Effect of Official Dollarization on a Small Open Economy: Ecuador

Bedri Kamil Onur Tas1 and Selahattin Togay2

Abstract This paper investigates the effects of dollarization on the macroeconomic performance of a small open economy, Ecuador, using a time series perspective. More specifically, we investigate how dollarization effects inflation, GDP, inflation uncertainty and money-price relationship in Ecuador. There are four main findings of this study. First, inflation is lower after dollarization. Second, GDP growth is higher after dollarization when several other factors like rising oil prices and increase in the equity markets of emerging economies are controlled. Third, inflation uncertainty measured by inflation variance through GARCH is lower during dollarization. Finally, money supply is endogenous after dollarization and exogenous before dollarization. A new test is proposed and implemented to investigate endogeneity of money. As a result, we conclude that dollarization improved the macroeconomic performance of Ecuador and changed the money-price relationship in Ecuador. The results are robust to different arguments about the empirical methodology.

(Corresponding Author) TOBB University of Economics and Technology, Department of Economics, Email: onurtas@etu.edu.tr
2

Gazi University Department of Economics, Email: stogay@gazi.edu.tr

1 Introduction The fact that many emerging countries faced currency crises with devastating negative economic effects caused an intense debate on exchange rate policies for emerging countries. Many economists proposed hard pegs (Summers (2000) and Fischer (2001)). Dollarization, in the sense that the country should abandon its national currency and adopt an advanced nation's currency as legal tender (US Dollar for Ecuador), has been implemented by many countries 3. Ecuador undertook official dollarization in March 20004 by dropping its own currency, the sucre, and adopting the US dollar. This paper empirically investigates the effects of dollarization on the macroeconomic performance of Ecuador. As mentioned by Jameson (2003a), Ecuador is the longest-lasting of the recent dollarizers and has the most extreme contemporary dollarization program. This makes Ecuador an excellent candidate to investigate the effects of dollarization on emerging economies. As presented in table I, we find that dollarization has significant positive effects on the economic performance of Ecuador 5. Table I Summary of the Empirical Results Effect of Dollarization CPI Inflation GDP Growth Inflation Uncertainty Lower after dollarization Higher after dollarization Lower after dollarization

Money-Price Relationship Money becomes endogenous after dollarization

The theoretical analyses of Calvo (2001) and Minda (2005) present that there are pros and cons of dollarization. Thus, extensive empirical analysis is required to investigate whether dollarization achieves what it promises: improvements in macroeconomic conditions through

3 4

Edwards and Magendzo (2003) lists 13 independent and 15 non-independent countries for the 1970-1998 period. After a serious economic crisis in 1998-1999 period, President Jamil Mahuad announced the dollarization program on January 9 2000. 5 During the current financial crisis in 2008, Ecuador has significant financial problems. The current financial crisis is out of the scope of this paper and available macroeconomic data for Ecuador ends at 2007.

financial and exchange rate stability. Alesina and Barro (2001, p.382), have argued that adopting another nations currency eliminates the inflation bias problem of discretionary monetary policy. Dornbusch (2001) indicates that countries that give up their currency will tend to grow faster than non-dollarized countries6. Starting from these theoretical arguments, we investigate the effects of dollarization on the macroeconomic performance of Ecuador using a time series perspective. More specifically, we investigate how dollarization effects inflation, GDP growth, inflation uncertainty and money-price relationship (endogeneity of money). There are four main results of this study. First, inflation is lower after dollarization. Second, GDP growth is higher after dollarization. These results persist after controlling other factors that might cause these improvements besides dollarization: oil production growth, rise in oil prices, increase in the MSCI emerging market index, improvement in institutional quality. Third, inflation uncertainty measured by inflation variance through GARCH is lower during dollarization. Finally, money supply is endogenous after dollarization and exogenous before dollarization. Quispe-Agnoli and Whisler (2006) indicate that the expected benefits of full dollarization include the elimination of exchange rate risk, contributing to the decline of the country risk premium and interest rates, as well as the reduction of the inflation rate and inflationary expectations. Some initial conditions could be relevant in the decision to implement official dollarization. Minda (2005) and Edwards and Magendzo (2006) observe that small countries with close trade or financial ties to the United States could favor official dollarization, as Panama did in 1904. Ecuador, El Salvador, and Panama, the largest countries that have implemented official dollarization, are still relatively small and are very open to U.S. trade and finance, with an average gross domestic product (GDP) of $11 billion (in 2000 dollars) and an average population of 7 million in 2004.

As mentioned in Edwards and Magendzo (2003), this growth effect is supposed to take place through two channels: (a) dollarization will mean lower interest rates, higher investment, and faster growth (Dornbusch, 2001); and (b), by eliminating exchange rate volatility, dollarization is supposed to encourage international trade and this, in turn, will result in faster growth. (Rose (2000), and Rose and van Wincoop (2001))

The empirical analysis about the effects of dollarization is limited. Previous studies in the literature are conducted using annual data. For example, Edwards and Magendzo (2006) examines the effect of dollarization using a yearly panel of 169 countries that covers 1970 through 1998. They find that GDP growth is not statistically different in dollarized and in non-dollarized ones. Since the dataset of Edwards and Magendzo (2006) ends in 1998, it does not identify Ecuador as dollarized. The dollarized countries in their dataset is mostly non-independent countries and independent countries are small countries. In their dataset 20 out of 169 countries are identified as dollarized. In our study, we use monthly and quarterly data and analyze the effects of dollarization on a relatively large independent country, Ecuador, using a time series perspective. The empirical analysis of dollarization is less complicated compared to the analysis of other policy changes because dollarization does not suffer from the epidemic case of fear of floating as identified by Calvo and Reinhart (2002). In other words, it is not possible for the countries to announce a policy and do not follow that policy. Calvo and Reinhart (2002) find that countries that say they follow their exchange rate to float mostly do not. In the case of dollarization, Ecuador dropped sucre and adopted the US dollar in March 2000. Since the sucre is not valid after March 2000, we can safely claim that dollarization is implemented exactly at this date. In other exchange rate regimes like floating and monetary policy regimes like inflation targeting the exact date of implementation and the degree of adoption might not be determined exactly which complicates the empirical analysis. The economic crisis in 1998-1999 period that led to dollarization in Ecuador initiated many empirical papers about macroeconomic performance of Ecuador. Nazmi (2001), Jacome (2004), Beckerman (2001), Beckerman (2002) and Martinez (2006) analyzed the process that led to this economic crisis. They argue that mainly institutional weaknesses, rigidities in public finances, and high financial dollarization amplified the financial crisis in Ecuador. Beckerman and Douglas (2002) and Solimano (2002) examines the pros and cons of dollarization in Ecuador. Jameson (2004) argues

