Banking Really Hasn't Changed Much Since the Panic
The concept of what I call "math as money" is often a difficult one to grasp, particularly when still anchored to the traditional view of money and banking. A bank is supposed to be a money multiplier, a trusted institution that takes in currency or hard money, puts it in a vault and then lends it out fractionally in the form of deposits. As economists still believe to this day, this type of system gives the central bank enormous powers and responsibilities as the sole determinant of "base money" (where exogenous hard money does not exist). That is how quantitative easing is largely viewed, as a tremendous increase in the amount of that so-called base money.
This isn't actually how banks work, thus by extension money, and it hasn't been that way for a very long time. As if it weren't completely obvious by now, monetary authorities have no idea what they are doing in large part because they are stuck in the past, thinking about ways to control a system that hasn't truly been the marginal source of additional resources and expansion since maybe as far back as the 1950's. The introduction of eurodollars was not just as an added marketplace for dollars trading overseas, it was the evolution of money itself where bank balance sheets would be the overriding factor for expansion or, as now, contraction.
The way in which banks exercise control over their own resources has likewise evolved, aided in large part by the adoption of regulations that through the last decades of the 20th century emphasized assets more than liabilities. These were the adoptions of capital ratios but more so the regulatory arbitrage opportunities by which banks could manage them. Before the 1980's, bank supervision was accomplished by experienced bank examiners using judgments, not numbers. The 1978 FDIC Manual of Examination Policies, for example, instructed the agency's examiners to be more holistic in their approach because:
"...capital ratios...are but a first approximation of a bank's ability to withstand adversity. A low capital ratio by itself is no more conclusive of a bank's weakness than a high ratio is of its invulnerability."
There is a great deal of timeless truth to that statement, but in the case of bank supervision it marked something of an endpoint for that approach. The industry had already undergone radical changes right down to the firms most representative of the traditional format itself: the S&L's and thrifts. Whereas prior to the 1970's banking had appeared highly stable with very few failures, the Great Inflation would change that, unleashing not just more of them but featuring larger firms like Franklin National Bank in 1974 and First Pennsylvania in 1980.
The causes of both higher bank risks and failures were viewed somewhat properly as the overhang of 1930's regulation, primarily Regulation Q's impositions at a time when interest rates were rising (overly plentiful money in the real economy). But there was much more to it than that, as monetary evolution was arguably a much bigger factor and one that was poorly understood often by the banks deploying the fruits of it. I wrote in April last year:
"As late as 1972, S&L's were absent from the nascent repo market, but by 1978 they had added about $4.8 billion in repo liabilities, totaling up to about 1% of the entire liability structure. By 1984, around the time Home State was collateralizing lending with securities that didn't exist, repo liabilities had expanded almost tenfold to $43 billion, or slightly less than 4% of a rapidly expanding balance sheet base. Total repo reliance would peak in Q1 1989, with almost $105 billion in liabilities, or just shy of 7% of total funding."
The S&L crisis itself was predicated in very large part upon those firms that had chosen to move away from traditional banking. The FDIC reported in 1984 that, "thrifts with annual growth rates of less than 15 percent had more than 80 percent of their liabilities in traditional retail deposits (generally in accounts of less than $100,000), the comparable figure for thrifts growing at rates in excess of 50 percent per year was only 59 percent." Those banks that were rapidly expanding were those moving to wholesale funding formats. As that was happening, you can almost appreciate just how the supervisory model of bank regulation was increasingly out of place; it no longer fit the times.
Rather than update and retrain examiners for modern money, regulation was left largely to the banks themselves. Politicians and "experts" didn't believe that was the case, of course, as that was the whole point of moving to the math. In other words, by instituting "constraints" like capital ratios it was believed that banks could do what they wanted to do while still allowing for a relatively solid determination about whatever that might be. If regulators couldn't understand and appreciate how they were getting funded (the liability side) then at least they would track what banks did with all that funding regardless of where it came from.
The regulatory environment began to adopt this approach in 1981 coincident to the double dip recession that marked the end of the Great Inflation. Federal banking agencies introduced explicit capital requirements, though initially the definition of what could be counted as "capital" varied by agency. These preliminary steps intended to measure a crude leverage ratio of "primary capital" to average total assets. The Federal Reserve and OCC began with a minimum ratio of 6% for community banks and 5% for larger regional firms (national banks were at the time largely still prohibited).
