Unlisted Assets and Enterprise Risk Management

Transcription

1 Unlisted Assets and Enterprise Risk Management Emilian Belev, CFA, ARPM May 2014

2 The Unlisted Assets Problem MPT is not common jargon in real estate investment departments Unlisted investment experts tend to think of risk in terms of first distributional moment. Location, Location is not an obvious quantity Estimation of duration of cash flows pertaining to unlisted assets are often over-simplified in the practice of ALM for pensions and insurance Broad asset class indexes, or marketable-proxies, attempt to battle the quantification aspect while compromising on investment s idiosyncratic nature and appraisal bias

3 The Global Total Portfolio Risk problem Multiple portfolios with diverse characteristics Across countries, across asset classes Asset classes such as convertible bonds and derivatives have complex properties Asset classes such as private equity, real estate, infrastructure, have no visible pricing, return, or risk information

4 Segregated solutions One approach: model asset class separately and aggregate risk only through covariance matrix: Overall covariance matrix resembles a chessboard where each square is a sub-model segment of the covariance matrix As aggregate number of factors increases relative to the limited number of observations, the matrix becomes unstable Fixing the problem by extending the historic observation period of the sub-models discounts the importance of the dynamics in the market place embedded in more recent observations Illiquid Assets Compound the Total Portfolio Problem. Current Illiquid modeling practice produces yet another silo factor set in the total portfolio covariance matrix

5 Fully Integrated Approach Employ a single parsimonious factor model across all asset classes across all geographies All investable assets are related to the same consistent set of factors, so interrelationships are easily observed and understood Limited number of factors allows for stable estimation of factor relationships, and fluid regime shifts

6 6. The EE Risk Model Everything Everywhere (EE) Risk Model factors: Five regional indices Six sector indices with global constituents Return of Salomon Smith Barney WGBI Percent changes in oil prices Dividend Yield factor relative performance of value vs. growth stocks Market Development factor investor confidence measure through differences in performance of stocks in developed and developing countries Size Effect confidence measure through differences in performance of companies of different market capitalization 69 factors - dummy variables - indicating the denomination currency of a security

7 7. EE Model Structure Factors (contd.) Parallel Shift in Treasury Term Structure of Interest Rates Changes in Slope of Term Structure Twist Changes in Curvature of Term Structure Butterfly The three treasury factors correspond to the a, b, and c coefficients in the relationship into which treasury yield data is fit: Y = (a) + (b X) + [c (X)^2], X - time horizon for respective point on the yield curve

8 8. EE Universe Global Equities Global Sovereign and Corporate Bonds (2013 PRMIA Award for New Frontiers in Risk Management) Asset-Backed Securities, Structured products, and customized coverage Municipal Bonds Futures, forwards, options on equities, currencies, interest rates, and bonds, Interest Rate and Credit Default Swaps Mutual and Hedge Funds Real Estate, Infrastructure, and Private Equity

9 Infrastructure

10 Northfield s Infrastructure Approach A granular project-specific approach Each project is analyzed in respect to: Deterministic cash flows from existing and future usage/contracts of use of the infrastructure facility Usage specifics, contract provisions, capacity, credit quality of contract counterparties, operating cash flows Risks related to financial gearing Floating rate, fixed rate, interest-only, balloon clauses, and prepayment behavior Risks of future fluctuations in cash flows from business All components have risk exposures to common risk factors plus idiosyncratic risks

11 Northfield s Infrastructure Approach: An Airport Example (cont d) The business cash flow volatility component is modeled using portfolios of appropriately chosen businesses. For an airport those will be: - Airline Gate Operators - Retail Companies Fully-granular approach: - Use time series of operating cash flows from that segment (business sub-portfolio) of the infrastructure project and estimate sensitivities to risk model factors Generic approach: - Use the volatility of the equity of the contractual counterparties (Airlines, Retailers), to estimate the volatility of demand for project capacity by adjusting for metrics of facility usage and for three types of leverage: - Financial Leverage - Operating Leverage (Fixed Costs) - Operating Leverage (Variable Cost per Unit of Output)

12 Northfield s Infrastructure Approach: An Airport Example (cont d) Percent of Total Holdings Airport Example 12/21/2012 Port. Ending Market Portfolio Symbol Composite Assets Value Weight Total Direct A_1_0_1 Gate Lease A_1_1_1 Gate Lease A_1_2_1 Gate Lease A_1_3_1 Gate Lease A_1_4_1 Gate Lease A_1_5_1 Gate Lease A_1_6_1 Gate Lease A_1_7_1 Gate Lease OFFSEC RNT_AIRPORTGATE Rent volitility synthetic S security (Gate RNT_AIRPORTRETA Rent volitility synthetic IL security (Reta A_0_0_1 Retail Lease A_0_1_1 Retail Lease A_0_2_1 Retail Lease A_0_3_1 Retail Lease A_0_4_1 Retail Lease A_0_5_1 Retail Lease A_0_6_1 Retail Lease A_0_7_1 Retail Lease A_0_8_1 Retail Lease

