7 On the Welfare Costs of Systemic Risk, Financial Instability, and Financial Crises
- Omotunde Johnson
- Published Date:
- April 2002
In 1986, when the mammoth nuclear reactor at Chernobyl in the Ukraine broke down, the damage was immeasurable. Initially, some 30 people were killed; more than 200 people were seriously exposed to radiation; and the entire area around Chernobyl of 25,000 square kilometers was contaminated for years. Not only was the subsequent damage almost impossible to estimate; so was the probability of an accident of that enormity. A systemic financial crisis is similar to that of a nuclear accident in that it is nearly impossible to predict and the damages suffered as a consequence are extremely difficult to measure.2
In discussing the costs of systemic risks, the approach throughout this paper has been of a scientific nature, with the focus on the crucially important but limited issue of just how fast and to what extent a large default in the system spreads. The difficulty of accurate cost assessment lies not only in the fact that the welfare cost of systemic risk may well be exorbitant, but the estimation of the probability is nearly impossible.
Of course, most systemic financial crises involve the transfer of funds electronically. There are many excellent monographs and treatises on this subject, particularly on the relationship between electronic transfers and systemic risks (e.g., Johnson, 1998; Sato, 1998, 1999) On this same subject, David Humphrey (1986) explored how an initial adverse shock was transmitted through the transactions network to a systemic problem. In a series of works following the direction of his research (e.g., Furfine, 1999; McAndrews and Wasilyew, 1995), we can appreciate how this linkage is affected by the magnitude of initial shocks and by the degree of interdependence among banks and the mode of regulation. (For a review of the current literature, see Kaufman, 1994 and 2000.)
The organization of this paper is relatively straightforward. It will first explore the possibility for estimating the incidence of systemic risk, an effort that is tantamount to navigating the ocean without a chart. The reason is that, although one could say an electronic network is relatively stable (according to the above noted literature on the subject), the calculation of the magnitude of possible losses on the national economy in question is not easy. Rather than detailing the transmission of settlement risk into systemic risk, the focus will be on the effect of a systemic crisis from the standpoint of a country’s national economy and demonstrating how the expected value of the incidence of systemic risk can be substantial relative to the scale of a national economy.
By definition, systemic risk almost defies the idea of being able to insure risk, simply because of the magnitude of damage in any widespread financial crisis. This phenomenon is exactly the problem encountered in the case of Chernobyl (and, for that matter, any other nuclear accident). The expected value of systemic risk is substantial in a world where technological errors and malicious acts, like those of computer hackers, could easily destroy an entire financial system. To avoid such a realization, prevention is all important, and fail-safe devices are encouraged despite the costs involved.
The paper will then deal with more conventional risks—or the welfare costs of financial instability and financial crises. In recent years, many examples of these kinds of costs have occurred in currency crises, monetary instability, and exchange rate fluctuations. Examples can be seen in Europe before the monetary integration resulting in the euro’s introduction in 1999, the devaluation of the Mexican peso in 1995 and the subsequent turmoil in Latin American financial markets, the monetary confusion in the Baltic countries, Russia, and the other countries of the former Soviet Union during the 1990s, and notably the currency crises throughout Asia beginning in 1997. In most situations, costs of financial instability can be assessed by price stability and income, as well as the loss of employment due to recession.
After discussing the cost of price fluctuations to a national economy, I will address the opportunity cost of adopting a certain exchange rate regime. For example, Hong Kong Special Administrative Region of the People’s Republic of China (SAR) incurred the cost of high interest and recession in order to keep its Currency Board System (CBS). Chile, after the 1980s, and Malaysia, after 1998, succeeded in isolating the effect of inflows and outflows (respectively) of capital, but this effect was accompanied by the loss in consumption smoothing patterns in growth. Korea and Thailand suffered deep troughs of recession after the Asian currency crisis. For both countries, I will address the question of just how high the cost of currency problems was.
The basic approach to determining the welfare cost of either systemic financial stability, or crisis risk is essentially to calculate the loss of consumers’ and producers’ surplus in the Harberger’s triangle (Haberger, 1964). One can compare this loss with the cost of prevention of any kind of risk. If the (opportunity) cost of prevention is extremely high given the existing risks, one should be satisfied with the status quo. This is, of course, the general approach to efficiently minimize the incidence of risk to society. For a similar discussion on electronic money, see the “Introduction” of Solomon (1991).
Welfare Cost of Systemic Risk
The magnitude of the welfare loss from a possible systemic crisis of a modern electronic payment system is difficult to estimate. The loss from a systemic crisis is, by definition, universal to the system and can last for a long period of time. It will affect the totality of a national economy and may even spread to the global economy with a substantial impact. However, since a disastrous breakdown of the electronic payment network has not yet occurred (except for cases of near-miss accidents), we do not know by what objective probability a systemic crisis can occur.
As we saw with Chernobyl, nuclear accidents create severe damage to human lives and health, industrial facilities, and the entire environment over an extensive period of time. One cannot possibly assess the damage in a single dimension, but must view it over many dimensions—the overall magnitude being enormous. Moreover, as noted previously, it is nearly impossible to assess the probability of such an accident, although we do know, of course, that serious nuclear accidents have occurred and will continue to occur.
