[legutierr]

What you describe is more an iterated prisoner's dilemma rather than a static game. What's interesting is that it has been demonstrated in controlled circumstances that the most successful strategy in an iterated prisoner's dilemma game is tit-for-tat [1], a strategy that any participating party can choose to adopt at any time.

The fact is, corporate-labor relations is not a monolith. Although some (or even most) participants in the labor market may be engaged in betray-betray behavior today, that doesn't mean that such relationships are universal or will come to dominate the labor market over time. Given that tit-for-tat is theoretically the most successful strategy, companies that adopt generous performance-based policies with their employees should enjoy long-term competitive advantage against any rivals who have chosen less-friendly employee policies. If this model is predictive, the advantages of a "betray only" strategy for corporations will be relatively short-lived, because the inevitable betray-betray relationships will make such companies less competitive than entities that adopt a "tit-for-tat" strategy.

[1] http://en.wikipedia.org/wiki/Tit_for_tat