OCTOBER 17, 2019 Crisis Program Fresno County Employees’ Retirement Association Introduction

FCERA Crisis Risk Program 2 October 17, 2019 Defining Crisis

Time (Months) Return Peak-to- . S&P 500 falls 30% or more Crisis (%) Bottom Trough Recovery Total Great Depression -83 Jun-32 34 150 184 Great Financial Crisis -51 Feb-09 16 36 52 . Five times since 1926 9/11 -45 Sep-02 25 48 73 Oil Crisis -43 Sep-74 21 20 41 . Median : 45% Black Monday -30 Nov-87 3 17 20 . Median peak-to-trough: 21 months

Source: Ibbotson Associates, Standard & Poor’s, as of 6/30/2019

FCERA Crisis Risk Program 3 October 17, 2019 What is a Crisis Risk Program?

1. Supplemental strategic planning 2. Alternative view of risk 3. Informs 4. Reduce crisis losses

FCERA Crisis Risk Program 4 October 17, 2019 Process

1. Historical Analysis 2. Strategy Discussions 3. Conclusion

FCERA Crisis Risk Program 5 October 17, 2019 Historical Analysis

FCERA Crisis Risk Program 6 October 17, 2019 Steps

1. Start with all- 2. Add a risk mitigating strategy 3. Consider cost/benefit 4. Rank and select strategies

FCERA Crisis Risk Program 7 October 17, 2019 Candidate Strategies

1. Core Bonds 2. Risk Parity 3. Long Treasuries 4. Mortgages 5. Trend Following 6. Precious Metals 7. Derivatives 8. Other

FCERA Crisis Risk Program 8 October 17, 2019 Measuring Downside Protection

Annual Crisis Return . 93-Year History Return Jun-32 Feb-09 Sep-02 Sep-74 Nov-87 How much crisis loss did a All Stock Portfolio 10.1 -83 -51 -45 -43 -30 protected portfolio avoid? -? +? +? +? +? +? Protected Portfolio ? ? ? ? ? ? . How much long-term annual return did that cost?

. Also, plot portfolio returns All Stock Portfolio Protected Portfolio (vertical axis) against crisis losses 20 20 (horizontal axis) 10 10 0 0 . Slope (x coefficient) of the -10 -10 regression line provides a single -20 -20 measure of historical effectiveness Nov-87 -30 -30 . Slope of 1 = no protection -40 -40 Sep-02 Sep-74 -50 -50 . Slope of 0 = complete protection Feb-09 -60 -60 Portfolio (%) Return Portfolio Portfolio (%) Return Portfolio y = 1x + 0 -70 y = ?x + ? -70 R² = 1 -80 R² = ? -80 Jun-32 -90 -90 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 S&P 500 Drawdown (%) S&P 500 Drawdown (%)

Source: Ibbotson Associates, Standard & Poor’s, as of 6/30/2019

FCERA Crisis Risk Program 9 October 17, 2019 Core Bonds

Annual Crisis Return . 93-Year History Return Jun-32 Feb-09 Sep-02 Sep-74 Nov-87 The most “cost-effective” strategy All Stock Portfolio 10.1 -83 -51 -45 -43 -30 -1.5 +21 +20 +24 +15 +13 . The foundation of any Crisis Risk Program Sixty-Forty Portfolio 8.6 -62 -31 -21 -28 -17 . Remaining strategies are added on

. Give-up a sixth of long-term All Stock Portfolio Sixty-Forty Portfolio pure equity return 20 20 10 10 . Cut crisis losses by 0 0 one-fourth to one-half -10 -10 Nov-87 . Slope of regression line goes -20 Sep-02 -20 Sep-74 from 1.00 to 0.65 Nov-87 -30 Feb-09 -30 -40 -40 Sep-02 Sep-74 Feb-09 -50 -50 -60 -60 Portfolio (%) Return Portfolio Jun-32 Portfolio (%) Return Portfolio -70 y = 0.6521x + 0.4326 -70 -80 R² = 0.9576 -80 Jun-32 -90 -90 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 S&P 500 Drawdown (%) S&P 500 Drawdown (%)

