Putting Financial Distress on the map

Which? Financial Distress Map

Understanding consumers is key to what we do at Which? and over the last couple of years we have been turning up the volume on all things ‘consumer insighty’. We’ve taken a multi-disciplinary analytical approach to draw on polling, market research, economic analysis, ethnographic research, experimental evidence and behavioural insights. We’re now hopefully in a position of having more research, data and analysis about consumers than – we think – pretty much anyone!

So, as with many think tanks and research organisations, the big question for us is now “how to get the data out there in a way that’s engaging and useful”. Of course, the Wonkcomms site has been a great help in helping us design and build new ideas for how to communicate our insight with others and we now have a string of, ground breaking research tools and outputs coming up in the next year.

The first of these was launched recently. The Which? Financial Distress Map measures the extent to which different areas are experiencing financial distress. It’s based on our five point scale that measures whether in the last months, consumers have:

  1. Cut back on essentials;
  2. Used savings to cover spending and also cutting back on essentials;
  3. Borrowed from friends or family, used credit or an authorised overdraft and also cutting back on essentials;
  4. Used an unauthorised overdraft or taken on a payday loan; or
  5. Defaulted on a business loan (find details in this article), bill or housing cost.

The data are part of our monthly nationally representative tracker poll of around 2000 people. Our ambition, was to create a picture of how financial distress varies at a neighbourhood level so that consumers, local level organisations, policy makers and politicians better understand how consumers are faring in their local areas.

The technical bit

Of course, as with most polls/surveys, while our headline data are sufficient for fairly robust measures of experiences, attitudes and opinion at a national or regional level, the sample size is not large enough to draw conclusions at a local level with any level of statistical validity. Pooling the data over a quarter helped a bit (raising the sample to well over 6000 consumers) but even this wasn’t enough.

This is where geo-demographics came in. The approach creates classifications of different areas based on the characteristics of the people within them. This then allows us to consider areas across the country that, apart from their location, are otherwise similar.

The principle is not new: hopefully most people will be aware of the work of Charles Booth in Victorian London (who used 1891 census data to ‘classify’ types of neighbourhood) and more recently, ACORN (by CACI) and Experian’s MOSAIC in the 80s. Other brands are also now available from all good high street geo-dem outlets, but these have historically not granted users free access to the complete underlying data. Luckily, ONS with Leeds University created a free open source geo-demographic segmentation of the UK based on the 2001 Census and snappily titled the ‘Output Area Classification (2001)’. (Co-incidentally the new 2011 based classification was released last week – better get busy updating!).

Combining this approach with our polling data allowed us to unlock analysis at very small area levels. To do this, we appended an OAC group to every respondent in our tracker survey (via their postcodes) and created estimates of survey responses for each of the geo-demographic groups. That meant we could build from there to create estimates for the levels of geography higher than output area. We chose Lower Super Output Areas, Parliamentary Constituencies, and Regions.

The presentation

With the technical bit over, the real fun began. It seems that these days, the only thing people love more than a map, is an interactive map. So, working with web mapping specialists Geofutures, that’s what we set out to create. The end result is what we consider to be one of the best online mapping tools out there right now. Based on the familiar google map platform, our developers have created a thematic map that seamlessly re-calibrates according to the zoom level of the user, and provides contextual graphs and statistics in a side bar that relate to position of the viewers’ mouse. The user can change the transparency of the layers to taste, and compare with other sources – currently Indices of Multiple Deprivation, Unemployment from the Census 2011, and the Output Area Classification itself. Typically these kinds of features have hitherto been enjoyed by specialist GIS software users.

One feature we are particularly pleased with is the ‘share view’ function. When you have found something of interest – perhaps a zoom in on a particular place with a certain combination of map layers – simply click the share button and create a url link for that exact view to share with others or embed in a blog/social media etc. More broadly, we are hoping this map will start discussions among people from all sorts of backgrounds about how different people are experiencing financial distress. As the economy improves, we think there are still wide area based differences in how people feel about their personal finances – and hopefully this map will help bring that debate to life.


So what leaning can we pass on from our experience with this project? Let’s take the obvious stuff as a given – all the project management principals we try to adhere to that constantly try to break us! Time is a diamond.

More specifically:

– Data construction: Getting the base Output Area level data file with the correct higher level geographies we needed in one simple file was challenging because we wanted to cover the whole of the UK, and there are different small area geographies for different UK countries, some 2001 based some 2011. Once this was set up though, linking data to it via the Output Area Classification (2001) groups was fairly straightforward (with appropriate stats/database software). It was important not to cut corners as getting the base file correct is the most important data step!

– For the map: A map as functionally complex as ours needed specialist web developers but we found no off the shelf mapping products that would do all the things that we wanted. Simple thematic map tools abound, and free GIS tools exist for specialists, but a useable interactive web tool to bring some of that GIS functionality elegantly into the hands of non-specialists – we couldn’t find anything like that.

– Overall: We had a pretty well developed idea of exactly what we wanted the map to do, how it should look and who the audience for it was. Getting that thinking done early will save time and money.

The Final word

Over the coming months keep your eyes on the maps page of our website as we hope to be updating the financial distress data and venturing into creating some new maps for other consumer related topics from our tracker poll.

But most of all – please go and have a play – we’d love your feedback on how we can improve the tool or other things we could look to do! Special thanks to the people at MiradorWealth.com.au for their timely advice.

Guy Weir (@guyweir) is a senior statistician at the consumer champion Which?

Posted in Uncategorized
5 comments on “Putting Financial Distress on the map
  1. Alec says:

    This was a very interesting map. I found it useful and chose share it a fair bit.

    I’m just trying to interpret the figures a bit. You ask whether or not people have had to use there savings: Its established that many people on low-incomes do not have savings (since they can’t spare the money) and having no savings is an indicator of “financial exclusion.” People without savings turn more quickly to payday lenders or other lenders of last resort in such cases.

    I feel like this has been measured because I can see higher rates of payday lending in Liverpool but lower rates of using savings, however, if no one answered to “used their savings” you might regard the area as being in lower financial distress… Is that correct?

    • guyweir says:

      Hi Alec. From our survey, the financial distress measure (aka squeezometer) classifies households according to the most severe of the 5 levels outlined.

      So regardless of whether or not they dipping into savings, if they were using payday loans they would be counted at level 4.

      If they had said they were just cutting back (level 1) and no more, then they would be classified at level 1.

      So the interpretation for the saving percentages would be – x% in this area are estimated to have used savings to cover spending and are also cutting back on essentials as the most severe form of financial distress experienced in the last month.

      Hope this clarifies,


      • Alec says:

        Ah I see! 🙂 I had it in my head they were all given some equal weighting.

        Thank you very much for your swift reply.

  2. […] 2014 has been no exception. Think tanks have used lego to explain social mobility, interactive maps to plot financial distress, infographics to tell visual stories on topics from public spending to […]

  3. […] 2014 has been no exception. Think tanks have used lego to explain social mobility, interactive maps to plot financial distress, infographics to tell visual stories on topics from public spending […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: