Predicting turning points in the rent cycle[1]

Predicting Turning Points in the Rent
Cycle Using the Natural Vacancy Rate –
An Applied Study of the Dublin Office Market*
John McCartney, CSO.
Paper read before the Statistical and Social Inquiry Society
of Ireland, 2nd
November 2010
(JEL Classification R33)
*
The author would like to thank Seán Lyons of the ESRI and an anonymous referee for helpful comments
on an earlier draft of this paper. Thanks also Clare Eriksson of Jones Lang LaSalle and Thomas Conefrey
of the ESRI for providing data assistance. The views expressed herein are those of the author and this
article in no way purports to represent the opinions, views or policies of any other individual or institution.
Any errors are the author’s. E-mail: j.mccartney@yahoo.ie

1
Abstract
All property markets have a natural vacancy rate (NVR). Theory suggests that when the
actual vacancy rate exceeds this NVR the market will be unbalanced. Then, rents will fall to
restore equilibrium. Conversely, when vacancies are below the NVR, rents will rise. By
establishing a ‘tipping point’ between positive and negative rental growth, the NVR is a
useful tool for analysing market cycles.
Standard econometric practice is to model the NVR as a constant. However, this paper finds
that there was a large structural shift in the natural vacancy rate for Dublin office space in
the late 1990s. Allowing for this, two discrete NVRs are estimated for Dublin. For the 1978-
1998 period, the NVR is calculated at 5.2%. This is broadly in line with rule-of-thumb
estimates used by industry practitioners. However, the estimated NVR for the 1999-2009
period is 15.0 % - more than twice the 7% rate that is normally assumed.
One interpretation is that Dublin’s office market is set to recover sooner than expected.
Currently, the actual vacancy rate is just over 23%. Rents will only stop falling when this
figure has been reduced into line with the NVR. For a given rate of net absorption, this will
be achieved more quickly if the NVR is 15% than if it were 7%.
Section 1 - Introduction
The vacancy rate promises to be a key variable for understanding commercial
property markets because it encapsulates both supply and demand conditions.
However, while it is commonly expected that a high vacancy rate will exert
downward pressure on rents, applying this expectation to practical market analysis is
less straightforward than one might imagine. In particular, a problem arises in
defining exactly what is meant by a ‘high’ vacancy rate. Previous studies have
demonstrated that the same vacancy rate (say 8% or 10%) can be consistent with
rental growth in one market and falling rents in another (Krainer 2001, Voith and
Crone, 1988). This suggests that, in itself, the absolute vacancy rate is not the best
indicator of market conditions. Instead the gap between the actual vacancy rate and
some market-specific ‘natural’ rate of vacancies is a better predictor of rental
dynamics (see Sanderson, Farrelly and Thoday, 2006). Conceptually, the natural
vacancy rate (NVR) is that which corresponds with market equilibrium and rent
stability. When the actual vacancy rate exceeds this level the market is ‘overbuilt’
and rents will fall. Conversely, when vacancies are below their natural level, theory
suggests that rental growth will occur.
Because the NVR establishes a ‘tipping point’ between positive and negative rental
growth it is potentially a useful tool for analysing commercial property cycles.
Despite this, however, the NVR does not feature prominently in European
commercial property research, and it is seldom referred to in Ireland. Moreover, on
those occasions when the NVR is utilised, its value is often approximated on a ‘rule of
thumb’ basis (Sanderson, Farrelly and Thoday, 2006). These approximations can be
highly arbitrary, resulting in potentially misleading analyses.
This article formally estimates the natural vacancy rate for Dublin’s office market
using an econometric approach. In doing so it follows a methodology that is well-
established in the US literature. However, in a refinement of the traditional approach
which treats the NVR as a constant, this paper allows for changes in the natural
vacancy rate over time. The results of this exercise suggest that, until 1998, the NVR

2
for Dublin offices was around 5.2%. Thereafter, however, it increased dramatically
and is now estimated at 15.0%. This is more than twice the level generally assumed
by analysts as ‘natural’ for the Dublin office market. One important implication of
this finding is that the Dublin office market may recover sooner than expected.
Relative to the true NVR the market is less ‘overbuilt’ than is commonly perceived.
Therefore, for a given rate of net absorption, rents are likely to stabilise earlier than
was previously anticipated.
The remainder of this paper is structured as follows: Section 2 outlines the theoretical
underpinnings of the NVR concept and reviews those factors that may influence the
natural level of vacancies in any office market. The following section describes the
standard econometric methodology for estimating an NVR, while Section 4 sets out
the empirical framework used to estimate the NVR for Dublin offices. The results of
the estimation are presented in Section 5 and, in Section 6, the implications of these
findings are discussed before a brief conclusion.
Section 2 – The Theory of Natural Vacancy Rates
The Natural Vacancy Rate Hypothesis has its unlikely origins in labour market
economics (Grenadier 1995, Voith and Crone 1988). A standard labour market theory
is that overall unemployment can be decomposed into its voluntary and involuntary
components. The voluntary component includes people who are out of work at any
time because they are in the process of changing jobs (frictional unemployment). It
also includes people who could get work at the current market wage but who choose
to keep searching for jobs that pay above the norm. At market equilibrium, everyone
who wants a job at the current market wage can have one – i.e. there is no forced
unemployment. However, some voluntary unemployment is inevitable because there
are always people moving between jobs and searching for better offers. Therefore we
should expect some level of unemployment even when the market is in equilibrium.
This is the natural rate of unemployment.1
Periodically, actual unemployment will
deviate from this natural level. When this occurs – for example due to business cycles
which affect the demand for labour – theory suggests that real wages will adjust to
restore equilibrium (Hendershott, MacGregor and Tse, 2002).
Application of this theory to the market for rented property was first proposed by
Blank and Winnick in 1953, and since then it has become an established tool for real
estate analysis, particularly in America (see Hendershott, MacGregor and Tse 2002,
Rosen and Smith 1983). In essence, it is argued that, even when property markets are
in equilibrium, we should expect some vacant space. There are two reasons for this.
Firstly, just as frictional unemployment is a characteristic of labour markets, frictional
vacancies are an inevitable feature of property markets. At any given time some
buildings will be temporarily vacant, either because they are between lettings or
because they are newly built and awaiting their first occupancy (see Gabriel and
Nothaft, 2001). Secondly, just as some workers choose not to accept jobs at the going
rate because they hope to get better offers, some landlords will choose not to let their
properties at current market rents because they hope to find tenants who are prepared
to pay more. Landlords could pursue this objective by conducting extended searches
for tenants whose specific requirements most closely match the characteristics of their
1
Similar explanations of the natural rate of unemployment can be found in most basic economics textbooks – see,
for example Begg, Fischer and Dornbusch (2002).

