These objects of contempt are now our best chance of feeding the world

I suggest you sit down before you read this. Robert Mugabe is right. At last week's global food summit he was the only leader to speak of "the importance of land in agricultural production and food security". Countries should follow Zimbabwe's lead, he said, in democratising ownership.

Of course the old bastard has done just the opposite. He has evicted his opponents and given land to his supporters. He has failed to support the new settlements with credit or expertise, with the result that farming in Zimbabwe has collapsed. The country was in desperate need of land reform when Mugabe became president. It remains in desperate need of land reform today.

But he is right in theory. Though the rich world's governments won't hear it, the issue of whether or not the world will be fed is partly a function of ownership. This reflects an unexpected discovery. It was first made in 1962 by the Nobel economist Amartya Sen, and has since been confirmed by dozens of studies. There is an inverse relationship between the size of farms and the amount of crops they produce per hectare. The smaller they are, the greater the yield.

In some cases, the difference is enormous. A recent study of farming in Turkey, for example, found that farms of less than one hectare are 20 times as productive as farms of more than 10 hectares. Sen's observation has been tested in India, Pakistan, Nepal, Malaysia, Thailand, Java, the Philippines, Brazil, Colombia and Paraguay. It appears to hold almost everywhere.

The finding would be surprising in any industry, as we have come to associate efficiency with scale. In farming it seems particularly odd, because small producers are less likely to own machinery, less likely to have capital or access to credit, and less likely to know about the latest techniques.

There's a good deal of controversy about why this relationship exists. Some researchers argued that it was the result of a statistical artefact: fertile soils support higher populations than barren lands, so farm size could be a result of productivity, rather than the other way around. But further studies have shown that the inverse relationship holds across an area of fertile land. Moreover, it works even in countries such as Brazil, where the biggest farmers have grabbed the best land.

The most plausible explanation is that small farmers use more labour per hectare than big farmers. Their workforce largely consists of members of their own families, which means that labour costs are lower than on large farms (they don't have to spend money recruiting or supervising workers), while the quality of the work is higher. With more labour, farmers can cultivate their land more intensively: they spend more time terracing and building irrigation systems; they sow again immediately after the harvest; and they might grow several crops in the same field.

In the early days of the green revolution, this relationship seemed to go into reverse: the bigger farms, with access to credit, were able to invest in new varieties and boost their yields. But as the new varieties have spread to smaller farmers, the inverse relationship has reasserted itself. If governments are serious about feeding the world, they should be breaking up large landholdings, redistributing them to the poor and concentrating their research and their funding on supporting small farms.

There are plenty of other reasons for defending small farmers in poor countries. The economic miracles in South Korea, Taiwan and Japan arose from their land reform programmes. Peasant farmers used the cash they made to build small businesses. The same thing seems to have happened in China, though it was delayed for 40 years by collectivisation and the Great Leap Backwards: the economic benefits of the redistribution that began in 1949 were not felt until the early 80s. Growth based on small farms tends to be more equitable than growth built around capital-intensive industries. Though their land is used intensively, the total ecological impact of smallholdings is lower. When small farms are bought up by big ones, the displaced workers move into new land to try to scratch out a living. I once followed evicted peasants from the Brazilian state of Maranhão 2,000 miles across the Amazon to the land of the Yanomami people, then watched them rip it apart.

But the prejudice against small farmers is unchallengeable. It gives rise to the oddest insult in the English language: when you call someone a peasant, you are accusing them of being self-reliant and productive. Peasants are detested by capitalists and communists alike. Both have sought to seize peasants' land, and have a powerful vested interest in demeaning and demonising them. In its profile of Turkey, the country whose small farmers are 20 times more productive than its large ones, the UN's Food and Agriculture Organisation states that, as a result of small landholdings, "farm output ... remains low". The OECD states: "Stopping land fragmentation ... and consolidating the highly fragmented land is indispensable for raising agricultural productivity." Neither body provides any supporting evidence. A rootless, half-starved labouring class suits capital very well.

Like Mugabe, the donor countries and the big international bodies loudly demand that small farmers be supported, while quietly shafting them. Last week's Rome food summit agreed "to help farmers, particularly small-scale producers, increase production and integrate with local, regional, and international markets". But when, earlier this year, the International Assessment of Agricultural Knowledge proposed a means of doing just this, the US, Australia and Canada refused to endorse it as it offended big business, while the United Kingdom remains the only country that won't reveal whether or not it supports the study.

Big business is killing small farming. By extending intellectual property rights over every aspect of production, and by developing plants that either won't breed true or don't reproduce at all, big business ensures that only those with access to capital can cultivate. As it captures both the wholesale and retail markets, it seeks to reduce its transaction costs by engaging only with major sellers. If you think that supermarkets are giving farmers in the UK a hard time, you should see what they are doing to growers in the poor world. As developing countries sweep away street markets and hawkers' stalls and replace them with superstores and glossy malls, the most productive farmers lose their customers and are forced to sell up. The rich nations support this process by demanding access for their companies. Their agricultural subsidies still help their own large farmers to compete unfairly with the small producers of the poor world.

This leads to an interesting conclusion. For many years, well-meaning liberals have supported the fair trade movement because of the benefits it delivers directly to the people it buys from. But the structure of the global food market is changing so rapidly that fair trade is now becoming one of the few means by which small farmers in poor nations might survive. A shift from small to large farms will cause a major decline in global production, just as food supplies become tight. Fair trade might now be necessary not only as a means of redistributing income, but also to feed the world.

George Monbiot