inflation report – september 2018 - bcb.gov.br · rate depends on the conditioning selic rate...

September 2018 \ Banco Central do Brasil \ Inflation Report \ 53

Inflation outlook 2This chapter of the Inflation Report analyses the inflation outlook up to 2021, thus covering all years for which the National Monetary Council (CMN) sets inflation targets.

Projections presented herein use the information set available at the last meeting of the Copom held on September 18th and 19th, 2018. For the conditioning paths used in the projections, especially those arising from the Focus survey carried out by the Banco Central do Brasil, the cutoff date is September 14th, 2018, unless otherwise indicated.

Conditional projections for inflation are presented in four scenarios, depending on the conditioning path used for the exchange and Selic rates over the projection horizon. The conditioning paths may be derived from expectations extracted from the Focus survey or paths in which the values of these variables remain constant throughout the projection horizon.

The first scenario assumes constant Selic and interest rates during the projection horizon, while the second scenario supposes paths extracted from the Focus survey for these two variables.

Two other scenarios – namely “hybrid” scenarios – are also presented. The first scenario assumes a constant Selic rate and exchange rate extracted from the Focus survey, while the second assumes the Selic rate from the Focus survey and a constant exchange rate.

It is worth noting that the conditional inflation projections disclosed in this Report contemplate probability intervals that embody the degree of uncertainty present at the aforementioned cutoff date. The projections depend not only on the assumptions about interest rates and exchange rates but also on a set of assumptions about the behavior of exogenous variables.

In its decision‑making process, the Copom analyzes a broad set of variables and models, in which it exercises judgments based on the available

54 \ Inflation Report \ Banco Central do Brasil \ September 2018

information set. In presenting some scenarios that inform its deliberations, the Copom seeks to increase the transparency of monetary policy decisions, contributing to its effectiveness in controlling inflation, which is its primary objective.

2.1 Revisions and short-term projections

In the quarter ending in August, consumer inflation as measured by the IPCA was 0.04 p.p. below the baseline scenario presented in the previous Inflation Report (Table 2.1). The IPCA registered high volatility in the period, reaching the highest rate for the month of June since 1995 and the first negative variation for the month of August over the last 20 years.

The effect of the halt in the transportation sector on food prices was higher than anticipated, accounting for a significant share of the surprise of 0.26 p.p. in the IPCA of June and July. Conversely, the IPCA of August was lower than projected (‑0.09 percent against 0.20 percent), thus offsetting the initial surprise and confirming the evaluation that the halt represented only a temporary shock. It should be emphasized the more favorable evolution of prices of ethanol, airfare prices, and food in August.

In this Report, short‑term projections in the scenarios with constant Selic and exchange rates and with Selic and exchange rates from the Focus Survey show a difference of 0.2 p.p., reflecting the distinct trajectories expected for the exchange rates. Thus, exceptionally in this Report, projections are presented according to these two scenarios.

Short‑term projections associated with the scenario with constant Selic and exchange rates point to respective variations of 0.40 percent, 0.43 percent and 0.24 percent for the IPCA from September to November 2018. If this trajectory materializes, the increase of 1.07 percent for the IPCA in the quarter would be higher than that observed in the same period of 2017 (0.86 percent), implying a rise in the 12‑month inflation rate from 4.19 percent, in August, to 4.41 percent in November (Table 2.2). In the scenario with Selic and exchange rates extracted from the Focus Survey, the projections for the same months are 0.38 percent, 0.34 percent, and 0.15 percent, respectively. If that is the case, the quarterly

Table 2.1 – IPCA – Inflationary surprise

% change

2018

Jun Jul Aug Quarter12 months up to Aug

Copom's scenario1/ 1,06 0,27 0,20 1,54 4,23

Realized IPCA 1,26 0,33 -0,09 1,50 4,19

Surprise 0,20 0,06 -0,29 -0,04 -0,04

Sources: IBGE and BCB

1/ Scenario at June 2018 Inflation Report cutoff date.

Table 2.2 – IPCA – Short-term projections –

Copom's scenarios

% change

Sep Oct Nov Quarter12 months up to Nov

Constant Selic and exchange

rates 0,40 0,43 0,24 1,07 4,41

Selic and exchange rate from

Focus survey 0,38 0,34 0,15 0,87 4,20

Sources: IBGE and BCB

1/ Scenarios at the cutoff date.

2018


inflation would reach 0.87 percent in November, and 4.20 percent in the 12‑month period, a level similar to that of August.

Even though limited by the high slack of production factors and by inflation expectations anchored in the target27, the recent exchange devaluation tends to raise consumer inflation over the next months, mainly through the direct impact on fuel prices. In the scenario with the exchange rate extracted from the Focus survey, this movement is mitigated by the trajectory of exchange appreciation expected until the end of this year. Additionally, inflation in the next quarter should also be impacted by the end of the favorable seasonality of prices of ethanol and in natura food, as well as by sharper increases in airfare prices in September and October. In the opposite direction, it should be noted the maintenance of the process of normalization of livestock prices following the dissipation of the effects of the halt in the transportation sector.

