________________________________________________________________________________________________ 1 Bridging the Gap: The Impact of Compensatory Measures on 1 Mountain Farming in Piedmont 2 Moino Francesca1, Giacardi Alessandro2 , Califano Giovanbattista3 and Cagliero Roberto1 3 1 CREA PB, Turin, Italy 4 2 Università di Roma La Sapienza, Rome, Italy 5 3 Università degli Studi di Napoli Federico II, Naples, Italy 6 Corresponding Author: Francesca Moino (francesca.moino@crea.gov.it) 7 8 This article has been accepted for publication and undergone full peer review but has not been through 9 the copyediting, typesetting, pagination and proofreading process, which may lead to differences 10 between this version and the Version of Record. 11 Please cite this article as: 12 Moino F., Giacardi A., Califano G., Cagliero R. (2025). Bridging the Gap: The Impact of 13 Compensatory Measures on Mountain Farming in Piedmont, Bio-Based and Applied Economics, 14 Just Accepted. DOI:10.36253/bae-16852 15 16 Acknowledgments 17 This work was funded by the European Union Next-GenerationEU: (Call PRIN 2022- Italian Ministry of University 18 and Research (MUR) - RECIPE - cod. 20224ZHNXE). 19 Highlights 20 21 ● An income gap exists between mountain and non-mountain cattle farms in Piedmont 22 ● Rural development program support helps reduce but does not fully eliminate this economic 23 disparity 24 ● Addressing income inequalities would require more than doubling per-hectare compensatory 25 allowance payments 26 ● Redistribution of support should prioritize farm type and labor over land area 27 28 29 Abstract 30 31 This study examines the impact of the Rural Development Program (RDP) on reducing income 32 disparities between farms in mountainous and non-mountainous areas in Piedmont, Italy. Using 33 Farm Accounting Data Network data from 2012–2022, the analysis focuses on cattle, sheep and 34 goats, and fruit farms, with 525 farms (3,171 observations; 36% in mountainous areas). A pooled 35 ________________________________________________________________________________________________ 2 multivariate regression assesses income disparities excluding RDP support, RDP’s effectiveness in 36 mitigating gaps, and the role of compensatory allowance. 37 The findings indicate that significant income disparities are primarily observed in small farms 38 specialized in cattle and sheep and goats, with mountain farms facing a net shortfall of €1,319 and 39 €2,384 per hectare, respectively. While compensatory allowance support helps reduce this gap—by 40 8.93% for cattle farms and 5.28% for sheep and goat farms—a substantial disparity remains. 41 Bridging the gap entirely would require doubling compensatory payments to €340 per hectare, 42 though alternative strategies are discussed. 43 44 Keywords: FADN, compensatory allowance, mountain areas, income gap 45 46 JEL Classification codes: C18, C54, Q18, R58 47 48 1. INTRODUCTION 49 Mountain agriculture in Europe faces critical challenges due to the natural constraints of these 50 regions, often classified as Areas with Natural Constraints (ANCs) under Regulation (EU) No. 51 1305/2013, Art. 32.1 Previously termed Less Favoured Areas (LFAs) under the Common 52 Agricultural Policy (CAP), these regions face challenges such as steep slopes, adverse climates, and 53 poorer soil fertility, which collectively restrict arable land and require labour-intensive farming 54 practices. These limitations lead to higher production costs and reduce profitability, causing many 55 farmers in mountainous regions to abandon agricultural activities (Giannakis & Bruggeman, 2015; 56 Strijker, 2004). Such abandonment contributes to depopulation, economic decline, and the loss of 57 agro-biodiversity and traditional landscapes (Cesaro & Marongiu, 2013). To counteract these adverse 58 trends, the European Union (EU) has historically provided targeted support to ANC within the CAP 59 1 See Regulation (EU) No 1305/2013 of the European Parliament and of the Council of 17 December 2013 on support for rural development by the European Agricultural Fund for Rural Development (EAFRD) and repealing Council Regulation (EC) No 1698/2005. OJ L 347, 20.12.2013, p. 487. ________________________________________________________________________________________________ 3 framework, aiming to mitigate the biophysical disadvantages and sustain agricultural livelihoods in 60 ANCs. 61 The effectiveness of ANC and LFA support measures has been widely studied, with mixed 62 findings (Romagnoli et al., 2021). For instance, Borsotto et al. (2010) questioned the effectiveness of 63 these subsidies across different European contexts, while Ferto et al. (2022) and Klima et al. (2020) 64 observed heterogeneity in outcomes based on regional applications. Oxousi (2012) conducted a 65 comparative analysis on farm performance, finding that while ANC support plays a key role in the 66 profitability of mountain and other disadvantaged farms, mountainous farms still struggle compared 67 to those in less favoured, but more productive, areas. Similarly, Wieliczko et al. (2018) identified 68 significant disparities in production efficiency and economic performance between farms receiving 69 ANC support and those that do not, indicating that while ANC payments partially help mitigate 70 economic disadvantages, they do not fully compensate for them. 71 As the ANC policy evolved, the 2014–2020 Rural Development Program (RDP) introduced 72 more refined eligibility criteria to direct support toward regions with genuine biophysical constraints. 73 ANC support remains a crucial tool for maintaining agricultural activities, preserving rural 74 landscapes, and preventing land abandonment in vulnerable regions (Whitaker, 2024; Veveris et al., 75 2014). In mountainous regions, these payments significantly contribute to rural sustainability, by 76 retaining population, enhancing business viability, and preserving environmental and territorial 77 integrity (Cooper et al., 2006; Dax et al., 2021). 78 While the impact of ANC support has been widely studied, limited research has assessed its 79 effectiveness in reducing economic disparities between mountainous and non-mountainous farms in 80 specific regional contexts. This study aims to address this gap by analyzing the adequacy of 81 compensatory allowance (CA) in mitigating income inequalities in mountain agriculture, focusing on 82 the Piedmont region. Unlike previous research, which has often relied on cross-country or aggregated 83 assessments (Poláková, 2019), this study adopts a context-specific approach using Farm Accountancy 84 Data Network (FADN) data. By evaluating the financial resources required to ensure that 85 ________________________________________________________________________________________________ 4 compensatory measures effectively offset the economic disadvantages faced by mountain farmers, 86 this research provides a robust policy assessment and contributes to a more precise understanding of 87 rural development interventions (Romagnoli et al., 2021). 