IIASA

Home

About

Foreword

Introduction

Research

Arguments

In-depth Analyses

All Data

     Tables

     Charts / Figures

     Thematic Maps

FAQ

Summary

Conclusions

Resources

Bibliography

Web Links

Index

Other

Feedback

Thanks

Help

Presentation

2nd Revision

Introduction

 
Arguments - Prediction Error
Population Growth
A high (absolute) error range in total population projections for China is unavoidable. Due to the enormous population size, the projected number of people greatly depends on slight variations in fertility and mortality assumptions. To get a clear picture of China's future food demand, ongoing efforts in population monitoring and high-quality vital statistics are absolutely essential.
 
As population projections are important for estimating China's future food demand, we have to ask the following questions:
WB00860_.gif (262 bytes) Why is it impossible to predict China's population in 2025 or 2050 within a few million people?
WB00860_.gif (262 bytes) Is there anything that can be done to reduce the prediction error?
blank_3.gif (810 bytes) blank_3.gif (810 bytes) blank_3.gif (810 bytes)
 
Why are such projections impossible? Tables & Charts
It is important to understand that the high (absolute) error margin in population projections for China has nothing to do with the reliability of population data or the methodological quality of projection models. China's demographic statistics are quite detailed and sufficiently reliable (for a developing country); the methods of population projection usually applied are robust and adequate.
The real problem is the huge size of the initial population. Even if the relative error of a prediction is quite small, the absolute error in projecting the total population can be in the range of tens or even hundreds of millions.
It is of course important to run a population projection with reliable base data, but even if the base data were perfect, the large absolute error range could not be avoided. No one can predict future changes in fertility rates with the two-digit accuracy that would be necessary to minimize the absolute error to a few million people.
The 1994 and 1998 revisions of the low variant UN population projections for China illustrates this point well. In the 1998 revision, the UN Population Division used an only a slightly lower fertility assumption than in the 1994 UN projection. The projected total population, however, was more than 50 million larger than in the 1994 projection. (For details, see Chart 1 and Table 1).
In the past, these difficulties of projecting a huge population did not overly concern demographers, because they were traditionally interested in growth "rates" and other "relative" demographic measures. However, in the context of food studies, it is the absolute number of people not the growth rates of the population that are most relevant.  Real people, not "growth rates," have to be supplied with food. With the geophysical conditions of food production in China, it is critical if the population is some 50 million larger or smaller - even if for a population projection this error range is quite small in relative terms.
UN Population Projections, 1994 & 1996
Chart 1

The impact of fertility assumptions on total population projections in China
Table 1

What can be done in this situation?  
The inevitable range of uncertainty in population projections for China is of high relevance for policymaking. Decision makers must be aware that in the case of China no one can predict the total population for the year 2025 (or even 2050) within an error margin of a few million people. For food security planning it is essential that population projections are replicated frequently. Only continuous monitoring of the most recent changes in fertility, mortality, and migration trends - translated into population projections - can provide a reasonably solid database for estimating China's future food demand.  
 
   
Related Arguments

Population:   Trends     Impact    Data Quality    Prediction Error    Intervention Possibilities    Intervention Costs

 
Revision 2.0 (First revision published in 1999)  - Copyright 2011 by Gerhard K. Heilig. All rights reserved. (First revision: Copyright 1999 by IIASA.)