Helping SMEs Climb the Export Quality Ladder - but how? Evidence from Tunisia

The treatment group receives access to:

- an information workshop held by international experts,
- a consultancy consisting of an individual diagnostic and a group-based follow-up over 4 months, and
- a small matching grant of up to 1000€/firm with a 50% self-contribution to spend on the quality assurance activities recommended in the diagnostic.

2021-06-21

2021-12-31

Data source: firm surveys; if possible combined with administrative data on firm exports.

Primary hypothesis 1: The program enables firms to export to more and higher quality destinations and/or potentially realising a quality premium on export product price.

1: Export probability (0,1): including and excluding indirect export through export-import agencies or sales to multinational firms
2: Export sales value in 2022 as inverse hyperbolic sine transformation following Mckenzie et al. AEARCTR-0003109
3: Change in unit price of main export product relative to baseline (winsorized at the 95 percentile)
4: Number of foreign clients with whom the company has business relationships (signed contracts, repeated shipments)
5: Number of export destination countries
6: A standardised export outcomes index for outcomes 1-4 based on Mckenzie et al. AEARCTR-0003109

In addition, we measure the following intermediate export indicators to capture changes in export preparation (given changes in export outcomes may take longer than 1 year to materialise) and export perception (to account for Breinlich et al. 2017 and Kim et al. 2018 who find information interventions discouraged non-exporters):

1: Export preparation index (based on Kim et al. 2018, with additional answer options)
2: Export perception index (based on Breinlich et al. 2017)

If possible to access administrative data:
1: Number of distinct products exported in 2022 (6-digit product classification) based on Mckenzie et al. AEARCTR-0003109
2: Number of distinct product-country combinations in 2022 (6-digit product classification) based on Mckenzie et al. AEARCTR-0003109

Primary hypothesis 2: The programs increase firms knowledge and adoption/use of quality standards and quality assurance services for exporting.
1: Knowledge: Self-constructured Quality infrastructure knowledge index based on quality infrastructure knowledge measurement tool
2: Perception of quality management: likelihood to express positive views on costs and complexity of quality management (in line with Breinlich et al. 2017 and Kim et al. 2018 one can also expect the opposite, namely that participants perceive quality management even more complicated after the treatment)
3: Use/adoption: 3.1.: Likelihood to state to convince clients through a formal (reference to quality standards etc.) rather than informal (reference to experience, superior taste, functionality etc.) definition of the quality of their products. 3.2.: Adoption of a higher number of selected quality management practices based on Mckenzie/Woodruff/Bloom et al. 3.3.: Likelihood to control inputs from suppliers, respond to technical requirements from clients
4: Changes and normalised levels in firms annual expenditure on quality assurance

Secondary hypothesis 1: The program helps firms innovation thanks to (more) product tests necessary for conformity assessment with foreign technical regulation and international quality standards.

1: Proxy for direct program effects. 1.1.: Likelihood of testing products in an internal or external accredited laboratory. 1.2.: Likelihood of modifying products or production processes due to/thanks to test results (based on an example)

2: Proxy for learning-from-exporting: likelihood to have engaged in any of the four innovation modes documented in the Oslo Manual and Community Innovation Survey

Secondary hypothesis 2: The program helps firms increase their productivity based on the adoption of quality standards and quality assurances, such as product tests and measurement instrument calibration.

Labor productivity
Total factor productivity

Other outcomes
Inverse hyperbolic sine of annual profits
Inverse hyperbolic sine of annual sales
Number of employees

For sampling, an open, public communication campaign was run through direct emailing by the project team on the basis of the national registry of industrial firms (API), a sponsored Facebook/Linkedin campaign, and official political partner institutions websites and email lists. Interested SMEs could sign up through the website of the program through a short registration form. In total, 266 SMEs signed up after two month of communication between December 2020 and January 2021, among which 214 complied with the eligibility criteria (6-200 employers, industrial activity, exporting or intention to export in the future, not being part of a foreign multinational company).


Among the eligible firms, 107 firms were allocated to a treatment and 107 firms to a control group on June 8th 2021. The randomisation was stratified based on four sectoral groups and the certification status of the company. In four strata we further subdivided into smaller strata if there were sufficient observations and high variation in the percentage of sales firms’ realised with exports. Given the relatively small sample size, we also re-randomised based on the rule that none of 7 baselines covariates (e.g. number of export countries, total export sales in Tunisian Dinar in 2020 etc.) were independently or jointly significantly different for treatment and control group( “big stick” method as documented in Mckenzie and Bruhn 2009, Morgan and Rubin 2012, Soares and Wu 1983).

The allocation of firms in the treatment and control group was announced at registration for the program and communicated publicly on the program's website to ensure transparency and equality of opportunity. The randomisation was done by the researchers in Stata, recorded and made public on the project’s website.

Public computer randomisation

Firms

No

0

214

~ 107 treatment group, 107 control group