When you look at national accounts, one striking fact is that the labor income share is remarkably stable in each country. And the average level of this labor income share varies quite dramatically across countries. These differences vanish to a large extend once you allocate proprietor's income (these are business owners) to labor income and capital income, but some differences remain.
Carlos Bethencourt and Fernando Perera-Tallo try to explain these differences. They build a model where workers can choose to produce or predate. If the labor income share is high, they rather produce, if it is low, predating is more interesting. With an increase in productivity, the labor income share increases, and this amplifies the impact on output as fewer people predate. This mechanism can thus make it easier to explain GDP/capita differences across countries from total factor productivity differences.
But for all this to happen, you need an elasticity of substitution between capital and labor in the production function below one, or the labor income share would not change. Empirical evidence for that is not that compelling though (hence the attractiveness of the Cobb-Douglas production function). And if the elasticity is above one, all results are reversed. But assume it is lower than one for a moment. Then it means that any attempt to improve institutions to make predation less rewarding should improve output. That would be anti-corruption campaigns, for example. Yet, I am not convinced that corruption actually lowers output. As I posted before, the empirical literature is not conclusive on this. And that empirical result would be consistent with the elasticity of substitution being close to one, if this model were not to be rejected.