Market Expectations System Information up to June 2016 Frequently Asked Questions Series
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
Market Expectations
System Information up to June 2016
Frequently Asked Questions Series
2
“Frequently Asked Questions” Series Central Bank of Brazil 1. Interest Rates and Bank Spreads 2. Price Indices in Brazil 3. Monetary Policy Committee (Copom) 4. Fiscal Data 5. Regulated Prices 6. Public Securities and Public Debt Management 7. Brazilian Payments System 8. External Accounts 9. Country Risk 10. Inflation Targeting Regime in Brazil 11. Functions of the Central Bank of Brazil 12. Reserve Requirements 13. Market Expectations System 14. Monetary Policy, Open Market Operations and FX Swap 15. Economic Indicators 16. Banking Credit in Brazil 17. Gross Domestic Product (GDP)
Economic Policy Deputy Governor Carlos Viana de Carvalho
Team André Barbosa Coutinho Marques Carolina Freitas Pereira Mayrink Henrique de Godoy Moreira e Costa Luciana Valle Rosa Roppa Luiza Betina Petroll Rodrigues Manuela Moreira de Souza Maria Cláudia Gomes P. S. Gutierrez Márcio Magalhães Janot
Coordinator Renato Jansson Rosek Authoring and publishing Investor Relations and Special Studies Department (Gerin) Brasília-DF
This publishing is part of the Financial Education Program of the Central Bank of Brazil (BCB).
3
Market Expectations System
This paper is part of the Banco Central do Brasil’s “Frequently Asked Questions” series (FAQ). This series, which is produced by the BCB’s Investor Relations and Special Studies Department (Gerin), provides information on economic topics of interest to investors and the general public.
The Banco Central do Brasil (BCB) is producing this series as part of its ongoing efforts to enhance the transparency of the Brazilian economic policy and the effectiveness in communicating its actions.
4
Contents
General View ............................................................................................................................................ 6
1. What is the information collected by the Market Expectations System? .......................................................... 6
2. What is the importance of market expectations? .................................................................................................. 6
Expectations collection ........................................................................................................................... 7
3. When was the Market Expectations System implemented? ............................................................................... 7
4. Who can provide information (expectations) to the System? .............................................................................. 7
5. How many institutions participate as information providers? ............................................................................ 8
6. How is the market information collected? .............................................................................................................. 8
7. Which expectations are collected, and in which format are they informed? ..................................................... 8
8. What is the validity of the data recorded in the System? ................................................................................... 10
9. Are the participants obliged to provide information? ......................................................................................... 11
10. What happens when an institution ceases to inform its projections? ......................................................... 11
Disclosed information ........................................................................................................................... 11
11. What are the main products published? ........................................................................................................... 11
12. What information does the Focus - Market Report provide? ........................................................................ 12
13. Are changes in the median of an indicator always caused by changes in the surveyed participants’ expectations? ...................................................................................................................................................................... 12
14. How is the expectation for the inflation accumulated over the next-12 months calculated? ................. 13
15. How is the expectation for the inflation accumulated over the next 12 months, smoothed, calculated? 15
16. How can I obtain daily time series of market expectations collected by the Market Expectations System? ............................................................................................................................................................................... 19
17. Is it possible to query individual information? ................................................................................................. 19
Top Five ................................................................................................................................................... 20
18. What is the incentive for the institutions to maintain the expectations updated? .................................... 20
19. What are reference dates? .................................................................................................................................. 20
20. How are the Top 5 institutions classified, for which variables and in which horizons? ............................ 20
21. Is there any precondition for the institutions to participate in the Top 5? .................................................. 22
22. Which is the formula used for the calculation of the Short Run Top 5 monthly rankings? ...................... 23
23. Which is the formula used for the calculation of the Medium-Run Top 5 rankings? ................................ 24
24. Which is the formula used for the calculation of the Long-Run Top 5 rankings, annually released? ..... 25
5
25. What are the formulas used for the calculation of the Annual Top 5 Rankings (Short Run and Medium rankings)? ............................................................................................................................................................................ 25
6
Market Expectations System
General View 1. What is the information collected by the Market Expectations System?
They are projections prepared by institutions operating in the financial market, such as banks, asset managers and consultants and, in some cases, companies of the real sector that have specialized teams who produce estimates for the main macroeconomic variables, in order to assist the decision making by professionals of the institution itself as well as by its external customers. The projections, in many cases with the help of econometric modeling, are performed for variables related to economic activity, interest rates and exchange rates, the variation of price indices, the balance of payments and the fiscal sector of the Brazilian economy.
Expectations can be collected in several ways, such as conducting surveys with market professionals or consumers, or from information extracted from financial assets such as inflation-linked bonds. The Market Expectations System, developed by the Central Bank of Brazil (BCB), is a system for the collection of market economists’ macroeconomic projections.
2. What is the importance of market expectations?
The market expectations are important inputs to the monetary policy decisions. Moreover, the availability of their statistics to the general public, through the BCB's website, enables enterprises and citizens to be aware about what the market agents are projecting, thereby constituting an important tool for planning their actions in the short, medium and long term.
7
Expectations collection 3. When was the Market Expectations System implemented?
The market expectations survey started in May 1999. During this period, the BCB's economic area developed studies and econometric models for the projection of inflation and other macroeconomic variables, in order to provide the necessary technical tools for the formal implementation in June 1999 of the inflation targeting regime. The market expectations for inflation showed, then, strong disparity due to the uncertain backdrop that followed the collapse of the exchange rate anchor.
