Inflation Targeting and Disinflation Costs in Emerging Market Economies Martin Stojanovikj (Corresponding author), We thank the Editor and the two anonymous Referees for their helpful comments and suggestions, which have greatly improved the paper. Of course, the usual disclaimer applies. The corresponding author gratefully acknowledges the financial support from... Deusto Business School, University of Deusto, Postal address: Hermanos Aguirre Kalea 2, 48014 Bilbao, Spain E-mail: martin.stojanovik@deusto.es Goran Petrevski (Corresponding author), Ss. Cyril and Methodius University in Skopje, Faculty of Economics, Postal address: Krste Misirkov Blvd. 9V, 1000 Skopje E-mail: goran@eccf.ukim.edu.mk Abstract In this paper, we study whether adopting inflation targeting in emerging market economies affects the output costs of disinflation, controlling for a number of additional factors. Based on a sample of 40 emerging market economies during 1990-2017, we provide evidence that adopting inflation targeting is not associated with lower sacrifice ratios in emerging market economies. Specifically, we show that, controlling for the macroeconomic and institutional environment in EMEs, the choice of monetary regimes does not matter for disinflation costs. We also find that, when starting from low to moderate initial inflation, the speed of disinflation (shock therapy versus gradual disinflation) does not matter in these economies. Moreover, we show that trade openness is associated with lower sacrifice ratios, while we obtain opposite results for central bank independence. However, the impact of these factors on sacrifice ratios is rather small. Our main findings are robust to alternative classifications of the inflation targeting regime, alternative definitions of disinflation episodes, different peak levels of trend inflation rate, and across various specifications of the empirical model. Key words: Inflation targeting, Sacrifice ratio, Disinflation, Emerging market economies. JEL codes: E52, E58. 1. Introduction Standard macroeconomic theory argues that disinflation policy is costly in the sense that, based on the short-run Phillips curve, reducing inflation is usually associated with a decline in economic activity. Although there are various theoretical explanations of disinflation costs, most of them stem from the imperfect credibility of disinflation policies as well as the presence of nominal or real price and wage rigidities (for instance, see Romer 2001). The concept of sacrifice ratios provides a summary measure of the real costs of disinflation policy. Specifically, the sacrifice ratio represents a simple measure of the cumulative output loss for reducing inflation by one percentage point (Cecchetti and Rich 2001). During the past three decades, an increasing number of industrialized countries and emerging market economies (EMEs) have adopted inflation targeting (IT) as a monetary policy strategy. Within this regime, the central bank focuses on price stability as its predominant policy goal, by announcing explicit numerical targets for medium-term inflation. The proponents of this monetary policy strategy argue that it offers several benefits for EMEs: it enhances central bank credibility, it reduces inflation persistence, it helps anchoring inflation expectations, it contains a high degree of flexibility, it enables policy makers to cope with short-run circumstances and adverse shocks, and it involves lower economic costs in the face of policy failures (Bernanke and Mishkin 1997, Batini and Laxton 2007, Mishkin and Schmidt-Hebbel 2002, Mishkin and Schmidt-Hebbel 2007). As a result, the above features suggest that IT may be superior compared to alternative monetary policy strategies in the sense that it reduces the inflation-output volatility and leads to lower disinflation costs. This paper investigates the relationship between IT and sacrifice ratios in 40 EMEs during 1990-2017, controlling for a number of additional factors, which are referred to as common determinants of the sacrifice ratio in the empirical literature. In calculating sacrifice ratios, we employ a slightly modified version of the approach used in Ball (1994). The main findings from our study can be summarized as follows: first, we provide strong evidence that adopting IT is not associated with lower sacrifice ratios in EMEs; second, the choice of monetary regimes does not matter for disinflation costs in these countries; third, when starting from low to moderate initial inflation, the speed of disinflation (shock therapy versus gradual disinflation) does not impact on disinflation costs; fourth, we show that trade openness is associated with lower sacrifice ratios, while more independent central banks tend to increase sacrifice ratios, though both effects are not economically important. Our main findings are robust to alternative classifications of the IT regime, alternative definitions of disinflation episodes, different peak levels of trend inflation, and across various specifications of the empirical model. Yet, we find that the results are sensitive to the filtering method employed for estimating trend output. The rest of the paper is structured as follows: Section 2 reviews the relevant literature; Section 3 explains the issues related to the data collection process and the calculation of sacrifice ratios; Section 4 presents the results from the baseline specification of the empirical model accompanied by various robustness checks and a brief discussion of the main findings, while the conclusions and policy implications are summarized in Section 5. 2. Literature review The empirical literature on sacrifice ratios has been pioneered by Ball (1994), who proposed a simple measure of the sacrifice ratio, calculated by dividing the cumulated output loss by the change in trend inflation during each disinflation episode. This method, although far from being flawless, has been widely used in empirical research. In these regards, two additional variations of the original approach proposed by Ball (1994) have emerged in the subsequent literature. In an attempt to take account of the long-lasting effects accompanying each disinflation episode, Zhang (2005) proposes a slightly modified measure of the sacrifice ratio, relying upon the Hodrick-Prescott (1997) filter, and assuming that potential output grows throughout the particular episode at the rate implied by the Hodrick-Prescott filter at the beginning of the disinflationary episode. Further on, Hofstetter (2008) builds upon this approach with the additional assumption that output is at its trend level one year prior to the year the disinflation episode starts, thus trying to capture an even larger portion of the longer-lasting effects of disinflation. In addition to Ball’s procedure and its variants, there are other possible approaches to measure disinflation costs. For instance, one approach is based on the estimates of the Phillips curve, whose slope measures the inflation-output trade-off, i.e., the response of output to the changes in inflation (Andersen and Wascher 1999, Gordon et al. 1982, Gordon 2013, Hutchinson and Walsh 1998). An alternative method employs a structural VAR model to estimate the effects of monetary policy on output and inflation (based on the impulse responses of output and inflation to a monetary policy shock) and then the sacrifice ratio is calculated as a ratio between the cumulative impulse responses of output and inflation over some specified horizon (Cecchetti and Rich 2001, Belke and Böing 2014). Over time, a large body of empirical literature has emerged dealing with the determinants of sacrifice ratios in both advanced and developing countries. Some of the papers investigate the determinants of sacrifice ratios in general (Hofstetter 2008, Katayama et al. 2019, and Senda and Smith 2008), while others focus on various specific factors, such as central bank independence (Baltensperger and Kugler 2000, Brumm and Krashevski 2003, Daniels et al. 2005, Diana and Sidiropoulos 2004, Jordan 1997), trade openness (Bowdler 2009, Daniels and VanHoose 2006, 2009, 2013, Temple 2002), central bank transparency (Chortareas et al. 2003), political factors (Caporale 2011, Caporale and Caporale 2008), labor market institutions (Bowdler and Nunziata 2010, Daniels et al. 2006), and fiscal variables (Durham 2001). However, within this strand of empirical literature there is much less evidence on the relationship between IT and the sacrifice ratio, and this is especially true for EMEs. In what follows, we provide a brief overview of the main findings from these studies, while the details are given in Table A3 of the Appendix. The experience of OECD countries provides a unanimous conclusion that IT does not lead to lower sacrifice ratios. For instance, Debelle (1996) calculates the sacrifice ratios for New Zealand, Canada, and Australia, showing that, in all cases, sacrifice ratios in the post-1989 period are higher than before. Laubach and Posen (1997) study eight advanced countries and find that in all IT-countries sacrifice ratios were larger than the average from previous disinflation episodes. In contrast, the most recent sacrifice ratios in non-targeting countries were lower than those during previous disinflation episodes. Similarly, in their study of disinflation costs in 13 OECD countries, Almeida and Goodhart (1998) find that both IT-countries and non-targeters had higher sacrifice ratios in the 1990s than in the 1980s. For the 1990s, the simple comparison between IT-countries and non-targeters (without providing formal econometric evidence) shows that the former countries have higher sacrifice ratios measured in unemployment terms. In terms of output, the sacrifice ratio is lower in IT-countries than in non-targeters, but this is true for both the 1980s and 1990s. However, if the UK is excluded from the sample, then the sacrifice ratios do not differ across IT-countries and non-targeters. Based on the estimates from the short-run Phillips curves for 21 OECD countries, Chortareas et al. (2003), too, provide evidence that IT does not affect sacrifice ratios. Also, the often-cited study of Ball and Sheridan (2004) shows that this monetary regime appears to have had no significant effect on the disinflation process in 20 OECD countries. Finally, Roux and Hofstetter (2014) show that IT reduces sacrifice ratios only if disinflation is slow (lasting more than three years), while it is irrelevant in fast disinflations. It seems that Gonçalves and Carvalho (2009) is the only study suggesting that adopting IT in OECD countries makes disinflation policy less costly. However, Brito (2010) criticizes their methodological approach that compares disinflations in inflation-targeting countries with non-simultaneous disinflations in non-targeting countries which occurred under different macroeconomic conditions. Therefore, controlling for the common trends in economic conditions, he obtains opposite results, i.e., disinflation costs under the IT regime are even larger compared to non-targeters. Empirical literature utilizing mixed samples with both advanced and developing countries also fails to provide firm evidence on the effectiveness of IT in reducing disinflation costs. For instance, Cecchetti and Ehrmann (2002) find that the average cross-country sacrifice ratio of inflation targeters is larger than the average sacrifice ratio in non-targeting countries though within both groups of countries the estimated sacrifice ratios vary considerably. Following the same approach and a similar sample, Corbo et al. (2002) show that IT countries had similar GDP-based sacrifice ratios before and after the adoption of IT. Gonçalves and Carvalho (2008) examine the