|Dealing with volatile external finances at source: the rôle of preventive capital account regulations
José Antonio Ocampo and José Gabriel Palma
Paper to be published in J A Ocampo and J Stiglitz (eds.), Capital markets Liberalization in Developing Countries,
Oxford University Press, 2006.
Since the 1970s, developing countries have faced significant problems with their capital accounts in the face of rapidly expanding but highly volatile and unregulated international financial liquidity. In the case of Latin America, few practical economic problems have been as complex as the question of how to handle sharp externally-generated financial cycles in economies that have some of the most open capital accounts in the Third World. These financial cycles have been characterized by a phase of large inflows (such as those created by bank lending in the 1970s and by portfolio inflows during the 1990s), followed by sharp reversals and large negative transfers (1980s and 1999-2003).1 A similar process was experienced by the Asian economies in the 1990s, leading to a crash in 1997 that brought the world financial system to the brink of a global crisis.
For developing countries, the crucial issue in this regard is whether domestic markets can deal with these sharp financial cycles on their own; and in particular, whether domestic financial and asset markets can absorb positive and negative external shocks of this magnitude without manias, panics and crashes. A closely related issue is whether the real economy has the capacity to adjust effectively to the rapidly changing economic environment that results from these cycles, and the cost of these adjustments. In terms of policy in a capital-importing developing country, it is not obvious what the optimal policy-mix for dealing with these cycles would be; and, specifically from the point of view of this paper, there is the question of what would be the role of preventive capital account regulation on capital flows in this policy-mix.
Unfortunately, recent experiences of developing countries with open capital accounts have shown that these sharp financial cycles are highly likely to lead to financial crises. There are three basic reasons for this. First, during periods of surges of inflows the incentive mechanisms and resource allocation dynamics of domestic financial markets have failed under the pressures generated by the increased liquidity brought about by these surges.2 As a result, borrowers and lenders ended up accumulating more risk than was privately, let alone socially, efficient. This risk has become evident in the alternate phase of the cycle, that of the ‘sudden stop’ in external financing.3 Second, the real economy has found it extremely difficult to deal with financial cycles of the magnitude experienced by countries with open capital accounts. Indeed, in developing economies characterised by significant currency mismatches in their financial portfolios some typical adjustment mechanisms -- such as the relative price (real exchange) adjustment -- have failed when faced with sharp changes in external liquidity; instead of helping to bring these economies back to equilibrium, these adjustment mechanisms have tended to augment the cycle through their pro-cyclical wealth effects. And third, in financially-liberalised economies governments and central banks have found their degrees of freedom to adopt counter-cyclical macroeconomic policies to offset the effects of these externally generated cycles seriously limited; indeed, market forces have often pushed them into opposite, pro-cyclical, macroeconomic policies.
These recurrent financial crises in developing countries have opened up political and academic debates at least on two fronts. First, on the issue of whether these crises have been the result of market interferences (mostly in the form of unhelpful government interventions that distorted the working of otherwise efficient financial and other markets), or the result of market failures proper; second, on the policy responses most likely to help developing countries to absorb the sharp swings in capital inflows characteristic of the last two decades.
Regarding the first issue, those who argue that it was mainly exogenous destabilizing mechanisms that led developing countries into recent financial crises have placed their emphasis on both the moral hazards created by government deposit insurance and bailouts by international institutions, and the ‘crony-mechanisms’ that have often distorted the access to finance. In turn, those who emphasize the existence of endogenous market failures are basically concerned with the study of how the combination of a particular type of international financial market and a particular form of financial liberalization in developing countries may have led to the creation of an economic environment in which the maximization-cum-equilibrium process failed. One of the issues at the centre of the resulting debate regarding the policy options most appropriate to reduce the likelihood of financial crises is the relative effectiveness of price-based and quantity-based controls of capital inflows.
This paper reviews the experience of three developing countries -- Chile, Colombia and Malaysia -- in this regard. After having opened up their capital accounts, these countries decided to fine-tune their integration into international financial markets via the reintroduction of capital controls in order to deal with the 1990s surge in private capital inflows, particularly short-term portfolio inflows. The next section of this paper looks at the dynamics of capital flows into developing countries, and presents some contrasting experiences of Latin America and East Asia. The next analyses the nature and effects of regulations on capital inflows in these three countries during the 1990s, in particular their effects on the magnitude and composition of capital flows, on the macroeconomic policy space that the authorities enjoyed under these circumstances, and on asset prices.
II. THE DYNAMICS OF CAPITAL FLOWS: THE CONTRAST BETWEEN LATIN AMERICA AND EAST ASIA
One of the most clear stylised facts of developing countries’ access to international finance is that when international financial markets have opened up their doors to these countries they have done so in the form of sharp cycles (e.g., the 1820s, 1860s, 1890s, 1920s, 1970s and 1990s). Figure 1 shows this phenomenon during the last three decades.
total = total net capital inflows; private = net private inflows. In this and other graphs below that show values in terms of 3-year moving averages, the value shown for the last year is the actual value of the year.
Source: IMF (2004a). With respect to Africa, as the IMF data-set only includes information for Sub-Saharan Africa, in order to have an estimate for the whole continent (and despite the problems of mixing data from IMF and World Bank sources), data for North Africa (Algeria, Morocco, Tunisia and Egypt) was added to that of Sub-Saharan Africa using information from World Bank (2004).
