A. The Crawling Exchange Rate Band and Inflation Targeting

From the mid 1980s and until September of 1999, the exchange rate of the Chilean peso was kept within a crawling band defined around an implicit target for the RER, the central parity exchange rate (CPER) (see Figure 1). The CPER crawled with daily adjustments make to reflect the estimated inflation rate differential between Chile and its main trading partners⁵. Initially, the last month’s CPI inflation rate was used to represent domestic inflation, while a projection for trading partners’ WPI measured in U.S. dollar represented external inflation. Undoubtedly, this form of calculating the CPER contributed to inflation inertia already present in the Chilean economy due to widespread indexation of wages and key prices. Only in 1998, the lagged inflation rate was replaced by the inflation target in the calculation of the CPER.

The Exchange Rate Band¹

Citation: IMF Working Papers 2005, 132; 10.5089/9781451861518.001.A001

¹Chilean pesos per U.S. dollar, monthly averages

• Download Figure
• Download figure as PowerPoint slide

View Full Size

The Exchange Rate Band¹

Citation: IMF Working Papers 2005, 132; 10.5089/9781451861518.001.A001

¹Chilean pesos per U.S. dollar, monthly averages

• Download Figure
• Download figure as PowerPoint slide

The Exchange Rate Band¹

Citation: IMF Working Papers 2005, 132; 10.5089/9781451861518.001.A001

¹Chilean pesos per U.S. dollar, monthly averages

• Download Figure
• Download figure as PowerPoint slide

For most of the period of its application, the band did not fully work as a space within which the value of the Chilean peso floated in response to market pressures. The exchange rate repeatedly hit the edges of the band; in the 1980’s the band ceiling, in most of the 1991-97 period the band floor, and again the ceiling for a short period in 1998 after the transitory narrowing of the band.⁶ The modifications of the exchange rate band usually implied a band widening or otherwise an accommodation of market pressures, with the single exception of the measures implemented in June 1998. The band width was increased from +/- 5% in 1990 to +/- 10% in 1992, and to +/-12.5% in 1997. In June of 1998, however, the band was transitorily narrowed to +2% and -3.65% around the CPER. This anomalous episode ended in September 1998 when the Central Bank announced the continuous widening of the exchange rate band from an initial band width of +/-4%. The continuous widening of the band ended in December 1999 with the free floating of the Chilean peso.

Several redefinitions of the level of the CPER also helped put some distance between the prevailing market rate and the edges of the band, and to alleviate the pressures on foreign exchange intervention. In 1992, the CPER was redefined from the U.S. dollar to a basket of currencies that included, in addition to the dollar, the deutsche mark (DM) and the Japanese yen (JY). The inherent exchange risk due to the co-movements of the band and the international value of the dollar against the DM and the JY was expected to act as a wedge between the market exchange rate and the band limits, i.e. the floor during the period of surging capital inflows. The DM and the JY progressively gained relative importance in the definition of the CPER throughout most of the period so that the currency composition of the exchange rate basket approached that of the Chilean international trade with the different currency areas and at the same time resulted in a more volatile CPER against the U.S. dollar. However, in 1997 the currency basket was modified sharply increasing back the weight of the U.S. dollar, in a moment in which the resulting recalculated CPER was 2% more appreciated. This measure helped to release some of the existing market pressures towards peso appreciation. However, the changes in the currency basket and the associated policy actions taken were not effective in separating the market exchange rate from the band’s edges.

With an increasingly flexible exchange rate and a band that targeted the RER, an alternative nominal anchor was required, not surprisingly then, in 1991, the newly independent central bank put into effect an inflation targeting system. The long term objective was the reduction of the inflation rate that initially stood at Chile’s long-term average in the 20th century (around 30 percent), to the single digit rates prevailing in advanced economies. Clear signals of the policy intentions were given through annual inflation targets, which were consistently met. The effort proved successful and the inflation rate fell to single digit rates in 1996, allowing for the announcement of the steady-state inflation target beginning in 1999, a range between 2 and 4 percent. (Figure 2).

Inflation Targeting

Citation: IMF Working Papers 2005, 132; 10.5089/9781451861518.001.A001

Source: Central Bank of Chile.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Inflation Targeting

Citation: IMF Working Papers 2005, 132; 10.5089/9781451861518.001.A001

Source: Central Bank of Chile.

• Download Figure
• Download figure as PowerPoint slide

Inflation Targeting

Citation: IMF Working Papers 2005, 132; 10.5089/9781451861518.001.A001

Source: Central Bank of Chile.

• Download Figure
• Download figure as PowerPoint slide

A binding exchange rate band that limited currency appreciation resulted in additional capital inflows that complicated the task of monetary policy. This conflict between the exchange rate and monetary policy was reflected in the massive intervention in the foreign exchange market. The Central Bank increased its net foreign exchange position by more than US$17 billion from the beginning of the 1990s to the maximum point in 1997, with NIR peaking at US$ 20 billion, more than 12 months of imports and 25% of GDP. To preserve the control of monetary conditions, such massive forex intervention had to be sterilized through the issuance of Central Bank bonds and bills, thus contributing to the quasi fiscal losses of the Central Bank⁷.

The “impossible trinity” indicates that to have an independent monetary policy and a pegged exchange rate, integration with international capital markets should be restricted. In order to pursue an independent monetary policy directed at reducing inflation while pegging the exchange rate within a band, controls had to be applied to international capital flows (NKF).

B. Foreign Exchange and Capital Account Regulations

The capital account regulations evolved over time in an opportunistic fashion following changes in external financing conditions and exchange rate policy to provide some support to the objectives of monetary policy. The foreign exchange and capital flow regulations applied in Chile throughout the 1980s and 1990s represented a form of adaptation to an external environment characterized by wide swing in important variables. They were backed by a well-developed institutional system that could efficiently enforce them. The rules were transparent and non-discriminatory, published in the Central Bank Compendium of Foreign Exchange Regulations, leaving room for very little discretion in their application. The system of information and control included the registration of foreign exchange operations carried out in the formal exchange market (MCF), through which all capital account transactions had to be channeled.⁸ Other forex transactions could be affected in the legal but informal foreign exchange market, at a freely floating exchange rate that at times was significantly different from the official exchange rate. Forex market segmentation was characteristic of the 1980s, and as a result, Chile had to obtain waivers from the IMF Board for not fulfilling the conditions under Article VIII of the Fund Articles of Agreement. The main exchange restrictions included the obligation to sell in the banking system the foreign exchange proceeds from export and capital inflows within strict time limits.⁹

The non-financial private sector was always authorized to acquire foreign exchange in the rapidly expanding informal market. The spread between the rates in the formal and informal foreign exchange markets is a good indicator of the degree with which exchange restrictions were applied (see Figure 3). The spread, which in the 1980s typically fluctuated around 20 percent, fell in the early 1990s to around 5 percent. Following important liberalization measures, the spread fell below 2 percent in 1995 and finally disappeared completely in April 1997, when all exchange restrictions were discontinued.

The Informal Exchange Rate Margin of the Peso

(Spread of the informal and formal market exchange rates of Chilean peso vis-à-vis the U.S.dollar, percent.)

Citation: IMF Working Papers 2005, 132; 10.5089/9781451861518.001.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

The Informal Exchange Rate Margin of the Peso

(Spread of the informal and formal market exchange rates of Chilean peso vis-à-vis the U.S.dollar, percent.)

