Tuesday, August 9, 2011

10/08/2011: Was US markets panic behind Irish banks shares crash?

I've just crunched through some interesting data on VIX and Irish Financials index IFIN and there are some interesting results.

To remind you - VIX is in effect a market-based metric of risk in the US markets.

The main premise advanced by the proponents of the argument that US financial crisis drove Irish financial crisis is that panics in the US have caused irrationally pessimistic revaluations of the Irish financial equities and thus led to the collapse of the banks shares in H2 2007- H2 2009.

To assess this, I divided daily data from VIX and IFIN into three periods. Pre-crisis period covers data from January 2000 through July 2007. Financial crisis period covers data from August 2007 through December 2009, while Sovereign crisis period runs from January 2010 through today.

Given the nature of data, VIX data for intraday spreads is only available since September 2003.

Table below summarizes core stats on the data:
Several features worth highlighting in the above:
  • IFIN data shows declining positive skew over the evolution of the crises, while VIX shows growing positive skew. This suggests that rising US risk aversion (VIX) was becoming structural over time as crises progressed from financials to sovereigns, while Irish financials were moving from positively skewed distribution in the pre-crisis period (positive non-risk premium to Irish financials) to progressively smaller positive skew in the crises periods. This is not consistent with the risk spillover from the US to Ireland story.
  • Intraday variation in Irish financials remains smaller than in VIX, but shows qualitatively similar dynamics to VIX. However, increase in intraday variation during the crises is much stronger in the Irish financials than in VIX, which again suggests that risk pricing in the US markets had little to do with Irish financials risk-pricing. Notice that intraday spreads are highly non-normal in their distribution with third and fourth moments off the charts.
  • 1-month dynamic correlations between VIX and IFIN remained negative across all periods (implying that rising US risk was associated with falling IFIN valuations), but relatively weak (at maximum mode of 0.35 on average). median correlations show a bit more dynamism during the crisis, rising from -0.41 in pre-crisis period to -0.51 during the Financial crisis period and declining to -0.45 in Sovereign crisis period. However, these are not dramatic either. In fact, positive skewness was reinforced during the Financial crisis period, while negative kurtosis declined in absolute value.

Chart above summarises the entire series of data, showing historically relatively weak, but negative (as expected) correlation between the values of Irish financial shares and the risk levels in the US markets.

Chart below breaks this down into three periods:
What's interesting in the above chart is that:
  1. Correlation remains negative but explanatory power significantly declines in the period of Financial Crisis (so the picture is the opposite of the claim that the US 'panic' spilled over into Irish markets), while the slope remains relatively stable.
  2. More interestingly, the relationship completely disappears since the onset of the Sovereign crisis. basically, once the IFIN hit 4,000 levels, there is no longer any meaningful connection between Irish financial shares prices and risk attitudes or perceptions in the US markets. Guess what - that magic number was reached around 29/09/2008.
Chart below plots 1mo dynamic correlations between VIX and IFIN
While correlations tend to stay, on average, in the negative territory, as the table above shows, they are not significantly large. In fact, overall during the Financial crisis period there were 318 instances of the correlation equal to or exceeding (in mode) 0.5 - or 51% of the time. In pre-crisis period this number was 42% and during the Sovereign crisis so far - 45%. But there is a slight problem in interpreting this 51% as the spillover effect from the US. During the Financial crisis period, pre-Lehman collapse, higher correlations took place 58% of the time, while post-Lehman collapse they took place 45% of the time. So overall, it appears that US risk attitudes (aka 'panics') were more related to adverse movements in IFIN before the Big Panic took place than during and after the Big Lehman's Panic set on.

Interestingly, there is also no evidence that changes/volatility in the US attitudes to risk had any significant serious impact (adverse or not) on volatlity Irish financial shares valuations, as shown in the chart below:
In no period in our data is there a strong relationship between changes (volatility) in US risk attitudes and the Irish financial shares valuations volatility.

A note of caution - these are simple tests. The data shows a number of problems that require serious econometric modeling, but overall, so far, there is no strong evidence to support the proposition that Irish banks shares or financial shares have been significantly and systematically adversely impacted by the US 'panic' or by 'Lehman collapse'. Our banks problems seem to be largely... our banks own problems...

No comments: