Let's Talk About: Basel III

The Fed last night unleashed eight zillion pages of Basel III implementation on the universe and I'm tempted to be like "open thread, tell us about your hopes and fears for capital regulation." So do that! Or don't because it is super boring, that is also a valid approach. Still I guess we should discuss. Starting slow though. Banks have to have capital, meaning that they have to fund some of their assets with things that are long-lived and loss-absorbing, like common equity, rather than with things that have to be paid back soon and at face value. The reason for this is that the rest of banks' assets are funded with things that we really do want to be paid back soon and at face value, like deposits, and if the value of those assets declines you don't want those deposits to be wiped out. The rules say that you need capital equal to a percentage of your assets. The game is deciding (1) what that percentage is, (2) what is capital (proceeds from selling common stock, and actual earnings, yes, but, like, deferred tax assets?), and (3) how you count assets (you might want more capital to shield you from losses in, say, social media stocks than you would to shield you from losses in Treasury bonds, so regulators use "risk-weighted assets," so that $1 of corporate bonds counts as $1 of assets, $1 of Treasuries counts as $0 of assets, and $1 of Facebook stock counts as $3 of assets*). Anyway, here are the required capital levels:
Author:
Publish date:

The Fed last night unleashed eight zillion pages of Basel III implementation on the universe and I'm tempted to be like "open thread, tell us about your hopes and fears for capital regulation." So do that! Or don't because it is super boring, that is also a valid approach. Still I guess we should discuss.

Starting slow though. Banks have to have capital, meaning that they have to fund some of their assets with things that are long-lived and loss-absorbing, like common equity, rather than with things that have to be paid back soon and at face value. The reason for this is that the rest of banks' assets are funded with things that we really do want to be paid back soon and at face value, like deposits, and if the value of those assets declines you don't want those deposits to be wiped out.

The rules say that you need capital equal to a percentage of your assets. The game is deciding (1) what that percentage is, (2) what is capital (proceeds from selling common stock, and actual earnings, yes, but, like, deferred tax assets?), and (3) how you count assets (you might want more capital to shield you from losses in, say, social media stocks than you would to shield you from losses in Treasury bonds, so regulators use "risk-weighted assets," so that $1 of corporate bonds counts as $1 of assets, $1 of Treasuries counts as $0 of assets, and $1 of Facebook stock counts as $3 of assets*).

Anyway, here are the required capital levels:

All banking organizations would be subject to the following minimum regulatory capital requirements: a common equity tier 1 capital ratio of 4.5 percent (newly introduced requirement), a tier 1 capital ratio of 6 percent (increased from the current requirement of 4 percent), a total capital ratio of 8 percent of risk-weighted assets (unchanged from the current requirement), and a tier 1 leverage ratio of 4 percent.

Common equity is longer-lived and more loss-absorbing than tier 1, which is longer-lived and more loss-absorbing than other capital, but the gist is that for every $1 of equity-like stuff you can make about $12.50 of loans, adjusted for risk. There are tons of supplements and caveats to that, including most notably a 1-to-2.5% surcharge for globally important banks. Banks tend to whine about wishing they had to have less capital, because the lower their capital requirements the more of their profits they can give back to shareholders, and shareholders, being shareholders, like money. Some people think that this whining and the reasoning behind it are bullshit but that is a somewhat weird view because I have met shareholders and they do in fact like money.

There are zillions of things here so let's pick one at random. A change in the new rules is that they will require the mark-to-market of available-for-sale securities through capital. One simple thing to think is that banks should be encouraged to do the business of maturity transformation by taking in demand deposits and lending those deposits out in long-term loans, and that this business sits awkwardly with the need to have capital equal to a constant percentage of the market value of those loans. If you take $90 in deposits and raise $10 in capital and lend $100 for 10 years at 5% interest, and suddenly interest rates go to 6% because the economy feels riskier,** then all of a sudden your loans are worth like $93 and you have only $3 in capital.*** And if rates go to 6.5% you are in some sense bankrupt. Even though in some other sense you are not - if nobody panics and withdraws their deposits, and everyone pays back their loans with interest, you actually end up perfectly solvent and with $10 in capital. Plus I guess you get the interest.

