Evidence and data
Use saving rates, investment rates, depreciation, capital-output ratios, project efficiency, and capacity utilization. ICOR evidence must be checked against capital quality.
Calibration or measurement
The warranted growth rate depends on s and v. A single ICOR hides sector composition, public-investment quality, imports, and maintenance needs.
Boundaries
- Fixed coefficients rule out substitution.
- Technological progress is absent unless added externally.
- Knife-edge dynamics are a warning, not a full business-cycle model.
Use guidance
- When sufficient
- Early-development capital-accumulation arithmetic when the incremental capital-output ratio (ICOR) is treated as approximately fixed over a short planning horizon and the question is about the saving rate needed to sustain a target growth rate. The warranted-rate identity g = s/v remains a useful back-of-envelope tool for development planners comparing investment needs across projects with similar capital intensities (Harrod 1939; Domar 1946).
- When sketch only
- Do not use as a predictive model of growth or as a guide to investment allocation. The fixed capital-output ratio rules out any substitution between capital and labor as relative prices change, and the absence of technology means no productivity gains can substitute for capital accumulation. Almost every growth experience since 1960 involves a capital-output ratio that moves rather than staying fixed.
- When to switch
- Switch to Solow (theoretical:solow-growth) for any question where factor substitution, depreciation, or steady-state dynamics are relevant, which is most policy questions. Switch to an endogenous-growth model (Romer 1990, Aghion-Howitt) when innovation incentives, human capital, or knowledge spillovers are the growth driver.
- Falsification signal
- Cross-country growth data consistently show that countries with similar saving rates but different capital-output ratios grow at different rates, and countries with similar ICORs but different institutions diverge sharply. If the ICOR were a fixed structural parameter, none of those patterns would occur. The Solow residual literature since the 1960s is, in aggregate, a continuous falsification of the fixed-coefficient assumption.