This chapter investigates financial contagion from the United States to eight African frontier markets during the 2007–09 financial crisis using a dynamic conditional correlation-based methodology. The crisis is explored from the perspectives of it being, first, a single long-period event and, second, a series of cumulative short-period subevents. For comparative purposes, contagion from the United States to a set of seven developed markets is also examined. We find strong evidence of contagion in African frontier markets during the early phases of the crisis. In contrast, contagion in developed markets was observed to occur more slowly. We speculate that the differences found are consistent with herding behavior occurring at a faster rate in the frontier markets than in their developed market counterparts.
Table 6.1
Descriptive Statistics over the Period 01/01/07–10/15/2009
Country | Mean Daily Return (%) | Standard Deviation (%) | Correlation with United Statesa | Skewness | Excess Kurtosis | |
Frontier Africa | United State | −0.036 | 1.937 | 1 | −0.161 | 5.799 |
Botswana | 0.017 | 0.569 | −0.014 | 1.263 | 21.198 | |
Côte d’Ivoire | 0.026 | 25.272b | −0.054 | 0.039 | 331.493b | |
Mauritius | 0.054 | 1.295 | 0.009 | 0.067 | 6.713 | |
Morocco | 0.020 | 1.051 | −0.027 | −0.570 | 4.119 | |
Namibia | −0.022 | 3.397 | 0.341 | 0.882 | 113.549c | |
Nigeria | −0.058 | 1.333 | −0.040 | 0.060 | 0.603 | |
Tunisia | 0.087 | 0.656 | 0.126 | −0.586 | 8.986 | |
Zambia | 0.062 | 1.280 | −0.015 | 0.475 | 8.578 | |
Developed | Canada | −0.014 | 1.854 | 0.653 | −0.526 | 5.100 |
France | −0.053 | 1.916 | 0.660 | 0.207 | 5.435 | |
Germany | −0.020 | 1.870 | 0.594 | 0.245 | 5.999 | |
Italy | −0.078 | 1.567 | 0.484 | −0.408 | 4.825 | |
Spain | −0.035 | 1.828 | 0.578 | 0.013 | 4.927 | |
United Kingdom | −0.028 | 1.746 | 0.663 | −0.026 | 4.658 |
a Constant correlations are estimated for the precrisis period of Jan. 1, 2007–Sep. 14, 2008, only. This period is used in order to identify precrisis differences in the level of integration with the US market.
b This unadjusted data needs to be treated with caution for reasons identified in the description associated with Figure 6.2b.
c This high value is associated with a spike in returns toward the end of the crisis period.
Table 6.2
Comparison-of-Means Based Contagion Tests: Long Crisis Period
African Markets | Stable Period | Crisis Period | Test for Equality of Variancesb | T-test for Equality of Meansc,d | Evidence of Contagione | |||
Mean Corr.a | Standard Deviation | Mean Corr.a | Standard Deviation | F | Sig | |||
Botswana | −0.028 | 0.035 | 0.011 | 0.043 | 30.776 | 0.000 | 12.060*** | Yes |
Côte d’Ivoire | 0.028 | 0.035 | 0.036 | 0.074 | 187.983 | 0.000 | 1.659** | Yes |
Mauritius | 0.043 | 0.000 | 0.043 | 0.000 | 24.590 | 0.000 | 1.881** | Yes |
Morocco | 0.011 | 0.036 | 0.017 | 0.081 | 159.192 | 0.000 | 1.060 | No |
Namibia | 0.342 | 0.217 | 0.318 | 0.093 | 164.047 | 0.000 | −1.965 | No |
Nigeria | 0.043 | 0.000 | 0.043 | 0.000 | 4.982 | 0.026 | 0.904 | No |
Tunisia | 0.066 | 0.002 | 0.066 | 0.002 | 1.811 | 0.179 | 0.935 | No |
Zambia | −0.016 | 0.000 | −0.016 | 0.000 | 8.478 | 0.004 | −1.269 | No |
Developed Markets | ||||||||
Canada | 0.763 | 0.112 | 0.828 | 0.046 | 121.101 | 0.000 | 10.486*** | Yes |
France | 0.713 | 0.053 | 0.745 | 0.053 | 4.051 | 0.045 | 7.795*** | Yes |
Germany | 0.664 | 0.059 | 0.769 | 0.030 | 67.056 | 0.000 | 30.620*** | Yes |
Italy | 0.531 | 0.101 | 0.567 | 0.052 | 123.984 | 0.000 | 6.169*** | Yes |
Spain | 0.602 | 0.065 | 0.704 | 0.046 | 34.937 | 0.000 | 23.574*** | Yes |
United Kingdom | 0.665 | 0.033 | 0.720 | 0.028 | 3.672 | 0.056 | 22.150*** | Yes |
a Mean daily time-varying correlation over the period.
b Levine’s test of variance equality; where rejection of the null hypothesis indicates inequality.
c The t-value reported is the form appropriate for the variance identified.
d *Significant at 10%, **significant at 5%, ***significant at 1%.
e Contagion is defined as statistical significance at 5% level. Note that the test results for Mauritius, Nigeria, Tunisia, and Zambia need to be treated with caution due to an insignificant alpha in the DCC model.