A remarkable feature of the 1990s cycles is that while the 1970s surge in inflows came after half a century after the previous boom (1920s), the one in 1990s took place less than a decade after the 1982 debt crisis (see Figure 2).
Sources: 1930-1949, B. Stallings (1990; this source only including inflows from the USA); 1950-1969, ECLAC Statistical Division (during this period, portfolio inflows include a very small amount of government bonds because Latin American countries report their balance of payments according to the IMF methodology, revision 5; in this methodology, under ‘net portfolio inflows’ public and private sector bonds are reported together); and 1970-2002, IMF (2004a).
In the case of Latin America, one well-known reason for this rapid return of inflows is that the US government had no choice but to act in the late 1980s to sort out the problems left by the 1970s cycle and the subsequent problems of the highly exposed US banks. This eventually led to the Brady Plan, which effectively created a secondary market for Latin American securities; this new market not only helped US banks to unload their Latin American debt but also facilitated the insertion of Latin America -- and emerging markets in general -- into the booming security markets that had been developing in the industrial world in the 1980s.
Although these capital flow cycles were basically external from the point of view of the developing world, domestic policies did play a role in the specific timing of the capital account booms, in their composition and in their effect on the recipient countries. Figure 3 shows the remarkable turnaround in the capital account of some Latin American and East Asian countries after their respective processes of domestic financial and capital-account liberalizations; this striking turnaround generated the dynamics that eventually led to financial crises in these countries.
Col = Colombia; Mex = Mexico; Arg = Argentina; Malay = Malaysia; Thai = Thailand.
In each case the period ‘between’ covers the years between financial liberalisation and financial crisis -- Chile, 1975-82; Colombia, 1993-1998; Mexico, 1988-94; Brazil, 1992-98; Argentina, 1991-2001; and Korea, Malaysia and Thailand, 1988-96. The period ‘before’ covers a similar number of years before financial liberalisation (in the case of East Asia and Argentina, however, as the period ‘before’ would have included years preceding the previous 1982 debt crisis, only years since 1982 have been included).
Source: World Bank (2004).
It is also widely recognized that the macroeconomic policy followed by different countries during periods of booming inflows also influenced the vulnerability of countries once a sharp turnaround in the capital account had taken place. This paper deals with precisely that positive role that preventive regulations on capital inflows had on a group of developing countries willing to apply them.4
The cyclical movement of the aggregate net inflows hides the sharply different dynamics of the three main components of private capital inflows and, in particular, their sequential booms: first it was credit-inflows (particularly bank lending), then portfolio-inflows and finally FDI-inflows. This is illustrated in Figure 4 for Latin America through the evolution of the net transfer of resources.
LATIN AMERICA: Net Transfer of Resources and its
Net transfer of resources = net inflows minus net factor payments; as mentioned above, net inflows includes ‘errors and omissions’ but excludes the use of IMF credit, IMF loans and exceptional financing. 3-year moving averages.
Source: ECLAC’s Statistical Division.
In the case of bank lending, the boom of the 1970s was followed by an extended period of net resource outflows that has lasted more than two decades. Indeed, strikingly, this source continued to generate strong negative resource flows to Latin America even when the Asian economies were experiencing a surge in bank lending between 1990 and 1996.
At the same time, the net transfer of resources accumulated during the boom periods -- US$(2000) 236 billion from 1972 to 1981 and US$(2000) 143 billion from 1991 to 2000 -- are moderate compared with both aggregate net inflows during these years (US$ 499 billion and US 638 billion, respectively), and the large amount of resources lost during the periods of negative transfers (US$ 649 between 1982 and 1990 and US$ 209 billion just between 2001 and 2004)5. And, even then, these two periods positive transfers (1972-1981 and 1991-2000) are among only a very few years out of the last two centuries when the net transfer of resources has actually been positive for Latin America.
A comparison of the dynamics of the relationship between changes in current and capital accounts in Latin America and East Asia also reveals two crucial differences between these two regions, differences that are critical for an analysis of the appropriate macroeconomic and regulatory policies. One is the asymmetric nature of the current account cycle in Latin America; i.e., Latin America’s inability to generate current account surpluses even at times of domestic recession and net capital outflows (leading to foreign exchange reserves depletion, debt rescheduling and defaults).6 East Asia, of course, does not suffer from this problem being perfectly able to generate substantial current account surpluses when required (e.g., late 1980s and after the 1997 financial crisis; see Figure 5). This Latin American asymmetry, which has tended to truncate the current account cycle in the up-swing, is mainly the result both of the problems generated by the composition of its exports and of its endemic under-investment in export-diversification. One of he main implications of this difficulty in generating current account surpluses is that it inevitably affects its capacity to repay principal, thus increasing the likelihood of running into Ponzi-type finance.
NET PRIVATE INFLOWS AND CURRENT ACCOUNT
East Asia (3): Korea, Malaysia and Thailand.
Sources: Latin America, 1950-1969, ECLAC’s Statistical Division; 1970-2002, World Bank (2004). East Asia, (2004; data available only from 1975).
