A. Database

Our database includes 106 countries partitioned into seven regions: North America (NA), Europe, Oceania, Asia, Latin America and the Caribbean (LAC), Middle East and North Africa (MENA), and Sub-Saharan Africa (SSA). The groupings of countries by region are especially useful to identify a “regional factor” given countries that are geographically close to each other are likely to be affected by similar types of (region-specific) shocks.

The primary source of our data series is the World Bank’s World Development Indicators. The database comprises annual data over the period 1960–2010. Real GDP, real private consumption, and real fixed investment correspond to the measures of national output, consumption, and investment, respectively. Using multiple macroeconomic aggregates, rather than just output, allows us to derive more robust measures of global, regional and national business cycles. All variables are measured at constant national prices. We compute the growth rates and remove the mean from each series. Appendix I shows the distribution of countries across seven regions.

In order to examine how global and regional cycles have evolved over time, we divide our sample into two distinct periods—the pre-globalization period (1960–84) and the globalization period (1985–2010). In addition to having roughly equal number of observations in each sub-period, there are three reasons for this demarcation. First, global trade and financial flows have increased markedly since the mid-1980s.¹⁰ In addition, regional economic linkages have become much stronger during the second period as evidenced by the rapid increase in regional trade agreements as discussed earlier.

Second, after a period of stable growth during the 1960s, the first period witnessed a set of common shocks associated with large fluctuations in the price of oil in the 1970s and synchronized contractionary monetary policies in the major industrial economies in the early 1980s. This demarcation is essential for differentiating the impact of these common shocks from that of globalization on the degree of business cycle comovement. Third, the beginning of the globalization period coincides with a structural decline in the volatility of business cycles in both advanced and developing countries until the financial crisis of 2008–09.¹¹

B. Econometric Model

We employ a dynamic factor model to study the evolution of global and regional business cycles over time. Dynamic factor models have recently become a popular econometric tool for quantifying the degree of comovement among macroeconomic time series as some new estimation methods have been developed to perform factor analysis in large datasets.¹² The motivation underlying these models is that the covariance or comovement between (observable) macroeconomic time series across countries is the result of a relationship between these variables and a small number of unobservable common factors, which can be thought of as the main forces driving economic activity. These factors can capture common fluctuations across the entire dataset (i.e., the world) or across subsets of the data (i.e., regions).¹³ Dynamic factor models are particularly useful for characterizing the degree and evolution of synchronization in various dimensions without making strong identifying assumptions to disentangle different types of common shocks.

In addition, factor models have the advantage that they are motivated by DSGE models, as was first noted by Sargent (1989) for a general equilibrium model. Recent research on international business cycles shows that one can view the dynamic factor model we employ as a reduced-form solution of a standard open-economy DSGE model as the data generated from that model has a representation of a dynamic factor model (Crucini, Kose, and Otrok, 2011). The typical DSGE model suffers from the curse of dimensionality, limiting the number of shocks and number of driving variables that can be analyzed using such models. The advantages of the dynamic factor model employed here are that it is parsimonious and allows one to consider larger, more interesting datasets than if one was estimating a multi-country DSGE model.

Our dynamic factor model features: (i) a global factor common to all variables (and all countries) in the system; (ii) a factor common to each region; (iii) a country factor common to all variables in each country; and (iv) an idiosyncratic component for each series of each country. Since our primary interest is in comovement across all variables in all countries (or regions), we do not include separate factors for each of the macroeconomic aggregates (including factors in yet another dimension would also make the model intractable for the large number of countries we study). The dynamic relationships in the model are captured by modeling each factor and idiosyncratic component as an autoregressive process.

Specifically, let ${\text{Y}}_{\text{t}}^{\text{i,j,k}}$ denote the growth rate of the i^th observable variable in the j^th country of region k. We have three variables (output, consumption and investment) per country (indexed by i), seven regions (indexed by k), and 106 countries (indexed by j). The model can then be written as:

where ϕ^i,j,k(L) and ϕ^m(L) are lag polynomial operators, ${\text{v}}_{\text{t}}^{\text{i,j,k}}$ are distributed $\text{N}(0,{\sigma }_{\text{i,j,k}}^2),{\mathrm{\mu }}_{\text{t}}^{\text{m}}$ are distributed $\text{N}(0,{\sigma }_{\text{m}}^2),$ and the innovation terms ${\mathrm{\mu }}_{\text{t}}^{\text{m}}$ and ${\text{v}}_{\text{t}}^{\text{i,j,k}}$ are mutually orthogonal across all equations and variables in the system. The β parameters are called factor loadings and capture the sensitivity of each observable variable to the latent factors (denoted by ${\text{f}}_{\text{t}}^{\text{m}}$; we have a total of 114 factors (1 global factor + 7 regional factors + 106 country factors) indexed by m). For each variable, the estimated factor loadings quantify the extent to which that variable moves with the global factor, the regional factor, and the country-specific factor, respectively. The lag polynomials can in principle be of different order; however, for simplicity and parsimony, we restrict them to be AR(3) for each factor and idiosyncratic term. Since we are using annual data, this should capture most spillovers, either contemporaneous or lagged, across variables and countries.

The model is estimated using Bayesian techniques as described in Kose, Otrok, and Whiteman (2003) and Otrok and Whiteman (1998). There are two related identification problems in the estimation of the model given by equations (1)–(3): neither the signs nor the scales of the factors and the factor loadings are separately identified. We identify the signs by requiring one of the factor loadings to be positive for each of the factors. In particular, we impose the conditions that the factor loading for the global factor is positive for U.S. output; that country factors have positive factor loadings for the output of each country; and factors for each region have positive loadings for the output of the first country listed in each region in Appendix I.¹⁴ Following Sargent and Sims (1977) and Stock and Watson (1989), we identify the scales by assuming that each ${\sigma }_{\text{m}}^2$ equals a constant. The constant is chosen based on the scale of the data, so that the innovation variance is equal to the average innovation variance of a set of univariate autoregressions on each time series. The results are not sensitive to this normalization.¹⁵

There is a rich literature on large dynamic factor models that is closely related to our work (e.g., Forni, Hallin, Lippi, and Reichlin, 2000; Stock and Watson, 2002; Doz, Giannone, and Reichlin, 2008). Our approach is related to this literature, but a virtue of our methodology is that we can easily employ a factor structure where there are simultaneously estimated factors using large and small cross-sections of the data without relying on asymptotics. In our parametric approach, we can estimate factors for small cross-sections of the data (e.g., individual countries) conditional on larger cross-sections (i.e., global and regional factors).

C. Variance Decompositions

To measure the importance of each factor for explaining the volatility of macroeconomic aggregates, variance decompositions are calculated. The formula for the variance decomposition is derived by applying the variance operator to each equation in the system. This provides an empirical assessment of how much of a country’s business cycle fluctuations are associated with global fluctuations or fluctuations among countries in each region. We estimate the share of the variance of each macroeconomic variable attributable to each of the three factors and the idiosyncratic component. With orthogonal factors, the variance of the growth rate of the observable quantity ${\text{Y}}_{\text{t}}^{\text{i,j,k}}$ can be written as follows:

Then, the fraction of volatility due to, say, the regional factor k would be:

These measures are calculated at each pass of the Markov chain; dispersion in their posterior distributions reflects uncertainty regarding their magnitudes.

