Achieving Sustainable Development Goals: The Case of Farms in Poland

Abstract

The aim of this study is to assess the factors determining the achievement of the Sustainable Development Goals (SDGs) in the context of improving the productivity of agricultural inputs—land, labor, and capital—and increasing farm income, using logistic regression. The analysis is based on primary data collected in 2023 during a pilot survey of 150 farms in the Central Pomerania region of Poland, conducted using the CATI (Computer-Assisted Telephone Interview) technique and a standardized questionnaire. The study examined factors related to farm income growth and to the productivity of land, labor, and capital. Logistic regression was applied to test 28 independent variables grouped into four categories: farm characteristics, production-related characteristics, farm asset-related characteristics, and household characteristics. Income growth was found to be associated with production-related and household characteristics. Land productivity showed associations with variables from all four categories, reflecting the complexity of this outcome. In the case of labor productivity, only household characteristics were significant, underscoring the strong connections between agricultural holdings and farming households. Capital productivity was associated with household characteristics, production-related characteristics, and farm asset-related characteristics. These findings highlight the need for an interdisciplinary approach to the sustainable development of agricultural holdings, integrating economic, production, and social dimensions.

1. Introduction

Sustainable development is a multidimensional concept that integrates the economic, ecological, social, and institutional subsystems into a cohesive whole, taking into account their mutual interactions [1]. The Sustainable Development Goals (SDGs) were set out in the plan Transforming our World: The 2030 Agenda for Sustainable Development [2], adopted in 2015 by 193 member states of the United Nations (UN). This document outlines 17 SDGs that address critical global challenges and focus on five fundamental themes: people, planet, prosperity, peace, and partnership [2].

This paper focuses on the sustainable development of agricultural holdings. In agriculture, sustainable development holds a distinctive position, encompassing the economic, social, and environmental dimensions of food production, land cultivation, livestock breeding, and farming and consumption [3].

In this context, it should be emphasized that sustainable agriculture encompasses not only environmental aspects. It also aims to increase farm efficiency and profitability, thereby increasing the income of farmers and their households. Land, labor, and capital are the fundamental factors of agricultural production, as confirmed by previous empirical research [4,5,6]. The efficiency with which these resources are used determines the level of agricultural income [7]. Therefore, increasing the productivity of these resources—land, labor, and capital—can contribute to better economic performance of farms and higher income. However, such improvements must be pursued in a way that aligns with broader sustainability objectives. As Chiarella et al. [5] point out, livelihoods play a fundamental role in sustainability, and an appropriate balance between productivity, conservation goals, and livelihoods is essential for sustainable land use and the achievement of long-term sustainable development goals.

This multidimensional nature of sustainability is reflected in the diversity of current research on the topic. Existing studies address agricultural sustainability indicators (e.g., [8,9,10,11]); the social dimension of sustainability [12]; assessments of agricultural sustainability in the European Union (EU) [10]; measurements of environmental sustainability at the farm level [13]; and analyses of production factor productivity that integrate environmental impacts through Agricultural Green Total Factor Productivity (AGTFP) [14,15].

The UN SDGs highlight the need to increase farmers’ incomes and to improve the productivity of land, labor, and capital. Consequently, an important research focus is the evaluation of SDG implementation in low-income regions such as Sub-Saharan Africa, South Asia, and Latin America [16,17,18,19]. A large proportion of the world’s poorest people are smallholder farmers living in Sub-Saharan Africa [20]. In both Sub-Saharan Africa and South Asia, small farms dominate, yet their output remains low due to factors such as declining soil fertility, limited use of external inputs, unfavorable policies, weak market and institutional frameworks, pests and diseases, and the impacts of climate change [21]. Similarly, Latin America and the Caribbean have experienced persistent income inequality alongside high poverty rates in rural areas, a widening rural–urban income gap and a shrinking rural population [22]. Therefore, increasing the productivity of smallholder farmers in these regions is crucial for poverty eradication and for achieving the SDGs [20]. At the same time, evidence suggests that the least developed countries in Sub-Saharan Africa and South Asia face significant challenges in implementing SDGs and in monitoring progress [23].

In the European context, enhancing agricultural productivity is also a core objective of the EU’s Common Agricultural Policy [24]. This underscores the importance of examining the factors associated with productivity growth and income increases in farming, particularly within the framework of advancing the SDGs.

Previous studies indicate that numerous factors are associated with increases in the productivity of basic production inputs—land, labor, and capital—which, in turn, are linked to higher agricultural income. These include both endogenous factors—directly related to the farm—and exogenous factors—external and independent of its operations.

The first group includes conditions such as the farm size [5,25,26]; resources available to the farm (land, machinery and equipment, labor force); sources of financing and the efficiency of their use [27,28,29,30,31,32,33]; location [32]; production type (specialization) [25,34]; and the value of agricultural production [7,29,35,36]. Given the strong links between the farm and the farmer’s household, these factors also encompass socio-economic characteristics of the farmer and their household, including age, gender, education level, household size and structure, and stage of household development [25,37,38].

