Capital flows to Brazil in the nineties: macroeconomic aspects and the effectiveness of capital controls.
The history of the resumption of capital flows to the Brazilian economy since the early nineties is intertwined with the success of the current stabilization plan, the Real Plan of July 1994. The very large volume of reserves worked and still works as the (short term) insurance policy of the exchange rate, which anchored the new currency. Nevertheless, the large volume of capital flows has prompted the government to try to fine-tune its size and composition.
Here we characterize those flows and analyze their main determinants. A simple econometric model finds that the interest rate differential has been, since 1991, the main determinant of the capital inflows. We then carefully review the empirical evidence on the interest rate differential between the Brazilian bond markets and abroad.
The restrictions on capital flows are reviewed and their effectiveness is analyzed. Finally, we sum up with a discussion of the macroeconomic causes and consequences of the capital flows, and the prospects for the future. The continued success of the current stabilization plan will in great portion depend on what will happen to the capital flows.
I. EMPIRICAL DESCRIPTION OF CAPITAL FLOWS TO BRAZIL IN THE NINETIES
This section describes the capital movements to and from Brazil in the nineties. In order to provide a short historical perspective, the late eighties were also included in the data. This provides a very sharp contrast between those years - when the foreign debt problem was a huge constraint on the Brazilian economy - and the nineties, when the resumption of capital flows, together with the foreign debt renegotiation significantly relieved the external constraint, vis-a-vis the earlier period.
Brazil, like many developing countries, witnessed a revival of capital inflows in the nineties. Chart I shows the capital account balance since 1987 (both in USS millions and as a percent of GDP), displaying the remarkable reversal since 1992. Chart 2 shows the aggregate net transfers to Brazil, defined as the sum of the capital account balance, the interest payments (negative, because Brazil is a net debtor), and the change in short-term liabilities flows (basically arrears). One can detect the same patterns as in the capital account (reversal since 1992), but with a much more marked increase in 1995. This is because in 1992, US$16.6 billion of the inflows took the form of refinancing,(1) which were used to pay back the arrears (US$14,253 million of arrears were repaid in 1992). In 1995, only US$0.5 billion of arrears were repaid, and there was a net inflow of US$19,667 million in short-term capital. One can detect the new pattern of the aggregate net transfers in 1995 and 1996, when around 3 percent of GDP was transferred to Brazil.
Therefore, the global picture is that Brazil has overcome the shortage of capital inflows since 1992. Moreover, since 1995 those flows, mainly short term, have become excessive, creating many problems for monetary and exchange rate policy, as analyzed below.
Chart 3 displays the behavior of medium and long-term capital movements.(2) The net figures are presented for the three main components: investments, financing, and currency loans (explained below). Those net figures, together with the net total of medium and long-term capital movements, are presented in bars referring to the first axis scale. The gross movements are presented in lines, referring to the second axis scale.
The reversal of the total net figure in 1992 is due both to foreign investments and currency loans, while the burden of the foreign debt shows up in a steadily negative figure for the item financing. Both the inflows and outflows grew substantially during the nineties, with the latter almost reaching the US$65 billion figure in 1996.
Chart 4 displays the behavior of net foreign investment (direct and portfolio) and reinvestment in Brazil. For the net figures, once again, 1992 is a turning point. The decrease in the 1995 figure may be attributed to the effects of the Mexican crisis. In the first quarter of 1995, US$3,352 million of foreign investment (mainly portfolio investment) flowed out of the country (the year-end figure was positive, US$4,670 million). In 1996, the growth trend that started in 1992 resumed.
The most striking feature of Chart 4, however, is the enormous growth in both inflows and outflows. As Charts 5 and 6 make clear, the main source of such growth was the portfolio investment which was multiplied by a factor greater than 30 between 1991 and 1994/5 (see Chart 5). Portfolio investment will be further analyzed later.
Foreign direct investment (see Chart 6) also displayed an upward trend since 1994, with a marked increase in 1996. This is explained in part by a change in the tax law that regulated profit remittance abroad. In 1993, the law regarding profit remittance was changed, so that profits paid regular domestic corporate income tax (48%), and dividends remitted abroad were taxed at 15 percent.(3) The resulting joint tax burden (56%) was high by international standards. Capital gains remitted abroad paid a 25 percent tax. Starting 1996, only corporate income tax (30.6%) is charged, which harmonized the Brazilian tax code with its counterparts in the Mercosul. Privatization revenues also account for the increase in foreign direct investment last year.
The continuation of the trend in foreign direct investment will depend on the strength of the privatization process and the public-private partnerships in areas previously restricted to state-owned firms (oil, energy generation, telecommunications, etc.). Forecasts hover around US$16 billion for 1997 and the following years. However, there have been recent claims in the financial press that fixed income investments have been disguised as direct investment, in order to avoid the "entrance" tax (7%, explained below) charged by the Brazilian government. We will further expand on this point below.
B2. Financing and Currency Loans
Chart 7 shows the performance of the Financing item, composed of the multilateral organizations (BIRD, IDB, and IFC), bilateral organizations and suppliers' and buyers' medium and long-term credits.(4)
The better performance of the capital account in the nineties allowed for an increase in the differential between amortizations and disbursements to Brazil, making the item Financing steadily negative in recent years. Nevertheless, in steady state, one would presume that a developing country like Brazil should be a net receiver of official transfers. Therefore, when that happens, this will represent an alternative (better) source of external savings to the short-term capital flows that have entered the Brazilian economy in the first half of the nineties, with positive impact on the stabilization plan.
Currency loans, on the other hand, substantially increased as a result of the resumption of capital flows in the nineties, as shown in Chart 8. The decomposition of those disbursements reveals that the main instrument for this increase was the issuance of notes in the international markets, with large volumes starting in 1992. This stands in sharp contrast with the previous decades, when bank loans accounted for most of the disbursements.
Chart 9 summarizes the main facts regarding the composition of capital movements to and from Brazil since the late eighties. The solid line is total medium and long-term capital movements (see footnote 2 for the difference between this figure and the capital account balance), with the by now familiar pattern of becoming very positive and increasing after 1991. This line is decomposed of its main factors (in columns). On the positive side, the main factors responding for the growth of medium and long-term capital movements in recent years are Foreign Portfolio Investments, Currency Loans - Notes, and Foreign Direct Investments. On the negative side, the main factors are Financing (repayments to official agencies), Currency Loans - Banks (repayments to banks), and Brazilian Investments abroad. The dotted line is the short-term capital movements, which also turned positive in 1992, but reached the very high level of almost US$20 billion in 1995, thereby surpassing the medium and long-term capital movements in that year. Note that the Brazilian Central Bank bulletin classifies portfolio investments as medium and long-term capital movements, not short term ones. If we classify portfolio flows as short-term capital movements, the predominance of short-term capital inflows in 1995 becomes much more striking as seen in Table 1. In 1996, however, short-term capital flows (exclusive of portfolio investment) decreased substantially, while the medium and long-term flows increased to the point of increasing the capital account (see Chart 1).
