Our randomization sample will comprise approximately 270 SMEs in the leather goods and footwear industry using clutch motors on at least two stitching machines and no energy-efficient “servo” motors. Our design will include two treatment arms and one control arm. Treatment arm 1 (T1) will be the “information and installation” group, and treatment arm 2 (T2) will be the “information only” group. There will also be a control group (C). The intervention will have four stages. At baseline, we will conduct a survey and elicit managers’ beliefs and willingness to pay for a servo motor. To do so, we will provide minimal information about what a servo motor is to all three groups, including the control group. We will also collect information on producers' social network links.
To elicit beliefs about cost savings, we will use methods described in Delavande, Gine, McKenzie (JDE, 2011) to elicit the probability distributions that managers assign to three things: (a) how much electricity per hour of use their existing stitching machines (with clutch motors) consume, (b) how much electricity per hour of use they believe the same machines with servo motors would consume, and (c) how much the servo motors would reduce their total monthly electricity costs. In each elicitation, we will provide firm owners with a number of beans (representing quantiles of the probability distribution) which they will put in bins (representing ranges of values of electricity use or savings).
To elicit willingness-to-pay in an incentive-compatible way, we will use a Becker-DeGroot-Marschak (BDM) mechanism (Berry et al, JPE 2020). In this procedure, we will first inform a manager that we will offer a cash grant for their participation in the study. We will then ask the manager to indicate a purchase decision for a full range of prices, or equivalently we will ask them the highest price at which they will purchase the new motor (“bid price”). A random number generator will then choose a price (from different distributions for different groups, as described below). If the random price is lower than the manager’s bid price, it will be subtracted from the participation payment and the manager will receive the servo motor.
In the BDM procedure, the distribution from which prices are drawn must have full support (to ensure that individuals have an incentive to state their true valuation) but need not necessarily be the same across groups. We will use different distributions for the three groups. For firms in Treatment 1 (T1), we will draw from a distribution that has almost all weight at very low prices (e.g. very near zero). For firms in Treatment 2 (T2) and Control (C), we will draw from a distribution that has almost all weight at a very high prices, at which we expect no firms will be willing to buy the new motor. We thus expect that all (or almost all) firms in T1 and no (or almost no) firms in T2 and C will acquire the new motor at this stage.
After the BDM procedure, firms in T2 will receive intensive information about the benefits of servo motors, including expert explanations on electricity cost savings and ease of installation on existing stitching machines using videos on an iPad. Firms in T1 will receive the information received by T2 and will also receive free installation of the servo motor (assuming it is purchased in the BDM procedure, as described above) and two electricity meters, one on the machine with the servo motor and one on a machine with a clutch motor.
Following this, we will conduct the same beliefs elicitation and BDM procedure described earlier in one midline and one endline visit, but with prices drawn from the same distribution for all three groups. We will also conduct surveys during all visits to understand adoption, electricity cost savings, and overall firm performance.
Berry, James, Greg Fischer, and Raymond Guiteras. "Eliciting and utilizing willingness to pay: Evidence from field trials in Northern Ghana." Journal of Political Economy 128, no. 4 (2020): 1436-1473.
Delavande, Adeline, Xavier Giné, and David McKenzie. "Measuring subjective expectations in developing countries: A critical review and new evidence." Journal of development economics 94, no. 2 (2011): 151-163.