An Experimental Component Index for the CPI: From Annual Computer Data to Monthly Data on Other Goods
指出CPI成分指数因忽略退出商品（多为价格下降的过时商品）而产生选择偏差，基于Pakes（2003）提出新的特征价格方法，在三个CPI样本中显著修正了该偏差。
The CPI component indices are obtained from comparing price quotes at a given store in different periods. If we omit comparisons from goods in the store in the initial, but not in the comparison, period we generate a selection bias: goods that exit are disproportionately obsolete goods that have falling prices. Building on Pakes (2003), we explain why standard hedonic predictions for second-period prices of exiting goods do not account for this bias. New hedonic methods are derived, shown to have desirable properties, and are applied to three CPI samples where they generate significant selection corrections.