Business Cycles synchronicity and income levels: has globalization brought us closer than ever? Rosmy Jean Louis* Daniel Simons Abstract



Download 124.98 Kb.
Page1/12
Date09.08.2018
Size124.98 Kb.
  1   2   3   4   5   6   7   8   9   ...   12
Business Cycles synchronicity and income levels: has globalization brought us closer than ever?
Rosmy Jean Louis*

Daniel Simons**
Abstract
Research on business cycle linkages show a tendency to model countries of relatively the same income level jointly. However, the issue of whether countries with the same income level move along the same business cycles has not been investigated formally. The recent push by regional/economic bloc of countries toward the adoption of one currency tends to suggest an interdependence of macroeconomic activities but not necessarily a common cycle. In this paper, we use real income per capita from the United Nations Statistical Database based on income classification by the World Bank and the nonparametric measure of cycle synchronicity proposed by Mink, Jacobs, and de Haan (2007) to develop a two-step approach to investigate the linkage between business cycles and income levels. First, we examine the business cycles of each category of countries to determine whether 1) each group of countries follows its own dynamics and is therefore subjected to the same business cycle and 2) these cycles are independent of each other across groups. Second, we use panel data analysis in search for an explanation of the synchronicity of the cycles observed. The overall results indicate that high income per capita countries (HICs) tend to be guided by stronger similarity in business cycles than countries in the middle (MICs) and low income (LICs) groups. The synchronicity ratios are on average 51, 50, 54, 73, and 100 percent for the LICs, low middle income countries (LMCs), upper middle income countries (UMCs), HICs - OECD, and HICs - non-OECD countries, respectively. We also determined that across groups, the wavelength was common in most of the countries suggesting the existence of a common world cycle. The results from the robust fixed effects estimation show that real oil price is consistently significant in explaining the synchronicity of output gaps.
Short Abstract

Research on business cycle linkages shows a tendency to model countries of relatively the same income levels jointly. However, the issue of whether these countries move along the same business cycles has not been formally investigated in the literature. In this paper, we take this approach and investigate whether each group of countries follows its own dynamics and is therefore subjected to the same business cycle and whether these cycles are independent of each other across income groups. Results indicate that high income per capita countries (HICs) tend to be guided by stronger similarity in business cycles than countries in the middle (MICs) and low income (LICs) groups. In search for an explanation of the business cycles synchronicity observed, panel data analysis was explored. The results from the robust fixed effects estimation show neither trade openness nor shocks to consumption underlie international business cycle synchronization, but rather shocks to oil prices.


