Forecasting Brazilian Output in the Presence of Breaks: A Comparison of Linear and Nonlinear Models

Abstract

Standard-Nutzungsbedingungen:Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden.Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen.Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. The ability of each model in forecasting out-of-sample the growth rates of GDP is examined. The forecasting ability of the two models is also compared with linear specifications. The authors find that nonlinear models display the best forecasting performance and that specifications including the presence of structural breaks are important in obtaining a representation of the Brazilian business cycle. Terms of use: Documents inJEL classification: C32, E32

Business Cycle in the Industrial Production of Brazilian States

“…The MSMH(2)-AR(0) formulation, however, had few specification problems. Chauvet(2002), for instance, also used the MSM(2)-AR(0) model to estimate the business cycle of Brazilian GDP in order to overcome the problem with the structural breaks in the series. 16 The standard likelihood ratio test carried out to determine the functional form of univariate models, that is, regarding MSMH(2)-AR(p) as unrestricted model, and MSM(2)-AR(p) as restricted model, indicates that for SP, in AR(1) and AR(0) formulations, and for PR in 13 MSM(k)-AR(p) models with k>2 and p>4 were tested, but they had convergence problems.…”

“…As we can see, the industry of SP and MG have 5 recessions, RS has 6, RJ presents 8 and PR has11 recessions, where special attention should be paid to the early 1980s, which was marked by a generalized reduction in the industrial activity, and to the economic plans implemented in the second half of that decade, which stagnated production until mid-1992. 18 The criterion adopted here to characterize an economic cycle was based on the proposal of the NBER and the same used in Chauvet(2002), that is, a recession (expansion) corresponds to a reduction (increase) in activity with a minimum length of 2 quarters . The author found that Brazil's GDP had 8 recessions and 9 expansions between the second quarter of 1980 and the first quarter of 2000, which were caused by external shocks.…”

“…Chauvet(2002) also found out that the Brazilian economy started the 1980s in recession and that between 1983 and 1987 it experienced 16 quarters of expansion.…”

“…These econometric techniques have been seldom applied to assess the behavior of Brazilian macroeconomic series, except for the studies conducted by Chauvet (2002), Chauvet, Lima and Vasquez (2002), Barros (1993), Engle and Issler (1993) and Picchetti and Toledo (2002). Just as underscored in Pinheiro and Almeida (1995) and Pacheco (1999), the Brazilian industry has a different production dynamics among Brazilian states, and the regime shift methodology proves adequate for characterizing not only these differences, but also for explaining the effects of an adverse shock to the regional industry.…”

Leading Indicators for the Capital Goods Industry

“…There are some other related articles, which do not aim to predict business cycle turning points, but to temporally project the aggregate activity instead. For example, Ferreira et al (2005) use diffuse linear and nonlinear indicators to predict the rate of growth of the Brazilian GDP, and Chauvet et al (2006) compare the forecast ability for GDP growth in Brazil using several specifications of Markov regime switching models.…”