Initially, about fifty financial institutions and consultancies were contacted and informed to the BCB by phone, fax or email of the annual projections for the main price indices (IGP-DI, IGP-M and IPA), besides the GDP growth. Subsequently, the survey on market expectations became more sophisticated, due to the larger number of participants, the incorporation of new variables (other indicators on inflation, exchange rate, interest rate, fiscal data and the Balance of Payments variables) and the monthly or quarterly data, in addition to the annual data.
With the growing importance of monitoring expectations for the monetary policy, in November 2001 the Market Expectations System's webpage was created. It has as main objectives:
to ensure the greater speed and safety in the collection process and to minimize the occurrence of information errors. The data started to be provided online, at any time, by institutions previously accredited,
which received from the BCB a specific password for the access to the IT platform. In 2010, a new set of improvements enhanced the integrity to the system, as the existing
modules were integrated on a single IT platform, with unique processing, thus increasing the processing agility and maintaining the reliability of their routines and statistics.
4. Who can provide information (expectations) to the System?
The Market Expectations System is an online tool, developed for the web environment, which can be accessed through the link http://www.bcb.gov.br/expectativas, restricted to the Investor Relations and Special Studies Department (Gerin) of the BCB and previously accredited institutions.
New institutions can only be included in the survey by the System administrators after the applicant informs the respective registration data and of the assignment of specific login and password. In principle, any entity (banks and other financial institutions, nonfinancial companies, consultancies, class associations, universities, etc.) may request participation in the research, as long as they have a specialized team on macroeconomic projections and update their projections regularly.
8
5. How many institutions participate as information providers?
Currently, about 130 logins are enabled, mostly banks, asset managers, dealers and brokers, as well as consulting and other non-financial companies. In the universe of qualified institutions, most of them update their expectations weekly.
6. How is the market information collected?
Gerin is the department responsible for the administration of the Market Expectations System, a web interface, where financial institutions, consulting firms and companies of the non-financial sector, previously authorized, inform their expectations for various macroeconomic variables.
The access can be done at any time, with no pre-set schedule for updates. However, data reported after 5:00 pm only impact the statistics produced by the System on the following business day. This happens because every business day at 5:00 pm, the information is consolidated and several statistics are generated - averages, medians, standard deviations, coefficients of variation, minimum and maximum values of the expectations recorded by the participants.
The System also has checks for the consistency and robustness of the data. Certain variables can be entered directly or calculated from other variables. For example, the expectation for the trade balance may be informed or calculated by the System based on the expectations for exports and imports in the same period. If such data have previously been entered, the System calculates the trade balance and blocks the projection of this variable, to ensure consistency of information.
7. Which expectations are collected, and in which format are they informed?
Table 1 shows the 24 variables whose expectations are collected by the Market Expectations System. As shown in the table, the System receives daily a limited number of monthly, quarterly and annual expectations for each variable. In addition, the type of information inserted in the System varies.
9
Table 1– Variables with Expectations Collected by the Market Expectations System
Variable type Variable to be forecasted
Number of projections collected by the System
Type of projection inserted in the System
Monthly Quar-terly Annually Monthly
Quarterly/ Annually
Price index IGP-DI, IGP-M, IPA-M, IPA-DI, IPCA, IPCA-15 or INPC 18 0 5 [a] m/m % [b] y/y % [c]
Regulated Prices 0 0 5 y/y % [c]
Economic activity
Industrial Production 18 0 5 [a] y/y % [c] y/y % [c] GDP, Agriculture GDP, Manufacture GDP or Service GDP
0 6 5 [a] y/y % [c]
FX Rate (R$/US$)
Ptax- end-of-period 18 [d] 0 5 [e] [e] Ptax- annual average 0 0 5 Annual average
Target for Selic rate
Selic Target - end-of-period 18 [d] [f] 0 5 [e] [e] Selic Target – annual average 0 0 5 Annual average
External Sector
Exports and Imports 0 0 5 Value in US$ billions Trade Balance 0 0 5 [g]
Balance of payment
Current Account Result 0 0 5 Value in US$ billions Direct Investment Liabilities [h] 0 0 5
Fiscal
Primary Result of the Consolidated Public Sector 0 0 5
% of GDP Nominal Result of the Consolidated Public Sector 0 0 5
Net Public Sector Debt 0 0 5 [a] As part of the solutions for checking the internal consistency of data, the field on the annual projection for the current year is calculated automatically if there are monthly/quarterly information for every month/quarter of the year. In this case, the System will calculate the annual figure, considering the effective results already released and the monthly/quarterly projections informed. If some of the monthly/quarterly projections is missing, the System allows the participant to inform its annual projection. [b] m/m % = change in comparison to the previous month, without seasonal adjustment. [c] y/y % = change in comparison to the [same period of the] previous year, without seasonal adjustment. [d] Monthly information related to December (or the last month in which there is a Copom meeting, in the case of Selic) is automatically assigned to the end of the annual period, and vice versa, and the annual average is calculated in case all relevant monthly expectations are informed. [e] Value in the last day of the period. [f] From the moment the Copom calendar for following year is known, projections for the Selic rate for the months with no meetings cannot be registered in the system. [g] The input of exports and imports data eliminates the need of information on the trade balance. [h] Up to March 2015, the collected variable was "Foreign Direct Investment". Click here for more details on the methodological change (in Portuguese). Note: Statistics for the annual inflation may differ from the monthly inflation statistics, accumulated in the year, since the samples may be different, given that a few institutions inform only annual projections, others only the monthly ones, and there is no obligation to inform every month. Thus, the samples of institutions that inform each monthly projection and each annual projection are different. Furthermore, the accumulation of the monthly medians would not necessarily be equal to the respective annual median, even if the samples were the same. Source: BCB.
