IRRs from the NCREIF Database · IRRs from the NCREIF Database Jeffrey D. Fisher, Ph.D. NCREIF Consultant Professor Emeritus, Indiana University DRAFT 12/6/13. Calculating IRRs with

IRRs from the NCREIF DatabaseJeffrey D. Fisher, Ph.D.

NCREIF ConsultantProfessor Emeritus, Indiana University

DRAFT 12/6/13

Calculating IRRs with the NCREIF Database

• Quick Review of IRR vs. TWR• NPI is based on Time Weighted Returns (TWR)• Each quarter a return is calculated as if the property was purchased at the

beginning of the quarter and sold at the end of the quarter.• This is like an IRR for the quarter assuming NOI is monthly and CapEx mid

quarter (hence the infamous denominator to approximate a quarterly IRR)• Relies on having a value each quarter• Quarterly returns chain linked giving equal weight to each quarterly return

(regardless of dollar amount invested).• An IRR uses a beginning investment amount, cash flows each quarter, and a

value at the end of a holding period (sale price or market value)• The amount of cash flow each quarter impacts the IRR (dollar weighted).

-10%

-8%

-6%

-4%

-2%

0%

2%

4%

6%

8%

1978

1979

1980

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

2006

2007

2008

2009

2010

2011

2012

2013

Total Quarterly Return

Average Return

NPI – Time Weighted Quarterly Returns

Quarterly AnnualMean 2.20% 9.10%Std Dev 2.24% 4.48%

Time series std dev

0

500

1000

1500

2000

2500

3000

3500

4000

4500

-69.

9%-6

0.7%

-51.

5%-4

2.3%

-33.

0%-2

3.8%

-14.

6%-5

.3%

3.9%

13.1

%22

.3%

31.6

%40

.8%

50.0

%59

.2%

68.5

%77

.7%

86.9

%96

.2%

105.

4%11

4.6%

Return Distribution for 2013-3

Cross Sectional Std Dev.

Quarterly Annual

Mean 2.82% 11.75%

Std Dev 7.18% 14.37%

NPI 2.59%

We can calculate IRRs using the NCREIFdatabase• We know the beginning value (initial market value when a property

enters the index)• We know the cash flows each quarter (NOI, Capex, Partial Sales)• We have the sale price or an ending market value if not sold.• But still several ways of calculating

1. Pooled Cash Flows• Add all the actual cash flows for each property in the portfolio• May have different starting and ending periods for each property.• Calculate an IRR on the total cash flows.• Often used for venture capital funds• This would reflect the IRR for the entire portfolio and reflect the

dollar amount invested each quarter

Alternative ways of calculating an IRR

Alternative ways of calculating an IRR2. IRR for a ‘typical’ property in the NPI – Geltner approach (IRR-Based Property-LevelPerformance Attribution Journal of Portfolio Management, Summer 2003)

• Select start period with an investment of $100• Create a capital index using market value changes from a starting period using

those properties in the NPI each quarter. (Composition changes over time.)Price change = (Ending MV – Beginning MV) / Beginning MV

• Apply the cash flow for properties in the NPI each quarter to the capital indexvalue that quarter to get a cash flow each quarter.

Cash Flow Return = (NOI – Capex + Partial Sales) / Beginning Market ValueApply the cash flow for properties in the NPI each quarter to the capitalindex

• Use the capital index to determine value at end of a holding period• Calculate the IRR on these cash flows

Comments on Geltner Approach:

• Not an IRR for the NPI as if it was a portfolio considering howmuch is invested each quarter.

• Weighted more by higher value properties in the NPI for aparticular quarter and influenced more by properties in the indexlonger.

• If the index is value weighted the results are identical to thepooled cash flow approach ONLY IF there are no acquisitions ordispositions during the selected time period.

