Network-dependent dynamics of innovation and production

Complex networks have been demonstrated to play an important role in many economic systems. This thesis aims at further integrating the science of complex networks with economic theory with a focus on dynamic aspects of two types of economic networks: technology/innovation and production networks. Technological evolution is often described as a recursive process whereby the recombination of existing components leads to new or improved technological components. While this idea implies a network structure of technological interdependencies, very little has been done to establish empirically that these interrelations help predict future innovation dynamics. We propose a simple model of network-dependent knowledge creation that we calibrate to patent data. We find that simple time series models yield fairly good predictive performance and represent a good benchmark for alternative forecasting models. Incorporating information on the network of patent citations, however, can substantially improve patenting forecasts. The second part of this thesis studies the dynamics of macroeconomic variables using production network models. We propose a macroeconomic model which we calibrate to the UK economy to quantify sectoral and aggregate impacts of the Covid-19 pandemic. Beside the production function assumption and the estimation of first-order shocks, we find that inventories and input bottlenecks are crucial in economic prediction. To better understand the driving forces behind these predictions, we study simpler dynamic input-output models. We suggest a simple mathematical optimization procedure that allows us to determine lower bounds of shock propagation. We find that the minimal shock propagation can be substantial, but much smaller than when compared to decentralized rationing behaviors of firms. In particular, the estimated economic impact strongly depends on the density of the underlying production network.