Lots of variables here make this a complex topic. In a Newtonian fluid (no VI Improvers or similar large polymers), there would be a direct correlation between kinematic viscosity and HTHS viscosity. With VI Improvers, this correlation is often broken because the large molecules in the VI Improvers "flex" or compress under pressure (shear), sort of like a sponge. Therefore VI Improvers will increase the kinematic viscosity, but lose some of this thickening effect under high shear rates of the HTHS test. The amount of this molecular flexing or "temporary shear loss" depends on the quantity and type of VI Improver used, which varies from oil to oil. This is why you can find an oil with both a higher kinematic viscosity and a lower HTHS viscosity than another.
To help understand this, think of base oils as being steel balls and VI Improvers as being a rubber coating on these steel balls. Now, to make a straight 30 weight oil you would use a large (10 cSt) steel ball with no rubber coating. To make a 5W-30 oil, you would start with a smaller (6 cSt) steel ball and add a rubber layer until the final ball is the same size as the straight 30 wt ball. Both balls are now the same size (same Kinematic viscosity), but when you squeeze this balls under pressure (HTHS viscosity) they will behave differently - the rubber coated steel ball will give (lose viscosity) while the uncoated steel ball will not. How much the rubber coated ball compresses depends on the thickness of the rubber layer (quantity of VI Improver)
and the hardness of the rubber layer (type of VI Improver).
Kinematic viscosity is measured with no pressure (shear) and will give the same reading for both a straight 30 wt oil and a 5W-30 oil. HTHS viscosity is measured under high shear rates and will give a lower number for the 5W-30 oil (rubber coated ball) because it compresses the large VI Improver molecules. As the 5W-30 oil compresses in the high shear areas of the engine it becomes a thinner oil and poses less internal frictional resistance.
As for the effect of these viscosities on fuel economy, most frictional losses occur in the bearings and ring/cylinder wall interface. Both of these areas are under high shear rates, so all else being equal the HTHS viscosity
should correlate better with mileage than kinematic viscosity. Of course, all things are rarely equal, so friction modifiers, polar base oils, VI Improver quantity and type, engine shear rates, and temperature will have some influence on this correlation. Furthermore, if the HTHS viscosity gets too low, friction can
increase as parts move into elastohydrodynamic or boundary regimes (Stribeck curve), so the correlation of HTHS viscosity to fuel economy is only valid within a range.
So in conclusion, fuel economy generally correlates to HTHS viscosity, except when it doesn't.
Tom NJ