Dialog recently hosted Barry Sandison, Deputy Secretary (Health, Compliance & Information), Commonwealth Department of Human Services (DHS), to reprise a presentation he gave to the Teradata Summit in Sydney in August 2014, for a dozen Queensland Government senior executives from a number of agencies.

Department of Human Services (DHS) has a gigantic database consisting of customer data about every Australian. This data is held within Medicare, Centrelink, and Child Support Agency systems. These systems administer over $150 billion of government outlays in 932 million transactions per year.

Barry outlined the opportunities available from the use of these valuable datasets with a primary focus on Centrelink, and gave examples of how they can be applied to a range of community, welfare and government service delivery functions, and potentially many other applications while ensuring appropriate management of privacy and confidentiality.
DHS has built an integrated enterprise data warehouse containing over 10 years’ history of these datasets. This provides powerful insight into the behaviours, demographics and evolving needs of the Australian population.

While the administrative datasets are fundamental to the integrity of the whole payments process, they also provide a longitudinal story of the physical movements, family changes, welfare needs, employment events, financial well-being and the use of Medicare services for almost every Australian resident.

More importantly they provide a picture of the evolving trends and circumstances of Australian citizens. This ability to understand what trends and changes are happening with the population is an enormously powerful tool for policy-makers, service delivery agents and advocates for government programs and reform. Combined with big data techniques and analytical tools, this data can provide enormous benefit to the broader community.

While the concept of “open data” is still largely new in the Australian environment, the Federal Government in particular is leading the way in sending “signals” to the states and territories that these datasets have the potential to add enormous value to the processes of state and local government and also the general community and business. The drive to have data placed on the open data website data.gov.au is a good example of the change taking place.

By way of comparison, retailers collect massive volumes of data on what we buy in their stores (e.g. FlyBuys), which is further aggregated by specialised data aggregators such as IRi Aztec and Acxiom to provide a “whole of market” view on product usage and very detailed market segments of consumer behaviour, enabling retailers to understand our preferences and likely purchase intentions. The banking industry similarly has massive volumes of data on customer’s financial behaviours and financial means, enabling them to improve services to their customers.

Looking beyond the daily operations of managing DHS payment processes, examples of where these administrative datasets can provide unexpected value include the following:

1. Foster care & child safety.

Analysis of historical data provides insight into how the situation of foster care children evolves over time, even into adulthood with a specific example of being able to identify the extent to which they might rely on welfare when they are older? The longitudinal data provides the opportunity to understand some of the aspects of foster care. While social workers may provide anecdotal evidence, hard data validates the results in support of informed policy making and subsequent funding.

By understanding the factors and conditions that make a difference, the data can be used to allow policy and service delivery experts to pose questions on where possible early intervention might provide real benefits collectively or to individual cases, thereby enabling better future outcomes for children, and identifying where funding can be best targeted for the most vulnerable. 

2. Public housing

Citizens in public housing may be recipients of other welfare. Rental payments are usually based on ability to pay and waiting lists used to allocate housing priorities. These programs ultimately require funding from Government, so full consideration of all factors must be given to ensure tenants not only have appropriate accommodation, but also pay at the rate deemed correct for their circumstances.

As an example, it is a simple matter to determine the average number of tenants at a particular address. Differences of significance between DHS databases and State held databases of public housing can allow further detailed analysis to identify anomalies.

3. Child support agency and debt management

The total debt managed by the Child Support Agency is a significant amount and is essentially monies owing that should have been paid directly by improvident parents for the support of children. The various demographics, ability to pay and patterns of payment can be analysed to establish best approaches to improve conformance to child support payment rules. Work is being undertaken to test the capacity of the system to predict imminent poor payment practices, thereby improving management of the debt portfolio.

4. Service delivery planning

Where and how people interact with the various DHS services (store front, telephone and internet/mobile device), what services they seek, and appreciating where current and forecasted needs exist, allows government agencies to deliver their services most effectively where they are needed. The sharing of data between agencies with common participants between agencies allows the range of government services available to be managed better.

