Although this is a more complex model than that given in Implications of a Simple Mathematical Model for the Population Debate, it is not as complex as it might be, as it was set up quickly. It does however give several useful results.
The model looks at the population over 10 year intervals.
For the model:
A spreadsheet giving graphs and data which can be adjusted is given here - pop2.xls
a) the age band at which immigration takes place makes a difference in the longevity of shifting the dependency ratio downwards. The lower the age bands for immigration, the more long term the shift down. As my analysis of timeshift data shows, we are fortunate that existing
b) a fixed death rate (as in the assumptions) means the average levels off (contrary to the simple model). This is a hypothesis which needs testing, and adjusting, but appears to fit with my analysis of timeshift data as a rough approximation. In my model, it is actually less severe and comes in later than than the actual data would warrant.
c) a small amount of immigration per year (e.g. 200) gives a markedly significant shift in dependency ratio downwards.
d) if immigration ceases, the dependency ratio will increase to well beyond that of a no immigration policy. However, this is long term. For instance immigration of 2000 (21-30) for 20 years (200 per year), and then ceasing does not cause this to take place until 70 years have elapsed.
Population figures need to be regularly baselined against any model to improve it (see for example, my study on 1996 against 2002)