Although a systemic risk may not be as disastrous as a nuclear accident, the systemic risk of payment systems resembles the risk of a nuclear accident. The incidence of an accident can be devastating, although predicting the probability of such an occurrence is extremely difficult. Neither the risk of a nuclear accident nor the risk of a systemic financial crisis is easily met by insurance. In the case of nuclear accidents, the international legal and insurance communities still attempt to provide some insurance scheme to mitigate damages from nuclear accidents. This may create a serious moral hazard, but the risk is so large that diversification is necessary after the incident occurs. In the case of the United States, the Nuclear Regulatory Commission requires that each atomic generator plant insure the minimum amount of $1.06 billion (Marrone, 1993). In Germany, where the data for premiums were available, the cap on the coverage or protection was DM 1 billion, which was paid by the insurance premium pool (amounting to DM 0-200 million), the mutual association of the power plant operators (DM 201-500 million) and the state guarantee without premium (DM 501-1,000 million). The premium in 1993 was DM 16.5 million. In the context of systemic financial risk, it is difficult to determine what insurance (if any) should be available, or whether, in fact, the idea should be dismissed altogether as a scheme encouraging moral hazard.
In cases of systemic financial crises, the incidence has fortunately been contained. Sato (1998) mentions some examples of some “near misses”: the Bank of New York’s computer troubles for thirty hours in November 1985; the effects of the U.S. stock market crash in October 1987; and the repercussions from the bankruptcy of Yamaichi Securities in 1997. These cases did not develop into universal currency or financial crises. They show, however, that many different kinds of risks have always occurred.
Three examples are most indicative of the self-magnifying nature of a systemic crisis (Sato, 1998; and White, 2000). In each case, the incidence of trouble is barely contained. In 1990, after Drexel Burnham Lambert, Inc. went bankrupt, Drexel could not even engage in repurchase agreement operations (repo operations) with the collateral of U.S. treasury bonds. In spite of the fact that operations were on a risk-free DVP (delivery-versus-payment) basis, clients of the bank refused to accept them. The famous case of the Herstatt risk in 1974 was created by the difference in trading hours around the globe. Similar crises were about to occur when Libya froze foreign assets in 1986 and when the Bank of Credit and Commerce International (BCCI) went bankrupt in 1991. The lessons learned from the Herstatt Bank incident prevented crises from occurring in this case. Finally, the Long-Term Capital Management hedge fund (LTCM), founded by John Meriweather, former head of fixed-income trading for Salomon Brothers in New York, became exposed to major risks in 1998, which subsequently spread to other markets. These examples provide cases for the problems of financial instability, or crises, as well as the problems of a systemic crisis. At the same time, the bailout of LTCM, which is discussed later in the paper, is a case where information from one or a small number of institutions was transmitted through financial linkages across similar institutions worldwide (Furfine, 1999).
Simulation Studies of Systemic Risks
Humphrey (1984 and 1986) presented innovative analyses of the process of transmissions from a settlement failure. His simulations used actual data of transactions and traced the effect of the removal of all payments to and from a large settling participant. Then, he showed that the unwinding processes of credit-debit operations could cause serious deterioration in the net debt position of financial institutions and, in one of his examples (Humphrey, 1986), he demonstrated how roughly 50 institutions could end up in a day with net debit positions exceeding their capital. In the meantime, McAndrews and Wasilyew (1995) conducted generalized simulations to study the systemic risk in terms of bankrupt banks and their nonperforming assets, assuming an identical size of fund but under alternative conditions. The larger the number of participants is, the greater the variance of the size of payments, and the larger the probability of interaction among banks is, the lower the probability that banks will survive. If the number of banks is more than 50, regardless of the degree of interaction, a bank ceases to exist with more than 80 percent probability after an initial shock. This study, indeed, presents a depressing picture of systemic banking risk.
Furfine (1999) gives a revisionist view to this idea. First, he makes an important distinction between the two types of systemic risks. “The first type is the risk that some financial shock causes a set of markets or institutions to simultaneously fail to function efficiently. The second type is the risk of failure of one or a small number of institutions will be transmitted to others due to explicit financial linkages across institutions.” Basically, the first type of risk was treated by the bank-run model of Diamond and Dybvig (1983), and the second type was analyzed by the simulations mentioned by David Humphrey in the preceding paragraph.
By taking the buffer-stock function of the federal funds market into account, Furfine (1999) presents simulations for the second type of risk. According to his studies, the degree of contagion that takes place when triggered by a fall of a major bank depends on such factors as bilateral federal funds exposures, the loss rate of a bankrupt bank, and the capitalization of other banks. The incidence of bank failures becomes small when loan facilities from the federal funds market mitigate the incidence of fund shortages among banks. Assuming a generous (at least according to Furfine) nonrecovery rate of 40 percent, his simulation results show that the fall of a significant bank will lead to a failure of two to six other important banks, and only 0.8 percent of total commercial banking assets. It is an important contribution, in that he has demonstrated that the buffer-stock mechanism through federal funds could reduce the systemic cost of transmission of a sudden withdrawal of funds from the system.
Although Furfine describes these mitigating circumstances, we unfortunately cannot afford to be complacent about systemic financial risk. Furfine, himself, actually mentions some caveats concerning factors not considered in his simulations. In this analysis, two crucial factors should be considered in assessing the welfare cost of systemic risks.
First, the failure of a bank triggers anxieties in depositors throughout the market—often on a global basis. The Diamond and Dybvig (1983) model is concerned with the contagion of fears among depositors to a particular bank, which is classified by Furfine as the first type of systemic risk. In many cases, however, the failure of a bank may not end with just the anxieties over the soundness of the particular bank. A bank failure or a payment failure may trigger a “panic” in the attitude of depositors in the market and may cause a chain of negative repercussions throughout the system. The process will create many shocks, with the initial ones becoming dominant, thus acting as triggers to the destruction of the entire financial system. In other words, an initial shock to a particular bank may replicate itself through exogenous shocks to the total system. As Kiyotaki and Moore (1997) analyze, a chain of mistrust will create a chain of failure and bankruptcy, which will subsequently cause macroeconomic stagnation.