A. Global and Regional Cycles

How important is the global business cycle? As Table 1 shows, the global factor explains a significant fraction of business cycle fluctuations in all three macroeconomic variables over the period 1960–2010, implying that there is indeed a sizeable world business cycle. The global factor on average accounts for 10 percent of output growth variation among all countries in the sample. It also explains roughly 9 percent and 5 percent of the volatility of growth rates of consumption and investment, respectively. Although the variance shares attributed to the global factor appear to be small at first sight, it is important to remember that the global (common) factor across the three macroeconomic aggregates is for a very large and diverse set of countries.

Table 1.

Variance Decompositions (1960–2010)

Notes: The model is estimated over the full sample period (1960–2010) and the variance decompositions are computed for each country, for output, consumption and investment. The cross-sectional mean of the variance share (in percent) attributable to the relevant factor is reported in each cell. The cross-sectional means are calculated for the world and relevant region indicated in the first column. The column marked “Global+Regional” denotes the sums of the average variance shares of the global and regional factors.

Table 1.

Variance Decompositions (1960–2010)

			Factors
Variables / Regions		Global	Regional	Global + Regional	Country	Idiosyncratic
Output
	World	10.40	12.86	23.26	40.79	34.46
	North America	16.64	33.59	50.23	36.79	12.42
	Europe	26.85	21.62	48.46	33.07	18.08
	Ocenia	7.45	3.60	11.04	60.49	25.90
	Asia	8.92	21.36	30.28	46.74	22.32
	Latin America and the Caribbean	8.00	17.42	25.42	36.85	36.95
	Sub-Saharan Africa	4.58	2.52	7.10	41.40	48.95
	Middle East and North Africa	5.99	5.95	11.94	52.58	32.61
Consumption
	World	8.52	7.44	15.96	34.27	48.43
	North America	17.87	27.41	45.29	35.85	17.83
	Europe	26.31	7.23	33.54	31.39	34.41
	Ocenia	4.53	6.11	10.64	35.29	50.22
	Asia	8.12	11.89	20.01	31.85	47.51
	Latin America and the Caribbean	4.72	11.00	15.72	34.98	48.61
	Sub-Saharan Africa	2.81	2.83	5.64	35.31	57.06
	Middle East and North Africa	1.30	4.04	5.35	38.12	54.07
Investment
	World	4.95	11.00	15.95	27.06	55.73
	North America	5.47	34.58	40.05	44.81	13.80
	Europe	12.72	22.73	35.45	34.98	29.18
	Ocenia	0.18	8.23	8.40	64.88	23.81
	Asia	5.35	10.91	16.27	35.05	48.19
	Latin America and the Caribbean	3.16	14.48	17.64	28.59	52.99
	Sub-Saharan Africa	2.90	1.96	4.86	14.40	78.89
	Middle East and North Africa	1.15	7.37	8.52	31.47	57.05

Notes: The model is estimated over the full sample period (1960–2010) and the variance decompositions are computed for each country, for output, consumption and investment. The cross-sectional mean of the variance share (in percent) attributable to the relevant factor is reported in each cell. The cross-sectional means are calculated for the world and relevant region indicated in the first column. The column marked “Global+Regional” denotes the sums of the average variance shares of the global and regional factors.

View Table

Table 1.

Variance Decompositions (1960–2010)

			Factors
Variables / Regions		Global	Regional	Global + Regional	Country	Idiosyncratic
Output
	World	10.40	12.86	23.26	40.79	34.46
	North America	16.64	33.59	50.23	36.79	12.42
	Europe	26.85	21.62	48.46	33.07	18.08
	Ocenia	7.45	3.60	11.04	60.49	25.90
	Asia	8.92	21.36	30.28	46.74	22.32
	Latin America and the Caribbean	8.00	17.42	25.42	36.85	36.95
	Sub-Saharan Africa	4.58	2.52	7.10	41.40	48.95
	Middle East and North Africa	5.99	5.95	11.94	52.58	32.61
Consumption
	World	8.52	7.44	15.96	34.27	48.43
	North America	17.87	27.41	45.29	35.85	17.83
	Europe	26.31	7.23	33.54	31.39	34.41
	Ocenia	4.53	6.11	10.64	35.29	50.22
	Asia	8.12	11.89	20.01	31.85	47.51
	Latin America and the Caribbean	4.72	11.00	15.72	34.98	48.61
	Sub-Saharan Africa	2.81	2.83	5.64	35.31	57.06
	Middle East and North Africa	1.30	4.04	5.35	38.12	54.07
Investment
	World	4.95	11.00	15.95	27.06	55.73
	North America	5.47	34.58	40.05	44.81	13.80
	Europe	12.72	22.73	35.45	34.98	29.18
	Ocenia	0.18	8.23	8.40	64.88	23.81
	Asia	5.35	10.91	16.27	35.05	48.19
	Latin America and the Caribbean	3.16	14.48	17.64	28.59	52.99
	Sub-Saharan Africa	2.90	1.96	4.86	14.40	78.89
	Middle East and North Africa	1.15	7.37	8.52	31.47	57.05

Notes: The model is estimated over the full sample period (1960–2010) and the variance decompositions are computed for each country, for output, consumption and investment. The cross-sectional mean of the variance share (in percent) attributable to the relevant factor is reported in each cell. The cross-sectional means are calculated for the world and relevant region indicated in the first column. The column marked “Global+Regional” denotes the sums of the average variance shares of the global and regional factors.

The global factor tends to be important in each region, but on average it plays a more dominant role in explaining business cycles in advanced country regions (North America and Europe) with stronger intra-regional trade and financial linkages than those regions with emerging market and developing economies (Asia, LAC, SSA, and MENA). The global factor on average explains about 17 (27) percent of output fluctuations in North America (Europe). In contrast, in the latter group of regions, the global factor explains between 5 percent (in SSA) to 9 percent (in Asia) of output fluctuations. The global factor is also associated with a substantial share of the variance in consumption and investment growth among regions of advanced countries, accounting on average for about 25 percent and 12 percent of the total variance of these variables, respectively. These shares are also much larger than the corresponding numbers for other regions.

How important are regional cycles? The results indicate that the regional factor on average plays a slightly more important role than the global factor in explaining output fluctuations. Table 1 shows that the regional factor accounts for about 13 percent of output growth fluctuations in the full sample. The fraction of output variance explained by the regional factor is larger than that of the global factor in three of the seven regions. The regional factor, like the global factor, tends to be more important for advanced country regions. On average, it accounts for about 34 percent and 22 percent of output growth fluctuations in North America and Europe, respectively. While the regional factor explains a relatively smaller share of variance of output for Oceania, SSA and MENA, it explains a significant part of variation for Asia and LAC.

The regional factor also appears to be a major driver of business cycles in consumption and investment. While it accounts for only a slightly small share of consumption variation than does the global factor, its contribution to investment volatility is two times larger than that of the global factor. In six out of the seven regions, the regional factor is more important than the global one in explaining the volatility of consumption and investment.

We also examine the overall synchronization of global business cycles by using the sum of the variance contributions of the global and regional factors. The rankings of different regions remain much the same, although the magnitudes are of course larger. Among North America and Europe, the total contribution of these two factors averages between 48 and 50 percent for output, 34 and 45 percent for consumption, and 35 and 40 percent for investment. For other regions, the average for output variance is between 7 percent (for SSA) and 30 percent (for Asia).

Our results also indicate that the factor loadings associated with output and consumption growth on the global factor are positive for 70 percent and 83 percent of countries, respectively (i.e., the posterior distributions of the factor loadings have very little mass in symmetric intervals about zero). These findings imply that positive developments in the U.S. economy are generally associated with higher growth in the rest of the world since the global factor is identified by a positive factor loading for U.S. output growth.