[TABULAR DATA FOR TABLE 1 OMITTED]
III. MAIN DETERMINANTS
Besides the good prospects for the economy in the medium and long run, which have attracted both portfolio and direct investment, one of the main determinants of capital inflows in the nineties has been the extremely high interest differential between Brazil and the developed economies.(5) Therefore, we start by considering the interest rate differentials between Brazil and the U.S.
A. Interest Rate
On the last day of September 1991, the very low level of foreign reserves prompted the Brazilian Central Bank to adopt a combined strategy of devaluing the currency by 15% and raise interest rates to a steady level above international rates. After the devaluation, the exchange rate policy of daily devaluations (crawling-peg) assured investors that the government aimed at keeping a stable real exchange rate.(6) Given the large interest rate differential between the Brazilian and international rates, short-term capital started pouring into the country to gain arbitrage profits.(7)
Among the several possible measures of interest rate differential (nominal, real, etc.), we follow Frankel (1991) in adopting the covered interest differential (CID) as our measure of attractiveness of domestic bond markets to foreign investors. The CID is defined as the residual once both the forward discount and the international interest rate are deducted from the domestic interest rate. That is, the CID represents the extra gain above and beyond the international interest rate a foreign investor would have by investing in the Brazilian bond market without incurring an exchange rate risk (the investor is already covered in the futures exchange market), had conditions of free capital movements prevailed (no taxes, quotas, and other restrictions).
Theoretically, in a fully integrated international capital market, arbitrage would drive this differential to zero. This is what makes Frankel (1991) suggest the CID as the most adequate measure of capital markets' integration. He shows that positive measures of CID are related to restrictions to capital inflows, while negative measures of CID are related to restrictions to capital outflow.
A1. The Covered Interest Differential
As always, when one wants to implement a theoretical concept like the CID, there are several empirical choices that have to be made. We start by considering monthly data for the period 1991 to 1995 (see Chart 10). We use the monthly overnight interest rate as Brazilian domestic interest rate (i). To compute the forward discount, we resort to the U.S. dollar futures market in Brazil, since no liquid forward market exists. For each month, we use the expected devaluation signaled by the futures market at the first day of that month (f).(8) The foreign interest rate is the U.S. Treasury bill ([i.sup.*]). Therefore, the CID (in percent per year) is computed through Equation 1 below.
CID = [1 + i/(1+f) x (1 +[i.sup.*]))] x 100 (1)
In Chart 10, the left-hand sale refers to both the domestic interest rate and the expected devaluation. Note that the unit is percent per month. The right-hand scale refers to the U.S. T-bills rate and the CID, both in percent per year.
The message from Chart 10 is very clear; starting at the end of 1991, interest rates were raised to very high levels. This, together with the policy of keeping the real exchange rate stable (until the Real Plan), gave foreign investors a very high yield, especially when the interest rates abroad were unattractive in the early nineties and, less so, nowadays. The capital inflows caused by this interest rate wedge not only solved the scarcity of foreign reserves problem, but also posed the opposite problem, that of an overabundance of foreign reserves.
After the Real Plan, in July 1994, the exchange rate was allowed to float for a short while. Given the capital inflows, the Real appreciated considerably, both in real and in nominal terms, giving extra gains to foreign investors with a stake in the Brazilian markets. Note, however, the only the expected part of this gain shows up in Chart 10. Since the first three months of the Real Plan - when a nominal (and real) appreciation occurred - a positive forward discount prevailed in the futures market, extra gains on top of those shown in Chart 10 existed for arbitrageurs. Except for December 1994, the CID has signaled very attractive arbitrage opportunities to foreign investors also during the Real Plan.
In the context of an equilibrium model, the CID is a measure of the country risk premium, "because it captures all barriers to integration of financial markets across national boundaries: transaction costs, information costs, capital controls, tax laws that discriminate by country of residence, default risk, and risk of future capital controls" (Frankel, 1991). Since one has more than one way to invest in fixed income in Brazil without incurring an exchange rate risk, several different measures of the CID exist. We now consider higher frequency (daily) data on the CID since August 1994. We offer three different ways of computing the CID, and compare the systematic differences between those.
A2. Computed with US$ Futures
For the domestic interest rate data, we use the daily data forecast by the futures market for the compound interbank rate (CID) for the next three months. For the nominal expected devaluation, the exchange futures market data are used. The foreign interest rate used is the LIBOR (in U.S. dollars). In Chart 11, this is the jagged solid line that is above the others most of the time.
A3. Computed with Brazilian Bonds Issued in US$
Another possible way to compute the CID is to deduct the foreign interest rate (LIBOR, in this case) from the secondary market yield on internationally issued Brazilian bonds' rates. For this purpose, we use the most liquid Brazilian bond, the IDU.(9) The CID computed with the IDU is the smooth dotted line that is below the others for most of the time.
A4. Computed with Brazilian Bonds Indexed to US$
A third possible way to compute the CID is to deduct the foreign interest rate from the domestically issued Brazilian bonds with an index clause to the U.S. dollar (NTN-D). These securities are indexed to the U.S. dollar, but redeemed in domestic currency (RS), unlike the Brady bonds. We were only able to find the data for the rates paid on these securities at the auctions, which are usually held every fortnight. Maturities range from three to 24 months, the latter being fully placed in the Central Bank portfolio. We omitted the rates for the securities placed directly with the Central Bank, thereby considering only the placements with the public. These are the (black) columns in Chart 11.
A5. Comparison of the Three Different Proxies of the Country Risk for Brazil
Chart 11 shows the three different measures of the CID: the first uses the exchange futures market to cover for the exchange rate risk implicit in Brazilian domestic bonds and deducts the LIBOR (US$); the second takes the spread between the rate of internationally traded Brazilian foreign debt (IDU) and the LIBOR (US$); and the last takes the spread between the rate of exchange-rate-linked domestic debt (NTN-D) and the LIBOR (US$).
It is noteworthy that those three ways of measuring the country risk for Brazil differ systematically. Except for the period of the exchange rate crisis of March 1995, the measure of CID constructed with the domestic futures market lies above all the others. The differentials between the different measures of CID, however, have decreased substantially, and almost disappeared between the NTN-D and futures CIDs. However, the differential between the IDU and the domestically-traded bonds is still large.