There is no really good way to address this; it is just sort of a core problem of banking. If you say "mark everything to market," then market fears have a procyclical destabilizing effect, as increases in rates lead to bank undercapitalization lead to panic lead to bankruptcies lead to increases in rates etc. If you say "mark nothing to market and hold everything at historical cost" then banks' accounts are pure fiction and you have no idea how if they can withstand losses and that leads to panics too. So you try to have a medium where you can believe the accounts but they don't blow up at any sign of trouble.

The way that medium works is, loosely, that some things get put on the mark-to-market side, where they are things that banks need to be able to rely on to sell to cushion losses and if they can't be sold for $100 any more the you have to treat them as being worth $93, and other things get put on the historical-cost side, where they are things that are in some sense protected from the market by the intermediation of banks, and so it doesn't matter if today you can only sell them for $93, they still go on at $100. This division is sort of made on the basis of investment intent (if a bank plans to hold the thing to maturity, then it doesn't mark to market, if it plans to use it for trading, then it does) and also sort of on pragmatic computation bases (if it's got a market to mark it to, mark it to market, if it's a nontraded direct loan to a client, don't). Available-for-sale securities are securities that you're not actively in the business of buying and selling each day (trading securities, which are valued on a mark-to-market basis), and that you haven't just chucked in your vault until they mature (hold-to-maturity investments, which are valued at historic cost, basically), but are somewhere in between. They're available for sale but not exactly up for sale. Under US accounting rules mark-to-market gains and losses on these securities don't go through income, but they do show up as "other comprehensive income."

In addition to the intent and pragmatic reasons for classifying things as mark to market or not, there's also a social-purpose basis, where there are some things that we want banks to shield from the market, and other things that we don't. In some ways the London Whale, who focused so much attention on the Volcker Rule, is better viewed as a parable about available-for-sale securities. The Whale swam in mark-to-market waters, while his bosses waded in the warm beaches of AFS. But when he lost a bit of money, which flows through accounting income and reduces capital, they stepped in to sell some appreciated AFS securities to**** boost accounting income and capital. And they had $375 billion of AFS securities, of which some $200bn was in mortgage-backed securities (half of it agencies), $50bn was in foreign government bonds, and $60bn was in corporate bonds.

JPMorgan likes those things for various reasons; one reason is that they are capital advantaged. Here, for example, is the risk weighting applied to residential mortgages by the standard approach in the proposed rules (Category 1 is basically well underwritten first-lien, Category 2 is other):*****

Corporate loans are even simpler - in the standard approach, they get 100% risk weight (pages 27-28). That means that, for every dollar that a bank is lending to businesses or homeowners, it has $0.35-$2 of risk-weighted assets and so needs at least 2.8 to 16 cents of total capital (at 8%). Compare that to, say:

  • buying Spanish government bonds, which get an incredible 0% risk weight even under the new rules,****** or
  • buying agency MBS, which get a 20% risk weight under the new rules (page 23), or
  • buying non-agency MBS, which ... under the old rules often looked better than making mortgages, but under the new rules looks worse.*******

So if you want to optimize for capital, those 6.04%-yielding 10-year Spanish bonds look pretty good, while a 30-year mortgage with a 4 handle looks less good. The bonds and MBS have other advantages as well: a handful of people in New York and London can, when they're not busy screaming at each other, put $375bn to work pretty easily (though not without some liquidity concerns), whereas actually lending out $375bn to homeowners and companies requires lending officers and underwriters to go out and talk to those people. If you are a bank, there are many considerations that drive you out of banking and towards trading.

If you think that a goal of capital regulation should be to nudge banks toward banking, then the size of JPMorgan's AFS portfolio, and its arguable use as a trading rather than investing portfolio, is troubling. One thing that you might think is along the lines of:

  • I get that banks should make illiquid loans to real humans and companies and hold capital against them in some static-ish way, because that's kind of what banking is,
  • and I get that banks should have liquid investments that they can use to cover losses and hold capital against them in some mark-to-market-ish way, because that makes them safer for depositors,
  • but why do they need to have like 20% of their assets in sort-of liquid securities that are sometimes sold to cover losses but nonetheless have low and static capital requirements?