The above simulations are mostly concerned with contagion through the payment system. These studies are important. In order to obtain a global view on the welfare cost, however, one cannot neglect the process of contagion through the psychology of market participants, as well as through electronic networks.
Another consideration is that although systemic risks can be, and should be, reduced by various devices like the cap on debt positions and the buffer-stock mechanism, one cannot ignore the inadvertent type of risk and the risk from Internet terrorists. Indeed, the fact that the financial system is critically dependent upon an electronic device has its own inherent strengths. For example, during the great earthquake in Kobe, Japan in 1995, all mail service and banks were closed in the region, and telephones were unreliable. Only the Internet proved to be a robust means of communication.
Given this positive side, however, what happens if this electronic device becomes technically disrupted or falls into the wrong hands? The examples of a teenage hacker who successfully disabled the Yahoo system by virtually shutting down the computer and a couple who allegedly generated the “I LOVE YOU” virus, which temporarily shut down many corporate and government operations, show that some types of computer literacy could be applied to disrupt the electronic payment system completely. The Committee on Payment and Settlement Systems (CPSS) states, in one of the Core Principles for Systemically Important Payment Systems, “The system should ensure a high degree of security and operational reliability and should have contingency arrangement for timely completion of daily processing.” Behind such general principles, many complex elements are obscured. Some are discussed below.
A Macroeconomic Assessment of the Expected Value of Systemic Failure Costs
Deviating slightly from the above simulation studies, it is interesting to pose two hypothetical questions. What will be the loss to the national economy if a systemic risk unfortunately is realized? And, what is the probability of such an occurrence?
According to Bertaut and Iyigun (1999), the Trans-European Automated Real-Time Gross-Settlement Express Transfer (TARGET) system processed a daily average of more than 150,000 transactions valued at EUR966 billion ($863 billion) during the first quarter of 1999. This amounted to about 13 percent of the annual GDP ($6.6 trillion in 1998) of the European Union. In the United States, the Federal Reserve System’s Fedwire service processed on a daily average basis more than 400,000 transactions valued at $1,343 billion. This was almost 16 percent of the annual GDP ($8.5 trillion in 1998) of the U.S. economy. Meanwhile, CHIPS (Clearing House Inter-Bank Payments System) also handles a large amount of funds. Any serious settlement failure within CHIPS would have serious repercussions for the financial system at large.
Take the U.S. Fedwire. The total amount of transactions taking place on the Fedwire in a week is approximately equivalent to the entire annual GDP. Assume that the average duration of a system breakdown is one week. This amount, equivalent to the entire GDP—some $8.5 trillion—will be affected. What will be the welfare cost of this amount? A five percent interest rate may be taken as the minimum per transaction welfare cost in the flow dimension.3 If an amount of payment is unrecoverable at the moment of systemic crisis, the total amount may not be lost. Some of it may be delivered later. Thus, the interest cost accruable for a week—that is, 5 percent divided by 52 (number of weeks in a year)—could be considered one estimate for the welfare loss.
$8.5 x.05 ÷ 52 = $ 0.008 trillion = $8 billion.
If a systemic crisis were to hit the system, then the magnitude would be as large as $8 billion, or equivalent to about 0.1 percent of GDP.
Figure 7.1 is a simple demand and supply diagram for money. The real money balance is depicted as a function of the nominal exchange rate that is on the vertical axis. AD is the demand curve for money, and the consumers’ or money holders’ surplus is drawn as the area of triangle ABC. If the electronic money reduces the cost of transaction from OB to OB′, then the money holders’ surplus increases by the area of echelon BCD′B′. When electronic money is destroyed, the damage is not limited to the echelon but all the benefit of money A′B′D′ will be lost. The rectangular area BOMC is the welfare cost because we do not know the shape of the demand curve. If the elasticity of money demand is small, ABC could be much larger than BOMC. We take the benefit of monetary service BOMC as a benchmark, which corresponds to the linear demand case with the elasticity of money demand of one-half.4.
Figure 7.1.Money Demand and Supply
What is the probability of a systemic crisis? We have no objective answers to this question. Objective probability can never, or hardly ever, be calculated. Subjective probability will fluctuate; it will be high after the near crisis, and low at other times. Taking a somewhat bold position, assume that the probability distribution of systemic risk occurrence is something like the Gaussian distribution. Assume further that a near crisis like Herstatt mentioned above (Sato, 1999) is triggered by extreme shocks that have a magnitude of more than twice the standard deviation (with probability density of 4 percent); and assume that a devastating crisis takes place when critical shocks reach a magnitude of more than three times the standard deviation (with probability density of 0.1 percent). If a near miss then occurs once a year, a catastrophe will occur with a probability of about one-fortieth.
With this improbable chain of supposition, one first obtains the cost of a systemic crisis when it does occur. The expected cost is the multiple of the 0.5 percent (calculated above) of the GDP of the United States multiplied by the probability of 1/40 (0.025), that is, 0.0025 percent of GDP, or about $200 million for the U.S. economy.
One could criticize the fact that most of the suppositions are constructed in such a way as to yield an overestimate. For instance, the period of the disruption may be a matter of days, hours, or even seconds. A high recovery ratio of deficient claims, together with the federal funds flexibility will generate a much smaller incidence (Furfine, 1999) of systemic crisis. The probability assessment of near crises adopted here may be an overestimate. Correction of these considerations would reduce the expected value of systemic cost to a great extent.
