Promise versus TypeError for input parameters

What are the cons and pros of implementing a promise function/method in the following way? -

Return a promise only if all input parameters pass validation. Otherwise, throw TypeError.

---

I used both of the following approaches in the past...

• Simply throwing an error: throw new TypeError('Argument "bla-bla" is invalid...')
• Returning a rejection: return Promise.reject(new TypeError('Argument "bla-bla" is invalid...'))

Now I am refactoring a library for that, to make it more consistent, and want to know the good/bad sides of either approaches first.