2.2 Conditional projections

The exchange rate used in the scenarios that assume a constant value for this variable changed from R$3.70/US$, in the June 2018 Inflation Report, to R$4.15/US$28 (Figure 2.1). The median of expectations for the year‑end exchange rate, extracted from the Focus survey of September 14th, 2018, when compared to the June 15th, 2018 values used in the June 2018 Inflation Report, increased for all years: from R$3.63/US$ to R$3.83/US$ for 2018, from R$3.60/US$ to R$3.75/US$ for 2019, and from R$3.60/US$ to R$3,70/US$ for 2020. For the end of 2021, the exchange rate is expected to be at R$3.80/US$.

As for the Selic rate, the value assumed in the scenarios using a constant rate was kept at 6.50 percent p.a. (Figure 2.2). Consistent with this Selic rate path and the increase in risk premia, the projected 360‑day pre‑DI swap rate, after a significant rise in the third quarter, is expected to decline in the following two quarters, stabilizing thereafter.

The median of expectations for the Selic rate extracted from the Focus survey – under the same comparison basis – between June 15th, 2018, and

27/ See box “Exchange rate pass‑through from the perspective of a semi‑structural model” in this Report.28/ Value obtained by the usual procedure of rounding the average quotation of the R$/US$ exchange rate observed during the five

business days ended on the Friday prior to the Copom meeting.

Figure 2.1 – Exchange rate assumptions for projections

Note: Values refer to quarterly averages.

3.0

3.2

3.4

3.6

3.8

4.0

4.2

4.4

I2018

II III IV I2019

II III IV I2020

II III IV I2021

II III IV

Focus survey (IR Jun 18)Constant (IR Jun 18)Constant (IR Sep 18)Focus survey (IR Sep 18)

R$/US$

Figure 2.2 – Selic rate target assumptions for projections

Note: Values refer to quarterly averages.

6

7

8

9

I2018

II III IV I2019

II III IV I2020

II III IV I2021

II III IV I2022

II III IV

Focus survey (IR Jun 18)Constant (IR Jun 18)Constant (IR Sep 18)Focus survey (IR Sep 18)

% p.a.


September 14th, 2018, remained at 6.50 percent p.a. for the end of 2018, and 8.00 p.a. for the end of 2019, rising to 8.13 percent p.a. for the end of 2020 and returning to 8.0 percent p.a. for the end of 2021, remaining at this level until the end of 2022 (Figure 2.2)29. In this conditioning path, the Selic rate begins to increase in May 2019, reaching 8.00 percent p.a. in October 2019. Consistent with this path for the Selic rate and the trajectory of risk premia, the projected rate of the 360‑day pre‑DI swap keeps the upward trend begun in 2018Q2, becoming relatively stable from the end of 2019 onwards.

The projections presented herein still depend on considerations about the evolution of the necessary reforms and adjustments in the economy. Its effects on projections are captured through asset prices, the degree of uncertainty, the expectations from the Focus survey, and their effect on the structural interest rate of the economy. In addition to these channels, fiscal policy influences the conditional projections for inflation through impulse effects on the aggregate demand.

These projections also embed the understanding that the process of structural reforms – such as fiscal and lending reforms – contributes to the gradual reduction of the structural interest rate.

The projections for the IPCA variation accumulated in four quarters were based on the combination of the above short‑term projections and conditioning paths. These projections are based on a set of models and information available, combined with the exercise of judgment.

The central projection associated with the scenario that combines constant interest and exchange rates over the entire projection horizon indicates that the inflation accumulated in four quarters, after reaching 4.39 percent in 2018Q2, increases to around 4.4 percent towards the end of the year (Figure 2.3 and Tables 2.3 and 2.4). The projected inflation reaches a peak of about 5.0 percent in 2019Q1, also influenced by the discarding of the unusually low quarterly inflation rate in 2018Q1. Projected inflation for the ending of 2019 lies around 4.5 percent, and, for 2020 and 2021, stays at 4.2 percent. In this scenario, the projections for administered price

29/ As described in the box “Small‑scale aggregate model – 2017” (June 2017 Inflation Report), the path of the 360‑day pre‑DI swap rate depends on the conditioning Selic rate path for the same period and the premium path (difference between the swap rate and the expected rate for the Selic). Therefore, the swap rate throughout 2021 also depends on the Selic trajectory over 2022.

