Forecasting behaviour of exchange rate after the annouce of switching from peg to float
- Thread starter Oussab
- Start date
- Joined
- 8/28/08
- Messages
- 339
- Points
- 53
Not really. Whatever model you choose would rely heavily on your own assumptions with respect to jump size and volatility after jump. Normally the OTC FX forwards market will point to where the market expects the currency to jump to in the case of a known float in the future. FX options with expirations linked to the forwards post float would measure the volatility the market expects the currency to follow after the jump occurs. In a perfect world, the FX forward contract would already perfectly encompass the jump in spot, so the volatility of the forwards should be jump-less, so to speak, and would be a guide for how the market expects the currency to behave once spot is floating. This isn't totally true, though, because part of the current volatility of the forwards may be in figuring out where the currency will settle if the jump size is really hard to know. Your job will be even more difficult because I don't believe FX options on morocco vs anything trade at all so volatility data may not exist. I'm not sure about forwards, hopefully forward point data or outright forward data on those exist. I would first study the case of EURCZK, which is a more liquid case of what you are describing. Spot is pegged (well, floored) now, but the central bank has said it will let it float in the not too distant future. FX forward and volatility data will exist for you to look at and play around with. Once you get comfortable with that case, I would then make assumptions about how much more volatile you'd think the moroccan currency (don't know what it's even normally crossed with) will be than EURCZK by looking at the current spread in volatility between EURCZK and a currency more similar to the moroccan currency that is more liquidly traded today.
- Joined
- 1/27/10
- Messages
- 1,037
- Points
- 173
I'm not sure I agree. EURCZK shows clear regime change. MADUSD much more volatile generally, but looks more-or-less stationary. Is it MADUSD you're studying?
- Joined
- 8/28/08
- Messages
- 339
- Points
- 53
What I meant to say is that EURCZK will have different dynamics than EURMAD, say, once it is floating, because CZK has a different economy, trading parters, and central banking from MAD - so simply looking at EURCZK without making some adjustment for it being a different currency cross would be incomplete. The level of volatility of EURMAD will just be different from EURCZK, so you have to make a reasonable adjustment up or down from EURCZK volatility levels in making your estimate. I was suggesting trying to find a reasonable EUR/CCY proxy that is currently liquidly traded, which you feel would more closely resemble what you think EURMAD given CCY's more similar dynamics to MAD than CZK. Then you could make an argument that after the jump, the volatility will be EURCZK+/-EURCCY volatility.
In any case, I apologize for not having a deeper background in MAD. I just looked it up and it seems it's more of a soft peg in the form of a managed basket of 60% EUR and 40% USD. That's more similar to what Russia had for RUB, which was a managed basket of 45% EUR and 55% USD, which they abolished in 2014. Russia abolished it in part because the shift in the price of crude oil is a more important factor in its terms of trade than either of those other two currencies, and the weakening of the price of oil meant Russia would have had a really hard time managing its currency reserves under a soft peg. The move really just meant oil became a stronger factor in determining currency movements, as well as political moves like US sanctions following the invasion of Crimea, so RUB simply became a more volatile currency but without really any crazy jump to speak of. Morocco is going to be a bit different. First of all, the it looks like 75% of its foreign denominated debt is in EUR and the balance mostly USD, and its export market shares look similar. That would usually mean a good basket for them to be smoothing is 75% EUR and 25% USD, and it did used to do that with a basket of 80% EUR and 20% USD. However, it moved to its current 60%/40% weighting to allow for a global factor from the big USD move. The important thing is that they've already done that as a step toward free floating. I haven't looked into the trend in central bank reserves - I'll leave that to you - but barring a large trend of declining FX reserves, I would estimate that the volatility of MAD after the free float will look fairly similar to the volatility of MAD today. It's fairly intuitive to say that the volatility purely due to a free float will result in a greater than or equal to volatility of the currency today, but the effect shouldn't be enormous given its largest two components of EUR and USD are already in the price. However, if there has been a large trend in FX reserves declining even after the basket was revised to 60%/40%, then you could see a faster depreciation in the MAD and a significantly higher level of volatility going forward. No stochastic model will help you find an answer to this sort of question. You can try looking up complex macroeconomic models for forecasting floating exchange rate volatility following a managed float, but I wouldn't put much faith into that either - if literature even exists on the subject.
In any case, I apologize for not having a deeper background in MAD. I just looked it up and it seems it's more of a soft peg in the form of a managed basket of 60% EUR and 40% USD. That's more similar to what Russia had for RUB, which was a managed basket of 45% EUR and 55% USD, which they abolished in 2014. Russia abolished it in part because the shift in the price of crude oil is a more important factor in its terms of trade than either of those other two currencies, and the weakening of the price of oil meant Russia would have had a really hard time managing its currency reserves under a soft peg. The move really just meant oil became a stronger factor in determining currency movements, as well as political moves like US sanctions following the invasion of Crimea, so RUB simply became a more volatile currency but without really any crazy jump to speak of. Morocco is going to be a bit different. First of all, the it looks like 75% of its foreign denominated debt is in EUR and the balance mostly USD, and its export market shares look similar. That would usually mean a good basket for them to be smoothing is 75% EUR and 25% USD, and it did used to do that with a basket of 80% EUR and 20% USD. However, it moved to its current 60%/40% weighting to allow for a global factor from the big USD move. The important thing is that they've already done that as a step toward free floating. I haven't looked into the trend in central bank reserves - I'll leave that to you - but barring a large trend of declining FX reserves, I would estimate that the volatility of MAD after the free float will look fairly similar to the volatility of MAD today. It's fairly intuitive to say that the volatility purely due to a free float will result in a greater than or equal to volatility of the currency today, but the effect shouldn't be enormous given its largest two components of EUR and USD are already in the price. However, if there has been a large trend in FX reserves declining even after the basket was revised to 60%/40%, then you could see a faster depreciation in the MAD and a significantly higher level of volatility going forward. No stochastic model will help you find an answer to this sort of question. You can try looking up complex macroeconomic models for forecasting floating exchange rate volatility following a managed float, but I wouldn't put much faith into that either - if literature even exists on the subject.