On the other hand, one could argue the opposite position as well. The crisis may not end in a matter of days, hours, and so forth. Moreover, the contagion effect caused by consumer psychology is not taken into consideration in this hypothetical consideration. The above estimate should be multiplied by the contagion multiplier that measures the extent that an initial shock triggers new independent withdrawals to the overall system (not necessarily to the same bank hit by the initial shock) by the depositors who are panic stricken by the shock. If an initial bankruptcy creates withdrawals of multiple K, then one should multiply the above estimate of welfare cost by K.
Finally, one should seriously consider not only the mechanical or technological breakdown of systems but also criminal break-ins. If a computer “geek” could render the Yahoo system dysfunctional, then it may not be just a fantasy that somebody could use his knowledge of artificial intelligence for pecuniary advantage.
These considerations provide strong grounds for building a buffer mechanism against unexpected and sudden payment failures, as well as against contagious withdrawals. A fail-safe mechanism against technical dysfunction, as well as criminal intrusions into the system needs to be developed. Many devices are proposed to reduce the “narrow” systemic risk caused by the payment mechanism. For example, the following devices have been considered (Sato, 1998):
delay the delivery of funds until interbank settlements are completed;
limit the membership exclusively to sound banks; and
put an upper limit to the daily exposure of banks on a bilateral basis and/or on a system-wide net debit basis.
Risk sharing through insurance would be another way to minimize the chain reactions of debt. Insurance would give participants the prospect for security in the face of an emergency and probably help to provide a smoother resolution of systemic crises and minicrises. On the other hand, moral hazard problems would always pose a dilemma for insurance. The word “systemic” by itself implies that the risks are not easy to address through diversification or pooling. Thus, the function of insurance is essentially limited. From a systemic and global point of view, the system should aim at providing more resilient short-term financing measures, protection from technical failure, and security from criminal incursion.
Incidentally, efforts to circumvent the Y2K problem offer some suggestions on the cost of the attempt to prevent systemic risks. The President’s Council on Year 2000 Conversion (2000) describes a similar effort to endeavor to prevent the electronic crisis of the payment system. According to the Council’s report, “(s)ome people were predicting that government agency failures alone would send the U.S. economy into a deep recession.” The Council made extensive efforts in various forms, such as improving information sharing, creating a Global Y2K Network, and creating a Coordination “center”, while at the same time distributing a booklet, “Y2K and You,” that featured checklists for a smooth change of date on the advent of the millennium.
The cost of this organized effort appears hard to assess. The introduction (page 3) of the Y2K report states, “Many predicted that the final price tag for the United States Government alone would top $20 billion. … The range was illustrated by frequently cited estimates of $300 billion to $600 billion for the worldwide cost.” Meanwhile, the retrospective account (p. 16) notes, “Cumulative agency estimates for the costs to solve the Y2K problem increased over four years from under $3 billion to the $8.5 billion that was actually spent. This was still significantly less than the $20 to $30 billion estimated by outsiders.”
No doubt, the same intensity, if not the same amount, of effort spent on preventing the Y2K problem could be spent on the prevention of a systemic risk. If such an effort were effected, it would significantly improve the buffer mechanism against the systemic risk. If the estimate of $ 1 billion per year as the expected value of the systemic risk for the United States were to have any basis, at least the prevention expense of a similar magnitude to the Y2K funds could be spent every ten years. In order to avoid a serious systemic disaster, a government or the international community may as well launch (instead of the Y2K program) a “Guarding Electronic Money (GEM)” program to engage in international cooperative activities to prevent systemic risks.
As Johnson (1998) and Johnson and others (1998) indicate, changes in the payment system will make the task of monetary policy more challenging—among other reasons, because the control of monetary aggregates will be more difficult for the central bank, at a minimum by changing the money multiplier. The difficulty is compounded when changes in the payment system also lead to changes in the velocity of money. Such problems add to the seriousness of the main concern of this paper—namely, the welfare costs of systemic risks, because possible errors in monetary policy will escalate the likelihood of crises. Specialists can no doubt provide further analysis of this subtle but important issue. Just one example, though, is worth noting—that is, where conventional problems in international finance begin to appear quite differently in the presence of electronic payments. It happens that discussions of dollarization or currency board arrangements are now quite popular. However, the welfare effects of these currency arrangements can differ critically, depending on whether or not a substantial part of the means of payment is in the form of electronic money. Thus, the benefit of currency integration and dollarization should be judged differently, depending on the degree of advancement in electronic transactions.
Welfare Costs of Financial Instability and Financial Crisis
Let us begin this section by exploring the distinction between financial “instability” and financial “crisis.” “Instability” seems to suggest a situation where financial variables such as prices and exchange rates fluctuate back and forth. Exchange rates under the floating rate system among industrialized countries can be quite stable or unstable. In the latter case, exchange rates are normally thought to fluctuate in both directions. A part of future variance of fluctuations can be forecast, but its major part cannot. These fluctuations in financial variables make it necessary for economic agents to insure themselves against expected fluctuations, and cope with the unexpected fluctuations. These can be regarded as the costs of financial instability.
On the other hand, “crisis” seems to imply a situation where financial variables move suddenly and precipitously, often contrary to previous expectations, and more or less in a single direction. During a crisis, the calculations and insurance costs will increase as in the case of financial instability, but the incidence does not halt there. In crises, changes in financial variables are so abrupt and drastic that they force change in the monetary system or in the international monetary regime. Hyperinflationary situations in many Latin American countries without currency convertibility in the 1980s, the chaos in the world currency market after the breakdown of the Bretton Woods regime in the early 1970s, and economic problems in Asian countries after the 1997 currency crisis are all examples of cases of financial crisis. The public loses its trust in monetary policy, in the international monetary regime, and in the payment system itself. Thus, financial crises may generate a systemic crisis by contagion of anxieties through withdrawals to the total system, as is discussed in the first section.
