(Regular readers of this blog will know that I consider that to be a good thing.)

A father wants to encourage his daughter to take on a challenging new piece for the piano. Both father and daughter agree that it would be worth the effort if she could be expected to succeed but the daughter is pessimistic about her abilities. The father isn’t sure either but, having taught her older sister to play that piece, he knows a little more than the daughter about the difficulties involved and he is optimistic. How can he encourage her?

He has a problem because she sees through his cheap talk. She knows that he wants her to give it a try and so she knows that he would say whatever it would take to boost her confidence, even if he was really just as pessimistic as she.

However, things are different if she has reference-dependent utility. In particular suppose that, perhaps because of the way she has been raised, once her own expectations have been lifted she would stubbornly persist, even in the face of failure, for fear of falling below those expectations. (Indeed this might be why she was reluctant to take on the challenge at the beginning.)

Paradoxically, this behavioral quirk solves the father’s credibility problem. Because the last thing he wants is for his daughter to drive herself to tears in a futile attempt to learn a too-difficult piece. Now if he really thinks she’s not good enough, he won’t want to create any illusions. He wouldn’t even suggest it. So when he does suggest it and he does encourage her, she knows that he really believes in her. And he should know, so she tries.

That’s the idea in the job market paper of Edoardo Grillo from Princeton.