effect of IT on sacrifice ratios in both OECD and developing countries. Their study shows that, generally, IT is associated with lower sacrifice ratio with the favorable effects being stronger in OECD countries, i.e., the results for the developing countries are not robust to the model specification. Tunali (2008) studies how various features of the IT regime affect the sacrifice ratio and obtain mixed results: the aggregate IT index lowers the sacrifice ratios only for the whole sample, but not for the two sub-samples; in addition, he finds that none of the specific features of IT (policy focus, accountability, instrument independence, financing government deficits, the use of forecasting and simulation methods etc.) are not statistically significant. Working with a large sample of 189 countries, Mazumder (2014) finds that the effects of IT are statistically insignificant in both OECD and developing countries. On methodological grounds, the study in Ardakani et al. (2018) differs sharply from the rest of the empirical literature by estimating the average treatment effects of IT based on the propensity score methodology. However, they fail to provide convincing evidence on the potential benefits of IT as a disinflation device, i.e., their results appear to be very sensitive to the specific estimation technique employed: the estimates form the semiparametric propensity score matching method imply that IT lowers the sacrifice ratios in developed economies only, and not in EMEs; yet, when employing the nonparametric series propensity score model, they find that IT reduces the sacrifice ratio in the full sample, but not in the individual sub-samples; finally, in the parametric propensity score method the effects of IT are insignificant in the full sample as well as in the two sub-samples. Magkonis and Zekente (2020) estimate the sacrifice ratios for a panel of 42 countries employing the Bayesian model averaging methodology. They find IT does not affect the sacrifice ratios, and this is true for both OECD and non-OECD countries. On the other hand, Gonçalves and Carvalho (2008) show that IT is generally associated with lower sacrifice ratios with the beneficial effects being larger in OECD countries. Recently, Sethi and Acharya (2019) study whether IT enhanced central bank credibility in 13 Asian countries, finding that IT is associated with lower sacrifice ratios. Finally, the empirical research of the effects of IT on sacrifice ratios focusing exclusively on EMEs is very scarce. To our best knowledge, Brito and Bystedt (2010) is the only study within this strand. In this respect, employing the Philips curve-based approach for a sample of EMEs, they provide evidence that the effects of IT are not statistically significant, thus failing to provide empirical support to the presumed beneficial effects of IT in EMEs. Therefore, summarizing the literature review, one can potentially conclude that although the main features of the IT regime imply that it could be a less costly device for controlling inflation, the empirical support to this proposition seems to be rather weak. 3. Data and methodology 3.1. Calculation of sacrifice ratios The sacrifice ratio provides an approximation of the real output costs associated with disinflation. As mentioned above, there are several approaches to measure disinflation costs. For instance, Hutchison and Walsh (1998) measure the sacrifice ratio based on the Phillips curve. One issue with this particular method is that it limits the inflation-output trade-off to be equal during the periods of accelerating inflations and the periods of disinflation, which is not the case for our sample of countries where different factors are responsible for the movement of inflation in opposite directions. Cecchetti and Rich (2001) employ three different structural Vector Autoregression (VAR) models to calculate the sacrifice ratio. On the other hand, Ball (1994) relies on trend inflation to identify peaks and troughs, i.e., the period from a particular peak to trough is identified as a disinflation episode. Furthermore, given the tendency of the Hodrick-Prescott filter to minimize the deviations of actual output from the trend, Ball (1994) proposes a log-linear method of deriving these deviations. Summing them for all the years comprising a particular disinflation episode and dividing by the change in trend inflation during the same period results in the sacrifice ratio. This method, although not unsusceptible to criticism, has been widely used in empirical literature. In our study, we follow Ball (1994) in calculating the sacrifice ratio with a slight modification, which reflects the specific features of disinflation episodes in EMEs, and potentially increases the computational power and accuracy within our sample. Working with annual data, particularly for trend inflation, provides a challenge in adopting the original approach of Ball (1994) who calculates the sacrifice ratios based quarterly data of inflation, which are not available for the majority of EMEs in our sample. Therefore, we follow Mazumder (2014) in using annual data for inflation, and then calculating trend inflation as a 3-year centered moving average. From an economic standpoint, this methodological tweak suits our analysis, since it provides a theoretically adequate setup for estimating the effect of adopting IT on the sacrifice ratio resting on the premise that inflation targets are implemented in the medium-term of two to three years (Hammond 2012). In this way, we allow for the central bank behavior in each period to be both forward and backward looking and more adapted to the focal period of the central bank policy actions (the medium-term). In addition, Mazumder (2014) provides evidence of the proximity of the sacrifice ratio measurements from this approach to the original sacrifice ratios in Ball (1994). In addition, we follow Ball (1994) and identify disinflation episodes as periods in which trend inflation rate falls from peak to trough for at least two percentage points. Subsequently, a peak occurs in the year in which the trend inflation rate is higher than the rate in both the preceding and succeeding years, while a trough occurs in the year in which the trend inflation is lower than the rate in both the preceding and succeeding years. Using this methodology, we have identified 86 disinflation episodes in 40 EMEs during 1990-2017. However, the distribution of disinflation episodes across monetary regimes is largely unbalanced with only 16 such episodes within the IT regime, which is due to the fact that this monetary strategy did not exist before the 1990s. The comparison with Mishkin and Savastano (2002), Hofstetter (2008), and Mazumder (2014) shows that we are able to identify the majority of disinflation episodes across the same countries in all three datasets. Sacrifice ratios are calculated as the sum of the deviations in actual output from trend output divided by the change in trend inflation rate over each disinflation episode. Ball (1994) estimates trend output based on the following three assumptions: 1) output is at its trend level at the start of a disinflation episode, 2) output is at its trend level four quarters after an inflation trough, and 3) output grows log-linearly between these two points when trend and actual output are equal. We identify our trend output level following the first two assumptions. We suspect that the approximation of a log-linear growth rate in trend output between these two points provides smaller output values throughout the disinflation period for the sample of EMEs, thus producing a downward bias on the estimates. This is indeed the case, i.e., the estimates obtained by the original Ball’s method always exhibit smaller magnitude, although, with respect to the statistical significance of the coefficients, the results are identical to our baseline estimates. Empirical literature points to the well-known stylized fact that output volatility in EMEs is much higher compared to OECD countries (Ramey and Ramey 1995, Blanchard and Simon 2001, Kose et al. 2003a and 2003b). A logarithmic transformation of the sort provided by Ball (1994) tends to offer a more adequate approximation of smaller values, while potentially underestimating the size of the output gap in our sample characterized with higher output volatility. Hence, we measure the sacrifice ratio by calculating the output gap for each year during a disinflation episode as the difference between trend and actual output over trend output, instead of working with log-levels of output. Although it potentially improves the accuracy of our calculations of the sacrifice ratio, the results provided in Section 4 show that this modification does not alter the results in any significant way. However, the sacrifice ratios calculated by the aforementioned modification and the original Ball’s method are not robust when experimenting with Hodrick-Prescott and Hamilton (2018) filters for estimating trend output. This finding is not surprising since it is known that the estimates of sacrifice ratios are sensitive to the specific method applied in the estimation of potential output. In these regards, a number of papers demonstrate that the magnitude of sacrifice ratios differs considerably when trend output is estimated by alternative methods (see Temple 2002, Senda and Smith 2008, and Mazumder 2014). 3.2. Data issues Due to data availability, we work with annual data and our sample covers 40 EMEs during 1990-2017. Specifically, the sample consists of the following countries: Algeria, Botswana, Brazil, Chile, China, Colombia, Croatia, Costa Rica, Cote d’Ivoire, Czech Republic, Dominican Republic, Ecuador, Egypt, El Salvador, Ghana, Guatemala, Hungary, India, Indonesia, Israel, Jordan, Lebanon, Malaysia, Mexico, Morocco, Nigeria, Pakistan, Philippines, Poland, Russia, Serbia, Singapore, South Africa, South Korea, Tanzania, Thailand, Tunisia, Turkey, Ukraine, and Uruguay. As mentioned above, our sample consists of 86 disinflation episodes in total (for the details, see Table 2 in the Appendix) of which 16 occurred under the IT regime. However, in our baseline estimations we work with a restricted sample comprising only disinflation episodes that meet two criteria: at least 2 percentage points change in trend inflation, and with peak inflation below 20% and 30%, respectively. Based on these two cumulative criteria, our effective sample is reduced to 77 disinflation episodes with peak inflation below 30%, and 64 episodes with peak inflation below 20%. In both cases, there are 16 disinflation episodes under the IT regime. With respect to the monetary policy strategies, according to the IMF’s official classification of exchange arrangements and monetary policy frameworks (IMF 2018), 55% of our whole sample consists of inflation targeters (22 countries). The rest of the sample is quite heterogenous with 4 countries implementing monetary targeting, 7 exchange-rate targeters (conventional peg, crawling peg, and stabilized arrangements), 5 countries conduct monetary policy without explicit nominal anchors (multiple indicator-based approach), while 2 countries do not have separate legal tender (official dollarization). However, this classification should be considered rather tentative as most of the countries with multiple indicator-based monetary policy simultaneously have de facto exchange rate targets. Historically, most of the countries within our sample have experimented either with some variant of exchange rate targets (e.g., hard pegs, crawling pegs, exchange rate bands) or with monetary targeting (Cobham and Dibeh 2009, Corbo 2002, Kuttner and Posen 2001). Table 1 presents the basic descriptive statistics of the main