IS IBC-Br, ESTIMATED INDEX FOR GDP FROM THE CENTRAL BANK OF BRAZIL 1, A GOOD APPROXIMATION TO QUARTERLY BRAZILIAN GDP? 2

The IBC-Br is the result of a faster and more easily constructed statistical survey by the Central Bank. A similar index is constructed by he Federal Reserve of Philadelphia. It is mainly used to assist in the construction of short-term public policies. It is often called by the press the “GDP preview”. More recently it is being used by the business sector as an element for cost and sales forecasting. It was developed as a parallel indicator to the quarterly GDP to reflect monthly macro-economic activity with a minimum disclosure lag. It is published usually during the first week of the present month estimating the macro-economy of two months ago. Being extremely valuable as a convenient representation of GDP for both business and government, the IBC-Br should be investigated regularly, even by private citizens, to avoid recurring errors.

Even though both macroeconomic measures are estimates of the same economic phenomenon but with different data sources, naturally there are statistical discrepancies that appear. This may happen because, among other problems, inaccurate numerical data and errors from data collection are unavoidable. The question is, do these discrepancies invalidate the utility of IBC-Br as a quick short-term substitute for official GDP which is published only 4 times a year and with a rather long delay? In other words, to what extent do they approach being similar allowing IBC-Br to be a quick representation of GDP?

DATA FROM THE CENTRAL BANK OF BRAZIL

PERCENTAGE CHANGES IN QUARTERLY GDP and IBC-Br SINCE THE FIRST QUARTER OF 2003 UNTIL THE SECOND QUARTER OF 2023.

The plot is not very enlightening but it does serve as a guide to some further comments. Are the measures coincidental? When short-term IBC-Br changes direction, does this represent a signal for the same change in GDP? It would be worthwhile to plot the difference between the two series as a measure of error either in terms of overestmation (growth of IBC-Br greater than GDP) or underestimation (IBC-Br smaller than GDP).

IBC-Br BY QUARTER PERCENTAGE ERROR OVERESTIMATION (IBC-Br > GNP)

A better way of seeing the error, month by month, is through the estimation of the difference between IBC-Br and GDP, a measure of error. See the plot above. Note the relatively small errors from about 2009 up to 2019. During this period, estimation of IBC-Br was relatively coherent with GNP.

PERCENTAGE CHANGE IBC-Br x GDP

In the above plot, each point represents the movement of the IBC-Br index of quarterly data and the movement of the corresponding index for quarterly GDP. If IBC-Br goes up in a certain period then this should indicate that official GDP should also show an increase when more accurate and available data are gathered. In a perfect world, for perfect estimates of GDP and IBC-Br, the data points should run along the black diagonal line where the two measures are equal. Reality is of course different. To make the analysis clearer, the plot has been divided into four quadrants.

The fourth quadrant (IV) points out estimation errors that are beneficent (and a relief to policymakers) in light of GDP increasing in spite of the fact that estimation for IBC-Br was pessimistic. Any error is undesirable but in this case the error is considered minor. However, in the 80 trimesters present in the plot, 15 are in the fourth quadrant which seems to be a rather large number. The second quadrant also represents errors but these are the catastrophic kind where the change in IBC-Br was positive (optimistic) but the change in GDP was unfortunately negative. The optimism that results from an estimated positive change in IBC-Br is disqualified by the fall in GDP. This is the error that should be avoided at all costs. There are only 5 errors of this kind in the total of 80 trimestres. It would appear that the Central Bank is not so unfavorable to the errors of quadrant IV where disappointing short-term estimates are eventually overcome by the relief of an increasing GNP. However, on the other hand, the Central Bank is very aware of the huge disappointment that comes from the false optimism (IBC-Br up) cut short by an uncomfortable reality (GNP down) in quadrant II.

TRUE   quadrants counts percent
TRUE 1         I     47   58.75
TRUE 2 II(error)      5    6.25
TRUE 3       III     13   16.25
TRUE 4 IV(error)     15   18.75

CONCLUSIONS

We conclude that the IBC-Br could be wrong about 25% of the time, not a small error by any means. It is common to respect analyses of economic models for errors that occur 10% of the time or less. We must always remember that one unique economic variable is never sufficient to explain the macro-economic environment but that a unique variable in conjunction with a host of others may contribute to a better understanding of economic phenomenon. IBC-Br, use with caution.

  1. IBC-Br Very short-term and monthly Central Bank Economic Activity Index. https://www3.bcb.gov.br/sgspub/localizarseries/localizarSeries.do?method=prepararTelaLocalizarSeries↩︎

  2. GDPtrim longer-term Central Bank Economic Activity Index https://www3.bcb.gov.br/sgspub/localizarseries/localizarSeries.do?method=prepararTelaLocalizarSeries↩︎