15/11/2012 13:22 GMT | Updated 15/01/2013 05:12 GMT

Dear George - About Your 'Shares-for-Rights' Proposal...

Weakening employee rights is not likely to boost innovation or the employment rate (much), and will have no impact at all on unemployment and economic growth.

Would allowing companies to offer (in some cases impose) the government's 'shares-for-rights' scheme to (upon) their employees help the economy recover? Would it aid the labour market, innovation, growth? George, in your speech to the Tory party conference you suggested it would - these questions strike me as an empirical matter, so I made a maths.

[What follows is inspired by outstanding blog posts on the economics of such matters, by the likes of FlipChartRick, Chris Dillow and others, and the political implications as set out by folks like Gareth Epps].

Rick asks whether, instead of hindering innovation, employee protection might enhance it. Citing evidence from American states that have varying levels of employee protection, he points out that there was "a strong [positive] correlation between innovation and the presence of employment protection." I extended this analysis the best I could using readily-available data, and compared a global innovation index (compiled by INSEAD business school and WIPO) to OECD employee protection indicators. According to Figure 1, there appears to be a very, very weak negative correlation between employee protection and innovation score.*

Figure 1

Being a simpleton, I ran a similar comparison between employee protection and some other measures of economic activity, hoping to help answer some of the questions above. Again the comparison is between countries and pits employee protection rights against publicly available data on a few economic indicators.

Figures 2, 3, 4 and 5 show that there is no relationship between the strength of employee protection and measures of economic output (Gross Domestic Product per capita or per hour worked), or between the former and unemployment. The only measure with which employee rights was correlated (again, very weekly, coincidentally with an identical R-squared value of 0.19) was the employment rate - considering that I haven't corrected for the fact I am carrying out multiple comparisons, this relationship is likely to be non-significant.**

Figure 2

Figure 3

Figure 4

Figure 5

So, to answer our questions, according to this data, the answer is simply no: no, weakening employee rights is not likely to boost innovation or the employment rate (much), and will have no impact at all on unemployment and economic growth.

Why does all that maths matter? Because the government is proposing to weaken workers' rights in exchange for them being handed shares (of unknown/unknowable value) in the very company from which they can be fired at will. With no evidence as to how this will improve the economy (please, George, if your Treasury boffins have stats that contradict my admittedly crude analysis, let me know).

As many bloggers, journalists and politicians have fisked the proposals in more detail, I will leave my graphs to speak for themselves - from them I see very little evidence for the case to dilute employee protection, and stand by a certain Mr Vince Cable when he described the Beecroft proposals to do away with nearly all workers' rights earlier in the year: it's the wrong approach.

Thanks George, and best regards,


*The correlation is weak enough (the "R-squared value" is only 0.19, where 1 would indicate a perfect correlation between two variables and 0 means they aren't related at all) for us to question whether it is valid or relevant.

**Using my limited maths-fu I ran a simple linear regression analysis in each case, which showed there was a significant slope to the line (albeit with an R-squared of 0.19) when comparing employee rights with both innovation and employment rate. However, I know that what I should have done is a different statistical test, probably some sort of ANOVA, including a correction for multiple comparisons (to get around the fact that if we do lots of tests, one or two might appear positive by chance). I need advice on which tests would be appropriate - please tweet @prateekbuch, comment below, or email prateek [dot] buch [at] socialliberal [dot] net