The reason is that while VFD is modern and sounds cool, it is a really poor substitute (which is true for a lot of modern electronic gizmos).
Like all machines, an electro motor has an ideal range of operational speed. Electro motors are admitteldy much closer to "perfect" machines than for example combustion engines (which cannot run slower than about 800 RPM at all, and only start to deliver serious torque at about 3,000 RPM), but they still aren't perfect. There is still a "right" and a "wrong" way to use them.
For drilling, you will usually want net speeds from as low as 100 (possibly 50) RPM up to about 3,000 (possibly 4,000) RPM, and you want this speed as steady as possible, with no vibrations, sudden decelerations, locking or other undesirable artefacts.
While torque in an electro motor very favourably starts -- even at very low speeds -- with an almost perfectly straight, high plateau up to about 4000 RPM (after which it slowly decreases to about 50% at 8,000 RPM), things are not nearly as ideal when it comes to current, power, and power efficiency.
Electro motors have their peak power plateau between approximately 4,000 and 12,000 RPM. Before that, power output (unsurprisingly, being the product of torque and rotational speed) goes up linearly with rotational speed. Power input, on the other hand, is exceedingly high on very slow rotating motors and even goes down as output goes up.
Efficiency increases with rotational speed following a "kind of" negative exponential saturation curve reaching a plateau of around 70-95% depending on the quality of the motor (actually the curve goes down again at high speeds, but for "reasonable" speeds we may assume that it looks like a saturation curve).
A motor running at 20% of its maximum speed is usually less than half as efficient as a motor running at 60% of its maximum speed.
That suggests that a motor should preferrably run faster (indeed, much faster) than needed -- ideally near the end of its torque plateau -- and then have the speed reduced mechanically to whatever is desired. It may lose a small amount of torque if it runs too fast beyond the max-torque plateau (gaining efficiency in return), that torque will however be regained in excess from the speed reduction, and it will be much more power efficient. Which also means less heat and less wear on the motor.
The starting current, which may be forbiddingly high on a powerful motor, is significantly reduced, and rotational speed is a lot more constant in presence of any environmental effects that might occur (inhomogeneous material, bit locking, power grid fluctuations, whatever).
A speed-reduced motor is therefore win-win in every respect, compared to a slower turning motor.
Now, why belts, and no gear? Gear makes a lot more noise, and it's more expensive, too. Belt-driven drill presses are silent enough to be used without hearing protection (I rarely ever recommend not using hearing protection with any kind of machine, but I can lead a normal conversation with my drill press running without being disturbed in any way).
Add to that the fact that variable speed electronics are delicate circuits which may (will) eventually fail whereas a mechanical belt has hardly a way of failing.
Before someone yells out: "Sure, a belts can rupture!", as an anecdote from my life, I got my drill press from a bankrupt tool manufacturing company 24 years ago (the drill was used for making tools, they didn't make drill presses). It had been in use for around 20 years at that time already, but apart from obvious usage marks it was in good shape, and it's still working with the same belt, same everything today.
In the very worst case, a ruptured belt is easily replaced, too, whereas broken VFD logic means "trashcan" to anyone who doesn't have a degree in electric engineering.
Belts provide the desired mechanics as efficiently and silently as possible, they cost next to nothing, and they practically last forever.
