A topic we have covered before is the evolution of the market place for Corporate Bond trading - "Matching engine for corporate bonds?". Since the demise of Lehman Brothers we have seen the banks pull back capital from the corporate bond flow trading business and a subsequent drastic reduction in liquidity. This post is a modest proposal for the structure of how a new Fixed Income market could work.
Firstly, let's establish some principles of business:
- The aim is not to disintermediate banks and drive them out of this line of business.
- The aim is not to disintermediate interdealer brokers and drive them out of this line of business.
- The purpose of the new market is to be a fair and open place for buyers and sellers of bonds to transact.
- All participants are welcome, as long as they respect the rules.
- Best execution is the goal
The market model here is based upon observations of market behaviour within the buy-side and the sell-side in a number of European firms.
Per the earlier post on "Bond fund distance, liquidity and trading" we see that there is an opportunity for a type of matching to occur even when there is no direct match. So, let's illustrate this more clearly:
Model 1 - "Buy-side to buy-side in complete size via one or two brokers as agent or principal"
- Buy-side A enters an order into the new venue matching engine to buy 8m of the Vodafone bond 1.25% of 26/09/2017. This may or may not have a limit price.
- Buy-side B enters an order into the new venue matching engine to sell 8m of the Vodafone bond 1.25% of 26/09/2017. This may or may not have a limit price.
As a first pass, the matching engine looks for sellers of any bond with buyers and buyers of any bond with sellers. In this case we have a perfect match for size.
The issue is now - did either firm place a limit price on their order? If there were one or more limit prices then the new venue matching engine will have to start a process of negotiation between the two firms.
I see this as being anonymous and probably facilitated by one or more banks. So the buy-sides do not enter into direct negotiation, one or two banks act as agent or principal in this process. If a deal can be done then the trade is executed and anonymously given up, so neither buy-side is aware of the identity of the other.
Model 2- "Buy-side to buy-side in partial size via one or two brokers as agent or principal"
- Buy-side C enters an order into the new venue matching engine to buy 8m of the Vodafone bond 1.25% of 26/09/2017. This may or may not have a limit price.
- Buy-side D enters an order into the new venue matching engine to sell 7m of the Vodafone bond 1.25% of 26/09/2017. This may or may not have a limit price.
In this case we have a size imbalance. Buy-side B is selling 7m and buy-side A wants to buy 8m. The issue is then - will buy-side A be willing to trade on a reduced size of order or will they wait for a seller of 8m? If they are willing to trade in 7m then the process looks like "Model 1"
Model 3 - Buy-side to retail investors via one or two brokers as agent or principal"
- Buy-side E enters an order into the new venue matching engine to buy 8m of the Vodafone bond 1.25% of 26/09/2017. This may or may not have a limit price.
- Retail investors 1 to 80 enter orders into the new venue matching engine to each sell 100k of the Vodafone bond 1.25% of 26/09/2017. These orders may or may not have a limit price.
When aggregated in size the retail investors match the institutional size order. At this point the institution needs to make a bid to take all 8m of the bonds at one price.
This will require a bank or other institution to face the 80 retail investors since most real-money long-only institutional asset managers will have no desire to process 80 tickets for a trade of 8m.
Again, I see this as being anonymous and probably facilitated by one or more banks. So the retail investors and buy-side E do not enter into direct negotiation, one or two banks act as agent or principal in this process.
Model 4 - Buy-side to near match buy-side via one or two brokers as agent or principal"
- Buy-side F enters an order into the new venue matching engine to buy 10m of the Vodafone bond 1.25% of 26/09/2017. This may or may not have a limit price.
- Buy-side G enters an order into the new venue matching engine to sell 10m of the Vodafone bond 1.625% of 20/03/2017.. This may or may not have a limit price.
Here we return to the process advocated in "Matching engine for corporate bonds?". As suggested, the new venue matching engine will maintain a comprehensive database of instruments.
When an order for a specific bond is received the order will contain a number of fields - such as ISIN, Size, Side - the usual economic information needed for any order. For this new venue though there will be a set of new attributes:
- Will accept less than full execution?
- Minimum % of order accepted
- Limit price
- Publish limit price
- Will accept attribute based near matches
- Maximum number of contra-trades accepted
So, when the order from buy-side F is received, if the correct flags are set, then the matching engine will look for possible contra-trades that are for a different ISIN.
In the event of a near match being found the matching engine will transmit information to both parties to allow them to evaluate the alternative on offer. This would be rather like a traditional equity market IOI workflow.
So, what's different here?
- No one is disintermediated.
- Retail and institutional can trade against each other if they choose.
- Near matches are possible based on configurable parameters.
- Anonymity is controlled closely
Of course, that's just an overview and there are a lot of details to be resolved, but most of the key features are listed here.
Comments
Post a Comment