Why?
Because real signals are often not sine and bandwidth does not abruptly stop at 200MHz. There are details like small ringing, etc that will be washed out with too low sampling rate.
Same for bandwidth: more bandwidth == less amplitude distortion (and more picked noise). And scopes normally have more bandwidth than advertised.
So, I don't see the reason why there "must" be any relation between bandwidth and sampling rate. Each parameter independently improves capturing. What makes sense to me is to have a "balanced" solution in terms of BW, sampling rate and price.
Interestingly enough, oscilloscopes of the past used to have a different BW / Rate ratio. Like, a 500MHz scope with just 1Gsa/s. Not anymore nowadays. May be because it was not a sweet price point, or that scopes relied on "equivalent" sampling.
Bandwidth mean sinewave bandwidth of course.
If people have NOT sinewave signal what base frequency is 200MHz and he use 200MHz BW oskilloscope and he think or wants he must also see these harmonics or other over 200MHz components in signal he is just telling he need more lessons.
It is essential to understand basics before start using tools. Even if tool is hammer or oscilloscope or particle accelerator.
200MHz oscilloscope is not for measure or display right at all even 1Hz signal if it means example 1Hz square wave and need measure its frequency components up to example 1GHz for measure its rising edge.
Oscilloscopes mostly have gaussian type BW shape.
Perhaps 200MHz nameplate sscope -3dB point ( input BNC) is example something between 200 to 250MHz. -6dB perhaps 230 - 280MHz or something like this.
It is not first time when I tell that instead of hack scope for more wide BW it is perhaps more wise to modify these for rejected BW specially so that BW shape is more like brickwall type than slow decay gaussian. Just for avoid aliasing what is many times lot of under estimated and poorly understood thing.
And this must say. 500MSa/s 200MHz oscilloscope is really easy with signals what have over 250MHz frequency components in signal under test and it leads aliasing and this is math, it is not oscilloscope brand. User need understand how to avoid these traps and not be fooled by scope screen or displayed numbers in these cases. This need knowledge and experience (as all things need).
Also due to Sinc interpolation parameter compromises some not so nice things start more or less below
fNyquist.