Ref #20460898 v1.8
What Drives Rents in
New Zealand? National
and Regional Analysis
Alan Bentley*, Enzo Cassino* and Nam Ngo**
Housing Technical Working Group
August 2023
*Reserve Bank of New Zealand
**Ministry of Housing and Urban Development
1
1 What Drives Rents in New Zealand? National and Regional Analysis
Contents
Executive Summary ____________________________________________________________________________ 2
Key Findings ___________________________________________________________________________________ 3
Introduction ___________________________________________________________________________________ 3
Key Facts of Rents in New Zealand ___________________________________________________________ 4
Literature Review ______________________________________________________________________________ 8
Theoretical Framework of the Rental Market _________________________________________________ 9
Data and Methodology_______________________________________________________________________ 10
Data Sources 10
Methodology 11
Results ________________________________________________________________________________________ 12
National Results 12
Regional Results 15
Simple Panel Regressions, Using Annual Change 15
Including Period Dummy Variables 17
Including Regional Interactions and Pooled Regressions 18
Conclusion ____________________________________________________________________________________ 20
References ____________________________________________________________________________________ 21
Appendix – Additional Regression Results __________________________________________________ 23
National Results 23
Regional Results 24
The analysis in this paper was carried out by the Housing Technical Working Group (HTWG). We would like to
thank HTWG colleagues for helpful comments on the analysis and earlier drafts of the paper, especially
Dominick Stephens, Chris McDonald, Andrew Coleman, Chris Parker, Frances Krsinich, Tyler Smith and
Talosaga Talosaga.
2
2 What Drives Rents in New Zealand? National and Regional Analysis
Executive Summary
This paper aims to provide an initial framework to improve our understanding of the factors that
impact housing rentals in New Zealand. This framework is useful for several reasons. Firstly, rents
provide a better signal of the balance of supply and demand for dwellings than house prices do.
This is because rents do not reflect expectations for future gains as house prices do. Secondly,
providing a better understanding of rent drivers can lead to better government policy as renters
typically pay a larger proportion of their incomes on housing costs than owner occupiers and so
are more vulnerable to large movements in housing costs. Thirdly, forecasting rents can also
improve the accuracy of house price forecasts, as they are one of the factors that influence house
prices. Finally, the framework helps us to test theories of how land and housing markets operate.
We find that, over the past 20 years, nominal wage inflation and the relative supply and demand
of dwellings are the two key drivers of rent inflation at both the national and regional level,
through impacting tenants’ ability and willingness to pay, and the availability of rental properties
respectively. When the effect of other factors is excluded, a 1 percent increase in nominal wages
leads directly to a 1 percent increase in new tenancy rents. A 1 percent increase in people per
dwelling,
1
leads to a 1.5 percent increase in rents at the national level. These results also hold using
Stats NZ’s estimate of rental inflation for all tenancies, albeit with lower magnitudes, as rents for
existing tenancies are typically less volatile than new ones.
We also find that rent inflation at the national level is affected by mortgage rates (increasing
mortgage rates increase rents) and the unemployment rate (increasing unemployment decreases
rents). However, their contributions are smaller and less significant than wages and the physical
supply and demand of dwellings. The small effect of mortgage rates on rents is consistent with
previous analysis done by the Housing Technical Working Group on the impact of land supply
restrictions. When land supply is highly constrained, we would expect financial factors, such as
interest rates, to have a greater impact on house prices than rents.
Expanding the research to consider regional differences, we find the identified key drivers of rental:
growth in wages or household income, and physical supply and demand, still explain changes in
rents at a regional council level. However, if we allow the model to assume these factors have
different sized impacts in different regions it improves the model’s predictive ability, suggesting a
variety of region-specific factors affect how much these drivers impact on rent inflation. These
differences may include local planning restrictions, infrastructure levels, and capacity of the local
construction industry to respond to changes in demand.
____________
1
People per dwelling is used as an indicator of the relative supply and demand of dwellings. We assume an increase in people per dwellings suggests demand is growing more
quickly than supply.
3
3 What Drives Rents in New Zealand? National and Regional Analysis
Key Findings
1. Understanding the key drivers of rents is important to monitor and assess the balance of
supply and demand in the housing market, improve the accuracy of house price forecasts, and
identify potential hot spots at the regional level.
2. Wage inflation and relative supply and demand of dwellings (measured by people per
dwelling) are the two key drivers of rent inflation for new tenancies at the national level.
3. Mortgage interest rates positively affect rents but relatively little, and the relationship is not
robust across model specifications. This is consistent with previous analysis done by the
Housing Technical Working Group on the impact of land supply restrictions.
4. The identified key drivers are robust to local circumstances, although unobserved region-
specific factors can dampen or magnify the effects in particular regions.
Introduction
New Zealand rents have received growing attention as the proportion of people who rent has
been increasing since the early 1990s. This paper aims to provide an initial framework to improve
our understanding of the factors that impact housing rentals in New Zealand.
2
This analysis is
useful for several reasons. Firstly, rents provide a better signal of the balance of supply and
demand for dwellings than house prices do. This is because rents do not reflect expectation for
future gains as house prices do. Secondly, providing a better understanding of rent drivers can
lead to better government policy as renters typically pay a larger proportion of their incomes on
housing costs than owner occupiers and so they are more vulnerable to large movements in
housing costs. Thirdly, forecasting rents can also improve the accuracy of house price forecasts, as
they are one of the factors that influence house prices. Finally, the framework helps us to test
theories of how land and housing markets operate.
____________
2 Throughout this paper we use rents as shorthand for actual rentals for housing. We do not consider or model imputed rents on owner-occupied housing.
