International Choice Modelling Conference, International Choice Modelling Conference 2017

Font Size: 
Estimating the preferences for ecosystem services and plantation characteristics: the case of private investments in Mozambique
Anna-Kaisa Kosenius, Maja Bleyer, Matleena Kniivilä, Paula Horne

Last modified: 28 March 2017


Land use changes through private agricultural investments affect local ecosystem services and livelihood possibilities. This paper focuses on tree plantations in Mozambique, the effect on ecosystem services, and public preferences for alternative plantation options, investigated using choice modelling. In addition to preferences for plantations, the study investigates livelihood activities, employment status, contacts with forest companies, perceived impacts of plantations on ecosystem services and livelihoods, and household characteristics. The study sites included villages in the neighborhood of four investment sites and one control site without current plantations, situated in the northern Mozambique.

In choice tasks, four attributes described the selected ecosystem services and characteristics of plantations illustrated with pictures. The ecosystem services studied in a survey were the availability of water (current, more available, less available) and the availability of firewood (current, easier, more difficult). The plantation was described with the distance from one’s own farm plot (far away, close to at least one farm plot) and the type and the location of land converted to plantation (outside of the village, abandoned land in the village, productive land in the village). The findings in the previous literature were exploited as preliminary information on parameter priors, and the experimental design was created with Ngene Software using Bayesian D-efficient design. In order to encourage respondents to pay attention to attributes and levels of plantation options, the presentation of choice tasks took two steps. First, the respondent chose the preferred option out of two plantation options (described in terms of ecosystem services and plantation characteristics compared to the current state) and second, they stated whether the current state would be preferable.

After three focus group test rounds, the data were collected in summer 2013 with face-to-face household interviews, applying the systematic sampling procedure, and interpreted by a native person with valuable cultural knowledge. Out of 242 households contacted, 218 answered the survey, corresponding to the response rate of 90%. Further, 206 responses were usable for the analysis. The data were analyzed with the conditional logit (CL) model and the random parameters logit (RPL) model using 500 Halton draws. The models allowed for random preference variation.

Regarding the preferences for plantation characteristics and the availability of ecosystem services, the public perceived the prevention of the decline in water availability the most important aspect. In accordance with theoretical expectations, avoiding the decline in water availability was valued higher than the improvement of the water availability. Interestingly though, improving the access to firewood was considered equally important to preventing a more difficult access to firewood. Regarding the location, the plantations situated far away were preferred to plantations close to one’s own farmland. Converting abandoned land and forestry land to plantations were preferred to the conversion on productive land. The water availability and the distant location of the plantation were perceived more important than an easier access to firewood and all other characteristics.

A more in-depth analysis was performed in order to study the relative importance of plantation characteristics and ecosystem services and the respondent-specific characteristic affecting the preferences. The preference heterogeneity was modeled with the RPL with heterogeneity in the means of random parameters, using 500 Halton draws. The stated knowledge of plantations induced stronger preference for the location far away from farmland. In comparison to males, females’ valuation of the more difficult access to firewood was lower. No preference heterogeneity was found in relation to the improvement of the access to firewood.

The basic assumption of choice modelling, that is, the respondents make their choices based on all presented attributes, may be violated either due to trials to reduce cognitive burden in the choice task or true preferences. The follow-up questions allowed for the investigation of the serial non-attendance of the attributes which may give additional insight into the perceived importance or relevance of attributes. First, the attribute describing the availability of water was the focused on by most respondents (74%). On contrast, a low share of respondents focused on the type of land converted to the plantation (5%). The further investigation of the perceived difficulty of choice tasks shows that one third (38%) of the respondents perceived the tasks very difficult or difficult, while almost a half (48%) of the respondents perceived the tasks easy or very easy. This may indicate that the low attendance to the land type attribute may be to some extent related to the task difficulty.

The results on the relative importance of ecosystem services and plantation characteristics perceived by the general public serve for the decisions concerning future plantations in Mozambique as well as in other developing countries. The analysis also shows that it is worth paying attention to the choice task difficulty and alternative attribute processing strategies when aiming at deepening the understanding of the relative importance of the studied aspects as well as when assessing the validity of the choice modelling results.

Conference registration is required in order to view papers.