88 The choice of Piedmont, a region in northern Italy, as a case study is justified by its diverse 89 agricultural landscape, which includes both mountain and lowland areas, allowing for a meaningful 90 comparison between farms operating under different environmental constraints. Covering over 52% 91 mountainous territory, Piedmont includes both the Apennine and Alpine ranges, where depopulation, 92 agricultural abandonment, and the difficulties faced by small, isolated communities are pronounced 93 (Ferlaino, 2019). Livestock farming, central to Piedmont’s mountainous areas, relies on limited 94 resources and small, family-run farms, which differ substantially from lowland agriculture (ISTAT, 95 2020). EU programs have increasingly addressed these issues, although the region’s administrative 96 fragmentation, characterised by numerous small municipalities, complicates local development 97 management. This makes Piedmont a compelling case for understanding the complexities of 98 mountain development and the critical role of targeted policies in fostering sustainability. 99 To properly interpret the methodology and findings, it is essential to understand the structure 100 of the CA under the RDP, established under Regulation (EC) No. 1698/2005. During the 2007–2013 101 programming period, the CA was implemented through Measure 211, while in the 2014–2020 period, 102 it was integrated into Measure 13.1, both aimed at providing direct financial support to farmers in 103 mountainous areas to offset the economic disadvantages imposed by challenging terrain and climate 104 conditions. Measure 13.1 allocated approximately 60 million euros annually to mountain farmers in 105 Piedmont, succeeding Measure 211, which had provided 52.5 million euros in the previous 106 programming period (Cagliero et al., 2018; NUVAL Piemonte, 2016). These measures play a crucial 107 role in sustaining agriculture, supporting socio-economic stability, and preserving landscapes in 108 disadvantaged areas (Ferlaino, 2019; Regione Piemonte, 2016). However, the compensatory 109 allowance is not automatically granted to all farmers but is subject to eligibility criteria based on land 110 use and farm characteristics. The amount of support varies depending on the agricultural system and 111 ________________________________________________________________________________________________ 5 the level of land disadvantage. Additionally, to prioritize small and medium-sized farms, payments 112 decrease as farm size increases: farms with less than 20 hectares receive 100% of the allowance, those 113 between 20 and 40 hectares receive 70%, those between 40 and 70 hectares receive 40%, those 114 between 70 and 1000 hectares receive 10%, while farms over 100 hectares are not eligible for support. 115 This tiered approach reflects the strategic importance of maintaining agricultural activity in 116 mountainous areas and ensuring that support is directed toward those most affected by structural 117 disadvantages (Regione Piemonte, 2016). 118 The results of this study are specific to the Piedmont region, as they reflect the particular 119 agricultural, economic, and policy context of this territory. While the per-hectare compensatory 120 allowance is a common feature of ANC support across the EU, its implementation varies in structure, 121 budget allocation, and eligibility criteria, depending on national and regional regulations (e.g., 122 differences in budget distribution, farm eligibility rules, and administrative procedures). 123 Consequently, the findings may not be directly replicable in other regions, as variations in policy 124 design could influence the effectiveness of CA in addressing economic disparities. However, this 125 study offers a methodological framework that can be adapted for analyzing other mountain regions, 126 particularly within the Alpine arc, where similar agricultural and economic constraints exist. 127 Expanding the analysis to additional regions would facilitate a broader evaluation of ANC policy 128 effectiveness and identify potential areas for policy refinement. 129 The paper is organised as follows: Section 2 presents the data and methodology, including analytical 130 frameworks; Section 3 reports the results of the policy impact analysis; Section 4 discusses findings 131 with an emphasis on policy implications; and Section 5 concludes, suggesting directions for future 132 research and policy improvements. 133 ________________________________________________________________________________________________ 6 2. DATA AND RESEARCH METHODOLOGY 134 2.1. DATA COLLECTION 135 The analysis is based on data from the FADN, which collects standardised data annually, 136 offering microeconomic insights into income trends and the structural economic dynamics of farms 137 across Europe. For this study, we used data from the Italian survey, which includes approximately 138 11,000 farms each year2. 139 From this dataset, we selected farms located in the Piedmont region with a standard output3 140 value exceeding €8,000, in order to focus on commercial farms. We further restricted the sample to 141 those farms engaged in farming types (FT)4 typical of the Piedmont mountainous area, specifically 142 farms specialised in cattle, sheep and goats, and fruit production. The analysis covers a ten-year 143 period, from 2012 to 2022. 144 To distinguish between farms in mountainous and non-mountainous areas, we used the receipt 145 of the CA as a classification criterion. Farms receiving this support were classified as located in 146 mountainous areas, while those not receiving it were considered non-mountainous. This approach 147 was adopted since the FADN database defines farm altitude based on the farm center, which may not 148 always correspond to the actual location of the arable land. Using CA support as a proxy provided a 149 more reliable way to identify mountain farms. Moreover, this classification appears consistent, as 150 available data show that, on average, 88% of the total eligible regional hectares received the payment 151 during the years considered5. In addition, farms larger than 100 hectares were excluded from the 152 2 https://rica.crea.gov.it/index.php?lang=en 3 The standard output is the average monetary value of the agricultural output at the farm-gate price of each agricultural product (crop or livestock) in a given region. It is calculated by Member States per hectare or per head of livestock, by using basic data for a reference period of 5 successive years. 