A. Background

1. Uruguay has won the battle against its very high inflation observed between the 1960s and early 1990s (Figure 1). After exceeding 130 percent in the mid-1980s, inflation gradually declined to single digits by the late 1990s. The progress was blown temporarily off course by the currency and financial crisis of 2002 that pushed inflation above 20 percent. But after peaking at 28 percent in March 2003, inflation declined to single digits by 2004 and it has remained in single digits since then (7.3 percent on average in January 2004–September 2012), marking the longest period of single digit inflation in recorded history.

View Full Size

Monetary Policy, Exchange Rates, Wage, and Inflation Dynamics

Citation: IMF Staff Country Reports 2013, 109; 10.5089/9781484395301.002.A001

Sources: Banco Central del Uruguay and Fund staff calculations.1/ Central Bank foreign exchange operations with the banking system (in millions of US dollars). The exchange rate indexes equal 100 in January 2007 (Positive values mean depreciation).

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Consumer Price Index

(12-month change, in percent)

Citation: IMF Staff Country Reports 2013, 109; 10.5089/9781484395301.002.A001

Source: Instituto Nacional de Estadistica.

• Download Figure
• Download figure as PowerPoint slide

2. However, inflation and its expectations have remained stubbornly above the authorities’ official inflation target range (4–6 percent). Following the move to a floating exchange rate regime in 2005, Uruguay gradually transitioned to a new monetary policy framework. The interest rate became the main monetary policy instrument in September 2007.² However, unlike the experience of other countries that moved to IT, Uruguay’s inflation and inflation expectations have persistently exceeded the officially established target range.³ Moreover, the magnitude of Uruguay’s persistent overshooting of inflation expectations relative to target exceeds those of its peers.⁴

View Full Size

Source: BCU, Consensus Forecast and IMF Staff estimates.Note: Blue area corresponds to the official target range. Consensus forecast expectations correspond to the average 12-month ahead inflation expectations. BCU expectations correspond to the 18-month ahead median inflation expectations.

• Download Figure
• Download figure as PowerPoint slide

3. Reducing inflation is now a top policy priority. The authorities have publicly expressed concerns about the level of inflation and its deviation from the official target (see quotes above). Despite a slowdown in economic activity and a substantial tightening of policy inflation has not subsided. In recent months headline inflation has ticked up from 7.8 to 9.1 percent (12-month basis through October 2012). Moreover, core inflation has slowly crawled up, increasing from 7.0 to 9.1 percent between July 2011 and October 2012 (Annex Figure 1). This rise in inflation has prompted “moral suasion” actions by the authorities on supermarkets to freeze or reduce the prices of certain consumer goods.⁵

View Full Size

Inflation and Inflation Target in Selected Economies, 2011

(In percent)

Citation: IMF Staff Country Reports 2013, 109; 10.5089/9781484395301.002.A001

Source: Central Banks.Note: Bars show the official target ranges, with the horizontal line marking the mid-point. Dots show end-year inflation for 2011.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Source: BCU, Central Banks, IMF-IFS, Haver analytics, IMF staff estimates.Note: Inflation dynamics around the time of the introduction of inflation targeting. Horizontal axis: Months. Month zero corresponds to the start of inflation targeting. Shaded blue area corresponds to the 25th to 75th interquantile range of the distribution. The date of the introduction of inflation targeting in Uruguay is September 2007, which corresponds to the date when BCU started using the policy rate as instrument. Six month moving average for inflation dynamics. The following countries are included in the sample: Australia, Brazil, Canada, Chile, Colombia, Czech Republic, Guatemala, Hungary, Indonesia, Israel, Mexico, New Zealand, Norway, Peru, Philippines, Poland, Romania, South Africa, South Korea, Sweden, Thailand, Turkey, United Kingdom, and Uruguay.

• Download Figure
• Download figure as PowerPoint slide

4. Reducing inflation and anchoring inflation expectations is important for several reasons. Entrenching stable inflation and inflation expectations within the target range would better support the process of de-dollarization in the economy, lower the cost of public debt in local currency, promote financial deepening, and reduce the need for indexation in the economy for financial contracts and wages. It would also create more space for easing monetary policy in response to an economic downturn or strong capital inflows.⁶

5. So why are inflation and its expectations stuck above target? As we will discuss below, our main conclusion is that despite the monetary tightening seen over the past two years in the form of policy rates, higher reserve requirements, or the introduction of marginal reserve requirements, the monetary policy setting has remained, as in other countries, cautious about downside risks associated with global conditions, financial stability considerations, buoyant capital flows and concerns about large exchange rate movements. Unfortunately, for Uruguay it has also coincided with widespread wage indexation practices, and has taken place at an early stage of the introduction of the IT regime, when inflation expectations have not yet converged to the target and the perception by the private sector about the commitment to inflation target has not been fully established. As a result, inflation and its inflation expectations have slowly been crawling up.

