Salience and Timely Compliance: Experimental Evidence from the Enforcement of Speeding Tickets

We use cover letters to augment a legal text. Different cover letters aim at increasing the salience of different enforcement parameters.

To analyze how the salience of deadline and penalties affects behavior, we conduct a field experiment on the enforcement of speeding tickets. The trial will be conducted in cooperation with the local authorities in the town of Ricany in the Czech Republic. The town has speeding radar systems installed along several roads. Each radar system measures the speed, takes an electronic photo and automatically identifies the number plates of the passing cars. Traffic authorities collect the information from the radars, identify the cars driving above the speed limit and calculate the difference between the speed limit and the car's speed (applying some institutionalized tolerance rule). Based on this speed, the relevant penalty is included in a letter that is then sent to the driver.
Similar as in other countries, the penalties for speeding increase stepwise in the speed measured by the radar. A driver is exposed to a small penalty for a speed of up to 20km/h above the limit, an intermediate penalty for a speed of 21-40km/h above the limit, and a high penalty for speeding with more than 40km/h above the limit. Our analysis will explore these thresholds (in particular, the first one at 20km/h). The penalties include fines which must be paid within 15 days after receiving the ticket, otherwise the due amount increases and additional penalties within an official demerit point system are applied.
To enforce payment, authorities send a summons to the registered address of the vehicle's owner either by regular mail or by e-mail through an official e-government platform called “databox”. The use of databox is compulsory for corporations but optional for private individuals, but if they choose to use it, the summons are sent through this channel.
The notifications contain, among extensive legal extracts, information about the time and date of the traffic violation, the amount of the fine, the payment deadline and some information on the consequences of missing the deadline. The notification used by the authorities is lengthy, fairly convoluted, and full of legal terms. One can therefore question, whether the text fully conveys the relevant information regarding the penalty, the deadline for paying the fine and the consequences of missing it. Obviously, these two attributes should play a crucial role in the individuals' decision regarding payment execution and its timing.

2017-12-01

2019-08-03

We will calculate the effect of our treatments on the payment decision of the individuals by comparing payment rates across treatments for full compliance (i.e., for the payment of all fines). Our primary outcomes will be based on (1) timely compliance (pre-deadline payments) and (2) payments within 30 days. Moreover, (3) we consider the exact timing of the (first) payment.

Ad 1+2: To analyze the treatments' effect on compliance we will first estimate binary outcome models (using LPM). For our key outcome, we consider both payments within 15 (pre-deadline!) and 30 days. In all cases we estimate the model with and without controls.

Ad 3: Our analysis of the timing of payments mainly focuses on the period before the deadline (15 days after receiving the speeding ticket). However, to assess the treatments impact on the timing of payments we will consider duration models with an outcome (based on the timing of the first-payment) defined over a longer periods (up to 100 days). Estimates will be based on Cox proportional hazard model and/or parametric hazard models (Weibull). To assess payment hazards (conditional payment probabilities for day t), we will parametrically and non-parametrically analyze hazards over time for the different treatment groups.

Note that our analysis focuses on Intention-to-Treat (ITT) effects, as some mailings will be probably not delivered and, among those that are delivered, not all will be opened and properly read. (To the extent that our outcome data allow us to assess these problems, we will also comment on the Treatment-on-Treated (TOT) effect size.)

In addition to full payments, we will also consider partial payments ("any"-payment-made-indicators). We will again assess these for the time frames from above (payments within 15 and 30 days). To capture mid-run treatment effects on compliance, however, we will also estimate models for our outcome variables (full/partial compliance) defined over a longer time frame of 60 and 100 days.

If we are able to merge our data with information on future driving/speeding behavior, we will also examine treatment effects on subsequent driving behavior.

The binary outcome models are again estimated with LPM. To assess if treatment effects differ between full and partial compliance, we will compare these estimates to results obtained for our primary outcome.

We will make use of the information about ticket receivers in our dataset to analyze the difference in our outcomes for private and corporate car owners. Furthermore, based on the information on zip codes, we split the sample into local vs non-local car owners. We also analyze differential effects for different speed levels keeping the amount of the fine constant, and, conditional on data availability we will consider heterogeneous effects between first-time speeders and individuals with past speeding infractions.

As part of this heterogeneity analysis, we will make use of the observable characteristics to explore if type heterogeneity drives the pattern in hazard rates.

Using the fact that the fine is a stepwise function of the speed at which individuals were driving, and that the application of fines higher than the lowest one is discretionary, we implement a regression discontinuity design around the first speed threshold that triggers an increase in the fine (using speed as the running variable), to analyze the effect of the size of the fine on timely payment. For this we estimate non-parametrically an equation with dummies capturing total or partial payments within different periods of time on the left hand side and the measured speed, a dummy equal to one if the vehicle's speed is above the cutoff, and a function that captures how the outcome variable varies with the speed (this function is allowed to differ on either side of the cutoff), on the left hand side.

We will estimate these equations for the full sample first. Depending on the number of observations per treatment, we will perform the analysis for the control group sample only. Furthermore, and also contingent on the number of observations, we will investigate if there are treatment differences by including interactions with the treatment dummies to assess if the impact of the higher fines differs across the different treatments.

