Determinants of the profitability of poultry farming in the Gazipur district of Bangladesh

Abstract

Nowadays, smallholder poultry producers in Bangladesh face substantial challenges in operating their businesses profitably. This study looked at the factors that influence the profitability of poultry farming in Bangladesh's Gazipur area, employing primary data collected from 200 poultry farmers with 100 broiler and 100-layer producers. Econometric models were used to analyze the survey data. The findings demonstrate that the economic viability of poultry farming in the study areas was subject to various socioeconomic factors. The estimated parameters that significantly and positively influenced the profitability were the cost of hired labor and veterinary services with the coefficient value of 0.16 and 1.35, respectively. Conversely, the profitability exhibited an inverse relationship with education level, family size, costs of tools and equipment, and day-old chicks, with parameter estimates of 0.85, 1.73, 1.07, and 1.47, respectively. As the cost components in this study are found to have a dubious effect on profitability, specific cost-reduction strategies are required at both national and local levels to assist farmers in adhering to their budgetary constraints and enhancing their net farm profit. In addition, innovation in the poultry sector should be made readily available at an affordable cost, and factors that impede profitability, such as large family size and higher education, should be converted to assets through adequate training.

1. Introduction

Bangladesh is predominantly an agrarian country. Hence, the expansion of the livestock and poultry sub-sector is regarded as a dynamic element in alleviating rural poverty by generating employment and income for the extensive rural populace. Over 150,000 commercial poultry farms in Bangladesh employ 6 to 8 million individuals, predominantly unemployed youth and women [1].

Apart from generating employment and providing livelihood opportunities, this sub-sector of agriculture also serves as a good source of protein and nutrition [2,3]. It accounts for 1.85 % of GDP [4] and 35.25 % of the annual meat supply [5]. Since the 1990s, this subsector grew at 2.8 % annually [6]. Poultry meat and eggs are the most affordable sources of animal protein, and all social, religious, and economic groups widely embrace them. Reports indicated that per capita consumption of poultry meat and eggs will rise by 26 % and 41 %, respectively, over the next five years [5]. Consequently, the poultry business necessitates an investment of around 4117.65 million USD1 to meet the augmented demand [1].

In Bangladesh, animal-based protein consumption is inadequate compared to other regions globally. Approximately 70 % of the population in this country experiences malnutrition, and nearly 60 % of families fail to consume sufficient protein [5]. A recent joint assessment by the WHO and FAO indicates that per capita meat consumption in Bangladesh is approximately 15.23 kg annually, while the requirement is around 43.8 kg. This value indicates a 65.23 % annual deficit between demand and supply, necessitating a 34 % increase in broiler production and a 49 % increase in layer production, respectively [7].

Over the past two to three decades, urbanization, changes in socioeconomic status, and a heightened focus on nutritional needs have enabled a notable shift from subsistence to commercial poultry farming in Bangladesh. The profession is no longer viewed exclusively as a means for the impoverished and is recognized as an essential tool for reducing rural-to-urban migration [8]. The growth of the poultry industry is often impeded by decreasing profitability, resulting from various issues that discourage many farmers, young individuals, and women from pursuing poultry farming as a feasible and sustainable source of income. The lack of affordable raw materials, expertise, effective management skills, access to loans, disease outbreaks, insufficient market structures, and farmers' inability to understand the potential of value chain operations are commonly responsible for their losses [5,6]. These factors are not unique to all farms and differ across geographical locations. This research investigates the relationship between profitability and the factors influencing its fluctuations in the study areas.

Research indicates that smallholder poultry producers encounter considerable difficulties in achieving farm profit. Consequently, many researchers have evaluated farm profitability levels through different metrics and recorded its determinants globally [[9], [10], [11]]. Most of this research was conducted in developing countries, primarily focusing on either layer [12] or broiler farms [[13], [14], [15], [16], [17], [18], [19], [20], [21]].

The majority of research conducted in Bangladesh has concentrated on broiler farming [5,6,8,[22], [23], [24], [25]], with only a few studies addressing both enterprises [26]. Besides examining the profitability scenario of poultry producers, researchers [27] also examined the profitability of poultry feed mills and their backward and forward linkages.

The literature review indicates that, although numerous studies have been conducted on the highlighted area both globally and in Bangladesh, their emphasis has primarily been restricted to descriptive assessments of profitability, such as benefit-cost and return (BCR) analyses, which have focused exclusively on farm-specific factors [5,6,[9], [10], [11],25,26]. The functional analysis of profitability, incorporating both farm-specific and farmers' socio-demographic characteristics, remains insufficiently addressed, highlighting a notable research gap. The studies employed small sample sizes, ranging from 30 to 180 samples, except for [20], which utilized secondary data sources from 1508 broiler farms. From that pursuit, this study may represent a comprehensive analysis of functional aspects, utilizing a substantial sample size of 300 from both broiler and layer farms. Analyzing the internal and external factors influencing the financial status of poultry producers is essential for informed decision-making, economic advancement, and developing sustainable strategies to enhance overall performance [28]. An in-depth analysis of margins and profitability will allow policymakers to determine the necessary interventions for smallholder poultry farmers in Bangladesh to improve outcomes.

The data for this study was collected before COVID-19, allowing for a comparison of the profitability of poultry farmers in the study regions before and after the pandemic. Profitability analysis is crucial from a policy standpoint due to its implications for the growth of the poultry sector. Robust expansion generally ensures economies of scale for poultry producers, even without adequate infrastructure and governmental support. This initiative may attract potential investors and emerging entrepreneurs to consider this industry a viable revenue source.

