Second, much of the overcapacity needs to be scrapped because demand won't recover to yesterday's levels. The bubble exaggerated demand in many industries. Automobile, IT and financial services stand out in this regard. Credit won't be as cheap in the future. Auto demand will reflect that. The industry may shrink by one-third.

The bankruptcy of Chrysler was the first step. GM may be next. Profitability in the financial sector was exaggerated by more than 100 percent at the bubble's peak, leading to outrageous capital expenditures for IT. As the financial sector shrinks and profitability normalizes, IT demand will remain at much lower levels, even as global demand recovers. Finally, the financial sector has to shrink enormously, possibly by up to 50 percent. That bubble-inspired demand simply won't be back.

The macro concept of excess capacity as a check against inflation will not work this time. Pumping in liquidity while relying on this flimsy concept would lead the global economy down a dangerous path. I expected stagflation as an end-game for the Greenspan era in 2006. It seems the world is sliding down this path without any resistance. After two decades of easy living during the Greenspan era, policymakers, markets and workers all want an easy solution -- a euphemism for a free lunch -- as a way out of an economic downturn. A free lunch may seem likely during special periods, defying economic logic. But the longer a free lunch lasts, the more pain will be inflicted during the inevitable adjustment.

Instead of overindulging in Keynesian stimuli, the world -- especially China -- should focus on reform. I have written on the need for China to shift demand from the export sector to households through wealth redistribution. For example, distributing the shares of state-owned enterprises evenly among the Chinese population would spark a decade-long economic boom. But we should also focus on improving efficiency -- a necessary condition for a new high growth cycle in China.

China has experienced high growth for the past three decades. Good policy mix was the catalyst. But that was a necessary, not sufficient, condition. Starting from a low base in the 1980s and '90s gave good policy ample room for effectiveness. The low base could be understood in terms of wage levels, urbanization or export penetration.

The last element has played the dominant role on the demand side. By convincing manufacturers to relocate while building support infrastructure, China's exports achieved a nearly 20 percent annual growth rate. The low base effect, however, will no longer apply in the future. In terms of the share of GDP in value-added, China's exports are now by far the largest in the world among large economies, although some economies have larger export-to-GDP ratios due to cross-border shipping of parts and components.

On the supply side, economies of scale linked to infrastructure network construction gave an enormous boost to productivity. The key was the so-called network effect. Building a highway improves efficiency by decreasing transportation costs along the route. Building a network squares efficiency per unit of investment by lowering the cost of transporting goods from one point throughout an entire area. China built highway, telecom and electricity networks for the first time over the past decade. As networks were completed, productivity improved enormously by lowering production costs. China's low costs were as much due to low wages as infrastructure development.

In the past, the market was puzzled by what seemed to be a conflict in China marked by strong macroeconomic performance despite poorly performing businesses (such as firms with low profitability and few sustainable assets, such as intellectual property rights). In the 1990s, most investors thought China's growth would fall apart quickly due to underperforming businesses. That didn't happen because benefits from infrastructure development spread through the economy by decreasing production costs and boosting export competitiveness. The upside from economic growth went disproportionately into asset inflation and government revenues. Corporate profits and labor income didn't benefit as much. The productivity gains from infrastructure development were critical to sustaining good macro, despite bad micro.

Productivity gains in the future will be more difficult. China's infrastructure networks have been formed. Expansion won't generate the same gains. The reason is that rising economies of scale when new networks are being built eventually turn into diminishing economies of scale. For example, if one builds a highway to alleviate congestion on an existing highway, the economic benefit is proportional to the investment. If one builds a highway next to one that's not congested, the benefit is probably less than the investment. Only when a highway is built for the first time is the benefit exponential.