The Long-Run Effects of Psychotherapy on Depression and Beliefs∗

We revisit two clinical trials in India that randomized depressed adults (n=775) to a brief, inexpensive course of psychotherapy or a control condition. Four to five years later, we find substantially lower rates (12 percentage points) and severity (0.15 standard deviations) of depression in the treatment group. The control group underestimates the impacts of the effective intervention, but experiencing therapy increases beliefs about treatment efficacy. We investigate the effects of therapy on self-confidence regarding ability at a paid work task, testing contrasting predictions of depressive realism versus self-protective overconfidence. We find that therapy made people’s beliefs about themselves more accurate by reducing their tendency to react more to positive than negative feedback, thereby reducing overconfidence. ∗We dedicate this paper to our coauthor Bhargav Bhat, who passed away during the course of this work. We thank our entire field team in Goa for excellent research assistance, especially Siddhant Gokhale, Advait Rajagopalan, Gunjita Gupta and Anahita Karandikar. We thank our research managers at the Behavioral Development Lab, Adrien Pawlik and Erik Hausen. We thank Benedict Weobong for his early support of our project. We thank all our study participants for their time and patience. We received IRB approval from Princeton. The experiment was pre-registered on the AEA registry, number AEARCTR-0003823. †Bhat: Sangath; de Quidt: Institute for International Economic Studies, Stockholm University, CAGE, CEPR, CESifo, ThReD (Jonathan.deQuidt@iies.su.se); Haushofer: Stockholm University (johannes.haushofer@ne.su.se); Patel: Harvard and Sangath (Vikram_Patel@hms.harvard.edu); Rao: Harvard and NBER (grao@fas.harvard.edu); Schilbach: MIT and NBER (fschilb@mit.edu) Vautrey: MIT (vautrey@mit.edu)