'Better than perceived? Correcting misperceptions of central bank forecast accuracy'
(AsPredicted #107,388)
Download PDF

---

---

1) Have any data been collected for this study already?
It's complicated. We have already collected some data but explain in Question 8 why readers may consider this a valid pre-registration nevertheless.

2) What's the main question being asked or hypothesis being tested in this study?
Main question: How does trust influence attention and people's expectations?
Second question: Can we foster trust in European Central Bank by correcting misperceptions about its projection accuracy?

Directed hypotheses:
1) Correcting misperceptions increases trust in the provider of information.
2) Trust and attention are positively related.
3) Trust mediates the impact of information on expectations.

3) Describe the key dependent variable(s) specifying how they will be measured.
There are four dependent variables.

I elicit Euro area inflation expectations of German households for 2024 twice. The first elicitation asks for point expectations while the second question asks for the supports of the distribution. Third, I elicit people's overall trust (on a scale between 1 to 10) in the European Central Bank. Lastly, I measure the time respondents spend on reading the information I provide them as well as the time they spend on answering the second inflation expectations question.

4) How many and which conditions will participants be assigned to?
All groups receive information before submitting their second inflation expectations and the question related to trust.

There are four experimental conditions: (i) Active control, (ii) Projection, (iii) Passive correction, (iv) Correction. The first group receives information on population growth in Germany between 2010-21 (i.e., 2%). The second group receives information on the ECB-staff projection of Euro area inflation for 2024. The third group also receives this projection but in addition they are informed about the historical projection accuracy (between 2001-2021). The last group receives the same information as (iii) but subjects in this group are first asked to evaluate forecast accuracy of ECB in the past and then provided with the historical projection accuracy along with the last projection for 2024 as feedback.

5) Specify exactly which analyses you will conduct to examine the main question/hypothesis.
I will regress the following specifications:

trust_ECB(i) = a_0 + a_j * I(Treatment=j) + e(i) (equation 1)

where I(Treatment=j) is an indicator variable for each treatment j={ii, iii, iv}.

posterior(i) - prior(i) = b_0 * (signal - prior(i)) + b_j * (signal - prior(i)) * I(Treatment=j) + e(i) (equation 2)

where posterior and prior refers to the midpoint of supports of inflation expectations in the second question and the point expectation in the first question, respectively. Signal refers to population growth in control group and the inflation projection in the treatment groups.

I will run mediation analyses with posterior expectations (either midpoint or range) or time spent on information as dependent variables, trust in ECB as the mediator and the treatment indicators as the explanatory variables.

I will run all regressions with OLS and quartile regressions and with robust standard errors.

6) Describe exactly how outliers will be defined and handled, and your precise rule(s) for excluding observations.
1) I will perform the same analyses with winsorized inflation expectation variables at 5%. If the results are significantly different from the full sample, I will use the winsorized data for the main analysis.
2) I will exclude any respondent who provided maximum equal to minimum in inflation expectations.
3) I will exclude all respondent with survey response time longer than 2 hours and treatment text reading time shorter than 3 seconds.

7) How many observations will be collected or what will determine sample size?
No need to justify decision, but be precise about exactly how the number will be determined.
The sample comes from the Bundesbank's Online Panel of Households Survey. Bundesbank determines the sample size. The full sample is estimated to be between 3000-4000 subjects.

8) Anything else you would like to pre-register?
(e.g., secondary analyses, variables collected for exploratory purposes, unusual analyses planned?)
Data collection has began but data has not been provided to me when I submitted this pre-registration. As of the writing of this pre-registration, the single author of this pre-registration was the only author involved in this research project.

Robustness checks:
I will include demographic controls to every regression I run.
I will estimate equation 1 again by replacing the dependent variable with an indicator variable for the trust in ECB with logit.
I will estimate equation 1 with a reduced panel sample by including the latest "trust in ECB" variable from the BOP-HH survey (June 2022) along with its interaction to the treatment among the regressors.
I will accompany the mediation analysis for trust with the moderating effect of the beliefs on ECB's projection accuracy.