Empirical approach to risk management strategies of Mediterranean mussel farmers in Greece

Questionnaires were distributed to all Greek mussel farmers between November 2008 and February 2009 and completed under the guidance of advisers during personal interviews and site visits. The survey targeted the whole industry, with an estimated total production capacity of 45 403 tons shared by 523 operating farms. Prior to its large scale use, the questionnaire was extensively tested for the risk terminology used. This aimed at structuring the questions and eliminating linguistic uncertainty. The questions about “sources of risks” and “risk management strategies” were based on the opinion of four mussel farming experts, and pre-tested on five farmers with a strong educational background and experience profiles before being presented to the respondents.

The research design and the structure and drafting of the questionnaire were based on the empirical study by Meuwissen et al. (2001) carried out for livestock producers in the Netherlands. Seven questions refer to the “socioeconomic features of farmers” (farm area, production, full-time/part-time occupation and respondents’ age, education and work experience). To understand “mussel farmers’ perception of risk attitudes”, five questions were used (willingness to take risk in production, in marketing, in farming in general; financial issues – more than my colleagues and whether “I am willing to take more risks than other farmers”), the last of which “willingness to take risks in farming in general” was included as a consistency check elsewhere in the questionnaire. Since all statements measure attitudes toward risks relative to other farmers, the term “relative risk attitude” was used according to Patrick and Musser (1997) and Meuwissen et al. (2001).

Thirty-three questions refer to the “sources of risks” and cover “in-farm production risk” (five questions: IDs 1–5), “technological risks” (three questions: IDs 6–8), “ex-farm economic risks” (three questions: IDs 9–11), “ex-farm production risks” (six questions: IDs 12–17), “customer perception risks” (five questions: IDs 18–22), “ex-farm risks related to government support” (four questions: IDs 23–26), “financial risks” (three questions: IDs 27–29) and “risks related to family situation” (four questions: IDs 30–33).

Fifteen questions refer to “risk management strategies” and cover “in-farm investments” (five questions: IDs 1–6), “ex-farm investments” (two questions: IDs 7–10), “insurance” (three questions: IDs 11–13) and “selling price” (two questions: IDs 14–15).

Finally, three questions (What risks do you consider manageable? For what types of risks would you like to purchase insurance? What type of risk could be covered by public/government support?) constitute the open-ended questions of the questionnaire. The answers were then categorized and presented as percentages of the responses.

The questions under respondents’ perception of their risk attitude, sources of risks and risk management strategies were to be answered on a Likert type scale of 1 to 5: 1 (disagree) to 5 (agree) for respondents’ perception of their risk attitude; 1 (no impact) to 5 (very high impact) for sources of risks; and 1 (not relevant) to 5 (very relevant) for risk management strategies.

Data analysis starts with descriptive statistics of questionnaire responses, followed by principal component analysis to reduce the large number of examined variables to a smaller representative group, thus simplifying the evaluation process.

The adequacy of the available sample was checked using the Kaiser–Meyer–Olkin (KMO) test and Barlett's test of sphericity. Orthogonal varimax rotation was used to ensure (extract) the maximum independence between factors. Factors’ scores were used for subsequent multivariate regression analysis to determine their impact on risk perception of mussel farmers and their risk management strategies.

To assess the relative importance of each question under the categories respondents’ perception of their risk attitude, sources of risk and risk management strategies, responses were ranked by their mean (Mi) in descending order: Mi ≥ 4 = important; 3 ≤ Mi < 4 = high moderate; 2 ≤ Mi < 3 = low moderate; Mi < 2 = low.

In order to reduce the number of variables and possible collinearity among the variables, factor analysis (FA) was applied on normalized and standardized data for variables of each category: (a) socioeconomic features of the farm and respondent, respondents’ perception of their risk attitude (SER), (b) sources of risks (SR) and (c) risk management strategies (RMS).