And one sensible compromise you might do there is to leave the risk weightings for agency MBS and whatever low, so that banks are encouraged to put their rainy-day money in them rather than junk bonds, but to mark those securities to market so that, if they're actually there as rainy-day money, they accurately measure how much rainy-day money there is. And if at the margin that has the effect of pushing banks out of mark-to-market secondary-market investing and toward non-mark-to-market lending, that may be exactly what you want.

Or not, of course, because sometimes lending goes terribly wrong too.********

So, yeah, hopes and fears. Other good topics for thought include (1) is this news given that nothing happens until 2019 and (2) how do we feel about efforts to get credit ratings out of risk-weighting? I feel good about it - replacing arbitrary and silly classifications like ratings with more candidly arbitrary and silly classifications like OECD rankings and mechanical subordination formulas for securitizations seems helpful in a Nassim-Taleb-not-pretending-to-know-what-you-don't-know sort of way.

Fed Basel III proposals and whatnots [FRB]
Federal Reserve set to unveil capital proposals [FT]
Capital Rule Is One Size Fits All [WSJ]

* See pages 27-28, 19, and 88-89 of the proposed rules. This statement is only approximately true but good enough for our purposes.

** In my simple model, "interest rates" sort of means spreads, shut up.

*** I pulled out my HP 12C to do this bond math and had a twinge of regret about not taking the CFA Level II last weekend. Thanks to everyone who offered to let me cheat for them though!

**** Tendentious "to." "With the possibly unintended effect of," perhaps.

***** Important: this is the Standardized Approach, basically for smaller banks; JPMorgan would use the Advanced Approach, which is more complicated. Numbers in the text are illustration from the standardized one, to get a directional sense; they're not what big banks actually do.

****** See pages 19-22 here. The new rules will use the OECD country risk classifications; countries with the best classifications (0 or 1) get a zero risk weight. The Fed drily notes:

The agencies recognize that CRCs have certain limitations. Although the OECD has published a general description of the methodology for CRC determinations, the methodology is largely principles-based and does not provide details regarding the specific information and data considered to support a CRC. Additionally, while the OECD reviews qualitative factors for each sovereign on a monthly basis, quantitative financial and economic information used to assign CRCs is available only annually in some cases, and payment performance is updated quarterly. Also, OECD-member sovereigns that are defined to be “high-income countries” by the World Bank are assigned a CRC of zero, the most favorable classification. Despite these limitations, the agencies consider CRCs to be a reasonable alternative to credit ratings for sovereign exposures and the proposed CRC methodology to be more granular and risk sensitive than the current risk-weighting methodology based on OECD membership.

That "high-income countries" thing means that Greece has a zero classification, which would give it a zero risk weight except that the proposed rules have an overriding 150% risk weighting for sovereigns who've defaulted in the last five years. So they went from 0 to 150% risk weight in a day, suggesting that OECD ratings have even more goofy-jump-risk than S&P/Moody's ratings.

******* Read complicated rules at pages 80-83 here, and recall that's the simplified approach for small banks. The useful math-free takeaway is this:

The SSFA methodology begins with “KG” the weighted-average risk weight of the underlying exposures .... At the inception of a securitization, the SSFA as proposed would require more capital on a transaction-wide basis than would be required if the pool of assets had not been securitized. That is, if the banking organization held every tranche of a securitization, its overall capital charge would be greater than if the banking organization held the underlying assets in portfolio. The agencies believe this overall outcome is important in reducing the likelihood of regulatory capital arbitrage through securitizations.

******** This book contains the best factoid ever, which is that Kerry Killinger would ask others at WaMu "What would a real company do in this situation?"