Price Indices The inflation indicators IGP-DI, IGP-M, IPA-DI, IPA-M, IPCA, IPCA-15, INPC and IPC-Fipe
follow the same rules, as shown in Table 1.
10
As part of the solutions for checking the internal consistency of data, the annual expectation field is calculated automatically if there are monthly information for all the months of the year. In this case, the System will calculate the cumulative inflation in the year, considering the effective results already released and the monthly projections informed, recording the annual projection of the participant. If some of the monthly projections is missing, the System allows the participant to inform its annual expectation.
Is worth mentioning that the statistics for the annual inflation may differ from the monthly inflation statistics, accumulated in the year, since the samples may be different, given that a few institutions inform only annual projections, others only the monthly ones, and there is no obligation to inform every month. Thus, the samples of institutions that inform each monthly projection and each annual projection are different. Furthermore, the accumulation of the monthly medians would not necessarily be equal to the respective annual median, even if the samples were the same.
Economic Activity Similar to what happens to price indices, for GDP (and industrial production), the projection for
the current year is automatically calculated, if all quarterly (monthly) expectations have been informed. Otherwise, the institution may inform the projection for the current year. As soon as the effective quarterly (monthly) data are released, they are incorporated to the annual result calculation for the respective year. For both variables, projections are informed as percentage growth over the same period of the previous year.
Exchange Rate and Selic Target Rate For these indicators, monthly information related to December (or the last month in which
there is a Copom meeting, in the case of Selic target rate) is automatically assigned to the end of the annual period, and vice versa, and the annual average is calculated in case all relevant monthly expectations are informed. From the moment the Copom calendar for following year is known, projections for the Selic target rate for the months with no meetings cannot be registered in the System.
8. What is the validity of the data recorded in the System?
The System only considers the data reported in the last thirty days. That is, if an institution does not inform their expectations again within thirty days, the System automatically saves the value in the data base, but it will stop considering it in the calculation of the statistics. The aim of the 30 days limit is to prevent the statistics from loading outdated projections. So even if expectations have not undergone any modification, the qualified institution must confirm the data within a range of thirty days to keep it active in the statistics calculation.
11
After thirty days without update, expectations remain in the system, with access through historical series for Gerin and the participating institution, but are not included in the calculation anymore.
9. Are the participants obliged to provide information?
There is no mandatory provision of information. There are participating institutions that specialize in forecasting some variables and do not provide projections for the others.
10. What happens when an institution ceases to inform its projections?
Gerin periodically checks whether the information provided by the System participants is updated or not, and blocks the input of those which have been for at least six months without any activity. Thus, the participant's occasional return to the system will depend on a specific contact requesting the reinstatement of the previous status.
Disclosed information 11. What are the main products published?
From the data collected, the System can calculate in real time the sample statistics, generating daily reports to the Board. In addition, it generates weekly the "Focus - Market Report" ("Market Readout"), available to the public at the BCB's internet website, in Portuguese and English.
The statistics produced by the system and released by the the BCB include the median, mean, standard deviation, coefficient of variation, maximum and minimum for all variables collected, including those of the Top 5 institutions group for the possible horizons - up to 18 months, 6 semesters or five years ahead, as the case may be. Because it is less subject to swings of the extremes, the median expectation is the statistic more closely monitored and disclosed (including the recent graphical evolution) in the "Focus - Market Report" every Monday regularly at 8:30 am, with the data collected until the previous Friday. The historical series of statistics are available since January 2000 in the BCB's website on the internet also updated weekly on Mondays, along with disclosure of "Focus - Market Report".
Besides the Focus - Market Readout, other reports are published at the BCB's website, such as the Top 5 ranking (on a monthly and annual basis), and the Focus - Frequency Distributions (this one just in Portuguese), which compares in three recent moments the shape of the frequency distributions relative to the projections for the annual IPCA and for the Selic rate twelve months ahead.
Other reports, used by the BCB Board, are generated periodically.
12
12. What information does the Focus - Market Report provide?
The Market Report provides the following information, together with the graphical evolution of the indicators and of their weekly behavior:
The median market expectations for the inflation over the next 12 months, smoothed, for the IPCA, IGP-DI, IGP-M and IPC -Fipe;
The median market expectations for the next month for which IPCA and IGP-DI have not yet been published and for the subsequent month (IPCA, IGP-DI, IGP-M, IPC-Fipe, exchange rate at period-end and Selic target rate at period-end);
The median expectations for the next year for which IPCA and IGP-DI have not yet been published and for the subsequent year (IPCA, IGP-DI, IGP-M, IPC-Fipe, exchange rates average and at period-end, Selic target rate average and at period-end, net public sector debt/GDP, real GDP growth, industrial production growth, current account of the Balance of Payments, Trade Balance, Direct Investment Liabilities and % change in administered prices);
The mean and the median expectations of the Top 5 institutions of short and medium term for the next month for which IPCA and IGP-DI have not yet been published and the subsequent month (IPCA, IGP-DI, IGP-M, exchange rate at period-end and Selic target rate at period-end);
The mean and the median expectations of the Top 5 institutions of short and medium term for the next year for which IPCA and IGP-DI have not yet been published and for the subsequent year (IPCA, IGP-DI, IGP-M, exchange rate at year-end and Selic target rate at period-end).
It is important to highlight that the Focus Report does not present BCB projections, but consolidated expectations from the market agents that participate on the Market Expectations System. Besides, the emphasis on the medians, and not on the averages, is based on the fact that the median is less influenced by extreme values. The historical series of every statistics calculated by the System may be accessed here.
13. Are changes in the median of an indicator always caused by changes in the surveyed participants’ expectations?
No. When a monthly or quarterly indicator is released, the Market Expectations System includes this value in the calculation of the annual indicator since the day of its release, for each participant, thus affecting statistics of that day. Therefore, even with no changes provided by the participants in their monthly expectations on the day of the release of the monthly indicator, the annual indicator changes individually and automatically, proportionally to the difference between the released value for the monthly indicator and its respective expectation. This occurs for every participant that has all monthly expectations for the remaining months of that year, and the annual statistics consequently changes.
13
Let’s suppose, for example, that on a certain day when a monthly IPCA is to be released a participant has a monthly expectation of 0.5% for the previous month (the one the IPCA to be released refers to), this institution has projections for all the remaining months of the current year, and its projection for the annual IPCA is 5%, calculated by the System as a result of the so far released monthly IPCA changes and the expectations of this institution for the remaining months of the current year. Let’s also suppose that the monthly IPCA is effectively 1.5%, and not 0.5% as projected by the institution. At the end of the same day the System recalculates the projection of this institution for the annual IPCA replacing the monthly projection of 0.5% by the effective 1.5%, and the result is higher than the previous one, even with no changes provided by the institution in any monthly projection for the remaining months of the current year. The same happens for every institution having monthly projections for all the remaining months of the current year. Therefore, averages, medians and other annual statistics will change. Then, when there is a surprise in a monthly indicator, such as price indices and industrial production, the respective expected annual indicator may change in a proportion as big as this surprise, even though no individual change has been provided in the monthly expectations of the participants, regarding the remaining months of the year. The same occurs for quarterly variables such as GDP growth. Besides that, a similar behavior occurs for the projected annual average of exchange and Selic target rates, and for inflation for the next 12 months, regarding price indices.
14. How is the expectation for the inflation accumulated over the next-12 months calculated?
The expectation for the inflation accumulated over the next 12-months is calculated daily for each institution which have expectations for all twelve months of the period considered. The calculation is done by accumulating expectations for monthly inflation.
Figure 1 shows the steps followed by the System to calculate the expectation for inflation accumulated over the next 12-months and their descriptive statistics.
Figure 1 – Expectation for the Inflation Accumulated Over the Next 12 months – Calculation Steps for a Given Price Index
Step 1: make a list of registered institutions and their monthly expectations for the given price index, valid on the day of the calculus (day “d”).
Result for day “d”: a list with ”n” institutions registered in the Market Expectations System: institution 1, institution 2, ..., institution n.
14
Step 2: separate the institutions which have 12 consecutive monthly expectations valid, the first of which being the next month (the first not-yet-released value for inflation) Result for day “d”: a list with “z” institutions and their respective expectations for monthly inflation z ≤ n each institution has 12 monthly expectations which will be used in the calculus: e1, e2, e3, ..., e12
Step 3: calculate, for each of the z institutions, the inflation accumulated over the next 12 months For each institution, the System calculates: [(1 +
𝑒1
100) ∗ (1 +
𝑒2
100) ∗ … ∗ (1 +
𝑒12
100) − 1] ∗ 100 = E = institution’s expectation for the inflation accumulated over
the next 12 months, measured for a given price index, on the day “d” Table 2 illustrates the calculation made in this step Repeat this calculation “z” times (once for each institution) Result for day “d”: “z” expectations for inflation accumulated over the next 12 months (one for each
institution) Step 4: calculate the descriptive statistics The System calculates mean, median, maximum, minimum, standard deviation and coefficient of variation of the "z" expectations calculated in Step 3. Total after the calculation: 6 descriptive statistics of expectations valid on day “d” This step is identical for all variables collected by the Market Expectation System Source: Gerin. These steps are the same for the calculation of inflation measured by any of the price indices collected. These steps are followed daily by the System.
The descriptive statistics for the inflation accumulated over the next 12 months, measured by a given price index, may change daily, for the following three reasons:
The number of institutions with valid expectations changes when: > New institutions are registered in the System and start to inform their expectations; > Some institutions cease to have valid monthly expectations (see Question 7);
Expectations change: some institutions log in the System and change their monthly expectations;
The months considered change: when a monthly inflation figure is released (by IBGE, FGV or Fipe), the 12-month-interval taken into account for the calculation changes, and therefore the expectation for the inflation accumulated over the next 12 months may also change.
Taking the IPCA as example, suppose: we are in July 2016, the latest information released by the IBGE was the the index in June 2016, published on
July 8th,
15
the next IPCA release, the July index, is scheduled to be released on August 10th, a given institution (“X”) has, in July 15th, the monthly expectations for IPCA shown in Table
2, all in percentages. The expectation of institution “X” for the accumulated inflation for the next twelve months, in
July 15th, would be the multiplicand of these variations, in decimal format (1 +𝑒𝑥𝑝𝑒𝑐𝑡𝑎𝑡𝑖𝑜𝑛
100) of 0.77
up to 0.35 (12 months ahead) which results 5.2397%.
Table 2 – Expectation for Inflation Accumulated Over the Next 12 months – Example for Institution “X”
Month to which the
expectation refers to
Institution ”X”’s expectations (valid on July 15th) for the
monthly IPCA inflation (1+ expectation/100)
Expectation (on July 15th) for inflation accumulated over the next 12 months (multiplicand of the previous column):
from July/16 to June/17 jul/16 0.77 1.0077 (1.0077)*(1.0089) ago/16 0.89 1.0089 set/16 0.31 1.0031 *(1.0031) *(1.0030) out/16 0.30 1.0030 nov/16 0.40 1.0040 *(1.0040) *(1.0040) dez/16 0.40 1.0040 jan/17 0.40 1.0040 *(1.0040) *(1.0035) fev/17 0.35 1.0035 mar/17 0.30 1.0030 *(1.0030) *(1.0030) abr/17 0.30 1.0030 mai/17 0.35 1.0035 *(1.0035) *(1.0035) - 1 jun/17 0.35 1.0035 = 5.2397%
Source: Gerin. Expectations shown in this table are hypothetical.
