Determinants of the Sacrifice Ratio: Evidence from

OECD and non-OECD countries

Sandeep Mazumder∗

April 2012

Abstract

This paper measures sacrifice ratios for all countries in the world over an approximately forty

year time period, in addition to exploring the determinants of worldwide sacrifice ratios. We test the

most commonly-cited determinants: the speed of disinflation, openness, central bank independence,

inflation targeting, and political factors for both OECD and non-OECD countries. We find that

the speed of disinflation is the most important determinant of OECD sacrifice ratios, but puzzlingly

has no effect on non-OECD nations’ disinflation costs. Instead we find evidence that greater central

bank independence and more openness are associated with lower non-OECD sacrifice ratios. We

also find that the ratio of government debt to GDP–a variable that is not important when it comes

to OECD countries–is highly significant for non-OECD economies. Specifically, we find that higher

indebtedness is associated with lower sacrifice ratios in non-OECD nations, suggesting that greater

levels of debt do not lead to higher expectations of inflation.

JEL Classification: E31, E32, E58, E52, E42.

Keywords: Sacrifice ratios; disinflation.

∗Address: Department of Economics - Carswell Hall - Wake Forest University - Box 7505 - Winston Salem, NC 27109.Email Addresses: mazumds@wfu.edu.

1

1 Introduction

The authority that conducts monetary policy in any given nation typically faces the task of con-

trolling the rate of inflation. For example, in the United States, the Federal Reserve implemented

contractionary monetary policy to deal with inflation in the early 1970s, the mid-1970s, and the early

1980s. However, such disinflationary policy does not come without cost. The cost of disinflation can

be quantified by the sacrifice ratio, which measures the amount of real GDP that must be given up

in order to reduce the inflation rate by one percentage point. This therefore raises two key issues

that macroeconomists face: first, we wish to measure the magnitude of output losses that are incurred

during disinflation episodes, and second, we wish to examine the determinants of the sacrifice ratio.

To that end, a great deal of research has been devoted to both measuring the sacrifice ratio and

examining its determinants. The seminal paper in this body of research is Ball (1994), who was

instrumental in establishing a method of selecting time periods that we can label as “disinflationary.”

Thereafter, Ball proceeds to specify a method in which the sacrifice ratio can be computed, namely

by computing the ratio of the sum of deviations of trend output from actual output, to the change

in trend inflation over the given disinflation episode. This method of computing sacrifice ratios is by

far and away the most widely-accepted and commonly-used technique of estimating disinflation costs,

and continues to be used by prominent papers in the literature. We therefore proceed in this paper

to use the Ball (1994) method of computing sacrifice ratios for our sample of countries.

Ball (1994) also examines the determinants of the sacrifice ratio, and finds that the speed of dis-

inflation (as determined by the amount of disinflation and length of the disinflation episode) and

flexible wage-setting are both responsible for lowering the costs of disinflation, while he argues that

the initial level of inflation at the start of a disinflation episode is not quantitatively important. Other

researchers have also examined a large variety of factors that may possibly be important for determin-

ing the sacrifice ratio. Key factors that are tested in the literature are: the degree of trade openness

(Temple, 2002), the amount of central bank independence (Jordan, 1997), the role of inflation target-

ing (Goncalves and Carvalho, 2009), and political factors such as ideology and democracy (Caporale,

2011).

Despite the various determinants of the sacrifice ratio that are tested, one overarching theme

appears in the literature. Almost all of these papers focus their analysis exclusively on advanced,

industrialized, or specifically, OECD economies. Meanwhile, the question of what determines the

sacrifice ratio has not been adequately addressed for the developing and emerging economies of the

2

world. Examination of non-OECD sacrifice ratio data remains to be an important but under-researched

area in the literature. Thus the goal of this paper is to fill this void by measuring the sacrifice ratio

on a far larger scale than has been done in the literature. In particular, this paper computes the

sacrifice ratio for every single country in the world over an approximately forty year time period.1

This represents a far wider scope of countries than is typically considered in the literature, as well as

a lengthier period of time.

This then allows us to investigate the determinants of the sacrifice ratio for both OECD and

non-OECD economies. We examine all of the typical determinants that have been suggested by the

existing literature that focuses primarily on OECD countries. Namely, we test whether the change

in inflation, the length of the disinflation episode, the initial level of trend inflation, the degree of

openness, the independence of the central bank, and political factors are important in determining

the sacrifice ratio. We find that the speed of disinflation–i.e. the amount of disinflation and length of

the episode–is always economically and statistically significant in determining OECD sacrifice ratios,

while there is little robust evidence of any other important determinants.

However we find a puzzle emerges when we test for the determinants of non-OECD sacrifice ratios:

the speed of disinflation is not a significant determinant of non-OECD sacrifice ratios. Instead, we find

that higher central bank independence and greater trade openness both are negatively and significantly

associated with the costs of disinflation in non-OECD economies. In other words, there is evidence that

a “credibility bonus” does exist in non-OECD economies, where greater separation between monetary

policymakers and the political regime allows for reductions in inflation at lower output costs. This

suggests that greater independence of monetary authorities is a policy that should be encouraged

in developing and emerging economies. We also find that greater openness in developing economies

enables disinflation to take place in a less costly manner. This agrees with the Romer (1993) hypothesis

that increases in output that come from an increase in inflation are negatively related to openness,

implying a negative relationship between openness and the sacrifice ratio. While Romer applied this

idea to developed economies, we find evidence that this prediction holds true for developing and

emerging economies as well.

We also examine the role of the ratio of central government debt to GDP, and find this to be

a highly robust determinant of the sacrifice ratio in non-OECD countries. This is true of various

measures of debt-to-GDP ratios. Somewhat curiously, however, higher levels of debt in non-OECD

1The number of countries in the world varies according to the source. For example, at the time of writing, the UnitedNations report 193 member countries, whereas the U.S. State Department lists 195 countries. This paper examines 189countries.

3

economies are associated with lower sacrifice ratios. This suggests two things to us. First, although

output typically falls during a disinflation episode, a higher level of government debt enables output

to be stabilized and perhaps actually increased, even during a disinflation. This appears to indicate

that the rise in government debt is accompanied by some other factor that enables output to expand

while inflation falls, such as an increase in productivity or some other favorable supply shock. Second,

in theory we would expect higher levels of indebtedness to lead to higher inflation expectations and

thus larger output costs of disinflation. Therefore the finding of a negative impact of debt on the

sacrifice ratio in non-OECD nations suggests that higher levels of debt have not led to higher inflation

expectations, which has enabled these economies to mitigate the output costs of disinflation. The

exact mechanism by which this occurs remains an interesting area for future work.

The rest of this paper as organized as follows: section 2 briefly reviews the relevant literature

concerning sacrifice ratios, and section 3 presents our estimates of worldwide sacrifice ratios. Sections

4 and 5 then examine the determinants of the output costs of disinflation in OECD and non-OECD

countries respectively. Finally we conclude in section 6.

2 Existing Literature

2.1 Measurement of the Sacrifice Ratio

The most prominent method by which sacrifice ratios are computed in the literature follows the work

of Ball (1994). The first step taken is to identify disinflation episodes.2 Trend inflation is defined as

a centered, eight-quarter moving average of actual quarterly inflation. In other words trend inflation

for a given year is the average of the four quarters of inflation of that year and the two quarters on

either side. To identify disinflation episodes, we then identify peaks and troughs in trend inflation.

An inflation peak is one where trend inflation in year t is higher than trend inflation in years t − 1

and t+ 1, and an inflation trough is one where trend inflation in year t is lower than trend inflation

in years t − 1 and t + 1. Finally, Ball defines a disinflation episode as one that starts an inflation

peak and ends at an inflation trough, where trend inflation falls at least 2 percentage points (or 1.5

percentage points with annual data). This rule of looking at disinflations where inflation falls by at

least 1.5 percentage points is designed to separate significant aggregate demand policy changes from

smaller changes that result from shocks (Senda and Smith, 2008).

The sacrifice ratio is then computed as the ratio of the sum of deviations between trend output

2In this paper we will use annual data, hence we focus on the annual data method of computing sacrifice ratios.

4

and actual output, to the change in trend inflation over the disinflation episode. Trend output is

measured in the following way, following Ball (1994). First we assume that output is at its trend (or

natural level) at the beginning of a disinflation episode. Second, we assume that output returns to

trend one year after the end of the episode. The purpose of this second assumption is to capture

the persistent effects of disinflation. In other words, output appears to return to trend with a lag.

Finally, Ball assumes that trend output grows in a log-linear fashion between the inflation peak and

the year after the end of the disinflation episode. The numerator of the sacrifice ratio is then the sum

of the differences between this log-linear fitted line for trend output and the log of actual output. The

denominator of the sacrifice ratio is then simply the change in trend inflation over the course of the

disinflation episode.

Some researchers estimate the sacrifice ratio based on a Phillips curve (such as Hutchison and

Walsh, 1998 and Fischer, 1996), and use this model to derive a relationship between output and

inflation. However the major problem with this approach is that it constrains the inflation-output

tradeoff to be the same during periods of disinflation as it is during periods of accelerating inflation. If

the sacrifice ratio is influenced by factors that are unique to disinflations, this constraint becomes an

invalid one. For this reason, the vast majority of the literature now stays clear of estimating sacrifice

ratios based on the Phillips curve.

Others have also estimated the sacrifice ratio in different ways, but most of these alternative

methods are essentially variations of Ball (1994). Two prime examples of this are Zhang (2005) and

Hofstetter (2008). Zhang (2005) argues that the Ball (1994) method ignores long-lived effects that can

accompany a disinflation episode. Zhang instead measures trend output by calculating a Hodrick and

Prescott (1997) (HP) filter of log real GDP, and then computes the growth rate of the HP filter. Zhang

then assumes that potential output grows at this rate at the beginning of the disinflation episode, thus

giving an alternative method of computing trend output. Hofstetter (2008) is a further extension of

Zhang (2005) that agrees with the notion of long-lived effects, but instead assumes that output is at

trend one year prior to the start of a disinflation episode. Regardless of the minor variations that

are assumed, the Ball (1994) technique of measuring sacrifice ratios is widely-deemed to be the best

existing method.

2.2 Determinants of the Sacrifice Ratio

Ball (1994) finds that the main determinant of the sacrifice ratio is the speed of disinflation, measured

by the ratio of the change in trend inflation to the length of the episode. In particular, he finds that

5

the sacrifice ratio is lower when disinflation is quick, suggesting that the “cold-turkey” approach to

disinflation may be best if welfare is measured by output losses. In addition, Ball finds that disinflation

costs tend to be lower in countries where wage-setting tends to be more flexible. However, he finds

that the initial level of inflation during a disinflation episode and the degree of trade openness are not

important determinants of the sacrifice ratio.

Subsequently a great deal of research has investigated the impact of other factors on the sacrifice

ratio. One such factor that has received a lot of attention is trade openness. Ball (1994) briefly

investigates this matter himself by measuring openness as the ratio of imports to GNP, and finds

statistically insignificant effects on the sacrifice ratio. Temple (2002) also examines the relationship

between openness and the sacrifice ratio, citing the Romer (1993) argument that more open economies

have lower inflation. Temple finds that there is a weakly negative correlation between inflation and

openness, and if anything there is only a weak relationship between openness and the output costs of

disinflation. Daniels et al. (2005) argue that the relationship between openness and the sacrifice ratio

must also account for the degree of central bank independence. Using the Franzese (2002) measure

of central bank independence, they find that the coefficient on openness becomes positive when the

interaction between central bank independence and openness is included in the model. Daniels et al.

