Quality Upgrading in Ugandan Carpentry

Producers do not understand basic practices that lead to higher-quality production. Furthermore, they are not incentivized to gain this knowledge because consumers do not recognize and reward them for higher quality. Therefore, we designed interventions on both sides of the market.

On the supply side, we organized a three-day training (boarding included) for carpenters focusing on critical aspects of quality production. The training provides guidance on improving the quality of the two most common joints, teaches best practices for timber seasoning, and explains how to enhance furniture finishing. The first training started on January 8, 2024 and the last one ended on February 17, 2024.

The demand-side intervention aims to provide information to consumers. Half of the carpenters invited for the training were eligible for an information board. If they attended the training, they received a personalized board that (i) informs the customers about the carpenter's attendance and (ii) outlines the aspects they have been trained on. The information boards were delivered on March 3, 2024; it is at the carpenter’s discretion to keep them up.

2024-01-08

2024-03-03

The main outcome of interest is the physical quality of items produced by carpenters.

We will also confirm the posters are available in the workshops, as we otherwise cannot expect an impact from the information treatment.

We measure quality as an index of physical attributes measured using our quality assessment tool.

The availability of posters is a dummy variable indicating the poster was still visible in the workshop during the endline survey (this takes place about seven months after the poster distribution).

Prices, sales, and customer flow.

Prices: The price at which products have been sold.
Sales: Total sales of carpentry products in the last one / three months.
Customer flow: The number of customers who came to the shop in the last 7 days

In November and December 2021, we conducted a census in the major towns in five districts. From the 1,549 firms on the census listing, we (randomly) sampled 750 firms (i) employing at least two carpenters or three employees and (ii) who have been in existence for at least one year.

In June-August 2022, we conducted an extensive baseline survey with 748 enterprises. We collected standard information on the owner/manager and the enterprise. In addition, enumerators were trained to assess the workmanship quality of one finished item, either awaiting collection or on display for sale. We developed this quality assessment tool in the spring of 2022 in collaboration with two local master carpenters, i.e., carpentry teachers at renowned vocational schools.

In September 2022, we split the 708 firms who consented to participate in the high-frequency surveys into groups of 250 and 458 workshops, stratified by district and firm size. We surveyed the sample of 250 carpenters four times at bi-weekly intervals in October and November. We also did quality assessments of up to six products and left paper ledgers to record a targeted number of customer interactions. The remaining workshops were surveyed twice.

In the spring of 2023, we implemented a framed field experiment in two furniture markets. We procured 10 sets of identically designed small tables from nine carpenters. Three master carpenters assessed the workmanship quality of the tables. Enumerators then recruited 727 consumers and 170 industry insiders (mostly carpenters and middlemen) who visited the market. These participants were shown a selection of five tables and asked to rank the five tables by workmanship quality and to state their willingness to pay for each. Participants were then randomized to receive either a short (3-5 minutes) training on assessing workmanship quality or an active control (placebo) intervention on the benefits of purchasing Ugandan-made wood furniture rather than imported furniture. We then ask participants to update their rankings and willingness to pay. The final ranking exercise is incentivized as participants are paid a bonus that depends on how closely the rankings match those of the master carpenters.

We implemented the skills training and information board interventions in the first calendar quarter of 2024. Note that the information on the board is similar to that provided during the consumer quality experiment (which positively impacted knowledge and willingness to pay). The interventions were implemented as a randomized controlled trial. Carpenters in the sample are clustered into small geographic areas of varying radii according to the density of carpentry workshops around them. These constitute plausible "markets," and treatment status was assigned at this level, anticipating spillovers in consumer information treatment across carpenters serving the same market. The randomization is stratified on region and cluster (market) size. Based on this protocol, carpenters are assigned to one of three treatment arms: (C) pure control, (T1) training only (supply-side constraint easing), and (T2) training + information board (supply- and demand-side constraint easing).

In the summer of 2024, we conducted a second round of high-frequency surveys with the group of 250 carpenters. We did the same exercise as during part I: two months of bi-weekly interviews, quality assessments of up to six different items per visit, and the collection of ledgers.

Finally, we conducted an endline survey in November-December 2024. We received the data in April 2025.

Not available

To minimize spillovers in the skills and information board interventions, we clustered the carpenters into ‘markets’. We randomized at the cluster level, stratified by district and cluster size.

A cluster (or market)

Yes

264

Of the 750 carpenters we interviewed at baseline, 729 consented to participate in the randomization.

Control: 242 firms in 88 clusters
Treatment 1 (training only): 245 firms in 88 clusters
Treatment 2 (training + poster): 242 firms in 88 clusters