Credibility and Inflation Targeting in Chile - CiteSeerX

Credibility and Inflation Targeting in Chile1 Luis F. Céspedes Research Department Central Bank of Chile Claudio Soto Research Department Central Bank of Chile First draft October 2005

Abstract In this paper we present some new evidence that indicates that the inflationary dynamics for the Chilean economy have changed in recent years. We show that price rigidity has increased while at the same time the degree of indexation based on past inflation has decreased over time. We also show that the pass-through from exchange rate to traded good inflation has also decreased in recent years. We argue that these changes are related to credibility gains by the monetary policy regime that has improved its tradeoff. Consistent with this hypothesis we show that the policy rule that characterizes the behavior of the Central Bank of Chile has become more aggressive in fighting inflation deviations from the target, and also more forwardlooking.

JEL classification: E52, E58

1

Prepared for the Ninth Annual Conference of the Central Bank of Chile “Monetary Policy under Inflation Targeting”, Santiago, Chile, October 20 and 21. We thank comments by Doug Laxton, Felipe Morandé and the efficient collaboration of Marcelo Ochoa. The views expressed are those of the authors and not necessarily those of the Central Bank of Chile.

1

I.

Introduction

After a long history of high and volatile inflation, at the beginning of the 1990s the Central Bank of Chile (CBC) began implementing its monetary policy by announcing yearly targets for inflation. While not fully an inflation targeting (IT) regime, as there was an explicit objective for the exchange rate, this new framework was the first step towards a full-fledge inflation targeting setup. One year before the first announced inflation target in 1990, the CBC was granted with autonomy. Moreover, the law granting autonomy to the CBC explicitly stated that the main objective for the monetary policy is to ensure price stability. Some authors have argued that this new institutional framework contributed in a fundamental way to lower inflation to its current level of around 3% by enhancing the credibility of the monetary policy (Corbo, 1998; Morandé, 2003; Schmidt-Hebbel and Werner, 2002).2 Credibility affects the dynamics of price adjustments and the underling process followed by inflation, and thus it determines the tradeoffs faced by the monetary authority when implementing its policy. In particular, a more credible monetary authority should face an improved tradeoff between inflation and output stabilization, making it less costly to carry out a stabilization process. In this paper we provide new evidence of changes in the dynamics of the Chilean inflationary process in recent years. Based on a New Keynesian Phillips curve, we show that price rigidity has increased in recent years while at the same time the degree of indexation in the economy--based on past inflation--has decreased. We also show that the exchange rate pass-through into traded goods inflation has decreased. Our findings are consistent with the idea that credibility on the monetary policy has increased over time. Specifically, we argue that as the monetary policy has become more credible, costly price adjustments have been carried out less frequently, and the prevalence of indexation based on past inflation has decreased. These changes in the inflationary process triggered by an enhanced credibility on the monetary policy may have had significant repercussions in the way monetary policy is implemented in Chile. As we show in Céspedes and Soto (2005), when credibility is low, a central bank that is concerned with the sacrifice ratio along a disinflationary process may be less aggressive when implementing its monetary policy in order to avoid large output losses. As it gains credibility, it may fight inflation deviation from target strongly.

2

Among other candidates for the success of the disinflationary process in Chile, De Gregorio (2003) and García (2003) argue that favorable productivity shocks played an important role.

2

Along this line, we present evidence on a structural change in the policy rule that characterizes the conduction of the monetary policy. Our evidence is consistent with the idea that over the last years the monetary policy in Chile has been operating in a "higher" credibility environment. In particular, we show that the monetary policy rule has become more forward-looking in terms of inflation and more aggressive in fighting deviations of inflation from the target. The paper is organized as follows. The next section briefly describes the different phases of the monetary policy regime in Chile since 1985. In particular, we discuss the two phases of the IT regime that started in 1991. In section 3 we discuss what credibility means and present some preliminary evidence on the change in the degree of credibility of the CBC. In section 4 we show how this change in the degree of credibility would have affect the way the monetary policy is conducted in Chile. The fifth section presents estimated policy rules for different subsamples. Finally, section 6 concludes.

II.

Evolution of the Inflation Targeting Framework in Chile

In the period 1991—2004, the Chilean economy grew at an average rate of 5.5% per year. The inflation rate fell from levels close to 30% at the beginning of 1991 to stationary levels around 3% by the end of the 1990s, and it has fluctuated around that figure since then (see Figure 1). During this period, the Central Bank of Chile began to announce an explicit target for inflation, a policy that has been credited with providing a sound guide for conducting monetary policy.3 Nevertheless, the whole monetary policy framework underwent significant changes along the way. In the first phase (1991–99), the macroeconomic framework included not only inflation targets, but also targets for the current account deficit and a managed exchange rate. In the second phase (2000 to date), the Central Bank implemented a full-fledged inflation-targeting regime in which inflation was the main policy objective, with no other explicit targets. The Central Bank of Chile announced the first target range for inflation in September 1990, after being granted independence in 1989. The initial target, a 15–20% target range for December 1991, was defined for CPI inflation. The initial inflation target was set over a rather short time horizon and represented a strong commitment to reducing inflation. The short-term inflation target reflected the need for generating credibility for the new regime. Previous poor inflation performance led to a widespread indexation of the economy and high inflation expectations (see Schmidt-Hebbel and Tapia, 2002). The stabilization process was very gradual to avoid high output losses in the context of this widespread indexation and low credibility (Massad, 2003).

3

In addition to annual inflation targets, the Central Bank implemented a target band for the exchange rate during the fist phase (1991–99). The band was perceived as the appropriate instrument for achieving the objective of a normal functioning of the external payments system. The Central Bank also set targets for the current account deficit. To retain the possibility of managing the exchange rate with monetary policy independence, the Central Bank maintained regulations on the capital account, including a non-remunerated reserve requirement for capital inflows. Exchange rate market interventions were conducted during this period to sustain the exchange rate band. Significant modifications to the exchange rate band were also applied, including adjustments in its width and one-time realignments. Since 2000, Chile has operated a flexible inflation-targeting regime with the objective of keeping the CPI inflation rate between 2 and 4 percent over a two-year horizon. The move toward a fullfledged inflation-targeting framework was seen as the natural step after reaching the steady-state level of inflation and establishing sufficient credibility. It was triggered, in part, by the macroeconomic outcomes of the Asian crisis. GDP growth fell significantly in 1998—99, while the annual inflation rate dropped from 4.6% in 1998 to 2.3% in 1999. These events led the monetary policy authority to embark on a substantial enhancement of its macroeconomic framework. The main new elements were the adoption of a free-floating exchange rate regime, the deepening of the foreign exchange derivative market, and the total opening of the capital account (see Morandé, 2002; Céspedes et al. 2005). In addition, transparency increased significantly with the publication of a regular inflation report and the public release of policy meeting minutes.

