Scott Sumner argues that nominal interest rates are near zero because monetary policy (specifically expected future monetary policy) is too tight. He argues that tight (expected future) monetary policy makes expected inflation low, which makes nominal rates low. He also argues that tight (expected future) monetary policy makes real rates low as well. I want to re-state Scott Sumner's argument in terms of a standard ISLM model. I know Scott doesn't like the ISLM model, but I do. And more economists understand the standard ISLM model than perhaps understand Scott's verbal reasoning.
Scott talks about monetary policy affecting nominal GDP. I am going to talk about monetary policy affecting the position of the Aggregate Demand curve. My way of talking is both more conventional and more accurate. (Because the effect of a shift in AD on NGDP depends on the slope of the Aggregate Supply curve, unless the AD curve is unit elastic, which it probably isn't, and because price times quantity is conceptually distinct from price times quantity demanded).
If a loosening of expected future monetary policy caused the expected future Aggregate Demand curve to shift to the right, this would have two effects on the position of the current AD curve.
First, if the expected future AD curve shifted right, this would cause the expected future price level to increase, which would increase the current expected rate of inflation, which would increase the wedge between current nominal and real interest rates, which would increase current AD. There are three ways of showing this in an ISLM diagram, all equivalent. If you put the nominal interest rate on the vertical axis, a 1% (i.e. 1 percentage point) increase in expected inflation shifts the IS curve vertically up by 1%. If you put the real interest rate on the vertical axis, a 1% increase in expected inflation shifts the LM curve vertically down by 1%. If you put both nominal and real interest rates on the vertical axis, then expected inflation creates a vertical wedge between the IS and LM curves, on the right of their intersection, and a 1% increase in expected inflation increases the size of the wedge by 1%. This point is usually well-understood by any student of second-year macro, so I won't dwell on it more.
Second, if the expected future AD curve shifts right, this would cause the expected future level of real output to increase (unless the future AS curve were vertical). There are (at least) two reasons why an increase in expected future real output (real income) would shift the current IS curve right. First, an increase in expected future real output and income would increase current demand for consumption (reduce current saving), which would shift the current IS curve right. Second, an increase in expected future real output (and real output demanded) would increase the expected profitability of current investment, which would shift the current IS curve right. (A third effect would be if an increase in expected future real output and income caused current asset prices to increase, which might increase current investment and consumption.)
One way to define the current "natural rate of interest" in the ISLM model is the real rate of interest at which the IS curve intersects the natural rate of output (or "full employment", if you prefer). An increase in expected inflation reduces the real rate of interest relative to the natural rate of interest, even if the nominal rate of interest is stuck at zero. Nearly everybody understands that first point. But an increase in expected future real output, by shifting the current IS curve right, will raise the current natural rate of interest relative to any given actual real rate of interest. Fewer people understand that second point.
And it's to help people understand that second point that I have written this post.
If expected future monetary policy is too tight, and so expected future AD is too low, that lowers the nominal rate of interest relative to the real rate of interest, but it lowers the natural real rate of interest as well. That is two reasons why an expected future monetary policy that is too tight can force nominal rates down to zero, yet still leave the rate of interest above the natural rate.
The standard ISLM model can be used to exposit the conventional view that monetary policy is powerless to increase AD when nominal interest rates are zero. But the standard ISLM model can also be used to exposit Scott Sumner's "unconventional" view that the power of a misguided (expected future) monetary policy is precisely what makes (current) monetary policy appear to be powerless.