Year Q

2018 III 4.4 4.4 4.4 4.4

2018 IV 4.50 4.4 4.1 4.4 4.1

2019 I 5.0 4.5 5.0 4.5

2019 II 4.3 3.7 4.3 3.7

2019 III 4.5 3.9 4.5 3.9

2019 IV 4.25 4.5 4.0 4.4 4.1

2020 I 4.4 3.9 4.2 4.1

2020 II 4.3 3.8 4.0 4.1

2020 III 4.2 3.7 3.9 4.0

2020 IV 4.00 4.2 3.6 3.8 4.0

2021 I 4.2 3.6 3.8 4.0

2021 II 4.2 3.6 3.7 4.1

2021 III 4.2 3.7 3.7 4.2

2021 IV 3.75 4.2 3.8 3.7 4.3

Note: Year-on-year IPCA inflation (%).

Table 2.3 – Central projections in different scenarios

Constant Selic and exchange

rates

Selic and exchange rates from

Focus survey

Exchange rate from

Focus survey and constant Selic rate

Selic rate from Focus survey and constant exchange

rate

Inflation target

Figure 2.3 – Projected inflation – Scenario with constant

Selic and exchange rates

Inflation fan chart


0

2

4

6

8

I2018

II III IV I2019

II III IV I2020

II III IV I2021

II III IV

%


inflation are around 8.3 percent for 2018, 5.7 percent for 2019 and 4.3 percent for 2020 and 2021.

In this scenario, the estimated probabilities of inflation exceeding the upper and lower limits of the target tolerance range are close to 0 and 1 percent in 2018, 19 and 11 percent for 2019, 18 and 11 percent for 2020, and 23 and 8 percent for 2021, respectively.

In comparison with the June 2018 Inflation Report (Table 2.5), projections for 2018, 2019 and 2020 in the scenario with constant interest and exchange rates increased by around 0.2 p.p., 0.4 p.p. and 0.1 p.p., respectively.

The exchange rate depreciation in 2018Q3 was the main factor driving the increase in the projections in comparison with the June Inflation Report. The exchange rate rose from an average of R$3.61/US$ in the second quarter to around R$4.15/US$ in the week prior to the September Copom meeting (217th meeting). Considering the quarterly average, the assumed constant exchange rate for the projection horizon also implies an exchange depreciation for the final quarter of 2018 when compared with the previous quarter, with direct impacts on inflation mainly concentrated between the end of 2018 and the beginning of 2019. The inertial inflation mechanisms are also a factor that contributes to the ascent of inflation projections in 2019 in comparison with the June Inflation Report.

In the same direction, it should be emphasized the contribution of rising inflation projection for administered prices, not only because of the upper trend for the exchange rate but also the trajectory of other items, like electric power rates.

The constant trajectory assumed for the Selic rate in this scenario – stimulative over the entire horizon under consideration – also contributes to maintaining inflation projections above the target, especially for longer horizons.

On the other hand, significant factors contribute to mitigate the effects of the exchange rate depreciation30 and bring inflation to a downward trajectory, such as inflation expectations anchored in the target, the high slack in production factors, the slower recovery path of economic activity and the less intense effects of exchange rate depreciation over time.

30/ See box “Exchange rate pass‑through from the perspective of a semi‑structural model” in this Report.

Table 2.4 – Projected inflation – Scenario with constant

Central projection and probability intervals

50%

Year Q 30%

10%

Central

2018 III 4.3 4.3 4.4 4.4 4.4 4.5 4.5

2018 IV 4.0 4.2 4.3 4.4 4.5 4.6 4.8

2019 I 4.4 4.7 4.9 5.0 5.1 5.3 5.6

2019 II 3.5 3.9 4.2 4.3 4.4 4.7 5.1

2019 III 3.6 4.0 4.3 4.5 4.7 5.0 5.4

2019 IV 3.6 4.0 4.3 4.5 4.7 5.0 5.4

2020 I 3.5 3.9 4.2 4.4 4.6 4.9 5.3

2020 II 3.4 3.8 4.1 4.3 4.5 4.8 5.2

2020 III 3.3 3.7 4.0 4.2 4.4 4.7 5.1

2020 IV 3.3 3.7 4.0 4.2 4.4 4.7 5.1

2021 I 3.3 3.7 4.0 4.2 4.4 4.7 5.1

2021 II 3.3 3.7 4.0 4.2 4.4 4.7 5.1

2021 III 3.3 3.7 4.0 4.2 4.4 4.7 5.1

2021 IV 3.3 3.7 4.0 4.2 4.4 4.7 5.1


Selic and exchange rates

Year Q June Inflation Report September Inflation Report

2018 III 4.3 4.4

2018 IV 4.2 4.4

2019 I 4.8 5.0

2019 II 3.9 4.3

2019 III 3.8 4.5

2019 IV 4.1 4.5

2020 I 4.2 4.4

2020 II 4.2 4.3

2020 III 4.1 4.2

2020 IV 4.1 4.2

2021 I 4.2

2021 II 4.2

2021 III 4.2

2021 IV 4.2


Table 2.5 – Projections in the previous and current Inflation Reports – Scenario with constant Selic and exchange rates


In comparison with the inflation projections of the August Copom meeting (216th Meeting), there was an increase of approximately 0.2 p.p. for 2018 and 0.4 p.p. for 2019 (see Minutes of the 216th Meeting).

In the scenario with interest and exchange rates from the Focus survey, the central projection indicates that the inflation accumulated in four quarters ends 2018 at approximately 4.1 percent. After reaching a maximum of 4.5 percent in 2019Q1, the year‑on‑year inflation ends 2019 at around 4.0 percent, 2020 at 3.6 percent and 2021 at 3.8 percent (Figure 2.4 and Table 2.6). In this scenario, the projections for administered prices inflation are 7.7 percent for 2018, 5.4 percent for 2019, 3.8 percent for 2020, and 4.2 percent for 2021.

In this scenario, compared to the June 2018 Inflation Report (Table 2.7), inflation projections for 2018 declined, mainly due to the negative inflation surprise accumulated from June to August and the lower projections for the last months of the year. For 2019, projections increased around 0.3 p.p., mostly due to the revision of some items of administered prices. With regard to 2020, the exchange rate appreciation foreseen by the Focus survey during the year, unlike the trajectory foreseen in the previous Inflation Report, contributes for a decline of around 0.1 p.p. in the projections.

In comparison with the scenario that uses constant Selic and exchange rates (Table 2.3), projected inflation is always lower from 2018Q4 onwards as of 2018Q4. For shorter horizons, the main factor conditioning this trajectory is the exchange rate. Unlike the scenario that uses constant exchange rate, the Focus survey foresees a trajectory of exchange rate appreciation from 2018Q4 to 2020Q4, followed by an exchange rate depreciation from 2021Q1 onwards, which reduces the difference in projections between both scenarios.31 The increase in the Selic rate predicted in the Focus survey, which anticipates an increase in the pre‑DI swap rate, also acts to reduce inflation projections, since it implies a moderation of economic activity.

When compared to the June Inflation Report, the difference between projections according to both scenarios accentuates, due to the increased difference in the exchange rate trajectories, as shown by Figure 2.1.

31/ See box “Past variation and exchange rate predictability” in this Report.

Figure 2.4 – Projected inflation – Scenario with Selic

and exchange rates from Focus survey

Inflation fan chart


0

2

4

6

8

I2018

II III IV I2019

II III IV I2020

II III IV I2021

II III IV

%


50%

Year Q 30%

10%

Central

2018 III 4.3 4.3 4.4 4.4 4.4 4.5 4.5

2018 IV 3.7 3.9 4.0 4.1 4.2 4.3 4.5

2019 I 3.9 4.2 4.4 4.5 4.6 4.8 5.1

2019 II 2.9 3.3 3.6 3.7 3.8 4.1 4.5

2019 III 3.0 3.4 3.7 3.9 4.1 4.4 4.8

2019 IV 3.1 3.5 3.8 4.0 4.2 4.5 4.9

2020 I 3.0 3.4 3.7 3.9 4.1 4.4 4.8

2020 II 2.9 3.3 3.6 3.8 4.0 4.3 4.7

2020 III 2.8 3.2 3.5 3.7 3.9 4.2 4.6

2020 IV 2.7 3.1 3.4 3.6 3.8 4.1 4.5

2021 I 2.7 3.1 3.4 3.6 3.8 4.1 4.5

2021 II 2.7 3.1 3.4 3.6 3.8 4.1 4.5

2021 III 2.8 3.2 3.5 3.7 3.9 4.2 4.6

2021 IV 2.9 3.3 3.6 3.8 4.0 4.3 4.7


Table 2.6 – Projected inflation – Scenario with Selic and exchange rates from Focus survey


In this scenario with Selic and exchange rates extracted from the Focus survey, the estimated probabilities of inflation exceeding the upper and lower limits of the target tolerance range are close to 0 and 2 percent in 2018, 10 and 18 percent in 2019, 9 and 21 percent in 2020, and 15 and 13 percent in 2021, respectively.

In the hybrid scenario with constant exchange rate and Selic rate from the Focus survey, inflation projections are around 4.4 percent, 4.4 percent, 3.8 percent and 3.7 percent for 2018, 2019, 2020 and 2021, respectively (Table 2.8). In comparison with the scenario that, alternatively, uses constant Selic rate (Table 2.3), the projections decline from 2019Q4 onwards, reflecting the upward Selic rate trajectory (and pre‑DI swap rate).

Finally, in the hybrid scenario with exchange rate from the Focus survey and constant Selic rate, inflation projections are approximately 4.1 percent, 4.1 percent, 4.0 percent and 4.3 percent for 2018, 2019, 2020 and 2021, respectively (Table 2.9). In comparison with the scenario that alternatively uses constant exchange rates (Table 2.3), the projections are lower between 2018Q4 and 2021Q2, as a result of the exchange appreciation path extracted from the Focus survey until the end of 2020. For 2021, the projections increase due to the exchange rate depreciation foresaw for that year.