Source: Ibbotson Associates, Standard & Poor’s, as of 6/30/2019

FCERA Crisis Risk Program 10 October 17, 2019 Long Treasuries

Annual Crisis Return Replacing a bit more with pure 93-Year History Return Jun-32 Feb-09 Sep-02 Sep-74 Nov-87 All Stock Portfolio 10.1 -83 -51 -45 -43 -30 provided significant -1.5 +21 +20 +24 +15 +13 additional protection at a modest cost in Sixty-Forty Portfolio 8.6 -62 -31 -21 -28 -17 terms of long-term historical return -0.2 +5 +4 +4 +2 +2 90% Sixty-Forty, 8.4 -57 -27 -17 -26 -15 10% Long Treasuries

. Again, historical results 90% Sixty-Forty, 10% Long Treasuries Sixty-Forty Portfolio benefited from secular 20 20 downward trend in yields 10 10 0 0 -10 Nov-87 -10 Nov-87 Sep-02 -20 Sep-02 -20 Feb-09 Sep-74 Sep-74 -30 Feb-09 -30 -40 -40 Return (%) Return -50 -50 Jun-32 -60 -60 (%) Return Portfolio Jun-32 Portfolio Portfolio y = 0.5877x + 0.4202 -70 y = 0.6521x + 0.4326 -70 R² = 0.9576 R² = 0.931 -80 -80 -90 -90 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 S&P 500 Drawdown (%) S&P 500 Drawdown (%)

Source: Ibbotson Associates, Standard & Poor’s, as of 6/30/2019

FCERA Crisis Risk Program 11 October 17, 2019 Mortgages

Annual Crisis Return 32-Year History Return Feb-09 Sep-02 Nov-87 Evenly splitting the allocation All Stock Portfolio 10.2 -51 -45 -30 between Long Treasuries and MBS is -1.3 +20 +24 +13 historically equivalent to a pure Long Sixty-Forty Portfolio 8.9 -31 -21 -17 Treasury allocation +0.0 +4 +4 +2 90% Sixty-Forty, 8.9 -27 -17 -15 10% Long Treasuries -0.1 +0 +0 +0 90% Sixty-Forty, 8.8 -27 -17 -15 10% Long Treas + MBS

. High quality pass-through MBS held their value in the last three crises

. Mortgage spread helps compensate for current low Treasury yields

Source: Ibbotson Associates, Standard & Poor’s, Bloomberg Barclays, as of 6/30/2019

FCERA Crisis Risk Program 12 October 17, 2019 Risk Parity

Annual Crisis Return 93-Year History Return Jun-32 Feb-09 Sep-02 Sep-74 Nov-87 A simple 2-Asset Risk Parity model provided All Stock Portfolio 10.1 -83 -51 -45 -43 -30 significant additional protection in the 3 -1.5 +21 +20 +24 +15 +13 biggest crises, and actually added some Sixty-Forty Portfolio 8.6 -62 -31 -21 -28 -17 long-term return, i.e. had negative cost +0.2 +2 +2 +3 +0 +0 90% Sixty-Forty, 8.8 -60 -29 -18 -28 -17 10% 2-Asset Risk Parity

. Caveat 1: to create full history, 90% Sixty-Forty, 10% 2-Asset Risk Parity Sixty-Forty Portfolio this model is only and 20 20 bonds, i.e. no commodities 10 10 (more on this later) 0 0 -10 -10 . Caveat 2: historical results Nov-87 Nov-87 Sep-02 -20 -20 benefitted from secular Sep-02 Sep-74 Sep-74 downward trend in yields Feb-09 -30 Feb-09 -30 -40 -40 Return (%) Return -50 -50 -60 -60 Jun-32 (%) Return Portfolio Jun-32 Portfolio Portfolio y = 0.6287x + 0.3905 -70 y = 0.6521x + 0.4326 -70 R² = 0.9396 -80 R² = 0.9576 -80 -90 -90 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 S&P 500 Drawdown (%) S&P 500 Drawdown (%)