Possible explanations involve default risk, taxes, exchange rate spread risk and the effect of the term structure. The NTN-Ds are usually longer than the longest futures contract available. Therefore, the lower CID when NTN-Ds are used may be forecasting a decrease in the interest rate differential in the future, as has indeed happened over a large portion of the period studied. If this interpretation is correct, the current equalization of the two measures of the CID (see Chart 11) may be signaling that the market no longer forecasts (or supports) further decreases in the differential between the two CIDs in the future. That is, with the current restrictions on capital inflows, interest rates may be reaching a lower threshold regarding capital inflows.
The differential between the IDU CID and the two other measures may be due to three factors. The first factor is the restrictions analyzed in this paper, among them the "entrance" tax on fixed-income investments. Also, domestic fixed-income investments pay a 15% withholding income tax.
The second factor is that the latter two measures involve securities that are redeemed in domestic currency, although the payments are indexed to the U.S. dollar. However, not all investments can use the "free segment" exchange rate which is the exchange rate used as the index. The other official exchange rate market, the "floating" market, has in recent years been kept by the Central Bank very close to the "free segment" one, but this may change in the event of an exchange rate crisis. Therefore, this "exchange-rate-spread" risk may be responsible for part of the differential between the measures of CID. Again, if this hypothesis is true, one may learn about the investors' expectations on the sustainability of the exchange rate by the behavior of this spread over time.
The third factor, the default risk, may signal that investors see the Brazilian government more committed to honoring the securities traded abroad (e.g., the IDU) than those traded at home (e.g., the NTN-D).(10)
The behavior of the differential between the IDU CID and the other two measures of the CID after the "tequilazzo" is very interesting. The IDU rate increases after the Mexican crisis, reflecting the natural increase of the premium required by investors to hold Brazilian bonds. The CID computed with the domestic rates, however, fall, denoting that during the aftermath of the crisis Brazilian domestic interest rates did not rise enough to keep the previous level of the CID. Accordingly, capital started to exit the country, as shown in Chart 12. Since neither the nominal nor the real interest differential changed much in that period, this evidence confirms that the CID is the best proxy for the attractiveness of domestic fixed income investments to foreign investors.
B. Other Determinants: An Econometric Exercise
Undoubtedly, many other factors besides the interest rate differential have attracted capital to the Brazilian economy. The renegotiation of the external debt, and the success of the Real Plan in keeping a low level of inflation for two years have already attracted many foreign investors with intentions of making more permanent investments in Brazil. This occurred despite the very low ratings given to Brazil by the main international credit rating agencies, Moody's and Standard & Poors (Silveira, 1996).
In order to access the capital flows determinants we followed the suggestion of Laban and Larrain (1994),(11) and used the following variables as regressors: the covered interest differential (CID), the average investment/GDP ratio in the previous four quarters (IGDPR4Q), the rate of GDP growth (GDPGR), and the current account deficit/GDP ratio (CADGDPR). For the Chilean case, Laban and Larrain (1994) use the foreign debt/GDP ratio to proxy for country risk. Surprisingly and unfortunately, there are no quarterly series for the Brazilian external debt, since the Central Bank stopped computing them on a regular basis in the nineties.
Appendix 1 presents the main results of the econometric exercise. We used quarterly data from 1985:1 to 1995:4. All regressions use OLS with the standard errors corrected by White's (1986) heteroskedasticity-autocorrelation consistent covariance matrix estimator. Table 2 contains the statistics of 40 regressions: eight combinations of the regressors for each of the five dependent variables: TOTLRUN (total of long and medium-run net flows, including officials loans and portfolio investment), SRCF (short-run capital flows, excluding portfolio investment), TOTNOF (total of long, medium, and short-run net flows, excluding only official loans), TOTSR (total of short-run capital flows including net portfolio investments), NOFLR (total of long and medium-run net flows, excluding official loans and portfolio investment).
Since 1991 represents a major mark in terms of the Brazilian integration in the world financial markets, with the liberalization of portfolio flows through the creation of the Annex IV (see Capital Controls section below for explanation) and several other measures that liberalized the capital account, we created a dummy for it (DANNEXIV). It is 0 for all quarters before 1991:1 and 1 afterwards. Since the liberalization may also have affected the explanatory variables on the capital flows (the slopes), we also introduced in the regressions the interaction [TABULAR DATA FOR TABLE 2 OMITTED] (the product) of this dummy variable (DANNEXIV) with each explanatory variable. Those have the same name as the original variable with a D at the beginning.
In Appendix 1 each regression occupies two lines, and is identified by a number from 1 to 40. They differ in terms of the dependent variable, whose acronym is displayed immediately after the number, and the combination of regressors used. When a coefficient and respective t-statistic do not appear, that means that it was excluded from that regression. Besides the coefficients' estimates and respective t-statistics, we report the R squared, the R bar squared, the Durbin-Watson statistic for first-order autocorrelation, and the Ljung-Box statistic's level of significance (for higher order autocorrelation).
The Chow test for structural change in 1991 of both slopes and intercept is performed through a test of exclusion of all the dummy variables included (see Johnston, 1984, pp. 225-233). The level of significance of the test (p-value), whole null hypothesis is that all dummies' coefficients are zero, is reported below the Durbin-Watson statistic. One can easily see that for all the cases, the null hypothesis of joint insignificance of all dummies is easily rejected at the 1 percent significance level. This means that, as expected, there has been a structural change in the behavior of capital flows after 1991.
Regressions 1 to 8, which report the results for the dependent variable TOTLRUN (total of long and medium-run net flows, including official loans and portfolio investment) show that only the covered interest differential is significant after 1991 in a few regressions (DCID). No other variable shows up significantly.
Regressions 9 to 16, which report the results for the dependent variable SRCF (short-run capital flows, excluding portfolio investment), reveal that besides covered interest differential after 1991 (DCID), also the investment/GDP ratio is significant, both for the whole period (AGDPR4Q) and for the change in the slope in 1991 (DIGDPR4Q). The dummy for the constant (DANNEXIV) is also up significantly in a few regressions.
Regressions 17 to 24, which report the results for the dependent variable TOTNOF (total of long and medium-run net flows, excluding officials loans but including portfolio investment), also show that besides covered interest differential after 1991 (DCID), the investment/GDP ratio is significant, both for the whole period (IGDPR4Q) and for the change in the slope in 1991 (DIGDPRO4Q). The constant and its dummy (DANNEXIV) are also significant in a few regressions. For that variable, the highest Rs squared and Rs bar squared are achieved.
Regressions 25 to 32, which report the results for the dependent variable TOTSR (total of short-run capital flows, including net portfolio investments), reveal that besides covered interest differential after 1991 (DCID), the investment/GDP ratio (IGDPR4Q) is also significant, but the hypothesis of no slope change in 1991 for the investment variable is not rejected.