Keywords: Economic Development, Business Cycles, Income per capita, Panel Data Analysis, economic integration and globalization
JEL Classification: C0, E0, F1, F3
Introduction
The research on business cycle linkages that abounds in the literature shows a tendency to model countries of relatively the same degree of economic development jointly. The works of Gerlach (1988), Backus and Kehoe (1992), Backus et al. (1995), Norrbin and Schlagenhauf (1996), Gregory et al. (1997), Artis and Zhang (1997), Artis, Kontolemis, and Osborn (1997), Bergman et al. (1998), Mills and Holmes (1999), Clark and Shin (2000), Lumsdaine and Prasad (2003), Artis, Krolzig, and Toro (2004), Stock and Watson (2003, 2005), Chauvet and Yu (2006), and Crucini, Kose, and Otrok (2011) are examples of such practice. All these studies used OECD countries in their search for an international business cycle.
There have also been some attempts to jointly model countries of different economic backgrounds (Crucini 1997, 1999; Aguiar and Gopinath 2007, among others). For example, Mendoza (1995) and Kose (2002) document the similarity in business cycle features of developed and developing countries, Jean Louis (2004) and Jean Louis and Simons (2005, 2007) investigate the business cycle linkages between North American countries but could not conclude that Mexico shares a common cycle with the United States and Canada combined. Mejia-Reyes (1999) models the United States along with the major Latin American economies and arrives at a similar conclusion. The business cycles of the United States and most of these countries are idiosyncratic, on a pairwise comparison basis. Focusing on Asia, Girardin (2002) found little evidence of symmetry between Japan and other South East Asian countries’ business cycle in a comparison of univariate results. Kose, Otrok, and Whiteman (2003) is the most comprehensive of the studies cited in the literature thus far. They use a Bayesian dynamic latent factor model to estimate common components in macroeconomic aggregates such as output, consumption, and investment for a sample of 60 countries covering seven regions of the world. They find that a common world factor is an important source of volatility for aggregates in most countries whereas region-specific shocks only play a minor role. In their view, this constitutes evidence on the existence of a world business cycle.
Although current studies on international business cycle linkages implicitly give indication that there is a link between business cycle synchronization and levels of economic development, this research question has not been formally addressed in the literature. There are at least three reasons for exploring this issue. First, there is the idea that an all Americas’ Monetary Union could be a stronger economic bloc to compete with the European Union and other rising economic powers such as India and China. Second, there is the debate of “One World, One Money?” without abolishing national currencies, which revisits the idea of a global money proposed by Keynes in 1944 (Mundell, 1995; 2001; Friedman 2001; Starr 2006). Third, there has been a proliferation of trade agreements among countries around the globe. On all accounts, our research contributes to the overall debate.
Be it because of competition at the world level that gives rise to the creation of economic blocs or because of globalization of markets that might necessitate a world currency to facilitate international transactions, a study on business cycle linkages that accounts for the level of economic development within and across blocs is enriching for the ongoing debate. Moreover, business cycle synchronization is a prerequisite, in line with Mundell (1961), for countries that contemplate higher forms of economic integration beyond customs union. Countries forming those blocs must be subjected to similar shocks, hence common cycle, in order for a “one-size-fits-all” monetary policy to be effective for each member of the group. The absence of a common cycle in these unions may lead to severe complications from monetary policies for the member nations.
Although there have been some research (Alesina and Barro 2002; Alesina, Barro, and Tenreyro 2003) in the literature that explore the benefits of currency unions for countries of different sizes and degree of specialization in the production of goods and services, it still remains a subject of contention whether industrialized and less-developed nations could find a mutually beneficial agreement. Disparities between the two groups of countries in America and other continents are very well pronounced. In this paper we use the nonparametric measure of cycle synchronicity proposed by Mink, Jacobs, and de Haan (2007) to develop a two-step approach to investigate the linkage between business cycles and income levels. First, we examine the business cycles of each category of countries to determine whether 1) each group of countries follows its own dynamics and is therefore subjected to the same business cycle and 2) whether these cycles are independent of each other across groups. Second, we use panel data analysis in search for an explanation of the synchronicity of the cycles observed. We extracted data on real income per capita for the 217 countries included in the National Accounts Main Aggregates of the United Nations Statistical Database and classified these countries by categories of income as per the World Development indicators. The preliminary results indicate that High Income per capita countries (HICs) tend to be guided by stronger similarity in business cycles than countries in the Middle (MICs) and Low income (LICs) groups. The synchronicity ratios (SRs) are on average 51, 50, 54, 73, and 100 percent for the LICs, Low middle income countries, Upper middle income countries, HICs - OECD, and HICs - non-OECD countries, respectively. We also determined that across groups, the wavelength was common in most of the countries suggesting the existence of a common world cycle. Section 2 presents the methodology. Section 3 deals with the data and results and Section 4 concludes the paper.
2. Methodology
The starting point towards uncovering business cycle synchronicity or lack thereof across countries based on their levels of income per capita is the determination of the measure of the business cycle itself. To this end, we use the Hodrick and Prescott (1997) nonparametric filter to decompose real output per capita into a trend and a cycle, where the trend is the potential output and the cycle is the deviation of actual output from its potential level. Output gap is calculated as the ratio of cycle over trend for each country. Once output gap is determined, a number of techniques are available in the literature to investigate the extent of co-movement of the cycles. These include Markov-switching vector autoregression decomposition as in Artis, Krolzig, and Toro (2004), Krolzig (1997a, 1997b, 2005), Krolzig and Sensier (2000), cointegration analysis as in Engle and Granger (1987), Stock and Watson (1988), Johansen (1988, 1991), test for common features as Engel and Kozicki (1993), tests for common trends and common cycles as in Beveridge and Nelson (1981), Vahid and Engle (1993), and Engle and Issler (1995).1 In addition to the conventional correlation, a few non-parametric tests have been developed recently to measure business cycle synchronization. For example, Kalemli-Ozcan, Papaioannou, and Peydró (2011) use three different measures to investigate the linkage between financial integration and business cycle synchronization (SYNCHi,j,t) for 20 OECD countries on a bilateral basis. The first measure of synchronization, which follows Giannone, Lenza, and Reichlin (2009) is defined as the negative of the absolute value of the differential growth rate of real GDP per capita country pairs (i, j) over time:
(1)
The second measure, SYNCH2i,j,t which is based on Morgan, Rime, and Strahan (2004), consists in estimating the real GDP per capita growth on country fixed-effects and year fixed-effects for each country to obtain a residual whose absolute value is used to construct the business cycle synchronization proxy as the negative of the absolute value of the differential between two countries.
(2)
These residuals account for cross-country and across-year mean growth in real GDP per capita fluctuations:
(3)

