Benchmarking: Field Evidence from Singapore

We will randomly assign the food courts to two groups, and only enroll individually-owned vendors into the experiment. All the enrolled vendors within each food centre would be assigned to the same manipulation condition.

A. Control group;
B. Performance benchmarking group;
C. Practice benchmarking group;
D. Performance with practice benchmarking group.

Stall owners in all groups will be given a report on four measures of performance -- sales volume (number of plates sold per week), average cost per plate, profit per plate, and profit per owner working hour (the latter calculated as sales revenue minus total cost divided by total hours worked by the stall owner).

Apart from performance indicators, the report to stall owners in the performance benchmarking group will also include their performance relative to the first, second, and third quartiles of all stalls selling cooked food. The report to stall owners in the practice benchmarking group will also include three managerial or technology best practices, randomly selected from the ten practices identified by the regression analysis. The report to the combined performance with practice benchmarking group will include both performance and practice benchmarking information.

2019-08-01

2019-08-15

Sales revenue; cost; owner productivity; management and technology practices.

1. Owner productivity: We measure food stall productivity by owner productivity, calculated as the sales revenue minus total cost divided by the number of owner-hours worked.
2. Management practices: Include whether vendors keep systematic accounting records, adopt incentive-based salary schemes, get Facebook/Instagram accounts and so forth.
3. Technology practices: Include whether vendors use automatic kitchen equipment, install digital ordering systems and E-payment, adopt calling pager and so forth.

We plan to conduct the randomized control trial in about 15 food courts in Singapore. We plan to enroll 20-50 individual vendors in each food court to participate in our study.

At Time 0, we will conduct a pre-intervention survey to collect baseline information: sales revenue, cost, working procedures, management practices, adoption of technology, personal background, and psychological profiles.

We will compile the information from the pre-intervention survey and then calculate performance indicators as well as benchmarking information for all vendors. We will then randomly assign the food courts into control group, performance benchmarking group, practice benchmarking group, and performance with practice benchmarking group.

At Time 1, upon completing data collection and analysis, we will contact the four groups and provide the corresponding information (control or treatment respectively).

3 month (Time 2), 12 months (Time 3) and 18 months (Time 4) respectively after the control/intervention visit, we will return to interview all participating vendors to collect information about changes in productivity and management and technology practices.

Randomization done in office by a computer.

Randomization by each row of stalls within one food court.

Yes

15 food courts, with 20-50 vendors from each food court.

vendors.

20% vendors control, 25% vendors performance benchmarking, 25% vendors under practice benchmarking, and 30% performance with statistics benchmarking.