4 Classify farms based on their typological affinities, which reflect the similarity of each agricultural activity to others. The TF are determined by the relative significance of various enterprises within each farm. This relative significance is quantitatively measured as the proportion of each enterprise's standard output to the total standard output of the farms. 5 https://servizi.regione.piemonte.it/catalogo/anagrafe-agricola-piemonte-dati-sintesi; PSR Piemonte – dati di sintesi | Servizionline https://servizi.regione.piemonte.it/catalogo/anagrafe-agricola-piemonte-dati-sintesi https://servizi.regione.piemonte.it/catalogo/psr-piemonte-dati-sintesi https://servizi.regione.piemonte.it/catalogo/psr-piemonte-dati-sintesi ________________________________________________________________________________________________ 7 analysis, as they are not eligible for CA payments. Including them would have led to a 153 misclassification of their geographical location and distorted the estimation. 154 Based on this criterion, the final sample consists of an unbalanced panel dataset of 525 farms, 155 resulting in a total of 3,171 observations, of which 36.14% are located in mountainous areas. Table 1 156 presents the distribution of the sample by FT, farm scale and geographic location. The majority of the 157 sample comprises farms specialising in cattle, followed by those engaged in fruit production, and 158 lastly, sheep and goat farming. Farms specialised in sheep and goats are the only group with a higher 159 number of observations in mountainous areas compared to plains. 160 161 Table 1. Sample size categorised by Farming Type, area and farm scale. 162 Farm Scale Cattle Sheep and goats Fruit Total Non mountain area Small 364 60 546 970 Large 853 21 181 1,055 Mountain area Small 262 39 228 529 Large 486 67 64 617 Total 1,965 187 1,019 3,171 Note: Small-scale = UAA < 20 ha; Large-scale = UAA between 20 and 100 ha 163 2.2. ANALYTICAL FRAMEWORK 164 The rationale behind the CA support measure assumes that farms located in mountainous areas 165 face a significant income disparity compared to those in non-mountainous areas. The first step of our 166 analysis aimed to empirically test whether this income disparity exists when excluding the effects of 167 the CA. Subsequently, we assessed the extent to which the CA narrows the income gap between 168 mountain and non-mountain farms. 169 To conduct this analysis, we used a pooled multivariate regression model. This model was 170 selected because it allows for the simultaneous estimation of the effects of several explanatory 171 variables on multiple dependent variables over time, making it particularly suitable for unbalanced 172 ________________________________________________________________________________________________ 8 panel data such as ours (Baltagi, 2008). The analysis was structured around two key dependent 173 variables: net income (NI) per hectare of utilized agricultural area (UAA) excluding CA support, and 174 overall NI per hectare of UAA. This distinction enabled us to isolate the specific contribution of CA 175 payments in narrowing income disparities. 176 Looking at the descriptive statistics in Table 2, we observe that, on average, the NI excluding 177 CA support is 41.16% lower for mountain farms compared to non mountain farms. However, when 178 considering the overall NI, the income gap narrows to 37.76%. 179 180 Table 2. Descriptive statistics of dependent variables 181 NI excluding CA/ha NI/ha Non mountain Mountain Non mountain Mountain Farm scale M (€) SD M (€) SD M (€) SD M (€) SD Cattle Small 3,228 7,556 1,115 2,182 3,228 7,556 1,251 2,193 Large 2,310 2,655 1,140 1,830 2,310 2,655 1,251 1,832 Sheep and goats Small 2,079 9,849 1,006 3,124 2,079 9,849 1,142 3,132 Large 355 356 793 1,016 355 356 908 1,047 Fruit Small 4,421 4,826 4,344 3,985 4,421 4,826 4,488 3,992 Large 5,129 5,594 3,878 3,350 5,129 5,594 4,016 3,365 Overall 3,265 5,120 1,906 2,925 3,265 5,120 2,032 2,936 Note: Small-scale = UAA < 20 ha; Large-scale = UAA between 20 and 100 ha 182 For the independent variables, we included the farm’s geographical location that distinguishes 183 whether the farm is in a mountainous or non-mountainous area, as this is the primary focus of our 184 study. Additionally, we considered several control variables, selected based on previous studies that 185 identified key characteristics from the FADN database as significant determinants of farm income 186 (Andrejovská & Glova, 2022; Kryszak et al., 2021; Ryś-Jurek, 2019; Średzińska, 2018; Strzelecka et 187 al., 2018). These variables include a time variable to control for yearly trends, farm characteristics 188 such as economic dimension (ED) of the farm, measured in terms of standard output, the number of 189 farm labour units, the percentage of family labour, the hectares of UAA, the specific FT, the organic 190 certification and some personal characteristics of the farmer such as age, gender and level of education 191 ________________________________________________________________________________________________ 9 (Table 3 and 4). An interaction term between FT, farm location and farm scale were included to 192 capture any differential effects that may arise from the combination of these three factors. This 193 approach enables us to consider how the limitations imposed by mountainous terrain may affect each 194 FT and farm scale differently. 195 The regression model assumes that the error terms for the two dependent variables are identical, 196 reflecting the premise that residual farm income, after controlling for the relevant variables, should 197 behave similarly across the two income measures. Given that the same farms are observed across 198 multiple years, we addressed the potential issue of autocorrelation by clustering the standard errors 199 at the farm level. This adjustment corrects for any non-independence of observations within each 200 farm over time, enhancing the robustness of our estimates. 