that orthodox perspectives are inadequate to explain dollarization decision of Ecuador and uses a post-keynesian institutionalism explanatory framework. Dean (2003) compares Canada and Ecuador from six different aspects: dangers of liability dollarization, currency and default risk premiums on interest rates, lower predictability of domestic money supply, stronger monetary shocks, currency substitution irreversibility and impotence of exchange rate policy. He conducts a descriptive analysis and concludes that dollarization is suitable for Ecuador but not Canada. Hanke (2003) argues that dollarization is suitable for Ecuador but the following refoms should be conducted for successful implementation of dollarization: financial integration with international financial system, fiscal transparency and control, tax simplification, supermajority voting, deregulation and privatization. On the other side of the debate, some studies argue that dollarization is not favorable for Ecuador. Jameson (2003a and 2003b) argues that dollarization is not suitable for Ecuador and will lead Ecuador to crisis since it does not solve any fundamental economic problems about the structure of the economy. Jameson (2003a) examines how a process of de-dollarization might be implemented. Jameson (2003b) analyzes the macroeconomic performance of Ecuador using yearly data from 1997 till 2002. This paper argues that macroeconomic performance of Ecuador has improved during 2000-2002 period. Dollarization played a role in this improvement by encouraging both private and capital flows. From a political-economic standpoint, dollarization has succeeded in providing access to international dollar resources. Jameson (2003b) indicates that although dollarization improved the economic performance of Ecuador, the fundamental structural economic problems like political instability and disappearance of independent monetary policy remain. These problems leave Ecuador susceptible to crisis. These studies use only yearly data because of data limitations and do not conduct a time series investigation of macroeconomic performance of Ecuador.

Our analysis differs from other related studies in the literature in several aspects. First of all, this study is the first extensive time-series analysis of the effects of dollarization on the macroeconomic performance of Ecuador. We use high frequency data to investigate effects of dollarization. Earlier studies could not conduct these studies because of data limitations7,8. To verify that our results are robust and correctly gauge the effect of dollarization, we use control variables like oil price, institutional quality and MSCI Emerging Markets Index. Also, the alternative QuandtAndrews Statistic methodology makes sure that our empirical results correctly measure the effects of dollarization and not any other change that occurred before dollarization. Second, to best of our knowledge, this is the first study that analyzes the impact of dollarization on inflation uncertainty using a GARCH model. Third, we investigate the impact of dollarization on money-price relationship in Ecuador which has not been done in the literature before. Fourth, we propose and implement a test of endogeneity of money in the context of GMM and instrumental variables, C statistic, which has not been used in the literature before to investigate money-price relationship. We use this methodology to investigate the endogeneity of money in Ecuador before and after dollarization. Compared to previous methods9, the C statistic method directly examines whether money is endogenously determined or not. To sum up, using several time-series methods, we contribute to better understanding of the macroeconomic effects of dollarization using monthly and quarterly data. The rest of the paper is organized as follows: In section 2, we describe the conditions that led to dollarization. Section 3 explains the data and methodology used in this study. In section 4, we present the results about the macroeconomic performance of Ecuador namely inflation and output

Those studies did not have enough monthly and yearly data points after dollarization to conduct time-series analysis. Alesina and Barro (2001) indicates that data availability limits empirical research about monetary unification. 8 Soto (2009) investigates the impact of dollarization on the labor market in Ecuador. Specifically, he investigates the following research questions: What are the causes of the slow growth in employment after dollarization in Ecuador? and How does employment respond to shocks in production, wages, and interest rates?. 9 Previous studies mostly implement Granger causality methodology which only approximately tests endogeneity of money.

growth. In section 5, we analyze the effects of dollarization on inflation uncertainty. In section 6, we examine the impact of dollarization on money-price relationship in Ecuador. Finally, section 6 presents the concluding remarks and policy implications of the results. 2 Ecuador Before Dollarization Although Ecuador met few of the criteria for entering an optimal currency area in 2000 (Panizza, Stein and Talvi 2003) it adopted dollarization. It was a small, open economy and its central bank had low credibility which made it suitable for dollarization. The deterioration in economic performance after 1997 forced Ecuador to adopt dollarization. The process that led to the economic crisis in 19981999 period can be summarized by the following factors: a) Political Instability: The conflicts between the mountainous and coastal regions caused instability and decreased the ability of the country to implement structural reforms. This is caused by the fact that the monetary and fiscal policy actions could have opposite effects on these two regions. Jacome (2004) indicates that the depreciation of the real exchange rate had favorable effects on the coast whereas has adverse effects on the mountainous regions. An important indicator of the instability is presented by Beckerman (2002); there were 5 different cabinets and 7 different finance ministers during the 1995-1998 period. b) Natural Disasters: Ecuadorian economy suffered significantly because of several natural disasters like earthquakes, volcanic eruptions and storms. These had adverse effects on revenues from tourism and energy production. For example, El Nino storm in 1998 had devastating impact on production and infrastructure. c) High Dependence on Export (Oil) Revenues: The public revenues and export revenues of Ecuador depend highly on oil revenues. Thus, significant movements in oil prices causes distortions on public finance. For example, the decrease of oil prices to $9 per barrel had adverse effects on export

revenues and public finance. Table II presents two most important items of export revenues, oil and banana. The structure of production significantly changed after dollarization. One of the most significant feature of the Ecuadorian economy is the high share of oil production in GDP. This feature leaves Ecuador susceptible to oil price shocks. (Table II about here.) d) High Degree of Financial Dollarization: After 1992 the Central Bank of Ecuador started a stabilization program based on pegged exchange rate. Between 1994 and 1998 the credibility of this system deteriorated and high levels of inflation and exchange rate uncertainty caused a significant increase in financial dollarization. In that period the saving accounts in dollars surpassed the national foreign reserves10. The fourth factor is eliminated after dollarization but other three factors currently exist in Ecuador. Also, the production structure of Ecuador changed significantly. As displayed in Table III, the portion of mining and oil production significantly increased. (Table III about here.) Additionally, the structure of investment and savings changed significantly after dollarization as presented in figure I below. Figure I indicates that dollarization had positive impacts on the structure of savings and investment. Investment and national savings increased whereas foreign savings significantly diminished. This represents the increased trust in the Ecuadorian economy.

