How Should Policy Respond?

We start from the presumption that capital flows, like trade flows, are beneficial to a country and that policy should facilitate them. In particular, there should be an acceptance that an appreciated exchange rate is part and parcel of absorbing the capital flows and bringing about the transfer of real goods and services. But where there are significant interest differentials, there is a likelihood of excessive inflows as countries become more financially integrated, and these flows are likely to be volatile. A country may not be able to absorb, in a beneficial way, all the foreign capital that it attracts, or may want to iron out some of the surges (Schadler et al., 1993).

The policy responses might be grouped into two categories: those preventative measures done before the crisis, and those done to try to rescue or ameliorate the unfolding crisis. There is, however, substantial overlap, with some policy measures having a role before and after the crisis, with this role sometimes taking a different form. Let’s start by enumerating the policy options. First, the “before” options:

• “Sand in the wheels” to discourage inflow
• Capital controls or taxes on inflow (e.g., unremunerated reserve requirements)
• Strong prudential measures
• Hedging
• Fiscal policy
• Run current account surpluses
• Intervention to prevent the exchange rate overshooting prior to a crisis

Now, the “morning after” options:

• Higher interest rates to defend the currency
• Capital controls on outflows
• Better bankruptcy provisions and management of the private sector foreign debt
• Intervention to support the currency and soften the absorption adjustment

As foreign exchange market intervention overlaps these two phases, we leave to a separate section the important discussion of how this intervention will influence the central bank’s balance sheet.

1. Before the Crisis: Prevention

The two broad approaches here are to try to limit the inflow, and to prevent the exchange rate from overshooting in its appreciation.

(a) Sand in the Wheels

It would be possible to discourage the inflow by introducing various types of “sand in the wheels”: unstable politics, arbitrary administrative or judicial decisions, poorly functioning institutions, obscure information and random market processes, resulting in wide and unpredictable fluctuations in the exchange rate. It goes without saying that policy should be aimed at removing such imperfections, not using them as a policy instrument to solve a problem of excessive inflows. This kind of “sand in the wheels” is simply inefficient and denies the emerging country the benefit of the cheaper capital available overseas.

We noted, in Section B.1(b) above, that maintaining exchange rate uncertainty and volatility is one way of discouraging inflows. While this sort of disruption is widely accepted as the main explanation for time-varying risk, it seems sub-optimal.²³

There is another price-based mechanism at work. The foreign investment bids up the price of domestic assets (not just equities, but debt and property). This helps foreigners achieve portfolio equilibrium as the yield on the assets is driven down towards the foreign interest rate. There are, however, two disadvantages for the recipient country. First, as asset prices rise, an asset bubble becomes likely. Second, domestic investment is encouraged by the asset price increase (Tobin’s “q” operates), so the stance of monetary policy is undermined.

(b) Taxes on Inflows

Is there nothing better available in the policy armory to restrain excessive inflows? If we see the problem in terms of a price differential between the return on capital at home and abroad, policy might aim to ration the inflow while at the same time ensuring that the recipient country gets the full benefit of the fact that capital is available more cheaply in the world market: this is, after all, the usual benefit of globalization. If rationing is needed (and this is an issue of absorptive capacity), then a tax on inflows seems worth exploring, as it does the job and gives the benefit of the price differential to the home country (although, as usual, capacity to administer such a tax is an issue). The first measure might be to ensure that the foreign investment is fully taxed in the recipient country. International tax treaties aimed at avoiding double taxation tend to shift taxation out of the recipient country (where at most there is a smaller withholding tax), perhaps to some tax haven.²⁴ This may be hard to change, but would at least ensure than taxation does not act as a distortion working against macro-economic stability. A comprehensive capital gains tax would seem to have the same virtue. While a rigorously enforced capital-gains tax may not prevent an asset bubble from forming, it may exercise some constraint and the revenue will help clean up the damage when it bursts.

