The described behavior is expected. Internally the numeric text box uses two inputs, one for the formatted value and another for editing the value. Initially the input with formatted value is displayed, once the input is focused the formatted input is hidden and the editor input is displayed.
Basically the event is attached to the editor input, so the click is fired after the numeric is focused (in edit mode). For better understanding, please refer to the following screencast:
A possible workaround is to attach the same handler to both focus and click events.