Attitudes Toward Risk and the Adoption of New Technologies Among Small Producers in Arid, Rural Regions: The case of San Juan, Argentina

Leopoldo Allub
Researcher, CONICET (Argentina)
Pasaje Costín 28(Rivadavia), 5400 San Juan, Argentina
e-mail: lallub@arnet.com.ar

Abstract

This article analyzes the independent effects of attitudes toward risk and income diversification on the degree of adoption of new farming technologies among small producers from the province of San Juan (Argentina). As a first step, we present a multiple regression model of decisional behavior in which "risk aversion" and "income diversification" are considered as variables that predict the adoption of new technologies (NT). Risk aversion --a subjective measure of risk -- is a scaled measurement that required respondents to choose from among different aspects of farming activities that they considered risky. Together with income diversification--an objective measure of risk – both variables emerge as the most important predictors of the adoption of NT. Later we develop a second model that attempts to identify the factors explaining risk aversion. Socioeconomic stratum emerges as the most important predictor for variations in the farmers' aversion to risk. Additional variables tested in the model were not significantly correlated to Risk Aversion.

Key words: Risk aversion, adoption of new technologies, diversification of income sources, multivariate sociological model, arid zones small farmers

Introduction

The present article seeks to assess if the differences in the observed levels of adopting new technologies (NT) are the result of variations in attitudes towards risk and to determine the influence of the following variables: agriculturalists' perceptions of the ecological conditions of their property, diversification of income sources, level of involvement and participation in rural development programs and socioeconomic stratum. Our data are derived from a sample of small rural producers in San Juan, Argentina.

The problem is of considerable theoretical and practical importance. A better understanding and explanation of the ways in which technological innovations are incorporated (or not) by small rural producers as part of their productive strategies enables us to formulate more efficient and effective rural development programs.

This paper proceeds in the following order: (1) Data sources and methodology; (2) determinants of technology transfers: hypothesis and variables; (3) regression models: a) variables that explain the adoption of NT, b) factors that increase or decrease uncertainty (risk aversion); (4) conclusions and implications for social policy.

Data Sources and Methodology

The study that gave rise to this article was based on data provided by the Agriculture and Livestock Social Program (PSA) in the province of San Juan. Located in an arid climate in the northern reaches of Argentina, east of the Andes, this agricultural system depends on irrigation for its production. The PSA falls under the direction of the National Ministery of Agriculture, Livestock, Fishing, and Nutrition.

An important characteristic of these producers, and the reason they have been selected for this study, is that they have never previously received technical assistance from any organization, public or private. Additionally, they have never had access to credit until the PSA program began. This presents us with ideal conditions, similar to a "natural experiment", to test the validity of competing theories of technology transfers.

Our unit of analysis for this study was the farming household. The "firm" or family business, as it is also known, consists of its productive unity, in that family labor is applied to the operation of their properties or plot, even if the producer may not be the actual owner (Nakajima, 1986).

The Theoretical Model

Our working hypothesis deals with the variables determining NT diffusion rates among small rural producers, it applies to the explanation of social behavior within this sphere only. We do not know if our theory has wider application, that is the duty of future research to determine. In this study, our purpose is to test two hypotheses. The first hypothesis can be formulated in the following manner:

H₁: The rate of NT adoption is inversely correlated with the level of risk aversion (RAI) and the degree of diversity of income sources (multiple-employment).

The second hypothesis can be stated thus:

H_2: The level of risk aversion (RAI) is inversely correlated with the producers´s socioeconomic stratum (SES), their perception of the property's agro-economic qualities, and the level of participation/involvement in PSA's economic development programs.

Together, these hypotheses form a system, given the fact that the dependent variable in the second equation is the independent variable in the first equation. To treat this as a "natural system" allows us to estimate each equation independently, as is the practice with causal models and path analysis (Blalock, 1964; Mason and Halter, in: Blalock, Jr. , 1985: 137-157).

Model 1 dependent variable: NT Adoption (Y_NT)

In our study we measured this variable utilizing a 10-item scale, consisting of four dimensions, that refers as much to the producers' subjective knowledge as to their objective agricultural practices.

