We study how taxes and subsidies affect portfolio choices in a laboratory experiment. We find highly significant differences after intervention, even though the net income is identical in all our treatments and thus the decision pattern of investors should be constant. In particular, we observe that the willingness to invest in the risky asset decreases markedly when an income tax has to be paid or when a subsidy is paid. We investigate this result further in a range of variations of the baseline experiment and find our main result to be largely robust. However, as we reduce the number of states of nature the bias weakens considerably.