Inflated job claims don’t help policy
Sniff Test: When numbers don't stack up, policy ambitions head south
When you recognise that the only way to really increase the nation’s wealth – as measured by GDP/capita – is by increasing productivity, making the case for innovation is pretty easy. It also isn’t hard to realise that innovation is going to be about new technologies; either ICT itself or ICT-enabled.
That good case can easily be destroyed by making inflated and inaccurate claims. Not all policy advisers and policy makers will recognise them. Those that do will dismiss you. Those that don’t can wind up repeating the claim and then be embarrassed themselves.
At InnovationAus.com, we like to check these claims. That’s what we did when StartupAUS chief executive officer Peter Bradd claimed last week that “It’s a well-known fact that high-growth startups generate 70-90 per cent of new jobs.” The claim was reported by The Australian, the ABC, and Brisbane Business News.
Inquiries to StartupAUS to identify the source of this “well known fact” provided the response that it is “a well-known statistic in the industry.” It was published in the 2015 Global Women Entrepreneur Leaders Scorecard. That report does make the claim asserting “With start-ups generating 70-90 per cent of all new jobs, entrepreneurs represent the answer to that challenge.” However, it provides no source for the data. The fact that Michael Dell made the same assertion at the United Nations also isn’t a data point.
StartupAUS’s Crossroads report cites US research that finds “Three million new jobs are added to the US economy each year by new firms, while over an extended period existing firms have been net job destroyers, losing a total of one million jobs per year.”
These aren’t surprising statistics, though, if the composition of new business entities is recognised. The Australian data is contained in the ABS Catalogue No. 8165.0. Some simple stats from that include:
- Over a quarter of a million new business are formed each year (from 2010-11 to 2013-14, it was 294,210; 287,521; 239,229 and 284,153).
- 17.8 per cent of the new entities in 2013-14 were in construction, the next highest at 12.5 per cent were professional services. Others above 5 per cent were agriculture, retail, accommodation and food, transport, and rental and real estate.
- The smaller the firm the lower the likely survival rate; of firms formed in 2010-11 only 58.1 per cent of those employing one to four people survived to June 2014, 65.4 per cent of those employing five to 19, 69.3 per cent of those employing 20 to 199 and 76.3 per cent of those employing over 200.
Many startups will have been formally employed persons moving into a contract model.
The Crossroads report also states that the Startup Economy report completed by PwC “projected that high growth technology companies could contribute 4 per cent of GDP (or $109 billion) and add 540,000 jobs to the Australian economy by 2033.” This is a misrepresentation.
PwC did not start with a base case and add some policy decisions; it started with an assumed goal of 4 per cent of GDP and attempted to construct a pathway to it. It creates the circular argument that says “if we imagine high growth technology companies could create 4 per cent of GDP we need to do these things, therefore we need to do these things because the sector will generate 4 per cent of GDP.”
StartupAUS isn’t the only group to rely on data points of unknown provenance. The Australian Industry Group in a report on lifting STEM skills said “It has been estimated that 75 per cent of the fastest growing occupations require STEM skills and knowledge.”
The Chief scientist even misattributed this and claimed “The Australian Industry Group recently estimated that 75 per cent of the fastest growing occupations in Australia require STEM skills and knowledge.“
More recently this year QUT has released research, saying “The research by accounting firm PWC lists the skills of critical thinking and problem solving developed in science, technology, engineering and mathematics (STEM) education as necessary survival skills in the future workplace.” The release goes on to say “PWC says 75 per cent of the fastest growing occupations need STEM skills.”
At last we have someone claiming there is research! However the research report from PwC includes no detail to substantiate the 75 per cent claim, only a reference to an academic paper from 2011. PwC also makes the same claim relying on the same academic paper on their own website promoting Fuelling NextGen digital innovation through education.
The academic paper referred to was a study on the value of integrating the teaching of STEM, published in the Journal of STEM education. Its version of the claim is “According to the US Department of Education (2007), 75 per cent of the fastest growing occupations require significant science or mathematics training.” The cited paper is the Report of the Academic Competitiveness Council dated May 2007.
Unfortunately searches in this document for references to occupations draws a blank.
Other reports refer to US Department of Labor projections saying “According to the projections by the U.S. Department of Labor for 2014, of the 20 fastest growing occupations, 15 of them require significant mathematics or science preparation.” The only dataset that matches the description doesn’t provide the kind of sense that the earlier claims made – especially since 20 occupations is a very small sample.
The conclusion is that the 75 per cent claim simply is unsupported by evidence. That doesn’t mean we shouldn’t care about STEM skills. The argument just needs to be made better.
Governments are being asked to make “evidence based policy”, but that shouldn’t be fulfilled by using evidence of doubtful provenance and validity. Thankfully neither of the cases above take us to the territory being explored by the beauty industry and “clinical tests.” But it has been done in the policy space.
A recent egregious case emerged in the UK with reports that over half of all UK businesses didn’t understand the digital economy. In that case the survey had been prepared by a firm that advertised one of its services as “PR surveys for brand exposure - Generate content and news angles with a … survey, and secure exposure for your brand. Our survey team can help draft questions, find news angles, design infographics, write & distribute your story.”
Every claim made that refers to a piece of data really needs to be queried. Good innovation policy will follow from good evidence.