Unique offerings to meet your needs
It’s all about how we can help you.
Spectrum Analysis Specialist Consulting Services
Consulting services, ABS Census analysis and projections, NZ Census and analysis, media requests and referrals.
Are you not quite sure exactly what sort of data, mapping or geographic modelling, information or reporting you need?
We are here to help.
We have vast experience in:
- all aspects of retail property
- corporate, retail or franchise network planning
- defining territories, exclusion zones and marketing areas
- site evaluation tools, surveys and reports
- enrolment and marketing strategies for schools, colleges and universities
- clubs and member associations reports
- Geographic Information Systems (GIS) , geopsatial analysis and data visualisation
- lease negotiation and management (in collaboration with our partnering businesses)
Australian Bureau of Statistics (ABS) Census Analysis and Projections
Are you interested in analysing information that has been collected by the Australian Bureau of Statistics (ABS) Census conducted every five years or calculating some projections?
The first release of census data is typically scheduled six months after the census night and the second release of census data is typically scheduled one year after the census night. Population figures and household counts can change significantly in five years in growth corridors of metropolitan and regional areas and gaining early access to these locations can be very lucrative for securing cheaper sites or for significantly improving strategic planning decisions.
Because population counts data is not available between census collection periods, we can produce Small Area Population and Household Projections.
The ABS produces its household projections, by classifying households into four types according to their structure including:
- separate house
- semi-detached, row or terrace house or townhouse
- flat, unit or apartment
- and other
Building permit data ignores the ‘Other’ type, but this is typically a very small proportion of the households in any SA1 (~1% on average), so the omission is not considered likely to have a material effect on accuracy.
The use of building permits to increase household counts has three flaws:
- it ignores when a single dwelling is demolished and replaced with multiple dwellings
- it cannot detect if an area has a net decrease in
households (which would be a very rare but possible event)
- it assumes that the dwelling is completed in the same period as the permit was issued
The first two of these flaws are likely to have only a small effect, but when we do our analysis, this effect is investigated and, if necessary, corrected through a reality check against the population estimates with respect to expected persons-per-household ratios.
The third issue is a matter of timing and may in some cases cause the estimates to be ‘early’ in their projection of household figures, but since the estimates themselves lag by at least six months in most of these cases reality will have caught up with the projection.
Household projections for each post-census year are projected using the formula:
Household Estimate = Base Household Count + Number of Housing Approvals by Housing Structure Type Classification
Further details about the use of the formula follow:
- for producing the small area household projections for June after the census, the previous census small area household counts
are used as the starting point. For the rest of the
years, projected small area household counts are used as the starting point
- data for two housing structure type classifications are added to the base household counts, these housing structure types are: separate houses and other residential (semi-detached, row or terrace houses, townhouses, flats, units or apartments)
- while there is no equivalent for household estimates of the role that the Estimated Resident Population (ERP) plays for the population estimates (ie. providing a benchmark), the household estimate is still ‘reality checked’ against the population estimates to ensure that the data flaws noted in the Assumptions section above have not created
significant errors. The reality check involves testing that the persons-per-household ratios using the updated estimates do not stray far from the ratios according to the census. A correlation analysis confirms that this is the case
The ABS produces its population projections by studying the demographic trends of the past few years in both Australia and overseas and through consultation with various other parties.
At the highest level, the procedure to project the population figures from one Census year through to each post-census year is to first obtain an ‘un-calibrated’ estimate as detailed below, then calibrate that figure against the ABS’s Estimated Resident Population (ERP) (where possible), which is published for every Local Government Area (LGA) in Australia.
This calibration ensures that the sum total of population counts of all the small areas (SA1) within a large area (LGA) match with the ABS projections, and thus provide a certain level of confidence in the accuracy of projections.
