MSA Geographic Allocations, Property Selection, & Performance Attribution in Public & Private Real Estate Markets

Transcription

1 + * MSA Geographic Allocations, Property Selection, & Performance Attribution in Public & Private Real Estate Markets David Ling, Andy Naranjo, and Ben Scheick NCREIF s Summer 2015 Conference

2 Where We Are Headed... Objectives of current research Differences in public & private geographic (MSA) concentrations Including allocations in gateway markets Impact of MSA allocations on relative performance Formal attribution analysis Key takeaways My spin/bias (if time allows)

3 Background & Motivation Returns in both private & public commercial real estate (CRE) markets should be driven at least in the longrun by CFs produced by underlying properties But private & public CRE often produce substantially different returns (and risk) At least according to standard return indices Especially over short time horizons

4 The GREAT DEBATE: Are equity REITs stocks or real estate?

5 One Way to Address the Question Do REITs & private direct investments perform similarly? Public market: % annually Difference = 92 bps annually Private market: 9.61% annually Problem: not comparing apples to apples!

6 Public vs. Private Market Performance Comparisons should account for differences in: Financial leverage Property Mix Management Fees E.g., Pagliari et al. (2005); Riddiough et al. (2005); Ling & Naranjo (2015)

7 We Therefore: Remove effects of leverage from firm-level equity REIT returns Unlevering is done quarterly (see paper for details) Exclude REITs that do not invest in core property types

8 # of Core REITs as % of Total 70.0% 60.0% 50.0% 40.0% 30.0% Indicates importance of excluding noncore REITs from comparative analysis 20.0% 10.0% 0.0% % Core Focused REITs

9 We Therefore: Remove effects of leverage from firm-level equity REIT returns Unlevering is done quarterly (see paper for details) Exclude REITs that do not invest in core property types Use same property-type weights as found in NPI index to create our aggregate core REIT return index, and Adjust NPI returns downward to reflect mgmt. expenses

10 We Therefore: Remove effects of leverage from firm-level equity REIT returns Unlevering is done quarterly (see paper for details) Exclude REITs that do not invest in core property types Use same property-type weights as found in NPI index to create our aggregate core REIT return index, and Adjust NPI returns downward to reflect mgmt. expenses Public market: 9.02% annually Difference = 23 bps annually Private market: 8.79% annually

11 Public vs. Private Market Performance What is still missing from these adjustments? MSA location of REIT properties vs. NPI properties! Performance differences may reflect differences in portfolio allocations across MSAs

12 Research Objectives Quantify differences in MSA portfolio allocations across public & private CRE markets Examine extent to which differences in MSA concentrations affect relative return performance Use formal attribution analysis to attribute the source of performance differences to: MSA allocations Property selection & management within MSAs Or some combination of the two (interaction)

13 Preview of Results Document significant differences in public vs. private MSA concentrations Adjusting NPI returns for differences in MSA concentrations leads to a (small) decline in core NPI performance i.e., NPI MSA allocations were superior to REIT MSA allocations This reweighting of NPI returns increases the magnitude of equity REIT outperformance during our sample period Decision to allocate to a particular MSA is relatively less important than manager s ability to select/manage properties within an MSA

14 Differences in MSA Allocations/Weights Data: SNL Real Estate Properties Database 500,000+ property year observations in our dataset from Measuring REIT MSA Concentrations: Sort each property held by each REIT at beginning of each year into an MSA Compute % of the REIT portfolio in each MSA at beginning of each year Use book value, adjusted cost basis, SF, and # of properties as weights Do this separately for 4 core property types (apartments, office, industrial, retail), as well as for aggregate core series Follow similar procedure to compute % of NCREIF portfolio in each MSA at the beginning of each year Based on market value

15 1 st Focus: Gateway City Allocations Much has been written & said by industry professionals about the desirability of investing in gateway MSAs Boston, New York, Washington, D.C., Chicago, Los Angeles, and San Francisco These MSAs are thought to have significant investment advantages over remaining 300-plus MSAs More liquidity & Constraints on production of new supply that puts upward pressure on rents Different allocations to these markets may be an important determinant of relative performance

16 Gateway % Allocations by Property Type

17 Value-Weighted Gateway NPI Returns minus Non-Gateway NPI Returns Apartment Industrial Office Retail Aggregate: Core Properties These are quarterly differences x 4 =1.05 percentage points (105 basis points) per year

18 Value-Weighted Gateway NPI Returns minus Non-Gateway NPI Returns Apartment Industrial Office Retail Aggregate: Core Properties

19 Differences in MSA Allocations 2013: REITs: < 1% NCREIF: 7.0% Significant difference in proportion of holdings within a particular year Magnitude of differences is time varying REIT weights are based on book value of properties

20 Differences in MSA Allocations 2003: REITs: 27.1% NCREIF: 11.2% Significant differences in holdings Again, magnitude of differences is time varying

21 Differences in MSA Allocations Differences in relative concentrations vary across property types within MSAs In LA, REITs did not dominate office from (compare to industrial)