One preemptive response to excessive surges of foreign capital might be inflow controls– Chilean-style unremunerated reserve requirements (URR).²⁵ Mainstream discussion of these still has the flavor that, just as “real men don’t eat quiche,” serious countries do not have URR controls. This seems puzzling, as objective assessments show them to have been modestly successful over the policy horizon²⁶ and they seem closely tailored to the requirement to discourage the least useful and most disruptive form of inflow–short term funds. The negative consensus surrounding URR has been unhelpful to their effective use. Financial markets, carrying the Impossible Trinity baggage, were unanimously critical when Thailand attempted to introduce URR in December 2006, triggering outflows. If URR are a legitimate policy response, they require more in principle support from the IMF (and some technical help in implementation might have helped, as well). To the extent that they are often thought to be effective only for relatively short periods of time (until markets find easy ways to by-pass them), such measures might be thought of as being relevant to surges and the cyclical issues (i.e., trying to get more of the impact of monetary policy back to the interest rate instrument) rather than the structural issue.

(c) Prudential Controls

To the extent that the foreign inflows are coming through the domestic financial system, there seem many opportunities for stronger prudential controls, driven by the by-now-wellestablished fact that prudential problems in the downswing of the cycle were largely created during the upswing.²⁷ Policy should be bold enough not only to recognize the incipient problems, but to act on them. There is a good case for prudential regulations preventing or greatly limiting the role of the core financial institutions (banks) in intermediating the foreign inflows. So one answer to the second leg of the “twin crises”–the collapse of the banking system–may be to prevent the banks (and their subsidiaries) from acting as intermediaries for the inflow, and subject their customers’ whole-of-balance-sheet exposures to detailed prudential scrutiny and proper reserving practices. As often occurs, doctrinal or philosophical views get in the way of good policy. In this case, there is a commonly voiced argument that prudential measures should not be used for macro-economic purposes, but this is a misunderstanding of the nature of the problems: it is a prudential problem which also happens to have macro-economic implications.

A more radical option would be to build a core of narrow banks (i.e., banks which would hold only government securities as assets). This would not only assure the safety of a core group of deposits, but could assure the stability of a basic payments system in the event of a financial crisis (for a proposal along these lines for Indonesia, see Grenville, 2004(b)).

It might be worth noting that the types of deposit insurance widely introduced in Asia after the crisis (where small depositors are covered, but the larger depositors who make up twothirds or more of the volume of depositors’ funds are not protected) would have no effect in ameliorating a systemic crisis (see Grenville, 2006).

(d) Hedging

In the aftermath of the Asian crisis, there was a strong suggestion that the crisis could have been averted or greatly mitigated if only domestic borrowers had hedged their foreign exchange exposure beforehand.²⁸ To evaluate this, we need to separate the three different channels that come into play in a crisis. First, the exchange rate falls and this is a threat to inflation (which can lead the central bank to raise interest rates at a time when the economy is already weak). Second, the capital outflow requires an adjustment in the current account position to fit with the new (reduced) availability of external funding (we might call this the absorption effect). Third, the exchange rate fall administers a balance sheet loss to anyone with a currency exposure (which was so damaging to domestic corporations in the Asian crisis). With this three-fold distinction in mind, we can evaluate the effect of hedging. While it can shift the exposure around, the exchange rate vulnerability remains: if there is large capital inflow, then someone–either domestic or foreign–has taken on a currency mismatch. If hedging shifts the exposure from one resident to another, there would seem to be little macro-effect. If the exposure is shifted to foreigners, this shifts the balance sheet exposure to them and softens the effect of the crisis on domestic corporations. This may mitigate the crisis, but the remaining two effects–exchange rate fall and the need for current account adjustment through the absorption effect–remain, and may even be more severe. Foreigners will attempt to cut their exposure when the currency comes under threat, pushing the exchange rate down and creating the same pressure on inflation and the same need for current account adjustment in response to capital reversal.