13 Northfield s Infrastructure Approach: An Airport Example (cont d) Summary Report AIRPORT_EXAMPLE 12/31/2012 NIS EE Model Factor Contribution (Std Dev) Total Risk (Std Dev) Percent Contribution Stock Specific Risk % Factor Risk Factor Contribution (Std Dev) Region % Super Sectors % Economic % Fundamental % Blind Factors -- Currency -- Curve %

14 Private Equity

15 Private Equity Is there a separate asset class as Private Equity that requires distinctly different treatment from public equity Or is it just Equity with some overlay characteristics like: Higher volatility due to illiquidity Lockout provisions Unfunded commitments Investors consider PE to be different ONLY because it does not have frequent and regular published returns series like public equity Hence we cannot consider it to be a different asset class We need to use public equity as baseline and use appropriate adjustments to reflect the above features

16 Private Equity (cont d) Liquidity Premium for Private Equity varies in the ranges 3-12% (Franzoni et al), with the long term value gravitating towards 5%, and also varies across industries, and different types of PE: MBO LBO VC Acquisitions, etc. Mean-variance utility E(U) = exp( mean [1/(2*RAP)] * sigma^2} suggests that the marginal market investors will be indifferent to buy a public market portfolio vs. a PE investment if the increase in the mean is matched by a commensurate increase in undiversifiable volatility. Consequently, risk parameters of analogous pubic investments can be adjusted to reflect private equity illqiquidity premium

17 Private Equity (cont d) Apart from illqiquidity driven by less transparency / thinner market considerations at the deal level, asset owners need to account for two more features at the PE fund level: Lockout provisions in PE fund investments. These pose time limits how soon the funds can be withdrawn. There is a real option application to estimating the cost of distress sale of an inordinate amount of liquid assets to make up for cash needs. That can be translated into volatility-equivalent adjustment to underling deal risk parameters. Cash calls from PE fund managers. Those can be either optional or firmly committed, and can be modelled respectively as a call option or a forward position on the underlying fund, with commensurate leverage effect on risk exposure from the PE investment.

18 Private Equity (cont d) Model each fund holding as a composite asset comprised of privately held businesses Each private company is attributed with the risk characteristics (factor exposures and asset specific risk) of the industry and country cohort Those characteristics get modified to account for leverage, maturity of the business and investment category illiquidity premium. At the fund level, the aggregate risk characteristics get modified to account for a stochastic shortfall respective to lockout provisions, and unfunded commitments (cash calls)

19 Application in Risk Management Broad risk drivers generated by a global risk model provide two very important dimensions of flexibility in risk management The intersection of the model factors used for alternative investments and marketable securities, allows for any degree of total portfolio risk adjustments, even if part of the risk comes from highly illiquid assets. E.g. the interest rate risk from a airport s long term lease cash flows can be hedged using a bond futures contract. An oil pipe facility near Houston is effectively a bet on the oil industry which can be balanced with selling stock in oil companies. A retail center exposure to a major anchor tenant can be fine-tuned by manipulating weights of the stocks and bonds of the same or similar retailers. The risk model can tell us by how much. The granularity by infrastructure project and property allows for a tailored analysis of the impact of adding new investments and selling existing ones; something not possible using an index-based approach.

20 Application in Risk Management (cont d) Other benefits of a broad factor-based model approach: The model output can be readily integrated in an Enterprise Risk Management system alongside all other components of the total portfolio The data estimation results can be integrated in any risk model / platform as long as it also has a sufficiently broad factor set. This can be accomplished in two ways: Translate the native Alternative Investments risk model factor exposures into generic risk model factor exposures, for a Markowitz-style cross-asset class risk estimation and optimization on any risk platform Generate economic scenario values based on the particular risk model output, that can be imported on scenario-based risk system, where distributional estimation of the total portfolio can be performed, inclusive of the difficult alternative assets.

21 Portfolio Contents Overview

22

23 Expected Tail Loss

24 Factor Variance Contribution

25 Total Portfolio Segment Correlations

26 Questions and Answers Session For follow up questions please contact Northfield s Illiquid asset modeling team at: Nick Cutler, Business Manager, Real Estate, Infrastructure, & Private Equity - cutler@northinfo.com Emilian Belev, CFA, Head, Enterprise Risk Analytics Emilian@northinfo.com Rick Gold, Senior Real Estate Analyst rgold@northinfo.com

27 References DiBartolomeo, D., Gold, R., Baldwin, K., Belev, E. A New Approach to Real Estate Risk. Northfield Working Papers Belev, E., Gold, R.. The Pitfalls of an Index-Based Approach to Managing Real Estate Investment Risk. Northfield Working Papers Ross, Stephen A., The Arbitrage Theory of Capital Asset Pricing, Journal of Economic Theory, 1976, 13, pp Sharpe, W., Capital asset prices: A theory of market equilibrium under conditions of risk, Journal of Finance, September Franzoni, F., Nowak, E., Phalippou, L., Private Equity Performance and Liquidity Risk, Network for Studies on Pension, Aging, and Retirement Discussion Papers, May 2011