Let's work through this like a true statistician.
For a given pair of individuals at a particular time, there is a probability p of unprotected sex resulting in a pregnancy. Thus the probability of no pregnancy resulting is 1 - p. Defining the events
A = unprotected sex results in pregnancy
B = flawed condom,
our events are independent. Therefore, the probability P(AB) of sex with a condom resulting in pregnancy is equal to
P(A) * P(B) = 0.02p, assuming P(B) = 0.02,
and thus the probability of not getting pregnant when protection is used is 1 - 0.02p. Taking the ratio of avoiding pregnancy when protected versus unprotected,
(1 - 0.02p) / (1 - p)
clearly depends on the value p. Using the value p = 0.85 quoted in another post results in only a scale factor of about 6.5, nowhere near 98. Lower, and probably more realistic, values of p will give us even a lower ratio.