Reduced-form tests of scale effects in markets with search, run when aggregate matching functions are estimated, may miss important scale effects at the micro level, because of the reactions of job searchers. A semi-structural model is developed and estimated on a British sample, testing for scale effects on the offer arrival rate and the wage offer distribution. When contrasting London with the rest of the country we find scale effects in wage offers. But the larger market delivers higher realized wages and not more matches, because the scale effects on matches are offset by the response of reservation wages.