(2005) reason that the Barro and Gordon (1983) framework that Romer (1993) relies upon is based

on purely competitive markets, whereas a framework that assumes imperfectly competitive markets

(such as in Razin and Yuen, 2002) actually implies that we should expect a positive relationship

between openness and the sacrifice ratio. However Bowdler (2009) lends further empirical evidence

to the negative impact of openness on the sacrifice ratio, where his analysis accounts for the type of

exchange rate regime.3 In almost all cases, however, the evidence suggests that openness has at best

only a weak impact on the sacrifice ratio.

The sacrifice ratio also provides a useful way to test the impact of central bank independence.

Although he does not explicitly estimate sacrifice ratios, Walsh (1995) argues that greater central

bank independence is associated with higher sacrifice ratios. Walsh interprets this as a benefit of

an independent central bank, since they can have a bigger effect on the real economy. In other

words, an independent central bank can effectively insulate the economy during a recession if greater

independence is associated with higher output losses when trying to reduce the inflation rate. Indeed,

several authors have found a positive relationship between central bank independence and the sacrifice

ratio. For instance, Fischer (1996), Jordan (1997), and Down (2004) all find that greater independence

3Temple (2002), Daniels et al. (2005), and Bowdler (2009) all measure openness as the share of imports to GDP.

6

of central banks is associated with larger costs of disinflation for OECD and other advanced economies.

However, there is also contrasting evidence, such as in Brumm and Krashevski (2003) and Diana

and Sidiropoulos (2004), who find a negative association between central bank independence and

disinflation costs when considering alternative estimation methodologies and different control variables

respectively.4 Interestingly, Cukierman (2002) argues that regardless of the sign of relationship between

central bank independence and the sacrifice ratio, greater independence of central banks increases the

probability that inflation targets are met, and welfare is consequently improved.

Therefore a recent strand of the literature has also examined the role of inflation targeting on

the sacrifice ratio. In response to previous authors, such as Bernanke et al. (1999) and Ball and

Sheridan (2003), who find no existence of a credibility bonus for countries who have adopted inflation

targeting, Goncalves and Carvalho (2008, 2009) examine the impact of inflation targeting on the

sacrifice ratio. They find that among the OECD nations, those who are inflation targeters have lower

sacrifice ratios and thus lower output losses during disinflation episodes. Furthermore, this result

accounts for the potential problem of self-selection by first examining whether countries with higher

average past inflation rates and lower debt levels are more likely to adopted inflation targeting in the

first place. However Brito (2010) finds that the Goncalves and Carvalho (2009) result is based on

a narrow set of OECD disinflations, and when the sample is expanded to a broader set of OECD

disinflations there is no longer any discernible impact of inflation targeting on the sacrifice ratio.

Indeed, Brito argues that even within the Goncalves and Carvalho (2009) sample, inflation targeting

does not matter if one controls for differences in common economic conditions both before and after

the mid-1990s.

The final class of sacrifice ratio determinants that is sometimes considered in the literature is

political factors. For example, Caporale and Caporale (2008) examine whether the governance of

countries’ political leaders matters. They find that right-wing governments have lower costs of disin-

flation than left-wing governments, perhaps due to greater credibility since right-wing governments are

often thought to have stronger anti-inflation reputations. Caporale (2011) also tests whether political

institutions impact the costs of disinflations with a focus on Latin American and Caribbean economies.

Caporale finds that more right-wing governments have less costly disinflations, while more democratic

governments also tend to have lower sacrifice ratios.

4Additionally, Chortareas et al. (2002) find that more transparent central banks have lower sacrifice ratios.

7

3 Data and Sacrifice Ratio Estimates

Inflation is measured as the percentage change in the annual consumer price index (CPI), which is

also the literature’s preferred measure of prices when it comes to measuring trend inflation. The

data for the CPI and real GDP are obtained for 189 countries for the period of 1969 to 2009 from the

International Macroeconomics Data Set of the U.S. Department of Agriculture (USDA). The USDA in

turn obtains their data from the World Bank’s World Development Indicators, the IMF’s International

Financial Statistics, and the USDA’s Economic Research Service. Aside from having a large sample

size, the advantage of looking at all the countries in the world is that we can examine the sacrifice ratio

and its determinants for both OECD and non-OECD economies. Almost all of the prevailing literature

has only used OECD data, even though significant differences may exist between the advanced and

developing economies of the world.5

We proceed to estimate the sacrifice ratio in this paper based on the most commonly-used method

in the literature, namely Ball (1994).6 We identify and measure the sacrifice ratio in the exact same way

as described in section 2.1 with one small modification: since our data is annual and not quarterly, we

define trend inflation as a centered, three-year moving average, as opposed to a centered, eight-quarter

moving average.7 However, we find this makes little difference to the measurement of the sacrifice ratio.

For instance, comparing our sacrifice ratio to Ball (1994)’s–where we focus on disinflation episodes

under both methods that start within a year of each other and have the same length of episodes to

within a year of each other–we find a mean sacrifice ratio for OECD nations of 1.15, compared to

1.25 from Ball (1994). Indeed, the correlation between the two approaches of measuring the sacrifice

ratio is very high and positive with a magnitude of 0.80. For non-OECD economies, we find a lower

mean value of the sacrifice ratio (0.43), but with far greater variance than that exhibited by OECD

economies.8 Tables A1 and A2 in the appendix list the data for our computed sacrifice ratios for

all nations. Overall we identify 78 disinflation episodes for OECD countries and 348 for non-OECD

economies.

5Temple (2002) adds 21 developing economies to his dataset, while Bowdler (2009) and Chortareas et al. (2002) add22 non-OECD countries to their sample. In addition, Hofstetter (2008) and Caporale (2011) focus entirely on LatinAmerica and the Caribbean. While these exercises are useful, they do an incomplete job of examining the non-OECDeconomies in the world.

6It is quite standard in the literature to focus on Ball (1994) as the sole measure of the sacrifice ratio. See, forexample, Jordan (1997), Temple (2002), Chortareas et al. (2002), Brumm and Krashevski (2003), Down (2004), Dianaand Sidiropoulos (2004), Daniels et al. (2005), Caporale and Caporale (2008), Goncalves and Carvalho (2009), Bowdler(2009), and Brito (2010).

7Just as in Ball (1994), we also limit disinflation episodes to those where trend inflation at the peak is less than twentypercent.

8The standard deviation of sacrifice ratios is 2.21 and 6.60 for OECD and non-OECD countries respectively.

8

We also gather several variables that will be used as potential determinants of the sacrifice ratio.

∆π is the change in trend inflation over the disinflation episode, Length is the length in years of

each disinflation episode, and π0 is the initial level of trend inflation at the start of episode. Ideally

we would like to have data on nominal wage rigidity, but unfortunately this is not available for the

country coverage and time horizon considered in this paper. Instead we proxy for nominal rigidities

in a similar way to Hofstetter (2008), who defines “inflation history,” as the average of the inflation

rate of the ten years preceding the disinflation episode. Thus we measure inflation history based on

ten years preceding the episode, and we also compute a five year inflation history, denoted πH10 and

πH5 respectively.

For the degree of trade openness, Openness, we measure the ratio of imports to real GDP using

import data from the IMF’s IFS. We also use the Penn World Tables measure (OpennessPWT ) as a

robustness check. Our data on central bank independence, CBI, are taken from Polillo and Guillen

(2005), who in turn use Cukierman (1992) to define central bank independence. The main advantage

of this data set is that it allows for some time variation in central bank independence, whereas the

majority of the literature that estimates CBI does so using fixed point estimates for the whole sample

period.9 However the downside is that Polillo and Guillen (2005)’s data covers the period of 1989

to 2000, which limits our sample size when examining the role of central bank independence on the

sacrifice ratio. To deal with this issue, we also define CBI which is CBI with the additional assumption

that the level of central bank independence that existed in 2000 was also true of the years 2001 to

2008.

To measure inflation targeting, we define a dummy variable IT which is equal to 1 for inflation

targeting economies, and 0 otherwise.10 In addition, we also define IT which is a dummy variable equal

to 1 if the country was an inflation targeter before the disinflation episode began, and 0 otherwise.

The advantage of this definition of inflation targeting is that it further pinpoints the exact impact

that adoption of inflation targeting has on the output costs of disinflation by refining the timing of

adoption. We also define a dummy variable PIT which is equal to 1 for countries that are considered

“potential” inflation targeters, and 0 otherwise. These potential inflation targeters are identified using

IMF (2006), Epstein (2007), and Stone (2003).

The political factors we consider in this paper are the political orientation of the country at the

9For an example of recent research that does this and also provides a review of earlier work, see Crowe and Meade(2008).

10The list of inflation targeters is compiled from Roger (2010), IMF (2006), Epstein (2007), and Central Bank ofIceland (2007). Also note that members of the European Monetary Union are automatically not counted as inflationtargeters as outlined in IMF (2006).

9

start of the disinflation episode, Left, and the level of democracy at the beginning of the episode,

Polity2. Left is an index taken from Beck et al. (2001) denoting the ruling party’s political orientation

in a country for a given year, where a greater value of Left indicates more left-wing governments.

Polity2 is an index of democracy, where a higher value indicates a greater level of democracy. The data

for democracy are obtained from the Center for Systemic Peace (Heston et al., 2011). In addition, we

will also consider debt-to-GDP ratios as a potential determinant of the sacrifice ratio. For robustness

we consider two alternative measures of debt-to-GDP ratios: the measures computed by Reinhart and

Rogoff (2011) and Jaimovich and Panizza (2010), denoted DebtRR and DebtJP respectively. Finally,

estimation is conducted using ordinary least squares (OLS) with heteroskedasticity and autocorrelation

consistent (HAC) standard errors.

4 Determinants of the Sacrifice Ratio in the OECD

4.1 Speed of Disinflation, Initial Inflation

Ball (1994) finds that the sacrifice ratio in OECD economies is decreasing in the speed of disinflation,

but the level of initial inflation at the outset of a disinflation episode does not have an important

impact on the sacrifice ratio. Many other authors, such as Fischer (1996), Jordan (1997), Brumm and

Krashevski (2003), and Hofstetter (2008) also confirm the importance of the speed of disinflation on

the sacrifice ratio, but there are a handful of papers that argue that the initial level of trend inflation,

π0, is negative and significant, such as Zhang (2005) and Goncalves and Carvalho (2009).

Table 1(a) contains our results that examine the impact of the speed of disinflation and amount of

initial inflation on the sacrifice ratio for OECD economies. In column (1) we regress the sacrifice ratio,

SR, on a constant and Speed = ∆π/Length. We agree with the prevailing arguments in the literature

that the speed of disinflation is indeed highly negative and significant in this sacrifice ratio regression.

Moreover, running the regression with the components of Speed, i.e. ∆π and Length, separately does

not alter this result. In fact the fit of the model, in terms of the R2, becomes much higher in regression

(2). We find that a one percentage fall in the rate of trend inflation is associated with an decrease in

SR of approximately 0.3, while an additional year of disinflation tends to raise the sacrifice ratio by

about 0.9. Therefore we conclude that the sacrifice ratio is lower in OECD nations when disinflation is

quick. In other words, there is some justification to a “cold-turkey” approach to disinflation, if welfare

is maximized by minimizing losses in total output over the course of a disinflation episode.