Table 6.3
Comparison-of-Means Based Contagion Tests: Subcrisis Periods
African Markets | Stable Period | Subcrisis Period 1 | Cumulative Impact of Subcrisis Periods 1–2 | Cumulative Impact of Subcrisis Periods 1–3 | Cumulative Impact of Subcrisis Periods 1–4 | |||||||||||||
ρ | σ | ρ | σ | t-testa,b | Ctn.c | ρ | σ | t-testa,b | Ctn.c | ρ | σ | t-testa,b | Ctn.c | ρ | σ | t-testa,b | Ctn.c | |
Botswana | −0.028 | 0.035 | 0.012 | 0.045 | 4.697*** | Y | 0.014 | 0.041 | 5.449*** | Y | 0.020 | 0.038 | 7.067*** | Y | 0.045 | 0.043 | 13.139*** | Y |
Côte d’Ivoire | 0.028 | 0.035 | 0.103 | 0.013 | 21.56*** | Y | 0.113 | 0.026 | 15.02*** | Y | 0.119 | 0.026 | 13.610*** | Y | 0.110 | 0.025 | 19.900*** | Y |
Mauritius | 0.043 | 0.000 | 0.043 | 0.000 | 0.229 | N | 0.043 | 0.000 | 1.994** | Y | 0.043 | 0.000 | 2.806*** | Y | 0.043 | 0.000 | 3.813*** | Y |
Morocco | 0.011 | 0.036 | 0.131 | 0.030 | 13.85*** | Y | 0.139 | 0.033 | 16.55*** | Y | 0.144 | 0.031 | 19.220*** | Y | 0.152 | 0.028 | 25.290*** | Y |
Namibia | 0.342 | 0.217 | 0.367 | 0.037 | 1.839** | Y | 0.366 | 0.034 | 1.863** | Y | 0.367 | 0.031 | 2.083** | Y | 0.369 | 0.026 | 2.3170** | Y |
Nigeria | 0.043 | 0.000 | 0.043 | 0.000 | 1.312* | N | 0.043 | 0.000 | 2.198** | Y | 0.043 | 0.000 | 1.719** | Y | 0.043 | 0.000 | 0.354 | N |
Tunisia | 0.066 | 0.002 | 0.067 | 0.004 | 1.017 | N | 0.067 | 0.004 | 1.212 | N | 0.066 | 0.003 | 0.849 | N | 0.066 | 0.003 | 0.798 | N |
Zambia | −0.016 | 0.000 | −0.016 | 0.000 | 0.225 | N | −0.016 | 0.000 | 0.252 | N | −0.016 | 0.000 | 0.276 | N | −0.016 | 0.000 | 0.338 | N |
Developed Markets | ||||||||||||||||||
Canada | 0.763 | 0.112 | 0.717 | 0.049 | −3.676 | N | 0.730 | 0.054 | −2.655 | N | 0.744 | 0.056 | −1.541 | N | 0.770 | 0.056 | 0.719 | N |
France | 0.713 | 0.053 | 0.666 | 0.015 | −10.962 | N | 0.672 | 0.018 | −9.231 | N | 0.678 | 0.020 | −7.912 | N | 0.724 | 0.027 | −6.478 | N |
Germany | 0.664 | 0.059 | 0.708 | 0.020 | 8.195*** | Y | 0.716 | 0.025 | 9.107*** | Y | 0.724 | 0.027 | 10.600*** | Y | 0.737 | 0.028 | 14.88*** | Y |
Italy | 0.531 | 0.101 | 0.477 | 0.026 | −6.935 | N | 0.484 | 0.029 | −5.991 | N | 0.494 | 0.033 | −4.670 | N | 0.516 | 0.039 | −1.939 | N |
Spain | 0.602 | 0.065 | 0.617 | 0.020 | 2.622*** | Y | 0.617 | 0.019 | 3.008*** | Y | 0.622 | 0.020 | 4.080*** | Y | 0.630 | 0.019 | 6.469*** | Y |
United Kingdom | 0.665 | 0.033 | 0.665 | 0.013 | −0.111 | N | 0.669 | 0.014 | 1.001 | N | 0.673 | 0.015 | 2.338*** | Y | 0.681 | 0.017 | 5.547*** | Y |
a As in Table 6.2, the t-value reported is appropriate for the form of variance identified.
b *Significant at 10%, **significant at 5%, ***significant at 1%.
c Contagion (Ctn.) is defined as statistical significance at 5% level. Note that the test results for Mauritius, Nigeria, Tunisia, and Zambia need to be treated with caution due to an insignificant alpha in the DCC model. For example, contagion shown in periods 1–2 in respect to Nigeria reflects changes in correlation beyond three decimal places.