Source: Ibbotson Associates, Standard & Poor’s, as of 6/30/2019

FCERA Crisis Risk Program 13 October 17, 2019 Trend Following

Annual Crisis Return 19-Year History Return Feb-09 Sep-02 An index of Trend Following Commodities Trading All Stock Portfolio 5.4 -51 -45 Advisors (CTAs) provided significant additional +0.3 +20 +24 protection in the 2 most recent crisis, along with some Sixty-Forty Portfolio 5.7 -31 -21 additional long-term return +0.4 +5 +6 90% Sixty-Forty, 6.1 -26 -15 10% Trend Following

. Shorter history = less 90% Sixty-Forty, 10% Trend Following Sixty-Forty Portfolio confidence 20 20 10 10 . May not work during short, 0 0 sharp crises like 1987 -10 -10 Sep-02 . Can be implemented without -20 Sep-02 -20 Feb-09 fund structure -30 Feb-09 -30 -40 -40 Return (%) Return -50 -50 -60 -60 Portfolio (%) Return Portfolio Portfolio Portfolio -70 y = 0.4286x - 0.2872 y = 0.5518x - 0.0092 -70 R² = 0.9361 -80 R² = 0.9767 -80 -90 -90 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 S&P 500 Drawdown (%) S&P 500 Drawdown (%)

Source: Ibbotson Associates, Standard & Poor’s, Eurekahedge, as of 6/30/2019

FCERA Crisis Risk Program 14 October 17, 2019 Precious Metals

Annual Crisis Return 32-Year History Return Feb-09 Sep-02 Nov-87 A sizable allocation to gold provided meaningful All Stock Portfolio 10.2 -51 -45 -30 additional protection, but was a significant drag on -1.3 +20 +24 +13 long term return Sixty-Forty Portfolio 8.9 -31 -21 -17 -0.3 +4 +3 +2 90% Sixty-Forty, 8.6 -27 -18 -15 10% Gold

. Gold had a 4.1% annualized 90% Sixty-Forty, 10% Gold Sixty-Forty Portfolio return over 32 years, vs. 7.6% 20 20 for Long Treasuries 10 10 0 0 . Source of return is political and -10 -10 Nov-87 behavioral, not productive Sep-02 Nov-87 -20 Sep-02 -20 assets Feb-09 -30 Feb-09 -30 . Benefit: doesn’t add rate risk -40 -40 Return (%) Return -50 -50 . Additional benefit: inflation -60 -60 protection (%) Return Portfolio Portfolio Portfolio y = 0.4766x - 0.1294 -70 y = 0.5559x - 0.1118 -70 R² = 0.9498 -80 R² = 0.9672 -80 -90 -90 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 S&P 500 Drawdown (%) S&P 500 Drawdown (%)

Source: Ibbotson Associates, Standard & Poor’s, as of 6/30/2019

FCERA Crisis Risk Program 15 October 17, 2019 Derivatives

Annual Crisis Return 32-Year History Return Feb-09 Sep-02 Nov-87 Simple options strategies provided significant All Stock Portfolio 10.2 -51 -45 -30 additional protection but were relatively expensive -1.3 +20 +24 +13 Sixty-Forty Portfolio 8.9 -31 -21 -17 -1.2 +2 +7 +5 Sixty-Forty + 7.7 -29 -14 -12 Costless Collar

. Option sellers require volatility Sixty-Forty + Costless Collar Sixty-Forty Portfolio 20 20 10 10 . Cost may be reduced by more 0 0 sophisticated strategies -10 -10 Sep-02 Nov-87 Nov-87 . Protection is increased by more -20 Sep-02 -20 Feb-09 frequent adjustment of hedging -30 Feb-09 -30 position -40 -40 Return (%) Return -50 -50 -60 -60 Portfolio (%) Return Portfolio Portfolio Portfolio y = 0.4527x + 0.0497 -70 y = 0.5559x - 0.1118 -70 R² = 0.8687 -80 R² = 0.9672 -80 -90 -90 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 S&P 500 Drawdown (%) S&P 500 Drawdown (%)

Source: Ibbotson Associates, Standard & Poor’s, Cboe, as of 6/30/2019

FCERA Crisis Risk Program 16 October 17, 2019 Other Strategies

1. Low Volatility Equity • Has produced excess risk-adjusted return • Also makes sense for crisis risk, but not much history • Significant beta = relatively small effect