Citation: IMF Working Papers 2005, 132; 10.5089/9781451861518.001.A001

• Download Figure
• Download figure as PowerPoint slide

The Informal Exchange Rate Margin of the Peso

(Spread of the informal and formal market exchange rates of Chilean peso vis-à-vis the U.S.dollar, percent.)

Citation: IMF Working Papers 2005, 132; 10.5089/9781451861518.001.A001

• Download Figure
• Download figure as PowerPoint slide

Under the pressures towards peso appreciation, the main regulation to capital inflows was introduced in 1991 in the form of an unremunerated reserve requirement (URR) on capital inflows.¹⁰ The URR consisted in compulsory and non-remunerated deposit in foreign exchange that had to be maintained at the Central Bank for a period of one year. The URR was applied to almost all debt creating flows and to some portfolio inflows increasing the cost of external financing through the covered channels, and particular that of short-term indebtedness.

Circumvention reduced the effectiveness of the URR, and a continuous confusion regarding the aim of the regulations —between improving the composition of capital flows and containing the size of total flows—, weakened the policy response. Although coverage of the URR was extended over time, certain flows that were heavily used for circumvention remained exempted, being the most important foreign direct investment (FDI) and import suppliers’ loans. In addition to the legal circumvention, URR avoidance could have taken the form of illegal evasion by way of capital flowing into Chile outside of the regulations. But according to the balance of payments errors and omissions and other assets flows, that represent an estimation of the repatriation of non-identified assets and are a proxy for non-registered inflows, remained significant but stable over time, not showing any pattern or reaction that would be indicative of their use for circumventing the URR.

To the extent that the financing costs associated to the URR could be avoided, the arbitrage would become more intense and the effectiveness of the URR weaker. The URR coverage fluctuated around 50% of capital inflows, without showing a systematic trend.¹¹ Actions of economic agents modified the composition of external financing, partly to elude the URR, while changes in the regulation closed loopholes and improved the effectiveness of the instrument. Immediately after its imposition in 1991, less than 40% of the gross inflows were covered, the extensions and adjustments of 1992 increased the URR coverage to 60% of gross capital inflows, but then the coverage fell to 30% by 1994. New policy adjustments in 1995 and 1996 once again increased the coverage to 40%, keeping it around that value for the remainder of the URR period. The partial coverage of the URR may have originated microeconomic costs by further differentiating the access to external financing.¹² Through a generalization of the URR coverage, perfectly viable on technical grounds, it was possible to eliminate most of the leaks and strengthen the effectiveness of the mechanism while reducing its microeconomic costs. Closing the main loopholes on FDI, suppliers’ credit and direct trade financing, the URR coverage would have doubled, as additional flows for a value of US$ 4,000 million per year in 1996 and 1997 would have been covered by the mechanism.¹³

The efforts towards a more generalized application of the URR and related regulations were discontinued after the Asian crisis and related events implied the end of the surge in capital inflows and the beginning of a drought in external financing.¹⁴ In 1998, after the very sharp reduction in capital inflows —to Chile and to emerging markets in general— the URR rate was reduced first to 10% and then to zero. In the following year, the minimum withholding period for foreign investment was eliminated and, finally, two years later all remaining restrictions and regulations were discontinued and transformed into a system of statistical information requirements. At the end of the process of gradual liberalization, the degree of international financial integration of the Chilean economy, measured on the basis of foreign assets and liabilities, was higher than the average for emerging market economies with investment grade.¹⁵ In addition, Chile’s debt indicators improved continuously over the period and remained at the top of emerging markets.

C. Policy Responses to Surges and Sudden Stops

The real exchange rate is a key relative price for macroeconomic stability and resource allocation in emerging market economies. The RER volatility is a characteristic of emerging market economies that affects more severely those that are less advanced in the process of integration to the global economy. While Poland and Colombia appear to have a highly volatile RER, Singapore and Malaysia present more stable conditions, and Chile and Philippines are in an intermediate position. See Table 1.

Table 1.

International Comparison on the Volatility of RER

In percent

Note: Author calculation on the basis of IMF data for monthly real exchange rates. With IMF data, an increase in the RER index represents an appreciation.

Table 1.

International Comparison on the Volatility of RER

In percent

Real Exchange Rate Variability Coefficient (St. Dev./Mean)
Sample	Colombia	Chile	Malaysia	Philippines Poland		Singapore
1990 - 2003	15.17%	10.67%	9.08%	10.86%	21.49%	5.10%
1990 - 1997	16.62%	10.42%	5.02%	10.83%	20.09%	5.58%
1998 - 2003	13.02%	10.62%	4.85%	7.50%	8.18%	4.09%
Real Exchange Rate Cumm Deviations from Mean over 48 months
Max	17.81%	12.72%	8.80%	12.54%	22.63%	6.71%
Min	-15.37%	-9.80%	-9.74%	-8.56%	-26.02%	-3.99%
Std Deviation	9.83%	7.11%	6.51%	6.20%	13.62%	3.21%

Note: Author calculation on the basis of IMF data for monthly real exchange rates. With IMF data, an increase in the RER index represents an appreciation.

View Table

Table 1.

International Comparison on the Volatility of RER

In percent

Real Exchange Rate Variability Coefficient (St. Dev./Mean)
Sample	Colombia	Chile	Malaysia	Philippines Poland		Singapore
1990 - 2003	15.17%	10.67%	9.08%	10.86%	21.49%	5.10%
1990 - 1997	16.62%	10.42%	5.02%	10.83%	20.09%	5.58%
1998 - 2003	13.02%	10.62%	4.85%	7.50%	8.18%	4.09%
Real Exchange Rate Cumm Deviations from Mean over 48 months
Max	17.81%	12.72%	8.80%	12.54%	22.63%	6.71%
Min	-15.37%	-9.80%	-9.74%	-8.56%	-26.02%	-3.99%
Std Deviation	9.83%	7.11%	6.51%	6.20%	13.62%	3.21%

Note: Author calculation on the basis of IMF data for monthly real exchange rates. With IMF data, an increase in the RER index represents an appreciation.

Chile has kept a relatively constant RER volatility in the pre and post Asian crisis periods, while the RER volatility in all the other countries considered have been falling. This feature is revealing, given the change in Chilean policy from volatility repression to volatility adaptation in the latter period, as was implicit in the liberalization and float. The implication would be that during the period of volatility repression, Chilean policies reduced RER volatility, and that the volatility adaptation of the second period has taken place during a time of relative calmed waters. It is also worth noting in this respect that using a volatility indicator on the basis of deviations of the RER from its long term mean over a period of rolling 48 months, Chile’s RER volatility also appears higher than that of Singapore, Malaysia and now also the Philippines, but lower than Colombia and Poland. Moreover, the deviations of the rolling 48 month averages with respect to the long term average in Chile appears more tilted towards deviations caused by excessive appreciation (12.72%) than excessive depreciation (-9.8%).