2.3 Monetary policy conduct and balance of risks

Recent indicators of economic activity point to the recovery of the Brazilian economy at a more gradual pace than envisaged early this year.

The global outlook remains challenging, with reduction of risk appetite towards emerging economies. The main risks continue to be associated with normalization of interest rates in some advanced economies and with uncertainty regarding global trade.

Inflation expectations for 2018 and 2019 collected by the Focus survey are around 4.1 percent. For 2020 and 2021, expectations are around 4.0% and 3.9%, respectively.

Year Q June Inflation Report September Inflation Report

2018 III 4.3 4.4

2018 IV 4.2 4.1

2019 I 4.7 4.5

2019 II 3.6 3.7

2019 III 3.6 3.9

2019 IV 3.7 4.0

2020 I 3.8 3.9

2020 II 3.8 3.8

2020 III 3.7 3.7

2020 IV 3.7 3.6

2021 I 3.6

2021 II 3.6

2021 III 3.7

2021 IV 3.8


Table 2.7 – Projections in the previous and current Inflation Reports - Scenario with Selic and exchange rates from Focus survey


50%

Year Q 30%

10%

Central

2018 III 4.3 4.3 4.4 4.4 4.4 4.5 4.5

2018 IV 4.0 4.2 4.3 4.4 4.5 4.6 4.8

2019 I 4.4 4.7 4.9 5.0 5.1 5.3 5.6

2019 II 3.5 3.9 4.2 4.3 4.4 4.7 5.1

2019 III 3.6 4.0 4.3 4.5 4.7 5.0 5.4

2019 IV 3.5 3.9 4.2 4.4 4.6 4.9 5.3

2020 I 3.3 3.7 4.0 4.2 4.4 4.7 5.1

2020 II 3.1 3.5 3.8 4.0 4.2 4.5 4.9

2020 III 3.0 3.4 3.7 3.9 4.1 4.4 4.8

2020 IV 2.9 3.3 3.6 3.8 4.0 4.3 4.7

2021 I 2.9 3.3 3.6 3.8 4.0 4.3 4.7

2021 II 2.8 3.2 3.5 3.7 3.9 4.2 4.6

2021 III 2.8 3.2 3.5 3.7 3.9 4.2 4.6

2021 IV 2.8 3.2 3.5 3.7 3.9 4.2 4.6


Table 2.8 – Projected inflation – Scenario with Selic

rate from Focus survey and constant exchange rate


The Committee evaluates that various measures of underlying inflation are running at appropriate levels. This includes the components that are most sensitive to the business cycle and monetary policy.

In its most recent meeting (217th Meeting), the Copom unanimously decided to maintain the Selic rate at 6.50 percent p.a. The Committee judges that this decision reflects its baseline scenario for prospective inflation and the associated balance of risks and is consistent with the convergence of inflation to target over the relevant horizon for the conduct of monetary policy, which includes the calendar year of 2019.

At that meeting, the Copom communicated that its baseline scenario for inflation encompasses risk factors in both directions. On the one hand, (i) the high level of economic slack may lead to a lower‑than‑expected prospective inflation trajectory. On the other hand, (ii) frustration of expectations regarding the continuation of reforms and necessary adjustments in the Brazilian economy may affect risk premia and increase the path for inflation over the relevant horizon for the conduct of monetary policy. This risk intensifies in case (iii) the global outlook for emerging economies deteriorates. The Committee judges that the latter risks have increased.

The Copom stresses that the evolution of reforms and necessary adjustments in the Brazilian economy is essential to maintain low inflation in the medium and long run, for the reduction of its structural interest rate, and for sustainable economic recovery. The Committee stresses that the perception of continuation of the reform agenda affects current expectations and macroeconomic projections.

The Copom judges that it should base its decisions on the evolution of inflation projections and expectations, of the balance of risks, and of economic activity. Shocks that produce relative price changes should only lead to a monetary policy response to their possible second‑round effects (i.e., to the propagation to prices in the economy that are not directly affected by the shock). It is through such second‑round effects that these shocks may affect inflation projections and expectations, and change the balance of risks. These effects may be mitigated by the level of economic slack and by inflation expectations anchored around the targets. Therefore, there is no mechanical relationship between recent shocks and the conduct of monetary policy.