Related

Citi Will Try The Stress Test Again With A $9bn Stock Buyback

More stress tests, bleargh. I guess the news is that Citi "failed", though I can't get all that excited by that because it didn't exactly "fail" in the sense of now it's being forced to raise capital / broken up / burned to the ground. Instead it failed assuming it follows the capital plan it submitted to the Fed, which is clearly a capital-lowering rather than capital-raising plan. I ballpark it at $10bn of share repurchases and dividends,* which is ... well, it's pretty big for Citi. So they can just not do that then. Or not do quite as much of that, which seems to be their plan: In light of the Federal Reserve’s actions, Citi will submit a revised Capital Plan to the Federal Reserve later this year, as required by the applicable regulations. The Federal Reserve advised Citi that it has no objection to our continuing the existing dividend levels on our preferred stock and our common stock, and we plan to do so, subject to approval by the Board of Directors each quarter. The Federal Reserve also advised that it has no objection to Citi redeeming certain series of outstanding trust preferred securities, as Citi proposed in its Capital Plan. We plan to engage further with the Federal Reserve to understand their new stress loss models. We strongly encourage the public release of these models and the associated benchmarks and assumptions. We believe greater transparency in this process will best serve all banking institutions and their shareholders as well as the international regulatory community and market participants, and will encourage a level playing field globally. There are at least two ha! moments in that snotty last paragraph. First there's the fact that the Fed had planned to release the stress test results on Thursday and got gun-jumped by Jamie Dimon. So much for Fed transparency. But also, specifically, as people are all running around suing each other about the Fed maybe kind of encouraging bank CEOs to hide material information from investors, it is odd that the Fed would have the stress test results and sit on them for two days. Imagine the scenario where Jamie Dimon, Vikram Pandit, and the Fed all know that JPM passed and was going to do a largeish buyback, while Citi failed and was going to do a ... I guess somewhat smaller buyback - and they didn't tell anyone from today until Thursday. If you sold JPM to buy C today, wouldn't you be kind of annoyed?**

Banks Prove That They Are Not Too Big To Fail By Saying "We Can Fail" On A Piece Of Paper, Moving On

One way you could spend this slow week is reading the "living wills" submitted by a bunch of banks telling regulators how to wind them up if they go under. Don't, though: they're about the most boring and least informative things imaginable and I am angry that I read them.* Here for instance is how JPMorgan would wind itself up if left to its own devices**: (1) It would just file for bankruptcy and stiff its non-deposit creditors (at the holding company and then, if necessary, at the bank). (2) If after stiffing its non-deposit creditors it didn't have enough money to pay its depositors it would sell its highly attractive businesses in a competitive sale to willing buyers who would pay top dollar. This seems wrong, no? And not just in the sense of "in my opinion that would be sort of difficult, what with people freaking out about JPMorgan going bankrupt and its highly attractive businesses having landing it in, um, bankruptcy." It's wrong in the sense that it's the opposite of having a plan for dealing with banks being "too big to fail": it's premised on an assumption that the bank is not too big to fail. If JPMorgan runs into trouble that it can't get out of without taxpayer support, it'll just file for bankruptcy like anybody else. Depositors will be repaid (if they're under FDIC limits); non-depositor creditors will be screwed just like they would be on a failure of Second Community Bank of Kenosha.

TARP Charts!

The Federal Reserve has this new paper out about TARP that does a bit of highly suggestive eyebrow raising about some banks that shall remain nameless. They start from the awkward fact that TARP wanted everything in one bag but didn't want the bag to be heavy, or as they put it: The conflicted nature of the TARP objectives reflects the tension between different approaches to the financial crisis. While recapitalization was directed at returning banks to a position of financial stability, these banks were also expected to provide macro-stabilization by converting their new cash into risky loans. TARP was a use of public tax-payer funds and some public opinion argued that the funds should be used to make loans, so that the benefit of the funds would be passed through directly to consumers and businesses. So you might reasonably ask: were TARP funds locked in the vault to return the recipient banks to financial health, or blown on loans to risky ventures, or other? Well, here is Figure 1 (aggregate commercial and industrial loans from commercial banks in the U.S.): So ... not loaned then. But that's not important! The authors are actually looking not primarily at aggregate amounts of loans but at riskiness of loans and here's what they get: