Skip to main content
Log in

On the accuracy of private forecasts of inflation and growth in Brazil

  • Published:
Journal of Economics and Finance Aims and scope Submit manuscript

Abstract

This study evaluates the accuracy of the private-sector forecasts of inflation and growth in industrial production collected by the Brazilian Central Bank (BCB). In addition to examining directional accuracy, we utilize as benchmarks both naïve and univariate autoregressive moving-average (ARMA) forecast and test rationality under flexible loss to allow for the possibility of asymmetric loss. Our analysis yields three important findings: First, the private forecasts are directionally accurate. Second, they are superior to the naïve forecasts and are either superior or as accurate as the ARMA forecasts. Third, the private forecasts are generally rational under either asymmetric or symmetric loss. Such findings point to the success of the BCB in anchoring private expectations. Given their importance as monetary policy inputs, we conclude that private inflation and growth forecasts are of value to the BCB for policymaking.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2

Similar content being viewed by others

Notes

  1. See, among others, Romer and Romer (2000), Gavin and Mandal (2001), and Groen et al. (2009).

  2. In order to have a reasonable number of observations for our analysis, we have used the “rate of change in industrial production index” as a measure of growth (instead of the rate of change in real GDP which is measured on a quarterly basis).

  3. The BCB collects the forecasts on a daily basis. The information available in the system up to 5 pm of a particular day is used to calculate the median forecast for that day. Daily consensus (median) forecasts of inflation and growth in industrial production are available at www.bcb.gov.br/

  4. The actual data for the IPCA for month t-1 is released between the 4th and the 14th day of month t. The actual data for the industrial production index for month t-2 is released within the first 11 days of month t. The forecasts utilized in this paper are made on the 15th through 18th day of month t.

  5. See Management Reports 2008 and Management Reports 2009 available on the Central Bank of Brazil website).

  6. Our findings indicate some anomalies. For instance, the symmetry and rationality test results at some forecast horizons are different from those at other forecast horizons. Such anomalies (which are common in forecasting) may be explained by the fact that our study uses the consensus forecasts since the data for individual forecasters are not available. As for symmetry, Timmermann (2006) points out that forecast combination may average out idiosyncratic variation in loss functions. As such, the individual forecasts may imply asymmetric loss but the consensus forecast may imply symmetric loss. As for rationality, Dovern and Weisser’s (2008) findings, using the forecasts for G7 countries, indicate that the consensus forecasts fail to be rational even though the majority of individual forecasts are rational.

References

  • Baghestani H, Marchon C (2012) An evaluation of private forecasts of interest rate targets in Brazil. Econ Lett 115:352–355

    Article  Google Scholar 

  • Bernanke B, Woodford M (1997) Inflation forecasts and monetary policy. J Money, Credit, Bank 29:653–684

    Article  Google Scholar 

  • Carvalho FA, Minella A (2012) Survey forecasts in Brazil: a prismatic assessment of epidemiology, performance, and determinants. J Int Money Finance 31:1371–1391

    Article  Google Scholar 

  • Diebold FX (2007) Elements of Forecasting, 4th edn. South-Western, Cincinnati

    Google Scholar 

  • Diebold FX, Lopez JA (1996) Forecast evaluation and combination. In: Maddala GS, Rao CR (eds) Handbook of statistics, vol 14. North–Holland, Amsterdam, pp 241–268

    Google Scholar 

  • Diebold FX, Mariano RS (1995) Comparing predictive accuracy. J Bus Econ Stat 13:253–263

    Google Scholar 

  • Dovern J, Weisser J (2008) Are they really rational? Assessing professional macroeconomic forecasts from the G7 countries. Kiel Working Papers No. 1447, Kiel Institute for the World Economy, Kiel

  • Elliott G, Komunjer I, Timmermann A (2005) Estimation and testing of forecast rationality under flexible loss. Rev Econ Stud 72:1107–1125

    Article  Google Scholar 

  • Elliott G, Komunjer I, Timmermann A (2008) Biases in macroeconomic forecasts: Irrationality or asymmetric loss. J Eur Econ Assoc 6:122–157

    Article  Google Scholar 

  • Gavin WT, Mandal RJ (2001) Forecasting inflation and growth: do private forecasts match those of policymakers? Bus Econ 36:13–20

    Google Scholar 

  • Groen JJJ, Kapetanios G, Price S (2009) A real time evaluation of Bank of England forecasts for inflation and growth. Int J Forecast 25:74–80

    Article  Google Scholar 

  • Marques ABC, Fachada P, Cavalcanti DC (2003) Sistema Banco Central de Expectativas de Mercado. Notas Técnicas do Banco Central do Brasil, no. 36, Junho

  • Newey WK, West KD (1987) A simple, positive semi-definite, heteroskedasticity and autocorrelation consistent covariance matrix. Econometrica 55:703–708