50%

Year Q 30%

10%

Central

2018 III 4.3 4.3 4.4 4.4 4.4 4.5 4.5

2018 IV 3.7 3.9 4.0 4.1 4.2 4.3 4.5

2019 I 3.9 4.2 4.4 4.5 4.6 4.8 5.1

2019 II 2.9 3.3 3.6 3.7 3.8 4.1 4.5

2019 III 3.0 3.4 3.7 3.9 4.1 4.4 4.8

2019 IV 3.2 3.6 3.9 4.1 4.3 4.6 5.0

2020 I 3.2 3.6 3.9 4.1 4.3 4.6 5.0

2020 II 3.2 3.6 3.9 4.1 4.3 4.6 5.0

2020 III 3.1 3.5 3.8 4.0 4.2 4.5 4.9

2020 IV 3.1 3.5 3.8 4.0 4.2 4.5 4.9

2021 I 3.1 3.5 3.8 4.0 4.2 4.5 4.9

2021 II 3.2 3.6 3.9 4.1 4.3 4.6 5.0

2021 III 3.3 3.7 4.0 4.2 4.4 4.7 5.1

2021 IV 3.4 3.8 4.1 4.3 4.5 4.8 5.2


Table 2.9 - Projected inflation – Scenario with exchange rate from Focus survey and constant Selic rate


The Copom reiterates that economic conditions still prescribe stimulative monetary policy, i.e., interest rates below the structural level. This stimulus will begin to be removed gradually if the outlook for inflation at the relevant horizon for the conduct of monetary policy and/or its balance of risks worsen.

In the Copom’s assessment, the evolution of the baseline scenario and of the balance of risks prescribes keeping the Selic rate at its current level. The Copom emphasizes that the next steps in the conduct of monetary policy will continue to depend on the evolution of economic activity, the balance of risks, and on inflation projections and expectations.


Exchange rate pass-through from the perspective of a semi-structural model

The magnitude of the exchange rate pass‑through to the consumer’s price level depends on several factors and may vary over time.1 Mapping these factors contributes to a better understanding of the dynamics of inflation.

This box presents a study on the determinants of the magnitude of the exchange rate pass‑through in Brazil from the perspective of a small‑scale semi‑structural model. We investigate how factors such as magnitude of depreciation, economic cycle, anchoring of expectations and operating margin of firms affect the intensity of the exchange rate pass‑through.

Semi-structural modelling and the exchange rate pass-through

The BCB uses several models to project macroeconomic variables, to construct scenarios and to simulate the effects of economic policies, aiming to assisting the decision‑making process of the Monetary Policy Committee (Copom). In the case of the semi‑structural models, the exchange rate pass‑through to domestic price inflation is captured by a term on the Phillips curve that represents the external inflation expressed in Brazilian reais.2 In this box, we present estimates of a set of models that seeks to capture, in the Phillips curve of market prices, the role of possible factors that affect the degree of exchange rate pass‑through in the Brazilian economy. It is worth noting that the exchange rate pass‑through also operates in BCB models through administered prices, for which there are specific models. In these models, the exchange rate pass‑through to administered prices depends primarily on the rules for adjusting certain items.3 For this reason, the analysis in this box focuses on the exchange rate pass‑through to market price inflation.

The effect of the exchange rate variation on inflation is usually estimated through a parameter that reflects, in the sample considered, the average exchange rate variation pass‑through to inflation, as usually occurs with any variable of a standard regression model. However, at each moment, the exchange rate pass‑through may detach from the average historical pass‑through due to several factors, such as the economic cycle and possible asymmetries in the level of pass‑through between appreciation and exchange depreciation.

In order to evaluate possible determinants of the level of exchange rate pass‑through to market price inflation in the Brazilian economy, a small‑scale aggregate semi‑structural model was used. The model is formed by the following equations: a Phillips curve for market price inflation; an IS curve, which describes the trajectory of the output gap; a curve for the 360‑day pre‑DI swap premium; and a Taylor rule.

1/ See, e.g., Goldfajn and Werlang (2000), Correa and Minella (2010), Frankel et al. (2012), IMF (2016), and de Mendonça and Tiberto (2017).

2/ In the semi‑structural models, the external inflation is represented by the variation of a commodity price index in US dollars converted into national currency by the corresponding exchange rate. For more details see box “Small‑scale aggregate model – 2017” in the June 2017 Inflation Report, and “Small‑scale model of disaggregated price – 2018” in the June 2018 Inflation Report.

3/ See box “Revision of the medium‑term projection models for administered prices”, in the September 2017 Inflation Report.


The Phillips curve for market prices inflation is represented by:4

where is the market prices inflation of the IPCA; is the expectation in time t regarding the IPCA inflation i quarters ahead; is the IPCA inflation; is an imported inflation measure; is an output gap measure; is the control variable p; and is the error term. The parameters are estimated by imposing the verticality constraint of the Phillips curve in the long run, .

Different specifications of the Phillips curve were tested in order to analyze how the exchange pass‑through is affected by different non‑linearities. The following possible determinants of exchange rate pass‑through were examined: (i) the stage of the economic cycle; (ii) the level of anchoring of inflation expectations; (iii) the magnitude of the exchange rate depreciation; and (iv) the ability of firms to absorb cost shocks, represented by their operating margin.