Chile’s real exchange rate depreciated over 60 percent from 1981 to the en of that decade because of the drought in external financing that followed the Latin American debt crisis of the early 1980s. Immediately after that, and during a period of large capital inflows to emerging markets, the Chilean peso appreciated strongly almost reaching in 1997 the levels of the early 1980s. However, another period of scarce foreign financing as result of the Asian and Russian crisis of 1997-1998, generated yet another strong real depreciation through March 2003, after which a new period of real appreciation has begun (see Figure 4).¹⁶ The real appreciation episodes followed a sustained surge in net capital inflows (NKF) and the depreciations a sudden stop. Three year moving average NKF peaked in 1981 and again in 1997, while the sudden stops in external financing that followed the surges are somewhat hidden in the graph because of the moving averages used for NKF. In any case under this metric, minimum capital inflows depicting the final part of the sudden stop were registered in 1988 and again in 2001 and 2003.

Capital Inflows and Real Exchange Rate

Citation: IMF Working Papers 2005, 132; 10.5089/9781451861518.001.A001

Note: An increase in the RER Index represents a real depreciation.Source: Central Bank of Chile.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Capital Inflows and Real Exchange Rate

Citation: IMF Working Papers 2005, 132; 10.5089/9781451861518.001.A001

Note: An increase in the RER Index represents a real depreciation.Source: Central Bank of Chile.

• Download Figure
• Download figure as PowerPoint slide

Capital Inflows and Real Exchange Rate

Citation: IMF Working Papers 2005, 132; 10.5089/9781451861518.001.A001

Note: An increase in the RER Index represents a real depreciation.Source: Central Bank of Chile.

• Download Figure
• Download figure as PowerPoint slide

The swings in the real exchange rate have a dual in the significant deviation from what could be considered sustainable use of external financing. During the real appreciation of the 1990s Chile’s current account deficit deteriorated, up to a point that doubts about its sustainability aroused, and then a reversal in the current account took place with a few years of surpluses. In 1990-97, private domestic expenditure (¾ of domestic demand) doubled after growing at an annual average rate of 10%. Reduction in the current account deficit was initiated in 1998 through a contraction in domestic demand, and continued its adjustment in 1999 and beyond, reaching almost balance in 2001-02 despite the unfavorable terms of trade.

The rapid expansion of private spending in 1991-1997 co-existed with a high level of the domestic interest rate, the average of the 90-day deposit rate in 1991-97 stood at 6.5%, much higher than the average rate of the second half of the 1980s or than the rate observed since the Asian crisis (1998-2000). Inflation fell, and in addition, a stream for fiscal surpluses averaging 2 percent of GDP also took place in the same period. All these elements eliminate expansionary monetary or fiscal policy as the possible origin of the over heating of the economy. External conditions seem a more likely origin of the macroeconomic swings, including the improvement in the terms of trade. However, the effects of terms of trade in domestic demand would be of little significance to the extent that foreign financing were available to confront what ends up being temporary income changes.¹⁷ To understand the macroeconomic ups and downs affecting Chile, the role of capital flows is crucial, thus we should focus on them.

A surge in net capital inflows to Chile took place in the first half of the 1990s and during this period annual net capital inflows averaged 7.3 percent of GDP. In addition, the cost of external financing for domestic borrowers in Chile fell significantly, from 1.8% over LIBOR for the average Chilean borrower in 1991 to 0.7% above LIBOR in 1996-97.¹⁸

The surge ended abruptly in mid 1998, inducing a large depreciation of the peso and a sharp tightening of monetary policy to prevent an acceleration of inflation. The fiscal balance deteriorated to a small deficit, mostly as a result of lower revenues associated to the recession that affected Chile from mid 1998 to mid 1999, being clearly a consequence rather than a cause of the sudden stop.

A common policy reaction to confront the effects on the real exchange rate of the capital flow volatility is the defense of some exchange rate level using sterilized intervention. However, such intervention does not modify the incentives for the private sector to take positions in foreign exchange and it is even possible that it may even favor additional speculation by rising domestic interest rates and by generating the expectations of a bigger pay-off when the resistance to the exchange rate adjustment is defeated. Speculators know that the defense of the exchange rate through sterilized intervention is costly, resulting in large open foreign asset position for the Central Bank that involve significant risks and generated sizeable losses.¹⁹

Non-sterilized intervention modifies incentives against holding liability positions in foreign currency and may be effective in containing the capital flow surge. The defense of the exchange rate band through non-sterilized intervention results in a more accommodative monetary policy that lowers interest rates and reduces the attraction of capital inflows. But such policy direction may also result in a far more expansionary monetary policy stance than the one consistent with the inflation targets. In Chile’s experience of the 1990s, targeting the exchange rate by means of non-sterilized intervention would most likely have implied missing the inflationary target. Note that after the sudden stop in 1998, non-sterilized intervention was successfully used in the defense of the then narrowed exchange rate band, however, at the cost of an extremely high real interest rate, a drop in economic activity, and, more importantly a sharp reduction of the core inflation rate from 6.0 percent in June 1998 to 2.1 percent in December 1999, well below that years target of slightly above 4.3 percent (see Figure II. 2).

A more neutral Fiscal Policy aimed at a stable structural fiscal surplus rather than a stable nominal surplus would have helped to moderate aggregate demand expansion and thus help to limit the RER appreciation.²⁰ However, effectively controlling the excessive expansion of private spending would have required steep expenditure cuts or pro-cyclical tax increases, and would have resulted in even larger fiscal surpluses. Such demanding fiscal policies never had sufficient political backing.

The options to confront the surge of capital inflows were two, either imposing and reinforcing restrictions on capital inflows, or allowing the exchange rate to float.

Accepting the importance of meeting the inflation target implied the need to keep the independent monetary policy focused on inflation rather than on the exchange rate, while the political constraints that limit the size of the fiscal surpluses eliminated a possible additional contribution from fiscal policy. Floating would, after all, have allowed monetary policy to regain full control over inflation and have eliminated the speculations surrounding band adjustments and discrete jumps in the exchange rate. But at the same time, it would have implied to fully realize the effects of the surges on capital flows on the RER, and this was subject to the fear of floating.²¹ An episode of ample RER volatility inducing financial distress was one that nobody wanted to see happen and doubts on the resiliency of the financing system to such event were still present.

Without the pre-conditions for a successful float, capital inflow regulations were used to confront the capital flow surge in the early and mid 1990s, with the URR being the most relevant. Since the term of the compulsory deposit was one year, irrespective of the maturity of the foreign financing, the financial cost of the URR (fc) was higher the shorter the maturity of the loan, being prohibitive for very short maturities and almost nil for long ones. Up to a certain value of the domestic-foreign expected spread of returns, the URR did eliminate arbitrage opportunities. However, when the expected spread exceeded the financial cost of the URR, the mechanism did not eliminate the incentives for arbitrage. The latter would have been the case under expectations of a sudden currency appreciation, or when the efforts of monetary policy to contain the expansion of domestic spending were intensified.²² Despite its limitations, the URR was effective in containing capital inflows and had a positive and statistically significant effect on the real interest rate giving additional room of maneuver to monetary policy.²³

D. Conditions to Liberalize and Float

As the gradual liberalization process advanced, the approach of the Chilean authorities to confront external volatility changed emphasis. The aim of using policy instruments to repress capital flow and exchange rate volatility was gradually replaced by the aim of adapting the economy and in particular the financial system to increased external volatility by meeting the conditions required for a successful float. The preservation of monetary independence under the constraints of the impossible trinity, required that financial opening had to be advanced in parallel with exchange rate flexibility, and capital account liberalization became associated to exchange rate floating.