6. The rest of the paper is organized as follows. Section B examines the monetary policy stance by comparing the interest rate behavior with that predicted by Taylor-rules. Section C then examines whether inflation expectations are well anchored or not. Next, Section D takes a closer look at the main drivers of inflation by estimating a Philips curve. With these elements in place, Section E examines potential changes to the communication framework to help BCU’s control over inflation expectations. A final section concludes.

B. The Monetary Policy Stance

7. The central bank of Uruguay (BCU) has tightened monetary policy over the past two years. It raised the policy rate by a cumulative 275 basis points, increased average reserve requirements, and introduced marginal reserve requirements (Annex Figure 1).

8. Has this tightening aligned the monetary policy stance with the inflation goals? To answer this question we estimate a Taylor-type interest rate rule using quarterly data over the period 1997–2012. In addition, we calibrate a Taylor rule with standard coefficients used in the literature (Taylor, 1993). The assessments based on these rules are complex given the uncertainty about which assumption to use for some parameters (e.g. the level and growth rate of potential GDP, long-term inflation expectations, or their corresponding weights in the rule), nonetheless they provide a useful benchmark to assess the stance of monetary policy.

9. The estimated specification is as follows:

Where i_t is the monetary policy interest rate in period t, E_t (π_t+4|I_t) is the expected 4-quarter ahead CPI inflation, π^* the inflation target, E_t(y_t – y_t^*) is the expected output gap, with y^* denoting potential output, defined as the Hodrick-Prescott (HP) trend.⁷ Finally, ρ(L) is a lag operator. The interest rate rule is estimated using instrumental variable-general method of moments (IV-GMM) and includes two lags of the interest rate (see Clarida, Gali and Gertler, 1998 and 2000). Three lags of all the independent variables and the interest rate are used as instruments. This approach deals with possible endogeneity bias as forward-looking variables are obtained from a linear combination of lagged variables (i.e. the instruments). Estimates are reported in Table 1. As reported the policy rule satisfies the Taylor principle (according to which the optimal policy response to a rise in inflation is to increase interest rates sufficiently so as to induce an increase of real interest rates).

Table 1.

Interest Rate Rule – IV GMM Regression

Source: Fund staff estimates.

Table 1.

Interest Rate Rule – IV GMM Regression

Dependent variable: policy rate
β	γ	ρ(1)	ρ(2)	α
1.56	0.52	0.60	−0.02	−8.29

Source: Fund staff estimates.

View Table

Table 1.

Interest Rate Rule – IV GMM Regression

Dependent variable: policy rate
β	γ	ρ(1)	ρ(2)	α
1.56	0.52	0.60	−0.02	−8.29

Source: Fund staff estimates.

10. A standard Taylor rule is also calibrated. Specifically, we calibrate a rule of the form i_t = c + α(π_t − π^*) + (β(y_t − y_t^*), where c is the real neutral rate calculated as the sum of the upper limit of the official target range (6 percent) and potential real GDP (4 percent), while α and β are calibrated to be 1.5 and 0.5 (a similar calibration is done in BIS, 2010).

11. The results suggest that the actual policy rate has been systematically below the policy rate implied by the rules during the period that followed the 2008–09 global economic crisis. Moreover, during this period, inflation expectations have consistently been above the target.⁸ The gap between the predicted and the actual policy rate might be attributable to factors that are ignored in this mechanical rule (see the discussion below). The widening gap is mainly the result of a sustained increase in the estimated rule-based rate that was not accompanied by increases in the actual rate. This finding is in line with those of Magud and Tsounta (2012) based on a wide array of methodologies.

12. But why has the interest rate gap widened after the 2008–09 crisis? In Uruguay, monetary policy has sought to balance inflation objectives with economic developments, including concerns about exchange rate appreciation.⁹ Such a widening is not exclusive to Uruguay; many other IT countries (e.g. Mexico and Brazil) have also seen such widening (see, for example, BIS, 2010; Magud and Tsounta, 2012, Taylor, 2012; and Hofmann and Bogdanova, 2012). For many of these countries, this widening has to do with the economic uncertainty related to the global crisis and the need to balance inflation objectives with other objectives—e.g., financial stability, growth, capital flows, and the exchange rate (Borio, 2012; BIS, 2010; Magud and Tsounta, 2012; Taylor, 2012; and Tovar, 2010). However, in most of these countries inflation and its expectations are relatively well anchored within the target range.