The remainder of the paper is structured as follows: Section 3 presents the detailed methodology of the study. The results and the associated discussions are the subject of Section 4, and Section 5 concludes with appropriate policy recommendations, followed by the study's limitations and the potential for future research.

2. Methodology of the study

2.1. Sampling technique and data collection

The data for this study was obtained via a household survey employing a scheduled interview schedule from July to August 2019, utilizing a simple random sampling method. This district was selected intentionally due to the presence of a substantial number of chicken farms and their associated enterprises. Approximately 680,782 households engage in poultry rearing alongside 1,309,652 poultry enterprises in this district [4,29,30]. The finite population correction method was employed to ascertain the sample size for this population, as referenced in Refs. [31,32]. The formula to find the appropriate sample size is-

$$n=\frac{N}{\left(1+N{e}^2\right)}$$

Where n = sample size, N = total number of households in the area, and e = desired margin of error. Using the above equation, the sample size was calculated at 100, considering a 10 % margin of the population. Based on that result, 200 interviews were recorded, including 100 broiler and 100-layer farmers from the Gazipur district of Bangladesh. To obtain the required information, this study adopted a semi-structured interview schedule, including both open- and closed-type questions pertaining to the production and profitability of the poultry sector in the study area. The final interview schedule was developed after a successful pre-testing of the interview schedule from 12 respondents and one key informant of the study area. A face-to-face interview was performed to gather the respondents' non-personal information. During interviews, participants have posed questions in multiple formats to evaluate the consistency of their answers. This endeavour aided in establishing a robust evaluation of the survey instrument's validity.

Data on the necessary factors were gathered to determine farmers’ socioeconomic traits and the profitability of raising chickens in the research area. These included details about the farm and the farmers themselves, such as age, education, family size, amount of land, and farming experience, as well as the various expenses incurred by the farmers in raising chickens, such as costs for tools and equipment, day-old chicks, feed, litter, labor hired, electricity, and transportation. These variables were selected after an exhaustive review of the literature.

In order to determine whether the return from farming had a more significant influence on profitability than the costs associated with farming, the net return from farming was also computed and added as a variable in the functional analysis. Moreover, an extended analysis of the poultry farmers’ socioeconomic characteristics, drawing comparisons with local and national data and various statistical measures, established the validity and reliability of the research data. Obtained data were then transferred to a spreadsheet for econometric analysis upon the multicollinearity assessment. No multicollinearity among the variables indicated minimum response bias in the data collection process.

The farmer or respondent was also called the farm owner or household head in this study. Therefore, the average costs incurred per farm indicates the cost incurred by each farm owner cum respondent cum farmer in this study. Data were analyzed using both simple descriptive as well as the most straightforward functional analysis, that is the multiple regression model as described in section 2.2.

2.2. Functional analysis

According to basic economic theory, a farm's profitability is determined by the number of farms and farmers' characteristics [20]. Therefore, to identify the factors that affected the profitability of the poultry farms studied, this study conducted multiple regression analysis, the simplest functional model of examining the connection by comparing two or more independent variables and a dependent variable. This model was chosen over all other methods due to its simplicity and potential to determine the factors while assuming static OLS assumptions.

The basic structural equation of this model has been adopted from [33, pp. 188–231] but is not elaborated in this study as it is not a unique analytical approach. It is presented as follows:

$$GrossMargin{\left(GM\right)}_j={\beta }_0+{\beta }_i{X}_i+{\mu }_j$$ (1)

Where,

GM = Average Gross margin (i.e., profitability) of $jth$ number of poultry farmers;

$X_i$ = Numbers of farms and farmers’ socioeconomic characteristics (examples include age, education, number of family members, expertise in farming, size of the farm, expenses for tools and equipment, chicklings, feed, litter, labor employed, veterinary care, and medications, electricity, transportation, and availability of credit);

${\beta }_0$ = Intercept;

${\beta }_i$ = Coefficients associated with the variables $X_i$;

${\mu }_j$ = Random error

$i$ = 1, 2, … … … … … … … … … … … … … … … … … ….15

$j$= 1, 2, … … … … … … … … … … … … … … … … … …0.200

Production factors are usually linearly related to the outcome of poultry production, while socioeconomic characteristics of producers and external factors are considered non-linear [33]. To address this issue, assuming that not every explanatory factor in equation (1) is linearly correlated with the dependent variable (i.e., GM), a natural logarithm is used on both sides of equation (1) to make it linear as shown in equation (2):

$${lnGM}_j={\beta }_0+{\beta }_i\mathit{ln}{X}_i+{\mu }_j$$ (2)

Finally, the results of this model estimation through a multiple regression approach have been obtained using the STATA 12 statistical package. As more than one explanatory variable was incorporated to assess the outcome of poultry production, the basic assumptions of multiple regression (no multicollinearity, homoscedasticity, normality) are fulfilled.

In order to validate the results obtained from OLS, several other analyses, such as Variable Selection Method (VSM), Principal Component Analysis (PCA) and Stepwise Regression (SR), were performed on the existing dataset. The structural equations of PCA and SR were adopted from Refs. [34,35], which are not elaborated in this study for convenience.

All possible variables related to poultry production and marketing were assessed, including socio-demographic factors, production factors, marketing-related factors, and accessibility-related factors. Additionally, the robustness of the dataset was checked by adding new variables, such as marketing cost (generated by adding the cost of hired labor and transportation, as these costs were very insignificant) and removing the initially estimated insignificant variables one by one, such as the cost of feed, litter, electricity, transportation, etc. In summary, the model was run separately using production-related, marketing-related, and sociodemographic factors to get accurate results. The aim was to check whether the estimated results were sensitive to the changes in the model specification as well as to provide the best policy recommendations generated from multiple analyses.