A closely related debate goes under the catchy title of “original sin” (Eichengreen et al., 2005) which puts the currency denomination of foreign debt as the central issue. Hausmann (1999) explains the difference between Mexico and Australia (both big foreign borrowers, but one fragile and the other not) in terms of the ability of Australia to borrow in its own currency, while Mexico (having “original sin”) had to borrow in dollars, leaving its borrowers vulnerable to an exchange rate depreciation. This raises the same issues as discussed in the previous paragraph. Unless it can be shown that foreign investors are more stable holders of currency exposure than domestic borrowers, the vulnerabilities remain, whoever holds the exposure.²⁹

Our analysis questions the conventional wisdom of encouraging countries to shift the exchange risk to foreigners, thus ridding themselves of “original sin.” Certainly, this shifts the balance sheet damage of a depreciation to foreigners. But the country and its investors pay a significant premium for this risk shifting. Just as a Japanese investor would have been much better off by investing in Australian dollars,³⁰ an Australian borrower would have been significantly better off borrowing in yen over this period. Shifting the currency risk to foreigners gives them the benefit of the difference between the low international rates and the high domestic rates. Why is this universally regarded as good policy?

(e) Fiscal Surplus

The one policy prescription which seems to achieve wide support in theory if not in practice is to respond to excessive capital inflow by shifting the budget in the direction of surplus.³¹ This prescription seems to rely on the Mundell-Fleming IS/LM framework: a budget surplus will shift the IS to the left, lowering interest rates and discouraging capital inflows. This seems to fail on two levels. First, the IS/LM framework no longer captures the way monetary policy operates. The authorities set the short-term interest rate and have no reason to change this in the face of a large budget surplus and a leftward shift of the IS. Longer-term interest rates are set by the Wicksellian natural rate, which does not change. Even if interest rates did fall, the capital flows facing the countries of East Asia seem to be fairly interestinelastic, as they are now dominated by FDI (including direct purchase of assets such as infrastructure) and portfolio flows into equities. If this is the right framework, then the extra savings from the budget will shift the saving/investment balance and, pari passu, the current account towards surplus. If the same quantity of capital inflow has to be brought into equilibrium with a smaller current account deficit, this would seem to put upward pressure on the exchange rate, the exact opposite of the desired result.³²

(f) Run Current Account Surpluses

In their broad order of magnitude, the capital inflows into East Asia in the past five years have been around the same as in the first half of the 1990s, but their absorption has been fundamentally different. In the 1990s, for better or for worse, there were corresponding current account deficits, so the capital flows were transferred in terms of real goods and services. In contrast, and perhaps reflecting the trauma of the Asian crisis, these countries have run current account surpluses for the past decade, so the net inflows have, roughly speaking, gone straight into official foreign exchange reserves.

This makes the countries much less risk prone (see the informal risk ranking in The Economist on 17 November, 2007, page 76). This has been achieved by a combination of restrained growth, lack-luster investment climate and exchange rates which have been held down by intervention. This may be a natural reaction and aftermath to the crisis, and a significant reserve build-up can be justified (see below). As a longer-term strategy, however, it is mercantilist and ignores the need for maximizing growth opportunities.

(g) Intervention

Can the authorities use foreign exchange market intervention to prevent the exchange rate from rising too much in the pre-crisis period? If it does not rise too much, then it will not fall much. This is where policymakers need some operational notion of what the “right” exchange rate is. Probably the least palatable message that comes out of this discussion is that the authorities should be ready to allow the exchange rate to appreciate. They need to resist opposing the ongoing underlying structural appreciation and the appreciation which is the normal part of monetary policy during the upswing of the cycle. If they can identify any further overshooting, there is a fair chance that intervention will, at least, do no harm and will turn out to be profitable for the central bank. Topping and tailing the cyclical overshooting of the exchange rate seems not only possible, but desirable. This is not a doctrinal issue, simply one of operational capacity. Whether or not it changes the path of the exchange rate much, it gets policy to focus on the right issue–has the exchange rate overshot? The justifiable concern that the exchange rate may overshoot would suggest some variant on the Williamson band-basket-crawl (BBC) (see Williamson, 2000).³³ This has, in a fairly mechanical form, some of the characteristics of the Singapore exchange rate approach, which permits quite aggressive and determined intervention, but normally only when the exchange rate has moved significantly away from what is seen as the medium-term equilibrium.³⁴

Of course, any intervention has to be kept consistent with the monetary stance but, as we noted in Section 2, this is less difficult in practice than the Impossible Trinity implies.³⁵ The threat to the stance of monetary policy is more likely to come from a reluctance to keep interest rates at the proper level, rather than the use of intervention in the foreign exchange market.

2. Managing a Crisis

So much for prevention. When this fails and the “sudden stop” is impending or has begun, central banks have three possible responses: raise interest rates, intervene in the foreign exchange market, or impose capital controls.

(a) Interest Rates

Higher interest rates can help to retain fleeing capital, but not often, and never when the exchange rate fall is accompanied by a financial crisis (Goldfarjn and Gupta, 1999). During the Asian crisis, the reversals in Thailand and Indonesia were dramatic, and could not be countered by any realistically acceptable rise in interest rates. At an intuitive level, the central problem is that the prospect of an imminent depreciation will always outweigh the investors’ higher running return. We shouldn’t have had to learn this lesson in 1997: in 1992 the UK was unable to defend the sterling peg because the market knew that an interest rate defense was too politically painful to be maintained, and in the same year Sweden tried to impose 500 percent interest rates to defend the krona, ultimately unsuccessfully.

(b) Intervention

There is very little support for foreign exchange intervention in the academic literature, and it takes a brave (some would say foolhardy) central bank to stand against a serious bout of capital outflow. Nevertheless, this is what reserves are for, and if the authorities are not ready to use their reserves, then why bother to have them in the first place? Intervention has (at least) two aims:

• First, to discourage capital outflow by supporting the exchange rate.
• Second, to finance a continuation of the current account position, so as to avoid a forced turnaround which, in turn would force a sharp contraction in GDP.

While in practice these two aims are inexorably interwoven, they should be judged separately. Even if the intervention has no effect on the path of the exchange rate, intervention might well be justified by the extra time it buys for the absorption adjustment process to take place.

Why does intervention get such bad (academic) press? Once again it is tempting to put some of the blame on the strong presumption, held by many analysts, that the market always provides the right answer. Perhaps a stronger reason is that history provides plenty of examples of futile defenses of unsustainable exchange rates. The test, however, is not to lump together all the attempted defenses and to try to distil a single answer on whether intervention “works,” but to identify the circumstances in which it could work, and test these. This is, unfortunately, not easy: we cannot know the counterfactual path of the exchange rate and there is an intractable identification problem, in that we cannot distinguish between the policy reaction (intervene when the exchange rate is falling) and policy failure (the exchange rate falls despite intervention).

What we do know is that some central banks have consistently made a handsome profit over time by attempting to “lop the peaks and fill the troughs” (see Andrew and Broadbent, 2004). Others such as Japan have had good (if unacknowledged) success. Whether they succeeded in loping and filling is impossible to prove, but their profits suggest, at least, that private arbitrageurs are “leaving money on the table.” The experience of Singapore during the crisis suggests that a well-functioning economy can protect itself against depreciation over-shooting through intervention. The key, in this and other successes, is for the authorities to allow the exchange rate to move a significant distance before attempting a determined and well-resourced defense (and even then to be prepared to shift the defensive lines rather than be overwhelmed).³⁶ This takes a high degree of expertise and experience, backed by good administrative arrangements. Not every country will be able to emulate Singapore’s success.

Whatever the arguments about the effectiveness of intervention in influencing the path of the exchange rate, there will still be a case for using reserves to smooth the absorption adjustment in a crisis, and in Section 4(e) below we will return to the issue of what is a sensible level of reserves to hold for this purpose.