B. Country-Specific Cycles

We next turn our attention to the roles played by country and idiosyncratic factors in explaining business cycles. Both of these play important roles in driving business cycles around the world (Table 1). The country factor is on average more important in explaining output variation than is the idiosyncratic factor (41 percent vs. 34 percent), but the reverse is true for fluctuations in consumption (34 percent vs. 48 percent) and investment (27 percent vs. 56 percent). Given that consumption and investment series tend to have larger measurement problems than those of output, it is normal that fluctuations in consumption and investment exhibit larger idiosyncratic components.

With respect to different regions, the results indicate that as countries get more exposed to global trade and financial flows and establish stronger intra-regional economic linkages, the global and region-specific factors increasingly become more relevant in explaining national business cycles at the expense of the country and idiosyncratic factors. For example, country and idiosyncratic factors together explain about 50 percent of output variation in North America and Europe, but their share is on average more than 80 percent in five other regions.

How different are the factors explaining fluctuations in output and consumption? The total contribution of the global and regional factors is greater for output than for consumption for the sample as a whole and also for each region. Indeed, Figure 1 shows that this is true even when we look at the global and regional factors by themselves. This implies that, on average, country-specific and idiosyncratic factors play a more important role in explaining consumption fluctuations than in the case for output fluctuations. This result is very similar to a well-known stylized fact in the literature that, contrary to the predictions of conventional theoretical models of international business cycles, output is more highly correlated across countries than consumption (Backus, Kehoe, and Kydland (1995) refer to this as the “quantity anomaly”).

Average Variance Explained by Global and Regional Factors (1960–2010)

Citation: IMF Working Papers 2013, 019; 10.5089/9781557753281.001.A001

Notes: We estimate the model over the full sample period (1960–2010) and compute the variance decompositions for each country, separately for output, consumption and investment. Each bar then represents the cross-sectional mean of the variance share attributable to the relevant factor for that particular variable. The cross-sectional means are calculated over the full sample of countries. The bar marked “Global+Regional” represents the sum of the average variance shares of the global and regional factors for each variable.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Average Variance Explained by Global and Regional Factors (1960–2010)

Citation: IMF Working Papers 2013, 019; 10.5089/9781557753281.001.A001

Notes: We estimate the model over the full sample period (1960–2010) and compute the variance decompositions for each country, separately for output, consumption and investment. Each bar then represents the cross-sectional mean of the variance share attributable to the relevant factor for that particular variable. The cross-sectional means are calculated over the full sample of countries. The bar marked “Global+Regional” represents the sum of the average variance shares of the global and regional factors for each variable.

• Download Figure
• Download figure as PowerPoint slide

Average Variance Explained by Global and Regional Factors (1960–2010)

Citation: IMF Working Papers 2013, 019; 10.5089/9781557753281.001.A001

Notes: We estimate the model over the full sample period (1960–2010) and compute the variance decompositions for each country, separately for output, consumption and investment. Each bar then represents the cross-sectional mean of the variance share attributable to the relevant factor for that particular variable. The cross-sectional means are calculated over the full sample of countries. The bar marked “Global+Regional” represents the sum of the average variance shares of the global and regional factors for each variable.

• Download Figure
• Download figure as PowerPoint slide

As one would expect, the contribution of the idiosyncratic factor in regions populated with relatively less developed countries is greater than that of any other factor. This is true for all variables, but especially so for investment, where on average the idiosyncratic factor accounts for 79 percent of fluctuations in SSA and 57 percent of fluctuations in MENA. This result implies that investment fluctuations in these regions do not appear to be influenced much by either national or global business cycles.

C. Global, Regional, and Country Factors

We also study the temporal evolution of the global, regional and country factors. These factors are able to capture the main cyclical episodes of the past fifty years. For example, the estimated global factor closely matches the major peaks and troughs observed in the global business cycle, including the recessions in 1974–75, the early 1980s, the early 1990s, the slowdown in the early 2000s, and the recent global recession.¹⁷ Moreover, there is considerable overlap in the evolution of the global factor and U.S. growth, especially during U.S. recessions.

Regional factors also reflect some of the major region specific episodes. For example, the Asian regional factor captures the Asian financial crisis, the LAC factor is consistent with the debt crisis of the early 1980s, and the North American and European factors exhibit the sharp contraction associated with the global financial crisis of 2008–09. We also examine relationships among the global factor, regional factors, country factors, and output in some select countries. These factors play different roles at different points in time and around the globe. In some episodes, the country factor is more strongly reflective of domestic economic activity, while in others the domestic growth appears to be influenced by movements in the global and/or regional factors.

Do global business cycles exhibit more persistent behavior than do the regional and national cycles? In order to measure persistence of cycles, we calculate the first-order autocorrelation of each factor. The global factor has large and positive autocorrelation (0.60), and compared to the autocorrelations of the regional factors and most of the country factors, it is much more persistent. The average persistence is about 0.20 for regional factors while it is around 0.10 for country factors. These results suggest that the global factor accounts for most of the persistent (or low frequency) comovement across countries whereas the global and regional factors appear to be responsible for the higher frequency comovement.¹⁸

A. Evolution of Global and Regional Cycles

Tables 2–3 show the variance decompositions based on models estimated separately for the pre-globalization (1960–84) and globalization (1985–2010) periods, respectively. A number of commentators argue that because of the forces of globalization, national business cycles should have become more synchronized over time. Contrary to these popular arguments, the average contribution of the global factor to output fluctuations registers a sizeable decline in the second sub-period, from 13 percent to 9 percent for the full sample of countries. The same pattern holds for consumption fluctuations while the importance of the global factor registers a slight increase in fluctuations in investment (Figure 2).

Table 2.

Variance Decompositions (1960–84)

Notes: The model is estimated over the 1960–84 period and the variance decompositions are computed for each country, for output, consumption and investment. The cross-sectional mean of the variance share (in percent) attributable to the relevant factor is reported in each cell. The cross-sectional means are calculated for the world and relevant region indicated in the first column. The column marked “Global+Regional” denotes the sums of the average variance shares of the global and regional factors.

Table 2.

Variance Decompositions (1960–84)

		Factors
Variables / Regions		Global	Regional	Global + Regional	Country	Idiosyncratic
Output
	World	13.40	10.68	24.08	41.00	32.56
	North America	22.46	24.28	46.74	39.02	12.62
	Europe	27.72	18.28	46.00	30.46	22.13
	Ocenia	11.82	4.37	16.19	52.93	27.30
	Asia	11.45	7.71	19.16	51.41	27.05
	Latin America and the Caribbean	14.53	16.41	30.94	37.55	29.81
	Sub-Saharan Africa	6.68	4.61	11.29	41.65	44.19
	Middle East and North Africa	8.05	6.99	15.04	50.78	30.22
Consumption
	World	10.57	8.24	18.81	36.95	41.78
	North America	27.69	16.10	43.79	38.99	15.40
	Europe	25.15	9.09	34.24	30.16	33.87
	Ocenia	4.92	5.29	10.21	34.45	50.31
	Asia	9.84	7.13	16.96	38.98	41.35
	Latin America and the Caribbean	10.23	11.18	21.41	35.96	41.00
	Sub-Saharan Africa	4.54	6.38	10.92	40.20	45.97
	Middle East and North Africa	1.21	6.58	7.79	36.77	51.92
Investment
	World	6.12	9.55	15.67	27.99	54.00
	North America	7.41	32.15	39.55	46.33	10.95
	Europe	11.55	18.18	29.73	35.64	32.98
	Ocenia	1.27	6.23	7.50	62.99	23.33
	Asia	6.23	7.60	13.82	34.12	49.42
	Latin America and the Caribbean	5.65	11.39	17.04	28.76	52.55
	Sub-Saharan Africa	4.58	3.16	7.73	16.06	73.86
	Middle East and North Africa	2.18	9.60	11.79	35.71	48.53

Notes: The model is estimated over the 1960–84 period and the variance decompositions are computed for each country, for output, consumption and investment. The cross-sectional mean of the variance share (in percent) attributable to the relevant factor is reported in each cell. The cross-sectional means are calculated for the world and relevant region indicated in the first column. The column marked “Global+Regional” denotes the sums of the average variance shares of the global and regional factors.

