Endogenous technical advance and the stochastic trend in output: A neoclassical approach

In the recent endogenous growth literature, the trend in output is stochastic, while investment in scientific knowledge is treated as analogous to a capital stock. This paper models both research and development (R&D) and disembodied technical advance using neo-classical equations, which specify the equilibrium stocks as a function of prices and output. The elasticity of R&D in the production function, estimated using factor shares, is time-varying and increasing. Returns to scale are increasing in all factors. The most important aspect of the model is that the equilibrium rate of technical advance is not cyclically invariant, but co-moves with output at business cycle frequencies. This allows the production function to behave as a stochastic trend. Growth accounting demonstrates that long-term variations in the rate of productivity growth have been associated mainly with low-frequency fluctuations in the rate of disembodied technical advance, with secondary effects from the stocks of R&D and physical capital. During the high productivity period 1948-1972, technical advance contributed 1.48 percentage points per year to output, with 0.98 accounted for by R&D. The productivity slowdown of the mid-1970s was accounted by a gradual decline in the effect of R&D, and a more abrupt collapse in disembodied technology. The subsequent productivity revival was associated with two accelerations in disembodied technical advance, the first beginning in the early 1980s, and the second in the mid-1990s. The production function is simulated in a small econometric model. Model simulations find that exogenous shocks to the inflation rate generate significant variations in growth by inducing fluctuations in disembodied technology.