3
particular buildings. Alternatively, they could simply bide their time and wait for
overall market rents to rise above current levels.
The proportion of total space that remains vacant due to these factors – frictional
vacancies and landlords waiting for better offers - is known as the natural vacancy
rate (NVR). Because this rate of vacancies is compatible with market equilibrium
(i.e. every landlord who wants to let space at the current market rate can do so), the
NVR is the vacancy rate which is consistent with rent stability (Jud and Frew, 1990).
As in labour markets, the actual vacancy rate in commercial property markets often
deviates from this natural rate. These digressions are partly driven by exogenous
business cycles. However, they are also driven by endogenous cycles in construction
development - Long lead-times in commercial building contribute to spells of over-
development followed by periods of sharp construction retrenchment.2
At times when
the market is over-supplied – either because of excessive construction or a slump in
occupier demand - vacancies rise above their natural rate and rents gravitate
downwards to restore equilibrium. In contrast, when the market is tight the actual
vacancy rate falls below its natural level and then rents increase (Hendershott, 1995).
At all times, however, the natural vacancy rate acts as the fulcrum around which
rental dynamics are determined.
Given the theory outlined above, it is reasonable to argue that the NVR in any office
market will be influenced by two sets of variables; those that that affect frictional
vacancies and those that influence landlords’ propensity to hold-out for higher rents.
Looking at the first category, the length of the standard office lease is likely to impact
upon frictional vacancies. In markets where lease terms are long and break-options
are infrequent and/or costly to invoke, tenant turnover and frictional vacancies should
be low. Holding everything else equal, this is consistent with a lower NVR
(Grenadier 1995, Rosen and Smith 1983, Voith and Crone 1988, Wheaton and Torto,
1988).
Conversely, however, frictional vacancies should be positively related to the extent of
speculative building. This is because, by its nature, speculatively built
accommodation can lie vacant for a period prior to its first letting. The extent of
speculative development partly depends on factors related to the local building
industry. It tends to be higher in markets where construction activity is strong (de
Leeuw and Ekanem 1971, Rosen and Smith 1983). It is also expected to be higher
where there is vigorous economic growth as this gives developers confidence that
they will be able to find tenants upon completion. Finally, where there are large
economies of scale in capacity construction, developers may find it profitable to build
more office space than they can pre-let (see McDonald, 2000). In addition to these
factors, the extent of speculative building may also depend on market structure.
Where occupiers have generic accommodation needs and their average space
requirements are low, developers may have great difficulty getting tenants to commit
to pre-lets. Here, they may have no choice but to build speculatively (see McCartney,
2008). Furthermore, where market supply is elastic (i.e. where developers can
2
See Barras and Ferguson (1987), Phyrr, Roulac and Born (1999). The Dublin office market closely follows this
classic cyclical pattern. MacLaran and O’Connell (2007, 2003), MacLaran (1993) and MacLaran, MacLaran and
Malone (1987) provide detailed descriptions of previous development cycles in the Dublin office market, while
McCartney (2008) presents an econometric analysis of Dublin development cycles over the 1976-2007 period.

4
quickly deliver new offices in response to price signals), construction is more likely to
occur ahead of demand (Rosen and Smith 1983, Krainer 2001). Factors affecting the
elasticity of supply include access to development credit, the nature of the local
planning system and the availability of open building land; where these variables
facilitate speculative construction, frictional vacancies and the NVR should be higher
(Krainer, 2001).
As outlined above, the NVR not only includes frictional vacancies, but also vacancies
arising from landlords’ decisions to hold-out for higher rents. One strategy that
landlords can pursue in order to achieve higher rents is to seek out tenants who are
willing to pay more because their accommodation needs closely match the
characteristics of the particular buildings on offer. Making these ideal matches often
requires extended search activities and so, at an aggregate market level, this strategy
is consistent with a high NVR. In practice, the extent to which landlords are prepared
to undertake prolonged searches depends on the expected returns from finding the
right tenant. These, in turn, are influenced by the heterogeneity of office occupiers.
Where there is a wide variation in the rent that different occupiers are prepared to pay
for the same property, the expected returns to search activity are high. Under these
conditions the NVR should be high because more space will be held vacant to
facilitate search (Grenadier 1995, McDonald 2000). In addition to diversity among
tenants, however, qualitative variation in the office stock may also affect the natural
vacancy rate. Where buildings are highly standardised and locational segmentation is
low, landlords have little to gain by engaging in extensive search; Irrespective of
occupier tastes, there is no reason for a tenant to pay more for one office than another
under these conditions. As a result, landlords will be less likely to hold vacant space.
Conversely, where the market contains an array of idiosyncratic properties the returns
to search activity will be higher, and so should the NVR (Jud and Frew, 1990).
Searching out tenants that provide the best match for their buildings is one strategy
that landlords can adopt to achieve higher rents. Another approach is simply to delay
letting their properties in the hope that overall market rents will rise. Insofar as this
involves landlords voluntarily holding vacant space it contributes directly to the
natural vacancy rate. Obviously, rental expectations will have a critical bearing on
the extent to which this ‘hoarding’ strategy is adopted - where landlords are optimistic
about rents they have a stronger incentive to withhold space. Therefore a higher NVR
is expected where economic and rental growth have traditionally been stronger
(Rosen, 1984).
In addition to rent expectations, the terms of the standard lease contract may also
influence hoarding. Firstly, the mechanism for rent revisions will be important.
Where revisions are determined by market review real rents can increase substantially
over the term of a lease. Here, the opportunity cost of signing contracts will be low
and landlords will be disinclined to hold empty space. However, where escalation
occurs through indexation the scope for real rental growth within leases will be quite
restricted. Here, landlords who foresee a rise in market rents will assume higher
opportunity costs from getting ‘locked-in’, and they will be more likely to carry
vacant space.
Secondly, the scheduling of rent revisions will be important. Where indexation is
used, rents are generally escalated annually. However, where market reviews are the

5
norm, the frequency of revisions differs. In France, for example, it is common for
reviews to occur every three years, whereas in Ireland market reviews normally take
place at five-yearly intervals.3
Clearly, in a rising market, landlords will envisage
higher opportunity costs where reviews take place less often. Consequently, they are
more likely to hoard empty space and the NVR should be higher (see Ball, Lizieri and
MacGregor, 1998).
A third contract-related factor that may influence the NVR is the issue of upward-only
rent reviews. Heretofore upward-only clauses have been standard in Dublin office
leases. By providing a sustained advantage to landlords who time their lettings to
coincide with the top of the market, these arrangements encourage hoarding and are
consistent with a higher NVR. During Ireland’s current economic crisis upward-only
reviews have become highly contentious and were prohibited on new leases from 28th
February 2010.4
As yet, it is too early to observe the impact of this intervention. In
theory, however, it is likely to exert a downward influence on the NVR for Dublin
offices.
Finally, it should be clear from the preceding discussion that real interest rates may
also play a key role in determining the natural vacancy rate. Not only will high
interest rates deter speculative building, they will also increase the cost to landlords of
carrying empty office stock (Ball, Lizieri and MacGregor, 1998). All else equal,
therefore, we should expect a lower NVR in markets with historically high interest
rates.
Few studies have attempted to empirically estimate the influence of these variables on
the NVR. However, those that are available tend to confirm the relationships
hypothesised above.5
Rosen and Smith (1983) found that tenant mobility was
positively related to the NVR in a range of American markets. Their analysis also
confirmed that NVRs were higher in areas of rapid construction. These findings
support the notion that friction leads to a higher natural vacancy rate. Similarly
Sanderson, Farrelly and Thoday (2006) found that average real GDP growth
positively influenced the NVR in European, Asian and American office locations.
This is consistent with the arguments that positive rent expectations encourage
speculative development and increase landlords’ propensity to withhold vacant space.
Finally, there is also support for the theory that market heterogeneity is associated
with higher NVRs – Rosen and Smith (1983) found that higher rent dispersion had a
positive relationship with the equilibrium vacancy rate.
Table 1 summarises the above discussion. It lists the variables that are expected to
influence frictional vacancies and landlords’ decisions to maintain un-let space
respectively. By extension, these factors are also expected to influence the NVR.
3
Berwin Leighton Paisner (2009) provide a detailed comparison of typical commercial property lease
arrangements across Europe.
4
S.I. No. 471 of 2009 – Land and Conveyancing Law Reform Act (2009) (Commencement) (Section 132) Order
2009.
5
The general methodological approach is to estimate the NVR for each individual location using a rent adjustment
equation populated with time-series data. The estimated NVRs then become the dependent variable in a cross
sectional analysis. This approach clearly imposes very heavy data requirements, hence the rarity of these studies.