22 Differences in MSA Allocations More evidence of large, time-varying differences in allocations

23 Do These Differences in MSA Allocations Affect Return Performance? Requires us to reweight NPI MSA-level returns (by property type) using MSA weights of REIT portfolio Allows for return comparison of MSA identical portfolios Various REIT weight measures used Note: Can t reweight REIT MSA-level returns using NPI MSA weights Why? Because we do not have MSA-level REIT returns

24 These are quarterly differences: Impact of Reweighting NPI MSA Returns? Mean core NPI return performance falls slightly i.e., using REIT MSA allocations drags down core NPI performance Cross-property mean differences: apartments, flat; industrial & office, positive; retail, negative) Table 4: Reweighted NPI Returns minus Unadjusted NPI Returns

25 Reweighted NPI Raw NPI Quarterly differences can be large & persistent! Figure 9 Panel B: Industrial Property Type Difference is +105 bps in 2009:Q1 Positive difference from 2002:Q2-2004:Q2 averaged 32 bps per quarter

26 Formal Attribution Analysis Question: What portion of REIT outperformance (during our sample) can be attributed to MSA allocations vs. property selection within MSAs? R REIT f,t R NPI f,t = allocation + selection + interaction

27 Attribution Analysis R REIT f,t R NPI f,t = allocation + selection + interaction Superior allocation (strategy): buying/selling in the right markets (MSAs) at the right time Superior selection (execution): Buying low selling high (acquisition team adds value) Superior asset/property management Interaction: Allocation affects selection; selection affects allocation

28 Attribution Analysis Main Result: Allocation effect constitutes small portion of total return difference Table 5 Attribution Analysis using NPI as Benchmark Relatively small allocation effect

29 Attribution Analysis Main Result: Allocation effect constitutes small portion of total return difference Table 5 Attribution Analysis using NPI as Benchmark Same relative story in recent years

30 Attribution Analysis Main Result: Allocation effect constitutes small portion of total return difference Table 5 Attribution Analysis using NPI as Benchmark

31 Attribution Analysis Main Result: Allocation effect constitutes small portion of total return difference Table 5 Attribution Analysis using NPI as Benchmark Again, small allocation effect Implication: Decision to allocate to particular MSA much less important than manager s ability to select (and manage) properties within each MSA!

32 Manager Level Attribution Analysis Can be done at the individual NCREIF manager level to explain performance relative to NPI benchmark Can be done at individual REIT level to explain performance relative to NPI benchmark Using Equity Residential (EQR) as example Table 8 Attribution Analysis using NPI as Benchmark Panel A: Return series Unlevered REIT Return (EQR) Raw NPI Returns - Apartment Reweighted NPI Returns (using EQR Weights)

33 Firm-Level Attribution Analysis Using Equity Residential (EQR) as example Table 8 Attribution Analysis using NPI as Benchmark Panel A: Return series Unlevered REIT Return (EQR) Raw NPI Returns - Apartment Reweighted NPI Returns (using EQR Weights) Panel B: Attribution analysis Unlevered REIT return (EQR) minus raw NPI - Apartment Pure allocation effect Selection effect plus interaction

34 Key Takeaways There are significant differences in MSA portfolio allocations between public & private markets Including gateway allocations These differences affect relative performance Especially over short time horizons However attribution analysis reveals that property selection (and management?) within a MSA is more important than allocation across MSAs Framework of benchmark adjustments & attribution analysis can also be applied at the manager/firm level

35 My Own Take/Spin Diversification is a powerful tool for increasing the risk-adjusted returns of portfolios It is the one free lunch in financial economics so geographic diversification is important But a naïve diversification strategy of randomly allocating investment capital across MSAs is likely to be as effective, and much less expensive, as an investment strategy that relies on return (or rent) forecasts aggregated to the MSA level I have always thought industry professionals spend too much time on MSA selection Perhaps because that is the level of aggregation where the data have been consistently and reliably collected

36 My Own Take/Spin A Random Walk Down Main Street: Can Experts Predict Returns on Commercial Real Estate (Ling, 2005) In addition to pursuing a naïve diversification strategy, we recommend investors focus on deal specifics, including the analysis of the immediate market area, due diligence and underwriting, and negotiations. In short, we recommend investors focus on both big picture diversification issues and effective deal execution. The acquisition and analysis of MSA level forecasts should be a casual hobby, not an activity upon which valuable resources are expended

37 + * MSA Geographic Allocations, Property Selection, & Performance Attribution in Public & Private Real Estate Markets David Ling, Andy Naranjo, and Ben Scheick NCREIF s Summer 2015 Conference

38 Additional Adjustments that May Improve Return Comparisons? Omit REITs from core that are too focused (?) on development Development is a noncore activity; may distort comparison with NCREIF managers May reduce relative performance of REITs Omit REITs that are too exposed (?) to Class B (or lower) properties May reduce relative performance of REITs Adjust for differences in liquidity Would enhance relative performance of REITs

39 Attribution Analysis R REIT f,t R NPI f,t = allocation + selection + interaction Allocation: portion of return differential due to differences in MSA allocations Mathematically: Difference between reweighted NPI & unadjusted NPI If REIT mangers generally pursue a top-down investment strategy (MSA selection then property selection), we can attribute remaining return difference to selection plus interaction effects Note: we cannot isolate pure selection effect since we do not have REIT returns at the MSA level