Factor analysis (FA) is a multivariate analysis method that aims at explaining the correlation between a large set of variables in terms of a small number of underlying independent factors. Principal factor analysis was used to extract factor loadings.

Each principal factor (PFi) represents a distinct cluster of intercorrelated variables. Thus, two variables of high magnitude in the same factor are highly correlated. The varimax rotation method (orthogonal rotation) was used to make each factor uniquely defined as a distinct cluster of intercorrelated variables. Factor loadings indicate the weight of each variable on its corresponding axis, while per factor scores (SPFi) are the linear result of initial variables with respect to this factor. The sign of the loading indicates the direction of the relationship between the factor and the variable as well as between intercorrelated variables in each PFi. The variation incorporated in each factor is expressed by its eigenvalue. Factors with eigenvalues > 1 (cut-off point for extracted factors) account for more variance than original variables of standardized data (Hair et al. 1998). The resulting data sets of the factor analysis FA (per factor scores: SPFi) on SER (socioeconomic & risk attitude), SR (risk sources) and RMS (risk management strategy) variables were SER factors, SR factors and RMS factors, respectively.

Multivariate analysis of variance (MANOVA) and post hoc Tukey HSD test were applied to identify significant differences in SER, SR and RMS scores in relation to the legal status of enterprises and their culture systems.

Path analysis is an extension of multiple regression analysis, which is regarded as a special case of structural equation modeling. The path model aims to overcome potential shortcomings in conventional models, e.g. linear regressions, by transforming causal effects into direct and indirect effects.

In this study, path analysis was applied to estimate the magnitude and significance of hypothesized direct effects of SER and SR on RMS and indirect effects of SER through SR on RMS (Fig. 1). Path coefficients were standardized regression coefficients in a system of linear regression equations (Wrigth 1934): SRK=Bk.iSERi+ek′R2kandRMSj=Bj.iSERi+Bj.kSRk+ej′R2j \matrix{ {S{R_K} = {B_{k.i}}SE{R_i} + {e_{k'}}{R^2}_k} \cr {{\rm{and}}} \cr {RM{S_j} = {B_{j.i}}SE{R_i} + {B_{j.k}}S{R_k} + {e_{j'}}{R^2}_j} \cr } where i, k and j are the SER factor, the SR factor and the RMS factor, respectively, B's are standardized regression coefficients, e is the random error term and R² is the determination coefficient estimated by the least squares regression techniques (multi-regression analysis – MGR). R² expresses the proportion of variance in a given dependent variable explained by an independent variable to the total variance of all dependent variables. The B coefficient is defined as the ratio of the standard deviation of the effect due to a given cause to the total standard deviation of the effect. For RMS, the direct effects are B_j.i and B_j.k (1 − B_k.i) and the indirect effect is (B_k.i) (B_j.k), while for SR only direct effects (B_k.i) are recorded. The total effect of each SER on each RMS through various SRs is the sum of direct and indirect effects of the involved variables (SER and SRs).

Figure 1

Significant factors used in the final model were selected through the backward stepwise variable selection method (F-to-remove; p ≤ 0.05; Zar 1999). In addition, the variance inflation factor (VIF) was used to estimate the level of collinearity among the explanatory variables in the multi-regression analysis. High or low VIF values represent high or low levels of collinearity, respectively, with the usual cut-off value of VIF = 10. A VIF value close to 1 indicates that the variables are uncorrelated (r ≈ 0; Hair et al. 1998). Analyses were performed using SPSS 17.0 (SPSS Inc., Chicago IL, USA).

The largest proportion of respondents (40%) were middle-aged farmers (41–50 yrs), followed by 21% of older farmers (51–60 yrs), 19% aged 31–40 years, 11% approaching retirement age (> 61 yrs), and 9% newcomers (18–30 yrs). The majority of respondents (61%) were secondary school graduates, while 27% were university graduates, and the remaining 12% completed primary education. The number of part-time employees (2.73 ± 1.81) was higher than the number of full-time employees (1.25 ± 1.60) per mussel farm (Table 1).

