1. In complex systems “more” is different

2. Positive feedback loops grow exponentially

The flash crash occurred when one exchanges system clock got overloaded and was slightly mis-aligning bid/ask times stamps.  The HFT traders interpreted the mis-stamped quotes as an arbitrage opportunity sending more messages and activity into the system further mis-aligning the time stamp and increasing the perceived opportunity. see Nanex report

3. Normal accidents are caused by common errors

Summary & Flash Crash 2.0

Explanatory pictures and video:

• Share

Watch the full episode. See more Need To Know.

HT: Barry ritholtz for this

Pretty amazing video given how big the amounts are and how little investigative work has been done here.  The amounts involved dwarf anything else discussed to date.  The $12.8 trillion has a significant amount of “back stop” money that wasn’t called upon but was effectively guaranteed and with a lot of swap lines etc.

The freeze up in the money markets was obviously looming as a cascading failure.  To learn about  large scale failure in complex systems check out the excellent and fun to read book ubiquity: why catastrophes happen.

Well done to Bloomberg to rock the boat and dig a bit into this.  Please note that the brave ladies and lads at CNBC, Fox business etc. who are busy posing as business journalists really haven’t even touched these issues, much less challenged the fed with an FOIA request.  Investigative or real journalism stops at the edge of the newsdesk and echo chamber for the TV folks.  Please: Reuters, WSJ pick up this torch there is probably Pulitzer material here.

The fact remains that a huge opaque system, the Fed is at work trying to sort out a mess that may be growing quietly larger.  The low rates are helping balance sheets, but it is a race between free FED spread and loan deteriorations.

Only history will correctly identify whether the stealthy approach has prevented a broader crisis of faith in the system in 2008 averting a temporary liquidity issue or whether the game is only perpetuating a problem under the popular extend and pretend sobriquet.

Que sera sera. read when money dies

Disclosure: Long N. Krone, Gold

• Share

Risk management is understanding and controlling exposure to bad outcomes. Risk managers often focus on extreme events that may overwhelm a system. Scenarios simulate extreme events and expected price outcomes. In non-natural disasters, the worst outcomes often occur when the inputs and processes seem to operate normally, but wrong processes are in place. Wrong process risk is undermanaged.

Risk exposure occurs when a manager exposes cash or the balance sheet to a (return generating) system such as an equity, bond, business line or some other system with an expected net positive cash flow.  This exposure is to a system made up of inputs, processes and outputs.

Value investors are famous for discussing the capricious nature of price as an occasionally flawed output relative to value.  The value thesis separates inputs and asset value creating process from the output of short-term market price.  The argument being that “value” shows up in price over a time and short-term price may be a “wrong output”. The return generating system needs time to create value, which will eventually be reflected in the “right” price.

Risk managers overly focus on system outputs, usually price

Many risk tools model price outputs. Fewer risk management approaches focus on monitoring return generating inputs and processes.  Price output fixation is likely due to the fact that sources of return generation are often poorly understood, differ significantly across exposures and are difficult to quantify.

System outputs such as price and its variance become easy proxies for an exposure’s risk.  True risk exposure may be better managed by looking at the exposure as a system to be monitored including the inputs and return generating processes.

The bigger Flash Crash issue: market integrity

An example of overly focusing on price is the recent Flash Crash and lesser known ongoing mini-crashes[1].  The Flash Crash is commonly misunderstood relative to its real market system risk impact.  Price is easily discussed, but masks the more germane point of market system integrity as often the crashed shares revert to near pre-crash levels.

The Flash crash and ongoing mini-crashes reflect a change in the underlying equity market process.  Price may vary so quickly from value that it causes market risk of a more subtle but profound type.   This risk is market integrity risk relating to the primary economic function of an equity market.

The primary role of an equity market is to act as a system for price discovery and efficient transactions.  Ongoing flash crashes send signals to participants that the equity market system may be malfunctioning in its price discovery role.   This perceived failure could lead to less “real” liquidity, exacerbating current flash crash inducing processes.

Publicly traded firms trade at a premium to privately held firms due to the faith in the liquidity and fair pricing of the public market.  If the public market system is believed to have an increased risk component, the equity premium of all participants could suffer.  It remains to be seen if micro-second messaging and execution is a fair trade off for even slight declines in the overall public market equity premium.  These points are raised to indicate an example of over focus on an event and price outcome such as the flash crash and its outputs that may miss the bigger point, i.e. the integrity of the system process.  Maybe high frequency trading is the wrong process for the public equity market system.

Managing risk involves a system’s inputs, processes and outputs

The risk manager is told that an exposure or line of business is changing and new risk management systems need to be put in place, these systems typically measure an amount of exposure and ongoing price.  The systems are stress tested using scenarios to determine that they are fit for purpose.

The exposure and systems are brought on line and the system churns away, hopefully with the desired positive cash flow output.  Inputs and return generating processes may not be revisited as long as the outputs of the system are looking “good” i.e. within tolerances and generating positive returns.

Systems failure: Event based & wrong process

Risk managers are familiar with operational or price event based failures. Event based failure involves a process overwhelmed by an input or the process breaking on its own.

Example: our payment clearing system was designed to handle 4 million TPS (transactions per second) but we had 7 million…

Price & operational risk management is typically about choosing event based scenarios to determine an acceptable stress/failure point for the exposure.   A business case is created based on scenarios describing input, process and expected outputs during a stress event.