201 Additionally, we addressed selection bias through inverse probability weighting (IPW). First, 202 we defined the propensity score as p(xᵢ) = Pr(Wᵢ = 1 | xᵢ, β), where Wᵢ is an indicator equal to 1 if a 203 farm is a “Mountain farm” (i.e., receiving CA support) and 0 otherwise. The vector β represents the 204 parameters, and xᵢ is a vector of observed covariates. The propensity score reflects the probability 205 that a farm receives CA support and is modelled as F[H(xᵢ)], where F is typically the logistic or 206 normal cumulative distribution function (Guo & Fraser, 2010). In this study, the propensity scores 207 were estimated using a logit model. Inverse probability weights were then computed by taking the 208 reciprocal of the estimated probability of the observed treatment status. For a treated unit (“Mountain 209 farm”), the weight is 1/p(xᵢ), while for a control unit (“Non mountain farm”) the weight is 1/(1 − 210 p(xᵢ)). These weights adjust for differences in covariate distributions between the two groups. Finally, 211 the computed weights were used in the multivariate regression (see Bellon et al., 2015) to estimate 212 the effect of being in a mountainous area on net income. This weighted regression aims to balance 213 the sample, mimicking a randomized experiment and providing less biased estimates (Guo & Fraser, 214 2010). 215 ________________________________________________________________________________________________ 10 After estimating the regression models, we conducted pairwise comparisons to identify 216 statistically significant differences in NI between mountain and non-mountain farms across the 217 various FT. 218 To further explore the specific role of the CA in reducing the income gap, we estimated the 219 required financial magnitude of CA payment to effectively address income disparities between 220 mountain and non-mountain farms. To achieve this, we conducted a pooled regression analysis, where 221 the dependent variable was net income per hectare of UAA, and the independent variables included 222 the CA payment per hectare of UAA, along with the same control variables used in the previous 223 analysis. Using the model estimates, we calculated how the marginal increase in CA payment would 224 affect the income of mountain farms, identifying the per-hectare payment needed to fully bridge the 225 income gap with non-mountain farms. 226 2.3. VARIABLES DESCRIPTION 227 The description of the variables reveals that, on average, farms in the dataset employ slightly 228 more than two workers, with an average of 2.07 labour units (LU), and rely predominantly on family 229 labour, which represents approximately 92.2% of the total workforce. The average age of the farm 230 managers is 53.9 years (Table 3). 231 Regarding ED, the majority of farms fall into the mid-range, with 40.24% of farms classified 232 within the ED category, having a standard output between €100,000 and €500,000 per year. Farms 233 with a standard output between €50,000 and €100,000 represent 24.38% of the dataset, while those 234 with an output between €25,000 and €50,000 account for 17.85%, and 13.88% fall within the €8,000–235 €25,000 range. Only a small proportion of farms exceed €500,000 in standard output (3.66%). In 236 terms of farm specialisation, 61.97% of farms focus on cattle farming, 32.13% specialise in fruit 237 production, and 5.97% are dedicated to sheep and goat farming. The majority of farms (63.56%) are 238 located in non-mountainous areas, while 36.14% operate in mountain regions. 239 ________________________________________________________________________________________________ 11 For farm scale, farms are classified as Small (< 20 hectares) and Large (20 to 100 hectares). 240 This classification reflects the structure of the Compensatory Allowance (CA), which is granted in 241 full to farms under 20 hectares, while farms above this threshold receive a proportionally reduced 242 payment. The sample is nearly evenly distributed, with 47.27% of farms classified as small and 243 52.73% as large. 244 In terms of farm management, 86.03% of the farm managers are male. Farm diversification is 245 relatively limited, with only 18.46% of farms engaged in activities beyond primary agricultural 246 production. Similarly, organic farming is a niche practice, with only 11.51% of farms certified as 247 organic. Education levels vary among farm managers, with 62.37% having completed secondary 248 school, and 22.31% holding a high school diploma. A smaller proportion has attained higher 249 education, with 1.58% holding a bachelor's degree and 0.06% an associate degree, while 13.68% have 250 primary or no formal education. The data spans multiple years, from 2012 to 2022, with a fairly 251 balanced distribution of observations across these years (Table 4). 252 253 Table 3. Description of the continuous independent variables 254 Description Unit of Measure M SD LU Farm’s labour unit n 2.07 1.96 FLU Percentage of family labour % 92.2 1.97 Age Farmer’s age n 53.9 12.03 255 256 257 Table 4. Description of the discrete allowance variables 258 Description n % ED € of Standard Output 8,000 – 25,000 440 13.88 25,000 – 50,000 566 17.85 50,000 - 100,000 773 24.38 100,000 – 500,000 1,276 40.24 ________________________________________________________________________________________________ 12 500,000 – 1,000,000 108 3.41 > 1,000,000 8 0.25 Farm scale Small (< 20 ha) 1,499 47.27 Large (20 to 100 ha) 1,672 52.73 FT Cattle 1.965 61.97 Sheep and goats 187 5.97 Fruit 1.019 32.13 Farm location Non mountain area 2,025 63.56 Mountain area 1,146 36.14 Gender Woman 443 13.97 Man 2,728 86.03 Diversification Yes 588 18.46 No 2,583 81.46 Organic Yes 365 11.51 No 2.806 88.49 Level of education No formal education 70 2.22 Primary school 362 11.46 Secondary school 1,971 62.37 High school diploma 705 22.31 Associate degree 2 0.06 Bachelor’s degree 50 1.58 Year 2012 316 9.97 2013 326 10.28 2014 316 9.97 2015 309 9.74 2016 314 9.90 2017 305 9.62 2018 304 9.59 2019 188 5.93 2020 186 5.87 2021 307 9.68 2022 300 9.46 259 3. RESULTS 260 The findings from the IPW-adjusted multivariate regression model, as outlined in Table 5, show 261 that the control variables are significant in the expected direction across the two equations. 262 Specifically, an increase in the Economic Dimension positively influences the dependent variables, 263 as does a marginal increase of UAA. Moreover, farms specialised in fruit production scored higher 264 on the dependent variables when compared to cattle farms. Focusing on the variable that distinguishes 265 ________________________________________________________________________________________________ 13 between farms located in mountainous areas and those not situated in mountainous areas, we observe 266 that in the two cases the coefficient is significant and negative. This suggests that there is a gap in net 267 income between farms located in mountainous areas and those not located in mountainous areas that 268 the CA support is unable to bridge. 