10

The ratio of saving accounts in dollars to aggregate savings raised from %13.3 in 1990 to %53.7 in 1999. Also, The ratio of loans in dollars to aggregate loans raised from %1.5 in 1990 to %66.5 in 1999. (Beckerman, 2002)

Saving and Investment 30 25 20 15 10

Unemployment Rate
16

14

12

10
5 0 -5

6
-10 99 00 01 02 03 04 05 06 07 08

4
National Savings/GDP Investment/GDP Foreign Savings/GDP

88

90

92

94

96

98

00

02

04

06

08

Figure I: Investment and savings and the unemployment rate in Ecuador Figure I also displays the unemployment rate in Ecuador. As indicated by Soto (2009), unemployment remained stubbornly high after dollarization. In this section we briefly summarized the Ecuadorian economy. In the following sections, we will investigate the effects of dollarization on the economic structure of Ecuador in detail. 3 Data and Methodology 3.1 Data: We use monthly and quarterly data for 1990-2007. The source of the data is IFS, the FRED database of the St. Louis Fed, Reuters Ecowin and Polity IV databases. The variables used in this study are: CPI inflation, real GDP, GDP growth, oil production growth, money supply, spot oil price, MSCI Emerging Markets Index growth and institutional quality measure of the Polity index. The data is explained in detail in the Appendix. Oil production growth, spot oil price, MSCI index growth and institutional quality are used as control variables in regressions that measure the effect of dollarization on real GDP and real GDP growth. 3.2 Methodology

This study investigates three important aspects of the Ecuadorian economy: macroeconomic performance (inflation and GDP growth), inflation uncertainty and money-price relationship. 3.2.1 Macroeconomic Performance Artificial regressions that are designed to measure the change before and after dollarization (March 2000) are used. We implement an OLS methodology with dollarization dummy to measure the exact effects of dollarization on macroeconomic performance variables of Ecuador. The dollarization dummy takes the value of 1 after March 2000 (2nd quarter of 2000 for quarterly data) and zero before March 2000. The same methodology is implemented by Mishkin and Schmidt-Hebbel (2007) to measure effects of inflation targeting on macroeconomic performance. The coefficient of the dollarization dummy variable gauges the exact effect of dollarization. We use several control variables to make sure that changes in GDP growth is caused by dollarization and not by other variables that can affect GDP. As a result, the following regression specification is estimated: Yt = 0 + 1Dt +2Ct +t (1) Where Yt denotes the macroeconomic variables. Dt is the dollarization dummy variable which takes the value 0 before dollarization (March 2000) and 1 after dollarization. Ct denotes the control variables other than dollarization that might affect the macroeconomic performance of Ecuador. One can argue that the changes in the macroeconomic performance of Ecuador might be caused by the improvement in the government institutions after the stabilization policy initiated after the 1998-1999 economic crisis. We use the Polity IV institutional quality measure as employed by Acemoglu and Johnson (2005) as a control variable11. This controls for the effect of changes in

11

Figure A.1 in the Appendix displays the institutional quality measure, constraint on executive, for 1990-2008 period. Interestingly, the institutional quality of Ecuador did not improve despite the stabilization policy actions taken after 1999.

institutional quality and helps us correctly measure the impact of dollarization on the macroeconomic performance of Ecuador12. Another important feature that we consider while implementing these empirical methods is the robustness of our results. Thus, we designed an alternative econometric specification and implemented several stability tests to test for changes in macroeconomic performance of Ecuador. Since 1 in equation 1 measures the change in the conditional mean of the macroeconomic variable (Yt), testing the stability of the following regression equation is equivalent of testing the changes in the conditional mean of those variables. Yt = 0 + 1Ct +ut (2)

Ones is a vector of ones. By testing the stability of 0 in equation 2 using Chow and QuandtAndrews breakpoint tests we can test the impact of dollarization on the conditional mean of these macroeconomic variables. The Chow test fits the equation separately for each subsample and examines whether there are significant differences in the estimated equations. To check for robustness of our results, we also implement the Quandt-Andrews Breakpoint Test which tests for one or more unknown structural breakpoints in the sample for a specified equation. The idea behind the Quandt-Andrews test is that a single Chow Breakpoint Test is performed at every observation between two dates, or observations13. The Quandt-Andrews test also provides us the most likely

12

Acemoglu and Johnson (2005) use the constraint on executive variable in Polity IV as a measure of institutional quality. They explain the variable as the following: A seven category scale, from 1 to 7, with a higher score indicating more constraint. Score of 1 indicates unlimited authority; score of 3 indicates slight to moderate limitations; score of 5 indicates substantial limitations; score of 7 indicates executive parity or subordination. Scores of 2, 4, and 6 indicate intermediate values. 13 The Quandt-Andrews test is essential to check for the robustness of our results. The date of the structural change is not imposed in the Quandt-Andrews test and the test searches for one or more unknown structural breakpoints in the sample. As a result, the test is immune to the argument that structural break tests have low power because the break date is chosen based on the knowledge of events that some change occurred. By using stability tests in the alternative regression specification we validate that our results are robust to these arguments.

dates of structural change. The Quandt-Andrews is selected because the structural break point is determined by the test and not imposed on the test. Thus, this test is immune to the argument that a known date of structural change is used and this might cause a bias. 3.2.2 Inflation Uncertainty As in Fountas (2001), we implement a GARCH (1,1) model 14 to investigate the impact of dollarization on inflation uncertainty. Consider an ARMA model of inflation, t, with time-varying conditional variance:

t is the information set available at time t. As in Andersen et al. (2003), we use the dummy variable as an explanatory variable in the variance equation and investigate effect of dollarization on inflation uncertainty. Andersen et al. (2003) use announcement dummies in the variance equation of exchange rates to measure the effects of macroeconomic announcements on exchange rate volatility. The coefficient of the dollarization dummy, t, shows the effect of dollarization on inflation uncertainty in Ecuador. 3.2.3 Money-Price Relationship In this study, we also make a methodological contribution by proposing an alternative test of endogeneity of money. Many studies like Pinga and Nelson (2001) and Ozmen (2003) use Granger causality or cointegration tests to investigate the relationship between money supply and price level or inflation. But these methods test causality not exogeneity of a variable. As argued in Togay and
14

Akaike and Schwarz information criterions pick GARCH (1,1). Table IX and section C of the Appendix display AIC and SBC values for alternative GARCH specifications.