Regressions 33 to 40, which report the results for the dependent variable NOFLR (total of long and medium-run net flows, excluding official loans and portfolio investment), show that the covered interest differential after 1991 (DCID) is significant in all eight regressions. The investment/GDP ratio after 1991 (DIGDPR4Q) is significant in all regressions in which it is included. The estimate of the GDP growth variable after 1991 (DGDPGR) is not significantly different from zero. The covered interest differential for the entire period shows up significantly, although with the wrong (negative) sign.(12)
Table 1 summarizes the main econometric results reported above. It presents only the regression with all independent variables for each measure of capital flows. In summary, the econometric results confirm that after 1991, when capital flows resumed, the main determinant was the interest rate differential. From the other variables tried in the regressions, only the investment/GDP ratio appears to have a positive and significant effect.
The fact that only after 1991 did the interest rate differential show up significantly in the regressions (through the variable DCID) is compatible with the explanation that the change of regime - represented by progress of the external debt renegotiation process and the opening up to foreign investment exemplified by, among other measures, the liberalization of the exchange rate transactions and the creation of the Annex IV was a necessary condition for the interest rate differential to attract foreign capital. Note also that the CID has components of both "pull" (the domestic interest rate) and "push" (the international interest rate, with a negative sign) factors.
III. CAPITAL CONTROLS
The massive capital inflows that started in the last quarter of 1991 eventually posed a macroeconomic problem to the government. To mitigate the short-term inflows towards investments in fixed income, mainly Brazilian government bonds, the government started to impose controls and restrictions on capital inflows in the second semester of 1993.
We begin by describing the succession of measures aimed at dealing with the excessive inflow of short-term foreign capital, and then evaluate their effectiveness in achieving the macroeconomic goals. One interesting aspect worth noting at the outset is that many of the measures described below simply allow for transactions that were previously restricted. After all, most of the legislation regarding the foreign sector was aimed at restricting capital outflows, while the problem of the 1990s was the exact opposite. However, instead of simply eliminating previous capital outflow restrictions, many of the measures described below just relax the previous limits, but keep them in place, probably as an insurance against a return to the previous situation of recurrent exchange rate crises.
A. Description of Legislation
In 1992 the main changes in regulation were still aimed at further opening the capital account, a process that had started during the late eighties. The additional income tax on profit and dividend remittance abroad was abolished. Foreign investors were allowed to invest in derivative markets. Firms were allowed to issue securities convertible into stocks abroad. The minimum length of stay of foreign capital invested through privatization auctions was reduced from 12 to six years. Regarding reinvestments in Brazil, foreign investors were no longer required to wait for two years before being able to sell assets purchased through the privatization process (Banco Central do Brasil, 1993).
In 1993, notwithstanding the implementation of several liberalizing measures on exchange markets, the government started a gradual process of "throwing sand on the wheels" of short-term capital inflows directed to fixed-income securities, in order to prevent further increases of the domestic government debt. According to the Brazilian Central Bank Annual Report (1994):
The impossibility of a more drastic reduction of the rate differential between domestic and foreign interest rates which would naturally discourage the inflow of foreign financial savings, resulted in measures that would make it possible to attenuate the monetary impact of the foreign sector, without interrupting the process of integration with international financial markets.(13)
In June 1993, the Central Bank expanded the minimum average amortization term of financial loans from 30 to 36 months. Furthermore, for purposes of the fiscal benefits related to the income tax on remittances of interest and other charges, the periods of these operations were increased from 60 to 96 months. The Central Bank also tried to induce delays in the inflow of export revenues by increasing the period for exchange contracting from 45 to 180 days after the actual shipment. In the case of export credit (advances on exchange contracts - ACCs), the maximum period between the inflow of resources and the shipment of the merchandise was decreased to 180 days (from 360). Regarding imports, it also allowed the anticipation of the exchange contracting in relation to the maturity of the liability abroad up to 180 days (before it used to be 45 days), unsuccessfully trying to make importers pay their dues in advance.
Banking regulation was also changed to prevent dollar denominated liabilities and allow for larger amounts of dollar denominated assets; the selling positions (dollar liabilities) defined on the basis of each bank's net worth were reduced by 50%, while buying positions (dollar assets) were increased from US$2 to US$10 million (excesses must be deposited at the Central Bank).
Since 1987, portfolio investment was fostered by the creation of specific channels that gave foreign investors exemption from domestic income tax on capital gains. The most widely used channel to invest in Brazilian stock and derivative markets is the so-called Annex IV - Securities Portfolios for Institutional Investors - created in May 1991. Only foreign institutional investors may invest in those portfolios. Examples of institutional investors that qualify for the use of the Annex IV are Pension Funds, Portfolios belonging to Financial Institutions, Insurance Companies, and Foreign Investment Funds. Despite the regulation, wealthy investors are known to have "individual funds" under Annex IV.
Given the tax exemptions of the investments under Annex IV, financial engineering was widely used to channel those funds to the high-interest-paying government debt. In August 1993, the National Monetary Council (CMN) forbade funds channeled through Annexes I to IV from being invested in fixed-yield bonds, including exchange NTN (a dollar linked Treasury bond) and commodity investment funds (which actually worked as fixed-income-like funds). Table 2 shows the effect of this regulation. In August 1993 there were US$1,720 million of Annex IV funds invested in fixed-income-like securities and funds when those investments were forbidden (the item "others" in Table 2). By September 1993 this figure had already dropped to zero. However, the alternative found by the market to circumvent the regulation and keep investing in fixed income was to invest in debentures. The figures for this investment item rose from US$275 million in August to US$1,284 in September, US$2,183 in October, and US$3,011 in November. Table 2 clearly shows how the item "debentures" replaced the item "others" (mainly the fixed-income-like securities and funds). In November, the CMN moved to close this loophole by also forbidding investments in debentures (only those already purchased - sometimes with maturities longer than five years - could be kept until maturity). It also created a specific channel for fixed income investments, the Foreign Capital Fixed Yield Funds, which levied a 5 percent "entrance" tax (IOF) on the initial exchange rate transaction. Financial loans in currency also started paying a 3 percent "entrance" tax (IOF).
The new round in the game between regulators and investment banks involved investments through derivative markets, which are fairly well developed in Brazil.(14) By December a new Central Bank measure was enacted forbidding a broader range of fixed-income-like securities, including investment strategies involving derivatives that led to predetermined returns, e.g., a box.(15)
The evaluation made then by the Brazilian Central Bank of those measures was that they
placed obstacles in the path of foreign capital entering the country with the exclusive purpose of seeking the earnings made possible by interest rate levels. At the same time, the structure that favors the inflow of resources to the stock market was preserved (Banco Central do Brasil, 1994).