This mathematical approach was further cemented by the International Lending and Supervision Act of 1983. The law pushed for further uniform definitions and standards to be developed by 1985, including a closer model to that which was being developed in Europe, including the "risk bucket" approach that the Federal Reserve Bank of New York had already proposed. These would all harmonize into the Basel Capital Accord in 1988.
Initially, the bucket model seems straightforward and even logical. A "riskier" asset as determined by regulators (the first problem) is given a higher risk-weighting (RWA - risk-weighted assets), which sounds appropriate enough. The initial Basel framework included four buckets but over time those have been expanded and the rules governing what goes where have been altered to give banks far more discretion in deciding RWA treatment.
The consequences were and still are enormous. For example, in the initial framework residential mortgages were assigned to the 50% bucket. However, "claims or guarantees" provided by "qualifying" banks and entities (primarily, at the start, the GSE's) would be assigned instead to the 20% bucket. Thus, a bank for a given amount of statutorily-defined "capital" could hold two and a half times more assets if they could "somehow" define those assets by the "claims and guarantees" of "qualifying" counterparties. In many ways, it was just that simple, as through the 1990's and 2000's the ability to manage these buckets became paramount for balance sheet expansion (capital efficiency).
By the initial outbreak of the crisis in 2007, the ability to manage RWA was as monetary as whatever futility the Federal Reserve or any other central bank offered or would offer. I wrote earlier this month about one specific but highly illuminating example:
"The answers start in the footnotes. In AIG's 2007 Annual Report filing, the company reported on Page 33 deep within the recesses of so much mind-numbing but relevant minutiae that it had written $527 billion, more than half a trillion, gross notional exposure into its AIGFP subsidiary super senior credit default swap portfolio. While that should have been enough of a wake up on its own, it was the disposition of the portfolio that truly starts to explain the hidden panic of 2008. Of that $527 billion gross notional, AIG detailed that $379 billion ‘were written to facilitate regulatory capital relief for financial institutions primarily in Europe.'"
Without knowing anything about credit default swaps, the intent is relatively easy to understand in general terms. These European banks were not interested in hedging per se, rather they were interested in AIG's ability to help them to "... appear to be less risky so that they could be specified as such in the regulatory framework of Basel II, leading them to be able to hold (vastly) more assets for a given amount of ‘capital.'" Math as money - offered CDS from a "qualifying" insurance company, the highly-rated insurance firm AIG, meant that European banks could reduce their RWA's by manipulating the bucket to which those assets would belong accomplishing a still unknown though likely massive systemic balance sheet expansion throughout the middle 2000's.
To hear Ben Bernanke or now Janet Yellen tell it (or Mario Draghi), all that is behind us now as the financial system is supposedly far more "resilient" for having learned the hard way. Policymakers don't often go into specifics, of course, typically keeping these assertions general enough to accomplish their goal of assuring the public without being tied to specific ways those assurances could be so easily disproven. "Math as money" is one of the numerous ongoing gray areas.
The "math" part is what was offered by AIG in the form of CDS, but that was just one example of how such math could be turned into money. The US Senate Permanent Subcommittee on Investigations (a chilling title, even if occasionally warranted) issued a report on JP Morgan's 2012 London Whale episode that shows it in action.
"In December 2011, JPMorgan Chase instructed the CIO to reduce its Risk Weighted Assets (RWA) to enable the bank, as a whole, to reduce its regulatory capital requirements. In response, in January 2012, rather than dispose of the high risk assets in the SCP - the most typical way to reduce RWA - the CIO launched a trading strategy that called for purchasing additional long credit derivatives to offset its short derivative positions and lower the CIO's RWA that way. That trading strategy not only ended up increasing the portfolio's size, risk, and RWA, but also, by taking the portfolio into a net long position, eliminated the hedging protections the SCP was originally supposed to provide."
Without the ability to hedge (even badly) the CIO office would have been forced to sell assets to meet the bank's reduced RWA requirement or budget. The easy and fluid ability to lay off risk on some other "qualified" counterparty, whatever that might mean, has immediate monetary consequences, but most especially during times where the inability to hedge into math as money is not easy and fluid. This is the functional, modern definition of liquidity, having, importantly, absolutely nothing to do with the bank reserves created as a byproduct of "quantitative easing."