The predominant farm type operated by the respondents (Table 1) was the long line system (92%), followed by hanging parks (8%). Most of the farms (44%) were operated under the legal status of sole proprietorship (self-employment), followed by 35% of general partnership (GP) and limited partnership (LP) in collaboration with other companies. Large entities such as limited partnership companies (5%) and Société Anonyme/Anonymous Company – SA (15%) had a small share.

Table 2 shows that FA identified three factors with eigenvalues higher than 1 (> 1), explaining 81.12% of the total variance. Using a cut-off value of 0.60 for factor loadings, factor 1 explains 43.86% of the variance by socioeconomic variables related to farm features (farm size, production capacity and full- or part-time labor force) and factor 2 explains 22.13% of the variance related to farm-manager education (education and age), with an opposite (reverse) relationship between the age of a mussel farmer and his educational background. Young farmers had more opportunities to access the educational system compared to older farmers as the country gradually modernized. Factor 3, explaining 15.12% of the variance, expresses the work experience of a farm manager and is related to his age and education.

All factors showed non-significant statistical differences between the legal status of enterprises and their culture systems (MANOVA; p < 0.1).

“Relative risk attitude” of mussel farmers was measured to investigate their willingness to take risk. The FA factor loading explained 72.35% of the total initial risk variation (Table 3). Farmers were comfortable with risk-taking in production (fieldwork, 63.27 ± 26.57%) and marketing (daily contacts with wholesalers, 62.45 ± 27.88%), i.e. areas they are more familiar with than financial issues (use of financial products, 48.57 ± 27.99%). Greek mussel farmers tended not to take financial risks as they did consider themselves unfamiliar with the processes involved, for example, applying for a bank loan to finance modernization, or flexibility in dealing with wholesalers (they prefer to sell their products through local and well-known wholesale agents rather than other new and most promising traders). Their attitude toward risk is above average (58.86 ± 24.26%), which coincides with their willingness to take more risk than others in the same business (59.59 ± 4.66%). Relative risk attitude showed significant statistical differences for the legal status of enterprises (MANOVA; p < 0.1), with the gradation of mean relative risk attitude being SA < GP and LP < self-employment (post hoc Tukey HSD test; p < 0.05). This may be a consequence of the fact that most mussel farmers are family businesses, which supports the observation that risk must be borne by individuals. Relative risk attitude showed statistically insignificant differences between farming systems (MANOVA; p > 0.1; Table 3).

Responses of mussel farmers to Likert-type questions are presented as a percentage distribution (%) of scores for different types of risk sources (Table 4). Table 4 shows the major identified sources of risk that affect mussel farming in descending order of importance. The five most important risk sources were ex-farm prices (mean value: 4.49 ± 0.82), operator's disability/health (4.20 ± 1.17), vessel availability (4.18 ± 1.47), HABs (4.12 ± 1.11), and farmer's family health (4.02 ± 1.13). Finally, scores from 27 to 33 represent risks with average values of less than 2.47 ± 1.37, estimated to have a low to moderate impact on mussel farming. These refer to pollution, environmental impact, sea rental (farm site leasing cost), illegal actions (i.e. cheating on mussel stock and equipment), environmental NGOs (protests against aquaculture activities, e.g. for conservation purposes), transport (delays in moving live/fresh mussel stock into the supply chain, which can cause losses, especially on long-distance export markets in summer), and diseases.

Using factor analysis (FA), the original 33 sources of risk (SR) were reduced to 10 (SR) explaining 79.95% of the total initial variance (Table 4). SR1 explaining 10.65% of the variance described as “health safety OR technology availability” is associated with mussel farmer's perception of the effect of health safety policies (positive association) and availability of grading machines and technology (negative association).

SR2, explaining 10.57% of the variance, is described as “personal welfare” and is associated with mussel farmer's perception of the effect of health status of the farmer and his family (risks: disability and loss of health of a farm operator and the health situation of farmer's family).