2. Discretionary re-allocation (“Trigger Fund”) • During crisis, use crisis risk assets to buy highly undervalued assets

FCERA Crisis Risk Program 17 October 17, 2019 Strategies Ranked

Strategy Protection Cost History Future Carry Active Risk (Expected (Improvement in (Change in (Crises Deviation from Regression) Historical Return) Observed) Benchmark)

Core Bonds 0.35 -1.5 5 Moderate Low Long Treasuries 0.06 -0.2 5 Low Low Mortgages 0.08 -0.1 3 Low Low Risk Parity 0.02 +0.2 5 Moderate Moderate Trend Following 0.12 +0.4 2 Moderate High Precious Metals 0.08 -0.3 3 None Low Derivatives 0.11 -1.2 3 Negative Moderate

. Strategies above the line are suitable for most institutions, and likely to be cost effective. . Strategies for further discussion:

1. Core Bonds

2. Long Treasuries

3. Risk Parity

4. Discretionary Re-allocation

FCERA Crisis Risk Program 18 October 17, 2019 The Most Important Program Element…

…is You!

. Some relative underperformance must be tolerated during normal times . “Cancelling the insurance” right before it pays off is worse than never having it

FCERA Crisis Risk Program 19 October 17, 2019 Strategy Discussions

FCERA Crisis Risk Program 20 October 17, 2019 For Each Strategy

1. What is it? 2. Why does it work?

FCERA Crisis Risk Program 21 October 17, 2019 Core Bonds

FCERA vs Core US Verus CMA Correlations Large FCERA LT Global Asset Class US TIPS 0.1 Target Agg US Treasury -0.3 Global Sovereign ex US 0.3 Credit (+ Bank Loans) 7% 30% 6% 25% Global Sovereign ex US Hdg -0.1 High Yield 3% 13% US Core -0.1 Gov 7% 30% 14% 60% US Core Plus 0.2 Securitized 3% 15% ST Govt/Credit 0.0 EMD 3% 13% Short-Term Credit 0.3 TIPS 3% 13% Long-Term Credit 0.2 US HY 0.6 Total 23% 100% 23% 100% Bank Loans 0.4 Global Credit 0.6 Global Credit Hdg 0.4 Alternative implementations: EMD USD 0.5 EMD Local 0.6 1. Restructure existing fixed income

. Reduce High Yield, Emerging Market Debt, & TIPS

. Increase allocation to Securitized Debt (MBS) 2. Fund new allocation to Core Bonds

Source: Bloomberg Barclays Global Family of Indices, June 2019

FCERA Crisis Risk Program 22 October 17, 2019 Long Treasuries

. Bond Index Stats “Flight to Quality” phenomenon => negative correlation between Treasuries and S&P 500 Equity Yield Duration Maturity Correl . Moving from credit to broad Treasuries achieves negative correlation but costs yield US Credit 3.09 10.9 7.4 0.3 US Treasury 1.92 7.9 6.4 -0.3 . Moving from broad Treasuries to Long Treasuries Long US Treasury 2.47 25.0 18.0 -0.3 recaptures yield but greatly increases duration/volatility

. Increased Treasury duration is beneficial during a crisis

. Increased Treasury duration is harmful during a recovery

. Increased Treasury duration is also harmful in inflationary times

. Alternative implementations:

1. Lengthen duration of existing allocation to sovereign debt

2. Fund new allocation to Long Treasuries

Source: Bloomberg Barclays Global Family of Indices, June 2019, and Ibbotson Associates

FCERA Crisis Risk Program 23 October 17, 2019 Simplest Possible Risk Parity – Step 1

Sixty- Forty

S&P 500 60% Core Bonds 40% T Bills 100% Standard Deviation 11.5%

Source: Ibbotson Associates, Standard & Poor’s, as of 6/30/2019

FCERA Crisis Risk Program 24 October 17, 2019 Simplest Possible Risk Parity – Step 2