It all further demonstrates how regulations in this way conform to the FDIC's 1978 Manual warning that, "A low capital ratio by itself is no more conclusive of a bank's weakness than a high ratio is of its invulnerability." In fact, at that time in 2012 there was something of a pandemic of bank math. Banks primarily in Europe were, as Deutsche Bank's then-co-CEO Anshu Jain said in September 2012, "under tremendous pressure to reduce risk-weighted assets." Left unsaid was just how much that might have been due to the fallout from AIG's (and others) downfall leaving a huge, systemic hole in offered math.
Deutsche Bank, for one, began immediately to correct the problem. Just a few months later in January 2013, Jain and his management team were pleased to announce that the bank had successfully cut €55 billion in RWA in the fourth quarter of 2012 alone. The effect was an immediate increase in the bank's core capital ratio to 8% from less than 6% (and insufficient) at the end of 2011. Of that €55 billion "improvement", €18 billion was derived from the "roll out of advanced models", an additional €8 billion was due to "data improvement exercises", and a further €15 billion from "portfolio optimization." Deutsche never provided much detail about that last one apart from language indicating "optimizing risk mitigation" from which we can infer at least more hedging.
Altogether, these new "maths" had the combined effect of an €8 billion capital infusion without a single euro having been committed or decommitted; all of it was balance sheet, well, voodoo. And though Deutsche was not alone in its RWA rejiggering at the time, again it was endemic throughout European banks but not only them (Morgan Stanley was was noted, too), it was perhaps its most blatant user. At that time, January 2013, DB's total RWA had been reduced to just 19% of its total assets. In other words, the full effect of all this math was to suggest that though Deutsche Bank still held an enormous amount of assets they weren't risky - at least in terms of regulatory treatment. As to actual risk, that may be another matter.
The RWA reduction compared to RBS, for example, whose RWA equaled 35% of total assets around the same time; Barclay's was at 24%; BNP Paribas 29%; Credit Suisse 23%. About the only bank in Europe in DB's neighborhood of regulatory "de-risking" by math was UBS at 15%. The Swiss bank, however, had already embarked upon a true reorientation, actually cutting the size of the bank, not just RWA math, and hugely reducing its exposures. At the end of 2011, for example, UBS reported CHF36 trillion in total gross notional interest rate swaps. That gross exposure had declined to CHF28 trillion by the end of 2012, and CHF11.6 trillion at the end of last year.
Deutsche Bank, on the contrary, chose a much different path, one of continued expansion determined to use the global wholesale banking industry's overall retreat as an opportunity to be positioned for the full recovery that was surely coming. I wrote last October:
"What Deutsche did to restore balance in the wake of 2013's changed QE profit circumstances was almost unique (Credit Suisse would try something similar). The bank loaded additional capital early last year , under circumstances that remain somewhat unclear, and then plowed ahead into as much risk as it could possibly source. In its press publications displaying those forward intentions in May 2014, the bank (this is somewhat unbelievable in hindsight, far more so than it was then) openly discussed how it expected to drive returns while maintaining leverage from new ventures into US junk and leveraged loans as well as emerging market debt."
They say nobody ever rings a bell at any market top, but this might have been as close as we will ever see. Though DB didn't look risky in RWA terms and still doesn't now, we have every reason to suspect reality hasn't arrived at the same determination. The problem of math as money is that the math isn't always so mathematical. By that I mean popular perception about the use of math infers precision and accuracy; myths that in 2007 and 2008 were exposed in far more than the revulsion from AIG's "regulatory capital relief." In the case of DB after its post-2011 math RWA trimming, however, there may be more immediate concerns. Though it sounds somewhat simple where banks are just "allowed" to choose whatever models and hedges they want, in truth these models and hedges still have to conform to deeper often complex regulatory standards.
If models that have been used to "de-risk" particular assets don't capture markets that are behaving "erratically", then their effectiveness in defining risk is called into question by regulations as well as operations. The same is true for hedges; you can assign CDS or some other technique to a particular asset, but if it doesn't perform as designed and expected, that hedge must be written down or written off leaving you again with a naked asset that demands more hedging (and effective hedging, which likely means higher costs to deploy especially given volatility changes; gamma), absorbing the change to the higher original RWA, or outright selling of the asset. None of these options are neutral, obviously, and often they have direct and immediate liquidity consequences.