variables included in the empirical model. As can be seen, inflation targeters and non-targeters differ sharply in terms of sacrifice ratios: while the average sacrifice ratio in the former group is 1.606, its mean value is only 0.189 in the latter group. Undoubtedly, this stark difference reflects the distribution of disinflation episodes within our sample: all disinflation episodes in IT-countries occurred starting from low to moderate initial inflation rates, whereas a number of disinflation episodes in the non-targeting countries started from moderate to high initial inflation rates. This is confirmed by the fact that both the mean initial inflation rate and the mean change in trend-inflation in non-targeters is much higher compared to IT-countries. It is well-known that disinflation costs differ widely when starting form high inflation versus disinflation from low or moderate inflation rates. Specifically, one would expect slowdowns in economic activity when bringing inflation down from low or moderate inflation but not so when stabilizing inflation from high inflation rates. Since the whole sample comprises disinflation episodes starting from high initial inflation rates, working with it could underestimate the sacrifice ratios. On the other hand, most of the disinflation episodes under the IT regime start from low or medium inflation rates, thus, being associated with larger sacrifice ratios. The combination of these two reasons might create a bias in the regression coefficient of IT. Because of this heterogeneity within our sample, we do not work with all the available data. Instead, we discard all the cases under high-inflation environment so that our empirical investigation is based on disinflation episodes starting with initial inflation rates below 20%-30%. On the other hand, one cannot observe any difference between the two groups of countries with respect to the length of disinflation episodes. This is more or less true for the control variables, i.e., IT-countries and non-targeters are very similar with respect to trade openness, central bank independence (CBI), debt-to-GDP ratios, and the exposure to foreign shocks. We consider these observations very important for our empirical analysis as they suggest that the differences in sacrifice ratios between inflation targeters and non-targeters are not a by-product of the differences in institutional and macroeconomic environment. Table 1. Descriptive statistics, 1990-2017 Inflation targeters Sacrifice ratio Change Length Inflation Open CBI Shocks Debt Mean 1.606 4.363 5.063 7.229 76.972 0.657 7.600 42.793 Standard deviation 2.583 1.797 1.389 3.356 34.948 0.224 9.196 23.501 Minimum -4.471 2.093 4.000 2.637 27.310 0.255 -12.381 2.150 Maximum 8.047 7.772 8.000 15.502 167.669 0.899 22.331 92.447 Non-targeters Mean 0.189 37.183 5.943 42.690 73.062 0.553 7.193 58.216 Standard deviation 2.583 196.012 3.054 196.062 43.457 0.197 10.041 32.877 Minimum -11.373 2.043 2.000 2.626 17.428 0.122 -12.381 7.690 Maximum 11.293 1646.750 16.000 1651.740 326.679 0.904 28.801 166.670 Note: Sacrifice ratio, Change, Initial inflation, Open, Shocks, and Debt are expressed in percentage points; Length is expressed in years; CBI is expressed in decimal points on a scale from 0 to 1. In the regression model, IT is measured as a binary variable (equal to 1 if country i is an inflation targeter in period t, and 0 otherwise). Here, we rely on the classification provided by Hammond (2012) in selecting the year in which a particular country decided to adopt this monetary strategy, notwithstanding the actual month in which the decision was officially implemented. The official dates of IT adoption are presented in Table A1 in the Appendix. Calvo and Mishkin (2003) argue that, when faced with weak fiscal and financial institutions, low credibility of monetary institutions, currency substitution and liability dollarization, and vulnerability to sudden stops of capital inflows, EMEs need a strong commitment to IT in order to exploit its benefits. In this regard, having a clear institutional commitment to the inflation target along with transparent and accountable monetary framework is what distinguishes fully-fledged targeters from the other countries adopting this regime. Hence, in the baseline model, we rely on the implementation of full-fledged IT, while in the section containing the robustness checks we also provide estimates based on the more flexible definition of the IT regime. In order to calculate the index of CBI, we rely on the dataset in Garriga (2016). The weighted CBI-index is more suitable for our analysis as it covers all the countries in our sample for the entire time period. The political variable included in the regression model is a categorical variable taking the values of 0, 1 or 2, referring to the governments led by left-wing parties, centrist governments, and right-wing governments, respectively. The data for this variable has been taken from Beck et al. (2001), while all the remaining data have been extracted from the World Bank World Development Indicators. 3.3. Model specification The baseline specification of our model is as follows: (1), where: is the sacrifice ratio for a particular disinflation episode; is the regression intercept; is a dummy variable for the IT regime whose values equal 1 if the country i is an inflation targeter during a particular disinflation episode, and 0 otherwise; are the control variables; while is the disturbance, following the normality assumption, . We estimate the specification by Ordinary Least Squares (OLS), using robust standard errors, which are consistent in the potential presence of heteroskedasticity and serial correlation. In order to control for the potential factors affecting sacrifice ratios, our regression model includes several variables, commonly present in empirical studies, such as: trade openness, domestic and external shocks, and political factors (Broner and Ventura 2006, Kraay and Ventura 2007, Loayza and Raddatz 2007). Ball (1994) identifies speed of disinflation (Speed) as the main determinant of sacrifice ratios. We measure Speed by the change in trend inflation over the duration of an inflation episode. The sign and the significance of the regression coefficient of this variable theoretically differ between the sharp regime-shift approach and the gradualist approach to disinflation: on the one hand, the traditional view is that gradual disinflation is less costly because it allows for an adjustment of wages and prices (Taylor 1983); on the other hand, Sargent (1983) argues that sharp shifts in monetary regimes decrease disinflation costs due to the enhanced policy credibility and the accompanying quick adjustment of inflation expectations. Therefore, the expected sign of this variable is ambiguous. Further on, we follow Ball (1994) in decomposing Speed into the change in trend inflation during a particular episode (Change) and the duration of each disinflation episode (Length). For OECD countries, the empirical studies generally confirm that the faster the disinflation process, the lower the output costs, i.e., credible disinflations are associated with smaller output losses (Ball 1994, Boschen and Weiss 2001, Diana and Sidiropoulos 2004, Daniels et al. 2005, Zhang 2005, Hofstetter 2008, Daniels and VanHoose 2009 and 2013, Gonçalves and Carvalho 2009, Mazumder 2014, Roux and Hofstetter 2014, and Katayama et al. 2019). On the other hand, Andersen and Wascher (1999) show that the speed of disinflation does not matter in OECD countries; Mazumder (2014) confirms this conclusion for non-OECD countries; Gonçalves and Carvalho (2008) obtain similar results for both developed and developing countries; while Caporale (2011) finds that the speed of disinflation may, in fact, increase sacrifice ratios. A priori, the association between CBI and disinflation costs is ambiguous. On the one hand, more independent central banks are capable of anchoring inflation expectations, which reduces disinflation costs. On the other hand, higher CBI should be associated with lower average inflation and lower inflation variability, which leads to less frequent price and wage adjustments, thus making disinflation costlier due to the flatter Phillips curve (Walsh 1995). The early empirical evidence (Debelle and Fisher 1993, Jordan 1997, Posen 1998) suggests that higher CBI increases the output loss of disinflation (the so-called “Credibility-Sacrifice Ratio Puzzle”). Some recent empirical studies (Baltensperger and Kugler 2000, Brumm and Krashevski 2003, Diana and Sidiropoulos 2004, and Mazumder 2014) find that CBI reduces output losses of disinflation in OECD countries, whereas Daniels et al. (2005), Daniels and VanHoose (2009, 2013) provide opposite evidence. The rationale for including trade openness (Open) in the specification is due to Romer (1993), who argues that openness affects the output-inflation trade-off by making the Phillips curve steeper. Specifically, in open economies, monetary contraction exerts a larger direct pressure on the price level through the exchange-rate appreciation. Consequently, following a negative monetary shock, the decline in inflation is larger in open economies, leading to lower sacrifice ratios. On the other hand, Daniels and VanHoose (2006) show that, in the presence of nominal rigidities, openness may increase the sacrifice ratio. As for the empirical evidence, Ball (1994), Temple (2002), Daniels et al. (2005), and Daniels and VanHoose (2013) find a negative or an insignificant relationship between openness and sacrifice ratio in OECD countries, while Mazumder (2014) provides support for this result in developing countries. Daniels and VanHoose (2009) provide evidence on the positive association between openness and sacrifice ratios. In small open economies, foreign shocks (Shocks) may have strong effects on domestic inflation, thus making the calculation of the sacrifice ratio biased. For instance, during a particular disinflation episode, a favorable supply shock may lead to a larger decline in inflation for a given monetary policy action. In this case, the reduced sacrifice ratio is partially caused by the favorable supply shock and not by monetary policy. Ball (1994) estimates the sacrifice ratio employing a method which smooths out supply shocks, identifying them as part of the error term. Nevertheless, he performs two types of estimations to check for supply shocks and fails to any bias in the results. As our sample consists of EMEs, we consider foreign shocks as an important determinant of the sacrifice ratio. It is well-known that these countries operate under unfavorable macroeconomic environment compared to the advanced economies, i.e., they are frequently exposed to adverse external shocks, which result in large inflation-output volatility (Fraga et al. 2003). Therefore, we control for the impact of foreign shocks by including the changes in oil prices as a regressor in the baseline specification. Andersen and Wascher (1999), Boschen and Weiss (2001) and Hofstetter (2008), too, include oil prices among the determinants of the sacrifice ratio, serving as a proxy variable that controls for the external shocks. Yet, the empirical evidence provided in these papers seems to be rather inconclusive. Initial inflation (Inflation) is included as a standard determinant of the sacrifice ratio. Both Lucas (1973) and Ball et al. (1988) show that trend inflation influences the output-inflation trade-off in the sense that higher inflation reduces the degree of downward nominal rigidity and thus steepens the Phillips curve. Therefore, higher initial inflation should be associated with lower disinflation costs. Ball (1994), Andersen and Wascher (1999), Temple (2002), Zhang (2005), Hofstetter (2008), Gonçalves and Carvalho (2008, 2009), and Mazumder (2014) all provide empirical support to the aforementioned proposition for both developed and developing countries. On the other hand, Hofstetter (2008), Daniels and VanHoose (2009), and Caporale (2011) fail to confirm these findings. Nominal wage rigidity is another control variable in our regression model. As mentioned above, in New Keynesian models, nominal rigidities exert critical influence on the inflation-output trade-off, the speed of disinflation, and the sacrifice ratio, respectively. In these regards, initial inflation may serve as a proxy for the extent of nominal rigidity since wage contracts tend to be shorter at high inflation rates. Due to the lack of available labor market data for EMEs, we adopt the convention put forth by Hofstetter (2008), who uses the 10-year inflation history as a proxy for nominal wage rigidity, expecting a negative association between this variable (Rigidity) and the sacrifice ratio. In principle, one should expect to find a positive relationship between government debt (Debt) and the sacrifice ratio. Indeed, Durham (2001), Brito (2010) and Roux and Hofstetter (2014) provide empirical evidence that supports this proposition for the advanced economies, though Durham (2001) and Mazumder (2014) find opposite results for the developing countries. Finally, following Caporale and Caporale (2008) and Caporale (2011), in order to control for the effects of political factors, we include a dummy variable (Party), which takes three values: zero if the government is controlled by left-wing parties, one in case of a centrist government, and two for right-wing parties. As right-wing governments are known to be inflation-averse, the expected coefficient in front of this variable is expected to be negative. 4. Estimation results and discussion 4.1. Estimation results As explained above, we estimate equation (1) based on the data meeting these two criteria: first, we consider only those disinflation episodes in which trend inflation has been reduced by at least 2 percentage points; and second, we discard all the disinflation episodes starting from high inflation rates, i.e., we retain only the disinflation episodes with peak inflation rates below 20% (30%). For instance, Dornbusch and Fischer (1993) and Burton and Fischer (1998) define episodes of moderate inflation as periods with inflation rates in the range of 15%-30%, lasting at least three years. These estimates, which should be viewed as our baseline results, are presented in Table 2. Specifically, regression estimates obtained from disinflation episodes with peak inflation below 20% are presented in columns 1-3, while the estimates from disinflation episodes with peak inflation below 30% are shown in columns 4-6. Table 2. Determinants of sacrifice ratios in EMEs, 1990-2017 (1) (2) (3) (4) (5) (6) (7) IT 0.015** (0.007) 0.016** (0.007) 0.015** (0.007) 0.016** (0.007) 0.016** (0.007) 0.015** (0.007) 0.011* (0.006) Change -0.059 (0.08) - - -0.024 (0.05) - - -0.096 (0.111) Length -0.002 (0.002) -0.002 (0.002) -0.002 (0.002) -0.002* (0.001) -0.002** (0.001) -0.002* (0.001) 0.001 (0.003) Inflation - - -0.066 (0.046) - - -0.039 (0.029) - CBI 0.042** (0.02) 0.038* (0.02) 0.042** (0.02) 0.031* (0.02) 0.028 (0.02) 0.035* (0.02) 0.045* (0.028) Party -0.005* (0.003) -0.004 (0.003) -0.005* (0.003) -0.004* (0.002) -0.004* (0.002) -0.005* (0.003) -0.008* (0.004) Open -0.029*** (0.008) -0.028*** (0.008) -0.031*** (0.008) -0.027*** (0.008) -0.027*** (0.008) -0.028*** (0.008) -0.017 (0.014) Shocks -0.001** (0.0005) -0.001** (0.0005) -0.003** (0.001) -0.001* (0.0006) -0.001* (0.0006) -0.002** (0.0001) -0.0003* (0.0002) Constant 0.031** (0.01) 0.029** (0.01) 0.036*** (0.013) 0.028** (0.01) 0.027** (0.01) 0.032*** (0.011) 0.019 (0.015) Sample size 52 52 52 63 63 63 39 Adjusted 0.35 0.36 0.36 0.30 0.31 0.31 0.06 Notes: OLS estimates with HAC standard errors in parentheses; ***, **, and * indicate 1%, 5% and 10% level of significance, respectively. As can be seen, in all cases, the effect of IT on the sacrifice ratio is positive and statistically significant. Yet, it can be observed that the magnitude of the regression coefficient of IT is very low, implying that the difference in sacrifice ratios between IT-countries and non-targeters are not economically important. In other words, controlling for the macroeconomic and institutional factors in EMEs, the choice of monetary regimes does not matter for disinflation costs. In these regards, the descriptive statistics presented in Table 1 reveals that these economies operate within more or less similar macroeconomic and institutional environment. Therefore, we believe that our main finding of the irrelevance of monetary strategies for disinflation costs is not surprising at all. On the other hand, an alternative interpretation of the above result is possible as it challenges the hypothesized advantage of IT as a less costly disinflation tool in comparison with the alternative monetary policy strategies. As regards the debate on shock therapy versus gradual disinflation, we find negative association between the difference in trend inflation during a disinflationary episode (Change) and sacrifice ratios, but the regression coefficient is both statistically insignificant and with low magnitude. We obtain the same results for the other two variables that are seen as important factors of disinflation costs – initial inflation (Inflation) and the duration of disinflation episode (Length): the former coefficient is negative though insignificant, whereas the latter is also negative, but it is virtually equal to zero and insignificant within the sub-sample consisting of disinflation episodes with peak inflation below 20% (columns 1-3). In sum, we cannot provide empirical evidence that the speed of disinflation affects disinflation costs when disinflation episodes start from low to moderate initial inflation. Considering the control variables, the positive and highly significant coefficient of CBI implies that disinflation seems to be more costly in EMEs with more independent central banks. This result reaffirms previous claims in the empirical literature put forth by Debelle and Fisher (1994), Jordan (1997), and Posen (1998), among others. According to Walsh (1995), with more independent central banks, nominal contracts tend to be longer reflecting the lower inflation expectations. As a result, higher CBI both shifts the Phillips curve inwards and makes it flatter, which increases the sacrifice ratio. Further on, the negative coefficient of Open suggests that EMEs which are more open to international trade experience lower disinflation costs though the magnitude of this effect is rather low. As for the adverse external shocks, as proxied by the changes in oil prices, although the coefficient of Shocks is negative and statistically significant, its magnitude is very low (virtually zero) which prevents us from concluding that external shocks affect sacrifice ratios. Similarly, we find negative association between Party and sacrifice ratios, but the regression coefficient is virtually equal to zero, implying that we cannot provide firm evidence on the impact of political ideology on disinflation costs. One source of criticism to the estimates presented in Table 2 is related to the composition of our sample, which consists of only 16 disinflation episodes under the IT regime so that it is dominated by the disinflation episodes under the alternative monetary policy strategies. As a result, one may suspect that the above results simply reflect the underrepresentation of disinflation episodes under the IT regime. Therefore, as a final safeguard against the possibility of biased estimates, we proceed by re-estimating equation (1) based on a sample in which the number of disinflation episodes are balanced across the IT-countries and non-targeters. To this end, we follow the commonly employed procedure in the empirical literature by excluding the non-targeting countries that are either poorer than the poorest IT-country (Ghana) or smaller than the smallest IT-country (Israel) within our sample. These estimates are presented in the last column of Table 2. Unsurprisingly, both the overall goodness-of-fit and the precision of regression parameters have been largely reduced due to the low sample. However, the main findings from our exercise remain basically unaffected. In particular, we are able to confirm the positive association between IT and sacrifice ratios, although the coefficient of IT is now significant at 10%. Also, despite the reduced sample, the rest of regression estimates remain more or less unchanged. Therefore, given the stability of our main findings, we are confident that they are not a mere artefact of the relatively low number of IT disinflation episodes within the sample. 4.2. Robustness analysis The main finding from Table 2 is that adopting IT does not have any material impact on disinflation costs in EMEs. In this section, we check the robustness of this conclusion by including additional control variables commonly found in the empirical literature, which could possibly add to the explanatory power of the regression model. Table 3. Robustness checks, 1990-2017 (1) (2) (3) (4) (5) (6) IT 0.010* (0.006) 0.014** (0.0064) 0.014** (0.0064) 0.016** (0.0069) 0.013* (0.0071) 0.017** (0.008) Change -0.025 (0.045) -0.043 (0.0445) -0.020 (0.0222) -0.023 (0.048) -0.006 (0.049) -0.022 (0.063) Length -0.002 (0.001) -0.001 (0.001) -0.002* (0.0010) -0.002* (0.0012) -0.002* (0.001) -0.002* (0.001) CBI 0.030* (0.018) 0.035** (0.0168) 0.033** (0.0159) 0.031* (0.018) 0.029 (0.0196) 0.032* (0.018) Party -0.004* (0.0025) -0.005* (0.0027) -0.005* (0.0159) -0.004* (0.0024) -0.006** (0.0027) -0.005* (0.0017) Open -0.028*** (0.009) -0.023** (0.0093) -0.022** (0.0092) -0.027*** (0.0088) -0.028*** (0.0084) -0.028*** (0.0091) Shocks -0.001 (0.009) 0.023 (0.0237) 0.018 (0.0217) -0.002* (0.0009) -0.002* (0.001) -0.001 (0.009) Debt - - - -0.002 (0.0133) - - Rigidity - - - - 0.002 (0.0014) - Asian crisis - - - - - 0.014 (0.011) Constant 0.027** (0.012) 0.012 (0.0108) 0.013 (0.0104) 0.029** (0.0133) 0.031** (0.011) 0.029** (0.01) Sample size 63 71 74 63 58 63 Adjusted 0.260 0.274 0.274 0.290 0.317 0.303 Notes: Sacrifice ratio is the dependent variable; OLS estimates with HAC standard errors in parentheses; ***, **, and * indicate 1%, 5% and 10% level of significance, respectively. Following Bernanke et al. (1999), in column (1) of Table 3, we estimate the same specification as before (working with maximum initial inflation of 30%), while allowing for a less formal implementation of the IT regime (the so-called soft IT). As can be seen, notwithstanding the actual definition, once again we obtain a positive and statistically significant association between IT and sacrifice ratios with a similar magnitude of the regression coefficient. In other words, according to these results, the actual form of implementing the IT regime (full-fledged versus lite-IT), does not influence our main findings. Compared to the advanced economies, disinflation episodes in EMEs are characterized by higher trend inflation. Therefore, we check the sensitivity of our results to alternative peak levels of trend inflation. For instance, Fischer et al. (2002) term an inflation episode involving inflation rates in the range of 25%-50% as moderate to high inflation, while Bruno and Easterly (1998) identify all two-year periods with inflation rates beyond 40% as inflation crises. In columns (2) and (3), we set the threshold for peak inflation at the start of disinflation episodes at 40% and 50%, respectively. As can be seen, working with different thresholds