View Full Size

Actual and benchmark policy rates

Percent

Citation: IMF Staff Country Reports 2013, 109; 10.5089/9781484395301.002.A001

Source: IMF Staff estimates.Note: The benchmark interest rate rule is calculated as i=c+α* (pi-pi*)+β*y, where c is the upper limit of the inflation target range (6 percent) plus potential real GDP growth (4 percent). The estimated rule is obtained by regressing the policy rate on its two lags, inflation expectations, and the output gap, over the period 1997Q1-2012Q2 as in Clarida, Gali, and Gertler, 2000.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Interest rate gap and the deviation of inflation expectations from target 1/

Percent

Citation: IMF Staff Country Reports 2013, 109; 10.5089/9781484395301.002.A001

Source: BCU, IMF staff estimates.Note: Red diamons depict the period 2007Q4-2009Q3; Blue diamonds depict the period 2009Q4-2012Q2.1/ Interest rate gap defined as the difference between the actual policy interest rate and an estimated/calibrated Taylor interestrate rule. The deviation of inflation expectations is measured relative to the center of the target range.

• Download Figure
• Download figure as PowerPoint slide

13. A policy rate persistently below the rule could de-anchor expectations. We examine this question in the next section.

C. Are Inflation Expectations Well Anchored?

14. For inflation expectations to be well-anchored, they need to be aligned with the inflation target and the underlying process of expectations must be independent of actual and lagged inflation. We evaluate these conditions empirically through a basic set of complementary analysis that assesses (i) whether inflation expectations and inflation dynamics are disconnected; and (ii) whether inflation expectations are anchored (or partly anchored) around a specific level (see Annex I for technical details).

15. The findings suggest that inflation expectations are influenced by actual inflation and they fluctuate around 7 percent. First, Granger causality tests suggest that inflation expectations are not completely disconnected from inflation dynamics—as we are unable to reject the null hypotheses that inflation does not Granger cause inflation expectations (Table 2, Panel A). Second, a complementary analysis (see Annex I for details) that examines whether inflation expectations can be described by a weighted average of a constant target, π^*, and past inflation finds that inflation expectations fluctuate around a level—7 percent—that exceeds the ceiling of the inflation target range. Moreover it also finds—in line with the Granger causality test—that inflation expectations are influenced by lagged inflation dynamics (Table 2, Panel B).

Table 2.

Are Inflation Expectations Well-Anchored?

Source: Fund staff estimates. Note: Estimates for 2004–2012. Granger causality tests based on 1 lags as determined by AIC, HQIC and SBIC information criteria. The anchor level and the degree of credibility are estimated as described in the Annex A.1.

Table 2.

Are Inflation Expectations Well-Anchored?

A.- Granger Causality Wald Tests
Dependent variable	Excluded variable	χ2	Prob > χ2
${\pi }_t^e$	πt	0.87	0.34
πt	${\pi }_t^e$	0.97	0.32
B.- Anchor level
${\pi }_t^e$ = λπ* + (1-λ)πt-1
π* = 7.00		λ = 0.84

Source: Fund staff estimates. Note: Estimates for 2004–2012. Granger causality tests based on 1 lags as determined by AIC, HQIC and SBIC information criteria. The anchor level and the degree of credibility are estimated as described in the Annex A.1.

View Table

Table 2.

Are Inflation Expectations Well-Anchored?

A.- Granger Causality Wald Tests
Dependent variable	Excluded variable	χ2	Prob > χ2
${\pi }_t^e$	πt	0.87	0.34
πt	${\pi }_t^e$	0.97	0.32
B.- Anchor level
${\pi }_t^e$ = λπ* + (1-λ)πt-1
π* = 7.00		λ = 0.84

Source: Fund staff estimates. Note: Estimates for 2004–2012. Granger causality tests based on 1 lags as determined by AIC, HQIC and SBIC information criteria. The anchor level and the degree of credibility are estimated as described in the Annex A.1.

D. Disentangling the Underlying Sources of Inflation Dynamics

16. What are the underlying sources of inflation in Uruguay? Specifically, to what extent are expectations, lagged inflation, and costs driving inflation? We frame this discussion in terms of whether inflation dynamics are the result of (i) the dependence of inflation on its own past (“intrinsic persistence”); (ii) the formation of expectations (“expectations-based persistence”) or (iii) fluctuations in the determinants of inflation, such as the output gap or marginal costs (“extrinsic persistence”).¹⁰ Disentangling these sources of inflation is complicated, as they are endogenous, and their relative importance also depends on the monetary policy regime and the policy reaction function (Fuhrer, 2011, and Altissimo et al., 2006).

