It's because internal econometrics are resulting in perverse incentives.
Defensive security is a cost center without clear, deterministic metrics for success. Let's say you pay X on defensive security (which is an oversimplification when you're talking about a cultural change, but that cultural change involves people learning how to pay attention to security, and paying attention is a form of man-hours, for which a cost can be calculated). If you don't get attacked, is it because the X you paid is high enough to deter/foil attackers, or could you have paid less and achieved the same result? If you are attacked and the attackers get past your defenses, is it because the X you paid wasn't enough, or if you had spent more, would the attackers have succeeded anyway, because of their relative power and motivation? For defensive security, it's very, very hard to justify to bean counters that X was the correct amount of money spend, no matter what the real outcome is, because it's hard to understand X's affect on that outcome.
Pentests which result in tickets/issues/etc. are much easier to justify. The company spent X on the pentest, and it got Y feedback in return. Simple, and effective, at least in the short-term.
It's part of the overall challenge that organizations face when they become metrics-driven. People choose the path of least resistance, so if you ask people to measure data, they'll measure the data that's easiest to measure. Data that's harder to measure - culture and social attitudes - becomes "not a priority" to measure.