for peak trend-inflation does not affect our main findings, i.e., the association between IT and the sacrifice ratio remains remarkably stable. In columns (4) and (5), we proceed by augmenting the baseline specification with additional control variables commonly employed in the empirical literature, while retaining our initial measure of IT, i.e., considering only full-fledged IT regimes. In column (4), we control for the fiscal position as proxied by the debt/GDP ratio. As can be seen, the coefficient of Debt is not statistically significant. Regarding nominal rigidities, we find a positive and significant effect, but the magnitude of regression coefficient is negligible (column 5). Since the data on nominal wage rigidity is not available for EMEs, we employ the proxy variable used in Hofstetter (2008) and Mazumder (2014), namely the 10-year inflation history prior to each disinflation episode. The rationale behind this approach is that the countries with a long history of high inflation tend to develop some mechanisms, allowing prices and wages to adjust frequently (for instance, indexation of nominal contracts). Therefore, these countries are characterized by a lower degree of nominal rigidities. However, the regression coefficient of Rigidity is counterintuitive, i.e., contrary to both theory and empirical evidence, thus potentially raising concerns with respect to the validity of this proxy variable. At the same time, an alternative interpretation of the above finding is possible: the long history of high inflation in EMEs is associated with deeply rooted inflation expectations and low credibility of disinflation policies, which results in high sacrifice ratios. In the last column of Table 3, we present the estimates from the regression model augmented by a dummy variable for the Asian financial crisis, which had major macroeconomic repercussions not only for the Asian economies included in our sample (Indonesia, the Philippines, South Korea, and Thailand), but also had spillover effects to Latin America (Brazil, Chile, Colombia, Mexico, and Peru). The inclusion of Asian crisis in our empirical model might be warranted for two reasons: on the one hand, the above-mentioned economies experienced negative growth rates during the Asian crisis, while on the on the other hand, a number of countries adopted the IT regime in the aftermath of the Asian crisis (such as Brazil, Colombia, Mexico, and South Korea). As a result, the omission of this variable could potentially affect both the calculation of sacrifice ratios and the estimate of the regression coefficient of IT. As can be seen in column (6), the dummy variable has the expected positive sign, but it is not statistically significant. More importantly, its inclusion in the empirical model does not have any material impact on the regression estimates, which remain comparable to those already obtained. In other words, once again, we are able to confirm the positive association between the IT regime and sacrifice ratios in EMEs. Since the formula for calculating the sacrifice ratio includes the deviations from potential output in the numerator, this procedure is potentially vulnerable to any errors in the estimates of potential output, which would be reflected in the magnitude of sacrifice ratios. Table 4 provides evidence on the sensitivity of our results with respect to the filter used in detrending output once again working with disinflation episodes up to 30% initial inflation. In column 1 we employ strictly the original method of Ball (1994) for calculating the sacrifice ratio. By comparing the results with those reported previously, we cannot detect any fundamental differences. Again, we find a positive association between sacrifice ratios and adopting the IT regime in EMEs though the magnitude of the coefficient of IT is now much smaller. Hence, we can confirm our previous finding that disinflation costs in EMEs do not vary with the monetary regime being in practice. We proceed checking the potential sensitivity of our results to the calculation of potential output by employing the Hodrick-Prescott filter (column 2) and the Hamilton filter (column 3). Ball (1994) argues that the Hodrick-Prescott filter has the tendency of minimizing the deviations between trend and cyclical output, thus exerting a downward bias on the values of sacrifice ratios. Indeed, we confirm that the effect of IT on sacrifice ratios is sensitive to the method for measuring the output gap. Specifically, when potential output is measured by applying the Hodrick-Prescott filter, the coefficient of IT loses its statistical significance. Also, when applying the Hamilton-filter, the magnitude of the regression coefficient of IT becomes much higher. Finally, note that the adjusted R-squared deteriorates sharply as we move from left to right across Table 4, supporting the findings in Ball (1994), Temple (2002), and Mazumder (2014) that the alternative methods for calculating the sacrifice ratio may be less accurate. Table 4. Robustness checks with alternative methods for estimating trend output (1) Ball (2) Hodrick-Prescott (3) Hamilton IT 0.007** (0.003) 0.016 (0.012) 0.015* (0.0089) Change -0.011 (0.0205) -0.071 (0.111) -0.045 (0.064) Length -0.0008* (0.0005) -0.0009* (0.0006) -0.0005 (0.0011) CBI 0.013* (0.0079) 0.024 (0.033) 0.032* (0.023) Party -0.002* (0.001) -0.005* (0.003) -0.003 (0.0038) Open -0.012*** (0.0036) 0.0006 (0.0102) -0.021** (0.008) Shocks -0.007* (0.004) -0.009*** (0.002) -0.002* (0.001) Constant 0.012** (0.0047) 0.002 (0.0172) 0.018 (0.016) Sample size 63 63 62 Adjusted 0.286 0.195 0.072 Notes: Sacrifice ratio is the dependent variable; OLS estimates with HAC standard errors in parentheses; ***, **, and * indicate 1%, 5% and 10% level of significance, respectively. Given that estimate equation (1) by OLS, the validity of regression estimates rests on the assumption of exogeneity of the right-hand side variables. However, since the implementation of IT is a deliberate decision made by policy makers, one may argue that it is the high cost of disinflation that motivates central banks to adopt IT. As a result, the regression results may not represent the true causal effect of IT on sacrifice ratios, but they may only point to the existence of a mere statistical association between them. We deal with the potential endogeneity of the IT regime by employing the Two Stage Least Squares (2SLS) estimation framework. In these regards, we use an instrumental variable strategy which exploits the geographical diffusion of the IT regime. This approach, initially developed in the political science literature, has been also employed in the empirical research on economic reforms (Buera, et al. 2011, Giuliano et al. 2013, Persson and Tabellini 2009), growth (Acemoglu et al. 2016), and fiscal rules (Caselli and Reynaud 2020). Similarly, Lucotte (2012) includes the popularity of IT (proxied by the number of countries that have already adopted IT) when assessing the probability of adopting IT. Therefore, in the first step of the 2SLS procedure we instrument the IT dummy variable by two alternative instruments: the number of neighboring countries that have adopted IT (IT-N), and the total number of countries in the world that have adopted IT by a given year (IT-T). The rationale behind this instrumentation approach is as follows: As argued in the political science literature, public policies spread across countries through different channels, such as competition, imitation, learning etc. (Dobbin et al. 2007, Shipan and Volden 2008). In other words, the adoption of IT in neighboring countries may induce the domestic country to introduce this monetary policy regime as well. We think that the proposition of geographical diffusion of public policy is consistent with the trend of rising global popularity of IT (Rose 2007). In addition, the above instrumentation strategy seems valid from an econometric point of view as the presence of IT in neighboring countries represents an exogenous source of variation in domestic monetary policy strategies, while at the same time it does not directly impact the sacrifice ratio in the domestic economy. In these regards, note that the above instrumental variables are correlated with the IT dummy: the correlation coefficient between IT and IT-N is 0.38, while the correlation coefficient between IT and IT-T is 0.51. Table 5. Sacrifice ratios and inflation targeting in EMEs: 2SLS estimation Second stage regression (1) (2) IT 0.083 (0.076) 0.015* (0.076) Change 0.0008 (0.002) -0.0002 (0.001) Length -0.002* (0.001) -0.002** (0.0008) CBI 0.016 (0.026) 0.021* (0.012) Party -0.005 (0.005) -0.004 (0.003) Open -0.024** (0.009) -0.026*** (0.006) Shocks -0.024 (0.063) 0.044 (0.032) Constant 0.010 (0.016) 0.022** (0.010) Sample size 63 52 Adjusted 0.07 0.19 First stage regression IT-N 0.073 (0.071) - IT-T - 0.010** (0.005) Change -0.010 (0.028) -0.017 (0.026) Length 0.003 (0.017) 0.020 (0.017) CBI 0.082 (0.287) -0.344 (0.274) Party 0.003 (0.062) 0.044 (0.061) Open 0.009 (0.514) -0.068 (0.111) Shocks 0.634 (0.515) 0.674 (0.500) Constant 0.083 (0.201) 0.089 (0.194) Adjusted 0.05 0.03 F-statistic 1.06 5.15 Notes: Sacrifice ratio is the dependent variable; 2SLS estimates with HAC standard errors in parentheses; ***, **, and * indicate 1%, 5% and 10% level of significance, respectively. The first-stage results presented in the bottom panel of Table 5 reveal that only one of the two alternative instruments, IT-T, is statistically significant (although the F-statistic is below the rule-of-thumb value of 10), suggesting that global popularity of IT might be one of the factors behind the adoption of this strategy in individual countries (column 2 of the bottom panel). Consistent with the first-stage estimates, the second-stage estimates (column 2 of the upper panel) confirm the positive association between IT and sacrifice ratios in EMEs. On the other hand, when we use IT-N as an instrument, we also find the positive association between sacrifice ratios and IT, though the regression coefficient of IT is not statistically significant. At the same time, the IT-N instrument in the first-stage regression is insignificant, too, implying that the implementation of IT in a country is not influenced by the presence of this monetary regime in the neighboring countries (the countries with common borders with the domestic economy). Therefore, the results obtained from the 2SLS estimation provide some support to the main findings from our baseline specification, which makes us confident that this conclusion is not a mere consequence of disregarding the potential endogeneity of the IT regime. 