View Table

Table 2.

Variance Decompositions (1960–84)

		Factors
Variables / Regions		Global	Regional	Global + Regional	Country	Idiosyncratic
Output
	World	13.40	10.68	24.08	41.00	32.56
	North America	22.46	24.28	46.74	39.02	12.62
	Europe	27.72	18.28	46.00	30.46	22.13
	Ocenia	11.82	4.37	16.19	52.93	27.30
	Asia	11.45	7.71	19.16	51.41	27.05
	Latin America and the Caribbean	14.53	16.41	30.94	37.55	29.81
	Sub-Saharan Africa	6.68	4.61	11.29	41.65	44.19
	Middle East and North Africa	8.05	6.99	15.04	50.78	30.22
Consumption
	World	10.57	8.24	18.81	36.95	41.78
	North America	27.69	16.10	43.79	38.99	15.40
	Europe	25.15	9.09	34.24	30.16	33.87
	Ocenia	4.92	5.29	10.21	34.45	50.31
	Asia	9.84	7.13	16.96	38.98	41.35
	Latin America and the Caribbean	10.23	11.18	21.41	35.96	41.00
	Sub-Saharan Africa	4.54	6.38	10.92	40.20	45.97
	Middle East and North Africa	1.21	6.58	7.79	36.77	51.92
Investment
	World	6.12	9.55	15.67	27.99	54.00
	North America	7.41	32.15	39.55	46.33	10.95
	Europe	11.55	18.18	29.73	35.64	32.98
	Ocenia	1.27	6.23	7.50	62.99	23.33
	Asia	6.23	7.60	13.82	34.12	49.42
	Latin America and the Caribbean	5.65	11.39	17.04	28.76	52.55
	Sub-Saharan Africa	4.58	3.16	7.73	16.06	73.86
	Middle East and North Africa	2.18	9.60	11.79	35.71	48.53

Notes: The model is estimated over the 1960–84 period and the variance decompositions are computed for each country, for output, consumption and investment. The cross-sectional mean of the variance share (in percent) attributable to the relevant factor is reported in each cell. The cross-sectional means are calculated for the world and relevant region indicated in the first column. The column marked “Global+Regional” denotes the sums of the average variance shares of the global and regional factors.

Table 3.

Variance Decompositions (1985–2010)

Notes: The model is estimated over the 1985–2010 period and the variance decompositions are computed for each country, for output, consumption and investment. The cross-sectional mean of the variance share (in percent) attributable to the relevant factor is reported in each cell. The cross-sectional means are calculated for the world and relevant region indicated in the first column. The column marked “Global+Regional” denotes the sums of the average variance shares of the global and regional factors.

Table 3.

Variance Decompositions (1985–2010)

		Factors
Variables / Regions		Global	Regional	Global + Regional	Country	Idiosyncratic
Output
	World	9.16	19.46	28.63	34.61	33.53
	North America	30.41	32.87	63.28	23.03	11.09
	Europe	19.85	40.25	60.10	19.69	19.06
	Ocenia	20.93	20.33	41.26	29.40	25.76
	Asia	5.08	32.49	37.56	33.88	26.14
	Latin America and the Caribbean	6.55	14.55	21.10	41.19	34.67
	Sub-Saharan Africa	5.70	7.64	13.34	36.56	45.82
	Middle East and North Africa	3.72	8.61	12.33	49.93	32.17
Consumption
	World	6.49	13.85	20.34	30.38	46.31
	North America	31.06	22.01	53.07	26.55	17.95
	Europe	14.16	29.50	43.66	18.74	35.55
	Ocenia	14.75	11.30	26.05	33.46	35.44
	Asia	2.21	24.18	26.39	27.84	43.74
	Latin America and the Caribbean	3.40	10.27	13.67	34.61	48.85
	Sub-Saharan Africa	4.49	4.38	8.86	33.28	54.56
	Middle East and North Africa	2.80	8.51	11.30	38.36	44.92
Investment
	World	8.58	15.72	24.30	29.12	43.79
	North America	24.96	30.17	55.12	33.77	9.77
	Europe	19.57	31.40	50.97	26.01	22.22
	Ocenia	16.36	29.21	45.57	34.59	16.02
	Asia	5.11	26.29	31.40	32.19	34.09
	Latin America and the Caribbean	5.89	12.12	18.01	37.06	42.25
	Sub-Saharan Africa	4.46	5.43	9.89	22.42	64.08
	Middle East and North Africa	7.31	7.36	14.67	36.74	43.31

Notes: The model is estimated over the 1985–2010 period and the variance decompositions are computed for each country, for output, consumption and investment. The cross-sectional mean of the variance share (in percent) attributable to the relevant factor is reported in each cell. The cross-sectional means are calculated for the world and relevant region indicated in the first column. The column marked “Global+Regional” denotes the sums of the average variance shares of the global and regional factors.

View Table

Table 3.

Variance Decompositions (1985–2010)

		Factors
Variables / Regions		Global	Regional	Global + Regional	Country	Idiosyncratic
Output
	World	9.16	19.46	28.63	34.61	33.53
	North America	30.41	32.87	63.28	23.03	11.09
	Europe	19.85	40.25	60.10	19.69	19.06
	Ocenia	20.93	20.33	41.26	29.40	25.76
	Asia	5.08	32.49	37.56	33.88	26.14
	Latin America and the Caribbean	6.55	14.55	21.10	41.19	34.67
	Sub-Saharan Africa	5.70	7.64	13.34	36.56	45.82
	Middle East and North Africa	3.72	8.61	12.33	49.93	32.17
Consumption
	World	6.49	13.85	20.34	30.38	46.31
	North America	31.06	22.01	53.07	26.55	17.95
	Europe	14.16	29.50	43.66	18.74	35.55
	Ocenia	14.75	11.30	26.05	33.46	35.44
	Asia	2.21	24.18	26.39	27.84	43.74
	Latin America and the Caribbean	3.40	10.27	13.67	34.61	48.85
	Sub-Saharan Africa	4.49	4.38	8.86	33.28	54.56
	Middle East and North Africa	2.80	8.51	11.30	38.36	44.92
Investment
	World	8.58	15.72	24.30	29.12	43.79
	North America	24.96	30.17	55.12	33.77	9.77
	Europe	19.57	31.40	50.97	26.01	22.22
	Ocenia	16.36	29.21	45.57	34.59	16.02
	Asia	5.11	26.29	31.40	32.19	34.09
	Latin America and the Caribbean	5.89	12.12	18.01	37.06	42.25
	Sub-Saharan Africa	4.46	5.43	9.89	22.42	64.08
	Middle East and North Africa	7.31	7.36	14.67	36.74	43.31

Notes: The model is estimated over the 1985–2010 period and the variance decompositions are computed for each country, for output, consumption and investment. The cross-sectional mean of the variance share (in percent) attributable to the relevant factor is reported in each cell. The cross-sectional means are calculated for the world and relevant region indicated in the first column. The column marked “Global+Regional” denotes the sums of the average variance shares of the global and regional factors.

