Itaú BBA - Revision of Itaú Unibanco’s Monthly GDP series; decline in October

Macro Vision

< Volver

Revision of Itaú Unibanco’s Monthly GDP series; decline in October

diciembre 18, 2017

The revised series include activity segments that were not covered by the original index.

This report details the adjustments made to the Itaú Unibanco monthly GDP series (PM-Itaú, previous acronym was PIBIU) to incorporate data points that were not published when the index was created.

The resulting series has greater quarterly correlation with the official GDP figure and includes activity segments that were not covered by the original index.

Itaú Unibanco monthly GDP[1] (PM-Itaú) fell 0.8% mom/sa in October. The decline was widespread among components. The 3-month moving average dropped 0.3%.

Revision incorporates new information

We are revising the proprietary Itaú Unibanco monthly GDP (PM-Itaú) series to incorporate additional groups of variables to the index starting in January 2013. The purpose is to enhance the index’s ability to assess the monthly evolution of economic activity in Brazil.

Some of the variables included were not available when the index was created in mid-2009:

-  The IBGE Monthly Service Survey (PMS), first published in January 2012, provides monthly figures on real revenues from segments that represented 30.2% of the service sector’s GDP in 2016. The survey allowed us to add the segment comprising information and communication.

- The IBGE National Household Survey (PNAD) was restructured in March 2012 and replaced the Monthly Employment Survey. Indicators in the monthly PNAD are used to assess two service segments (public administration and real estate activities — the latter was also absent from the original index) and the construction sector. One significant restriction is that the survey calculates a three-month moving average, but we chose to treat it as a monthly number based on our belief that the benefit of tracking activity overshadows the overlap effect with the previous two months and the following two months.

We also included i) the Central Bank Credit Note, which allows us to track another relevant service segment absent from the previous version (financial intermediation); ii) National Oil Production,   published by the National Petroleum Agency (ANP); and iii) Power Generation by Source, published by the National Operator of the Electric System (ONS). The two latter resources add value to our analysis of the mining and extraction segment as well as of the utility segment (gas and electricity, water and sewage, and urban sanitation).

Enhanced series has greater correlation with IBGE’s quarterly GDP numbers

Itaú Unibanco monthly GDP (PM-Itaú) advanced 1.4% between December 2016 and September 2017, led by agriculture and livestock output and retail (up by 4.8% and 4.7%, respectively).

The enhanced PM-Itaú series shows a greater quarterly correlation with the official GDP (by IBGE). The quarterly average for the PM-Itaú from 1Q13 to 3Q17 shows a r-square of 99.3% with the official quarterly GDP reading, vs. 97.2% in the comparison using Itaú Unibanco’s monthly GDP series before the revision. The improved correlation is evidenced, for instance, in the recession from 2014 to 2016: both the official GDP and the new monthly GDP series indicate that activity peaked in 1Q14, while our monthly GDP figure before the revision indicated that the peak took place in 3Q13. Additionally, the series before the revision indicated that the recession was deeper than it actually was, adding up to -10.1% from 1Q14 to 4Q16, vs. -8.2% according to the official GDP series and -7.9% in the new monthly GDP series.

October: Widespread decline

Itaú Unibanco monthly GDP (PM-Itaú) fell 0.8% mom/sa in October, leading the 3-month moving average to -0.3%. The yoy growth figure stands at 2.2%. In the quarter through October, the PM-Itaú declined 0.4% qoq/sa and is up 1.7% against the same quarter one year ago.

  

The breakdown by components shows the decline was widespread. Ten out of thirteen PM-Itaú components posted declines in October, led by Construction (-1.4%) and Retail (-1.3%). Information & Communication posted the strongest gain (0.7%).

Preliminary figures for November suggest a small gain in PM-Itaú.

 

Appendix 1: Technical Note

Itaú Unibanco monthly GDP tracks the level of economic activity in Brazil. Its good correlation with quarterly GDP (as calculated by the National Accounts System – Reference 2010 - IBGE) allows it to be used as a source to forecast GDP.

The indicator is made up of proxies for each supply-side component of quarterly GDP: Agriculture & Livestock; Extraction & Mining; Manufacturing; Electricity & Gas; Water & Sewage and Urban Sanitation; Construction; Retail; Transportation; Storage & Mail, Information & Communication; Financial & Insurance Activities and Related Services; Other Service Activities; Public Administration, Defense, Public Education, Healthcare & Social Security; and Taxes. These proxies are linked to a moving base using the previous year’s weights. The weights follow those used in the national quarterly accounts for each category. Applying the aggregation provides a moving base index for the Itaú Unibanco monthly GDP series, which is then transformed into a fixed base.

Because not all of the series that make up monthly GDP were established before 2003 (first year of the non-revised series of our activity indicator), new data and models were used starting in 2013. The series for 2003 to 2012 was recreated using the evolution of the original series in the annual comparison, based on the revised series for 2013.

The revised series starting in 2013 is then seasonally adjusted using the X-13 ARIMA-SEATS method (of the U.S. Census Bureau). Seasonal adjustment specifications are similar (despite differences in frequency) to those used by IBGE to seasonally adjust its quarterly GDP. For the period before 2013, the seasonally adjusted monthly change of the previous series (before the revision) was kept fixed.

Although the PM-Itaú quarterly average is close to IBGE’s GDP reading, we have not assumed the latter as just a monthly figure. The quarterly change in monthly GDP is not necessarily equal to the quarterly change in GDP calculated in the National Accounts.

 

Artur Manoel Passos
Alexandre Gomes da Cunha

 


[1] To get both our original and seasonally-adjusted series on the Bloomberg, type PIBIU or PIBISA . To get them on Broadcast, use ticker PIBIU or PIBIUD (seasonally adjusted). The series are also available at: https://www.itau.com.br/itaubba-en/economic-analysis/our-economic-series/spreadsheets



< Volver