why TCP BBR throughput still significantly affected by packet loss.

After reading this medium post [1] about BBR, where they simulate packet loss using the following command :

sudo tc qdisc add dev eth0 root netem loss 1.5%

But throughput drop from 8.55 Gbits/sec to 2.49 Gbits/sec with TCP BBR.

Since BBR is designed to respond only to actual congestion by modeling the real available bandwidth I don't understand why the throughput is reduced this much.

Is it because the model still assume the available bandwidth is this small ?

[1] https://medium.com/google-cloud/tcp-bbr-magic-dust-for-network-performance-57a5f1ccf437