4
4 What Drives Rents in New Zealand? National and Regional Analysis
Figure 1 – New Zealand rents in context
Overall rents have increased broadly in line with wage growth over the past two decades, albeit at
a faster rate than general inflation, as measured by the Consumers Price Index (Fig. 1). In contrast,
house prices have risen further than rents. The tri-agency Housing Technical Working Group
(HTWG)
3
identified these differing growth rates, amongst other things, as evidence to support their
conclusions presented in Assessment of the Housing System: with insights from the Hamilton-
Waikato Area (HTWG, 2022). The report demonstrated the relative importance of systematic
interest rate declines and the tax system, in the context of restricted land supply (land use rules,
regulations, and constraints), compared to the impact of dwelling supply relative to population
growth. The group concluded that physical dwelling supply has not been the major driver of house
prices over the past 20 years. If this had been the case, we would have expected house price
increases to have been more in line with the increase in rents. Underpinning this assessment is an
assumption that rents are influenced by the relative supply and demand of physical dwellings.
Research presented in this paper aims to further our understanding of the drivers of rents.
The rest of the paper is organized as follows: section 2 summarises key features of rents in New
Z
ealand, section 3 reviews the previous literature on modelling rents internationally and in New
Zealand, section 4 describes the theoretical framework for analyzing rents, section 5 describes the
data and modelling approach. The results are presented in section 6, followed by the conclusion in
section 7.
Key Facts of Rents in New Zealand
The share of New Zealand households who pay rent has increased significantly during the past
three decades, rising from about 23 percent in 1991 to 32 percent in 2018 (Stats NZ, 2020). The
____________
3 Members of the HTWG are affiliated with the Treasury, Reserve Bank of New Zealand, and Ministry of Housing and Urban Development.
5
5 What Drives Rents in New Zealand? National and Regional Analysis
associated decline in home ownership has been particularly acute for young adults, with the
proportion of New Zealanders aged 25 to 34 who are owner-occupiers declining from about 65
percent in 1988 to 35 percent in 2018 (Bentley, 2021). The number of households in rented
dwellings increased from about 290,000 in 1996 to 530,000 in 2018.
From a wellbeing perspective, rents matter since low-income households have little discretion over
their level of housing expenditure. Renters typically have lower incomes than owner-occupiers,
spend a greater share of their income on housing costs, and have lower material wealth (ibid).
New Zealand rental properties are typically of lower physical quality compared with owner-
occupied properties (New Zealand Treasury, 2022).
The usually resident population in New Zealand increased by over 1 million people during the
study period, from an estimated 4.0 million in June 2003 to 5.1 million in June 2022, at an average
growth rate of 1.3% per year. Over the same period, the number of dwellings also increased by
1.3% per year, by 500,000 to 2 million. However, these long-run growth rates hide periods of
mismatch between population and dwelling growth (Fig. 2). Variation over time in New Zealand’s
population growth rate is driven primarily by changes in net external migration. Notably, the
population was growing at a faster rate than dwellings during the period 2015–20, increasing the
number of people per dwelling to a high of 2.64 as at June 2020. Border restrictions over the
following two years curtailed population growth, whilst dwelling growth continued, reducing
people per dwelling to 2.56 as at June 2022. This is similar to the 2.58 people per dwelling as at
June 2003, at the beginning of our study period.
Figure 2 – Population and dwelling growth rates compared
Enhancements to the measurement of rent inflation, including the use of granular administrative
data and a new approach for quality-adjustment, has improved the reliability of rental price
indices. The changes also facilitated the introduction of an additional ‘flow’ series in 2019, which
shows price change for new tenancies (Stats NZ, 2019; Bentley, 2022a). In comparison, the ‘stock’
series, used as an input into the overall Consumers Price Index, measures price change across all
tenancies. Movements in the latter series tend to be lagged and less volatile (Fig. 3), reflecting the
stickiness of rents for sitting tenants, who can be said to enjoy a ‘tenancy discount’ (Bentley, 2021).
6
6 What Drives Rents in New Zealand? National and Regional Analysis
Figure 3 – New tenancy rent inflation compared with inflation for all rentals
Just over a third of rental households are in Auckland, another third in the major urban areas of
Canterbury, Wellington and Waikato, with the remainder in less densely populated regions (Fig. 4).
There is substantial regional variation in the proportion of households who are not owner-
occupiers, the vast majority of who pay rent.
4
Over 40 percent of households are non-
homeowners in Gisborne and Auckland, compared with less than 30 percent in Marlborough and
Tasman regions.
Figure 4 - Regional diversity in rental markets
____________
4 A small proportion of household who are not owner-occupiers do not pay rent.
7
7 What Drives Rents in New Zealand? National and Regional Analysis
National trends can hide regional diversity of rent inflation (Fig. 5). We found evidence of spatial
correlation in rent growth, where periods of higher price growth occurred in one region at the
same time as neighbouring regions.
Figure 5 – Regional rent inflation
Rent inflation has been above average in many provincial North Island regions
5
since about 2016.
Counter-balancing this has been below-average growth rates in Auckland, home to over a third of
renters. The Canterbury region experienced major earthquakes in 2010 and 2011, reducing the
number of dwellings available for habitation. Rents increased markedly over the following few
years, subsequently reversing once supply and demand were brought back into balance after the
rebuild. These key trends hint at the role of physical supply and demand for rental properties on
rents. Later we provide a more rigorous empirical assessment of this finding.
____________
5 Namely, Northland, Waikato, Bay of Plenty, Gisborne, Hawke’s Bay, Taranaki, Manawatu-Whanganui.
8
8 What Drives Rents in New Zealand? National and Regional Analysis
Literature Review
A number of different frameworks can be used to model the drivers of rents. Previous studies have
used a range of data sources. Some studies have used individual level rent data, whereas other
studies have used aggregate data. Individual level data is well suited to inferring the causal impact
of specific events or policy changes, such as changes in the level of the Accommodation
Supplement. However, aggregate rental data are more useful for estimating the relationship
between rents and macroeconomic factors such as wages and interest rates. Our study uses
aggregate level data as we are primarily interested in these macroeconomic relationships.
Previous studies have used a variety of modelling techniques to study rent growth. Studies focused
on identifying the impact of individual factors on rents tend to use single equation econometric
methods, such as Ordinary Least Squares (OLS) or panel data techniques. In contrast, studies that
concentrate on forecasting rents may use methods such as Vector Autoregressions (VARs). In this
study we are interested in the macroeconomic relationship between rents and other factors, so we
use OLS for the national level analysis and panel estimation for the regional analysis.
