The FSA consultative document “Strengthening Oversight and Regulation of Shadow Banking, A Policy Framework for Addressing Shadow Banking Risks in Securities Lending and Repos” has a lot to chew on. This post will focus on the idea they floated on minimum repo haircuts.
The underlying issue is pro-cyclicality. Regulators fear that buoyant markets will encourage lower haircuts and that, in turn, will incentivize ever-greater amounts of leverage. Afraid of bubbles, regulators wish to lean against what will appear to be rational exuberance before it leads to irrational. Corporate bonds and securitization products were mentioned in particular as having a “procyclical feature”.
The report wrote, “…Minimum regulatory haircuts for repos and securities financing transactions (whether bilateral, tri-party or CCP) may limit the build-up of excessive leverage and reduce procyclicality in the financial system via the financing of risky assets, in particular by entities not subject to prudential regulation…” and “Haircuts should be based on the long-run risk of the assets used as collateral and be calibrated at a high confidence level to cover potential declines in collateral values during liquidation…” The long-run is defined “…the maximum price decline used to derive the applicable haircut should be calculated using a time series of price data that covers at least one stress period…”
Haircuts, the FSB writes, should also be used to mitigate “wrong way risk”, large position liquidity risk, and any FX mismatch risk. The FSA is contemplating two different types of haircuts levels. Mandated minimum haircuts might either be set at “…relatively high levels that may typically be closer to actual market practices in normal times, which therefore would be more likely to be used in transaction activity…” or “…set numerical floors at a lower level, making clear that they are intended as a backstop to prevent excessive leverage…”
Below is a chart from the report that illustrates some repo levels.
The FSA “…proposes to exclude cash collateralized securities borrowing transactions where (i) the purpose of the transaction is to borrow the specific securities and (ii) the lender of the securities reinvests the cash collateral into a separate reinvestment fund and does not use it to finance the assets being lent…” So it looks like Sec Lenders can take a breath of relief.
There is some debate on whom exactly to include in the rules. The choices are (i) all market participants; (ii) exclude transactions between regulated financial intermediaries but include all else and (iii) regulate the exposures between regulated financial intermediates and other entities. The last option was recognized as the easiest to implement, although (i) and (ii) capture shadow banking activity – a hot button for sure — more thoroughly.
The FSB noted that the rules would have to be implemented globally in light of the ease of simply booking transactions outside of the reach of the rules. The report also recognized that Total Return Swaps also provide financing and need to be included. TRS “…can also be used to achieve similar economic objectives as repo and securities lending transactions (and possibly vice-versa). It is therefore desirable to harmonise implementation guidelines and numerical floors with the proposed regulation of non-centrally cleared derivatives…”
The report also acknowledged the problems of incorporating trade-by-trade haircut rules in an environment where haircuts are calculated on a portfolio basis (e.g. prime brokers). Good luck with that.
Wow. We don’t even know where to start here. First, pro-cyclicality is a macro / systemic issue. The fear is that the market as a whole becomes over leveraged and they don’t even know it. The FSB advocates for trade repositories and greater transparency, but it isn’t clear the information can be meaningfully processed much less interpreted in a timely manner. Even if trades are executed at the mandated haircuts, isn’t it the overall size of the market and position imbalances within that regulators should care about?
The haircuts on shorter instruments are lower than those on longer dated securities. That makes sense on a PV01 basis. But when one looks at how, for example, corporate bonds behave in a crisis; the conclusions might be thrown out the window. The jump to default risk in corporate paper means that when default risk rises, yield curves will invert and the bonds start trading like equity. No one cares what the yield is at that point, liquidity disappears. Longer dated corporate paper, on the other hand, often has some option value and retains better liquidity. That would argue for higher haircuts on shorter dated corporate paper versus longer bonds.
Looking at the haircut levels for securitized products, we wonder how they came up with those numbers. In the financial crisis, 16% for 5 year plus paper wouldn’t have come close to covering market declines on (then rated) AAA sub-prime backed bonds. Try 30% or 40%. And even then, finding a bid was torture.
We know that these haircuts are supposed to be minimums, but some dealer will peg their haircuts to those minimums and others, not wanting to lose market share, will follow. Minimums, when there is competition, often become maximums.
Perhaps one solution lies in copying the futures industry? Margin levels for futures can change on short notice based on market conditions. If the regulators want to mandate that markets control leverage, one way might be to be able to force changes in haircuts as markets change (even on term trades), keeping a floor to guard against procyclicality. Leveraged investors, knowing that haircuts will rise as a matter of course will keep more collateral on hand to avoid the risk of missing a margin call (and hence liquidation). This is what LCH already does on repos when market conditions become more volatile or individual members have credit problems (ok…rev up the flashbacks to MF Global about now). Some have called it “bad boy margin”.
Research has pointed out the irony of tri-party haircuts remaining stable during the financial crisis while bilateral haircuts were rising. It might be time better spent making sure that the stickiness in tri-party repo, caused in our opinion by the difficulty in changing documented-mandated levels, are more fluid.
We are wondering how all these objectives can be executed. Actually coming up with sensible haircut levels is the least of their worries. Most dealers probably already comply. But the cross-jurisdictional issues – insuring that trades cannot be done outside of the long arm of the law – are daunting. Regulating otherwise non or “lightly” regulated market participants will provide its own special challenges. Data collection, processing, and extracting actionable intelligence will require its own special ring of hell. Leaning against procyclicality is a fine concept, but getting it done is another thing.
A link to the FSB report is here.