17. The roles of these factors are evaluated by estimating Phillips curves for Uruguay. Regressions are run using quarterly data over the period 2004–12 (we also report estimates for 1997–2012 for completeness) using the Generalized Method of Moments to address potential endogeneity problems, as in Gali and Gertler (1999).¹¹ Specifically, we estimate a New Keynesian Phillips Curve (NKPC) formally summarized as follows:

Where the variables include lagged headline inflation, π_t; 12-month ahead Consensus Forecasts’ inflation expectations, E_tπ_t+i; and a measure of the output, unemployment gap, or marginal costs, x_t, which we capture by the percentage deviation of quarterly real GDP or of quarterly unemployment from its trend—as obtained from a Hodrick-Prescott filter.¹² Marginal costs are proxied using real wages and the output gap (see Celasun, 2006). We proxy cost push shocks by the deviation of headline inflation from core inflation and the deviation of the real exchange rate from its underlying trend—which captures also for the relative price of tradables to non-tradable goods. Variables are instrumented using one to three lags of the variables. As is standard, all variables are de-meaned.

18. Results confirm that both intrinsic and expectation-based persistence are important in driving inflation dynamics. The coefficient for lagged inflation remains at just over 0.5 and inflation expectations have the correct sign and a magnitude similar to that of the coefficient for lagged inflation (Model 1 in Table 3). These results contrast somewhat with those reported by Gelos and Rossi (2008), who find inflation expectations to be the main driver of the inflation process during 1998–2006, with a limited role for lagged inflation. To some extent this is expected given the time period covered by their study, which includes the 2002 crisis. It is plausible that during crises episodes agents reassess their expectations and become more forward-looking. Our results, which focus on the post-crisis episode, suggest that the role of inflation expectations in driving inflation has become somewhat less robust. It also suggests that the effectiveness of monetary policy may be hampered by the intrinsic inflation persistence.

Table 3.

Phillips Curve Estimates¹

Note: Statistical significance * p<0.05; ** p<0.01; *** p<0.001 Source: BCU, INE and Haver Analytics. Fund staff estimates.

¹Sample ends in 2012Q2.

Table 3.

Phillips Curve Estimates¹

	2004Q1-2012Q2		1997Q1-2012Q2
Variable	1	2	3	4
Lagged CPI	0.55 ***	0.57 ***	0.52 ***	0.62 ***
Inflation expectations	0.50 *	0.20	0.39 ***	0.36 **
Output gap	0.26 **		−0.44 ***
Unemployment gap		−0.08		−0.94 ***
Real Wages	16.81 ***	7.64 *	5.4	5.94
Exchange rate overvaluation	−0.09 *	−0.04 *	−0.01	−0.04
Constant	0.66	0.22	0.57 ***	0.21 ***
R2	0.58	0.61	0.88	0.88

Note: Statistical significance * p<0.05; ** p<0.01; *** p<0.001 Source: BCU, INE and Haver Analytics. Fund staff estimates.

¹Sample ends in 2012Q2.

View Table

Table 3.

Phillips Curve Estimates¹

	2004Q1-2012Q2		1997Q1-2012Q2
Variable	1	2	3	4
Lagged CPI	0.55 ***	0.57 ***	0.52 ***	0.62 ***
Inflation expectations	0.50 *	0.20	0.39 ***	0.36 **
Output gap	0.26 **		−0.44 ***
Unemployment gap		−0.08		−0.94 ***
Real Wages	16.81 ***	7.64 *	5.4	5.94
Exchange rate overvaluation	−0.09 *	−0.04 *	−0.01	−0.04
Constant	0.66	0.22	0.57 ***	0.21 ***
R2	0.58	0.61	0.88	0.88

Note: Statistical significance * p<0.05; ** p<0.01; *** p<0.001 Source: BCU, INE and Haver Analytics. Fund staff estimates.

¹Sample ends in 2012Q2.

Table 4.

Summary Statistics of Inflation and its Components, 2004–2012¹

Sources: BCU. Fund staff estimates.

¹Sample ends in April 2012.

²Autoregressive coefficient from an AR(1) process.

Table 4.

Summary Statistics of Inflation and its Components, 2004–2012¹

Variable	Mean	Sth. Dev.	Min.	Max.	Persistence1
Headline	7.0	1.3	3.4	10.0	0.93
Core	6.6	2.1	1.4	12.0	0.90
Tradables	5.9	2.9	−0.2	12.4	0.94
Nontradables	7.7	1.2	5.6	10.7	0.84
Regulated	5.6	3.5	−1.3	13.8	0.94
Non-regulated	7.6	1.9	3.5	11.4	0.56

Sources: BCU. Fund staff estimates.

¹Sample ends in April 2012.

²Autoregressive coefficient from an AR(1) process.

View Table

Table 4.

Summary Statistics of Inflation and its Components, 2004–2012¹

Variable	Mean	Sth. Dev.	Min.	Max.	Persistence1
Headline	7.0	1.3	3.4	10.0	0.93
Core	6.6	2.1	1.4	12.0	0.90
Tradables	5.9	2.9	−0.2	12.4	0.94
Nontradables	7.7	1.2	5.6	10.7	0.84
Regulated	5.6	3.5	−1.3	13.8	0.94
Non-regulated	7.6	1.9	3.5	11.4	0.56

Sources: BCU. Fund staff estimates.

¹Sample ends in April 2012.

²Autoregressive coefficient from an AR(1) process.

Table 5:

Individual Countries Inflation Target

Note: CB = Central Bank; G = Government; H CPI = Headline CPI; P+T = Point with tolerance band.

¹In percentage points.

²Target proposed by central bank at start of 2012, pending cabinet approval.

Table 5:

Individual Countries Inflation Target

Country	Target Set By	Target Measure	Target 2012 1	Target Type	Multiple Targets	Target Horizone
Brazil	G and CB	H CPI	4.5 ± 2	P+T	2012 and 2013	Yearly Target
Chile	CB	H CPI	3 ± 1	P+T		Around two years
Colombia	CB	H CPI	2 - 4	Range		Medium term
Guatemala	CB	H CPI	4.5 ± 1	P+T	2012 and 2013	End of year
Mexico	CB	H CPI	3 ± 1	P+T		Medium term
Peru	CB	H CPI	2 ± 1	P+T		At all times
Uruguay	G and CB	H CPI	4 - 6	Range		18-month
Australia	G and CB	H CPI	2 - 3	Range		Medium term
Canada	G and CB	H CPI	2 (mid-point of 13)	P+T		Six-eight quarter; Current target2016.
New Zealand	G and CB	H CPI	1 - 3	Range		Medium term
Norway	G	H CPI	2.5	Point		Medium term
South Korea	CB (with GB)	H CPI	3 ± 1	P+T		Three years
Sweden	CB	H CPI	2	Point		Normally two years
United Kingdom	G and CB	H CPI	2	Point		At all times
Czech Republic	CB	H CPI	2 ± 1	P+T		Medium term, 12–18 months
Hungary	CB	H CPI	3	Point		Medium term
Indonesia	G and CB	H CPI	4.5 ± 1	P+T		Medium term
Israel	G and CB	H CPI	1 -3	Range		Within two years
Philippines	G and CB	H CPI	4 ± 1	P+T		Medium term (from 2012–2014)
Poland	CB	H CPI	2.5 ± 1	P+T		Medium term
Romania	G and CB	H CPI	3 ± 1	P+T		Medium-term target from 2013
Serbia	G and CB	H CPI	4 ± 1	P+T		Medium term
South Africa	G	H CPI	3 -6	Range		On a continuous basis
Thailand	G and CB	H CPI 2	3 ± 1.5 2	P+T	Target set annually	Eight quarters
Turkey	G and CB	H CPI	5 ± 2	P+T	2012 and 2013	Multi year (three years)

Note: CB = Central Bank; G = Government; H CPI = Headline CPI; P+T = Point with tolerance band.

¹In percentage points.

²Target proposed by central bank at start of 2012, pending cabinet approval.

View Table

Table 5:

Individual Countries Inflation Target

Country	Target Set By	Target Measure	Target 2012 1	Target Type	Multiple Targets	Target Horizone
Brazil	G and CB	H CPI	4.5 ± 2	P+T	2012 and 2013	Yearly Target
Chile	CB	H CPI	3 ± 1	P+T		Around two years
Colombia	CB	H CPI	2 - 4	Range		Medium term
Guatemala	CB	H CPI	4.5 ± 1	P+T	2012 and 2013	End of year
Mexico	CB	H CPI	3 ± 1	P+T		Medium term
Peru	CB	H CPI	2 ± 1	P+T		At all times
Uruguay	G and CB	H CPI	4 - 6	Range		18-month
Australia	G and CB	H CPI	2 - 3	Range		Medium term
Canada	G and CB	H CPI	2 (mid-point of 13)	P+T		Six-eight quarter; Current target2016.
New Zealand	G and CB	H CPI	1 - 3	Range		Medium term
Norway	G	H CPI	2.5	Point		Medium term
South Korea	CB (with GB)	H CPI	3 ± 1	P+T		Three years
Sweden	CB	H CPI	2	Point		Normally two years
United Kingdom	G and CB	H CPI	2	Point		At all times
Czech Republic	CB	H CPI	2 ± 1	P+T		Medium term, 12–18 months
Hungary	CB	H CPI	3	Point		Medium term
Indonesia	G and CB	H CPI	4.5 ± 1	P+T		Medium term
Israel	G and CB	H CPI	1 -3	Range		Within two years
Philippines	G and CB	H CPI	4 ± 1	P+T		Medium term (from 2012–2014)
Poland	CB	H CPI	2.5 ± 1	P+T		Medium term
Romania	G and CB	H CPI	3 ± 1	P+T		Medium-term target from 2013
Serbia	G and CB	H CPI	4 ± 1	P+T		Medium term
South Africa	G	H CPI	3 -6	Range		On a continuous basis
Thailand	G and CB	H CPI 2	3 ± 1.5 2	P+T	Target set annually	Eight quarters
Turkey	G and CB	H CPI	5 ± 2	P+T	2012 and 2013	Multi year (three years)

Note: CB = Central Bank; G = Government; H CPI = Headline CPI; P+T = Point with tolerance band.

