Predyct Decision Framework

As important a tool as Predyct’s ERM is it is not the entire solution. This is because risk confronts companies with two major but very different problems: 1) Volatility of surplus , and 2) Cost of capital. Volatility of surplus capital is captured by value-at-risk (VaR) and dynamic financial analysis (DFA) “type” simulation frameworks, but cost of capital is a corporate finance issue that focuses on capital structure. A firm’s capital structure if too leveraged will impose a high cost of equity capital that will squeeze a firm’s net earning margin relative to its competitors eliminating its embedded value (See embedded Value link). Being underleveraged will also impose a cost on the firm by lowering its return on capital relative to its competitors and thereby making it less competitive. A real risk solution must solve these two different issues in a completely consistent framework. Anything less is not adequate.

Predyct Analytics combines these two models in what we call our Predyct Decision Framework, or PDF. Our PDF includes Predyct’s ERM model (a VaR simulation model) to assess a firm’s volatility of surplus which identifies which risks (insurance, catastrophe, interest rate, FX, Operational, Liquidity and equity) contribute and in what proportions to its surplus volatility. We call this our Bottom-Up analysis. PDF also includes Predyct’s shareholder value model (SVM) that breaks a firm down into its constituent business activities and benchmarks each activity to a set of publicly traded peers to arrive at a value for the entire enterprise. SVM represents Predyct’s Top-Down analysis and identifies each business activity’s market to book (M/B), price earnings (P/E), cost of capital (Ke), return on equity (ROE) and market implied growth. It is Predyct’s position that managing value is as important a component of risk management as managing its volatility of surplus from which capital adequacy is determined. DFA and VaR type models do not identify a firms cost of capital or market implied growth—Factors that are required to be known in order to make optimal resource allocations. These factors are identified with market models such as the capital asset pricing model (CAPM) and its brethren (i.e., multifactor arbitrage pricing “type” models (APT), Modigliani & Miller (M&M), Black/Merton/Scholes (BMS): collectively, Predyct’s SVM). Without an understanding of a firm’s cost of capital and market implied growth potential in its different business activities firm management will tend to value the revenue it receives from its different business activities as being the same—and they are not all the same. In this way management will not make the correct resource allocation decisions, firm value will be impaired, cost of capital will increase and net earning margins will be diminished. A vicious cycle will evolve.

The other issue about DFA and VaR type models is that very rarely are they calibrated to accurately capture the correlation between firms’ assets, liabilities and assets and liabilities. Generally, these models tend to bucket large groups of assets and liabilities as well as their respective correlations. As such, these models may greatly underestimate correlation, risk and the required risk capital. SVM models that rely on peer analysis using publicly traded peers also have weaknesses in-so-far as they assume that the peers used are an accurate reflection of the company being assessed. This is why we feel strongly that both methods have to be utilized (as a cross check on the other) to obtain a fair assessment of a company’s risk and value. Incongruities of output between the two methods have to be studied and reconciled before conclusions can be reached.

Examples of where each method (as well as their collaboration) was instrumental in identifying the underlying cause of distress: (Links are provided to the specific case study mentioned.)

AIG was initially assessed in 2001/2002 using Predyct’s SVM. Later in 2007 we would use both SVM and Predyct’s ERM. We applied Predyct ERM to AIG’s domestic property and casualty company which is Company C in the benchmarking link on the Predyct ERM navigation bar. Lehman Brothers, Bear Stearns and Merrill Lynch were all assessed using Predyct’s SVM. AIG was too large and diverse a company to be assessed exclusively (from the outside) with a VaR type simulation model like Predyct ERM.

Company A2 kept allocating resources to its check printing business at the expense of its far more valuable (and natural successor) EFT business. This was diagnosed with SVM. Predyct ERM (or any simulation “type” model) would not have made this diagnosis.

Company B1’s increase in surplus volatility and cost of capital was initially predicted by our SVM when Company B1 acquired the HMO. This was later confirmed by Predyct’s ERM. However, solutions to the resultant problem of over leverage, cost of capital and stock decline could only be identified using SVM. Solutions that we identified included asset sales including the divestiture of either the HMO or the life company. The life company was divested.

Company B3 was diagnosed almost exclusively using Predyct ERM. This was an overcapitalization issue that was diagnosed largely based on the company’s claim payout patterns which can be highly idiosyncratic from one company to the next. Peer analysis (SVM) could have been used but would not have been nearly as exacting.

Company B4 had allowed its capitalization to run far below what was required to sustain its AA rating. Predict diagnosed its difficulties initially with SVM and subsequently fined tuned the final analysis using Predyct ERM.

Company B2 was initially diagnosed using Predyct’s ERM. SVM would not have been capable of diagnosing that 40% of its capital was being consumed by Homeowners Business and that’s its ROE in that business was negative.

Company B5 was diagnosed initially using Predyct ERM to identify how much capital was being consumed by fixed annuities and its coincident low ROE. But SVM was used to identify the value impact this was having on B5 as well as to assess whether this was a fixable business or not. SVM analysis revealed that fixed annuity companies across the market had low market values and very low growth indicating that divestiture was the best course.

Company C1 was diagnosed exclusively using Predyct’s SVM

Company C2 was diagnosed using a complimentary framework of both Predyct’s ERM and SVM. However, the prescient diagnosis was largely due to SVM since most of Company C2’S value decline was due to the future dilutive effects of options that the company gave away in the process of going public. It is highly unlikely that any DFA or VaR framework could have diagnosed this dilution since the options were not publicly traded or marked-to-market.

These are just a few examples from Predyct Analytics archives that we believe reveal the necessity of using both top-down SVM valuation as well as a bottom-up VaR simulation (Predyct ERM) approaches in order to make sound risk management and investment decisions.