Policing ‘troubled families’ through ‘algorithmic regulation’

These dynamic systems can handle unstructured, messy and unpredictable data and respond in real-time to new ways of acting on patterns in the data. Like a shoal of startled fish, the application of this heterogeneous data can sharply change direction at any moment. (McQuillan, 2015: 3)

There are tools available that can help authorities to identify the most vulnerable children and families in their area, even as their circumstances change. Technology can now automatically alert a social worker when a child starts to miss lessons or allow a youth offending practitioner to instantly pull up a history of school exclusions for a child they are visiting. (Neal, 2014)

An interesting, but generally overlooked, aspect of the Troubled Families Programme is the increasing reliance and emphasis on data related issues. I often come across articles on the internet from software companies and management consultancy firms offer ‘data solutions’ to some of the challenges presented by working with ‘complex’ families. The quote above by Phil Neal, MD of Private Eye favourite C(r)apita One is taken from this post on the Public Servant Daily here and other examples, including different ones by Neal, can be found here, here, here and here

I may never have got around to posting this blog, if I hadn’t been sent a brilliant and disturbing paper by Dan McQuillan on ‘Algorithmic states of exception’. It’s very difficult to summarise the paper, so I’m not going to try to do that in the hope that you’ll be sufficiently intrigued to read it yourselves, but McQuillan makes the point that new data structures and new types of database allow for a form of ‘algorithmic regulation’ which can be used to ‘predict’, ‘identify’ and ‘modify’ social problems as they are envisaged by the government. He uses the Troubled Families Programme as an example and when I read this section, I was instantly reminded of a paragraph from Donzelot’s classic text The policing of families. The two excerpts are below, both emphases are mine:

At the start, there are always the figures on delinquency, the statistics of offences committed by minors. Experts in criminology study this first layer and detect in the delinquent minors’ past, in the organization of families, the signs they have in common, the invariables of their situation, the first symptoms of their bad actions. With the help of these findings, the typical portrait of the future delinquent, the predelinquent, the child in danger of becoming dangerous, can be drawn up. An infrastructure of prevention will then be erected around him, and an educative machinery will be set in motion, a timely action capable of stopping him short of a criminal violation. Not only will he be an object of intervention, but by the same token, he will in turn become an object of knowledge. The family climate, the social context that causes a particular child to become a “risk,” will be thoroughly studied. The catalogue of these indications makes it possible to encompass all forms of maladjustment, so as to construct a second circle of prevention. (Donzelot 1977: 97-98)

The way society disciplines citizens through discourses of health, criminality, madness and security (Foucault, 1977) are given categorical foundations in the structures of data. Consider, for example, the category of ‘troubled families’ created by the Department for Communities and Local Government (Department for Communities and Local Government, 2014) to identify families as requiring specific forms of intervention from the agencies in contact with them. The 40,000 or so families whose ‘lives have been turned around’, by being assigned a single keyworker tasked with getting them in to work and their children back to school on a payment by results model, would have been identified through some operations on the data fields that make their existence legible to the government. In turn, various agencies and processes would have operated on those individuals as both effect and affect, as a created intensity of experiential state, in ways that would construct the subjectivity of membership of a so-called troubled family. These actions would, in turn, become new content for data fields and would form the substrate for future interventions. Thus, the proliferation of data does not simply hedge the privacy of enlightenment individuals but produces new subjectivities and forms of action. (McQuillan, 2015: 2)

Essentially, we have seen Donzelot’s prediction ‘re-booted’ and put in a cloud via a ‘technological turn’. McQuillan argues that, as a result of new data structures and increased contact with technology, ‘everyday life is becoming permeated by points of contact with algorithmic systems that can influence the friction or direction of our experience’ (2105: 5). When these technologies seep into the public sector and the realm of social control, it affords states the opportunity to ‘predict’ who will do what in the future, and establish procedures, sanctions and exclusions to stop them from doing so. Of course, some populations, such as those deemed to be ‘troubled’ will have more ‘points of contact’ with the state than others and some will already be ‘known’ to agents of the state.