To know more about price indices, see FAQ 2- Price Indices in Brazil.
15. How is the expectation for the inflation accumulated over the next 12 months, smoothed, calculated?
Expectation for the inflation accumulated over the next 12 months, smoothed, of each institution registered in the System is calculated from the expectations for the inflation accumulated over the next 12 months (calculated according to the steps shown in Question 14).
At each publication of a price index1, the twelve-month period (used for the calculation of inflation accumulated over the next 12 months) changes. The monthly inflation data just released ceases to compose the original set of twelve months. This value is replaced by the expectation following the 12th month of the original period (which would thus be the 13th month with respect to the original period).
1 IGP-DI, IGP-M, INPC, IPA-DI, IPA-M, IPCA, IPCA-15 or IPC-Fipe.
16
Often there is a significant difference between the expectation for the month that comes out and the expectation for the month that will be included in the calculation, implying a leap in the twelve-month inflation expectation. Chart 1 shows that some leaps occur when IPCA figures are released, exactly because of this difference.
Chart 1 – Median of Expectations for IPCA Inflation Accumulated over the Next 12 Months (July 2015 to June 2016)
Source: BCB’s Market Expectation System and IBGE.
In order to smooth out these jumps, the difference is distributed pro-rata between the day of
publication of a monthly indicator and the day of publication of the same indicator for the following month. The result is the expectation for the inflation accumulated over the next 12 months.
Continuing with the example of Question 14, suppose the maintenance of that institution's expectations from July 9th to August 10th (as shown in Table 3). The expectation for the inflation accumulated inflation over the next 12 months, from July 2016 to June 2017 (12th month ahead) is 5.2397%, as already explained. When July 2016 IPCA figure is released, on August 10th, the twelve-month period changes (out July 2016 and enters August 2017). This change involves replacing a percentage of 0.77 by a percentage of 0.50 (and the accumulated on this day, August 10th, would change from 5.2397% for 4.9577%, causing a jump in historical series).
5.55.75.96.16.36.56.76.97.17.3
Jul 1
5
Jul 1
5
Aug 1
5
Sep 1
5
Sep 1
5
Oct 1
5
Nov 1
5
Dec 1
5
Dec 1
5
Jan 1
6
Feb 1
6
Feb 1
6
Mar
16
Apr 1
6
May
16
May
16
Jun 1
6
%
median of expectations for the inflation accumulated over the next 12 monthsdates when IPCA (IBGE's CPI) was released
17
Table 3– Expectation for Inflation Accumulated over the Next 12 Months Smoothed – Example for Institution “X” Month to which the expectation refers to
Institution ”X”s expectations for the monthly IPCA inflation (stable from July 9th to August 10th)
(1+ expectations/100)
Expectation for inflation accumulated over the next 12 months (multiplicand of the previous column):
from July/16 to June/17 from August/16 to July/17
jul/16 0.77 1.0077 (1.0077)*(1.0089) ago/16 0.89 1.0089 (1.0089)*(1.0031) set/16 0.31 1.0031 *(1.0031) *(1.0030) out/16 0.30 1.0030 *(1.0030) *(1.0040) nov/16 0.40 1.0040 *(1.0040) *(1.0040) dez/16 0.40 1.0040 *(1.0040)*(1.0040) jan/17 0.40 1.0040 *(1.0040) *(1.0035) fev/17 0.35 1.0035 *(1.0035)*(1.0030) mar/17 0.30 1.0030 *(1.0030) *(1.0030) *(1.0030) *(1.0035) abr/17 0.30 1.0030 mai/17 0.35 1.0035 *(1.0035) *(1.0035) - 1 *(1.0035)*(1.0050) - 1 jun/17 0.35 1.0035 = 5.2397% jul/17 0.50 1.0050
=4.9577%
Source: Gerin.
The smoothed inflation distributes this leap between the consecutive publication dates, pro-
rata, until all changes are incorporated into the day August 10th. Thus, the leap is eliminated without loss of information.
The calculation steps are described in Figure 2. If there is a change of the monthly expectations between the dates of IPCA release, the differences are distributed pro rata in the same way.
Figure 2 - Expectation for the Inflation Accumulated over the Next 12 Months Smoothed – Calculation Steps for a Given Price Index
Step 1: take the “z” institutions listed in Step 2 of Figure 1. Separate those which have 13 monthly consecutive expectations valid on day “d”, for the given price index, the first of which being the next month (the first not-yet-released value for inflation) Result for day “d”: a list with “w” institutions and their respective 13 expectations for monthly inflation w ≤ z ≤ n
Step 2: : calculate, for each of the w institutions, the inflation accumulated over the next 12 months smoothed For each institution with 13 monthly expectations “ej” valid (j= 1, ..., 13) on day “d”, the System calculates:
{(1 +𝐸
100) ∗ [(1 +
𝑒13
100) ÷ (1 +
𝑒1
100)]
(𝑛𝑑𝑡𝑛𝑑𝑝⁄ )
− 1} ∗ 100 = S = institution’s expectation (on day “d”) for the
inflation accumulated over the next 12 months smoothed, measured by the given price index
18
“E” is the expectation for the inflation accumulated over the next 12 months, for this institution, price index,
on day “d”, as explained in Question 14; 𝑒13 is the intitution’s expectation for the monthly inflation 13 months ahead on day “d”; 𝑒1 is the intitution’s expectation for the monthly inflation 1 month ahead on day “d”; ndt is the number of days between the release of the last inflation figure and the day “d”; ndp is the number of days between the release of the last inflation figure and the next release of this given
price index; Result for day “d”: “w” expectations for inflation accumulated over the next 12 months smoothed (one for
each institution).
