Deterring moral hazard in social insurance: Experimental evidence from doctors’ leave prescription

Last registered on June 11, 2025
View Trial History

Pre-Trial

Trial Information

General Information

Title
Deterring moral hazard in social insurance: Experimental evidence from doctors’ leave prescription
RCT ID
AEARCTR-0016184
Initial registration date
June 06, 2025

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
June 11, 2025, 8:18 AM EDT

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

Locations

Country
Chile
Region
REGIÓN METROPOLITANA

Primary Investigator

Name
Pablo Celhay
Affiliation
Pontificia Universidad Catolica de Chile
Contact Primary Investigator

Other Primary Investigator(s)

Additional Trial Information

Status
Completed
Start date
2022-02-01
End date
2022-12-01
Keywords
Behavior, Health
Additional Keywords
Healthcare Economics; Insurance; Adverse Selection; Risk reclassification
JEL code(s)
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
This study implements a randomized controlled trial (RCT) to evaluate the behavioral response of high-prescribing physicians to automated notifications regarding their issuance of medical leave certificates in Chile. The intervention targets approximately 1,500 physicians identified as the top 2–3% emitters of medical licenses based on data from 2019–2021. The sample was stratified by baseline volume of license issuance and medical specialty, and randomly allocated into three groups: a control group and two treatment arms. Over an 11-month period (February–December 2022), physicians received real-time "pop-up" notifications through the IMED electronic license issuance platform. Treatment Group 1 received informational messages about their relative position in the national distribution of license issuers. Treatment Group 2 received similar messages with the addition of a warning regarding potential auditing, referencing relevant legal frameworks. The control group received no notifications. The study uses administrative records from FONASA, including the complete universe of medical licenses issued electronically or by paper from 2018 to 2022. Data include variables such as diagnosis code, encrypted physician and patient IDs, issuance date, and duration. Additional datasets provided demographic and professional information for balance checks and heterogeneity analyses. To estimate treatment effects, the empirical strategy involves fixed-effects panel regressions at the physician-month level, clustering standard errors at the randomization block level. The primary outcomes include the total number of licenses issued and days granted. The analysis also includes dynamic specifications to examine monthly treatment effects over time, and subgroup analyses based on diagnosis categories and physician characteristics.
External Link(s)

Registration Citation

Citation
Celhay, Pablo. 2025. "Deterring moral hazard in social insurance: Experimental evidence from doctors’ leave prescription." AEA RCT Registry. June 11. https://doi.org/10.1257/rct.16184-1.0
Sponsors & Partners
Experimental Details

Interventions

Intervention(s)
The intervention aimed to reduce excessive issuance of medical leave certificates by high-prescribing physicians in Chile’s public health insurance system (FONASA), using behavioral nudges delivered through existing digital infrastructure. It consisted of real-time, automated notifications integrated into an electronic platform used by physicians to issue medical licenses.

The intervention targeted approximately 1,500 physicians identified as the top 2–3% emitters of licenses from 2019 to 2021. Physicians were stratified by baseline issuance volume and specialty, then randomly assigned into one of three groups:

Control group: No intervention.

Treatment group 1 (Information message): Upon issuing their first license of the day, physicians received a pop-up notification stating their cumulative number of licenses issued in the year and their national percentile ranking.

Treatment group 2 (Information + regulatory warning): Physicians received the same notification as group 1, but with an added legal warning stating that their behavior could trigger an audit under existing health regulations (Art. 21 of DS 3 and Law 20.585).

Notifications were delivered daily and updated dynamically with new license counts, while percentile rankings remained fixed due to logistical constraints. The messages were intended to leverage social comparison and perceived regulatory oversight to influence prescribing behavior. The intervention was implemented for 11 months, from February to December 2022, without altering physicians’ ability to issue licenses or their clinical autonomy.
Intervention (Hidden)
Intervention Start Date
2022-02-01
Intervention End Date
2022-12-01

Primary Outcomes

Primary Outcomes (end points)
The key outcome variables in this experiment are designed to capture both the frequency and intensity of medical leave issuance by physicians.
Primary Outcomes (explanation)
The key outcome variables in this experiment are designed to capture both the frequency and intensity of medical leave issuance by physicians. The primary endpoints of interest are:

Total number of medical leave certificates issued per physician per month
– This measures whether the intervention led to a reduction in the quantity of licenses issued by high-prescribing doctors.

Total number of days of medical leave granted per physician per month
– This captures whether physicians adjusted the length of the leave they prescribed, in addition to the frequency.

Average duration of medical leave per license issued
– This metric allows detection of changes in intensity per episode, indicating whether doctors reduced the number of days prescribed per license.

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
This study uses a randomized controlled trial (RCT) to evaluate the impact of digital behavioral nudges on the issuance of medical leave certificates by high-emitting physicians in Chile's public health insurance system (FONASA).

Sample and Randomization:
The intervention targeted approximately 1,500 physicians identified as the top 2–3% emitters of medical licenses based on administrative data from 2019–2021. The sample was stratified by specialty and baseline license volume, then randomly assigned to one of three groups (n ≈ 500 per group):

Control Group: No notification.

Treatment Group 1 (Information Message): Received a daily pop-up notification via the IMED platform with their cumulative number of licenses issued in the year and their percentile ranking nationally.

Treatment Group 2 (Information + Regulatory Warning): Received the same notification as Group 1, with an added warning about potential audit under Chilean health regulations.

Intervention Period:
The intervention was implemented for 11 months, from February to December 2022. Notifications were triggered upon the physician’s first license issuance of the day.

Data and Measurement:
Outcomes are measured using administrative data from FONASA, covering all electronic and paper-issued medical licenses. The primary endpoints include the total number of licenses issued, total days granted, and average days per license. Additional data are used to verify balance across groups and assess heterogeneous effects by diagnosis and physician characteristics.

This design allows for causal identification of the impact of informational and regulatory messages on physician behavior within a high-frequency, real-world digital environment.
Experimental Design Details
Randomization Method
Randomization was conducted centrally by the research team using a computer algorithm. Physicians were stratified based on their baseline number of medical licenses issued (2019–2021) and medical specialty. Within each stratum, individuals were randomly assigned to one of three groups (Control, Treatment 1, or Treatment 2) using a reproducible random seed to ensure transparency and replicability. No public lottery or manual methods (e.g., coin flip) were used. The allocation was implemented prior to the intervention launch and remained fixed throughout the study period.
Randomization Unit
The unit of randomization is the individual physician. Each of the approximately 1,500 physicians identified as high emitters of medical leave certificates was independently randomized into one of three groups (Control, Treatment 1, or Treatment 2).

Randomization was stratified by two key variables:

Baseline volume of license issuance (quantiles based on 2019–2021 data)

Medical specialty

There was only one level of randomization—at the individual level. No group- or facility-level clustering was used in the assignment of treatments.
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
0
Sample size: planned number of observations
1500
Sample size (or number of clusters) by treatment arms
1500
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
Pontificia Universidad Católica de Chile
IRB Approval Date
2021-11-29
IRB Approval Number
211129003
Analysis Plan

Post-Trial

Post Trial Information

Study Withdrawal

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

Request Information

Intervention

Is the intervention completed?
No
Data Collection Complete
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials