You just can't get away with saying "well, we said 4% then, but things don't look as good now as they did then." These studies incorporate periods of time back to (typically) either 1926 (CRSP data) or 1870 (CRSP supplemented by Cowles data). The claim was NOT "based on how things look in 1999, 4% should be OK." The claim was "4% includes a fat margin of safety for all economic and market conditions ever previously experienced.
Either one of two things is true:
a) Financial data is so poorly behaved, that a sample of 89 or 145 years of past data cannot yield predictions that are more reliable than "1 / chosen estimate of maximum remaining lifetime,"
b) The financial and economic outlook in the U.S. in 2015 is not just worse than 1994, it is worse than any previous time in recorded U.S. financial history since 1870. Worse than the Long Depression of 1873-96, Panic of 1907, Great Depression of 1929-1945, Great Recession of 2008-2009, worse than any of those.
My opinion is that 4% worked during the Great Depression and World War II, and 3.7% worked even during 70s where we had double-digit inflation, rising interest rates, and a stock market that went nowhere for 16 years (DOW was 1000 in 1966, and still 1000 in 1982).
So people saying that going forward we should drop to 3% or even 2.5% are basically predicting the next 20-30 years will be absolutely horrible, economically speaking... Worse than anything we've ever seen in the last 150 years.
Me, I see 1 billion Chinese and Indians joining the middle-class over the next 20 years, and all wanting to buy Cokes and washing machines.
4% wasn't generated from the good times in the past... 6% would have worked fine for many of the 30-year periods in the last 100 years... 6% may actually "probably" be safe... (i.e. more than 50% of the time, 6% works)
And on the other extreme... If 2% isn't safe, then no one is safe... If 2% doesn't work, then we're in a world where money doesn't matter (Mad Max times).