Twitter barometer tells true cost of snow on London

Much has been said of the Federation of Small Business (FSB) estimate that the cost to the capital of freak snow shutting down business yesterday is running at £1bn a day. Investigation of their estimate shows a crude finger in the air assumption that 20% of the UK’s 32m workforce are laying dormant. I thought this deserved some looking into…

Twitter is increasingly being used by marketing types to figure out what the hell the populace is saying and subjects “trending” on Twitter often give a fairly clear picture of what is important today. It seemed reasonable to use Twitter as a cultural barometer of how many people were affected by the snow yesterday, and so calculate an improvement to the FSB’s estimate of the true cost to the economy.

The concept

People on Twitter tweet about being off work sick. We know from the Confederation of British Industry (CBI) how much it costs the economy when people are off sick.

People also tweet about being off work because of the snow. Assuming they’re pretty much as likely to tweet in either case, we can use the volumes of tweets about being off work to figure out our estimate of the cost of the snow.

The results

The method is detailed below, but for those interested in the headline figure, this estimate puts the cost to the UK economy of yesterday’s snow at:


The method

We equate these two ratios to get at the cost estimate:

cost per day due to work closed / cost per day due to sickness


tweets per day about work closed / tweets per day about being off sick

Finding the number of tweets about work being closed

Using it is possible to track back through time to find out how many people in the UK were tweeting using particular words. I searched for tweets on Monday 2nd Feb that were within 500 miles of the UK and contained the words “work” and “closed”.

I went through the list to find the relevant tweets. Any tweets about delays to work I counted as one hour lost time i.e. 1/8th or 0.125 of a day. I ended up with the figure that the number of lost days on Monday according to Twitter was:

41 7/8 or 41.875

Finding the number of tweets from people off sick on a given day

Using a similar method to above, I searched for tweets in the 10 days before yesterday that contained “off” and “sick” and were considered to be within 500 miles of the UK. I would have searched a greater time period, but the search service returned no results before Friday Jan 23rd.

A manual scan allowed me to find out how many people over this period had taken a day off sick, or mentioned one of their colleagues as being off sick (I assumed only one tweet would happen per person off sick).

From this, I calculated the mean and standard error range of the number of people per day off sick as seen through Twitter as:

2.2 (0.37, 4.03 3 s.f.)

The cost per day of people being off work due to sickness

The Confederation of British Industry put the cost in 2007 of 172m sick-days at £13.2bn. That makes the cost due to sickness in one day:

£36.2m (3 s.f.)

Putting it together

Using the initial formula and the above statistics, the cost estimate is given by:

cost per day due to sickness * tweets per day about work being closed / tweets per day about sickness

= £36.2m * 41.875 / 2.2 (0.37, 4.03)

= £690m (£4.1bn, £380m 2 s.f.)


  1. Posted February 3, 2009 at 6:28 pm | Permalink

    My opinion is that rather than costing millions the snow is really a net positive £ benefit to Britain. Let me explain.

    Forcing everyone to work from home gives time to reflect and get a little perspective. To avoid the continual churn of meetings and conference calls and commuting and actually take some time out to think. You’re well and fit so unlike being sick you can actually think. And the snow helps to change perspective as the familiar becomes new and exciting. I expect there’ll be a large increase in changes to the way teams work over the next week. We’ll never know!


  2. Martin Budden
    Posted February 3, 2009 at 11:02 pm | Permalink

    I’m with Nick. You calculation takes a very presenteeist viewpoint. It neglects the fact that many people managed to work at home and the fact that all those parents who spent the day playing with their kids in the snow will be more productive all week. Then there’s the fact that many people will put in extra hours during the week to make up lost time.

    My guess is the net effect to the economy is about zero.

  3. Posted February 4, 2009 at 1:20 pm | Permalink

    Actually, I’m with you both on this. I think the benefits of enforced different work patterns are myriad.

    However, I think it is an interesting exercise to compare the estimates obtained using this Tweet-based method and the FBS’s method.

    @martin – as it turns out, I factored in the fact that people might be working from home by not including tweets from people who said they were. In a sense, that makes this method less of a sledge-hammer approach than assuming 0% productivity from 20% of the workforce.

One Trackback/Pingback

  1. […] UK had a shit load of snow, which pretty much took out the whole country on the 2nd, a Monday. I wrote about a method I’d used to calculate the economic cost to the UK based on the Twitter chatter as a […]