The second difference between the regions in this respect is the contrasting time-sequence that characterizes the dynamics of the relationship between the current and the capital accounts. The issue here is the driving force behind the balance of payment cycle: does the current account ‘lead’ the capital account, or is it the other way round? This issue can be tested with the help of the Granger ‘time-precedence’ or ‘predictability’ test (normally -- and misleadingly -- called the Granger-’causality’ test). The results of this test show a major difference between Latin America and East Asia in this ‘chicken-and-egg’-type problem. In Latin America, changes in the capital account tend to precede (and are useful for the prediction of) changes in the current account, while in East Asia the time-dynamic is the opposite (see Appendix).7 This is an important finding from the point of view of the subject of this chapter because if the primary source of the Latin American financial cycle is an externally induced (i.e., mostly exogenous) change in the capital account, controls on capital inflows are more likely to be a ‘first-best’ counter-cyclical policy option if one wants to deal with the excesses of the financial cycle at its source.
In East Asia, meanwhile, the results of the test indicate a more macroeconomic textbook time-sequence: changes in the current account -- mostly the result of developments in the domestic real economy -- precede, and are a useful variable to predict, changes in the capital account. It therefore seems less clear that in this region controls on capital inflows should be the dominant component of a ‘first-best’ policy to deal with the inflow-problems at source. This does not mean that capital controls cannot help -- as discussed below, inflow-controls in Malaysia in 1994 were remarkably effective in the short run. It means that the ‘first-best’ policy for dealing with the excesses of the financial cycle in this case would be to deal with the domestic problems leading the changes in the current account; i.e., mainly the huge (and seemingly uncontrollable) levels of deficit-finance needed to sustain the levels of corporate investment required by their development strategies. For example, in the case of the Republic of Korea (mainly due to declining profitability -- a decline that had little to do with productivity trends and much with collapsing micro-electronic-prices8) the corporate sector had to finance its high, but relatively stable, levels of investment (of just under 30% of GDP) switching from own-profits to external finance. This change led to a rapid increase in the sectoral deficit of the corporate sector, from about 5% of GDP (1987) to nearly 20% of GNP (1996), which absorbed not only all the increase in the surplus of the capital account but also that of the household and government sectors as well (Palma, 2003).
Not surprisingly, Latin America and East Asia followed different routes to their respective financial crises of the late 1990s. Even though there was a similarity in terms of the surges in capital inflows and in the speed of credit expansion between these two regions, there also was a crucial difference in the use made of this additional credit. While in Latin America additional credit to the private sector was mainly directed towards increased consumption and asset speculation, in East Asia it was used to sustain high levels of corporate investment (in the face of falling profitability). As mentioned above, the difference was related to the macroeconomic dynamics that had led to booming capital inflows in the first place. In East Asia, it was mainly a finance-for-investment ‘endogenous pull’; in Latin America, instead, it was a more mixed response. First, this region also produced an ‘endogenous pull’ of inflows but this pull was very different nature than that of East Asia; it was based on the need to attract foreign finance to help service the region’s already huge foreign debt in the face of its inability to generate current account surpluses. Second, added to this, there was a crucial ‘exogenous push’ component in the surge of foreign capital: portfolio flows is search of new market outlets (see Figure 6).
Source: IMF (2004a).
Among the many macroeconomic challenges that emerged, of course, was how to absorb the surge of liquidity resulting from non-sterilised inflows; an age-old mechanism was followed: additional inflows eased the access and reduced the price and the transaction costs of liquidity. In turn, easy access to cheap credit fuelled expectations regarding the performance of the economy, a performance that was enhanced by the additional expenditure brought about by extra borrowing and availability of foreign exchange. Thus, for a while, improved expectations fed a self-fulfilling prophecy. As a result, from the point of view of this ‘exogenous push’ of foreign capital, in Latin America ‘over-lending’ and ‘over-borrowing’ were not only the result of a closely interrelated process, but also one that had a clear direction of causality: the propensity to ‘over-lend’ led to the propensity to ‘over-borrow’.
From this perspective, therefore, the path to financial crises in Latin America (except for Brazil9) started with a sudden surge in (mainly portfolio) inflows, an explosion of credit to the private sector and low levels of interest rates (after stabilization); all these produced rapid real-exchange appreciation, consumption booms, asset bubbles in stock and real estate markets, a reduction in savings, and a deterioration of current accounts (massive ones in several cases). In the meantime, foreign-debt levels soared and their term structure deteriorated. It did not take much for this Latin route of inflow-absorption to lead to a state of such financial fragility that a sudden collapse of confidence and withdrawal of finance were almost inevitable, leading to major financial crises.
In East Asia in turn there was also a massive surge in inflows, an increase in private credit and low interest rates, but there were no consumption booms and only more limited asset bubbles (none at all in Korea, and in Malaysia and Thailand asset bubbles that were not anywhere near as large as in Latin America). Rather, in the context of rapid technological change and declining profitability, there were high levels of investment that ended up producing remarkably high corporate debt/equity ratios. Despite stunning growth records, a high degree of competitiveness and fundamentals that although not perfect were the envy of most developing countries, this East Asian route to financial fragility also led to financial crisis.