One of the key results obtained by Ball (1994) concerns the impact of nominal wage rigidity on

10

the sacrifice ratio. Specifically, Ball finds that more flexible wage-setting economies tend to have

lower sacrifice ratios. Due to data limitations, we are not able to measure nominal wage rigidities for

the country coverage and time periods in this paper, so we follow Hofstetter (2008) who proxies for

nominal rigidities using inflation history. In particular, we define two measures of inflation history:

πH5 and πH

10 which are five- and ten-year inflation histories respectively. Columns (3) and (4) in turn

add these measures of inflation histories to our previous model, where we again find that the speed

of disinflation is highly negative and significant. However, we find that nominal rigidities are neither

economically nor statistically significant in these regressions.11

Finally in columns (5)-(8), we also add initial trend inflation, π0 to our model, where we find

that regressing the sacrifice ratio on a constant and π0 alone does indeed produce a negative and

significant coefficient. This suggests that a higher initial trend inflation rate may encourage more

frequent contract, and thus wage, renegotiations. This, in turn, could have the effect of reducing

nominal rigidities, meaning that a lower sacrifice ratio is obtained with a higher value of π0. However,

the finding of a negative and significant coefficient for π0 is not a robust result, since the statistical

significance disappears when we control for the speed of disinflation and nominal rigidities. Moreover,

the economic importance of π0 also falls by a substantial amount when we control for these other

factors. However the coefficients for ∆π and Length remain highly significant in all specifications

tested in Table 1(a), indicating that the speed of disinflation is an important determinant of OECD

sacrifice ratios.

4.2 Openness

A great deal of the literature examines the impact of trade openness on the sacrifice ratio. For example,

Temple (2002) extends the Romer (1993) hypothesis that more open economies are associated with

lower inflation by measuring the slope of the Phillips curve. If the slope of the Phillips curve is

genuinely steeper in more open economies, we can expect lower sacrifice ratios. Temple does indeed find

a negative relationship between openness and the sacrifice ratio, but also finds that this relationship

is weak at best. Ball (1994) goes one step further by declaring that no discernible relationship exists

between openness and disinflation costs.

In Table 2(a) we investigate this matter for our sample of OECD disinflations. Columns (1)-(4)

report the result using Openness as our measure of trade openness, and columns (5)-(8) re-run these

11This might imply that inflation history is not the best proxy for nominal wage rigidity, and is something that futurework can improve upon.

11

regressions using the Penn World Tables measure of openness, OpennessPWT . We start by regressing

SR on a constant and openness, and then we also run separate regressions that control for the speed of

disinflation, nominal rigidities, and the initial level of trend inflation. For the first measure of openness

we find a positive relationship with the sacrifice ratio, but one that is not statistically significant. For

the Penn World Tables measure of openness, we find no statistical significance, and further still the

coefficient on trade openness is essentially zero in all specifications. What is most striking about this

set of results is that openness is never statistically significant regardless of which specification we

examine. Therefore, we concur with Ball’s assessment that there is no association between the degree

of trade openness and the sacrifice ratio. In other words, OECD member nations that have more open

economies will not necessarily see smaller output costs of disinflations as a result of their openness. We

do continue to find, however, that the amount of disinflation and length of disinflation–which together

measure the speed of disinflation–continue to be highly significant determinants of the sacrifice ratio.

4.3 Central Bank Independence

One of the most researched areas in the literature concerning sacrifice ratios has to do with central

bank independence. Some authors, such as Jordan (1997), find that economies with more independent

central banks tend to have higher sacrifice ratios. At first glance, this might appear to be an argument

against central bank independence, but as Walsh (1995) argues, it may actually be an argument

in favor of more central bank independence. In particular, if greater central bank independence is

related to higher sacrifice ratios, this means that the monetary policy authority of the nation stands

to have greater influence on the real economy, which is a particularly useful property to have during

a recession or even when faced with the threat of a recession. Nonetheless, most researchers would

expect that central bank independence reduces the sacrifice ratio if credibility has been achieved. To

that end, some researchers do find a negative relationship between central bank independence and

sacrifice ratios, such as Diana and Sidiropoulos (2004). This suggests that more independence of

the central bank can produce a “credibility bonus.” In other words, one of the primary advantages

of having an independent central bank is that it allows greater credibility due to the separation of

monetary policymakers from political entities, thereby enabling central banks to fight inflation in a

more efficient manner.

We thus proceed to examine the role of central bank independence on our sample of OECD sacrifice

ratios. Column (1) in Table 3(a) shows the regression of SR on a constant and CBI (measured using

12

Polillo and Guillen, 2005) for OECD economies.12 From this model we see that CBI does have a

significant impact on the sacrifice ratio, and in a negative way. In other words, there is some evidence

of a credibility bonus from this regression result. However, this result is not robust to the addition of

∆π and Length to the model (column (2)) and also when including our proxy for nominal rigidities

(columns (3) and (4)). Instead it is the length of the disinflation episode that matters most, suggesting

that the number of years of the disinflation episode matters more than whether the central bank is

independent or not.

Finally, we also check the Daniels et al. (2005) hypothesis that the coefficient on openness is

positive in the sacrifice ratio regression when its interaction with CBI is included in the model. In

other words, trade openness alone might not have an impact on the sacrifice ratio, but it may be

the joint impact of trade openness together with central bank independence that can have significant

effects on the costs of disinflation. We check for this in columns (5)-(8) of Table 3, where (5) and (6)

regress SR on a constant, CBI, openness, and the interaction between CBI and openness (for our two

measures of trade openness), and columns (7) and (8) also then control for the speed of disinflation

as well. We find that including the interaction between central bank independence and openness does

tend to produce a positive coefficient for openness in OECD nations. Moreover in (6), we see that

OpennessPWT and the interaction between CBI and OpennessPWT are both statistically significant.

However, this significance is not robust to the inclusion of ∆π and Length to the model (columns (7)

and (8)), therefore we conclude that there is at best limited evidence that central bank independence

and openness together are important for OECD sacrifice ratios, at least from a statistical perspective.

This conclusion is also robust to the use of CBI which slightly extends our sample size (columns

(9) and (10)), where central bank independence and openness remain statistically insignificant once

controlling for ∆π and Length. From an economic perspective, the magnitude of the coefficients for

CBI and CBI are always large, but the sign varies according to the specification tested. We therefore

consider this an indeterminate result, meaning that CBI is not a significant determinant of sacrifice

ratios in advanced economies. Nonetheless, we continue to find that the speed of disinflation is a

significant determinant of OECD sacrifice ratios, even when we control for the level of central bank

independence.

12One thing to note is that the sample size is much smaller with our CBI regressions than with other potential

determinants that we test, due to limited availability of data for CBI . The use of CBI helps to somewhat alleviate thisproblem.

13

4.4 Inflation Targeting

The debate about whether inflation targeting is useful or not is one that remains unresolved among

macroeconomists, and to that end, we turn next to investigate the impact of inflation targeting on the

output costs of disinflation. One way in which inflation targeting will be noticeably useful is if it leads

to a reduction in the sacrifice ratio. Goncalves and Carvalho (2009) address this very issue, where

they find that inflation targeting has a negative and significant impact on the sacrifice ratio, once

controlling for the speed of disinflation and the initial level of inflation at the start of an episode.13

Indeed, Goncalves and Carvalho (2008) argue that inflation targeting countries in the OECD have

sacrifice ratios approximately one-third of the average of non-inflation targeting countries. Our data

agrees that inflation targeting economies have lower sacrifice ratios on average than that of non-

inflation targeting nations, but not in the order of magnitude as suggested in Goncalves and Carvalho

(2008).14

Our results in Table 4(a) suggest that inflation targeting does not have an important effect on

OECD sacrifice ratios. In (1) we regress SR on a constant and IT , in (2) we also add ∆π and Length,

and finally we add our proxies for nominal rigidities in (3) and (4). In all cases, the coefficient on

IT is never statistically significant, and in fact is frequently of unexpected sign (namely, positive).

It remains to be the case that the speed of disinflation is the most economically and statistically

significant determinant of OECD sacrifice ratios.

Bernanke et al. (1999) make the valid point, however, that the timing of inflation targeting adoption

may be important. In other words, instead of defining our inflation targeting dummy variable as being

equal to one for all countries that are inflation targeters, we should instead focus on countries who had

adopted inflation targeting before the disinflation episode begins. Thus we also examine the impact of

IT (defined in section 3) on the sacrifice ratio, the results of which can be seen in columns (5) and (6)

of Table 4(a). We find this new definition of our inflation targeting dummy variable does improve the

results in the sense that we now obtain the expected negative coefficient. In addition, the economic

importance of the IT coefficient is reasonable, where the results indicate that countries that adopt

an inflation target before the disinflation episode have lower sacrifice ratios in the amount of 0.56.

However we do not find this to be a statistically significant result.

Lastly, we also examine the impact of PIT in conjunction with IT , where PIT includes countries

13Goncalves and Carvalho (2009) also control for government debt and monetary policy transparency.14The mean value of the sacrifice ratio for OECD inflation targeters is 1.76 in our data, compared to 1.99 for non-

inflation targeters.

14

that are potential inflation targeters. In other words, perhaps it is not just the adoption of an inflation

target that matters, but the announcement of future adoption that may matter more, since firms,

households, output markets, and financial markets may all augment their expectations to include the

central bank’s future commitment to an inflation target. Columns (7) and (8) report these results, and

we find that PIT does play an important role in an economic sense, since a country that is a potential

inflation targeter has a lower sacrifice ratio in the region of 0.9-1.0. Indeed, the coefficient on PIT

is almost significant at the 5% level, but is not quite significant in either (7) or (8). Meanwhile, we

continue to obtain counter-intuitive positive signs for IT , though this is statistically indistinguishable

from zero.

Overall, we conclude that inflation targeting has not played an important role for OECD sacrifice

ratios, similar to what Brito (2010) finds, although there is some evidence that potential inflation

target adoption might be important. This suggests that the benefits of inflation target adoption–in

terms of lower output costs of disinflation–disappear after the adoption of the inflation target has

actually taken place.

4.5 Political Factors

The final class of determinants of OECD sacrifice ratios that is considered in the literature deals with

political factors. For instance, Caporale and Caporale (2008) find that more right-wing governments

in the OECD are associated with lower costs of disinflation. We therefore also check whether political

factors are important for our sample of OECD sacrifice ratios. Specifically we test whether political

orientation and level of democracy can help explain sacrifice ratios.

Table 5(a) displays our results for governance and democracy. In column (1) we regress SR on a

constant and Left, where a greater value of Left denotes governments that are more left-wing in terms

of their political orientation. We then also re-run this regression, while also controlling for the speed

of disinflation and nominal rigidities (columns (2)-(4)). In all cases, we find a negative coefficient for

Left, indicating that more left-wing governments tend to have lower sacrifice ratios. Caporale (2011)

argues that the very opposite happens, that more right-wing governments have greater anti-inflation

sentiments. Our results suggest that perhaps the Caporale result is not as strong as one might think

when applied to our sample of OECD sacrifice ratios, and in fact, it may be that more left-oriented

governments have lower output costs of disinflation. However this finding is not significant, since the

coefficient for Left is not significant in any of these regressions, while both ∆π and Length remain

highly statistically significant.

15

We also test Polity2 in our sacrifice ratio regression, which measures the level of democracy of the

nation in question, where a greater value of Polity2 denotes more democratic governments. Caporale

(2011) argues that more democratic governments tend to have lower sacrifice ratios since the greater

openness of such governments makes it less likely that policy changes come as a surprise. Therefore one

can expect expectations and prices to adjust more quickly since policy choices in a more democratic

government reflect, at least indirectly, the preferences of the country’s voters. However, our empirical

results in columns (5)-(8) indicate that Polity2 is not a significant determinant of OECD sacrifice

ratios, in neither an economic nor statistical sense. Finally, we check whether political orientation and

democracy together are important determinants of the sacrifice ratio in the OECD (columns (9) and

(10)), and find that these variables remain insignificant when tested together.

5 Determinants of the Sacrifice Ratio in non-OECD countries

5.1 A Puzzle

Our results for OECD economies suggest that the speed of disinflation, in terms of the amount of

disinflation and length of disinflation episode, is the most important determinant of the sacrifice

ratio. This result is robust to examining a variety of other factors, namely openness, central bank

independence, inflation targeting, and political factors. We next proceed to examine whether these

same factors can explain the behavior of sacrifice ratios in non-OECD economies.

