AN APPROACH TO THE ECONOMETRIC ESTIMATION OF ATTITUDES TO RISK IN AGRICULTURE: COMMENT

Bardsley and Harris (1987) test empirically the effects of changes In deterministic wealth and random income on the measure partial risk aversion. The paper, which is otherwise well written, failed to impose the relationship between the two effects and estimated them independently. Consequently, inconsistent estimates were derived for the elasticities of the measure of partial risk aversion with respect to wealth and to income. The purpose of this comment is to derive that relationship theoretically and to point out the resulting inconsistencies in the original paper.     

Stochastic efficiency analysis with risk aversion bounds: a correction

A recent paper by Hardaker et al. (The Australian Journal of Agricultural and Resource Economics, 48, 2004a, 253) and book by Hardaker et al. (Coping with Risk in Agriculture, 2004b) describe a procedure for determining an efficient set from among a set of random alternatives. This procedure, called stochastic efficiency with respect to a function (SERF), is claimed to make the same assumption concerning the risk aversion measures as does stochastic dominance with respect to a function (SDRF). This is claim is incorrect. SERF imposes an additional requirement on the risk aversion measures of the decision makers. Both procedures assume a lower and an upper bound on risk aversion, but SERF also assumes that all risk aversion measures are of the same functional form as these lower and upper bound functions. This additional strong requirement on risk preferences implies that the efficient set identified under SERF is usually smaller than that identified using SDRF.     

Potential demand for hedging by Australian wheat producers

The potential for hedging Australian wheat with the new Sydney Futures Exchange wheat contract is examined using a theoretical hedging model parametised from previous studies. The optimal hedging ratio for an ‘average’ wheat farmer was found to be zero under reasonable assumptions about transaction costs and based on previously published measures of risk aversion. The estimated optimal hedging ratios were found by simulation to be quite sensitive to assumptions about the degree of risk aversion. If farmers are significantly more risk averse than is currently believed, then there is likely to be an active interest in the new futures market.     

A study of economic efficiency of Iranian farmers in Ramjerd district: an application of stochastic programming

An extension of utility-efficient programming to the non-linear discrete stochastic programming method was developed and used in the analysis of the economic efficiency of a sample of farmers in Iran. The results indicate that it would be feasible to increase substantially farmers' total net revenue by increasing their economic efficiency in terms of technical and allocative efficiencies. The study further suggested that risk aversion plays an important role in farmers' behaviour. The sample farmers are risk averse and hence are likely to trade higher expected profits for lower risk. Understanding this characteristic is important for interventions intended to raise farm productivity and efficiency.     

Estimation and use of risk aversion coefficient for traditional dryland agriculture in western Sudan

This paper explores the importance of including risk when modeling subsistence-oriented agriculture in a developing nation. The specific objective is to measure the degree of risk aversion for typical farmers in the smallholder traditional agriculture of the Sudan. The procedure followed is to impute the farmer's risk aversion coefficient through a mathematical risk programming technique. Imputed farmers' risk aversion coefficients were used to validate the model specified and identify, for further analysis, a single risk optimal farm plan for each of the different farm situations studied.     

Risks and returns from soil conservation: evidence from low-income farms in the Philippines

This paper examines risks and returns associated with soil conservation on hillside farms in the Philippines. Stochastic efficiency analysis is combined with a heteroskedastic regression model to assess the impacts of contour hedgerows on low-income corn farms. Regression analysis indicates that, over time, contour hedgerows can improve yields up to 15% compared with conventional practices. The analysis also provides weak support tor a hypothesis that hedgerows are variance reducing. However, results show that the reduction in yield variability afforded by hedgerows is modest, and that yield variability may increase by as much as 5% as hedgerow intensity rises. Tests for stochastic dominance show that, compared with the conventional tillage system, hedgerows do not constitute an unambiguously dominant production strategy. Stochastic elticiency with respect to a function is used to identify a range lor the coefficient ol relative risk aversion within which hedgerows dominate conventional tillage. Results suggest this range would be rather high; hedgerows dominate the conventional cropping strategy only lor decision-makers with relative risk aversion coefficients in the range 3-5.5. Implications for soil conservation adoption in low-income settings are discussed.     

Non-parametric estimation of decision makers' risk aversion

A new non‐parametric method to estimate a decision maker's coefficient of absolute risk aversion from observed economic behaviour is explained. The method uses the expected value‐variance (E‐V) framework and quadratic programming. An empirical illustration is given using Norwegian farm‐level data.     

Comparing risk attitudes of organic and non-organic farmers with a Bayesian random coefficient model

Organic farming is usually considered to be more risky thanconventional farming, but the risk aversion of organic farmerscompared with that of conventional farmers has not been studied.Using a non-structural approach to risk estimation, a Bayesianrandom coefficient model is used to obtain individual Arrow–Prattcoefficients of absolute risk aversion for a sample of Dutchorganic and non-organic arable farmers. The model is estimatedusing Gibbs sampling. The results indicate that organic farmersare significantly less risk averse than their non-organic colleagues.     

AN ANALYSIS OF WILLINGNESS-TO-PAY FOR CROP INSURANCE*

In this paper a method for estimating a producer's willingness-to-pay for crop insurance is presented. The method includes formulae to capture the impact of crop insurance on the producer's expected income and variance of income. These impacts are evaluated in the context of a model of producer welfare which features both price and yield uncertainty, as well as risk aversion on the part of the producer. The method is applied to the Australian wheat industry and estimates of willingness-to-pay are shown to be relatively sensitive to the levels of coverage and yield variability.     

Farmer responses to changing risk aversion,enterprise variability and resource endowments

The focus of this article is on assessing how risk aversion, enterprise variability and resource endowments affect farm land‐use decisions and economic returns. A theoretical model of a two‐enterprise, two‐constraint farm is developed, and then, an empirical illustration for an Australian farm is provided. The methodology used builds on previous expected mean‐variance (EV) models by incorporating land and budget constraints. The Kuhn–Tucker conditions of the EV model are examined to highlight that changes in resource endowments have larger effects on economic returns, than do changes in risk aversion or enterprise gross margin variability. It was also found that combinations of enterprise mixes that do not use all available resources can produce higher economic returns, relative to some enterprise mixes that use all available resources.     

Risk-Taking Preferences of Farmers in Northern Thailand: Measurements and Implications

Risk-taking preferences were elicited from small semi-commercial farmers in Northern Thailand using an experimental procedure that included real monetary payoffs of meaningful magnitudes. A total of five sets of lotteries with increasing payoffs were offered. The farmers were found to be risk averters, and their preferences conformed to the hypothesis of increasing (nondecreasing) partial relative risk aversion. Using regression analysis, farmers' expected variation of rice yields and farm size were found to be directly related to a decrease in risk aversion, while the extent of multiple cropping, availability of non-land household assets, and tested mathematical ability were found to be indirectly related to a decrease in risk aversion. The variables expected variation of rice prices, farmers' age, and tested abstract ability scores were not related to risk-taking preferences.     

Impacts of risk aversion on whole-farm management in Syria

This article reports on a study of the impact of risk on farm management practices in northern Syria, focusing particularly on how these are affected by risk aversion and farm size. The study is based on production data from an eight-year field trial and on prices from market surveys. A large linear programming model is built, representing the eight years as observations from a discrete probability distribution. Risk aversion is modelled by inclusion of a utility function with constant relative risk aversion, represented using the DEMP/UEP approach.     

Stochastic efficiency analysis with risk aversion bounds: a simplified approach

A method of stochastic dominance analysis with respect to a function (SDRF) is described and illustrated. The method, called stochastic efficiency with respect to a function (SERF), orders a set of risky alternatives in terms of certainty equivalents for a specified range of attitudes to risk. It can be applied for conforming utility functions with risk attitudes defined by corresponding ranges of absolute, relative or partial risk aversion coefficients. Unlike conventional SDRF, SERF involves comparing each alternative with all the other alternatives simultaneously, not pairwise, and hence can produce a smaller efficient set than that found by simple pairwise SDRF over the same range of risk attitudes. Moreover, the method can be implemented in a simple spreadsheet with no special software needed.     

Incorporating risk in a positive mathematical programming framework: a dual approach

In this study we develop a new methodological proposal to incorporate risk into a farm‐level positive mathematical programming (PMP) model. We estimate simultaneously the farm nonlinear cost function and a farmer‐specific coefficient of absolute risk aversion as well as the resource shadow prices. The model is applied to a sample of representative arable crop farms from the Emilia‐Romagna region in Italy. The estimation results confirm the calibration ability of the model and reveal the values of the individual risk aversion coefficients. We use the model to simulate different scenarios of crop price volatility, in order to explore the potential risk management role of an agri‐environmental scheme.     