The second group, comprising exogenous factors, primarily includes natural conditions (e.g., [39]); prices of production inputs and sales prices of agricultural products (e.g., [29]); and transport infrastructure [40]. In the era of digitalization, the relevance of this dimension to agricultural productivity is increasingly recognized. Research by Zhang et al. [41] indicates that the digitalization of agriculture and rural areas is associated with improvements in agricultural green total factor productivity, mainly through enhancement of technical efficiency. Recent studies also examine the use of artificial intelligence to support agricultural efficiency and, in turn, potentially increase farmers’ incomes [42]. For farms located within the European Union (EU), it is additionally important to account for the role of direct support schemes in generating financial surpluses from agricultural activities, as direct payments constitute a substantial share of these farms’ income [43,44].

Undoubtedly, the resources of land, labor, and capital—forming the basis of a farm’s production potential—shape the possible scale of agricultural activity and are closely linked to both the value of production and the level of income achieved [45]. Therefore, progress toward the SDGs at the farm level can be supported, among other means, by enhancing the productivity of these resources. Figure 1 presents the relationships between increases in farm income, improvements in land, labor and capital productivity, and specific SDG targets, highlighting their relevance to the sustainable development of the agricultural sector.

Increased farmer incomes are linked to the achievement of targets 1.1 and 1.2 by reducing poverty both globally and relative to national poverty lines. They also support target 2.3, which aims to double the income of small-scale food producers, and target 10.1, which seeks to increase the incomes of the poorest 40% of the population at a rate exceeding the national average. The second factor, increased land productivity—measured as higher yields per unit area—is associated with target 2.3 on doubling agricultural productivity. It also supports target 2.4, which promotes the adoption of sustainable agricultural practices that increase productivity, and target 2.a, which focuses on investments in agricultural infrastructure and technologies that enhance production capacity. Closely related, higher labor productivity in agriculture—defined as higher output per worker—is linked to target 2.3, which addresses improvements in farmers’ productivity, and to target 8.2, which promotes economic efficiency through modernization, innovation, and more effective use of labor resources. Finally, improved capital efficiency in agriculture—reflected in higher returns on investments in machinery, infrastructure, and technology —is associated with target 2.3 on increasing agricultural production efficiency and with target 2.a on scaling up investment in agricultural infrastructure and research. It is also linked to target 9.3, which aims to improve access to financial services and markets for small agricultural enterprises, and target 9.4, which emphasizes infrastructure modernization and resource efficiency.

This study forms part of the aforementioned research stream, focusing on the first group of factors—endogenous factors—by considering both household characteristics and farm-specific conditions. It is worth noting that, in the context of sustainable development—particularly in agriculture—in Central and Eastern Europe, an ongoing discussion addresses differences in the achievement of goals across individual countries. However, previous research has focused primarily on variables based on aggregated data [46,47], including those related to farm characteristics [48]. A similar situation applies to the assessment of farm productivity [49,50]. Consequently, there is a research gap in linking the achievement of sustainable agricultural development goals—especially in the areas of income growth and productivity improvement—not only to farm production outcomes but also to household characteristics. This study contributes to that discussion. By using unique survey data, we linked increases in income and productivity with Farm Characteristics, Production-Related Characteristics, Farm Asset-Related Characteristics, and Household Characteristics for farms in Central Pomerania. To address this gap, the objective of this study is to assess the factors determining the achievement of the Sustainable Development Goals (SDGs) in the context of improving the productivity of agricultural inputs—land, labor, and capital—and increasing farm income, using logistic regression.

The remainder of the paper is structured as follows: Section 2 outlines the research methodology and data sources. Section 3 presents the findings of the empirical analysis. The final section provides a discussion of the main results, draws conclusions, and identifies potential directions for future research.

2. Materials and Methods

The analysis was based on data from a pilot study conducted in 2023 using the CATI (Computer-Assisted Telephone Interview) method with a standardized questionnaire [51]. The sample comprised 150 farms located in Central Pomerania, Poland. Respondents were selected according to the following criteria: (1) an individual farm managed by a natural person, (2) at least 1 hectare of agricultural land, and (3) the farm’s location within Central Pomerania. The study was exploratory in nature. The sample was not representative of the population of agricultural holdings in either Poland or the EU. Central Pomerania has certain specific agricultural characteristics [52]. Further research using a representative sample is therefore needed.

The research was conducted in four stages. In the first stage, factors determining an increase in farm income were assessed. The second stage examined improvements in land productivity, the third evaluated gains in labor productivity, and the fourth analyzed an increase in capital productivity. The Automated Statistical Description System was used to perform the exploratory analysis [53]. The objective was to identify qualitative variables (using Chi-square tests) and quantitative variables (using Mann–Whitney U tests) that significantly differentiated or were related to the selected dependent variables: INCR_INCOME, INCR_LAND_PROD, INCR_LAB_PROD, and INCR_CAP_PROD. These variables represent, respectively: an increase in farm income, an increase in production productivity, an increase in labor productivity, and an increase in capital productivity.

As a result of the tests, independent variables showing statistically significant relationships with at least one of the dependent variables were identified. This set included variables related to farm characteristics (LOC, ECO, EMP, AGRILAND, OTHER_LAND); features related to the production process (PROF (NOT PROFITABLE), PROF (LOW), PROF (MEDIUM), PROF (HIGH), PROD_STOCK, SALE_UNOFFICIAL, SALE_MARKET, RED_COMP); characteristics related to farm assets (BUILD_DECOM_PLAN, CONV_PLAN, SALE_LAND_COMP, BUILD_RENT_COMP, ASSETS_STOCK); and household characteristics (GEND (MEN), CAR_INS, DPH, CHANGE_EXP (INCREASE), CHANGE_EXP (DECREASE), EXP_HOUSE_ENERG, EXP_FOOD, EXP_HEALTH, EXP_REC_CULT, INC_RENT, DP_HOUS_EQUIP). Descriptions of all independent and dependent variables are provided in

Table 1. Catalogue of variables.

