Data: (in the excel file)
Please answer all questions in the attached pages. To handle them, each group will need to download actual interbank market information from Bloomberg, for every banking business day in the last 13 years (January 1st, 2005 through January 1st, 2020):
(i) Bid and asked exchange rates (spot and 1- and 6-month forwards) for two currency pairs: USD/EUR (Euro); JPY/USD (Yen).
(iii) The 1-month and 6-month interbank bid and asked interest rates (LIBID) for the following three currencies: US Dollar and Euro.
Question 1: Covered Interest Rate Parity using USD/EUR and JPY/USD (50 points)
Over the last 15 years, has covered interest rate parity held?
1. (30 points) As a rough first pass, you could intuitively use the relevant LIBOR and midpoints of the daily Forex rate bid-ask quotes to check that, each day in your sample period and for each currency pair (for both forward delivery dates), the forward premia/discounts were approximately equal to the relevant interest rate differentials:
Please provide graphs, tables, or any other material that will help to strengthen your point.
2.(10 points) For the EUR/USD forward rates, please also proceed more formally. You should check whether the EUR/USD forward rates are in line with the spot rates, after adjusting for the relevant interest rate differentials. Do not forget to take into account bid-ask spreads for exchange rates (Bid vs. Asked) as well as interest rates (LIBOR on the ask side vs. LIBID or another deposit rate).
Hint: verify whether there is any day in your 13-year sample when either of the following inequalities do not hold:
Notes (these are hints provided by Prof Robe):
If you choose Bloomberg’s LIBID as a proxy for unsecured interbank deposit rates, you’ll find out that Bloomberg’s LIBID figures are always 1/8th of 1% below the contemporaneous LIBOR quotes. The reason is that Bloomberg’s LIBID is not an actual market rate but is instead an indicative figure that Bloomberg computes by subtracting a fixed percentage (e.g., 0.125%) from the reported LIBOR quote (formerly, that reported by the British Bankers’ Association or BBA). In other words, while LIBOR is market-based (notwithstanding the controversy about LIBOR during and immediately after the Lehman crisis, LIBID is not.
A research and policy article by two economists from the Swiss National Bank (Mancini -Griffoli & Ranaldo, Limits to Arbitrage during the Crisis, e-copy on SSRN at: http://ssrn.com/abstract=1569504) provides a detailed analysis of these issues. In particular, it identifies two main ways to exploit deviations from CIRP. One “arb” strategy relies on secured interest rates; the other, on unsecured ones. Each approach has its own problems – in particular, data issues took the authors quite far. For the case, no one is expected to go to the lengths Mancini-Griffoli & Ranaldo have gone. Here are some suggestions for a practical solution within the context of the case:
For the case, no one is expected to go to the lengths Mancini-Griffoli & Ranaldo have gone. Here are some suggestions for a practical solution within the context of the case:
a. The starting point for a first possible workaround to the LIBID issue is the observation that, during the crisis, financial intermediaries hoarded cash or invested it in the safest possible assets. Intuitively, then, a bid interest rate like the interbank bid repo rate or a governmental treasury rate might provide a better measure than LIBID of the actual rate of return on short-term bank investment (at least during the crisis period).
b. Because interbank repo bid rates are notoriously difficult to obtain, you may want to still use LIBOR on the ask side of your CIRP formulas but to replace the LIBID “rates” with a market-based substitute – for example, by using “safe” government rates (e.g., the 30- and 182-day T-bill rates in the US). Alternatively, you may want to stick to LIBID but increase the spread you compute from LIBOR during periods of crisis (in that case, please explain how you set the spread during the crisis period).
c. A natural question is when you may “safely” resort to Bloomberg’s LIBID and when you need to use another interest rate quote.
(i) One way to identify the periods of concern, might be to use the LIBOR-OIS (overnight IR swap) spread, which is a good measure of market stress.
(ii) Another way might be to use the TED spread – when the latter is high, one may worry about the relevance of LIBID.
(iii) This extra information requires additional data pulls from Bloomberg. A third, less data-intensive, approach might be to use your knowledge of the financial crisis to pick the periods when you should worry about LIBID relevance – for example, see http://en.wikipedia.org/wiki/Subprime_crisis_impact_timeline
Question 2: Yen carry trades (40 points)
An investment advisor claims that, over the past 13 years, he made a lot of money for various investors by means of Yen carry trades, using the yen as the funding currency and the US dollar as the target currency. He proposes to “do the same” for you now, using highly leveraged positions. Specifically, he proposes to execute carry trades, using 1-month or 6-month loans that he would then roll over. He claims that he is “able to obtain interbank rates for his transactions.”
Please evaluate his claims of past success. To do so, you may assume that he is indeed able to trade at interbank rates.
Suggestion: you may wish to proceed by answering the following two questions:
a. Over the last 14 years, could he have made money from carry trades, using the yen as the funding currency (borrowing at LIBOR) and the US dollar as the target currency (depositing at USD LIBID or some other measure of the USD interbank deposit rate)?
Hint: Assume that he would have been continuously doing carry trades during the past 12 years, using 1-month or 6 month loans that you roll over. Describe the instruments he would have used to implement the carry trades, the roll-over strategy, gains/losses, etc.
b. What would the risks have been? Please show your work.
Hint: For example, what would have happened to his “bet” in Fall 2008? More generally, what does the distribution of the carry-trade gains/losses look like – like a normal distribution, or something else? If the latter, what is your interpretation? Is the mean statistically different from 0?