Investment in infrastructure now covers a multitude of things from bridges to roads, ports to railways, provision of power supplies to "social" projects involving schools, health care and waste services, says Michael Carapiet, executive director and joint head of Macquarie Bank corporate financing.

The market capitalisation of 400 stocks of companies that deal in infrastructure projects stands at an aggregate $2.4 trillion, he told the audience at the Macquarie Asian Infrastructure Forum, organised in Hong Kong last week by AsianInvestor and FinanceAsia magazines.

Building in the fact that these companies also raise debt as liabilities, it means that the amount of assets that these companies control is somewhere in the region of $5 trillion. Macquarie Bank believes this could double in the next three years. Furthermore, an equivalent volume of infrastructure assets are probably held in unlisted companies.

Infrastructure today is still a localised business. An international infrastructure provider, operator or financier is running an essential service for one locality, which must be tailored to the vagaries of the government, municipality and community.

ôGovernments used to want to privatise infrastructure when they were broke,ö Carapiet says. ôThis is no longer the case and they recognize that the private sector can provide services more efficiently. Some governments might place a certain sector off-limits, but there are plenty of other opportunities available. They want infrastructure to be provided smoothly, and are less concerned about external providers making a profit than in their ability to deliver the better services they promised. ö

That willingness must partly be due to the sheer volume of cash needed to pay for these mega projects. The World Bank estimates that globally, $400 billion needs to be spent annually on new infrastructure each year. Whilst international money can get the ball rolling, a lot of the money has to be found domestically for operators and providers dealing specifically at a local level.

In Korea, for example, there are 130 privately owned infrastructure companies with $50 billion invested. ôIn Korea another $200 billion needs to be spent on transport infrastructure,ö says Carapiet, ôA lot of that will be private money.ö

In China, Macquarie has counted over one hundred privately owned toll road companies, predominantly local corporations. Carapiet says the firm has been offered 20% minority stakes in such companies on many occasions, and whilst they have passed - Macquarie's model is to actively own and control infrastructure companies in its portfolios - it still thinks these are interesting opportunities for passive investors. Carapiet adds that to go into mainland China infrastructure without a local partner is "madness".

Many private toll roads in India are in preliminary discussion stage, perhaps ultimately to be developed with the backing of a government guarantee scheme. Woeful infrastructure in India is one of the principal restraints to IndiaÆs economic growth but this means huge investment opportunities loom.

Elsewhere, infrastructure has become a well-respected asset class only recently. The concept of infrastructure as an asset class has only taken root in the past two years in the United States and continental Europe. Yet in Australia, there almost seems to be more money for infrastructure than there are projects to soak it up.

Internal rates of return on infrastructure deals tend to follow a gently descending flight path in the low double digits, as projects progress through the typically defined lifecycle of greenfield, to ramp-up and finally into maturity. At the same time, increasing number of interested investors will bid up asset prices and ultimately hurt returns.

A recent entrant into the industry is private-equity players but Carapiet believes most lack the right model to remain serious players. ôPrivate equity firms might have a 25-30% target, but they still seem to be comfortable in the range that infrastructure can provide,ö says Carapiet. ôThey might only have a three-to-seven year horizon yet these types of operations can go on for half a century.ö