269 ________________________________________________________________________________________________ 14 270 Table 5. Results of the IPW-adjusted multivariate regression model. 271 NI excluding CA/ha NI/ha Coefficient Robust SE Coefficient Robust SE Location (Mountain) -1319.44** 603.97 -1201.49** 605.79 FT (vs Cattle) Sheep & goats 1737.47* 1003.23 1741.02* 1005.25 Fruit 2419.50*** 883.81 2421.13*** 884.60 FT × Location × Farm scale Cattle × Mountain × Large -2080.62** 821.62 -2065.50** 822.79 Cattle × Mountain × Small -832.43* 481.14 -867.59* 483.10 Sheep & goats × Mount × Large -4283.34*** 944.87 -4282.66*** 945.98 Sheep & goats × Mount × Small -1064.76 1018.55 -1057.00 1022.25 Sheep & goats × Non M × Small -2620.65*** 996.31 -2662.57*** 1001.69 Fruit × Mountain × Large -2069.54** 801.21 -2063.17** 800.91 Fruit × Mountain × Small -1513.35 939.55 -1513.56 947.54 Fruit × Non mountain × Small 86.07 927.07 91.64 930.49 LU 248.43** 116.99 247.35** 117.11 FLU 1179.13 1090.93 1161.00 1090.79 ED 1070.17*** 260.69 1060.88*** 260.84 Woman -298.93 299.37 -299.62 299.48 Diversification -18.91 426.15 -22.89 426.63 Organic 39.21 475.92 57.18 477.77 Education 16.50 209.12 10.93 208.72 Age (years) 24.56** 12.35 24.28** 12.30 Year (vs 2012) 2013 -66.35 353.20 -83.99 357.96 2014 145.63 405.84 141.99 405.14 2015 444.99 331.13 494.20 329.88 2016 1391.10** 690.63 1384.50** 690.78 ________________________________________________________________________________________________ 15 2017 374.81 428.84 367.44 428.68 2018 1072.66** 490.55 1068.52** 487.95 2019 325.40 583.24 320.91 584.57 2020 520.55 555.68 499.98 554.84 2021 103.94 395.84 107.05 394.49 2022 -156.24 386.01 -150.14 384.57 Constant -3252.46** 1416.46 -3188.35** 1413.39 Note: * p<0.10, ** p<0.05, *** p<0.01. 272 Post-hoc pairwise analyses indicate that, in the two regressions, the gap is not uniformly evident 273 across all FTs and farm scales. From the first regression, which has NI excluding the amount of CA 274 support received as the dependent variable, a negative income gap exists for small farms specialised 275 in cattle and in sheep and goats, while large farms in these same categories and those specialised in 276 fruit do not show a statistically significant gap. Focusing on small cattle farms, the estimated 277 magnitude of the gap excluding CA support is -€1,319.44 per hectare, while the estimated NI gap is 278 -€1,201.49 per hectare. For small farms specialised in sheep and goats, the estimated magnitude of 279 the gap excluding CA support is -€2,384.19 per hectare, which decreases to -€2,258.49 when 280 considering the estimated NI gap (Table 6). 281 282 Table 6. Results of the post-hoc analyses. 283 NI excluding CA NI χ² P-value Scale Gap (€/ha) Robust SE Gap (€/ha) Robust SE Cattle Small -1319.44 ** 603.97 -1201.49** 605.79 159.88 < .0001 Large -71.24 305.14 -3.586 301.425 68.93 < .0001 Sheep & goats Small -2384.19 *** 867.49 -2258.49 *** 870.54 101.85 < .0001 Large 343.25 386.87 418.60 397.00 9.31 0.002 Fruit Small -1233.37 754.99 -1109.84 757.62 147.01 < .0001 Large -763.25 946.99 653.55 953.66 0.48 0.486 Note: NI = Net income; CA = Compensatory allowance; Gap = estimate of the operating income difference between mountain and 284 non-mountain farms; ** p < .05, *** p<0.01. The last two columns refer to the Wald test between the two estimated gaps, where H0: 285 NI excluding CA = NI. 286 287 ________________________________________________________________________________________________ 16 Our analysis further proceeded to determine the required financial magnitude of CA payment 288 to effectively address income disparities between farms situated in mountainous and non-289 mountainous areas. The analysis was conducted exclusively on small farms specialising in cattle and 290 sheep and goats, as these were the farming types where the income gap was found to be negative and 291 statistically significant. As shown in Figure 1, mountain farms currently receive an average CA 292 payment of around €135/ha. To fully close the income gap between farms in mountainous areas and 293 those in non-mountainous areas, the CA payment would need to more than double, reaching 294 approximately €340/ha. 295 Figure 1. Estimated increase in CA required to bridge the net income gap between mountain and 296 non-mountain farms. The shaded area represents a 95% confidence interval. The dashed line depicts 297 the average CA received by mountain farms. 298 299 ________________________________________________________________________________________________ 17 4. DISCUSSION 300 The findings of this study provide a deeper understanding of the economic challenges faced 301 by farms in Piedmont’s mountainous areas, especially concerning income disparities when compared 302 to farms in non-mountainous regions. In addition, the study highlights the role of public support in 303 addressing these challenges, particularly evaluating the effectiveness of compensatory measures such 304 as the CA. 305 Our results confirm the existence of an income gap between farms situated in mountainous 306 and non-mountainous areas. This disparity is largely driven by socio-ecological challenges such as 307 steep terrain, adverse climate conditions, and limited competitiveness, which collectively increase the 308 risk of land abandonment in mountain regions (Dax, 2021). Additional challenges, including an 309 ageing farming population, limited technical training, and the prevalence of low-input systems, 310 further undermine the economic sustainability of agriculture in these areas (Giannakis & Bruggeman, 311 2015; Strijker, 2004). These vulnerabilities appear to be particularly pronounced in smaller farms, 312 which tend to have fewer resources to adapt to structural constraints. 