Earlier this week, Deutsche Bank's German cousin Commerzbank announced that it was contemplating suspending its dividend (since confirmed) as well as cutting 20% of its workforce. The former is a big deal since the bank just reinstituted it at the end of last year as an intentional signal that it had recovered from 2008 and its nationalization (it remains partly nationalized). In describing why Commerzbank might be taking such a radical, preventative step, the media seems to have fallen back on two particular concepts. The first is NIRP, acknowledging that was once believed to be "stimulus" is actually anti-stimulus in at least one way (leaving the others to still be discovered through reality). The second is this unspecified "market volatility" or "volatile markets." While perhaps seeming an innocuous description, there are terrible consequences hidden in the vagueness.
Volatility is the big one, the chief arbiter of math as money. As the Senate report on the London Whale showed, JP Morgan changed its VaR methodology for its CIO portfolios and on an expedited basis - even characterizing the process of approval as "hurried." The reason was volatility had picked up at the time, forcing calculations of expected volatility to threaten to increase RWA by among other things reducing the modeled effectiveness of its hedges. By changing the math of the VaR governing condition, the effect was to reduce the anticipated loss exposure by 50%, thus reducing (inappropriately) CIO's contribution to the bank's overall RWA.
Applying that same kind of scenario to DB (or Commerzbank, perhaps Credit Suisse and others?) in 2016, an environment that is at least comparable to what JP Morgan was facing in 2011 and 2012 (and as far as possible hedging is concerned, likely significantly more troubling with regard to systemic capacity), we can reasonably infer especially for a bank heavily dependent on models for having "de-risked" that there is a high degree of actual risk unwinding some unknown but significant part of that process. That negative pressure would be further amplified by an environment where total offered balance sheet capacity is limited; meaning where DB's ability to replace ineffective hedges with new ones might be constrained because the price or flexibility of what is currently offered is relatively uneconomical due to those systemic reductions in overall balance sheet capacity. We know without much doubt this is the case with systemic capacity since at least the middle of last year and remains ongoing by everything from negative swap spreads to hugely negative premiums on cross currency basis swaps.
Any problem for Deutsche as far as math as money is concerned is not strictly a problem for the single German bank. As we found starting in August 2007, balance sheet capacity has the unfortunate tendency to be reversed in irreversible fashion once it gets beyond a certain point. In other words, given that math capacity overall in the system is driven by a relative few number of global banks, the impairment of one starts the process of systemic impairment for all (again, this doesn't necessarily imply crash, just irreversibility). For DB, that may be "most" true as it has tended toward especially a high absolute degree of derivative exposure (offered capacity) as well as its relative position compared to where almost all of its peers have been drastically cutting back.
In what might prove to be the worst blow yet, the reason for such a possibly troubling volatility shock in math as money is QE itself; not what QE actually accomplished, rather what everyone thought it would. Even those banks with a more realistic economic assessment seemed to have modeled low financial volatility as a consequence of so much "money printing." The effect, then, of the past year and a half was to so thoroughly disprove that assumption as to destroy it. A world where QE isn't actually "money printing" isn't the same at all. The amount of balance sheet capacity demolished because of being forced into a more realistic financial taxonomy may be incalculable, but we can feel directly its consequences - none more so than Deutsche Bank at this moment.
It is an absolute falsehood that banking has changed since the panic. Regulations have changed the numbers but by and large the system survived if only in diminished form and flow, not at all the categorical shifts that are often claimed. The same vulnerabilities inherent in the pre-crisis eurodollar/wholesale system remain, most especially the regulatory emphasis on flawed math that was developed precisely because funding formats (liabilities) started to become utterly complex and convoluted. Shifting focus to credit losses on the asset side as if they were the biggest danger was in reality nothing more than an ill-conceived shortcut. The monetary system was radically overhauled (of its own volition and conduct) during the last half of the 20th century, but there is a universal truth that remains especially relevant in the 21st; money is still all about liabilities no matter what any regulatory regime decides.