# Choosing an exchange rate in a macroeconomic panel data set

I am constructing an investor sentiment index to determine the impact of investor sentiment on stock market crises. I am following the methodology in this paper, http://121.192.176.75/repec/upload/201312051625034821.pdf.

One of the components in the index is the exchange rate. The movement of exchange rate is closely related to international capital flows. A continuous appreciation of the domestic currency attracts more demand for domestic assets from international investors, leading to a higher investor sentiment index.

I am constructing the Index for 6 different countries; U.S., UK, China, Canada, Japan and Australia.

My question is: If I follow the method used in the paper and use the end of month national currency per USD (e.g. sterling per USD, Yen per USD etc..), what exchange rate should I use for the U.S?

Or should I use a common exchange rate for all 6 countries e.g. national currency per Euro?

Any help would be greatly appreciated.

Thanks

Sounds to me like you're looking for something like the USD index, see https://en.m.wikipedia.org/wiki/U.S._Dollar_Index

As for a common exchange rate, sounds to me like you're looking for a "risk" currency which is the currency into which, if you managed FX inventories, you'd convert your trade flows. In other words using USD as a risk currency then you can choose to unwind your GBP, EUR, AUD positions to USD as they happen and then manage your long and short exposures in terms of USD rather than all currency instruments. Depending on competitive spreads that would work for managing inventories. I don't think it would really do for a sentiment index.

My guess for a sentiment index is to use a weighted index for each of your 6 currencies relative to the other 5. Also, in terms of "countries" you've missed out the EU.

In other words for EU make up an index of EURUSD, EURGBP, EURCNY, EURCAD, EURJPY, EURAUD.

For USD use EURUSD, GBPUSD, USDCNY, USDCAD, USDJPY and AUDUSD.

etc.

• Thanks for your answer, its really helpful. A couple of things: I'm not using the EU in my study, so as you said I need to calculate a weighted index for each of the 6 currencies relative to the other 5. For the USD it will be GBPUSD, USDCNY, USDCAD, USDJPY and AUDUSD. The wikipedia link for the USD index calculates a weighted geometric mean. Is there are simple way to determine the weights? – Kai_M Jun 19 '16 at 23:16
• I can't help ask why not use EU in your study? From the point of view of EUR it's just a large trading bloc. – rupweb Jun 19 '16 at 23:20
• As for the weights in an index of how different currencies / countries relate together. Hmm, try looking at trade balance figures then assign weights? For example, GBPUSD trade balance is x and USDJPY trade balance is $y so you assess GBPUSD weight as$x / ($x +$y) etc – rupweb Jun 19 '16 at 23:23
• Time and data limitations - plus the paper that I cited in my original question used mostly EU countries. – Kai_M Jun 20 '16 at 10:04
• For the weights, what you said is pretty clear, I just want to be 100% sure. For each weight the numerator is the relevant trade balance (e.g. \$x for GBPUSD) and the denominator is the sum of the 5 trade balance values. Thanks for all your suggestions. – Kai_M Jun 20 '16 at 10:10