"Never prophesy, especially about the future." That nicely captures the perils of predictions - so nicely, indeed, that the saying or a version of it has been credited to numerous people, from the movie mogul Sam Goldwyn to baseball's Yogi Berra.

But in practice, we do have to prophesy, however imperfectly. Take climate change, an issue that involves assessing what could happen decades ahead, and how to respond to it. Or take defence planning: despite the difficulty of forecasting the nature of future conflicts, decisions have to be taken today that will affect how wars are fought for years to come (the F-13 Joint Strike Fighter, for example, the most expensive defence-industrial project ever, is planned to be a mainstay of American and Western air forces until at least 2060). Similarly long horizons are involved in planning for our energy needs and our pensions.

So we need to look at the long term. Where to begin? A good place to start is with population trends - which is why this is the subject of the first of the 20 essays brought together in Megachange: The World in 2050, a book published by The Economist this month.

The world's population is changing very fast. It took 250,000 years for it to reach 1 billion, around 1800. The latest billion, taking the number of people on the planet to 7 billion, took just a dozen years (a landmark the United Nations said was reached last October). By 2050 the global population will have risen to a little over 9 billion, according to the UN's central projections.

And by then the global population will be older (the median age will rise from 29 to 38) and more urban, with nearly 70% living in cities and towns, compared with just over 50% today. It will also be more African: about half the extra 2.3 billion people on the planet by 2050 will be in Africa. In 1950 Europe accounted for over a fifth of the world's population, and Africa for a tenth; those proportions are on their way to being reversed. By 2050, there will probably be nearly as many Nigerians (close to 400 million) as Americans.

Very broadly, from the point of view of population patterns, the world will fall into three groups between now and 2050. The first consists of younger-than-average countries where the share of the economically active population relative to that of dependent children and elderly will be very favourable. These countries will potentially enjoy a 'demographic dividend', if there is enough productive work for their large numbers of working-age people (or they could face instability if jobs are scarce). In this group are India, the Middle East and Africa.

In the second category of countries are those where the average age is rising, but not by much, and where the share of the working-age population relative to the young and old is deteriorating, but only modestly. The United States is in this group, as are Latin America and South-East Asia.

The third group - and the big losers from the demographic changes in the next four years - includes Europe, Japan and China. Japan will be the oldest society ever known, with as many dependents as people of working age. And China, thanks not least to the legacy of its one-child population, will start to age rapidly. By 2050 its population will be older not only that America's, but even than Europe's. China really is in a race to grow rich before it grows old.

All this has big consequences: for the economy, business, security, migration, health and the demands on resources, not to mention for culture and social change. It should inform many of the policy decisions taken today. The sooner we start preparing for the coming demographic changes and all that flows from them, the better our long-term prospects will be.

Megachange: The World in 2050 is available now.