SR3 explains 9.79% of the variance and is described as “financial risk” associated with mussel farmer's perception of the effect on the farm of the farmer's ability to manage interest rate changes and loan repayments.

SR4 explains 8.81% of the variance and is described as “environmental risk” associated with mussel farmer's perception of the effect of freshwater availability mainly through rainfall (resulting in high salinity and nutrient losses that indicate limitations on algal food availability in estuarine ecosystems where most of the mussel farms are located), diseases and general environmental impact.

SR5 explains 7.55% of the variance and is described as “market risk” associated with mussel farmer's perception of the effect of ex-farm prices, and negatively associated with seed recruitment.

SR6 explains 7.47% of the variance and is described as “social acceptance” associated with mussel farmer's perception of the effect of public opinion, usually influenced or induced by NGOs and the media highlighting malpractice by farms.

SR7 explains 6.91% of the variance and is described as “institutional” associated with mussel farmer's perception of the effect on the farm of policies related to transportation (new legal standards for the supply chain), absorption of supplies, environmental regulations and availability of new licenses.

SR8 explains 6.27% of the variance and is described as “production cost & labor availability” associated with mussel farmer's perception of the effect of production cost & labor availability.

SR9 explains 6.21% of the variance and is described as “HABs” associated with mussel farmer's perception of the effect of harmful algal blooms and government actions (bans etc.) on the farm.

SR10 explains 5.73% of the variance and is described as “biofouling & predators” associated with mussel farmer's perception of the effect of biofouling and predator attacks.

Health safety OR technology availability, financial risk, environmental risk, market risk, production cost & labor availability, and HABs showed statistically insignificant differences between the two mussel culture systems (MANOVA; p > 0.1). Mean score values of personal welfare, social acceptance, and institutional showed statistically significant differences for the legal status of enterprises (anonymous companies < other types of legal status; self-employment < general partnership GP and limited partnership LP; self-employment < anonymous companies, respectively; MANOVA; post hoc Tukey HSD test; p < 0.05). This may be due to the fact that most farmers are self-employed.

The mean values of social acceptance and biofouling & predators scores showed statistically significant differences between the two culture systems (long line < hanging parks; hanging parks < long line, respectively; MANOVA; post hoc Tukey HSD test; p < 0.1). Hanging parks are located within Natura 2000 protected zones, but they also have a strong social influence on local communities (mussel farming is a labor-intensive activity that provides employment, so farmers have significant political influence on local and national elections). Biofouling can be more easily managed in hanging parks compared to long lines (e.g. exposure of mussels to air to kill ascidians, etc.).

Table 5 shows the mean scores for the risk management strategies of Greek mussel farmers. The preferred strategy was the provision of financial and credit reserves (4.84 ± 0.43). Strategies with “low impact” (scores below 3) are ranked between 13 and 15. They include sale price contracts, personal insurance policies, and species diversification. All three are absent or of minimal relevance. Strategies that scored between high and low values (ranked from 2 to 12) have “moderate impact” with mean values considerably below the first option (3.65 ± 1.65 to 3.06 ± 1.77). Except for “financial and credit reserves”, all other variables from 2 to 11 are within 85% of importance in the ranking of farmers. These are not ways to intensify their core business, but to seek alternative ways to make a living. It can be concluded that mussel farming business is unstable. Moderate impact strategies include employment of farmers in other businesses (e.g. agribusiness, commerce), cost reduction and collaboration between farmers either horizontally (by sharing equipment, supplies, labor, etc.) or vertically in trade and commerce. Other moderate impact strategies include business diversification (in processing, fishing, distribution), participation in public support programs, off-farm investments (e.g. agritourism, stock market), boat insurance policy, adherence to strict hygiene rules, business insurance policy, and spatial diversification (geographic dispersion of the business).

Similar to the sources of risk, the 15 identified risk management strategy (RMS) variables were reduced by FA to five factors (RMS), explaining 66.96% of the total initial variance (Table 5).

