Demographic forecasting in an age of uncertainty

The observation by French philosopher and essayist, Paul Valéry, that the ‘future is not what it used to be’ has never seemed more prescient than in today’s turbulent times.

Although natural demographic change is, by definition, a slow process – simplistically, people are born, gradually move from place to place, grow older and eventually die – there are many ‘disruptors’ to this process which mean that the future never turns out quite as one thought it might.

These disruptors come in two main forms.

The first is events

In the last few years, we have witnessed several key events or exogenous shocks which could not have been foreseen, for example, Brexit, COVID-19, wars in Ukraine and Middle East or the cost-of-living crisis.

Unlike gradual natural demographic shifts, these events have the potential to cause sudden, unexpected and sometimes long-lasting impacts on demographic patterns.

This might be manifested as a temporary or sustained shift in birth rates, a change in movement within the UK, altered patterns and extent of international migration – and its effect on the age-sex structure of the population – or fluctuation in the number of excess deaths.

The second is uncertainty

The second disruptor is more intangible: the factors which shape our expectations, and in turn forecasts, but about which there is inherent uncertainty.

Models of demographic change draw on what has happened in the past to make assumptions about what might happen in the future. But what if those assumptions, or the data underpinning them, are wrong, contested or simply out of date?

Mitigating uncertainty is an important rationale for scenario modelling.

In the UK, there is no real-time universal population register constantly recording the number and whereabouts of all individuals. Population is counted through a decennial Census.

In between, population and its movement between areas or countries must be estimated. Official intercensal estimates can have flaws; equally, population counts have limitations. The 2021 Census was a contemporary snapshot, but one taken at a uniquely unrepresentative time.

Reviewing the evidence

Edge Analytics therefore advocates constant monitoring, review and evaluation of the latest evidence, revising assumptions and forecasts accordingly. There is never one correct answer for what the future will look like, but rather a range of plausible outcomes considering alternative input assumptions.

To conclude with another apposite quotation – “When the facts change, I change my mind.” – John Maynard Keynes