Diversification of Portfolio Risk Reconciling Theory and Observed

Diversification of Portfolio Risk: Reconciling Theory and Observed Weightings Cath Jackson Department of Town and Regional Planning University of Sheffield ERES Conference, Eindhoven, June 2011 1

Background • Markowitz (1952) set out principles underlying portfolio theory • Less than perfect positive correlation = diversification of risk • MPT set out Mean Variance Criterion – Efficient Frontier shows all possible (efficient) portfolios • Asset classes provide practical structure for investors – How define asset classes? – “care must be used in using and interpreting relations among aggregates. We cannot deal here with the problems and pitfalls of aggregation. ” (Markowitz, p. 91) ERES Conference, Eindhoven, June 2011 2

Asset classes? • US: consistently challenge NCREIF or Salomon Brothers areas – Hartzell et al. (1987); Malizia and Simons (1991); Mueller and Ziering (1992); Mueller (1993); Eichholz et al. (1995); Goetzmann and Wachter (1995); Ziering and Hess (1995); Nelson and Nelson (2003) • UK: consistently challenge Administrative Regions – Eichholz et al. (1995); Hoesli et al. (1997); Lee and Byrne (1998); Hamelink et al. (2000); Jackson (2002); Jackson and White (2005 a, b); Heydenreich (2010); Byrne and Lee (2011) • Local area characteristics and property market fundamentals more important for optimal risk diversification ERES Conference, Eindhoven, June 2011 3

Hindsight? • Ten years have passed. . . – what would have happened if strategies had changed? • Hamelink, Hoesli, Lizieri and Mac. Gregor (2000) (HHLM) – 157 markets, cluster analysis, estimated total returns, CB Hillier Parker – Nine relatively homogeneous clusters, mixed sectors – “conventional UK administrative and statistical regional classifications do not provide useful information in structuring a portfolio strategy” • Jackson (2002) retail, Jackson and White (2005 a) industrial, Jackson and White (2005) office (JJW) – 322 markets, cluster analysis, rental growth rates, IPD – Fifteen (sixteen) relatively homogeneous clusters, distinct sectors – Regions largely not evident, market and occupier factors important ERES Conference, Eindhoven, June 2011 4

Methods and Data • HHLM and JJW classifications compared – – – Original data to 1996/7; Now 1998+ Efficient frontiers developed Also compared to regional classes and sectoral classes Optimum weights compared to observed allocations (weighted by CV) Temporal performance/patterns: 1981 -1985, 1986 -1992, 1993 -1997 • IPD total returns data – – 73 markets common to HHLM, JJW and IPD (average fund holds 43) 23 retail, 18 industrial, 32 office markets No London retail Size of cluster membership affects variance ERES Conference, Eindhoven, June 2011 5

1998 -2007 15 Portfolio Return (%) 14 13 HHLM JJW Regions Sectors 12 11 10 5 6 7 8 Portfolio Risk (%) ERES Conference, Eindhoven, June 2011 9 10 11 6

HHLM weights High 6. 4 9. 5 21. 0 0. 0 46. 9 0. 0 31. 4 7. 1 13. 8 23. 6 0. 0 24. 1 0. 0 65. 3 0. 0 1. 8 13. 2 17. 7 0. 0 2. 0 0. 0 51. 9 0. 0 44. 5 0. 0 3. 6 0. 0 18. 5 0. 0 81. 5 0. 0 0. 0 100. 0 0. 0 Scottish Retail ERES Conference, Eindhoven, June 2011 Retail II 16. 2 Retail I London Fringe Mainly Industrial Indus London Office London Fringe Mainly Offices Low 3 Retail 4 Off / Indus City Offices Risk Peripheral Office and Industrial Return Northern Industrial and Office No patterns Central London Offices 1998 -2007 Weights (percentage in each cluster) 7

