Examine the parameters for your database instance: You generate an execution plan for this query and the plan table shows a nested loop join in the plan/ After query execution, you see that a hash…

SQL plan baselines cause the optimizer to reproduce a previously captured and accepted plan, which would enforce consistent plan selection rather than explain a divergence between EXPLAIN PLAN output and actual execution.

Extended statistics improve cardinality estimates for correlated or expression-based predicates during plan compilation, but they are applied equally to both EXPLAIN PLAN and actual execution and do not cause plan divergence between the two.

With OPTIMIZER_DYNAMIC_SAMPLING set to level 11 (automatic), Oracle performs dynamic sampling during the hard parse of the actual query execution, gathering cardinality statistics not present when EXPLAIN PLAN was run; these updated statistics can change the join cost calculations enough to favor a hash join over the originally estimated nested loop.

Adaptive execution plans allow runtime switching between nested loops and hash joins based on row count feedback, but the parameter settings shown indicate adaptive plans are not enabled, making automatic dynamic sampling the correct explanation for the observed behavior.