In January 1994, a new restriction was enacted, now targeting one government security (NTN-National Treasury Notes) which could be purchased by Annex IV funds under a broad interpretation of a Decree that listed the NTN as a privatization currency. Note in Table 1 how the volume of privatization currencies increases five-fold after August 1993. In March, the "entrance" tax levied on the Foreign Capital Fixed Yield Funds was extended to all portfolio investments, although the tax rate was initially set to zero percent for Annex IV funds. This was meant as a "clear signal as to the possibility of taxing these operations" (Banco Central do Brasil, 1995). The mechanism of automatic prior authorization of foreign loans was suspended, and renewal or extension of previous loans were also subject to the minimum terms of 36 or 96 months, which prevailed for new loans.
One the eve of the Real stabilization plan - June 30, 1994 - several additional restrictive measures were taken:
a). Prohibition of transformation of advances on exchange contracts (ACC) into anticipated (sic) export payments (short-term), when this results in the postponement of the regulatory period for shipment of the merchandise;
b). Increase of the minimum period of amortization of anticipated export payment operations registered at the Central Bank from 360 to 720 days;
c). 90 day suspension of inflows of foreign resources to be used in future capital increases and bridge investments for later conversion of debt into investments and, following that, indefinite suspension of such operations; and
d). 90 day suspension of the taking of foreign loans by public sector entities (Banco Central do Brasil, 1995).
With the Real Plan of July 1994, the Central Bank let the exchange rate float. Given the prevailing conditions at the time, the emphasis on very high interest rates led to the nominal (and real) appreciation of the real. Since capital flows subsisted, the Central Bank tried new measures, such as the increase from US$10 to US$50 million in the banks' buying position in the free-rate market (before they had to deposit any excess at the Central Bank).
In August 1994, a new round of measures at liberalizing exchange outflows were undertaken:
a). The possibility of contracting exchange for future liquidation in operations of a financial nature, an alternative previously permitted only in commercial operations;
b). Dispensation from the import license for the contracting of import exchange operations;
c). Permission for anticipated liquidation of foreign liabilities related to financial loans and financing registered at the Central Bank up to August 31, 1994, independently of the resources having completed the minimum period of permanence in the country; and
d). Free negotiation among the parties of the percentage of the value of imports to be financed in operations with terms of more than 360 days, thus permitting a larger volume of on sight payments that were previously limited to a maximum of 20% (Banco Central do Brasil, 1995).
In the attempt to stimulate demand for foreign currency, the value of transfers that the banks were permitted to carry out without Central Bank authorization for purposes of investment abroad by private nonfinancial legal entities was raised from US$1 million to US$5 million. Legal entities were also allowed to purchase real estate abroad, something previously restricted to individuals.
In September 1994, special investment funds abroad were allowed for Brazilian investors. Those funds must carry at least 60 percent of (internationally issued) Brazilian government securities. Also in September, the continuous appreciation of the real led the Central Bank to intervene in the exchange markets in late September, ending the short period of free flotation of the exchange rate.
By late October 1994, further restraints on capital inflows were enacted:
a). Reduction in the maximum period for the contracting of exchange prior to shipment and, consequently, of ACC operations, which dropped from 180 to 150 days in the case of exporters with a total value of contracted operations equal to or less than US$10 millions in the last 12 months (small scale); for medium and large scale exporters, the maximum period was reduced from 180 to 90 days; a maximum period of 30 days was set for products considered essential to internal supply;
b). The earmarking of exchange contracting operations to registration of exports, without permitting alteration of the merchandise to be exported. The purpose of this measure was to make it difficult to practice negotiation of export performance (Banco Central do Brasil, 1995).
Several other restrictions tried to prevent the increase in outstanding credit to Brazilian exports, well-known channel to avoid capital controls on inflows, including a 15% reserve requirement on ACCs to be deposited at the Central Bank. A 30% reserve requirement was imposed on contracts involving the taking over of the importer's obligations. The aim was to discourage importers from resorting to this financing mechanism offered by banks through withdrawals against credit lines abroad. In November, the rate was increased to 60%.
To further discourage capital inflows, the "entrance" tax was raised on most portfolio investments and loans: a) from 3% to 7%, in the case of loans; b) from 5% to 9%, in the case of investments in Foreign Capital Fixed Yield Funds; and c) from zero 1 percent, in the case of Annex IV investments. The minimum period for domestic loans under Resolution #63 (that regulates banks' foreign liabilities aimed at making domestic loans to firms) was raised from 90 to 540 days, and stricter requirements were put in place. Annex IV funds could no longer invest in money market funds (FAFs) or fixed-income privatization currencies. Pension funds were allowed to place up to 10 percent of their reserves in investment funds abroad. Privatization Funds-Foreign Capital were forbidden to invest in domestic debt. A massive liberalization of exchange transactions was undertaken. This was the status when the Mexican crisis hit in December 1994. The (rather mild) effects of the Mexican crisis required the government to undo a few of the previous measures aimed at increasing the demand for foreign currency. The term of 180 days for the closing of exchange prior to shipment (ACC) was reestablished, while the reserve requirement on that trade-finance instrument was suspended.
The worsening of the trade balance and a clumsy intervention of the Central Bank in the exchange markets by early March prompted a near exchange rate crisis, which required a sharp increase in domestic interest rates, along with several other measures that undid the previous restrictions, namely: a) reduction of the "entrance" tax rate from 7% to zero percent on foreign loans; from 9% to 5% on investments in Foreign Capital Fixed Yield Funds; and from 1 percent to zero percent on Annex IV investments; b) reduction of the minimum average term from 36 to 24 months for new financial loans and from 36 months to six months in the case of renewals or extensions of previous loans; and c) reduction of domestic relending under Resolution #63 minimum period from 540 to 90 days. The banks' buying position limit before they had to deposit any excess at the Central Bank was reduced from US$50 to US$5 million.
At the end of March 1995, the Brazilian real had undergone a 5.2% nominal devaluation, and a new exchange ban regime had been implemented, with frequent Central Bank interventions. This new regime aimed at "permitting a gradual devaluation of the 'real' against the dollar, without however providing the market any signals as to the speed or intensity of these devaluations" (Banco Central do Brasil, 1996).
As markets were convinced that the exchange rate regime was credible, massive capital inflows resumed as of July 1995. A new round of restrictions on capital inflows then took place in August 1995: a) the foreign loans "entrance" tax was raised from 0 percent to 5 percent; the Foreign Capital Fixed Yield Funds, from 5% to 7%; b) a 7 percent tax (IOF) on short-term financial transactions between institutions in the country and abroad in the floating rate segment (which was being used to circumvent the restrictions) was introduced; c) derivatives markets in Brazil were forbidden to foreign investors. Moral suasion was also a widely used method used by the Central Bank with the aim of controlling the inflows.
