Composite leading indicators of economic activity: An application to Rio de Janeiro's upstream oil and gas industry.
| Autor | Patrocinio, Rafael Goncalves |
-
Introduction
Anticipating upturns and downturns in business cycles is a key competitive advantage for companies and a powerful tool for policymakers. Since Mitchell and Burns (1938 and 1946), economic forecasting has attracted attention. A vast literature covers the measurement and the anticipation of economic activity in global, national, and sectoral levels. However, as Lucas (1977) defines, business cycles are easily misunderstood since they commonly refer to co-movements in different forms of economic activity, not just fluctuations in Gross Domestic Product (GDP).
In this paper, we construct a composite leading indicator (CLI) to the upstream oil and gas economic activity in the state of Rio de Janeiro. The starting point for a composite leading indicator is the choice of the target variable. In this sense, we choose IBGE's Pesquisa Industrial Mensal--Producao Fisica (Monthly industrial survey--physical production) for Rio de Janeiro's extractive industry as the target. The Pesquisa Industrial Mensal --Producao Fisica (PIM-PF) provides short-term fluctuations for extractive and transformation industries' real products since 1970. Since May 2014, PIM-PF has followed a new methodology; however, there was no break for the series dated back since 2002.
The upstream oil and gas sector is Rio de Janeiro's primary extractive industry and plays a key role in the state's economy. The sector is responsible for direct jobs, tax revenues, and investments, and provides a strategic raw material for economic and social development. Petroleum production royalties represented 18% of Rio de Janeiro's state revenue in 2018 (Secretaria da Fazenda do Rio de Janeiro, 2019). At the same time, it is also a multiplier for the state's economy, since service industries are intertwined with the upstream oil and gas sector. That said, anticipating changing business cycles in Rio de Janeiro's upstream oil and gas economic activity may contribute positively to a significant number of stakeholders.
The composite leading indicator in this paper follows the OECD's (2012) recommendation. The construction method has five steps: series pre-selection, seasonal adjustment, cycle identification, evaluation, and aggregation. First, we classify the potential leading indicators into three categories: rapidly responsive to economic activities, expectation-sensitive, and prime movers. We benefit from multiple public sources to gather the candidate leading series, such as ANP, BCB, CBOE, CNI, Eletrobras, EIA, FGV, Firjan, IBGE, OECD, Yahoo Finance, and others. Second, we remove seasonal patterns in the series and identify the cyles. Then, we evaluate the series best fit to integrate the composite leading indicator through four statistical tests: (i) cross-correlation (Hollauer, Issler, and Notini, 2009; Oliveira, 2016; NYU, 2017; Issler and Pimentel, 2019); (ii) quadratic probability score (Chauvet, 2000; Issler, Notini, and Rodrigues, 2009); (iii) Granger-causality (Issler, Notini, and Rodrigues, 2009; Oliveira, 2016); and (iv) probit (Morais and Chauvet, 2011). Finally, we use The Conference Board's (2001) aggregation method. Finally, we compare the predictive accuracy of autoregressive (AR) and vector-autoregressive (VAR) models--the latter includes lags of the CLI--using the Diebold-Mariano Test (Diebold and Mariano, 2002).
Our paper connects to the literature on measuring and analyzing business cycles in Brazil (see, for example, Chauvet, 2000; Picchetti and Toledo, 2002; Issler, Notini, and Rodrigues, 2009; Hollauer, Issler, e Notini, 2009; Morais and Chauvet, 2011; Oliveira, 2016; Campelo Jr., Issler and Pimentel, 2019; Ozyildirim, Sima-Friedman, Picchetti, and Lima, 2019). Specifically, we contribute to a more scarce literature that measures and analyzes cyclical aspects of specific sectors and industries, which has been done for industrial activity (Hollauer, Issler, and Notini, 2009), the capital goods industry (Chauvet and Morais, 2011), and the construction sector (Cruz and Colombo, 2018). To our knowledge, there is still a gap for the extractive industry. This paper provides the first composite leading indicator to Rio de Janeiro's upstream oil and gas industry economic activity.
The sections in this paper are organized as follows. Section 2 contains a brief review of international and Brazilian literature regarding composite leading indicators. Section 3 describes the methodology. Section 4 analyzes the main results of the paper and reports the assessment of the composite leading indicator constructed. Finally, Section 6 concludes with the main findings and further studies.
-
Review of literature
2.1 Composite leading indicators in the global landscape
Composite leading indicators research began back in the 1930s when Burns and Mitchell (1938 and 1946) developed a pioneering work concerning business cycles. They developed a list of leading, coincident and lagging series of economic activity in the United States in an effort to explain the recession that began in 1937. As stated by Mitchell and Burns:
"Business Cycles are a type of fluctuation found in the aggregate economic activity of nations that organize their work mainly in business enterprises. A cycle consists of expansions occurring at about the same time in many economic activities, followed by similarly general recessions, contractions, and revivals which merge into the expansion phase of the next cycle; this sequence of changes is recurrent but not periodic; in duration business cycles vary from more than one year to ten or twelve years." (Burns and Mitchell, 1946, p 3)
Early attempts to explain business cycles were not unanimously approved. Koopsmans (1947) considered Mitchell and Burns' work excessively empirical, but the authors disagreed and re-affirmed that the work was fundamentally based on the economic theory.
Moore (1961) added other series to the initial list proposed by Mitchell and Burns in the context of the post-war business transformation. Following this idea, Moore and Shiskin (1967) developed composite leading, coincident, and lagging indicators by a combination of the series. Recessions and expansions do not have a single explanation, as economic theory and experience have demonstrated. That said, a combination of series into a composite indicator may better explain signals from different sectors of the economy.
New methods for composite indicators have appeared successively, from heuristical to more complex models. The Conference Board (TCB) has published composite leading, coincident, and lagging indicators of economic activity in the United States since 1995, through a methodology that requires no estimation based on formal econometric models. Alternatively, Stock and Watson (1988a, 1988b, 1989, 1993) propose construction methods using sophisticated econometric and statistical techniques. However, a major failure in Stock and Watson's model was the missing US recession in 1990-1991 (Issler, 2009). Chauvet and Piger (2008) compare the real-time performance of business cycle dating methods and find that both the Bry-Boschan algorithm and a Markov-switching dynamic-factor model have accurately identified the NBER business cycle chronology over the past 30 years. Duarte, Issler and Spacov (2004) compare the state of the economy dating abilities from TCB's method and a series of alternative econometric-based models. Despite the simplicity present in TCB's method, the results are notably satisfactory.
Finally, since the 1970s, OECD has provided a system of composite leading indicators. OECD (2012) details the construction process in four steps: pre-selection, filtering, evaluation, and aggregation. This paper follows those steps, and they will be further described in Section 3. For further reading on leading indicators, we recommend Lahiri and Moore (1992) and Marcellino (2006). For in-depth analyses of business cycles, we suggest Diebold and Rudebusch (1999).
2.2 Composite leading indicators in Brazil
Contador (1977) and Contador and Ferraz (1999) developed pioneering work on coincident and leading indicators in Brazil, followed by Chauvet (2001, 2002) and Picchetti and Toledo (2002). In 2002, IBGE redesigned the Pesquisa Mensal do Emprego (Monthly Employment Survey) which characterized a major drawback in composite indicators research, since long-term series are crucial for business cycle analysis. Research and development of composite indicators in Brazil face major challenges regarding the frequency, timeliness, revision, and length of the series. Data should be available in monthly frequency, shortly after the period to which it refers, with no significant revision and no breaks in the life span. Despite these challenges, Duarte, Issler, and Spacov (2004) compare three alternative indicators for Brazilian economic activity and establish a recession chronology. Ellery Junior (2002) analyzes the empirical relationship between GNP and key factors, such as investment, work hours, consumption and productivity. Chauvet and Morais (2008) construct a composite leading indicator to anticipate downturns in the Brazilian business cycle through a time-varying autoregressive probit model.
Other business cycle studies have been developed for Brazil in order to understand sector specificities. Hollauer, Issler and Notini (2009) construct a composite coincident indicator for Brazilian industrial activity through a comparison of The Conference Board's (TCB) method, linear dynamic models and the Mariano-Murasawa approach. They conclude that TCB's method is superior for this purpose. Morais and Portugal (2008) develop a composite coincident indicator of industrial activity in the state of Rio Grande do Sul.
Even though composite coincident indicators have already been exhaustively studied in Brazil, there is still room to advance in composite leading indicators for sectoral analysis. To address this, Campelo Junior (2008) develops a composite leading...
Para continuar a ler
PEÇA SUA AVALIAÇÃOCOPYRIGHT GALE, Cengage Learning. All rights reserved.