Variance Explained by Global Factor

Citation: IMF Working Papers 2013, 019; 10.5089/9781557753281.001.A001

Notes: We estimate the model separately over the two periods, 1960–84 (yellow) and 1985–2010 (blue). We then compute the variance decompositions for each country for output, consumption and investment. Each bar then represents the cross-sectional mean of the variance share attributable to the regional factor for that particular variable in a given period. The cross-sectional means are calculated for the relevant regions.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Variance Explained by Global Factor

Citation: IMF Working Papers 2013, 019; 10.5089/9781557753281.001.A001

Notes: We estimate the model separately over the two periods, 1960–84 (yellow) and 1985–2010 (blue). We then compute the variance decompositions for each country for output, consumption and investment. Each bar then represents the cross-sectional mean of the variance share attributable to the regional factor for that particular variable in a given period. The cross-sectional means are calculated for the relevant regions.

• Download Figure
• Download figure as PowerPoint slide

Variance Explained by Global Factor

Citation: IMF Working Papers 2013, 019; 10.5089/9781557753281.001.A001

Notes: We estimate the model separately over the two periods, 1960–84 (yellow) and 1985–2010 (blue). We then compute the variance decompositions for each country for output, consumption and investment. Each bar then represents the cross-sectional mean of the variance share attributable to the regional factor for that particular variable in a given period. The cross-sectional means are calculated for the relevant regions.

• Download Figure
• Download figure as PowerPoint slide

These patterns also hold up and are in fact stronger in most cases when we evaluate the variance decompositions of output fluctuations for different regions (Figure 2). The global factor appears to play a smaller role in explaining business cycles in the second sub-period in all regions except North America and Oceania. The share of the variance of output due to the global factor increases from 22 (12) percent to 30 (21) percent in North America (Oceania) in the second period.

In contrast, the regional factor has on average played an increasingly more dominant role in explaining business cycles over time (Tables 2–3 and Figure 3). For example, the average share of the variance of output attributed to the regional factor has risen from 11 percent to 19 percent during the second period. These patterns are particularly strong among the following regions: North America, Europe, Oceania, and Asia regions. In particular, in the second sub-period, the regional factor has accounted for roughly one-third of output fluctuations in North America and Asia, and 40 percent in Europe, 20 percent in Oceania. In a similar fashion, the regional factor for SSA and MENA has played a more important role in the second period, but the increase in the variance share of output attributed to the regional factor is much smaller in these regions. In LAC, the importance of the regional factor does not change much over the two sub-periods.

Variance Explained by Regional Factors

Citation: IMF Working Papers 2013, 019; 10.5089/9781557753281.001.A001

Notes: We estimate the model separately over the two periods, 1960–84 (yellow) and 1985–2010 (blue). We then compute the variance decompositions for each country for output, consumption and investment. Each bar then represents the cross-sectional mean of the variance share attributable to the regional factors for that particular variable in a given period. The cross-sectional means are calculated for the relevant regions.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Variance Explained by Regional Factors

Citation: IMF Working Papers 2013, 019; 10.5089/9781557753281.001.A001

Notes: We estimate the model separately over the two periods, 1960–84 (yellow) and 1985–2010 (blue). We then compute the variance decompositions for each country for output, consumption and investment. Each bar then represents the cross-sectional mean of the variance share attributable to the regional factors for that particular variable in a given period. The cross-sectional means are calculated for the relevant regions.

• Download Figure
• Download figure as PowerPoint slide

Variance Explained by Regional Factors

Citation: IMF Working Papers 2013, 019; 10.5089/9781557753281.001.A001

Notes: We estimate the model separately over the two periods, 1960–84 (yellow) and 1985–2010 (blue). We then compute the variance decompositions for each country for output, consumption and investment. Each bar then represents the cross-sectional mean of the variance share attributable to the regional factors for that particular variable in a given period. The cross-sectional means are calculated for the relevant regions.

• Download Figure
• Download figure as PowerPoint slide

How can we explain these results? It is useful to answer this question in two steps. First, the decline in the importance of the global factor supports the interpretation that the strong business cycle synchronization observed during the 1970s and early 1980s reflected large common disturbances—the two oil price shocks—and the effects of correlated disturbances in the major advanced countries, notably the disinflationary monetary policy stance in the early 1980s. Although the latest financial crisis can also be seen as a massive global shock, its full impact on our variance decompositions has probably yet to be fully realized (as we have only three years of observations associated with the crisis).¹⁹ Moreover, as we indicated in our introduction, there was substantial variation in the performance of different regions during the latest crisis. These developments have collectively led to an overall decline in the importance of the global factor in explaining business cycles during the globalization period.

Second, there has been on average an increase in the importance of regional factors. In particular, regional linkages have become much more significant in regions where intra-regional trade and financial flows have increased substantially after the mid-1980s (North America, Europe, Oceania, and Asia). These regions have undertaken substantial steps to strengthen regional economic linkages during the second sub-period. For example, intra-regional trade and financial linkages have registered significant growth over the past quarter century in the North American region where the process of economic integration started in the mid-1980s and culminated with the ratification of the NAFTA in 1994.²⁰ During the past decade, intra-regional trade flows accounted for nearly 55 percent of total trade while intra-regional financial assets were about 20 percent of total assets in the North American region.²¹

One of the greatest regional integration projects of history took place in Europe with the eventual establishment of the European Union and creation of euro area. Intra-regional trade flows have constituted roughly 75 percent of total trade in Europe during the past decade. Intra-regional asset holdings went up from 55 percent to roughly 75 percent of total assets in Europe over the same period.

Regional integration in Asia has been mainly driven by the ASEAN (Association of South East Asian Nations) and complemented by a number of bilateral regional arrangements (Plummer, 2006). The region has seen a rapid increase in intra-regional trade and financial flows especially over the past decade. The share of intra-regional trade flows was around 55 percent over the past decade while intra-regional asset holdings increased to more than 5 percent of total assets. In the case of Oceania, the Australia-New Zealand Closer Economic Relations Agreement in 1983 marked the beginning of a period of rapid growth in intra-regional trade and financial linkages (Gonuguntla, 2010).²²

The nature of trade has also changed in these four regions over time. One of the major driving forces of the rapid growth in regional trade flows has been the acceleration of intra-industry trade. The surge in intra-industry trade linkages has been associated with vertical specialization of trade (Yi, 2003; De Backer and Yamano, 2007) which is associated with a higher degree of business cycle synchronization (di Giovanni and Levchenko, 2010; Ng, 2010). Moreover, during the second sub-period, countries in these regions have increased the pace of diversification of their industrial (and trade) bases. This has been accompanied by a greater degree of sectoral similarity across countries within regions further contributing to the increase in the degree of business cycle synchronization (Imbs, 2004; Kose and Prasad, 2010).

