Download file

Managerial Discretion in the Real Estate Sector:
An Empirical Evidence
Inês Pinto
ERES 2010 – Doctoral Colloqium
SDA Bocconi – School of Management, Milan, Italy
23rd-26th June 2010
Supervisor: Professor João Duque
RESEARCH OBJECTIVE
The aim of this study is to analyse if managerial discretion
is associated with management incentives to avoid
negative fair value changes.
On the other hand, we also attempt to determine which
fund-level factors can influence such behaviour.
2
RESEARCH OBJECTIVE
In order to determine the value of the property for subscriptions or
redemptions, fund managers may choose any value between the
acquisition cost and the average of the appraisal values attributed by
two independent appraisers with a minimum periodicity of two years.
We predict that fund managers have the opportunity to control the
financial information throughout the timing of the recognition of the
unrealized gains that arise from fund assets.
3
MOTIVATION/CONTRIBUTIONS
• Portuguese real estate investment funds represent an opportunity to
investigate the executive behaviour regarding accounting choice in a
sector particularly sensitive to managerial discretion.
• The results enhance our understanding of the concept of conservatism
in a different context. Managerial discretion may incentive fund
managers to have a less conservative attitude regarding accounting
choices.
• This research examines the importance, quality and attributes of
financial information in the real estate sector.
• This research can also supply relevant evidence to the question of
transparency of non-listed real estate vehicles compare to the public
real estate sector.
4
PRIOR RESEARCH
Managerial discretion is defined as the latitude of options that top
managers have in making strategic choices with impact on
organizational outcomes (Hambrick and Finkelstein, 1987).
Managerial discretion is influence by three sets of factors: task
environment, the organization's internal forces and the managerial
characteristics of the executive himself.
Despite some limitations, our work seek to represent an empirical
application of managerial discretion in a specific context.
5
PRIOR RESEARCH
Conservatism principle is captured by the accountants tendency to
recognize “bad news” more quickly than “good news”, being the
unrealized losses typically recognized earlier than unrealized gains.
(Basu, 1997).
Earnings management literature suggests that managers strategically
exercise managerial discretion in order to avoid earnings decreases,
earnings losses or negative earnings surprises (Burgstahler and Dichev,
1997; Degeorge et al., 1999; Matsumoto, 2002 and Burgstahler and
Eames, 2006).
6
HYPOTHESIS DEVELOPMENT
Hypothesis Development
Managerial discretion in real estate investment funds (REIFs) is
associated with management's incentives to avoid the
recognition of negative fair value changes of properties.
7
RESEARCH DESIGN
In order to test our hypothesis, we develop a research design similar to
Burgstahler & Dichev (1997) and Degeorge et al. (1999) based on the
analysis and comparison of the distributions of asset value changes
fixed by fund managers and those recommended by appraisers.
Therefore, we develop a variable called RDIF – Return Rates Difference
in order to estimate the discretionary behaviour of fund managers:
RDIFit  RIMPit  RAit
RIMPit is the annual rate of return implicit in two sequential appraisals at the
beginning of the year for property i.
RAit represents the annual property market value rate of return fixed by fund
managers for year t for the same property i.
8
RESEARCH DESIGN
If the value of RDIF is negative, then we can conclude that
the market value increment fixed by fund managers is higher
than the return estimated by appraisers.
Negative values of RDIF indicate situations where fund
managers are less cautious or prudent in unrealized gains
recognition than appraisers´ recommendation. Therefore, if our
hypothesis hold, we expect to observe a high frequency of
negative RDIF.
9
RESEARCH DESIGN
To verify if fund managers effectively use their discretion to avoid
the recognition of properties depreciation recommended by
appraisers, we conduct a regression of RA on RIMP in line with
Basu (1997) model for accounting conservatism.
Properties depreciations recommended by appraisers (RIMP<0)
are classified as “bad news”, while properties appreciations are
recognize as “good news”.
RAit    1RIMPit   2 DRIMPit  3 RIMPit * DRIMPit   it
RAit represents the asset value change fixed by fund manager;
RIMPit is the asset value change recommended by appraisers;
DRIMPit is a dummy variable with the value of 1 when there is a bad news (RIMP<0).
10
RESEARCH DESIGN
According to our hypothesis, the slope coefficient
from regression of RA on RIMP is expected to be lower in
the case of bad news sample because managers will be
less sensitive to this type of appraisers recommendation.
11
RESEARCH DESIGN
Finally, in order to investigate which fund characteristics can
influence the managers discretion, we seek to conduct a logit
regression modeling the probability of our variable RDIF being
negative for a specific property:
Pr ob( NEG  1)   0  1 X  
NEG is a dummy variable with a value of 1 for observations
with a negative RDIF and 0 for other observations and X is a
set of factors that could potentially influence the probability of
RDIF being negative (rented vs not rented, type of fund for
example).
12
PRELIMINARY RESULTS
• Sample, Data and Variables
– Real estate investment fund data was obtained from Portuguese
Securities Market Commission (CMVM – Comissão do Mercado de
Valores Mobiliários);
– The final sample includes 8.660 property-year observations between
2003 and 2008;
– 14 open-end REIFs and 137 closed-end REIFs
13
PRELIMINARY RESULTS
Descriptive Statistics:
• Both the annual return fixed by fund managers (RA) and the annual
returns implicit in appraisal estimates (RIMP) present a relatively stable
distribution during the sample period, being the sample mean of RA
(0.73%) slightly lower than the sample mean of RIMP (0.70%).