Table 6.4
Dummy Variable Based Robustness Tests for Contagion: All Periods
African Markets | Long Crisis Period | Subcrisis Period 1 | Cumulative Impact of Subcrisis Periods 1–2 | Cumulative Impact of Subcrisis Periods 1–3 | Cumulative Impact of Subcrisis Periods 1–4 | ||||||||||
Dummy Coefficient | t-valuea | Ctn.b | Dummy Coeff. | t-valuea | Ctn.b | Dummy Coeff. | t-valuea | Ctn.b | Dummy Coeff. | t-valuea | Ctn.b | Dummy Coeff. | t-valuea | Ctn.b | |
Botswana | 0.039 | 12.642*** | Y | 0.040 | 4.697*** | Y | 0.042 | 5.449*** | Y | 0.048 | 7.067*** | Y | 0.073 | 13.139*** | Y |
Côte d’Ivoire | 0.008 | 1.874** | Y | 0.075 | 8.912*** | Y | 0.085 | 11.316*** | Y | 0.092 | 13.609*** | Y | 0.082 | 15.377*** | Y |
Mauritius | 0.000 | 2.001** | Y | 0.000 | 0.229 | N | 0.000 | 3.746*** | Y | 0.108 | 7.346*** | Y | 0.130 | 11.229*** | Y |
Morocco | 0.006 | 1.221 | N | 0.120 | 13.846*** | Y | 0.128 | 16.553*** | Y | 0.000 | 4.861*** | Y | 0.000 | 5.699*** | Y |
Namibia | −0.024 | −1.710 | N | 0.025 | 0.513 | N | 0.024 | 0.548 | N | 0.132 | 19.224*** | Y | 0.141 | 25.291*** | Y |
Nigeria | 0.000 | 0.967 | N | 0.000 | 2.190** | Y | 0.000 | 2.198** | Y | 0.025 | 0.648 | N | 0.027 | 0.856 | N |
Tunisia | 0.000 | 0.935 | N | 0.001 | 2.332*** | Y | 0.001 | 2.514*** | Y | 0.000 | 1.719** | Y | 0.000 | 0.560 | N |
Zambia | 0.000 | −1.449 | N | 0.000 | 0.225 | N | 0.000 | 0.252 | N | 0.099 | 3.731*** | Y | 0.123 | 5.896*** | Y |
Developed Markets | |||||||||||||||
Canada | 0.066 | 9.090*** | Y | −0.046 | −1.802 | N | −0.033 | −1.418 | N | −0.018 | −0.873 | N | 0.007 | 0.427 | N |
France | 0.032 | 7.795*** | Y | −0.046 | −3.906 | N | −0.041 | −3.851 | N | −0.035 | −3.679 | N | −0.025 | −3.345 | N |
Germany | 0.106 | 27.332*** | Y | 0.045 | 3.387*** | Y | 0.053 | 4.470*** | Y | 0.060 | 5.653*** | Y | 0.073 | 8.628*** | Y |
Italy | 0.037 | 5.523*** | Y | −0.054 | −2.372 | N | −0.046 | −2.281 | N | −0.036 | −2.001 | N | −0.015 | −1.004 | N |
Spain | 0.102 | 22.071*** | Y | 0.014 | 0.981 | N | 0.015 | 1.136 | N | 0.020 | 1.660** | Y | 0.027 | 2.919*** | Y |
United Kingdom | 0.054 | 22.154*** | Y | 0.000 | −0.050 | N | 0.003 | 0.486 | N | 0.007 | 1.249 | N | 0.016 | 3.350*** | Y |
a *Significant at 10%, **significant at 5%, ***significant at 1%.
b Contagion (Ctn.) is defined as statistical significance at 5% level. Note that the same caveat needs to be applied as in Tables 6.2 and 6.3; test results for Mauritius, Nigeria, Tunisia, and Zambia need to be treated with caution due to an insignificant alpha in the DCC model.
18.117.98.250