Another distinction in the literature is whether to estimate the structural demand and supply curves
for rent or the combined reduced form relationship. Based on economic theory and the previous
empirical literature, the demand for rental housing depends on factors such as household income,
house prices, the number of people per dwelling and interest rates. The supply of rental housing
depends on factors including interest rates and inflation.
Compared to the extensive empirical work on drivers of house prices, research on drivers of rents
is much more limited, both internationally and in New Zealand.
6
The available studies have used a
variety of methods and focus on the role of different factors.
In the time series analysis, much of the empirical work has focussed on modelling US rents. Duca,
Muellbauer and Murphy (2016) model real rents (deflated by the PCE deflator). They find a positive
long-run impact of real incomes and house prices on rents, and a negative impact of user costs,
which incorporates the impact of lagged house price changes and interest rates. Dias and Duarte
(2019) use a structural VAR approach to examine the impact of US monetary policy on rents, and
find that after a tightening in monetary policy, house prices decline whereas rents increase,
indicating that monetary policy may influence households’ decision to own a house or rent.
Saunders and Tulip (2019) include equations for rents and rental vacancies in a broader model of
the Australian housing market. They find dwelling completions and changes in population drive
the rental vacancy rate, which in turn has a strong impact on rents. In addition, rents have a large
impact on house prices, along with interest rates and house price momentum. Howard and
Liebersohn (2021) use a spatial equilibrium framework to decompose the drivers of US rents, and
find an increasing role for demand to live in cities in which housing supply is inelastic.
Relatively few studies have analysed the New Zealand rental market. Coleman and Scobie (2009)
build a simple structural model with parameters of supply and demand for owner-occupied and
rental housing to assess the impact of various policy actions. They find a reduction in tax
concessions for landlords would increase rents and moderate house prices. In addition, lower
housing costs, such as through lower regulatory and consent costs, would reduce rents and house
prices. Lower real interest rates would reduce rents and increase house prices. The only study we
____________
6 For a broader review of the literature on modelling house prices and rents, see Duca, Muellbauer and Murphy (2021)
9
9 What Drives Rents in New Zealand? National and Regional Analysis
are aware of that has analysed the New Zealand rental market at a regional level is Nunns (2019).
His study develops a calibrated spatial equilibrium model to analyse the impact on regional house
prices and rents of rising housing demand arising from population growth, credit availability and
tax policy settings combined with constrained supply due to zoning rules. His analysis finds regions
with more binding supply constraints have experienced larger rent (and house price) increases in
response to migration shocks.
Theoretical Framework of the Rental Market
The theoretical framework for our model of rents is similar to the approaches used by Coleman
and Scobie (2009) and Watson (2013) for modelling the property market. Whereas those models
allow individuals to choose between renting and owner occupying, our model focuses only on the
rental segment of the property market. In the model, rents P
r
and the number of households that
are renting H
r
are determined by the demand for and supply of rental services.
We assume in the short-run the supply of rental properties is fixed. In longer term, the supply can
adjust as more (or less) people chose to become landlords or more (or fewer) houses are built to
rent to tenants
= (
,
, , ).
Hi
gher rents increase the return to landlords, increasing the supply of rental properties. Higher
house prices P
H
reduce the supply of rental properties as the yield to landlords is reduced. Higher
mortgage rates i reduce the supply of rentals as the cost of capital to invest in property rises.
Higher inflation reduces the supply of rentals as it increases landlords’ expenses.
The demand side of the rental market D
r
can be written as
= (
,
, , , ).
Higher rents relative to the cost of buying a house, which is captured by house prices and
mortgage interest rates, lowers the demand for rental properties. Higher household income Y
increases demand for rent by increasing renters’ ability to pay. We also include the average
number of people per dwelling as influencing the demand for rentals. People per dwelling is often
used as an indicator of the demand-supply imbalance in the rental market. However, the
theoretical relationship between rents and people per dwelling is not clear cut. An increase in
people per dwelling may reflect social factors, such as young people choosing to live with their
parents for longer. This would reflect a shift down in the rental demand curve, and put downward
pressure on rents. On the other hand, if people chose to live in large households in response to
rising rents, this represents a shift along the demand curve, and higher people per dwelling would
be positively correlated with rents. The estimated coefficient in our model captures the net impact
of both of these channels. As part of our empirical analysis we also examine the impact of splitting
people per dwelling into population growth and increases in the number of dwellings to examine
whether each factor has a different impact on rent growth.
10
10 What Drives Rents in New Zealand? National and Regional Analysis
We combine the supply and demand equations for rental services to solve the reduced form of
the model for rents and the number of households that are renting
= (, , ,
, )
a
nd
=
(
, , ,
, 
)
.
E
stimating the demand curve and the supply curve separately would allow us to identify the
impact of a shock that shifts the demand curve or a supply curve, such as an income shock or a
shock in the number of people per dwelling. However, in this study we are primarily interested in
using our models to understand the relationship between rents and other macroeconomic
variables. As a result, we concentrate on estimating the reduced form of the relationship between
rent growth and exogenous factors that impact it. The parameters we estimate are a combination
of the separate coefficients on each factor in the demand curve and the supply curve, so the
estimated coefficients represent the correlation of rents with each factor rather than necessarily the
causal relationships.
Data and Methodology
Data Sources
This section describes the data sources used for our empirical analysis. Based on the availability of
time series, our study period is from 2003Q4 to 2022Q2. Rates of change are calculated for all
variables: quarterly change is defined as the change from previous quarter; annual change is
defined as the change from the same period in the previous year.