Credit card example of business case and scenario Exposure sought: Personal credit card industry

• Input:We are allocating $3 billion and appropriate resources to credit cards
• Process:Credit spreads indicate 10.0% annual return on capital for 7 years
• Output:We anticipate an +$300m annual cash flow

Risk Scenario:

• Input:A -5% GDP & -300 bps spread contraction and behavior shifts…
• Process:Charge off rates increase +20% from baseline…
• Output:The business could temporarily show $200m/year losses

Wrong process failures are larger than extreme events

Wrong process failure occurs when everything is working correctly relative to inputs and processes but outputs are undesirable, the wrong process is being used.  Wrong model selection or utilization is a form of Wrong process.

Most structured finance was related to wrong process failure. The wrong risk assessment processes were being used. The dis-aggregation of information and diligence responsibilities in the securitization process lead downstream participants to believe they had quality assets. Downstream securitization due diligence processes as implemented worked well.  Unfortunately the choice of process involving shallow diligence with minimal discovery was flawed.

Wrong process risk is often more dangerous than extreme event risk. Wrong processes create the illusion of normally functioning systems with successful outputs (profits) in the short or medium term.  The false appearance of normal operational “profitable” behavior means that processes aren’t inspected or deeply questioned and challenged.  In finance, like businesses, the only axiom more dangerous than, “If it ain’t broke don’t fix it.”, is probably, “if it is making money, don’t question it.”  It is occasionally forgotten that humility and vigilance are often the price for excellence in finance.

The most insidious wrong processes take longer to show bad outcomes.  Unquestionable faith in an exposure often correlates to an exposures recent profitability.  The longer something is profitable, the more money and participants crowd in with similar systems.  At the same time each new participant questions the fundamental inputs and processes less and less. Group behavior and faith that “someone must be paying attention”, has the perverse effect of reducing input & process risk analysis at exactly the time that aggregate exposure risk is growing.

Systemically threatening Wrong process is difficult to root out of short term output focused organizational cultures.

Actionable Systems Risk

Actionable systems risk involves placing ongoing equal weighting and importance on a systems inputs, process and outcomes.  Every system has capacity constraints in inputs, processes and outputs.

The first indicator of trouble in a system occurs when the outputs vary significantly from the inputs. This is the law of the conservation of risk, risk doesn’t disappear, it is shifted sideways to someone else or forward in time.  If risk seems to disappear from a system, it may be indicative of a flawed understanding of the system.

Securitization is an example of this, a bundle of debt based payment streams is sliced (tranched) and presented in such a way that aggregate risk appears less than original risk.

Trading strategies or alternative asset classes requiring increased leverage or more input relative to an expected rate of return are another example of an exposure that may have shifted due to overcrowding.  These environments are highly unstable as the crowded process becomes more vulnerable to collapse.

Complexity & innovation in finance are antithetical to Actionable Systems Risk Thinking

In finance complexity is often seen to represent sophistication and innovation.  This sophistication backed by math models, ratings and or years of apparent success can lead to large problems. Complexity masks a system’s inputs, processes and/or outputs.

In the case of mortgage securitization, complexity masked inputs.  As profits rolled into the securitization sector, inputs and processes were questioned less.  Over time this lead to an almost blind faith in ratings downstream. Faith in the ratings system meant ignoring system inputs, namely the deteriorating quality of underwritten mortgages.

In alternative asset investments, risk management is mostly performed at the output and input level with processes such as security selection obscured.  Initial and ongoing due diligence and portfolio monitoring may allow the alternative asset risk manager to assess the inputs to a system (manager exposure).  Principal component analysis, peer benchmarking and other output-based analytics may be used as soft proxies for retrospective allocation process management.  The reality is that many alternative exposures by definition have idiosyncratic processes limiting risk management.

War games: The Wrong input and Wrong process audit

Due to the danger of wrong inputs and processes, it is may be time for the actionable systems risk manager to audit many exposures inputs and processes.  Independent, internal benchmarking of inputs, processes, capacities and systemic assumptions on a periodic basis would most likely be a worthwhile activity for many risk managers.

War games test expected behaviors and system responses by fully questioning the integrity of inputs and processes. Performing an annual or semi-annual wrong process and wrong input audit may make sense for many lines of business as a way of testing input and process assumptions.  Cross checking of inputs, stressing of processes and questioning those assumptions that appear “most” obvious may prove valuable.  At a minimum performing a wrong input, wrong process audit forces the manager to pose new questions and reframe old beliefs.

Disclosure: “No Positions”

• Share

Howard Schilit’s book, Accounting Shenanigans is a classic in terms of clear simple explanation of games CFO’s play.

In buying into a company as a shareholder, you are purchasing a piece of a value creation process.  The sustainability, duration and effectiveness of that process are the businesses realities.  Those realities are converted into Cash Flow, Balance Sheet and Income statements.  The assumptions used are management’s numeric story of the value creation process.  It is important to consider how the prism of accounting policy ranging from conservative to agressive is being used to present that value creating or destroying systems reality.

For fun I built a widget to naively weight the accounting policies of a company.  See screenshot above.  The excel file shown above and checklist is available just send me an e-mail.

I am put together a few simple checklists based on Schilit’s classic book Accounting Shenanigans.  Accounting won’t tell one why value shows up, but it will tell one what form it appears to be taking and a bit about how your management tells its tale.

Here are some of the tools you should use according to Schilit.

Documents used to find shenanigans & what to look for:

Auditors report

• Absence of opinion
• Qualified report
• Reputation of the auditor

Proxy statement

• Litigation
• Executive compensation
• Related party transactions

Footnotes from 10-k and 10-Q’s

• Accounting policies / changes
• Related party transactions
• Contingencies or commitments

President’s letter and transcripts

• Forthrightness and history of honesty

MD&A (part of the 10-K)

• Specific concise disclosures
• Consistent with Footnote disclosure

Form 8-K

• Disagreements over accounting policies

Registration statement (for IPO’s)

• Past performance
• Quality of management and directors

• Share