Terrestrial aquaculture and agriculture in Rondônia: an economic analysis

How to cite

Morello, Thiago; Addicott, Ethan; Pereda, Paula. 2026. "Terrestrial aquaculture and agriculture in Rondônia: an economic analysis." AEA RCT Registry. AEARCTR-0019218. https://doi.org/10.1257/rct.19218-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Thiago Morello* , Ethan Addicott , Paula Pereda

* Primary investigator

Registry ID
AEARCTR-0019218
DOI
https://doi.org/10.1257/rct.19218-1.0
First registered
Last updated
Status
In development
Trial dates
2026-07-27 – 2026-11-30
Countries
Brazil (State of Rondônia)
JEL codes
Q22, Q01, C25
Registry record
http://www.socialscienceregistry.org/trials/19218

Abstract

Fish-farming is growing worldwide, especially in developing countries, where it is becoming an alternative source of protein comparable to cattle rearing. Fish farming is also expanding in the Brazilian Amazon, where cattle rearing remains the main usage of already deforested land. Nevertheless, in such region, many barriers prevent fish-farming from becoming comparable to cattle in terms of farm profit and protein supply. The main barriers being the high market risk, due to a still incipient regional/national market, the lack of skilled labour, and the scale economy implied together with lacking credit access. Upon this background, we ask two research questions, having as our target subject the average Rondônia farmer with minimal fish-farming knowledge. First, what is the willingness to accept compensation (WTA) for up-taking a contract offered by a hypothetical new processing plant, as means to increase the share of farmland occupied with sustainable fish farming? Second, which are the marginal utilities for the five key attributes capturing the trade-off between (i) non-monetary costs of adopting good and sustainable fish-farming practices and (ii) better conditions for selling fish and purchasing inputs? These key attributes are: (i) required pond area to be built in the farm, (ii) good and sustainable fish-farming practices, (iii) practice non-compliance penalty, (iv) environmental sustainability label and (v) supplementary payment for fish delivered.
We test ten main hypotheses, namely: (1) the average WTA is statistically significant and larger than the maximum level of profit yielded by fish-farming in the study region, (2) the propensity for up-taking a contract is statistically equal among classes of fish-farmers, (3) marginal utility for required pond area is null, (4) marginal utility for good practices is null, (5) marginal utility for non-compliance-penalty is null, (6) marginal utility for the environmental sustainability label is null, (7) marginal utility for the supplementary payment is null, (8) marginal utility for a contract-based alternative (i.e., the reverse of the status-quo) is null, (9) smallholders not having fish-selling experience are not more likely to opt-out of contracts, and (10) cattle ranching and fish-farming are substitutes when taking all farmers as a whole, but become complements at large farm sizes. A sample of farms with satellite-detected ponds is randomly selected for interviews. All respondents are interviewed in-person and on-field by the research team and hired enumerators. By the time of this registration, only the pretest (pilot) survey was conducted (the full-fledged survey was planned to take place after this registration).

Intervention

Our intervention is a discrete choice experiment (DCE) conducted with a sample of farmers having a satellite-detected pond in their farm. The DCE is conducted as an in-person interview on-field in rural areas of Rondônia state, western Brazilian Amazon. The good valued is a contract requiring that sustainable fish farming be expanded in the farm and offering better sales and input access conditions. The background scenario is the arrival of a new fish processing plant in the region.
At baseline and endline, respondents are exposed to the DCE, which contains five attributes, presented as characteristics of a fish sales contract: (i) size of new earthen ponds required to be built in farm, (ii) good fish-farming practices required; (iii) (monetary) penalty for non-compliance with the practices, (iv) environmental sustainability label, (v) supplementary payment for fish delivered. These attributes are presented in five choice sets, each with three alternatives. Two of the alternatives are two contracts (different combinations of attributes' levels) and the third is the option for not signing a contract (status quo or "opt-out" option). There are also four attributes that do not vary across contract-based alternatives: (i) contracts' five-year term, (ii) fish to be sold coming exclusively from the new ponds, (iii) need to request an environmental license for the new pond, (iv) full coverage of the production cost. Besides choosing the most preferred alternative in the choice sets, the questionnaire requires data about (1) experience with fish farming, (2) attitude regarding contracts, (3) socioeconomic and land usage profile, (4) self-evaluation of the interview by the enumerator.

Experimental design

Regarding sampling, respondents were selected from a third-party (private company, Bussola.Farm) georeferrenced dataset containing spatial points (geographical coordinates) where water ponds were detected with high-resolution satellite imagery. The data was confidentially shared with the principal investigator (PI). The PI eliminated water ponds classified as inactive fishing ponds in the original data, and kept only the points within 15 km of the boundaries of the Ariquemes municipality (this municipality is a top producer of farmed fish of Rondônia state). From these points, the farms visited in the pretest and full-fledged survey were selected.
A D-efficient design was employed to select the levels of the five attributes for each contract-based alternative of the choice set (Stata's user-written dcreate package was used). The data will be analysed with mixed-logit (allowing for random marginal utility coefficients), latent class logit, conditional logit and (generalised) multinomial logit estimators (other estimators may be also used). Hypotheses will be tested based on point estimates and standard errors generated by the estimators.
In particular, we test ten main hypotheses, namely: (1) the average WTA is statistically significant and larger than the maximum level of profit yielded by fish-farming in the study region, (2) the propensity for up-taking a contract is statistically equal among classes of fish-farmers, (3) marginal utility for required pond area is null, (4) marginal utility for good practices is null, (5) marginal utility for non-compliance-penalty is null, (6) marginal utility for the environmental sustainability label is null, (7) marginal utility for the supplementary payment is null, (8) marginal utility for a contract-based alternative (i.e., the reverse of the status-quo) is null, (9) smallholders not having fish-selling experience are not more likely to opt-out of contracts, and (10) cattle ranching and fish-farming are substitutes when taking all farmers as a whole, but become complements at large farm sizes.

Design details

Randomization method
Randomization of sample done in office by a computer with the R software random numbers generator. DCE design based on Stata's dcreate user-written package, also done in the office.
Randomization unit
Satellite-detected pond.
Clusters
Discrete choice experiment: NA
Observations
Discrete choice experiment: 225
Arms
Discrete choice experiment: 225 (only one arm); this is the planned sample size; a smaller sample (of, for instance, N = 135), may be imposed by budgetary limitations and difficulties faced during on-field data collection (especially long distances).
Primary outcomes
(1) Willingness to accept monetary compensation to uptake a contract, (2) marginal values of the contract's attributes (pond area required, good practices required, non-compliance penalty, environmental sustainability label, supplementary payment; and as well as a binary variable identifying contract-based alternatives) and (3) contract uptake proportion.
Sponsors
Name: National Council of Technological and Scientific Development Location: Brazil Url: https://www.gov.br/cnpq/en
Partners
Name: Bussola.Farm Partner_type: private_company Url: https://bussola.farm/

Cited by

No citation data available for this registration yet.

Keywords