Birthday Attack against Cryptocurrency

One of my friends is creating his own cryptocurrency, just as a fun project, and he made some design choices that I think are insecure, but I personally don't have enough expertise to evaluate.

Assuming we're considering an secp256k1 private/public key pair and a Bitcoin style address (just a hash of the pubkey), would it be feasible or even possible to attempt a birthday attack against this system if the hash (let's assume sha256 for this example) was truncated to a 14 byte output?

Bitcoin truncates to 20 bytes by running RIPEMD160 on top of sha256, which would put a birthday attack in the realm of 2^80 attempts, but if I understand correctly, a keypair could be generated corresponding to this 14 byte address in 2^56 operations, is that correct?

That requires about $2^{57}$ hashes (and little memory), which is extremely feasible nowadays. The cost when it is wanted collision among hashes of public keys with known private keys is higher, because we need $2^{57}$ EC-multiplication-and-hash; but that's still feasible.