In this section, I will briefly consider what is involved as costs of financial instability and financial crisis. Instability and crisis will not be differentiated in a rigorous manner. After a brief discussion of the concepts of costs, I will illustrate the welfare cost of a financial instability or crisis situation by estimating the loss of real income that a country suffers when adopting an international monetary regime. The reasons for taking this approach are as follows. First, in order to assess the welfare cost of a monetary regime—for example, the Currency Board System of Argentina—one must consider what would have happened if Argentina had adopted a different monetary regime. Therefore, it is important to take account of the opportunity cost or the opportunity benefit of a system if one wants to measure the welfare costs brought about by financial instability or crisis. Second, by measuring the welfare cost in terms of the loss to real income, it is necessary to have some sense of comparison with the welfare loss of a systemic failure that was discussed previously.
What Is Lost Through Financial Instability and Financial Crises?
Conventional wisdom tells us that price fluctuations will create welfare loss. Economic agents such as producers, consumers, and intermediaries will generally lose by confronting price fluctuations, it is believed, because they have to incur more forecasting costs and insurance costs against price fluctuations. However, speculators will gain from price fluctuations, although it is not clear that their gain is more than the loss of other parties in the market. Moreover, it is a well-known fact that expected mean-preserving price variations are beneficial to consumers as well as to producers. Figure 7.2, in a partial equilibrium setting, compares the consumer surplus between a stable price and the mean-preserving fluctuating price. OB is the range of the two prices OB’ and OB″. It is easily apparent that the average of the areas AB′ C′ and AB″C″ is greater than the area of ABC. For instance, it is more advantageous for a consumer to have a seasonal variation of the price of cabbage than to have the price remain the same year round. The same diagram can be drawn for a producer with a rising supply curve, and thus one can come to the conclusion that fluctuating prices bring more profit to the producer. This property is generally known as the convexity of expenditure functions, or profit functions, with respect to prices. Therefore, the welfare cost of price and financial fluctuations results from the fact that the variation is not easily predicted.
Figure 7.2.Comparing Consumer Surpluses
Important costs in macroeconomics come from price-wage rigidity (e.g., Blanchard and Fischer, 1989). “Menu cost” is the term used to identify these costs. Its importance is clear if we consider the negotiating and settling cost of labor contracts. Because of the difficulty of changing prices and the difficulty of writing a contingent claim, exchange rate fluctuations and exogenous price fluctuations of other commodities may create output fluctuations of the product (of the production factor) that subsequently incurs the menu cost. Such output fluctuations may be optimal for the companies, given the cost of rewriting contracts. But for the economy as a whole, such an occurrence may constitute a serious output loss, in particular, in the period of recession. Because of uncertainty, some contracts that would be profitable are not negotiated. Then, the chain of credit is broken and welfare losses will develop, which are essentially the costs from instability.
In the case of a financial crisis, expectations are almost never met, with the result that the public loses trust in the financial system. The utility of holding money becomes negligible in the case of hyperinflation. In Figure 7.3, the inflation tax makes the cost of holding money prohibitively high, as in OB″. The amount of cash holding B″C″ will be far below what Milton Friedman called optimum, and the consumer surplus AB″C will be minimal. Because people do not inherently trust each other, the chain of trust relationships will be broken and long-term financial contracts will rarely be made (Kiyotaki and Moore, 1997). There is an asymmetric nature in financial contracts as is seen by comparing the effect of inflation with that of deflation. The value of collateral will be eroded during a recession and the downturn of trade cycles will yield a much deeper contraction of credit and production.
Figure 7.3.The Cost of Holding Money
Costs of Alternative Currency Regimes
Simple tools will be used to illustrate the above conceptual discussions with concrete examples, so as to appraise the relative welfare costs of financial instability and financial crises that have occurred in recent monetary history. Recent examples include the labor difficulties of 1992 before monetary integration in Europe, inflation and financial turmoil in Latin American countries, monetary confusion in the Baltic countries, Russia, and other countries of the former Soviet Union, and most recently, the Asian currency crises. In both the Latin American and Asian situations, the cost of financial instability, as well as financial crises, will be discussed. The maximum level of costs involved in financial turmoil will be measured by calculating the income loss incurred in the process of recovery from financial shocks.
In other words, a comparison will be drawn between the welfare loss of the countries due to instability or crises that depend on the international monetary regime they adopt. Here, it is important to consider both the direct cost of a regime to an economy and the opportunity cost (or gain) of adopting a certain exchange rate regime rather than other options. For example, Hong Kong SAR incurred the cost of high interest and recession in order to keep its Currency Board System (CBS). If, instead, Hong Kong SAR had adopted a flexible exchange rate regime it would have incurred another type of cost (or benefit). Chile and Malaysia might have succeeded for the time being in blocking the effect of inflows and outflows (respectively) of capital, but these gains were accompanied by the opportunity cost of giving up the benefit of free capital mobility needed for intertemporal allocation.
It is still useful in this case to recall the long-told “tri-lemma” or the impossible triangle of international finance. Fixed-exchange rates, capital mobility and monetary autonomy can hardly coexist. Sacrificing one or more of them usually incurs costs for the economy. Each item has its own merit. The fluctuation of exchange rates may add transaction costs and calculation costs. The absence of capital mobility impairs intertemporal resource allocation. The lack of monetary autonomy, of course, intensifies business cycles and the unemployment problem. Thus, no panacea exists in the choice of exchange rate regime. A country must sacrifice something.