III. CAPITAL ACCOUNT REGULATIONS IN CHILE, COLOMBIA AND MALAYSIA
A. The nature of the regulations
In a world of highly volatile and unregulated international liquidity, the basic task of capital controls is to help both avoid the pro-cyclical dynamics created by surges in inflows and to counteract the strong market-incentives to adopt pro-cyclical macroeconomic policies that tend to develop in economies that have chosen to integrate into international capital markets with open capital accounts (Ocampo, 2003a). From this perspective, there are two basic issues for developing countries; the first is how to deal with the remarkable volatility of capital flows, particularly in some of its components -- see Figure 6. The second, the fact that the liberalisation of capital accounts, instead of creating ‘automatic stabilisers’ vis-à-vis these volatile flows, they tend effectively to create (in the words of Stiglitz, 2003) ‘automatic destabilisers’. In this context, preventive capital account regulations attempt to deal both with the first problem at source, and to give monetary authorities more room for manoeuvre in order to deal with the second.
In the 1990s most countries in Latin America and some in East Asia opted for rapid liberalisation of their capital account. However, the experiences of China, India and Taiwan show that ‘unimaginative’ integration into international capital markets was a choice. Their alternative, a more selective path of participation in international capital markets, proved a more effective way of avoiding the pro-cyclical dynamics of unrestricted capital flows.10 What the experiences of Chile, Colombia and Malaysia have in common is the fact that although they initially opted for a process of rapid integration into international capital markets they later decided to ‘fine-tune’ this integration using different forms of preventive regulations to control unwanted capital inflows.
The central instrument used by Chile and Colombia was price-based regulation, namely unremunerated reserve requirements (URR) for most foreign-currency liabilities. The advantage of this system was that it created a simple, non-discretionary and preventive price-based incentive to regulate capital inflows; in particular, these measures penalized short-term foreign-currency liabilities (mainly portfolio-inflows). In the case of Chile, these price-based capital account regulations were established in June 1991 and strengthened in May 1992 (some other modifications were also made in between, particularly the closing of a major loophole: foreign currency term deposits in domestic institutions). Capital inflows were subject to a flat-rate foreign-currency deposit in the Central Bank, originally 20% but raised to 30% in May 1992; these unremunerated deposits continued until June 1998. These deposits were originally meant to be kept for a three-month period, but they were soon extended to 12 months. In 1995 these regulations were strengthened, particularly by extending them to some portfolio flows that until then had been exempted from reserve requirements (another major loophole in the system). There were also other less important changes throughout the period, particularly those aiming at closing loopholes and adjusting other secondary provisions.
In 1993 Colombia adopted the unremunerated reserve requirement system; initially, these reserves only applied to credits with maturities below a specified term (initially 18 months), with the amount of the deposit originally being inversely proportional to the term of the credit. It is interesting to emphasize that the introduction of this system was initially conceived to help liberalise the capital account, as it replaced a more traditional system of capital controls in which regulations were based on the final use of the loans.11 Reserve requirements were modified in Colombia more frequently than in Chile, including not only changes to the minimum maturity of loans subject to regulations but also to the rates and maturities of the reserve requirements (Ocampo and Tovar, 2003). In 1994 the regulatory system was strengthened on two occasions; initially the minimum maturity subject to the deposit was increased to three years and later again to five years. However, in early 1996 the regulatory system was put in reverse and the minimum maturity was lowered back to three years and a unique deposit rate was established; regulations were strengthened again in 1997, when the minimum maturity was restored to five years12; soon afterwards, the whole system was replaced by a scheme more clearly resembling that of Chile, a system which applied to all loans and thus eliminated the principle of a minimum maturity-period. The main difference from the Chilean system was that the deposit (originally 30% of the inflows, which had to be kept in the central bank for 18 months) was made in the local currency and was therefore not protected from devaluation.
In both Chile and Colombia, economic agents could opt to substitute the unremunerated deposit with a one-off payment to the Central Bank of a sum equivalent to the opportunity cost of the deposit. This made the regulation into a de facto ‘Tobin-type’ tax -- i.e., a fixed cost for external borrowing. However, by Tobin tax standards, the tax was very high: in Chile it was about 3% for one-year loans during most of the period, and tended to fluctuate in response to changes in certain macroeconomic factors, such as international interest rates. The level for Colombia was higher, being on average equivalent to 13.6% for one-year loans and 6.4% for three-year loans during the 1994-1998 period. In the case of Colombia, the domestic interest rate and devaluation expectations also determined the magnitude of the implicit tax. These taxes were meant to have a counter-cyclical role, which is why they were raised (particularly in Colombia) during periods of surges in capital inflows and lowered -- eventually to a zero rate in both countries -- when external conditions deteriorated following the 1997 East Asian crises and the 1998 Russian default.13
Although the central feature of the Chilean and Colombian system was price-based, other administrative regulations on capital flows complemented reserve requirements; to contrast them with price-based regulations, we will refer to them as ‘quantitative’ in nature. In Chile, all inflows (including FDI) were subject to one-year minimum-stay requirements (a requirement that was lifted in May 2000), and the issuing of ADRs and similar instruments were subject to minimum amounts of issues and adequate risk classification; they were also subject to direct approval by the central bank. In Colombia, the Superintendence of Securities regulated the amount of the funds that portfolio investors could bring into the country and their domestic use, as well as bond and ADR issues made by Colombian firms on foreign markets. And although, unlike in Chile, trade loans were exempt from reserve requirements, other types of regulation were used to control this type of borrowing (minimum repayment periods for imports, except capital goods). Finally, in both cases, the reserve requirement implied an obligation to register all loans at the central bank. In Colombia, this included short-term commercial credits, which prior to the regulation had not been subject to this requirement.