Kose (2008), the existing causality studies do not make a clear distinction between exogeneity and causality. Thus, the presence of causal relationship from price to money supply is neither necessary nor a sufficient condition for testing endogenous money hypothesis. As recommended by Baum et al. (2007), we implement a test of overidentifying restrictions to test endogeneity of money15. As shown in Hayashi (2000), a regressor is endogenous if it is not predetermined (i.e., not orthogonal to the error term), that is, if it does not satisfy the orthogonality condition. Following this argument, we test whether money is endogenous using the C statistic (also known as a GMM distance or difference-in-Sargan statistic). Under the null hypothesis that the specified endogenous regressors can actually be treated as exogenous, the test statistic is distributed as chi-squared with degrees of freedom equal to the number of regressors tested. The endogeneity test is, like the C statistic, defined as the difference of two Sargan-Hansen statistics: one for the equation with the smaller set of instruments, where the suspect regressor(s) are treated as endogenous, and one for the equation with the larger set of instruments, where the suspect regressors are treated as exogenous. Also like the C statistic, the estimated covariance matrix used guarantees a nonnegative test statistic. Under conditional homoskedasticity, this endogeneity test statistic is numerically equal to a Hausman test statistic; see Hayashi (2000, 233-234). We conduct the test for the pre and postdollarization periods and analyze whether money supply is endogenous in these periods or not. 4 Macroeconomic Performance of Ecuador and Dollarization This section investigates the effects of dollarization on inflation and GDP growth. Table IV presents the summary statistics of monthly inflation for dollarization and non-dollarization periods and the summary statistics for Real GDP and Real GDP growth using quarterly data. (Table IV about here.)
15

This test is also known variously as the Sargan statistic, Hansen J statistic, Sargan-Hansen J test. The value J of the GMM objective function evaluated at the efficient GMM estimator \beta_{EGMM} is distributed as \chi^{2} with (L K) degrees of freedom under the null hypothesis that the full set of orthogonality conditions are valid. (Baum et. al. (2007))

Following the methodology in Mishkin and Schmidt-Hebbel (2007), tables V and VI measure the change in inflation and GDP growth before dollarization and after dollarization. In table V of monthly data, the coefficient of the dollarization dummy variable is negative and significant for the inflation equation. Analysis of quarterly data in table VI indicates that the coefficient of dollarization dummy variable is negative and significant for the inflation equation. For the analysis of Real GDP and Real GDP growth, table VI displays alternative regression specifications with different sets of control variables. Table VI shows that the dollarization dummy is significant with positive coefficient for the Real GDP and Real GDP growth equations. Thus, we conclude that inflation is significantly lower in Ecuador after dollarization and Real GDP and Real GDP growth is significantly higher after dollarization. These results are robust to different regression specifications with alternative control variables. (Tables V and VI about here.) Tables V and VI present that after dollarization the macroeconomic conditions of Ecuador improved significantly. Different regression specifications control for the rise in the oil prices, oil production growth, the improvement in institutional quality and trend of growth in the emerging economies. Thus, independent of these factors, dollarization helped Ecuador to achieve better macroeconomic conditions. 4.1 Robustness There are several arguments that can be made about the empirical results of the paper presented above. First, one can argue that the regression methodology is responsible for the favorable results about dollarization. To analyze whether the results of tables V and VI are robust to selection of empirical methodology we implemented two alternative methods: Chow and Quandt-Andrews breakpoint tests. Table VII examines the robustness of the results presented in table V. Both the Chow and Quandt-Andrews tests reject the null hypothesis that there are no structural breaks in the

conditional mean of monthly inflation. Also, Quandt-Andrews test identifies February and March 2001 as the date of structural break which is the date of dollarization. (Table VII about here.) Table VIII presents the alternative Chow and Quandt-Andrews test statistics for all the regression specifications of table VI. For all regression specifications, the Chow test concludes that the conditional mean of the variables are different before and after the second quarter of 2000. The Quandt-Andrews test concludes that there is a structural break in the dataset and all of the break dates coincide with dollarization16. As a result, using alternative methods, we conclude that there has been a structural change in macroeconomic variables of Ecuador. In other words, the results of the paper are robust to alternative empirical methods. (Table VIII about here.) Second, it can be claimed that macroeconomic factors other than dollarization might have improved after March 2000 and this might have affected our results. As presented in tables II and III, oil production is an incremental component of the Ecuadorian economy. Thus, increases in oil prices and oil production have significant positive effects. So, we included oil prices and oil production as a control variables. Also, after the countries started to recover from the global crisis in 1999, all emerging countries experienced positive economic growth. To control for this effect we used MSCI Emerging Markets Index. After the crisis, important structural changes are undertaken in Ecuador as a part of the recovery program. To make sure that these improvements do not affect our empirical results we include institutional quality as a control variable in our regressions. We believe that these variables represent most of the important economic changes for the Ecuadorian economy other than dollarization. By using these variables as control variables, we confirm that the empirical methodology used in the paper correctly measure the direct effects of dollarization.
16

The test either exactly identifies 2nd quarter of 2000 as the break date or identifies a date after 2nd quarter of 2000.

Third, it is also possible that improvements that occurred before dollarization already had positive effects and the empirical results depict these effects associated with previous changes. The Quandt-Andrews test can be used to examine the robustness of out results to this argument. The test searches for an unknown structural breakpoint in the data17. Tables VII and VIII presents that all of the structural break dates identified by the test are either exactly first quarter of 2000 (March 2000 for monthly analysis) or after 2000 and very close to the date of dollarization adoption. All of the dates identified by the test are after dollarization meaning that the results of table V and VI are not biased by previous events before dollarization. Finally, empirical analysis of structural change is susceptible to the critique of Hansen (1992) who claims that ... the date of structural change may be selected by appeal to events known a priori. ... it is essential that the researcher can argue that the events are selected exogenously. The robustness check conducted using the Quandt-Andrews test also provides answers for this critique. Since the test first identifies the structural break date and then tests whether there is a break at that date the break date is exogenous. In other words, we did not impose a known date but the test itself picked that date. Thus, tables VII and VIII presents that the results of table V and VI are robust to the critique of Hansen. The Quandt-Andrews test identifies dates that exactly match the date of dollarization or dates that are 2 or 3 quarters ahead. To sum up, robustness analyses of the main results of the paper indicate that the results are not biased and robust to alternative specifications. 5 Inflation Uncertainty and Dollarization It has been found that inflation uncertainty significantly affects macroeconomic performance like inflation and GDP growth. Cukierman and Meltzer (1986) construct a game-theoretic model and show that higher inflation uncertainty will raise average inflation rate. Using GARCH methodology and Granger causality tests, Apergis (2004) provides empirical support that inflation uncertainty

17

The test is explained in detail in Hansen (2001) and Perron (2006).

