The Perils Of NGDP Targeting
Whereas CPI targeting central banks would respond to a positive supply chock that lowers consumer price inflation, such as the Internet-related productivity gains and Asian crisis related slump in commodity prices in the late 1990s by inflating more and thereby create an asset price bubble, like the "dot com bubble", a NGDP targeting central bank wouldn't as the decline in inflation would be offset by higher real GDP (RGDP).
During the period of 1991-95, inflation was on average 2.4%, a number that dropped to 1.7% during 1996-2000, creating the impression, given the CPI targeting paradigm that conditions became less inflationary. By contrast, NGDP growth increased from 5% to 6.1% during that period, indicating more inflation. It is thus clear that with NGDP targeting instead of CPI targeting, the "dot com bubble" might have been prevented or at least greatly contained.
However, the coming 5 year period, 2001 to 2005 (when the housing bubble was inflated), illustrates that the NGDP targeting paradigm can be problematic too. Whereas inflation increased from 1.7% back to the 2.4% level, NGDP growth fell to 4.9%. If you based your target on the conditions of the early 1990s, then the NGDP targeting central bank would have pursued basically the same catastrophic policies that Greenspan pursued then. So in that period, NGDP targeting would have been as bad or slightly worse than CPI targeting.
What's worse, NGDP targeting would under conditions of negative supply chocks pursue even more inflationary policies than CPI targeting. One example of this is Scott Sumner's insistence that the Bank of England should pursue an even more inflationary monetary policy, despite average annual inflation of 3.5% in Britain during the last 5 years. Pursuing such policies would further aggrevate the increasing imbalances in the British economy.