The Price of “Us First” - Discrimination in the Danish labour market

We will deliver leaflets advertising the services (e.g., house cleaning) of several workers who only differ in their origins (native vs. migrant, as signaled by their names) to a random sample of Danish households. The hourly rate charged by the workers will be randomly varied to generate a demand schedule. In addition, we will vary the workers’ skill levels across different treatment arms. The leaflets will indicate a phone number that interested customers can use to contact the workers. Differences in callback rates between different types of workers will be our primary outcome of interest. We will monitor callbacks for a period of 4 months following the distribution of the leaflets. The leaflets will be distributed in urban and rural areas. By assigning different phone numbers across locations, we will be able to investigate geographic heterogeneity.

2020-01-17

2020-05-15

The key outcome of interest is the callback rate from potential customers to different types of leaflets. Our aim is to measure differences in callback rates between native and migrant workers across a range of prices and skill levels.

Our experiment will consist of sending 16 types of leaflets to Danish households in randomly assigned non-overlapping areas. Workers’ origins (native vs. migrant) will be indicated using distinctive names.

Leaflet 1 – 8 (the “Price” treatment)

The main treatment (the “Price” treatment) will consist of 8 types of leaflet. The first four will be destined to urban areas (neighbourhoods of Copenhagen) and will be obtained from the combination of two workers (native vs. migrant) and two prices (high and low). The second four will be destined to rural areas and will be identical to the first four except for the phone numbers they will display (which is necessary to investigate heterogeneity between rural and urban areas).

Leaflet 9 – 16 (the “Skill” treatment)

The second treatment (the “Skill” treatment) will consist of an additional 8 types of leaflets, which will all be sent to urban areas. These leaflets will differ from the first 8 insofar as they will offer customers a choice between two different workers (customers will be presented with the phone numbers of two operators active in their areas and they can choose which one, if any, they would like to call). Crucially, the two workers will differ in their quality (i.e., skill level), as signaled by the ratings of previous customers. In the first leaflet, native and migrant worker have the same skills. In the second leaflet, the migrant has higher skills. In the third leaflet, the native has higher skills. The fourth leaflet replaces the migrant with another native and serves as a benchmark to measure the impact of differing customer reviews, independent of the workers’ origins. The remaining four leaflets are identical to the first four except for the fact that they swap the order in which the names of the workers appear on the leaflets.

The different types of leaflets will be assigned to randomly chosen areas by means of a computer programme. The randomization method ensures balance across a set of available covariates, including population size, share of females, average age for areas in Copenhagen, and population size, average age, employment rate, and share of university graduates for areas outside Copenhagen.

Different types of leaflets will be randomly assigned to distinct non-overlapping areas, whether urban (i.e., in Copenhagen) or rural (i.e., in the Sjaelland region outside Copenhagen), of similar size.

No

56,000 leaflets divided in 16 types and sent to randomly assigned areas.

56,000 leaflets

Price treatment

For the main treatment, we will have 18,000 leaflets with a native worker and 18,000 leaflets with a migrant worker. For each worker, 9,000 leaflets will indicate a low price and the remaining 9,000 will indicate a high price. Out of these 36,000 leaflets, 20,000 will be allocated to urban areas and 16,000 to rural areas.

Skill treatment

For the skill treatment, an additional 20,000 leaflets will be sent to urban areas. Each of the eight variations described above will be allocated 2,500 leaflets.