January 4th, 2019, 3:38 pm
I would argue in the following ad-hoc fashion:
I assume that the FRA pays (L - K) where L is some LIBOR rate and K is the fixed FRA rate and I assume that the FRA is discounted with LIBOR or a rate highly correlated to LIBOR. In the case where the FRA pays in advance (2) the DF should be rather independent of L (assuming the rate process to be more ore less Markovian). On the other hand, for the FRA paying in arrears, whenever the L is high, the DF will be high and vica versa. This means when I receive something it gets heavily discounted and when I pay something there is almost no discounting. I don't like that situation so I request compensation for it, namely a lower FRA rate (K2) than I would have requested in what you call the standard case (K1) - though in advance is the market standard for FRAs. That makes K2 > K1.
Does this make sense?