III.

Has Credibility Improved in Chile?

Answering this question is difficult, as there are no direct measures of “credibility”. On the one hand, there is no unanimous definition of the concept of credibility by itself. On the other, given different definitions of what credibility means, practical measures are not readily available as most definition involves an important subjective component.

a.

What credibility means?

In the academic literature credibility of a central bank is identified as incentive compatibility, precommitment or strong aversion towards inflation. In the classical Barro and Gordon (1983) paper a central bank is “credible” if it attains higher payoffs (utility) by following its promised actions rather than reneging. Therefore, it is expected that a central bank will do exactly what it promised to 3

See Corbo (1998) on the effects of inflation targeting on Chilean inflation dynamics in the 1990s.

4

do because that increases its own payoff. This logic is behind Walsh’s proposal for an optimal contract design for central bankers in order to make more efficient the fight against inflation (Walsh, 1995). By making incentive compatible the fight against inflation through an explicit contract, this mechanism ensures credibility in the promises of the monetary authority. Alternatively, if such contract does not exist, credibility could be reach by means of a (“credible”) pre-commitment “technology”. Finally, credibility has been associated with a strong aversion towards inflation. However, this definition seems a bit tautological. It is not clear how to determine whether the monetary authority really has a strong aversion towards inflation, unless one is able to determine how credible is this aversion towards inflation. Moreover, having a strong aversion towards inflation does not mean that a central bank has the means to reach its objectives. In a recent paper, Blinder surveys a group of central bankers and academics and asks why credibility is important and how can central banks enhance their credibility (Blinder, 1999). Although he doesn’t ask for a definition of credibility, he asks how close is the concept of credibility to “dedication to price stability”. Nearly 90% of the respondent answered that they were “quite closed related”. The main argument for why the CBC may have gain credibility over the 90s hinges precisely on this notion of credibility. The new organic constitutional law of 1989 not only guaranteed CBC autonomy from the government, but it also explicitly established as one of the two main objectives for the Bank to ensure the stability of the value of the national currency. In other words, one of the explicit objectives is price stability. Autonomy, however, may not be enough to guarantee credibility. Posen (1995) and Fischer (1994), for example, found a positive correlation between the sacrifice ratio and an index of central bank independence. This result seems to suggest that more autonomous central banks are not necessarily more credible. What could end up being more relevant is “to match deeds and words” (Blinder, 1999). Albagli and Schmidt-Hebbel (2004) report an international cross-section comparison of inflation performance for several IT central banks. They find that for different measures of inflation deviations from target, Chile ranks among the most accurate inflation targeters in the sample of 19 countries. Therefore, at least from an international perspective, the CBC has fulfilled its promises. The question is then whether this behavior has been consistent over time. In Figure 2 we present two measures of inflation deviations from target for Chile since 1991. In one case we consider the center of the target range as the objective for the authority. In the other, we assume that if inflation

5

is within the target range there is no deviation from target. In both cases we report the deviation and the absolute value of the deviation. From the figure is clear that at the beginning of the 90s the CBC was less accurate in hitting the target.4 It improved by mid 90s, when the economy entered a phase of sustained growth a low inflation. The performance of the CBC in terms of hitting the mid-point of the target range worsens after the Russian crisis, when inflation felt below target more that 2%. However, over the last years in general the CBC has managed to keep inflation within the target range.

b.

A direct measure of credibility

A direct measure of credibility is the difference between private inflation expectations and the announced target. We consider two measures of expected inflation. The first one is based on nominal-real interest rate differentials. The second one is a measure of expectation taken from survey data. This credibility measure is consistent with the one discussed by Faust and Svensson (2001) and Cukierman and Meltzer (1986). In particular, these last authors define credibility as the absolute value of the difference between the policymaker's plans and the private sector's beliefs about those plans. In this case, the smaller this difference, the higher the credibility of planned monetary policy. Figure 3a depicts the evolution of the difference between expected inflation constructed using market interest rates and the announced inflation target. From this figure we observe that at the beginning of the 90s credibility increased. However, in the midst of the Asian crisis expected inflation was way above the announced target, signaling a possible decrease in credibility. By the end of the 90s and during the last five years expected inflation has been lower than the mid-point of the announced target, but inside the target range (except for 2004). In other words, credibility of the announced target seems to have increased over the last years. There are two important criticisms to this measure of credibility based on market expectation extracted from interest rate differentials. First, the level and the volatility of inflation were much higher at the beginning of the 90s than at the end of the decade. That implies that the expected the inflationary premium implicit in the nominal interest rate was also higher at the beginning of the 90s. Therefore, is not all that clear that the difference between expected inflation and the announced target has decrease over time. Second, although nominal instruments exist since a long time, the

4

Some authors normalise the deviation of inflation from target by the inflation target level.

6

market turnover was pretty low until 2001 when the monetary policy was nominalized.5 In other words, the market for nominal instrument was not very liquid but until recently. Therefore, prices not necessarily reflect market expectations on inflation. In Figure 3b we present the difference between private inflation expectations and the announced target using survey data. Survey data on analyst expectations regarding future inflation from consensus forecast are available for Chile since 1993. When using this measure of expectations, we observe that expected inflation has been in general much closer to the mid-point of the inflation target. Until 2002, expected inflation was above the mid-point of the target (except for the year 1999). After that year, expected inflation has been below the mid-point of the target, but always inside the target range.6

IV.

Frequency of price adjustment, credibility and indexation

The preliminary evidence presented in the previous section hints that in fact there may have been gains in monetary policy credibility along the disinflationary process carried out by the CBC over the 90s. In this section we perform a more detailed analysis on how this change in the macro environment may have changed the inflationary dynamics. We start by estimating a New Keynesian Phillips curve for Chile and performing some stability test on some key semi-structural parameters. Then we evaluate whether the degree of import prices rigidity may have changed in a manner consistent with the hypothesis that the credibility of the Chilean monetary policy regime has risen. Our evidence below tends to support the idea that the exchange rate pass-through has decreased over the last years.7

a.

