Optimal firm behaviour in the context of technological progress and a business cycle (Q1202114)

This book analyses a variety of models of investment behaviour of firms. These are continuous time optimum control problems, typically finite horizon type, of increasing complexity introducing fluctuations of output demand and technological progress etc. The problems are deterministic (i.e. non-stochastic) and some are non-autonomous. The author draws attention to some distinctive features. A fair amount of effort has gone into interpretation of the costate variables, which in such problems are liable to be discontinuous, as shadow prices as in the programming literature and rates of change of the value function with respect to state variables. More importantly the model involving technological progress has complications in the constraints which makes the usual version of the Maximum Principle inapplicable. Without going into details it may look like: \[ \text{Max}_{u,s}\int^ \infty_ 0 e^{-u} F\{Q(t),\;x(t), u(t), t\}dt\tag{1} \] subject to (2) \(Q(t)=\int^ t_{\nu^{-1}(t)} G(u(\tau),t,\tau)d\tau\), (3) \(\dot x(t)=f\{Q(t),x(t),u(t),t\}\) for \(t\geq 0\); \(\dot\nu(t)=s(t)\) for \(t\geq n_ 0\), (4) \(g\{Q(t),x(t),v(t),u(t),s(t),t\}\geq 0\) for \(t\geq 0\) etc. .. \(x(0)\) is given and \(v(n_ 0)=0\). The author stresses that constraints of the type (2) and the fact that \(Q(t)\) enters as an argument in functions like \(g\) makes the problem complicated and standard versions of the Maximum Principle inapplicable. Under some assumptions regarding \(G(.)\), namely, \(G\{u(\tau),t,\tau\}=0\) for \((t,\tau)\) s.t. \(n_ 0<\tau<t<0\}\) and that there are functions \(G_ 1\) and \(G_ 2\) s.t. \(G(\cdot)=G_ 1\{u(\tau)\}G_ 2\{t,\tau\}\), he provides a ``heuristic'' derivation of an appropriate version of the Maximum Principle in Appendix 4. Under suitable concavity conditions the necessary conditions are also shown to be sufficient.