¹In percentage points.

²Target proposed by central bank at start of 2012, pending cabinet approval.

Table 6:

Decision Making in Inflation Targeting Central Banks

Source: Hammond (2012) and Central Banks. Note: CB = Central Bank; G = Government; H CPI = Headline CPI; P+T = Point with tolerance band.

¹Currently 7.

²Currently 7.

Table 6:

Decision Making in Inflation Targeting Central Banks

Country	Number on Policy Making Committee	External Members?	Meetings per Year	Governor’s Term (years)	Decision Making Process	Votes Published?
Brazil	8	No	8	No Fixed Term	Vote	Balance of Votes
Chile	5	No	12	5	Vote	Yes
Colombia	7	No	12	4	Vote	Majority or Unanimous
Guatemala	8	7	8	4	Vote	No
Mexico	5	No	8	6	Consensus	n/a
Peru	7	No	12	Term of Parliament	Vote	No
Uruguay	6	No	4	No Fixed Term		No
Australia	9	6	11	7	Consensus	n/a
Canada	6	No	8	7	Consensus	n/a
New Zealand	Governor	n/a	8	5	Governor decides	n/a
Norway	7	5	6	6	Consensus	n/a
South Korea	7	5	12	4	Vote	No
Sweden	6	No	6	6	Vote	Yes
United Kingdom	9	4	12	5	Vote	Yes
Czech Republic	7	No	8	6	Vote	Yes
Hungary	5 to 9 1	4	12	6	Vote	Yes
Indonesia	6 to 9 2	No	12	5	Consensus	n/a
Israel	6	3	12	5	Vote	Balance of Votes
Philippines	7	No	8	6	Vote	No
Poland	10	9	12	6	Vote	Yes in Inflation Report
Romania	9	5	8	5	Vote	No
Serbia	5	No	12	6	Vote	No
South Africa	7	No	6	5	Consensus	n/a
Thailand	7	4	8	5	Vote	Balance of Votes
Turkey	7	1	12	5	Vote	No

Source: Hammond (2012) and Central Banks. Note: CB = Central Bank; G = Government; H CPI = Headline CPI; P+T = Point with tolerance band.

¹Currently 7.

²Currently 7.

View Table

Table 6:

Decision Making in Inflation Targeting Central Banks

Country	Number on Policy Making Committee	External Members?	Meetings per Year	Governor’s Term (years)	Decision Making Process	Votes Published?
Brazil	8	No	8	No Fixed Term	Vote	Balance of Votes
Chile	5	No	12	5	Vote	Yes
Colombia	7	No	12	4	Vote	Majority or Unanimous
Guatemala	8	7	8	4	Vote	No
Mexico	5	No	8	6	Consensus	n/a
Peru	7	No	12	Term of Parliament	Vote	No
Uruguay	6	No	4	No Fixed Term		No
Australia	9	6	11	7	Consensus	n/a
Canada	6	No	8	7	Consensus	n/a
New Zealand	Governor	n/a	8	5	Governor decides	n/a
Norway	7	5	6	6	Consensus	n/a
South Korea	7	5	12	4	Vote	No
Sweden	6	No	6	6	Vote	Yes
United Kingdom	9	4	12	5	Vote	Yes
Czech Republic	7	No	8	6	Vote	Yes
Hungary	5 to 9 1	4	12	6	Vote	Yes
Indonesia	6 to 9 2	No	12	5	Consensus	n/a
Israel	6	3	12	5	Vote	Balance of Votes
Philippines	7	No	8	6	Vote	No
Poland	10	9	12	6	Vote	Yes in Inflation Report
Romania	9	5	8	5	Vote	No
Serbia	5	No	12	6	Vote	No
South Africa	7	No	6	5	Consensus	n/a
Thailand	7	4	8	5	Vote	Balance of Votes
Turkey	7	1	12	5	Vote	No

Source: Hammond (2012) and Central Banks. Note: CB = Central Bank; G = Government; H CPI = Headline CPI; P+T = Point with tolerance band.

¹Currently 7.

²Currently 7.