Variable Symbol	Variable Name	Category Explanation
Dependent Variables
INCR_INCOME	Increase in income	A dichotomous variable equal to 1 if total household income increased between 2018 and 2022, and 0 if it decreased or remained unchanged.
INCR_LAND_PROD	Increase in land productivity	A dichotomous variable equal to 1 if land productivity (production value per unit of farm area) increased between 2018 and 2022, and 0 if it decreased or remained unchanged.
INCR_LAB_PROD	Increase in labor productivity	A dichotomous variable equal to 1 if labor productivity (production value per number of people working on the farm) increased between 2018 and 2022, and 0 if it decreased or remained unchanged.
INCR_CAP_PROD	Increase in capital productivity	A dichotomous variable equal to 1 if capital productivity (production value per total farm asset value) increased between 2018 and 2022, and 0 if it decreased or remained unchanged.
Independent variables
Farm Characteristics
LOC	Location	Location of the farm (Pomeranian Voivodeship or West Pomeranian Voivodeship).
ECO	Organic farm	A dichotomous variable equal to 1 if the farm is organic, and 0 otherwise.
EMP	Number of employed persons	Number of persons employed on the farm, including the owner if they work on the farm.
AGRILAND	Area of agricultural land	Area of agricultural land [ha].
OTHER_LAND	Area of other land	Area of other land (excluding agricultural land, forests, and forested land) [ha].
Production-Related Characteristics
PROF (NOT PROFITABLE)	Not profitable agricultural production	A dichotomous variable equal to 1 if the farm recorded unprofitable agricultural production, and 0 if it achieved low, medium, or high profitability of agricultural production. The response reflects the respondent’s declaration regarding the profitability of agricultural production.
PROF (LOW)	Low profitability of agricultural production	A dichotomous variable equal to 1 if the farm recorded low profitability in agricultural production, and 0 if it achieved unprofitable, medium, or high profitability. The response reflects the respondent’s declaration regarding the profitability of agricultural production.
PROF (MEDIUM)	Medium profitability of agricultural production	A dichotomous variable equal to 1 if the farm recorded medium profitability in agricultural production, and 0 if it achieved unprofitable, low, or high profitability. The response reflects the respondent’s declaration regarding the profitability of agricultural production.
PROF (HIGH)	High profitability of agricultural production	A dichotomous variable equal to 1 if the farm recorded high profitability in agricultural production, and 0 if it achieved unprofitable, low, or medium profitability. The response reflects the respondent’s declaration regarding the profitability of agricultural production.
PROD_STOCK	Share of stock-designated products in total production value	Share of stock-designated products in total production value [in %].
SALE_UNOFFICIAL	Share of informal market sales in the total production value	Share of informal market sales in total production value [in %].
SALE_MARKET	Share of official market sales in the total production value	Share of official market sales in total production value [in %].
RED_COMP	Significant reduction in agricultural production	A dichotomous variable equal to 1 if the farm significantly reduced agricultural production over the past five years, and 0 otherwise.
Farm Asset-Related Characteristics
BUILD_DECOM_PLAN	Planned decommissioning of buildings	A dichotomous variable equal to 1 if the farm plans to demolish buildings previously used for agricultural production within the next five years, and 0 otherwise.
CONV_PLAN	Planned conversion of agricultural land into building plots	A dichotomous variable equal to 1 if the farm plans to convert agricultural land into building plots within the next five years, and 0 otherwise.
SALE_LAND_COMP	Agricultural land sale	A dichotomous variable equal to 1 if the farm sold agricultural land between 2018 and 2022, and 0 otherwise.
BUILD_RENT_COMP	Building rental	A dichotomous variable equal to 1 if the farm rented out buildings previously used for agricultural production between 2018 and 2022, and 0 otherwise.
ASSETS_STOCK	Share of agricultural product stocks in the farm’s total assets	Share of agricultural product stocks in the farm’s total assets [in %].
Household Characteristics
GEND (MEN)	Gender	A dichotomous variable equal to 1 if the farm is managed by a male, and 0 otherwise.
CAR_INS	Use of car insurance	A dichotomous variable equal to 1 if the household used car insurance (as a financial service) for at least one car owned by the household, and 0 otherwise.
DPH	Household life cycle stage	Household life cycle stage: single young person household; young couple/partnership without children; couple/partnership with preschool and/or school-aged children; middle-aged couple/partnership with adolescent or independent children; multigenerational household; couple/partnership after children have left the household; middle-aged single-person household; elderly single-person household.
CHANGE_EXP (INCREASE)	Change in household expenditures (steady increase)	A dichotomous variable equal to 1 if total household expenditures at the farm continuously increased between 2018 and 2022, and 0 otherwise.
CHANGE_EXP (DECREASE)	Change in household expenditures (steady decrease)	A dichotomous variable equal to 1 if total household expenditures at the farm continuously decreased over the past five years, and 0 otherwise.
EXP_HOUSE_ENERG	Share of housing and energy expenditures in total household expenditures	Share of housing and energy expenditures in total household expenditures [in %].
EXP_FOOD	Share of food and non-alcoholic beverages expenditures in total household expenditures	Share of food and non-alcoholic beverage expenditures in total household expenditures [in %].
EXP_HEALTH	Share of health expenditures in total household expenditures	Share of health expenditures in total household expenditures [in %].
EXP_REC_CULT	Share of recreation and culture expenditures in total household expenditures	Share of recreation and culture expenditures in total household expenditures [in %].
INC_RENT	Share of pension income in total household income	Share of pension income in total household income [in %].
DP_HOUS_EQUIP	Use of direct payments for household equipment	A dichotomous variable equal to 1 if the farm allocated direct payments to finance household equipment expenditures, and 0 otherwise.