In addition to the primary data, supplementary data were collected from various published and unpublished reports, policy papers, articles, the Ministry of Fisheries and Livestock, the Bangladesh Livestock Research Institute, the Bangladesh Bureau of Statistics, and other sources to strengthen the study's findings.

2.3. Overview of the functional variables and assumptions on their impact

$X_1=$ Age of the farmers (years): Age refers to the respondent who owns and runs the poultry farm. It serves as a proxy of experience. Older farmers are assumed to have a negative association with the profitability of poultry farms in the sense that they are not adaptive to innovation and technology adoption and are not aware of better farming practices. On the contrary, young farmers are more likely to learn better management practices. Therefore, the possibility of earning profit is higher for them [20]. Both possibilities can be sustained through model estimation. Therefore, the estimated nature of the coefficient could be either positive or negative.

$X_2=$ Education of the farmers (years of schooling): An educated farmer is likely to better manage the farm due to enhanced decision-making capability and earn a higher profit than an uneducated farmer [20]. Educated farmers may undertake poultry farming as an entrepreneurial initiative to prove their skills and capabilities of exploring new ideas and applying them to better management of their farms. Keeping this view in mind, a higher level of education is expected to be estimated with a positive sign.

$X_3=$ Family size of the farmers (number): Farming is generally regarded as a labor-intensive occupation [12]. Therefore, a large family size is considered advantageous because their involvement will require less hired labour, thereby less cost, and vice versa.

$X_4=$ Farming experience (years): In line with the assumption of the educational level, a farmer with extensive experience in farming is assumed to understand better and manage farming operations effectively and efficiently, which helps to generate a positive return [20]. In light of this, the model analysis anticipates that the anticipated sign of this variable will be positive.

$X_5=$ Landholding size of the farmers (acre): Large land holding size generally acts as a significant indicator of better financial performance of a farm. It also acts as a proxy for higher social status in rural areas. A large land size is also expected to achieve economies of scale from a large volume of business and less cost per unit of production [20]. On the other hand, a small land size indicates less economic viability for operating the farm. Therefore, this variable is hypothesized to be positively or negatively related to the farm's profitability.

$X_{6-13}=$ Veterinary services, electricity, day-old chicks, feed, litter, labor costs, tools, equipment, and maintenance costs (BDT): These items indicate the raw materials required to raise poultry on a farm. The costs incurred per farm imply the costs incurred by the farm owners as well as the respondents in running their poultry business. The supply of raw materials and their price significantly hampers the production process. Different authors identified different cost items as the significant cost of poultry production. For example [20], noted that feed is one of the primary expenses, accounting for as much as 70 % of the overall production cost. In contrast [12,21], recognized the cost of labor, veterinary services, tools, and equipment as vital cost items along with feed cost [34]. identified the primary cost components that might significantly lower the farming return, such as labor, feed, veterinary care, energy, and transportation. Consequently, the farm's profitability is anticipated to be adversely impacted by any cost involved in running the poultry farms.

$X_{14}=$ Access to credit: Agricultural loan accessibility and availability are critical to the efficient operation and improved financial performance of poultry farms. They facilitate entrepreneurship development and enable farmers to purchase adequate raw materials and adopt recent farm innovations [[35], [36], [37]]. Therefore, the binary nature of this variable is added to this analysis to see if farmers with adequate access to credit facilities were earning more profit from poultry farming than their counterparts. Moreover, increased access is supposed to be positively linked to the profitability of the poultry farm.

Before running the multiple regression model, both dependent and independent variables were plotted in scatter graphs to check whether they were normally distributed (Fig. 1).

Fig. 1.

Probability distribution graph of the selected explanatory variables.

Almost all the variables revealed that their relationship was not linear (Fig. 1). Therefore, a natural logarithm was used on each continuous variable to ensure its normal distribution in the analysis, which held the assumption of normality in a classical linear regression model (Fig. 2).

Fig. 2.

Correlation matrix of the selected log-based explanatory variables.

3. Results and discussion

3.1. Summary statistics of the analytical variables

The information on socioeconomic characteristics is beneficial for gaining insight into the sample profile when formulating effective policy interventions. Table 1 presents the primary socioeconomic attributes of the studied farmers and their farms, divided into two categories of participants: broiler and layer farmers.

Table 1.

Descriptive statistics of the selected socioeconomic variables used in functional analysis.

Variables	Broiler farmers (N = 100)		Layer farmers (N = 100)		Total sample (N = 200)		T-value	Expected sign
	Mean	Std.	Mean	Std.	Mean	Std.
Dependent variable
Gross return from poultry farming	149900 (1764)	147133 (1731)	672231 (7909)	831119 (9778)	411065 (4836)	650356 (7651)	−6.19∗∗∗
Independent variables
Age of the farmers (years)	40.08	12.10	41.25	10.63	40.67	11.38	−0.73	+/−
Level of education (years of schooling)	7.52	4.52	8.53	3.91	8.03	4.24	−1.69∗	+
Family size (no.)	4.66	1.58	4.86	1.39	4.76	1.49	−0.95	+
Work experience (years)	6.65	4.26	9.74	5.94	8.20	5.38	0.44	+
Landholding size (acre)	1.21	4.50	1.26	1.23	1.24	3.29	−0.11	+/−
Cost of tools, equipment, and maintenance (BDT)	12740 (150)	14026 (165)	18000 (212)	8563 (101)	15370 (181)	11887 (140)	−3.20∗∗∗	–
Cost of day-old chicks (BDT)	173304 (2039)	76107 (895)	47318 (557)	28371 (334)	110311 (1298)	85265 (12061)	15.51∗∗∗	–
Cost of feed (BDT)	717284 (8439)	309669 (3643)	1370526 (16124)	1335606 (15713)	1043905 (12281)	637627 (541983)	−4.87∗∗∗	–
Cost of litter use	25187 (296)	39049 (459)	17541 (206)	13559 (160)	21364 (251)	29406 (346)	1.85∗	–
Cost of hired labor (BDT)	7630 (90)	24434 (287)	27844 (328)	56599 (666)	17737 (209)	44647 (525)	−3.28∗∗∗	–
Cost of veterinary services (BDT)	74343 (875)	44267 (521)	127090 (1495)	138489 (1629)	100717 (1185)	105902 (1246)	−3.63∗∗∗	–
Cost of electricity (BDT)	12098 (142)	8829 (104)	14860 (.)	8225 (97)	13479 (159)	8623 (101)	−2.29∗∗	–
Cost of transportation (BDT)	4112 (48)	5075 (60)	6205 (73)	3552 (42)	5159 (61)	4186 (49)	−2.25∗∗	–
Annual household gross return (BDT)	1126275 (13250)	428979 (5047)	1658045 (19506)	1482072 (17436)	1705226 (20061)	1271392 (14958)	−7.22∗∗∗	+
Availability of credit (dummy: 1 = yes, 0 = no)	0.63	0.49	0.35	0.48	0.49	0.50	4.11∗∗∗	+