• Acquisitions and dispositions during the selected analysis periodimpact the IRR only through their impact on the average cash flowreturn and average price appreciation for the index

IRR Tool Replicating Geltner Methodology

3. Calculate IRR on each individual property

• Use actual cash flows from each individual property.• Initial value can be when it first enters database or a value at a later

point in time.• Ending value can be sale price or market value at the end of a holding

period if not sold yet.• Calculate the IRR on the above cash flows.• Can then calculate the average, standard deviation and other

distributional characteristics of these IRRs. Good for risk analysis.• The rest of this presentation deals with this approach

0

50

100

150

200

250

300

Frequency of Quarterly IRR

Quarterly Annual

Mean 2.32% 9.62%

Std Dev 3.41% 6.82%

Sold Properties from NPI

Property Type Return Standard Deviation

Apartment 9.70% 5.50%

Industrial 7.49% 7.24%

Office 6.66% 10.07%

Retail 8.06% 7.20%

0.00%

2.00%

4.00%

6.00%

8.00%

10.00%

12.00%

Apartment Industrial Office Retail

Return

Standard Deviation

Requested exercise from the RiskManagement Officer of a NCREIF Member• Assume all properties in the NPI as of the first quarter of 1990 are

purchased on that date• Properties that enter the NPI after that come in at initial market value• Sold properties leave when sold• Any properties unsold as of the 2nd quarter of 2013 are sold at their

market value• Exclude properties held less than a year (flips)• Calculate the IRR on these properties• Provide distributions for the IRR

0

100

200

300

400

500

600

700

-18%

-17%

-16%

-15%

-13%

-12%

-11%

-10% -8

%-7

%-6

%-5

%-4

%-2

%-1

% 0% 1% 3% 4% 5% 6% 8% 9% 10%

11%

12%

14%

15%

16%

IRR Frequency(Quarterly IRRs)

Apartments

Sample size 3,328

Quarterly AnnualMean 2.20% 9.09%Std Dev 2.09% 4.18%1st percentile -3.09% -11.78%3rd percentile -1.39% -5.44%5th percentile -0.71% -2.81%25th percentile 1.25% 5.08%50th percentile 2.08% 8.58%75th percentile 3.07% 12.85%95tth percentile 5.31% 23.02%

Note: Standard Deviation is a mixture of differentproperty holding periods and different uniqueproperties.

0100200300400500600700800900

1000

IRR Frequency(Quarterly IRRs) Sample size 6,069


Industrial

Office

0100200300400500600700800900



Retail

0

100

200

300

400

500

600

700



Higher Risk, Lower Return?!

Variation – extend sample of holding periods

• Still assume purchase in 1990• Assume sale after one year, after two years, after three years, etc.

until actually sold or sell at market value 2nd quarter of 2013• This generates 77,832 IRRs• Better dispersion of holding periods in the sample

0

2000

4000

6000

8000

10000

12000

14000

-58%

-53%

-48%

-43%

-38%

-32%

-27%

-22%

-17%

-12% -6

%-1

% 4% 9% 14%

19%

25%

30%

35%

40%

45%

50%

56%

61%

66%

71%

76%

82%

87%

92%

97%


0

2000

4000

6000

8000

10000

12000

14000

-11% -9% -8% -7% -6% -5% -4% -2% -1% 0% 1% 2% 3% 4% 6% 7% 8% 9% 10%


Same IRRs – justomitted the extremes

Probably reporting anomalies

Probablyreportinganomalies

Quarterly AnnualMean 1.67% 6.86%Std Dev 3.21% 6.43%5th percentile -2.94% -11.23%10th percentile -1.30% -5.09%25th percentile 0.60% 2.42%50th percentile 1.93% 7.95%75th percentile 2.94% 12.31%95tth percentile 5.27% 22.82%

Possible Next Steps

• For each property, allow purchase and sale in any combination ofstart and end dates

• Calculate IRRs for cohorts, e.g., start in 1995 and end in 2000• Again break down by property type and also region etc.• Use TBI to adjust when appraised value used for terminal value rather

than actual sale price?

IRRs from the NCREIF Database · IRRs from the NCREIF Database Jeffrey D. Fisher, Ph.D. NCREIF Consultant Professor Emeritus, Indiana University DRAFT 12/6/13. Calculating IRRs with

Documents