6
Table 1: Factors Expected to Influence the NVR
Factors Influencing Friction;
Variable Expected Influence Reason
• Lease length Shorter leases = higher NVR Higher tenant turnover
• Rate of new construction Higher completions = higher NVR Higher proportion of speculative development
• GNP growth Higher growth = higher NVR Higher rent expectations, higher proportion of
speculative development
• Average lot size Smaller tenant requirements =
higher NVR
Smaller lots difficult to pre-let ∴ low average
lot size = more speculative development
• Elasticity of supply More elastic supply = higher NVR Higher proportion of speculative development
• Real interest rates Lower rates = higher NVR Low rates encourage speculative development
Factors Influencing Landlords’ Propensity to Hold Vacant Space;
Variable Expected Influence Reason
• Heterogeneity of
occupier & office stock
More heterogeneous = higher NVR Higher expected returns to search activity
• GNP growth Higher growth = higher NVR Positive rent expectations ∴ ‘lock-in’ imposes
higher expected opportunity cost
• Revisions mechanism Indexation = higher NVR ‘Lock-in’ imposes higher expected
opportunity cost where indexation used
• Revisions frequency Lower frequency = higher NVR ‘Lock-in’ imposes higher expected
opportunity cost where revisions are
infrequent
• Real interest rates Lower rates = higher NVR Low rates reduce cost of holding vacant space
Section 3 – Methodological Approach
The theory set out above suggests that, in any office market, rents will fall when the
actual vacancy rate exceeds the natural vacancy rate for that specific market (which,
in turn, is determined by the variables listed in Table 1). Conversely, rents should rise
when the actual vacancy rate is below this NVR. Thus, rental growth can be
expressed as a function of the gap between the actual and natural vacancy rates;
)( NVRVRrR −=Δ (1)
…where ΔR is the change in rents, VR is the actual or observed vacancy rate and
NVR is the natural vacancy rate. In practice, many authors have modelled this
relationship using variations of the following basic equation; 6
ΔR = α+ β1VR (2)
From equation (2) it is straightforward to calculate the natural vacancy rate (see
McDonald 2000, Sanderson, Farrelly and Thoday 2006); At equilibrium, rental
6
Sanderson, Farrelly and Thoday (2006) estimate this basic equation for 29 office markets in Europe, Asia and the
USA. Other authors have used more complex variations with additional independent variables. Comprehensive
reviews of these models are provided by Ball, Lizieri and MacGregor (1998), Hendershott, MacGregor and Tse
(2002), McDonald (2000, 2002).

7
growth equals zero and the actual vacancy rate coincides with the natural vacancy
rate. Therefore, setting ΔR to zero and substituting NVR for VR, we solve to get;
NVR =
−α
β1
(3)
Upon estimation we expect a negative coefficient β1 and so equation (3) should result
in a positive NVR.
From equation (3) we can derive that the constant term in the previous equation
incorporates the NVR. Thus, assuming a negative sign on the vacancy rate
coefficient, the Y-axis intercept can be written as α = β1NVR. A correct
interpretation is that the known vacancy rate (VR) in equation (2) is measured with
reference to the natural vacancy rate, which is unknown a priori, but which is
embedded within the intercept (see Ball, Lizieri and MacGregor 1998, Shilling,
Corgel and Sirmans 1987, Rosen and Smith, 1983).
Implicit in this standard approach is an assumption that the NVR is constant over time
(see Ball, Lizieri and MacGregor 1998, Tse and Fischer 2003).7
Clearly, the validity
of this assertion depends upon there being no structural shift in the factors which
influence the equilibrium vacancy rate. While this may be plausible in some markets,
it is questionable in the case of Dublin. Over the period of this study Ireland’s
economy underwent a pronounced restructuring, there was a fundamental change in
the monetary policy framework, important developments took place in Dublin’s urban
planning context, and the term of the institutional office lease shortened significantly.
One implication is that a constant NVR cannot be taken for granted in Dublin. In
practice, this means that our rent equation should be assessed for parameter stability
and, if necessary, adjusted to allow for a time-varying NVR.
As set out above, equation (2) describes the basic stock-flow model of rent adjustment
that has traditionally been used to estimate natural vacancy rates. In recent years,
however, further refinements have been incorporated into this framework. For
example, current practice is to estimate rent adjustment equations using lags of the
actual vacancy rate because contemporaneous values may be endogenous.
Furthermore, as our understanding of the rent adjustment process has evolved,
additional explanatory variables have been added to improve fit and enhance the
dynamic properties of the model (see Hendershott 1995, Hendershott, MacGregor and
Tse 2002). Reflecting some of these nuances, this paper estimates the NVR for
Dublin’s office market using the following rent adjustment equation;
tttttt SIGNPVRRR εββββα +Δ−Δ−Δ+−=Δ −− 431211 (4)
In this specification, which is set out with the expected signs of the variables in place,
ΔRR is the annual change in real office rents and VR is the actual vacancy rate,
measured relative to the NVR in the constant term. As outlined above, VR is lagged
7
Only a small number of studies deviate from this assumption; Wheaton and Torto (1988) modelled the NVR as a
linear time trend, while Tse and Fischer (2003) estimated a time-varying parameter model to allow for time effects.
Sivitanides (1997) used drivers of rental expectations to model variability in the NVR, while Zhou (2008)
identified natural breaks in the data and estimated discrete NVRs for each time period.