∗∗∗p < 0.01, ∗∗ <0.05, ∗<0.1.

Figures in the parenthesis indicate values in USD.

Source: Field survey, 2019

According to Table 1, most poultry producers' average age was around 41 years. This age, however, differs from the estimate of [5], who calculated that the average poultry producers’ age was around 30 years. Moreover, this age represents the middle and potential working-age category, as several authors classified the age between 30 and 50 years in this category [13,17,22,26,38]. Age could indicate farming experience, meaning that farmers with more experience will likely possess better knowledge and expertise in producing layer and broiler chickens [[39], [40], [41], [42], [43]]. Compared to layer farmers (approximately 41 years), broiler farmers were one year younger (about 40 years). However, no notable difference is observed in the mean age of the farmers, as evidenced by the estimated T value of −0.73 presented in Table 1.

Regarding the completion of schooling, most respondents had up to 8 years of schooling, representing the secondary level of education in Bangladesh's context. On average, layers farmers were slightly more educated (about 8.53 years) than broiler farmers (about 7.52 years), as indicated by the statistically significant T-statistic of −1.63 at 0.01 level of significance (Table 1). Several authors also observed a similar finding [5,16,22,24] where most participants had a secondary education grade. This result, however, differs from Refs. [13,17,18], who revealed that more than 50 % of chicken producers had completed a tertiary level of education, which had a beneficial effect on managerial ability and the development of contemporary agricultural business management abilities. Education enables farm households to effectively manage their limited resources by promoting the adoption of innovative solutions and technical advancements [44,45]. Moreover, obtaining a higher education is anticipated to enhance farmers' capacity to make well-informed decisions for increased efficiency and production due to their higher capabilities, access to knowledge, and proficient farm planning [[46], [47], [48]].

On average, the family size of the poultry farmers consisted of five members. This statistic is somewhat more significant than the average family size for the country, which is 4.06, which suggests a medium-sized family in the context of Bangladesh [26]. Studies representing other underdeveloped and developing country's perspectives [[16], [17], [18]] indicate that the average family size of the poultry farmers was between four and six people, which suggests that the large pool of family labor would provide inexpensive labor for agriculture operations. Additionally, the number of hired laborers used by poultry farmers is influenced by family size; the more significant the family size, the less hired labor is used, and vice versa [40]. Even though the family size of layer farmers (4.86) was moderately higher than that of broiler farmers (4.66), no indicative difference is observed in the family size within the respondents' category, as shown by the statistically insignificant T-value of −0.95 (Table 1).

Table 1 demonstrates that both respondents were pretty experienced in their occupation, with nearly eight years of expertise and no statistical difference. This study finds consistency with other studies that poultry farmers’ work experience ranges between 5 and 10 years, enabling farmers to optimize their output given their extended knowledge of poultry rearing [13,15,22,26]. In contrast, it deviates from the findings of other studies that identified their respondents as either less experienced (<5 years) or extensively experienced (>10 years) in poultry farming [5,17,45]. More distinctive agricultural experience is expected to improve knowledge of various farm operations. Given the level of technology, farmers who have been in business for a long time have found out how to combine limited resources and thus might be aware of specific features or solutions to poultry production that their peers are unaware of [39,[44], [49], [50]]. Generally, more years of agribusiness experience assist farmers in better-identifying marketing risks, setting cost targets, and determining the optimum time to reach their goals, enhancing the possibilities of higher profits in poultry production [[51], [52], [53]].

On average, the land size of all respondents was about 1.24 acres. At the same time, no notable difference is observed between the farm sizes of the groups, as the T-value of −0.11 is found to be statistically insignificant. Moreover, this figure indicates the small farmer's category according to Ref. [54]. This estimation is in line with [5], who estimated the mean farm size of their respondents as 2.17 acres, indicating the small farming category, while it differs from Ref. [38], who reported the majority of their respondents as landless farmers, having approximately 0.48 acres of land.

