I raised some questions about the ultimate desirability of information manipulation, and Fuhai and Xiaojian responded with a thoughtful email. They agreed to let me share part of it here:
1. Our paper consists of two parts of messages, one positive (why there is media bias), while the other normative (what is the outcome of media bias). For the first part, media bias emerges as the unique perfect Bayesian equilibrium in our model. This provides an explanation on the phenomenon we observe from reality. Our abstract thus states that "This article provides a rationale for this tendency by using a modified International Environmental Agreement model with asymmetric information." By the Longman Dictionary of Contemporary English, rationale means "the reasons and principles on which a decision, plan, belief etc is based." Our "rationale" is essentially an explanation on why the media has incentives to accentuate or even exaggerate climate damage. It belongs to the approach of positive economics and is value neutral, up to this point.
2. Then we do have a "normative" analysis on the media bias. The main difficulty of the climate problem is that it is a global public problem and we lack an international government to regulate it; the strong free riding incentives lead to a serious under-participation in an IEA. We show that the media bias may have an ex post instrumental value as the over-pessimism from media bias may alleviate the under-participation problem to some extent. (In this sense, we are close to Dessi's (2008, AER) theory of cultural transmission and collective memory.) Meanwhile, we also address the issue of trust/credibility as people have Bayesian updating of beliefs in our perfect Bayesian equilibrium. We show that, ex ante (when there is uncertainty on the state of nature), the media bias could be beneficial or detrimental, due to the issue of credibility; as a result, the welfare implication is ambiguous.