8
one period because rents and vacancy rates may act simultaneously to clear the
market, potentially making a contemporaneous measure of vacancies endogenous. In
addition to the vacancy rate variable, equation (4) also includes real GNP growth
(GNP), the change in real interest rates (I) and the annual change in Dublin’s total
office stock (S). GNP growth is expected to have a positive effect on real rental
growth as it should ultimately lead to an increased demand for office space. In
contrast, a rise in real interest rates should reduce rental growth because this dampens
economic activity and office demand. Finally, growth in the office stock implies a
greater supply of office property which, ceteris paribus, is negatively associated with
rental growth. Within this framework, equilibrium occurs where real rental growth is
zero and where rents just cover the annual cost of capital (McDonald 2000). Here,
just enough new space is delivered to replace fully depreciated buildings and therefore
there is no net change in the office stock. At equilibrium, any movement in GNP or
real interest rates will provoke another round of rent adjustment and so they are set to
zero. Upon estimation of this equation, values for ˆα and ˆβ1 enable us to estimate a
natural vacancy rate for the Dublin office market as described above.
Section 4 – Empirical Framework
Commercial property research has traditionally been constrained by a scarcity of good
quality data (see Ball, Lizieri and MacGregor 1998, Hendershott et al. 1999). Data
series are usually shorter than would be expected in mainstream economic analysis,
while issues such as missing values and definitional discontinuities are common
(Dunse et al. 1998, Leishman 2003). Because of these constraints European office
markets are generally under-researched (D’Arcy et al. 1999, Brounen and Jennen
2009). If anything, these data challenges are more acute in Dublin than elsewhere
because Dublin’s office market is infantile compared with those in many American,
British and European cities. To illustrate this, only two purpose-built offices of more
than 186 sq m existed in the city by 1960. Moreover, ignoring owner-occupiers, a
sizable rental market did not emerge in Dublin until the 1970s (see Malone, 1981).
Notwithstanding these challenges, however, the dataset used in this analysis covers a
32 year period (1978-2009), representing three-and-a-half full cycles of the Dublin
office market. This constitutes a relatively long data series compared with many of
those used in the international literature. For example, Gardiner and Henneberry’s
rent equations for the UK regions were based on just eight years of data (1988, 1991).
Hendershott, Lizieri and Matysiak (1999) analysed London office rents using 20
annual observations, while Wheaton, Torto and Evans (1997) based their equation for
London on a 21-year sample. Tse and Fisher (2003) estimated rent adjustment
equations for London and Hong Kong using 22 and 23 years of data respectively,
while Rosen’s (1984) study of the San Francisco market and Hendershott’s (1995)
analysis of the Sydney office market also relied on 23 observations. Finally, Ng and
Higgins (2007) modelled office rents in Singapore using thirteen-and-a-half years of
semi-annual data (26 data points).
Rent equations for the Dublin office market have previously been developed by
several authors. As part of a wider comparative study, Sanderson, Farrelly and
Thoday (2006) estimated a simple rent adjustment equation for Dublin. However, this
was predicated on just 12 years of annual data. An earlier paper by D’Arcy,
McGough and Tsolacos (1999) used a 25 year dataset (1973-1997) to model real

9
rental growth. However, as their model did not include vacancy rates, it provides no
basis for calculating an NVR.8
Finally, McGreal et al. (2004) set out a model based
on 26 annual observations (1975-2000). However, because rents were estimated in
levels and their model did not include vacancy rates, it is also unsuitable for
estimating an NVR.
The dependent variable in this paper is derived from the office component of the
Jones Lang LaSalle Irish Property Index. Rents in this index are the Estimated Rental
Values (ERVs) of a portfolio of properties typical of those held by institutional
investors. This series, which is not adjusted for incentives such as rent-free periods or
capital contributions to fit-out costs, is deflated using the Consumer Price Index (CPI)
and entered into our model in differences. Vacancy rates come from Chartered
Surveyors Lisney and are measured as a percentage of Dublin’s total modern (post
1960) office stock. GNP growth is the annual change in real GNP, sourced from the
Central Statistics Office. As is standard in rent equations, this variable is lagged to
allow time for changes in economic activity to affect office demand and real rents. In
this case a one-year lag structure gave the best fit. Real interest rates are prime
lending rates supplied by the Central Bank and Financial Services Authority of
Ireland, deflated by the CPI and differenced. The annual change in Dublin’s office
Stock, measured in thousands of square metres, is derived from absolute figures
published by Lisney.
Section 5 – Results
Initially, equation (4) was estimated for the full period of analysis using Ordinary
Least Squares (OLS) regression. The results of this exercise are presented in Table 2
below.
Table 2: Results of Fixed-Effects Regression (1978-2009)
Coefficient T- Statistic
Intercept 4.875**
2.616
VRt-1 -1. 029***
-5.567
 GNPt-1 0.630***
4.249
 It -1.321***
-2.756
 St -0.011 -0.818
R2
= .792 2
= .760 DW = 1.569 F = 24.751***
**
P<0.05 ***
P<0.01
Reassuringly, all of the regressors in this equation take on the expected signs and,
with the exception of the Stock variable, are significant at 5% or higher. Reflecting
this, an R2
of .79 indicates that four-fifths of all variation in real rental growth is
explained by the model. However, there are some niggling concerns about this
regression. A general test of specification error, the Ramsey Reset Test, hints that the
model may be under-fitted.9
More worryingly, although the constant term is
significant at 5% (indicating a positive NVR), the natural vacancy rate implied by this
8
The authors reported that insufficient data was available at that time to include a vacancy rate variable.
9
F(1,25) = 6.305, P=0.019.

10
equation is just 4.74%.10
This seems implausibly low for several reasons. Firstly, the
actual vacancy rate in Dublin’s office market averaged 8.56% between 1978-2009. If
the NVR was lower than this then, on average, real rents should have gravitated
downwards. Instead, however, they have actually increased by almost three quarters
of a percent per annum over the 32-year period. This suggests that, if our theory is
correct, the true NVR should be higher than 8.56%.
A second reason for doubting the NVR implied by these OLS results is illustrated in
Table 3 below. This table shows the observed vacancy rate that prevailed in Dublin’s
office market on each occasion over the last 32 years when the market was in
equilibrium (i.e. when rental growth was zero). Clearly, on the three instances when
this occurred over the last decade, the actual vacancy rate was much higher than
4.75%. Given that the actual and natural vacancy rates should coincide when the
market is in equilibrium, this suggests that the true NVR for Dublin office space in
recent years is 12%-17% rather than 4.74%.
Table 3: Observed Vacancy Rate (%) at Previous Equilibrium Points
Year VR11
2008 14
2006 12
2001 17
1995 5
1991 6
1987 5
1981 5
Interestingly, however, Table 3 also contains an indication of what might have caused
our fixed-effects equation to under-estimate the true NVR. In contrast to the period
since 2001, the early years of our study (1978-1995) consistently saw low vacancies
when the market was in equilibrium. Based on this observation, it seems that
something fundamental may have changed between 1995-2001 which caused an
upward shift in the NVR. To explore this further, Chow Tests for parameter stability
were performed for all possible break-points between 1995 and 2001. The largest F
Statistic generated by this procedure occurred in 1999 and confirmed the hypothesis
that our unrestricted equations for the 1978-1998 and 1999-2009 sub-periods were
statistically different at 5%.12
While this establishes that the dynamic of rent adjustment in Dublin’s office market
altered around 1999, it remains unclear whether this change originates from a
difference in the intercept term or the slope coefficients. If the former is true, and/or
if the coefficient on the vacancy rate variable changed, then the NVR may have
shifted between periods. To investigate this in more detail, equation (4) was re-
estimated with the inclusion of five additional variables. The first was a dummy with
values of zero for the 1978-1998 period and one thereafter. The coefficient on this
10
Calculated as the negative of the intercept divided by the coefficient on the lagged vacancy rate – see Section 3
above.
11
As equilibrium can occur at any time during the year, vacancy rates in this table do not necessarily correspond
with their year-end values.
12
F(5, 21) = 3.58 vs. 5% critical value of 2.68.