Table 7:

Accountability and Transparency in Inflation Targeting Central Banks

Source: Hammond (2012) and Central Banks. Note: IR = Inflation Report; PC = Press conference; PR = Press release.

Table 7:

Accountability and Transparency in Inflation Targeting Central Banks

Country	Open Letter	Parliamentary Hearings?	Press Notice /Conference	Minutes	Votes	Inflation Report	Frequency
Brazil	Yes	Yes, six per year	PR+PC for IR	Yes, after eight days	Balance of votes	Yes	4
Chile	No	Yes, four times per year	PR	Yes, after two weeks	Yes	Yes	4
Colombia	No	Yes, twice yearly	PR, PC for IR	Yes, after two weeks	Majority or unanimous	Yes	4
Guatemala	No	Yes, twice a year	PR + PC	Yes, after four weeks	No	Yes	3
Mexico	No	Yes, not regularly	PR	Yes, after two weeks	n/a	Yes	4
Peru	No	Yes, once a year	Teleconference	No	No	Yes	4
Uruguay	No	No	PC+IR	No	No	Yes	4
Australia	No	Yes, twice yearly	Notice	Yes, after two weeks	n/a	Yes	4
Canada	No	Yes, twice yearly	PR+PC for IR	No	n/a	Yes	4
New Zealand	Other	Yes, four times per year	PR, PC for IR	No	n/a	Yes	4
Norway	No	Yes	PR + PC	No	n/a	Yes	3
South Korea	No	Yes	PR + PC	Yes, after six weeks	No	Yes	2
Sweden	No	Yes, twice yearly	PR	Yes, after two weeks	Yes	Yes	3 plus 3
United Kingdom	Yes	Yes, three per year	PR + PC for IR	Yes, after two weeks	Yes	Yes	4
Czech Republic	No	No (Report)	PR, PC for IR	Yes, after eight days	Yes	Yes	4
Hungary	No	Yes, once a year	PC	Yes	Yes	Yes	4
Indonesia	No	No	PR	No	n/a	Yes	4
Israel	No	Yes, twice yearly	PR	Yes, after two weeks	Balance of votes	Yes	2
Philippines	Yes	No	PR + PC	Yes, after four weeks	No	Yes	4
Poland	No	No	PR + PC	Yes, after three weeks	Yes in Inflation Report	Yes	4
Romania	No	No	PR + PC for IR	No	No	Yes	4
Serbia	Yes	No	PR + PC	No	No	Yes	4
South Africa	No	Yes, at least three per year	PR + PC	No	n/a	Yes	2
Thailand	Yes	No	PR + PC	Yes, after two weeks	Balance of votes	Yes	4
Turkey	Yes	Yes, twice a year	PR	Yes	No	Yes	4

Source: Hammond (2012) and Central Banks. Note: IR = Inflation Report; PC = Press conference; PR = Press release.

View Table

Table 7:

Accountability and Transparency in Inflation Targeting Central Banks

Country	Open Letter	Parliamentary Hearings?	Press Notice /Conference	Minutes	Votes	Inflation Report	Frequency
Brazil	Yes	Yes, six per year	PR+PC for IR	Yes, after eight days	Balance of votes	Yes	4
Chile	No	Yes, four times per year	PR	Yes, after two weeks	Yes	Yes	4
Colombia	No	Yes, twice yearly	PR, PC for IR	Yes, after two weeks	Majority or unanimous	Yes	4
Guatemala	No	Yes, twice a year	PR + PC	Yes, after four weeks	No	Yes	3
Mexico	No	Yes, not regularly	PR	Yes, after two weeks	n/a	Yes	4
Peru	No	Yes, once a year	Teleconference	No	No	Yes	4
Uruguay	No	No	PC+IR	No	No	Yes	4
Australia	No	Yes, twice yearly	Notice	Yes, after two weeks	n/a	Yes	4
Canada	No	Yes, twice yearly	PR+PC for IR	No	n/a	Yes	4
New Zealand	Other	Yes, four times per year	PR, PC for IR	No	n/a	Yes	4
Norway	No	Yes	PR + PC	No	n/a	Yes	3
South Korea	No	Yes	PR + PC	Yes, after six weeks	No	Yes	2
Sweden	No	Yes, twice yearly	PR	Yes, after two weeks	Yes	Yes	3 plus 3
United Kingdom	Yes	Yes, three per year	PR + PC for IR	Yes, after two weeks	Yes	Yes	4
Czech Republic	No	No (Report)	PR, PC for IR	Yes, after eight days	Yes	Yes	4
Hungary	No	Yes, once a year	PC	Yes	Yes	Yes	4
Indonesia	No	No	PR	No	n/a	Yes	4
Israel	No	Yes, twice yearly	PR	Yes, after two weeks	Balance of votes	Yes	2
Philippines	Yes	No	PR + PC	Yes, after four weeks	No	Yes	4
Poland	No	No	PR + PC	Yes, after three weeks	Yes in Inflation Report	Yes	4
Romania	No	No	PR + PC for IR	No	No	Yes	4
Serbia	Yes	No	PR + PC	No	No	Yes	4
South Africa	No	Yes, at least three per year	PR + PC	No	n/a	Yes	2
Thailand	Yes	No	PR + PC	Yes, after two weeks	Balance of votes	Yes	4
Turkey	Yes	Yes, twice a year	PR	Yes	No	Yes	4

Source: Hammond (2012) and Central Banks. Note: IR = Inflation Report; PC = Press conference; PR = Press release.

