By Aaron Krowne
Jim Willie’s latest public article, focusing on US Treasury interest rate manipulation, mentioned the fails-to-deliver (FTD) scandal of late 2008, extending (in at least some sense) to the present.
This got me wondering to just what extent FTDs in Treasuries are still occurring.
(For the uninitiated, FTDs are “market events” that signify that a trader who has sold a security and has promised to deliver it to the buyer at some specified future time fails to actually do so. Yes, this even happens with US Treasuries, and it happened in a big way during and around the 2008 market crash. You might think this should be just outright illegal, as it represents in effect counterfeiting an official US government obligation. But then you’d be a stickler for “the rules” and thus uncool. Ahem…)
So, I poked around a bit, and then wandered over to the DTCC (the Depository Trust Clearing Corporation, the quasi-governmental entity that for some reason gets to opaquely lord over all over-the-counter securities trading) to look at their data, which they are now publishing only in response to public unease over the 2008 flurry of FTDs. They have a nice cute little interactive graph app which tells you far less than you might expect — starting with the fact that it only goes back one year (if you wanted to see the 2008 crash in comparison to now or any time in the past 3 years, well, remember the “uncool” remark, poindexter?)
I found a couple further problems/unsettling points to wonder about with this data, however. Here is my screen capture of the past 1 year of data, as charted on the DTCC’s page with some of my own annotations.
The first thing that jumps out at you is that in March of this year, for some reason (not explained), the average daily level of failures triples, from a bit over two and a half billion dollars a day, to over seven and a half. Hmm. No notes about this from the DTCC, or anyone else for that matter, which is odd given that things in the debt world (especially US Treasury) are supposed to be getting calmer as we go along, and that the DTCC has supposedly implemented penalties that are supposed to make these failures less common.
The other weird thing is that the chart mysteriously gaps up in Y-scale (billions of dollars) from $7.65B to $19B. Now, maybe the DTCC just wanted to emphasize the data at the low end of the scale, but this isn’t really a technically legitimate way to do that. It leaves a gap of about $12B in the vertical scale, such that a $19B daily level looks almost exactly the same as a $7.6B level on the chart (an example is highlighted in the chart above, which should actually say the bars are over 2x apart, not “almost”). I’ve never heard of this being done before (not on purpose at least). The legitimate way to emphasize the low-end of the vertical scale is to use an exponential axis, which would actually be really helpful with this “spikey” data, as my plain version of the same data shows:
(click image for full sized version)
(Note that left to right on the X-axis is going back farther in time, the opposite of the DTCC chart)
I downloaded the source data for that from the DTCC. Interestingly, it only goes back 1 year as well.
Really, guys? Is this just a half-assed attempt at transparency, or is someone hiding something?