Source: own study.

. The selection of variables for the analysis was based on a review of the literature [5,7,25,26,27,28,29,30,31,32,33,34,35,36] and the authors’ previous research [54,55,56]. It is important to acknowledge the limitations of the survey-based research: all responses provided by the participants were declarative in nature, i.e., self-reported. Collected data referring to the situation of both the farms and the farmers’ households in 2022 [51]. The dependent variables represent growth in specific values and therefore refer to the five-year period preceding the survey year (2018–2022). This time frame makes it possible to assess consistent increases or decreases in income, as well as in the productivity of land, labor, and capital. A similar approach applies to the independent variables CHANGE_EXP (INCREASE) and CHANGE_EXP (DECREASE), which capture changes in household expenditures. In addition, the five-year period covers variables that capture completed divestment processes (SALE_LAND_COMP; BUILD_RENT_COMP) as well as planned ones (BUILD_DECOM_PLAN; CONV_PLAN). These processes typically represent long-term changes. All other variables refer to the year 2022.

To identify the strongest predictors for each of the dependent variables, an automatic predictor selection procedure was applied using the stepAIC algorithm from the “MASS” package [57] in the R environment [58]. Predictor selection was performed using the forward selection method, which involves iteratively adds to the model the predictor that minimizes the value of the Akaike Information Criterion (AIC)—a statistic that evaluates model fit while penalizing excessive complexity. A lower AIC value indicates a more parsimonious model. The procedure began with a model containing only the intercept and systematically tested all possible models with individual variables, adding predictors one by one until the model with the lowest AIC value was obtained. This process was carried out separately for each of the four dependent variables (INCR_INCOME, INCR_LAND_PROD, INCR_LAB_PROD, and INCR_CAP_PROD) using logistic regression.

Model diagnostics were performed to evaluate multicollinearity, the distribution of deviance residuals, and the agreement between observed and model-predicted values of the dependent variable. Multicollinearity analysis using the vif() function from the car package [59] for predictors in individual models indicated no evidence of multicollinearity (VIF < 1.28). Hosmer and Lemeshow goodness-of-fit tests, conducted with the hoslem.test() function from the ResourceSelection package [60], showed that in each case the observed values of the dependent variable closely matched the model-predicted values, with no significant differences detected. In addition, visual inspection of histograms of deviance residuals indicated that the residuals remained within ±3 standard deviations, meeting the assumption of acceptable deviation from model predictions. To evaluate the predictive performance of the logistic regression model, we applied a 10-fold cross-validation using the caret package in R (version 4.4.2) [61]. The dataset was randomly partitioned into 10 equal-sized folds, while preserving the class distribution (stratified sampling). In each iteration, the model was trained on nine folds and tested on the remaining fold. To address potential class imbalance, up-sampling of the minority class was performed within each resampling iteration. Model performance was assessed using the area under the ROC curve (AUC), sensitivity, and specificity, with AUC used as the primary optimization metric. The cross-validation results indicated that the analysis of out-of-sample folds exhibited moderate to high classification ability, both overall and for the individual classes coded as 1 (“yes”) and 0 (“no”). The coefficient assessing the overall classification accuracy (AUC) ranged between 0.77 and 0.86, the sensitivity coefficient ranged between 0.72 and 0.78, and the specificity coefficient ranged between 0.62 and 0.85. Detailed diagnostic results are provided in Appendix A.

3. Results

All farms included in the survey sold their products on the market, classifying them as commercial farms. The dominant production profile was crop specialization, reported by 85.33% of respondents. The surveyed farms showed significant variation in annual production value: 16.67% reported values below PLN 32,000; 38.00% were in the range of PLN 32,001–100,000; 22.00% reported between PLN 100,001–200,000; and 23.33% achieved values exceeding PLN 200,001. The financial structure indicates that, on average, 74.67% of farm operations were financed with own capital, with the remaining share covered by external sources. Fixed assets accounted for more than 86% of total assets, suggesting a strong capital base among the surveyed farms. The analysis of household structures showed that the dominant groups were young farmers in marriages or partnerships without children (38.00%) and young couples with preschool and/or school-aged children (24.67%). Monthly per capita household income was below PLN 1000 for 2.67% of respondents, between PLN 1001–1500 for 48.67%, between PLN 1501–2000 for 29.33%, and above PLN 2001 for 19.33% of respondents.

The first stage of the study examined factors associated with income growth in the surveyed farms. The results of the logistic regression analysis are presented in

Table 2. The impact of the identified predictors on the occurrence of the INCR_INCOME phenomenon.