The evolution of global and regional cycles is of course affected by various types of shocks, in addition to increased trade and financial integration. For example, common shocks, such as abrupt, unexpected changes in commodity prices or sharp fluctuations in asset prices in the major financial centers, can lead to a higher degree of synchronization of global or regional business cycles. Both the global and regional cycles can also be affected by cross-border spillovers of disturbances originating in a large economy. Moreover, global and regional business cycles can emerge simply because of correlated shocks, such as shocks associated with the implementation of similar policies. It is easy to see how these types of shocks have been influential in some of the regions that have experienced more pronounced regional cycles. For example, while the implementation of similar policies has contributed greatly to the convergence of national cycles in Europe over the period 1985–2010, cross-border spillovers originating in the United States and China have probably been important in explaining regional cycles in North America and Asia, respectively.

Not surprisingly, the importance of regional factors in explaining business cycles in less developed regions, such as SSA and MENA, which are also the regions relatively less exposed to the forces of globalization, has registered only a slight increase between the two periods. In LAC, the importance of the regional factor does not change much over the two sub-periods. This could reflect the region specific developments, such as the buildup in external debt, the subsequent debt crises in the early 1980s, and regional contagion following the financial crisis of the 1990s and early 2000s.

The total contribution of global and regional factors together to output fluctuations has, on average, has recorded a small increase from 24 percent in the first sub-period to 29 percent in the second sub-period (Figure 4). This is of course the consequence of a substantial increase in the relative importance of the regional factor. Since the total contribution of global and regional factors is a measure of the extent of co-movement across national business cycles, these results show that while there is some evidence implying that business cycles have become more synchronized over time, the increase in the change of variance explained by the global and regional factors is relatively small. However, as we present in Section VI, this small increase is entirely driven by the period of highly synchronized cyclical activity due to the global financial crisis of 2008–09.

Variance Explained by Global and Regional Factors

Citation: IMF Working Papers 2013, 019; 10.5089/9781557753281.001.A001

Notes: We estimate the model separately over the two periods, 1960–84 (yellow) and 1985–2010 (blue). We then compute the variance decompositions for each country, for output, consumption and investment in each of these two periods. Each bar then represents the cross-sectional mean of the variance share attributable to the sum of the global and regional factors for that particular variable in a given period.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Variance Explained by Global and Regional Factors

Citation: IMF Working Papers 2013, 019; 10.5089/9781557753281.001.A001

Notes: We estimate the model separately over the two periods, 1960–84 (yellow) and 1985–2010 (blue). We then compute the variance decompositions for each country, for output, consumption and investment in each of these two periods. Each bar then represents the cross-sectional mean of the variance share attributable to the sum of the global and regional factors for that particular variable in a given period.

• Download Figure
• Download figure as PowerPoint slide

Variance Explained by Global and Regional Factors

Citation: IMF Working Papers 2013, 019; 10.5089/9781557753281.001.A001

Notes: We estimate the model separately over the two periods, 1960–84 (yellow) and 1985–2010 (blue). We then compute the variance decompositions for each country, for output, consumption and investment in each of these two periods. Each bar then represents the cross-sectional mean of the variance share attributable to the sum of the global and regional factors for that particular variable in a given period.

• Download Figure
• Download figure as PowerPoint slide

How do our findings compare with the results in the literature? Most of the earlier studies have typically focused on just output or industrial production fluctuations and limited their analysis to specific regions. However, as we briefly discuss in Section II, these studies often report conflicting results. For example, some recent papers document that there is a distinct European business cycle while others argue the opposite (De Haan, Inklaar, Jong-a-Pin, 2008).²³ Other studies find regional cycles specific to East Asia and North America. Some of these studies resort to simple correlations (over much shorter time-periods and smaller samples than ours here) and report that business cycles within certain regions have become more correlated over time. Most of the studies are unable to properly account for the extent of regional business cycle comovement since they mostly rely on bilateral correlations, which are fraught with a number of problems. For example, deriving summary correlation measures from large datasets requires one to take averages across the estimated correlations, a procedure that can mask the presence of comovement across a subset of the data. One way to reduce the number of bivariate correlations is to specify a country or weighted aggregate to serve as the reference against which other countries’ correlations are computed. However, changes in the reference country/aggregate often lead to significantly different results. Such weighting schemes also inevitably give rise to questions about the weights and concerns that a large county may dominate the global/regional business cycle by virtue of its size when, in fact, that country may be disengaged from other countries. Moreover, static correlations cannot capture the dynamic properties of the data, such as autocorrelations and cross-autocorrelations across variables.

Factor models obviate these problems. They do not require one to average across variables or define a “numeraire” country. Instead, they identify the common component and, at the same time, detect how each country responds to that component. More importantly, factor models are flexible enough that multiple factors can be specified in a parsimonious way to capture the extent of synchronicity across the entire dataset as well as the synchronicity specific to subsets of the data, such as regions. Furthermore, since the factors are extracted simultaneously, we can assign a degree of relative importance to each type of factor.

Our findings are consistent with the results reported by studies that employ dynamic factor models. For example, Karagedikli and Thorsrud (2012) and Mumtaz, Simonelli, and Surico (2011) also report that regional business cycles have become more pronounced over time. We focus on a much larger dataset than these studies do.²⁴ Our large dataset, which allows us to consider the synchronization of cycles in many regions, our long sample, which covers a substantial period of the recent era of globalization, and our demarcation of the pre-globalization and globalization periods are essential in enabling us to identify the emergence of regional cycles during the second sub-period.²⁵

B. Synchronization of Cycles in Consumption and Investment

We next assess the evolution of variance shares attributable to different factors in explaining consumption and investment fluctuations. For the world as a whole, the increase in the variance contribution of regional factors to consumption fluctuations is on average about 6 percentage points (from 8 percent to 14 percent), but the joint share of the global and region-specific factors has not changed much (Tables 2–3). For the North American, European, Oceanian, and Asian regions, the two common factors jointly account for a substantially higher share of consumption fluctuations in the second sub-period whereas, for SSA and LAC, the share of these factors has on average declined over time. The share of consumption variation due to the country and idiosyncratic factors is much smaller in North America, Europe and Oceania than those of others. In particular, the average variance of consumption explained by these two factors is about 54 percent in these three regions, but it is 83 percent in the other four regions.

These findings indicate that the three regions populated mostly by advanced countries have been able to use globalization relatively more effectively to share national consumption risk. In addition to their relatively more advanced levels of financial development, these regions are more exposed to global trade and financial linkages in the second sub-period. These results are confirmed by some recent studies as well (Sorensen, Wu, and Zhu, 2007; Kose, Prasad, and Terrones, 2009). On the other hand, regions with relatively less developed domestic financial markets and with less exposure to cross-border linkages are probably at earlier stages of utilizing this benefit of globalization as their consumption fluctuations are still closely tied to national cycles.

The share of investment variance attributable to the global and regional factors increases for the world and for all the regions in the second sub-period. The variance of investment growth due to the global and regional factors rises from 16 percent in the first sub-period to 24 percent in the second sub-period for the full sample. These two factors on average account for 51 percent of investment variation in North America, Europe and Oceania and 16 percent in the other four regions in the second sub-period. One interpretation of this finding is that fluctuations in investment have been becoming more synchronized over time, especially in regions with more pronounced global and intra-regional linkages.

