The movie industry is an extreme case, but it turns out that the production of almost everything is concentrated in just a few centres.
The pattern of geographical concentration is most pronounced in the US. For example, two-thirds of car and auto parts production takes place in the mid-west, mainly in a few cities in Michigan, such as Detroit and Flint.
But the management guru Michael Porter has detailed many of the geographic clusters or concentrations in particular industries. He analyses, for example, the development of four industries since the Second World War - the printing press industry in Germany, patient monitoring equipment in the US, ceramic tiles in Italy and robotics in Japan.
When you start to think about this pattern of geographical specialisation, its prevalence becomes mesmerising. Next time you are in the supermarket, and are buying some everyday household item, look to see where it is made. Toothpaste? Ireland or Germany. Nappies? United States or France. Bacon? If not home produced, Denmark. There is only one big centre - or at most two or three - for anything that can be traded.
And so we have Hollywood (movies), Seattle and Toulouse (aircraft), City of London (financial services), Paris and Milan (couture) and Detroit and Wolfsburg (cars).
It was almost a century ago that the British economist Alfred Marshall noted that most industry was concentrated in specific districts: cutlery in Sheffield, cotton in the Manchester area and lace in Nottingham, not to mention coal in Newcastle.
Marshall also set out a fundamental economic explanation, which the economics profession ignored for the next 100 years, only recently returning to it. Paul Krugman at MIT and Brian Arthur at Stanford are among the pioneers of the modern version of economic geography.
As some of these researchers spell out, the world is becoming more like the one Marshall described. The forces for geographical concentration are becoming stronger as the economy becomes increasingly weightless - that is, as the goods and services with greatest economic value-added increasingly take a non-tangible form. There has always been only one Hollywood. Now there is also only one Microsoft, only one Novell, and before long there might be only one Citibank.
Marshall saw three explanations for the existence of geographical clusters: they have a pool of skilled labour for companies to choose from and a range of companies for workers to apply to; they create a network of suppliers and specialised services; and they foster the exchange of ideas and information, encouraging technical progress. Krugman andArthur have translated Marshall's arguments into the language of modern economics.
Geographical concentration is the result of what economists call increasing returns to scale, or what an engineer would describe as positive feedback. This means that the more a company does of something, the better it does. Expensive set-up costs, for example, generate economies of scale. A new aircraft engine costs $2bn-$3bn to design, develop, test and put into production, but the second one costs only $50m-$100m. Or there can be increasing returns in the marketing and distribution of a product: the bigger it gets, the better McDonald's does because customers know the brand and know what they are getting.
The most important kind of positive feedback in marketing exploits network externalities. This term just means that a product becomes more beneficial to its users the more other people are using it, because it sets a common standard. There are countless examples of one product squeezing another out of the market entirely because of network effects - for example, the VHS rather than the Beta standard for video cassette recorders, the spread of the DOS operating system for computers, the use of petrol or gasoline to power car engines.
Increasing returns or positive feedback create the basic virtuous circle that leads to the concentration of production in a relatively few places. Companies want to be in the place where there are the workers to make their product and customers to buy it.
Transport costs make a difference to the basic tendency towards geographical agglomeration. If they are too high, it will discourage an industry from clustering in one place.Equally, a reduction in transport costs can often permit greater concentration in one place - a fact very relevant to weightlessness, where transportation costs are nearly zero.
New technology will disperse this kind of geographical concentration. The view that modern computer and communications technologies will shift work out of cities iswidespread. There are certainly some examples of big companies dispersing their workforce in this way. Telephone operators, for example, are unlikely to sit together in one big building in the centre of town. Some sit in offices far removed from the big cities they mostly serve. Some work at home. Technology has made home working possible in many new areas.
But it is wrongheaded to predict that technology marks the end of economic concentration in urban centres. If production can occur anywhere, it might as well take place in the places where it has already started. The increasing weightlessness of economic activity is dramatically increasing the scope of increasing returns to scale and decreasing transport costs. Both tend to favour more rather than less clustering. There is therefore huge scope for more geographical concentration in many parts of the entertainment business or the media, and in professions where advice is increasingly being delivered via technology rather than in person.
Suppose Oxford University were to admit on-line students, delivering lectures and tutorials via computer and video links. It would capture a big share of the market for online studentships, and the university would grow. Already a big employer of academics, it would hire more and more of them. The city of Oxford would become even more of a centre for the education business than it already is.
An increasing proportion of business in the weightless world is becoming subject to increasing returns. Many of the products or services of the weightless economy are interdependent. They are used in groupings - computer operating systems, software, modems, telephone cables and mobile phones, for instance.
An equally important force is the prevalence in weightless economics of network externalities, the benefit that users of a product derive the more users there are, and infinite expandability. It is feasible for one company to supply a global market - you only have to think of Microsoft, supplying a huge proportion of the world market for operating system software for personal computers.
In addition, the cost of hardware and cabling, and the perceived importance of access to it, is leading to the wiring of our cities at an astonishing pace. Public buildings such as libraries and cyber-cafes have become the information superhighways. The information technology revolution, far from dispersing work, will make the new economy more urban than ever. And movies like The English Patient will continue to be a rarity.