Sixty- Levered De-levered Forty Core Bonds S&P 500

S&P 500 60% 62% Core Bonds 40% 246% T Bills -146% 38% 100% 100% 100% Standard Deviation 11.5% 11.5% 11.5%

Source: Ibbotson Associates, Standard & Poor’s, as of 6/30/2019

FCERA Crisis Risk Program 25 October 17, 2019 Simplest Possible Risk Parity – Step 3

Sixty- Levered De-levered 2-Asset Forty Core Bonds S&P 500 Weight Risk Parity

S&P 500 60% 62% x 67% = 42% Core Bonds 40% 246% x 67% = 165% T Bills -146% 38% -106% 100% 100% 100% 100% Standard Deviation 11.5% 11.5% 11.5% 11.5%

Source: Ibbotson Associates, Standard & Poor’s, as of 6/30/2019

FCERA Crisis Risk Program 26 October 17, 2019 Effectiveness of 2-Asset Risk Parity

With only 2 asset classes, the levered bond allocation had a strong protective effect during the last two crises

However, it gave little protection during the inflationary Oil Crisis

It had its own little crisis during the “War on Inflation”

Source: Ibbotson Associates, Standard & Poor’s, as of 6/30/2019

FCERA Crisis Risk Program 27 October 17, 2019 Effectiveness of 3-Asset Risk Parity

With the characteristic 3 asset classes (stocks, bonds, and commodities), our Risk Parity model would have disappointed during the Great Financial Crisis, as commodities prices went down and stayed down

However, it was better behaved during the War on Inflation

Source: Ibbotson Associates, Standard & Poor’s, as of 6/30/2019

FCERA Crisis Risk Program 28 October 17, 2019 Reality Check

With only 3 asset classes, our simple model tracks the new S&P 12% Parity Index very well over almost all of its (back-filled) history, with a correlation of 87%

Source: Ibbotson Associates, Standard & Poor’s, as of 6/30/2019

FCERA Crisis Risk Program 29 October 17, 2019 Risk Parity Conclusions

. The levered bonds are the part that works for crisis risk . Typical implementation is more diversified, so may disappoint . Best implementation for Crisis Risk Program: • TIPS better than commodities • Treasuries better than Credits • Passive better than active

FCERA Crisis Risk Program 30 October 17, 2019 Discretionary Re-Allocation (Trigger Fund)

. Scope 1. Between major asset classes (uncommon) 2. Within major asset class • Example: re-allocate from Treasuries to High Yield . Who “pulls the trigger”? 1. Board • Advantage: full control 2. Staff • Advantage: nimble 3. Manager • Advantage: market expertise

FCERA Crisis Risk Program 31 October 17, 2019 Conclusion

FCERA Crisis Risk Program 32 October 17, 2019 Summary

1. There are cost-effective ways to control crisis risk 2. Program requires long-term commitment 3. FCERA can start small • Trade some current yield for future protection and opportunity • Add Risk Parity allocation

FCERA Crisis Risk Program 33 October 17, 2019 Appendix: Introduction to Risk Parity

FCERA Crisis Risk Program 34 October 17, 2019 Overview

Risk parity is an approach to investment portfolio management which focuses on the allocation of risk, defined as volatility, rather than the allocation of capital.

Risk parity is not an asset class per se (in the way that equities, fixed income, and commodities are), but more of an approach to managing a group of asset classes. — Therefore, Risk Parity would not introduce any new asset classes into the Plan, but would instead introduce a new strategy to managing certain asset class exposures already in the Plan.

FCERA Crisis Risk Program 35 October 17, 2019 What is risk parity?

A long-only portfolio seeking to generate returns through persistent exposure to Global Equity, Global Fixed Income, and Global Inflation-Protected Assets

— Global Equities for upside participation in periods of strong growth

— Global Fixed Income for downside protection in periods of weaker growth

— Commodities and Global Inflation-Linked Bonds to preserve real rates of return in inflationary periods

Seeks balanced exposure for consistent performance across market environments.

Source: PanAgora , Inc.

