Finding the cost of Open Data 2


https://youtu.be/U-Y0SMitMpk

Apologies for the oddly targeted adverts in the video above but that’s the cost of free.

We don’t have an option to vote or even express opinions on individual items of government expenditure in the UK, so I can’t say I’d like to spend £10bn on rail subsidies or pay 2p in the pound tax on the NHS or on education or 0p on Trident nuclear weapons (I haven’t got a clue whether those figures are in the right ball park but hopefully you get my drift). If we did use a popularity based system for setting levels of public expenditure I expect we would get some pretty quirky results because everyone would want to pay less tax but would have their pet topics on which they would like to spend more money.

“What’s he rambling on about tax and government expenditure for? I didn’t come here for the Guardian leader” Chill – at some time most of us have taken a trip down this rabbit hole on expenditure and tax in regards to something we think government should be spending money on or not. Let’s try to apply this approach (flawed as it is) to Open Data.

How much do the general public as opposed to us geeks want Open Data? Well a review by Sciencewise of various surveys into public attitudes over the last 5 years suggests not too much enthusiasm or understanding

Public awareness of open data and its relevance is low, possibly because it is an abstract issue with unclear benefits to everyday life …

The drivers for opening up and sharing data are not necessarily clear to the public, and views on the benefits of open data are not universally shared between members of the public …

No doubt as public confusion about Open Data and personal data privacy is resolved many more people will be in favour of opening up data to aid transparency of government and policy making. However, I couldn’t find any indication in the report that the public recognised the potential for Open Data to fuel new businesses and services, understandable really as at the moment belief in the economic returns of Open Data within government is in the same faith based corner of policy making as some of the initiatives of Michael Gove.

The original work on the economic and social benefits of marginal cost pricing to support Open Data was “Models of Public Sector Information Provision via Trading Funds” by Newberry, Bently and Pollock (2008), this painted an exceptionally rosy picture of the benefits and has largely disappeared from discussion in recent years. A couple of months ago BIS finally published a report by ConsultingWhere and ACIL Talisman “Assessing the Value of OS OpenData™ to the Economy of Great Britain – Synopsis.” This report appears to have been sitting on the shelf for well over a year, perhaps the results did not quite paint the story that people were hoping for? The study projects forward the identified benefits from 2012/13 to 2016

“The study estimates that the OS OpenData initiative will deliver a net £13.0 million – £28.5 million increase in GDP in 2016. The main components of this increase are net productivity gains (£8.1 million – £18.2 million) and additional real tax revenues (£4.4 million – £8.3 million).”

This is hardly the frothy level of benefit that early advocates of opening up OS data kept quoting. BIS commented on the report in an appendix

“Some caution is required in using these findings. Although around 70 firms were contacted to participate in the interviews, only a small proportion of these (around 20) did so. The majority of those contacted were unable to articulate the benefits of using the data and therefore were not included in the study. Those that were able to articulate this were typically already aware of, and in some cases using, the data before its use became unrestricted. In essence this methodology assumes that the 20 are representative of those who are downloading data, but in practice as they are those who can articulate the benefits they may well be getting greater benefits than the average user. More generally firms which were able to respond to the interview questions may have different characteristics to other firms.

The study grosses up the results using sector multipliers and OS OpenData download records, applying a sensitivity to the resulting analysis to calculate an upper and lower bound estimate. However, there appears to be no evidence underpinning the choice of sensitivities applied to the analysis.”

That might be read as questioning the methodology and putting a damper on excitement about the net benefits of Open Data release. The author summed up by saying

“This is not an exact science since these models are driven by a series of assumptions and data extrapolations. Such results should be interpreted more qualitatively than quantitatively. i.e. they tell us that Ordnance Survey opendata has generated an economic benefit.”

Now maybe bus stops or health data or expenditure data or some other open data offer a bigger return, perhaps some who are more knowledgable will comment on these sectors. But remember that the Free Our Data campaign which helped to kick off the whole Open Data movement in the UK was largely focussed on the release of OS Data and it appears that  to date the benefits are less than we had been lead to expect.

Just to restate my position on opening up public data. Open Data has the potential to shine a light on government performance, policy and expenditure and is an essential tool in holding our government to account but it is yet to be proven that opening up data will unleash a torrent of innovation or economic activity that will significantly outweigh the costs of releasing the data.

Which brings me to the title of this ramble – this open data stuff is not without cost. Setting aside the lost revenue of the trading funds which, at least in the case of OS Open Data, it appears can be balanced by wider economic benefit, we have to recognise that for most government departments there are costs in releasing data. The National Audit Office report “Implementing transparency” (2012) summarises departmental staff costs for open data release

“Estimates provided for this review suggest that the additional staff costs of providing standard disclosures of pre-existing data range from £53,000 to £500,000 annually by department.”

“The National Policing Improvement Agency produced street-level crime maps at a cost of £300,000 and estimates the annual running costs for hosting, supporting and maintaining the crime map site at more than £150,000. The Agency has budgeted £216,000 in 2011-12 to further develop the website, including linking crime data to police and justice outcomes”

“The Department for International Development estimates that to deliver the commitment to provide full information on international development projects with a value of more than £500 by January 2011, it incurred capital costs of £250,000, administrative set-up costs of £156,000 and has ongoing annual running costs of £64,000”

In the big picture these are small sums of money but when almost every sector of society is impacted by cuts and austerity  surely we need a more clearly articulated assessment of the benefits and costs of opening up government data. Perhaps we need some Open Data about the costs and identified benefits of Open Data?


2 thoughts on “Finding the cost of Open Data

  • Rollo

    Enjoyed the blog post; you’ve clearly been doing some research. I was intrigued by the stats associated with the release of OD, as there appears to be a confusion between providing OD and providing data services. To include costs for creating and maintaining a national street level crime web site is, in my personal opinion, not justifiable if you are attempting to assess the cost of OD per se (which should focus on the compilation and supply of the data). There are other ways that the NPIA could have supplied that data – in the words TBL “raw data now” might apply. In the case of OS, OD is just that – data. Associated services (also free for non-com use) are accounted for separately (OnDemand/OpenSpace). None of that distracts from the main thrust of your blog however…it’s more of a case of the definition of OD being inflated in certain instances to suit certain arguments.

Comments are closed.