Therefore,


(4)

Simply put, this index measures how similar growth rates are between each pair of countries in any given year when we account for the average growth rate in each country and the average growth in each year.


The third measure, SYNCH3i,j,t, follows Imbs (2006) and Baxter and Kouparitsas (2005) in computing the 5-year correlation of the cyclical component of output obtained from Baxter and King (1999) Band-Pass filter.
Although the measures of synchronization used by Kalemli-Ozcan et al. (2011) are state of the art, simpler, not subject to the shortcoming of various filtering methods, and easy to grasp, the opportunity cost of using such methodology is too overwhelmed in terms of time since we have 217 countries. With 20 industrial countries, Kalemli-Ozcan et al. were able to investigate business cycle synchronization for 190 pairs of countries. Using similar methodology would require that we work with 23, 436 pairs of countries (the total number of combinations of size 2 taken from a set of size 217). We chose the nonparametric methodology proposed by Mink et al. (2007) as the second best alternative available to answer our research question. This measure is as flexible as Kalemli-Ozcan et al.’s in that it is easy to use, and can be calculated at every point in time in a bivariate or multivariate setting within/across sectors or countries to indicate whether cycles are synchronized or not. However, it first requires filtering to obtain the output gap and it is bounded between -1 and 1. We are not concerned about issues related to parameter heterogeneity raised by Kalemli-Ozcan et al. because our classification of the countries by income levels brings homogeneity to the groups.2 The main issue, nonetheless, remains the choice of an appropriate reference cycle against which synchronization can be assessed with individual cycles when dealing with multiple countries.3 Should we use the cycle of a leading developed economy such as the United States (US) or a weighted average of several advanced economies, or the cycle of a common factor? Facing this dilemma in their investigation of European business cycle synchronization, Camacho et al. (2006) understandably used bilateral comparison of cycles whereas Mink et al. (2007) used the median of all observed output gap. In this paper, we take a broader approach, we experiment with different criteria in choosing the reference cycle. We consider the country with the minimum average output gap over the sample period, the median average, the maximum average, the maximum median, and the median average real GDP per capita growth rates, and the US as an ad hoc country, which is known to have great influence on the world economy but was not chosen by the selection criteria.
In its simplest form, the bivariate version of the synchronicity measure between the reference cycle (gr,t) and the individual country cycle (gi,t) as proposed by Mink et al. is represented as follows:

(5)
This synchronicity measure takes the value of 1 when the reference cycle and the individual cycle have the same sign and -1 when they move in opposite directions. The percentage of time φi,r,t is 1 is a number that lies in the interval [0,1], a cut-off point of 0.50 indicates that the cycles are neither synchronous or asynchronous. We regroup all the countries with φi,r,t =1 more than 50 percent of the time to form the pool of countries with synchronized cycles within each income per capita group. These groups are then amalgamated in the search for a world business cycle.
The multivariate version of the synchronicity measure as per Mink et al. (2007) is given by:
(6)

Where N is the number of countries in a given pool of income per capita group. This equation tells us at each point in time the average synchronicity of the individual countries’ cycles with the reference cycle. Positive values of φt indicate the dominance of synchronous cycles over asynchronous cycles in relation to the reference country for a given year. We use both version of Mink et al. synchronicity measure in our investigation of the linkage between business cycles and income levels.



Download 124.98 Kb.

Share with your friends:
  1   2   3   4   5   6   7   8   9   ...   12




The database is protected by copyright ©sckool.org 2020
send message

    Main page