In many cases, exchange rate floating is a policy used as a last resort when confronting a traumatic event; however, those cases of disorderly exits have important handicaps to become successful floats as they are many times associated to balance of payments or financial crisis. A float consistent with the preservation of macroeconomic and financial stability requires meeting a number of pre-conditions, including opportunity, nominal anchor, exposure to exchange rate risk, and financial markets development. In that regard, the opportunity to float is relevant to maximize the probability of success. Allowing the currency to float during a surge of capital flows or just after of a sudden stop entails the risk of an over or under shooting of the real exchange rate that may endanger macroeconomic stability and result in misallocation of resource. It appears much preferable that the capital account regulations and the exchange rate band are lifted when the risk of a large and sudden appreciation or depreciation of the currency is minimal. In Chile, the capital account liberalization was completed after the economy had gained recognition for its macroeconomic and financial stability. The indicators of external solvency and liquidity, of inflation, fiscal consolidation, and domestic financial system strength were all quite improved after the debt crisis was overcome, reaching levels among the best of emerging market countries.

Floating implies loosing the exchange rate as a nominal anchor, thus entailing the risk that capital flows pressure or other shocks resulted in large depreciations (or appreciations) of the currency and lead to inflationary (or deflationary) spirals. Monetary policy should provide an alternative nominal anchor, but the independence of the monetary authority and the absence of fiscal dominance are preconditions for its effectiveness. With a credible monetary policy, the depreciation (or appreciation) is likely to be transitory and the effects on price stability can be satisfactorily dealt with at reasonable costs in terms of out put. In Chile, a credible monetary anchor, alternative to the exchange rate was successfully established by the independent Central Bank, fiscal dominance was completely eliminated and the external liquidity position strengthened before the currency was let to float.

The exchange rate volatility that could result from the liberalization of the capital account and floating could affect the health of a financial system with a large exposure to exchange rate risk. As a precondition to float, the foreign exchange and interest rate exposures of financial institutions should be limited to adequately cope with increased exchange rate volatility. Under floating there are no guarantees that the exchange rate will be stable, or that the interest rate will have no need to adjust. It is important, then, to ensure that financial institutions can cope by limiting their exposure to market risks using financial regulations or internal risk management plans. In addition, financial system borrowers carry exchange rate risks that can be transmitted to financial institutions through their credit exposure and become a source of systemic risk. The limits to credit risk should also be defined in consideration to the foreign currency exposure of debtors, and this secondary foreign exchange risk derived from the exposure of clients should also be explicitly monitored by financial institutions and supervisors, and considered in managing market risks.

In Chile, the liberalization of capital flows and the exchange rate float was advanced after implementing a system to prevent asset-liability mismatching in foreign currency. Limits for the asset liability mismatches of banks in currencies as well as in liquidity and interest rates were imposed in relation to the banks capital and reserves. Floating took place in Chile after the financial system was subject to a well-developed system of supervision and regulation, including special provisions for direct and indirect exposure of banks to exchange rate risk.

To help agents coping with exchange risk, the development of liquid local financial markets, particularly for foreign exchange instruments, spot and forward, is key. In this way, financial institutions and their clients would have access to markets in which their risk exposures can be covered. A continuous development of the local market for the coverage of exchange risk took place in Chile from 1995 on, with a continuous increase in the forward forex market deepness, not only in terms of larger daily average transactions but also in terms of a progressive extension in the contracts maturity. All this development increasingly allowed that foreign exchange risk could be shared and be transferred towards those parties better equipped to confront it.

Dollarization of domestic financial liabilities has been an important origin of currency mismatches in private portfolios and an impediment for floating in many countries in Latin America. In Chile, financial dollarization has not been important as the local financial market developed on the basis of domestic currency instruments. Nominal instrument at the short end of the maturity spectrum, and inflation indexed instruments for the long end. However, even in the absence of financial dollarization, currency mismatches were still relevant since a relatively large portion of the total financing need of the private sector was obtained through foreign debt, with many of the domestic borrowers being producers of non-tradable goods. The only way to reduce those mismatches is through deeper markets for financial instruments denominated in Chilean pesos, either by developing the local financial markets and enhancing foreign participation or through a further internationalization of the Chilean peso, by issuing in international markets foreign debt contracts and other financial instruments denominated in domestic currency.

Prospective reports on macroeconomic conditions and external vulnerabilities may provide a more clear risk evaluation to foreign investors, reducing information problems or surprises that may originate surges and sudden-stops in NKF. Transparency in informing markets should not only be limited to the policy framework, targets and instruments, but also encompass information on the evolution of macroeconomic and financial conditions. Although there are no guarantees that market behavior that leads to swings in capital flow like panics, mania or herd behavior will be avoided, information on policies, conditions and medium-term external prospects should contribute to market stability. In Chile, a communications policy of increasing transparency was implemented prior to floating including the publication of a monetary policy report three times a year, the publication of quarterly fiscal reports and the dissemination of the documents associated to the annual Article IV Consultation with the IMF.

A. Arbitrage and Real Exchange Rate

The real exchange rate can be presented as a financial variable that responds to the arbitrage of returns between assets denominated in domestic and foreign currency.²⁵ The value of today’s RER (LRER in log form) that would result from perfect arbitrage of returns is the arbitrage RER (arb[LRER(t)]). This value is given by the expected future exchange rate, E[LRER(t+1)], minus the domestic-foreign interest rate differential (rdom-rext)²⁶ as presented in equation (1).

Where rdom represents the short-term real interest rate in domestic currency paid by Central Bank of Chile in its short-term CPI indexed instruments and rext represents the short-term real interest in foreign currency relevant for the Chilean sovereign borrower. The foreign currency real interest rate is defined in equation (2) below, where libo90 is the 90 day libo rate for transactions in US dollars, EΠ *the expected US inflation rate, and srcl Chile’s sovereign risk premium. Given Chilean monetary policy of the last 15 years rdom can be considered a policy variable, while the evolution of rext is considered to be largely determined by international financial conditions and hence exogenous, dismissing the possible effects of policies on the sovereign spread since the focus is in the short-term.

Under perfect arbitrage conditions, equation (1) would guide the evolution of the RER, thus implying that the expected returns of holding assets in domestic or foreign currency would be the same, that is the interest differential would compensate any expected capital gains or losses resulting from RER realignments. However, it is possible that some profit opportunities remain unexploited and hence expected returns differ. On average, during the period under analysis (1990-2003), the annual returns of domestic currency assets exceeded those denominated in foreign currency by around 240 basis points.²⁷ The excess return for domestic currency assets is not associated to a higher counterpart risk, since domestic and foreign rates have been represented by the relevant borrowing rates for the Chilean “sovereign”. Behind the excess returns in domestic assets may be the effects of other risk premia —different from counterpart risk—that reflect limitations to domestic market liquidity, or the effect of policy actions directed at giving domestic monetary policy additional room for maneuvering, both through the use of the URR and by foreign exchange intervention of the Central Bank.²⁸

By definition, the arbitrage RER depends on the expected RER, which is an unknown variable. The estimation is then performed using two stage least squares assuming that the expectations on the RER are formed rationally using all the relevant available information.²⁹ The results are presented in table IV.1 and indicate that all the three different specifications for the financial RER model co-integrate according to the Engle Granger test.³⁰ The three of them present an important degree of error persistence as indicated by the low DW.

Table 2.

Chile: The Real Exchange Rate and Financial Arbitrage ¹

Method: Two Stage Least Squares (Monthly data 1990.1-2003.11)

Source: Estimations by the author.

