Discrimination After Hiring: Within-Firm Sorting, Slow Employer Learning, and the Cost to the Sorted Worker

Audit studies show that employers discriminate at hiring, and firm-sorting evidence shows that they segregate workers across employers. Less is known about discrimination after hiring, within the firm, because personnel data rarely observe initial assignment, later performance, and separation jointly. We use records from the South African Constabulary, a mounted police force whose pay was fixed within rank and class, to isolate the within-firm assignment margin. The cleaned sample contains 354 Afrikaner recruits and more than 10,000 other men enlisted between 1900 and 1908; 300 of the Afrikaner recruits enlisted after the May 1902 Treaty of Vereeniging. Afrikaners were placed in shorter contracts and lower job classes than observably similar non-Afrikaners. The wage gap disappears once assignment is held fixed. Within rank, character ratings are statistically equivalent across groups, and the early Afrikaner dismissal hazard attenuates sharply with tenure, consistent with slow employer learning. Voluntary separations peak in the second half of the first year, after the early dismissal threat has receded. The misallocation is costly on both sides: the firm loses 26,000 to 176,000 service-days, and the worker loses tenure, retention, and re-enlistment.