INCR_INCOME
Predictors	Odds Ratios	Std. Error	p
(Intercept)	2.365	2.611	0.436
PROF [NOT PROFITABLE]	0.000	0.000	0.995
PROF [LOW]	0.368	0.383	0.336
PROF [MEDIUM]	0.067	0.077	0.019
PROF [HIGH]	0.000	0.000	0.996
DP_HOUS_EQUIP	0.816	0.088	0.061
BUILD_RENT_COMP	0.000	0.000	0.996
SALE_LAND_COMP	0.270	0.219	0.107
CAR_INS	0.267	0.240	0.142
Observations	150
R2 Tjur	0.191
Deviance	56.204
AIC	74.204
AICc	75.490
log-Likelihood	−28.102

Source: own study.

.

The analysis of the estimated odds ratios (OR) and their significance for the occurrence of income growth (INCR_INCOME) revealed that medium agricultural production profitability (PROF_MEDIUM) reduced the likelihood of income growth by 93.3%, which is statistically significant. Moreover, allocating direct payments to household equipment expenditures (DP_HOUS_EQUIP) was linked to an approximately 18.4% lower likelihood of income growth, with the p-value at the threshold of statistical significance (p = 0.061). No other statistically significant effects were observed. The total variance explained by the model was R² = 0.191, indicating that the model accounts for 19.1% of the variability in the INCR_INCOME variable.

The second stage of the study examined factors associated with an increase in land productivity (INCR_LAND_PROD). The results of this analysis are presented in

Table 3. The impact of the identified predictors on the occurrence of the INCR_LAND_PROD phenomenon.

INCR_LAND_PROD
Predictors	Odds Ratios	Std. Error	p
(Intercept)	0.804	0.870	0.840
AGRILAND	1.090	0.054	0.079
GEND [MAN]	4.221	2.582	0.019
EXP_HOUSE_ENERG	0.908	0.044	0.046
SALE_LAND_COMP	0.386	0.182	0.043
PROD_STOCK	0.762	0.074	0.005
CAR_INS	0.350	0.188	0.050
ASSETS_STOCK	0.957	0.024	0.076
BUILD_RENT_COMP	3.914	2.961	0.071
BUILD_DECOM_PLAN	2.325	1.214	0.106
SALE_UNOFFICIAL	0.443	0.252	0.152
Observations	150
R2 Tjur	0.276
Deviance	127.505
AIC	149.505
AICc	151.418
log-Likelihood	−63.753

Source: own study.

.

The analysis of the odds ratios (OR) and their statistical significance for the dependent variable INCR_LAND_PROD (increase in land productivity) revealed several significant and borderline associations. An increase in the area of agricultural land (AGRILAND, OR = 1.090, p = 0.079) was linked to approximately a 9% higher likelihood of increased land productivity; this effect was at the threshold of statistical significance. Male gender (GEND_MEN) was associated with more than a fourfold higher likelihood of improved land productivity, with the result being statistically significant (p < 0.05). In contrast, a higher share of household expenditures on housing and energy (EXP_HOUSE_ENERG) corresponded to an approximately 9.2% lower likelihood of increased land productivity, which was statistically significant. The sale of agricultural land between 2018 and 2022 (SALE_LAND_COMP) reduces the likelihood of increased productivity by about 61.4%, which was also statistically significant. A significant negative effect was observed for the share of production allocated to on-farm reserves (PROD_STOCK), reducing the likelihood of increased land productivity by approximately 23.8%. Having car insurance (CAR_INS) was related to a 65% lower likelihood of increased productivity; however, this result is exactly at the threshold of statistical significance. In addition, the variable related to the share of agricultural product stocks in total assets (ASSETS_STOCK) showed no significant effect on the outcome, although the result fell within the range of a statistical tendency. Similarly, plans to rent out buildings previously used for agricultural production (BUILD_RENT_COMP) were associated with an almost fourfold higher likelihood of increased land productivity, but this result does not reach statistical significance, although it approaches the threshold of a statistical tendency. No other statistically significant effects were identified. The coefficient of determination for the model was R² = 0.276, indicating that the model explained 27.6% of the variability in the INCR_LAND_PROD variable.

The third stage of the study examined factors associated with an increase in labor productivity. The results of this analysis are presented in

Table 4. The impact of the identified predictors on the occurrence of the INCR_LAB_PROD phenomenon.

INCR_LAB_PROD
Predictors	Odds Ratios	Std. Error	p
(Intercept)	3.368	1.706	0.017
CHANGE_EXP [INCREASE]	0.410	0.181	0.044
CHANGE_EXP [DECREASE]	0.050	0.044	0.001
BUILD_DECOM_PLAN	0.461	0.240	0.138
INC_RENT	1.082	0.044	0.052
DP_HOUS_EQUIP	0.905	0.040	0.025
ASSET_ STOCK	1.028	0.019	0.121
CONV-PLAN	0.507	0.233	0.139
Observations	150
R2 Tjur	0.229
Deviance	169.518
AIC	185.518
AICc	186.539
log-Likelihood	−84.759

Source: own study.

.