In column (1) of Table 1(b) we regress SR on a constant, and Speed = ∆π/Length, and obtain

a positive coefficient. This presents a puzzle to us, since we expect the speed of disinflation to be

negatively related to the sacrifice ratio, which is precisely what we find for OECD sacrifice ratios. We

then also re-run the regression with ∆π and Length entered into the regression separately (column

(2)), and still obtain no statistical significance. Furthermore the coefficient for ∆π is positive, although

the small magnitude (0.11) indicates that this variable is not economically important for non-OECD

countries. In addition, we find that nominal rigidities are statistically indistinguishable from zero, and

in fact have coefficients magnitudes that are very close to zero (columns (3)-(4)). Controlling for the

initial level of inflation, π0, also makes no differences to these inferences (columns (5)-(8)). Therefore

the factors that are dominant in the literature, namely, the speed of disinflation, nominal rigidities,

and the initial level of trend inflation, play no important role in non-OECD economies. In other words,

the output costs of disinflation in the developing and emerging economies of the world behave rather

differently from those of advanced economies. More specifically, there is no evidence that a cold-turkey

16

disinflation would be less costly in non-OECD economies in terms of output losses, which is contrary

to what we find for OECD nations. At the very least, this brings to attention the need to pay closer

attention to disinflations in developing economies. As of yet, this represents an under-researched and

not well-understood area of the literature. Next, we proceed to investigate what other factors might

therefore be of more significance in determining non-OECD sacrifice ratios.

When examining the role of inflation targeting in the developing countries of the world, we find

no sign of this being an important determinant of the sacrifice ratio. Specifically, Table 4(b) tests IT ,

IT , and PIT (alongside IT ) in our sacrifice ratio regressions. We do not find any robust evidence

that inflation targeting is important for non-OECD economies. On the two occasions where we do see

some statistical significance (regressions (4) and (5)), we obtain incorrect coefficient signs. Therefore

we conclude that inflation targeting is not important for non-OECD economies, just as was the case

with OECD member nations. We also check whether political factors have played an important role

in determining the behavior of non-OECD sacrifice ratios, and the results in Table 5(b) suggest that

Left and Polity2 are not significant when entered as the only political variable in the sacrifice ratio

regression (columns (1)-(8)). However there is some evidence that when we control for both political

orientation and democracy, that Left has a positive and significant coefficient (column (9)). This

indicates that more left-wing governments have higher sacrifice ratios in developing economies, which

in turn implies that the more right-wing governments of the developing world have tougher stances

towards inflation, which agrees with the arguments of Caporale (2011) who specifically examines

Latin American and Caribbean economies. However this result is not robust to the ten-year measure

of inflation history, πH10 (column (10)). We also test whether trade openness is an important factor

behind non-OECD sacrifice ratios in Table 2(b). We do find one instance of a negative and significant

coefficient for OpennessPWT in regression (8), but this result is not robust to switching to the πH5 proxy

of nominal wage rigidity (column (6)). Therefore it appears, at least at first glance, that openness is

not a significant determinant of disinflation costs in non-OECD nations.

5.2 Central Bank Independence and Openness

However, this interpretation changes radically once we consider the role of central bank independence

on non-OECD sacrifice ratios, the results of which can be seen in Table 3(b).15 In regression (1), we

regress SR on a constant and CBI and find no statistical significance. Indeed, when we control for

15Also note that the R2

s for these results are the highest we obtain for non-OECD economies out of all results in thispaper.

17

∆π, Length, and nominal rigidities (proxied by πH5 and πH

10), we still find no statistical significance

(columns (2)-(5)). However, we find the results change dramatically if we follow the Daniels et al.

(2005) reasoning. Specifically, Daniels et al. (2005) argue that openness and central bank independence

should be jointly tested in the sacrifice ratio regression, and further still the interaction between the

two variables may also be important. The reason for testing openness and central bank independence

together lies in firms’ price-setting behavior. In particular, a higher degree of trade openness will tend

to lower firms’ ability to change prices in response to changes in money supply, thereby reducing the

incentive for a central bank with discretionary power to undertake inflationary policy. In other words,

the interaction between openness and central bank independence may be a key determinant of the

sacrifice ratio.

In column (5) of Table 3(b), we regress SR on a constant, CBI, Openness, and the interaction

between the two for non-OECD economies. We find a negative and significant coefficient for both

openness and central bank independence. Moreover, both coefficients are very large in magnitude,

suggesting that these variables are not only statistically important, but are extremely economically

relevant as well. These results suggest that greater central bank independence and greater openness

both have the effect of lowering sacrifice ratios in non-OECD economies. When we also control for

the speed of disinflation (column (7)), we find that openness remains negative and significant, while

CBI barely misses statistical significance at the 5% level. Nonetheless, the results clearly indicate

that both CBI and Openness are economically important determinants of non-OECD sacrifice ratios.

In regression (6), we again regress SR on a constant, central bank independence, openness, and

the interaction term, but using the Penn World Tables measure of openness. We again find that both

CBI and OpennessPWT are negative and significant, although the magnitude of the coefficient for

openness is smaller under the Penn World Tables measure. Indeed, the interaction between central

bank independence and openness is positive and significant.16 Daniels et al. (2005) find a negative

and significant coefficient on this interaction term when considering OECD economies, which they

suggest shows that greater openness reduces the effect of central bank independence on the output

costs of disinflation. Hence, we find that the opposite is true of the developing economies in the world:

greater openness enhances the effect of central bank independence. In other words, countries that

have a higher degree of trade openness are more likely to find that their central bank becomes more

effective in terms of reducing output losses during a disinflation episode, if they are more independent.

16Note that the interaction between CBI and our other measure of openness, Openness, is positive in regressions (5)and (7) as well, and is extremely close to significance at the 5% level in both specifications.

18

Indeed, this finding is robust to controlling for the speed of disinflation (column (8)), where CBI and

OpennessPWT continue to produce highly negative and significant coefficients.

In addition, we also repeat the regressions from (7) and (8) but with CBI as our measure of central

bank independence (columns (9) and (10)), which helps to extend our sample size, and we again

find that central bank independence and openness are highly negative and significant for non-OECD

economies, while their interaction is positive and significant. Thus the results from Table 3 suggest

that although openness and central bank independence are not statistically important for OECD

member nations, the opposite is true for non-OECD economies. For these developing and emerging

economies, greater openness and more independent central banks both tend to lower the output costs

of disinflation. Furthermore, greater openness in these developing and emerging economies enhances

the effect of central bank independence.

5.3 Debt-to-GDP Ratios

The final potential determinant that we will consider in this paper is one that has not received much

attention in the literature, namely the size of a nation’s central government debt, relative to GDP.

Goncalves and Carvalho (2008) is one paper that checks for the role of debt on the sacrifice ratio,

which is applied to OECD nations and a handful of developing economies. They find no impact of

debt on the output costs of disinflation. Although, Goncalves and Carvalho (2009) do not explicitly

include debt as a regressor in their final sacrifice ratio regression, they do implicitly include it by first

using debt-to-GDP ratios as a determinant of whether countries adopt inflation targeting in the first

place. Durham (2001) does explicitly test whether the debt-to-GDP ratio is an important factor that

affects the sacrifice ratio, where Durham anticipates a positive relationship between debt and output

cost of disinflation. The reasoning for this is that governments with substantial levels of debt may

face the temptation to monetize their debt, thereby also creating inflation. Therefore we may expect

to find a positive relationship between the stock of government debt and the sacrifice ratio, meaning

that countries with lower levels of debt will tend to have lower sacrifice ratios. Indeed, this is what

the empirical evidence of Durham (2001) supports when considering high-income countries, although

the author does find some evidence of a negative relationship between debt and sacrifice ratios for

low-income countries.

We thus proceed to examine the role of central government debt-to-GDP ratios on the sacrifice

ratio. Table 6(a) displays the results for OECD economies, where we find that debt is not an important

determinant of the sacrifice ratio, since we obtain no statistical significance in any specification. Indeed

19

the coefficient magnitudes themselves tend to be very close to zero. However, the same is not true of

non-OECD economies, the results of which can be seen in Table 6(b). In regressions (1)-(4) we test

the impact of the DebtRR measure of debt-to-GDP ratios on the sacrifice ratio. We find a negative

and significant coefficient in all four regressions, even when we control for the speed of disinflation

and nominal rigidities. In models (5)-(8), we then repeat these regressions except using the DebtJP

measure of debt-to-GDP instead. We obtain the exact same inferences as before: debt obtains a

highly negative and significant coefficient in all of these models tested. Lastly, we also control for

political orientation and democracy in (9) and (10), and find that political factors are not significant

determinants of non-OECD sacrifice ratios once we control for debt.

Thus we find that debt is a statistically significant factor that impacts the behavior of non-OECD

sacrifice ratios, although in a negative way. Moreover this finding of a negative and significant coeffi-

cient is a highly robust result. This outcome suggests two things to us. First, we usually expect output

to fall during a disinflation episode, and while this still may be true in many scenarios, it appears that a

higher level of government debt enables output to be stabilized–and further still, output may actually

rise–during disinflation episodes in non-OECD nations. This probably suggests that the increase in

government debt occurs simultaneously with some other factor that allows production to increase while

inflation falls. Candidates of these other factors may be an increase in productivity or another type of

favorable supply shock. This is an interesting issue that warrants future research. Second, we would

expect a priori that a higher level of debt is associated with higher inflation expectations, and thus

higher output costs of disinflation. Hence the finding of a negative coefficient for debt in non-OECD

economies indicates that higher levels of debt have not led to higher inflation expectations. As a

result, these developing economies have been able to reduce output losses during disinflation episodes,

despite a higher level of government debt. The exact mechanism by which inflation expectations are

influenced by debt in developing and emerging economies remains an area for future research.

6 Conclusion

The sacrifice ratio is something that has been estimated by many authors over several different time

periods. However, the existing body of research has focused almost exclusively on advanced, indus-

trialized, or OECD economies. This paper seeks to fill a gap in the literature by measuring sacrifice

ratios using the commonly-used Ball (1994) methodology, which we apply to the entire world over a

roughly forty year period.

20

We then seek to find the determinants of these sacrifice ratios, for both OECD and non-OECD

economies. For OECD economies, we find that the speed of disinflation is overwhelmingly the most

important determinant of the sacrifice ratio, and this finding is robust to examining a variety of

different specifications. In particular, there is strong evidence that output losses during a disinflation

episode are minimized if the disinflation occurs in a relatively quick fashion. In other words, the cold-

turkey approach to disinflation may be warranted if welfare is measured by minimizing total output

losses. Other determinants that are widely considered in the literature, namely the initial level of

trend inflation, openness, central bank independence, inflation targeting, and political factors do not

appear to be important determinants of the sacrifice ratio, although there is some limited evidence

that the potential adoption of an inflation target tends to reduce output losses during a disinflation

episode.

When examining the determinants of non-OECD sacrifice ratios, we find something of a puzzle: the

speed of disinflation–which is a highly robust determinant of OECD sacrifice ratios–has no significant

impact on disinflations in the developing and emerging economies of the world. In other words, the

cold-turkey approach to disinflation appears not to be warranted in non-OECD nations. Instead,

we find that both central bank independence and openness are important determinants of non-OECD

sacrifice ratios, where more open economies that also have more independent central banks tend to have

lower output costs of disinflation. In other words, a greater degree of trade and a greater separation

of monetary policymakers from political regimes in non-OECD countries can be beneficial in terms

of reducing sacrifice ratios. In addition, we unexpectedly find that non-OECD economies with higher

ratios of debt to GDP tend to have lower sacrifice ratios. This suggests that the mechanism by which

higher levels of debt is translated into inflation expectations is quite different in non-OECD economies

as we expect might happen in OECD nations. This remains an interesting area of research that future

research should seek to examine.

