Measuring Permanent Responses to Capital-Gains Tax Changes in Panel Data