11
variable measures any change in the intercept between periods. The other new
variables are interactive terms created by multiplying our dummy by each of the
existing slope variables. These “drifters” measure differences in the slope coefficients
between periods. The results of this regression are reported in Table 4 below.
Table 4: Results of Dummy Variable Regression
Intercept 9.024**
2.690
VRt-1 -1.637***
-4.128
 GNPt-1 0.931***
4.283
 It -0.921*
-2.064
 St -0.074*
-1.818
Dummy 17.162**
2.326
D.(VRt-1) -0.054 -0.108
D.(GNPt-1) -0.802**
-2.623
D.(It) -1.345 -1.193
D.(St) 0. 037 0.832
R2
= .888 2
= .840 DW = 1.894 F= 18.450***
*
P<0.10 **
P<0.05 ***
P<0.01
This analysis confirms that the NVR in Dublin’s office market did indeed shift
upwards after 1998. To see this, recall from Section 3 that the intercept in our rent
equation is simply α = β1NVR; i.e. the coefficient on lagged vacancies multiplied by
the NVR. Table 4 shows that the intercept term α changed significantly between
periods (the coefficient on the dummy variable is significantly different to zero).
However, the coefficient on the vacancy rate drifter was not significant, indicating a
constant β1. Logically, it follows that the observed change in our intercept must have
derived from a structural shift in the natural vacancy rate.
Interestingly, the GNP drifter was also significant in this analysis. A negative sign on
this variable indicates that Dublin office rents became less sensitive to changes in
economic growth after 1998. One interpretation is that the income elasticity of
demand for office property may have shifted downwards at this time.13
This could
have occurred for several reasons. On one hand, the occupied space-per-employee
ratio may have become compressed. Certainly, this hypothesis is plausible given the
relative growth rates of office-based employment and occupied space in the late
1990s. Between 1978-1996 office employment rose at a relatively constant rate,
averaging approximately 3% per annum. However, over the following five years
(1997-2001) it expanded at a much faster rate of 8.28% per annum.14
This compares
with an average increase in occupied space of 6.95% per annum in the same period.
An alternative possibility is that increased productivity weakened the link between
overall economic growth and the demand for office space. These issues merit a more
detailed analysis and are flagged for future research. For now, however, in order to
13
For a formal derivation of the income elasticity of demand for office space see Hendershott, Lizieri and
MacGregor (2010).
14
Over the period covered by this paper there is no continuous series for employment in the main office-based
sectors of the Irish economy (administrative public sector, financial services and ‘other business services’).
Figures reported are the author’s estimates based on an analysis of combined CSO sources.

12
derive a well-specified and parsimonious model, this GNP drifter was retained in our
final rent equation, along with the differential intercept and the original variables from
equation (4). However, the insignificant drifters in Table 4 were discarded to give the
final model presented below.15
Table 5: Results of Parsimonious Regression
Intercept 8.889***
4.375
Dummy 16.942***
4.064
VRt-1 -1. 724***
-7.449
 GNPt-1 0.821***
4.658
D.(GNPt-1) -0.501**
-2.345
 It -1.190***
-3.064
 St -0.049***
-3.247
R2
= .877 2
= .846 DW = 1.977 F = 28.479***
**
P<0.05 ***
P<0.01
As with the base model presented in Table 2, all of the variables in this regression
display the expected signs. However, the overall fit of the model is improved by
removing the restrictions that the intercept has to be constant between time periods
and that lagged GNP growth has a fixed-effect across the full estimation period. As a
result, the R2
increases by almost nine percentage points to 0.88 and all variables
become significant at 5% or better.16
Figure 1: Actual vs. Predicted Change in Real Rental Index
Figure 1 generally confirms the good fit of this model; predicted rental movements
track actual rent changes very closely over the entire period of analysis – a timeframe
15
To ensure that it was valid to drop the insignificant variables, an F-test of the joint restriction that the omitted
variables had zero coefficients was conducted; F(3,21) = 0.679 vs 1% critical value of 4.87.
16
Wheaton and Torto (1988), Sivitanides (1997) and Zhou (2008) also achieved improvements in the fit of their
models by allowing the NVR to vary through time.

13
covering three-and-a-half full cycles of the market. However, three specific features
of the graph are especially noteworthy. Firstly, it illustrates our model’s ability to
‘call’ turning points in the rental cycle. There have been six peaks and troughs in the
Dublin office market since 1978 and the equation presented in Table 5 predicts all of
these to within one year or better. Secondly, this graph underlines the robustness of
our equation by illustrating how accurately it models the behaviour of Dublin office
rents during the 2008-2009 market collapse - a period when real rents fell more than
twice as quickly as at any previous time. Finally, the graph suggests a statistically
well-behaved model with residuals distributed randomly over time and across the data
range. Indeed, this visual impression is confirmed through the standard diagnostic
tests. A Durbin-Watson statistic of 1.977 indicates no first order serial correlation in
the error terms17
, while second order autocorrelation is also ruled out by an LM test.18
The model displays no symptoms of multicollinearity while White19
and Breusch-
Pagan-Godfrey (BPG)20
tests for heteroscedasticity confirm a constant error variance.
Finally, the results of a Ramsey Reset Test on our final model indicate that earlier
concerns about possible misspecification have been eradicated.21
Given these characteristics the model presented above appears to provide a sound
basis for estimating the NVR for Dublin office property. The natural vacancy rate for
our earlier period is derived simply by dividing the negative of the intercept in Table
5 by the coefficient on lagged vacancies, as per the standard methodology set out in
Section 3. This indicates that the natural vacancy rate for Dublin offices in the 1978-
1998 period was; -8.89/-1.72 = 5.15%. This figure is approximately in line with
‘rule-of-thumb’ estimates of the NVR which have traditionally been used by local
industry practitioners.22
However, as shown below, these estimates have been
rendered obsolete by a pronounced shift in the NVR in the late 1990s.
The NVR for the 1999-2009 period can be calculated using a similar approach. This
time, however, because the intercept changed between periods, the correct procedure
is to add the constant from the earlier period to the differential intercept before
applying a negative sign and dividing by the coefficient on lagged vacancies. As
expected, this confirms a significantly higher NVR in the post 1999 period of;
-(8.89+16.94)/-1.72 = 14.98%
Section 6 – Discussion
The standard methodology for estimating a rent adjustment equation treats the NVR
as time-invariant. However, the empirical analysis herein confirms that the NVR in
Dublin’s office market did not remain constant over the 1978-2009 period. Instead it
experienced a distinct step-change, rising from approximately 5% prior to 1999 to
almost 15% thereafter.
17
d = 1.977 vs. 1% critical value of 1.601.
18
2
= 4.398 vs. 1% critical value of 9.210 (2df).
19
2
20
2
21
F(1,23) = 0.134, P = 0.717.
22
The most commonly quoted figure is 7% (Knight Frank Commercial Market Report, November 2009, DTZ
Sherry FitzGerald Dublin Office Market, Autumn 2009, Real Estate Opportunities PLC Annual Report, 2005).
However, many other commercial property agents, equity analysts, fund managers, bankers and property
consultants have quoted equilibrium vacancy rates in the 5-8% range – see, for example; The Irish Independent
15th
August 2001, 26th
June 2002, 25th
February 2004, The Sunday Business Post 22nd
February 2009.

