LEAVING one's home to settle in a foreign land requires courage or desperation. No wonder only a tiny fraction of humanity does so. Most migration takes place within countries, not between them, part of the great procession of people from country to town and from agriculture to industry. International migrants, defined as people who have lived outside their homeland for a year or more, account for under 3% of the world's population: a total, in 2000, of maybe 150m people, or rather less than the population of Brazil. Many more people--a much faster-growing group--move temporarily: to study, as tourists, or to work abroad under some special scheme for a while. However, the 1990s saw rapid growth in immigration almost everywhere, and because population growth is slowing sharply in many countries, immigrants and their children account for a rising share of it.

Counting migrants is horrendously difficult, even when they are legal. Definitions vary. Some countries keep population registers, others do not. The visitor who comes for a holiday may stay (legally or illegally) to work. Counting those who come is hard, and only Australia and New Zealand rigorously try to count those who leave. So nobody knows whether the rejected asylum-seeker or the illegal who has been told to leave has gone or stayed. But the overall picture is one of continuing growth in the late 1990s.

Between 1989 and 1998, gross flows of immigrants into America and into Europe (from outside the EU) were similar, relative to population size. About 1m people a year enter America legally, and some 500,000 illegally; about 1.2m a year enter the EU legally, and perhaps 500,000 illegally. In both America and Europe, immigration has become the main driver of population growth. In some places, the effects are dramatic. Some 36% of New York's present population is foreign-born, says Andrew Beveridge, a sociology professor at Queens College, New York: "It hasn't been that high since 1910," the last peak.

America at least thinks of itself as an immigrant land. But for many European countries the surge of arrivals in the 1990s came as a shock. For example, the Greek census of 2001 found that, of the 1m rise in the population in the previous decade (to 11m), only 40,000 was due to natural increase. "In a decade, Greece has jumped from being one of the world's least immigrant-dense countries to being nearly as immigrant-dense as the United States," notes Demetrios Papademetriou, co-director of the newly created Migration Policy Institute in Washington, DC.

Asia too saw a burst of immigration in the 1990s, propelled initially by the region's economic boom. Foreign workers accounted for an increasing share of the growth in the labour supply in the decade to the mid-1990s. Chris Manning, an economist at the Australian National University in Canberra, reckons that foreign workers made up more than half the growth in the less-skilled labour force in Malaysia, perhaps one-third of the growth in Thailand and 15-20% of the growth in Japan, South Korea and Taiwan.

What makes all these people move? In the past, governments often imported them. In Europe, migration in the 1950s and 1960s was by invitation: Britain's West Indians and Asians, for example, first came at the government's request. Britain's worst racial problems descend from the planned import of textile and industrial workers to northern England. Now the market lures the incomers, which may produce less disastrous results.

Three forces often combine to drive people abroad. The most powerful is the hope of economic gain. Alone, though, that may not be enough: a failing state, as in Somalia, Sri Lanka, Iraq or Afghanistan, also creates a powerful incentive to leave. Lastly, a network of friends and relatives lowers the barriers to migrating. Britain has many Bangladeshi immigrants, but most come from the single rural district of Sylhet. Many host countries "specialise" in importing people from particular areas: in Portugal, Brazilians account for 11% of foreigners settling there; in France, Moroccans and Algerians together make up 30% of incomers; and in Canada, the Chinese share of immigrants is more than 15%.

Most very poor countries send few people abroad. Immigration seems to start in earnest with the onset of industrialisation. It costs money to travel, and factory jobs provide it. That pattern emerges strikingly from a study by Frank Pieke of Oxford University of emigration from China's Fujian province. He describes how internal and overseas migration are intertwined. Typically, a woman from a family will go to work in a factory in a nearby province, supporting a man who then goes abroad and probably needs a few months to find himself a job.

Net immigration flows continue as long as there is a wide gap in income per head between sending and receiving countries. Calculations by the OECD for 1997 looked at GDP per head, adjusted for purchasing power, in the countries that sent immigrants to its rich members, and compared that figure with GDP per head in the host country. In all but one of its seven largest members, average annual income per person in the sending countries was less than half that of the host country.

Migrant flows peter out as incomes in sending and host country converge. Philip Martin, an economist from the University of California at Davis, talks of a "migration hump": emigration first rises in line with GDP per head and then begins to fall. Migration patterns in southern Europe in the 1980s suggested that the turning point at that time came at just under $4,000 a head. In a study for the European Commission last year of the prospective labour-market effect of EU enlargement, Herbert BrÜcker, of Berlin's German Institute for Economic Research (DIW), estimated that initially 335,000 people from the new members might move west each year, but that after ten years the flow would drop below 150,000 as incomes converged and the most footloose had gone. Net labour migration usually ends long before wages equalise in sending and host countries.

Migrants do not necessarily come to stay. They may want to work or study for a few months or years and then go home. But perversely, they are more likely to remain if they think that it will be hard to get back once they have left. "If you are very strict, you have more illegals," observes Germany's Ms SÜssmuth.

There has always been a return flow of migrants, even when going home meant a perilous return crossing of the Atlantic. According to Dan Griswold of the Cato Institute, a right-of-centre American think-tank, even in the first decade of the 20th century 25-30% of migrants eventually went home. And where migrants are free to come and go, many do not come in the first place. There is no significant net migration between the United States and Puerto Rico, despite free movement of labour. "It's expensive to be underemployed in America," explains Mr Griswold. But in Europe, with its safety net of welfare benefits, the incentives to have a go are greater.

Tougher border controls deter immigrants from returning home. A book co-authored by Douglas Massey of the University of Pennsylvania, "Beyond Smoke and Mirrors", describes how in the early 1960s the end of a programme to allow Mexicans to work temporarily in America led to a sharp rise in illegal immigrants. Another recent study, published in Population and Development Review, also links tighter enforcement to a switch from temporary to permanent migration. Its author, Wayne Cornelius, says the fees paid to coyotes, people who smuggle migrants, have risen sharply. He found that, when the median cost of a coyote's services was $237, 50% of male Mexican migrants went home after two years in the United States; but when it had risen to $711, only 38% went back. And, whereas the cost of getting in has risen(as have the numbers who die in the attempt), the cost of staying put has declined, because workplace inspections to catch illegals have almost ceased. The chance of being caught once in the country is a mere 1-2% a year, Mr Cornelius reckons. So "The current strategy of border enforcement is keeping more unauthorised migrants in the US than it is keeping out."

Tighter controls in Europe are probably creating similar incentives to stay rather than to commute or return. A complex, bureaucratic system designed to keep many willing workers away from eager employers is bound to breed corruption and distortion. And the way that rich countries select immigrants makes matters worse.

Chart to follow with future update.

PHOTO (COLOR): Innumerable