Glass Ceilings or Sticky Floors? Statistical Discrimination in a Dynamic Model of Hiring and Promotion

Abstract

I show that when two groups differ in (i) their average skill level, (ii) the precision with which they can signal their skill prior to entering the labour market, and/or (Hi) the frequency with which they have the opportunity to signal their skill prior to entering the labour market, then even if firms become increasingly informed regarding each worker's skill over time, equally skilled workers from different groups will have different likelihoods of making it to top jobs in the economy, even though there is no discrimination when it comes to promotion to these top jobs. In 1995, 95% of senior-level managers in the top Fortune 1000 industrial and 500 service companies in the US were male, 97% of whom were also white (Federal Glass Ceiling Commission, 1995). In the UK females make up just 8.3% of the senior judi ciary, 9.7% of top business leaders, and 9.1% of newspaper editors (EOC, 2005). At the end of the 2004 American football season, only 5 of the 32 NFL head coaches were black, and only 2 of the 117 Division I College football teams had black head coaches (Rosenburg, 2004). In 1999, there were only 68 female full professors at the 120 PhD granting economics departments in the US (CSWEP, 1999). This severe underrepre sentation of females and minorities at the highest levels of occupational achievement throughout a wide range of occupations has often been referred to as the 'Glass Ceiling'.1