increases inflation in the G7 countries. Grier and Perry (1998) find that a rise in inflation uncertainty significantly affects inflation in more than half of the countries they analyze. Elder (2004) theoretically and empirically investigates the effects of inflation uncertainty on real economic activity. He finds that a shock to inflation uncertainty decreases output growth. Friedman (1977) indicates that uncertainty about future inflation distorts the efficient allocation of resources and this leads to lower output. Stockman (1981) showed that anticipated inflation reduces the demand for real balances, implying that the demand for capital and output growth decreases. His results find empirical support in Zhang (2000). As a result, lower inflation uncertainty is crucial for economic well-being. In this section, we investigate whether dollarization helped Ecuador to achieve lower inflation uncertainty. As mentioned in Fountas (2001), autoregressive conditional heteroskedasticity (ARCH) and generalized ARCH (GARCH) approaches can be employed to proxy uncertainty. Fountas (2001) uses annual data and implements a GARCH (1,1) model to investigate inflation uncertainty in UK. Daal et al. (2005) use monthly inflation rates based on log differences of CPI. They implement PGARCH methodology to estimate inflation uncertainty and use Granger causality tests to investigate the relationship between inflation and inflation uncertainty for both developed and emerging countries. In this section, we implement the GARCH methodology described in section 2 to investigate the effect of dollarization on inflation uncertainty. First, we conduct an ARCH LM test and unit root tests to verify whether the necessary conditions for the implementation of the GARCH methodology are satisfied. The ARCH LM test shows the presence of ARCH effects for inflation18. Unit root tests reject the null hypothesis of a unit root for inflation at 1 percent. Table in Appendix B presents the test statistics which show that inflation does not have unit root.

18

The F statistic is 51.6 and the Obs*R-squared statistic is 41.3. Both statistics have p values of 0.00. Thus, we reject the the null hypothesis that there is no ARCH.

.14 Dollarization .12 .10

Monthly Inflation

.08 .06 .04 .02 .00 -.02 90 92 94 96 98 Year 00 02 04 06

Figure II: Variability of Inflation Figure II displays monthly inflation of Ecuador. It can be seen that variability of inflation is lower after dollarization (March 2000). Table IX displays the formal test of the impact of dollarization of inflation uncertainty. It shows different specifications results of GARCH(1,1) for inflation. We estimated various ARCH and GARCH models. The Akaike Information Criterion (AIC) and Schwarz Bayesian Criterion (SBC) of these alternative models are displayed in section C of the Appendix 19. GARCH(1,1) specification for the variance of inflation in Ecuador is selected by the SBC. Several studies of inflation uncertainty like Apergis (2004), Fountas (2001) and Grier and Perry (1998) implement GARCH(1,1) specifications. As in Andersen et al. (2003) and Fountas (2001), we investigate changes in the variance of inflation by using the variable of interest (dollarization dummy) as an explanatory variable in the variance equation. In all of the regression specifications of table IX, the coefficient of the dollarization dummy is significant and negative. This result concludes that inflation uncertainty (variance) is significantly lower after dollarization.

19

In all of the alternative models the dollarization dummy variable in the variance equation is significant with a negative coefficient.

(Table IX about here.) Figure III displays a graphical presentation of the results of Table IX. The conditional standard deviation of inflation is much lower after dollarization. As a result, the results of this section infer that dollarization helped Ecuador to achieve lower inflation uncertainty.

.10

.08

.06

.04

.02

.00 90 92 94 96 98 00 02 04 06

Conditional standard deviation

Figure III: Conditional Variance of Inflation Before and After Dollarization 6 Money-Price Relationship in Ecuador Is inflation a monetary phenomenon? This question has been extensively analyzed theoretically and empirically. Several studies in the literature investigate money-price relationship empirically: Belrs and Jones (1993) for Algeria; Pradhan and Subramanian (1998) for India; Sun and Ma (2004) for China and Pinga and Nelson (2001) for 26 countries. Vymyatnina (2006) conducts the analysis from a post-Keynesian perspective for Russia. To the best of our knowledge, there are no studies that investigates money-price relationship for Ecuador and this study is the first study that examines effect of dollarization on money-price relationship. Dollarization is expected to effect money-price relationship because in the full dollarization system a country abandons its monetary and exchange rate policies. The amount of money is determined internally (endogenously) according to balance of payments. In other words, as Schuler

(2005) argues the supply demand dynamics in the exchange rate market determines the money supply. This argument indicates that dollarization endogenizes money supply. In that context, we expect and find money supply to be endogenous after dollarization (after March 2000). Before dollarization, (1990-2000 period) there has been different exchange rate regimes in Ecuador. During pre-dollarization period, we expect and find money supply to be exogenous in Ecuador. This is partly caused by the fact that most of the money supply is determined by financing government expenditures and financing troubled banks20. In this section, we empirically investigate the validity of this argument. We conducted the endogeneity tests explained in section 3.2.3 using different measures of money supply: M1, M2 and reserve money. Different specifications and estimations techniques are implemented (2SLS, GMM). The lagged values of dependent and independent variables are used as instruments. The underidentification tests of all regressions have p-values of 0.00 indicating that instruments are significantly related with the endogenous variable. The Sargan and Hansen J statistics of all the regressions have p-values between 0.38 and 0.99. Thus, for all of the instrumental variable regression specifications, we accept the null hypothesis that the instruments are valid instruments, i.e., uncorrelated with the error term21. Table X displays the GMM test of exogeneity of monetary aggregates before and after dollarization. Table X shows that the null hypothesis that money is exogenous is accepted before dollarization and rejected after dollarization. Thus, we conclude that money is endogenous after dollarization for all measures of money supply. (Table X about here.)

20 21

For the 1995-1999 period the average cash deficit of Ecuador is 748.44 million dollars. (Source: IFS yearly statistics) The regression outputs of all the IV regressions are displayed in the previous versions of the paper. Since the results of the endogeneity tests are of interest, we do not show the regression results in this version to save space. The regression outputs are available from the authors upon request.

Finally, using the endogeneity tests conducted for M1, M2 and reserve money, we deduce that money is endogenous after dollarization and exogenous before dollarization in Ecuador. As explained above, dollarization endogenizes money. The results presented in table X provides empirical evidence for the arguments of Schuler (2005). This section expresses another significant impact of dollarization that has not been described in detail in the literature before. 7 Conclusion and Policy Implications This paper contributes to the literature by investigating the effects of dollarization using a time series perspective. More specifically, we investigate how dollarization effects inflation, GDP growth, inflation uncertainty and money-price relationship (endogeneity of money) in Ecuador. There are four main results of this study. First, inflation is lower after dollarization. Second, GDP growth is higher after dollarization. Third, inflation uncertainty measured by inflation variance through GARCH is lower during dollarization. Finally, money supply is endogenous after dollarization and exogenous before dollarization. As explained in Section 6, this is caused by the fact that the amount of money is determined internally (endogenously) according to balance of payments . The results of this paper has many policy implications. We provide evidence that dollarization improves macroeconomic performance of Ecuador by lowering inflation, increasing GDP growth and lowering inflation uncertainty. Endogeneity of money in Ecuador after dollarization raises some arguments about economic policy in Ecuador. Production in Ecuador highly depends on oil production. As displayed in table III, the percent of oil production in GDP is significantly higher for the 2000-2006 period. Thus, during the dollarization period the Ecuadorian economy depends more on oil production. Rising oil prices help Ecuadorian government financing. In that sense, endogenous money supply and inability of Ecuador to use money supply as a policy tool do not cause any serious economic problems when Ecuador is earning high oil revenues.