Table 1 further highlights a notable statistical difference in the costs incurred by broiler and layer farmers within the research areas. Their most significant expenses were buying feed, chicklings, veterinary assistance, litter, tools and equipment, electricity, labor, and transportation. When it came to tools and equipment, farmers spent over BDT 15370 (∼181 USD), but layer farmers spent significantly more (about BDT 18000∼212 USD) than broiler farmers (about BDT 12740∼150 USD). On the contrary, broiler farmers incurred a much higher cost of approximately BDT 173304 (∼181 USD) to purchase day-old chicks than layer farmers (approximately BDT 47318∼557 USD). In comparison, the mean cost incurred by all respondents is estimated at BDT 110311 (∼1298 USD). In the case of feed, both the groups incurred the highest amount of cost among all the cost components, of which the cost incurred by the layers farmers was comparatively much higher (about BDT 1370526∼16124 USD) than the broiler farmers (close to BDT 717284∼8439 USD) while it was about BDT 1035900 (∼12187 USD) for all the respondents. Farmers spent a mean amount of BDT 21364 (∼251 USD) for litter use purposes, while it was about BDT 25187 (∼296 USD) and BDT 17541 (∼206 USD) respectively for broiler and layers farmers. The comparative cost analysis for the hired labor use indicates that layer farmers incurred much higher costs (about BDT 27844∼327.58 USD) than broiler farmers. On average, the farmers spent approximately BDT 17737 (∼209 USD) for hired labor in their poultry farms in the study areas. Table 1 also reflects that poultry farmers’ costs for availing of the veterinary services were not negligible in the study areas. It was the third-highest cost incurred by them among all cost components. On average, farmers spent approximately BDT 100717 (∼1185 USD) on this item, while it was much higher for the layer farmers (nearly BDT 127090∼1495 USD) than broiler farmers (nearly BDT 74343∼875 USD). For the cost of electricity and transportation, farmers incurred a mean level of BDT 13479 (∼159 USD) and BDT 5159 (∼61 USD), respectively. However, they did not regard these items as significant costs of poultry farming. In both cases, layer farmers incurred higher costs than broiler farmers. Such a cost analysis is somewhat similar to Ref. [34], which states that the cost of layer production was almost 12 times higher than that of broiler production in the study areas.

Finally, only about half of the respondents had the opportunity to access credit in the study areas. In contrast, participation in these facilities was much higher among the broiler farmers than among the layer farmers. This result is consistent with previous studies by Refs. [13,15], which revealed that chicken farmers have trouble obtaining financing for their operations.

3.2. Factors determining the profitability of poultry production

Table 2 summarises the estimated results of the regression model. It identifies the elements that affected poultry farming's profitability in the study region. It is important to note that the nature of cost items and the extent of variation in the level of profitability differed between broiler and layer farmers. Therefore, it is crucial to design policy measures based on those factors that determine the profitability of poultry producers, who are mostly affected by changes in production, supply, and market conditions.

Table 2.

Computed outcomes from the multiple regression analysis.

Variable description	Coefficients	Robust std. error	t-value
Log of Age	−1.01	1.23	−0.80
Log of education	−0.85	0.37	−2.32∗∗
Log of family size	−1.73	1.01	−1.71∗
Log of experience	0.28	0.44	0.63
Log of farm size	−0.09	0.34	−0.26
Log of cost of tools and equipment	−1.07	0.55	−1.95∗∗
Log of cost of day-old chicks	−1.47	0.43	−3.40∗∗∗
Log of cost of feed	0.82	0.59	−1.38
Log of cost of litter	−1.19	0.98	−1.21
Log of cost of hired labor	0.16	0.05	3.25∗∗∗
Log of cost of veterinary services	1.35	0.82	1.66∗
Log of cost of electricity	0.21	0.53	0.40
Log of cost of transportation	−0.04	0.12	−0.37
Availability of credit (dummy: 1 = yes, 0 = no)	−0.01	0.68	0.02
Constant	28.69∗∗	11.72	2.45
F (14, 185)	3.99∗∗∗
R Squared	0.27
N	200
Heteroskedasticity test
Breusch-Pagan/Cook-Weisberg test statistic	59.04∗∗∗
Multicollinearity test
Variance inflation factor (VIF)	1.55

∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0-0.1.

Source: Field survey, 2019

The findings show that six of the fourteen explanatory factors employed in the model estimation significantly influenced the profitability of poultry farming in the research locations. The level of education, family size, costs of tools, and day-old chicks was negatively associated with farm profitability, while costs of hired labor and veterinary services were positively associated.

Unlike the prior assumption in section 2.3 regarding the impact of variable $X_2$, this study found that an increase in the level of education will decrease farm profitability by 85 %. The probable explanation of this result could be that an educated member is more likely to adopt innovation in farming practice, which requires increased investment and, thus, reduced profit. Even higher education increases the household's non-food expenditure, which may have resulted in such an outcome. However, this finding contradicts the study by Ref. [11], which estimated a positive dependence between a farm's financial performance determined by the Cost-Benefit Ratio (CBR) and years of schooling.

The coefficient of family size is estimated at −1.73, which implies that an additional member in the family will decrease the profitability of poultry farms by 173 %. It implies that adding to the number of family members is not advantageous in reducing hired labor costs. Instead, it increases the cost of living, which should ultimately be met by the profit earned from poultry farming. This finding is entirely unexpected compared to what was assumed earlier in section 2.3. Furthermore, it does not support the study by Refs. [12,55], who determined a positive association between family size and farm profitability.

Among the cost components, the costs of tools and equipment and day-old chick are estimated to have a negative dependence, with profitability having coefficient values of −1.07 and −1.47, respectively. The highly significant and negative coefficient values of these cost items confirm the assumptions made earlier in section 2.3 regarding the estimated impacts of these variables in the model estimation. To put it more concisely, if the cost of tools and equipment increases by 1 %, it reduces the profitability of poultry farms by 107 %. This result can be supported by Ref. [12] as these authors also claimed that the repair and maintenance cost significantly reduces farm profitability. The estimated negative parameter concerning the day-old chick cost implies that as this cost increases by 1 %, it will decrease profitability by 147 %. This finding can be justified in accordance with [56], that scarcity, as well as increased cost of day-old chicks, led to a surge in their price in the unstable poultry market, making it harder for farmers to manage risk-free investments, resulting in a decline in profit. This finding is backed by Ref. [21] and contradicts many studies such as [8,14,15,17], which revealed that increased cost of day-old chick would result in higher profits for poultry growers.

