New proposals for the quantification of qualitative survey data

Abstract

In this paper we deal with several issues related to the quantification of business surveys. In particular, we propose and compare new ways of scoring the ordinal responses concerning the qualitative assessment of the state of the economy, such as the spectral envelope and cumulative logit unobserved components models, and investigate the nature of seasonality in the series. We conclude with an evaluation of the type of business cycle fluctuations that is captured by the qualitative surveys.

Inflation expectations in India: Learning from household tendency surveys

“…Using the notion of spectral envelope, Proietti and Frale (2011) show that the balance statistic has a filtering role and extracts the underlying cycle imbedded in a time series of counts. Let individual 's true expectation about time period be * .…”

“…8 It allows us to compare, contrast, and combine measures of inflation expectations based on both qualitative data and quantitative responses. By 5 See, among others, Fishe and Lahiri (1981), Dasgupta and Lahiri (1992), Ash et al (1998), Nardo (2003), Kakoska and Teksoz (2007), Lui et al (2011), Proietti and Frale (2011), Breitung and Schmeling (2013), Vermeuleu (2014), Lahiri and Zhao (2015), and Kaufmann and Scheufele (2017) for developments in quantification methods over the last 30 years. 6 Curtin (2007Curtin ( , 2010 and Bruine de Bruin et al (2017) document systematic differences in reported inflation expectations across several surveys in the United States and offer several rationalizations.…”

“…We have four sets of estimates: two from purely qualitative data (the balance statistics and the HOPIT model estimates based on qualitative responses) and two based on quantitative expectations (balance statistics from implied qualitative responses and HOPIT estimates based qualitative responses imputed from numerical forecasts).24 See, for example,Dasgupta and Lahiri (1992),Proietti andFrale (2011), andBreitung andSchmeling (2013) Vermeulen (2014). shows that the balance statistic does well during stable periods, whereas the CP method works relatively better during volatile periods.25 Both the balance statistics and the HOPIT model based on reported qualitative responses show similar performance, with in-sample RMSE of 2.28 and 2.32, respectively.…”

Discussion of The class of CUB models: statistical foundations, inferential issues and empirical evidence

“…In repeated preferences and attitudes surveys an obvious extension is letting the preference and uncertainty parameters vary over time. A simple illustration is provided by the analysis of consumer opinion surveys, for which the CUB approach can provide an alternative way of quantifying the results, which typically take the form of the percentage of responses falling into ordinal categories; see also Proietti and Frale (2011).…”