Unstable oil prices and very low oil price (9 dollars per barrel in 1999) were among the reasons of economic crisis in 1999. Besides, as mentioned in Jameson (2003a and 2003b) the instable political conditions and fundamental structural problems still remain in Ecuador. These inadequate economic conditions for dollarization might lead Ecuador into another crisis unless economic reforms are made. To sum up, macroeconomic conditions significantly improved after dollarization but instable oil prices might be followed by a decline in oil prices which might lead to economic crisis. To avoid crisis, Ecuadorian authorities should rapidly conduct structural economic reforms which are compatible with dollarization. These reforms are required to maintain the improved macroeconomic conditions achieved by dollarization. APPENDIX A Data CPI Inflation: Inflation is calculated as the log difference of CPI. The source is the IFS. Real GDP: Level of GDP at 2000 prices in US Dollars is used as the Real GDP. It is from the IFS. GDP Growth: GDP growth is calculated as the log difference of Real GDP. Oil Production Growth: Log difference of quarterly total oil production which is taken from the IFS database is used. Money Supply: Reserve money, M1 and M2 are used for the analysis of money supply. The variables are from the IFS. Spot Oil Price: Crude oil price of the West Texas Intermediate. (FRED) MSCI Emerging Markets Index Growth: Index growth is calculated as the log difference of the index created by Morgan Stanley Capital International (MSCI). MSCI is designed to measure equity market performance in global emerging markets. (Reuters Ecowin) Institutional Quality: The constraint on executive variable in Polity IV index is used as a measure of institutional quality as employed by Acemoglu and Johnson (2005). They explain the variable as the following: A seven category scale, from 1 to 7, with a higher score indicating more constraint. Score of 1 indicates unlimited authority; score of 3 indicates slight to moderate limitations; score of 5 indicates substantial limitations; score of 7 indicates executive parity or subordination. Scores of 2, 4, and 6 indicate intermediate values.

The Polity IV dataset is available at http://www.systemicpeace.org/inscr/inscr.htm. Figure A.1 below presents the constraint on executive index of Ecuador for 1990-2008 period.

Figure A.1: Constraint on executive index of Ecuador B Unit Root Tests of Monthly CPI Inflation Test Statistic ADF = -6.7 Phillips-Perron = -6.7 P-Value 0.00 0.00

C Alternative Empirical Models for Inflation Uncertainty Akaike Info Criterion Schwarz Bayesian Criterion Akaike Info Criterion Schwarz Bayesian Criterion ARCH(1) ARCH(2) ARCH(3) ARCH(4) GARCH(1,1) GARCH(1,2) GARCH(1,3) GARCH(2,1) GARCH(2,2) GARCH(2,3) GARCH(3,1) GARCH(3,2) GARCH(3,3) -4.96 -5.04 -5.05 -5.05 -5.14 -5.13 -5.15 -5.14 -5.13 -5.16 -5.13 -5.12 -5.16 -4.93 -4.99 -4.99 -4.97 -5.09 -5.07 -5.08 -5.07 -5.05 -5.07 -5.05 -5.02 -5.05

References Acemoglu D , Johnson S (2005) Unbundling Institutions. Journal of Political Economy 113(5):949995 Alesina B, Barro RJ (2001) Dollarization. The American Economic Review, 91(2):381-385 Andersen T G, Bollerslev T et al (2003) Micro Efects of Macro Announcements: Real-Time Price Discovery In Foreign Exchange. American Economic Review 93(1):38-62 Apergis N (2004) Inflation, output growth, volatility and causality: evidence from panel data and the G7 countries. Economics Letters 83 (24):185-191 Baum C F , Scharer ME , Stillman S (2007) Enhanced routines for instrumental variables/GMM estimation and testing. Stata Journal 7(4):465-506 Beckerman P (2002) Longer-term origins of Ecuador's Predollarization Crisis. In: Beckerman P and Solimano A (edts) Crisis and Dollarization in Ecuador: Stability, Growth, and Social Equity , The World Bank, pp17-81

Beckerman P , Douglas HC (2002) Longer-term origins of Ecuador's Predollarization crisis. In: Beckerman P and Solimano A (edts) Crisis and Dollarization in Ecuador: Stability, Growth, and Social Equity , The World Bank, 81-127 Belrs A, Jones T (1993) Money inflation and causality in a financially repressed Economy: Algeria1970-1988. Applied Economics 25:473-480 Calvo G A (2001) Capital markets and the exchange rate with special reference to the dollarization debate in Latin America. Journal of Money, Credit, and Banking 33 (2):312-334 Calvo G A and C M Reinhart (2002) Fear of floating. Quarterly Journal of Economics,117 (2):379408 Cukierman A, Meltzer AH (1986) A Theory of Ambiguity, Credibility, and Inflation under Discretion and Asymmetric Information. Econometrica 54:1099-1128 Daal E, A.Naka , B. Sanchez (2005) Re-examining inflation and inflation uncertainty in developed and emerging countries, Economics Letters 89:180-186 Dean JW (2003) Why Ecuador was ripe for dollarization, but Canada is not. In: Rochon PL ,Seccareccia M (edts) Dollarizaton Lesson from Europe and Americas, Routledge. pp.165-175 Dornbusch R. (2001) Fewer monies, better monies. American Economic Review 91(2): 238-242 Edwards S, Magendzo.I (2003) Dollarization and Economic Performance: What do we really know?. International Journal of Finance and Economics 8: 351-363 Edwards S, Magendzo I (2006) Strict dollarization and economic performance: An empirical investigation. Journal of Money,Credit and Banking 38:269-82 Elder J (2004) Another Perspective on the Effects of Inflation Uncertainty. Journal of Money, Credit and Banking, 36:911-928 Fischer S ( 2001) Exchange Rate Regimes: Is the Bipolar View Correct? Finance and Development 38: 18-21 Friedman M (1977) Nobel Lecture: Inflation and Unemployment. The Journal of Political Economy 85:451-472 Fountas S (2001) The relationship between inflation and inflation uncertainty in the UK: 1885-1998. Economics Letters 74:77-83 Grier B. K. and M. J. Perry(1998) On inflation and inflation uncertainty in the G7 countries, Journal of International Money and Finance 17:671-689 Hanke S H (2003) Money and Rule of Law in Ecuador. Policy Reform 6(3):131-145 Hansen, B.E. (2001) The New Econometrics of Structural Change: Dating Breaks in U.S. Labor Productivity. Journal of Economic Perspectives 15(4): 117128 Hayashi F (2000) Econometrics. 1st ed. Princeton, NJ: Princeton University Press Jacome L I (2004) The Late 1990s Financial Crisis in Ecuador: Institutional Weaknesses, Fiscal Rigidities, and Financial Dollarization at Work.IMF Working Paper No. WP/04/12