(c) Capital Controls

The academic literature is similarly unenthusiastic about capital controls, although after to the Asian crisis there appeared to be increased support for inflow controls of the Chilean type, as mentioned above. It is hard to find any support at all for outflow controls, and again this may reflect the reality that the loudest voices come from the creditor countries. Despite the frequently heard assertions of the sanctity of debt,³⁷ it is equally hard to see the philosophical objection: every country has domestic bankruptcy rules which are invoked, in extremis, to sort out the relative rights of debtors and creditors when the debtor is insolvent. The subtlety here is that there are both private debtors and debtor countries, so the essence of the issue is how to keep the arrangement confined to the parties immediately involved. These “consenting adults” made an agreement, and when it falls apart, the effects should ideally be confined to them. Rapid recognition of bankruptcies in 1997 would have fundamentally altered the way the Asian crisis played out, especially in Indonesia. Private debtors would not have been in a position to buy foreign currency to stave off their creditors (and by so doing, drive down the exchange rate). Rather, their balance sheets (and their checkbooks) would have been in the hands of a bankruptcy administrator who, in due course, would have negotiated a settlement with the creditors.

Given the undoubted success of the Korean standstill on bank debt at the end of 1997 and the importance of this in restoring stability and confidence, it might be thought that this would become part of the normal policy armory. Not so. It is treated as a unique occurrence in unusual circumstances. It would have been impossible, it is said, to make deals with all the widespread creditors in the other cases. As a generalization, this is clearly nonsense: it could be done in the same way that domestic bankruptcy administrators work, by an administrator simply announcing that the business is insolvent, and having creditors come forward to register their claims, which are dealt with in good order. This would, however, require some international endorsement to avoid individual creditors jumping ahead in the line, and it has not been possible to get international endorsement of orderly debt resolution even in the far simpler sub-case of sovereign debt restructuring.

There is one further policy measure, related to foreign exchange intervention, which gets little discussion but seems to have been effective in Brazil in 1999 (see Bevilaqua and Azevedo, 2005). Rather than use its foreign exchange reserves to sell into the market, the government can issue debt denominated in dollars (through either new budget financing or by rolling over existing debt). This provides dollar-denominated assets which the market can use to provide currency cover for those who would otherwise have bought dollars in the foreign exchange market. Of course the government is taking on currency risk, so the Brazilian example should not be copied. This should only be done if the currency has overshot and is likely to appreciate.

3. Managing the Central Bank Balance Sheet

We have suggested, above, that intervention both to constrain overshooting in appreciation before the crisis, and to restrain the depreciation during the crisis, may well be justified. This involves significant foreign exchange reserves (see Figure 8 [ PDF 24.2KB | 1 pages ]), usually on the balance sheet of the central bank. This raises operational and policy issues which we address here.

Large capital inflows lead to intervention and increases in foreign exchange reserves (see Figure 9 [ PDF 24.2KB | 1 pages ]). One measure of the pressure on the exchange rate is shown in Figure 10 [ PDF 21.9KB | 1 pages ]. The reserve build-up presents two problems for the management of a central bank’s balance sheet. First, the central bank has a foreign exchange exposure, often very large, which threatens its capital position in the event of appreciation. Second, the earnings on these foreign exchange assets are often smaller than the cost of issuing the sterilization instrument, putting the central bank’s profit at risk.³⁸

Despite these two potential-cost factors, reserve holding may represent a proper policy choice: even where the central bank makes losses, the country as a whole may make offsetting gains. Alternatively, the investments may be thought of as a sensible selfinsurance policy against foreign capital flight. If we consider that the cost of the crisis in Indonesia is reflected in a level of income which is around one third lower than it would have been had the crisis been avoided, and that this is an ongoing loss (it was not a “V” shaped recession), if reserves could have been used to avoid or mitigate such a crisis, the return on reserve-holding would be very high. But no central bank wants to go, cap in hand, to the government for a recapitalization if either of these factors puts its solvency in question.³⁹

