The economics of #OpenData – are there any?

Last week the NAO published its Implementing Transparency report.

The objectives of transparency are summarised in the report as accountability, service improvement based on user choice and comparative data and the much sought but elusive economic growth from new products and services based on OpenData.

The Guardian in an uncharacteristically scathing piece summarised the report.

“Read between the lines of its [the NAO] report out today, Implementing Transparency, and you will see a government which has been chucking out tonnes of data, that no-one looks at and without a complete strategy. Oh and it’s cost an awful lot of money.”

Fair comment? I think so. It certainly seems that the rhetoric of transparency and accountability (possibly linked to the underlying ideology of reducing the size of government) has held sway over the bean counters in the Treasury. Of course not all of the benefits of transparency can be put into a profit and loss account but it does appear that common sense and value for money have not been important criteria in determining what data should be prioritised for release.

“As the scope of the transparency agenda has developed, the Cabinet Office has published examples of the benefits of public data initiatives to support the strategic case for transparency, for example on its Open for Business website, but has not yet systematically assessed the costs and benefits of the Government’s specific transparency initiatives.” (NAO)

Concern is raised about the quality, completeness and suitability of the data that has been published to inform better choice and accountability. I have ranted extensively about the publication of crime statistics (here, here and here) and questioned whether accountability will work to the benefit of all groups in society.

“With regards to community accountability, the police crime map provides much more detailed recorded crime information than was previously available. However, additional information is still needed, for example on police activity and resourcing locally, for residents to hold neighbourhood police services to account more fully.” (NAO)

There has been little attempt to measure or assess the benefits of transparency apart from some limited statistics on website visits. We learn that data.gov.uk has had 1.75m visits since its launch in January 2010, unfortunately over 80% of users “leave from either the home page or the data page on the website. This suggests that they are not accessing data during their visit ..” One of the highlights quoted is the 47m visits to police.uk between February and November 2011 which sounds pretty impressive until the Guardian pours some cold water over this OpenData triumph.

“We were interested in that 47m figure for the crime maps site and tested it using Nielsen data. There is no guidance on what exactly constitutes ‘visits’ – is it page views or unique users? Our figures show that while the site did get a lot of visitors when it was initially launched in February last year – and had a brief peak during the England riots last year (ironically, the data on the site is all historical, so visitors looking for riot offences would have been disappointed), in December it appeared to only have 47,000 viewers, looking at 364,000 pages”

In the interest of balance, it should be noted that access to the Depart for Education’s schools performance tool (note the DfE’s tool not an application built by the private sector) has increased by 84% in the period under review.

The report highlights three areas of risk – privacy, fraud and unintended consequences.

The privacy concerns have been extensively aired and it seems that just removing names and addresses or some simple aggregation will not prevent data being cross tabulated with other sources to deanonymise. The government has commissioned an independent review into privacy and transparency, it will be interesting to see what response this review gets from the Cabinet Office particularly with regard to the NHS data being offered to big pharmas.

Whilst the economic benefits of OpenData have been slow to materialise within the private sector apparently some enterprising villains have already identified new opportunities to defraud the public sector using OpenData

“… with increased transparency around contracts and payment details – fraud attempts to a value of £7 million directly related to transparency releases have been found in local government …” (NAO)

On the economic front government is flaunting a new, improved and bigger estimate of the economic contribution of public data is £16bn. The source for this estimate is (once again) based upon the opening up of geographic data in New Zealand

“A recent review of the literature on reusing public sector information put the value of public sector information in the European Union at around €140 billion a year. The author based the estimates on extrapolating from studies of the total economic impacts of geospatial information in Australia and New Zealand. Based on this review, the Government derived an estimate of £16 billion for the current total economic value of UK public sector information.” (NAO)

This sounds like “Pollock squared” to those of us of a sceptical bent and the NAO goes on to point out an alternative much lower estimate from the OFT

“The Office of Fair Trading produced an earlier study in 2006 on the commercial use of public sector information. They surveyed more than 400 public bodies to identify the income generated from public data release. They also commissioned research to estimate the economic value of UK public sector information. Based primarily on the survey results, the contractors estimated this value to be about £590 million in 2005.”

“Both studies have limitations. The Office of Fair Trading report notes that top-down approaches, such as that used in the EU-wide estimate, tend to overstate the economic value of public service information. This is because they do not factor in reasonable substitutes available if that information does not exist or is prohibitively expensive. Furthermore, the assumption of similar public sector information markets is crude given significant known differences between countries. However, the Office of Fair Trading report is likely to understate the economic value of public sector information.” (NAO)

So we don’t really have a good idea of the potential economic benefit. Fair enough, this is part of the “new economy” we are sailing in uncharted waters and should have some faith that economic benefit will follow on from the release of OpenData. But surely that would suggest a phased approach to the program of releasing OpenData allowing government to assess the benefits from accountability and new businesses along the way and to keep some control over costs? data.gov.uk currently costs £2m p.a. to run, there are also departmental costs to release data.

“The costs [per department] range from £53,000 to £500,000. These represent a lower bound for the cost of standard releases because they only capture staff costs and do not include, for example, costs of upgrading IT systems or payments to contractors.” (NAO)

No doubt my more technical friends will ridicule the higher end costs to publish data but inevitably there are staff costs involved in identifying, extracting and preparing data for publication. Anonymisation of personal data content could prove to be even more costly particularly if private sector consultants are needed.

The benefits sought from the transparency agenda, accountability and service improvement, are almost unarguable and are probably worth the £15m a year we may be spending on releasing data (2m for data.gov.uk plus 25 odd departments at 250k and perhaps the same again for Local Government and others). But let’s not be kidded by the unproven economic arguments of  OpenData champions in the academic and startup communities into believing that any data requested should be released regardless of the cost or anticipated usage.

If this were a school report (we would probably be at the end of the first year of secondary school) I can imagine the tone of the head of year’s summary

“Francis has made a good start at the OpenData Academy. He is an enthusiastic pupil who has displayed a positive attitude to his subject but needs to pay more attention to measuring the usage and benefits of OpenData. In the coming year he should reinforce his enthusiasm for transparency and  accountability with a sound economic framework that does not depend upon unsubstantiated forecasts of an economic miracle (advice that I also gave to his cousin George).”