¹Figures in parentheses correspond to P-values. Instruments Lrer(-1), fc, rdom-rext, Lfexint(-1), Volrdom, Volrer(-1), Volrext, Nkf_abm, Lcpr, Dumband

²Significant at 1 percent according to the Engle-Granger test for Co-integration.

Table 2.

Chile: The Real Exchange Rate and Financial Arbitrage ¹

Method: Two Stage Least Squares (Monthly data 1990.1-2003.11)

	Equation 1.1	Equation 1.2	Equation 1.3
Dependent Variable LRER
Constant	-0.8301 (0.100)	-0.2695 (0.001)	-0.1518 (0.029)
Arb[rer]	1.1203 (0.000)	1.0589 (0.000)	1.0340 (0.000)
Fc	0.3573 (0.384)	0.4358 (0.002)	-- (--)
Lfexint	0.0029 (0.867)	-- (--)	-- (--)
Nkf_abm	0.0002 (0.207)	-- (--)	-- (--)
Volrdom	0.3162 (0.528)	-- (--)	-- (--)
Volrer	0.2391 (0.357)	-- (--)	-- (--)
Volrext	-0.0115 (0.704)	-- (--)	-- (--)
Lcpr	0.0378 (0.246)	-- (--)	-- (--)
Dumband	0.0167 (0.249)	-- (--)	-- (--)
Adjusted R2	0.961	0.966	0.966
Durbin-Watson statistic	1.265	1.333	1.290
Dickey-Fuller Estimation Residuals	-8.721 c/	-9.102 c/	-8.890 c/

Source: Estimations by the author.

¹Figures in parentheses correspond to P-values. Instruments Lrer(-1), fc, rdom-rext, Lfexint(-1), Volrdom, Volrer(-1), Volrext, Nkf_abm, Lcpr, Dumband

²Significant at 1 percent according to the Engle-Granger test for Co-integration.

View Table

Table 2.

Chile: The Real Exchange Rate and Financial Arbitrage ¹

Method: Two Stage Least Squares (Monthly data 1990.1-2003.11)

	Equation 1.1	Equation 1.2	Equation 1.3
Dependent Variable LRER
Constant	-0.8301 (0.100)	-0.2695 (0.001)	-0.1518 (0.029)
Arb[rer]	1.1203 (0.000)	1.0589 (0.000)	1.0340 (0.000)
Fc	0.3573 (0.384)	0.4358 (0.002)	-- (--)
Lfexint	0.0029 (0.867)	-- (--)	-- (--)
Nkf_abm	0.0002 (0.207)	-- (--)	-- (--)
Volrdom	0.3162 (0.528)	-- (--)	-- (--)
Volrer	0.2391 (0.357)	-- (--)	-- (--)
Volrext	-0.0115 (0.704)	-- (--)	-- (--)
Lcpr	0.0378 (0.246)	-- (--)	-- (--)
Dumband	0.0167 (0.249)	-- (--)	-- (--)
Adjusted R2	0.961	0.966	0.966
Durbin-Watson statistic	1.265	1.333	1.290
Dickey-Fuller Estimation Residuals	-8.721 c/	-9.102 c/	-8.890 c/

Source: Estimations by the author.

¹Figures in parentheses correspond to P-values. Instruments Lrer(-1), fc, rdom-rext, Lfexint(-1), Volrdom, Volrer(-1), Volrext, Nkf_abm, Lcpr, Dumband

²Significant at 1 percent according to the Engle-Granger test for Co-integration.

The first attempt at empirically asses the effectiveness of financial policies in guiding the RER consist of verifying whether other variables can explain its evolution. In the first specification for the RER under arbitrage (1.1), the policy variables considered are the financial cost of the URR (fc), the central parity rate in log form (lcpr), a dummy for the period of application of the exchange rate band (dumband), and the cumulative foreign exchange intervention by the central bank in log form (lfexint). In addition to the policy variables, considered as explanatory variables are the volatility of domestic real returns (Volrdom), of the foreign currency real returns (Volrext) and of the RER (Volrer), and, finally the last explanatory variable of this first specification is the net foreign capital inflows to the largest Latin-American countries (Nkf_abm).³¹

In this first specification for the RER under arbitrage (equation (1.1)), the effects of several of the variables considered are not statistically significant. Eliminating those variables in steps with increasingly more demanding significance levels, only the financial cost of the URR (fc) and the arbitrage RER are left when the cut-off point of statistically significance gets to 5 percent. Those two are the only explanatory variables considered in the second specification (eq 1.2). The exclusion of the other variables (Lfexint, Nkf_abm, Volrdom, Volrext, Volrer, Dumband and Lcpr) is not rejected by the data at 10 percent significance.³² Moreover, the exclusion of the seven variables plus fc is rejected by the data at 10 percent significance.³³

In the third specification (1.3), the effect of fc is restricted to be equal to zero, so that the RER is regressed only on the constant and the arbitrage RER, however the exclusion of fc is rejected at 1 percent significance.³⁴ Consequently, the preferred specification (1.2) is the one that includes the effect of the URR financing cost as explanatory variable for the short-term evolution of the RER.

The financial approach tells us that given the expected future RER, and the domestic and foreign interest rate differential, the URR financing cost has a statistically significant effect on the RER. This effect consists in generating a more depreciated RER than the one that would result from arbitrage conditions in the absence of the URR. On the contrary, the foreign exchange intervention does not appear to have any statistically significant effect on the level of the RER within the arbitrage model.³⁵

The error-correction model (ECM) for the RER built around specification 1.2 does not include forex intervention or the URR financing cost as explanatory variables. The first difference is represented by a “d” preceding the variable, while the “p” values are presented between brackets below each estimated coefficient. Despite extending the lengths of the lags to 24, it was not possible to obtain an ECM estimation with a reasonable DW. In any case the results show a strong and very significant error correction response to the lagged residual of the co integration regression.

Table 3.

Chile: Error-Correction Model for RER

(Based on Equation. 1.2) Method: Ordinary least squares (1991.01-2003.12)

Table 3.

Chile: Error-Correction Model for RER

(Based on Equation. 1.2) Method: Ordinary least squares (1991.01-2003.12)

$[\mathit{\text{dRER}}(t)]=-0.0007-{0.1281}^{*}[\mathit{\text{drdom}}(t)-\mathit{\text{drext}}(t)]-{0.0004}^{*}\mathit{\text{dnkf}}(t)$

$\dots \dots \dots \dots \dots \dots \dots (0.092)\dots ‥(0.004)\dots \dots \dots \dots \dots \dots \dots \dots \dots \dots \dots ‥(0.001)$

$+{0.0439}^{*}[\mathit{\text{dRER}}(t-9)+\mathit{\text{dRER}}(t-11)]-{0.914}^{*}\mathit{\text{RESID}}(t-1)$

$‥(0.004)\dots \dots \dots \dots \dots \dots \dots \dots \dots \dots \dots \dots \dots (0.000)$

$\mathit{\text{Adj}}.R2..0.93\dots \dots \dots \dots \mathit{\text{DW}}0.23$

View Table

Table 3.