And then, serendipitously, DCLG published a document on Data Sharing Guidance and Principles, as an Annex to the new Financial Framework for the expanded phase of the Troubled Families Programme. The document suggests ways that the maximum amount of personal and familial data can be shared between service providers in a ‘legally compliant manner’ across each of the six criteria for being a ‘troubled family’. And, as if that wasn’t enough, a final section highlights how frontline workers can ‘nominate’ families where they believe there exist ‘problems of equivalent concern’:

These indicators enable nominations from professionals locally and, depending on the nature of the risk and seriousness of the circumstances, may be undertaken with or without the individual’s consent. In some cases, consent must be obtained by law before a nomination is made. However, in cases where consent is not prescribed by law, individuals should be made aware that their data is being shared and their consent should be sought wherever possible. However, this will be a matter for local assessment and professional judgment in the circumstances of each case (DCLG, 2015: 7)

Nothing to worry about there then….



Donzelot, J. (1997) The Policing of Families, Johns Hopkins: Baltimore

McQuillan, D (2015, forthcoming) ‘Algorithmic States of Exception’, European Journal of Cultural Studies, Vol 18, No 4-5, ISSN 1367-5494


Paul Garrett has also written previously about the ‘electronic turn’ in social work which highlights similar concerns and also highlights some of the potential benefits of social workers engaging with new forms of technology.

And, finally, many thanks to the individual who bought Dan McQuillan’s paper to my attention. Much appreciated.

The ‘troubled families’ (non) announcements

We’ve always known that these families cost an extraordinary amount of money … but now we’ve come up with the actual figures. Last year the state spent an estimated £9 billion on just 120,000 families … that is around £75,000 per family.

(David Cameron, Troubled Families Programme launch speech, 15 December 2011)

The publication of a DCLG report and the announcement last week that the Troubled Families Programme (TFP) had seen ‘More than 105,000 troubled families turned around saving taxpayers an estimated £1.2 billionhas been subjected to a bit of scrutiny over the past few days. Jonathan Portes has posted a pretty comprehensive blog here, the Mirror have covered the announcement in an online piece and I’ve done a couple of short posts here and here. But there are a few things which weren’t announced last week or which were not included in the report which also deserve a little bit of scrutiny. So here goes.

Firstly, the £1.2 billion ‘savings’ figure never features in the DCLG report The Benefits of the Troubled Families Programme to the Taxpayer (hereafter called the Benefits report),or the accompanying Methodology report. In fact, the Methodology report states very clearly on p4 that ‘care should be taken when directly comparing results between local authorities’ and ‘the results reported are not results from the national programme as a whole’. If the advice to take care  with the figures had been taken, the figure of £1.2 billion ‘savings’ surely wouldn’t have been featured so heavily in the press release and there must be a very good reason it didn’t feature in either official publication. Two other figures which are mentioned early in the press release don’t feature in either publication. The ‘average gross cost saving to the taxpayer per ‘troubled family’ is identified as £12,000 in the press release but in the report it is stated as £11,200. The press release also states that the average cost of intervention is £5,493 per family, but this figure doesn’t appear anywhere in the report either, which puts the figure at £5,214 (p4). In the DCLG TFP Financial Framework document published in 2012, it is estimated that the cost of intervention was expected to be around £10,000 (p7). This discrepancy or ‘saving’ was not highlighted last week.

When the number of families ‘turned around’ is reported so precisely, it appears strange that less ‘care’ is taken with the reporting of the financial elements of the programme. It should also be noted that the £1.2 billion figure is a gross saving – it doesn’t take into account the total cost of the interventions, which, if we take the figure in the official publication rather than the press release, comes to around £550 million. So, taking a leap of unjustified faith with these figures, the best estimate of the net savings at the current time, without any counterfactual, and assuming that every positive change that took place within ‘troubled families’ lives was directly attributable to the work of the TFP, is around £650 million.

