Portable Alpha: What it is and how it can be applied to Insurance Companies' Surplus Portfolios by P

1
Portable Alpha What it is and how it can be
applied to Insurance Companies' Surplus
Portfoliosby Pierre LarocheSenior
DirectorFinancial Engineeringfor The 2006
Stochastic Modeling Symposium and Investment
Seminar,Toronto, April 4 2006
2
1. DesjardinsGlobal Asset Management2. What
is Portable Alpha?3. PA, Portfolio Volatility
and Tracking Error4. Discussion5. An Example
3
Desjardins Global Asset Management (DeGAM)

• DeGAM is Mouvement Desjardins' main portfolio
  management subsidiary.
• CAD 41 Bn under management
• CAD 14,5 Bn Fixed Income
• CAD 11,5 Bn Repo
• CAD 9 Bn Multi-Management Retail
• CAD 5 Bn Funds of Hedge Funds
• CAD 1 Bn Multi-Management Institutional(Conve
  ntional Quantitative)

4
2. What is Portable Alpha?
Sample 125 US Defined Benefits Pension
Plans Source JP Morgan Chase, 2005
5
2. What is Portable Alpha?
"Portable Alpha refers to the process of
separating the alpha from the beta and then
applying it to other portfolios" (Kung and
Pohlman (2004)) To well understand PA, let's
define it as a portfolio enhancement strategy
rP excess returns (i.e. returns in excess of
the risk-free interest rate) of the enhanced
portfolio (E), the base portfolio (B) and the
(external) active management mandates (A) that
constitute the enhancement sources and where
,
6
2. What is Portable Alpha?
Since actively managed portfolios are usually
benched against an index portfolio, we can write
We then have
And since
Then
7
2. What is Portable Alpha?

• Many classical hedge fund strategies carry a mix
  of alpha and beta. Sometimes, the beta component
  is quite proheminent.
• For example, most Fixed Income Arbitrage hedge
  funds use a mix of the three following
  strategies
• Buying illiquid securities and/or with more
  credit risk and selling more liquid securities
  and/or with less credit risk.
• Riding the interest rate curve ("positive carry"
  positions).
• Selling interest rate volatility.
• This combination of these three pure beta bets
  has a circa 0.8 correlation with Tremont's Fixed
  Income Arbitrage Hedge Fund Index!

8
3. PA, Portfolio Volatility and Tracking Error

• Important issue Active portfolio management
  creates tracking error does that raise the
  enhanced portfolio's absolute risk (i.e.
  volatility)? Answer usually not.

We know that
Hence
Since TE must be non-negative, this last equation
requires that
.
So that if
then
9
3. PA, Portfolio Volatility and Tracking Error

• We tested the model over the Jan. 1996 Dec.
  2005 observation period for 269 managers
• 40 TSE Composite 103 SP500 78 MS EAFE
  48 MS Emerging markets

• The main results were
• The median value added is quite small
• The best value added are found for the tougher
  markets (EAFE and Emerging Markets)
• Most b are less than 1.0
• The managed portfolios' volatility is equal to
  that of their benchmark index
• The model is quite well validated by data

10
4. Discussion PMIT

• The main benefits arising from PMIT's (portable-)
  alpha and beta separation are
• The portfolio manager can stick to her strategic
  asset allocation target (in the pure beta
  indexed - portfolio)
• There is no need for modifying the management
  structure
• It simplifies the portfolio performance
  attribution
• It allows a simpler risk budgeting

11
4. Discussion Challenges

• Applying PMIT has several challenges
• The investment policy may have to be modified,
  especially regarding the funding decision
  (financing the pure alpha bets)
• The investment constraints (particularly
  regarding credit risk limits, leverage and
  duration) may be harder to define and monitor
• The risk involved may not be well understood
  (e.g. tracking error) nor measurable with enough
  precision (e.g. liquidity risk and transparency
  issues)
• The identification of reliable external managers
  with above-average alpha investment skills is not
  an easy task and may be costly
• The best alpha markets are often "difficult"
  markets

12
4. Discussion Modelling

• Stochastic Modeling Issues
• Expected added value, volatility and tracking
  error are sufficient no need to take care of
  the managed portfolios' correlations. Benefit
  lower dimension of the correlation
  matrix/Cholesky Factorization problem.
• Statistical challenges
• Short observation periods and style drifts
  (homogegeity of the sampled returns)
• Fundamental hypothesis may not be met (estimation
  and model risks)

13
4. Discussion Multi-Management

• There are many reasons to allocate external
  active management mandates to at least a dozen of
  different firms
• Average value added does not diminish and becomes
  more stable
• Portfolio volatility diminishes
• Tracking error diminishes a lot
• Information ratio rises substantially

14

• Discussion Multi-Management

15

• Discussion Multi-Management

16

• Discussion Multi-Management

17

• Discussion Multi-Management

18
Application to Insurance Companies' Surplus
Portfolio

• Important Makeover
• Active Management
• New Asset Classes (Emerging Markets Bonds, Real
  Return Bonds, Insurance-Linked Bonds,
  Commodities, etc.)
• Some Asset Classes Removed (i.e. "transformed")
• Overlays (fully financed) vs- lowering Money
  Market Allocation ( and lowering the duration of
  the other fixed income allocations)
• Main Challenges
• Investment Committee's Good Understanding of the
  Rationale
• Additional Risks and MCCSR Concerns

19
Question Period