Jameson KP (2003a) Dollarization in Latin America: Wave of the Future or Flight to the Past? Journal of Economic Issues 37:643-663 Jameson KP (2003b) Is It Possible to De-Dollarize? : The Case of Ecuador. International Journal of Political Economy 33 (1): 42-60 Jameson KP (2004) Dollarization in Ecuador: A Post-Keynesian Institutionalist Analysis. University of Utah Working Paper No: 2004-2005 Martinez GX (2006) The Political Economy of the Ecuadorian Financial Crisis. Cambridge Journal of Economics 30(4): 567-585. Minda A ( 2005 ) Full Dollarization: A Last Resort Solution to Financial Instability in Emerging Countries? The European Journal of Development Research 17 (2): 289 - 316 Mishkin FS, Schmidt-Hebbel K (2007) Does Inflation Targeting Make a Difference? NBER Working Paper No. 12876 Moore B (1988) Horizontalists and Verticalists: The macroeconomics of credit money, Cambridge: Cambridge University Pres. Nazmi N (2001) Failed reforms and economic collapse in Ecuador. The Quarterly Review of Economics and Finance 41:727-735. Ozmen E (2003) Testing the quantity theory of money in Greece, Applied Economics Letters 10:971-974. Pinga V E, Nelson G C (2001) Money prices and causality monetarist versus structuralist explanations using pooled country evidence. Applied Economics 33: 1271-1281. Pradhan B K, Subramanian A (1998) Money and prices some evidence from India. Applied Economics 30:821-827. Panizza U, Stein E, Talvi E (2003) Measuring costs and benefits of dollarization: An application to Central American and Caribbean countries. In: Yeyati E, Sturzenegger F (edts) Dollarization: Debates and Policy Alternative, Cambridge: The MIT Press, 133-200 Quispe-Agnoli MJ (2006) Official dollarization and the banking system in Ecuador and El Salvador. Economic Review 3:55-71 Perron, P. (2006) Dealing with Structural Breaks. pg. 278372 in Palgrave Handbook of Econometrics, Vol. 1: Econometric Theory, edited by K. Patterson and T. C. Mills. New York: Palgrave Macmillan. Rose AK (2000) One money, one market: estimating the effect of common currencies on trade. Economic Policy 15(30): 7-46 Rose AK, Wincoop VE (2001) National money as a barrier to international trade: the real case for currency union. American Economic Review 91(2): 386-390 Schuler K (2005) Some theory and history of dollarization. Cato Journal 25:115-12 Solimano A (2002) Crisis and dollarization : an overview In: Beckerman P and Solimano A (edts) Crisis and Dollarization in Ecuador: Stability, Growth, and Social Equity , The World Bank, pp 1-17

Soto R (2009) Dollarization, economic growth, and employment. Economics Letters 105: 4245 Stockman AC (1981) Anticipated inflation and the capital stock in a cash-in-advance economy. Journal of Monetary Economics 8:387-393 Summers LH (2000 ) International Financial crises: causes, prevention, and cures. American Economic Review 90(2): 1-16 Sun H, Ma Y (2004) Money and price relationship in China. Journal of Chinese Economic and Business Studies 2:225-247 Vymyatnina Y (2006) How much control Bank of Russia have over Money supply? Research in International Business and Finance 20:131-144 Zhang J (2000) Inflation and Growth: Pecuniary Transactions Costs and Qualitative Equivalence. Journal of Money, Credit and Banking 32:1-12 TABLES Table II Ratio of Oil and Banana Revenues to Total Revenues Year Crude Petroleum Bananas including plantains 1990 46.4 17.4 2001 37.1 18.2 2007 53.8 9.4 Table III Structure of GDP as Ratio of GDP (1990 = 100) Year Agriculture Mining (Including Oil) Industrial Production Services 1990-1999 11.6% 14.6% 58% 15.8% 2000-2006 8.4% 20.4% 5.6% 65.6% Source: The UNDP web site. Table IV Summary Statististics of CPI Inflation and GDP Growth in Different Periods
Mean Standard Deviation Real GDP CPI Inflation Real GDP Growth Mean Standard Dev. Mean Standard Dev. Whole Period 0.02 0.02 Nov. (Jan. 1990 2007) Whole Period 4374 578 Q3 (Q1 1994 0.008 2007) 0.016 Non-Dollarization 0.03 0.021 (Jan. 1990- March 2000) Non-Dollarization 3912 147 (Q1 1994 Q1 2000) 0.002 0.02 Dollarization 0.01 0.015 (April 2000 Nov. 2007) Dollarization 4758 517 (Q2 2000 Q3 2007) 0.01 0.01

Table V Analysis of Changes in Inflation (Monthly Data) CPI Inflation Dollarization_Dummy -0.02 (-7.74)** Lag of Dependent Constant 0.03 Variable (17)** R-Squared 0.22 Number of Obs. 212 -0.01 (-3.94)** 0.55 0.01 (9.7)** (0.002)** 0.46 211

Notes: Dollarization_Dummy =1 after Q1 2000. tstatistics are presented in parantheses. * denotes significant at 5 percent and ** denotes significant at 1 percent.

