Answer from cs61c-bx (Tuan Ha 16989532) for Question 1
Since some TLB uses full associativity, and it's small, miss rate is lower than L1 cache miss rates.