Rodrik (2006) sets out the cost of foreign exchange reserve holding for emerging countries as a whole, putting it at around 1 percent of GDP (see Figure 11 [ PDF 39.3KB | 1 pages ]). He sees this as a selfinsurance policy worth taking. Table 2 [ PDF 17.3KB | 1 pages ] illustrates the magnitudes of these two problems for a number of East Asian countries. The ongoing cost of financing reserve holdings seems quite modest (smaller than Rodrik’s estimates). Ho and McCauley (2008) confirm this view (see Figure 12 [ PDF 37.7KB | 1 pages ]).

The risk of valuation losses in the event of an appreciation seems more substantial. A hypothetical scenario that might provide the broad order of magnitude is to imagine a 20 percent appreciation of the yuan, impinging on foreign exchange reserves equal to half of GDP. In this case, the accounting loss for the central bank would be equal to 10 percent of GDP.⁴⁰ Table 2 [ PDF 17.3KB | 1 pages ] sets out some more detailed calculations.

In one sense, these capitalization and profit issues are accounting problems which could be handled by some inventive inter-governmental accounting–through the addition of some government bonds to the central bank balance sheet. Just as central bank losses arising from support of a failing banking system must be swiftly made good by the budget without threatening central bank independence, so too revaluation losses should be made good swiftly and without condition.

This leaves open the more important policy issue of whether these foreign exchange exposures are in the nation’s interest, and whether the investment in often-low-return assets is sensible. This issue seems unresolved in the literature, although we have moved beyond judging reserve adequacy by comparison with the imports bill (traditionally three months of imports), recognizing that the capital account is now where the threat lies.

One often-heard suggestion is the Guidotti Rule (Greenspan, 1999), which proposes that emerging countries should hold foreign exchange reserves equal to the debt that will fall due over the next year. If this is interpreted as the longer-term debt falling due over the next year, it might make some sense as insurance against difficulty in rolling over the long-term debt (and this may be the issue for Latin America). However, if the reserves are being held against the short-term debt liabilities (which would be the case in East Asia), it raises the issue of why the short-term debt was a good idea in the first place. The Guidotti rule is, however, a reminder that the old rules of thumb connecting recommended reserve holdings with imports are not relevant in a world where the shock comes to the capital account.

A more fruitful argument is found in Jeanne and Ranciere (2006), who note the role that reserves could have played in averting the dramatic fall in absorption which was forced on the crisis countries of Asia as they turned their current account deficits into surpluses in order to meet the funding constraint. They note that, in a large sample of “sudden stops,” the average output loss was 4.5 percent of GDP in the first year and 2.2 percent in the second. Their model requires input of parameters covering risk factors and other unknowns and allows for little interaction between the level of reserves and the likelihood of a sudden stop, but seems to be the basis for a sensible approach to assessing reserve levels.

All this does, however, leave a huge policy issue largely unaddressed in this discussion. It might be possible to explain the build up of emerging economies’ foreign exchange reserves in terms of self-insurance against volatile capital flows. But when they amount to more than one third of GDP for the countries taken together, and for the PRC, to more than half of GDP, is it sensible policy for this to continue? A current account surplus of over ten percent of GDP and growing suggests a lack of sustainability, and the size of the potential valuation losses is a reminder that, seen in terms of self-insurance, the premium may turn out to be high. One often-proffered answer to upward pressure on the exchange rate–tighten fiscal policy–seems inappropriate for the PRC, with its existing huge saving surplus.⁴¹ Whether the answer is found in further freeing of capital outflow (which might involve significant public participation, as in the case of Singapore), or stimulus to domestic consumption, or more significant appreciation of the currency, remains in the realm of future policy challenges.

Download this Discussion Paper [ PDF 557KB| 36 pages ].

Comment(s)

There are [0] comment(s) for this entry. Post a comment.