Cost Estimates in the Recent Financial Instability and Crises
In order to compare the welfare costs of various monetary situations, it is useful to separate these costs into the following categories:
2. Currency Board System (CBS);
3. hyperinflation caused by fiscal and monetary indulgence of the government with flexible exchange rates or fixed exchange rates accompanied by strict capital control;
4. fixed exchange rate regime among industrialized economies with capital mobility;
5. fixed or highly managed exchange rate regimes supported by capital control;
6. fixed exchange rates or highly managed exchange rates disrupted by currency devaluation and floating rates; and
7. flexible rates among major industrialized countries or regions.
Dollarization or the Use of Another Country’s Currency for Domestic Purposes
In Panama, Argentina, Mexico, and Cambodia, the dollar is used as a medium of exchange, in addition to the local currencies of these countries. The dollarization movement seeks to abolish the central bank and to make the dollar, or another country’s currency, the legal tender of the country.
The major component of welfare cost caused by dollarization is the income loss due to the sacrifice of monetary autonomy that is used to counteract recession or unemployment. The country that has adopted dollarization must tolerate the consequences of the economic policies of the country that issues the currency in domestic use (for instance, the United States), regardless of whether that country’s monetary policy is too tight or too expansionary for the country that has dollarized. Another component of the opportunity cost of dollarization is the loss of seigniorage revenue. In 1995, the Marshallian k for M1 in Argentina was about 0.06. If this M1 is issued by the monetary sector and held by the public at an interest rate of 3 percentage points lower than the short-term interest rate, then 0.18 percent of the GDP would be lost each year by adopting dollarization.
On the other hand, there is an opportunity gain that exists with adoption of dollarization, if otherwise the economy may experience hyperinflation, as in category (2). If the United States does not pursue a strongly inflationary policy, then the country adopting the dollar may be relieved from hyperinflation and any resulting systemic crisis that its domestic central bank would have caused. Compared with the CBS, dollarization increases the credibility of the fixed parity at the cost of seigniorage. In short, dollarization recovers the consumer surplus of money by sacrificing the producer surplus of money at the central bank. Once the economy is dollarized, it is an irreversible decision, except in the case of the political break-up among countries that took place in the ruble region of the Baltics, Russia, and other countries of the former Soviet Union, as well as in the former Austro-Hungarian Empire.
The Currency Board System
Similarly, the Currency Board System (CBS) enforces monetary discipline. Dollarization deprives the domestic monetary authority of its role as the issuer of currency. Under the CBS, the role of money issuer remains with the central bank, but the discretion of the monetary policy is strictly limited. Seigniorage continues, but is relatively restricted by the monetary rule of the CBS. Monetary policy is linked to the central bank of a foreign country that is supposed to have a more stable discipline. Accordingly, the cost is the limitation of the monetary policy that aims at stable employment.
In Argentina and Hong Kong SAR, this characteristic is well observed. The recent stagnation of the Argentine economy can be partly explained by the difficulty of relaxing monetary policy because of the limitations of the CBS. The cost of sustaining the CBS was reflected in the slow growth of the economy (Table 7.1). To defend the CBS against the attack of speculators, Argentina had to rely on the built-in swap agreement with international monetary organizations like the Inter-American Development Bank. On the other hand, hyperinflation, which eliminated the consumer surplus of money in the 1980s, was well contained. An opportunity gain existed from not returning to a hyperinflationary situation.
|MPI3||Hong Kong SAR||96.4||107.7||122.5||123.0||100.0||110.4||123.4||120.1||97.1||106.3||117.2||116.3|
IPI = Industrial Production Index.
MPI = Manufacturing Production Index.
CPI = Crude Petroleum Production Index.
|MPI||Hong Kong SAR||94.2||105.4||117.9||116.1||90.4||100.7||105.3||99.7||81.7||93.8||98.7|
IPI = Industrial Production Index.
MPI = Manufacturing Production Index.
CPI = Crude Petroleum Production Index.
IPI = Industrial Production Index.
MPI = Manufacturing Production Index.
CPI = Crude Petroleum Production Index.
Hong Kong SAR succeeded in keeping its Currency Board in the face of attacks during the Asian currency crisis sparked by the devaluation of the Thai baht in mid-1997. Under capital mobility, the cost of defending the CBS against speculative attacks was sizable. Hong Kong SAR could afford to continue the CBS, essentially because its international reserves were sufficiently ample. Still, Hong Kong SAR had to defend Hong Kong dollars in 1997 and 1998, while its GDP showed substantial decline, as did that of many Asian countries (Table 7.1). Biases can, and do, exist in evaluating the economic performance of industrial production, but Hong Kong SAR certainly has lost a substantial percentage of manufacturing production from its 1995 level. The magnitude of decline is shown by the following data. In the first quarter of 1998, the IPI (Industrial Production Index) was down to 90 percent of the 1995 level, and in the first quarter of 1999, the IPI was down to 80 percent of the 1995 level. In Argentina, the decline of the manufacturing index has not been so prominent, but the economy is still staggering (September 2000). Thus, the welfare cost of the CBS is evident.