In September 1995, the "entrance" tax (IOF) on currency loans was changed to provide an incentive to longer loans. A decreasing scale of taxes was adopted, inversely related to the loan maturity: 5% (two years or less), 4% (three years), 2% (four years), 1% (five years), zero percent (6 years or more).
In February 1996, another "package" of measures aimed at further restricting short-term capital inflows was enacted. For investments under Annexes I to IV, it forbade investments on TDA, OFND and Siderbras debentures (securities that provided fixed income results not previously excluded). The minimum average term for currency financial loans was put back to 36 months (new, renewals, or extensions). The funds under Resolution #63, while waiting in a domestic bank to be lent, cannot be invested in NTN-D (exchange-rate-linked domestic debt). A 5% "entrance" tax (IOF) was imposed on investments in Privatization Funds. Foreign investors (individuals or legal entities) were allowed to invest in Real Estate Funds and Emerging Firms Investment Mutual Funds, with a tax (10% or 5% for regular registered funds) on all withdrawals for periods shorter than one year.(16)
Appendix 2 presents a summary of the above legislation, as well as the "financial innovations" engineered by the financial market to bypass the controls. The next section evaluates the effectiveness of those controls.
B. Evaluation of the Effectiveness of Capital Controls
As shown in the previous section, in the three-year period that started in June 1993, there were 15 changes in regulations regarding capital flows. That is, almost one intervention every two months. Nevertheless, capital kept steadily flowing to Brazil, except for a short period around the Mexican crisis. Chart 12 displays the continuous accumulation of foreign reserves at the Central Bank, which reached US$60 billion by June 1996, almost 10% of GDP or 1.2 years of imports.
Effectiveness may have several meanings.(17) Here we are concerned with a counterfactual experiment: what would have been the capital inflows had Brazil imposed no restrictions on capital inflows and had all other factors remained constant? An econometric framework could be constructed to offer an answer to this question. However, given that the flows resumed in 1991 and the restrictions started in 1993, there is not much hope that a very powerful test can be constructed. We tried an index of capital controls. It started with zero, and, for each new round of measures (see Annex 2), it increased or decreased by one, depending on whether the measures further restricted or relaxed the capital controls. This index was added to the regressions described before, but its coefficient estimate was not statistically significant.
We claim, however, that the annual data presented in Charts 12 and 13, which display the performance of financial exchange transactions, represent a convincing argument that ways are being found by the financial markets to circumvent the restrictions. In 1996, there was a massive increase of direct investment (US$5.8 billion for the first three quarters of 1996). The financial press attributed a great portion of this increase to fixed-income investments, disguised as direct investments to avoid the restriction on capital inflows.
Therefore, it seems that neither the restrictions imposed on capital inflows nor the restrictions relaxed on capital outflows have been powerful enough to prevent massive net capital inflows. Laban and Larrain (1993) develop a model to show that the liberalization of capital outflows "may not be the appropriate policy to defend the real exchange rate in the presence of massive capital inflows because it is likely to strengthen those very capital inflows." The Brazilian case seems to corroborate their theoretical result.
Nevertheless, the government claim seems to be that the combination of very high interest rates with those restrictions actually bought the plan enough time to sustain the "exchange rate anchor" until fiscal and other reforms are implemented. This brings us to a new meaning of effectiveness, i.e., the possibility of achieving a "good" macroeconomic equilibrium that otherwise would not have prevailed in a multiple equilibrium model. Despite our previous argument against the success of the restrictions in preventing capital flows, this might indeed be a possibility, although one would like to have a formal political economy model and convincing data that the needed fiscal reforms are being implemented. For now, we waive our hands on the formal model, but present in the next section arguments that cast a few doubts on this optimistic line of reasoning.
V. A FEW MACROECONOMIC CONSEQUENCES OF THE RECENT CAPITAL FLOWS
The exchange-rate-anchor strategy followed by the Real Plan would not have been possible had the Central Bank not accumulated the foreign reserves before the plan (see Chart 12). The foreign reserves were and still are the insurance policy of the stabilization plan until the necessary fiscal reforms are undertaken. However, this strategy, which the government claims not to be inconsistent because it is only temporary, has many adverse effects. From those, we highlight the most important ones, which are the overvaluation of the currency, with the detrimental effects on the current account of the balance of payments, and the quasi-fiscal burden caused by the massive sterilization at very high interest rates.
A. Real Exchange Rate Appreciation
The usual real appreciation that accompanies stabilization plans was exacerbated in the Brazilian case by the massive capital inflows that occurred both before and after the Real Plan. As in all exchange-rate-anchored stabilization plans, there has been an endless discussion in Brazil about the size of the overvaluation of the currency. Given the well known disparity between the price of tradable and non-tradable goods in the early stages of the stabilization (see Bruno, 1993), the measures of the degree of overvaluation vary greatly depending on whether a domestic CPI or WPI is used as a proxy for domestic inflation (or unit labor cost, apparently the most suitable measure).
Since to measure the degree of overvaluation is not the subject of this paper, we will only note that most analysts point out that a current account deficit of no less than 3 percent of GDP will arise if the current real exchange rate is maintained. This is clearly a risky strategy in a world of volatile capital flows. On the other hand, if further progress is not attained on the fiscal side, a devaluation of the real would not save the stabilization plan. Of course, if fiscal deficits of the magnitude of the recent ones persist (nominal deficit of 7.4% and 6.0% of GDP for 1995 and 1996, respectively), no nominal anchor will prevail in the long run.
B. Increase in the Quasi-Fiscal Deficit and in the Domestic Debt Because of Massive Sterilization
The other macroeconomic consequence we highlight is the quasi-fiscal burden of the massive sterilization prompted by the capital inflows. Since foreign reserves held at the Central Bank are invested at international rates, the (ex ante) cost of those reserves is approximately the (very high) CID. This burden appears on the fiscal accounts as interest rate payments. Interest rate payments, however, were responsible only for one-fourth of the massive negative fiscal shift that occurred in 1995 (the operational fiscal balance moved from a 1.3% of GDP surplus in 1994 to almost 5.0% deficit in 1995, and to a 3.89% deficit in 1996). The other three-fourths occurred because of the worsening position of the accounts that enter the primary surplus, which fell from 5.2% of GDP in 1995 to only 0.4% in 1995.
