We present results on two different problems: the Lyapunov exponent of large, sparse random matrices and the problem of polymers on a Cayley tree with random complex weights. We give an analytic expression for the largest Lyapunov exponent of products of random sparse matrices, with random elements located at random positions in the matrix. This expression is obtained through an analogy with the problem of random directed polymers on a Cayley tree (i.e., in the mean field limit), which itself can be solved using its relationship with random energy models (REM and GREM). For the random polymer problem with complex weights we find that, in addition to the high- and the low-temperature phases which were already known in the case of positive weights, the mean field theory predicts a new phase (phase III) which is dominated by interference effects.