When giving a verdict on the validity of tensile test data it is necessary to take in account the extension measuring system. It seems that you have acquired stress/strain curve directly from machine which calculates strain from position of its crossheads, which is not accurate. Then obviously when 0.2% strain will be positioned on graph it will corresponds to wrong stress value. It is advised to use extensometer in case of metallic samples and optical strain measurement in case of composites. After this step you would be in position to compare your numerical data with test data.