Working Paper: NBER ID: w29950
Authors: Thomas Philippon
Abstract: Growth theory is based on the assumption of exponential total factor productivity (TFP) growth. Across countries and time periods I find that TFP growth is actually linear. The additive growth model, unlike the exponential one, provides useful long-term forecasts for TFP. For the distant past the model suggests piecewise linear evolutions where the size of TFP increments changes in the late 1600’s, the early 1800’s, and around 1930. For the distant future the model predicts ever increasing increments in standards of living but with falling real interest rates and growth rates that converge to zero. The model suggests stable TFP growth in the US, but a TFP slowdown in the Euro area since the late 1990s.
Keywords: Total Factor Productivity; Economic Growth; Additive Growth Model
JEL Codes: E22; N1; O11; O3; O4
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| TFP growth is linear (O40) | TFP increases by a constant increment (F16) |
| Additive growth model provides a more accurate representation of labor productivity (O49) | predicts correct nonlinear evolution of labor productivity (O49) |
| Exponential growth model fails to capture actual data (C59) | overpredicts future levels of productivity (O49) |
| Long-term properties of the additive growth model imply that growth rates may converge to zero (O41) | standards of living continue to increase at an accelerating rate (O49) |
| Perceived TFP slowdown is a result of misinterpretation of the growth model (O41) | advocates for the additive model as a more reliable framework (C52) |