The analysis of odds ratios (OR) and their statistical significance for the dependent variable INCR_LAB_PROD (increase in labor productivity) revealed several significant relationships. Continuous growth in household expenditures (CHANGE_EXP_INCREASE), led to approximately a 59% lower likelihood of increased labor productivity. This effect was statistically significant. An even stronger effect was observed for continuously decreasing expenditures (CHANGE_EXP_DECREASE, OR = 0.050, p = 0.001), reducing the likelihood of the outcome by about 95%, representing a highly statistically significant result. The share of pension income in total household income (INC_RENT) was linked to a slight rise in the likelihood of higher labor productivity—approximately 8.2%—with the effect being at the threshold of statistical significance. In contrast, an increase in the allocation of direct payments to household equipment expenditures (DP_HOUS_EQUIP) corresponded to a slight but statistically significant decrease in the likelihood of increased labor productivity—by approximately 9.5%. No other statistically significant relationships were observed. The coefficient of determination for the model was R² = 0.229, indicating that the model explained 22.9% of the variability in the INCR_LAB_PROD variable.

In the final stage of the study, the factors associated with an increase in capital productivity of the surveyed farms were examined. The results of this analysis are presented in

Table 5. The impact of the identified predictors on the occurrence of the INCR_CAP_PROD phenomenon.

INCR_CAP_PROD
Predictors	Odds Ratios	Std. Error	p
(Intercept)	1,621,408.400	12,641,329.453	0.067
EXP_REC_CULT	1.301	0.108	0.001
EXP_HEALTH	2.056	0.341	<0.001
CHANGE_EXP [INCREASE]	1.337	0.797	0.626
CHANGE_EXP [DECREASE]	0.011	0.019	0.013
PROF [NOT PROFITABLE]	0.405	0.609	0.548
PROF [LOW]	0.273	0.369	0.337
PROF [MEDIUM]	8.164	10.631	0.107
PROF [HIGH]	1.908	2.819	0.662
BUILD_RENT_COMP	4.335	3.887	0.102
SALE_MARKET	0.772	0.069	0.004
EXP_HOUSE_ENERG	1.131	0.058	0.016
CONV_PLAN	4.830	2.944	0.010
ASSETS_STOCK	1.042	0.029	0.135
Observations	150
R2 Tjur	0.418
Deviance	95.238
AIC	123.238
AICc	126.349
log-Likelihood	−47.619

Source: own study.

.

The analysis of odds ratios (OR) and their statistical significance for the dependent variable INCR_CAP_PROD (increase in capital productivity) revealed several significant relationships. An increase in the share of household expenditures on recreation and culture (EXP_REC_CULT) increased the likelihood of higher capital productivity by approximately 30.1%. This effect was highly statistically significant. An even stronger effect was observed for the share of household expenditures on health (EXP_HEALTH), more than doubling the likelihood of the outcome. In contrast, a continuous decrease in household expenditures (CHANGE_EXP_DECREASE) reduced the likelihood of increased capital productivity, indicating a strong negative effect. Furthermore, a higher share of official market sales in the total production value (SALE_MARKET) corresponded to a 22.8% lower likelihood of increased capital productivity, while an increase in the share of expenditures on housing and energy (EXP_HOUSE_ENERG) was related to a 13.1% higher likelihood. Additionally, farms planning to convert agricultural land into building plots within the next five years (CONV_PLAN), had more than a fourfold higher likelihood of increased capital productivity, which was also statistically significant. No other statistically significant effects were observed. The coefficient of determination for the model was R² = 0.418, indicating that the model explained 41.8% of the variability in the INCR_CAP_PROD variable.

The highest R² value was observed for the model describing the increase in capital productivity (0.418), indicating a very good fit. In contrast, the lowest fit was found for the model identifying factors associated with farm income growth. This likely reflects the stronger explanatory power of asset-related and production-related predictors, which are more directly aligned with the measurement of capital productivity. Given the exploratory nature of this pilot study, this difference should be interpreted with caution, as it may also partly result from sample-specific characteristics. These findings underscore the need for further research, particularly to explore in greater depth the relationship between household characteristics and income levels, as well as to incorporate macroeconomic factors.

4. Discussion and Conclusions

This study aimed to assess the factors determining the achievement of the Sustainable Development Goals (SDGs) in the context of improving the productivity of agricultural inputs—land, labor, and capital—and increasing farm income, using logistic regression. The analysis was based on primary data collected during a pilot study conducted in 2023 using the CATI method and a standardized questionnaire, from a sample of 150 farms in the Central Pomerania region of Poland.

The obtained results indicate the need for an interdisciplinary approach to the sustainable development of agricultural holdings [62,63]. The multidimensional nature of the SDGs in the agricultural context also underscores the importance of integrating endogenous economic conditions for achieving the SDGs with exogenous factors (e.g., local development policies, direct payment schemes), as well as environmental aspects (such as climate risk) and soil cultivation techniques, including fertilization. A broader, interdisciplinary perspective on these issues could also contribute to higher R² values in the estimated models. This study placed particular emphasis on factors associated with farm income growth and increases in the productivity of land, labor, and capital. To this end, logistic regression was applied to test 28 independent variables grouped into four categories: farm characteristics, production-related characteristics, farm asset-related characteristics, and household characteristics.

Table 6. Mapping of significant predictors from survey-based logistic regression models to relevant SDG Targets with Supporting Evidence.

