07 July 2014

Valuing the NZ Census

The New Zealand government has published Valuing the Census: A report prepared for Statistics New Zealand which quantifies the benefits to New Zealand from the use of census and population information. It is of interest to statisticians, economists and privacy specialists.

The report states
Statistics New Zealand has commissioned this report to estimate "what dollar value can we place on the benefits to New Zealand gained through the use of census and population statistics information?" This work fits within a much wider programme of engagement with census users to inform relative priorities, and will also provide much of the benchmark material from which an evaluation of the net benefits from changes to census frequency and/or collections methods can be made.
The Executive Summary indicates that
This report provides estimates of the dollar value to New Zealand gained through the use of census and associated population statistics information. The conclusion is clear: despite significant difficulties in developing a rigorous quantification, it is reasonable to conclude that the census delivers benefits well in excess of its direct costs.
The valuation task is complex, reflecting the fact that currently internationally there are no directly applicable models or approaches, and that there are costly hurdles in place to obtain precise estimates of user values for the information. As a consequence, this report utilises a range of approaches to valuation.
What does the census provide? The census provides information on people in New Zealand: it has surveyed the entire population every five years since 1881. As such it provides both a comprehensive picture and a linked time series dataset that has no direct comparators. In valuation terms this poses challenges, as the census' existence (undergirded by the statutory requirement for returns by all New Zealanders) and its generally free dissemination of results means that no market prices exist for direct outputs and that there has not been investment in any tool which closely mirrors the census. Indeed, a frequent response from users during this review is that if the census did not exist, key users would have worked together to create as near a replacement as possible. Uses of the census are diverse, with many applications that are indirect and/or embedded in other products and tools. For information at the level of overall population count with demographic characteristics, census data underpin long-term forecasting such as New Zealand's long-term fiscal position and the requirements for growth related infrastructure and housing. At the more detailed level utilising the Census' more detailed linkage to detailed demographic characteristics for defined geographic meshblocks allows firms and government agencies to identify target groups or, especially when coupled with the historical data, to better understand patterns and relationships such as achievement and earnings for Maori young people. Less direct linkages arise from the census' use in determining the frame for many other non-demographic surveys.
Reliable population based data and projections provide higher level benefits through the reduction in uncertainty for longer-term decisions and investments, and also provide an analytic basis for development of policy choices in some areas which otherwise involve difficult political choices. In an era of growing dynamism in family structures, the census provides one of the main tools to identify those patterns, in turn informing policy, service delivery, and investment choices.
Benefit quantification. As a consequence of the wide range of data uses and the complexity of valuing non-market transactions, this report gathers insights into possible valuations using a wide range of approaches. The core issue explored is in effect a valuation of the extra precision that census data provide over the multitude of other more partial measures. While only a few main areas of use are examined (as more detailed costing would be costly), the report also provides some guidance on the relative values in areas of use.
The main benefit areas quantified are:
  • the benefits from more accurate health funding allocations as funding is delivered more accurately to more needy areas; 
  • reductions in the costs associated with underutilised fixed capital investments, in both the public and private sectors, because of better information on their timing and location (infrastructure funded by central and local government, aged care, retail); 
  • benefits from improved precision and insight in policy making in a range of government agencies, especially for Maori and vulnerable groups; 
  • improvements in the value added by a range of firms which use census data in a wide variety of analyses provided to government and private sector firms; and 
  • gains from improved survey accuracy and reductions in sample size for private sector market research companies, and StatisticsNZ in respect of a range of other non-census products.
Overall benefit to New Zealand. Benefits are typically estimated at an annual level and then summed over a 25 period to provide a net present value. Given the difficulties in assessing values for many benefits, this report provides a set of reasonable ranges in which a value is likely to lie for some key benefit areas. A cost for carrying out the census, including compliance costs, has been deducted from these benefits to provide an overall net present value. Given then that the values included in the table represent only some eleven major areas of benefit out of the much larger range of unquantified benefits discussed, it seems reasonable to conclude that a lower bound for the Census' value to New Zealand is in a range as set out below ...
Using the most generally applicable discount rate of 8%, this suggests a net present value of close to $1 billion for the benefits to New Zealand gained through the use of census and population statistics information over the next 25 years. In other words, every dollar invested in the census generates a net benefit of five dollars in the economy. This value estimate though is not at the level of rigour applicable to assets recorded on an organisation's balance sheet. It does not include many of the uses discussed but not quantified.
There are many other direct and indirect uses of the census for which quantification has not been attempted but which are clearly highly valuable. The census is used for instance to determine the electoral boundaries for Maori seats, it forms the basis for the NZ deprivation index (widely used in a range of research and policy work aimed at helping New Zealand's most vulnerable people) and underlies work on the Long-Term Fiscal model which informs tax and expenditure policy choices affecting the next 10-50 years.
Indirect uses are also widespread. Economic models rely on robust demographic analysis. Another less obvious application is the use of census data as part of modelling work underlying the calculation of sustainable pathways for Regional Councils and the ecological modelling used to estimate potential future environmental loads and impacts. The difficulty and/or cost of identifying values on these mean it is not cost-effective to develop further, but a consequence is that the overall value of the census to New Zealand will be significantly above the quantified benefits outlined in this report.
Looking forward: use of this valuation. This report clearly indicates that the census provides value to New Zealand well in excess of its cost, but it does not address the issue of whether the current collection and analysis system provides the best value-for-money. It could be that net expected value might be greater if either some additional accuracy or new outputs could be produced (even involving an increased cost), or a combination of changes to the collection and processing systems along with changes to the types and quality of outputs produced was adopted.
This would require a much more detailed set of analyses, for which the information in this report provides a starting platform. This report provides guidance on some areas of high value, some indications of relative value, and identification of many key users which enables more targeted exploration for further stages of census development. For instance, consideration of a move in the timing of censuses to 10 yearly could be investigated on the basis of the difference in value (accuracy and timeliness) to users in key areas, weighed against the expected reduction in costs. This step will require clearer details of the potential changes in methodology and their consequences in terms of accuracy and cost than are currently available.