On the (in)feasibility of covered interest parity as a solution to the forward bias puzzle

Pippenger's CIP-based solution to the forward-bias puzzle: A rejoinder

“…The first measure is true by definition, whereas the second is true conditional upon CIP being satisfied. Baillie (2011) and Chang (2011) both make essentially this point, though arguably they overstate their case. Specifically, Baillie (2011) considers CIP an identity and Chang (2011) implicitly also does so when describing (3) as a tautology.…”

“…Pippenger's explanation for the forward-bias puzzle prompted several, largely critical, responses (Baillie, 2011;Chang, 2011;King, 2011;Müller, 2011) to which he has recently replied (Pippenger, 2011b). However, his defence of his model does not rebut the substance of the points raised against it.…”

“…However, his defence of his model does not rebut the substance of the points raised against it. In particular, when questioning the relevance of the artificially generated data series employed in my comment, Pippenger (2011b) makes a mistaken assumption about their statistical properties and, when tackling Baillie (2011) and Chang's (2011) concerns, he focuses on challenging some assumptions that are not absolutely essential to their main point. I shall expand on each of these points in Sections 2 and 3, respectively.…”

The solution to the forward-bias puzzle: Reply

“…In my replies to Baillie (2011) and King (2011) I provide the economic theory behind my econometric solution. Chang (2011) claims that my solution is "trivial" primarily for two reasons: 1 First, as discussed above, e t+1 is deterministic and therefore cannot be treated as a stochastic error term, which of course is a necessary component of a proper econometric specification. Second, and more importantly, Eq.…”

“…That claim provoked several comments. The editor has chosen to publish four of those comments: Baillie (2011), Chang (2011), King (2011) and Müller (2011). My reply to their comments starts with the easiest issues.…”