Chile: Error-Correction Model for RER

(Based on Equation. 1.2) Method: Ordinary least squares (1991.01-2003.12)

$[\mathit{\text{dRER}}(t)]=-0.0007-{0.1281}^{*}[\mathit{\text{drdom}}(t)-\mathit{\text{drext}}(t)]-{0.0004}^{*}\mathit{\text{dnkf}}(t)$

$\dots \dots \dots \dots \dots \dots \dots (0.092)\dots ‥(0.004)\dots \dots \dots \dots \dots \dots \dots \dots \dots \dots \dots ‥(0.001)$

$+{0.0439}^{*}[\mathit{\text{dRER}}(t-9)+\mathit{\text{dRER}}(t-11)]-{0.914}^{*}\mathit{\text{RESID}}(t-1)$

$‥(0.004)\dots \dots \dots \dots \dots \dots \dots \dots \dots \dots \dots \dots \dots (0.000)$

$\mathit{\text{Adj}}.R2..0.93\dots \dots \dots \dots \mathit{\text{DW}}0.23$

The URR appears to affect the RER in the context of a short-run financial model. However, these results are conditional on exchange rate expectations and the model does not give any indications on the variables that may affect the evolution of the RER over the medium- and long-term. To confront this issue a macroeconomic model for the determination of the RER is presented in the next section.

B. Real Exchange Rate (RER) as a Relative Price

We have already shown that the URR financing cost (fc) is the policy variable that has effects on the RER in the setting of a model for the international arbitrage condition. In this section we attempt to empirically asses the degree at which financial policy variables could modify the RER in the setting of a macroeconomic model for the relative price between tradable and non tradable goods. The RER plays a key role in the process of domestic macroeconomic adjustment responding to the fundamentals that affect relative supplies and demands. Empirical models for the RER can have different degrees of complexity, including sectoral, as well as domestic and foreign variables.³⁶ In a simple macroeconomic setting for quarterly data, five fundamentals are considered to define the equilibrium relative price:

• First, the net foreign asset position (NFA) represents wealth and affects domestic demand through inter temporal optimization. The larger the net foreign asset position and wealth, the higher the demand for goods in general. Higher domestic demand is expected to exert upward pressure on nontradables goods prices, thus lowering the relative price of tradable goods and appreciating the RER.

• Second, the terms of trade (LTT), or relative price of exported versus imported goods. An increase in terms of trade is associated with higher real domestic income and domestic demand, and thus with a higher demand for non tradable goods and in turn with a lower relative price of tradable goods (a more appreciated RER). However, the improvement in the terms of trade also affect the intra-temporal optimization of the consumption basket between importables and exportables, so that its impact on the RER is uncertain depending on the structure of preferences and on the relative strength of the income and substitution effects.

• Third, import tariffs are also a fundamental variable. In general, for a higher protection rate, the relative demand for tradable goods would be lower implying that the RER appreciates with the level of protection, but also preferences and substitution effects may complicate the analysis.

• Fourth, the relative productivity in tradable and non-tradable goods is another of the fundamental variables that determine the equilibrium relative price of tradable goods. Avoiding sectoral variables due to data limitations, total labor productivity (LPROD) is included as an explanatory variable. Productivity advances take place primarily among tradable goods and thus higher productivity is associated with a reduction in the relative production costs and relative price of tradable over non-tradable goods, and, consequently with an appreciation of the real exchange rate.

• Fifth, government spending, which is considered to be more intensive in non tradable goods relative to other forms of domestic demand, tends to appreciate the RER.

The regressions for the real exchange rate under the macro model are presented in Table 4. The first regression, 2.1, includes only the fundamentals as explanatory variables and all the estimated coefficients have the expected signs being all of them, with the exception of the one for the terms of trade, statistically significant at 1 percent. But the residuals show some degree of error persistence, and furthermore, the relationship in 2.1 does not co integrate according to the Engle Granger test which could be related to the omission of relevant explanatory variables.³⁷

Table 4.

Chile: Macro Model Estimations for Real Exchange Rate¹

Method: Two Stage least squares (1990.2-2003.3)

Source: Estimations by the author.

^1/Figures in parentheses correspond to P-values. Instruments: lrer(t-1), volrer(-1), rdom-rext-fc, lgdp(-1), lempl(-1), ltt, gobmy, nfa(-1), ltarif, fkn_abm, lfexint(t-1), volrdom, lacdo, dumband.

^2/Significant at 1 percent according to the Engle-Granger test for cointegration.

Table 4.

Chile: Macro Model Estimations for Real Exchange Rate¹

Method: Two Stage least squares (1990.2-2003.3)

	Equation 2.1	Equation 2.2	Equation 2.3
Dependent Variable: LRER Explanatory variables
Constant	18.1049 (0.000)	8.6977 (0.022)	16.0374 (0.000)
LPROD	-1.6484 (0.000)	-0.7872 (0.122)	-1.2907 (0.000)
NFA	-0.2099 (0.023)	0.00762 (0.971)	-0.2176 (0.005)
LTT	-0.1609 (0.207)	-0.0516 (0.812)	-0.3281 (0.003)
GOBMY	-2.0955 (0.004)	-1.2315 (0.3875)	-1.4774 (0.013)
LTARI	-6.8723 (0.000)	-0.5273 (0.8028)	-4.7295 (0.000)
FKN_ABM	-- (–)	-0.0010 (0.185)	-0.0014 (0.000)
Volrdom	-- (–)	-1.8649 (0.274)	-4.4666 (0.000)
rdom-rext-fc	-- (–)	1.0923 (0.394)	-- (–)
Lcpr	-- (–)	0.4921 (0.182)	-- (–)
Volrer	-- (–)	4.6939 (0.318)	-- (–)
Dumband	--(–)	-0.0430 (0.524)	-- (–)
Lfexint	-- (–)	-0.1043 (0.335)	-- (–)
Adjusted R2	0.702	0.707	0.832
Durbin-Watson statistic	1.598	1.843	1.941
Dickey-Fuller Estimation Residuals	-3.187	-6.798	-7.014 2

Source: Estimations by the author.

^1/Figures in parentheses correspond to P-values. Instruments: lrer(t-1), volrer(-1), rdom-rext-fc, lgdp(-1), lempl(-1), ltt, gobmy, nfa(-1), ltarif, fkn_abm, lfexint(t-1), volrdom, lacdo, dumband.

^2/Significant at 1 percent according to the Engle-Granger test for cointegration.

View Table

Table 4.

Chile: Macro Model Estimations for Real Exchange Rate¹

Method: Two Stage least squares (1990.2-2003.3)

	Equation 2.1	Equation 2.2	Equation 2.3
Dependent Variable: LRER Explanatory variables
Constant	18.1049 (0.000)	8.6977 (0.022)	16.0374 (0.000)
LPROD	-1.6484 (0.000)	-0.7872 (0.122)	-1.2907 (0.000)
NFA	-0.2099 (0.023)	0.00762 (0.971)	-0.2176 (0.005)
LTT	-0.1609 (0.207)	-0.0516 (0.812)	-0.3281 (0.003)
GOBMY	-2.0955 (0.004)	-1.2315 (0.3875)	-1.4774 (0.013)
LTARI	-6.8723 (0.000)	-0.5273 (0.8028)	-4.7295 (0.000)
FKN_ABM	-- (–)	-0.0010 (0.185)	-0.0014 (0.000)
Volrdom	-- (–)	-1.8649 (0.274)	-4.4666 (0.000)
rdom-rext-fc	-- (–)	1.0923 (0.394)	-- (–)
Lcpr	-- (–)	0.4921 (0.182)	-- (–)
Volrer	-- (–)	4.6939 (0.318)	-- (–)
Dumband	--(–)	-0.0430 (0.524)	-- (–)
Lfexint	-- (–)	-0.1043 (0.335)	-- (–)
Adjusted R2	0.702	0.707	0.832
Durbin-Watson statistic	1.598	1.843	1.941
Dickey-Fuller Estimation Residuals	-3.187	-6.798	-7.014 2

Source: Estimations by the author.

