The SIC-ness undermining the digital economy

Profile picture for user slauchlan By Stuart Lauchlan July 21, 2013
Summary:
A popular view of the UK digital economy is that it is small, dominated by start-ups and delivers low revenue and low employment. Not true, says an interesting new report which argues that the UK digital economy is being misunderstood by both business and policy makers thanks to reliance on a 68 year old method of measurement.

lam2
The UK digital economy?

A popular view of the UK digital economy is that it is small, dominated by start-ups and delivers low revenue and low employment.

Not true, says an interesting new report which argues that the UK digital economy is being misunderstood by both business and policy makers thanks to reliance on a 68 year old method of measurement.

2
The conclusion reached by the National Institute for Economic and Social Research (NIESR) and business tracker Growth Intelligence in the Mapping the UK's digital economy with Big Data report, is that the figures currently accepted by policy makers could be skewed by as much as 40%.

The study, commissioned by Google, suggests:

  • There are at least 270,000 companies that form the digital economy - far more than any government estimate, the latest being around 120,000.
  • Revenue reported by digital companies is growing 25% faster than that reported by non-digital companies
  • On average digital employers hire three more people - 15% more - than those employers who are not digital
  • How companies in traditional sectors, from architecture to manufacturing, are using and relying on technology to run their businesses: digital technology is no longer the sole preserve of start-ups and software companies
  • Despite the emphasis placed on initiatives such as Tech City, the areas with the highest concentration of digital companies are actually outside London and spread right across the country, in places like Aberdeen, Middlesbrough and Manchester.

Back to the war

So how come these numbers are more impressive than the official Whitehall-generated ones?

The NIESR team says it's because they've chosen not to use a 65 year-old Government classification system, known as Standard Industrial Classification (SIC) codes, but have instead tracked digital footprints left by companies online.

1
Flash back to post-war Britain. To help it draw up the right policies for jobs and growth, the UK government of 1948 took what was a very scientific approach as statisticians counted, classified and measured the economic activity of every business in the country.

To that end, they developed a set of SIC codes and the data they collected was used to shape policy in every aspect of the British economy. It was, as the NIESR report notes, Big Data before its time.

But flash forward to 2013 and this approach is outdated with a number of critical issues, argues the report:

  • There are puzzling inconsistencies in what comes under digital. ‘Software publishing’ (SIC 58290) is included, but ‘business and domestic software development’ (SIC 62012) is not. ‘Reproduction of video recording’ (SIC 18202) is included, but ‘video production activities’ (SIC 59112) is not.
  • The industrial mix is measured using large public datasets such as the Business Structure Database (BSD), which only includes companies paying VAT and/or those with at least one employee registered in the PAYE system As such, the BSD will miss out a large number of start-ups and early stage companies - many of which are likely to be digital companies.
  • SIC coding is often incomplete or uninformative in the data, or even missing. Individual businesses pick their own SIC code and are responsible for keeping it up to date. There is no clear incentive to classify correctly or to amend which means that many thousands of companies may have misleading SIC codes while anything up to 20% may have no SIC code at all.

In the report, it states:

These SIC codes are no longer adequate. Far from providing an accurate picture of the economy, they only serve to show us how much we don’t know. One in ten companies in the UK are now classified vaguely as ‘other’. One in five have no classification at all.

It is important not just for statisticians, economists and policymakers that we measure the economy accurately but for every one of us who are affected by the major economic decisions that Governments take based on that data.

Out of date?

170px-Vince_Cable,_March_2008
Vince Cable

Of wider concern is the implication that the UK is using an out-dated business classification system which in turn means that hundreds of thousands of digital companies are incorrectly identified by government and financial institutions.

With politicians, banks and insurers basing policy decision on SIC codes, it means that thousands of firms could be missing vital support, argues NIESR - and presumably therefore, Google.

Dr. Max Nathan, senior research fellow at NIESR, warns:

“Policymakers have identified the digital economy as one of the UK’s key economic strengths. That means they need to be aware of the true numbers of digital businesses around the country. The old image of tech businesses as start-ups that make no money is out of date too: using big data we show a broad array of active businesses selling digital products and services."

For his part, Business Secretary Vince Cable concedes:

"This is an interesting alternative report."

In truth the government does seem aware of the limitations of the SIC code approach. In June 2013, the Government estimated that there were 120,000 businesses in the digital economy, based on the information available in SIC codes.

But it attached a caveat to that conclusion:

We may not have an exact picture of the number of businesses in the information economy, or its employment, or the value it brings to the UK economy.

What's the alternative?

So that's the problem. Is the alternative any better?

3
Growth Intelligence says it enriches information from Companies House by gathering structured information from the Internet, as well as matching in data from other public resources (such as patents information).

Growth Intelligence classifications include 145 different sectors and 38 principle product groups, resulting in a potential classification of companies into 5510 categories compared to the possible 806 for SIC codes.

The firm says even this approach will underestimate the real size of the digital economy:

We only include companies who provide SIC codes, in order to benchmark conventional definitions of the digital economy against definitions based on Growth Intelligence’s classifications.

Second, by using registered companies, we undercount the number of business units ‘on the ground’ – since many companies have more than one office or branch.

There's a lot more information in the report on how the Growth Intelligence approach is delivered. Worth a look.

Verdict

"Statistics are like a drunk with a lampost: used more for support than illumination" (Sir Winston Churchill.)

Nonetheless, the claims made by NIESR and Growth Intelligence (and Google?) are both interesting and alarming.

I'm not picking up from Business Secretary Cable's rather understated canned comment that he's particularly inclined to do much about this.

It would be fascinating to hear how many digital start-ups and success stories concur with the conclusions of this report.

Further investigation needed.

Another quote to end: “If you can not measure it, you can not improve it.”  (UK mathematician Lord Kelvin.)