When David Cameron launched the TFP in 2011 he said the government had ‘come up with the actual figures’ for how much ‘troubled families’ cost. He said it was £9 billion – around £75,000 per family. Neither of these figures are mentioned in the Benefits report of the accompanying Methodology report. In early 2013, DCLG published two reports – The Fiscal Case for Working with Troubled Families: Analysis and Evidence on the Costs of Troubled Families to Government (which broke down the £9 billion figure into different policy areas) and The Cost of Troubled Families (which includes case studies of local authority estimates of the cost of their ‘troubled families’). Neither of these two documents are mentioned or referenced in the most recent reports, nor is anyone directed to them via any of the webpages related to last week’s announcements and publications.

The Benefits report suggests that, across the seven local authority areas used for the analysis, the average cost of the ‘troubled families’ they worked with (in the 12 months prior to intervention) was just £26,200 (p4). So the new ‘actual’ figure is around a third of the first ‘actual’ figure of £75,000 that the Prime Minister gave when he launched the TFP.

Another surprising omission form the information provided in the recent reports is the fact that two of the seven ‘exemplar’ local authorities who provided data for the Benefits report also provided costings for the Costs of Troubled Families report released in January 2013.

In Manchester, back in 2013, the city council estimated that they spent, on average around £74,000 per year on their ‘troubled families’ (p32) and predicted around £32,600 of savings from the TFP (p5). The figures released last week suggested that in the three years of the programme, the cost of ‘troubled families’ in Manchester in the 12 months prior to TFP intervention had been £58,238 in year 1, £37,075 in year 2 and £27, 793 in year 3. The average saving had been £6,240 in year 1, £7,632 in year 2 and £9,530 in year three (Annex A of the Benefits report). I’d argue these are quite substantial differences with neither the cost being anywhere near as high as anticipated (in year 2 it’s around half the expected cost and in year 3, just over a third) and the savings also, understandably, being severely lower than predicted.

In 2013, Wandsworth predicted savings of around £29,000 (p5) out of an expected cost of ‘troubled families’ of £32,000 (p27). The figures released last week suggested that ‘troubled families’ in Wandsorth cost on average only £17,895 in the 12 months prior to TFP intervention and the estimated savings were only £6,528 per ‘troubled family’ turned around (Annex A of the Benefits report).

But none of this is mentioned in any of the information released by DCLG last week, when it is surely highly relevant to understanding how the programme is progressing, or not progressing, in terms of its expected savings.

These are, effectively non-announcements – pieces of information which one could argue should have been published, discussed and explained or laid out for deliberation and scrutiny, if necessary. And it would have been very easy to do this, given that the independent evaluation has not reported any substantial findings yet and if the government hadn’t been so intent on telling everyone how perfect this programme is. Anyone would think the Prime Minister had made some kind of personal commitment to ‘turn around’ the lives of these families before the imminent election, saving bucket loads of ‘taxpayer’s money’ and transforming public service delivery at the same time…..



The Troubled Families Programme: the perfect social policy

Coverage of the recent announcement regarding the ‘success’ of the Troubled Families Programme (TFP) in turning round over 105,000 families appears to have missed a very interesting finding from the release of the local authority figures. The government press release accompanying the news states that

There are 117,910 families targeted under the government’s Troubled Families programme. For rounding purposes, however, the target is usually referred to as 120,000

The TFP has apparently achieved a 100% success rate in local authorities identifying exactly the same amount of troubled families in their area as the indicative number provided by the government – the total number of families identified and worked with by local authorities comes to 117,910. Every single local authority has identified every single ‘troubled family’ in its area and every single local authority has begun working with them. The practice on the ground, therefore, represents a perfect alignment with the information provided by politicians and civil servants in Whitehall. This is despite the data used to identify the number of ‘troubled families’ being 6 years out of date and compiled for completely different purposes and identifying potentially very different families.

The figures also highlight that local authorities have performed better than the DCLG expected they would. On page 2 of the Financial Framework guidance for the TFP, the following statement can be found:

You may not succeed in turning around every family that you work with, and therefore it is likely you will need to work with more families than your indicative numbers.

