The EBA just published the results of the 2016 transparency exercise. Yay! The EBA decided to deliver on its promise to present more detailed data of individual banks. (I hope the EBA now forgives me for posting a critical note on their bank disclosure plans earlier this year.)
I toyed a bit with the data do examine sovereign holdings of banks. Less of these holdings is better, it helps to break the toxic bank-sovereign loop.
Thanks to R and Excel, I was able to manage the data fairly quickly.
First of all, lots of sovereign holdings are classified as Available for Sale (AFS), an accounting classification that allows for the recognition of gains and losses in bank capital. The table below shows that EU banks that participated in the 2016 transparency exercise held 3,77tn euro of sovereigns, of which 1.65tn (43%) in AFS:
This is potentially tricky, given that losses in the AFS category will increasingly affect banks’ Common Equity Tier 1 ratio. The ECB, for example eliminated a filter that protected banks from losses on sovereign AFS debt. In addition, banks are increasingly exposed to swings in values of securities held in AFS, because the CRR phases out the old filter that protected banks from any volatility in their AFS portfolios. By end of this year, at least 60% of unrealised AFS gains and losses shall affect Common Equity Tier 1, next year that will be 70%.
More of a worry is that interest rate risk may affect AFS gains and losses going forward, something that barely attracts the attention of supervisors. But it surely will at some point.
The next table shows a breakdown of the geographical allocation of sovereigns by European banks. The first column shows the banks’ countries of origin. The numbers then show where these banks invest: Austrian banks hold 30% of their own sovereigns, and 0.8% of Belgian sovereigns. The elevated values on the diagonal should not surprise us: banks are happy to invest in home securities.
What did surprise me was the holdings of Italian sovereigns by two distinct countries: Belgium (15.4%) and Portugal (13%). These numbers are driven by Dexia for Belgium and by Novo banco for Portugal.
With the Italian referendum this weekend, this boggled my mind: why would these banks expose themselves so oddly?
 See Article 14 of the ECB Options and Discretions regulation.