No, he doesn't assume that. He assumes an average rate of inflation with the mean average being above zero, because it's always been that way. http://www.usinflationcalculator.com/inflation/historical-in...

When you're talking about decades of retirement, the compound annual growth rate of inflation is all that really matters.

There isn't clear consensus on inflation/deflation in the near term. While last year saw deflation, a lot of economists think that printing money like we are will cause unusual deflation.