A-typical year?


Atypical is an apt word for describing 2016. From the celebrity death rate to decisions at the ballot box in the UK and America that are fundamentally reshaping politics, there’s a definite sense of disruption.

And so it was in the labour market. Granted, 2016 wasn’t the year when atypical working patterns broke into the headlines. The accolades for that go to previous years: 2013 for the media-fuelled zero-hours contract (ZHC) boom, 2014 for the rapid rise of self-employment, and 2015 for the focus on tech-driven ‘gig economy’ work. But this was definitely the year when we started to ask serious questions about what atypical work really means for individuals and businesses. Think the exposure of working practices at Sports Direct and JD Sports warehouses, the Uber court case, and the launch of Matthew Taylor’s independent review of employment practices in the modern economy.

We’re used to headlines showing how these new forms of work are increasing. And it’s pretty clear that they are part of a permanent shift in our labour market, rather than a pragmatic but short-lived response to the downturn.

But the bigger issue we now need to understand is the lived experience of non-standard employment, and its effect on family incomes. When it comes to temping, agency working, self-employment and ZHCs, should we only care about the income effect of the variability of working time – not having a guarantee of the shifts or jobs you’ll do next week? For some people (not everyone) this is an acceptable trade-off for freedom and flexibility. But what if pay rates themselves are affected?

The major difficulty in answering this question (if we put aside self-employment – where the big challenge is that we have very little data on earnings at all) is that it’s hard to isolate the impact of atypical contracts themselves as opposed to the kinds of people on them and the jobs they do. It’s well-documented that atypical workers earn less than others. But we also know that they’re different in many ways. For example zero-hours workers are younger, less qualified and more likely to work in low-paying sectors. Is this what explains their lower pay?

Our new research gets us closer to answering this question. We conduct regression analysis which controls for a range of personal and job characteristics including gender, age, qualification level, occupation, industry and job tenure. We do this separately for permanent agency workers (building on our recent report on agency working), temps, ZHCs, and also part-time working (which, while not usually considered particularly atypical, is often thought to carry a pay penalty). In other words, this analysis attempts to compare the pay of zero-hour workers – for example – to that of non-ZHC workers with similar characteristics doing similar jobs.

Our results are shown in the table below, followed by our five key takeaways.


The first thing to note is that personal and job characteristics explain a large majority of the overall pay differential between various atypical workers and others. This is shown by the big reductions in effect size between the ‘raw’ and ‘all controls’ models, and the percentages in the final column. For example, the pay differential associated with temporary work falls 75 per cent – from -£3.08 to -77p – when we control for characteristics. This tells us that when we refer to the overall difference in pay between atypical workers and other employees, we’re mostly describing different people in different jobs.

Second, for part-time workers, personal and job characteristics appear to explain just about everything – the pay differential is no longer significant once we control for these. This doesn’t mean we should write the ‘part-time pay penalty’ off altogether though – evidence suggests that a significant effect endures for women but not men, something forthcoming Resolution Foundation research will explore.

Third, for the other three atypical employment types, a significant negative pay differential remains even with all our controls. The largest ‘precarious pay penalty’ is for ZHC workers, who our estimates suggest take home 6.6 per cent less than non-ZHC workers with similar characteristics doing similar jobs. For an average worker this could amount to £1,000 per year.

Fourth, we cannot be certain that the remaining pay penalty is all about working atypically. There may be other characteristics of atypical workers not captured in the data – like lower effort at work or a lack of very industry-specific skills – which explain some of the remaining difference. So more research – both quantitative and qualitative – is needed here. Nonetheless this analysis provides the clearest evidence to date that across the economy there is pay penalty stemming from these contractual forms themselves. In some cases at least, they may be being used to hold down wages for workers with less job security and lower bargaining power.

Fifth and finally, additional modelling suggests that these pay penalties are felt more acutely by the lowest-paid atypical workers. We use a technique called quantile regression to estimate pay distributions for atypical and non-atypical workers as if they had the same personal and job characteristics. Our results suggest distributions that are more unequal (once characteristics are controlled for) in each type of atypical work. For example, the ‘precarious pay penalty’ rises to at least 9.5 per cent for the bottom fifth of ZHC workers, and at least 3.8 per cent for the bottom fifth of permanent agency workers. We’ll continue to explore and interpret these results, but the suggestion is that the lowest paid are most-affected by the penalties we uncover.

Through this analysis, our wider ongoing study of agency work, and various high-profile investigations that are being undertaken by all wings of government, a picture is beginning to emerge of what atypical work really means for individuals and how it affects their living standards.

If 2016 was the year we started asking the important questions, the expectation is that 2017 will bring both answers and possible solutions.