A fourth hypothesis, derived from the preceding one, is that the more headquar- ters outsource their most complex, unstandardized functions, particularly those sub- ject to uncertain and changing markets, the freer they are to opt for any location, because less work actually done in the headquarters is subject to agglomeration econo-
mies. This further underlines that the key sector specifying the distinctive production advantages of global cities is the highly specialized and networked services sector. In developing this hypothesis I was responding to a very common notion that the number of headquarters is what specifies a global city. Empirically it may still be the case in many countries that the leading business center is also the leading concentration of headquarters, but this may well be because there is an absence of alternative locational options. But in countries with a well-developed infrastructure outside the leading busi-