Not one local authority has needed to work with more than their indicative number in order to ‘turn around’ all of their families. In fact, many local authorities can demonstrate a 100% success rate not just in identifying and working with ‘troubled families’ but in turning them around. Manchester, for example have identified, worked with and turned around a staggering 2385 ‘troubled families’. Not one has ‘slipped through the net’ or refused to engage with the programme. Leeds and Liverpool have a perfect success rate in each ‘turning around’ over 2000 ‘troubled families. By my reckoning, over 50 other local authorities across the country have been similarly ‘perfect’ in their TF work. Not one single case amongst those 50 odd councils where more ‘troubled families’ were identified or where a ‘troubled family’ has failed to have been turned around. It is also staggering that work with some of the most disadvantaged families who have allegedly been immune to all previous policy interventions and whose ‘troubles’ have existed ‘for generations’ has been so successful at a time of wide-ranging and deep cuts to local authorities and others public services. But, we must believe the government on this. For whilst the figures do not constitute ‘official statistics’, the government surely wouldn’t have published them if they knew them to be false or misleading….

So there we have it. The Troubled Families Programme. The perfect social policy.

No method in the ‘troubled families’ madness

DCLG published a couple of documents yesterday relating to the alleged ‘savings’ accrued to ‘the taxpayer’ from the Troubled Families Programme. I’m going to try and do a couple of blogs on different aspects of these publications and the ‘news’ contained within them, and this first one is going to look at the methodology behind the reports.

It would be churlish not to acknowledge that the publication of a methodology report represents something of a symbolic step forward, considering the contempt that has often been shown to methodology in other aspects of the programme, not least the identification of the 120,000 families in the first case. But that’s kind of where the good news ends.

So, firstly, the report The Benefits of the Troubled Families Programme to the Taxpayer and accompanying press release, which highlighted the estimated £1.2 billion saving from the Troubled Families Programme (TFP), were based on data collected from 7 of the 150 plus local authorities involved in the programme. The methodology document states that ‘the results reported are not from the national programme as a whole’ (p4) but this is not made clear in the press release which accompanied the report. We are also told that, of these seven, ‘five local authorities used a sample, while two local authorities collected data on their full cohort of families’ (p4). We are also informed that:

‘The data collection methods used represents a significant step forward in evaluating costs and benefits at local authority level. Local authorities sought to use the best available data and to maximise the representativeness of their local samples. However, some samples may not be entirely representative’ (p4).

But we are never actually told what the sampling procedure was, which was a similar problem with the data accompanying the DCLG Understanding Troubled Families report.

Read on a little further and we are told that ‘care should be taken when directly comparing results between local authorities’ (p4) which is, again, a familiar concern with data collected under the TFP. The DCLG press release which featured comments from Eric Pickles, Danny Alexander and Eric Pickles didn’t appear to exhibit much in the way of ‘care’ when it triumphantly proclaimed ‘the new figures showed that the programme had already saved taxpayers an estimated £1.2 billion’. Neither the report nor the methodology document actually mention the figure of £1.2 billion saved, but that didn’t stop the DCLG press ignoring the advice of its own report by extrapolating estimated costs from seven local authorities and applying the same flawed logic to over 150.

These are not official statistics that we are dealing with and local authorities have used their own ‘local administrative systems’ to collect data. Where this has not been possible ‘local authorities have used data provided by key workers’ (p4). So this ‘significant step forward’ includes using data that local authorities already had and asking a few workers to complete some forms – which could be used to help justify or jeopardize their precarious jobs – when they have no other way of finding information.  The report then goes on to acknowledge that ‘data collection limitations have prevented local authorities recording information on every measure’ (p4).One local authority managed to collect information for only 14 of 39 indicators and most achieved around between 22 and 29 out of 39.

As Jonathan Portes pointed out on Twitter, the fact that no impact assessment has been carried out shows that the governments assertions about ‘savings’ are ‘largely meaningless’. The methodology also notes that whilst the cost savings calculator does use ‘the best available ‘deadweight’ estimates’ (p5) to understand what might have happened without the TFP intervention, the report itself deals only with ‘gross’ figures and therefore assumes that all of the ‘savings’ across all of the families can be attributed to the TFP intervention. Which is quite ludicrous really. But then the whole thing is…

Best wishes,


As I said at the top of this post, I’ll hopefully do more blogs on this in the next couple of days on the language used within the main report and the lack of reference to previous ‘research’ and analysis on the ‘cost’ of ‘troubled families’.