21

References

Ball, L. (1994): “What Determines the Sacrifice Ratio?” in Monetary Policy, ed. by N. Mankiw,

The University of Chicago Press, 155–193.

Ball, L. and N. Sheridan (2003): “Does Inflation Targeting Matter?” NBER Working Papers

9577, National Bureau of Economic Research.

Barro, R. and D. Gordon (1983): “A Positive Theory of Monetary Policy in a Natural Rate

Model,” Journal of Political Economy, 91, 589–610.

Beck, T., G. Clarke, A. Groff, P. Keefer, and P. Walsh (2001): “New Tools in Comparative

Political Economy: The Database of Political Institutions,” The World Bank Economic Review, 15,

165–176.

Bernanke, B., T. Laubach, F. Mishkin, and A. Posen (1999): Inflation Targeting: Lessons

from the International Experience, Princeton University Press.

Bowdler, C. (2009): “Openness, exchange rate regimes and the Phillips curve,” Journal of Interna-

tional Money and Finance, 28, 148–160.

Brito, R. (2010): “Inflation Targeting Does Not Matter: Another Look at OECD Sacrifice Ratios,”

Journal of Money, Credit and Banking, 42, 1679–1688.

Brumm, H. and R. Krashevski (2003): “The Sacrifice Ratio and Central Bank Independence

Revisited,” Open Economies Review, 14, 157–168.

Caporale, B. and T. Caporale (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.

Central Bank of Iceland (2007): “Monetary Bulletin,” 2007-2, Central Bank of Iceland.

Chortareas, G., D. Stasavage, and G. Sterne (2002): “Monetary Policy Transparency, Inflation

and the Sacrifice Ratio,” International Journal of Finance & Economics, 7, 141–155.

Crowe, C. and E. Meade (2008): “Central Bank Independence and Transparency: Evolution and

Effectiveness,” IMF Working Papers 08/119, International Monetary Fund.

22

Cukierman, A. (1992): Central Bank Strategy, Credibility and Independence: Theory and Evidence,

MIT Press.

——— (2002): “Does a Higher Sacrifice Ratio Mean that Central Bank Independence is Excessive?”

Annals of Economics and Finance, 3, 1–25.

Daniels, J., F. Nourzad, and D. VanHoose (2005): “Openness, Central Bank Independence, and

the Sacrifice Ratio,” Journal of Money, Credit and Banking, 37, 371–379.

Diana, G. and M. Sidiropoulos (2004): “Central Bank Independence, Speed of Disinflation and

the Sacrifice Ratio,” Open Economies Review, 15, 385–402.

Down, I. (2004): “Central Bank Independence, Disinflations, and the Sacrifice Ratio,” Comparative

Political Studies, 37, 399–434.

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 Dis-

cussion Series 2001-47, Board of Governors of the Federal Reserve System.

Epstein, G. (2007): “Central banks, inflation targeting and employment creation,” Economic and

Labour Market Papers 2007/02, International Labour Office.

Fischer, A. (1996): “Central bank independence and sacrifice ratios,” Open Economies Review, 7,

5–18.

Franzese, R. (2002): Macroeconomic Policies of Developed Democracies, Cambridge University

Press.

Goncalves, C. and A. Carvalho (2008): “Inflation Targeting and the Sacrifice Ratio,” Revista

Brasileira de Economia, 62, 177–188.

——— (2009): “Inflation Targeting Matters: Evidence from OECD Economies’ Sacrifice Ratios,”

Journal of Money, Credit and Banking, 41, 233–243.

Heston, A., R. Summers, and B. Aten (2011): Penn World Table Version 7.0, Center for Inter-

national Comparisons of Production, Income and Prices at the University of Pennsylvania.

Hodrick, R. and E. Prescott (1997): “Postwar U.S. Business Cycles: An Empirical Investigation,”

Journal of Money, Credit and Banking, 29, 1–16.

23

Hofstetter, M. (2008): “Disinflations in Latin America and the Caribbean: A free lunch?” Journal

of Macroeconomics, 30, 327–345.

Hutchison, M. and C. Walsh (1998): “The Output-Inflation Tradeoff and Central Bank Reform:

Evidence from New Zealand,” Economic Journal, 108, 703–725.

IMF (2006): “Inflation Targeting and the IMF,” Report of the monetary and financial systems depart-

ment, policy and development review department, and research department, International Monetary

Fund.

Jaimovich, D. and U. Panizza (2010): “Public debt around the world: a new data set of central

government debt,” Applied Economics Letters, 17, 19–24.

Jordan, T. (1997): “Disinflation costs, accelerating inflation gains, and central bank independence,”

Weltwirtschaftliches Archiv, 133, 1–21.

Polillo, S. and M. Guillen (2005): “Globalization Pressures and the State: The Global Spread

of Central Bank Independence,” American Journal of Sociology, 110, 1764–1802.

Razin, A. and C. Yuen (2002): “The ‘New Keynesian’ Phillips curve: closed economy versus open

economy,” Economics Letters, 75, 1–9.

Reinhart, C. and K. Rogoff (2011): “From Financial Crash to Debt Crisis,” American Economic

Review, 101, 1676–1706.

Roger, S. (2010): “Inflation Targeting Turns 20,” Finance and development, International Monetary

Fund.

Romer, D. (1993): “Openness and Inflation: Theory and Evidence,” The Quarterly Journal of

Economics, 108, 869–903.

Senda, T. and J. Smith (2008): “Inflation History And The Sacrifice Ratio: Episode-Specific

Evidence,” Contemporary Economic Policy, 26, 409–419.

Stone, M. (2003): “Inflation Targeting Lite,” IMF Working Papers 03/12, International Monetary

Fund.

Temple, J. (2002): “Openness, Inflation, and the Phillips Curve: A Puzzle,” Journal of Money,

Credit and Banking, 34, 450–468.

24

Walsh, C. (1995): “Output-inflation tradeoffs and central bank independence,” FRBSF Economic

Letter.

Zhang, L. (2005): “Sacrifice Ratios with Long-Lived Effects,” International Finance, 8, 231–262.

25

Table 1: The Sacrifice Ratio, the Speed of Disinflation, and Initial Inflation

(a) OECD

Dependent Variable: SR (1) (2) (3) (4) (5) (6) (7) (8)

Constant3.68 -1.27 -1.47 -0.87 3.57 -0.79 -1.10 -1.08

(0.63)∗∗ (0.55)∗ (0.68)∗ (0.78) (0.55)∗∗ (0.82) (0.93) (0.97)

Speed=∆π/Length-1.92

(0.53)∗∗

∆π-0.34 -0.33 -0.34 -0.26 -0.27 -0.41

(0.07)∗∗ (0.07)∗∗ (0.08)∗∗ (0.10)∗ (0.11)∗ (0.16)∗

Length0.87 0.90 0.83 0.81 0.85 0.86

(0.11)∗∗ (0.13)∗∗ (0.13)∗∗ (0.14)∗∗ (0.15)∗∗ (0.15)∗∗

πH

5

-0.00 0.01(0.01) (0.01)

πH

10

-0.01 -0.01(0.01) (0.01)

π0

-0.19 -0.06 -0.06 0.06(0.04)∗∗ (0.05) (0.07) (0.11)

R2

0.153 0.409 0.411 0.368 0.151 0.407 0.406 0.358N 78 78 71 60 78 78 71 60

(b) Non-OECD

Dependent Variable: SR (1) (2) (3) (4) (5) (6) (7) (8)

Constant0.17 -1.24 -1.26 -0.088 -0.64 -1.73 -1.84 -0.26(0.44) (1.21) (1.22) (0.44) (0.92) (1.65) (1.74) (0.52)

Speed=∆π/Length0.17(0.30)

∆π0.11 0.14 -0.00 0.07 0.09 -0.02(0.14) (0.16) (0.06) (0.10) (0.12) (0.06)

Length0.22 0.19 0.05 0.25 0.22 0.06(0.21) (0.20) (0.17) (0.23) (0.22) (0.16)

πH

5

0.00 -0.00(0.01) (0.00)

πH

10

0.00 0.00(0.00) (0.00)

π0

0.10 0.06 0.08 0.02(0.11) (0.07) (0.08) (0.04)

R2

-0.002 0.006 0.004 -0.010 0.003 0.004 0.002 -0.013N 348 348 316 280 348 348 316 280

Note: SR is the sacrifice ratio, ∆π is the amount of disinflation (trend inflation at peak minus trend inflation at trough)over the disinflation episode, Length is the number of years of the disinflation episode, πH

5 and πH

10 are the average ofinflation in the previous 5 and 10 years respectively, and π0 is trend inflation at peak. OLS estimation is implementedwith HAC standard errors. ** and * denote significance at the 1 and 5 percent levels respectively.

26

Table 2: The Sacrifice Ratio and Openness

(a) OECD

Dependent Variable: SR (1) (2) (3) (4) (5) (6) (7) (8)

Constant1.59 1.93 -1.40 -1.13 -1.56 -0.95 -1.46 -1.02

(0.45)∗∗ (0.45)∗∗ (0.60)∗ (0.61) (1.15) (1.07) (1.25) (1.16)

Openness1.73 1.25 1.95 1.67(2.37) (1.43) (1.50) (1.71)

OpennessPWT -0.00 -0.00 -0.00 -0.00(0.01) (0.00) (0.00) (0.00)

∆π-0.32 -0.35 -0.26 -0.26 -0.39 -0.41

(0.08)∗∗ (0.07)∗∗ (0.12)∗ (0.11)∗ (0.18)∗ (0.16)∗

Length0.84 0.87 0.84 0.84 0.85 0.85

(0.13)∗∗ (0.11)∗∗ (0.17)∗∗ (0.15)∗∗ (0.18)∗∗ (0.15)∗∗

πH

5

-0.00 0.01(0.01) (0.01)

πH

10

-0.01 -0.01(0.01) (0.01)

π0

-0.03 -0.06 0.06 0.06(0.07) (0.07) (0.12) (0.11)

R2

-0.006 -0.013 0.358 0.402 0.355 0.398 0.301 0.346N 75 78 75 78 68 71 58 60

(b) Non-OECD

Dependent Variable: SR (1) (2) (3) (4) (5) (6) (7) (8)

Constant0.40 0.83 -1.45 -0.87 -1.84 -1.29 -0.15 0.666(0.48) (0.38)∗ (1.26) (1.67) (1.80) (2.30) (0.56) (0.639)

Openness0.22 0.34 0.18 0.41(0.54) (0.46) (0.49) (0.47)

OpennessPWT -0.00 -0.00 -0.01 -0.01(0.01) (0.01) (0.01) (0.00)∗

∆π0.12 0.10 0.12 0.09 0.01 -0.04(0.14) (0.14) (0.12) (0.13) (0.05) (0.05)

Length0.23 0.25 0.22 0.25 0.04 0.14(0.22) (0.19) (0.23) (0.19) (0.15) (0.15)

πH

5

0.01 0.00(0.01) (0.00)

πH

10

0.00 -0.00(0.00) (0.00)

π0

0.05 0.06 -0.01 0.01(0.08) (0.09) (0.04) (0.04)

R2

-0.003 -0.002 0.004 0.004 -0.001 -0.002 -0.016 0.012N 321 329 321 329 290 299 256 268

Note: Openness is trade openness measured by the ratio of imports to real GDP, and OpennessPWT is the Penn WorldTables’ (Heston et al., 2011) measure of trade openness, defined as the ratio of exports plus imports to real GDP percapita. OLS estimation is implemented with HAC standard errors. ** and * denote significance at the 1 and 5 percentlevels respectively.