Malaysia also offered major innovations in the area of capital account regulations in the 1990s, but relied more on provisions of a quantitative nature. After a surge of net private capital inflows that in relative terms climbed to heights that could probably claim a place in The Guinness Book of Records -- net private inflows (including errors and omissions) increased as a share of GDP from minus 3% in 1988 to 25% in 1993 (see Palma, 2000a) -- the Malaysian authorities decided to take radical action. To stop this surge dead in its tracks, in January 1994 this country adopted a series of drastic measures that were mostly quantitative in nature. It prohibited non-residents from buying a wide range of short-term securities, placed limits on non-trade-related liabilities of commercial banks, and prohibited commercial banks from making swaps and forward transactions with foreigners (see Nagera Bank, 1994; Park and Song, 1997; Ötker-Robe, 2000; and Palma, 2000a). Also, deposit interest rates were drastically reduced -- real deposit rates fell from an annual average of 4.2% in 1993 to one of minus 0.9% in 1994, and real lending rates from 6.2% to 1.8%, respectively (Palma, 2000a); this was done in order to reverse arbitrage flows, both passive and active ones.
As these measures were so drastic, and as they included such a strong quantitative component, the effect was not only immediate but also remarkable; so much so that as early as August of the same year, some of the controls were already beginning to be lifted, and by the end of the year most had disappeared. The Malaysian authorities thus seem to have developed some ‘overshooting’ anxiety.14 In fact, net private inflows fell in one year by no less than 18 percentage points of GDP! These measures seem to have been particularly effective vis-à-vis short-term flows, which fell by more than 13 percentage points of GDP in just one year. Although these restrictions were lifted relatively soon, some in August 1994 and others in January 1995, the shock effect they generated seems to have affected the expectations of economic agents in a more permanent way than the Latin American-style price-based mechanisms (see below).
The other Malaysian innovation came in 1998 with the East Asian financial crisis but applied to capital outflows. These regulations were aimed at limiting ringgit speculation by both residents and non-residents, particularly by eliminating offshore trading of the domestic currency. As in the case of the 1994 regulations, they effectively segmented access to the domestic financial transactions between residents and non-residents. In February 1999, a price-based instrument, an exit tax, which would be phased out in the following years, replaced this regulation. As these regulations affected outflows rather than inflows, we will not consider them in this paper.15
B. Policy objectives of regulations
As an extensive literature on the subject has emphasised, the accumulation of risks during periods of surges of capital inflows will depend not only on the flow imbalances that can eventually lead to unsustainable private and public-sector debts, but also on their effects on corporate balance-sheets and asset prices (especially stocks and real estate). Regulations on capital inflows, therefore, have three potential roles (Ocampo, 2003a, 2003b). The first is as a macroeconomic policy tool; the key aims are to provide some room for counter-cyclical macroeconomic policies, to help to cool down aggregate demand, and try to avoid the accumulation of unsustainable debt burdens. The second role is as a ‘liability policy’. At an aggregate level, they can help to avoid both risky corporate balance-sheet structures (especially due to excessive reliance on short-term external debts), and the worst effects of the volatility of capital inflows; at the firm level, capital account regulation can help avoid the maturity and currency mismatches typical of private sector financial structures in developing countries. Finally, a crucial role of capital controls is to help avoid asset bubbles, given the sharp cyclical pattern that characterizes asset prices in developing countries.
Viewed as a macroeconomic policy tool, regulations on capital inflows can provide some room to ‘lean against the wind’ during periods of financial euphoria, through the adoption of a contractionary monetary policy and/or reduced appreciation pressures. Furthermore, if they are effective in reducing the magnitude of inflows, they can reduce or eliminate the quasi-fiscal costs of sterilized foreign-exchange accumulation. Their role will depend on the dynamics of capital inflows and their relation to current account deficits and their domestic counterparts (savings and investment behaviour). If capital flows generate an ‘exogenous push’ on current account and domestic imbalances, as they seemed to have done in Latin America, they can help to control the direct source of the financial disturbance. If they respond to an ‘endogenous pull’, as in East Asia, they can work in an indirect way to limit the finance available for domestic investment. In either case, the effect of regulations may be limited by the fact that integration into capital markets tends to generate strong pressure to adopt pro-cyclical macroeconomic policies (in particular, as discussed above, to allow domestic interest rates to fall to unreasonable levels, or to allow the exchange rate to appreciate excessively).
Viewed as a ‘liability policy’, capital-account regulations take on board the fact that markets usually reward sound external debt profiles (Rodrik and Velasco, 2000). This phenomenon reflects the fact that during times of uncertainty markets tend to respond to gross (and not merely to net) financing requirements; this means that the rollover of short-term liabilities is not financially neutral. Under these circumstances, a loan and bond maturity profile that leans towards longer-term external obligations will reduce the risk of a balance of payments crisis. At the same time, on the equity side, foreign direct investment (FDI) -- properly defined -- should also be preferred to other portfolio flows, as it has proved in practice to be least volatile. Equity flows have the additional advantage that they allow some of the risks associated with the business cycle to be shared with foreign investors, and FDI may bring parallel benefits (access to technology and external markets). These benefits should be balanced against the generally higher costs of equity financing.