Labor cost is estimated to have a positive association with profitability. It implies that a 1 % increase in labor cost will increase the profitability of poultry farming by 16 %, which contradicts a prior expectation regarding variable $X_{10}$ presented in section 2.3. The plausible explanation of this result could be in line with [24,57], who reported a correlation between higher labor productivity and higher wage rate, which may have positively influenced the outcome. Conversely, higher labor cost is directly linked to high product prices, which can result in higher profitability [58]. Therefore, such a positive relationship is not surprising. However, this finding contradicts prior studies by Refs. [8,11,12,15,17,21], who noted a substantial inverse link between labor costs and agricultural profitability while it is consistent with [24].

The cost of veterinary services and medicine is the last item found to have statistical significance in terms of profitability. However, the coefficient sign of this variable is positive, implying that as this cost increases by 1 %, the profitability increases by 135 %. This result does not align with the earlier assumption regarding the probable impact of this described in section 2.3. A more plausible explanation of this impact could be that if farmers can efficiently afford better veterinary services, it will help increase poultry birds' growth and production without substantial weight loss or death. It makes sense because poultry maintenance needs a substantial dosage of drugs and medications for frequent vaccinations to achieve high production [15,18]. Furthermore, increased growth and increased sales result in higher profitability per farm. This finding is consistent with [[13], [14], [15],17], who revealed that higher drug costs result in higher poultry farmers’ gross revenue. However, it goes against [12,21] as they estimated a negative dependence between the cost of veterinary services and profitability.

Apart from the statistically significant variables, all other factors, including age, experience, farm size, costs of litter, electricity, transportation, and the availability of credit services, are identified as insignificant determinants of profitability with positive and negative implications. Several authors also estimated these variables as insignificant determinants of profitability. For example [28], identified age and credit access as insignificant determinants, while [5] identified age and family size as insignificant variables in their study. Additionally, credit needs were identified as a negligible indicator of profitability by Ref. [28]. Although feed cost was identified as the foremost and significant cost of poultry farming by many authors such as [12,13,15,17,20,21], this cost is estimated to have an insignificant positive impact in this study, which can be supported by Refs. [8,14,24].

Table 3 compares OLS, VSM, PCA, and SR results. It reveals that multiple analyses using the existing dataset validate the earlier findings obtained through the OLS method with practically the same sign and significance. Except for the cost of tools and equipment, which was previously identified as a significant determinant of profitability, the VSM approach reveals that the dependent variable under consideration were similarly influenced by the variables estimated by OLS (Table 3 and A2). SR approach, using statistical selection, retains the same significant variables as OLS, ensuring the robustness of the findings while excluding insignificant predictors [Table 3 and A3]. PCA illustrates strong positive loadings for hired labor, cost of tools and equipment, cost of veterinary services and negative loadings for the cost of day-old chicks in the first component, which explains most of the variability (23 %) (Table 3 and A4). Several other studies have also utilized a similar approach [[59], [60]]. In conclusion, applying referent models or adding and dropping variables has a negligible effect on this analysis.

Table 3.

Comparison of estimated parameters among OLS, VSM, PCA and SR approaches.

Significant Variables	Coefficient of OLS	Coefficient of VSM	PCA loading (significant component)	Coefficient of SR
Log of education	−0.85∗∗	−0.85∗∗∗	−0.55 (Comp3)	−0.82∗∗
Log of family size	−1.73∗	−1.77∗	0.17 (Comp1)	−1.73∗
Log of cost of tools and equipment	−1.07∗∗	−0.81	0.40 (Comp1)	−1.05∗∗
Log of cost of day-old chicks	−1.47∗∗∗	−1.54∗∗∗	−0.20 (Comp1)	−1.64∗∗∗
Log of cost of hired labor	0.16∗∗∗	0.13∗∗∗	0.14 (Comp1)	0.16∗∗∗
Log of cost of veterinary services	1.35∗	1.86∗∗∗	0.44 (Comp1)	1.32∗∗
R2/Cumulative proportion	0.27	0.24	0.47	0.25

∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1.

Table 3 also presents that the R² value for the OLS model is 0.27, indicating that the included variables explain 27 % of the variation in profitability. VSM and SR have an R² value of 0.24 and 0.25, respectively, which closely aligns with OLS. Although PCA does not provide an R² value, its first three components explain 47 % of the total variability (up to component 3 in Table A4). Thus, it indicates the ability of the selected variables to explain the profitability of poultry farmers in the study areas. Altogether, the VSM, SR and PCA validate the reliability of the OLS results.

3.3. Model test

Tests for heteroskedasticity and multicollinearity were conducted to choose the model that best suited the data. A model is thought to provide biased parameter estimates if the problems of multicollinearity and heteroskedasticity are not dealt with, making the regression coefficients misleading. Heteroskedasticity refers to a situation in which one or more predictor variables not included in the model are expected to change the variance of the estimated regression model's error term. Multicollinearity is the state in which two or more of the regressors used in model estimation are highly correlated to each other. It suggests that one of the least squares regression model assumptions is broken, namely that the variance of the disturbance term is not constant [61].