313 ________________________________________________________________________________________________ 18 Moreover, our study demonstrates that this income gap is not uniformly distributed across different 314 types of farming, confirming that distinct mountain farming types result in different economic 315 outcomes (Papić Milojević & Bogdanov, 2023). The income gap was found to be statistically 316 significant, and negative in small farms specialised in cattle and sheep and goats. Importantly, these 317 differences are further influenced by farm scale: small-scale farms within these specialisations show 318 a marked income disadvantage, whereas large farms within the same types often do not present a 319 significant income gap. The reason for this phenomenon could be attributed to the high fixed costs 320 borne by small farms specialized in animal breeding, which, combined with lower production 321 volumes, result in a decrease in profitability (Kuhl et al., 2019). At the same time, the absence of a 322 significant income gap for other farming types raises important policy considerations. If certain 323 sectors do not experience substantial economic disadvantages due to mountain constraints, this might 324 suggest that CA support should not be uniformly distributed, but rather adjusted according to the 325 specific needs of different farming types and scales. Current eligibility rules already distinguish 326 between small and large farms—granting full CA support only to those under 20 hectares—but our 327 findings show that this differentiation, while appropriate in principle, does not go far enough in 328 correcting income disparities. 329 Furthermore, our results indicate that the income gap in small farms specialized in animal 330 breeding is only partially bridged by CA support, which covers only a small portion of this gap. This 331 finding echoes previous concerns regarding the effectiveness of policies aimed at supporting 332 mountain farms in Italy (Whitaker, 2024). However, CA support remains insufficient to fully close 333 the gap. Previous studies have shown that the impact of ANC support is variable, depending on 334 context (Namiotko et al., 2017). While some research aligns with our findings, demonstrating ANC 335 supports’ limited role in bolstering mountain farm income (Ferto et al., 2022; Wieliczko et al., 2018), 336 other studies have highlighted its effectiveness in other agricultural types, such as cereals (Klima et 337 al., 2021), or in boosting income for organic farms, especially small-scale (Veveris et al., 2014). 338 ________________________________________________________________________________________________ 19 Even though current eligibility criteria consider factors such as the agricultural system, the level 339 of land disadvantage, and farm scale (Regione Piemonte, 2016), they do not fully capture the 340 economic heterogeneity across those factors and different farming types. This limitation suggests that 341 a more differentiated approach in the allocation of CA support could enhance its effectiveness. In 342 particular, redistributing resources from less-affected sectors to those that bear the highest costs of 343 mountain farming could provide a more targeted use of funds. 344 Notably, the current policy framework already recognises the importance of farm scale by 345 modulating CA support, however our findings indicate that this differentiation is not sufficient to 346 fully compensate for the income gap observed in small mountain farms. 347 To further reduce the income disparities, a multifaceted strategy should be considered. In 348 addition to revising CA distribution, policies aimed at enhancing productivity and market access for 349 mountain farms could be beneficial. Investments in infrastructure, technological innovation, and 350 training programs tailored to the needs of mountain farmers could improve efficiency and economic 351 resilience (Dax & Fischer, 2018; Pezzini, 2001). Furthermore, fostering cooperatives and producer 352 organizations may strengthen the bargaining power of mountain farmers, enabling them to capture a 353 larger share of market value and thereby reducing income disparities (Knickel et al., 2018). 354 Nonetheless, the income gap for farms specialised in animal breeding is not fully addressed 355 by CA support, and achieving full compensation would require more than doubling the per-hectare 356 payment. While increasing financial support could help, it is essential to evaluate its cost-357 effectiveness. A simple increase in payments may not be the most efficient approach, as it could lead 358 to budgetary constraints without necessarily addressing structural inefficiencies. Instead, a 359 combination of increased financial aid and complementary measures, such as targeted investments in 360 modernization, innovation, and value-chain integration, could yield better long-term results. Future 361 policy adjustments should consider the balance between financial sustainability and the actual impact 362 of interventions on mountain farm incomes. 363 ________________________________________________________________________________________________ 20 It is essential to consider a redistribution of support payments that reflects the varied economic 364 outcomes across different mountain FT, underscoring the need for sectoral policies tailored to these 365 differences (Papić Milojević & Bogdanov, 2023). A more tailored distribution of CA support, based 366 not only on land characteristics but also on the specific economic challenges of different farming 367 types, could lead to a more equitable and effective support mechanism. In particular, an equitable 368 distribution of CA support may be challenging to achieve with a flat per-hectare payment structure. 369 Allocating contributions based on labour units rather than area could offer a more equitable approach, 370 and presents a promising avenue for future research. Additionally, first-pillar direct payments could 371 play a critical role in this redistribution by facilitating a convergence process that reallocates resources 372 in favour of mountain areas (Tantari et al., 2017). 373 It is important to acknowledge a broader structural disadvantage that our analysis does not capture: 374 the limited range of viable production options in LFAs. Mountain farms are often constrained not 375 only by higher production costs but also by ecological and climatic factors that restrict crop and 376 livestock choices, limiting their ability to diversify or switch to more profitable activities. This implies 377 an additional layer of opportunity cost, which is not addressed through intra-sectoral profitability 378 comparisons alone. Capturing such constraints would require integrating agronomic feasibility 379 assessments and opportunity cost modeling into a broader analytical framework, an important, though 380 currently out-of-scope, direction for future research. However, we believe recognizing this limitation 381 is essential when designing compensation schemes and rural development strategies that aim to fully 382 reflect the multi-dimensional nature of disadvantage in LFAs. 383 Beyond public support, narrowing the income gap between mountain and non-mountain farms 384 might also require consumer recognition and willingness to pay a price premium for mountain-origin 385 products. This added value would create a more sustainable revenue stream for mountain farms, 386 helping to offset higher production costs and lower yields, and thus supporting the long-term viability 387 ________________________________________________________________________________________________ 21 of agriculture in these challenging areas (Cei et al., 2023; Staffolani et al., 2023; Mazzocchi & Sali, 388 2021). 