HiI guess this question has been posted several times before, yet I would be very grateful for any fedback.Everytime I price FX (path-dependent) exotics, I ponder which model I should use: LN, LV,or SV (the latter one being essentially a mixed LV SV model in my firm).I have read that the LV model is flawed in that it: (1) categorically underpredicts the future smile (2) assumes a 100% spot/vol correlation.Hence, given these flaws, I wonder why anyone would choose the LV model over the SV model. More specifially, I have the following questions:- what are the apparent drawbacks of the SV model?- if there are apparent differences in the price which I obtain by using LV vs. SV, are there any intuitive risks that such a discrepancy btwn the LV or the SV implies?- Given my view is that eg thee future 1mnth skew is equal to the current 1mnth, i.e. that the skew stays constant over time (for the given spot level), which model will best reflect this view? The SV model?- is there any intuitive reason why LV categorically underpredicts the future smile?- since I workin with EM currencies, some of which have massive RR ( eg IDR, PHP), is there anything in particular one should keep in mind. In particular, given these currencies with huge RR, which in my eyes imply a relatively deterministic surface, might it be more appropriate to use a LV instead of SV? It may be of note that i found that calibration of SV model to Vanillas in these currenies with highly assymetric distributions (as implied by the RR) often failsOne last thing: My current practice is that even where I see significant differences in LN vs LV/SV prices, I tend to make prices based on LN, and adjust the LN price by a 2nd order adjustment that takesinto account the smile/skew by looking at the dDelta/dVol & dVega/dVol, also in LN, (and applying the currenct value of teh Vanna & the dDelta/dVol)However, my concern is that the 2ndOrder-adjusted LN value that I obtain in this way again sometimes differs significantly from teh LV & the SV price, which essentially seems to imply I may be missing something when I look at the 2nd orders this way.But once again, why I can trade Vanna & dVega/dVol, its not clear to me how to lock in the LV price, or the SV price for that matter...Sorry for the many question, any feedback is much appreciatedCheers!However,

The paper "Stochastic Local Volatility" downloadable from this website discusses the two approaches (LV & SV) and even goes some ways towards unifying the two.QuoteStochastic and local volatility models have been regarded as alternative and competing approaches to the same unobservable quantity, the instantaneous volatility of the underlying asset. But it is only when one takes a restricted view of volatility dynamics that they appear to be different. Under more general assumptions the two approaches yield identical claim prices and hedge ratios.Also, for the pricing and hedging of FX options, this website may answer some of your questions.

Hi See also here:http://www.sungard.com/Reech/menus/docu ... ochvol.pdf