It depends on the process implied by "I'll have a number". What you'll have will at first be just $4$ balls, drawn in a particular order. You can then form a representation of a number using them, and then you can consider the number represented by that representation. You could concatenate the digits on the balls and then discard the result if it begins with $0$ because that's not a well-formed representation of a number. Or you could associate a string of digits beginning with $0$ with a number in the only natural way, by taking the $0$ to signify $0$ thousands, even though one wouldn't usually write it that way. In the first case, you could get $9\cdot9\cdot8\cdot7$ numbers; in the second case you could get $10\cdot9\cdot8\cdot7$ numbers. This is not a question of combinatorics; it's just that "I'll have a number" is not a very precise description.
I wouldn't be surprised if the actual formulation in the book is more precise. Generally speaking, if you ask questions about something in a book you don't understand or think might be wrong, it's a good idea to quote the book verbatim, since the problem may already be in your (re)interpretation of the question.