ABSTRACT Climatic variability, characterized by periodic droughts and El Niño events, and weather events such as frosts, affect the production and consumption decisions of rural households throughout the Andean region. Both are important abiotic factors affecting the choice and mix of crops and livestock, their productivity, and therefore household food security. Important food crops, such as potato and quinoa, and livestock are affected by these events. Our research aims to answer the following questions: 1) What have farmers developed as successful strategies to cope with climatic variation in the Andean region; 2) How do farmers currently use information from forecasts and local sources to make production and consumption decisions; and 3) What mechanisms and institutions facilitate or constrain the utilization of information about climatic risk.

Our unit of analysis is the rural farm household, in its relation with markets and their communities. We focus our research on decision making, analyzing household production strategies, the portfolio of economic activities, and the role of diversification in mitigating risk.. Our purpose is to identify distinct groups of farmer strategies, evaluate these in terms of risk, productivity and net returns. To measure the effect of drought and frost on production we compare data from three years in the case of Bolivia (1993, 1995, 1999), and two in the case of Peru, and contrast these outcomes with those generated by existing and modified crop and livestock models, to evaluate actual economic returns, and the potential returns of incorporating climate and weather forecasts.

We will conduct our research at two sites in the Altiplano region, Aroma in Bolivia, and Puno in Peru. Our research the first year focuses on identifying household production strategies and economic portfolios in Bolivia near an El Niño event, and compare results to previous years. Identified factors that affect use of information will be studied the second year through case studies. The approach to evaluate household strategies will be validated in Peru by conducting a similar survey the second year, in order to identify groups of farmers with distinct economic portfolios. We will also identify in Peru the factors affecting the use of climatic information and weather forecasts through in-depth case studies the third year. The research will be conducted in collaboration with the International Potato Center, the Centro de Investigación de Recursos Naturales y Medio Ambiente (CIRNMA), the United Nations Development Program Bolivia currently assessing the impact of El Niño 1997-1998, and PROINPA, a foundation promoting research on Andean crops, and farmers from rural communities in the region.
Period: August 1, 1999 through July 31, 2002

RESULTS FROM PRIOR RESEARCH Title: Sociological Analysis of Small Ruminant Production Systems in Bolivia and Kenya, Small Ruminant-Collaborative Research Support Project

Sustainable Agropastoral Systems on Marginal Lands in Bolivia
The Kenya Dual Purpose Goat
Agency: United States Agency for International Development USAID
Principal Investigators: Michael F. Nolan and Corinne Valdivia
Period Award: October 1 1994 to September 30 1995 and October 1 1995 to September 30 1996
Total Award: $132,324 and $179,530
Grant No: DAN-1328-G-00-0046-00 Subgrant Nos:105-12-17 and 105-12-16
Bolivia Research Relevance: The theoretical framework and methodologies evaluate household diversification and understand the role of economic activities in production and consumption were developed through this study. Two data bases were constructed, one for 1993 an average rainfall year and the second for 1995, a drought year. These two points in time allowed us to understand the role of access to land resources, labor, animals, stage of the life cycle, and assets in the identification of distinct household economic strategies. The project focused on the role of livestock. A finding of the study, relevant to this proposal, is that in crop livestock systems potato production is crucial as a food security mechanisms that will be undertaken by all the families surveyed, with no exceptions.

Kenya SR-CRSP Research Relevance: This study consisted of the social, economic and environmental evaluation of a technological package for resource poor households in semiarid conditions. Baseline and monitoring data from households in two agroecological sites were selected to evaluate the adaptation of a technological innovation. The household approach to analyze strategies in dealing with drought are being tested. A study on the role of networks and social capital was completed, and will inform our research on social and market mechanism to access climatic and weather information.

Title: Sociological Analysis of Small Ruminant Production Systems, SR-CRSP
in Peru, Morocco, Brazil, Kenya and Indonesia.
Agency: USAID Grant No DAN-1328-G-SS-4093-00 1979-1985
Principal Investigators: Michael F. Nolan and Jere L. Gilles
Period of Award: October 1 1979 through September 30 1985
Total Award: $1'150,000
Relevance: This was the sociological portion of an interdisciplinary research program that looked at Small Ruminant Production Systems world wide. Only results from Peru and Morocco are relevant to this proposal.

Research in Peru and Morocco focused on the role that indigenous knowledge and institutions play in adapting to climatic risk. In Peru research looked at response to the El Niño of 1983 in Piura and the Altiplano, and at indigenous resource management systems. Research in Morocco looked at the "agdal", Berber system for managing resources in drought prone areas.

