Lorenzo Bini Smaghi on the Impact of Basel III

Readers of this Blog will want to keep up with LBS’s thinking on how the new banking rules will affect the real economy. See his recent address to an Italian conference here.  (Surely the correlation coefficient of +0.95 for Ireland in slide 18 is a mistake?)  Overall, it seems, there will be little adverse effect on the real economy in the long run.

He is quoted in today’s Italian papers warning the banks that they must use their profits to bolster their reserves.

12 replies on “Lorenzo Bini Smaghi on the Impact of Basel III”

He is quoted in today’s Italian papers warning the banks that they must use their profits to bolster their reserves.

As we’ve learned to our cost over the last three years, most bank “profits” consist of paper money, stocks, “assets”, portfolios, etc. Any actual cash they make is given out in bonuses. This idea isn’t going to fly.

OMF; you are mixing up cash and profit, liquidity and solvency. This is the same sort of mistake the banks made. Cash is for a bank like dresses are for Brown Thomas ladies’ department. You have to have loads just to be able to open the business each morning but just because you have loads doesn’t necessarily mean you have made a bob.

His piece in Il Sole 24 suggests to me (a) more centralized control at ECB level and (b) no discounting applied to bonds – ‘the risk of contagion shouldn’t be underestimated’ (he takes a swipe at academics for just this). In La Repubblica two days previously Smaghi was quoted as saying something along the lines that Dublin was expected to provided an emergency safety net for its banks. For Minister Noonan and the boys it is back to the drawing board. Burden sharing off the table.

Stress tests EBA

“Differently from what some commentators seem to suggest, this time stress tests will be credible and more rigorous”.

Glad to hear that.


“Liquidity issues, however, should be addressed with different tools

Basel III rules have opted for specific quantitative provisions – outright liquidity ratios – and rightly so

This should improve the management of liquidity risk, which is a cash-flow risk, and cannot be fully addressed via capital requirements only”

So we put in 24 billion and we still have the cash-flow risk of 200 billion.
So what tools will Lorenzo provide for us?

“Surely the correlation coefficient of +0.95 for Ireland in slide 18 is a mistake?”

I’m not sure why you think so. Obviously the best-fitting line is steeper than the 45-degree line shown with the scatterplot, but to my untrained eye the relationship seems near-enough linear.

I get the impression that he thinks these high correlations are evidence for something. But what? The worst models are the ones that don’t get written down — the work of Accidental Theorists.

A 7-8% decline in EU investment on either model and no recovery in the EU one (slide 7). Just what Europe needs.

What the ECB seems unaware of, along with the Commission and its supporters, is how closely German growth over the last 20 years has resembled the Japanese ‘model’ more even than the US, itself experiencing its slowest-ever 20yr period of growth, certainly since the Great Depression.

This model is now being imposed on EU countries individually, beginning with Greece and Ireland, and will be collectively from September this year.

Why would such an achingly poor model of growth be adopted?

@Kevin Donoghue:
In the Irish case, the Bank CDS and Sovereign CDS fall pretty much on a 45 degree straight line up to the point where the latter reaches 600, then the slope of the line become almost vertical, as the Bank CDS rises very steeply but the Sovereign remains round 600. Over the whole range, the linear correlation cannot be 0.95! Is the significance of this that in the case of Ireland the markets have uncoupled bank risk from sovereign risk?

@Brendan Walsh

My comment mostly reflected my reluctance to trust my eye when it comes to guessing a correlation coefficient. You have vastly more experience so you’re probably right. But bearing in mind that there’s at least 300 observations in that scatterplot, I’d have to see the data to be convinced. A dozen or so stray points prompt the eye to see a curve with a positive second derivative, but they should not make much impression on the calculated correlation. Most of the observations are huddled together in clusters. I thought about looking for data, but the notes provided by LBS are not encouraging:

Page 17: “The chart presents average sovereign CDS (weighted with the capital key at the ECB) and bank CDS (simple average for 10 largest euro area banks) for daily observations in the period 1.1.2010 – 18.03.2011 and the correlation coefficient. Diagonal line at 45°.”

Page 18: “The chart presents sovereign CDS and median bank CDS for daily observations in the period 1.1.2010 – 18.03.2011, and the correlation coefficient.. Diagonal line at 45°. Data in basis points.”

The following may be foolish questions, but what the hell, this is a blog. Are any of the 10 largest euro area banks Irish, Greek or Portuguese? If not (as I suspect) what’s the point of these correlations? And why use the simple average bank CDS on Page 17 and the median on Page 18?

My jaundiced view of LBS is that he just got some staffer to throw a few pictures together, vaguely supporting the notion that the peripherals cannot hope to divorce bank risk from sovereign risk. Maybe we can’t, but I’d like to see a less tendentious thinker examine the case.

If Dear Lorenzo’s maths & stats are as dodgy as his logic – I wouldn’t be getting too excited … I won’t be wasting me time …

@Kevin Donoghue

According to Lorenzo’s logic the correlation should always be 1.00 – he acknowledges no distinction between banks and the state in any EZ country and both are to be perfectly twinned for ever more with the latter fully responsible for all ills, defects, sins, let alone debts of the former while the present executive board of The ECB holds sway well into the fulness of the florentine sense of the limits of limitless time.

@Kevin Donoghue
You are perfectly right to point out that the correlation could be as high as 0.94.
I should have made a different point, namely, that the linear correlation is meaningless when there is clearly a break in the relationship between the two variables.

Comments are closed.