Textes d’opinion

Does the 2016 census point the way to better social policy?

Peter Hicks | 4 avril 2015

Yesterday, Statistics Canada announced that Canadians will no longer be asked to fill in information about their income on the 2016 census. In its place the agency will simply obtain this information from tax records linked to each individual. This is big news.

While this will not fix all of the damage caused by the federal government’s 2010 decision to make the long-form portion of the census voluntary, the change (which extends a voluntary option which respondents have been able to select since 2006) is an important step in enhancing the accuracy and reliability of Canada’s socio-economic data.

Statistics Canada’s decision is confirmation that the administrative data which exists within various government departments can be leveraged in a powerful way. As illustrated by the change to the census, admin data can help fill in gaps among, or even replace data collected by, existing self-reported surveys. Indeed, for several years Statistics Canada has been diligently linking together tax records and various surveys like the Labour Force Survey in order to get a more comprehensive picture of how individuals transition through life.

The census decision is just a start. Administrative data holds great potential to radically improve how governments, businesses, and individuals use information, to make better, more effective decisions. For those interested in “evidence-based” policy this is what we ought to be talking about.

As I argue in a new essay for the Institute for Research on Public Policy, governments are just beginning to grapple with how to use the massive amount of data at their disposal.

Governments are still far behind the technological changes that have already reshaped many realms of society, but transformational change appears to be on the near horizon, not only in statistical information but in the very design of the programs themselves.

In today’s environment, citizens want customized programs and services that are aligned to their needs, preferences and goals, and that take account of the full course of life. Companies like Amazon and Netflix have been able to respond to some of these demands by developing a data and statistical infrastructure capable of tailoring recommended services to each user.

So why does much of government, particularly in the area of social policy, still look a lot like it did in the 1960s and 70s?

As I argue in the essay, one of the key reasons that our policy framework has failed to keep pace is because we lack the necessary statistical tools and evidence. Many of today’s surveys and censuses are based on technologies that pre-date the computer age and are designed (as with policy itself) to only look at the broadly defined needs of broadly defined groups of people in a broadly uniform manner at a single point in time.

Administrative data allows us to finally look at individuals in their complex economic, and social settings. Linking tax data with the census is a good start. However, if we want to affect a real cultural shift in government, we need a new framework for both statistical and policy development that can support these changes. In the essay I describe how such an approach could be applied in the area of social policy – and in a manner that does not infringe on privacy.

At the heart of this new model, which I refer to as the “enabling society”, would be a system of Big Statistics delivering detailed information to policy-makers, service providers and citizens themselves about the kind of interventions that have been shown to work best based on a given individual’s profile, needs and preferences. For example, someone who is contemplating going back to school or changing careers would be able to know in advance the probability of success in terms of prospective employment, earnings and job stability, given their present skills, past work history and future goals.

Generating this kind of real-time information requires governments to bring together lots of different overlapping sources of data on individuals—spread across many different departments and levels of government—and place them in a statistical framework in which analysts can start to map these linkages and assess the effectiveness of different kinds of interventions. Integrated into all aspects of the design, delivery and governance of social policy, this evidence-base will finally make it possible for governments to adopt a citizen-centred approach.

Since the death of the mandatory long-form census a lot has been said about the demise of evidence-based policy. These concerns are real, but in a way they also look backwards at a statistical system and policy environment ripe for change. Let’s have that debate.


Peter Hicks is a social policy consultant who was previously an assistant deputy minister in the Government of Canada. His essay, The Enabling Society, can be found at http://irpp.org/