14
To understand the possible causes of this dislocation, it is useful to distinguish
between structural changes and cyclical fluctuations. Zhou (2008) argues that normal
market cycles impact on the actual vacancy rate and cause it to deviate temporarily
from the NVR. In turn, these deviations trigger a rent adjustment process that returns
the market to equilibrium. In contrast, structural changes cause a shift in the NVR
itself. Zhou (2008) defines structural changes as those that are non-recurring, non-
predictable and have a long-term impact on the functioning of the market. These
structural changes may occur in any of the variables that affect the NVR; i.e. those
which alter the level of frictional vacancies, those which influence landlords’
preferences for vacant space, or those which affect both.
Several structural changes occurred around 1998/1999 which may have contributed to
a level-shift in the NVR for Dublin offices. Firstly, there was a fundamental change
in Ireland’s monetary policy framework when the Euro currency was introduced in
1999. This precipitated a once-off reduction in real interest rates which, in turn,
contributed to a major construction boom (Bergin et al. 2009, Williams 2006). This
was clearly evident in the Dublin office market from 1999 onwards. While average
new building completions ran at 47,568 sq m per annum between 1978-1998, output
increased almost fourfold to 174,478 sq m per year in the period 1999-2009. As a
result, Dublin’s total stock of modern office space rose by 129% in just 11 years.23
Much of this new development was speculative and net absorption – although buoyant
by historical standards at 110,101 sq m per annum - could not keep pace with
completions.24
Consequently, frictional vacancies increased dramatically during this
period and have remained elevated for the last decade. In addition to fuelling a
construction boom it is obvious that lower real interest rates after 1998 may also have
contributed to a higher NVR by reducing the cost to landlords of carrying empty
space.
Furthermore to its unprecedented scale, the geographical pattern of Dublin’s 1999-
2003 office building boom may also have contributed to the increased NVR. This
boom was unusual in that, for the first time, it saw a large-scale suburbanisation of
office development in Dublin (Bertz 2002, MacLaran and Killen 2002). This shift to
the suburbs was precipitated by a combination of push and pull factors. On one hand,
more stringent conservation policies in the Central Business District (CBD) and
increased competition for city-centre sites from alternative uses (e.g. hotel and
apartment development) pushed office development out of the traditional core area
towards more peripheral locations. On the other hand, a fragmentation of Dublin’s
planning administration acted as a pull factor because it encouraged suburban local
authorities to compete for rateable development by zoning increasing amounts of land
for office use (Bertz, 2002).25
This peripheralisation of office development may have
contributed to an increased NVR in two ways. Firstly, given that frictional vacancies
have traditionally been higher in Dublin’s suburban markets than in the city centre,
the increased weighting of suburban space within the overall office stock would have
led to a higher NVR. Secondly by opening up large tracts of relatively inexpensive
23
MacLaran and O’Connell (2003) provide a detailed description of ‘Dublin’s Fourth Office Development Boom’
which took place between 1999-2003. After a short hiatus Dublin’s fifth major office building boom followed
between 2007-2009.
24
Net absorption is simply the annual change in occupied space.
25
This phenomenon is known as “fiscal mercantilism”. Although the fragmentation of Dublin’s planning
administration into four local authorities occurred in 1994, the mid-1990s was a time of subdued development
activity, so the effects were not seen until 1999 when Dublin’s fourth office development boom began.

15
new building land, more facilitative planning arrangements in the suburbs increased
the elasticity of supply for office space. To the extent that this may have encouraged
speculative development, it is also likely to have contributed to the increased NVR.
The third major change was a rapid and pronounced restructuring of Ireland’s
economy away from manufacturing and towards services in the late 1990s. A recent
paper by Quill and Teahon (2010) indicates that, from the mid-1980s to the mid-
1990s, expansion in the Irish economy was largely driven by chemicals and
electronics manufacturing. Together, these industries accounted for 31% of average
annual GVA growth between 1987-1996. After that point, however, their
contribution faded to just 17%, while services took over as the new locomotive of
economic expansion. Thus, the contribution of financial, public and ‘other’ services
rose from 19% of average GVA growth in the 1987-1996 period to 37% between
1997-2006. Given that these sectors are intensive users of office space, this structural
shift generated positive rental expectations which may have encouraged more
speculative development and hoarding of vacant space.
Finally, it should be noted that changes to the typical office lease may also have been
important in the shift to a higher NVR. In particular, the term of the standard
institutional lease contract has declined from the traditional 35 years to less than 10
years at present (see O’Neill, 2009). As a result, tenant turnover has risen, causing a
permanent increase in the level of frictional vacancies.26
Overall, then, our finding of a sharp upward shift in the NVR for Dublin offices can
be plausibly explained by structural changes in the macroeconomic, urban planning
and institutional contexts of the late 1990s.
Having established that the NVR is now more than twice as high as was previously
assumed, a remaining question is how this revelation should affect our analysis of the
outlook for Dublin office rents. In this context Table 6 may be instructive. Currently,
the total modern office stock in Dublin is just under three-and-a-half million sq m
and, of this, 23.1% or approximately 808,000 sq m is vacant. Theory tells us that
rents will continue falling until this figure is reduced into line with its ‘natural’ level.
If we begin with the traditional assumption that the NVR is 7%, this means that
vacancies would have to be reduced to 244,848 sq m (i.e. 7% of the current office
stock) before we could expect the decline in real Dublin rents to be arrested. This
implies that an ‘overhang’ of 563,152 sq m of excess space needs to be digested
before the rent cycle will turn. To put some time frame on how long this might take
we can make a technical assumption that the net absorption of office space will
remain constant at its 32-year average of 69,181 sq m per annum. At this rate of
digestion it would take a further eight years for real Dublin office rents to stop falling.
However, if we reject the traditional assumption of a 7% NVR and accept that the true
figure is almost 15% as argued in this paper, then the outlook changes dramatically. In
this situation vacancies would only have to come down to 523,976 sq m (i.e. 14.98%
of the current total office stock) before rents should stop falling. From current levels,
26
Tax incentives for commercial property development are another area where structural change may have
occurred. However, because the main interventions took place more than a decade earlier, through the Urban
Renewal and Finance Acts of 1986, this is considered an unlikely source of the NVR shift observed in 1999. For
discussion of these tax incentives see Williams (2006) and references therein.

16
this means that only 284,024 sq m of vacant space would need to be digested and,
assuming the same rate of absorption, this would be achieved in approximately four
years.
Table 6: Impact of a Higher NVR on The Timeline for Rental Recovery
7% 14.98%
Total Stock 3,497,835 3,497,835
Vacant Stock 808,000 808,000
Vacancy Target27
244,848 523,976
Overhang 563,152 284,024
Assumed Absorption 69,181 69,181
Years to Recovery28
8.14 4.12
The above analysis clearly demonstrates the central proposition of this paper; If the
NVR for Dublin office property is higher than previously thought then the market
faces a shorter journey back to equilibrium and, all else equal, a recovery should
emerge sooner than expected. However, it should be understood that the timelines
suggested in Table 6 are not intended as forecasts per se. Rather, their purpose is
merely to emphasise the importance of the NVR in determining the position of the
market relative to its equilibrium. If we wish to derive more considered forecasts we
must remember that the timeline to recovery depends not just on the distance that the
market has to travel back to equilibrium, but also on the speed at which it makes this
transition. This transition speed is largely determined by the rate of net absorption, a
detailed analysis of which is beyond the scope of our paper. Nonetheless, it is worth
briefly considering how different rates of absorption might hypothetically affect the
timeline to a recovery in Dublin office rents.
Figure 2: Net Absorption of Dublin Office Space 1999-2010(f)
27
Calculated as the Total Stock multiplied by the relevant NVR percentage.
28
Calculated as the ‘Overhang’ divided by Annual Absorption.