From the point of view of domestic balance-sheets, the mix of lack of depth in developing countries’ financial markets and lack of external financing in their domestic currencies lead agents in the private and public sectors to hold variable mixes of maturity and currency mismatches in their portfolios. These mismatches tend to be interrelated. Indeed, due to the limited availability of domestic financing -- and long-term domestic financing in particular -- agents may be inclined to borrow abroad if external financing is easily available, but will then accumulate currency mismatches if their revenues are generated in domestic currencies (e.g., if they operate in the non-tradable sector). Capital account regulations can thus help avoid currency mismatches, particularly if they establish different rules on access to external capital markets by firms in the non-tradable sector (or, to the extent that this can serve as an imperfect proxy, by residents). Domestic prudential regulations that take into account the particular currency mismatch that characterise non-tradable sectors can serve as a partial substitute in this regard (see Ocampo, 2003; and Rojas-Suárez in this volume). To the extent that maturity and currency mismatches are interrelated, a more comprehensive first-best solution, obviously, has to aim also at deepening domestic financial markets.
Recent literature has also emphasized the crucial role that asset prices play in the business cycle. In developing countries, a surge in capital inflows often generates over-optimism (or ‘irrational exuberance’, to use the terminology coined by Robert Shiller and Alan Greenspan) and, as such, may well end up producing excessively high market valuation of stocks and real estate. Portfolio inflows may directly fuel stock price booms and may lead to speculative financing to profit from stock-market revaluation. In turn, rising real estate prices may generate excessive investment in residential construction as well as help fuel the boom in domestic financing, based on overpriced loan collaterals. These booms can produce increased financial fragility; and when there is (in the case of developing countries, a practically inevitable) sudden stop in external financing, this would lead to large price corrections; also, after a collapse of asset prices banks may find that their loans portfolio lacks adequate collaterals and this may even fuel further downward corrections in prices if they decide (or need) to ‘fire-sell’ the collaterals they had received in lieu of debts.
A look at the rationale behind the imposition of regulations of capital inflows in Chile, Colombia and Malaysia indicates that authorities were explicitly aiming to increase their room for macroeconomic policy manoeuvre and to change the composition of external finance towards less volatile flows. Averting asset bubbles seem to have figured less prominently in their explicit policy objectives. In the following sections, we will explore the extent to which there is evidence that they were successful in their task.
C. Effects on the magnitude of capital inflows
Figure 7.A shows the level and composition by source of net private capital inflows in Chile before, during and after capital controls. As is fairly evident from the graph, capital controls in Chile seem to have had a significant but rather short-term effect, at least in terms of levels. By 1994, the 1991 reduction seems to have evaporated, even in the face of the strengthened regulations adopted in May 1992, and the reduction brought about by the 1995 strengthening of controls seems to have lasted for only one year. Of course, we will never know what levels these inflows would have reached had it not been for these controls, but the evidence seems to indicate that private inflows did bounce back after having been affected briefly by the imposition of controls. In terms of volume, then, these controls seem to have had the effect of ‘speed bumps’ rather than permanent restrictions (Palma, 2000a). There were also changes in the composition of capital flows towards foreign direct investment, but this was a broader regional (and global) trend, which also occurred in countries that did not open their capital accounts in any significant way, and those that did open it but did not rely on preventive capital account regulations.
The ‘speed-bump’ character of inflow regulations in Chile is even clearer in Figure 7.B. In terms of volume, net equity securities and other investments, went from generating large negative flows in 1988 to large positive flows in 1990. The 1991 regulations did interrupt this process, but only by a year, and their strengthening in 1992 does not seem to have prevented a renewed capital account boom. The 1995 regulations generated a new lull that, as had happened in 1991, only lasted for a year.
Sources: Panel A, World Bank (2004); and Panel B, IMF (2004b). See respective sources for different definition of components of private capital flows.
The fact that the effects of regulations on capital inflows were only temporary may, of course, not be independent from the level of the implicit tax, which was much lower in Chile than that in Colombia. Also, as we have seen, loopholes in the regulations were a constant source of concern and led to a series of measures aimed at closing them.
In Colombia, frequent changes in the regulations make annual data an unreliable clue to their effect on capital flows. So Figure 8 concentrates on the most widely used series in Colombia to analyse capital flows: private capital flows involving cash transactions (thus eliminating those flows that are tied to trade transactions or involve investment in kind, both of which were not subject to the URR). As the Figure makes clear, the major turning points in the dynamics of private capital flows in Colombia during the 1990s followed the timing of policy decisions closely
PRIVATE CASH CAPITAL FLOWS, COLOMBIA
Indeed, the capital account liberalization, which came in two steps (February 1992 and September 1993),16 led to a boom in external financing. The strengthening of controls in March 1994 had only moderate effects, but the stronger decisions adopted in August 1994 generated, with a lag, a sharp reversal of the trend. The lag was associated to the fact that there were broadly-based expectations, based on policy announcements by the new Administration (that took power in August 1994) that controls would be strengthened; this led to a speculative wave of registrations of new loans in anticipation of such policy change. The less restrictive regulations introduced in early 1996 prompted, in turn, a rapid increase in capital inflows. By late 1996, speculative attacks threatened to break the floor of the currency exchange band (as reflected by the peak observed in the data for December 1996). The series of regulations imposed early in 1997 reversed that trend again, bringing capital inflows to more moderate levels; these levels were maintained until early 1999, when the Brazilian crisis triggered rapid capital outflows.