The dimensions considered in the NT-acceptance scale were the following: 1) Gains reinvested to enhance productivity (chemical fertilizers, pesticides, etc.); 2) Property management technologies (labor, product diversification, intensification of production, irrigation management, etc.); 3) Mechanization of productive processes (machinery, tools, etc.); 4) Commercialization of products. For each one of these dimensions we developed a list of questions and asked the producers if they had gained specific knowledge pertaining to these dimensions since they began participating in PSA programs. If they responded affirmatively, we immediately asked if they had applied this knowledge in their productive efforts. This scale, or index, is the result of adding all the assigned values from each item corresponding to PSA learning plus the actual application on the farm.[1]

Model 1 predictor variables:

a) Risk Aversion (RAI)

Adoption is the final stage in the process of technological change ( diffusion is a stage of education that comes before adoption) There are two traditions among studies analyzing the variables influencing NT adoption: economics and sociology. In Africa during the mid-1960's, and later in Latin America and Asia, agricultural economists from academia and international organizations began undertaking detailed, microeconomic studies of domestic agricultural systems. These studies utilized stratified, random samples; repeated, structured interviews; and, frequently, closed-question surveys. Interviews were conducted using direct measures of variables such as yield per hectare, etc. The dominant trend was to analyze the objective agricultural practices with the express purpose of devising proposals to advise agriculturalists on what they should do (Norman, 1992).

However, while economists have disregarded the influence of non-economic variables, non-economists have dismissed the role of economic factors. For sociologists, attitudes, interest, entertainment, social networks, and others attributes of the adopters play an important role in determining the rate at which innovations are adopted (Coleman, Katz, and Menzel, 1957; Burt, 1987). By contrast, profit-maximization has played nearly an exclusive explanatory role in the vast majority of economic research. Unfortunately, while there have been significant contributions to each discipline's understanding of this dynamic process of change, the knowledge has not been cumulative. Researchers have concentrated on technological diffusions, contexts for diffusion, explanatory variables, and methodologies compatible with their respective disciplines despite the fact that these processes of change are not limited to, or by, any one discipline in particular.

Among agricultural economists, the neo-classical school is dominant. This perspective claims that low productivity in agricultural systems is due to a lack of NT incorporation. Adoption of innovative technologies is considered a key factor in breaking the cycle of rural economic stagnation and improving living standards in the countryside. The clearest example of how this theory has been applied is the so-called "green revolution," which sought to achieve rural development through the adoption of modernizing technologies.

In contrast to the neo-classical approach, other authors have identified an alternative underlying principal that patterns peasant behavior: "risk aversion," or uncertainty (Lipton, 1968). This evasive action is related to the conditions of uncertainty in which the rural production processes is embedded. Due to the limitations imposed by the social and natural environment--like water scarcity, rainfall patterns, etc. which are frequent obstacles in arid regions--the decision-making process for small rural producers takes place under such uncertain conditions that it is impossible to determine the probability of a given outcome.

Viewed in this light, the so-called "traditional practices" are no longer understood as economically irrational attitudes but highly evolved, time-tested strategies for minimizing risk. In other words, these are stable strategies for avoiding total loss and the subsequent disintegration of coordinated productive activities. Therefore, the decisive force influencing the decision-making process of rural productive entities is uncertainty. Seen from this contrary viewpoint, it is risk-aversion and not the neo-classical principle of profit-maximization that drives this process. Throughout, these economic actors behave more like risk-minimizers than greedy speculators, no matter the potential gains (Bunge, 1999: 343). The practical implication is that fewer resources are devoted to risky activities given the fact that a simple crop failure can threaten a family's livelihood.

In our estimation, "risk-aversion' is a psycho-economic concept. Even though economic characteristics are "objective" and real, they are also "interpreted" or perceived in distinct fashions by different actors, which elicits a variety of expectations. It is our position that the mental dimension does not stand in opposition to the behavioral aspect, they can be both "cause" (independent variable) and "effect" (dependent variable) at the same time. This means that a realistic theory of social action should not adopt a behavioralist viewpoint that concentrates solely on manifest activities while completely ignoring the subjective dimension underlying much of this action. Consequently, we should use a construct that utilizes both dimensions--we can always find objective indicators for these mental or "ideational" processes.[2]

In this study, risk-aversion is conceptualized as an endogenous variable that is related to the uncertainty accompanying change (the behavioralist version is discussed in the following paragraph.). In operationalizing our variable, we utilized some of Ghosh indicators as a reference to construct a risk-averse index or scale (RAI). (Ghosh , et al. 1994, 269-278 ). In the process, we reformulated some items in order to adapt them to the problematic environment facing farmers from our region of focus. Typically, risk-aversion is treated as a "latent" or inferred variable representing a psychological propensity that cannot be observed through direct experience but only through observable indicators.[3]

b) Diversification of income sources (INCDIVER)/ and multiple-employment

In our study we also consider the variables diversification of income sources/multiple-employment as behavioralist perspectives of risk-aversion. This assumes that, in order for this variable to be relevant, a certain amount of objective and subjective uncertainty must exist when these farmers try out new technologies like crops, fertilizers, hybrids, new tools, techniques, etc. (Feder and O'Mara, 1981). Since instability and risk are prevalent in semi-arid regions, rural families manage production and consumption-related risk by employing strategies wherever possible to diversify their income sources and/or the composition of their crops in order to ensure their standard of living. Thereby, they avoid risking everything they own on an uncertain course of action (Norman, 1973: 3-20; Valdivia and Jette, 1996 : 1329; Valdivia and Jetté, 1996b : 1). Such strategies attempt to create more predictable outcomes by reducing income fluctuations. In this study we used the concept of "diversification" in operational terms, defined by the number of productive activities in the family's economic portfolio and the contribution to total earnings (in cash and kind) from farming and non-farm production (Valdivia and Jette, 1996 : 1332). We constructed an index for the diversification of income sources by taking account of the monetary income made up of all household activities.

Empirical Models

Continuing, in this section we will present two regression models to test the hypotheses and explore themes previously discussed. The models assume that the causal relationships are linear and unidirectional in the sense that the independent variables are "causes" of NT adoption rates and not the reverse.

TABLE 1: Model 1: Regressions for Risk-Aversion (RAI) and Income Diversification/Multiple-Employment (INDIVER) on NT Adoption

a Predictors: (Constant), Risk aversion (RAI) b Predictors: (Constant), Risk aversion (RAI), Income diversification (INDIVER)

In model 1, we inserted the variables "risk-aversion" and "income diversification," one of Ghosh's indicators. The "stepwise regression" procedure was used to estimate each variable's ability to explain the variations in NT adoption rates.

The output shows the results of fitting a multiple linear-regression model to describe the relationship between NT adoption and two independent variables. In the stepwise regression the first variable selected is risk-aversion (RAI). According to the test, risk aversion explains 22.7% of the variation in NT adoption.

After adding the variable Income Diversification in the second step, the model's prediction rate improved by 17.2 [4]. The coefficient of determination, R², indicates that the model as fitted explains 39.9207% of the variability in NT adoption. One should also note that the highest P-value in the independent variables is .001, belonging to Income Diversification. indicating that the relationship is statistically significant at the greater than 99% confidence level. Consequently, we accept both variables into the model. The final equation representing this relation has the following values:

NT adoption = 5, 2799 -. 470* risk aversion - .415* income diversification + e (1)

In which the term e is the amount of unexplained variation on NT Adoption given for variables that have not been explicitly included in the theoretical system

The Analysis of Variance table (ANOVA) shows that the p-value is less than 0.01, indicating that the relationship is statistically significant given a 99% confidence interval.

The regression coefficients that measure the deviations from the regression line, are B_rai= - .470 y B_idifi = -.415. This indicates that the independent variables RAI and IDIFI are in fact negatively correlated with NT adoption. The Durbin-Watson statistic (DW = 2.273) tests residuals to determine if there is significant correlations based on the variables under consideration. Since the DW is greater than 1.4, there probably is not any serious autocorrelation in the residuals

Model 2: Regressions for socioeconomic stratum of the producer (SES);the level of farmer involvement or participation in PSA (INDIPAR), the agriculturalist's perception of the agroecological qualities of their properties (LANDQUAL) as variables that increase (or decrease) risk-aversion.

The great variety of human motivations underpinning much of the activity is produced in a social context that is conducive to modeling the attitudes and actions of the relevant social agents (agriculturalists). If risk-aversion is an individual personality trait, these people act within various social subsystems (most of which are preexisting) that contributes to the modeling of their beliefs and actions. According to this perspective, the relevant factors to consider in determining the causes of risk-aversion (or reduction) are the following: the socioeconomic stratum of the producer (as distinguished from socioeconomic status); level of farmer involvement or participation in rural development programs (an indirect measure for access to information); perception of the agroecological conditions of the properties.[5]

1) Farmer participation and/or involvement with PSA.

The neo-classical assumption is that information costs and agent knowledge is at or close to zero. However, this is a highly unrealistic supposition. In practice, knowledge acquisition is practically inaccessible to small owners. On this basis we formulated a working hypothesis that greater involvement/participation with PSA (an indicator of support-service quality) is related to lower levels of risk-aversion , and, consequently, higher NT adoption rates.

2) Perception of the agro-ecological quality of the land

Our theoretical assumption underlying the incorporation of this variable into the model is that in the moment in which the agriculturalists makes the decision to adopt a NT, they hold a certain understanding about the conditions of the natural environment they work in. This perception determines in large part the range of technological options under consideration. In arid regions where irrigation water is scarce, fluctuation (instability) and risk are prevalent considerations. Consequently, in constructing the scale for the "Perception of Agro-ecological Quality" we took into account the factors linked to the efficient use of irrigation water on the farm, according to the characteristics of the land under cultivation by the small producers.[6]

TABLE 2: Causal Model 2: Regressions for Farmer's socioeconomic stratum, Farmer participation and/or involvement with PSA and Perception of the agro-ecological conditions of the land on Risk aversion

a Predictors: (Constant), socioeconomic stratum b. Dependent variable NT adoption

These results enable us to see the advantage of adjusting the linear regression model in order to describe the relationship between the producer's socioeconomic stratum , their level of participation and /or involvement with the PSA, and their perception of the agro-ecological qualities of their land . The value for P in the Analysis of Variance table (ANOVA) is less than 0.001. This indicates a statistically significant relationship for both variables at the 99% confidence level.

The coefficient of determination, R², indicates that SES explains 29.23% of the variability in RAI. The Durbin-Watson statistical test showed a value of 1.4, indicating that RAI was not significantly correlated with participation and /or involvement with the PSA and their perception of the agro-ecological qualities of their land. These variables were excluded from the model as explanatory variables. The following equation describes the model:

Y _RAI = 4.6367 – 0. 5259* SES + e

In which the term e is the amount of unexplained variation on RAI given for variables that have not been explicitly included in the theoretical system

A special comment is deserved for the absence of an observed relationship between "participation and/or involvement in agricultural programs" and aversion to risk. One possible explanation may be the link between the efficacy of activities offered by the support program in question.

Finally, despite recurrent evidence in the literature and contrary to our expectations that risk-aversion varies greatly based on the farmer's perception of the agroecological characteristics of their holdings, our quantitative models failed to identify a statistically significant relationship between these factors. These results suggest a general need for developing a more adequate measure for this relationship.

Conclusions and Implication for Social Policy

The results concerning the factors determining the adoption of new technologies are the following:

1) We confirmed the hypothesis that the rate of NT adoption was negatively correlated with both the degree of risk-aversion and the degree of income diversification/multiple-employment among the small producers in our study.

Corollary: NT adoption in general and NT transfer programs in particular will have a greater impact to the degree they are able to reduce the level of uncertainty of outcomes in the adoption of new technologies.

2) We confirmed the hypothesis that the degree of risk-aversion is negatively correlated with the farmer's socioeconomic stratum. Specifically, producers with elevated earnings, that farm their own property, and that are more educated display greater acceptance of risk in their interview responses than do producers with lower incomes, that rent the land they cultivate, and who have less formal education.

Corollary: If the agriculturalist's earnings increase and they devote a portion to ensuring their family's welfare, this will reduce their level of risk-aversion (or it will raise the uncertainty threshold). Consequently, this will produce a greater propensity to adopt NT (In turn, this should result in greater productivity and higher earnings.).

3) Income diversification is negatively correlated with the adoption of NTs. This could be due to a diverse set of causes that requires further research. (This variable was found not to be related to risk-aversion ).

4) Contrary to our hopes, greater involvement and participation in agricultural support programs studied here do not appear to have a sufficiently significant impact on reducing the levels of uncertainty associated with NT adoption. This finding may indicate inefficiencies or the ineffectiveness in realizing the principal objectives of the development program. This proposition could be generalized to cover any scenario for NT transfer. In our study, producers regarded much of the "know how" to be too expensive to put into practice, which brings us to the crucial role that credit plays plus cost-benefits analysis would play in the final decision .

5) Finally, we regard policies aimed at making it easier for small farmers to gain access to material resource that generally require a large investment--like land, machinery, tools, etc.--can result in large failures if the farmers lack the requisite knowledge of how to use them. Gaining this information is only possible by attaining a certain level of specialized knowledge (via education). Owning the land that the agriculturalists work (owner-farmer) is a factor facilitating NT adoption, but to be more effective it should operate jointly with increased family incomes and higher educational attainment for the producers.

Corollary: Adequate( education and) training is absolutely indispensable for the successful adoption of new technologies and the improving the income and living conditions of rural families.

References

Blalock, Hubert Jr. (1964) Causal Inference in Non-Experimental Research, Chapel Hill: University of North Carolina Press.

Bunge, M. (1999) Las Ciencias Sociales en Discusión, Una Perspectiva Filosófica, Buenos Aires: Editorial Sudamericana,.

Burt, Ronald S. (1987): "Social Contagion and Innovation", American Journal of Sociology vol.92, 1287-1335

Coleman, J.S., E. Katz and H. Menzel (1957): "The diffusion of an innovation among physicians", Sociometry XX: 253-70.

Feder, Gershon and Gerald T. O’Mara (1981): "Farm Size and the Diffusion of Green Revolution Technology" , Economic Development and Cultural Change, 30,1: 59-76

Ghosh , Soumen J. , Thomas McGuckin, and Subal C. Kumbhakar (1994): "Technological Efficiency, Risk Attitude, and the Adoption of New Technology: The Case of the U.S. Dairy Industry", Technological Forcasting and Social Change 46, 269-278

Lipton, M. (1968) :"The theory of optimising peasant", Journal of Development Studies, 4 (3) : 327-351

Mason Robert and Halter A. "The Application of a System of Simultaneous Equations to an Innovation Diffusion Model " in : H.M.Blalock Jr. (Editor) ( 1985) Causal Models in the Social Sciences. New York: Aldine de Gruyter.

Nakajima, C., (1986): Subjetive Equilibrium Theory of the Farm Household, Amsterdam, Elsevier Press

Norman, D.W. (1973): Rationalising Mixed Cropping under Indigenous Conditions: The Example of Northern Nigeria, Journal of Development Studies, 3-21

Norman, D.W. (l992) "Household Economics and Community Dynamics" in: Corinne Valdivia (Editor) (1992) Sustainable Crop-Livestock Systems for the Bolivian Highlands , Proceedings of an Sr-CRSP Workshop, University of Missouri Columbia.

Valdivia, Corinne, E.E.Dunn, and Christian Jetté (1996) : "Diversification as a Risk Management Strategy in an Andean Agropastoral Community"Amer. J. Agr.Econ. 78 : 1329-1334

Valdivia, Corinne, and Christian Jetté (1996 b) "Peasant Households in Semi-Arid San José: Confronting Risk Through Diversification Strategies". Tech. Re. IBTA181/49/SR-CRS47, La Paz, Bolivia.

Acknowledgements

The present article is the result of a project named Peasant Strategies and Technological Change: A case of technology transfer for small rural producers in an arid region. It was part of a larger project under the direction of the Program for Ecodevelopment in Arid & Semi-arid Lands. The author is very grateful to Pablo Rodriguez B., a fellow at CONICET, for his substantial contribution in collecting and processing the data, and to Miguel Murmis (CONICET), Carlos Waisman (UCSD), Fernando Cortés (COLMEX), and Harold Dregne (TTU) for comments and their suggestions on earlier drafts of this paper The conclusions and any errors contained herein are my own and not the fault of my colleagues or contributors.

Endnotes

¹ "I am going to ask you some questions regarding farming practices and productivity reinvestment. We would like to know if you gained this knowledge through PSA and if you have already applied this knowledge in practice. For example: 1) Seed varieties or hybrids? 2) Fertilizers or manure? 3) Pesticides, herbicides, etc. 4) Any product or crop in particular that you were not previously familiar with? 5) Better irrigation usage? 6) To make drainage ditches? 7) How to improve cultivation techniques or how to work the land? 8) To grow a different crop than you were growing? 9) To use machinery and/or equipment to work the fields? 10) Different timing to plant or harvest than you practiced in the past (including "double harvest" or replanting)? 11) Other – for example, other ways to sell your products, etc. (please indicate).

² There is great variation in the meaning of the concept, risk-aversion. This fact impedes the comparison of cases and drawing general conclusions. Nevertheless, while there are a variety of methods for measuring risk, they are all related in some way. The way to test "utility" is to see if it offers a reasonable basis for predicting the agriculturalist's behavior.

³ "Continuing, we're going to mention some options that producers tend to encounter when making decisions related to their crops. Notice that for each statement there are two possible responses. Please, pay attention and be frank in choosing the response closest to your personal opinion. If a question is unclear, feel free to tell us so we can clarify it for you: 1)Choose between a crop with a potential for big gains as well as losses OR various crops that assure you against losses but offer small gains; 2) Fixed crop prices OR fluctuating prices that can bring big losses or big gains; 3) High production even at unstable prices OR low production at high prices; 4) A stable job that doesn't pay much OR an unstable job where you could earn a lot depending on your personal effort; 5) Have family members help with work OR have money to hire seasonal wage laborers; 6) Work the farm utilizing only your personal savings OR request credit or a loan from the bank; 7) Purchase irrigation water from another irrigator if necessary in order to cultivate more land OR cultivate only the amount of land that you have water for; 8) Two stable jobs that earn less overall OR unstable jobs that can earn you more; 9) A job where you can have great success or complete failure OR a job in which you will never have great success or complete failure; 10) Request credit from the National Bank, a cooperative, OR a private bank.

⁴ We constructed an index for the diversification of income sources (INDIVER) by taking account of the monetary income made up of all household activities. The index varies between 0 and 1 with the latter value representing a very concentrated income source (Valdivia, Dunn and Jette, 1996: 1332). Monetary income is composed of cash earnings from all household activities. Non-monetary income is composed of the following items: a) income in kind coming from the subsistence economy, and b) unpaid work performed by members of the family household engaged in activities related to their property. In our study, despite the enormous difficulties measuring subsistence consumption and its contribution to household income, the following query was included in the questionnaire: Q. 7 Do you have other resources to maintain yourself and your family if you don't have money? For example, a garden, chickens, pigs, etc. ? How important are these resources to your family income? 1) Very important 2) More or less important 3) Not important. Values for income equivalent to 10% of minimum legal salary for a rural peasant ($3,718) for those who answered "More or less important" , and 20% for those that answered "Very important" were assigned.

⁵ Nevertheless, the limited number of cases in our sample (N=51) limited the number of variables incorporated in the regression model . For this reason, and based on the previously mentioned indicators, we constructed the SES index calculated in the following formula: (Propjorn + Indipro + Eduuno) / 3.

⁶ In your opinion, is the quality of the lands you farm good, average, or poor? (check the correct box for each question): 1)Amount of salt 2)Capacity to "make the irrigation water last" 3)Ease of terrain for irrigation 4) Content of organic matter, humus, dark soil, etc. 5) Amount of resilient weeds (perennials) 6) Other qualities (please indicate)

SEA News issue #8

The SEA News index is at http://welcome.to/seanews

Copyright © Leopoldo Allub 2000
Last revised: May 21, 2003.