A firm hiring new workers has a choice: it can either hire permanentquot; workers, which means entering into a quot;life-time employmentquot; (long term) contract with them, or quot;temporaryquot; ones, who can be hired and fired according to current needs. We assume that the latter are less productive and/or more expensive per unit of output and ask what is the optimal composition of the labor force under these assumptions. Extending Pindyck's (1988) model of irreversible investment, we show that the optimal proportion of permanent workers is high if the firm is either large, faces high demand growth, operates in a reasonably stable environment, or uses a labor intensive technology. Previous studies suggest that tenure is indeed higher in larger firms in the US and Japan and that reliance of Japanese firms on permanent workers increased throughout the high growth 1950s and 60s, and declined thereafter. However, these findings do not distinguish between ex-post permanent workers, i.e. those who stayed with their firms for a prolonged period of time, but who could have been laid-off if needed, and the ex-ante permanent workers, who were promised lifetime employment (LTE). We, therefore, look for complementary practices, which would likely be adopted alongside LTE, and not otherwise. Our review of the existing literature indicates that Japanese labor practices adopted during the early 1950's, when LTE was introduced, were complementary to it. We also find evidence that at least some US and European non-unionized firms adopted practices complementary to LTE. The adoption of these practices is consistent with the model's results relating LTE to the environment in which these firms operated. Finally, we examine the cost of the lifetime employment contract relative to the optimal mix of lifetime and temporary contracts, i.e. the magnitude of the value of flexibility and show that under certain circumstances the adoption of LTE for all the workers can have a significant negative impact on the value of the firm
