Abstract

In the current fast-evolving economic environment where the pandemic crisis was followed by the Russian war of aggression against Ukraine and then by the rapid surge of inflation, the novel Consumer Expectations Survey, initiated in 2020 by the ECB, has been providing timely information on consumer beliefs and expectations as well as the possibility to monitor trends in consumer economic and financial behaviour in conjuncture with monetary policy. However, the benefits of such rich new data collection come also with methodological challenges that require the development of innovative tools and indicators to monitor various data quality aspects of the survey. Measuring expectations remains a challenge, not only because it requires respondents to engage with concepts that might not always be easy to understand (e.g., inflation). Survey tenure, questionnaire design and the mode of collection might have an impact on their replies as well. For example, the mode of collection of CES, which is an online survey, might increase respondents engagement by allowing for an easily accessible survey through smartphones, but it makes it more difficult to reach certain subgroups of the population like older respondents. In addition to this, having a strong panel component enhances the possibility to analyse individual changes over time while, at the same time, respondents might learn through time or adopt strategic response behaviour. In this circumstance, it is essential to have an efficient rotation algorithm in place, including mechanisms to smooth recruitment of respondents over time, to minimise the impact of potential sample composition effects on key aggregate results. In this paper, we firstly describe the survey design and sample composition. We then highlight important statistical aspects such as panel management, survey tenure, selective attrition and response behaviour and analyse their interplay through regressions, proposing new approaches to ensure the production of high-quality and timely data on consumer behaviour and expectations. In this way, we will shed light on important quality aspects of online surveys and further contribute to the growing literature on the eliciting of expectations that are increasingly important for policy makers and researchers.