27

Table 3: The Sacrifice Ratio and Central Bank Independence

(a) OECD

Dep. Var.: SR (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Constant4.94 -0.83 -1.26 -1.22 1.53 0.93 -1.61 -0.76 -0.13 -0.20

(1.36)∗∗ (1.40) (1.96) (1.94) (2.81) (2.37) (2.80) (2.22) (2.63) (1.80)

CBI-5.28 -3.17 -3.34 -3.31 3.19 5.26 -1.70 -2.93(2.51)∗ (1.90) (2.00) (2.00) (6.46) (6.00) (6.07) (6.25)

CBI-3.67 -1.71(4.65) (4.15)

Openness15.35 4.79 -0.07(13.36) (13.12) (11.73)

OpennessPWT 0.07 -0.00 0.00(0.03)∗ (0.03) (0.02)

CBI× -38.12 -8.62Openness (27.80) (25.96)CBI× -0.17 -0.00

OpennessPWT (0.07)∗ (0.08)

CBI× 5.22Openness (19.44)

CBI× -0.00OpennessPWT (0.04)

∆π-0.29 -0.45 -0.44 -0.28 -0.30 -0.38 -0.37

(0.09)∗∗ (0.30) (0.27) (0.10)∗∗ (0.10)∗∗ (0.11)∗∗ (0.12)∗∗

Length1.01 1.05 1.05 1.01 1.02 0.97 0.92

(0.17)∗∗ (0.22)∗∗ (0.21)∗∗ (0.20)∗∗ (0.21)∗∗ (0.19)∗∗ (0.18)∗∗

πH

5

-0.05(0.03)

πH

10

-0.02(0.01)∗

π0

0.20 0.18(0.24) (0.20)

R2

0.072 0.514 0.489 0.492 -0.001 0.103 0.421 0.474 0.341 0.381N 29 29 29 29 28 29 28 29 36 37

(b) Non-OECD

Dep. Var.: SR (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Constant1.42 1.52 1.03 0.86 8.35 10.09 8.87 10.20 5.86 7.38(1.56) (1.78) (2.22) (2.31) (3.06)∗ (2.85)∗∗ (3.71)∗ (3.18)∗∗ (2.05)∗∗ (2.41)∗∗

CBI-1.39 -1.25 -1.79 -0.05 -21.32 -25.93 -20.98 -24.86(3.25) (3.25) (2.96) (4.24) (10.02)∗ (8.93)∗∗ (10.58) (8.59)∗∗

CBI-12.01 -16.74(4.33)∗∗ (5.65)∗∗

Openness-31.11 -32.76 -24.80

(12.85)∗ (13.14)∗ (7.45)∗∗

OpennessPWT -0.15 -0.17 -0.13(0.05)∗∗ (0.04)∗∗ (0.03)∗∗

CBI× 84.20 83.28Openness (42.80) (44.44)CBI× 0.42 0.41

OpennessPWT (0.15)∗ (0.13)∗∗

CBI× 57.59Openness (18.36)∗∗

CBI× 0.31OpennessPWT (0.09)∗∗

∆π-0.18 -0.26 -0.27 -0.22 -0.41 -0.23 -0.38(0.25) (0.22) (0.24) (0.12) (0.26) (0.10)∗ (0.20)

Length0.16 0.19 0.14 0.23 0.57 0.26 0.55(0.49) (0.49) (0.49) (0.28) (0.33) (0.22) (0.30)

πH

5

0.01(0.05)

πH

10

-0.00(0.00)

π0

0.09 0.10(0.17) (0.19)

R2

-0.031 -0.079 -0.154 -0.145 0.139 0.096 0.119 0.154 0.158 0.181N 32 32 32 32 31 32 31 32 43 44

Note: CBI is an index of central bank independence taken from Polillo and Guillen (2005), and CBI is CBI extended to 2008.OLS estimation is implemented with HAC standard errors. ** and * denote significance at the 1 and 5 percent levels respectively.

28

Table 4: The Sacrifice Ratio and Inflation Targeting

(a) OECD

Dependent Variable: SR (1) (2) (3) (4) (5) (6) (7) (8)

Constant1.99 -1.68 -1.62 -1.76 -0.87 -0.81 -1.46 -1.57

(0.31)∗∗ (0.51)∗∗ (0.82) (0.88) (0.99) (1.11) (0.83) (0.91)

IT-0.23 0.57 0.62 0.81 0.56 0.74(0.43) (0.33) (0.37) (0.47) (0.37) (0.47)

IT-0.56 -0.56(1.41) (1.50)

PIT-1.02 -0.92(0.54) (0.57)

∆π-0.35 -0.29 -0.44 -0.25 -0.38 -0.29 -0.42

(0.07)∗∗ (0.10)∗∗ (0.15)∗∗ (0.10)∗ (0.16)∗ (0.10)∗∗ (0.15)∗∗

Length0.92 0.91 0.94 0.83 0.83 0.91 0.93

(0.11)∗∗ (0.14)∗∗ (0.15)∗∗ (0.16)∗∗ (0.17)∗∗ (0.14)∗∗ (0.14)∗∗

πH

5

-0.00 0.01 0.00(0.01) (0.013) (0.01)

πH

10

-0.01 -0.01 -0.01(0.01) (0.01) (0.01)

π0

-0.04 0.06 -0.08 0.03 -0.06 0.04(0.06) (0.10) (0.06) (0.11) (0.06) (0.11)

R2

-0.011 0.416 0.415 0.374 0.400 0.349 0.414 0.370N 78 78 71 60 71 60 71 60

(b) Non-OECD

Dependent Variable: SR (1) (2) (3) (4) (5) (6) (7) (8)

Constant0.40 -1.26 -1.82 -0.22 -1.84 -0.27 -1.88 -0.230(0.38) (1.20) (1.77) (0.51) (1.74) (0.52) (1.84) (0.523)

IT0.64 0.58 0.44 0.97 0.30 0.94(0.50) (0.52) (0.78) (0.41)∗ (0.97) (0.48)

IT0.71 0.73

(0.30)∗ (0.41)

PIT-0.52 -0.09(0.85) (0.51)

∆π0.11 0.09 -0.01 0.09 -0.02 0.08 -0.014(0.14) (0.12) (0.06) (0.12) (0.06) (0.10) (0.064)

Length0.22 0.22 0.05 0.22 0.06 0.25 0.058(0.21) (0.23) (0.16) (0.22) (0.16) (0.26) (0.174)

πH

5

-0.00 -0.00 -0.00(0.00) (0.00) (0.00)

πH

10

0.00 0.00 0.00(0.00) (0.00) (0.00)

π0

0.07 0.01 0.08 0.02 0.08 0.02(0.08) (0.04) (0.08) (0.04) (0.10) (0.04)

R2

-0.002 0.004 -0.001 -0.012 -0.001 -0.016 -0.003 -0.016N 348 348 316 280 316 280 316 280

Note: IT a dummy variable equal to 1 for inflation targeters (0 otherwise), IT is a dummy variable equal to 1 if thecountry was an inflation targeter before the disinflation episode began (0 otherwise), and PIT is a dummy variable equalto 1 for potential inflation targeters (0 otherwise). OLS estimation is implemented with HAC standard errors. ** and *denote significance at the 1 and 5 percent levels respectively.

29

Table 5: The Sacrifice Ratio and Political Factors

(a) OECD

Dep. Var.: SR (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Constant2.58 -0.98 -0.65 -0.57 1.00 -1.49 -1.20 1.76 -0.69 2.05

(0.51)∗∗ (0.73) (1.04) (1.08) (0.49)∗ (0.69)∗ (1.20) (2.80) (1.21) (2.69)

Left-0.35 -0.29 -0.29 -0.31 -0.26 -0.25(0.25) (0.19) (0.20) (0.24) (0.19) (0.23)

Polity20.09 -0.00 -0.03 -0.35 -0.04 -0.33(0.05) (0.05) (0.07) (0.29) (0.07) (0.28)

∆π-0.33 -0.27 -0.40 -0.32 -0.26 -0.34 -0.26 -0.33

(0.07)∗∗ (0.11)∗ (0.16)∗ (0.07)∗∗ (0.12)∗ (0.18) (0.12)∗ (0.17)

Length0.90 0.85 0.86 0.89 0.89 0.93 0.88 0.92

(0.12)∗∗ (0.15)∗∗ (0.15)∗∗ (0.12)∗∗ (0.17)∗∗ (0.17)∗∗ (0.17)∗∗ (0.17)∗∗

πH

5

-0.00 0.01 -0.00(0.01) (0.01) (0.01)

πH

10

-0.01 -0.01 -0.01(0.01) (0.01) (0.01)

π0

-0.04 0.06 -0.05 0.01 -0.04 0.01(0.07) (0.11) (0.07) (0.12) (0.07) (0.12)

R2

0.010 0.428 0.413 0.366 -0.005 0.407 0.408 0.378 0.413 0.380N 71 71 71 60 74 74 67 57 67 57

(b) Non-OECD

Dep. Var.: SR (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Constant0.55 -1.47 -2.02 -0.39 0.05 0.03 0.09 -0.13 -0.14 -0.26(0.70) (1.13) (1.74) (0.55) (0.16) (0.53) (0.63) (0.57) (0.61) (0.58)

Left0.08 0.07 0.06 0.26 0.36 0.26(0.29) (0.30) (0.31) (0.14) (0.15)∗ (0.16)

Polity20.03 0.03 0.02 0.01 0.00 -0.00(0.02) (0.02) (0.03) (0.03) (0.03) (0.04)

∆π0.10 0.05 -0.06 -0.01 -0.01 -0.02 -0.02 -0.02(0.17) (0.12) (0.05) (0.04) (0.05) (0.05) (0.05) (0.05)

Length0.31 0.34 0.17 0.03 0.00 0.00 0.00 -0.00(0.22) (0.25) (0.18) (0.15) (0.16) (0.17) (0.17) (0.17)

πH

5

-0.00 0.00 0.00(0.00) (0.00) (0.00)

πH

10

-0.00 0.00 0.00(0.00) (0.00) (0.00)

π0

0.07 0.00 0.00 0.03 -0.01 0.02(0.09) (0.04) (0.04) (0.04) (0.04) (0.04)

R2

-0.004 0.003 -0.003 0.001 0.000 -0.007 -0.019 -0.022 0.001 -0.015N 269 269 269 237 254 254 229 202 226 199

Note: Left is an index denoting the party orientation of a country, where a greater value for Left indicates more left-wing governments (data from Beck et al., 2001). Polity2 is an index of democracy (data from the Center for SystemicPeace), where a higher value of Polity2 indicates a greater level of democracy. OLS estimation is implemented withHAC standard errors. ** and * denote significance at the 1 and 5 percent levels respectively.