However, these findings imply different outcomes than those offered by the predictions of standard international business cycle models. The basic theory implies that stronger trade and financial linkages translate into lower investment correlations across countries. Specifically, when restrictions on capital and current account transactions are relaxed in these models, capital and other resources rapidly move to countries with more favorable technology shocks because of the standard “resource shifting” effect in standard business cycle models (see Backus and others, 1995; Ravn and Mazzenga, 2004; and Heathcote and Perri, 2002).²⁶

The results with respect to the synchronization of consumption and investment fluctuations suggest that the data paint a different picture than do the standard business cycle models. In particular, the forces of globalization have not translated into an increase in the degree of risk sharing achieved by all of the regions in our sample as would be expected on the basis of standard models. In addition, contrary to the predictions of these models, the importance of global and regional factors for investment fluctuations has risen during the period of globalization, implying a higher degree of cross-country comovement of fluctuations in this variable.

A. Country-Specific Results

Do the averages we presented so far accurately reflect the sources of business cycle variation at the country level? In order to address this question, we examine the relative contributions of global and regional factors to each of the macroeconomic variables for each country. Figures 5–6 present the relative contributions of the global and regional factors to output fluctuations in individual countries in four regions: North America, Europe, Asia and Oceania. These regions are especially interesting cases as they experience a substantial increase in the importance of the regional factor in the second period. Moreover, the global factor becomes less influential in all of these regions except North America. We present the contributions of the global and regional factors separately for the two sub-periods. We also show the posterior coverage intervals (of length two standard deviations) around the posterior means of the estimated variance contributions. Non-overlapping posterior coverage intervals indicate statistically significant changes between the two periods.

Output Variance Explained by Global Factor North America and Europe

Citation: IMF Working Papers 2013, 019; 10.5089/9781557753281.001.A001

Notes: We estimate the model separately over the two periods, 1960–84 and 1985–2010. For each country, we then show the posterior means of the share of the variance of output growth fluctuations accounted for. We also show the corresponding posterior coverage intervals of length two standard deviations (%).

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Output Variance Explained by Global Factor North America and Europe

Citation: IMF Working Papers 2013, 019; 10.5089/9781557753281.001.A001

Notes: We estimate the model separately over the two periods, 1960–84 and 1985–2010. For each country, we then show the posterior means of the share of the variance of output growth fluctuations accounted for. We also show the corresponding posterior coverage intervals of length two standard deviations (%).

• Download Figure
• Download figure as PowerPoint slide

Output Variance Explained by Global Factor North America and Europe

Citation: IMF Working Papers 2013, 019; 10.5089/9781557753281.001.A001

Notes: We estimate the model separately over the two periods, 1960–84 and 1985–2010. For each country, we then show the posterior means of the share of the variance of output growth fluctuations accounted for. We also show the corresponding posterior coverage intervals of length two standard deviations (%).

• Download Figure
• Download figure as PowerPoint slide

Output Variance Explained by Regional Factor North America and Europe

Citation: IMF Working Papers 2013, 019; 10.5089/9781557753281.001.A001

Notes: See notes of Figure 5.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Output Variance Explained by Regional Factor North America and Europe

Citation: IMF Working Papers 2013, 019; 10.5089/9781557753281.001.A001

Notes: See notes of Figure 5.

• Download Figure
• Download figure as PowerPoint slide

Output Variance Explained by Regional Factor North America and Europe

Citation: IMF Working Papers 2013, 019; 10.5089/9781557753281.001.A001

Notes: See notes of Figure 5.

• Download Figure
• Download figure as PowerPoint slide

The variance contribution of the global factor drops from the first period to the second for 16 countries, remains unchanged for 16 others, and increases for only 7 countries, for North America, Europe, Asia and Oceania. In the case of North America, the global factor becomes more important in the United States, but it does not become significantly more influential in Canada and Mexico. When we examine the importance of the regional factor over time, we observe a different pattern: it goes up for 27 countries, stays unchanged for 8, and declines for 4 for these four regions.

These patterns are quite similar when we look at the three other regions as well, with the relative importance of the global factor going up for only 12 countries but declining for 27 of them. The relative importance of the regional factor, by contrast, rises for 23 and falls for 15 countries. There is no significant change in the share explained by the global or regional factors in other countries. For the world as a whole, the share of the global factor rises in the second sub-period in 21 countries, remains unchanged in 42 countries, and declines in 43 countries. For the world as a whole, the share of the regional factor rises in the second sub-period in 50 countries, remains unchanged in 37 countries, and declines in 19 countries. These findings indicate that our headline results with respect to the evolution of the importance of the global and regional factors over time are confirmed by the country-specific analysis.²⁷

B. Demarcation of Time and Configuration of Regions

We next analyze the robustness of our results to alternative break points in the sample and changes in configurations of regions. In Section III, we have already discussed a variety of reasons why 1985 is a logical break point to analyze the evolutionary changes in the roles played by global and regional factors in explaining business cycles. We also estimated the model for alternative sample periods using 1983 and 1987 as break points. The results for these sub-samples lead to qualitatively similar conclusions (Table 4). In addition, the individual variance decomposition patterns documented in Figures 5–6 remain essentially the same, confirming that our results do not hinge on the exact break date.²⁸

Table 4.

Variance Decompositions for the World (Different Breaks)

Notes: The model is estimated over the periods stated in the first column and the variance decompositions are computed for each country, for output, consumption and investment. The cross-sectional mean of the variance share (in percent) attributable to the relevant factor is reported in each cell. The cross-sectional means are calculated for the world. The column marked “Global+Regional” denotes the sums of the average variance shares of the global and regional factors.

Table 4.

Variance Decompositions for the World (Different Breaks)

			Factors
Variables / Regions			Global	Regional	Global + Regional	Country	Idiosyncratic
Benchmark
	1960-1984
		Output	13.40	10.68	24.08	41.00	32.56
		Consumption	10.57	8.24	18.81	36.95	41.78
		Investment	6.12	9.55	15.67	27.99	54.00
	1985-2010
		Output	9.16	19.46	28.63	34.61	33.53
		Consumption	6.49	13.85	20.34	30.38	46.31
		Investment	8.58	15.72	24.30	29.12	43.79
Break=1983
	1960-1983
		Output	13.91	9.71	23.63	41.04	32.77
		Consumption	11.29	8.19	19.48	36.38	41.36
		Investment	6.50	9.12	15.62	27.65	54.13
	1984-2010
		Output	8.19	20.45	28.64	35.67	32.98
		Consumption	5.16	15.40	20.56	30.30	46.31
		Investment	6.65	18.16	24.82	28.13	44.48
Break=1987
	1960-1987
		Output	13.72	9.07	22.79	42.22	32.80
		Consumption	10.39	6.79	17.18	37.38	43.09
		Investment	5.89	9.23	15.11	28.09	54.66
	1988-2010
		Output	13.74	18.76	32.51	32.88	32.40
		Consumption	8.30	15.96	24.26	29.47	44.21
		Investment	12.15	14.81	26.96	27.94	43.02

Notes: The model is estimated over the periods stated in the first column and the variance decompositions are computed for each country, for output, consumption and investment. The cross-sectional mean of the variance share (in percent) attributable to the relevant factor is reported in each cell. The cross-sectional means are calculated for the world. The column marked “Global+Regional” denotes the sums of the average variance shares of the global and regional factors.

View Table

Table 4.