5. Crisis management

Emergencies and community crises such as bushfire and flood often require emergency payments to be allocated appropriately. The DHS datasets provide the means to determine quickly who may require such payments and to establish the veracity of claims.

6. Compliance

Effective management of the massive Government welfare and health outlays, ensuring they are provided to the right people at the right time in the right amount is vital to ensure judicious use of taxpayer funds, for example with Medicare costs.

The ability to identify anomalies, in particular those of a fraudulent nature, or more simply of the need to influence behaviours (for example) of doctors billing to Medicare or Child Support payers.

7. Policy & program development

The development of new Government policies and programs is best achieved through data driven analysis and the production of business cases that are based on facts. The Productivity Commission and the Department of Finance in particular demand these qualities of all new programs. DHS data combined with other relevant data can ensure decisions are made that are founded on facts.

In previous times, policy development processes often relied on limited samples of data to model outcomes. DHS has the ability to use the “whole dataset” to provide a complete picture to support the needs of a planned program.

Enabling greater value from DHS and Government data

The examples above clearly indicate surprising and significant value far beyond the core administrative functions that the DHS datasets were originally designed to perform.

How can the States and Territories, other Government entities, non-government organisations (NGOs) and the private sector benefit from these DHS datasets, realising new and unique value relevant to their own strategy and operations and benefit their audience?

There is serious momentum growing around this opportunity. It is noteworthy that a holder of another massive Government dataset, the Australian Tax Office, recently indicated a desire to share its data de-identified for whole-of-government and public use.

What are the blockers?

The blockers most often raised are those of privacy, data security and confidentiality.

While privacy is a pervasive consideration in any application of Government datasets, there are proven methods to de-identify data to allow it to be used for broader community benefit.

If questions in the Government context are assessed through analysis of a “whole” dataset, the results produced will provide better answers to those questions, without identifying specific individuals. Such questions typically analyse data about many hundreds or thousands of people to whom ultimate benefit from a strategic decision may result.

No single Government agency has a complete or comprehensive picture of all citizens (customers) to support today’s needs or the planning of future services. Integration and reconciliation of multiple agency datasets will allow Government and other organisations to gain significant value and strong insight into emerging trends and the needs of the wider populace.

So how do we start to realise value from these datasets?

Four steps are required to realise value:

1. Understand. What are the core challenges and opportunities for which analysis of data can make a real difference? What can we improve through better insight of our core services and customers? What might be the potential benefits of better insight into our core challenges and opportunities?
2. Discover. What datasets in combination with our own existing data can give us new insights into our core operations? Understanding what other agencies and external organisations our core customers interact with may be a clue to what other datasets might provide that “surprising value”. In particular, datasets with a longitudinal basis should be of enormous interest.
3. Source. Engage and establish ethical and secure data sharing agreements with relevant data providers. Create working models of how the different datasets integrate to realise greater value.
4. Analyse. Only when data is correctly integrated and prepared for analysis can a diverse range of analytical processes be delivered. Such processes include reporting, detailed analysis, modelling, forecasting, prediction, monitoring, data mining, trend detection and prescription to produce high value information, enabling informed decisions to be made, addressing the challenges and opportunities identified in step 1.

To ensure that value is enduring, these processes must be embedded into the organisation’s culture. Barry Sandison and his team have created this culture very successfully in DHS.

With the help of data management partners such as Dialog and Catapult BI and leading technology organisations such as Teradata, organisations can greatly increase the operational value of their own datasets as well as delivering surprising new value through analytical capabilities, to the benefit of the community as a whole.

Catapult BI and Dialog are highly experienced with the data science and technology aspects needed to guide customers to achieve great outcomes, providing the management, data and technology perspectives, to allow this value to come to fruition.

Richard Green can be contacted at richard_green@catapultBI.com.