Re-assessing the South African household inflation expectations survey through a sequential mixed methods approach
Thesis (MCom)--Stellenbosch University, 2018.
ENGLISH SUMMARY : Many central banks rely on survey-based measures of household inflation expectations as a gauge of policy credibility and an input for forecasting inflation. An accurate measure of the public’s inflation expectations is therefore an essential instrument to guide policy. The public’s perceived ignorance of inflation informed the choice of wording in the South African household inflation expectations survey, which launched in 2000. Based on the international critique against the use of simplified survey wording, this study evaluates the validity of the Bureau for Economic Research’s (BER) household inflation expectations survey question. The study investigates whether South African households understand the term inflation, and the impact of providing a historical inflation number to the respondents in the survey question. This is done through a sequential mixed methods approach. Semi-structured, qualitative interviews were used to explore the public’s understanding and perceptions of inflation. This was followed by a demographically representative, quantitative survey, which was used to reach generalisable conclusions. The results of both the qualitative and quantitative stages of the study suggest that South African households have a far greater understanding of the term ‘inflation’ than previously assumed, although a large proportion do not understand the term adequately. Furthermore, when respondents are provided with historic information on inflation (as is done in the BER survey), survey wording has almost no impact on the public’s inflation expectations. In the absence of an anchor number, survey wording is found to be (somewhat) more important. Without the anchor number, the mean inflation expectation is lower than when respondents are asked about their expectation for ‘prices in general’. Both the descriptive statistics and the econometric analysis of this study show that the socio-economic characteristics of respondents play a role in influencing their inflation expectations. However, if outliers are removed from the data, the socio-economic factors have less influence on inflation expectations.
AFRIKAANSE OPSOMMING : Geen opsomming beskikbaar.