^1/Figures in parentheses correspond to P-values. Instruments: lrer(t-1), volrer(-1), rdom-rext-fc, lgdp(-1), lempl(-1), ltt, gobmy, nfa(-1), ltarif, fkn_abm, lfexint(t-1), volrdom, lacdo, dumband.

^2/Significant at 1 percent according to the Engle-Granger test for cointegration.

The fundamental variables do not suffice to explain the swings of the quarterly RER in Chile over the last 15 years. Several additional explanatory variables related to financing conditions or forms of policy intervention that affect financing conditions were considered. The second regression, equation (2.2), includes in addition to the fundamentals, seven additional explanatory variables, the financial variables are:

• First, the supply of foreign financing measured by net capital flows to major Latin-American countries (nkf_abm), that should exert an appreciating effect on the RER

• Second, the domestic-foreign real interest rate differential adjusted by the URR financing cost fc, (rdom-rext-fc), represents the domestic pull that attracts capital inflows and is also expected to result in an appreciating response of the RER;

• Third, the foreign exchange intervention (Lfexint) that represents the cumulative acquisition of net foreign assets by the Central Bank. Such intervention should have a depreciating effect on the RER;

• Fourth and fifth, the presence of the foreign exchange band (Dumband), that could either prevent appreciations or depreciations, but most likely the former during the band yeas; and the central parity rate, that could serve as a point of attraction and guide for the RER;

• And finally, the volatilities of the domestic real returns (Volrdom) and of the real exchange rate (Volrer)

Adding the seven financial variables we obtain equation 2.2 and the data rejects the exclusion of those additional variables from the RER equation at 1 percent significance.³⁸ Even though the Durbin-Watson test improves to 1.84, the relationship represented by the regression 2.2 does not co-integrate at 10 percent significance according to the Engle Granger test.³⁹ Moreover, the seven additional variables result in multicollinearity that distorts the value and statistical significance of the estimated parameters.⁴⁰

The third regression, equation 2.3, is defined by the exclusion of some of the additional explanatory variables in 2.2. The five variables to be excluded were selected on the basis of their t-statistics, using increasingly more demanding statistical significance. The variables excluded were rdom-rext-fc, Lfexint, Lcpr, Dumband, and Volrer, and the results indicate that such exclusion is not rejected by the data at 10 percent significance.⁴¹ All the explanatory variables in 2.3 have the appropriate signs and are statistically different from zero at 1 percent significance. This relationship co-integrates at 1 percent significance according to the Engle Granger test ⁴², and the problem of residual autocorrelation appears to have been corrected, DW=1.94.

From the results, it is clear that the financial policy variables (rdom, fc and Lfexint) did not play an active role in explaining the RER swings by compensating for the effects of the capital flow variables. It is still possible, however, that the policy variables may have been relevant in affecting the RER by modifying the adjustment path towards its trend. That will be analyzed in the next section.

C. Policy Effectiveness and the RER

The error correction model (ECM) for the RER presented in table III.4 was developed in association with the co integration equation 2.3, where dRER is the first difference of the dependent variable (Real Exchange Rate in logs) and dXi are the first differences of the fundamental and financial variables included as explanatory in the macro model for the RER. Finally RESID(-1) is the lagged estimated residual in the co-integration equation 2.3.

The explanatory variables effectively included refer to current and lagged changes in fundamentals: in productivity (dprodl), in terms of trade (dtt) in net foreign assets relative to GDP (dnfa), in import tariffs (dtari), and in government spending relative to GDP (dgobmy). The explanatory variables also refer to financing conditions, including changes in the foreign capital flows to Latin American countries (df_abm); foreign exchange intervention (dfexi), and changes in the URR adjusted interest rate differential (drdom-drext-dfc). All the estimated coefficients associated to the selected explanatory variables are statistically significant at 1 percent and the regression presents no autocorrelation. Finally, the adjustment velocity of the real exchange rate towards its equilibrium trend is rather slow allowing for some persistence of the effects.

To asses the impact of key variables on the RER, total multiplier effects for relevant changes in those variables were constructed on the basis of the error correction model estimation results, and presented in Table 5. Those variables that are only included in the ECM but not in the co integration relationship like fc and lfexint, have only transitory effects on the trajectory of the RER but do not modify its equilibrium value.

Table 5.

Chile: Error-Correction Model for Real Exchange Rate (Based on Equation 2.3)^1/

Method: Ordinary least squares (1991.1-2003.3)

Source: Estimations by the author.

^1/Figures in parentheses correspond to P-values.

Table 5.

Chile: Error-Correction Model for Real Exchange Rate (Based on Equation 2.3)^1/

Method: Ordinary least squares (1991.1-2003.3)

$\begin{array}{l}[\mathit{\text{dRER}}(t)]=\underset{(0.000)}{-{0.5608}^{*}\mathit{\text{dProdl}}}\underset{0.0000)}{-{0.7634}^{*}\mathit{\text{dprodl}}(t-\mathit{1})}\underset{(0.000)}{-{0.1992}^{*}[\mathit{\text{dtt}}(t)+\mathit{\text{dtt}}(t-\mathit{1})]}\end{array}$

$\begin{array}{l}\underset{(0.0000)}{-{0.1524}^{*}[\mathit{\text{dnfa}}(t)+\mathit{\text{dnfa}}(t-2)]}\underset{(0.0000)}{-{0.8136}^{*}[\mathit{\text{dgobmy}}+\mathit{\text{dgobmy}}(t-2)]}\underset{(0.000)}{-{1.5064}^{*}\mathit{\text{dgobmy}}(t-1)}\end{array}$

$\begin{array}{l}\underset{(0.0000)}{-{3.715}^{*}\mathit{\text{dtari}}(t-\mathit{2})}\underset{(0.0000)}{-{0.0006}^{*}[\mathit{\text{df}}\_\mathit{\text{abm}}(t)+\mathit{\text{df}}\_\mathit{\text{abm}}(t-\mathit{2})+\mathit{\text{df}}\_\mathit{\text{abm}}(t-\mathit{3})]}\end{array}$

$\begin{array}{l}\underset{(0.0000)}{{0.6065}^{*}[\mathit{\text{drdom}}(t-3)-\mathit{\text{drext}}(t-3)-\mathit{\text{dfc}}(t-3)+\mathit{\text{drdom}}(t-4)-\mathit{\text{drext}}(t-4)-\mathit{\text{dfc}}(t-4)]}\end{array}$

$\begin{array}{l}\underset{(0.0000)}{-{2.0631}^{*}[\mathit{\text{drext}}(t)-\mathit{\text{drext}}(t-2)]}\underset{(0.0001)}{-{0.1022}^{*}[\mathit{\text{dfexi}}(t-1)-\mathit{\text{dfexi}}(t-3)]}\underset{.(0.0000)}{-{0.3074}^{*}\mathit{\text{RESID}}(-1)}\end{array}$

Adj R-Squared 0.78    Durbin Watson 2.06

Source: Estimations by the author.