Figure 9 shows the very strong surge in private capital inflows experienced by Malaysia between 1988 and 1993. This was the background to the decision adopted by the Malaysian authorities to impose strict controls on capital inflows at the beginning of 1994. As we have seen, and unlike the Chilean and Colombian experiments with capital account regulation, the key characteristic of these controls was their quantitative (administrative) character. Also, domestic interest rates were reduced to reverse arbitrage flows.
Source: IMF (2004a; includes ‘errors and omissions;). See source for definition of components of private capital flows.
As mentioned above, these measures were so drastic that the effect was immediate. In fact, net private inflows fell in one year by 18 percentage points of GDP; the main component of this fall was short-term flows, which fell by an amount equivalent to more than 13 percentage points of GDP.17 In part due to the very success of these controls, and in part due to strong political pressure from the domestic and foreign financial systems, some of the controls began to be lifted as early as August of the same year; by January most had disappeared.
Although capital flows recovered with the lifting of controls, the recovery was relatively mild compared, for example, with the recovery of net private inflows in Chile after 1995. Moreover, the recovery took place only in ‘other’ inflows, leaving net portfolio inflows still in a negative net figure; in turn, ‘errors and omissions’ changed from a large positive net figure in 1993 to a large negative one in both 1995 and 1996. Thus, this quantitative short-sharp-shock had stronger short-term and longer-lasting effects than the continuing (and strengthening) Chilean or Colombian price-based controls.
One of the main peculiarities of the Malaysian case is the large size of the balance of payments ‘errors and omissions’ item. This phenomenon is relevant not only because it reveals pre-1994 deficiencies in Malaysia’s Central Bank accounting practices, but also because with controls in place they first disappeared and then became negative. The relevance of this is that one of the most repeated criticisms of controls is that they tend to be ineffective because capital will always find ways of bypassing them. In Malaysia it seems to have been the other way round: with controls came a successful tightening of procedures of recording inflows and a massive reduction, rather than an increase, in this item.18
In any case, not all elements of the macroeconomic package were dismantled at the end of 1994: low interest rates were maintained as part of residual policy package to disincentive a possible rapid return of private capital inflows after the end of quantitative restrictions. This certainly helped to maintain the volume of arbitrage inflows at a relatively low level, but may have helped to fuel the real estate bubble of 1996, which made the 1997 crisis much worse than it would have been otherwise (see below).
D. The broader macroeconomic effects
While there is little room to doubt the capacity of the 1994 Malaysian controls to affect capital flows, the price-based capital account regulations of Chile and Colombia have generated a great deal of debate. In particular, contrary to the broad agreement on the positive impact they had on external debt profiles (see below), their effectiveness as a macroeconomic policy tool has been subject to a great deal of controversy.19
Nonetheless, judging from the solid evidence that exists with respect to the sensitivity of capital flows to interest rate spreads in both Chile and Colombia, it can be asserted that URR do influence the volume of capital flows at given interest rate spreads.20 This may reflect the fact that available mechanisms for evading or eluding regulations are costly, and that national firms’ access to external funds is not independent from their maturities.21 In Colombia, where these regulations were modified more extensively over the 1990s, there is strong evidence that increases in URR reduced overall flows; this evidence is consistent with the more qualitative evidence of the links between the timing of major turnarounds in the capital flows and policy decisions regarding the capital account (Ocampo and Tovar, 1998 and 2003). In any case, as mentioned above, a significant part of the history of these regulations, particularly in Chile, was associated with the closing of regulatory loopholes.
However, the macroeconomic effect of price-based regulations cannot be judged only on the basis of their effect on the magnitude of capital flows. Indeed, one of the explicit policy objective of their introduction in Chile was allowing interest rates to remain at levels that were higher than parities; at the same time, at least according to some analysts, URR were meant to avoid further exchange rate appreciation. In turn, in Colombia, the strengthening of such regulations in 1994 was aimed at both opening up the space for a contractionary monetary policy (as the central bank judged that the credit and demand boom of 1992-1994 was generating undesirable effects), and reversing the exchange rate appreciation that the country had been experiencing.
In terms of interest rates, the available econometric evidence regarding the effects of capital account regulations is less controversial. It indicates that URR were able to increase domestic interest rates in both Chile and Colombia.22 The evidence of the effects of URR on exchange rates is more mixed, but this may well reflect more the difficulties inherent in exchange rate modelling (see Williamson, 2000, ch. 4), or the inability of URR to affect exchange rates at the levels they were actually applied (rather than any intrinsic ineffectiveness of URRs vis-à-vis exchange rates.)
The links between the different macroeconomic effects of regulations is also clear in the evolution of the regulatory instruments themselves. Thus, in Chile a basic problem of regulations was the variability of the rules pertaining to the exchange rate, since the lower limits of the exchange rate bands were changed several times during the period of controls (1991 to 1999). During capital account booms, this gave rise to a ‘safe bet’ for agents bringing in capital, since when the exchange rate neared the floor of the band (in pesos per dollar) it was highly likely that the floor would be adjusted downwards. In Colombia, the main problem was the frequency of the changes made in reserve requirements. Changes that were foreseen by the market (particularly those of August 1994) led to speculatory movements that reduced the effectiveness of such measures for some time.
