This is a directional hypothesis test. The bank claims P = 0.40, but the sample data = 170/500 = 0.34.

Ho: P = 0.40

Ha: P < 0.40

Z = (p - P) / Sp

where

"small" p = 0.34

"big" P = 0.40

Sp = sqrt(P*q/n) --> q = 1 - P = 0.6, n = 500

Once you find Z, determine its probability in the normal distribution tables (this is the p-value).

If it's really small (say < .05), you have enough evidence to reject the bank's claim.