Violence & Capital Flight

Violent election year
Generates enough fear,
Speeding clandestine capital flight
If level of significance is right.

Global Financial Integrity has recently brought out its another report on illicit financial flow. Unlike previous reports, this time it underscores trade invoicing took place between 2008 and 2017. Illicit Financial Flow was broad and covered many areas. This report concentrates on trade mismatches. However, I did not get the relevant data for Bangladesh up to 2018. Much of the recent data were not available.

The 2015 report was complete and I found it handy back then. Immediately after the 2015 report , I tried to find out any possible link between political violence and illicit financial flow. Was there a correlation between the two? There was a little bit of correlation.

I still hinged on that 2015 data and looked on whether election years have any differential effect on illicit financial flow. For the political violence data, I relied on Odhikar, an NGO works on human rights issues. And from 2015 report I gleaned the data on illicit financial flow.

The regression function I constructed looked like this:

lnIFF_i = a₁ + bPolViol_i + a₂ D_i

where lnIFF_i= log natural of illicit financial flow,
PolViol_i= victims of political violence,
D_i= 1 when the years are election years; prior and after years fall here.
= 0 when the years are normal years.

Here a₂, coefficient of dummy variable, captures the differential effect of election years.

Since the data is time series in nature, I checked for autocorrelation. Durbin Watson statistic prior to natural log transformation reported 1.40 for 10 observations and 1 explanatory variable.

Then I carried out the semilogarithmic regression. At 5% level of significance, the model did not turn out to be significant. F= 3.339, p=0.095 (for degrees of freedom 2 and 7) . However it is significant at 10% level of significance. The intercept, a₁=8.273, appeared to be significant (t=44.79, p=0.000000721). The slope coefficient of political violence, b=0.000013, turned out to be insignificant (t=1.099, p= 0.308). The coefficient for dummy variable, a₂=0.4210 was found to be insignificant (t= 2.116, p=0.0720). However, it was significant at 10% level of significance.

Stat analysis did not find any evidence of differential effect of election years on illicit financial flow. What is interesting is that if the level of significance is raised at 10% level of significance then there is evidence that election years have some kind of differential effect on illicit financial flow. However, slope coefficient of political violence is still insignificant.

Let's interpret the result. At 10% level of significance, mean illicit financial flow of election years was higher than non-election years by 50.97%. As election years were more violent, as reflected in the political violence stat, more people laundered clandestine money abroad.

Based on higher level of significance , there is statistical evidence that violent election years accelerated the process of illicit financial flow. Clearly panic has its own economics. Sowing panic has intended /unintended consequences. However, finding conspiracy theory is not my goal. But my endeavor to see effect of violent election years on clandestine capital flight was not a failure.