Hyperinflation Under Flexible Rates or Under Fixed Rates with Strong Capital Control
Because of the lack of monetary and fiscal discipline, many economies plunged into hyperinflation. (Figuratively speaking, many inflationary Latin American countries suffered from government failure, while Asian economies suffered from private or market failure.) The welfare cost of hyperinflation is comparable to the welfare cost of the systemic crisis considered earlier in the paper. In the case of systemic crisis, the function of money is seriously disrupted because of clearing troubles, bank runs, and Internet theft. As is explained in Figure 7.3, the function of money is impaired because people would hardly hold money because of the rapid depreciation rate. To make a bold statement, suppose a country has the Marshallian (M1) k of 0.3 under normal times with an interest rate of 5 percent, but it has reduced its money holding to k of 0.06 under an inflation rate of 100 percent. Then, the welfare loss of the Harberger money triangle will be 11.9 percent of GDP.
Fixed Exchange Rate Regime Among Industrialized Economies with Capital Mobility
As the tri-lemma thesis indicates, such a regime among industrialized economies with capital mobility is a difficult combination to achieve. The European Union challenged this objective and succeeded in creating a purely single currency, the euro, in 2000. The emergence of credibility among the participants was crucial and, in the end, an ideal situation. On the other hand, EU countries’ unemployment rates are still high. The real exchange rate of the euro to the dollar or to the rest of the world has lost more than 20 percent since January 1999 (as of September 2000). Does this mean that the market does not appreciate the benefit of a single currency in Europe, or that the future U.S. economy is gaining such credibility because of its unprecedented technological boom?
We can determine the welfare cost of the decline in growth among European countries after the currency turmoil of 1992 but, unfortunately, space prevents me from expanding on this idea in this paper. However, the magnitude of the losses appears to be fairly small relative to the magnitude of losses in Asia.
Fixed or Highly Managed Exchange Rate Regime Supported by Capital Control
Chile attempted to curb the inflow of capital by imposing requirements for deposits without interest. This policy prevents excess inflows of capital, based on the notion that the excessive trade deficit is unfavorable to economic development. This capital policy seemed to have succeeded for a while. Needless to say, distortion in the capital market involves welfare loss. The amount of loss could be calculated by the welfare echelon trapezoid of WXYZ shown in Figure 7.4
Figure 7.4.Welfare Loss from Capital Controls
Malaysia attempted to halt or restrict the outflow of capital by imposing an embargo on capital outflow. This was not a repellant but rather a hostage policy. As Table 7.1 indicates, Malaysia achieved a reasonable recovery under this policy of capital control. Aside from the loss from distortion, it is difficult to rank the Malaysian policy definitely below the IMF policy imposed on the Asian development strategy with flexible exchange rates and capital mobility.
Chile, on the other hand, attempted to halt the inflow of capital. This policy appeared to be successful until the end of 1997, as is seen in Table 7.1 Chile has currently lost the growth momentum it had achieved and has suspended capital control.
Fixed or Highly Managed Exchange Rates Disrupted by Currency Devaluation and Floating Rates
In Mexico beginning in 1994 and in Thailand, Indonesia, and South Korea, all in 1997, the attempts to defend the parity against speculative attacks failed and the countries’ currencies succumbed to a floating regime. Exchange rates precipitated. All countries listed in Table 7.1 experienced declines in GDP, which has since rebounded in every country except Indonesia, where the recovery has been slower, partly because of political turmoil. The welfare loss due to the financial crisis can be measured, at a maximum, by the gap of the actual growth path and the potential growth path as indicated. This is a maximum estimate because other causes certainly contributed to the decline.
The highly managed exchange rate or the fixed rate can be compared to an extremely tight rod of a steam engine. Such a rod forcibly suppresses the exit of steam pressure [exchange rates] until the engine breaks. Afterwards, because of the past compression, [the force of the steam causes] exchange rates to fluctuate widely, resulting in a sizable resource loss. Table 7.1 shows that Korea and Indonesia experienced at the trough in 1999 a decline of IPI more than 20 percent from the previous peak. In Thailand, the gap was 15 percent or so. The recession was a long one for Indonesia (lasting 10 quarters) and the economy is just now recovering (as of September 2000). Thailand’s recession was of a medium length (about eight quarters), while the recession in Korea was far shorter (about four quarters). In reviewing Table 7.1, we can see that by integrating the loss of production with the trend line, we can obtain the measure of the welfare cost in any country that faced a financial crisis.
Flexible Rates Among Major Industrialized Countries or Regions
Flexible rates that operate for some time are akin to a safety valve that dissipates pressure occasionally. Also, the ensuing speculation from one-sided expectations are difficult to assess. It is hard to gauge whether exchange rates among the dollar, the euro, the yen, and other currencies have settled down to levels that, on average, reflect the equilibrium real exchange rates. At least, however, the fluctuation among them has resulted in, at most, financial “instability,” and never in financial “crisis.” In the early 1970s, exchange rates fluctuated widely. Market agents learned much and provided counterbalancing expectations against bubble-like movements of currencies. The depth of the market that is capable of homogenizing a variety of expectations helped the stability of the exchange market, although coordinated interventions might also have contributed somewhat to the stability.
In the first section of this paper, I stressed the potential danger of a systemic crisis in the hypercybernetic world with system-wide psychological contagion among depositors, mechanical failures, and Internet terrorists. As was seen, the incidence of a systemic crisis can be substantial, while its probability may also be great.
In the second section of the paper, I showed that one has to sacrifice at least one or two of the three objectives (fixed exchange rates, capital mobility, and monetary autonomy) in any system. The decision as to what to sacrifice depends on the nature of national policy preference and the nature of the economic disturbances a country faces. The welfare cost derived from adopting a regime should always be compared with the opportunity gain of adopting the present system—that is, the cost of adopting an alternative system. From this point of view, it is difficult to rank economic performances among various international monetary regimes—for example, among the IMF implemented flexible exchange rates with capital mobility, more or less fixed exchange rates with capital control, and the Currency Board System.