30

Table 6: The Sacrifice Ratio and Debt-to-GDP Ratios

(a) OECD

Dep. Var.: SR (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Constant1.97 -1.18 -1.33 -1.41 2.05 -0.92 -1.00 -1.15 3.17 2.25

(0.49)∗∗ (0.72) (1.26) (1.37) (0.45)∗∗ (0.66) (1.01) (1.00) (2.30) (2.63)

DebtRR0.00 -0.01 -0.00 -0.01 -0.00(0.01) (0.01) (0.01) (0.01) (0.01)

DebtJP -0.00 -0.00 0.00 0.00 0.00(0.01) (0.01) (0.01) (0.01) (0.01)

Left-0.07 -0.24(0.23) (0.20)

Polity2-0.55 -0.36

(0.26)∗ (0.31)

∆π-0.32 -0.28 -0.52 -0.31 -0.29 -0.42 -0.28 -0.27

(0.08)∗∗ (0.13)∗ (0.20)∗ (0.08)∗∗ (0.16) (0.18)∗ (0.07)∗∗ (0.08)∗∗

Length0.88 0.89 0.91 0.79 0.79 0.80 0.99 0.86

(0.14)∗∗ (0.20)∗∗ (0.19)∗∗ (0.13)∗∗ (0.17)∗∗ (0.17)∗∗ (0.14)∗∗ (0.16)∗∗

πH

5

0.02 -0.00(0.04) (0.02)

πH

10

0.06 -0.02(0.04) (0.02)

π0

-0.04 0.08 -0.01 0.11(0.07) (0.14) (0.10) (0.12)

R2

-0.016 0.370 0.355 0.297 -0.016 0.340 0.325 0.305 0.454 0.37N 65 65 58 49 61 61 58 54 57 54

(b) Non-OECD

Dep. Var.: SR (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Constant1.05 2.31 1.86 2.09 0.76 0.66 0.45 0.70 2.32 0.69

(0.37)∗∗ (1.01)∗ (1.41) (1.48) (0.44) (0.66) (0.89) (1.01) (1.07)∗ (0.77)

DebtRR -0.01 -0.01 -0.02 -0.02 -0.02(0.00)∗∗ (0.00)∗∗ (0.00)∗∗ (0.01)∗∗ (0.00)∗∗

DebtJP -0.01 -0.01 -0.01 -0.01 -0.01(0.00)∗∗ (0.00)∗∗ (0.00)∗∗ (0.00)∗∗ (0.00)∗

Left0.18 -0.06(0.22) (0.22)

Polity20.03 0.08(0.04) (0.04)

∆π-0.07 -0.14 -0.17 0.00 0.01 -0.01 -0.08 0.05(0.11) (0.13) (0.14) (0.06) (0.07) (0.07) (0.16) (0.07)

Length-0.19 -0.13 -0.14 0.02 0.03 -0.01 -0.16 -0.12(0.32) (0.34) (0.36) (0.18) (0.18) (0.20) (0.37) (0.20)

πH

5

0.00 0.00(0.00) (0.00)

πH

10

0.00 -0.00(0.00) (0.00)

π0

0.07 0.09 0.00 0.02(0.10) (0.10) (0.07) (0.07)

R2

0.037 0.045 0.034 0.049 0.027 0.015 -0.003 -0.007 0.039 0.025N 76 76 64 59 160 160 152 138 64 126

Note: DebtRR is the ratio of total gross central government debt to GDP (data from Reinhart and Rogoff, 2011), andDebtJP is the ratio of central government debt to GDP (data from Jaimovich and Panizza, 2010). OLS estimation isimplemented with HAC standard errors. ** and * denote significance at the 1 and 5 percent levels respectively.

31

Appendix: Sacrifice Ratio Tables

This appendix displays each of the disinflation episodes for OECD nations (Table A1) and non-OECD

nations (Table A2) in our sample from 1972 to 2008, using annual data. The tables display the onset

of each disinflation episode, the length of the episode, the rate of trend inflation at the peak, the

amount of disinflation, and the associated sacrifice ratio.

32

Table A1: OECD Sacrifice Ratios

CountryStart of Length Initial Trend Change in Trend SacrificeEpisode (years) Inflation (%) Inflation (%) Ratio

Australia 1975 5 14.52 5.45 0.341981 5 10.32 3.68 2.431987 7 8.25 6.67 0.671995 4 3.05 2.19 0.17

Austria 1974 5 8.50 4.24 1.531981 7 6.19 4.51 0.771992 7 3.66 2.73 3.51

Belgium 1975 5 11.54 6.34 -0.091982 6 8.01 6.67 1.581990 9 3.26 2.02 7.08

Canada 1974 4 9.76 1.62 1.401981 5 11.15 6.99 1.801990 4 5.13 3.95 2.77

Czech Republic 1997 7 9.32 7.74 2.34Denmark 1981 6 11.40 7.31 0.12

1988 7 4.44 2.66 2.74Estonia 2001 3 4.45 1.79 0.29Finland 1975 5 16.27 7.32 2.24

1981 7 11.06 7.03 0.931989 7 5.85 4.96 8.072001 4 2.50 1.86 1.18

France 1975 3 11.65 2.20 -0.391981 7 12.95 10.11 0.841990 9 3.37 2.58 7.78

Germany 1974 5 6.62 3.13 2.741981 7 5.67 5.26 2.141993 7 4.07 3.09 0.91

Greece 1974 4 18.59 5.90 -0.111991 10 18.57 15.51 2.85

Hungary 1979 4 7.65 1.67 0.94Iceland 2001 4 4.89 1.82 6.48Ireland 1975 4 18.89 6.37 0.31

1980 8 14.44 10.28 1.091990 5 7.71 5.22 3.322001 4 5.03 2.33 -0.65

Israel 1990 5 18.81 7.70 -0.011995 6 11.22 8.74 -0.832001 4 2.64 2.11 5.07

Italy 1975 4 17.60 2.52 -0.311981 7 19.06 13.84 0.931990 4 6.33 1.77 0.221994 5 4.57 2.68 -0.61

Japan 1974 6 15.52 10.28 0.191980 8 5.48 5.02 3.331990 6 2.87 2.64 0.161997 5 0.86 1.63 3.09

33

CountryStart of Length Initial Trend Change in Trend SacrificeEpisode (years) Inflation (%) Inflation (%) Ratio

Luxembourg 1975 5 11.54 6.33 -0.411982 6 8.00 6.65 2.121990 9 3.25 2.02 5.75

Netherlands 1975 4 9.63 4.71 -0.891981 7 6.39 6.35 1.602001 5 3.26 1.91 3.77

New Zealand 1981 3 16.31 7.51 0.081986 8 14.76 13.41 2.441995 4 2.60 1.80 1.06

Norway 1975 4 10.10 2.78 -0.891981 5 11.97 5.59 1.481987 7 7.54 5.53 3.83

Portugal 1990 9 12.31 9.91 2.31Slovakia 1994 4 10.64 4.44 -0.69

2000 3 9.98 3.58 0.182003 4 6.46 3.50 2.27

Slovenia 2001 5 8.26 5.41 1.09South Korea 1991 4 8.04 2.85 2.00

1997 4 5.63 3.26 4.86Spain 1990 9 6.48 4.44 6.46Sweden 1981 7 11.47 6.71 0.82

1990 4 8.75 5.71 1.341994 5 3.13 2.85 -1.01

Switzerland 1974 4 8.41 7.05 1.971981 7 5.39 4.03 3.131991 7 5.10 4.65 2.88

United Kingdom 1975 4 18.89 6.37 0.071980 8 14.44 10.28 2.901990 5 7.71 5.22 2.39

United States 1974 4 8.80 2.17 3.731980 7 11.71 8.66 1.421990 9 4.82 2.79 9.07

34

Table A2: Non-OECD Sacrifice Ratios

CountryStart of Length Initial Trend Change in Trend SacrificeEpisode (years) Inflation (%) Inflation (%) Ratio

Afghanistan 2004 3 16.53 6.40 -0.19Algeria 1978 6 13.60 6.72 0.58

1985 4 10.32 2.77 2.06Antigua and Barbuda 1977 2 13.60 2.52 0.59

1981 5 15.60 13.19 2.03Argentina 2003 3 14.58 6.26 -0.09Armenia 2003 4 4.27 1.63 5.19Bahamas 1974 4 9.64 5.12 8.94

1980 5 10.77 6.58 5.411991 8 5.84 4.48 8.05

Bahrain 1981 6 8.03 10.26 3.121996 5 1.56 2.63 1.29

Bangladesh 1994 4 8.15 2.64 0.301998 4 6.72 4.14 -0.23

Barbados 1980 7 14.06 11.20 3.701990 5 5.17 4.15 3.831996 4 3.99 3.08 -2.14

Belize 1974 4 17.75 14.17 0.661982 5 7.69 5.36 5.231995 5 3.96 4.45 5.46

Benin 1991 2 6.94 2.06 -0.111995 4 19.30 16.12 -0.122001 3 3.55 1.93 -1.64

Bermuda 1991 9 4.84 2.67 4.64Bhutan 1982 5 12.62 6.55 3.12

1992 5 13.16 4.89 0.231997 6 8.63 5.97 -0.22

Bolivia 1991 4 15.72 7.42 0.921995 8 10.25 8.30 -1.84

Bosnia Herzegovina 1998 6 8.53 8.06 -1.12Botswana 1975 3 13.89 2.58 1.27

1980 6 13.93 5.04 -1.731992 9 14.09 6.45 -0.072001 3 7.73 7.20 -0.01

Brazil 2002 6 10.00 5.51 1.72Brunei 1982 3 5.55 3.36 -0.79

1994 6 4.23 3.99 -1.00Bulgaria 2001 4 7.83 3.32 0.41

Burkina Faso 1978 3 13.65 3.64 0.161981 7 10.44 11.47 -0.141988 5 3.21 5.37 2.161995 2 8.74 4.34 -0.291997 4 4.51 3.30 1.78

Burundi 1975 2 12.76 2.96 0.541983 4 9.44 5.25 0.731990 2 9.22 3.28 -0.642004 2 10.68 2.63 0.16

Cambodia 1997 5 9.35 8.74 -0.22Cameroon 1974 7 13.72 4.77 3.28

1983 3 13.75 4.54 -0.141986 5 9.81 9.97 -0.161995 5 15.99 13.90 -0.162001 4 2.83 1.87 -1.29

35

CountryStart of Length Initial Trend Change in Trend SacrificeEpisode (years) Inflation (%) Inflation (%) Ratio

Cape Verde Islands 1983 6 12.14 7.98 -1.691990 4 8.25 4.13 2.361996 6 7.62 6.70 -1.192002 3 2.14 2.24 0.55

Central African Republic 1976 5 18.90 12.35 -1.331983 6 10.15 13.50 -1.871990 3 -0.58 1.58 0.561995 4 15.83 16.39 0.682002 2 3.43 1.97 2.34

Chad 1981 2 8.67 2.91 1.611983 4 9.83 14.45 -1.291989 2 3.68 4.09 1.521991 2 -0.22 2.56 -4.832001 3 7.15 7.79 2.14

China 1979 5 5.28 3.25 5.711989 3 14.01 8.76 1.091995 6 17.72 18.18 -0.29

Costa Rica 1995 8 18.08 8.14 -0.11Cote D’Ivoire 1978 7 19.00 15.07 1.09

1987 4 7.85 7.21 -0.501995 6 14.29 11.78 -2.90

Croatia 1999 5 5.04 3.73 0.44Cuba 1993 5 16.83 21.67 -0.04Cyprus 1976 2 8.44 2.24 -1.71

1980 8 11.24 8.76 2.081992 7 5.47 2.98 2.06

Djibouti 1982 4 7.03 6.91 -0.581989 2 8.75 3.45 -1.291995 8 5.27 3.80 7.33

Dominica 1990 5 4.99 4.02 -2.07Dominican Republic 1974 4 14.24 6.21 -0.84

1980 3 11.16 4.23 0.121996 2 8.74 2.57 -0.16

Ecuador 1974 6 17.23 5.61 -1.26Egypt 1981 2 15.32 1.58 -0.73

1990 4 19.28 8.01 0.431994 8 11.99 9.43 -0.93

El Salvador 1975 3 14.34 3.67 0.071980 4 15.63 3.45 6.54

Equatorial Guinea 1989 3 3.18 5.46 1.271994 5 19.06 15.28 2.422002 5 7.91 3.38 -6.87

Eritrea 1991 4 17.67 8.19 0.881996 4 11.80 4.60 -7.26

Ethiopia 1981 3 4.18 2.85 -1.151984 4 14.40 17.32 1.911992 2 13.54 4.50 -1.011994 4 13.29 13.33 -1.071998 4 4.28 6.25 -0.56

Fiji 1973 2 15.85 2.97 -0.341975 4 13.00 6.03 1.881980 6 11.16 7.32 1.731989 7 8.71 6.70 0.581997 5 4.04 2.00 -0.49