The Phillips curve and the persistence of inflation

We estimate a New Keynesian Phillips Curve (NKPC) were parameters have a semi-structural interpretation. We emphasize that parameters are not completely structural. In fact, as we show below, changes in the macro environment seem to have had an impact on the value of some of the “deep” parameters that characterize the NKPC (see also Rudd and Whelan, 2005). In particular, we show that the fraction of firms adjusting optimally prices each period decreased at the end of the

5

Together with defining as its monetary policy instrument an overnight nominal interest rate, the CBC also introduced in August 2001 a set of nominal instruments to help setting benchmarks for that market. 6 For comparison with previous periods, we are considering one year ahead expected inflation. However, after 2001, with the introduction of a steady-state inflation target, the monetary authority explicitly states that its goal is to keep inflation within the target range in a horizon of 12 to 24 months. 7 García and Restrepo (1999), Schimidt-Hebbel and Werner (2002) have documented the existence of a change in the pass-through coefficient for the Chilean economy using different setups.

7

90s. We also show that the weight given to the announced inflation target for those firms that do not optimally adjust prices in a particular period increased over the last years. We argue this evidence is consistent with the view that higher credibility of the CBC in its commitment with price stability (low inflation) changed the way in which firms set prices. The theoretical NKPC comes from a standard model of monopolistic competitive firms that adjust prices infrequently (see Galí and Gertler (1999) for a full derivation of the NKPC). We extend the basic model in Galí and Gertler to allow for “passive” price adjustment for all those firms that do not optimally optimize each period. Notice that in our setup firms change prices every period, either optimally or by following this “passive” adjustment. The logic behind this dual pricing mechanism is based on the idea of that menu costs for price adjustments are not related to the cost of changing prices by itself, but rather they refer to all those cost that an optimizing procedure involves (collecting information on market conditions, costs, forecasting, etc). Given those menu cost, firms will go through a re-optimization process only infrequently (see Christiano, Eichenbaun and Evans (2005)). There are two reasons for why this framework is appropriate to describe the Chilean inflationary dynamics. First, the notion of passive adjustment is very plausible to occur in a high inflation environment—as in Chile at the beginning of the 90s—where the information content of prices it is difficult to extract.8 The second reason is that the passive adjustment setting allows us to formally introduce an indexation mechanism into the Phillips curve. In the case of Chile, indexation has been a phenomenon prevalent for many years (see Landerretche, Lefort and Valdés, 2002). The passive adjustment mechanism for those firms that do not optimally adjust their prices consists in upgrading prices in proportion to a geometric average of past inflation and the announced target for inflation. The passive adjustment rule can be expressed as i

(1)

Γ t,i =∏ (1+π t + j −1 ) (1+π ttar +j) κ

1−κ

j=1

where Γ t,i is the percentage change in price of a firm that is not able to optimally adjust between t and t+i, π t + j −1 is the inflation rate in t+j-1 and π ttar + j is the inflation target set by the authority for the period t+j. This updating rule implies that whenever firms do not receive a signal they adjust their prices by a geometric average of the inflation target set by the authority and past inflation. Parameter κ is a measure of the degree of persistency of inflation and can be associated to the 8

Calvo, Celasum and Kumholf (2001) and by Mankiw and Reiss (2001) have developed pricing models in which the pricing strategies may be related to a passive adjustment mechanisms.

8

credibility of the target set by the authority. The larger is this parameter, the larger the weight given to past inflation and the lower the weight to the inflation target. Therefore, the lower is the credibility of the announcement. The Phillips curve with the Calvo model and this updating rule is given by, m t + γ E πl t +1 + γ πl t −1 + ζ πl t =λξ mc f t b t

(2)

m t are marginal costs, where πl t = π t − π ttar corresponds to the deviation of inflation from target, mc

and λ (κ ,θ , β ) = (1-θ )(1-θβ ) / (θ (1+κβ ) ) , γ f (κ ,θ , β ) = β / (1+κβ ) , γ b (κ ,θ , β ) = κ / (1+κβ ) .

Parameter β is the subjective discount factor. The term ζ t is a function of changes in the inflation target.9 Notice that parameter κ defines the backward looking component in the Phillips curve. In other words, this parameter also is a measure of the degree of indexation in the economy. Parameter θ is the probability that a firm will keep the current price until the next period. It also corresponds to the fraction of firms that do not optimally adjust prices in a particular period. We consider this parameter also as a measure of credibility. In the standard Calvo model, the probability of adjusting prices --or the fraction of firms setting optimally prices each period-- has no economic interpretation besides being a measure of the degree of price stickiness. In our case, we assume that firms will adjust prices more often (i.e. a larger fraction of firms adjusting optimally prices each period) when future inflation is more uncertain and the compromise of the central bank regarding inflation is perceived to be weak. Parameters θ , β and κ are estimated by GMM with quarterly data from 1991:1 to 2004:4 using the following orthogonality condition derived from (2), (3)

{(

) }

m t − θβπl t +1 − θκπl t −1 + θ (1 + θβ )ζ z Et θ (1 + θβ )πl t - (1-θ )(1 − θβ )ξ mc t t

where zt is a vector of instruments that includes three lags of the inflation deviation from target, the real marginal cost deviation from trend, and the output gap (from t-3 to t-5).10 Benchmark results for the estimated hybrid model are presented in Table 1. We present the estimated values of parameters θ , β, and λ under four different specification for the marginal cost, and assuming alternatively that capital is freely mobile and firm specific. We also report the In steady state, with a constant inflation target, ζ t = 0. In order to check the relevance of the instrument set used in our regressions we test null hypothesis that the coefficients on all the instruments are jointly zero in the first stage of the estimation. The F-statistic, the

