Economics Research Seminar

"A Behavioural Cost-of-Living Index"

Standard cost-of-living indices have a welfare interpretation under the assumption that consumers optimise correctly. I relax this assumption using the limited attention model of Gabaix (2014). I embed this behaviour in index-number theory and derive Afriat-type revealed-preference conditions and a computable  cost-of-living index that imparts a behavioural interpretation to the Laspeyres price index. In household scanner data, limited attention expands the share of households permitting welfare analysis by 27–55 percentage points. I then field a survey to directly elicit perceived prices. Price perceptions are more accurate among respondents with greater numeracy, purchasing familiarity, and lower cognitive uncertainty. Combined with aggregate expenditures, these prices imply a limited-attention cost-of-living increase of 7.2\% versus 6.5\% classically. To study the potential for information to correct misperceptions, I run a pre-registered experiment and find that a feedback treatment closes roughly one-fifth of the sample's mean perceived inflation gap. My results show that imperfect price perceptions, through distorting consumer choices, can materially affect inflation and cost-of-living measurement.