Consequently, the relationship between the variables utilized in model estimation (i.e., multicollinearity) was examined using the Variance Inflation Factor (VIF). According to the rule of thumb, a value of VIF equal to 1 implies no correlation, 4 implies moderate correlation, and 10 indicates strong multicollinearity among variables [20]. In the case of solid multicollinearity, the model needs to drop variables with VIF values greater than 10. Fortunately, the average VIF value of the variables used was estimated at 1.55, which implied that the selected variables were free of the multicollinearity problem. The individual VIF values of variables $X_1-X_{14}$ were 1.41, 1.42, 1.17, 1.47, 1.45, 1.68, 1.68, 2.00, 1.48, 1.22, 2.02, 1.89, 1.52, and 1.25, respectively, which indicates that the estimated model is free from multicollinearity.

Conversely, the value of the Breusch-Pagan test statistic was determined to examine the existence of heteroskedasticity in STATA. At the 1 % significance level, the test yielded a significant result of 46.78. It suggested that it is possible to reject the constant variance null hypothesis. This study reported a robust standard error to address the heteroskedasticity issue, as suggested by Ref. [62]. According to Ref. [62], the best way to deal with heteroskedasticity is to use robust standard error because it provides t-statistics similar to the exact t-distribution if the sample size is large. Another way to deal with heteroskedasticity is to use the weighted least squares. However, this approach has some limitations as it changes parameter estimates. If the assigned weights are incorrect, it provides biased parameter estimates. If correct, it provides smaller standard errors than OLS with robust standard errors.2 Therefore, this study utilized robust standard error of the model estimates, presented in Table 2.

4. Conclusion

The role of the poultry sector is acknowledged as an effective and vibrant means for alleviating poverty and addressing malnutrition. Poultry farming in Bangladesh presents significant opportunities for rural households, providing affordable protein sources, enhancing nutritional status, and much more. This subsector holds significant importance within agriculture and frequently emphasizes the investment made by farmers across various livestock enterprises. Consequently, previous investigations mainly concentrated on a cost-benefit evaluation of poultry farming, particularly broiler farming in Bangladesh and globally. Nonetheless, there was a lack of a systematic method for pinpointing the elements that influence the profitability of poultry farms in the existing literature. This study aimed to identify the factors influencing poultry farming, including both broiler and layer enterprises, to address the current research gap.

The results indicate that various socioeconomic factors influenced the financial sustainability of poultry farming in the selected study areas. The analysis revealed that factors like the expenses associated with hired labor and veterinary services significantly and positively impacted profitability. Conversely, elements such as educational background, family size, expenses for tools and equipment, and the cost of chicklings had a significant and adverse impact on profitability. Consequently, it is advised that the government prioritize efforts to lower the costs associated with various raw materials for production, especially regarding the expenses related to the purchase and maintenance of tools and equipment, as well as chicklings. While the relationship between hired labor and veterinary services and profitability is positive, these findings contrast sharply with the expectations outlined in the literature review.

Furthermore, the findings of this study reveal an unresolved impact of various cost elements on a farm's profitability. Nonetheless, the government needs to intervene to reduce production costs and support the development of small-scale poultry producers in Bangladesh. Contract farming with breeding facilities is proposed as a viable alternative to lower the expenses associated with acquiring day-old chicks. In Bangladesh, poultry farmers typically receive agricultural credit amounting to 3906 USD for 1000 broiler birds and around 14529 USD for 1000-layer birds, excluding the costs associated with shed preparation, in accordance with the agricultural credit policy set by the Central Bank [[63], pp. 77–78]. Nonetheless, it falls short of the calculated values from this study, which were around 5079 USD and 18469 USD, respectively, for every 1000 broiler and layer birds. The data suggests that the credit amount falls short for poultry farmers, considering the magnitude of their production efforts. Consequently, expanding the agricultural credit range is essential to provide the necessary resources for initiating or sustaining small-scale poultry farms. One potential approach could involve incorporating poultry farmers into the Small and Medium Enterprise (SME) scheme, where the credit amount varies from 588 to 58824 USD [[64], p. 7]. Accessible credit with advantageous conditions can assist farmers in covering production expenses while also allowing them to allocate additional resources towards feed or chickling production independently or to enhance vertical integration via contract farming.

Furthermore, it is essential for farmers not only to prioritize cost reduction strategies but also to examine the elements that impede the profitability of poultry farming in the study areas, including large family sizes. Nonetheless, they can undergo training and participate in practical experiences through extension programs, which will improve their understanding and abilities in effective farm management while minimizing negative environmental effects. While higher education is often seen as a barrier, since educated farmers tend to embrace innovations and improved management practices that require greater investment, it can be transformed into a valuable resource if these innovations are accessible at a reasonable price. Additionally, a comprehensive and easily accessible government support service is necessary to foster the growth of the small-scale poultry industry in Bangladesh.

This study indicates that the COVID-19 pandemic hindered extensive data collection across various study regions. Consequently, the projected outcomes and suggestions might differ had a more extensive dataset been utilized. The samples were categorized into two distinct groups: broiler farmers and layer farmers, both originating from the same study region. Nonetheless, this type of categorization may lead to potential bias because of the chance of spillover effects. One approach to managing this spillover effect is to analyze the groups of broiler and layer farmers across various study regions. Nonetheless, as previously noted, the COVID-19 pandemic hindered the data collection process. Therefore, this study utilized cross-sectional data.

Additional studies could be carried out over a longer timeframe utilizing various geographical areas. A panel dataset can be established using this survey as the foundational study in the Gazipur district of Bangladesh, allowing for comparisons of results and assessments of reliability over time. Participants did not maintain any written documentation of their agricultural practices and relied entirely on their recollections to furnish the required details. Consequently, the data obtained do not reflect the true figures but rather the hypothetical or computed values derived from their experience, underscoring another limitation of the study.