9

10

9

estimated value of parameter κ, which measures the degree at which firms index their prices to past inflation. The estimated share of the backward-looking component γ b is statistically significant in all specifications and it is about 0.45. This figure is slightly smaller than the one reported by Agenor and Bayraktar (2003) for Chile in their study of the inflation dynamics in middle income countries. These authors estimate a non-structural Phillips curve that includes both a backward and a forwardlooking component, and found that the backward-looking component is about 0.52. Unlike our case, Agenor and Bayraktar use several lag of the output gap (up to 3 for Chile) as the driving force for inflation. The estimated values for parameter θ under this hybrid specification of the NKPC lie in the range 0.8-0.9. This implies that firms adjust prices on average every 7 quarters. In the case of discount factor β our empirical results indicate that the estimated value for this parameter are somewhat low. As Galí and Lopez-Salido (2002) our specification for marginal cost with firm-specific capital delivers a lower values for θ , implying shorter average duration of price stickiness. Finally notice that for all cases the overidentifying restrictions are satisfied. Table 1 also presents the results for the specification for marginal cost that assumes firm-specific capital (results assuming capital mobility are similar).11 Parameter κ is consistently estimated in the range 0.75-0.82 implying that within the sample period, firms weighted more past inflation rather than announced inflation targets to passively adjust their prices. Not surprising, these figures imply a much stronger role for the backward looking component of inflation in the Chilean case than the estimated by Galí and Gertler (1999) and López-Salido (2001) for the Euro area. However, they are quite similar to the estimates of for the U.S. by Galí and Gertler, and Galí Gertler and López-Salido. We turn now to the issue of the stability of the parameters. 12 Following Céspedes, Ochoa and Soto (2005) we consider a predictive test for structural change with unknown breakpoint developed by Ghysels and Hall (1990), Ghysels et. al. (1997) and Guay (2003). This test consists on estimating the parameter vector for a first subsample and then evaluating the moment conditions for the second associated p-value and the adjusted R² from the first stage of these regressions allowed us to reject the null hypothesis that the instruments are jointly irrelevant. In generally the adjusted R² is over 0.5. 11 We perform a robustness exercise to address the importance of the backward looking component of the Phillips curve. We add additional lags of inflation to the hybrid model. Additional lags of inflation turn out to be non-significant. Neither is the sum of the three additional lag of inflation. 12 This issue has not been formally analyzed in the literature. Jondeau and Le Gihan (2005) test the stability of their Phillips curve estimates but examine only the reduced (linear) form parameters and use Wald-type tests which have some important drawbacks as they point out.

10

subsample at these parameter values. The subsample is then increased by one observation at the time. Each time the statistic PR (“PRedicted”) is constructed.13 Table 2 presents the estimated predictive tests (supremum: supPR; average: avgPR and exponential: expPR) along with the date for which the largest PR test is obtained. The PR-type tests can be divided into a test of structural change for the vector of parameters and a test of the stability of the overidentifying restrictions (Sovell, 1996). We report the PR1-type statistics that tests both parameter and overidentifying restriction stability and, the PR2-type tests for parameter variations only. The results show that for the four specifications of marginal costs, we cannot reject the existence of a breakpoint. Furthermore, the PR1 and the PR2 tests estimate consistently the date of the breakpoint around the first quarters of 2001 which is close to the date that the inflation target reached its stationary annual value of 3%. We also report the estimated values of the parameter κ and the duration of price stickiness before the breakpoint date. It is noteworthy that in most cases the first subsample value of κ is larger than the estimated parameter using the whole sample suggesting that after the break point in 2001 there are more firms updating their prices according to the inflation target. Regarding price stickiness, we find that the estimated duration of price stickiness is smaller before the breakpoint, as expected. In order to further assess the potential direction of the structural change in the inflationary process in the Chilean economy, we run a set of regressions starting with the period 1991.I-1997.IV and then adding one extra observation at a time. For these calculations we use real marginal costs

computed based on a production function with overhead labor and firm specific capital. The results indicate that price rigidity, captured by the parameter θ, has been increasing in recent years (see Figure 4). In particular, the estimated duration of price rigidity went from 2 quarters, for the period previous to 2000, to around 3 quarters for the whole period. Also, we find that the parameter that measures the degree by which firms adjusts their current prices based on previous inflation (κ) has been decreasing over time (see Figure 5). Consequently, the fraction of firms adjusting their prices using the inflation target has increased over time.

13

For our purposes this approach has several advantages over alternative approaches like the Wald-type tests proposed in the work of Andrews (1993) and Andrews and Ploberger (1994). Firstly, we only use first subsample estimates of the parameters, which allow us to test the presence of a break even when the second subsample contains a few observations and parameter estimates are not feasible. Indeed, it is a common drawback of Wald-type tests that they cannot be applied to detect structural instability at the end of the sample. Secondly, we do not set a priori orthogonality conditions equal to zero in the second subsample thus, we avoid rejecting stability when in fact the parameters were stable but there was certain type of misspecification (e.g., omitted variables). A brief description of the test can be found in Céspedes, Ochoa and Soto (2005).

11

The evidence presented here is consistent with the view that some parameters are policy dependent (Rudd and Whelan, 2005). In particular, we argue the increase in the price rigidity parameter and the decrease in the degree of indexation to past inflation is consistent with the hypothesis that monetary policy in Chile has become more credible. These results are also in line with the empirical and theoretical view that more credible monetary policy are negatively correlated with inflation persistence (see Taylor, 2000; and Sargent, 1999).

b.

Changes in the Exchange Rate Pass-Through

Broad empirical literature has shown that the pass-through from exchange rate into prices is not full in the short-run. There is wide empirical evidence of deviations from the law of one price for import prices for various countries (Campa and Goldberg, 2002). There is also evidence that the response of inflation (CPI) to changes in the exchange rate is less than one to one in the short-run (Borensztein and De Gregorio, 1999; and Goldfajn and Werlang, 2002). In this subsection we provide new evidence regarding the evolution of the pass-through in Chile during the last 15 years using the New Keynesian Phillips curve as analytical framework (Monacelli, 2002). Then we argue that the results obtained are consistent with an increase in the credibility of the monetary policy framework. In general, the pass-through from exchange rate to prices may be imperfect in the short-run because of features of the market structure as in Dornbusch (1977), or because of nominal rigidities. In particular, allowing for pricing to market (Betts and Devereux, 1996), if a fraction of import prices are sticky in the domestic currency then changes in the exchange rate will not be completely trespassed into prices. In other words, the pass-through from exchange rate to import prices will be incomplete. Therefore, changes in price rigidity will also imply changes in the exchange rate passthrough. Since we are interested in the pass-through as a macro phenomena, we estimate the effect of changes in the nominal exchange into imported goods inflation based on a sticky price model similar to the one used to derive the Phillips curve from the previous section. Following Monacelli (2002) we assume that different varieties of imported goods are bought in the international market by domestic retailing firms that later sell them domestically. We assume that each retailing firm has monopoly power on a particular imported variety, and adjust their prices infrequently with an exogenous probability 1- θ M each period. For simplicity, we also assume that the distribution

12

service provided by retailing firms does not require the utilization of any input.14 A particular retailing firm selling variety zM that is able to adjust its price in period t chooses a new price,

PMnew,t ( zM ) , to maximize its expected profits, ∞ ⎡ P new ( z ) − St + i PM* ,t + i ( zM ) ⎤ Et ∑θ Mi Λ t ,i ⎢ M ,t M CM ,t + i ( z M ) ⎥ , Pt + i i =0 ⎣ ⎦

subject to the domestic demand for that variety, CM ,t + i ( zM ) . Variable St is the nominal exchange rate, and PM* ,t ( zM ) is the international price of variety zM expressed in foreign currency. From the first order condition for this problem we can establish the following relationship for imported goods inflation, π M ,t , and the logarithm of the price of imported goods abroad expressed in domestic * currency, st + pM,t , relative to the price of import goods in the domestic market, pM,t ,