Mussel farmers seek “off-farm employment OR apply strict hygienic rules” (RMS1, explaining 15.46% of the total variance) through off-farm investment and employment, and alternatively on-farm adherence to strict hygiene and environmental rules.

In addition, farmers use “intra-company measures” through negotiation of price contracts for sales, spatial diversification into other species and participation in government supporting programs (RMS2 explaining 14.12% of the variance).

RMS3, explaining 13.58% of the variance and described as “insurance”, refers to farmers’ willingness to purchase personal, business and boat insurance.

RMS4, explaining 12.98% of the variance and described as “integration”, refers to horizontal and vertical integration between farmers (production and trade) and alternatively to production at the lowest possible cost.

Finally, RMS5, explaining 10.82% of the variance and described as “geographic dispersion OR business diversification”, refers to strategies of geographically dispersing the business or grouping it together (business diversification) with other activities such as fishing, processing and distribution networks.

Only the mean of Insurance scores showed statistically significant differences for the legal status (self-employment < general partnership GP and limited partnership LP) and culture systems (long line < hanging parks; MANOVA; post hoc Tukey HSD test; p < 0.1).

Responses to the open-ended questions (Table 6) show that all mussel farmers (100%) were familiar with risk management associated with daily farming activities during their routine operations. They expressed interest in purchasing insurance tailored for their boats (44.9%) and protection from weather impacts (14.3%). Farmers also want public funds to be used to compensate for losses caused by extended harvest bans due to harmful algal blooms (79.6%), predation attacks (51.0%), pollution (26.5%), diseases (8.2%) and illegal actions (8.2%).

The path analysis framework (Fig. 1) demonstrates the relative links (B) between the risk management strategies, the risk factors and the socioeconomic status of Greek mussel farmers.

Figure 2 shows the significant relationships through the paths. The explained variance of RMS variables by the SER and SR variables ranged from 21% (R² = 0.21) for “Geographic dispersion OR Business diversification” to 56% (R² = 0.56) for “Intra-company measures”. On the other hand, the explained variance of the SR variables by the SER variables ranged from 0% (R² = 0) for “Institutional” and “Biofouling & predators” to 26% (R² = 0.26) for “Health safety OR Technology availability”. The VIF ranged from 1 to 1.53 and indicated that the independent variables were uncorrelated.

Figure 2

The total, non-causal, direct and indirect effects and the association effect (r = correlation coefficient, estimated by MGR as zero order coefficient) of each independent variable with each dependent variable (Table 7) were estimated using the B values shown in Figure 2.

The total association of risk management strategy factors (RMS) “Off-farm employment OR Applying strict hygienic rules” is statistically significant with risk factors (SRs) “Health safety OR Technology availability” (0.272), “Environmental risk” (−0.511) and with “Production cost & Labor availability” (0.287). The direct and indirect effects of “Health safety OR Technology availability” were 0.272 and 0.138 through socioeconomic factors SER “Relative risk attitude” and the percentage of the total absolute effect (PTAE) 66.22% and 33.77%, respectively, of “Environmental risk” were −0.51 (PTAE =71.42%) and −0.20 (PTAE = 28.57%) through SER “Work experience” and of “Production cost & Labor availability” 0.287 (PTAE = 77.51%) and −0.08 (PTAE = 22.48%) through SER “Work experience”.

Total association of RMS “Intra-company measures” is statistically significant with SRs “Health safety OR Technology availability” (−0.414),“Environmental risk” (−0.338) and with “Social acceptance” (−0.240) and “Institutional” (−0.286), as well as with SERs “Relative risk attitude” (−0.355) and “Farm features” (−0.195). The direct and indirect effects of “Relative risk attitude” were −0.34 (PTAE = 100%) and 0, respectively, of “Health safety OR Technology availability” −0.26 (PTAE = 66.22%) and −0.13 (PTAE = 33.77%) through SER “Relative risk attitude”, of “Environmental risk” −0.33 (PTAE = 71.42%) and −0.13 (PTAE = 28.57%) through SER “Work experience”, of “Social acceptance” −0.28 (PTAE = 77.51%) and −0.08 (PTAE = 22.48%) through SER “Work experience”, while the direct effects of “Institutional” and “Farm features” were −0.39 (PTAE = 100%) and −0.21 (PTAE = 100%), respectively.

