Corporate Financial Reporting: Experimental Evidence and Economic Implications

Last registered on July 26, 2022

View Trial History

Pre-Trial

Trial Information

General Information

Title

Corporate Financial Reporting: Experimental Evidence and Economic Implications

RCT ID

AEARCTR-0009794

Initial registration date

July 25, 2022

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published

July 26, 2022, 1:39 PM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Country

Germany

Region

all

Primary Investigator

Name

Davud Rostam-Afschar

Affiliation

University of Mannheim

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Yuhan Liu

PI Affiliation

University of Mannheim

Contact Investigator

PI Name

Jannis Bischof

PI Affiliation

University of Mannheim

Contact Investigator

Additional Trial Information

Status

In development

Start date

2022-08-01

End date

2024-03-15

Keywords

Behavior, Crime, Violence, & Conflict, Finance & Microfinance, Firms & Productivity

Additional Keywords

earnings management, earnings quality, survey methods, indirect questioning, item count technique, list experiment, randomized response, social desirability bias

JEL code(s)

G30, G32, M14, M41, D91, C83

Secondary IDs

Prior work

This trial does not extend or rely on any prior RCTs.

Abstract

In this project, we study whether firms manage their earnings and the key drivers that influence the probability of firms conducting earnings management. Companies take various measures to achieve their earnings targets. Financial reporting regulations give firm decision makers (some) flexibility in the way they present information allowing them to manipulate figures deliberately and actively according to their own objectives. The prevalence of earnings management, however, is difficult to quantify since archival data is silent on the decision-making process. Survey methods may overcome this problem. However, given the sensitive nature of earnings management, issues such as measurement error (non-response, recall bias, social desirability bias) may arise. To elicit truthful answers to sensitive questions, we run a double list experiment (Droitcour et al., 1991) and compare the estimates of the proportion of firms that engage in earnings management to those from direct and indirect questioning.

External Link(s)

Registration Citation

Citation

Bischof, Jannis, Yuhan Liu and Davud Rostam-Afschar. 2022. "Corporate Financial Reporting: Experimental Evidence and Economic Implications." AEA RCT Registry. July 26. https://doi.org/10.1257/rct.9794-1.0

Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Experimental Details

Interventions

Intervention(s)

Our experiment consists of two baseline lists (with/without sensitive item included), a direct question about own engagement in earnings management, and an indirect question about earnings management regarding other companies. Survey participants are randomly assigned to different groups that contain different sets of questions. In the list experiment, the respondent is only asked to provide a total number of activities that the firm engaged in from the listed activities, not which ones.

Intervention Start Date

2022-08-01

Intervention End Date

2023-01-20

Primary Outcomes

Primary Outcomes (end points)

• proportions of earnings management determined by the total number of items respondents provided from the list experiment
• economic implications (management actions, investments, hiring)

Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)

• time frequency of conducting earnings management from the direct question
• time frequency of conducting earnings management from the indirect question

Secondary Outcomes (explanation)

Experimental Design

Survey participants are randomly assigned to four different groups that receive different sets of questions. In the list experiment, the respondent is only asked to provide the total number of the listed items they have engaged in in the past, not which ones. The averaged value of the two mean-difference estimators is an estimation for the proportion of respondents who have managed earnings.

Experimental Design Details

We conduct a double list experiment because we expect to obtain a lower variance of the estimate compared to a single list experiment. As respondents are only asked to provide a total number of applicable items in the list experiment, variance of the responses is of critical concern.

In a single list experiment, the variance of the estimator depends on the variance of the sensitive item and the covariance between the sensitive item and the baseline list. However, this covariance may be large. In a double list experiment, on the other hand, if respondents react in an indifferent way when the sensitive item is included in the lists, the variance of the averaged difference-in-means estimators will only depend on the variance of the sensitive item, and the variance of the two baseline lists as well as the covariance between the two baseline lists (Glynn, 2013).

According to Glynn (2013), the optimal design of baseline lists should fulfill the following two conditions:
1) negatively correlated items are included in each list to minimize the variance of the list; and
2) the two baseline lists are positively correlated to maximize the covariance between the baseline lists.

Randomization Method

Randomization done by a computer through Qualtrics

Randomization Unit

Firm

Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters

No clusters

Sample size: planned number of observations

Approx. 1,800 questionnaires with valid responses by firm representatives

Sample size (or number of clusters) by treatment arms

Each treatment arm has 450 observations of firm representatives

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

We follow Glynn (2013) for sample-size calculations. We conduct the power calculation based on the pretested baseline lists. We sent the baseline lists to student assistants and they were told to behave as CEOs in real companies which operate in certain industries. 25 valid answers were returned. If the answers from the student assistants are representative and firm respondents do not change their behaviors when the sensitive item regarding earnings management is included in the list, 703 firm responses are needed if the goal is to achieve a half-width of 0.08 and a 5% significance level. Note that each group has equal sample size, i.e., Group 1 and Group 2 should have 352 valid responses each.

Supporting Documents and Materials

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

IRB