(4)

π M ,t =

(1-θ M )(1-θ M β ) θM

(s

t

+ pM* ,t − pM ,t ) + β Etπ M ,t +1

We estimate equation 4 by GMM using as a proxy of the price of imported goods in foreign currency the index of unitary value of imports (IVUM). The domestic price of imported goods is proxied by the price index corresponding to tradable goods included in the CPI basket. Therefore, our measure of inflation corresponds to the quarterly change in the tradable good price index. Our results indicate that the relative price of imports abroad is a significant driving force of the tradable inflation rate and that the forward-looking component is also very relevant. Moreover, the coefficient associated to price rigidity, θ M , is positive and significant (see Table 3). This value leads to an estimated duration of price stickiness of around 5 quarters. In order to test for potential structural breaks we consider the predictive test for structural change discussed in the previous section. The supPR test indicates that there is a structural break around the first quarter of 2001. Again, in order to assess the scope of this change we estimate a set of regression starting with the period 1991.I-1997.IV and then adding one extra observation at a time. The results from this exercise indicate that during the 90s, the average price rigidity was close to 2.5 quarters, half the value for the whole sample.15

14

We also estimate a version of the model assuming that the retailing service provided by these firms requires some labor. Results did not significantly change (see specifications 2 to 4 in table 3) 15 Acknowledging the fact that the number of observations may be low, we divided the sample in two parts: from 1991.I to 1997.IV and from 1998.I to 2004.IV. We find results consistent with the hypothesis of a change in price rigidity. In particular, we find that the average duration of price stickiness during the second period is almost twice the duration during the first period.

13

These results indicate that during the "low inflation" period for the Chilean economy, after 1999, the short-run pass-through has been lower then in the previous period. Taylor (2000) has argued that in low inflation environments the pass-through coefficient should be lower as low inflation may be associated to less persistent changes in costs. From the perspective of the theoretical framework we use, it can be argued that higher credibility in monetary policy, interpreted as a credible commitment to low and stable inflation rate around 3% has reduced the frequency of price changes.16 One special feature of the change in the pass-through coefficient in Chile is that it seems to have occurred not at the moment in which inflation rates reached low values, around 1998, but sometime later. We interpret this result as indicating that only when monetary policy was credible, there was a change in the pass-through coefficient. This evidence is related to evidence presented by Bailliu and Fujji (2004) that indicates that the exchange rate pass-through coefficients for a group of industrialized economies declined following the inflation stabilization in the early 1990s but did not decline following a similar episode in the early 1980s. One interpretation for this result is that the disinflationary process in the 90s was perceived as more credible than the previous one.

V.

Credibility and the Monetary Policy

In the previous section we show that the frequency of price adjustments, the degree of indexation to past inflation and the exchange rate pass-through have decreased over time, in particular after 2000. We interpreted this evidence as support to the view that monetary policy has become more credible over the last years. There is broad literature that shows that when the monetary authority is not fully credible its tradeoff between output and inflation stabilization is worst than when credibility is larger. In fact, in the simplest version of the New Keynesian Phillips curve, a fully credible stabilization process can be carried out without any output losses. In Céspedes and Soto (2005) we present a model were we compare the sacrifice ratio implied by different monetary policy rules under alternative assumptions regarding the credibility of a disinflationary shock. In particular, we show that if the central bank is “strict” regarding the fulfillment of its inflation target, then inflation stabilization may generate significant output losses and large sacrifice ratios for low credibility levels. Figure 7 depicts the sacrifice ratio--accumulated output losses relative to accumulated reduction in inflation--for a disinflationary shock as a function 16

Céspedes and Valdés (2005) provide evidence that shows that more independent central banks, which are associated with more credible monetary policy, have indeed lower pass-through coefficients.

14

of credibility, measured by a Kalman gain coefficient, under two alternative policy rule.17 A baseline rule, that corresponds to a Taylor rule, and an alternative policy rule that has a strong feedback response of the interest rate to deviation of inflation from target. From the figure is clear that the sacrifice ratio falls as the credibility of the inflation target rises (that is, with a higher Kalman gain coefficient). What is more interesting is that the tougher rule has a higher sacrifice ratio than the baseline Taylor rule when credibility is low. As credibility increases, however, the sacrifice ratio of the alternative rule drops below the sacrifice ratio under the baseline policy rule. In other words, as credibility increases, a central bank can reduce inflation faster without having to sacrifice too much output. In what follows we investigate whether there have been changes in the way the monetary policy is conducted in Chile that are consistent with the previous analysis. In concrete we estimate monetary policy interest rate rules for two periods that roughly coincide with the two phases described above in section II: the first quarter of 1991 to the fourth quarter of 1997; and the first quarter of 1998 to the fourth quarter of 2003.18 As discussed earlier, the first phase was characterized by short-run horizons for the inflation targets, a managed exchange rate, and a target for the current account. We capture these features by assuming that the real interest rate ( rt ) set by the Central Bank during the first period of inflation targeting was a function of current inflation, the output gap, and the current difference between the real exchange rate and its equilibrium level.19 In particular, we estimate the following relationship: (5)

rt = (1- ρ0 ) r + ρ0 rt -1 + (1- ρ0 ) ωπ,0 ( πt - πttar ) + (1- ρ0 ) ωy,0 gt -1 + (1- ρ0 ) ωe,0 qtmis + ut ,

where gt−1 corresponds to the output gap in t–1, qtmis is the real exchange rate misalignment, and ut is a linear combination of prediction errors and policy innovations.20 Since the term ut could be

17

The model in Céspedes and Soto (2005) follows the approach by Erceg and Levin (1999) to model imperfect credibility assuming that private agents must solve a signal extraction problem in order to sort out whether a shock to the inflation target is permanent or transitory. To solve for the signal extraction problem they use the Kalman filter. The larger the Kalman gain coefficient, the faster they understand the nature of the shock to the target (i.e. the more credible is a permanent shock to the inflation target). 18 We do not split the sample exactly in 1999 in order to have enough data to estimate the monetary policy rule for the second phase. 19 In our estimations, we use an ex-post real interest rate for the period 2000–03. 20 The output gap measure corresponds to the difference between effective output and a trend measure for output. This trend measure is obtained using the HP filter. The real exchange rate misalignment corresponds to the difference between the effective real exchange rate and the equilibrium real exchange rate. The equilibrium real exchange rate is computed by applying the HP filter to the effective real exchange rate. A positive value for qtmis implies that the real exchange rate is undervalued relative to its equilibrium level.

