Standard errors in parentheses are based on 1000 bootstrap replications. * denotes statistical significance at 5% level and ** denotes significance at 10% level.
1 The findings of Gregory et al. (1997), are based on Kalman filtering and dynamic factor analysis, Clark and Shin (2000) on VAR factor model, Gerlach (1988) on spectral methods, Stockman (1988) on error correction method, and Norrbin and Schlagenhauf (1996) on dynamic factor model. Most of these techniques are not practical for our study since we have a large number of variables. For example, a VAR or VECM with 217 countries would be very cumbersome, if possible at all.
2 It is worth highlighting that we do use the methodology of Kalemli-Ozcan et al. in another paper that we are currently writing where the country of references are chosen based on the degree of trade openness as opposed to output gap. This shortcut can be seen as a combination of Mink et al. and Kalemli-Ozcan et al. methodologies where the output gap is replaced with growth rate of output but the reference cycle is still there. It would be interesting to see whether we arrive at results similar to those of this paper.
3 As Basher (2010) points out, such dilemma does not necessarily exists in sectoral analysis since the choice of the reference cycle is at times pretty straightforward. The methodology of our paper is similar to Basher’s but the scope of our work is by far wider. Basher’s investigation into the decoupling of the oil sector from the non-oil sector covers only three of the six Gulf Cooperation Council (GCC) countries (Kuwait, Qatar, and Saudi Arabia) whereas our paper covers the 217 countries of the National Accounts Main Aggregates of the United Nations Statistical Databases.
4 The use of the term is convenient; it is not intended to imply that all economies in the group are experiencing similar development or that other economies have reached a preferred or final stage of development. Classification by income does not necessarily reflect development status.
5 It is important to note that our pairing differs from that of Kalemli-Ozcan et al. (2010) and others. The reference cycle is the same for all individual countries whereas in Kalemli-Ozcan et al. there is no reference cycle since both countries forming the pair can vary within the pool.
7 The grouping of countries based on the Median reference cycle is not incorporated in Table 5 because this reference is considered as a fictitious country. The country selected by the minimum average gap criterion and the US as an ad hoc are used as references for the Median pool.
8 It would be quite counterintuitive here to argue that the lesser synchronicity of the cycles for the 1990s onward is a good thing since it enables countries to buy insurance against bad times as supported by the risk sharing literature. The reason is simply that less developed countries have little resources to allocate to capital markets.
9 We think an anonymous participant at the International Macro/Macroeconomie Internationale Session of the 44th Annual Conference of the Canadian Economics Association May 28-30, 2010 for making these valuable suggestions, which definitely brought more robustness to our findings.
10 The linkage between consumption (C) and output (Y) can be understood in the following terms: increase in domestic interest rate above foreign interest rate leads to a decrease in domestic current consumption as individuals shift a larger portion of their income to savings, thereby decreasing domestic output. In foreign countries, output will also fall as households increase their purchase of domestic bonds due to the higher interest rate. Hence, the synchronization of the business cycle observed [page 160 macroeconomics: a modern approach by R. Barro 1997, and Baxter and Crucini 1993].
11 Results for the sub-sample periods showed that lagged real oil prices plays a consistent significant role in explaining synchronicity. These results are available on request.
12For brevity, the full results for both the full and subsamples estimations are not presented here but are available on request
13 For brevity, results for the entire samples and the subsamples are not fully presented in this paper but can be furnished upon request.