Stats NZ’s Rental Price Index (RPI)
7
is used at the national level. This is a quality-adjusted price
index of rental inflation, derived from the Ministry of Business, Innovation and Employment
(MBIE)’s Tenancy Bond data (Stats NZ, 2019; Bentley, 2022a). The ‘flow’ series is used, which shows
the change in rents for new tenancies, except where it is stated that the ‘stock’ series is used. The
latter shows rent changes across all renters, including sitting tenants. Regional council level RPIs
are sourced from the Ministry of Housing and Urban Development (HUD). These use the same
quality adjustment methodology as Stats NZ’s national RPI (HUD, 2022). CoreLogic’s House Price
Index (HPI) has been sourced from HUD. The index methodology uses a Sales Price Appraisal
Ratio, a quality adjustment approach widely used in New Zealand for HPIs (ibid). Consumers Price
Index less rentals for housing subgroup is a non-standard series available from Stats NZ’s infoshare
service (table reference: CPI017AA).
Wage measures are sourced from Stats NZ: average weekly earnings from the Quarterly
Employment Survey (QES), a sample survey of about 4,000 enterprises; median earnings per job
from the quarterly Linked employer-employee data (LEED), created from administrative pay-as-
you-earn income tax data (Stats NZ, 2021). QES data is used for national level regressions. Since
the survey does not contain regional breakdowns, LEED data is used for regional regressions.
Modelled household disposable income time series are sourced from HUD. These use quarterly
LEED earnings benchmarked to annual household survey estimates of disposable (after tax)
____________
7 Backcast using longer timeseries from Ministry of Housing and Urban Development prior to Stats NZ RPI, which starts in November 2006.
11
11 What Drives Rents in New Zealand? National and Regional Analysis
income (HUD, 2022). Unemployment rates are sourced from Stats NZ’s Household Labour Force
Survey, a sample survey of about 16,000 households (Stats NZ, 2021).
Stats NZ’s national and subnational population estimates are based on Censuses of Population
and Dwellings (Stats NZ, 2022a; Stats NZ, 2022b). Intercensal estimates are derived from registered
births, deaths, and net migration of residents. Dwelling estimates are based on Censuses of
Population and Dwellings, interpolated and extrapolated using Stats NZ’s building consents data
(see Bentley, 2022b, for a full description of the methodology used).
Mortgage rates are sourced from the Reserve Bank of New Zealand (RBNZ). Two series were
explored, both are a simple average of rates advertised to new customers by registered banks.
Standard rates are used for the floating mortgage rate (RBNZ, 2023a), whereas the 2-year fixed
mortgage rate refers to ‘special’ rates offered to borrowers who meet certain conditions as
specified by the bank (RBNZ, 2023b)
8
.
Methodology
For the national-level analysis, we use a simple regression model of rental inflation on the potential
determinants, and use Ordinary Least Squares (OLS) to estimate model coefficients.

=
β


+
Where the change in rent  at time is a function of
(
= 1, … ,
)
variables .
For national-level analysis, we use the general-to-specific approach to estimate the models and
include both the contemporaneous and lagged values of the explanatory variables. We use
quarterly data in order to ensure sufficient sample size for the number of variables. However, there
are concerns that for rental inflation and several of the explanatory variables, quarterly data may
include too much noise and hide the underlying signal. Therefore, we test our model with annual
changes instead of quarterly changes. To maintain the sample size but also avoid serial correlation,
we include the lag of four instead of one for all variables.
For regional analysis, we use panel regressions to jointly estimate relationships for all regions,

=



+




+


+
,
w
here region fixed-effects
are estimated for all but one region
9
1 . The structure imposes
the same relationships between dependent and independent variables for all regions, but allows
for a different constant for each region. We also investigate: (i) the impact of including the
interaction of regional fixed-effects with the other variables; and (ii) omitting the region fixed-
effects in a pooled regression.
____________
8 Backcast using standard rates prior to the start of special rate series in 2017
9 One region is omitted to identify the model.
12
12 What Drives Rents in New Zealand? National and Regional Analysis
Results
National Results
In this section we present the results of the aggregate national level modelling. There are a
number of findings about the impact of macroeconomic factors on rent growth.
All else equal, an increase in nominal wages leads directly into a 1-to-1 ratio increase in rents, as
shown in all columns of Table 1. The correlation is stronger contemporaneously, but we also find
that lagged wage inflation contributes to rental inflation.
In terms of relative supply and demand, a 1 percent increase in people per dwelling leads to a 1.5
per cent increase in rents (Table 1). There is some limited evidence suggesting that the higher the
increase in the supply/demand gap, the stronger the wage-rent relationship, due to competition
for rental properties allowing landlords to capitalize on renters’ wage gains. The interaction term
between wage and people per dwelling in table A.2 is positive but not statistically significant.
Across all model specifications in Table 1, the unemployment rate is negatively correlated with
rental inflation, i.e an increase in unemployment rate would lead to a decrease in rental inflation.
There are two possible explanations for this. Firstly, better job security can encourage people to
form new and smaller households, which in turn, increases the demand for rental properties. For
example, young adults may be more inclined to leave the family home when unemployment is
low. Secondly, a strong labour market and positive economic outlook would ensure tenants’
current and future ability to pay, allowing landlords to raise rents.
