Opportunities

Understand NDIS demand in your area

The NDIS Demand Map is a data analytics tool that provides information to help providers to grow in the NDIS.

Use the NDIS Demand Map to help understand the services and locations NDIS participants may choose by 2025.

What is the NDIS Demand Map?

The NDIS Demand Map provides an up to date forecast of NDIS demand by postcode across Australia. The forecast estimates the expected demand when participants have minimal barriers to accessing services.

Use the map to find out:

  • How many NDIS participants are expected to live in a postcode.
  • How much participants are expected to spend to meet their needs, and on what types of support.
  • How many workers may be required to meet participant needs and preferences.

What are the forecasts based on?

The forecasts use de-identified NDIS and the Australian Bureau of Statistics' 2021 census data to predict what the NDIS will look like by 2025.

The forecast will be updated regularly based on new NDIS data as it becomes available.

It is important to note that the forecast is based on participant spending patterns to date, in particular the spending patterns of experienced participants in developed markets. This pattern may change in the future if participant's preferences change.

Demand map forecasts

How should you use the forecasts?

The Demand Map provides a forecast range for future participant numbers, spending and workforce. This range reflects the inevitable uncertainty in making forecasts, particularly in areas where the NDIS has not been operating for long or areas with low populations. We estimate that the actual value will fall somewhere within the forecasted range 90% of the time.

Demand map screenshot 1

The forecast and current demand for neighbouring postcodes can be combined using the 'Select any bordering postcodes you are interested in' feature. This is particularly useful when the ranges are large due to low populations in the desired postcode. However, users should not attempt to independently sum results across postcodes without using this feature. The ranges for individual postcodes will not sum to the actual estimated range as the forecasts are not independent of one another.

We expect 90% of postcodes to fall within the reported ranges; but not necessarily the midpoint of the range.

Website users should not sum categories or sum totals across postcodes, as this will produce inaccurate results.

Users should make their own assessments when using this data.

Tell us your views

We welcome any feedback you have on how we can improve the Demand Map. Please send your views through our feedback form.

Important information in interpreting forecast ranges

In interpreting forecast ranges, Demand Map users should note:

Forecast certainty

  • These ranges are forecasts for the scheme in June 2025. The ranges are not a measure of actual participants in the scheme currently.
  • Forecasts, by their nature, have an element of uncertainty. The ranges provided estimate the uncertainty about the participants that will access the scheme, what their needs are, where they will live and what they will choose to spend on what types of support and services. This uncertainty is particularly prevalent in less populated areas, or in areas where the NDIS is has not been operating for long.
  • Forecasts do not take account of unforeseen changes (i.e., changes in demand due to implementation of new operational processes by the NDIA or legislative reform) and assume that the current data generating process continues.
  • We estimate that the actual value will fall within the forecasted range 90% of the time; but not necessarily the midpoint of the range. This means we estimate that, 10% of the time, the actual number of participants, participant spending and workforce will be different to what is predicted. It is not possible to say with certainty which postcodes will be different to what is predicted.
  • However, not every source of uncertainty can be accounted for, which may result in the estimated prediction intervals being too narrow.
  • The model accounts for the different spending patterns of participants in accessing Supported Independent Living (SIL). Due to the sometimes-sparse number of SIL participants in any given postcode, there is considerable uncertainty in the SIL participant forecasts. As a result, the 'Daily Living Support in Shared Accommodation' and 'Home-base Support Workers' categories may have considerably wider forecast ranges than other spend or worker categories.

Low population areas

  • For regions with low populations, caution should be taken when interpreting the ranges. Where the number of participants is expected to be low, an incorrect prediction could be the difference between no spend in a postcode and substantial spend in a postcode. The large range of possible values is a reflection of the difficulty in providing an accurate forecast for these postcode. In these cases, confidence could be increased by combining neighbouring postcodes.
  • For postcodes with very low resident populations (less than 100 people recorded in the 2021 Census) or small postcodes where the forecast is not reliable (for example the participant forecast exceeds 20 per cent of the Census population), forecast ranges are not provided at this stage.
  • To ensure privacy, forecast ranges will not be provided for postcodes with participant and workforce populations below 10.

Adding postcodes

  • The ranges for individual postcodes will not sum to the estimated ranges if forecast ranges are summed between multiple postcodes by the user as the forecasts are not independent of one another.
  • As a result, website users should not independently sum categories or sum totals across postcodes themselves, as this will produce inaccurate results.
  • The forecast and current demand for neighbouring postcodes can be combined using the 'Select any bordering postcodes you are interested in' feature.

