Vacant Dwelling Indicators based on Metered Electricity Consumption 2021

Methodology

This publication provides indicators of residential vacancy using electricity consumption data. For the purposes of these vacancy indicators, dwellings were considered to be vacant where - for an individual quarter - they had less than 180kwh of consumption recorded in data provided by ESB Networks, for at least four consecutive quarters (for the reference quarter and at least three consecutive prior quarters). 

The denominator for all vacancy rates included in this publication is the number of unique ESB Networks connections where there is a consumption figure recorded (including where this is zero) for the relevant quarter. Vacancy rates could change over time as dwellings have low consumption for a fourth quarter or have consumption above the threshold - a change in the numerator. They can also change as the denominator increases due to new dwellings or re-connections of previously terminated or deactivated connections; or decreases with dwellings going into dereliction. Buildings converted from commercial to residential and vice versa would affect this also. 

Analysis and linkage in this publication were undertaken by the CSO for statistical purposes in line with the Statistics Act, 1993 and the CSO Data Protocol.

Before using personal administrative data for statistical purposes, the CSO removes all identifying personal information, including the Eircode and Meter Point Reference Number (MPRN), and creates pseudonymised Protected Identifier Keys (PIKs). The PIK is a unique and non-identifiable number which is internal to the CSO. Using the PIKs enables the CSO to link and analyse data for statistical purposes, while protecting the security and confidentiality of the individual data. All records in the matched datasets are pseudonymised and the results are in the form of statistical aggregates which do not identify any individuals.

Rounding on Graphs and Tables

In some instances, numbers in the graphs and tables which form proportional distributions have been adjusted to ensure that the sum of the figures when rounded to one decimal place still equals 100.

Data Sources and Linkages

ESB Networks Consumption Data

Data have been provided to the CSO by ESB networks detailing electricity customers and quarterly consumption. For this analysis only customers in the tariff categories DG1 and DG2 have been examined. DG1 is urban domestic and DG2 is rural domestic. A small number of records have been removed for the purpose of the analysis here where there addresses indicate that they are not for residential use. These are mainly farm buildings.

Electricity consumption is measured in Kilowatt Hours (kWh). A kilowatt hour is a unit of energy equivalent to one kilowatt of power sustained over an hour.

Meter readings data were converted to quarterly consumption estimates by ESB Networks. Meter readings can be actual or estimated. Actual readings can be from a meter reader or from the customer. The data file provided to the CSO did not provide information on whether the consumption in a period was based on an actual or estimated reading. 

Each electricity connection has a unique MPRN. Each unique MPRN is considered a unique dwelling for the purpose of this analysis.

Building Energy Rating (BER) Assessments Data compiled by the Sustainable Energy Authority Ireland (SEAI)

These data include assessments done for the Building Energy Rating (BER). The data available to the CSO and used in this analysis includes assessments from 2009 to 2021.  Buildings are rated from A to G where A-rated homes are the most energy efficient.

The BER dataset contains the same MPRN identifier as in the ESB Networks consumption data and thus allows for accurate linkage of the datasets. Not all ESB networks connections will have had BER assessments as these are predominately necessary just for sales and lettings. Therefore, it is known that the analysis by BER rating, dwelling type and period of construction does not show a full representation of the total dwelling stock as certain types of dwellings are more likely to have had BER assessments because of being sold and rented out in recent years, including for new builds which are likely to be overrepresented in the matched dataset.

Table 3.1 details match rates between ESB Networks connections and BER assessments. Where a dwelling has more than one BER assessment is it matched to the most recent assessment. Matches are only made to BER assessments made in years before and including the year of consumption that is being examined.

Local Property Tax (LPT)

This dataset details local property tax (LPT) from all active residential properties on 1^st November 2021. This is used within this analysis to determine ownership type for breakdown of vacant and all dwellings as shown in Table 3.2. The LPT has been linked to the ESB Networks consumption data via pseudonymised Protected Identifier Key (PIK) based on the Eircode of properties. The LPT dataset received from Revenue has Eircodes provided for 88.2% of records. The ESB Networks data has Eircode provided for 64.3% of records. Where Eircodes were not provided, the CSO has assigned Eircodes through geocoding processes. Properties whereby the Eircode has not been able to be assigned with sufficient accuracy or where there is the same Eircode for multiple dwellings have not been matched. For all dwellings in Q4 2021, 63.3% have been matched between ESB Networks data and LPT and so are included within the ownership analysis. It can be harder to accurately assign Eircodes where they have non-unique addresses, which mainly occurs in rural areas. This means that the matched population isn’t fully representative of the total housing stock.

Census of Population Analysis (COPA)

The COPA is a pseudonymised copy of the Census of Population 2016 dataset held internally within the CSO for analysis purposes. It contains Census attribute information for individuals and households. This dataset has a pseudonymised PIK version of Eircode allowing linkage with this variable from the ESB Networks consumption data. Linkages were made for active connections in Q4 2021 where the ESB Networks Eircode was geocoded with sufficient accuracy and for records which had been assigned a unique Eircode. Further to this, matches were only made for dwellings which were occupied in the 2016 Census. Dwellings built after 2016 Census would clearly not be able to be matched. Just under a half - 49.6% - of records in Q4 2021 could be matched. 

Definitions

Urban-Rural

The Urban-Rural classification used within this publication relates to the tariff category (DUoS) given by ESB Networks. DG1 is urban domestic and DG2 is rural domestic. 

The DUoS is determined by factors such as the type of use, maximum import capacity (MIC), and the nature of the connecting network. A domestic customer premises is defined as any premises whose main purpose is that of a single residence and where the connection agreement is with a private individual with an MIC up to 29,000 voltamperes.

Urban domestic connections are defined as domestic connections that are fed from three-phase overhead or underground low voltage network. Rural domestic connections are defined as domestic connections that are fed from single-phase overhead network. 

Geography

All geographical analysis is based on latest available boundaries. This includes analysis related to time periods prior to boundary changes. For example, Local Authority and Local Electoral Area (LEA) level analysis in Cork is based on property location in relation to the new Cork City boundaries after changes in May 2019. Local Authorities, LEAs, and Electoral Divisions (EDs) are assigned through geocoding of addresses undertaken within the CSO (including using Eircodes provided within the ESB Networks dataset). It was not possible to assign these geographies to all addresses with sufficient accuracy. These addresses are not included within the relevant geographical analysis. 

Comparisons with Census Vacancy Rates

Direct comparisons between vacancy rates obtained from the Census of Population and within this publication are not possible due to definitional differences. The Census classes dwellings as vacant when they are unoccupied as of Census night, are not used as a holiday home and are not usually inhabited by occupants who are temporarily absent at the time of Census. This is a point in time measure, while vacancy as indicated in this publication is based on at least four quarters of low electricity consumption. Although the Census vacancy measure includes dwellings that have been vacant for longer periods of time, it also includes dwellings which are unoccupied for potentially very short periods such as between rental tenancies.

Although still based off different definitions, it is possible to compare percentage of dwellings vacant at both the 2016 and 2022 Censuses and comparable timeframes from the methodology used within this publication. From this methodology it is also possible to detail the percentage of dwellings vacant from Q2 2016 through Q4 2021 - where a dwelling was vacant by definitions used here in Q2 2016 and had less than 180kwh quarterly consumption all the way through to Q4 2021. See Table 3.3.