The total association of RMS “Insurance” is statistically significant with SRs “Health safety OR Technology availability” (−0.43), “Personal welfare” (0.32) and with “Social acceptance” (0.216) and “HABs” (−0.24). The direct and indirect effects of “Health safety OR Technology availability” were −0.43 (PTAE = 66.22%) and −0.21 (PTAE = 33.77%) through SER “Relative risk attitude”, of “Personal welfare” 0.320 (PTAE = 63.29%) and 0.186 (PTAE = 36.70%) through SER “Relative risk attitude”, of “HABs” −0.24 (PTAE = 78.74%) and 0.065 (PTAE = 21.25%) through SER “Work experience”, while the direct effects of “Social acceptance” were 0.216 (PTAE = 100%).

The total association of RMS “Collaboration OR Production at lowest cost” is statistically significant with SRs “Health safety OR Technology availability” (0.219), “Biofouling & predators” (−0.329) and with “Social acceptance” (−0.372) and “HABs” (0.240). The direct and indirect effects of “Health safety OR Technology availability” were 04218 (PTAE = 66.22%) and 0.111 (PTAE = 33.77%) through SER “Relative risk attitude”, of “Social acceptance” −0.37 (PTAE = 77.51%) and −0.10 (PTAE = 22.48%) through SER “Work experience”, of “HABs” −0.24 (PTAE = 78.74%) and 0.06 (PTAE = 21.25%) through SER “Work experience”, while the direct effects of “Biofouling & predators” were −0.32 (PTAE = 100%).

The total association of RMS “Geographic dispersion OR Business diversification” was statistically significant with SRs “Biofouling & predators” (−0.271), “Personal welfare” (0.120), as well as with SERs “Relative risk attitude” (−0.224). The direct and indirect effects of “Personal welfare” were 0.312 (PTAE = 67.56%) and 0.150 (PTAE = 32.43%) through SER “Relative risk attitude”, while the direct effects of “Biofouling & predators” and “Relative risk Attitude” were −0.30 (PTAE = 100%) and −0.40 (PTAE = 100%), respectively. Furthermore, Table 8 shows the results of the path analysis and the total absolute effects of the relative risk attitude factor, the socioeconomic factors (SER) and sources of risks factors (SR) on risk management strategy factors (RMS) for Greek mussel farmers. For “Off-farm employment” or “Applying strict hygienic rules,” the dominant absolute effect was given by the “Environmental risk” (0.510); for “Intra-company measures” – by “Relative risk attitude” (0.475) and “Institutional” (0.391); for “Insurance” – by “Relative risk attitude” (0.405) and “Health safety OR Technology availability” (0.431); for “Collaboration OR Production at lowest cost” – by “Social acceptance” (0.391) and “Biofouling & predators” (0.328); and for “Geographic dispersion OR Business diversification” – by “Relative risk attitude” (0.551).

Most respondents agreed that fluctuations in ex-farm prices are the main source of risk. This is critical for the viability of most farms, because while production costs have increased considerably (oil price, taxes, and wages), ex-farm prices of mussels have remained rather stable for over a decade (Theodorou et al. 2011; Theodorou et al. 2014). Ex-farm prices may constitute a major source of risk in exported seafood (e.g. Danish mussels: Nguyen 2012a,b; Thong 2012; Vietnamese catfish: Le & Cheong 2010; Norway salmon: Bergfjord 2009), in contrast to products that support well-established local demands (e.g. oysters in France: Le Bihan et al. 2013). This situation explains the sensitivity of the Greek industry to ex-farm prices, as the major part of the mussel production is for export (Theodorou et al. 2011). Mussel farming is a labor-intensive activity and requires a lot of physical work by a farmer, often under extreme weather conditions on a vessel. As most Greek mussel farms are micro-enterprises (Theodorou et al. 2014) and mussel farm owners are directly involved in the fieldwork, the health status of the operator and his family is a critical risk factor for business sustainability. Vessel availability (a suitable boat equipped with modern tools such as star wheels, French type grading machines, etc.) is the third major source of risk for mussel farming. Without these tools, it is very difficult for farmers to compete on the market and maintain production (Theodorou et al. 2011; Theodorou et al. 2015).