Step 3: calculate the descriptive statistics The System calculates mean, median, maximum, minimum, standard deviation and coefficient of variation of the "w" expectations calculated in Step 2. Total after the calculation: 6 descriptive statistics of expectations valid on day “d”
Source: Gerin. These steps are the same for the calculation of inflation measured by any of the price indices collected. These steps are followed daily by the system.
Chart 2 compares IPCA Inflation Accumulated over the Next 12 Months Smoothed to Non-Smoothed, showing the effect of smoothing over the leaps of original series.
Chart 2 –Medians of Projections for the IPCA Inflation Accumulated over the Next 12 Months (July 2015 to June 2016) – Smoothed vs. Non-Smoothed
Fonte: Gerin.
To know more about price indices, see FAQ 2- Price Indices in Brazil.
5.55.75.96.16.36.56.76.97.17.3
Jul 1
5
Jul 1
5
Sep 1
5
Oct 1
5
Nov 1
5
Dec 1
5
Jan 1
6
Feb 1
6
Mar
16
Apr 1
6
May
16
Jun 1
6
%
median of expectations for the inflation accumulated over the next 12 months
median of expectations for inflation accumulated over the next 12 months, smoothed
19
16. How can I obtain daily time series of market expectations collected by the Market Expectations System?
The BCB's Market Expectations System collects market expectations and produces descriptive statistics (mean, median, maximum, minimum, standard deviation and coefficient of variation) of these expectations. These descriptive statistics form daily time series, which are available online to the public.
Figure 3 shows the initial screen consultation of historical series.
Figure 3 - Market Expectations System - Consolidated Statistics Query Screen
Source: BCB-Gerin.
17. Is it possible to query individual information?
No. The confidentiality of information is guaranteed. Only the institution itself, besides Gerin, as the system administrator, has access to the individual data.
20
Top Five 18. What is the incentive for the institutions to maintain the expectations
updated?
With the aim of encouraging the improvement of the predictive ability of the survey participants and rewarding their analytical effort, the BCB prepares the Top 5 ranking, a classification system of the institutions based on the accuracy of their projections for the short, medium and long term. The medians and means of the projections informed by the five institutions with the best performance (the Top 5) are disclosed in the "Focus-Market Report".
19. What are reference dates?
Regarding the variables subject to the Top 5 rankings (IPCA, IGP-DI, IGP-M, exchange rate and Selic target rate), the error or deviation from the actual values of these variables are calculated based on expectations valid on specific dates, the “reference dates”. On these dates, the participant must have valid expectations registered in the system, i.e. reported in the last 30 days, otherwise some penalties will apply. Furthermore, the absence of valid projections (three monthly and one annual) on the last reference date for each period (or pair of dates in the cases of exchange and Selic rates) implies the exclusion of the participant from the respective monthly ranking.
Consequently, there is a greater update of information by the participants on these reference dates. The reference dates are the following:
IPCA: last business day preceding the date of release of the IPCA-15; IGP-DI: last business day preceding the date of release of the IGP-M 2nd 10-day period; IGP-M: last business day preceding the date of release of the IGP-M 1st 10-day period; Exchange Rate: last day of the previous month and last business day before or on the 15th
of the current month; Selic rate: last business day before or on the Wednesday of the week preceding the Copom
meeting of the reference period and the last business day before or on the 4th Wednesday preceding the Copom meeting of the reference period.
To be considered on a certain date, reference or not, the expectations must be included in the Expectations System before 5:00 pm, at the latest. At 5:00 pm, the statistics are generated and data registered thereafter will impact the statistics only on the next business day.
20. How are the Top 5 institutions classified, for which variables and in which horizons?
21
The Top 5 rankings of short and medium term are published every month. In the short-term ranking, one evaluates the accuracy of the projections with a one month lag in comparison to the actual value released for the indicator, over the last 6 months (
Figure 4, first diagram). The medium term ranking considers the average accuracy of the projections over three consecutive 4-month periods in relation to the effective results over three months - the reference month and the two preceding months (
Figure 4, second diagram). The long-term ranking considers the accuracy of the projections reported along 12 months for the annual index published in January of the following year (
Figure 4, third diagram). For the rankings of short, medium and long term, the statistics related to the five institutions with better rating (or more institutions, if any ties) make up the historical series published on the BCB website under the reference "Top 5 Institutions".
Figure 4 – Top 5 Short Term
Medium Term
Long Term
month: n-5 n-4 n-3 n-2 n-1 n
projections (weight):
actual:
1111 1 1
month: n-5 n-4 n-3 n-2 n-1 n
projections (weight):
actual:
1234
4 3 2 1
34 12
month: Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec YEAR
projections (weight):
actual:
2468 1359 712 11 10
22
Source: BCB.
