The coronavirus pandemic exposes some fundamental shortcomings in the accepted methods used to estimate productivity, notably the failure to adjust for variations in the utilisation of capital. In a time of national lockdown, the consequent introduction of furloughing (workers away from jobs but still being paid) and a massive shift to homeworking, capital utilisation is expected to fall rapidly. Official measures of productivity, including those produced by the UK Office for National Statistics (ONS), have not historically taken into account variations in capital utilisation over time. In this case, Multi-Factor Productivity (MFP) appears to fall too far, since measured capital input is near constant. There is no internationally agreed method to adjust for capital utilisation; although the literature offers a number of options, none are widely accepted due to conceptual, data availability and data quality issues. We offer an extension to an existing approach of using labour hours worked as a proxy for capital hours worked, overcoming conceptual issues by matching worker types (occupations) to capital types (assets). We use data from the US O*NET database, mapped to UK occupation codes, to inform the matching of UK occupation codes to assets, then measure the hours worked of those occupations relative to a baseline in order to measure deviations in capital utilisation by asset. We also introduce a conceptual framework to apply these adjustments, noting that not all assets will be subject to variation in utilisation to the same degree. We test a number of sensitivities in the methods, including methods to construct the baseline and the degree of variation allowed for each asset. Our central estimate shows a decline in capital utilisation of around 9% in the UK market sector in the height of the pandemic, recovering over half of this by the end of 2020. This subdues, but does not eliminate, the fall in MFP through 2020.
