Seller versus Producer concentration: incorporating the impact of foreign trade

The last decade has witnessed many studies pointing to increasing concentration of industries in the USA and beyond. Some have interpreted this as evidence of a pervasive decline in the intensity of competition. We are more cautious - in not necessarily equating high concentration with soft competition - but we do not deny that concentration remains a potentially valuable metric when tracking the evolution of markets over time. Our concern in this paper is with a particularly important measurement issue which has been overlooked in most previous studies: typically, concentration is measured, with data from business registers or censuses, on the size distribution of producers in a given industry rather than the concentration of sellers in its associated market. No account is taken of importers while exports are not subtracted from domestic turnover/production, even though, for domestic consumer choice, the former may count for a lot but the latter for nothing. The major reason for this seeming neglect is the absence of harmonised production and trade data observed at the firm level for most countries. This is certainly true for the UK, and the purpose of this paper is to present a second-best bounds approach for adjusting estimates of producer concentration into seller concentration which requires only industry-level data on imports and exports and their geographical dispersion over partner trading countries. As an illustration this is applied with striking results for a sample of 119 UK manufacturing 4-digit level, 1998-2018. We show that the main result found in most previous UK studies – a distinct upward trend in typical producer concentration – does not apply for trade-adjusted seller concentration, and the incidence of industries which would be defined as “concentrated†or “highly concentrated†using conventional anti-trust policy definitions is much reduced.