17
So far, we have used the simplistic assumption that absorption will remain constant at
its long-run level. However, as shown in Figure 2, net absorption in the Dublin office
market was negative in 2009 and, although it recovered somewhat in 2010, it remains
very subdued by historical standards. If we envisage that Ireland’s economy will
develop in line with the ESRI’s Low Growth macroeconomic scenario29
it may be
reasonable to assume that net absorption might average 90% of its long-run mean
value over the medium term. Under this assumption just 62,263 sq m of vacant space
would be digested each year and, as shown in Table 7, it would take 4.56 years to
absorb the current excess. This would postpone a recovery in Dublin office rents until
mid-2015.
Table 7: Sensitivity of Recovery Timelines to Net Absorption Rates
90% LR
Average
LR
Average
120% LR
Average
Overhang 284,024 284,024 284,024
Assumed Absorption 62,263 69,181 83,000
Years to Recovery 4.56 4.12 3.42
In contrast, if we assume that economic growth will be in line with the ESRI’s High
Growth Scenario then it may be reasonable to argue that, by 2015, absorption will
recover to average levels seen in the decade prior to the recession. Assuming steady
progress towards this target, an average absorption rate of 83,000 sq m per year is
implied – approximately 20% above the long run average. At that rate of net
absorption the market would attain equilibrium in less than three and a half years,
meaning that real rental growth could resume from the middle of 2014.
While the figures in Table 7 provide more nuanced suggestions about how Dublin
office rents might develop under alternative macroeconomic scenarios, some
important caveats still need to be borne in mind. Firstly, these suggestions are
predicated on the assumption that Dublin’s total office stock will remain constant in
the medium term. Certainly, it is unlikely that any further office completions will
occur for a number of years after 2010.30
However, it is possible that the total office
stock could contract as we move towards 2013-2014, particularly if the High Growth
Scenario is realised. This is because there are a number of dated office buildings in
the city centre that are ripe for redevelopment and which would be temporarily
withdrawn from the office stock if construction works were to commence. Assuming
that these buildings were vacant prior to redevelopment, the immediate effect of this
would be to reduce the gap between the actual and the natural vacancy rates, thereby
shortening the timeline to a market recovery.
A further caveat is that, given the current economic turmoil, the NVR for Dublin
offices could move again, just as it did in the 1990s. Clearly, this would affect the
accuracy of any forecasts predicated on a rigid or outdated view of the NVR. Indeed,
even at this stage it is possible to envisage several factors that could potentially cause
29
Given the high degree of economic uncertainty that prevails, a recent ESRI paper set out two alternative
recovery paths for the Irish economy – a High Growth Scenario and a Low Growth Scenario. See Recovery
Scenarios for Ireland: An Update (Bergin et al., 2010).
30
This is corroborated by recent agency reports which indicate that the current development cycle in the Dublin
office market will terminate after Q4 2010 - see CBRE Marketview Dublin Office, Q3 2010, Lisney Office Update,
Autumn 2010, for example.

18
a structural shift in the NVR for Dublin office space. Firstly, Bergin et al. (2009,
2010) argue that the current recession is likely to result in a permanent loss of
economic output relative to the country’s previous potential. If so, then new office
development and frictional vacancies could be permanently reduced, while more
pessimistic rental expectations would reduce the incentive to hold vacant space. In a
similar vein, the abolition of upward-only rent reviews could also reduce the hoarding
of vacant space. These factors would potentially exert downward pressure on the
NVR. Conversely, however, the shortening of lease terms could lead to a higher
NVR by increasing tenant turnover. Finally, NAMA, given its scale, could affect the
NVR in ways that are not yet known. For all of these reasons it is important that
market analysts continue monitor the Natural Vacancy Rate over time.
Two final provisos relate to the interpretation of the analysis herein. Firstly, due to
data restrictions it is only possible to analyse the market at an aggregated level. This
may obscure trends and developments within particular sub-markets. For example,
as the city centre was the first target for office developers in the 1960s, it now
contains a high concentration of older buildings.31
Therefore, when the next
construction cycle begins, space withdrawals due to redevelopment could initially
lead to a more rapid tightening of the city centre market and a quicker pick-up in rents
there than elsewhere. Finally, it should be noted that the dependent variable in this
paper is real rather than nominal rental growth. If, as expected, future inflation is
positive, nominal rents are likely to begin growing some time in advance of real rents.
Section 7 – Conclusions
Commercial real estate has played a major role in the boom - and subsequent bust - of
Ireland’s economy. Yet, despite the obvious importance of this sector, there has been
little empirical analysis of Ireland’s commercial property markets. This paper focuses
on one specific aspect of commercial property – office rents. Its primary purpose is
not to generate rental forecasts. Instead its aims to provide a theoretical and
methodological framework within which practitioners can form educated judgements
about the position of the market relative to its equilibrium. That notwithstanding, the
paper makes some suggestions about the likely timeline to a recovery in Dublin office
rents. Based on the revelation that Dublin’s Natural Vacancy Rate is now around
15%, and assuming a net consumption of office space which is broadly in line with
the ESRI’s macroeconomic scenarios, widespread growth in real office rents is likely
to resume sometime between mid-2014 and mid-2015.
31
See McDonald (1985) for a detailed history of early office development in Dublin.

19
REFERENCES
Ball, M., C. Lizieri and B.D. MacGregor (1998) The Economics of Commercial
Property Markets. London: Routledge.
Barras R. and D. Ferguson (1987) “Dynamic Modelling of the Building Cycle: 1,
Theoretical Framework”, in Environment and Planning A, Vol. 19 (4) 493-520.
Begg, D., S. Fischer and R. Dornbusch (2005) Economics, 5th
Edition. London:
McGraw-Hill.
Bertz, S. (2002) “The Peripheralisation of Office Development in the Dublin
Metropolitan Area – The Interrelationship Between Planning and Development
Interests”, in Irish Geography, Vol. 35(2) 197-212.
Bergin, A., T. Conefrey, J. FitzGerald, and I. Kearney (2009) “Recovery Scenarios for
Ireland”, Research Series No.7, May 2009. Dublin: ESRI.
Bergin, A., T. Conefrey, J. FitzGerald, and I. Kearney (2010) “Recovery Scenarios for
Ireland: An Update”, Special Article in Quarterly Economic Commentary, Summer
2010. Dublin: ESRI.
Berwin, Leighton, Paisner (2009) Commercial Leases in Europe – The Essential
Guide. London: EuroProperty.
Blank, D. M., and L. Winnick (1953). “The Structure of the Housing Market”, in
Quarterly Journal of Economics, 67, 181–203.
Brounen, D. and M. Jennen (2009) “Local Office Market Dynamics – A Tale of Ten
Cities”, in Journal of Real Estate Finance and Economics, Vol. 39, 385-402.
CBRE (Various Years) Marketview, Dublin Office. Dublin: CBRE.
D’Arcy, E., McGough, T., & Tsolacos, S. (1999). “An Econometric Analysis and
Forecasts of the Office Rental Cycle in the Dublin Area”. Journal of Property
Research, 16(4), 309–321.
De Leeuw, F. and N.F. Ekanem (1971) “The Supply of Rental Housing”, in American
Economic Review, Vol. 61, 806-17.
Dunse, N. C.Jones, A. Orr and H. Tarbet (1998) “The Extent and Limitations of Local
Commercial Property Market Data”, in Journal of Property Valuation and Investment,
Vol. 16(5) 455-473.
DTZ Sherry FitzGerald (Various Years) Dublin Office Market. Dublin: DTZ Sherry
FitzGerald.
Gabriel , S.A. and F.E. Nothaft (2001) “Rental Housing Markets, The Incidence and
Duration of Vacancy and the Natural Vacancy Rate”, in Journal of Urban Economics,
Vol. 49(1), 121-149.