BertautCarol C. and Murat F.Iyigun1999 “The Launch of the Euro,” Federal Reserve Bulletin Vol. 85 (October) pp. 655–66.
BlanchardOlivier J. and StanleyFischer1989Lectures on Macroeconomics (Cambridge, Massachusetts: MIT Press).
CorriganG.1981 “The Risk of a Financial Crisis,” in The Risk of Economic Crisised. by MartinFeldstein (Chicago: University of Chicago Press) pp. 44–53.
Committee on Payment and Settlement Systems2000 “Core Principles for Systematically Important Payment Systems,” Publication No. 34 (Basel: Bank for International Settlements).
DangelmaierPaul1993 “Nuclear Liability Insurance in the Federal Republic of Germany,” in Nuclear Accidents: Liability and Guarantees (Paris: Organization for Economic Cooperation and Development).
DiamondDouglas and Phil H.Dybvig1983 “Bank Runs, Deposit Insurance, and Liquidity,” Journal of Political Economy Vol. 91 (June) pp. 401–19.
FurfineCraig H.1999 “Interbank Exposures: Quantifying the Risk of Contagion,” BIS Working Paper No. 70 (Basel: Bank for International Settlements).
HarbergerA.C.1964 “Taxation, Resource Allocation and Welfare,” in The Role of Direct and Indirect Taxes in the Federal Reserve System (Princeton, New Jersey: Princeton University Press, for the National Bureau of Economic Research).
HumphreyDavid B.1984 “The U.S. Payments System: Costs, Pricing, Competition and Risk,” Monograph Series in Finance and Economics No. 1/2 (New York: Salomon Brothers Center for the Study of Financial Institutions, New York University).
HumphreyDavid B.1986 “Payment Finality and Risk of Settlement Failure,” in Technology and the Regulation of Financial Marketsed. by AnthonySaunders and Lawrence J.White (Lexington, Massachusetts: Lexington Books).
HumphreyDavid B.1994 “Payment Systems: Principles, Practice, and Improvements,” Technical Paper No. 260 (Washington: World Bank).
HumphreyDavid B.Robert H.Keppler and FernandoMontes-Negret1997Cost Recovery and Pricing of Payment Service: Theory Methods and Experience (Washington: World Bank).
HumphreyDavid B.Robert H.Keppler and FernandoMontes-Negret1998 “The Payment System and Monetary Policy,” IMF Paper on Policy Analysis and Assessment 98/4 (Washington: International Monetary Fund).
JohnsonOmotunde E.G. and others1998Payment Systems Monetary Policy and the Role of the Central Bank (Washington: International Monetary Fund).
KaufmanGeorge G.1994 “Bank Contagion: A Review of the Theory and Evidence,” Journal of Financial Services Research Vol. 8 (April) pp. 123–150.
KaufmanGeorge G.2000 “Banking and Currency Crises and Systemic Risk: A Taxonomy and Review,” Financial Markets Institutions and Instruments Vol. 9 (May) pp. 69–131.
KiyotakiNobuhiro and JohnMoore1997 “Credit Chains” (unpublished; London: London School of Economics).
MarroneJoseph1993 “Closing the Circle of Protection for the Public: The Evolution of the System in the United States,” in Nuclear Accidents: Liability and Guarantees (Paris: Organization for Economic Cooperation and Development).
McAndrewsJames J. and GeorgeWasilyew1995 “Simulations of Failure in a Payment System,” Working Paper No. 95-19 (Philadelphia: Federal Reserve Bank of Philadelphia).
OECD Nuclear Energy Agency (NEA) and The International Atomic Energy Agency1993Nuclear Accidents: Liability and Guarantees (Paris: Organization for Economic Cooperation and Development).
President’s Council on Year 2000 Conversion2000The Journey to Y2K: Final Report of the President’s Council on Year 2000 Conversion (Washington).
SatoSetsuya1998Kessai System o Design suru (Designing a Payment System) (Tokyo: Sigmabase Capital).
SatoSetsuya1999Money Shinkaron (An Evolutionary Theory of Money) (Tokyo: Sigmabase Capital).
SatoSetsuya and David B.Humphrey1995 “Transforming Payment Systems: Meeting the Needs of Emerging Market Economies,” Discussion Paper No. 291 (Washington: World Bank).
SaundersAnthony and Lawrence J.Whiteeds. 1986Technology and the Regulation of Financial Markets (Lexington, Massachusetts: Lexington Books).
SolomonElinor H.1997Virtual Money: Understanding the Power and Risks of Money’s High-Speed Journey into Electronic Space (Oxford: Oxford University Press).
SolomonElinor H.ed. 1991Electric Money Flows: A Molding of a New Financial center (Boston: Kluwer).
SummersBruce J.ed. 1994The Payment System: Design Management and Supervision (Washington: International Monetary Fund).
I am much indebted to Setsuya Sato for his valuable suggestions and to Fumiko Takeda for her invaluable research assistance. Conversations with Nobuhiro Kiyotaki and Carolyn Beaudin were also most useful.
Corrigan (1991) makes this initial observation, comparing systemic risk to a nuclear accident, as cited by Kaufman (1994).
As Humphrey, Keppler, and Montes-Negret (1997) demonstrate, the electronic system economizes the transactions cost of, say 2 percent, but the total consumers’ or money holders’ surplus will be eradicated by a systemic crisis.
This would be the case if the demand for money is a linear function of the interest rate, the elasticity of the demand for money with respect to the rate of interest is one-half, and if the Harberger triangle equals the market value of the monetary service, namely the product of the money demand and the interest rate.