15

correlated with actual values of inflation and the output gap, we estimate this relationship using GMM, as suggested by Clarida, Galí, and Gertler (2000). The results indicate that the specification of the monetary policy rule in equation (5) is a good description of monetary policy actions for the period 1991–97 (see Table 4). The estimated coefficients have the expected sign and are significant in all cases. In addition to responding to deviations of inflation from the target, the CBC responded to deviations of output from its trend value and to the exchange rate. As discussed previously, the CBC also targeted the current account during this period.21 When we include 1998–2003 in the sample, our estimations of the policy reaction function in equation (5) yield poor results. Most coefficients become insignificant or have the wrong sign (or both). This may reflect the change in the inflation-targeting framework starting in 1999. In order to test the possibility of a structural change, we use the test for structural change proposed by Ghysels et. al. (1997) and Guay (2003) and discussed in the context of the Phillips curve estimation above. The results of the test indicate that there is a structural break and that this break occurs around the first quarter of the year 2000 (Table 5). To account for this structural change, we estimate the following policy reaction function: (6)

tar ⎤⎦ + (1- ρ1 ) ωy,1 gt -1 + ut , rt = (1- ρ1 ) r + ρ1 rt -1 + (1- ρ1 ) ωπ,1 ⎡⎣ Et ( πt+4 ) - πt+4

using quarterly data from 1998:1 to 2003:4. As we said, we do not split the sample exactly in break point detected with the PR test in order to have enough data to estimate the monetary policy rule for the second phase. The estimations of this last policy rule indicate that in the full-fledged inflation-targeting phase, the monetary authority has become more forward looking in terms of inflation. In contrast to the previous phase, the evidence indicates that the Central Bank is willing to allow deviation of current inflation from the inflation target as long as it does not have an impact on future inflation. Additionally, the coefficient associated with inflation, ω π,1 , increased relative to corresponding coefficient in the interest rate rule for the previous inflation-targeting phase. This stronger response of the interest rate to inflation deviations is consistent with a central bank that takes advantage of an

21

To test how this could have been reflected in the policy rule (other than through the exchange rate), we included the current account in our estimations. The coefficient associated with the current account balance turned out to be negative and significant, indicating that the Central Bank responded actively to changes in the current account. In particular, increases in the current account deficit beyond the target level triggered increases in the interest rate in order to cool down domestic expenditure.

16

improved trade-off between inflation and output, possibly as a consequence of higher credibility or commitment to controlling inflation.

VI.

Conclusions

In this paper we present some new evidence that indicates that the inflationary dynamics for the Chilean economy have changed in recent years. We show that price rigidity has increased while at the same time the degree of indexation based on past inflation has decreased over time. We also show that the pass-through from exchange rate to traded good inflation has also decreased in recent years. We argue that these changes are related to credibility gains by the monetary policy regime that has improved its tradeoff. Consistent with this hypothesis we show that the policy rule that characterizes the behavior of the Central Bank of Chile has become more aggressive in fighting inflation deviations from the target, and also more forward-looking.

[TO BE COMPLETED]

17

References [To be completed] Albagli, E. and K. Schmidt-Hebbel 2004 “By How Much and Why do Inflation Targeters Miss

Their Targets?” Bailliu, J., and E. Fujii. 2004. “Exchange Rate Pass-Through and the Inflation Environment in Industrialized Countries: An Empirical Investigation.” Working paper 2004-12. Ottawa: Bank of Canada. Bernanke, B. 1986. “Alternative Explanations of the Money-Income Correlation.” Carnegie-

Rochester Conference Series on Public Policy 25: 49–99. Bernanke, B., T. Laubasch, F. Mishkin, and A. Posen. 1999. Inflation Targeting: Lessons from the

International Experience. Princeton University Press. Blinder, A. (1999) “Central Bank Credibility: Why do We Care and How Do We Build It?” NBER Working Paper 7161 Calderón, C., R. Duncan, and K. Schmidt-Hebbel. 2004. “The Role of Credibility in the Cyclical Properties of Macroeconomic Policies in Emerging Economies.” Review of World

Economics 140(4): 613–33. Calvo, G. 1983. “Staggered Prices in a Utility-Maximizing Framework.” Journal of Monetary

Economics 12(3): 383–98. Céspedes, L. F., I. Goldfajn, P. Lowe, and R. Valdés. 2004. “Policy Responses to External Shocks: The Experience of Australia, Brazil, and Chile.” External Vulnerability and Prevention

Policies, edited by R. Caballero, C. Calderón, and L. F. Céspedes. Santiago: Central Bank of Chile. Céspedes, L. F. and C. Soto 2005 “Inflation Targeting and Credibility in Emerging Markets: Lessons from the Chilean Experience” International Finance, Forthcoming Céspedes, L. F., M. Ochoa and C. Soto 2005 “An Estimated New Keynesian Phillips curve for Chile” CBC Working Paper 333, Central Bank of Chile. Choudhri, E., and D. Hakura. 2001. “Exchange Rate Pass-through to Domestic Prices: Does the Inflationary Environment Matter?” Working paper WP/01/194. Washington: International Monetary Fund.

18

Clarida, R., J. Galí, and M. Gertler. 1999. “The Science of Monetary Policy: A New Keynesian Perspective.” Journal of Economic Literature 37(4): 1661–707. ———. 2000. “Monetary Policy Rules and Macroeconomic Stability: Evidence and Some Theory.”

Quarterly Journal of Economics 115(1): 147–80. Corbo, V. 1998. “Reaching One-Digit Inflation: The Chilean Experience.” Journal of Applied

Economics 1(1): 153–64. Corbo, V., O. Landerretche, and K. Schmidt-Hebbel. 2002. “Does Inflation Targeting Make a Difference.” In Inflation Targeting: Design, Performance, Challenges, edited by N. Loayza and R. Soto. Santiago: Central Bank of Chile. Cukierman A., and Meltzer A. (1986), "A Theory of Ambiguity, Credibility, and Inflation under Discretion and Asymmetric Information," Econometrica vol. 54, issue 5, pp. 1099-128. Devereux, M. B., C. Engel, and P. E. Storgaard. 2004. “Endogenous Exchange Rate Pass-through When Nominal Prices Are Set in Advance.” Journal of International Economics 63(2): 263–91. Erceg, C., and A. T. Levin. 2003. “Imperfect Credibility and Inflation Persistence.” Journal of

Monetary Economics 50(4): 915–44. Faust, J. and Svensson, L. E. (2001), 'Transparency and Credibility: Monetary Policy with Unobservable Goals," International Economic Review vol. 42, issue 2, pp. 369-97. Fraga, E., I. Goldfajn, and A. Minella. 2004. “Inflation Targeting in Emerging Market Economies.” In NBER Macroeconomics Annual 2003, edited by Mark Gertler and Kenneth Rogoff. MIT Press. Galí, J. and M. Gertler 1999, "Inflation Dynamics: A Structural Analysis," Journal of Monetary Economics 44, 195-222. Galí, J, M. Gertler and D. López-Salido 2001, "European Inflation Dynamics," European Economic Review, vol. 45, 1237-1270. Galí, J., and T. Monacelli. 2005. “Monetary Policy and Exchange Rate Volatility in a Small Open Economy.” Review of Economic Studies 72(3): 707–34. Huh, C. G., and K. J. Lansing. 2000. “Expectations, Credibility, and Disinflation in a Small Macroeconomic Model.” Journal of Economics and Business 52(1–2): 51–86.

