Comments on:
The Impact of Currency Carry Trade Activity
on the Transmission of Monetary Policy

World Finance & Banking Symposium, Vilnius

Clinton Watkins
Akita International University, Japan

December 2023

Summary

  • Carry is a pervasive factor in currency markets.

    • A deeper understanding would be useful for investors & policy makers.

  • I enjoyed reading the paper.

    • Valuable and robust contributions to the literature.
    • Thorough analysis, logical and well justified.
    • Complex but the authors explained their approach well.
    • Well written.

  • I have a few comments and questions.

Overview

  • A simplification of the approach and results (apologies for any misrepresentation!).
flowchart LR
  subgraph a [VAR model]
  A("FOMC Monetary Policy <br> Announcement <br> (↑ policy tightening)") 
  A ==> B("Conventional <br> Monetary Policy <br> Shock <br> (↑ FF, ↓ SP500)")
  A ==> C("Information <br> Shock <br> (↑ FF, ↑ SP500)")
  B ==> D("↑ USD")
  C ==> E("↓ USD*")
  end
  subgraph b [Threshold VAR model]
  D ==> F("↓ Typical Carry Investment Currency")
  F === G("Depreciate more when involved in the <br> carry trade (high NOI), <br> amplified by flight to safety.")
  D ==> H("↓ Typical Carry Funding Currency")
  H === I("Depreciate less when involved in the <br> carry trade (low NOI), <br> offset by safe haven demand.")
  E ==> J("↑ Typical Carry Investment Currency")
  J === K("Appreciate significantly when involved <br> in the carry trade (high NOI), supported <br> by positive risk taking sentiment. <br> No change if not investment currency.")
  E ==> L("↓↓ Typical Carry Funding Currency*")
  L === M("Depreciate when involved in the <br> carry trade (Low NOI). No change when <br> not used as a funding currency.")
  end
  style a fill:#eaeee7
  style b fill:#d2d6cf

Monetary policy pre/post 2000

  • The sample covers some quite different periods in monetary policy:

  • Are the results robust over these different periods?

  • VARs are interpreted regarding positive FOMC policy shocks. Are the dynamics symmetric regarding negative policy shocks?

  • Are there implications for the shock types and dynamics when policy is below/above the neutral rate?

Figure 1a: Exchange Rates and Interest Rate Differentials.

Red: 12/2003-12/2006, blue: 12/1998-12/2000.

Monetary policy pre/post 2000

Federal Funds Effective Rate 12/1998-12/2000 and 12/2003-12/2006

Monetary policy versus information shocks

  • The authors determine whether an FOMC announcement delivers a conventional monetary policy shock verus an information shock using the reaction in 3-month Fed Funds futures and the S&P500 index, drawing on studies such as Jarociński and Karadi (2020).

    • Monetary policy shock: ↑ FF (yield) and ↓ SP500.
    • Information shock: ↑ FF (yield) and ↑ SP500.

  • The FF reflects an interest rate reaction to for specific maturity at a specific time in the future, while the stock reaction (theoretically) looks forward unbounded by maturity.

    • Yield curve slope can change in response to monetary policy announcements, e.g. higher 3-month and lower 2-year yield when a tightening cycle is expected to peak.

  • Consider a term structure of interest rate reactions to take full account of interest rate expectations at different maturities?

    • Using a policy rate expectation in three months for the monetary policy shock and capturing everything else (information shock) via stock prices may be convenient but it seems an artificial construct. Are rate expectations adequately modelled?

  • Is the post FOMC stock market move a reaction or merely execution of orders delayed until after the FOMC (as a risk event)?

Regime 1, Regime 3…

  • The authors use net open interest (NOI) as a proxy for carry trade activity (in each non-US currency where CFTC futures positioning information is available).

    • A reasonable proxy for carry exposure given the lack of alternatives.
    • However, does not necessarily represent carry as a motivation to trade.

  • Under regime 1 (3), non-US curencies have a relatively low (high) NOI reflecting that they are funding or short (investment or long) currencies against the USD.

  • At the same time, some parts of the paper note that regime 1 reflects low carry trade activity and regime 3 reflects high carry trade activity (in general).

    • I may be confused, but does not regime 1 (3) reflect high carry trade activity for typical funding (investment) currencies?

    • Table 4, Panel B shows the results for when the foreign currencies are in the funding regime (e.g sell JPY and buy USD) and is presented as when there is low carry trade activity.

      • Wouldn’t Panel B reflect the period of high carry trade activity for low yield currencies like the JPY?

Regime 1, Regime 3…