Doesn't running 87 cut timing and result in reduced efficiency? I'd be curious if the reduction in efficiency offsets the savings per gallon? I'd guess it's probably about a wash.
I'm pretty sure the timing is only cut if the sensors detect knock/pinging so, if you're not mashing the pedal to the floor to induce that, it won't retard timing and you won't notice a difference. I regularly put both 89 and 91 in my R and I notice no difference in MPG around town nor when I go on long trips regardless of which I choose and I record all my data so I have decades of fuel usage data on all my cars to back up my average MPG.
My recent experiment has been to drive around in D rather than S which we know is a different shift map and the car short shifts like there's no tomorrow. It's possible the knock sensors are detecting ping I can't hear but the overall effect of changing to D is that my MPG has increased about 4-5 MPG around town. That's an extra 70 miles per tank at around-town speeds and, for me, an extra 9 or 10 days before full ups. That's significant and I'm running 89 lately - it's $0.20 or more cheaper per gallon.
Extrapolating that up to 65 or 75 MPH, I have a feeling (without taking the time to calculate it) the wind resistance between those speeds have more to do with MPG than octane by a wide margin.
Lastly, my historical data shows that long trips on the I-5 in California gets several miles per gallon more going 72 than going 76. On an 18 gallon tank in my Corvette, I could go an extra 70 miles or about an hour of driving more by dropping my speed only 4 MPH. Whether that's significant enough for anyone here is up to them. But for me, it was worth the slightly later arrival time and a few bucks more in my wallet.