19. Extrinsic persistence, as captured by the output gap and real wages, is statistically significant (Models 1 and 3 in Table 3). The role of real wages is quite relevant given the degree of wage indexation. Results are somewhat less satisfactory when using the unemployment gap, which turns out to be statistically insignificant in the most recent sample, but quite relevant for the longer sample. Also, of the two cost-push shock measures, only the real exchange rate measure is significant (with real exchange rate appreciations contributing to lower inflation). The deviation of headline from core inflation did not result in significant results and are not reported.

20. Our analysis suggest that the inflation process in Uruguay is driven by both past and expected inflation. Moreover, inflation is influenced by costs, in particular the degree of spare capacity (as captured by the output gap) and labor costs (as captured by wages). In recent years, up to 90 percent of collective wage agreements include clauses with ex-post corrections for the deviation of actual from expected inflation (Melgar et. al., 2011). This is likely to have feedback effects on inflation and its expectations. Although our model does not have a formal test for assessing the relevance of wage indexation on inflation, its effects are captured by the coefficient on lagged inflation in a similar way that wage indexation was captured by lagged inflation in the wage Phillips curve estimated by Melgar et al., (2011). Overall, wage indexation may help explain why lagged inflation remains an important driver of inflation dynamics. Finally, the exchange rate is found to have an effect on inflation. In particular, it appears that a narrowing but persistent undervaluation of the real exchange rate since 2002–03 has contributed to higher inflation.

E. What Can Be Done to Strengthen the Monetary Policy Framework?

21. The BCU could improve further some aspects of the monetary policy framework to increase its influence over inflation expectations. In this regard, BCU could consider to:

• Communicate in a clearer manner the likely future direction of monetary policy. Open and transparent communication can enhance policy effectiveness. It has become standard practice for central banks under inflation targeting regimes to indicate the rationale behind policy actions and the expected outcomes, and to give forward guidance on future policy actions. Although BCU has made important progress in some of these areas, it could strengthen its guidance on future policy actions by publishing in its statements a more detailed assessment of its “bias” with respect to future changes in monetary policy. This would help the BCU influence inflation expectations better.

• Publish its conditional forecasts of inflation along with an explanation of the risks surrounding the forecast. Most inflation targeting countries publish an inflation forecast, usually quarterly, and many even publish core inflation forecasts (Hammond, 2012; Fracasso, et al 2003). Publishing inflation forecasts would help the BCU communicate to the public its views about future inflation and how it will converge to the target range.¹³

• Ensure timely communication with the market. More frequent meetings of the monetary policy committee (COPOM) could also help provide better guidance to the market about the stance of monetary policy and future policy directions, and thus help anchor inflation expectations around the target. BCU has four policy committee meetings a year, almost half the number in other inflation targeting countries in the region and the rest of the world.¹⁴

View Full Size

Decision Making at Central Banks

Number of policy committee meetings per year

Citation: IMF Staff Country Reports 2013, 109; 10.5089/9781484395301.002.A001

Source: IMF Staff on the basis of Hammond (2012).

• Download Figure
• Download figure as PowerPoint slide

F. Conclusions

22. Uruguay’s inflation and inflation expectations exceed the inflation target and the gap has been widening in recent years. This paper has argued that one reason for this increasing gap is that the stance of monetary policy has deviated from that implied by the Taylor rule as well as a rule estimated for Uruguay using past data.

23. To help bring inflation and its expectations to the mid-point of the target BCU needs to maintain a tightening bias. The tightening pace should depend on the evolution of the economy, including the output gap, credit dynamics, and the exchange rate.

24. In addition, the BCU could also strengthen its communication. It can take a more determined, clear, and transparent ‘action path’ that explains how inflation will be brought to the center of the target range. Given the widened deviation of inflation from target, bringing inflation and its expectations back to the center of the target range might be somewhat more demanding today than a few years ago. Thus, in addition to a continued tightening of the monetary policy stance, and stronger monetary policy communications, concerted effort on other fronts—including prudent wage increases and counter-cyclical fiscal policy would also be helpful.