Measuring changes in consumer confidence

Consumer confidence indicators are surveyed monthly and each month concern different individuals. This complicates a straightforward interpretation of shifts in confidence. First, it is not clear how many respondents switch from and to negative, neutral and positive opinions in consecutive months. Second, reported net changes in confidence may be largely driven by the different respondent samples used over time. The proposed methodology addresses both issues. It involves estimating unobserved switching between negative, neutral and positive opinions for what can be thought of as being the same set of individuals. Next, a new change-in-confidence measure is developed from these switching proportions and the associated confidence bounds are computed for testing purposes. Applications to US and Dutch confidence data show that US respondents tend to switch attitudes more often than their Dutch counterparts do. Furthermore, the illustrations show that monthly changes in consumer confidence are not often significantly different from zero. Hence, claims about increased or decreased confidence should be made with care.