You will never wake to the headline, "World Runs Out of Oil."

Rather, global oil production will rise, reach one or more peaks, and decline. Well before production declines to very low levels, the peak will mark a point of no return that will be a watershed in the economic history of the 21st century. For the first time, industrial economies will be forced to a lower-quality energy source. And this decline will affect every aspect of modern life.

The notion of a world speeding towards a peak in oil production was made famous by the geologist M. King Hubbert. In the late 1950s and early 1960s, Hubbert used a simple bell-shaped curve to forecast the annual rate of production in the lower 48 U.S. states. At a time when oil production was increasing rapidly, Hubbert forecast that it would peak in about a decade (1965-1970) and decline thereafter. Despite provoking nearly unanimous derision, his forecast was remarkably accurate. Oil production peaked in 1970 and declined fairly steadily thereafter. A similar bell-shaped pattern appears in several other oil producing nations, such as Norway, the United Kingdom, and Egypt.

Subsequent research indicates that Hubbert's forecast was part genius and part luck. U.S. oil production is determined by the costs of production, the price of oil, and the quantity of oil "shut in" by the Texas Railroad Commission, which aimed to stabilize prices by opening and closing oil wells in Texas to ensure a balance between supply and demand from the 1930s through the early 1970s. Had prices evolved over some alternative path or had the Commission controlled production using some other criterion, Hubbert's prediction probably would have been less accurate.

The element of luck has been overlooked by those who use Hubbert's method to forecast the peak in global oil production. Their forecasts have consistently erred, suggesting an imminent peak, only to be revised when production continued to rise after the predicted date. Hubbert's methodology cannot predict the peak in global oil production because it mistakes the price-induced slowing of oil consumption during the 1970s and 1980s for the effects of resource depletion.

The genius in Hubbert's approach stems from a simple aspect of his bell-shaped curve: relatively large uncertainties about recoverable oil supply have relatively little effect on the timing of the peak. For example, updating Hubbert's analysis through 2003 and including Alaskan production indicates that about 230 billion barrels will be produced from fields in the United States, which is nearly 30 percent more than Hubbert's original estimate of 171 billion barrels. Despite this increase, the timing of the peak "backcast" hardly changes. Put simply, compared to pessimistic assessments, optimistic estimates for the amount of oil that remains only postpone the peak slightly. Given this fact, I can confidently state that the peak in global oil production will occur in my lifetime (I am 48).

The peak in global oil production marks a fundamental change in supply. Prior to the peak, production can increase significantly with little or no increase in price. This is possible because most of the world's supply is found in a few very large fields. For example, there are more than 14,000 oil fields in the United States. Of these, the largest 100 contain nearly 40 percent of total supply. Increasing production from these large fields is relatively inexpensive. But once these large fields are depleted, they are replaced with fields that are one-tenth or one-hundredth their size. These high-cost fields reduce the profitability of production even at higher prices.

The importance of production costs is illustrated by the history of U.S. production. Oil production in the lower 48 states increased more than ten-fold between 1900 and 1970, but the real price of oil barely increased. After 1970, real oil prices doubled and then tripled. This price increase caused drilling to double. Nonetheless, production declined nearly 20 percent. As a result, the oil and gas sector increased its fraction of national investment without increasing its contribution to GDP -- in effect, hundreds of billions of dollars were flushed down a dry hole.

The economic effects of the peak go beyond spending more at the pump. Because oil readily comes from the ground and is easily refined, it generates a large "energy surplus," which is the difference between the energy obtained and the energy used to obtain it. The large energy surplus powers the non-energy sectors of the economy, such that goods can be imported and exported at little extra cost, people can live far from work, and a small fraction of the workforce can feed those that produce the goods and services we associate with modernity. All of this may change following the global peak in oil production. After the peak, each barrel of oil will require more energy to extract, leaving less to power the non-energy sectors of the economy.

No alternative fuel now being researched generates a greater surplus or can be used more efficiently than oil. This reduction in the energy surplus differentiates the peak in global oil production from previous energy transitions. As society changed from wood to coal and from coal to oil, each new energy resource was "better" than its predecessor. It could be used more efficiently and it generated a greater surplus.

This creates an additional difficulty for the inevitable transition away from oil. Alternative fuels can generate an energy surplus large enough to power the U.S. and world economies, but to do so the infrastructure for the alternative fuel needs to be larger than the current oil infrastructure. If 1 Btu (British thermal unit) of oil could be used to extract 50 Btu of new oil from the ground (which was the ratio at the U.S. peak), most alternatives currently produce 2-10 Btu per Btu invested. The infrastructure for such alternatives would need to be five to twenty-five times larger than the current oil infrastructure.

The expanded infrastructure requires a timely transition. If the infrastructure for the alternative energy source is put in place before the peak arrives, the energy used to do so will have a relatively small impact on non-energy sectors. Conversely, if society waits until the peak, constructing the large infrastructure for the alternative fuel will siphon large amounts of energy from the non-energy sectors of the economy at the very time that the total supply and energy surplus from oil is shrinking. In short, society has to pay the costs for the transition. We can pay them now, while we have oil in the bank, or we can pay them later, when our oil bank account is emptying.

Economists often assure us that the competitive market will induce the needed investments in a timely fashion. I am less sanguine. The markets' ability to anticipate the timing of the peak and the rate of decline is limited by a lack of transparency in the world oil market. Estimates from the Organization of the Petroleum Exporting Countries (OPEC) of its proven reserves are a mix of geology and politics. This uncertainty is critical because much of the oil produced between now and the peak (and beyond) will come from OPEC. As such, the market cannot know how much oil remains and therefore cannot cause prices to rise in anticipation of the peak.

The market therefore needs help to ensure that the entrepreneurial spirit will manage the transition from oil. But not the kind embodied in the Energy Policy Act of 2005. No serious person can believe that it will help. The current bill demonstrates that Republicans and Democrats have the same view of energy policy: they just give tax money to different groups. Sound policy should instead establish an economic environment that increases the economic returns and reduces the risk to long-term research and development on alternative energies. Policy should impose a large Btu or carbon tax on energy that is phased in over a long period, perhaps 20 years. This would signal entrepreneurs that there will be a market for alternative energies.

Furthermore, increases in the energy tax should be offset by reducing other taxes, such as payroll or corporate taxes. Economic studies show that such an approach can generate a win-win solution--reduce energy use (and the environmental damages not paid by users), stimulate research and development on alternative energies, and speed economic growth. Notice that the tax does not pick technologies--that will be left to the market, which is smarter than any politician (or economist!)

Government policy aimed at the next energy transition must strive for economic efficiency, but efficiency cannot be the sole criterion. The potential for large impacts may force policy makers to rely heavily on the precautionary principle (see p. 30), which compares the costs of being correct against those of being incorrect. We know that oil production will peak within our lifetime, we are pretty sure that market prices will not anticipate this peak, and we know that not having alternatives in place at the time of the peak will have tremendous economic and social consequences.

So if society does too much now to stimulate alternative energies, as opposed to later, there will be some loss of economic efficiency. But if society does too little now, as opposed to later, the effects could be disastrous. Under these conditions, doing too little now in the name of economic efficiency will appear in hindsight as rearranging deck chairs on the Titanic.

Robert K. Kaufmann is an author, a professor at the Center for Energy & Environmental Studies at Boston University and a consultant to the Japan National Oil Corporation, the European Central Bank, and the U.S. government.