^1/Figures in parentheses correspond to P-values.

View Table

Table 5.

Chile: Error-Correction Model for Real Exchange Rate (Based on Equation 2.3)^1/

Method: Ordinary least squares (1991.1-2003.3)

$\begin{array}{l}[\mathit{\text{dRER}}(t)]=\underset{(0.000)}{-{0.5608}^{*}\mathit{\text{dProdl}}}\underset{0.0000)}{-{0.7634}^{*}\mathit{\text{dprodl}}(t-\mathit{1})}\underset{(0.000)}{-{0.1992}^{*}[\mathit{\text{dtt}}(t)+\mathit{\text{dtt}}(t-\mathit{1})]}\end{array}$

$\begin{array}{l}\underset{(0.0000)}{-{0.1524}^{*}[\mathit{\text{dnfa}}(t)+\mathit{\text{dnfa}}(t-2)]}\underset{(0.0000)}{-{0.8136}^{*}[\mathit{\text{dgobmy}}+\mathit{\text{dgobmy}}(t-2)]}\underset{(0.000)}{-{1.5064}^{*}\mathit{\text{dgobmy}}(t-1)}\end{array}$

$\begin{array}{l}\underset{(0.0000)}{-{3.715}^{*}\mathit{\text{dtari}}(t-\mathit{2})}\underset{(0.0000)}{-{0.0006}^{*}[\mathit{\text{df}}\_\mathit{\text{abm}}(t)+\mathit{\text{df}}\_\mathit{\text{abm}}(t-\mathit{2})+\mathit{\text{df}}\_\mathit{\text{abm}}(t-\mathit{3})]}\end{array}$

$\begin{array}{l}\underset{(0.0000)}{{0.6065}^{*}[\mathit{\text{drdom}}(t-3)-\mathit{\text{drext}}(t-3)-\mathit{\text{dfc}}(t-3)+\mathit{\text{drdom}}(t-4)-\mathit{\text{drext}}(t-4)-\mathit{\text{dfc}}(t-4)]}\end{array}$

$\begin{array}{l}\underset{(0.0000)}{-{2.0631}^{*}[\mathit{\text{drext}}(t)-\mathit{\text{drext}}(t-2)]}\underset{(0.0001)}{-{0.1022}^{*}[\mathit{\text{dfexi}}(t-1)-\mathit{\text{dfexi}}(t-3)]}\underset{.(0.0000)}{-{0.3074}^{*}\mathit{\text{RESID}}(-1)}\end{array}$

Adj R-Squared 0.78    Durbin Watson 2.06

Source: Estimations by the author.

^1/Figures in parentheses correspond to P-values.

For the URR financing cost, the sustained change in the policy variable (fc) was defined as 100 basic points. An increase in the URR financing cost of that magnitude depreciated the real exchange rate by 0.6 percent after three quarters, but after ten quarters the total effect on the RER fell to 0.1 percent completely disappearing afterwards. The real domestic interest rate had an effect of the same magnitude, but in the opposite direction of the fc. In the case of the intervention in the foreign exchange market, the sustained policy change was defined as 10 percent of the intervention stock. Such effect depreciated the RER by 1 percent after one quarter, but then the effect vanishes very rapidly, and even it was reversed to an appreciation of 0.5 percent after three quarters.

Table 6.

Chile: Effects on RER of Sustained Changes in Key Variables

In percent

Source: Estimations by the author on the basis of Table 5. and Equation (2.3).

Table 6.

Chile: Effects on RER of Sustained Changes in Key Variables

In percent

	DFC	DFEXI	DGOBMY	DF_ABM	DTARI	DTT
delta	100	10	1	10	1	1
Periods
1	0.00	1.02	-0.45	-0.80	-1.45	-0.44
3	0.61	-0.53	-4.94	-2.04	-5.42	-0.38
6	0.49	-0.18	-4.80	-1.59	-4.96	-0.35
10	0.11	-0.04	-4.75	-1.42	-4.78	-0.33
20	0.00	0.00	-4.73	-1.37	-4.73	-0.33

Source: Estimations by the author on the basis of Table 5. and Equation (2.3).

View Table

Table 6.

Chile: Effects on RER of Sustained Changes in Key Variables

In percent

	DFC	DFEXI	DGOBMY	DF_ABM	DTARI	DTT
delta	100	10	1	10	1	1
Periods
1	0.00	1.02	-0.45	-0.80	-1.45	-0.44
3	0.61	-0.53	-4.94	-2.04	-5.42	-0.38
6	0.49	-0.18	-4.80	-1.59	-4.96	-0.35
10	0.11	-0.04	-4.75	-1.42	-4.78	-0.33
20	0.00	0.00	-4.73	-1.37	-4.73	-0.33

Source: Estimations by the author on the basis of Table 5. and Equation (2.3).

The financial policy variables have limited effects on the RER. The effect of the URR appears to have been enough to compensate for the wider domestic/foreign interest rate differential associated to the cyclical upswing, but completely insufficient to face the push of capital flows surge. On the other hand, the contribution of the foreign exchange intervention on the RER appears to be insignificant and short-lasting. Not only the effectiveness of the URR on the RER appears to be much stronger and long-lasting that of the intervention in the foreign exchange market, but also appears to be more efficient when considering the fiscal costs involved. While the URR yielded 0.2 to 0.3 percent of GDP in revenue to the Central Bank during its application, the fiscal cost of using foreign exchange intervention is not irrelevant. International reserves yield the libo real rate and to the extent the net purchases of foreign exchange had to be sterilized, they were financed at the domestic real rate (rdom). The net return of such an operation, finance international reserves positions domestically have been negative, being the historical average -400 basis points excluding the effect of changes in the real exchange rate⁴³. In recent periods the cost of domestically financing NIR has fallen reflecting lower sovereign spreads and a more accommodative domestic monetary policy.

The effect of the financial policy variables was insufficient to confront the external push of capital flows during the surge. The increase of capital flows to Latin-American countries, nkf_abm, that took place during the time of the surge of inflows resulted in a strong and long lasting appreciation of the RER. The additional US$10 billion in nkf_abm would have appreciated the RER by a cumulative 2 percent after 3 quarters, and by 1.4 percent over the long-term. Furthermore, a 1 percent increase in the terms of trade resulted in a real appreciation of 0.4 percent in the short-run and 0.3 percent in the longer term.

According to the estimation results, the policy variables that could be used to confront the RER appreciation are the import tariffs and government spending. Of course those policies have their own objectives, so that may not be available to manage the RER, but their potential contribution should not be dismissed and could in some cases be consider an additional benefit of pursuing certain structural policy directions. In this regard, the reduction of import tariffs during a period of capital flow surges add an element of RER stabilization to the welfare benefits associated to trade opening, a 1 percent reduction in the tariff index may depreciate the RER by 5 percent over the long term. Moreover, the use of a counter cyclical fiscal policy not also contributes to output stability, but in addition would help to avoid a rapidly growing government spending during the boom that may end up aggravating the RER appreciation resulting from the surge in capital flows. A reduction of 1 percent of GDP in government spending would depreciate the RER by almost 5 percent over the long term. No doubt both responses appear to be extremely strong and may represent an over estimation. However, they are indicative of the relative importance of structural policies over financial policies in determining the value of the real exchange rate in the medium- and long-term.