Field
Trial Title
|
Before
What's Driving Uninsurance? Pay-When-You-Drive Contracts and Market Failures in Auto Insurance
|
After
Prepaid Auto Insurance Contracts and Willingness-to-Pay for Lower Liquidity Requirements
|
Field
Abstract
|
Before
Despite a universal insurance mandate, 30 million drivers in the United States do not carry the minimum automobile liability insurance required by law. Traditional auto insurance contracts charge a fixed monthly premium to cover unlimited driving, making them more attractive for higher mileage drivers (adverse selection) and encouraging excess driving on the margin (moral hazard). We introduce a Pay-When-You-Drive (PWYD) contract designed to increase take-up among low income uninsured drivers, which charges an incremental insurance premium per minute of driving. We randomize auto insurance contract offers to uninsured drivers in California (where 15% of drivers lack insurance), varying the flexibility of the contract (traditional monthly premium driving versus a per-minute premium), the amount of the upfront payment required, and the price of per-minute premiums. The design will test for liquidity constraints and expected risk as explanations for driving without insurance, and evaluate the potential of PWYD contracts to increase insurance take-up among uninsured drivers. Additional price variation will allow us to estimate the elasticity of take-up and insured driving with respect to price, and to test for the presence of adverse selection and moral hazard in auto insurance markets.
|
After
Despite a universal insurance mandate, 30 million drivers in the United States do not carry the minimum automobile insurance required by law. Traditional contracts pool high and low frequency drivers and require large upfront payments to enroll. High upfront premiums may make these contracts unappealing to low income drivers, who also drive fewer miles on average. We introduce a flexible "prepaid" auto insurance contract designed to increase take-up among uninsured drivers by lowering liquidity requirements and charging drivers an incremental premium per day of driving. We randomize auto insurance contract offers to uninsured drivers in California (where 15% of drivers lack insurance), varying the flexibility of the contract (traditional versus prepaid), the price of coverage, and quantity discounts for longer coverage terms. The design tests the potential of flexible prepaid contracts to increase insurance take-up among uninsured drivers, estimates willingness-to-pay for lower liquidity requirements, and explores potential barriers to insurance take-up.
|
Field
Trial Start Date
|
Before
March 14, 2018
|
After
August 01, 2018
|
Field
Trial End Date
|
Before
October 09, 2018
|
After
August 01, 2019
|
Field
Last Published
|
Before
March 12, 2018 10:27 AM
|
After
August 23, 2018 01:59 PM
|
Field
Intervention (Public)
|
Before
We will introduce a novel auto insurance contract into the minimum liability auto insurance market in California. In traditional contracts, drivers pay one monthly lump sum premium which covers all of their driving for the month. We will introduce a new "pay-when-you-drive" (PWYD) contract that charge drivers per minute of driving with no up front monthly premium. The contract offered, required deposit, and the actuarially adjusted per minute premium will be experimentally varied as described in the Experimental Design section.
PWYD insurance will be activated and deactivated using SMS messages. We will additionally passively monitor the driving behavior (driving time and safety) for subsamples of each treatment group using a phone-based safe driving application and on-board diagnostic devices.
|
After
We introduce a novel auto insurance contract into the minimum liability auto insurance market in California. In traditional contracts, drivers pay one lump sum premium which covers all of their driving for the period of coverage (often three or six months). We will introduce a new daily insurance contract with no up front monthly premium which will allow drivers to pay for coverage only on days they are driving. The contract offered, price of coverage, and quantity discounts will be experimentally varied as described in the Experimental Design section.
Insurance coverage can be paused and reactivated using SMS messages. We will additionally passively monitor the driving behavior (driving time and safety) for subsamples of each treatment group using a phone-based safe driving application and on-board diagnostic devices.
|
Field
Intervention Start Date
|
Before
March 14, 2018
|
After
August 01, 2018
|
Field
Intervention End Date
|
Before
October 09, 2018
|
After
August 01, 2019
|
Field
Primary Outcomes (End Points)
|
Before
(1) Insurance Take-Up (Binary): Whether an individual accepted the quoted insurance offer and enrolled in the plan (defined as signing the contract and making the initial payment/deposit).
(2) Insurance Enrollment (Positive Integer): The number of days the individual was enrolled in their insurance plan.
(3) Total Amount Spend on Insurance (Continuous): Total spent on insurance during the three month study period.
(4) Activations (Positive Integer): The number of times a driver texts to activate their insurance.
(5) Insured Driving Time (Continuous): The number of minutes PWYD insurance is active (sum of time between activation and deactivation over all activations).
|
After
(1) Insurance Take-Up (Binary): Whether an individual accepted the quoted insurance offer and enrolled in the plan (defined as signing the contract and making the initial payment/deposit).