Bans on mussel harvesting due to harmful algal blooms (HABs) are also a major source of risk, as they prevent any marketing plans, especially during harvest periods (Theodorou et al. 2011, 2020). On the other hand, and unlike most sources of risk in fish aquaculture (Georgiadis et al. 2001; Murray & Peeler 2005; Peeler et al. 2007), diseases are less important as a source of risk since there has never been any serious case of losses (Karagiannis & Angelidis 2007; Karagiannis et al. 2013). Similar findings were reported by Ahsan and Roth (2010) for the emerging mussel sector in Denmark, which is a relatively new industry without any disease problem yet. The opposite situation is observed in France, where shellfish pathogens and summer mortality due to high temperatures have destroyed the sector's production several times (Huvet et al. 2010; Garcia et al. 2011; Pernet et al. 2012). Consequently, bivalve shellfish farmers assess sources of risk based on their own local experience.

Mussel transportation losses are also of low impact. Most Greek mussels are sold alive and it takes 1–3 days to deliver them to markets, including export markets (Angelidis 2007a; Theodorou et al. 2019). Nonetheless, these types of losses are limited and not considered catastrophic events.

Political pressure from non-governmental environmental groups on the sector is moderate, because operations are environmentally friendly and sustainable. However, when sites are located within Natura 2000 Protected Areas, new rules in legislation affecting the code of practice may introduce a future risk (Angelidis 2007b; Konstantinou et al. 2012; Latinopoulos et al. 2012; Karagiannis et al. 2013). Illegal actions against farms (i.e. poaching, destruction of installations) always carry a risk of loss, but so far this has been limited, representing isolated incidents and with manageable seasonal costs (not catastrophic). Sea rentals for this type of activity are affordable in the long term and not considered to be a significant risk.

The most preferred response of Greek mussel farmers to a risk management strategy (with limited variability and standard deviation of < 1) is to build up financial reserves to cope with unforeseen adversities in order to survive financially until the next season. Individual practices may include personal or family bank savings, a bank loan obtained through a long-term good business relationship, or keeping the farm in a good financial condition. Prioritizing liquidity and solvency is also considered very important by Danish mussel farmers (Ahsan & Roth 2010). Certainty of their income from alternative sources, such as off-farm employment, is preferred as a risk mitigation strategy. Mussel farming is a seasonal activity and extra cash from other sources can cover possible losses due to “uncertain” production. Complementary work in different occupations offers some security by reducing overall income variability (Dickey & Theodosiou 2006). These responses are not surprising, as pluri-activity and multiple job holding are structural features of farm households in Greece, and usually involve more household members than just the farmer (Kizos 2010; Kizos et al. 2011).

In Denmark, producing at the lowest possible cost is considered the primary risk management strategy (Ahsan & Roth 2010). Although Greek mussel farmers do not have the strong cooperative mentality of Danish producers (Ahsan & Roth 2010), they suggest that horizontal cooperation of farmers could mitigate losses, achieve benefits of scale, and reduce financial risk by lowering operating costs and increasing fixed capital depreciation (Cush & Varley 2013; Theodorou et al. 2014). Finally, diversification seems to be the least priority for Greek mussel farmers, as their traditional background does not allow them to easily adopt novel technologies or open up to new markets. Furthermore, the limited suitable space available in Greece for new bivalve species, such as benthic clams or oysters, explains the limited preference for this strategy.

