A multi-country DSGE model with endogenous growth for policy analysis. The model features multiple countries, multiple sectors and multiple policy levers. The model features an international technological frontier, advanced by incremental process innovation, based on the work of Holden (2016; 2017). Product innovation absorbs long-run scale effects, while firm entry absorbs much of the impact of short run fluctuations in demand on process innovation. Slow international diffusion of productivity improvements is driven by complementarities between human capital and advanced technologies. Intuitively, new human capital is needed to understand new technologies. The model is estimated on data from 1870-2013 for six regions.
Holden (2016) produces a model capable of reconciling large medium-frequency movements with long-run trend reversion, even in the presence of varying population growth. To preserve our ability to match these facts and to enable us to capture the global nature of technology, we embed the core of the Holden (2016) model as a multi-nationally owned sector within the model we present here. The Holden (2016) model removes the long-run scale effect via duplication of process innovation over industries, with the measure of industries being advanced by product innovation. Since the measure of industries cannot respond quickly to fluctuations in demand, a further margin is needed to ensure that there is not implausibly large variance at frequency zero. In the Holden (2016) model, this is entry of firms into each industry. The model generates medium-frequency dynamics in productivity as the returns to invention are higher in good times, and endogenously new industries end up with relatively high productivity. Thus an increase in demand leads to more relatively new industries and hence higher aggregate productivity.
Growth within the aforementioned multi-nationally owned sector will drive long-run growth in all countries. This is preferable to modelling a different technology level in each country, with countries attempting to catch-up towards the maximum across countries, as that would introduce intractable non-differentiabilities (the maximum). Our approach also avoids the counter-factual implications of the alternative of making global technology some composite of national components: an improvement in scythe design from a third world nation should not have an impact on productivity in countries no longer using scythes in farming.
Despite technological growth coming from a multi-nationally owned sector, our model will still be able to generate substantial differences in growth in output per hour across countries, thanks to differing levels of human capital. While human capital accumulation cannot be the driver of long-run growth—a farmer with a PhD and a scythe is likely to be as productive as a scythe-bearing farmer without a PhD—new human capital does appear to be necessary to take advantage of frontier productivity growth. Indeed, broadly construed, human capital is precisely what is needed for adoption: to adopt a new technology, I must first understand it, i.e. I must have the relevant human capital. In our model, countries with below equilibrium levels of human capital can potentially experience faster growth than that of the technological leader, as they bring their human capital stock up to the necessary level to deal with advanced technologies. The gradual accumulation of human capital will be key to the model’s ability to generate slow catch-up in output per hour across countries.