Table 1: Baseline model
Rent inflation
(nominal, flow)
Rent inflation
(nominal, flow)
Rent inflation
(nominal, flow)
Rent inflation
(nominal, flow)
Predictors
(1)
(2)
(3)
(4)
Rent inflation (lagged) -0.19
(0.12)
-0.19
*
(0.11)
-0.17
(0.11)
-0.15
(0.11)
Wage growth
0.54
***
(0.11)
0.56
***
(0.08)
0.54
***
(0.09)
0.54
***
(0.09)
Wage growth (lagged)
0.35
***
(0.12)
0.36
***
(0.11)
0.39
***
(0.12)
0.39
***
(0.12)
People per dwelling -0.51
(0.78)
People per dwelling (lagged) 1.97
**
(0.86)
1.73
***
(0.52)
1.63
***
(0.53)
1.51
***
(0.53)
Mortgage rate 0.02
(0.02)
Mortgage rate (lagged)
0.02
(0.02)
0.03
(0.02)
0.03
(0.02)
Inflation excluding rents -0.12
(0.20)
13
13 What Drives Rents in New Zealand? National and Regional Analysis
Rent inflation
(nominal, flow)
Rent inflation
(nominal, flow)
Rent inflation
(nominal, flow)
Rent inflation
(nominal, flow)
Inflation excluding rents (lagged)
0.15
(0.20)
0.14
(0.17)
0.13
(0.17)
0.15
(0.17)
Unemployment rate
0.01
(0.01)
Unemployment rate (lagged)
-0.02
(0.01)
-0.02
(0.01)
-0.02
(0.01)
-0.02
*
(0.01)
House price inflation -0.01
(0.06)
House price inflation (lagged) 0.08
(0.07)
0.08
*
(0.04)
Observations 73 73 73 73
R2 / R2 adjusted 0.489 / 0.376 0.470 / 0.404 0.440 / 0.380 0.423 / 0.371
We test the robustness of the results by using population growth and dwelling growth separately
instead of the change in people per dwelling, as shown in the third column of table 2. The signs
and magnitudes of the regression coefficients are as expected. All else equal, an increase in
population is positively correlated with rent inflation, while an increase in dwellings is negatively
correlated with rental inflation. As the Wald test cannot reject the null hypothesis of equal
coefficients, we conclude that the contributions of population and dwellings growth towards rental
inflation are equally important, thus combining both into the people per dwelling variable as in the
baseline model is plausible.
To check the robustness of the results found with people per dwellings, we also calculate the
vacancy rate of rental properties as an alternative indicator of the supply-demand imbalance. We
use unit record bond data to create a measure of the time between tenancies at property level as
a proxy for the vacancy rate. Column 4 in table 2 show the results from using the vacancy rate
instead of people per dwelling to capture the relative supply-demand balance. These results are
consistent with the baseline model. An increase in the vacancy rate signals the easing of pressure
in the rental market and leads to a decrease in rental inflation.
Another finding is that the sensitivity of rent inflation to mortgage interest rates is positive.
However, the sensitivity is quite small and is not always statistically significant across model
specifications. There are several possible explanations for the sensitivity of rent inflation to
mortgage rates. For example, first home buyers may delay buying due to rising mortgage
unaffordability, increasing demand for rental property. Higher financing costs and restricted land
markets may limit the supply response to increased demand for rentals when interest rates rise,
putting further pressure on rent inflation. There may also be feedback loops in the banking sector,
which can limit the supply response of rental properties to lower interest rates. As supply begins to
increase relative to demand this will increase vacancy rates and reduce yields for property
investors. This may lessen banks’ appetite to lend for further rental property development.
14
14 What Drives Rents in New Zealand? National and Regional Analysis
Across all specifications in table 1, the impact of general inflation beyond that already captured by
the nominal wage coefficient, measured by CPI less rents, is positive but not statistically and
economically significant.
Table 2 Alternative measure of relative supply and demand
Rent inflation
(nominal, flow)
Rent inflation
(nominal, flow)
Rent inflation
(nominal, flow)
Rent inflation
(nominal, flow)
Predictors
(1)
(2)
(3)
(4)
Wage 0.56
***
(0.08)
0.52
***
(0.08)
0.54
***
(0.09)
0.34
***
(0.09)
Wage (lagged) 0.36
***
(0.11)
0.39
***
(0.11)
0.39
***
(0.12)
0.28
**
(0.12)
People per dwelling
(lagged)
1.73
***
(0.52)
Adults per dwelling
(lagged)
1.46
***
(0.46)
Population growth
(lagged)
1.68
***
(0.58)
Dwellings growth
(lagged)
-1.43
(0.97)
Change in vacant
time (lagged)
-0.01
***
(0.00)
Observations 73 73 73 73
R2 / R2 adjusted
0.470 / 0.404
0.445 / 0.385
0.441 / 0.371
0.457 / 0.398
* p<0.1 ** p<0.05 *** p<0.01
In column 3 of table 3, we look at the impact real wage on real rents by deflating both by the CPI
excluding rents. All regression coefficients retain their expected signs and significances which
suggest that the impact of wages on rents is direct rather than through inflation.
While the flow RPI, created using new tenancies, is a good real-time indicator of rental inflation,
the stock measure of RPI is considered a more comprehensive and reliable representation of the
situation facing most tenants, who remain in their current rental properties
. In column 2 of table 3,
we test our model using stock rental inflation as the dependent variable.
The results are consistent with the flow measure. All regression coefficients retain their expected
signs and significances. However, the magnitudes are smaller, which may be because rental
inflation for existing tenancies is typically less volatile than rental inflation for new tenancies,
because existing rents only tend to adjust infrequently.
15
15 What Drives Rents in New Zealand? National and Regional Analysis
Table 3: Alternative measure of rental inflation
Rent inflation
(nominal, flow)
Rent inflation
(nominal, stock)
Rent inflation
(real, flow)
Predictors (1) (2) (3)
Rent inflation (lagged) -0.17
(0.11)
0.65
***
(0.10)
-0.27
**
(0.11)
Wage growth
0.54
***
(0.09)
0.09
***
(0.02)
0.62
***
(0.08)
Wage growth (lagged) 0.39
***
(0.12)
0.09
***
(0.03)
0.38
***
(0.12)
People per dwelling (lagged)
1.63
***
(0.53)
0.30
**
(0.14)
1.99
***
(0.53)
Mortgage rate (lagged) 0.03
(0.02)
0.01
(0.00)
0.03
(0.02)
Inflation excluding rents (lagged) 0.13
(0.17)
0.02
(0.04)
Unemployment rate (lagged) -0.02
(0.01)
-0.00
(0.00)
-0.01
(0.01)
Observations 73 61 72
R
2
/ R
2
adjusted 0.440 / 0.380 0.573 / 0.516 0.560 / 0.520
In Model 3, both rents and wages are deflated using CPI excluding rents
Regional Results
In this section we extend the analysis to consider data for the 16 regional council areas of New
Zealand. Our motivation for investigating regional rent data is two-fold: (i) to increase the amount
of data available for modelling, and hence incorporate additional variation in the data due to
region-specific events; and (ii) to understand differences and commonalities between regions.