Title: Rockefeller Dissertation Scholarship Dekha Sheikh
Agency: Rockefeller Foundation
Principal Investigators: Maury E. Bredahl and Corinne Valdivia
Period of the Award: July 1998-June1999.
Total Award: $17,873
Relevance: This dissertation research project will measure the sources of income variability and possibilities of smoothing consumption in a semi-arid and a sub-humid environment of Kenya. Its relevance to this project is the development of a method to measure diversification that incorporates probabilities of loss, and a method to measure covariant activities for multi variate regression analysis. A doctoral student is being guided in this research by the P.Is.

Title: Returns to Research Investments in the Indonesia Small Ruminant Collaborative Research Support Program (SR-CRSP)
Agency: USAID through the Bean Cowpea CRSP
Principal Investigator: Corinne Valdivia
Period of the Award: October 1 1992 - September 30 1993
Total Award: 20,000
Relevance: Social returns to research of management technologies were assessed with a consumer producer economic surplus model. This model is useful to measure the potential benefits or gains resulting from increased productivity. It allows incorporation of market characteristics in the measurement of the social returns of public investments. The methodology may be applied to measure the impact of improved climate and weather forecasts.

Title: Sustainable land use in the Andes
Agency: Several through the International Potato Center
Principal Investigator: Roberto Quiroz
Period of the Award: 1998-2002
Project No 14,CIP
Relevance: To evaluate alternative land use management strategies and intensification, developing linkages with CONDESAN (consortium for the development of the Andes) and with the Global Mountain Program. This project consists of several sub-projects, including the development of decision support tools, the use of remote sensing, GIS, and modeling to monitor land use in semiarid arid Andes; Improving the mixed crop-livestock systems in the Andean Eco-region to identify low risk interventions. These activities are on-going and relevant to our proposal because of the models developed and those that are being improved. Linkages will be built with the regional participants in this project to share our research results.

STATEMENT OF WORK

The Problem

Introduction
Climatic variation and the risk that it entails, is a key feature of agriculture in the semi-arid tropics (Gill, 1991). Although farmers strive to develop strategies to minimize the impacts of these variations, low productivity and food insecurity are a result. In the Andes, producers are faced with periodic drought, frosts and El Niño events (Washington-Allen, 1993). These events shape the management of crops and livestock, agricultural productivity and total rural income. The research outlined below addresses three questions related to this issue: 1) What have farmers developed as successful strategies to cope with climatic variation in the Andean region; 2) How do farmers currently use information from forecasts and local sources to make production and consumption decisions; and 3) What mechanisms and institutions facilitate or constrain the utilization of information about climatic risk. This research will identify current household strategies to cope with risk and describe how climatic data is used to reduce losses in order to improve food security in Andes.

Climatic perturbations in the Andean Region
Agriculture is the main source of income for rural people in the Altiplano of Bolivia and Peru, and climatic perturbations are the determining factor for agricultural production in the region. (Le Tacon, 1989). The region is characterized with high diurnal temperature variations, frost risks, low and irregular precipitation and high risks of drought during the growing season (Francois et al., n.d.). Droughts also increase the occurrence of frost, compounding the risks faced by farmers. Scientists and farmers associate typical El Niño Southern Oscillation (ENSO) events with drought in this region (Woodman 1997; Washington-Allen, 1993; Céspedes and Rodríguez 1995; Ccama, 1991).

Potatoes are the main staple and an important cash crop in the region (Ccama, 1991; Montes de Oca, 1992). The production of commercial varieties of potatoes is especially sensitive to droughts and frost. Local varieties, are more resistant to these perturbations and used for subsistence. Work is currently being conducted to improve the prediction of precipitation in the Lake Titicaca Basin. Preliminary results show that it is possible to forecast rainfall from January through March several months in advance (Personal communication with Pablo Lagos, Instituto Geofisico del Perú). These advances create the possibility of improving farm decision making in the region in ways that will enhance food security. However, will these forecasts improve or change decision making?

Relevance
Andean agricultural production systems are sustained by a combination of crop and livestock activities. The Andean region is marked by food shortages and has some of the lowest levels of per capita caloric intake in this hemisphere (van Haeften 1996a). Two thirds of the rural residents of Peru's highlands are poor, nearly half of them are extremely poor, and their Bolivian counterparts live under similar conditions(van Haeften, 1996b). The extreme poverty of the region and periodic natural disasters have forced thousands of people to migrate to urban slums and to the jungle. Often these migrants provide the labor force for drug production in the jungle.

A study conducted in three agropastoral communities in the Altiplano of Bolivia showed 30% of children under five to be malnourished (Murillo and Markowitz,1995). Their nutritional status is directly linked to the ability of their parents to produce food, and the ability to manage drought and frost is a key element to improving the quality of life of rural families. Balancing livestock, with the production of forages, food crops such as potato and quinoa production, is an important part of any strategy designed to improve food security in the Andean region. Systems research shows that sheep in this region are particularly important in the most vulnerable segments of the population (Valdivia, Dunn and Jetté, 1996). Livestock are not only an important source of income in the risky Andean environment, but they are used to reduce the risks inherent in crop production and provide the reserves needed to recover from natural disasters. The manipulation of the family's crop and livestock portfolios is an important risk management strategy.

The value of forecasts in decision making in a peasant agriculture context
Farmers in the Andes use their traditional knowledge to predict the level of rainfall for the next season, some times eight months in advance. Weather observations at specified times, as well as other signs, such as birds and insects, are used to forecast climate (Hatch, 1981). The ability to predict correctly is important because it affects the quantity of fallow fields that will be plowed, the type of crops and varieties that will be planted, and the type of inputs that will be used (Huanca, Markowitz and Jetté, 1994; Céspedes and Rodríguez, 1995). In other words, the production strategy and corresponding portfolio are defined by those predictions. Market prices, the quantity and quality of land labor and animals, are important, especially in determining the proportions that will be kept for consumption and that will be sold (Valdivia et al. 1995), but climate is the driving force.

Climatic information is valuable when it contributes to decisions that improve earnings or reduce losses (Nicholls,1996). Most studies on the use of forecasts have concentrated on temperate developed countries, with reliable markets for inputs and commodities and insurance. They have focused more on the benefits of forecasts rather than the costs of using them (Nicholls,1996).

This study considers the value of information under conditions of food insecurity, imperfect or incomplete markets, unequal access to information, and production decisions that are conditioned on ensuring a certain level of food security (Kusterer, 1989; Ellis, 1993; Valdivia et al., 1996; Dunn et al., 1996). In the Andes farmers cannot rely on markets to access food and inputs on a timely and affordable way. As a consequence, any objective of maximizing net returns is conditioned on not endangering their present level of welfare. This means that most farmers will not engage in activities that will endanger their present living standards (Kusterer, 1989).

We propose that access to climate and weather forecasts will reduce uncertainty, and allow farmers to choose portfolios that improve their returns. This study will contribute to measuring the value of improved information for management of resources and household portfolios and will identify the groups that have the capacity to use this information. Production and consumption decisions of farmers and their families, and the role of information and traditional knowledge (Hatch, 1981), are necessary to determine the ways in which climatic predictions may be used to improve production strategies shifting the portfolio to more reliable crop varieties and agricultural activities. The next section addresses first what we know about decisions in semiarid environments, and what we have learned about the Altiplano.

Theoretical Considerations

Household decisions in an uncertain environment
Rural households in many developing countries are characterized as peasant producers (Ellis, 1993), meaning that their production decisions affect their consumption decisions and viceversa; the markets in which they participate are unreliable and incomplete (Ellis, 1993); many of the resources that they manage are accessed through non-market relations. Many uncertainties affect their daily decisions, caused by climate, economic, political and idiosyncratic events. Farmers, because of these conditions and their low levels of income are in various degrees, mostly risk averse (Sadoulet and de Janvry, 1995). No insurance markets exist to, for example, protect them from droughts or the fall of market prices. Under uncertainty, farmers manage risk ex-ante by diversifying the portfolio of economic activities to smooth income through the year (Dunn et al. 1996), conditioned on their level of wealth, weather sensitivity, and ability to cope with negative outcomes (Alderman and Paxson; Carter, 1997; Valdivia et al, 1996; Bromley and Chavas, 1989; Panter-Brick and Eggerman, 1997). Activities may include off- farm employment (Low, 1986), spatial diversification of crops, staggered planting, and choice of economic activities that are not covariant (Reardon et al, 1992; Robinson and Barry, 1987). Rosenzweig and Binswanger (1993) found that in riskier environments, portfolio assets less sensitive to weather but less profitable were chosen. On the other hand others propose that diversification may be a means to maximize use of all resources available to the household (Ellis, 1993; Reardon et al., 1992; Valdivia et al., 1996).

The degree of covariance between various income earning activities should be low or negative in order for the portfolio to better insure against risk (Reardon et al, 1992; Robinson and Barry, 1987). A question is whether information would improve portfolio selection improve household welfare or wether wealth and stage in the lifestyle inhibit the choice of portfolios (Valdivia et al., 1996; Dunn et al., 1996; Reardon et al., 1992 ). Other researchers argue that reduction of risk can increase income by increasing specialization (Bromley and Chavas, 1989 ). Will forecasts have this effect? Studying changes in farmer portfolios is thus an important window for understanding how weather information influences production decisions.

Local Knowledge Systems
The ability of farmers to use climatic forecasts to increase food security depends on the nature of their local knowledge systems. Forecasts must be in a form that is meaningful and useful to them and farmers must have alternatives available to them that allow them to adjust to climatic fluctuations. Local knowledge systems are different than scientific knowledge systems. Scientific knowledge systems are organized to produce general principles that are applicable across time and space. Local knowledge systems are organized to explain single cases at a point in time (Kloppenberg, 1991). Local knowledge systems incorporate traditional or indigenous knowledge as well as scientific knowledge. For example, Andean farmers may use traditional forecasting methods like assessing the condition of certain cacti on a given saints day (Hatch 1981) while they also use weather information they hear on the radio (Woodman, 1998).

Social Capital
Historically, the scientific on debate social adjustments to change has focused on either cultural or structural constraints. The former see differences in adaptation due to values and belief systems that are transmitted across generations. Often traditionalism or a "culture of poverty" are seen as barriers to change (Lewis, 1968). The latter approach emphasizes the structure of the economy and the economic resources that a family possesses. However, recently there has been a focus on social relationships and more specifically social capital (Wall et al, 1998:316). Social capital describes those networks and social relationships that give persons advantages in competitive situations (Coleman, 1988). These may include intimate relations such as family ones as well as those with neighbors or casual acquaintances. The latter type of relationship has received considerable attention since the publication of Granovetter's paper on the "strength of weak ties" in 1973. This paper demonstrated the importance of casual acquaintances as sources of information and as links to resources. Trust and norms of interaction may reduce risks and transaction costs significantly. Social capital is not fungible however, a given form of social capital may be valuable in some situations and harmful or useless in others (Coleman, 1988,s98). The above research described studies that indicated the importance of social capital in developed economies, but their impact may even be greater in the Altiplano where markets and climate are unreliable and where traditional institutions and informal relations are more prominent than in North America. Participation in community institutions may be an important source of social capital in the region. Narayan and Pritchett (1996) have developed indicators of social capital in Africa and social capital has been found to be an important predictor of household survival in rural Russia, (O'Brien et al, 1997).

Transactions Costs
Markets that farmers in the Andes face tend to fail (Ellis, 1993; Sadoulet and de Janvry,1995) or are non existent, as mentioned above. As a result many decide to produce for their own consumption rather than rely on the market. Failure of the market takes place when the farmer prefers to produce for subsistence than make use of the market (Sadoulet and de Janvry, 1995), because the difference between the subjective value of their own production and market price is very high. Production for consumption indicates that it is cheaper for a farmer to produce than to purchase in the market. Transactions costs include factors that increase transportation costs, imperfect markets, asymmetric information. Use of climate and weather forecasts may be limited by these costs. Differences in access to the market and effective prices do to transactions costs, conditions access to climate and weather forecasts, and therefore its use. Farmers with lower transactions costs will are able to use more weather and climate information in their decisions. Sadoulet and de Janvry propose that both transactions costs and assets, explain different income strategies and predict social differentiation (1995). We propose that the different strategies, are conditioned by distinct access to information and ability to withstand a risky event.

Findings on Households and Decision Making in the Altiplano of Peru and Bolivia
Small holder farmers and their families have always confronted climatic perturbations in the Andes. Uncertainty and risk are in their production and consumption decisions. For centuries people in this region have used diversification as an ex ante risk management strategy (Martínez-Castilla 1992; Kervyn, 1988). In a normal rainfall cropping year our research found that households pursue a diversification strategy that is conditioned by stage in the life cycle, and that all families grew potatoes and quinoa for consumption. Older families depended more on transfers and younger groups were more diversified (Valdivia and Jetté, 1997). The latter diversified to cattle production if wealthier, retaining both adults in the farm households (Dunn et al., 1994). Those with fewer resources had sold their labor off the farm (Dunn et al.,1994).

During the low rainfall period of 1994-1995 a second survey was applied to evaluate if there was a change in strategy, and what permitted this change. Frost also affected production, therefore the strategies reflected actions to confront both events. There was an overall increase in forage production, dairy, and sheep, all believed to withstand better the abiotic effects (Valdivia, forthcoming). Networks facilitated access and control of resources, more effective than property in explaining farmer strategies (Espejo and Jetté, 1995). Off-farm employment was an important diversification strategy for those without dairy. Both are activities that are not covariant with crops.

Farmers have incorporated many activities into their portfolios, and current strategies reflect the technological innovations that have been available to them in the last 30 years (Markowitz and Valdivia, 1995). Understanding how farmers access weather information, and how it affects their choice of crops and use of inputs. Measuring the costs of changing decisions, as reflected in their economic portfolio, through the change in the net income of each identified strategy will allow us to determine the impact of information (Phillips, 1996).