Continuing my brief summaries of some of the papers presented in the December’18 Emerging scholars conference (see the first one), I bring a paper on how capital regulation affects the repo market by Antonis Kotidis and Neeltje van Horen.
I have talked about the leverage ratio several times in this blog. Ages ago I discussed a different paper looking at the same issue (here and here), and I also talked about my own paper looking at the derivatives market (here and here). This is a topic that I find particularly interesting given how the leverage ratio—and, in general, post-crisis regulation—differs from the previous regulations.
I had my issues with Allahrakta et al. (2018), especially in terms of identification. Since my posts in 2016 it has been published in the Journal of Financial Intermediation, which is a great journal. My main issue was that they did not have a clear shock to exploit. In fact, looking at the data, they observed a break in the series around the end of November 2012, when nothing happened. They concluded that dealers were realising at that time that the LR would affect their repo business. But this is a backward reasoning: they start by assuming the LR affects repo, and then they provide an explanation on why we see this effect happening when no regulation is being announced or implemented.
Kotidis and van Horen, on the other hand, pick a particular change of the leverage ratio in the UK which provides a much better identification. From January 2017 onward, the biggest banks in the UK had to calculate the leverage ratio by using a daily average rather than their situation at the end of the month. This is an important change. Most repo transactions are overnight; this means that if a requirement, such as the leverage ratio, depends on the state of the balance sheet at the end of the month, a bank can simply reduce the amount of repos for one day to reduce the requirement. This Bloomberg article, where I commented, picked on this issue.
Repo transactions are particularly taxing in terms of the leverage ratio due to their nature. They utilise balance sheet—and hence the leverage ratio applies—but they are secured and have a very low return. Therefore, they are a great candidate to be affected by this requirement.
From January 2017 onwards, as I was saying, UK banks could no longer use the window-dressing “trick”. In order to identify the effect, they look at clients that enter into (bilateral) repo transactions with a UK dealer and a non-UK dealer. This is important because different dealers might have different clients, and thus if we observe that the total repo portfolio of UK dealers drops after the regulation it could be due to their clients demanding less, not because the dealers themselves reacted. But by fixing the client (i.e., by using client fixed effects), then the effect that they pick is much more likely to come from the dealer.
They find that UK dealers did reduce liquidity in the repo markets after this reform, and this affected mainly small clients rather than the big ones. The reduction comes in terms of volume and pricing, but not haircuts and maturity. They also find evidence that non-UK dealers stepped in to substitute for the drop.
This is a great addition to the set of papers looking at the effects of post-crisis regulation. We are still figuring these things out—many of the new regulations have just been introduced in the past year or two—but we are understanding more and more. Hopefully central banks will not lose their appetite to figure out which regulations are working as intended and which ones might have unintended consequences.