Principal investigator: Viviane Pires Ribeiro
Article title: The Recent Experience of PRONAF in Pernambuco: An Analysis Using Propensity Score
Article authors: André Matos Magalhães, Raul Silveira Neto, Fernando de Mendonça Dias and Alexandre Rands Barros
Location of the intervention: Pernambuco, Brazil
Sample size: 4.500 family farmers
Main theme: Economic Policy and Governance
Type of Intervention: Impact of PRONAF on income
Primary variable of interest: Income
Evaluation method: Propensity Score Method
Evaluation Context
The National Program for Strengthening Family Farming (PRONAF) was created in 1996 by the federal government and since then it has been one of the main Brazilian policies in the social and agricultural areas. One of the program's main objectives is to reduce poverty affecting family farmers in the country, thus seeking to ensure access to affordable credit for small farmers and integrate them into other rural development policies, such as support for infrastructure development and technical assistance. In the period from 1998 to 2001 alone, the program signed 2,543 million contracts throughout Brazil, spending more than 6,9 billion reais.
The PRONAF program divides farmers eligible for credit into four groups (A, B, C, and D), classified according to their gross annual income and the amount financed. Group A consists exclusively of those settled under the National Agrarian Reform Program, regardless of their income bracket. Between 1998 and 2000, the other groups were classified according to the following gross annual income ranges: from R$ 1.500,00 to R$ 1.500,00 for farmers in group B; from R$ 5 to R$ 8.000,00 for farmers in group C; and from R$ 8.000,00 to R$ 27.500,00 for farmers in group D.
Intervention Details
Magalhães et al. (2006) evaluate the experience of PRONAF in the Brazilian state of Pernambuco. The research is conducted using data from a field study carried out by FADE-UFPE with family farmers in Pernambuco, during the months of August to November 2001. The interviews were conducted in more than 60 municipalities located in the state, totaling approximately 4.500 small farmers. Two-thirds of the sample consisted of PRONAF beneficiaries and one-third consisted of non-beneficiaries, but with a socioeconomic profile similar to the first group and residing in the same region as the beneficiaries. This second group was used as a control group and served as a basis for comparisons with the beneficiaries.
The impact of PRONAF is evaluated through three specific variables: the value of production, the value of production per hectare, and the value of production per person engaged in production. These variables were chosen by the authors because they are representative of two of the main objectives pursued in the study: i) income growth, denoted by the growth in the value of production; ii) growth in the capacity of family farmers, represented by the growth in the value of production per hectare and per person engaged in production.
Methodology Details
Effectiveness analysis, according to Magalhães et al. (2006), seeks to verify whether the public policy was able to generate the expected results. In this sense, the authors analyze the effectiveness of PRONAF (National Program for Strengthening Family Farming) through its impacts on selected production variables. The impacts are considered separately for PRONAF groups (B, C, and D) and for the total sample. Therefore, estimates of the program's impact on the value of production, the value of production per hectare, and the value of production per person on the rural property are obtained for the total and for the three groups considered in PRONAF using three estimators: the difference of means without control, the difference of means from Least Squares regressions, and the estimator of... Propensity Score.
Results
The results found by Magalhães et al. (2006) indicate that PRONAF was not very effective in Pernambuco until 2001. When the authors consider the differences in characteristics and probabilities of participation in the program, the results indicate that the program has not generated a significant impact for the target audience in Pernambuco.
The estimates obtained from the comparison without any control for the variable "production value" indicate a positive result for the program, but this result, for all groups and for the entire population, is fully explained by the different characteristics of the agents and the different probabilities of participation: the differences decrease when the Least Squares estimator is considered and disappear when the Propensity Score estimator is considered. Thus, the evidence indicates that the program was not effective in its effects on the value of production, neither for the entire population nor for any of the groups considered for the period investigated.
For the variable "value of production per hectare," the estimates indicate that when the entire universe is considered, a negative effect of the program emerges, present in the evidence generated by all three estimators. However, when considering the evidence by groups, only the rural properties in the group, in their estimates with controls, do not present the negative result obtained in the aggregation. This result suggests that the program has encouraged a more extensive occupation of the property without, however, having led to increases in production at the same rate, thus evidencing a loss of productivity.
Estimates of the variable "value of production per person" follow the same pattern obtained for those of "value of production." Although positive, they decrease in value as more controls are introduced into the estimates.
Lessons in Public Policy
What is the impact of PRONAF on the income of beneficiaries residing in Pernambuco? The study conducted by Magalhães et al. (2006) suggests that the impact of PRONAF on the income and productivity of its beneficiaries was quite reduced and even negative in some cases. According to the authors, this result is consistent with those found in other studies on the program, except for those conducted in Southern Brazil. Thus, these contradictory results suggest that there may be a conflict faced by a program aimed at family farmers who have or may have participation in markets, but which also aims at social policies for the entire group of family farmers. Therefore, the selection bias, the effectiveness problems, and the reduced effects on income/productivity in states where family farming is not integrated into agribusiness, seen in light of other evaluations, suggest that the program may be succeeding only in localities where this integration is already established, with timid results in others.
References
MAGALHÃES, André Matos et al. The recent experience of PRONAF in Pernambuco: an analysis using propensity score. Applied Economics, v. 10, n. 1, p. 57-74, 2006.