Furthermore, the burden of the interest rate payments can no longer be blamed only on capital inflows, as the fiscal situation has worsened considerably since the start of the stabilization plan. Chart 14 shows the growth of the domestic net federal debt.(18) Until mid-1995, all the growth of the debt since January 1991 could be fully explained by two factors: the unfreezing of the blocked bank accounts which occurred in the Collor administration,(19) and the accumulation of foreign reserves. Note the dramatic changes after July 1995. Until december 1996, a gap of almost US$70 billion opened up between the net federal debt and the sum of the other two series. This is the consequence of the deterioration of the fiscal accounts, although some outlays may not show up immediately in the fiscal accounts, such as loans to troubled financial institutions.(20)
Capital flows to Brazil resumed in the early nineties. Brazil, as many other developing economies, profited from favorable external factors (see Calvo et al., 1992). The liberalization of exchange flows and the renegotiation of the foreign debt allowed the economy to place itself as one of the main recipients of foreign capital flows. With the success of the current Real Plan, this trend became even stronger.
Nevertheless, the main determinant of foreign capital flows has been the huge interest rate differential between domestic and international interest rates. This differential, which is maintained to guarantee the domestic consistency of the stabilization plan until further fiscal reforms are enacted, has attracted massive flows of short-term speculative capital.
We use the covered interest differential (CID) as the preferred measure of the attractiveness of Brazilian bonds to foreign investors. The CID represents the extra gain that an investor would have by investing in the Brazilian bond market instead of fixed income abroad, already discounting the exchange rate risk. The econometric results show that, until 1995, the huge interest rate differential has been the main factor responsible for the massive capital flows to Brazil in the nineties.
We surveyed and analyzed the restrictions to those short-term capital inflows. The main conclusion is that these restrictions have not been able to prevent foreign capital from profiting from arbitrage opportunities with Brazilian bonds, although there are not enough observations to determine how much capital would have flown in had the restrictions not been in place.
Capital flows have exacerbated the substantial real appreciation of the domestic currency since the Real Plan of July 1994. This has harmed the current account balance, raising doubts as to the long-term sustainability of the exchange-rate anchor. Given the very high level of foreign reserves (almost 10% of GDP or 1.2 years of imports), the government claimed to have enough buffers to confront a reversal of capital flows. This remains to be seen, but the only sure thing is that the current fiscal deficit is not consistent with inflation stabilization in the long run.
Because of the sterilization undertaken, the inflows of foreign capital have also provoked an increase of the domestic debt and of the quasi-fiscal deficit. This complicates the fiscal situation.
In sum, since the end of 1991, with only a few minor interruptions, Brazil has received massive foreign capital inflows. Until 1995, those flows were not predominantly caused by bright investment opportunities in fixed capital or even in the stock market, but by an enormous interest rate differential that was generated both by very high domestic rates and by low rates abroad in most of the previous five years. These flows resulted in a major accumulation of foreign reserves, which served as an insurance policy of the Real Plan, permitting the Brazilian government to keep an exchange rate anchor for a very long time.
However, the easy (but costly) access to foreign savings has a detrimental incentive effect on the government as to its determination to push forward economic reforms needed to balance the budget, open the economy, reform the tax and pension systems, privatize state-owned companies and banks, allow free entry to sectors previously restricted to government enterprises (oil, telecommunications, and infrastructure), reform the public sector, among others. All those reforms, and, above all, the fiscal restriction needed to balance the budget are very costly in political terms.
The capital flows to Brazil represent, therefore, a blessing and a curse. They are a blessing because without them the Real Plan would not have subsisted. They are a curse because, as the recent political economy literature has documented, structural reforms are usually a result of crises, not good times. By making foreign savings available, capital flows reduce the sense of urgency of the structural reforms, thereby jeopardizing the ultimate success of the stabilization plan.
To pursue the structural reforms despite the unfavorable short-term political trade-off is, therefore, the main challenge to the current administration. As shown in the Mexican case of December 1994, investors may change their minds extremely fast as to the likelihood of the success of a stabilization policy, once the structural reforms are not being carried on. Another important stylized fact of the Mexican experience is that the nationals, not the foreigners, were the first to leave Mexican investments (see, among others, Frankel and Schmukler, 1996). The capital flows that have served so far as a backbone of the stabilization plan may very well turn into a very hard punishment if the notoriously volatile investors' confidence is threatened.
Acknowledgment: This paper was written for the Seminar "El Retorno de los Capitales Privados a America Latina," held in Santiago de Chile, July 30 and 31, 1996. We thank Felipe Larrain and other participants of the Seminar for their comments. The superb research assistant work done by Marcus Vinicius Valpassos is acknowledged. Ilan Goldfajn provided invaluable comments. The Brazilian Central Bank through its research department (DEPEC) provided us with unpublished data. All errors are ours.
[TABULAR DATA FOR APPENDIX 1 OMITTED]
* Direct all correspondence to: Marcio G.P. Garcia, Departmento de Economia, PUC/Rio, Rua Marques de Sao Vicente, 225, Rio de Janeiro, RJ 22453-900 Brazil.
1. The main items refinanced were US$9.5 billion with the Club of Paris, and US$7.1 of unpaid interest to banks (through IDU bonuses) (Banco Central do Brasil, 1992).
2. These figures come from Table IV. 12, Capital Movement, of the Brazilian Central Bank Bulletin (April 1996), and are not directly comparable to those presented in the Capital Account of the Balance of Payments (Table IV. 1). The main differences are that the following items are not included in Table IV. 12: short-term capital flows; inflows due to refinancing; and inflows and outflows of Brazilian residents (except for the item investments). Note also that the Brazilian Central Bank includes portfolio investment in medium and long-term capital movements.
3. Tax rates vary among firms; the 48% tax rate was the actual tax rate paid by a large U.S. oil firm in Brazil.
4. The suppliers and buyers' short-term credits (less than 360 days) enter the short-term capital account of the balance of payments.
5. Calvo, Leidermau, and Reinhart (1992) conclude that external factors (basically low interest rates in the U.s.) have been dominant for the capital flows to Latin American countries, up to 1992.
6. The ex post result was a real appreciation, which further increased the arbitrage gains.
7. For a detailed description of this period, see Carneiro and Garcia (1994).
8. As is well known, futures prices are not unbiased expectations of the future spot price (Hodrick, 1987, has a very comprehensive review of the unbiasedness tests for exchange forward prices until then). However, what we need to compute, the CID, is exactly what is provided by the futures market, a hedge against the exchange rate risk.
9. The IDU Bonds are bonds issued by The Federal Republic of Brazil under the terms of the Brady refinancing agreement. The issue size was US$7,200 million and the issue date was November 20, 1992. The maturity is January 1, 2001, and the average life is five years. Most of its payments are still due in the term of the current president, which makes this Brady bond more of a fixed income security than an equity.
