The Systemic Effects of Benchmarking

Transcription

1 1 The Systemic Effects of Benchmarking Boston University Joint work with: Diogo Duarte, Boston University Keith Lee, Boston University Securities Markets: Trends, Risks and Policies Conference February 2016

2 The Systemic Effects of Benchmarking 2 Asset management Asset management is a very large industry that moves a substantial amount of money on a day to day basis in the global financial markets One special feature of the asset management industry is that managers are rewarded for returns that beat the market as measured by a benchmark Therefore, asset managers have incentives to invest differently than predicted by the standard mean-variance theory As long as the music is playing, you ve got to get up and dance. Charles Prince, CEO Citigroup

3 The Systemic Effects of Benchmarking 3 Can benchmarking induce trading behavior by asset managers that results in increased risks and instabilities in financial markets?

4 The Systemic Effects of Benchmarking 4 Our results We consider an exchange economy with one retail and two institutional investors, a benchmark and an alternative stock, and short-term borrowing and lending The retail investor is myopic Institutional investors have incentives to gain higher exposure to the benchmark stock Buffa, Vayanos & Wolley (2015), Cuoco & Kaniel (2011)

5 The Systemic Effects of Benchmarking 5 Our results Strong benchmarking incen0ves Benchmark underperforms High por5olio sensi0vi0es for ins0tu0onal investors Benchmark outperforms Low por5olio sensi0vi0es for ins0tu0onal investors Fire sales following nega0ve news Buy-and-hold strategies for ins0tu0onal investors Large vola0lity due to high trading volume Low trading volume and low vola0lity Large value-at-risk-for retail investor Low value-at-risk-for retail investor

6 The Systemic Effects of Benchmarking 6 Our results Benchmarking is welfare reducing for retail investors We establish this by looking at the equivalent variation of consumption, which is negative in all states for retail investors However, benchmarking does not affect the long-term performance of any investor In the long run, the most patient investor wins out Yan (2008)

7 The Systemic Effects of Benchmarking 7 Implications Our results contribute to the recent discussion on the systemic importance of the asset management industry FSOC (2014), Haldane (2014), IMF (2015), OFR (2013) We show that benchmarking may incentivize asset managers to invest in ways that have systemic effects on retail investors Nonetheless, tail risks do not materialize in the long run It is imperative for regulators to formulate precise goals for a regulation of the asset management industry

8 The Systemic Effects of Benchmarking 8 Implications Our results also show that the regulation of asset management needs to be designed differently than the regulation of banks Jones (2015) In our model, retail investors are exposed to high tail risk whenever institutional investors are exposed to low tail risk Regulatory tools that target the tail risk exposure of institutional investors, such as stress tests and value-at-risk reporting, may not be able to identify systemic threats to retail investors

9 The Systemic Effects of Benchmarking 9 The model There are two stocks, S 1 and S 2, with claims to dividends: dd 1,t D 1,t = µ 1 dt + σ 1 dz t + J 1 dn t, dd 2,t D 2,t = µ 2 dt + σ 2 dz t + J 2 dn t Here, Z is a standard Brownian motion, and N is a Poisson process with jump rate λ There is also riskless borrowing and lending at the stochastic interest rate r t

10 The Systemic Effects of Benchmarking 10 Investors The retail investor is a standard myopic mean-variance optimizer with log-utility Institutional investors benchmark against asset 2 This means that states of the world in which asset 2 outperforms relative to asset 1 are expensive states of the world for institutional investors In such states, it is difficult for institutional investors to beat the benchmark States of the world in which asset 2 outperforms relative to asset 1 carry high marginal utility for institutional investors

11 The Systemic Effects of Benchmarking 11 Institutional investors In order to capture these effects, we assume that institutional investor j {A, B} solves the following optimization problem subject to her budget constraint: [ ] sup E e ρjt (1 + I j s t ) log c j tdt (c j t,πj t ) t 0 0 Here, s t = D 2,t D 1,t +D 2,t benchmark is the fraction of dividends attributed to the I j > 0 parametrizes the importance of the benchmark in the utility of institutional investors

12 The Systemic Effects of Benchmarking 12 Our model formulation We can interpret consumption of institutional investors as managerial compensation Our formulation captures unique features of managerial compensation (Basak & Pavlova (2013)) Managerial compensation depends on the benchmark Marginal utility of compensation increases when the benchmark increases Managerial compensation may increase when the benchmark underperforms

13 The Systemic Effects of Benchmarking 13 Equilibrium Stocks 1 and 2 are in positive supply of one unit Investor j owns a fraction α j of wealth at time 0: W j 0 = αj W 0 General equilibrium quantities are computed semi-analytically using a variant of the methodology of Martin (2013) Preferences ρ A = 0.03 ρ B = 0.02 ρ R = 0.01 I A = I B [0, 100] Dividends µ 1 = µ 2 = 0.08 λ = 0.05 σ 1 = σ 2 = 0.17 J 1 = 0 J 2 = 0.1 D 1,0 = 2 D 2,0 [0.01, 10] Initial Wealth α A = α B = α R = 0.25