36

CountryStart of Length Initial Trend Change in Trend SacrificeEpisode (years) Inflation (%) Inflation (%) Ratio

French Polynesia 1974 3 12.90 3.73 1.901979 7 12.28 12.79 0.991988 2 5.26 6.39 -0.061992 2 7.72 2.82 0.121994 4 5.02 4.77 1.902002 3 5.16 4.08 0.33

Gabon 1983 4 10.61 17.64 -2.171989 4 8.57 11.87 -0.731995 4 19.71 22.68 -0.832001 2 8.60 9.82 0.13

Gambia 1976 5 15.78 11.79 -0.022002 5 19.51 15.82 -0.43

Georgia 2000 2 9.38 4.53 0.07Ghana 2004 2 18.14 5.88 0.05Grenada 1977 7 18.49 15.04 108.31

1988 2 6.67 1.95 -0.211990 2 4.77 1.56 -0.821992 2 4.48 1.67 1.611994 5 2.94 1.86 10.38

Guatemala 1974 4 14.48 4.04 1.141980 4 11.20 8.45 0.03

Guinea 1992 7 18.53 15.20 -0.09Guinea Bissau 1974 4 8.85 5.22 0.20

1979 2 7.38 1.51 11.312001 3 5.10 4.87 2.40

Guyana 1974 3 19.47 15.95 -0.681979 4 11.04 5.19 -4.04

Haiti 1974 4 18.15 14.54 0.461980 4 13.91 5.91 1.141984 4 9.10 10.45 0.05

Honduras 1974 4 8.81 2.45 4.521980 7 13.20 9.80 1.96

Hong Kong 1978 2 2.89 3.03 -0.221982 4 10.62 6.48 0.311989 11 10.44 13.44 -0.83

India 1973 5 17.33 16.26 0.431982 4 10.96 3.42 -0.721991 4 11.54 2.61 1.751997 6 9.79 5.83 0.41

Indonesia 1980 7 15.51 8.90 -0.372002 3 9.99 2.23 0.31

Iran 1976 4 17.14 2.85 -7.882003 3 15.19 1.81 0.40

Jamaica 2004 3 13.08 2.02 0.01Jordan 1975 3 11.29 1.60 -1.68

1980 7 12.45 11.56 -0.391990 5 18.11 15.05 -0.121997 4 4.21 3.19 1.18

Kazakhstan 2000 4 10.00 3.60 -1.60Kenya 1975 6 16.13 4.98 -0.23

1981 7 15.38 7.56 3.661997 6 8.98 3.14 -1.93

Kiribati & Tuvalu 1992 6 5.47 4.00 8.042002 4 3.67 4.53 0.13

37

CountryStart of Length Initial Trend Change in Trend SacrificeEpisode (years) Inflation (%) Inflation (%) Ratio

Kuwait 1977 4 7.94 10.79 -5.601981 7 7.36 6.33 1.041990 4 7.71 7.18 5.651995 4 2.92 1.65 1.60

Laos 2003 4 12.20 6.03 0.37Lebanon 2003 2 2.95 1.99 -1.29Lesotho 1976 3 14.17 5.35 -3.15

1981 3 15.02 6.09 1.211987 4 14.95 3.15 -3.201992 4 14.21 3.72 2.551996 5 10.84 9.43 -0.352003 3 15.20 10.31 0.43

Libya 1977 2 13.71 4.58 1.001979 2 11.17 6.06 -1.771983 6 11.11 7.13 17.401992 10 10.78 17.98 -4.97

Macau 1989 5 9.59 2.70 -3.281994 7 7.17 9.44 0.28

Macedonia 2001 4 4.71 4.34 1.05Madagascar 1974 4 12.14 7.26 -0.44

1987 4 18.78 9.00 -1.692001 2 11.63 4.42 2.59

Malawi 1975 5 12.18 9.80 -0.96Malaysia 1974 3 10.79 6.82 0.98

1981 6 7.40 6.74 -1.931997 6 3.81 2.40 10.69

Maldives Islands 1992 4 17.24 12.20 0.571996 4 6.44 6.32 -1.68

Mali 1986 7 3.21 4.78 -0.291995 5 14.47 13.76 0.002001 3 3.18 2.99 -0.64

Malta 1974 2 7.91 2.38 -0.451980 5 11.47 11.99 -0.231994 4 4.23 1.72 -2.00

Marshall Islands 1980 3 12.26 4.47 2.941983 5 9.67 6.23 4.461990 6 4.93 2.18 -19.52

Mauritania 1974 4 14.73 9.77 0.211985 3 9.26 3.62 -1.061992 4 8.38 3.26 -0.671997 5 5.78 1.83 -2.48

Mauritius 1974 4 19.10 8.88 -0.361989 4 11.77 4.39 -0.351998 5 6.85 3.41 -1.38

Micronesia 1994 4 3.58 3.48 -2.83Mongolia 1980 3 16.31 5.19 0.31Morocco 1975 2 11.33 1.65 -1.33

1981 4 10.81 2.02 -1.081985 4 9.64 6.86 -0.321994 5 5.48 3.99 2.34

Mozambique 2003 4 14.29 4.77 0.02Namibia 1984 4 16.77 3.61 -0.02

1988 4 15.05 8.57 1.211994 4 12.31 4.65 -0.052004 4 12.16 4.78 -0.27

38

CountryStart of Length Initial Trend Change in Trend SacrificeEpisode (years) Inflation (%) Inflation (%) Ratio

Namibia 1984 4 16.77 3.61 -0.021988 4 15.05 8.57 1.211994 4 12.31 4.65 -0.052004 4 12.16 4.78 -0.27

Nepal 1973 5 13.21 8.50 -0.251981 4 12.51 4.75 2.001987 3 12.91 4.22 -0.681991 4 13.65 5.82 0.201997 5 8.16 5.43 -1.11

New Caledonia 1974 4 12.92 9.53 0.461980 3 12.26 4.47 1.171983 4 9.67 4.35 0.681987 4 6.43 2.86 -14.261992 7 5.70 3.66 -3.91

Nicaragua 1997 6 11.22 6.21 -2.04Niger 1977 3 18.96 9.74 -1.53

1981 7 14.95 18.72 2.651989 4 -1.67 2.82 -0.181995 6 17.30 15.76 -0.172001 3 3.18 2.75 -0.75

Nigeria 2004 4 15.63 7.23 -0.04Oman 1990 4 3.54 3.19 -2.48

1995 7 1.26 2.21 0.83Pakistan 1980 7 10.70 6.09 -0.87

1995 8 11.70 8.58 0.59Palau 1987 4 5.75 2.89 0.97

1991 3 4.67 3.09 4.421999 2 10.55 2.57 0.252001 3 10.49 5.20 0.17

Panama 1972 5 6.59 2.90 -3.111980 8 9.70 9.29 -9.20

Papua New Guinea 1974 5 14.00 8.63 1.971980 7 8.63 4.46 4.091990 4 6.13 2.09 -1.511999 8 14.70 13.07 4.50

Paraguay 1973 4 15.74 8.89 0.452002 4 10.67 3.76 0.52

Philippines 1973 4 19.65 11.03 0.401980 3 16.27 5.16 -0.291990 5 13.92 6.16 2.321997 5 8.22 5.01 1.61

Republic of South Africa 1981 3 14.52 1.69 3.161986 4 17.04 3.09 -1.451990 11 14.80 9.39 4.372002 4 6.82 5.35 0.63

Rwanda 1978 2 7.75 3.48 -0.381983 5 8.06 9.44 1.392005 2 11.71 1.70 0.72

Samoa 1974 4 15.18 7.99 -0.811989 2 10.07 3.44 0.321991 2 7.48 4.51 0.471993 2 7.61 3.98 1.121995 2 4.85 1.74 -1.791997 3 4.82 3.67 -0.172003 2 8.17 2.07 0.142005 2 7.32 3.59 0.45

39

CountryStart of Length Initial Trend Change in Trend SacrificeEpisode (years) Inflation (%) Inflation (%) Ratio

Saudi Arabia 1979 7 4.25 6.86 -1.341990 4 2.66 2.14 -7.431995 6 2.22 3.41 -1.65

Senegal 1974 4 19.85 14.57 -1.551983 6 13.59 15.43 0.531995 5 14.30 13.40 0.10

Seychelles 1984 3 3.66 2.44 -2.221991 6 3.04 3.28 2.812000 4 6.20 3.75 -0.82

Singapore 1974 3 14.85 13.56 0.261980 7 6.93 7.06 -2.541990 10 3.08 2.71 -9.81

Solomon Islands 1975 3 11.06 4.68 -0.951981 4 14.16 5.21 -1.721988 4 14.22 2.71 0.961995 2 11.55 1.72 -1.411997 4 10.75 3.43 -6.612002 4 9.00 1.88 0.17

Sri Lanka 1974 3 9.52 6.46 -0.151981 6 18.31 12.59 -0.521989 6 15.68 6.39 0.141997 3 11.62 4.88 0.252002 2 10.01 2.19 -0.14

St. Kitts and Nevis 1981 4 11.79 7.94 2.161991 2 6.19 2.13 0.751993 3 5.20 4.97 0.271998 4 5.42 3.32 -1.10

St. Lucia 1990 4 4.42 1.89 -2.101995 3 3.94 2.57 1.231999 4 3.47 2.59 2.71

St. Vincent and the Grenadines 1977 3 15.51 3.65 -2.531980 5 12.83 8.79 0.121990 5 6.03 4.75 3.241997 5 2.33 1.72 -6.07

Suriname 1976 2 15.78 5.74 -0.031978 7 12.94 11.79 1.571988 2 13.73 5.46 -0.96

Swaziland 1974 2 14.29 1.67 -4.541980 4 18.40 6.62 0.781986 5 15.86 6.00 -1.781994 5 12.69 5.59 1.272001 5 10.06 5.44 -0.40

Syria 1974 5 17.28 12.70 -2.411981 3 17.33 7.44 -0.841994 6 15.98 17.04 -1.07

Taiwan 1980 6 15.03 14.86 1.041995 4 3.62 2.70 -0.65

Thailand 1974 3 15.05 9.36 0.481980 6 14.09 12.37 -0.151997 4 6.48 5.31 3.77

Togo 1976 4 17.37 10.61 0.231981 5 14.38 14.79 1.451987 2 1.34 1.65 -0.742001 3 2.96 2.12 -0.52

40

CountryStart of Length Initial Trend Change in Trend SacrificeEpisode (years) Inflation (%) Inflation (%) Ratio

Tonga 1981 3 16.04 9.12 -0.321986 3 14.37 8.15 -0.311991 4 9.41 8.27 0.572003 4 11.00 4.11 -2.94

Trinidad and Tobago 1974 4 17.94 7.04 0.961980 7 15.51 6.82 -7.701989 4 10.07 3.04 -2.171993 4 8.69 4.61 -0.68

Tunisia 1985 5 10.26 2.86 4.711990 4 7.49 2.65 -1.341994 7 4.98 2.44 1.41

Uganda 1995 5 8.52 5.38 -0.562000 2 3.81 2.31 0.15

United Arab Emirates 1993 6 5.57 4.40 -4.56Uruguay 2003 4 14.17 7.77 -0.74Vanuatu 1981 4 14.92 12.17 -1.07

1988 8 10.85 9.04 -5.51Venezuela 1980 4 16.65 7.46 0.24Vietnam 1997 5 5.38 4.82 0.93

West Bank 1995 6 11.93 7.82 -6.72Yemen 2005 2 15.05 1.53 -0.07Zaire 1979 10 13.54 11.73 -1.01

1996 7 9.18 3.05 15.29Zimbabwe 1975 4 12.50 3.02 2.11

1981 2 11.47 2.44 -0.211983 5 9.98 7.52 1.71

41