What is and isn't captured in the forecasts

  • Current in-kind spend is no longer displayed in separate spend categories as the proportion of in-kind supports have significantly reduced. This also allows us to better highlight the type of support being received.
  • Disabilities Services Officers now fall into the Support Coordinators category as the two categories are no longer distinguishable. Improvements in therapy data have allowed Physiotherapy to be placed into its own category.

Workforce estimates

  • Currently no actual data on the number of workers in the NDIS is available to inform a forecast on the workforce that will be needed in the future. To produce this forecast, the model estimates the workforce through analysis of participant spending, using assumptions on the share of NDIS payments paid as labour costs. These assumptions are based on stakeholder interviews.
  • Workforce estimates are provided for a number of occupation groups. Some occupations with low forecast numbers and/or more uncertainty about the specific occupation have been grouped together into "other". Other may include occupations such as nurses, social workers, podiatrists, human resource managers, nutrition professionals, manufacturers and fitness instructors.

Participant spending categories

  • Forecasts of participant spending are broken down across 12 categories. These categories group the NDIS registration groups for simplicity and modelling robustness. Details on which registration group maps to which website category can be found in Appendix A.

Other

  • The forecast utilises the most recent postcode data provided by Australia Post. For some postcodes that are situated on state borders, the Australian Post dataset does not recognise postcodes directly bordering in adjoining states. In these instances you will need to directly click on the postcode required to see NDIS demand data.

Technical information

Forecasted Demand

The Demand Map uses a 'Deep Autoregressive Neural Network' (DeepAR) machine-learning technique to forecast NDIS participation, and a 'Multivariate Regression' (MR) to forecast the expected annual spend and required workforce to deliver these services. The model draws on census data and NDIS administrative data from over 2,500 postcodes in Australia to project the number of participants in each postcode. The expected value of spend per person for different categories is estimated for these participants using the MR model. Only summary statistics at a postcode level are used in modelling, such that individual data observations are not linked across sources. Job estimates are inferred from a projected volume of services to be delivered to these participants.

The DeepAR model is a global forecasting method. It leverages the historical change in participant numbers as well as census data to predict the future number of participants in a postcode. The ranges produced for the model are directly calculated as part of the modelling procedure, which directly estimates the models variance.

The MR spend model estimates the relationship between postcode participant characteristics, other demographic characteristics and spend per person, for each spend category. The model is then used to predict the average predicted spend necessary to meet the demand of a mature participant in a mature market. The MR model allows for correlation in errors for different categories of spend to ensure more robust modelling. A similar procedure is used to estimate the expected workforce.

To ensure robustness, modelling is evaluated out of sample through step ahead cross-validation. This procedure trains the model on past values, and evaluates performance against the remaining subsample. The average performance of the model is compared to benchmark forecasts and the range of the forecast is evaluated to test whether the 90% of values are falling within the predicted range.

Current Demand

Current demand is defined as the demand for NDIS services over the 12 months period from 1 March 2022 to 28 February 2023, including: the total spend on NDIS services provided during the specified period by postcode, and the number of participants with active plans by postcode at the end of the period (28 February 2023).

There is a toggle made available on the Demand Map to allow users to switch between current and forecast demand figures.

To protect privacy, current demand figures are not available for postcodes with very low resident populations (less than 100 people as per the 2021 census) or when less than 10 NDIS participants reside in the postcode.

Disclaimer

These forecasts have been produced by the Department of Social Services. Every effort has been made to provide the most current, correct and clearly expressed information possible on this site. Nonetheless, inadvertent errors can occur and applicable laws, rules and regulations may change. The information contained on this site is general and is not intended to serve as professional advice. No warranty is given in relation to the accuracy or reliability of any information. Users should not act or fail to act on the basis of information contained herein. Users should not rely on the information for any business, commercial or other purpose, and are strongly encouraged to seek professional advice concerning the information provided on this site before making any decision. Users should not rely on sum categories or sum totals across postcodes, as this will produce inaccurate results. Users should not use this tool for the purpose of re-identification. All contributors to this site disclaim all and any liability to any person or organisation in respect of anything, or in consequence of anything, done or omitted to be done by any person, organisation or other user in reliance, whether in whole or in part, upon any information contained herein.

No part of the text or graphics on this site may be reproduced or transmitted in any form or by any means, electronic or mechanical, including by photocopying, facsimile transmission, recording, re-keying or using any information, storage and retrieval system. Certain links on the site lead to resources located on servers maintained by third parties. As such, no representations are made as to the accuracy, currency or any other aspect of the information contained on such servers or the timely, accurate or complete transmittal of such information.