4.3. Discussion The main message from our empirical investigation is that the implementation of IT as a disinflation device in EMEs may not be costless at all. Specifically, we show that, controlling for certain institutional, political, and macroeconomic conditions, both IT and other monetary regimes result in similar sacrifice ratios. As shown above in the literature review, this finding should not be regarded peculiar at all. Quite the contrary, our results are consistent with a number of empirical studies focusing on both advanced and developing economies, which also challenge the effectiveness of IT as a disinflation device (for instance, see Ardakani et al. 2018, Brito and Bystedt 2010, Cecchetti and Ehrmann 2002, Laubach and Posen 1997, Magkonis and Zekente 2020, and Mazumder 2014). We think that our finding can be rationalized as follows: In principle, the proponents of IT emphasize that IT is expected to result in lower disinflation costs through its favorable effects on inflation expectations. In these regards, it is argued that the firm commitment to announced inflation target improves the credibility of monetary policy, while at the same time, the close communication with financial markets participants, and the high degree of transparency in general, are also helpful in anchoring inflation expectations (Agénor 2002, Batini and Laxton 2007). Given the central importance of inflation expectations for the slope of the inflation-output trade-off, it should be emphasized that the beneficial effect of IT on inflation expectations is critical for its success in reducing the output costs during a disinflation episode. However, the accumulated empirical evidence cast doubts about the effectiveness of IT in anchoring inflation expectations. For instance, Levin et al. (2004) provide some evidence for the effectiveness of IT in anchoring long-run inflation expectations in industrialized countries, though it is not the case with short- and medium-run inflation expectations. These findings imply that, even when inflation targets are effective in anchoring inflation expectations, their beneficial effects materialize with a lag of several years. Also, they employ the event-study approach to five EMEs and conclude that both short- and long-term inflation expectations in these countries did not change markedly after the introduction of IT. In addition, Angeriz and Arestis (2007), Ball and Sheridan (2004), Castelnuovo et al. (2003), Cecchetti and Hakkio (2010), Filardo and Genberg (2009) and Willard (2012) all find that the announcements of explicit quantitative inflation targets does not matter for inflation expectations. The findings from the related strand in empirical literature equally challenge the proposition that IT may be an effective tool for reducing inflation persistence. For instance, Ball and Sheridan (2004), Gadea and Mayoral (2006), Hossain (2014), Levin and Piger (2004), Siklos (1999), and Willard (2012) show that IT does not lower inflation persistence in OECD countries, while Arsić et al. (2022), Edwards (2007), Filardo and Genberg (2009), and Siklos (2008) provide similar evidence working with mixed samples of advanced countries and/or EMEs. Working with a large sample of 91 countries, Vega and Winkelried (2005) provide some evidence that IT reduces inflation persistence in both advanced and developing countries, but the effect is both small and diminishes very quickly. Therefore, empirical research does not provide convincing evidence that IT itself serves as an effective tool for stabilizing inflation expectations and for reducing inflation persistence. In this regard, we find that the difference between monetary regimes (IT versus the alternative monetary regimes) does not matter for the sacrifice ratios in EMEs, implying that it is difficult to claim the superiority of IT over the alternative nominal anchors. For instance, Benati (2008) studies inflation persistence in five advanced economies under different monetary regimes and finds that, although inflation persistence declined after the adoption of IT, inflation exhibits very low persistence under the alternative nominal anchors, too, implying that IT cannot be regarded superior to other monetary regimes. Similarly, based on micro-level data, Broz and Plouffe (2010) show that exchange-rate pegs are effective in reducing private sector’s concern with inflation expectations, while inflation targets are not. In addition, even when IT does succeed in stabilizing inflation expectations, its impact appears with a considerable lag extending up to several years, which suggests that its favorable effects are largely dependent on the credibility of monetary policy. As argued in IMF (2006), it seems that the successful implementation of IT depends more on credibility than on technical factors. In these regards, a number of papers demonstrate the imperfect credibility of inflation targets during the initial phase of the implementation of IT (Amer and Freeman 1995, Corbo et al. 2002, Debelle 1996, Schaechter et al. 2000, Valera et al. 2017). For obvious reasons, the credibility problem is magnified during the disinflation process when actual inflation converges to the long-run inflation targets with a considerable time lag. Under these conditions, imperfect credibility, accompanied by backward-looking expectations, results in frequent breaches of inflation targets, further loss in credibility, and negative output gaps (Fraga et al. 2003). Moreover, the endogenous evolution of central bank credibility is a function of past target misses as well as the reaction to supply shocks (Agénor 2002). There are certain additional reasons behind the lower credibility of inflation targets in the EMEs. Experience shows that many EMEs have not met all the precondition required for successful implementation of IT, at least in the initial phase. Indeed, the specific institutional and macroeconomic environment in these countries often complicates the design and implementation of the IT regime. For instance, the presence of fiscal dominance, weak banking systems as well as the long historical experience with high inflation, which are common feature in many EMEs, reduce the credibility of their central banks, thus, undermining the effectiveness of monetary policy notwithstanding whether it is implemented within the IT or other regimes. Also, since most of the EMEs are small open economies, their central banks must be concerned with both exchange rate fluctuations and inflation targets simultaneously. For these reasons, during the initial phase of introducing IT, several countries, such as Chile, Hungary, Israel, and Poland, combined inflation with exchange rate targets. However, this practice of dual objectives, might create confusion in the public, thus, compromising the credibility of inflation targets (Amato and Gerlach 2002, Batini and Laxton 2007, Fraga et al. 2003, Freedman and Ötker-Robe 2009, Jonas and Mishkin 2007, Masson et al. 1997, Mishkin 2000 and 2004, Mishkin and Savastano 2002, Mishkin and Schmidt-Hebbel 2002, Schmidt-Hebbel and Carrasco 2016). In these regards, Baxa et al. (2015) provide empirical evidence that, when IT accompanied by targets for the exchange rate, it does not automatically change the dynamics of the inflation process, i.e., it might be unsuccessful in reducing inflation inertia. In fact, due to imperfect credibility of disinflation policies, it is not surprising that both the advanced economies and EMEs have introduced IT only after they had already reduced inflation to some relatively low levels (Masson et al. 1997, Mishkin and Posen 1997). In these regards, Mishkin and Posen (1997) analyze the experiences with IT in New Zealand, Canada, and the UK, pointing out that IT might have served to lock-in the gains from previous disinflation rather than facilitating disinflation. Similarly, Bernanke et al. (1999) and Ball and Sheridan (2004) also show that there is no credibility bonus for the countries that have adopted IT as their monetary strategy, concluding that adopting IT appears to have no significant impact on the disinflation process. 5. Conclusions In this paper, we employ a slightly modified version of the approach used in Ball (1994) to study the relationship between IT and sacrifice ratios in EMEs during 1990-2017, controlling for a number of additional factors, which may affect the cost of disinflation. We provide robust evidence that both IT and other monetary regimes are associated with similar sacrifice ratios in EMEs, thus, challenging the proposition that adopting IT in EMEs may result in lower output costs during disinflation episodes. In addition, we find that, when starting from low to moderate initial inflation, the speed of disinflation (quick versus gradual disinflation) does not matter in EMEs. Also, we show that trade openness is associated with lower sacrifice ratios, while more independent central banks lead to higher sacrifice ratios though the magnitude of these effects is modest. Our main findings are robust to alternative classifications of the IT regime (full-fledged versus lite-IT), alternative definitions of disinflation episodes, different peak levels of trend inflation rate, and across various specifications of the empirical model. In addition, we confirm our main findings by employing an instrumental-variable estimation procedure, which accounts for the possible endogeneity of the IT regime. Finally, we find that the results are sensitive to the filtering method employed for estimating trend output, i.e., the regression estimates lose their statistical significance when we apply both the Hodrick-Prescott and the Hamilton filter. Although we find that IT is not associated with lower disinflation costs in EMEs, we do not interpret our results as invalidating IT as a framework for conducting monetary policy. Instead, our empirical evidence only implies that IT may not be superior to other monetary regimes during the disinflation periods, when the central bank attempts to bring inflation down to the long-term inflation target. For instance, this concern becomes relevant during the recent global surge of inflation, which deviates considerably from the announced inflation targets. Our findings imply that it might not be appropriate for the central banks to respond aggressively to the deviations from the inflation targets. In other words, under these circumstances, the optimal policy response would be to increase the length of the time horizon in the inflation forecast. Alternatively, inflation targeters in EMEs could adopt temporarily higher inflation targets or wider bands around the central target. However, in order to avoid the possible adverse impact on central bank credibility and inflation expectations, the new inflation targets or their bands should be complemented by careful communication with the public. In this respect, the experience of Brazil in 2003 provides a useful example of how to cope successfully with cost-push inflationary pressures. Faced with a supply shock, the central bank of Brazil temporarily abandoned the previous inflation target and adopted a new one, accompanied by an explanation for this move and by declaring the path of the return to the long-run inflation target (Fraga et al. 2003, Mishkin 2004). References Acemoglu, D., Naidu, S., Restrepo, P., and Robinson, J. A. (2019). Democracy does cause growth. Journal of Political Economy, 127(1), 47-100. Agénor, P.