To get an accurate answer you probably won't be able to get around using a proper pricer and comparing the two methods. To contrast the two approaches:
- FX Forwards: convert all cashflows from CCY1 to CCY2 using the interpolated FX forwards, then discount all payments at a single $YTM_{CCY2}$.
- XCCY: create a cross currency swap where the paying CCY1 leg features the bond's coupon payments and receiving CCY2 leg the target coupon rate. To account for the richness/cheapness of the bond, choose the premia of swap to be $100-P_{bond}$. You could also enter this on a matched-maturity (rather than par-par) basis where you strike at the mid market rate (zero NPV). Finally, solve for the fixed rate $R_{CCY2}$ on the CCY2 leg.
The main risks will indeed be the forward points for method 1 and the XCCY basis for method 2. However, depending on the payment frequency of the bond, you might need to factor in several other curves as well, e.g. 3s6s or 3s12s basis or OIS curves. You don't need to forecast the 3m XCCY basis per se since you a have a full XCCY basis curve up to 30 years or so. This is what your pricer will rely on.
The hedge cost differential would then just be the difference $c = YTM_{CCY2} - R_{CCY2}$. You could also compare the implied basis from covered interest rate parity versus the quoted XCCY basis. Generally the two should align relatively well though.
If you want to do a comparison in spread terms then you could use method 2 and swap into a floating interest rate instead which would give you a XCCY ASW (asset swap spread). This can be directly compared to the local currency ASW (which is close enough to the Z-Spread). For the first method, I've never seen an FX implied Z-Spread being used (although it's mathematically possibly of course).
There are many books that cover these topics in detail but perhaps a bank primer such as this one is sufficient.