10. This conclusion seems to disagree with the practice of the credit rating companies. In most cases those companies give a better credit rating to the "sovereign risk" of government securities issued in domestic currency than to the "sovereign risk" of government securities issued in foreign currency. For example, the "sovereign rating" for Portuguese securities issued in foreign currency is AA-, while it is AAA for Portuguese securities issued in domestic currency (Silveira, 1996).
11. "Both GDP growth and the investment GDP ratio are expected to have a positive effect on capital inflows. Clearly, capital would be more attracted to come into a growing economy, and one that is investing strongly. Finally, the coefficient on the current account deficit is expected to be positive, because a higher deficit will likely require more external financing" (Laban and Larrain, 1994, p. 12).
12. To partially address the criticism that the estimates may suffer from simultaneous equations bias, all regressions were rerun with the dependent variables lagged one quarter. The qualitative results did not change.
13. Interest rates were kept at very high levels to control aggregate demand in view of the lack of further fiscal adjustment.
14. For a description of derivatives in Brazil, see Braga (1996).
15. A box strategy consists of trading four options, two calls and two puts, so that the payment at the maturity date is fixed. Since the payment is fixed at the maturity date, a no-arbitrage argument leads to the conclusion that the return on the whole strategy must equal the riskless rate of return. In the Brazilian case, this is the rate on the interbank funds market (CDI).
16. By the end of 1996, the government withdrew many of the previous measures aimed at restricting the inflow of short-term capital.
17. Surveying the academic literature on capital controls, Dooley (1995) writes:
Empirical work on the "effectiveness" of capital controls has suffered from the lack of a widely accepted definition of what constitutes an effective control program. At one end of the spectrum, evidence of effectiveness has been defined as the ability to detect over extended time periods different average behavior of selected economic variables for countries with and without capital controls programs. At the other extreme, effectiveness has been defined as the ability to maintain an inconsistent macroeconomic policy regime forever.
18. This is the "Divida Mobiliaria Federal" minus the securities traded with the states and municipalities (LBC-Es).
19. That is, compulsory savings were transformed into voluntary savings.
20. In an appraisal of the Real Plan during its second birthday festivities, Bacha (1996) presented the following table.
Public Sector Net Debt as Percent of GDP: Brazil Item Dec-94 Dec-95 Apr-96 (1) Federal Government Net Debt 12.3 14.1 14.3 Gross Debt 31.4 35.2 36.6 Domestic Debt 18.1 23.3 25.5 In Treasury Securities 11.2 16.3 19.6 Foreign Debt 13.3 11.9 11.1 Credit 19.1 21.1 22.3 Domestic Credit 11.9 13.2 14.0 Given by the Central Bank 5.3 6.9 7.5 Foreign Reserves 7.2 7.9 8.3 (2) States and Municipalities' Net Debt 9.5 11.1 11.6 (3) State-owned Companies' Net Debt 6.7 7.0 6.8 Public Sector Net Debt (1 +2+3) 28.5 32.2 32.8 GDP (R$ billions) 537.3 656.3 689.8
Bacha concludes that from the 840 basic points (bp) increase in the net federal debt in securities from December 1994 and April 1996, only 200 bp were attributable to the fiscal deficit during that period (this is the increase in the net federal debt). The remaining 640 bp are equally split between the decrease in other less expensive types of debt (monetary base and foreign debt), and the purchase of assets by the federal government in the form of loans to public and private domestic financial institutions (220 bp) and foreign reserves (110 bp).
Bacha, E. L. 19906. Plano Real: Uma Segunda Avaliacao. Mimeo, Instituto deconomia da UFRJ.
Banco Central do Brasil. 1993. Relatorio do Banco Central do Brasil: 1992, Brasilia.
Banco Central do Brasil. 1904. Relatorio do Banco Central do Brasil: 1993, Brasilia.
Banco Central do Brasil. 1995. Relatorio do Banco Central do Brasil: 1994, Brasilia.
Banco Central do Brasil. 1996. Relatorio do Banco Central do Brasil: 1995, Brasilia.
Braga, R. 1996. "Derivatives Markets in Brazil: An Overview," The Journal of Derivatives, 4 (1).
Bruno, M. 1993. Crisis, Stabilization and Economic Reform: Therapy by Consensus. Oxford: Clarendon Press.
Calvo, G. A., L. Leiderman, and C. M. Reinhart. 1992. Capital Inflows and Real Exchange Appreciation in Latin America: The Role of External Factors. IMF Working Paper WP/92/62.
Calvo, G. A., L. Leiderman, and C. M. Reinhart. 1994. "The Capital Inflows Problem: Concepts and Issues," Contemporary Economic Policy, XII, July.
Carneiro, D. and M. Garcia. 1994. "Flujos de Capital y Control Monetario bajo Sustitucion Domestica de Dinero: La Reciente Experiencia Brasilena," in Afluencia de Capitales y Establizacion en America Latina, edited by R. Steiner. Bogota, Colombia: Fedesarollo.
Dooley, M. 1995. A Survey of Academic Literature on Controls over International Capital Transactions. NBER Working Paper #5352, Cambridge, Massachusetts.
Frankel, J. 1991. "Quantifying International Capital Mobility in the 1980s," in National Saving and Economic Performance, edited by B. Bernheim and J. Shoven. Chicago: National Bureau of Economic Research, The University of Chicago Press, Chicago.
Frankel, J. and S. Schmukler. 1996. Country Fund Discounts and the Mexican Crisis of December 1994: Did Local Residents Turn Pessimistic Before International Investors? Paper presented at the NBER Summer Meetings, Cambridge.
Hodrick, R. 1987. The Empirical Evidence on the Efficiency of Forward and Futures Foreign Exchange Markets. New York: Harwood Academic Publishers.
Johnston, J. 1984. Econometric Methods, 3rd ed. New York: McGraw-Hill.
Laban, R. and F. Larrain. 1993. Can a Liberalization of Capital Outflows Increase Net Capital Inflows? Documento de Trabajo [N.sup.o] 155, Instituto de Economia, Santiago.
Laban, R. and F. Larrain. 1994. What Drives Capital Inflows ? Lessons from the Recent Chilean Experience. Documento de Trabajo [N.sup.o] 168, Instituto de Economia, Santiago.
Silveira, S. 1996. "Panorama Macro Financeiro: Comentarios," Macrometrica, Rio de Janeiro, October 17.
White, H. 1986. Asymptotic Theory for Econometricians. New York: Academic Press.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||includes appendix|
|Author:||Garcia, Marcio G.P.; Barcinski, Alexandre|
|Publication:||Quarterly Review of Economics and Finance|
|Date:||Sep 22, 1998|
|Previous Article:||Do foreign investors stimulate or inhibit stock market development in Latin America?|
|Next Article:||Risk-capital inflows, inflation, and macroeconomic policy in India, 1993-95.|