Variation in factor shares, extensively documented in recent years, implies that standard growth accounting exercises are plagued by measurement issues. First, the standard assumption of constant shares generates a bias in the estimation of the contribution of factors to economic growth. Second, the effect that changes in factor shares have on output depends on the relative abundance of factors and, for this reason, having correct units of measurement for all factors is imperative. We perform a growth accounting exercise that incorporates factor share variation and solves the measurement issue. Results suggest that (i) the correct units of measurement are significantly lower than standard ones for the stocks of physical, natural and human capital per worker (ii) changes in factor shares have an important effect on the growth rates of income per worker for several countries and (iii) the marginal productivity of all factors is positively correlated with per worker income.

We present an endogenous growth model where innovations are factor saving and model the choice of technologies in an Overlapping Generations framework. Markets are competitive and factor prices are determined by marginal productivity of factors; therefore, the income share of reproducible factors increases with the stage of development. Beyond the standard results of this type of model we find that (i) without bequests long run growth is not possible, (ii) if the economy presents long run growth then intra generation inequality may last forever but if the economy does not present long run growth then in steady state there is no intra generation inequality (iii) when the economy is open, the pattern of capital flows depends not only on the relative abundance of factors but also on the technologies and, for this reason, capital may not flow from rich to poor economies (iv) consistently, capital flows may not help to break poverty traps.

We consider a decentralized version of the neoclassical growth model where labor share is chosen by workers to maximize their long run (permanent) wages. In this framework, if the labor share increases relative to the competitive share, workers capture a larger share of a smaller total income in the steady-state. This is because the incentives to invest are lower and the steady-state capital to labor ratio is lower. We find that the “Golden Rule” labor share is equal to the elasticity of output with respect to labor. This is precisely what would obtain under the assumption of competitive factor markets. We also consider the model with two classes of workers: organized and unorganized. In this case, organized labor may choose a higher than competitive share and the difference is economically significant for plausible parameter values. Furthermore, relative to the Cobb-Douglas case, organized labor chooses a higher share for the empirically relevant case of an elasticity of substitution less than unity. We also analyze versions of the model with endogenous skill acquisition and capitalists with bargaining power.