Im studying the expected impact of the floating regime on FX options (usd mad ; eur mad)
- Joined
- 8/28/08
- Messages
- 339
- Points
- 53
Agreed. I hadn't realized MAD wasn't really pegged but was instead a managed basket of EUR and USD, which is obviously why it's more volatile than a proper floor like EURCZK. RUB is therefore a better test case.
- Joined
- 8/28/08
- Messages
- 339
- Points
- 53
Russia abolished its fairly similar managed basket (as described in my previous post) on 10-Nov-2014. Why not have a look at the before-and-after of EURRUB and USDRUB option implied volatility and FX spot and forward realized volatility? All that data is readily accessible on a Bloomberg terminal. Like I said before, though, that effect will be muddied by the price of crude oil plunging and various political headlines raising RUB volatility as well.
- Joined
- 8/28/08
- Messages
- 339
- Points
- 53
You can't, really, as you'd have to know what came out when which is nearly impossible to track now. The good news is that a lot of the political stuff had already been in the works - the annexation of Crimea happened in Q1 2014, well before 10-Nov. Headlines were still moving back and forth at that point but the messy stuff had already happened.
Really the biggest difference would have been the pressure the CBR felt on its FX reserves due to the price of oil declining. You can try amending the time series of onshore spot EURRUB and USDRUB by stripping out regression beta moves in spot due to the change in the first future oil price, and then calculating the realized volatility of the amended series before and after the shift, I suppose. I'm not a time series analysis expert. Or you could just forget about the whole oil thing due to the price drop having been mid-way through its decline by the time they abandoned the basket - but it does look like a lot of the drop took place after the basket ended, especially if you consider percent changes rather than absolute differences (percent is what matters).
With respect to transaction volumes being different... I don't know. Maybe there is some literature comparing the differences between FX volatility in pairs with differing average daily volumes. I haven't encountered this sort of problem, never having really dealt in 'frontier' currencies. I would only expect this problem to matter if the flow of FX reserves out of the BKAM has been truly 'one-way'. If the BKAM has been draining its FX reserves consistently over time (again, especially since the re-weighting of the basket) to maintain stability of its currency relative to the basket, then I would be worried about the low transaction volume for this particular exercise. The reason being that if the central bank has been the sole liquidity provider in buying MAD, then where does MAD settle once the central bank stops buying? On the other hand, assuming the BKAM still has an adequate level of reserves once they move to a free float, it's likely that the BKAM would continue to buy MAD to slow the pace of its depreciation. Unfortunately these sorts of mechanisms aren't model-friendly. Also, I have no idea what the FX reserve situation is at the BKAM, so that's something you'd have to check.
The problem is that , the BKAM will not implement the floating regime directly it'll pass over a process before pure floating regime , BKAM will first widen the boundaries between which currency will be floating so BKAM will still be present on the FX market to ensure that the currency is between the boundaries . But in my internship the work that i should do is to evaluate the impact of an eventual floating regime on the FX options , so i'll have to simulate the path of the exchange rate
- Joined
- 8/28/08
- Messages
- 339
- Points
- 53
Sure, this is a very common approach for a central bank. Singapore and China both do this. Options on those currencies trade liquidly via the USD without much regard to the widened boundaries, as pressure tends to build on trade weighted terms and the boundaries are floating from the floating basket components which are built from those trade weighted terms. China is a bit more complex given it has needed to drain FX reserves heavily, but Singapore is a fairly benign example of this at work.
Simulating the path of the exchange rate once you determine what inputs you want to use is the most trivial part of this exercise. The hard bit is how to figure out what inputs to use. You can get as fancy as you want here. You can use models with higher order characteristics like vol-of-vol, spot-vol correlation, jumps with whatever frequency and size, etc. The point here is that the model will just spit out what you put into it rather than using the model to imply dynamics from known volatility surfaces, so its really more a matter of choosing a model (I'd suggest going simple) and figuring out what inputs for the variables you'd like to use within that model.
- Joined
- 8/28/08
- Messages
- 339
- Points
- 53
Well, the most important inputs will be incorporated within a Gaussian framework. You need to create a sensible forward curve to be the expected drift of the pair. And then you need to determine what volatility the process should follow. Those are the two most important things. From there, if you want to move into a stochastic volatility kind of framework, I would estimate the correlation between spot and volatility, as well as the volatility of volatility and incorporate non-zero inputs for those two things into your model. I wouldn't get any more complex beyond that.
- Joined
- 8/28/08
- Messages
- 339
- Points
- 53
Sure, SABR sounds good.
- Joined
- 8/28/08
- Messages
- 339
- Points
- 53
I think you're misunderstanding me. I was just recommending that you look at what happened to EURRUB and USDRUB currency volatility after Russia lifted a similarly managed currency basket. It's the closest comparable you've got, and it's recent and has a lot of available data around it. Obviously they aren't the same country and their economies are different, but the way their central banks managed their currencies is similar.
- Joined
- 8/28/08
- Messages
- 339
- Points
- 53
No, I'm not a professor. There is a lot of data available on a Bloomberg terminal. If you don't have one, I'm not sure.