Appraisal estimates register a higher variability with an inter-quartile
range of 2.67% vs 0.47% for property annual returns, suggesting that
appraisers seem to adjust their estimates more quickly than fund
managers.
• In 2008, we can observe a substantial decrease in both returns, being
the sample mean return implicit in appraisal recommendations of
0.12%, evidencing the ongoing period of turbulence of financial
markets.
14
PRELIMINARY RESULTS
(ii) Incentives of managerial discretion
If fund managers use their discretion to strategically manage asset
value changes, we expect that manager’s asset revaluations exceed
appraisal estimates.
400
200
0
Frequency
600
800
Histogram for Return Rate Difference – RDIF
-.15
-.1
-.05
0
RENDIF
.05
.1
.15
15
PRELIMINARY RESULTS
We can observe more frequent negative scores of RDIF (positively
skewed). Statistical test similar to Burgstahler & Dichev (1997)
evidences that negative scores of RDIF occur more frequently than
would be expected given a smooth distribution (the standardized
difference for the two intervals immediately to the left of zero is 4.0
and 2.6).
Therefore, we have some evidence that the asset value changes
fixed by fund managers seem to be higher than appraisers´
recommendations in the intervals near zero.
16
PRELIMINARY RESULTS
To examine the distribution of RDIF conditional on annual return
implicit in appraisal estimates (RIMP), we sort observations on the
variable RIMP to form equal-sized portfolios of 475 observations per
portfolio. Some of the results can be observed in the following table:
Descriptive Statistics of RDIF´s portfolios formed by the sign of variable RIMP
Annual RIMP
<-2.36%
-2.36% to -1.01%
-1.01% to -0.47%
-0.47% to 0%
0%
0% to 0.39%
0.39% to 0.88%
0.88% to 1.54%
1.54% to 2.65%
2.65% to 4.28%
>4.28%
Portfolios
-4
-3
-2
-1
0
1
2
3
4
5
6
Mean RDIF
-1.74%
-1.17%
-0.77%
-0.35%
0.00%
0.20%
0.45%
0.90%
1.38%
1.90%
2.80%
Mean RA
-1.07%
-0.06%
0.03%
0.61%
0.04%
0.84%
0.71%
0.87%
1.22%
1.70%
2.44%
Perc.Neg.RDIF
61.52%
78.10%
82.48%
80.76%
13.90%
32.00%
39.81%
36.00%
31.24%
34.10%
29.90%
We can observe that for the first portfolio to the left of zero, despite
the negative valuations of appraisers (mean of RIMP is -0.15%), the
mean of RA is of 0.61%.
17
PRELIMINARY RESULTS
We also computed the percentage of negative values of our variable
RDIF for each of the years of our sample, together with the annual
capital growth of the Portuguese IPD index. We report the results in
the following table:
Temporal trend of the percentage of negative RDIF
Year
Perc.
Neg.RDIF
IPD
C.Growth
2004
2005
2006
2007
2008
44.88%
38.25%
34.06%
26.84%
42.68%
3.40%
3.10%
5.50%
5.80%
-3.30%
Despite the low number of observations available, we observe a
negative correlation between the percentage of negative RDIF and the
Portuguese IPD index.
18
PRELIMINARY RESULTS
The results from the regression about the asymmetry of timeliness in
asset value changes recognition are presented in the following table:
Pooled Regression of RA on RIMP
According to our expectation β3 is negative. This coefficient measures
the impact of bad news relative to good news in the property capital
return fixed by fund managers (RA).
19
PRELIMINARY RESULTS
(ii) Incentives of managerial discretion
We predict that managerial discretion become stronger for properties
that are not rented. We conduct similar tests dividing our property
sample into two subgroups: property rented and not rented.
300
200
0
100
Frequency
400
500
Histogram for Return Rate Difference – RDIF for Rented and Not Rented Properties
-.1
-.05
0
RENDIF - Properties Rented
.05
.1
.15
-.15
-.1
-.05
0
RENDIF - Properties Not Rented
.05
.1
.15
100
50
0
Frequency
150
200
-.15
20
PRELIMINARY RESULTS
Nevertheless, results do not confirm our prediction. We also conducted
splitting the sample into open-end and closed-end funds. However, we
were not able to see any significant difference in fund managers´
behaviour.
The weak evidence in finding factors that can motivate fund
managers´ behavior can be related to other determinants not included
in this first analysis.
Therefore, we seek to perform the logit regressions in order to
modeling the probability of RDIF being negative.
Pr ob( NEG  1)   0  1 X  
21
CONCLUSIONS
Our results suggest that fund managers are less cautious
or prudent in unrealized gains recognition than appraisers´
recommendation. There is evidence of an asymmetry in the
timeliness of asset value changes recognition. When
appraisers recommend depreciating a property (“bad
news”), fund managers seem to postpone the recognition
of this information.
22
FURTHER RESEARCH
• Investigate if fund managers use this discretion opportunistically
in order to increase fund performance. Based on a fund-held
sample, we test different hypothesis to examine if fund
characteristics as fund type, dimension or fund vacancy rates
can stimulate the management of net asset value. After
modeling how annual revaluations increments of funds
properties would be reported in the absence of earnings
management, we estimate a discretionary accrual proxy as the
difference between the reported revaluation increment and its
expected value obtained by the model.
• Analyze the possible economic implications of fund managers
behavior. Compare the use of this hybrid accounting system with
the adoption of a fair value method.
23