If you dig deeper into pgpool-II you will find that it does not have failover logic. Its intention is to pool connections and distribute query load among replicas, but it cannot differentiate node failure from network partition and cannot promote a standby to master in the case of failure.