19

Massad, C. 2003. “Políticas del Banco Central de Chile 1997–2003.” Santiago: Central Bank of Chile. Mishkin, F., and K. Schmidt-Hebbel. 2001. “One Decade of Inflation Targeting in the World: What Do We Know and What Do We Need to Know?” Working paper W8397. Cambridge, Mass.: National Bureau of Economic Research. Monacelli, T. 2003. “Monetary Policy in a Low Pass-through Environment” Working paper 227. Frankfurt: European Central Bank. Morandé, F. 2002. “A Decade of Inflation Targeting in Chile.” In Inflation Targeting: Design,

Performance, Challenges, edited by N. Loayza and R. Soto. Santiago: Central Bank of Chile. Rudd, J. and K. Whelan. 2003. “Inflation Targets, Credibility, and Persistence in a Simple StickyPrice Framework”. FEDS Working Paper No. 2003-43 (July). Schmidt-Hebbel K., and M. Tapia. 2002. “Monetary Policy Implementation and Results in Twenty Inflation-Targeting Countries.” Working paper 166. Schmidt-Hebbel, K and A. Werner. 2002. “Inflation Targeting in Brazil, Chile, and Mexico: Performance, Credibility, and the Exchange Rate.” Working Paper 171. Santiago: Central

Bank of Chile. Svensson, L. 2000. “Open-Economy Inflation Targeting.” Journal of International Economics 50(1): 155–83. Taylor, J. 2000. “Low Inflation, Pass-through, and Pricing Power of Firms.” European Economic

Review 44(7): 1398–408. Valdés, R. 1998. “Efectos de la política monetaria en Chile.” Cuadernos de Economía 35(104): 97– 125. Walsh, C 1995 “Optimal Contract for Central Bankers” American Economic Review 85(1): 150167.

20

Figure 1 Chile 1991 – 2005: Effective Inflation and Inflation Target:

30.0% 27.0% 24.0% 21.0% 18.0% 15.0% 12.0% 9.0% 6.0% 3.0% 0.0% -3.0% 1991

1994

1997

Inflation target floor

2000

Inflation target ceiling

21

2003

Effective Inflation

Figure 2 Chile 1991-2005: Inflation deviation from target

Deviations of inflation from mid-point target

1990q1

0

-4

-2

2

0

% 4

%

2

6

4

6

8

Absolute deviations of inflation from mid-point target

1995q1 2000q1 Time period 1989q1 2004q4

2005q1

1990q1

2005q1

Absolute deviations of inflation from target range

1990q1

0

-2

1

0

% 2

%

2

3

4

4

Deviations of inflation from target range

1995q1 2000q1 Time period 1989q1 2004q4

1995q1 2000q1 Time period 1989q1 2004q4

2005q1

22

1990q1

1995q1 2000q1 Time period 1989q1 2004q4

2005q1

Figure 3a Chile 1991-2005: Inflation target credibility (expected inflation measured as nominal-real interest rate differentials)

-2

0

5

0

10

% 2

15

4

20

6

25

Deviations of inflation expectations from mid-point target

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Inflation expectations

1990

Inflation target mid-point

1995 2000 Time period 1993-2005

2005

Absolute deviations of inflation expectations from target range

0

0

2

2

%

%

4

4

6

6

Absolute deviations of inflation expectations from mid-point target

1990

1995 2000 Time period 1993-2005

1990

2005

23

1995 2000 Time period 1993-2005

2005

Figure 3b Chile 1991-2005: Inflation target credibility (expected inflation from consensus forecast)

-1

0

-.5

5

% 0

10

.5

1

15

Deviations of inflation expectations from mid-point target

1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Inflation expectations

1990

Inflation target mid-point

2005

Absolute deviations of inflation expectations from target range

0

0

.2

.2

.4

%

% .4

.6

.6

.8

1

.8

Absolute deviations of inflation expectations from mid-point target

1995 2000 Time period 1993-2005

1990

1995 2000 Time period 1993-2005

2005

24

1990

1995 2000 Time period 1993-2005

2005

Figure 4 Price rigidity Phillips curve ( θ ) 0.75

0.70

0.65

0.60

0.55

0.50

0.45 1998

1998

1999

1999

2000

2000

2001

2001

2002

2002

2003

2003

2004

Figure 5

Weight of past inflation in the price adjustment passive rule ( κ ) 1.1

1.0

0.9

0.8

0.7

0.6

0.5 1998

1998

1999

1999

2000

2000

25

2001

2001

2002

2002

2003

2003

2004

Figure 6 Price rigidity Phillips curve of traded goods ( θ M ) 0.91 0.90 0.90 0.89 0.89 0.88 0.88 0.87 0.87 0.86 0.86 0.85 1998

1998

1999

1999

2000

2000

2001

26

2001

2002

2002

2003

2003

2004

Figure 7 Sacrifice Ratio and Credibility 13 Baseline policy rule Alternative policy rule

Accumulated output loss/Accumulated inflation reduction

12

11

10

9

8

7

6

0

0.1

0.2

0.3

0.4 0.5 0.6 Kalman gain parameter

0.7

0.8

0.9

1

Note: The Kalman gain parameter is a measure of credibility. The larger this parameter, the more credible is a permanent disinflationary shock. Source: Céspedes and Soto (2005)

27

Table 1 Benchmark Phillips curve estimation

Model

θ β κ λ γf γb τ1 τ2 D J-Test p-value

Cobb-Douglas

Overhead Labor

ξ=1 0.894 (0.06) 0.976 (0.19) 0.737 (0.12) 0.009 (0.00) 0.568 (0.03) 0.429 (0.01) 0.418 (0.09) -0.429 (0.01) 9.428 (5.34) 7.882 (0.93)