We conduct a randomized control trial with the universe of recipients of speeding violation tickets during the period of the study. Each speeding case is randomly allocated into a control group or one of three treatments. Due to technical constraints at our cooperation partner, the randomization will not use any stratification.

Our experimental approach is based on a 2 x 2 factorial design. All speeders in all groups receive the legal notification message, i.e., the usual summons used by the enforcement authority. Three treatment groups [T1-T3] will receive an additional cover letter. The control group [T0] does not get any cover letter. The cover letter includes a simple text that aims at communicating in a clear way either the deadline, the costs associated with not paying the fine on time, or both. (Note that a subset of speeders, in particular, cars registered for companies that use an electronic way of communicating with authorities, will receive the cover letter (as well as the legal notification) as a PDF.)

An English versions of the different cover letters' texts are provided below:


[T1] Deadline Salience:

Dear Sir/Madam,

We summon you to pay the prescribed amount for a speeding violation. We encourage you to carefully read the information contained in the attached pages and take appropriate action.

Please pay the amount in full and make sure it is credited to the city's account within 15 days after receiving this summons.

The city office of Ricany.

[T2] Penalty Salience:

Dear Sir/Madam,

We summon you to pay the prescribed amount for a speeding violation. We encourage you to carefully read the information contained in the attached pages and take appropriate action.

If you do not pay the whole amount the office will continue investigating the offense. The amount that you will potentially have to pay may be as high as CZK 2,500. (For medium-severity speeding violations i.e., for a speed of more than 20km/h above the limit, this part reads 'as high as CZK 5,000'). In addition, the driver may be added points within the demerit point system.

The city office of Ricany.


[T3] Deadline + Penalty Salience:

Dear Sir/Madam,

We summon you to pay the prescribed amount for a speeding violation. We encourage you to carefully read the information contained in the attached pages and take appropriate action.

Please pay the amount in full and make sure it is credited to the city's account within 15 days after receiving this summons.

If you do not pay the whole amount the office will continue investigating the offense. The amount that you will potentially have to pay may be as high as CZK 2,500. (This part is again adjusted for medium-severity speeding violations, see [T2]). In addition, the driver may be added points within the demerit point system.

The city office of Ricany.

The randomization will be made by a computer (using software installed at the municipality office). The software draws a random number and, based on this number, assigns each case to one of the four conditions T0, T1, T2 or T3 (with a rate of 1/4 each).

Speeding ticket

No

At this stage, there is unfortunately considerable ambiguity regarding the length of the trial (at least 9 month) and regarding the number of observations collected during this time period (speeders who will get a ticket). Based on what we know, this could be anywhere between 5,000 to 20,000 speeding tickets.

We would aim at 20,000 speeding tickets, but -- as pointed out above -- there is considerable ambiguity whether we could reach this number.

Sample Size / 4. I.e., if we would reach the planned 20,000 experimental speeding tickets, this would be 5,000 per treatment.

Given the ambiguity about the number of observations and the size of the (deadline) treatment effects, we did not conduct a fully fledged power analysis.

This paper studies the enforcement of fines. We randomly assign 80,000 speeding tickets to treatments that increase the salience of the payment deadline, late penalties, or both. Stressing the penalties significantly and persistently increases payment rates. Emphasizing only the deadline is not effective. The findings from the RCT are consistent with a survey experiment which documents the treatments' impact on priors about parameters of the compliance problem. Exploiting discontinuous variation in fines, we then document a strong price responsiveness: a 1% increase in the payment obligation induces a 0.23 percentage point decrease in timely compliance. This semi-elasticity suggests that the impact of the salience nudges is equivalent to the effect of a 4-9% reduction in fines.

Dusek, Libor, Nicolas Pardo, and Christian Traxler (2020), Salience, Incentives, and Timely Compliance: Evidence from Speeding Tickets, Discussion Paper 2020/09, Max Planck Institute for Research on Collective Goods.

https://ideas.repec.org/p/mpg/wpaper/2020_09.html

This paper studies the enforcement of fines, and, in particular, the effects of simplification and salience nudges on timely payments. In a randomized controlled trial, we add cover letters to 80,000 payment notifications for speeding. The letters increase the salience of the payment deadline, the late penalty, or both. Emphasizing only the deadline is not effective. Stressing the late penalty significantly and persistently increases payment rates. The effect is largest if both parameters are made salient. The most effective treatment yields a net revenue gain that covers approximately 25 percent of the labor costs of the ticket administration personnel. A survey experiment documents how the salience nudges alter prior (mis)perceptions about the communicated parameters. The survey results rationalize the differential effects of the treatments and, together with the evidence from the RCT, offer a broader framework for explaining why certain nudges are effective in some contexts but fail in others.

Dušek, L., Pardo, N. and Traxler, C. (2022), Salience and Timely Compliance: Evidence from Speeding Tickets. Journal of Policy Analysis and Management, Vol. 41, p.426-449. https://doi.org/10.1002/pam.22387

https://onlinelibrary.wiley.com/doi/full/10.1002/pam.22387