Why NBBO Economics Have Become Distorted

We recently studied how trades speed around the marketplace — typically starting from a broker order in Secaucus, traveling at the speed of fiber to take orders on exchanges around the market, then causing a reaction function at the speed of microwave.

Today, we are looking at how quote updates typically flow around the market. 

What we find is that primary listing exchanges set the new National Best Bid and Offer (NBBO) most of the time. Then, most venues see orders arriving at a fairly consistent rate over time, but some venues have a rapid increase in quotes at the new NBBO within the first millisecond.

Whether that’s good or bad for market structure and routers is an interesting question.

Listing venues set the most NBBOs

We look at who sets new NBBOs using venue timestamps across all exchanges. That removes any delays reporting new trades back to the SIP. 

We see that primary exchanges set the new NBBO the majority of the time. That shouldn’t be surprising, as primary exchanges also need to compete for company listings, and companies compare them based on things that reduce their costs of capital, such as consistent quotes and tight spreads.

For example, Nasdaq market makers set quotes in Nasdaq-listed stocks close to 58% of the time. In contrast, all other exchanges combined improve the NBBO for Nasdaq stocks less than 43% of the time.

Chart 1: More than half of the time new NBBO price is set by the primary listing exchange 

Other venues join NBBO at different rates and speeds

What does tend to happen more consistently is that other venues join (or copy) the NBBO prices that have been set on the primary.

Importantly, we don’t see all venues rushing to copy the NBBO quotes at exactly the speed of light. Instead, most exchanges see a consistent arrival of new orders as time (x-axis) progresses.

But there are a few interesting exceptions:

• Sharp jumps show a rapid copying of new quotes. LTSE, and later IEX, stand out, with both having a sharp jump in copy quotes both well within 1 millisecond.
• Height of the line shows that IEX stands out, especially for more liquid stocks in the Nasdaq-100 (Chart 2a), where it has more quotes copying the NBBO than any other exchange — by a significant margin — despite several other exchanges actually providing the market with more liquidity (ARCA, EDGX and BATS).
• Line thickness shows the breadth of stocks in the universe where quotes are copied. IEX and LTSE stand out again, with copy-quotes on far more stocks, giving the appearance of widespread liquidity to investors, especially versus their trading market share (Chart 3).

Chart 2a: IEX send orders copying NBBO quotes for liquid (Nasdaq-100) stocks

Interestingly, the patterns change very little when we look at less liquid stocks.

Given there are about 3,300 companies in the chart below (versus just 100 in the chart above), the increase in total quotes is relatively small. It’s probably true that, with these smaller companies, they are also less profitable to quote. But these are also the tickers that need liquidity support the most. Even the SIP revenue allocation formula is tilted in favor of quotes in these stocks.

What the data shows is that for these less liquid stocks, IEX and LTSE experienced an even sharper jump in the number of copying quotes. However, as you will see in Chart 3, these large numbers of quotes don’t eventually lead to trades.

Chart 2b: There is less interest in copying quotes for less liquid, smaller-cap, stocks

Is this good or bad for market structure?

This matters in different ways to different participants.

For brokers and traders, it reduces the exclusive need for speed to be at the top of the queue. That’s because, as long as you can find a venue with no order on that venue, you can be at the top of their queue. That, in turn, helps increase spread capture and profitability for traders on those exchanges. 

But fragmentation adds other costs for brokers, including more connections and more complicated routing. It also adds costs to investors through higher opportunity costs.  

There are other costs of diluting queue priority, too.

The system doesn’t reward competitive quotes that lead to trades

More importantly, fragmentation of quoting is bad for the market makers actually setting the quotes in the first place. Their business is to profit from spread capture, but copy quotes make that less likely.

Perhaps even worse, the way regulated data economics works in the U.S. adds to inefficiencies. The SIP revenue allocation formula was designed to reward quotes and trades “equally” regardless of who set those prices more often. Back before there were over a dozen exchanges, with some protected quotes using speed bumps, it was set up to reward all quotes equally. 

Looking at recent SIP revenue shows that some exchanges seem to earn a lot of quote revenues, without actually doing any trading. In some exchanges, there are clear costs to exchanges, like rebates, to reward market makers providing those quotes, but in other cases the cost to exchanges and benefits to market makers are less transparent.

Chart 3: LTSE and IEX send large number of orders copying the NBBO quotes 

SIP quote revenues can add to tens of millions of dollars for some exchanges. This is an artificial incentive that supports fragmentation, without actually making the market more competitive and cheaper for investors.

We need to make sure the NBBO is good at protecting investors and issuers

Most of us seem to agree that the NBBO is important for investors and issuers. Academic research also suggests that tight spreads reduce costs of capital and increase liquidity. That, in turn, helps make the U.S. market more attractive than many markets around the world. 

We spend a lot of time arguing about the benefits of a “public” and affordable NBBO. Perhaps we also need to make sure that the economics also fairly reward the venues and traders setting those quotes. That could even help make markets more efficient.