-R. (2002). Monetary Policy under Flexible Exchange Rates: An Introduction to Inflation Targeting. In: N. Loayza and R. Soto, eds., Ten Years of Inflation Targeting: Design, Performance, Challenges. Santiago: Central Bank of Chile, 79-169. Almeida, A., and Goodhart, C. A. E. (1998). Does the Adoption of Inflation Targets Affect Central Bank Behaviour? Banca Nazionale del Lavoro Quarterly Review, 51(204), 19-107. Amato, J. D., and Gerlach, S. (2002). Inflation targeting in emerging market and transition economies: Lessons after a decade. European Economic Review, 46(4-5), 781-790. Amer, J., and Freeman, R. T. (1995). Inflation Targeting in the 1990s: The Experiences of New Zealand, Canada, and the UK. Journal of Economics and Business, 47(2), 165-192. Andersen, P. S., and Wascher, W. L. (1999). Sacrifice Ratios and the Conduct of Monetary Policy in Conditions of Low Inflation. BIS Working Papers No. 82. Basel: Bank of International Settlements. Angeriz, A., and Arestis, P. (2007). Assessing the performance of ‘inflation targeting lite’ countries. The World Economy, 30(11), 1621-1645. Ardakani, M. O., Kishor, K. N., and Song, S. (2018). Re-evaluating the effectiveness of inflation targeting, Journal of Economic Dynamics and Control, 90, 76-97. Arsić, M., Mladenović, Z., and Nojković, A. (2022). Macroeconomic performance of inflation targeting in European and Asian emerging economies. Journal of Policy Modeling, 44(3), 675-700. Ball, L. (1994). What Determines the Sacrifice Ratio? In: N. Gregory Mankiw, ed., Monetary Policy. Chicago: The University of Chicago Press, 155-182. Ball, L. Mankiw G.N., Romer, D. (1988). The New Keynesian Economics and the Output – Inflation Trade-off. Brookings Papers on Economic Activity, 1, 1-65. Ball, L. M., and Sheridan, N. (2004). Does Inflation Targeting Matter? In: Ben S. Bernanke and Michael Woodford, eds., The Inflation-Targeting Debate. Chicago: University of Chicago Press, 249-282. Baltensperger, E., and Kugler, P. (2000). Central bank independence and sacrifice ratios: Some further considerations. Open Economies Review, 11, 111-125. Batini, N., and Laxton, D. (2007). Under What Conditions Can Inflation Targeting Be Adopted? The Experience of Emerging Markets. In: Frederic Mishkin and Klaus Schmidt-Hebbel, eds., Monetary Policy under Inflation Targeting. Chile: Central Bank of Chile, 467-506. Baxa, J., Plašil, M., and Vašíček, B. (2015). Changes in inflation dynamics under inflation targeting? Evidence from Central European countries. Economic Modelling, 44, 116-130. Beck, T. Clarke, G., Groff A., Keefer, P., and Walsh, P. (2001). New tools and new tests in comparative political economy - the database of political institutions. The World Bank Economic Review, 15(1), 165-176. Belke, A., and Böing, T. (2014). Sacrifice ratios for Euro area countries: new evidence on the costs of price stability. Australian Economic Review, 47(4), 455-471. Benati, L. (2008). Investigating Inflation Persistence across Monetary Regimes, Quarterly Journal of Economics, 123(3): 1005-1060. Bernanke B., and Mishkin, S. F. (1997). Inflation Targeting: A New Framework for Monetary Policy? Journal of Economic Perspectives, 11(2), 97-116. Bernanke, B., Laubach, R., Mishkin, S. F., Posen A. (1999). Inflation Targeting: Lessons from the International Experience, Princeton University Press. Blanchard, O., and Simon, J. (2001). The Long and Large Decline in U.S. Output Volatility. MIT Department of Economics Working Papers. Boschen, J., and Weiss, C. (2001). The Ex-Ante Credibility of Disinflation Policy and the Cost of Reducing Inflation. Journal of Macroeconomics, 23(3), 323-347. Bowdler, C. (2009). Openness, exchange rate regimes and the Phillips curve. Journal of International Money and Finance, 28(1), 148-160. Bowdler, C., and Nunziata, L. (2010). Labor market structures and the sacrifice ratio. Journal of Macroeconomics, 32(3), 816-826 Brito, R. D. (2010). Inflation targeting does not matter: another look at OECD sacrifice ratios. Journal of Money, Credit and Banking, 42(8), 1679-1688. Brito, R. D., and Bystedt, B. (2010). Inflation targeting in emerging economies: Panel evidence. Journal of Development Economics 91, 198-210. Broner, F., and Ventura J. (2006). Globalization and Risk Sharing. Discussion Paper 5820. Centre for Economic Policy Research, London, United Kingdom. Broz, J. L., and Plouffe, M. (2010). The effectiveness of monetary policy anchors: Firm-level evidence. International Organization, 64(4), 695-717. Bruno, M., and Easterly W. (1998). Inflation Crisis and Long-Run Growth. Journal of Monetary Economics, 41, 3-26. Brumm, H., and Krashevski R. (2003). The Sacrifice Ratio and Central Bank Independence Revisited. Open Economies Review 14, 157-168. Buera, F., Monge-Naranjo, A., and Primiceri, G. (2011). Learning the Wealth of Nations, Econometrica, 79(1), 1-45. Burton, D., and Fischer, S. (1998). Ending Moderate Inflations. In: Moderate Inflation: The Experience of Transition Economies. International Monetary Fund and Bank of Hungary. Calvo, G., and Mishkin, F. (2003). The mirage of exchange rate regimes for emerging market countries. Journal of Economic Perspectives, 17(4), 99-118. Caporale, B., and Caporale T. (2008). Political Regimes and the Cost of Disinflation. Journal of Money, Credit and Banking, 40, 1541–1554. Caporale, T. (2011). Government Ideology, Democracy and the Sacrifice Ratio: Evidence from Latin American and Caribbean Disinflations. Open Economics Journal, 4, 39–43. Caselli, F., and Reynaud, J. (2020). Do fiscal rules cause better fiscal balances? A new instrumental variable strategy. European Journal of Political Economy, 63, 101873. Castelnuovo, E., Nicoletti-Altimari, S., and Rodriguez-Palenzuela, D. (2003). Definition of price stability, range and point inflation targets: the anchoring of long-term inflation expectations, ECB Working Paper, No. 273. Frankfurt A. M.: European Central Bank. Cecchetti, S., and Ehrmann, M. (2002). Does Inflation Targeting Increase Output Volatility? An International Comparison of Policymakers Preferences and Outcomes. In: N. Loayza and K. Shmidt-Hebbel, eds., Monetary Policy: Rules and Transmission Mechanisms. Santiago: Central Bank of Chile, 247-274. Cecchetti, S. G., and Hakkio, C. (2010). Inflation targeting and private sector forecasts. In: D. Cobham,  Ø. Eitrheim, S. Gerlach, and J. F. Qvigstad, eds., Twenty years of inflation targeting. Cambridge: Cambridge University Press, 306-336. Cecchetti, S., and Rich, R. (2001). Structural estimates of the US sacrifice ratio. Journal of Business and Economics Statistics, 19(4), 416-427. Chortareas, G., Stasavage, D., and Sterne, G. (2003). Does Monetary Policy Transparency Reduce Disinfation Costs? Manchester School, 71(5), 521-540. Cobham, D., and Dibeh, G., eds. (2009). Monetary policy and central banking in the Middle East and North Africa. Routledge. Corbo, V. (2002). Monetary policy in Latin America in the 1990s. Series on Central Banking, Analysis, Economic Policies, No. 4. Santiago: Central Bank of Chile. Corbo, V., Landarretche, O., and Schmidt-Hebbel, K. (2002). Does Inflation Targeting Make a Difference? In: N. Loayza and R. Soto, eds., Ten Years of Inflation Targeting: Design, Performance, Challenges. Santiago: Central Bank of Chile, 221-269. Daniels, J. P., Nourzad, F., and VanHoose, D. D. (2005). Openness, central bank independence, and the sacrifice ratio. Journal of Money, Credit, and Banking, 37(2), 371-379. Daniels, J., Nourzad, F., and VanHoose, D. (2006) Openness, centralized wage bargaining, and inflation. European Journal of Political Economy, 22, 969–988 Daniels, J. P., and VanHoose, D. D. (2006). Openness, the sacrifice ratio, and inflation: Is there a puzzle? Journal of International Money and Finance, 25(8), 1336-1347. Daniels, J. P., and VanHoose, D. D. (2009). Trade openness, capital mobility, and the sacrifice ratio. Open Economies Review, 20(4), 473-487. Daniels, J. P., and VanHoose, D. D. (2013). Exchange-rate pass through, openness, and the sacrifice ratio. Journal of International Money and Finance, 36, 131-150. Debelle, G. (1996). The ends of three small inflations: Australia, New Zealand, and Canada. Canadian Public Policy/Analyse de politiques, 22(1), 56-78. Debelle, G., and Fischer, S. (1994). How Independent Should a Central Bank Be? In Conference Series [Proceedings], Vol. 38, Federal Reserve Bank of Boston, 195-225. Diana, G., and Sidiropoulos, M. (2004). Central bank independence, speed of disinflation and the sacrifice ratio. Open Economies Review, 15(4), 385-402. Dobbin, F., Simmons, B., and Garrett, G. (2007). The Global Diffusion of Public Policies: Social Construction, Coercion, Competition, or Learning? Annual Review of Sociology, 33(1), 449-472. Dornbusch, R., and Fisher S. (1993). Moderate Inflation. The World Bank Economic Review, 7(1), 1-44. Durham, J. (2001). Sacrifice ratios and monetary policy credibility: do smaller budget deficits, inflation-indexed debt, and inflation targets lower disinflation costs? Finance and Economics Discussion Series 2001-47, Board of Governors of the Federal Reserve System Edwards, S. (2007). The relationship between exchange rates and inflation targeting revisited. In F. Mishkin and K. Schmidt-Hebbel, eds., Monetary Policy under Inflation Targeting. Santiago: Banco Central de Chile, 373-413. Filardo, A., and Genberg, H. (2009). Targeting Inflation in Asia and the Pacific: Lessons from the Recent Past. BIS Working Paper Series No. 52. Basel: Bank for International Settlements. Fischer, S., Sahay, R., and Végh C. (2002). Modern Hyper - and High Inflations. Journal of Economic Literature, 40, 837-880 Fraga, A., Goldfajn, I., and Minella, A. (2003). Inflation Targeting in Emerging Market Economies. In: Mark Gertler and Kenneth Rogoff, eds., NBER Macroeconomics Annual 2003, Vol. 18. Cambridge, Massachusetts: The MIT Press, 365-400. Freedman, C., and Ötker-Robe, I. (2009). Country Experiences with the Introduction and Implementation of Inflation Targeting. IMF Working Paper 09/161. Washington D. C.: International Monetary Fund. Gadea, M. D., and Mayoral, L. (2006). The persistence of inflation in OECD countries: A fractionally integrated approach. International Journal of Central Banking, 4, 51-104. Garriga, C. A. (2016). Central Bank Independence in the World: A New Data Set. International Relations, 42(5), 849-868. Giuliano, P., Mishra, P., and Spilimbergo, A. (2013). Democracy and reforms: evidence from a new dataset. American Economic Journal: Macroeconomics, 5(4), 179-204. Gonçalves, C. E. S., and Carvalho, A. (2008). Inflation targeting and the sacrifice ratio. Revista Brasileira de Economia, 62(2), 177-188. Gonçalves, C. E. S., and Carvalho, A. (2009). Inflation targeting matters: Evidence from OECD economies' sacrifice ratios. Journal of Money, Credit and Banking, 41(1), 233-243. Gordon, R. J. (2013). The Phillips curve is alive and well: Inflation and the NAIRU during the slow recovery. Technical report, National Bureau of Economic Research. Gordon, R. J., King, S. R., and Modigliani, F. (1982). The output cost of disinflation in traditional and vector autoregressive models. Brookings Papers on Economic Activity, 1, 205-244. Hamilton, D. J. (2018). Why You Should Never Use the Hodrick-Prescott Filter? Review of Economics and Statistics, 100(5), 831-843. Hammond, G. (2012). State of the Art of Inflation Targeting. Bank of England – Centre for Central Banking Studies. Hodrick, R., and Prescott, E. (1997). Postwar U.S. Business Cycles: An Empirical Investigation. Journal of Money, Credit and Banking, 29, 1–16. Hofstetter, M. (2008). Disinflations in Latin America and the Caribbean: A free lunch? Journal of Macroeconomics, 30(1), 327–345. Hossain, A. A. (2014). Monetary policy, inflation, and inflation volatility in Australia. Journal of Post Keynesian Economics, 36(4), 745-780. Hutchinson M., and Walsh C. (1998). The output inflation trade-off and central bank reform: evidence from New Zealand. Economic Journal, 108(448), 703-725 IMF (2006). Inflation Targeting and the IMF. Washington DC: International Monetary Fund. IMF (2018). Annual report on exchange arrangements and exchange restrictions 2017. Washington DC: International Monetary Fund. Jonas, J., and Mishkin, F. S. (2007). Inflation Targeting in Transition Economies: Experience and Prospects. In: F. S. Mishkin, Monetary Policy Strategy. Cambridge, Massachusetts and London: The MIT Press, 345-403. Jordan, T. (1997). Disinflation costs, accelerating inflation gains, and central bank independence. Weltwirtschaftliches Archiv, 133, 1-21 Katayama, H., Ponomareva, N., and Sharma, M. (2019). What determines the sacrifice ratio? A Bayesian model averaging approach. Oxford Bulletin of Economics and Statistics, 81(5), 960-988. Kose, M. A., Otrok C., and Whiteman, C. (2003a). International Business Cycles: World, Region, and Country-Specific Factors. American Economic Review, Vol. 93, No. 4, 1216–39. Kose, M. A., Prasad, E. S., and Terrones, M. E. (2003b). Financial Integration and Macroeconomic Volatility. IMF Staff Papers, 50, 119-42. Kraay, A., and Ventura J. (2007). Comparative