Variance Decompositions for the World (Different Breaks)

			Factors
Variables / Regions			Global	Regional	Global + Regional	Country	Idiosyncratic
Benchmark
	1960-1984
		Output	13.40	10.68	24.08	41.00	32.56
		Consumption	10.57	8.24	18.81	36.95	41.78
		Investment	6.12	9.55	15.67	27.99	54.00
	1985-2010
		Output	9.16	19.46	28.63	34.61	33.53
		Consumption	6.49	13.85	20.34	30.38	46.31
		Investment	8.58	15.72	24.30	29.12	43.79
Break=1983
	1960-1983
		Output	13.91	9.71	23.63	41.04	32.77
		Consumption	11.29	8.19	19.48	36.38	41.36
		Investment	6.50	9.12	15.62	27.65	54.13
	1984-2010
		Output	8.19	20.45	28.64	35.67	32.98
		Consumption	5.16	15.40	20.56	30.30	46.31
		Investment	6.65	18.16	24.82	28.13	44.48
Break=1987
	1960-1987
		Output	13.72	9.07	22.79	42.22	32.80
		Consumption	10.39	6.79	17.18	37.38	43.09
		Investment	5.89	9.23	15.11	28.09	54.66
	1988-2010
		Output	13.74	18.76	32.51	32.88	32.40
		Consumption	8.30	15.96	24.26	29.47	44.21
		Investment	12.15	14.81	26.96	27.94	43.02

Notes: The model is estimated over the periods stated in the first column and the variance decompositions are computed for each country, for output, consumption and investment. The cross-sectional mean of the variance share (in percent) attributable to the relevant factor is reported in each cell. The cross-sectional means are calculated for the world. The column marked “Global+Regional” denotes the sums of the average variance shares of the global and regional factors.

We also analyze how our main results change in response to variations in the country composition of regions. We consider three separate experiments involving the following changes: moving Mexico from North America to the LAC region; moving Turkey from Europe to the MENA region; and moving Israel from MENA to Europe. Although these experiments slightly change the quantitative results, our qualitative findings remain intact.²⁹

C. Crises and Synchronization

It is natural to study whether our results are driven by crises since periods of crises often coincide with highly synchronized business cycles in regions (or groups of countries). The crises episodes can potentially bias our findings and lead to changes in the shares of variance captured by different types of factors. During the second sub-period, the most prominent widespread crises have of course been the Asian financial crisis of 1997–98, which directly affected a handful of countries in the region, and the global financial crisis of 2008–09, which led to sharp contractions in a number of advanced countries.

We undertake three experiments to study the impact of these periods. First, we exclude the four countries (Korea, Malaysia, Philippines, Thailand) directly affected by the Asian crisis from our sample and re-estimate the model. Second, we use the original model estimates and then compute the variance contributions of different factors for the Asian region excluding the crisis countries. Finally, to account for the impact of the global financial crisis, we consider two complementary experiments by excluding the last three years from the second sub-sample (2008–10) and by extending the sample to 2015 using forecasts of the data for the 2011–15 period.

The results of these experiments indicate that our headline findings are preserved. However, the last experiment suggests that the global financial crisis affects some of our other results (Table 5). In particular, we reported in the previous section a small increase in the total contribution of global and regional factors together to output fluctuations in the second sub-period. This finding appears to be driven by the global financial crisis. If the second sub-period includes only the 1985–07 period, the variance share due to the global factor decreases more in the second sub-period leading to a slight decline in the total variance share of output due to the global and regional factors.

Table 5.

Variance Decompositions for the World (Different End Years)

Notes: The model is estimated over the periods stated in the first column and the variance decompositions are computed for each country, for output, consumption and investment. The cross-sectional mean of the variance share (in percent) attributable to the relevant factor is reported in each cell. The cross-sectional means are calculated for the world. The column marked “Global+Regional” denotes the sums of the average variance shares of the global and regional factors. Data for 2011–15 are the forecasts from the IMF’s World Economic Outlook (April, 2011).

Table 5.

Variance Decompositions for the World (Different End Years)

			Factors
Variables / Regions			Global	Regional	Global + Regional	Country	Idiosyncratic
Benchmark
	1960-1984
		Output	13.40	10.68	24.08	41.00	32.56
		Consumption	10.57	8.24	18.81	36.95	41.78
		Investment	6.12	9.55	15.67	27.99	54.00
	1985-2010
		Output	9.16	19.46	28.63	34.61	33.53
		Consumption	6.49	13.85	20.34	30.38	46.31
		Investment	8.58	15.72	24.30	29.12	43.79
End Year=2007
	1985-2007
		Output	6.25	17.24	23.50	36.99	34.13
		Consumption	4.76	12.89	17.65	32.23	45.31
		Investment	5.01	14.54	19.55	32.43	43.58
End Year=2015
	1985-2015
		Output	10.82	19.76	30.58	33.87	33.48
		Consumption	6.70	13.74	20.44	29.66	47.93
		Investment	10.25	14.52	24.77	27.35	45.99

Notes: The model is estimated over the periods stated in the first column and the variance decompositions are computed for each country, for output, consumption and investment. The cross-sectional mean of the variance share (in percent) attributable to the relevant factor is reported in each cell. The cross-sectional means are calculated for the world. The column marked “Global+Regional” denotes the sums of the average variance shares of the global and regional factors. Data for 2011–15 are the forecasts from the IMF’s World Economic Outlook (April, 2011).

View Table

Table 5.

Variance Decompositions for the World (Different End Years)

			Factors
Variables / Regions			Global	Regional	Global + Regional	Country	Idiosyncratic
Benchmark
	1960-1984
		Output	13.40	10.68	24.08	41.00	32.56
		Consumption	10.57	8.24	18.81	36.95	41.78
		Investment	6.12	9.55	15.67	27.99	54.00
	1985-2010
		Output	9.16	19.46	28.63	34.61	33.53
		Consumption	6.49	13.85	20.34	30.38	46.31
		Investment	8.58	15.72	24.30	29.12	43.79
End Year=2007
	1985-2007
		Output	6.25	17.24	23.50	36.99	34.13
		Consumption	4.76	12.89	17.65	32.23	45.31
		Investment	5.01	14.54	19.55	32.43	43.58
End Year=2015
	1985-2015
		Output	10.82	19.76	30.58	33.87	33.48
		Consumption	6.70	13.74	20.44	29.66	47.93
		Investment	10.25	14.52	24.77	27.35	45.99

Notes: The model is estimated over the periods stated in the first column and the variance decompositions are computed for each country, for output, consumption and investment. The cross-sectional mean of the variance share (in percent) attributable to the relevant factor is reported in each cell. The cross-sectional means are calculated for the world. The column marked “Global+Regional” denotes the sums of the average variance shares of the global and regional factors. Data for 2011–15 are the forecasts from the IMF’s World Economic Outlook (April, 2011).

We also analyze the possible effects of the global financial crisis on our results considering a longer time period. Specifically, we extend our sample by including forecasted values of the three macroeconomic variables for the period 2011–15 (these forecasts are based on the IMF’s World Economic Outlook, April 2011). The results point to similar conclusions we get from the 1985–2010 sample.

D. Dynamics of the Global Factor

Although our factor structure is dynamic, the observable variables depend on the factors only contemporaneously in the model. If the global factor has lagged effects on the observables, it is possible that our model understates the importance of the global factor. Given the dynamic nature of our factors, we think that these effects are likely to be small. Moreover, since we use annual data, the lagged effect would have to be fairly slow moving to have an impact on the results. We address the potential for lagged effects by regressing each observable variable on the global factor and two lags of the factor. The coefficients on the first lag were significant for only a small fraction of the time series we have.³⁰ We conclude that lagged responses are of minor importance and would not affect our main conclusions about the relative importance of the global and regional factors.