ξ=1 0.913 (0.04) 0.923 (0.13) 0.739 (0.08) 0.009 (0.00) 0.549 (0.03) 0.439 (0.01) 0.405 (0.06) -0.439 (0.01) 11.442 (4.64) 5.826 (0.98)

ξ≠1 0.715 (0.03) 0.975 (0.19) 0.737 (0.12) 0.070 (0.03) 0.568 (0.03) 0.429 (0.01) 0.418 (0.09) -0.429 (0.01) 3.511 (0.41) 7.882 (0.93)

ξ≠1 0.709 (0.03) 0.923 (0.13) 0.739 (0.08) 0.084 (0.02) 0.549 (0.03) 0.439 (0.01) 0.405 (0.06) -0.439 (0.01) 3.435 (0.30) 5.826 (0.98)

CES, σ=0.5 ξ=1 0.939 (0.03) 0.970 (0.11) 0.700 (0.08) 0.003 (0.00) 0.578 (0.02) 0.417 (0.01) 0.404 (0.01) -0.417 (0.01) 16.288 (8.09) 6.115 (0.98)

ξ≠1 0.806 (0.05) 0.970 (0.11) 0.700 (0.08) 0.031 (0.02) 0.578 (0.02) 0.417 (0.01) 0.404 (0.01) -0.417 (0.01) 5.151 (1.35) 6.119 (0.98)

CES, σ=1.5 ξ=1 0.897 (0.05) 0.960 (0.18) 0.742 (0.12) 0.009 (0.00) 0.561 (0.03) 0.433 (0.01) 0.416 (0.09) -0.433 (0.01) 12.188 (5.17) 7.276 (0.95)

ξ≠1 0.800 (0.04) 0.960 (0.18) 0.742 (0.12) 0.034 (0.01) 0.561 (0.03) 0.433 (0.01) 0.416 (0.09) -0.433 (0.01) 4.994 (1.02) 7.274 (0.95)

CES for an open CES for an open economy, σ=0.5 economy, σ=1.5 ξ=1 0.883 (0.15) 0.982 (0.15) 0.707 (0.10) 0.010 (0.00) 0.580 (0.03) 0.417 (0.01) 0.410 (0.07) 0.417 (0.01) 8.550 (3.46) 7.449 (0.96)

ξ≠1 0.678 (0.02) 0.982 (0.15) 0.707 (0.10) 0.093 (0.03) 0.580 (0.03) 0.417 (0.01) 0.410 (0.07) 0.417 (0.01) 3.109 (0.22) 7.449 (0.96)

ξ=1 ξ≠1 0.903 0.813 (0.04) (0.03) 0.971 0.971 (0.11) (0.11) 0.700 0.700 (0.08) (0.08) 0.008 0.029 (0.00) (0.01) 0.578 0.578 (0.02) (0.02) 0.417 0.417 (0.01) (0.01) 0.405 0.405 (0.06) (0.06) -0.417 -0.417 (0.01) (0.01) 10.339 5.353 (3.92) (0.90) 6.514 6.514 (0.98) (0.98)

Note: Standard errors based on a Newey-West covariance matrix robust to serial correlation up to 12 lags within .parenthesis. Column D reports the estimated duration of price stickiness, and J the Hansen test of the overidentifying restrictions (below in parenthesis we report the p-value). The set of instruments includes five lags of inflation deviation from its target, four lags of real marginal costs, detrended output, four lags of demeaned nominal monetary policy rate and three lags of detrended terms of trade.

28

Table 2 PR test for structural breaks in the Hybrid Phillips curve Sup PR1 Avg PR1 Cobb-Douglas PR test 136.37* 58.63* Estimated beak point date 2001:2 κ1= 0.816 D1= 2.68 Overhead labor PR test 267.18* 104.85* Estimated beakpoint date 2001:3 κ1= 0.922 D1= 2.58 CES PR test

84.59* 17.38** Estimated beakpoint date 2001:2 κ1= 0.729 D1= 3,96

CES for an open economy PR test 69.88* 26.88* Estimated beakpoint date 2001:2 κ1= 0.836 D1= 3,2

Exp PR1

Sup PR2

Avg PR2

Exp PR2

65.05*

79.09*

34.86*

36.44*

130.60*

260.63*

85.51*

127.18*

39.15*

46.66*

9.31

20.20*

32.09*

41.85*

13.06**

17.90*

Note: The table reports predictive tests for the null hypothesis of structural stability along with the estimated breakpoint date and the value of .κ and the estimated using Monte Carlo simulations with 10,000 replications *Indicates that the statistics is significant at the 1% level. **Indicates that the statistics is significant at the 5% level.

29

Table 3 Tradable Inflation Phillips Curve (1)

(2)

(3)

(4)

β

0.993 (0.01)***

0.906 (0.03)***

0.955 (0.02)***

0.966 (0.02)***

θM

0.806 (0.05)***

0.897 (0.01)***

0.848 (0.01)***

0.852 (0.01)***

λ

0.048 (0.03)*

0.022 (0.00)***

0.034 (0.01)***

0.030 (0.01)***

J-Test p-value

4.375 (0.99)

4.814 (0.99)

6.370 (0.99)

6.398 (0.99)

Sample

1991.1-2004.4

1991.1-2004.4

1991.1-2004.4

1991.1-2004.4

0.25

0.25

0.50

0.25

0.50

0.25

1041.2

1553.4

171.5

443.7

2000q1

1999q2

1998q4

1999q4

Labor share Elas. subt. laborforeign good SupPR test Estimated break date

30

Table 4 Monetary Policy Rules: Chile 1991-2004 Dependent variable: Monetary policy real interest rate

(1) ρ0

(2)

0.72 (0.07)***

ρ1

0.66 (0.05)***

ωπ,0

0.46 (0.25)*

ωπ,1

3.19 (0.93)***

ωy,0

0.86 (0.18)***

ωy,1

0.97 (0.42)**

ωq,0

0.31 (0.10)***

Period

1991.1-1997.4

1998.1-2003.4

0.79

0.90

2.80 (0.42)

1.97 (0.37)

R2 J-statistic p-value

*** significant at 1% level, ** significant at 5% level, * significant at 10% level. GMM estimation. Standard errors in parenthesis.

Table 5 PR test for Monetary Policy Rule Stability

PR test

Sup PR

Avg PR

Exp PR

174.48

71.27

83.94

Estimated beak point: 2000q1 Note: The PR tests were calculated using a Newey-West covariance matrix robust to serial correlation up to12 lags

31