14 The Systemic Effects of Benchmarking 14 s 0 = 0.1 s 0 = 0.5 Value-at-risk of retail investor s 0 = 0.1 s 0 = Benchmark importance parameter Benchmark importance parameter Investor R Aggregate Value at risk s 0 = 0.1 s 0 = 0.8 s 0 = 0.5 Value at risk s 0 = 0.8 s 0 = 0.1 s 0 = Benchmark importance parameter Benchmark importance parameter

15 The Systemic Effects of Benchmarking 15 Value-at-risk of institutional investors Investor A Investor B Value at risk s 0 = 0.8 s 0 = 0.1 s 0 = 0.5 Value at risk s 0 = 0.8 s 0 = 0.1 s 0 = Benchmark importance parameter Benchmark importance parameter Investor R Aggregate s 0 = s 0 = 0.8

16 The Systemic Effects of Benchmarking 16 Portfolios weights for stocks Benchmark stock Portfolio weight of investor A I A =I B = 5 I A =I B = 0 Portfolio weight of investor B I A =I B = 5 I A =I B = 0 Portfolio weight of investor R I A =I B = 0 I A =I B = 5 Benchmark dividend ratio Benchmark dividend ratio Benchmark dividend ratio Alternative stock Portfolio weight of investor A I A =I B = 0 I A =I B = 5 Portfolio weight of investor B I A =I B = 0 I A =I B = 5 Portfolio weight of investor R I A =I B = 5 I A =I B = 0 Benchmark dividend ratio Benchmark dividend ratio Benchmark dividend ratio

17 The Systemic Effects of Benchmarking 17 Portfolio volatilities Investor A Investor B Investor R Portfolio volatility I A =I B = 0 I A =I B = 5 Portfolio volatility I A =I B = 0 I A =I B = 5 Portfolio volatility of investor R I A =I B = 0 I A =I B = 5 Benchmark dividend ratio Benchmark dividend ratio Benchmark dividend ratio

18 The Systemic Effects of Benchmarking 18 Survival All investors survive in the long run t if at least one dividend process has positive expected growth rate In this case, the most patient investor dominates in terms on relative wealth as t (Yan (2008)) Benchmarking does not affect the long-term performance of our investors

19 The Systemic Effects of Benchmarking 19 Welfare impact We evaluate the welfare impact of benchmarking by analyzing the equivalent variation of consumption How much more does an investor need to consume in a world without benchmarking incentives to achieve the same utility as in a world with benchmarking incentives? When there are no benchmarking incentives, all investors are myopic with log-utilities Therefore, a negative equivalent variation of consumption implies a welfare loss

20 The Systemic Effects of Benchmarking 20 Equivalent variation of consumption Investor R Equivalent variation of consumption I A =I B = 0 I A =I B = 5 Benchmark dividend ratio

21 The Systemic Effects of Benchmarking 21 Summary Benchmarking may induce trading behavior by institutional investors that exposes a retail investor and the aggregate market to tail risk Benchmarking increases the portfolio sensitivities of institutional investors in states of the world in which the benchmark underperforms In such states, institutional investors trade large volumes and execute fire sales, which raises market volatility and exposes the retail investor to tail risk

22 The Systemic Effects of Benchmarking 22 Summary There is no impact on the long-term performance of our investors However, benchmarking is welfare reducing for the retail investor These results have important implications for the regulation of the asset management industry

23 The Systemic Effects of Benchmarking 23 References Basak, Suleyman & Anna Pavlova (2013), Asset prices and institutional investors, American Economic Review 103, Buffa, Andrea, Dimitri Vayanos & Paul Wolley (2015), Asset management contracts and equilibrium prices. Working Paper, Boston University and London School of Economics. Cuoco, Domenico & Ron Kaniel (2011), Equilibrium prices in the presence of delegated portfolio management, Journal of Financial Economics 101(2), FSOC (2014), Notice fsoc Haldane, Andrew G. (2014), The age of asset management?. Speech given at the London Business School, London, United Kingdom.

24 The Systemic Effects of Benchmarking 24 IMF (2015), Financial system stability assessment. International Monetary Fund Country Report No. 15/170. Jones, Brad (2015), Asset bubbles: Re-thinking policy for the age of asset management. IMF Working Paper. Martin, Ian (2013), The Lucas Orchard, Econometrica 81, OFR (2013), Asset management and financial stability. Yan, Hongjun (2008), Natural selection in financial markets: Does it work?, Management Science 54(11),

25 The Systemic Effects of Benchmarking 25 Thank you!

26 The Systemic Effects of Benchmarking 26 Fire sales Investor A Non benchmark stock Substitution Total Income Benchmark stock Substitution Income Total Safe asset Total Income Substitution Benchmark importance parameter Benchmark importance parameter Investor R Benchmark importance parameter Non benchmark stock Substitution Total Income Benchmark stock Income Substitution Total Safe asset 4e 04 2e 04 0e+00 Substitution Income Total Benchmark importance parameter Benchmark importance parameter Benchmark importance parameter

27 The Systemic Effects of Benchmarking 27 Excess consumption of institutional investor Excess consumption of institutional investor I A =I B = 5 I A =I B = 0 Excess consumption of institutional investor I A =I B = 5 I A =I B = Benchmark dividend ratio Benchmark dividend level