    Article  Google Scholar 

  • Pearce DK (1979) Comparing survey and rational measures of expected inflation: forecast performance and interest rate effects. J Money Credit Bank 11:447–456

    Article  Google Scholar 

  • Romer CD, Romer DH (2000) Federal reserve information and the behavior of interest rates. Am Econ Rev 90:429–457

    Article  Google Scholar 

  • Sinclair TM, Stekler HO, Kitzinger L (2010) Directional forecasts of GDP and inflation: a joint evaluation with an application to Federal Reserve predictions. Appl Econ 42:2289–2297

    Article  Google Scholar 

  • Stekler HO (1994) Are economic forecasts valuable? J Forecast 13:495–505

    Article  Google Scholar 

  • Timmermann A (2006) Forecast Combinations. In: Elliott G, Granger CWJ, Timmermann A (eds) Handbook of economic forecasting. North–Holland, Amsterdam, pp 135–196

    Chapter  Google Scholar 

Download references

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Hamid Baghestani.

Appendix A

Appendix A

This appendix describes how the autoregressive moving-average (ARMA) forecasts of inflation and growth are generated. To start with, we use the augmented Dicky-Fuller (ADF) test to examine the unit root behavior of the series. The results indicate that the unit root null hypothesis is rejected for both series (for inflation, the MacKinnon p-value of the calculated ADF test statistic for 1996.01–2002.12 is 0.001, and for growth, the p-value for 1996.01–2002.11 is 0.040). We utilize the inflation (growth) series for 1996.01–2002.12 (1996.01–2002.11) to estimate the autocorrelation and partial autocorrelation functions. Using these estimates and the Schwarz criterion, we then select the following ARMA models (absolute t-values are in parentheses).

  • CPI inflation (P t ):

    $$ \begin{array}{l}{P}_t=\underset{(7.40)}{0.590}+\underset{(9.95)}{\left(1+0.774\left.B\right)\right.}{\widehat{u}}_t\hfill \\ {}\begin{array}{cccc}\hfill \hfill & \hfill \hfill & \hfill {\mathrm{R}}^2=0.38,\hfill & \hfill p-\mathrm{value}\kern0.5em \mathrm{of}\kern0.5em Q\left[12\right]=0.575.\hfill \end{array}\hfill \end{array} $$
  • Growth in industrial production (G t ):

    $$ \begin{array}{l}\left(1-\underset{(3.05)}{0323B}-\underset{(2.74)}{0295{B}^2}-\underset{(1.55)}{0.166{B}^3}\right){G}_t=\underset{(1.61)}{3.378}+{\widehat{u}}_t\hfill \\ {}\begin{array}{cccc}\hfill \hfill & \hfill \hfill & \hfill {\mathrm{R}}^2=0.49,\hfill & \hfill p-\mathrm{value}\kern0.5em \mathrm{of}\kern0.5em Q\left[12\right]=0.196.\hfill \end{array}\hfill \end{array} $$

As can be seen, the parameter estimates are generally significant. Also, the p-values of calculated Ljung-Box Q-statistics are above 0.10, indicating that the residual series are white noise and thus the ARMA models are correctly specified.

We employ the above 1996.01–2002.12 ARMA model estimates for inflation to generate the forecasts for 2003.01–2004.01. These forecasts correspond to the current-month and one- through twelve-month-ahead private forecasts made in 2003.01. Re-estimating the model for 1996.01–2003.01, we use the updated parameter estimates to generate the forecasts for 2003.02–2004.02. Again, these forecasts correspond to the current-month and one- through twelve-month-ahead private forecasts made in 2003.02. This procedure is repeated until the last set of forecasts is generated. Note that, at the time of the survey, the participants do know the actual inflation for the prior month.

For growth, we employ the above 1996.01–2002.11 ARMA model to generate the forecasts for 2002.12–2004.01 (a total of 14 forecasts). The last 13 forecasts correspond to the current-month and one- through twelve-month-ahead private forecasts made in 2003.01. Re-estimating the model for 1996.01–2002.12, we use the updated parameter estimates to generate the forecasts for 2003.01–2004.02 (a total of 14 forecasts). Again, the last 13 forecasts correspond to the current-month and one- through twelve-month-ahead private forecasts made in 2003.02. This procedure is repeated until the last set of forecasts is generated. Note that, at the time of the survey, the participants do not yet know the actual growth for the prior month.

Rights and permissions

Reprints and permissions

About this article

Cite this article

Baghestani, H., Marchon, C. On the accuracy of private forecasts of inflation and growth in Brazil. J Econ Finan 39, 370–381 (2015). https://doi.org/10.1007/s12197-013-9263-1

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s12197-013-9263-1

Keywords

JEL Classification

Navigation