Comments on Forbes, Fratzscher, Kostka and Straub

1
Comments on Forbes, Fratzscher, Kostka and Straub
Bubble Thy Neighbor Portfolio Effects

• Andrew K. Rose
• Berkeley-Haas, CEPR and NBER

2
Key Message

• Brazils bond taxes lower
• Brazilian bond holding
• Equity holdings in control risk countries
  (Colombia, Indonesia, Peru, Thailand .)
• Focus in this paper cross-country
• Also interesting cross-asset
• Brazilian bond taxes lowers Brazilian equity
  holdings more than Brazilian bonds! (Table 6)

3
The Good

• Good Question
• Interesting, original, topical, important
• New Data Set
• Choice of Brazil
• Good tax variation
• Important country, possibility of spillovers
• Very Careful, Thorough Work
• Frustratingly so (for discussant)
• Nice Quantitative Summary of Effects
• Most Caveats included

4
Not as Good

• How scalable is the finding?
• Interview evidence from equity investors makes it
  look like taxes should be irrelevant (makes one
  suspicious of a fishing expedition)
• Ex is three months optimal?
• Ex why use change in IOF for just bonds?
• Little motivation for externalities from
  interviews
• Is stronger coverage of equity (4x bonds) the
  reason for findings?
• Data coverage, not true effect?
• Some (mild) over-statement
• But most results shown, warts and all authors
  very aware of sensitivities

5
Nit-Picking?

• If Brazil weight falls, then something else must
  rise (possibly cash)
• Odd that effects of IOF are
• 16.6-19.7 for global equity funds
• Only 3.1-4.8 for global emerging market equity
• Only 0.1-2.1 for Latin American regional equity
• Only 5-5.1 for Global emerging market bonds
• Would have expected opposite!

6
Most Externality Estimates Small

• Ex discussion (p19) of Table 4 (key!) This
  indicates that there are no significant
  externalities from changes in the IOF on average
  portfolio allocations to all other countries in
  the sample.

7
Bubble thy Neighbor?

• Wheres the evidence of bubbles?
• If pecuniary externality with complete markets,
  effects offset each other Pareto efficiency
• Reasonable in financial markets?
• Bubble a loaded word

8
A Worry Low Effect of Benchmark

• Looks like 11 effect in Figure 1
• Very far from 1 in empirics (ß.7)
• Why so low? Measurement Error? Something that
  requires IV?
• First Iron Law of Econometrics coefficients
  biased down
• Why not impose coefficient of 1 on benchmark?

9
Referee Questions/Points

• Why start in 2006, not earlier?
• Trade Diversion analogy a red herring
• Trade diversion with discriminatory taxes,
  differential prices (RTAs are not multilateral)
• Irrelevant here!
• Box-Cox test for levels vs. logs
• If errors iid in (1), should be MA(1) in (2)
• More generally test for dynamics?
• Especially with transitory taxes?
• Is (3) necessary?
• Usually dont care about covariances between
  regressors
• Descriptive statistics in Table 3 would help

10
My Take Believable on Brazil

• Evidence of Externalities Weak
• Most direct evidence is negative why so little
  effect on bonds?
• Little in Tables 4-6
• Relatively little evidence on equities too
• Appeal to Signaling Story (Bartolini-Drazen)
• But no direct evidence of signaling
• Just a consistent interpretation