Thus, in broader terms, the usefulness of capital controls as a macroeconomic policy tool must be evaluated first on their ability to give monetary authorities improved policy-choice -- particularly between reducing capital flows (and thus the money supply or the magnitude of sterilization) and increasing domestic interest rate spreads; and between higher domestic interest rates and exchange rate devaluation.23 The fact that policy choice and other macroeconomic factors play a role brings into focus the fact that URRs are a complement to, rather than a substitute for, other macroeconomic policies; this fact was explicitly recognized in both Latin American experiments with price-based controls. In Chile, these other macroeconomic policies were applied more effectively, particularly in relation to fiscal policy.
The macroeconomic effect of capital regulations is shown in Figure 10 in a simple diagram that relates capital inflows and the interest rate spread under imperfect capital mobility. Regulations shift the relationship between both variables downwards, allowing either lower levels of inflows (and sterilization) for a given level of interest rate spread (i.e., a movement from A to B), or higher interest rates/exchange rate devaluation for a given level of inflows (a movement from A to C). The final equilibrium (including, of course, an intermediate outcome between B and C) will depend on other policy variables and macroeconomic conditions.
Given the multiple channels through which the URR can affect the economy, the effectiveness of these regulations can be best measured by a broad index of ‘monetary pressures’ that includes the three possible channels through which capital inflows can affect the economy: the accumulation of international reserves, exchange rate appreciation and a reduction in interest rates. Figure 11 provides an index similar to others used in the relevant literature, in which the weights for the three indicators are their standard deviation during the period analysed.
Index of Expansionary Monetary Pressure
An inspection of the graph indicates that the 1994 Malaysian controls were extremely effective in reversing the strong expansionary effect of capital surges in previous years, particularly in 1993. The price-based capital-account regulations of Chile and Colombia had weaker effects, particularly in the first case. Indeed, the introduction of such regulations in Chile in June 1991 and their strengthening in May 1992 was not accompanied by a reversal of the expansionary trend (though the index remained stagnant for a few months);24 those instituted in July 1995 had a more discernible effect. In Colombia, which used price-based regulations more aggressively, the effects were stronger. In particular, the movement in the index of expansionary pressures is closely tied to the changes in capital-account regulations in 1993-1997. In both Chile and Colombia the capital account turned contractionary in 1998, with the reduction in the URR having only a negligible effect on this trend.
As Table 1 indicates, the mix in the evolution of the three components of the index varied significantly from one episode to another, even within the same country. Thus, taking Malaysia as a standard, controls had major effects on international reserves accumulation with an additional moderate impact on the exchange rate, whereas the policy decision to reduce interest rates supported the reduction of capital inflows. The effects of the June 1991 Chilean decisions are unclear, but those of May 1992 allowed a higher interest-rate-cum-devaluation mix but were incapable of affecting reserve accumulation. In turn, those of July 1995 did have a broadly desirable outcome: they stopped reserve accumulation as well as exchange rate appreciation while allowing authorities to maintain fairly high interest rates. In Colombia, the August 1994 regulations allowed the authorities to undertake a contractionary monetary policy while avoiding exchange rate appreciation. In the context of the continuation (or, more precisely, a gradual weakening) of such contractionary monetary policy, the decision to loosen regulations in February 1996 generated an avalanche of reserve accumulation and exchange rate appreciation, which in turn was stopped by the return to controls in January 1997 (or, to be more accurate, by a series of decisions between that month and May 1997).
Both Figure 11 and Table 1 indicate, however, that much of these macroeconomic effects were temporary in nature.
Change in Key Variables Preceding and Following
Major Capital Control Episodes
In Table 1, in particular, the macroeconomic effects of controls are hardly visible in the second year after regulations were imposed or strengthened. In Palma’s terminology used above (2000a), they operated more as ‘speed bumps’ rather than as more permanent (speed) restrictions. Of course, this may be associated with the fact that in several cases the regulations were designed to be temporary, or that their strong initial effect prompted authorities to take immediate steps in the opposite direction. This was particularly the case of the January 1994 Malaysian controls, and the August 1994 Colombian ones. In yet another episode (the strengthening of regulations in Colombia starting in January 1997), the effects were overwhelmed by new events (the Asian and Russian crises). Thus, the temporary character of the effects is more clearly perceived in the history of Chilean regulations.
However, more broadly, this is reflected in the fact that none of the three countries was able to avoid the cycle that developing countries experienced in the 1990s: rapid expansion up to the Asian crisis followed by a contraction when faced with the events that began with the 1997 Asian crisis. In this respect, there is a sharp contrast between the experience of these countries and those of China, India and Taiwan, which followed a strategy of keeping quantitative controls and only liberalizing them very gradually. This, together with the stronger effects of Malaysian vs. Chilean/Colombian regulations, are clear indications of the stronger macroeconomic effects of ‘quantitative’ vs. ‘price-based’ regulations.
E. Effects on the composition of capital inflows
Figure 12.A brings to fore a fact that is less controversial about the effects of inflow regulations in the three countries: they were effective in helping them to maintain a debt profile with a low share of short-term debt, a fact that proved to be very beneficial during the turbulent events of 1997-1998. Indeed, in some cases, some regulations led to changes in trends that proved permanent. This is particularly true of the decisions adopted by Chile and Colombia in the mid-1990s. The 1994 Malaysian regulations also interrupted a rising trend of short-term borrowing that was only partially reversed in the following years. As we saw above, in all three countries there is also clear evidence that regulations did not have any adverse effects on FDI.
Short Term External Debt