(2) Insurance Enrollment (Positive Integer): The number of days the individual was enrolled in their insurance plan.
(3) Total Amount Spend on Insurance (Continuous): Total spent on insurance during the three month study period.
(4) Activations (Positive Integer): The number of days insurance coverage was active.
|
Field
Experimental Design (Public)
|
Before
We will randomly offer drivers applying for minimum liability auto insurance one of four contract offers:
A. Traditional Insurance Plan
B. High Deposit / Low Price PWYD Plan
C1. Low Deposit / Low Price PWYD Plan
C2/C2d. Low Deposit / High Price PWYD Plan
Conditional on enrolling in C2/C2d, half of these drivers will receive a discounted per minute premium to the low price PWYD plan.
|
After
We will randomly offer drivers applying for minimum liability auto insurance either a traditional contract offer (control) or a daily auto insurance plan. Those assigned to the daily auto insurance plan will be randomly assigned to one of three price tiers (low, medium, high) and a quantity discount (either no discount or discount).
The medium price will be the actuarially fair price. The low and high prices will be 20% deviations from that price (down and up, respectively). The discounts will be 2 "free days" (14% off) if they purchase 14 days and 6 "free days" (20%) off if they purchase 30 days of coverage.
|
Field
Randomization Method
|
Before
We will pre-randomize the initial offer in blocks of 50 applicants. We will additionally pre-randomize within each treatment group of enrolled drivers.
|
After
We will pre-randomize the initial offer in blocks of 49 applicants. We will additionally pre-randomize within each treatment group of enrolled drivers.
|
Field
Randomization Unit
|
Before
We will randomize each individual to a treatment group, blocking the randomization every 50 applicants.
|
After
We will randomize each individual to a treatment group, blocking the randomization every 49 applicants.
|
Field
Planned Number of Observations
|
Before
The study sample will be uninsured drivers in California shopping for minimum liability auto insurance coverage. We are targeting 10,000 applications, which will result in several thousand drivers (depending on take-up rates from the first stage of the study). We will have a pool of funding which we will apply to subject acquisition or providing insurance for individuals who enroll. The budget constraint will bind and we will stop soliciting applications when forecasted outlays exhaust the pool of funding. The number of individuals we can include in the study thus depends on the realizations of the following cost parameters:
- Price per lead (this is a market rate, per-click price that will fluctuate)
- Conversion rates (the share of individuals directed to our website who apply)
- Take-up rates (the share of drivers offered a quote who choose to enroll)
- Insurance costs (traditional premium less spending on PWYD insurance) for those who enroll insurance
|
After
The study sample will be uninsured drivers in California shopping for minimum liability auto insurance coverage. We are targeting 3,000-5,000 applications, which will result in several hundred drivers (depending on take-up rates from the first stage of the study). We will have a pool of funding which we will apply to subject acquisition or providing insurance for individuals who enroll. The budget constraint will bind and we will stop soliciting applications when forecasted outlays exhaust the pool of funding. The number of individuals we can include in the study thus depends on the realizations of the following cost parameters:
- Price per lead (this is a market rate, per-click price that will fluctuate)
- Conversion rates (the share of individuals directed to our website who apply)
- Take-up rates (the share of drivers offered a quote who choose to enroll)
- Insurance costs (traditional premium less spending on daily insurance) for those who enroll insurance
|
Field
Sample size (or number of clusters) by treatment arms
|
Before
We will allocate 20% of applicants to the control group (a traditional plan), 16% to a pay-when-you-drive plan requiring an initial deposit the amount of the traditional plan premium, 20% to the low price pay-when-you-drive plan, and 22% to the high price pay-when-you-drive plan without a discount, and 22% to the high price pay-when-you-drive plan (discounted to the low price after enrollment).
The number of drivers for the driving outcomes will depend on the take-up rates and the number of drivers assigned to each treatment group.
|
After
We will allocate drivers uniformly across the treatment groups for the take-up experiment. The number of drivers for the driving outcomes will depend on the take-up rates and the number of drivers assigned to each treatment group.
|
Field
Secondary Outcomes (End Points)
|
Before
(6) Activations by Time of Day (Positive Integer)
(7) Insured Driving Time by Time of Day (Continuous)
(8) Observed Driving Time (Continuous): For individuals in subsamples who install the phone-based application or on-board diagnostic device to passively track their total driving time, we will be able to measure their observed driving time in addition to their active insured driving time. We can use the comparison between this outcome and reported insured driving time to estimate the share of driving that is uninsured.
|
After
(5) Observed Driving Time (Continuous): For individuals in subsamples who install the phone-based application or on-board diagnostic device to passively track their total driving time, we will be able to measure their observed driving time on both days their coverage is active and days coverage is inactive.
|