Poverty, Risk Aversion, and Path Dependence in Low-Income Countries: Experimental Evidence from Ethiopia

In most low-income countries, rural households depend on mixed rain-fed agriculture/livestock production, which is very risky. Due to numerous market failures, there are few ways to shift risks to third parties. The literature has focused on what determines the responses of households in such environments. Of special concern are path dependencies in which households experiencing failure are prone to further failure and potential poverty traps. This paper estimates levels and determinants of risk aversion in the highlands of Ethiopia. We find high risk aversion and evidence that constraints have important impacts on risk-averting behavior with perhaps significant implications for long-term poverty. The results also suggest the possibility of path dependence and offer insights into links between risk aversion and poverty traps.     

Risk Averters that Love Risk? Marginal Risk Aversion in Comparison to a Reference Gamble

We propose an analytical distinction between standard risk aversion based on the valuation of a single gamble and marginal risk aversion based on the change in valuation between two gambles. We measure marginal risk aversion in two dimensions—mean and variance. Data from a field experiment is used to study marginal risk aversion. Our results suggest that individuals rely on a reference gamble when assessing marginal risk. Individual responses to marginal changes in mean and variance are nearly identical in direction and magnitude—suggesting that information on both standard and marginal risk aversion is needed to accurately model behavior.     

THE EFFECT OF FERTILISER ON RISK: A HETEROSCEDASTIC PRODUCTION FUNCTION WITH MEASURABLE STOCHASTIC INPUTS

The sources of production risk are many and diverse in nature. Estimating risk as a black box, without explicit recognition of its sources, can lead to inferior estimates of optimal inputs under risk aversion. In this paper, a method is presented for estimating production functions with measurable stochastic inputs and for generating the parameters of the probability distributions of yield for various environments and input levels. Based on this method, it appears that moderate risk aversion can account for a 6.7 per cent to 16.7 per cent reduction in nitrogen use (relative to the risk-neutral solution) for selected rice producing areas of the Philippines. Estimating optimal inputs without environment specific information about the sources of risk leads to large errors. This underscores the value of collecting information about the sources of risk and of exercising caution when such information is not available.     

Should farmers invest in financial assets as a risk management strategy? Some evidence from New Zealand*

This study explores the potential for risk reduction by New Zealand farmers through the diversification of their farm asset portfolios to include financial investments such as ordinary industrial shares, government bonds and bank bills. Low correlations between rates of return on farm and these financial assets suggest that significant reduction of income variability might follow their inclusion in farmers’ portfolios. Stochastic efficiency analysis is used to analyse alternative portfolios of ordinary shares, government bonds and bank bills and New Zealand farmland, using coefficients of absolute risk aversion derived from a negative exponential utility function. The results suggest that those farmers showing high degrees of risk aversion would gain utility by including financial assets in their portfolios. Deregulation of the New Zealand economy in the 1980s appeared to reduce the potential gains from diversification. Bonds rather than ordinary shares are the main contributors to portfolios which maximise utility for individuals classified as ‘somewhat’ risk averse.     

THE INFLUENCE OF RISK AVERSION ON THE UPTAKE OF SET-ASIDE: A MOTAD AND CRP APPROACH

This paper applies MOTAD and CRP models to the farm set-aside decision and presents a framework for analysis which examines the influence of risk on the land-idling decision. The results suggest that set-aside payment at a rate of £180-£200 per hectare would be attractive only at relatively high levels of risk aversion on an East Anglian specialist cereal farm and with moderate risk aversion on a Northumberland mixed farm. However, at identical levels of risk aversion, the area set aside was greater on the East Anglian farm. A survey to ascertain farmers' risk: income weightings is suggested as a useful adjunct to this work.     

Estimating Intertemporal Preferences for Natural Resource Allocation

In this article, we show how the degree of risk aversion, discounting, and preference for intertemporal substitution for a natural resource manager can be structurally estimated within a recursive utility framework. We focus on the management of a reservoir in California, and test the data for consistency with a recursive utility model specification versus standard time-additive separability. The results show that the data are consistent with a risk-averse manager with recursive preferences. The data also reject time-additive separability, with or without risk aversion, such as the standard constant relative risk aversion utility model. The improvement in model fit when recursive preferences are used is notable.