In the video, he states:
What is self-esteem? It is a word (actually it is two words). It is sort of something you think you might have or not have. But, that's a figure of speech. That's not an empirical phenomenon.
Thus, by stating "self-esteem does not exist", he is implying it cannot be empirically observed, i.e., it is not something that can be observed reliably and systematically.
From there on out, he starts arguing why this is so. Some key points:
- Reported self-esteem measures are pretty appalling.
- You can already model the concept of self-esteem with the Big 5. I presume he is implying here the concept is superfluous and implications of 'low or high self-esteem' are modeled better using Big 5.
- The introduction of the concept has lead to undesirable consequence of advocating for self-esteem. But, there is no reason to believe that increasing your self-esteem is a good idea. "It is not at haste that you should have a good opinion of yourself in every bloody situation."
He continues to give examples for these points, e.g., "self-esteem might become disconnected from the actual accomplishment", but I believe this is outside of scope what you are asking for here.