How good a picture can Euro GC Pooling provide of the interbank euro GC repo market?

An interesting academic paper has recently suggested that Eurex Repo’s Euro GC Pooling (EGCP) system ‘allows us to comprehensively investigate the bilateral CCP-based part of the interbank segment of the euro GC repo market’ and makes it possible to ‘accurately document and analyse developments in the bilateral CCP-based interbank repo market’.[1] This is of topical interest, as the idea has been floated that the EGCP euro repo rate index, in the form of the STOXX GC Pooling Index, could provide a credible alternative to unsecured interbank deposit indexes such as EURIBOR and LIBOR.

Just how representative is EGCP of the bilateral/non-triparty CCP-cleared interbank euro GC repo market segment?[2]

The academic paper notes that the ECB money market survey measured the non-triparty CCP-cleared interbank euro GC repo segment in 2012 at EUR 210 billion average turnover per day (all figures are double-sided, ie repo plus reverse repo). EGCP turnover (also double-sided) averaged EUR 65 billion or about 31% of the ECB figure. This is a large share but not predominant. It also needs to be remembered that the ECB survey is only took a sample of the non-triparty CCP-cleared interbank euro GC repo market inasmuch as it covered 172 banks, not all of which trade repos. So, the true market share of EGCP was somewhat lower than 31%.

What might be the size of the whole non-triparty CCP-cleared interbank euro GC repo market? We know that non-triparty CCP-cleared interbank euro GC repo is traded on BrokerTec and MTS. Over 2012, this business averaged a total of about EUR 100 billion a day (double-sided). This seems to be about twice the authors’ estimate for BrokerTec and MTS (see Fig.2 on p34 of their paper), but still within the ECB estimate of EUR 210 billion.

However, if the business underlying the RepoFunding Rates published by BrokerTec and MTS is taken as a proxy for non-triparty CCP-cleared interbank euro GC, this market segment in 2012 would have been some EUR 246 billion per day on average (double-sided) just for the German, French and Italian markets, plus the EUR 65 billion for EGCP. On this basis, EGCP is 11%.[3]

On the basis of relative size, therefore, there must be doubt about just how representative EGCP can be of the triparty CCP-cleared GC interbank euro repo market segment. The quality of EGCP as a wider market proxy is also undermined by the lack of geographical diversity in the user base of EGCP. It is still dominated by German banks (44%). There are no Italian users at all.

The collateral composition of EGCP is also out of line with the wider interbank market. First, EGCP’s ECB collateral basket contains some 7,500 securities, of which, two-thirds are non-government issues. Although they are all ECB-eligible and subject to concentration limits, given that the collateral management system will select the cheapest-to-deliver and that the basket is far wider than would be found in the interbank market, particularly the non-triparty segment, the collateral mix held by net buyers in EGCP is almost certainly more diversified and of lower average quality than the wider interbank market. The difference will be accentuated by the fact that the ECB ECB basket excludes Italian and Spanish government bonds (in fact, it only includes six Eurozone government bonds).

Given these fundamental differences in coverage, it is doubtful as to whether EGCP can provide an accurate picture of non-triparty CCP-cleared interbank euro GC repo. There must be serious doubts as to whether the GC Pooling Index is an accurate measure of the cost of interbank euro GC repo funding. Even an index that was reasonably representative of the non-triparty CCP-cleared interbank euro GC repo market segment would suffer from the fact that it excluded the non-CCP segment. This includes the whole of the voice-brokered market segment repos (some 19% of the overall market according to the ECB money market survey, albeit mainly for term) and a share of the directly-negotiated segment, some of which is interbank GC.

The GC Pooling Index currently struggles to provide a valid insight into the cost of interbank repo funding beyond one-day tenors. EGCP, like all electronic trading systems, is overwhelmingly overnight, tom/next or spot/next. The academic paper unconsciously exaggerates the importance of term repos in EGCP by using term-adjusted tenors. And the use of Eurepo to fill gaps merely compounds the problem, given that this index is moribund.[4]

But the real problem may be that the STOXX GC Pooling Index may be attempting what is, at the moment, an impossible task. Since 2008, the euro repo market has been fragmented by the re-emergence of significant risk premia between eurozone sovereigns. The consequent price-tiering means that there is currently no such thing as a euro GC repo rate.

[1]  ‘The Euro Interbank Repo Market’ by Loriano Mancini, Angelo Ranaldo and Jan Wrampelmeyer University of St Gallen School of Finance Working Paper 2013/6).

[2]  The term ‘bilateral’ is used by the ECB to describe non-triparty transactions. They are using bilateral to refer to the method of settlement, ie whether this takes place directly between banks or their custodians, or is outsourced to a triparty agent. But the term ‘bilateral’ is also widely used to describe the face-to-face relationship of counterparties (essentially trading in an OTC market) and whether a CCP is intermediating. Moreover, the application of the term ‘bilateral’ to ATS (automatic trading systems) or electronic markets is strictly-speaking incorrect, as these are in fact ‘multilateral’ markets, in that users can see and participate in the market at the same time. And contrary to the suggestion in the paper, EGCP is part of the triparty market. This system consists of: (1) an ATS (an automatic trading system or electronic platform) — Eurex Repo; (2) a CCP — Eurex Clearing; and (3) a triparty agent — Clearstream — which manages the allocation and management of collateral.

[3]  All business on BrokerTec and MTS is interbank by virtue of the membership criteria of these ATS and most is CCP-cleared. The selection of rates to be input into the RepoFund Rates produced what can be argued is a reasonable proxy for GC repo rates. It needs to be remembered that trades in ‘specific’ securities can be GC. They only cease to be GC if they go on special. The RepoFunds Rate selection algorithm removes most, if not all, specials.

[4]  The STOXX GC Pooling Index faces the same problem and has to use quotes rather than traded rates.

Leave a Reply