Un-calibrated small area population for each post-census year have been projected using the formula:
Un-calibrated Small Area Population Estimate = Base Population + Births – Deaths + ∑(Average Number of People Belonging to a Housing Structure Type Classification)i* (Number of Housing Approvals by Housing Structure Type Classification)i
Further details about the use of the formula follow:
- for producing the un-calibrated small area population projections for the June after the census, the census small area population counts are used as the base population
- for the rest of the years, calibrated small area population count projections are used as the base population
- the birth rates and death rates are published by the ABS at large area levels. For each small area, the number of births and deaths are calculated using the birth and death rate of the large area they belong to and applied on the base population. Birth and death rates are not available yet for the current year and estimated birth and death rates (based on the birth and death rates of the past years) are used as substitutes
- ‘i’ takes the values 1 to 2. The ABS classifies households as three types (as above, excluding other). However, the building approval
data from the ABS combines the latter two categories into one category. Therefore, these two categories are used in all analysis. The average numbers of people belonging to each housing structure type are calculated for each large area using the Expanded Community Profile (ECP) from the census year. National median of the average number of people belonging to a certain housing structure type aree substituted in instances where a certain large area happened to be missing one or more types of households. For each small area, the average numbers of people belonging to a certain housing structure type are drawn from the average number of people belonging to a certain type in the larger area which contains the smaller area
- the ABS provides building approval data by housing structure type at the small area (SA1) level. The product of the average number of people belonging to each housing structure type and the new housing of that type that have been approved for an area are calculated for each
housing structure type, and are added to the base population
Our Household and Population Projections
We use a similiar projection process that involves:
- reviewing the birth rate and mortality rate of the area, overseas and interstate migration, the number of building approvals by building type classification in the area and the average number of people residing in each building type for the population change calculation
- the birth rate and the death rate of small areas Statistical Area 1 (SA1) will be the same as the larger area to which they belong (Statistical Local Area (SLA))
- the trend the average number of people associated with each housing structure type will continue for the projection period. For example, if the previous census tells us that 1.5 persons on average live in flats, units or apartments in a certain area, then this trend will continue for the next few years. Also, the average number of people belonging to each housing structure type in a small area SA1 will mirror that for the larger area SLA to which it belongs
- we look at the number of households in an area in the most recent census and then apply an increase (if any) based on the number of building permits granted in each SA1
- use the Population and Household Projections and calibrate them against the ABS’s Estimated Resident Population (ERP) that is published for every Local Government Area (LGA) in Australia. We ensure that the sum total of population counts of the small areas (Statistical Area 1 or SA1) within each LGA match with the ABS projections to confirm the accuracy of projections
Within certain licensing conditions, we can use our six step data analysis process to:
- create population estimate spatial tables for post-census years at the SA1, postcode or suburb level
- create population estimates in Excel (or other data format) tables for post-census years at the SA1, postcode or suburb level
- develop customised maps for an area of interest showing the population change
A typical table would include the name of a suburb, state, current census population figure and population estimate for the month and year you request.
New Zealand Census of Population and Dwellings
We have access to demographic information at various geographical levels starting from areas as small as a Statistical Standard Meshblock, the smallest area used to collect and present statistics which is made of about 60 people in rural areas and 110 in urban areas.
These meshblocks are aggregated to form larger geographical areas such as area units, territorial authorities, regional councils, urban areas, wards and constituencies.
We have access to Stats NZ detailed data on the following subjects:
- population and dwellings counts
- age and sex
- cultural diversity – including ethnic groups, birthplace and languages spoken
- marital status
- work – including unemployment rates and occupation
- families – including types and composition
- postcode boundaries
- street layers
We can provide the facts and data for corporate, retail and franchise networks to develop their strategic plan for expanding into New Zealand in the next two to five years or alternatively, consider relocations and/or closures of under performing sites
We can use our six step data analysis process to:
- create a database of the existing network (if it exists) by surveying all current sites, gather information on internal and external features, visibility of the store, generators (such as chemists, newsagents and supermarkets) and competition in the area (positive and negative) that is approved by you before being conducted by our experienced staff.
- add to this, information from Stats NZ’s Census of Population and Dwellings, Stats NZ’s Longitudinal Business Database and the Property Council of New Zealand – New Zealand Shopping Centre Directory and if available, sales data from the last 12 months to be used as the dependent variable to create a prediction figure at the end of the analysis for each site in the network
- segment sites into homogenous groups to sample those that are common in the network (metropolitan strip is separate to CBD shopping centre)
- analyse key drivers having the greatest influence on store performance and short list the top drivers to analyse further. These variable form the starting point of the modelling process
- build sales prediction models with a statistically significant model using a mathematical equation of the best drivers (usually one driver for each characteristic of internal, generator, competitor and one indicating the size of the shopping precinct). The number of drivers in the model is typically driven by the sample size. Usually around one independent variable for every 8-10 sites with a minimum of 3-5 variables and a maximum of 10-12 variables
- once key variables for a site have been entered, the model returns a prediction of dollar sales for assessing a site’s potential shown graphically in a scatterplot of actual versus predicted
- the final model is built to assess all the stores in the network using your actual saes figures and the predicted sales figures displayed on the scatterplot with each site categorised into four broad categories of poor (tail), good (leave alone), under (operations review) and excellent (star) performer sites with details on why a site is performing well
- conduct pre-entry study with local, national and state level summaries of target customers, growth patterns and maps showcasing demographically rich areas for your products and services
- provide territory planning and territory mapping services based on demographic data
- compare and prioritise ‘New Build’ areas
- prepare site potential reports for new developments or redevelopments
- compare specific area demographics to National or State averages
- study the trend in demographics of interest
- geocode addresses to analyse businesses, trends, penetration and competition
- prepare Target Market Indexing (TMI)
- assist with strategic planning from a spatial viewpoint
- upgrade models when required
- colour code demographic data on maps and specific data packs similar to our other Data Packs with Australian information
- we can also provide marketing, enrolment and strategic planning information for schools, colleges and universities in New Zealand
Our Managing Director, Peter Buckingham is regularly invited to provide conference speaking presentations, specialist workshops, article content for magazines, audio podcast interviews and more.
You can see a range of content here via these links:
Please contact Spectrum Analysis if you would like us to prepare some information for you.