It's got nowt to do with power. With a different resolution mechanic you can easily have low-powered stuff that doesn't have this problem. It's entirely the resolution mechanic's fault.
I'm not talking about overall power level, but power
difference between characters of different levels and-or skills.
If the highly-skilled higher-level character has an 80% chance of succeeding at task X while the unskilled low-level character has a 10% chance of succeeding at the same task, the way I'm defining it that's a +70% difference between those two characters.
Unless I've been reading you wrong (always a possibility!) you'd rather see the high-skill high-stat high-level character have a 100% chance of success while the low-level low-stat unskilled guy has 0% chance; ie. a 100% difference (which is the maximum possible). In other words, you're looking for a steep power curve.
And when comparing between otherwise identical characters of adjacent level, the size of that % difference taken as an average across all levels (to account for uneven jumps, tier advances, etc.) generally indicates the steepness of the power curve built into that system.
A real quick way to eyeball this at least for combat purposes is to take a typical low-grade monster (let's say, four Orcs) and figure out the range of character levels to which they'd present a viable threat without being either pushovers or a near-guaranteed TPK; and then to take a more significant monster (say, two Frost Giants) and repeat this process. A wider range of character levels points to a flatter overall power curve.
Another quick eyeball test is how well the system handles characters of different levels running in the same party. Flatter power curve = more tolerance for level variance.
3e had a very steep curve. 1e and 5e don't.