The coefficient of the exchange rate pass‑through is modeled as a function of the variables of interest to analyze how the level of exchange rate pass‑through varies with these determinants. Thus, this parameter becomes variant over time:

where

, that is, it is the value of the output gap when it is in the negative field and

zero, otherwise;

, being the analysts’ expectation in quarter t for

inflation 24 months ahead, 12‑month accumulated, based on the Focus survey,5 and the interpolated measure for the inflation target 24 months ahead6, 7;

, being the variation of the nominal exchange

rate (R$/US$) and is a term that seeks to capture the differential between domestic and external inflation consistent with the long‑term conditions in the modeling of the Phillips curve for market prices; and is the moving average of the last period(s) of the cyclical component of the operating margin of the companies in the Brasil 100 index (IBrX 100).

In some specifications, instead of using continuous values to measure the output gap, a dummy variable was used to indicate the periods when the output gap was negative. On the other hand, the term

4/ The specification of the IS curves and the curve for the 360 day pre‑DI swap premium is set out in the “Small‑scale aggregate model – 2017” box of the June 2017 Inflation Report. In its turn, Taylor’s rule relates the Selic rate to expected inflation deviations from its target, to the output gap, to the Selic equilibrium level, and to the Selic self‑regulatory terms.

5/ The series was built by means of interpolations. When monthly inflation expectations beyond a calendar year ahead were available, the interpolation between inflation expectation for two calendar years ahead and the last monthly expectation of available inflation, accumulated in twelve months, was made. Otherwise, the interpolation between the expectation of inflation for two calendar years ahead and the expectation of inflation for a calendar year ahead was made.

6/ In cases where the expectation was below the target, the variable assumed a zero value.7/ Since inflation targets are defined only for calendar years, the series was built by interpolating the target for two calendar years

ahead and one calendar year ahead by using the target values that had been announced at the time. When necessary, the target was extrapolated by using the hypothesis that the value for the longest horizon announced would remain until the second calendar year ahead.


assumes non‑zero values only when the depreciation of the nominal exchange rate exceeds the term of the domestic and external inflation differential.

Replacing equation (2) by equation (1), we have:8

The estimation of the system of equations was performed using Generalized Methods of Moments (MMG), with quarterly data and considering in the sample at least the period between 2003Q1 and 2018Q19. We considered 23 specifications, which are differentiated by the set of variables determinant of the exchange rate pass‑through used. In some estimates, all variables of interest were included, while in others, only a subset of possible determinants of the pass‑through was considered. The variety of specifications is intended to counterbalance the uncertainty inherent to any model estimation method, and especially in the case of this study, it helps mitigating the level of uncertainty arising from the relatively small sample size used. For the purpose of presenting the results, the specifications were grouped into nine subgroups, corresponding to the different possible combinations of sets of variables of interest.

Results

Table 1 presents the effects of the variables of interest on the exchange rate pass‑through. For each of the subgroups of estimated models, the minimum and maximum values of each coefficient within the subgroup are presented.10 The coefficients presented in the table correspond to the values estimated for the parameters and of equation (3). Thus, they indicate in how many percentage points (p.p.) the pass‑through coefficient is affected by a variation of 1 p.p. in each variable of interest.11 Therefore, comparisons between the impacts of the variables of interest on the exchange rate pass‑through depend on the estimated coefficients as well as on the magnitude and unity of the variables of interest.

In all specifications, the coefficients of the variables of interest were statistically significant and presented the expected signal. Thus, the estimates show that the negative output gap and the anchoring of the inflation expectations contribute to reduce the exchange rate pass‑through, while the reduction of the operating margin of the firms and the magnitude of the exchange rate depreciation are factors that increase the exchange rate pass‑through.

The results suggest that the magnitude of the exchange rate depreciation affects the level of the pass‑through: the higher the depreciation, the higher the exchange rate pass‑through. The subgroups that also include the measure of inflation expectations (5, 7 and 9) indicate that, for every 1.0 p.p. of increase in depreciation, the exchange rate pass‑through increases in the range of 0.11 p.p. to 0.19 p.p.

8/ Note that when the term is positive, it multiplies another variable (imported inflation) that already includes the exchange rate variation, resulting in a quadratic term in the magnitude of the exchange rate depreciation. For the purposes of comparability, the lag structure used for the product gap and the exchange rate variation did not vary between specifications.

9/ The start of the sample varies according to the availability of the variables of interest used in each specification.10/ The range of values of each coefficient took into account the point value of the parameter in each estimation instead of its

confidence interval.11/ In cases where the variable is replaced by a dummy, the parameter indicates the impact, in percentage points, of the

presence of a negative gap on the exchange rate pass‑through coefficient.


