seignorage through an immediate jump in the money supply and a constant subsequent growth rate, because this reproduces the same consumption allocation the economy would have if the deficit were financed by lump-sum taxes instead (Section 5). The authors “demonstrate that the optimal monetary policy in our economy is to abandon the fixed exchange rate regime as soon as new information about the deficit arrives,” a result that follows from three features of their model – exogenous output, a cash-in-advance constraint that applies only to consumption, and seignorage rebated lump-sum to households – and they caution that “this result reflects, in part, the absence of nominal rigidities in our model” and abstracts from non-indexed domestic debt and unhedged foreign-currency loans (Section 5).

Using pre-crisis estimates of non-performing loan ratios together with data on total credit to the private sector, the authors construct rough estimates of governments’ implicit liabilities to the financial sector and find these liabilities were far larger in the crisis countries than in the non-crisis comparison group (Section 6). Non-performing loan ratios ranged “from a low of 14 percent in the Philippines to a high of 19 percent in Thailand” among crisis countries, versus “roughly 4 percent” in non-crisis Hong Kong, Singapore, and Taiwan (Table 6); the resulting implicit liabilities reached roughly 22-30 percent of GDP for Korea and Thailand, numbers the authors use directly to calibrate the parameter Omega/theta in their model (Section 6, Table 6).

Using monthly stock-market indices of financial-sector value relative to industrial/manufacturing/commerce sector indices, the authors show the financial sector’s relative value had been declining well before each country’s crisis date, especially in Korea and Thailand (Section 6, Table 10). In Korea the banking index fell 67 percent from its August 1991 peak to the October 1997 crisis date (64 percent relative to the manufacturing index); in Thailand the finance index fell 92 percent from its January 1994 peak to the July 1997 crisis date (80 percent relative to the commerce index); smaller but still substantial declines appear for the Philippines and Malaysia (Table 10). The authors conclude that “private agents in these four countries understood the potentially fragile nature of their banking systems,” and infer from the decline in bank equity values (rather than no change) that markets expected the government to protect depositors and creditors but not fully insure bank shareholders (Section 6).

The authors flag four specific shortcomings: the model reveals the state of the banking system to agents instantaneously rather than gradually over time; the decision to abandon the fixed rate is modeled with a simple debt-threshold rule rather than the political process that actually governs such decisions; nominal rigidities are abstracted from entirely; and the model does not address how the collapse of financial intermediaries itself affects future output and tax revenue (Section 7, “Conclusions”). They write that “obtaining this information took time and resources” in reality, that “this decision is shaped by political considerations” in practice, that the abstraction from nominal rigidities means those rigidities “will almost surely have an impact on the nature of optimal monetary policy,” and that they “took as given” the state in which banks collapse without modeling its effect on future tax revenue, flagging this as a direction for future work (Section 7).