The paper examines the labour quality explanation of the employer size?wage gap: larger firms pay higher wages because they employ more skilled workers. Most previous studies control for unobserved skills of workers using longitudinal data and the fixed effects estimator thus relying on a questionable assumption of time-invariant unobserved individual heterogeneity. This paper releases this assumption by using a sample of workers who simultaneously hold two jobs; hence, identification is achieved by differencing across two jobs held at the same time rather than in different periods. A caveat of this approach is possible heterogeneity of the two jobs; this issue is discussed in details in the paper. Based on data from the UK Quarterly Labour Force Survey, this study finds little support for the labour quality explanation: controlling for unobserved skills in the sample of moonlighters does not reduce the estimate of the wage gap.