Since January 2009, a new annual category has been designed for the short and medium term rankings, considering, for each institution, a linear transformation of the deviations used for the monthly calculation in the short and medium term rankings, in such a way that the for each month institution that obtains the lowest monthly absolute deviation in a certain ranking for a given variable and in a given period is awarded 10 points, the institution that obtains the largest absolute monthly deviation in the same ranking for the same variable and the same term is assigned zero point, and the other institutions are awarded points, interpolated, between zero and 10. After the monthly points have been calculated, the average of the monthly values for each participant in the calendar year that has just ended is calculated and the annual rankings are determined. In order to be considered in these rankings, the institutions must have been classified into at least six monthly rankings in the calendar year (or four, in the case of the Selic rate) in each horizon (having started their projections, therefore, until the last day of the sixth month of the 12 months) and still: a) in the case of short-term, the deviations of the parcels referring to each monthly ranking calculation are equal to the average absolute deviation of the participating institutions for each of the dates prior to the start of their projections, according to the deviation that would have been assigned to these institutions if they had been included in the calculation of the ranking for each of these months - the linear transformation to the new ranking follows the pattern already mentioned; and b) in the case of the medium-term, instead of the average absolute deviation, the maximum absolute deviation is used. In this way, the consistence with the calculation procedures used for the monthly Top 5 of short and medium term is maintained. Thus, for each of the 12 months (or 8 meetings for the Selic) of the calendar year, variable (IPCA, IGP-DI, IGP-M, Exchange Rate and Selic) and maturity (short and medium) the institutions have scores ranging from zero to 10 and the average of these figures is the base variable for calculating the annual ranking. In the case of the Selic rate, instead of twelve months (and at least six participations), eight meetings (and at least four participations) should be considered. The intermediate parcels for the calculations of the deviations and the average notes are always rounded to the fourth decimal place.
21. Is there any precondition for the institutions to participate in the Top 5?
Some criteria applicable to all types of classification of the participating institutions have been defined for the purpose of imposing penalties for institutions that do not comply with the minimum requirements of transparency and timeliness in their projections: when calculating a given monthly ranking, the institutions that have not confirmed or updated within 30 days prior to the final reference date (or to each of the dates that make up the last couple of reference dates, in the case of the exchange rate and the Selic - details below), and at least three monthly projections and one annual, are excluded from the ranking.
23
For example, in the calculation of the short term ranking for the exchange rate in January 2007, which is released in February 2007, the institutions that do not have valid expectations on the two dates that make up the last couple of reference dates for January, that is, on December 29, 2006 (last working day of the previous month), and on January 15, 2007, are not classified. If the system identifies, for any institution, that there is no valid expectation on those dates, this institution does not participate in this ranking, for this variable. Valid expectations on a certain day refers to those included in the Expectations System and which have been confirmed or updated within the 30 calendar days prior to that date.
22. Which is the formula used for the calculation of the Short Run Top 5 monthly rankings?
The equation for calculating the monthly deviations used in the preparation of short-term Top 5 rankings is presented below.
Each institution "R" receives a penalty R CP, which is a simple average of the penalties received in each of the six months considered for the Top 5 Short Term.
In the Short Term monthly ranking, the institutions are classified based on R CP, defined in the Equation 1.
Equation 1:
𝜓𝑅𝐶𝑃 = { ∑ {(𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑝𝑒𝑛𝑎𝑙𝑡𝑦)𝑑𝑡∙ (1 − 𝑗𝑑𝑡
) + 𝑗𝑑𝑡
𝑁
𝑡=𝑁−5
∙ [(𝑚𝑎𝑥𝑖𝑚𝑢𝑚 𝑝𝑒𝑛𝑎𝑙𝑡𝑦)𝑑𝑡∙ (1 − 𝑘𝑑𝑡
) + (𝐸𝑑𝑡𝜏𝑡
𝑅 − 𝜏𝑡) ∙ 𝑘𝑑𝑡]}} ÷ 6
where: R CP = penalty assigned to the Institution R;
t = month for which the deviation is calculated; N= month related to the last reference date of the calculation period;
td = reference date of in the month t;
R
ttdE = projection of the institution R which is valid in
td for t (in the cases of exchange and
Selic rates, there are two reference dates);
t= actual result of the surveyed variable in month t;
tdpenalty) (average = absolute average deviation of the valid projections on td , for
t, in
relation to the actual result in month t;
tdpenalty) (maximum = maximum deviation of the valid projections on td , for
t, in relation to
the actual result in month t;
tdk = 0, when the institution does not have valid projection ontd ;
24
=1, when the institution does not have valid projection ontd ;
tdj = 0, if
td precedes the day on which the institution first recorded its projections for
(even if the institution recorded blank cells as its projections); = 1, if
td is equal or after the day on which the institution was enabled to inform its
projections for .
23. Which is the formula used for the calculation of the Medium-Run Top 5 rankings?
The equation for calculating the monthly deviations used in the preparation of medium-term Top 5 rankings is presented below.
In the Medium Term monthly ranking, the institutions are classified based on: R MP, as defined in the Equation 2.
Equation 2:
𝜓𝑅𝑀𝑃 = { ∑ (𝑁 − 𝑡 + 1)
𝑁
𝑡=𝑁−3
∙ ∑[(𝑚𝑎𝑥𝑖𝑚𝑢𝑚 𝑝𝑒𝑛𝑎𝑙𝑡𝑦)𝑑𝑡−𝑤+1∙ (1 − 𝑘𝑑𝑡−𝑤+1
) + (𝐸𝑑𝑡−𝑤+1𝜏𝑁−𝑤+1
𝑅 − 𝜏𝑁−𝑤+1)
3
𝑤=1
⋅ 𝑘𝑑𝑡−𝑤+1]} ÷ 30
where R MP= penalty assigned to the institution R;
t = month for which the deviation is calculated; N= month related to the last reference date of the calculation period; w= set of projections for the same monthly indicator;
1w-td = reference date of in the month t-w+1; R
wN 1d 1w-tE
= projection of the institution R which is valid on 1w-td for 1wN (in the cases of Exchange and Selic rates, there are two reference dates);
1wN = actual result of the surveyed variable in the month N-w+1; 1w-tdpenalty) (maximum
= maximum absolute deviation of the projections valid on 1w-td , for 1wN , in relation to the actual result;
1w-tdk
= 0, when the institution does not have valid projection on 1w-td ; = 1, when the institution has valid projection on 1w-td .