Observed (CV) 33. 1 33. 7 34. 7 35. 5 33. 0 30. 3 30. 2 30. 3 33. 2 35. 5 3. 6 4. 1 4. 0 3. 5 3. 7 4. 0 3. 5 3. 3 2. 9 2. 1 3. 6 3. 7 4. 0 4. 8 5. 1 5. 0 5. 1 4. 8 4. 7 Retail II 5. 9 5. 8 5. 4 4. 7 5. 3 5. 4 5. 0 4. 4 4. 3 London Fringe Mainly Offices London Fringe Mainly Industrial 12. 3 12. 0 11. 7 12. 5 11. 9 10. 1 9. 7 10. 6 12. 7 13. 3 Retail I 18. 7 18. 0 18. 5 18. 7 19. 9 21. 1 20. 8 20. 7 20. 0 Central London Offices 3. 6 3. 4 3. 6 3. 7 4. 0 4. 5 4. 9 5. 5 4. 9 Scottish Retail 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 City Offices Northern Industrial and Office Peripheral Office and Industrial 1998 -2007 Weights (percentage in each cluster) 2. 6 2. 5 2. 3 2. 2 2. 4 2. 7 3. 0 2. 8 2. 4 2. 2 16. 5 16. 9 16. 3 15. 6 16. 8 17. 4 16. 7 14. 0 12. 9

Temporal stability 1981 -85: ‘Medium’ variance 16 11 6 0 1 2 3 4 5 6 7 8 1986 -92: ‘High’ variance 17 12 7 5 8 11 1993 -97: ‘Low’ variance 14 14 9 4 5 6 ERES Conference, Eindhoven, June 2011 7 8 9 10 9

Retail 7: Northern, large, declining Industrial 1: Regionally diverse Industrial 2: Regionally diverse / southern Industrial 4: South east and eastern I Industrial 5: South east and eastern II Industrial 6: London and south east Office 1: Southern and regional Office 2: Peripheral London and southern Office 3: London 0. 0 0. 8 0. 0 0. 6 0. 9 0. 0 0. 0 76. 1 73. 8 51. 4 25. 6 12. 7 0. 0 0. 0 0. 0 0. 0 1. 2 13. 4 19. 6 54. 1 82. 0 100. 0 0. 0 2. 4 3. 7 0. 0 0. 0 8. 8 0. 0 0. 0 52. 0 51. 7 51. 5 32. 1 19. 7 6. 9 0. 0 0. 0 13. 1 27. 7 45. 4 100. 0 0. 0 23. 8 31. 7 39. 6 64. 9 73. 0 81. 3 91. 5 13. 7 0. 0 9. 9 20. 1 46. 6 51. 7 55. 0 54. 6 0. 0 0. 0 0. 0 15. 1 5. 0 0. 0 86. 3 100. 0 0. 0 0. 0 0. 0 68. 9 68. 2 67. 2 53. 4 35. 2 17. 3 0. 0 0. 0 31. 1 21. 9 12. 8 0. 0 0. 0 0. 0 0. 0 0. 0 Medium Retail 6: Southern, good employment 0. 0 20. 4 47. 4 61. 0 67. 8 45. 9 18. 0 0. 0 3. 0 6. 7 8. 6 4. 8 0. 0 0. 0 High Retail 5: Northern 0. 0 0. 0 24. 2 16. 6 9. 0 0. 0 Low Retail 4: Southern, large, service centres High Retail 3: Southern, mixed employment High Low Retail 2: Southern, good employment High Low Retail 1: Large/medium Return Risk Suggested weightings Low