FCERA Crisis Risk Program 36 October 17, 2019 Portfolio construction

The Traditional Approach to portfolio construction The Risk Parity approach would — Accepts the volatility level of the asset classes available in the marketplace apply to a — Constructs a portfolio with those asset classes portion of the Plan, and the — Each asset class contributes a different level of risk to the portfolio traditional approach would Return is the objective and risk is an output apply to the remainder of the Plan. The Risk Parity Approach to portfolio construction

— Uses leverage to adjust the volatility of each asset class to similar levels

— Weights the asset class exposures so that each asset class contributes a similar amount of expected volatility to the portfolio

— Each asset class contributes similar levels of risk to the portfolio

Risk is the objective and return is an output

FCERA Crisis Risk Program 37 October 17, 2019 Traditional portfolio

CAPITAL AND RISK ALLOCATION The Traditional Approach 100% may result in the Plan 10% 12% closely tracking the 3% outcome of the equity 80% market. 35%

60%

85% 40%

55% 20%

0% Capital Allocation Risk Allocation

Equity Fixed Income Real Assets

Source: PanAgora Asset Management, Inc.

FCERA Crisis Risk Program 38 October 17, 2019 Risk parity portfolio

CAPITAL AND RISK ALLOCATION The Risk Parity Approach, 250% through the use of 40% leverage, weights assets so that all investments 200% influence the portfolio similarly.

150%

166%

100% 20%

40% 50%

44% 40% 0% Capital Allocation Risk Allocation

Equity Fixed Income Real Assets

Source: PanAgora Asset Management, Inc.

FCERA Crisis Risk Program 39 October 17, 2019 Leverage in portfolios

Just like Risk Parity, most investments contain leverage In Risk Parity, leverage is — Public and Private Equity, Real Estate, Infrastructure, and Hedge Funds generally — Leverage is embedded throughout the portfolio but often encapsulated obtained through the use — (or in the case of Risk Parity – commingled fund) allows it to be non-recourse leverage of derivatives.

Risks of leverage can be controlled with good risk management

— Levering a diversified portfolio of liquid assets

— Active rebalancing to target constant and balanced

— Maintaining high levels of unencumbered cash

— Utilizing instrument leverage through exchange-traded and centrally cleared futures (not borrowed leverage)

FCERA Crisis Risk Program 40 October 17, 2019 Risk parity overview

ANNUAL PERFORMANCE 12 MO. EXCESS PERFORMANCE VS. GLOBAL EQUITIES , 40% 30% 20% 10% 0% -10%

Excess Return, % -20% -30% -40% -60% -40% -20% 0% 20% 40% 60% 80% Benchmark Return, % 363 YR.MONTH ROLLING RO ING CORR. CORR ATIONW/ GLOBAL EQUITIES 0.8 0.7 0.6 0.5 0.4 0.3 Correlation 0.2 0.1 0.0 2005 2006 2006 2007 2007 2008 2008 2009 2009 2010 2010 2011 2011 2012 2012 2013 2013 2014 2014 2015

MSCI ACWI IMI NR USD Salient Risk Parity Index 10 Vol

As of March 31, 2016 Benchmark: Salient Risk Parity Index 10 vol

FCERA Crisis Risk Program 41 October 17, 2019 Appendix: Intro to Trend Following CTA

FCERA Crisis Risk Program 42 October 17, 2019 Commodity Trading Advisor (CTA)

— Approach: constructs signals based on technical indicators or proprietary analytical models. Strategies will differ based on time horizon, markets traded, and research capabilities. — Portfolio: Traditionally futures and forwards, but not uncommon to see cash equities used as well. Constructed and managed systematically by optimization process, usually with higher leverage than other strategies. — Example: trades around short-term momentum signals in commodity markets

PERFORMANCE AND RISK – LAST 5 YEARS CORRELATION AND BETA – LAST 5 YEARS (1.0) (0.5) 0.0 0.5 1.0 1.5 14 12 Beta 10 BC US Agg. 8 Beta MSCI ACWI 6 4 Correlation BC US Agg. 2 Correlation 0 MSCI ACWI (2) Annualized Return % Annualized Std. Dev. % HFRI Macro Systematic MSCI ACWI BC US Agg. HFRI Macro Systematic

FCERA Crisis Risk Program 43 October 17, 2019