Striving of for Quality - Experimental Evidence of the Impact of Quality Certification on Firms and Networks in Rwanda

Our study will provide rigorous causal evidence on the impact of certification of the Rwanda Standards Board’s (RSB) Zamukana Ubuziranenge (ZU, Kinyarwanda for "Striving for Quality") program. The ZU program offers structured technical assistance to small and medium enterprises (SMEs) to meet national and international quality standards, including S-mark and ISO certifications. RSB begins the intervention by conducting a “Gap Assessment,’’ benchmarking current practices against such standards. This benchmarking activity will include lab testing of products where relevant.

The RSB will then deliver the ZU program to treated firms, including: i) a customized roadmap to certification; (ii) hands-on support to establish compliance systems; (iii) a post-assistance assessment of certification readiness; and (iv) full subsidization of certification fees.

2026-03-15

2028-12-31

1. Efficacy of RSB ZU program on obtaining quality certifications
2. Impact on market access
3) Impact on firm performance: revenue, employment, and gross profits.

1. Efficacy of RSB ZU program on obtaining quality certifications. This outcome will be measured using the RSB certification registry, tracking whether treatment firms are more likely to obtain quality certifications (ISO, S-mark, HACCP).

2. Impact on firms’ market access: (i) international market access, as measured by the probability of exporting, export volumes, the number of destinations, and how quality-sensitive the destinations are; and (ii) domestic market access, as measured by the number of domestic buyers and the quality-sensitivity of these buyers (as proxied by being a multinational firm, exporter, or large buyer).. These outcomes will be measured using administrative tax records from the Rwanda Revenue Authority (RRA). More specifically, international market access will be measured using customs records, whereas domestic market access will be measured using domestic firm-to-firm sales data from Value Added Tax records.

3. Impact on firms’ performance: revenue, employment, gross profits. These outcomes will be measured using administrative tax records from the Rwanda Revenue Authority (RRA). Revenue information will be obtained from the Corporate Income Tax records. Employment information will be obtained from Pay-As-You-Earn payroll data, which may be complemented with surveys and/or Rwanda Social Security Board data. Gross profits information will be obtained from the Corporate Income Tax records.

1. Signaling vs. quality upgrading

2. Effect on buyer-supplier production networks

1. Signaling vs. quality upgrading. We aim to distinguish between two channels: the impact of signaling existing quality levels and the impact of quality upgrading. For this purpose, we will use three measures of quality: (i) gap assessment reports, (ii) product quality lab tests; and (iii) survey measures of quality. We will use these measures to estimate treatment effects on quality. We will also examine heterogeneity by baseline measures of quality. The impact on firms with relatively high baseline quality is likely to be mostly due to signaling, whereas the impact on firms with low baseline quality will be a mix of signaling and quality upgrading.

2. Effect on buyer-supplier production networks. We will analyze spillover effects on the performance of incumbent suppliers and buyers of treated firms, as well as effects on treated firms’ selection of suppliers. This information will be obtained from firm-to-firm transaction data obtained from the Value Added Tax records of the RRA. This information may be complemented with Electronic Billing Machines (EBM) data from the RRA, if data access is granted.

The RCT is being implemented in six steps:
1. Identification of Target Firms. Using a combination of Rwanda Revenue Authority (RRA) administrative data, lists of firms shared with us by the RSB, and supplementary lists from the Private Sector Foundation of Rwanda and the Rwanda Ministry of Commerce, we compile a list of target firms to recruit.
2. Recruitment Campaign and Screening. The RSB and researchers will jointly recruit firms in sector-based batches. In Wave 1, for which recruitment is in progress, we focus on agribusiness firms, generating a list from all sources of around 1,000 firms. Waves 2 and 3 will target other sectors, likely including other types of light manufacturing firms, following RSB priorities for each specific program cycle. The recruitment campaigns will inform firms about the ZU program, explain the application process, and encourage participation. Of the 1,000 firms we included in Wave 1 recruitment, we have thus far generated 149 applications to the ZU program, and continue to receive applications. The RSB will then provide an initial screening of applicants to determine whether the firms are eligible for the ZU program. The pool of eligible firms will comprise our sample for that wave.
3. Lab Testing, Gap Assessment, and Baseline Survey. With all firms in our sample, the RSB will then conduct a “Gap Assessment,’’ benchmarking current practices against standards. This benchmarking activity will include lab testing of products where necessary. The researchers will also conduct a baseline survey. All three of these steps – Gap Assessment, lab testing, and baseline survey – will be conducted with the entire sample (treatment and control firms) before randomization.
4. Randomization. Fifty percent of the firms will then be randomly selected via a lottery to receive the ZU program (treatment), with the rest serving as controls. This ensures equitable allocation of limited slots while enabling rigorous evaluation.
5. ZU Program Implementation. The RSB will deliver the ZU program to treated firms, including: i) a customized roadmap to certification; (ii) hands-on support to establish compliance systems; (iii) a post-assistance assessment of certification readiness; and (iv) payment of certification fees.
6. Impact Measurement. We will assess the program’s causal impact using rich administrative data from the RRA and RSB, as well as original firm surveys run at baseline and 12 months after each wave’s program completion. With this data, we will measure the effectiveness of this major public policy program, and answer our key research questions:
i. What is the impact of certification on firm’s market access, as well as other outcomes (sales, profits, employment)?
ii. Does certification drive quality upgrading or mainly signal existing quality
iii. How does certification affect buyer-supplier networks, through network restructuring and upgrading spillovers?

Not available

Randomization will be carried out by the research team using a computer.

Randomization will be carried out at the firm-level

No

We will not randomize in clusters

600 firms

300 treatment firms and 300 control firms. The RCT will be rolled out in three waves with an approximately equal split.

Power Calculations and Sample Size. The RSB has shared with us 15 years of historical certification and firm-level quality testing records, which we have merged with administrative tax records at the RRA to conduct an initial (propensity score matched) difference-in-differences analysis, comparing firms in the year they received their first certification through the RSB to matched firms that have never been certified. Results for revenues and employment are provided above. They hint at large impacts of certification, with point estimates suggesting 20-35% increases in revenues and 20-30% increases in employment two years after certification. We use these results, along with additional administrative RRA data and information from our recruitment survey of firms, to conduct our power analysis. The average revenue of firms who in our survey express interest in obtaining a certification is 247M RWF and the standard deviation is 241M RWF. Given that we have access to panel administrative data, we carry out our power calculations assuming 3 baseline measurements and 3 follow-up measurements, which show within-firm correlation of revenues across periods of 0.76. Our estimates suggest with a sample size of 600 firms, we will be able to detect an effect size as small as 0.09 standard deviations. This is smaller than the most conservative estimated treatment effects found in the observational literature on the impact of ISO certificates (which typically use propensity score matching): 0.16-0.27 standard deviation increase on total exports (Sun and Ouyang, 2010; Volpe Martincus et al., 2008), 0.24 standard deviation increase on labor productivity (Calza et al., 2019) and a 0.5 standard deviation improvement in management practices (Alfaro Serrano, 2019). This is also smaller than the 20% increase in revenue (0.13 SD) effects we find above in our own setting in Rwanda using our difference-in-differences estimate.