Optimal Policies in the Dasgupta-Heal-Solow-Stiglitz Model under Nonconstant Returns to Scale