Dependent Variables	Factor	Independent Variables	Direction of Effect	Specific SDG Targets	Evidence from Analysis
INCR_INCOME	Production-Related Characteristics	PROF [MEDIUM]	-	Target 1.1 of SDG 1 Target 1.2 of SDG 1 Target 2.3 of SDG 2 Target 10.1 of SDG 10	Medium agricultural production profitability was associated with a lower likelihood of income growth.
INCR_LAND_PROD	Production-Related Characteristics	PROD_STOCK	-	Target 2.3 of SDG 2 Target 2.4 of SDG 2 Target 2.a of SDG 2	The share of production allocated to on-farm reserves was associated with a lower likelihood of increased land productivity.
	Farm Asset-Related Characteristics	SALE_LAND_COMP	-		The sale of agricultural land between 2018 and 2022 was associated with a lower likelihood of increased land productivity.
	Household Characteristics	GEND [MAN]	+		Being male was associated with a more than fourfold higher likelihood of improved land productivity.
		EXP_HOUSE_ENERG	-		Continuous growth in household expenditures was associated with a lower likelihood of increased labor productivity.
INCR_LAB_PROD	Household Characteristics	CHANGE_EXP [INCREASE]	-	Target 2.3 of SDG 2 Target 8.2 of SDG 8	Continuous increase in household expenditures was associated with a lower likelihood of increased labor productivity.
		CHANGE_EXP [DECREASE]	-		Continuous decrease in household expenditures was associated with a lower likelihood of increased labor productivity.
		DP_HOUS_EQUIP	-		An increase in the allocation of direct payments to household equipment expenditures was associated with a lower likelihood of increased labor productivity.
INCR_CAP_PROD	Household Characteristics	EXP_REC_CULT	+	Target 2.3 of SDG 2 Target 2.a of SDG 2 Target 9.3 of SDG 9 Target 9.4 of SDG 9	An increase in the share of household expenditures on recreation and culture was associated with a higher likelihood of increased capital productivity.
		EXP_HEALTH	+		An increase in the share of household expenditures on health was associated with a higher likelihood of increased capital productivity.
		CHANGE_EXP [DECREASE]	-		A continuous decrease in household expenditures was associated with a lower likelihood of increased capital productivity.
		EXP_HOUSE_ENERG	+		An increase in the share of expenditures on housing and energy was associated with a higher likelihood of capital productivity growth.
	Production-Related Characteristics	SALE_MARKET	-		A higher share of official market sales in the total production value was associated with a lower likelihood of increased capital productivity.
	Farm Asset-Related Characteristics	CONV_PLAN	+		Farms planning to convert agricultural land into building plots within the next five years were associated with a more than fourfold higher likelihood of increased capital productivity.

Source: own study.

presents the results of the logistic regression models in the context of achieving the SDGs.

In the first stage, factors associated with farm income growth were examined. Two variables showed a negative association: medium agricultural production profitability and the allocation of direct payments to household equipment expenditures. The former reflects production-related characteristics, and previous studies have confirmed that the value of agricultural production is an important determinant of farm income [7,29,36]. The latter represents household characteristics. Although direct payments are intended to support farm income, their allocation to household equipment may be linked to reduced operational development in the surveyed farms. The value of production may therefore be associated with income growth and, consequently, with poverty reduction and the doubling of incomes for small-scale food producers (target 1.1 of SDG 1; target 1.2 of SDG 1; target 2.3 of SDG 2; target 10.1 of SDG 10).

In the second stage of the study, variables associated with increase in land productivity were identified. Among the variables reflecting farm characteristics, the area of agricultural land was statistically significant. A larger surveyed farm size was associated with higher production levels and appeared to be linked to greater productivity [64]. Among the production-related characteristics, a statistically significant association was found for the share of production allocated to on-farm reserves in total production value. This may reflect the fact that a large proportion of the analyzed farms in Poland operate with a mixed production profile [65]. Farm asset-related characteristics also showed associations with land productivity, specifically the planned rental of buildings previously used for agricultural production within the next five years, and the sale of agricultural land between 2018 and 2022. These findings underline the relevance of assessing changes in agriculture over the long term. Household characteristics showing statistically significant relationships with land productivity included being male, the share of household expenditures on housing and energy in total expenditures, and the use of car insurance by the household. This is consistent with studies presented in the literature highlighting the strong interconnection between the farming household and the agricultural holding [66]. Our findings highlight the importance of considering farmers’ household characteristics, as well as investment and divestment activities, in the context of achieving the Sustainable Development Goals, particularly target 2.3 of SDG 2, target 2.4 of SDG 2, and target 2.a of SDG 2.

The third stage of the study examined the relationships between the identified factors increases in labor productivity. In this case, the significant role of the farming household in the development of the surveyed agricultural holding was particularly evident. Only variables related to household characteristics were statistically significant: continuous increases in household expenditures, continuous decreases in household expenditures, the share of pension income in total household income, and the allocation of direct payments to household equipment expenditures. Consistent with the findings for land productivity, these results point to a strong association between agricultural activity and the economic status of the surveyed households. An increase in labor productivity in agriculture contributes to the achievement of target 2.3 of SDG 2 and target 8.2 of SDG 8. Similar patterns were observed in the analysis of factors related to capital productivity. Statistically significant household characteristics included the share of household expenditures on recreation and culture, the share of expenditures on health, continuous decreases in household expenditures, and the increase in the share of expenditures on housing and energy. In addition, capital productivity of the surveyed farms showed associations with production-related characteristics (the share of official market sales in total production value) and farm asset-related characteristics (plans to convert agricultural land into building plots within the next five years). An increase in capital productivity contributes to the achievement of target 2.3 of SDG 2, target 2.a of SDG 2, target 9.3 of SDG 9, and target 9.4 of SDG 9. It is worth noting that the likelihood of increased capital productivity among the surveyed farms was 4.83 times higher for those planning to convert agricultural land into building plots, compared with those without such plans (OR = 4.830), holding other factors constant. This association may be related to greater entrepreneurial orientation among farm managers, which aligns with SDG target 2.3 (increasing the productivity and incomes of small-scale producers). The higher market value of building plots relative to agricultural land may raise farm capital productivity, although this effect is likely to be short-term. In the long run, however, the conversion of agricultural land could reduce the productive potential of agriculture, including at the global scale, potentially hindering the achievement of other SDG targets. Further research is therefore needed to examine how to balance the individual economic benefits for farms with the broader global costs of converting agricultural land into building plots.