Among the strategies with moderate or very low impact, it is worth noting that private insurance policies are not a priority. Corporate structures, unlike farmers personally, wish to buy insurance as a common business policy because they can afford it.

However, Greek producers want to have insurance for their vessels and/or against the weather impact on mussel farm installations and equipment, and animal stock. French oyster farmers hold similar views on risk-transfer mechanisms (Le Bihan et al. 2013). They also support the request for specific customized hedging products for the European bivalve shellfish sector, as the market has so far failed to cover this “gap”.

Greek mussel farmers are reluctant to enter into agreements with wholesalers offering them long-term contracts with stable prices. This attitude can be attributed to: a) the unsuitability of these “modern tools” for the needs of the sector; b) the complicated structure and questionable reliability of these contracts; c) farmers’ suspicion of modern business tools.

Greek mussel farmers suggested that bearable risks for them are those related to conventional self-protection mechanisms within the farm, as they feel familiar with their routine daily practices to prevent losses. This is consistent with their responses to the questionnaire, which show that they are accustomed to face risks in production and in general farming activities, making every effort to maximize income. It is a common attitude in agribusiness worldwide that farmers are confident in supervising their own production activities (Meuwissen et al. 2001; Le & Cheong 2010).

The need for public compensation for the losses of Greek mussel farmers arises from causes of disasters that are not directly related to their activity. These may include irreversible phenomena or major unpredictable disasters due to weather conditions (e.g. tsunami, heat waves, anoxia, or radioactivity), extended harvest bans due to harmful algal blooms, pollution (e.g. oil spills), predator attacks (e.g. sea turtles), diseases, and illegal actions (e.g. sewage and radioactive waste discharges). Hanging parks are more prone to disasters compared to long lines, because they are located in shallow waters near estuaries.

There is currently no insurance policy for the Greek mussel farming sector, rendering the business exposed to operational risks (Theodorou et al. 2020; Theodorou & Tzovenis 2021). At the European level, compensation is only available through the European Fisheries Fund in case of major disasters concerning the protection of livestock or human health. Through Article 57 of EU Regulation 508/2014, the EU further extends the scope of insurance compensation (covered by the European Maritime and Fisheries Fund) for aquaculture stock losses due to causes such as natural disasters, weather impacts, water quality, and diseases. At a global level, oil pollution losses could be covered by the International Oil Pollution Compensation (IOPC) fund (Le Bihan et al. 2013).

The socioeconomic profile of the farm/farmer has no influence on farmer’ decisions. Mussel farming is a socially supported, local aquaculture activity in Greece that provides a supplementary income to coastal society members (Zanou et al. 2005). In addition, the sector is hampered in certain regions by the limited availability of locations with suitable environmental conditions for mussel growth, giving niche producers a competitive advantage. The latter attributes much of the sector's development to local governance, as most mussel farmers are local people (Latinopoulos et al. 2012; Konstantinou et al. 2012). People with a common cultural heritage convey unified perception across generations, thus limiting the scope for variation in mentality. Such shared beliefs tend to integrate with prevailing socioeconomic principles through either formal rules or informal norms of behavior (North 1993). Therefore, the relative risk attitude of mussel farmers significantly affects most of their risk management strategy choices, and their explained variation R² is rather limited (between 0.21 and 0.58). Consequently, despite recognizing the market and financial risks as major sources of risk, they do not adopt a specific strategy to mitigate their effects.

The survey covering the period from November 2008 to February 2009 addressed mussel farmers’ needs for risk-sharing strategies prior to the onset of the Greek financial crisis. The study is still relevant, but the current approach of farmers has to be investigated in the new business environment to identify the industry's adaptation and its likely new needs (Theodoridis et al. 2017; Theodorou & Tzovenis 2017; Le Bihan et al. 2020; Avdelas et al. 2021; Lupo et al. 2021).