Uncertainty analysis of the Limits to Growth model: sensitivity is high, but trends are stable

Abstract

Since publication, the Limits to Growth model has received both praise and criticism. One criticism is the model’s sensitivity to input error. We have performed an uncertainty analysis to see in retrospect if the model’s sensitivity was of concern. The results show that standard deviations of output variables are high. The general trends of the variables, however, are predictable, with very similarly shaped trend lines. Trajectories indicating a favourable future for humankind (i. e., without a severe decline in population and resources) stay in areas of low probability.Uncertainty analysis is an important step in determining the reliability of a model. Models which are used to determine policies or guide decisions must be reliable to ensure sound choices are made. The Limits to Growth model by Donella Meadows and colleagues was one of the first computer models to investigate global issues of population growth and resource constraints. The model received much attention and criticism, sometimes being accused of being too sensitive to variations in input parameters. This paper studies the model’s sensitivity to input error through an uncertainty analysis, and examines if this sort of analysis could have affected the debate surrounding the model’s reliability and usefulness. Results showed that given the data used to calibrate the model, the output was susceptible to large variations, with the population variable returning a normalised standard deviation of 0.43. However, despite input error, the trends of the variables remain predictable.