20
Gardiner, C. and J. Henneberry (1988) “The Development of a Simple Regional
Office Rent Prediction Model”, in Journal of Valuation, Vol. 7 (1988).
Gardiner, C. and J. Henneberry (1991) “Predicting Regional Office Rents Using
Habit-Persistence Theories”, in Journal of Property Valuation and Investment, Vol.
9(3).
Grenadier, S. R. (1995). “Local and National Determinants of Office Vacancies.”
Journal of Urban Economics, 37, 57–71. (1), 31–47.
Hendershott, P. H. (1995) “Real Effective Rent Determination: Evidence From the
Sydney Office Market”, in Journal of Property Research, Vol. 12 127-135.
Hendershott, P. H. (1996). “Rental Adjustment and Valuation in Overbuilt Markets:
Evidence From the Sydney Office Market”. Journal of Urban Economics, 39, 51–67.
Hendershott, P. H., Lizieri, C. M., & Matysiak, G. A. (1999). “The Workings of the
London Office Market”, in Real Estate Economics, 27(2), 365–387.
Hendershott, P. H., MacGregor, B., & Tse, R. (2002). “Estimation of the Rental
Adjustment Process”, in Real Estate Economics, 30(2), 165–183.
Hendershott, P. H., C.M. Lizieri and B.D. MacGregor (2010) “Asymmetric
Adjustment in the London Office Market” in Real Estate Economics, Vol. 41(1), 80-
101.
Jones Lang LaSalle (Various Years) On Point – Irish Property Index. Dublin: Jones
Lang LaSalle.
Jud, G.D. and J. Frew (1990) “Atypicality and the Natural Vacancy Rate Hypothesis”,
in AREUEA Journal, Vol. 18(3) 294-301.
Knight Frank (2009) Commercial Market Report, November. Dublin: Knight Frank.
Krainer, J. (2001) “Natural Vacancy Rates in Commercial Real Estate Markets”, in
FRBSF Economic Letters, No. 2001-27, 5th
October 2001. San Francisco: FRBSF.
Leishman, C. (2003) Real Estate Market Research and Analysis. Palgrave McMillan:
Basingstoke.
Lisney (Various Years) Rental Indices. Dublin: Lisney.
Lisney (Various Years) Annual Review. Dublin: Lisney.
Lisney (Various Years) Dublin Offices. Dublin: Lisney.
MacLaran, A., M. MacLaran and P. Malone (1987) “Property Cycles in Dublin: The
Anatomy of Boom and Slump in the Industrial and Office Property Sectors”, in The
Economic and Social Review, Vol. 18(4) 237-256.

21
MacLaran, A. (1996) “Office Development in Dublin and the Tax Incentive Areas”,
in Irish Geography Vol. 29(2) 49-54.
MacLaran, A. and J. Killen (2002) “The Suburbanisation of Office Development in
Dublin and its Transport Implications”, in Journal of Irish Urban Studies, Vol.1(1)
21-35.
MacLaran, A. and O’Connell, R. (2007) “Dublin’s Fifth Office Development Boom”,
in Journal of Irish Urban Studies, 2007 Vol. 6, 179-186.
MacLaran, A. and O’Connell, R. (2003) “Dublin’s Fourth Office Development
Boom”, in Journal of Irish Urban Studies, Vol. 2(2), 85-91.
Malone, P. (1981) Office Development in Dublin 1960-1980. Dublin: Department of
Geography, Trinity College.
McCartney, J. (2008) “An Empirical Analysis of Development Cycles in the Dublin
Office Market 1976-2008” in Quarterly Economic Commentary, Winter. Dublin:
ESRI.
McDonald, F. (1985) The Destruction of Dublin. Dublin: Gill and Macmillan.
McDonald, J. F. (2000). “Rent, Vacancy and Equilibrium in Real Estate Markets”.
Journal of Real Estate Practice and Education, 3(1), 55–69.
McDonald, J. F. (2002). “A Survey of Econometric Models of Office Markets”.
Journal of Real Estate Literature, 10(2), 223–242.
McGreal, W.S., J. Berry, C. McParland and B. Turner (2004) “Urban Regeneration,
Property Performance and Office Markets in Dublin”, in Journal of Property
Investment and Finance, Vol. 22(2) 162-172.
Ng, B.F. and D. Higgins (2007) “Modelling the Commercial Property Market: An
Empirical Study of the Singapore Office Market”, in Pacific Rim Property Research
Journal, Vol. 13(2) 176-192.
O’Neill, F. (2009) “A New Model for Commercial Property”, in Finance, Vol. 23 (3),
p. 13-15).
Phyrr, S.A., S.E. Roulac and W.L. Born (1999) “Real Estate Cycles and Their
Strategic Implications for Investors and Portfolio Managers in the Global Economy”,
in Journal of Real Estate Research, Vol. 18(1).
Quill, P., and P. Teahon (2010) “Structural Economic Change in Ireland 1957-2006;
Statistics, Context and Analysis”, in Journal of the Statistical and Social Enquiry
Society of Ireland, Vol.39 2009/2010.
Real Estate Opportunities PLC (2005) Annual Report.

22
Rosen, K.T. and L.B. Smith (1983) “The Price Adjustment Process for Rental
Housing and the Natural Vacancy Rate”, in The American Economic Review, Vol.
73(4) 779-786.
Rosen, K.T. (1984) “Toward a Model of the Office Building Sector”, in AREUEA
Journal, Vol. 12(3) 261-269.
Sanderson, B., K. Farrelly and C. Thoday (2006) “Natural Vacancy Rates in Global
Office Markets”, in Journal of Property Investment and Finance, Vol. 24(6) 490-520.
Shilling, J. D., Sirmans, C. F., & Corgel, J. B. (1987). “Price Adjustment Process for
Rental Office Space”, in Journal of Urban Economics, 22, 90–100.
Sivitanides, P.S. (1997) “The Rent Adjustment Process and the Structural Vacancy
Rate in the Commercial Real Estate Market”, in Journal of Real Estate Research,
Vol.13(2), 195-209.
Tse, R.Y.C. and D. Fischer (2003) “Estimating Natural Vacancy Rates in Office
Markets Using a Time-Varying Model”, in Journal of Real Estate Literature, Vol.
11(1) 37-45.
Voith, R. and T. Crone (1988) “National Vacancy Rates and the Persistence of Shocks
in U.S. Office Markets”, in AREUEA Journal, Vol.16(4) 437-458.
Wheaton, W. C., & Torto, R. G. (1988). “Vacancy Rates and the Future of Office
Rents”, in Journal of the American Real Estate and Urban Economics Association,
16(4), 430–436.
Wheaton, W., R. Torto and P. Evans (1997) “The Cyclic Behaviour of the Greater
London Office Market”, in Journal of Real Estate Finance and Economics, Vol.
15(1).
Williams, B. (2006) “Fiscal Incentives and Urban Regeneration in Dublin 1986-
2005”, in Journal of Property Investment and Finance, Vol.24(6).
Zhou, J. (2008) “Estimating Natural Vacancy Rates with Unknown Break-Points for
the Chicago Rental Housing Market”, in Journal of Housing Research, Vol. 17 (1), pp
61-74.

Predicting turning points in the rent cycle[1]

More Related Content

Similar to Predicting turning points in the rent cycle[1] (20)

More from ExSite (20)

Recently uploaded (20)

Predicting turning points in the rent cycle[1]