NE NW&M Scot SE SW Wales WM Y&H Retail Indus Office 62. 3 61. 5 59. 9 59. 8 58. 8 60. 8 63. 7 65. 5 63. 5 59. 9 54. 5 48. 8 47. 8 45. 6 46. 1 46. 5 46. 2 47. 4 48. 0 48. 5 49. 9 47. 0 42. 8 42. 1 43. 3 0. 8 1. 3 1. 2 1. 1 0. 9 0. 8 1. 0 1. 2 1. 3 1. 4 1. 2 1. 1 1. 2 1. 4 1. 6 3. 7 3. 8 3. 6 4. 0 3. 5 2. 7 2. 6 2. 7 3. 1 4. 3 5. 0 5. 4 5. 8 6. 0 6. 3 6. 4 6. 3 6. 2 6. 0 6. 8 7. 3 7. 5 6. 9 7. 8 8. 0 7. 8 7. 7 7. 2 6. 7 7. 3 8. 0 8. 7 10. 2 11. 0 10. 6 11. 2 11. 1 10. 7 10. 8 11. 3 10. 6 10. 4 11. 1 10. 9 10. 4 6. 7 6. 6 7. 7 8. 2 8. 0 7. 7 7. 5 8. 2 8. 7 8. 9 9. 1 8. 8 8. 9 8. 1 8. 5 8. 6 8. 4 8. 2 8. 4 8. 5 8. 6 2. 8 3. 2 3. 3 3. 4 3. 3 2. 9 2. 8 3. 1 3. 5 3. 9 4. 2 4. 1 3. 9 3. 8 3. 2 2. 8 2. 6 2. 3 2. 1 2. 5 2. 8 3. 2 1. 1 1. 0 0. 8 0. 7 0. 8 1. 0 1. 1 1. 4 1. 5 1. 6 1. 3 1. 4 1. 5 1. 8 2. 0 4. 3 4. 1 3. 7 3. 2 2. 7 2. 9 3. 8 5. 2 5. 9 6. 4 6. 3 6. 0 6. 2 6. 9 6. 8 6. 6 6. 3 7. 3 8. 1 8. 0 5. 1 4. 9 5. 0 4. 8 4. 7 4. 3 4. 2 4. 1 4. 0 4. 3 4. 8 5. 4 5. 7 6. 4 6. 6 6. 9 6. 6 6. 3 5. 9 5. 6 6. 1 6. 7 6. 9 9. 8 10. 4 10. 9 12. 1 13. 9 14. 1 13. 6 12. 6 11. 3 11. 8 13. 8 15. 0 16. 2 16. 6 17. 2 18. 4 19. 2 17. 6 15. 9 16. 7 18. 9 20. 0 18. 5 7. 3 7. 1 6. 5 5. 8 5. 1 4. 7 5. 3 5. 6 6. 7 8. 0 8. 1 8. 7 9. 3 9. 1 9. 0 9. 4 9. 5 10. 3 11. 7 13. 2 13. 6 13. 8 82. 7 82. 4 82. 0 81. 3 80. 9 81. 7 82. 7 83. 4 82. 6 79. 5 76. 9 75. 7 74. 0 73. 7 72. 5 71. 6 71. 4 72. 8 73. 7 71. 7 67. 9 66. 4 67. 6 Medium London 1. 7 1. 6 2. 0 1. 9 2. 1 2. 3 2. 5 2. 6 3. 4 3. 6 3. 7 3. 8 3. 3 3. 0 3. 2 3. 5 3. 9 4. 0 3. 8 High EM 3. 0 3. 3 3. 5 3. 6 4. 0 4. 1 4. 0 4. 5 4. 9 4. 7 5. 3 4. 2 4. 7 4. 4 3. 7 4. 0 3. 6 4. 1 4. 3 4. 2 3. 8 3. 9 Low East Observed weightings 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Conclusions • Markowitz’s (1952) theory of portfolio selection helps guide optimal asset selection • Grouping of assets into classes subject to ongoing debate • Theory used to test HHLM and JJW classifications – Recognising importance of asset performance at local level • Efficient Frontiers (almost) always offer superior strategies • Prudent to examine weightings underlying efficient frontiers – Narrow allocations, heavy weights in few classes – Marked contrast to observed (aggregate) allocations ERES Conference, Eindhoven, June 2011 12

Conclusions • Reconciling theory and observed weightings. . . – Data? Sample? – Benchmarking? – Do investors affect the market, so others ‘have to’ follow? • Gabaix et al. (2006) quantified endogenous effect of investor buy/sell decisions in thin equities markets • Baum et al. (2000) suggest investors affect markets • Henneberry and Roberts (2008) suggest they may be endogenous • Dunse et al. (2007) empirical determine statistically significant effect of investor transactions on yields • Fisher et al. (2004) explore the importance of “transaction cycles” to institutional investors • Can theory and observed weightings be reconciled? ERES Conference, Eindhoven, June 2011 13

Diversification of Portfolio Risk: Reconciling Theory and Observed Weightings Cath Jackson Department of Town and Regional Planning University of Sheffield ERES Conference, Eindhoven, June 2011 14