25
24. Which is the formula used for the calculation of the Long-Run Top 5 rankings, annually released?
The equation for calculating the monthly deviations used in the preparation of long-term Top 5 rankings is presented below.
In the Long Term monthly ranking, the institutions are classified based on: R LP, as defined in Equation 3.
Equation 3:
𝜓𝑅𝐿𝑃 = { ∑ (𝑁 − 𝑡 + 1) ∙ [(𝑚𝑎𝑥𝑖𝑚𝑢𝑚 𝑝𝑒𝑛𝑎𝑙𝑡𝑦)𝑑𝑡∙ (1 − 𝑘𝑑𝑡
) + (𝐸𝑑𝑡𝜏𝑅 − 𝜏) ⋅ 𝑘𝑑𝑡
]
𝑁
𝑡=𝑁−11
} ÷ 78
where: R LP= penalty assigned to the institution R;
t = month for which the deviation is calculated; N= month related to the last reference date of the calculation period;
td = reference date of in month t; R
tdE = projection of the institution R which is valid on td for (in the case of Exchange and Selic rates, there are two reference dates);
= actual annual result f the surveyed variable ; td)penalty (maximum = maximum absolute deviation of the projections valid on td , for , in
relation to the actual result; tdk = 0, when the institution does not have valid projection on td ;
=1, when the institution has valid projection on td .
25. What are the formulas used for the calculation of the Annual Top 5 Rankings (Short Run and Medium rankings)?
Equations for the Calculation of the Annual Rankings Short Term
In the Short Term annual ranking, the institutions are classified based on CPNBR , as defined in the Equation 4, for each variable surveyed .
Equation 4:
𝑁𝐵𝑅𝐶𝑃 = { ∑ (𝑁𝑜𝑡𝑒𝑅𝐶𝑃𝑚)
12
𝑚=1
} ÷ 12
26
Equation 5: 𝑁𝑜𝑡𝑎𝑅𝐶𝑃𝑚 = 10 ∙ [𝜓𝑅𝐶𝑃𝑚 − 𝜓𝐶𝑃𝑚𝑚𝑎𝑥] ÷ [𝜓𝐶𝑃𝑚𝑚𝑖𝑛 − 𝜓𝐶𝑃𝑚𝑚𝑎𝑥]
where: R CPm= penalty assigned to the institution R, for the month m, as per Equation 1;
CPmmax= maximum penalty of the institutions, for the month m, as per Equation 1; CPmmin= minimum penalty of the institutions, for the month m, as per Equation 1. If there is no R CPm for any month ”m” in the 12 months period (or 8 meetings, in the case
of the Selic rate), it should be used for this month “m”, *R CPm, defined as follows:
Equation 6:
𝜓𝑅∗𝐶𝑃𝑚 = { ∑ (𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑝𝑒𝑛𝑎𝑙𝑡𝑦)𝑑𝑡
𝑁
𝑡=𝑁−5
} ÷ 6
where: t = month for which the deviation is calculated; N= month concerning the last reference date of the calculation period;
td = reference date of the surveyed variable in month t; tdpenalty) (average = average absolute deviation of the valid projections on td , for t , in
relation to the result occurred in month t. Medium Term
In the Medium Term annual ranking, the institutions are classified based on MPNBR , as defined in the Equation 7, for each variable surveyed .
Equation 7:
𝑁𝐵𝑅𝑀𝑃 = { ∑ (𝑁𝑜𝑡𝑒𝑅𝑀𝑃𝑚)
12
𝑚=1
} ÷ 12
Equation 8:
𝑁𝑜𝑡𝑒𝑅𝑀𝑃𝑚 = 10 ∙ [𝜓𝑅𝑀𝑃𝑚 − 𝜓𝑀𝑃𝑚𝑚𝑎𝑥] ÷ [𝜓𝑀𝑃𝑚𝑚𝑖𝑛 − 𝜓𝑀𝑃𝑚𝑚𝑎𝑥]
where: R MPm= penalty assigned to the institution R, for the month m, as per Equation 2;
MPmmax= maximum penalty of the institutions, for the month m, as per Equation 2; MPmmin= minimum penalty of the institutions, for the month m, as per Equation 2. If there is no R MPm for any month ”m” in the 12 months period (or 8 meetings, in the case
of the Selic rate), it should be used for this month “m”, *R MPm, defined as follows: Equation 9:
27
𝜓𝑅∗𝑀𝑃𝑚 = { ∑ (𝑁 − 𝑡 + 1) ∙ ∑[(𝑚𝑎𝑥𝑖𝑚𝑢𝑚 𝑝𝑒𝑛𝑎𝑙𝑡𝑦)𝑑𝑡−𝑤+1]
3
𝑤=1
𝑁
𝑡=𝑁−3
} ÷ 30
where: t = month for which the deviation is calculated; N= month concerning the last reference date of the calculation period; w= a set of projections for the same monthly indicator;
1w-td = reference date of the surveyed variable in month t-w+1; 1w-tdpenalty) (maximum
= maximum absolute deviation of forecasts valid on 1w-td , for 1wN , in relation to the result occurred.
Related Documents