A disadvantage of using regional data is the increase in statistical volatility due to smaller quantities
of data in any given region compared with modelling the country as a whole. This ‘noise’ lead us
to focus on modelling the annual change rather than the quarterly change of rents. To avoid
issues with multicollinearity we retained only one observation per year (June).
Simple Panel Regressions, Using Annual Change
A simple model (Table 4: Model R1) with only change in household income and people per
dwelling, explains just over 31% of the variation in regional rents. Splitting people per dwelling into
separate coefficients (people; dwellings) allows us to see the modelled impacts separately (Model
R2), and examination of the statistical significance of the variables shows the relative importance of
the explanatory variables, in order of priority: income, population, dwellings.
16
16 What Drives Rents in New Zealand? National and Regional Analysis
Household disposable income is a marginally stronger predictor than personal earnings (Model
R2: R-squared 29.2%; compared with Model R3: 31.6%). This makes intuitive sense since household
income represents the resources of that households collectively have available for rent and other
expenditure.
Including some national-level explanatory variables, CPI inflation and mortgage rates (average 2-
year fixed special rates), we observe that price inflation looks to already be captured in (nominal)
household income growth. Mortgage rates are significant (Models R4 & R5) and improve the R-
squared to nearly 35%.
Table 4 – Regional panel regressions (annual change)
Model R1 Model R2 Model R3 Model R4 Model R5 Model R6
Predictors
Estimates
Estimates
Estimates
Estimates
Estimates
Estimates
Household income
1.25
***
(0.12)
1.30
***
(0.14)
1.19
***
(0.14)
1.21
***
(0.14)
1.00
***
(0.13)
People per dwelling 0.80
***
(0.23)
Population
0.74
**
(0.24)
1.10
***
(0.25)
1.08
***
(0.26)
1.02
***
(0.25)
0.54
*
(0.24)
Dwellings
-1.15
*
(0.52)
-0.79
(0.51)
-1.44
**
(0.52)
-1.38
**
(0.51)
-2.92
***
(0.50)
Median earnings
1.43
***
(0.16)
CPI
0.10
(0.15)
Mortgage rate (2-
year fixed)
0.02
**
(0.01)
0.02
***
(0.01)
0.02
**
(0.01)
Period 2009-15
-0.03
***
(0.01)
Period post-2016
0.01
(0.00)
Observations 288 288 288 288 288 288
R2 / R2 adjusted 0.314 /
0.271
0.316 /
0.270
0.292 /
0.245
0.349 /
0.301
0.348 /
0.302
0.492 /
0.452
Regional dummies and intercept not shown.
17
17 What Drives Rents in New Zealand? National and Regional Analysis
Including Period Dummy Variables
Exploratory data analysis found differences in the relationship between rents and other variables
over time. Dummy variables for three periods of similar duration, but differing rates of population
growth (pre-2009, 2009-15, post-2016),10 are found to be significant and increase the overall
explanatory power (Table 4: Model R6). This suggests compounding factors may be at play during
sustained periods of imbalance between the supply and demand of rental properties. Visual
inspection of the long run relationship between people per dwelling and rents also illuminates
variation in the strength of the relationships over time (Fig. 6).
Figure 6 – Temporal and spatial differences in people-per-dwelling and rent growth correlations
____________
10 Period mean annual population growths: 2002-08 (‘pre-2009’): 1.3%; 2009-15: 1.1%; 2016-2022 (‘post-2016’): 1.5%
18
18 What Drives Rents in New Zealand? National and Regional Analysis
Including Regional Interactions and Pooled Regressions
The models so far have included regional dummy variables, which creates panel fixed-effects: that
is, the model allows for an overall higher or lower RPI growth for each region, as a constant factor
across all time periods. Alternatively, by including interactions of the regional dummies with other
variables (Table 5: Model R7) the model allows for the estimation of region-specific coefficients for
other explanatory variables. Conceptually, this is equivalent to running separate regression models
for each region. Such models may be most appropriate if we are interested in the best coefficient
estimates for a particular region.
Conversely, if we are attempting to understand the general relationship between the factors in the
model and RPI growth we can remove the regional dummy variables and run a pooled regression
(Model R8). In this case we are using the regional data to increase the number of data points
available for modelling, without a need to understand the impact region-specific effects. We found
that the pooled regression coefficients are similar to those in the panel regressions suggesting the
key drivers are robust to local circumstances in different regions.
Table 5 – Regional interactions and pooled regressions (annual change)
Model R5
Model R7 (interactions)
Model R8 (pooled)
Predictors Estimates Estimates Estimates
Household income 1.21
***
(0.14)
0.55
(0.76)
1.26
***
(0.13)
Population 1.02
***
(0.25)
0.15
(0.91)
1.00
***
(0.23)
Dwellings
-1.38
**
(0.51)
-2.13
(1.32)
-1.92
***
(0.41)
Mortgage rate (2-year fixed) 0.02
***
(0.01)
0.02
***
(0.01)
0.02
***
(0.01)
Observations 288 288 288
R2 / R2 adjusted 0.348 / 0.302 0.488 / 0.342 0.320 / 0.311
Regional dummies and intercept not shown.
Including regional interactions markedly improves model fit (R-squared increases from 34.8% to
48.8%), suggesting a variety of unobserved region-specific factors affect the magnitude of impacts
on rent inflation (Table 5; Fig. 7). These differences may include local planning restrictions,
infrastructure, and capacity of the local construction industry to respond to changes in demand.
Including house prices as an explanatory variable (Appendix Table B1) is significant and has a
positive relationship. This could indicate that many of the unobserved region-specific factors act
on the housing market as a whole, rather than just the rental market. It may also reflect less
incentive to make properties available for rent in a rising market. Conversely, in a falling market
time to sell would be expected to increase which may lead to an increase in the rental stock, and
therefore downward pressure on prices.
