Cauchy distribution has no well defined expectation (and if we took the principal value of the corresponding integral, it would be finite)... Does this count as an 'unbounded mean'? I'm not sure what alternative hypothesis to finite expectation was assumed in the question...