Advantage and the Cross-Section of Business Cycles. Journal of the European Economic Association, 5(6), 1300-1333. Kuttner, K. N., and Posen, A. S. (2001). Beyond bipolar: A three-dimensional assessment of monetary frameworks. International Journal of Finance and Economics, 6(4), 369-387. Laubach, T., and Posen, A. S. (1997). Some Comparative Evidence on the Effectiveness of Inflation Targeting. Research Paper No. 9714, New York: Federal Reserve Bank of New York. Levin, A. T., Natalucci, F. M., and Piger, J. M. (2004). The Macroeconomic Effects of Inflation Targeting. Federal Reserve Bank of St. Louis Review, 86(4), 51-80. Levin, A. T., and Piger, J. M. (2004). Is Inflation Persistence Intrinsic in Industrial Economies? Working Paper No. 334, Frankfurt a/M: European Central Bank. Loayza, N., and Raddatz C. (2007). The Structural Determinants of External Vulnerability. World Bank Economic Review, 21(3), 359-387. Lucas R. (1973). Some International Evidence on Output-Inflation Trade-Offs, American Economic Review, 63(3), 326-334. Lucotte, Y. (2012). Adoption of inflation targeting and tax revenue performance in emerging markets economies: an empirical investigation. Economic Systems, 36(4), 609-628. Magkonis, G., and Zekente, K-M. (2020). Inflation-output trade-off: Old measures, new determinants? Journal of Macroeconomics, 65, 103217. Masson, P. R., Savastano, M. A., and Sharma, S., (1997). The Scope for Inflation Targeting in Developing Countries. IMF Working Paper 97/130. Washington DC: International Monetary Fund. Mazumder, S. (2014). Determinants of the Sacrifice Ratio: Evidence from OECD and non-OECD Countries, Economic Modelling, 40, 117-135. Mishkin, F. S. (2000). Inflation Targeting in Emerging Market Countries. American Economic Review, 90(2), 105-109. Mishkin, F. S. (2004). Can Inflation Targeting Work in Emerging Market Countries. NBER Working Paper No. 10646. Mishkin. S. F., and Posen, A. S. (1997), “Inflation Targeting: Lesson from four Countries.” Federal Reserve Bank of New York Economic Policy Review, 3, 9-117. Mishkin, S. F., and Savastano, A. M. (2002). Monetary Policy Strategies for Emerging Market Countries: Lessons from Latin America. Comparative Economic Studies, 44, 45-82. Mishkin. S. F., and Schmidt-Hebbel, K. (2002). A Decade of Inflation Targeting in the World: What Do We Know and What Do We Need to Know? In: Inflation Targeting: Design, Performance, Challenges, 2002. Santiago: Central Bank of Chile, 171-219. Mishkin. S. F., and Schmidt-Hebbel, K. (2007). Monetary Policy under Inflation Targeting: An Introduction. In: Frederic S. Mishkin and Klaus Schmidt-Hebbel, eds., Monetary Policy under Inflation Targeting. Santiago, Chile: Central Bank of Chile, 1-22. Persson, T., and Tabellini, G. (2009). Democratic capital: The nexus of political and economic change. American Economic Journal: Macroeconomics, 1(2), 88-126. Posen, A. S. (1998). Central Bank Independence and Disinflationary Credibility: A Missing Link? Oxford Economic Papers, 50(3), 335-59. Ramey, G., and Ramey, V. (1995). Cross-Country Evidence on the Link between Volatility and Growth. American Economic Review, 85(5), 1138–51. Romer, R. (2001). Advanced Macroeconomics (second edition). Boston etc.: McGraw Hill. Romer, D. (1993). Openness and Inflation: Theory and Evidence. Quarterly Journal of Economics, 158(4), 869-903. Rose, A. K. (2007). A stable international monetary system emerges: Inflation targeting is Bretton Woods, reversed. Journal of International Money and Finance, 26(5), 663-681. Roux, N. D., and Hofstetter, M. (2014). Sacrifice ratios and inflation targeting: the role of credibility. International Finance, 17(3), 381-401. Sargent T. (1983). Stopping Moderate Inflations: The Methods of Poincare and Thatcher. In: Dornbusch and Simonsen (eds.), Inflation, Debt, and Indexation, Cambridge, Mass., MIT Press. Schaechter, A., Stone, M. R., and Zelmer, M. D. (2000). Adopting Inflation Targeting: Practical Issues for Emerging Market Countries. IMF Occasional Paper No. 202. Washington D. C.: International Monetary Fund. Schmidt-Hebbel, K., and Carrasco, M. (2016). The past and future of inflation targeting: implications for emerging-market and developing economies. In Monetary policy in India. New Delhi: Springer, 583-622. Senda, T., and Smith, J. K. (2008). Inflation History and the Sacrifice Ratio: Episode‐Specific Evidence. Contemporary Economic Policy, 26(3), 409-419. Sethi, D., and Acharya, D. (2019). Credibility of inflation targeting: some recent Asian evidence. Economic Change and Restructuring, 52(3), 203-219. Shipan, C. R., and Volden, C. (2008). The Mechanisms of Policy Diffusion. American Journal of Political Science, 52(4), 840-857. Siklos, P. (1999). Inflation-Target Design: Changing Inflation Performance and Persistence in Industrial Countries. Federal Reserve Bank of St Louis Review, 81(2), 47-58. Siklos, P. L. (2008). Inflation Targeting around the World. Emerging Markets Finance and Trade, 44(6), 17-37. Taylor, J. (1983). Union Wage Settlements during a Disinflation. American Economic Review, 73, 981-993. Temple, J. (2002). Openness, inflation, and the Phillips curve: a puzzle. Journal of Money, Credit and Banking, 34(2), 450-468. Tunali, D. (2008). Inflation Targeting and the Sacrifice Ratio: The Effect of Monetary Policy Design. Rutgers University (mimeo). Valera, H. G. A., Holmes, M. J., and Hassan, G. M. (2017). How credible is inflation targeting in Asia? A quantile unit root perspective. Economic Modelling, 60, 194-210. Vega, M., and Winkelried, D. (2005). Inflation Targeting and Inflation Behavior: A Successful Story? International Journal of Central Banking, 1(3), 153-175. Walsh, C. E. (1995). Central bank independence and the costs of disinflation in the European Community. In: B. Eichengreen, J. Frieden and J. von Hagen (eds.) Monetary and Fiscal Policy in an Integrated Europe. Heidelberg, New York, and London: Springer, pp. 12-37. Willard, L. B. (2012). Does inflation targeting matter? A reassessment. Applied Economics, 44(17), 2231-2244. Zhang, L. (2005). Sacrifice Ratios with Long-Lived Effects. International Finance, 8, 231-262.  APPENDIX Table A1. Dates of adopting inflation targeting in selected EMEs Country Year of adoption Brazil 1999 Chile 1999 Colombia 1999 Czech Republic 1998 Dominican Republic 2012 Ghana 2007 Guatemala 2005 Hungary 2001 India 2016 Indonesia 2005 Israel 1997 Mexico 1999 Philippines 2002 Poland 1998 Russia 2014 South Africa 2000 South Korea 1998 Thailand 2000 Turkey 2006 Serbia 2009 Ukraine 2016 Uruguay 2007 Note: Authors’ compilation from various sources.  Table A2. Sacrifice Ratios in selected EMEs (1990-2017) Country Start of the disinflation episode Length of a disinflation episode (in years) Initial trend inflation (%) Change in trend inflation (in percentage points) Sacrifice Ratio Disinflation episodes under the IT regime Brazil 2002 6 10.001 5.500 1.8669 Chile 2007 4 5.505 3.804 1.8859 Colombia 2007 7 5.611 2.916 3.4991 Czech Republic 1997 8 9.351 7.772 3.7939 Czech Republic 2007 4 3.915 2.445 0.8896 Ghana 2008 4 15.502 6.648 1.3257 Guatemala 2007 4 8.245 4.268 0.5628 Hungary 2011 5 4.812 4.777 1.8937 Indonesia 2006 6 9.989 5.066 0.0454 Israel 2001 4 2.637 2.093 3.4621 Poland 2010 6 3.538 4.033 -0.13 Serbia 2012 4 8.721 7.188 0.3189 South Africa 2002 4 6.958 5.420 0.7625 South Africa 2008 4 7.832 2.897 1.9512 South Korea 2009 7 3.456 2.472 -4.471 Turkey 2007 4 9.599 2.503 8.0468 Disinflation episodes under other monetary regimes Algeria 1993 9 27.086 25.091 0.9633 Algeria 2011 4 5.775 2.124 0.4315 Botswana 1992 9 14.088 6.451 -1.893 Brazil 1993 6 1651.74 1646.75 -0.011 Chile 1990 14 21.616 19.498 -4.345 China 1994 6 18.553 19.162 -0.494 China 2011 6 3.783 2.106 -2.002 Colombia 1991 16 25.850 23.887 1.2526 Costa Rica 1991 3 23.180 8.146 -0.547 Costa Rica 1995 8 18.082 8.125 -0.029 Costa Rica 2005 12 12.528 11.724 -2.85 Cote d’Ivoire 1995 6 14.285 11.754 -2.509 Cote d’Ivoire 2012 4 2.933 2.125 -0.588 Croatia 1999 5 5.009 3.177 1.2166 Croatia 2007 4 4.055 2.169 -0.053 Croatia 2012 4 2.626 3.239 2.2001 Dominican Republic 1990 4 46.066 40.143 0.0134 Dominican Republic 1996 2 8.744 2.568 -0.394 Dominican Republic 2003 4 28.045 22.076 0.3031 Dominican Republic 2007 2 8.120 2.043 -0.546 Dominican Republic 2011 5 6.196 4.371 2.1366 Ecuador 2000 7 62.005 59.349 -0.039 Egypt 1990 4 19.255 7.962 0.425 Egypt 1994 8 11.995 9.432 -1.035 Egypt 2009 4 13.781 4.917 -0.833 El Salvador 2007 4 5.108 2.653 1.4483 Ghana 1996 4 44.636 27.227 -0.032 Ghana 2002 2 24.799 6.760 0.028 Ghana 2004 3 18.139 5.884 -0.252 Guatemala 1990 10 28.592 22.657 -0.065 Hungary 1991 3 28.947 7.285 0.8136 Hungary 1995 11 23.547 18.802 1.4873 India 1991 4 11.543 2.610 1.4634 India 1997 6 9.791 5.830 -0.48 Indonesia 1998 4 28.385 19.355 0.246 Indonesia 2002 3 10.052 2.294 0.377 Israel 1990 5 18.809 7.697 -0.316 Israel 1995 6 11.223 8.748 -0.852 Jordan 1990 5 16.687 13.625 -0.306 Jordan 1997 4 4.210 3.195 1.1443 Jordan 2007 2 8.322 2.330 -0.356 Jordan 2009 2 6.023 3.267 0.0411 Lebanon 2012 4 5.458 6.351 -0.351 Malaysia 1997 6 3.807 2.401 11.2926 Mexico 1996 11 30.000 26.139 -1.012 Morocco 1994 5 5.843 3.990 2.0571 Morocco 2007 4 3.011 2.045 -0.575 Nigeria 1994 6 62.344 54.495 0.1263 Nigeria 2004 4 15.631 7.231 0.0978 Nigeria 2011 4 12.259 3.743 -0.665 Pakistan 1995 8 11.695 8.577 0.5612 Philippines 1990 5 14.561 5.385 2.5983 Philippines 1995 9 8.231 4.951 2.1133 Russia 1999 8 44.744 34.290 0.1378 Russia 2008 6 11.588 5.038 3.415 Serbia 2000 5 69.526 57.186 0.1959 Serbia 2005 7 12.957 4.753 -6.564 Singapore 1990 10 3.077 2.707 -11.373 South Africa 1990 11 14.795 9.388 4.2744 South Korea 1991 4 8.039 2.857 2.0136 South Korea 1997 4 5.626 3.246 4.1347 Tanzania 1994 11 28.930 23.905 1.8824 Thailand 1997 4 6.475 5.307 3.0825 Tunisia 1990 4 7.493 2.649 -1.427 Tunisia 1994 7 4.984 2.439 0.058 Turkey 1995 11 91.580 82.788 0.4786 Ukraine 2000 4 20.948 15.953 0.3655 Ukraine 2008 5 17.982 15.220 1.764 Uruguay 1990 11 98.315 93.388 -0.956 Uruguay 2003 4 14.170 7.766 0.075  Table A3. IT and sacrifice ratios – summary of the empirical findings. Study Data and methodology Main findings Almeida and Goodhart (1998) 13 OECD countries during 1981-1997; simple comparison (x) Ardakani et al. (2018) 98 advanced and developing countries during 1998-2013; propensity score matching (-) in advanced countries (x) in developing countries Brito (2010) 24 OECD countries during 1990-2005; OLS (x) Brito and Bystedt (2010) 46 developing countries during 1980-2006; GMM (x) Chortareas et al. (2003) 21 OECD countries during 1990-2000, SUR (x) Gonçalves and Carvalho (2008) 40 OECD and developing countries, 1980-2006; OLS (-) Gonçalves and Carvalho (2009) 30 OECD, 1980-2006; OLS, Heckman’s two-stage procedure (-) Corbo et al. (2002) 23 advanced and developing countries during 1980-2000; simple comparison (x) Debelle (1996) NZ, Australia, and Canada during 1974-1993; simple comparison (+) Laubach and Posen (1997) 8 advanced countries during 1971-1993; OLS (+) Magkonis and Zekente (2020) 42 OECD and non-OECD countries during 1975-2015; Bayesian model averaging (x) in both samples Mazumder (2014) 189 advanced and developing countries during 1972-2007; OLS and fixed-effects (x) in OECD countries (x) in developing countries (+) in high-income countries (x) in middle-income countries (x) in low-income countries Roux and Hofstetter (2014) OECD countries during 1990-2006; OLS (-) Sethi and Acharya (2019) 13 Asian countries during 1970-2014; OLS (-) Tunali (2008) 53 OECD and developing countries during 1990s-2007; 2SLS (x) in both samples Note: “+” indicates that IT increases the sacrifice ratio; “-“ indicates that IT reduces the sacrifice ratio; “x” indicates either mixed evidence or statistically insignificant results. 24