The results align with the existing body of research on sustainable development, particularly emphasizing the need to integrate its economic dimension with social aspects. The fact that the analyzed dependent variables were associated with factors reflecting the characteristics of agricultural holdings may indicate a potential shift of financial resources from farms toward non-productive purposes, specifically household consumption. On the other hand, it also underlines the possible importance of investing in human capital. Another relevant aspect concerns the stability of farm operations. The findings suggest that sustainable agricultural development may benefit from not only stabilizing farm income but also ensuring greater predictability in household expenditures. Both increases and decreases in household expenditures were found to be associated with changes in labor productivity in agricultural holdings. Sustainable development depends on the long-term viability of agricultural activities, without constant adjustments in farm operations or household economic conditions. A further issue involves the conversion of agricultural land into building plots. Sustainable agricultural development may require optimizing land protection by maintaining or reconsidering the agricultural use of farmland.

Given the cross-sectional and observational nature of the data, these associations do not imply causation, and unobserved confounding factors may influence the observed relationships.

The findings provide valuable insights into the factors associated with the income and productivity of the surveyed farms in the context of achieving sustainable development goals. Nevertheless, several limitations should be acknowledged. First, the analysis is based on data obtained from a specific sample of farms, which limits the generalizability of the results to the broader farming population. The study is exploratory in nature, and a follow-up investigation using a representative sample is planned. Moreover, the research relied on a structured interview questionnaire, and the use of self-reported data may introduce potential biases including incomplete understanding of questions, the tendency to provide socially desirable answers, and recall bias, particularly for variables referring to past events, activities, or expenditures. Another limitation of the study is the absence of objective performance measures, which may reduce the precision of the productivity and income indicators. These methodological constraints could influence the strength and direction of the observed associations; therefore, the results should be interpreted with caution. Additionally, some extreme odds ratio values observed in the models likely stem from small sample sizes within certain categorical predictor groups. While recoding these categories could mitigate quasi-complete separation, doing so substantially diminished the statistical significance of key predictors and the overall explanatory capacity of the models. For this reason, the original coding scheme was retained. Furthermore, given the exploratory nature of this pilot study and the limited sample size, no correction for multiple comparisons (e.g., Bonferroni or FDR) was applied. Consequently, the risk of Type I errors (false positives) is elevated, and borderline p-values should be interpreted as indicative rather than conclusive.

Nevertheless, it is important to highlight the advantages of conducting survey-based research among farmers. Farmers represent a socio-economic group that makes distinct financial and investment decisions [67]. Survey research provides direct access to farmers’ opinions and experiences, enabling the collection of data on various aspects of farm functioning—particularly the interactions between the farm and the farmer’s household. The questionnaire used in this study was developed based on previous research experience [54,55,56,68,69,70,71,72,73,74] as well as studies using unique data collected within the Farm Sustainability Data Network [26,75,76,77]. This approach offers a comprehensive perspective on the economic conditions underlying the achievement of SDGs in agriculture.

The findings of this study align with the objectives set out in major EU policies on sustainable agriculture. The Common Agricultural Policy (CAP) for 2023–2027 places strong emphasis on enhancing farm productivity and resilience, supporting farmers’ incomes, protecting natural resources, and fostering rural development [78]. These findings also align with the goals of the European Green Deal and the Farm to Fork Strategy, both of which advocate for the sustainable transformation of food systems towards climate neutrality, public health, and economic sustainability [79,80]. The observed associations between household characteristics and farm productivity and income underscore the importance of integrating socio-economic factors into rural development policy frameworks.

This study contributes both to practical applications—particularly in agricultural policy and advisory services—and to the scientific literature on sustainable farm development and the determinants of income generation in farming households. The development of the surveyed farms, in terms of income growth and the efficiency of land, labor, and capital utilization, depends on the scale of available resources, their effective use, the stability of the economic situation, and human capital, which often originates from the farming household. The findings may serve as an important source of information for institutions involved in agricultural advisory services. They indicate the need for systematic education of farm owners on the conditions required for sustainable agricultural development. Such training should address not only modern agricultural technologies but also financial and investment decision-making within farms. The results further contribute to the ongoing research on sustainable agricultural development [26,76] and to studies on financial and investment decisions made by farms and farming households [55,56,77]. Moreover, they underscore the importance of linking these two research areas, which represents a promising direction for further investigation. There is also a clear need to examine investment processes related to sustainable development and their sources of financing.