389 It is important to acknowledge, however, that the empirical strategy employed in this study 390 does not allow for causal inference. Although we apply statistical techniques to adjust for observable 391 differences between farms, unobserved confounding factors may still influence the results. As such, 392 the findings should be interpreted as associations rather than causal effects. Policy implications 393 should be considered exploratory, offering indications rather than definitive prescriptions. 394 395 5. CONCLUSIONS 396 This study aimed to assess the effectiveness of RDP’s measures to bridge the income gap 397 between farms located in mountainous and non-mountainous areas in Piedmont. The results show 398 that a statistically significant and negative income gap exists only in small farms specialized in cattle 399 and sheep and goats. Net of the CA support provided through the RDP, the average shortfall amounts 400 to €1,319.44 per hectare for cattle farms and €2,384.19 per hectare for sheep and goat farms. 401 Compensatory allowance support helps reduce this gap by 8.93% for cattle farms and 5.28% for sheep 402 and goat farms, though a substantial disparity remains. 403 While increasing compensatory allowances (CAs) could help narrow income disparities in 404 mountain areas, our findings suggest that this measure alone is neither financially sustainable nor 405 sufficient to address the broader challenges these regions face. A more promising approach involves 406 embedding CAs within integrated territorial development strategies that tackle structural and 407 systemic constraints. For example, policy experiences in France during the early 2000s highlight the 408 value of bundled interventions, whereby CAs were linked to participation in agri-environmental or 409 organic farming schemes—promoting both economic viability and environmental stewardship. At 410 the regional level, the Strategia per le Montagne del Piemonte (DGR 27/02/2023) offers a concrete 411 ________________________________________________________________________________________________ 22 illustration of how multi-sectoral action plans can support youth entrepreneurship, workforce 412 training, and the valorization of ecosystem services alongside traditional income support (Regione 413 Piemonte, 2022; 2023). Additionally, shifting from per-hectare to per-labor-unit payments could 414 enhance the equity and effectiveness of support, better aligning aid with actual farm effort and 415 viability—though this would require overcoming significant administrative and WTO-related 416 hurdles. Ultimately, advancing the cost-effectiveness and legitimacy of rural policy in mountain areas 417 calls for a transition from a logic of compensation to one of strategic investment, in line with the 418 broader vision of the Strategia Nazionale per le Aree Interne, recognizing mountain farming as a 419 cornerstone of territorial resilience and social cohesion 420 In general, the findings indicate that CA support does not fully bridge the income gap, even 421 when supplemented by contributions from other RDP measures. To completely close this gap, CA 422 support would need to more than double, increasing from €135/ha to €340/ha, which would require 423 a substantial rise in allocated resources. Given the limited economic feasibility of this approach, a 424 more viable solution would involve not only increasing contributions but, more importantly, 425 redistributing them. All in all, mountainous areas constitute approximately 45% of Piedmont’s total 426 land area and 30% of its agricultural land, making them crucial for the region’s economic and 427 environmental sustainability. The study serves as an exploratory analysis, with the intention of 428 expanding the research to additional Alpine regions in future studies. 429 430 ACKNOWLEDGMENTS 431 This work was funded by the European Union Next-GenerationEU: (Call PRIN 2022- Italian Ministry of 432 University and Research (MUR) - RECIPE - cod. 20224ZHNXE). Also, we would like to thank the two 433 anonymous reviewers for their thoughtful comments and efforts towards improving the initial draft of our 434 manuscript. 435 436 ________________________________________________________________________________________________ 23 REFERENCES 437 Andrejovská, A., & Glova, J. (2022). Sustainability of farms in EU countries in the context of income 438 indicators: Regression analysis based on a new classification. Agriculture, 12(11), 1884. 439 Baltagi, B. H., (2008). Econometric analysis of panel data (Vol. 4, pp. 135-145). Chichester: Wiley. 440 Bellon, M. R., Gotor, E., & Caracciolo, F. (2015). Assessing the effectiveness of projects supporting 441 on-farm conservation of native crops: Evidence from the high Andes of South America. World 442 development, 70, 162-176. 443 Borsotto, P., Cagliero, R., & Trione, S. (2010). Le zone svantaggiate e le altre zone con svantaggi 444 specifici. In Le politiche comunitarie per lo sviluppo rurale. Il quadro degli interventi in Italia 445 (pp. 127-142). INEA. 446 Cagliero, R., Iacono, R., Licciardo, F., Prandi, T., & Rossi, N. (2018). La montagna e le zone 447 svantaggiate nei Programmi di Sviluppo Rurale: Una valutazione delle indennità compensative 448 attraverso la Rica. In Economia Agro-Alimentare (Vol. 20, Issue 3). 449 Cei, L., Defrancesco, E., Gatto, P., & Pagliacci, F. (2023). Pay more for me, I’m from the mountains! 450 The role of the EU Mountain Product term and other credence attributes in consumers’ 451 valuation of lamb meat. Agricultural and Food Economics, 11(1), 12. 452 Cesaro, L., & Marongiu, S. (2013). L’agricoltura di montagna svantaggio o risorsa? RRN Magazine, 453 (6), 6-9. 454 Cooper, T. , Baldock, D. , Rayment, M. , Kuhmonen, T. , Terluin, I. , Swales, V. , Poux, X. , 455 Zakeossian, D. , & Farmer, M. (2006). An Evaluation of the Less Favoured Area Measure in 456 the 25 Member States of the European Union. IEEP Brussels Report for Directorate-General 457 Agriculture. 458 Dax, T., Schroll, K., Machold, I., Derszniak-Noirjean, M., Schuh, B., & Gaupp-Berghausen, M. 459 (2021). Land abandonment in mountain areas of the EU: An inevitable side effect of farming 460 modernization and neglected t hreat to sustainable land use. Land, 10(6). 461 Dax, T., & Fischer, M. (2018). An alternative policy approach to rural development in regions facing 462 population decline. European Planning Studies, 26(2), 297-315. 463 Direzione Agricoltura - Regione Piemonte. Allegato ai Quaderni della Regione Piemonte – Collana 464 “Agricoltura”. Direttore Responsabile: Luciano Conterno. 465 Ferlaino, F., Rota, F. S., Dematteis, G., et al. (2019). Le montagne del Piemonte. Curato da Fiorenzo 466 Ferlaino, Francesca Silvia Rota, e Giuseppe Dematteis. Istituto di Ricerche Economico Sociali 467 del Piemonte (IRES). Via Nizza 18, 10125 Torino. 468 Fertő, I., Bojnec, Š., & Podruzsik, S. (2022). Do subsidies decrease the farm income inequality in 469 Hungary?. AGRIS on-line Papers in Economics and Informatics, 14(2), 49-56. 470 Fraschetti, L., Lupia, F., and Storti, D. (2017). Le zone agricole soggette a limitazioni naturali nelle 471 politiche comunitarie. Agriregionieuropa anno 13 n°51, Dic. 472 Giannakis, E., & Bruggeman, A. (2015). The highly variable economic performance of European 473 agriculture. Land use policy, 45, 26-35. 474 Guo, S., & Fraser, M. W. (2010). Propensity score analysis: statistical methods and applications. 475 Psychometrika, 75(4), 775. 476 ________________________________________________________________________________________________ 24 ISTAT (2020) "Annuario statistico italiano – Territorio". [Online] Available: 477 https://www.istat.it/it/files//2020/12/C01.pdf [Accessed: Nov. 01, 2024]. 478 Klima, K., Kliszcz, A., Puła, J., & Lepiarczyk, A. (2020). Yield and profitability of crop production 479 in mountain less favoured areas. Agronomy, 10(5), 700. 480 Knickel, K., Redman, M., Darnhofer, I., Ashkenazy, A., Chebach, T. C., Šūmane, S., ... & Rogge, E. 481 (2018). Between aspirations and reality: Making farming, food systems and rural areas more 482 resilient, sustainable and equitable. Journal of rural studies, 59, 197-210. 483 Kryszak, Ł., Guth, M., & Czyżewski, B. (2021). Determinants of farm profitability in the EU regions. 484 Does farm size matter?. Agricultural Economics/Zemědělská Ekonomika, 67(3). 485 Kühl, S., Flach, L., & Gauly, M. (2020). Economic assessment of small-scale mountain dairy farms 486 in South Tyrol depending on feed intake and breed. Italian Journal of Animal Science, 19(1), 487 41-50. 488 Mazzocchi, C., & Sali, G. (2022). Supporting mountain agriculture through “mountain product” label: 489 A choice experiment approach. Environment, Development and Sustainability, 24(1), 701-723. 490 Namiotko, V., Góral, J., & Soliwoda, M. (2017). The economic situation of farms located in less 491 favoured areas on the example of Lithuania and Poland. Agricultural and Resource Economics: 492 International Scientific E-Journal, 3(4), 5-19. 493 NUVAL Piemonte. (2016). Scheda di Valutazione Misura 211: Indennità a favore degli agricoltori 494 delle zone montane - Rapporto di valutazione ex post del PSR 2007-2013 Regione Piemonte. 495 Regione Piemonte - Direzione Risorse Finanziarie e Patrimonio. 496 Oxouzi, E., Melfou, K., Galea, M., & Papanagiotou, E. (2012). Economic performance and crop farm 497 efficiency in mountainous and other less favoured areas in Greece. Bulgarian Journal of 498 Agricultural Science, 18(6), 846-853. 499 Papić Milojević, R., & Bogdanov, N. (2024). Typology of farms in areas with natural constraints--500 diversity of livelihood strategies and their determinants. Applied Ecology & Environmental 501 Research, 22(2). 502 Pezzini, M. (2001). Rural policy lessons from OECD countries. International Regional Science 503 Review, 24(1), 134-145. 504 Poláková, J. (2019). Subsidies to less favoured areas in the Czech Republic: Why do they matter? 505 Prague Economic Papers, 28(4). 506 Regione Piemonte. (2016). Guida al PSR 2014-2020: Attività di informazione realizzata nell’ambito 507 del piano di comunicazione del Programma di sviluppo rurale 2014-2020, Misura 20 – 508 Assistenza tecnica. 509 Regione Piemonte. (2023). Piano di Azione 2024. Strategia per le Montagne del Piemonte. Visioni 510 per i territori montani. Dicembre 2023. 511 Regione Piemonte. (2022). Documento di Strategia Regionale per lo Sviluppo Sostenibile 512 (SRSvS). Luglio 2022. 513 Regulation (EU) No 1305/2013 of the European Parliament and of the Council of 17 December 2013 514 on support for rural development by the European Agricultural Fund for Rural Development 515 (EAFRD) and repealing Council Regulation (EC) No 1698/2005. OJ L 347, 20.12.2013, p. 487. 516 Regulation (EC) No 1698/2005 of 20 September 2005 on support for rural development by the 517 European Agricultural Fund for Rural Development (EAFRD) 518 ________________________________________________________________________________________________ 25 Romagnoli, L., Giaccio, V., Mastronardi, L., & Forleo, M. B. (2021). Highlighting the drivers of 519 Italian diversified farms efficiency: A Two-stage DEA-panel Tobit analysis. Sustainability, 520 13(23), 12949. 521 Ryś-Jurek, R. (2019). Determinants of family farm income depending on farm size. Roczniki 522 (Annals), 2019(3). 523 Średzińska, J. (2018). Determinants of the Income of Farms in EU Countries. Studia Oeconomica 524 Posnaniensia, 6(2), 54-65. 525 Staffolani, G., Bentivoglio, D., & Finco, A. (2022). Consumers’ purchasing determinants towards 526 mountain food products. Sustainability, 14(14), 8282. 527 Strijker, D. (2005). Marginal lands in Europe—causes of decline. Basic and Applied Ecology, 6(2), 528 99-106. 529 Strzelecka, A., Zawadzka, D., & Kurdyś-Kujawska, A. (2018). Production Potential and Income of 530 Agricultural Holdings in Poland. Regional Barometer. Analyses & Prognoses, 16(3), 137-144. 531 Storti, D. (2013). Le zone agricole svantaggiate: ieri, oggi, domani. Agriregionieuropa. 532 Tantari, A., Persia, F., & Cardillo, C. (2019). Assessing the impact of direct payments convergence 533 on farm income inequality: The case of Italian farms. Journal of International Food & 534 Agribusiness Marketing, 31(4), 417-428. 535 Veveris, A., Lakovskis, P., & Benga, E. (2014). Economic Aspects Of Lfa And Organic Farming 536 Payments In Latvia. Management Theory & Studies for Rural Business & Infrastructure 537 Development, 36. 538 Whitaker, S. H. (2024). The impact of government policies and regulations on the subjective well-539 being of farmers in two rural mountain areas of Italy. Agriculture and Human Values, 1-19. 540 Wieliczko, B., Kurdyś-Kujawska, A., & Sompolska-Rzechuła, A. (2018). Economic component of 541 ANC payments. Example of the farms in Poland. Research For Rural Development, 2. 542 543