19
19 What Drives Rents in New Zealand? National and Regional Analysis
Figure 7 – Model fit, with and without regional interactions
In Model R2 we found evidence that contemporaneous population changes have a stronger
impact on rents than contemporaneous dwelling changes. Investigating the impact of including
lagged variables (Appendix Table B2) we found dwelling changes also impact rents with one– and
two–year lags. This supports a view that a capacity-constrained construction industry takes time to
respond to very strong population growth.
As a robustness check on the annual change rates used in the regional regressions reported, we
investigated using rates of change over longer time horizons. The models were found to fit the
data better over longer time horizons (Appendix Table B3). This may be due to (i) less noise in the
series, and/or (ii) that the explanatory variables affect rents over a longer time horizon.
20
20 What Drives Rents in New Zealand? National and Regional Analysis
Conclusion
Recent data innovations have facilitated the empirical research presented in this paper. We have
focused our study on rent growth for new tenancies (a leading indicator of inflation for all
tenancies). A new tenancy (‘flow’) Rental Price Index (RPI) has been produced monthly for all of
New Zealand, since February 2019 (Stats NZ, 2019). The new index methodology, designed for use
with granular administrative rent microdata, has also enable the construction of region-level RPIs
underpinning our regional analysis.
The primary finding from our study is that income growth and relative supply and demand of
dwellings have been the key drivers of rents in New Zealand over the past 20 years. The impact of
mortgage rates has been smaller, and not statistically significant at the national level. Mortgage
rates were statistically significant in the regional regressions, but with small coefficients. These
empirical findings are consistent with the Housing Technical Working Group’s (HTWG, 2022)
conclusion that restrictions on the supply of land for urban use mean that financial factors, such as
interest rates, will have a greater impact on house prices than rents. These results provide evidence
to support another Housing Technical Working Group conclusion that the New Zealand land
market in aggregate lies somewhere on a spectrum between completely abundant and completely
restricted supply, with variation in the degree of restrictiveness across the regions. In theory, in a
completely restricted land supply, interest rate changes should have no impact on rents.
This research is part of a broader programme of work being undertaken by the Housing Technical
Working Group, which includes: developing a suite of direct indicators to help quantify the
magnitude of land supply restrictions in New Zealand; and developing a better understanding of
the implications of New Zealand’s tax system on the housing market.
We expect that the analysis presented in this paper will be a useful starting point for a variety of
housing market related policy applications, such as: understanding the likely impact of changes in
incomes, dwelling supply, and/or population growth on rents; and to build macroeconomic
forecasts of rents (and by implication aid house price forecasts). Further research could include a
look at the drivers of rents at the lower end of the market and the relationship with measures of
unmet housing need. Low-income households often have little discretion over their level of
housing expenditure meaning rent is often the first call on income. In this light, we hope improving
our understanding of factors that drive rents will be useful for developing public policy.
21
21 What Drives Rents in New Zealand? National and Regional Analysis
References
Bentley, A. (2021). Sticky Rents and the Affordability of Rentals for Housing in New Zealand. New
Zealand Population Review, 47, pp.145-170. Population Association of New Zealand.
Bentley, A. (2022a). Rentals for housing: A property fixed-effects estimator of inflation from
administrative data. Journal of Official Statistics, 38(1), pp.187-211. https://doi.org/10.2478/jos-2022
-
0009
Bentley, A. (2022b). Timely Rental Price Indices for thin markets: Revisiting a chained property
fixed-effects estimator. Paper presented at the 17
th
meeting of the Ottawa Group on Price Indices,
Rome, 7-10 June 2022. Available at www.ottawagroup.org
Coleman, A and Scobie, G (2009) “A Simple Model of Housing Rental and Ownership with Policy
Simulations”, Working Paper 09/05, New Zealand Treasury
Dias, D. and Duarte, J (2019) “Monetary Policy, Housing Rents, and Inflation Dynamics”,
International Finance Discussion Paper Number 1248, Board of Governors of the Federal Reserve
System
Duca, J, Muellbauer, J and Murphy, A (2016) “How Mortgage Finance Reform Could Affect
Housing”, American Economic Review: Papers and Proceedings, 106(5), pp.620-624
Duca, J, Muellbauer, J and Murphy, A (2021) “What Drives House Price Cycles? International
Experience and Policy Issues”, Journal of Economic Literature, 59(3), pp.773-864
Housing Technical Working Group (2022). Assessment of the Housing System: with insights from
the Hamilton-Waikato Area. August 2022. Available at
Assessment of the Housing System: with
insights from the Hamilton-Waikato Area (treasury.govt.nz)
Howard, G and Liebersohn, J (2021) “Why is the rent so darn high? The role of growing demand to
live in housing-supply-inelastic cities”, Journal of Urban Economics, Vol.124 pp.1-14
Ministry of Housing and Urban Development (2022). Change in housing affordability indicators:
Concepts, sources, and methods. November 2022. Available at www.hud.govt.nz
New Zealand Treasury (2022). Trends in Wellbeing in Aotearoa New Zealand: 2000-2020.
Background Paper for the 2022 Wellbeing Report. April 2022. Available from
https://www.treasury.govt.nz
Nunns, P. (2019) “The Causes and Economic Consequences of Rising Regional House Prices in New
Zealand”, Working Paper No.005, Centre for Applied Research in Economics, University of
Auckland
Reserve Bank of New Zealand (2023a). New residential mortgage standard interest rates (B20).
May 2023. Available at www.rbnz.govt.nz
Reserve Bank of New Zealand (2023b). New residential mortgage special interest rates (B21). May
2023. Available at www.rbnz.govt.nz
22
22 What Drives Rents in New Zealand? National and Regional Analysis
Saunders, T and Tulip, P (2019) “A Model of the Australian Housing Market”, Research Discussion
Paper RDP 2019-01, Reserve Bank of Australia
Stats NZ (2019). New methodology for rental prices in the CPI. February 2019. Available at
www.stats.govt.nz
Stats NZ (2020). Housing in Aotearoa: 2020. December 2020. Available at www.stats.govt.nz
Stats NZ (2021). User guide for Stats NZ's employment measures. March 2021. Available at
www.stats.govt.nz
Stats NZ (2022a). National population estimates: At 30 June 2022. August 2022. Available at
www.stats.govt.nz
Stats NZ (2022b). Subnational population estimates: At 30 June 2022 (provisional). October 2022.
Available at www.stats.govt.nz
Watson, E (2013) “A Closer Look at some of the Supply and Demand Factors Influencing
Residential Property Markets”, Analytical Note 2013/11, Reserve Bank of New Zealand
23
23 What Drives Rents in New Zealand? National and Regional Analysis
Appendix – Additional Regression Results
National Results
Table A.1: Results using annual change
Annual rent inflation (nominal, flow)
Predictors (1)
Rent inflation (t-4) -0.54
***
(0.14)
Wage
0.62
***
(0.10)
Wage (t-4) 0.75
***
(0.14)
People per dwelling (t-4) 1.18
***
(0.38)
Floating mortgage rate (t-4)
0.01
(0.02)
Inflation excluding rents (t-4) -0.37
***
(0.12)
Unemployment rate (t-4) -0.03
**
(0.01)
Observations 67
R
2
/ R
2
adjusted 0.508 / 0.449
Table A.2: Interaction between wage growth and relative supply and demand
Quarterly rent inflation (nominal, flow)
Predictors (1)
Wage (rolling two-period average) 0.98
***
(0.15)
People per dwelling (lagged) 1.57
***
(0.54)
Wage x People per dwelling
0.04
(0.62)
Observations 73
R
2
/ R
2
adjusted 0.430 / 0.368
All variables enter the model as percentage change from the previous quarter.
* p<0.1 ** p<0.05 *** p<0.01
24
24 What Drives Rents in New Zealand? National and Regional Analysis
Regional Results
Table B1 - Panel regressions, including house prices (annual change)
Model R5 Model R9 Model R10 Model R11 Model R12
Predictors Estimates Estimates Estimates Estimates Estimates
Household income 1.21
***
(0.14)
1.00
***
(0.13)
0.94
***
(0.13)
0.85
***
(0.13)
0.82
***
(0.13)
Population
1.02
***
(0.25)
0.54
*
(0.24)
0.53
*
(0.21)
0.81
***
(0.22)
0.57
*
(0.23)
Dwellings
-1.38
**
(0.51)
-2.92
***
(0.50)
-1.21
**
(0.46)
-1.44
**
(0.45)
-2.50
***
(0.48)
Mortgage rate (2-year
fixed)
0.02
***
(0.01)
0.02
**
(0.01)
0.02
***
(0.01)
0.02
***
(0.01)
Period 2009-15
-0.03
***
(0.01)
-0.02
***
(0.01)
Period post-2016
0.01
(0.00)
0.01
(0.00)
House Price Index
0.15
***
(0.02)
0.15
***
(0.02)
0.10
***
(0.02)
Observations 288 288 288 288 288
R2 / R2 adjusted 0.348 /
0.302
0.492 /
0.452
0.456 /
0.417
0.489 /
0.451
0.542 /
0.504
Regional dummies and intercept not shown.
25
25 What Drives Rents in New Zealand? National and Regional Analysis
Table B2 - Panel regressions with lagged variables (annual change)
Model
R5
Model
R13
Model
R14
Model
R15
Model
R16
Model
R17
Model
R18
Predictors Estimates Estimates Estimates Estimates Estimates Estimates Estimates
Household income 1.21
***
(0.14)
1.11
***
(0.15)
1.13
***
(0.14)
1.26
***
(0.13)
0.86
***
(0.13)
Population
1.02
***
(0.25)
0.98
**
(0.30)
0.97
***
(0.24)
0.84
***
(0.23)
0.84
***
(0.22)
Dwellings -1.38
**
(0.51)
-1.71
**
(0.56)
-1.64
**
(0.53)
-1.68
**
(0.52)
-1.62
***
(0.47)
Mortgage rate (2-
year fixed)
0.02
***
(0.01)
0.02
(0.02)
0.02
***
(0.01)
0.02
***
(0.01)
Household income
(1-year lagged)
0.27
(0.15)
0.08
(0.15)
Population (1-year
lagged)
-0.10
(0.27)
0.03
(0.31)
Dwellings (1-year
lagged)
1.87
**
(0.57)
1.06
(0.54)
1.17
*
(0.49)
Mortgage rate (1-
year lagged)
0.02
**
(0.01)
-0.00
(0.02)
Household income
(2-year lagged)
0.21
(0.15)
Population (2-year
lagged)
0.15
(0.28)
Dwellings (2-year
lagged)
1.45
*
(0.58)
Mortgage rate (2-
year lagged)
0.02
**
(0.01)
Rental Price Index
(1-year lagged)
0.24
***
(0.05)
0.11
*
(0.05)
House Price Index
0.14
***
(0.02)
Observations 288 287 286 287 287 287 287
R2 / R2 adjusted 0.348 /
0.302
0.188 /
0.130
0.155 /
0.094
0.363 /
0.308
0.362 /
0.315
0.367 /
0.322
0.498 /
0.458
Regional dummies and intercept not shown.
26
26 What Drives Rents in New Zealand? National and Regional Analysis
Table B3 - Panel regressions with longer time horizons
1 year 2 year 3 year 4 year
Predictors Estimates Estimates Estimates Estimates
Household income 1.21
***
(0.14)
1.66
***
(0.18)
1.56
***
(0.23)
1.48
***
(0.26)
Population 1.02
***
(0.25)
0.84
***
(0.30)
1.25
***
(0.40)
1.34
***
(0.47)
Dwellings -1.38
***
(0.51)
-1.43
**
(0.65)
-1.01
(0.78)
-1.07
(1.08)
Observations 288 144 96 64
R
2
/ R
2
adjusted 0.348 / 0.302 0.489 / 0.415 0.517 / 0.404 0.559 / 0.383
Regional dummies and intercept not shown