Clarus Financial Technology

Liquidity on the Bloomberg SEF

Chris Barnes

Liquidity

In this third article in a series, we are back to the theme of liquidity, as we take a deep dive into some data analysis from SDRView Researcher.

After reviewing the previous post on liquidity measures, I realised that we can split our data not only by tenor but also source. This allows us to look at volumes traded in specific instruments, such as a 10 Year USD IRS, and compare volume and price metrics for trades reported to the Bloomberg SDR versus those reported to the DTCC. This therefore allows us to analyse trades done on the Bloomberg SEF versus those done in the broader, on-SEF market.

So whilst the last blog concluded that liquidity was much greater on-SEF than off-SEF, we also have the opportunity to assess liquidity on a particular SEF versus the rest of the on-SEF market.

Compelling stuff when trying to decide, pre-trade, which SEF to utilise.

Volumes

First off, we need to ensure we are looking at a valid data set. From SDRView Researcher, we can use the preset “Tenor” view and split it by source:

USD Swaps by Tenor and Source

(API Call: sdrview.clarusft.com/rest/api/…. Requires a valid API key).

Showing;

Market Share

And below we can quickly change this same query into Risk terms (changing the display from Notional to DV01) and take a look at the BSEF market share per tenor:

Bloomberg SEF Market Share for USD Swaps by Tenor

That chart is almost worth a blog unto itself, but the quick highlights are:

My gut take on those figures is that they may be over-stating the true market share of Bloomberg’s SEF – mainly because all volumes are capped at the reporting threshold, and conventional wisdom would state that the IDB SEFs see larger ticket sizes in these standardised instruments. We’ve talked about this previously in terms of block trades – but not looked at what percentage of those block trades are non-spot starting, tailored transactions. That might make an interesting follow-up next week….stay tuned.

Liquidity

Nit-picking details aside, I wanted to answer a simple question – does liquidity on the Bloomberg SEF warrant this impressive market share? To answer that, I leveraged the work we did previously when we looked at price dispersion measures, as per the recent BoE staff report on SEF liquidity. For those interested, the measures of liquidity we use are price dispersion indices, calculated as per the below:

\( \tag {1} DispVW_{i,t} = \sqrt{\sum\limits_{k=1}^{N_{i,t}}\frac{Vlm_{k,i,t}}{Vlm_{i,t}}(\frac {P_{k,i,t}-\bar{P_{i,t}}}{\bar{P_{i,t}}})^2}\)

where;
\(N_{i,t}\) is the total number of trades executed for contract i on day t, e.g. how many 10y trades occurred on the 8th of January 2016?
\(P_{k,i,t}\) is the execution price of transaction k, i.e. the price of a particular 10y trade on the 8th January 2016.
\( \bar{P_{i,t}}\) is the average execution price on contract i and day t, e.g. what was the average price of all 10y trades done on 8th January 2016.
\( Vlm_{k,i,t}\) is the volume of transaction k e.g. the size of the 10 year trade we are looking at on 8th January and
\(Vlm_{i,t}=\sum_{k}Vlm_{k,i,t}\) is the total volume for contract i on day t. e.g. the total volume of 10 years traded on the 8th January.

We also calculated a second measure, replacing the 10 year average price with the 10 year closing mid-price as per the “JNS” measure from this original paper. We take our SDR fixings for the mid-price.

Generally speaking, we would like price dispersion to be as low as possible. This means that any particular trade has had a minimal impact on average prices that day, and seeing as we volume weight the contribution of each trade, larger trades have passed through the market without any meaningful impact.

A Quick Caveat

We acknowledge that there will be some deviation from true averages and true mids due to the mixture of CME-cleared and LCH-cleared swaps with the emergence of the CCP basis during 2015. However, for the purposes of this blog, let’s assume that the mixture of CME and LCH cleared swaps is broadly comparable between trades done on BSEF and the wider SEF market. (Feel free to comment below or contact us if you have a strong opinion (or data!) on that front).

Liquidity Results

Background in place, let’s look at some data and some charts. Generally speaking, when considering the numbers, we can say that “liquidity” was greater on BSEF when it had a lower price dispersion than the market as a whole.

Admittedly, this broad brush approach can play into the hands of a single SEF if it sees very little volume across a small number of tickets, and hence all trades occur close to its’ average price for the day.

However, as evidenced earlier – and because we are looking at 10 year USD swaps – this is not the case for our data sample.

Daily

So, let’s look at the daily histories of price dispersion on the BSEF vs the wider SEF market.

Click to enlarge:

 

 

 

 

 

 

Showing;

Monthly

Now, consistent with much of the industry – including our own Swaps’ reviews – let’s collate this data on a monthly basis. The left hand chart shows the cumulative Volume Weighted price dispersion (in basis points) per venue, whilst the right hand chart shows the JNS measure which takes price dispersion relative to the end of day “mid” price instead of a daily average price.

Click to enlarge:

 

 

 

 

 

 

Showing;

Overall

Finally, let’s take a look at the number of days, each month, that BSEF enjoyed a lower Price Dispersion reading, and hence beneficial liquidity conditions, compared to the rest of the SEF market:

 

 

 

 

 

 

 

Showing;

In Summary

It’s been one of the longer posts today, but I find the analysis particularly interesting:

Stay informed with our FREE newsletter, subscribe here.

Exit mobile version