Estimates indicate that the degree of anchoring of expectations plays a central role in determining the exchange rate pass‑through. For each increase of 1.0 p.p. in the degree of disanchoring, the pass‑through coefficient is between 1.5 p.p. and 6.7 p.p. These results highlight the importance of anchoring expectations for inflation control. When expectations are anchored, the impact of exchange rate depreciations on inflation expectations is mitigated, reducing the exchange rate pass‑through to prices. A key element to the anchoring of expectations is the central bank credibility and the economic policy regime.

Regarding the business operating margin cycle, estimates suggest that for every 1.0 p.p. increase in this indicator, the exchange rate transfer coefficient decreases between 0.4 p.p and 1.5 p.p. This evidence is compatible with the hypothesis that, when their operating margin is higher, firms are more able to absorb cost pressures, moderating the transfer of exchange rate depreciation to prices. On the other hand, when the operating margin is lower, firms have less space to absorb exchange rate variations, which tends to increase the degree of pass‑through.

Another important factor in determining the magnitude of the exchange rate pass‑through is the stage of the economic cycle. The estimates show that, as the output gap becomes more negative, the pass‑through coefficient decreases. Specifications that include this variable suggest that in the case of a negative output gap of 1.0%, the exchange rate pass‑through is 2.1 p.p. to 9.1 p.p. lower than when the economy is operating above or at its potential. These estimates involve both specifications that use a dummy variable for the negative gap, and those that consider the magnitude of the negative gap. This result is consistent with the hypothesis that, in times of low economic activity, firms find it more difficult to pass on exchange rate depreciation to the consumer.

Based on the results presented in Table 1, it is possible to identify how, in each period of time, the analyzed factors contributed to the evolution of the exchange rate pass‑through. Figure 2 shows the relative contribution of each of the determinants of the variable part of the exchange rate pass‑through coefficient ( ) to the exchange rate pass‑through in each time period. The contribution of each


12/ In this exercise, it was assumed that the exchange rate of September 2018 would be equal to the average rate observed in the five business days ended on the Friday prior to the September 2018 Copom meeting (R$4.15/US$).

determinant was calculated by multiplying each coefficient by the value of its respective determinant and then normalizing the result so the sum of the contributions is equal to 1 in the 2003Q1. Positive (negative) values indicate factors that, at that time, contributed to a larger (smaller) transfer.

Applying the model to the recent period – which refers to the second and third quarters of 2018 – one can observe that, although the magnitude of the exchange rate depreciation acts for a higher exchange rate pass‑through, the anchoring of expectations, the position in the economic cycle and the operating margin of companies contribute to reduce the level of the exchange rate pass‑through when compared to other periods.12 Considering the magnitude of the contributions of each factor, it is the effect of factors that reduce the exchange rate pass‑through that preponderates.

Conclusion

This box explores how the exchange rate coefficient can be affected by different conjunctures. In particular, the results reinforce the importance of anchoring inflation expectations for the conduct of monetary policy, since they show that the exchange rate pass‑through is more contained in the anchored expectations environment.

As has been emphasized in other Inflation Reports, models are tools to support the decision‑making process and the economic analysis and should be combined with the use of a broad set of information and the exercise of judgment.

References

CORREA, A. S. and MINELLA, A. (2010): “Nonlinear Mechanisms of the Exchange Rate Pass‑Through: A Phillips Curve Model with Threshold for Brazil.” Revista Brasileira de Economia v.64 n.3, p. 231‑243, Jul‑Sep.

DE MENDONÇA, F. H. and TIBERTO, P. B. (2017): “Effect of credibility and exchange rate pass‑through on inflation: An assessment for developing countries.” International Review of Economics and Finance 50, p. 196–244.

Figure 1 – Contribution of factors that impact the magnitude of the exchange rate pass-through

Values normalized so that factor contributions sum to 1 in the first quarter of 2003

-2.0

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

I2003

III I2004

III I2005

III I2006

III I2007

III I2008

III I2009

III I2010

III I2011

III I2012

III I2013

III I2014

III I2015

III I2016

III I2017

III I2018

III

Unanchoring of inflation expectations Output gap Magnitude of depreciation

Companies' operating margin Sum of factors


FRANKEL, J., PARSLEY, D. e WEI, S. (2012): “Slow Pass‑through Around the World: A New Import for Developing Countries?” Open Econ Rev 23, p. 213–251.

GOLDFAJN, I. and WERLANG, S. (2000): “The pass‑through from depreciation to inflation: A panel study.” Banco Central do Brasil, Working Paper Series 5.

IMF (2016): “Regional Economic Outlook – Western Hemisphere: Exchange Rate Pass‑Through in Latin America.” International Monetary Fund, April, chapter 4.

inflation report – september 2018 - bcb.gov.br · rate depends on the conditioning selic rate...

Documents