Table VI Analysis of Changes in Macroeconomic Variables (Quarterly Data)

CPI Inflation (1) Dollarization Dummy Lag of Dependent Variable Oil Production Growth World Oil Price MSCI Emerging Market Index Growth Institutional Quality(Polity IV) Constant R-Squared 0.09 (12)** 0.28 0.04 (3.7)** 0.59 -0.02 (-1.51) 0.16 (1.93) 0.61 3912.2 (49.5)** 0.54 108 (1.22)** 0.99 3429.2 (57.8)** 0.88 3914 (50.6)** 0.54 25.2 (11.4)** 216.4 (0.96) -0.06 (-5.1)** (2) -0.03 (-3.3)** 0.62 (7.2)** (3) -0.05 (-3.14)** 0.63 (7.33)** (1) 846.2 (7.9)** (2) 67.4 (2.7)** 0.98 (43.8)** -38 (-0.15) 24.13 (9.42)** 255.97 (0.96) 360 (1.71) 942.4 (0.64) 0.85 0.002 (0.69) 0.09 0.002 (0.66) 0.09 0.004 (0.89) 0.1 0.001 (0.4) 0.16 0.11 51 -0.000 (-0.58) -0.003 (-0.17) Real GDP (3) 234.6 (3.1)** (4) 795.9 (7.45)** (5) 580 (2.76)** (1) 0.01 (2.3)** (2) 0.01 (2.05)* 0.03 (0.22) 0.02 (1.82) -0.000 (-0.69) 0.016 (0.75) 0.00 (0.96) Real GDP Growth (3) 0.01 (2.08)* (4) 0.01 (2.73)** (5) 0.01 (2.11)*

Number of Obs. 69 68 68 55 54 55 53 52 54 53 54 52 Notes: Dollarization_Dummy =1 after Q1 2000. t-statistics are presented in parantheses. * denotes significant at 5 percent and ** denotes significant at 1 percent.

Table VII Stability Test Results of the Coefficient of the Constant for Monthly Regressions (Robustness tests of table V)

Null Hypothesis: No breaks at March 2000 Chow Test Statistic (1) F-statistic Log likelihood ratio Wald Statistic (2)

52.47 14.56 47.28 14.28 (0.00) (0.00) 52.47 14.56 (0.00) (0.00) Notes: P-values of the test statistics are (0.00) (0.00) displayed in parantheses. (1) is the model with
the contant term and (2) is the model with both the constant term first lag of inflation.

Null Hypothesis: No breakpoints within trimmed data Number of breaks compared: 148 Quandt-Andrews Statistic (1) Maximum LR F-statistic Maximum LR F-statistic (2001M04) Exp LR F-statistic (2001M02) Ave LR F-statistic 107.09 (2)

26.91 (0.00) 49.61 9.73 (0.00) 38.93 8.59 (0.00) (0.00) Note: P-values of the test statistics are displayed in parantheses. (1) is the model with the contant term and (2) is (0.00) the model with (0.00) both the
constant term first lag of inflation.

Table VIII Stability Test Results of the Coefficient of the Constant for Quarterly Regressions (Robustness tests of table VI)
CPI Inflation (1) Chow Test Statistic F-statistic 14.21 (0.00) 13.28 (0.00) 14.21 (0.00) 10.82 (0.00) 10.42 (0.00) 10.82 (0.00) 9.65 (0.00) 9.55 (0.00) 9.65 (0.00) 50.34 (0.00) 36.22 (0.00) 50.34 (0.00) 5.54 (0.02) 5.57 (0.02) 5.54 (0.02) 4.77 (0.04) 4.93 (0.03) 4.77 (0.03) 41.14 (0.00) 36.12 (0.00) 49.14 (0.00) 9.76E-15 (1.00) 1.14E-13 (1.00) 9.76E-15 (1.00) 4.44 (0.04) 4.42 (0.04) 4.44 (0.04) 3.45 (0.07) 3.54 (0.06) 3.45 (0.06) 3.33 (0.07) 3.49 (0.06) 3.33 (0.07) 4.89 (0.03) 4.95 (0.03) 4.89 (0.03) (2) (3) (1) (2) Real GDP (3) (4) (5) (1) Real GDP Growth (2) (3) (4)

Log likelihood ratio

Wald Statistic

Quandt-Andrews Statistic Maximum LR 67.72 F-statistic (0.00)

14.25 (0.00)

13.15 (0.00)

206.97 (0.00)

13.83 (0.00)

43.74 (0.00)

212.32 (0.00)

36.66 (0.00)

5.06 (0.29)

4.01 (0.37)

4.21 (0.34)

5.31 (0.22)

Break Period 2001Q2 2001Q2 2000Q3 2003Q3 2003Q3 2001Q4 2003Q3 2001Q3 2000Q2 2000Q2 2000Q2 1999Q4 Determined Notes: P-values of the test statistics are displayed in parantheses. The stability statistics of each regression specification in table VII is displayed. The breakpoint is the first quarter of 2000 for the Chow statistic. 48 breaks are compared with the Andrews-Quandt test for inflation and 45 breaks are compared for real GDP. The stability tests can not be calculated for the fifth regression specification of Real GDP Growth.

Table IX (GARCH(1,1)) Inflation Uncertainty and Dollarization

Inflation Equation (1) (2) 0.0046 (0.02)* 0.74 (0.00)** (3) 0.004 (0.34) 0.78 (0.00)** (4) 0.03 (0.5) 0.97 (0.00)** -0.72 (0.00)** 0.00003 (0.00)** 0.05 (0.00)** 0.87 (0.00)** -0.00003 (0.00)**

C AR(1) MA(1)

Variance Equation 0.001 0.00005 (0.01)** (0.00)** 0.12 (0.02)* 0.8 (0.00)** -0.0001 (0.01)** 0.017 (0.14) 0.88 (0.00)** -0.00005 (0.00)**

0.0002 (0.00)** 0.03 (0.55) 0.2 (0.18) -0.0002 (0.00)**

RESID(-1)^2

GARCH(-1)

DOLLARIZATIONDUMMY

0.002 (0.00) Adjusted R-square 0.36 0.36 0.4 AIC -5.2 -5.92 -5.99 -6.06 SIC -5.1 -5.83 -5.88 -5.96 Notes: (P-values are presented under the coefficients in parantheses.) CPI INFLATION

Table X Endogeneity Test of Money: The C statistic (also known as a GMM distance or difference-in-Sargan statistic)
Before March 2000 After March 2000 Ho: M1 is Exogenous Test Statistic Chi-sq P-Value Test Statistic Chi-sq P-Value 0.477 0.4896 48.2 0.000 0.006 0.94 16.7 0.000 0.5 0.48 6 0.01 0.12 0.73 4.5 0.03 Ho: M2 is Exogenous 0.96 47.9 0.43 16.7 0.84 5.2 0.43 4.6

Regressor 2SLS without lag Inflation 2SLS with lag Inflation GMM without lag Inflation GMM with lag Inflation

2SLS without lag Inflation 2SLS with lag Inflation GMM without lag Inflation GMM with lag Inflation

0.003 0.6 0.042 0.6

0.000 0.000 0.02 0.03

Regressor 2SLS without lag Inflation 2SLS with lag Inflation GMM without lag Inflation GMM with lag Inflation

Test Statistic 1.14 2.5 0.84 2

Ho: Reserve Money is Exogenous Chi-sq P-Value Test Statistic Chi-sq P-Value 0.27 53 0.00 0.11 22.5 0.00 0.36 4.5 0.03 0.16 7.1 0.01