One of the things that make bonds unique relative to stocks is the way compensation is collected for trading. The lions share of equity trades involve an explicit commission charged to the client. Be it a flat rate per ticket, as is common among online and retail brokerage, or a cents per share commission, as is common among institutional trading arrangements, this figure is explicitly reported on trade confirmations and has become a source of competition among brokers.
Fixed income securities generally incorporate a mark-up or mark-down in their price to compensate dealers for trading activities. That is to say that dealers usually adjust the price of the security rather than separately charge a commission. For example, they may be able to buy a bond for 100 in the Dealer to Dealer market. Rather than sell the bond to their retail customer at 100 with a commission of 0.10, they simply sell the bond to their customer at 100.10. The economics of the trade are identical either way. This is just meant to illustrate that there are differences in the industry conventions.
This nuance also leads to differences between the way trades are reported to the respective tapes. All equity trades are reported without commission (which is not to say without compensation as there are other ways to make money than just commissions, but that is well beyond the scope of this piece). However, many fixed income trades are reported with commissions embedded. Beginning July, 2016 the fixed income trade reporting services included a field that designates whether a trade has a commission charged separately or if the price includes a mark-up for customer buys or mark-down for customer sells.
As the names imply, a retail customer purchase that has been marked-up will have a higher price than the price the dealer paid to acquire bonds to sell to that customer. And vice versa. When a retail customer sells a bond, they receive a price that has been marked-down below where the dealer is able to sell the bonds to other dealers. There is nothing unusual about this activity. Again, this is just a way of charging commissions for brokerage services.
Where the activity becomes unusual is when the apparent compensation on a riskless principal trade turns negative. In that scenario, a broker is effectively paying a customer to do a trade. They lose money, but, presumably they make it up in volume. For the month of May 2017 BondWave calculates that 163 riskless principal trades were initially reported with a negative mark-up or mark-down. Purchases by retail customers were executed at lower prices than where the bonds were sourced in the dealer to dealer market. And sells to retail customers were executed at higher prices than where the bonds were sold in the dealer to dealer market.
Of course, the simplest explanation for this is likely to be the correct explanation. It is most likely that these trades were simply reported incorrectly. However, the table showing the top 10 negative markups/downs also shows that 5 of the trades were corrected anywhere from 1 to 20 days after the initial trade report. The corrections either fixed the price discrepancy or changed the side of the market so that the mark-up/down calculated becomes positive.
However, of the top 10 negative mark-ups and mark-downs, 50% have thus far gone uncorrected. Why would that be? Again, the simplest explanation. There are few consequences to making a mistake when reporting trades to the fixed income tapes, so there are few incentives to fix mistakes.
This is a topic we touched on in A Land Where Typos Still Exist. When it comes to trade reporting for corporate and municipal bonds, with a single exception (dealer to dealer trades in corporate bonds), there is no connection between the price or size reported to the tape and the price or size used for settlement purposes. So, if there is a mistake in a trade report there is rarely an economic incentive to repair the data.
There are clear negative implications to this system. The fixed income trade tape is used by regulators to monitor for potentially abusive practices and soon it will be used to calculate and report to retail customers via trade confirmations the estimated mark-up/down earned by their dealers. Since the confirmation is a legal document any inaccuracies are potentially litigable.
In our first two pieces we have focused on what appear to be uncorrected mistakes on the trade tapes. The first set of typos led to the appearance of negative bid/offer spreads. This piece focused on what we believe are likely typos leading to apparent negative compensation for brokers. In our next piece we will explore the consequences of having an opaque market structure for fixed income securities. Specifically, we will consider situations where the actual bid/offer spread is negative. And we will see that this situation can even exist for the most informed participants in the market, the dealers.
Paul Daley is Managing Director in BondWaves Information Lab