CRediT authorship contribution statement

G.M. Monirul Alam: Validation, Supervision, Project administration, Data curation, Conceptualization. Mst Tania Parvin: Writing – original draft, Validation, Supervision, Project administration, Funding acquisition, Formal analysis, Data curation, Conceptualization. Md Ruhul Amin: Resources, Formal analysis. Jarin Saiyara Promee: Writing – review & editing.

5. Ethical standard

Ethical approval for this research was obtained from the Research Management Committee (RMC) of Bangabandhu Sheikh Mujibur Rahman Agricultural University (BSMRAU), which is renamed as Gazipur Agricultural University (GAU) as of 13 February 2025. The entire investigation was carried out in accordance with the standards of the Helsinki Declaration regarding the survey of human subjects. The survey was conducted anonymously and voluntarily, and verbal consent was obtained from all potential respondents before the interview. The data collected for this study were handled with strict confidentiality.

Data availability statement

The data associated with this study has not been deposited into a publicly available repository. However, it will be made available on request.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix.

Table A1.

Results of the functional analysis according to the study groups

Variable description	Broiler farmers		Layer farmers
	Coefficients	Robust std. error	Coefficients	Std. error
Log of age	−0.55	1.90	0.13	0.38
Log of education	−1.32∗∗	0.61	−0.09	0.15
Log of family size	−3.08∗	1.69	−0.03	0.35
Log of experience	0.16	0.98	−0.16	0.13
Log of farm size	−0.14	0.60	0.04	0.10
Log of cost of tools and equipment	−2.17∗∗	0.89	0.19	0.24
Log of cost of day-old chicks	0.46	1.25	0.68∗∗∗	0.23
Log of cost of feed	−0.52	1.61	−0.38∗∗	0.19
Log of cost of litter	−0.80	1.46	−0.58	0.39
Log of cost of hired labor	0.24∗∗	0.12	−0.01	0.02
Log of cost of veterinary services	1.11	1.09	−1.23∗∗∗	0.29
Log of cost of electricity	0.71	0.83	0.09	0.20
Log of cost of transportation	−0.13	0.16	−0.16	0.17
Availability of credit facilities (dummy: 1 = yes, 0 = no)	0.49	1.07	−0.39	0.23
Constant	29.25	18.99	1.08	4.20
F (14, 85)	1.90∗∗		4.08∗∗∗
R Squared	0.21		0.40
N	100		100
Heteroskedasticity test
Breusch-Pagan/Cook-Weisberg test statistic	11.89∗∗∗		1.01
Multicollinearity test
Variance inflation factor (VIF)	1.50		1.91

∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1.

Source: Field survey, 2019

Table A2.

Results of VSM

Variable description	Coefficients	Robust std. error	t-value
Log of education	−0.85	0.32	−2.71∗∗∗
Log of family size	−1.77	0.96	−1.83∗
Log of cost of tools and equipment	−0.81	0.56	−1.44
Log of cost of day-old chicks	−1.54	0.32	−4.77∗∗∗
Log of cost of hired labor	0.13	0.05	2.76∗∗∗
Log of cost of veterinary services	1.86	0.61	3.05∗∗∗
Constant	18.86∗∗	8.63	2.18∗∗
F (6, 193)	8.78∗∗∗
R Squared	0.24
N	200

∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1.

Table A3.

Results of the SR

Variables	Coefficient	Std.	Err.	t	P > t	[95 %
Log of education	−0.82	0.35	−2.34	0.02	−1.51	−0.13
Log of family size	−1.72	0.90	−1.92	0.06	−3.48	0.05
Log of cost of feed	0.93	0.47	1.96	0.05	−0.01	1.86
Log of cost of veterinary services	1.32	0.56	2.34	0.02	0.21	2.43
Log of cost of hired labor	0.16	0.06	2.56	0.01	0.04	0.27
Log of cost of tools and equipment	−1.05	0.45	−2.36	0.02	−1.93	−0.17
Log of cost of day-old chicks	−1.64	0.31	−5.29	0.00	−2.25	−1.03

Table A4.

Result of PCA

Variables	Comp1	Comp2	Comp3	Comp4	Comp5	Comp6	Comp7	Comp8	Comp9	Comp10	Comp11	Comp12	Comp13	Comp14	Unexplained
Log of gross return from poultry farming	0.18	−0.45	−0.06	0.01	−0.31	0.20	0.48	0.21	0.42	−0.20	0.16	0.13	0.31	0.04	0
Log of age	0.08	−0.25	0.56	0.08	0.01	0.12	−0.48	−0.28	0.25	0.16	0.09	−0.23	0.37	0.00	0
Log of education	0.11	0.20	−0.56	0.26	0.04	−0.05	−0.29	0.30	0.17	0.38	0.29	−0.12	0.34	−0.06	0
Log of family size	0.17	0.09	0.20	0.19	0.66	−0.25	0.52	−0.11	−0.04	0.07	0.22	−0.10	0.20	−0.09	0
Log of experience	0.30	−0.04	0.36	0.09	0.22	0.03	−0.25	0.69	0.02	−0.08	0.02	0.32	−0.25	−0.09	0
Log of farm size	0.23	0.16	0.03	0.50	−0.20	−0.49	−0.02	−0.05	0.24	−0.24	−0.42	−0.05	0.01	0.31	0
Log of cost of tools and equipment	0.40	0.21	−0.07	0.00	−0.11	0.06	−0.19	−0.26	−0.39	−0.49	0.29	0.33	0.31	−0.08	0
Log of cost of day-old chicks	−0.20	0.49	0.28	−0.06	−0.11	0.28	0.19	0.24	−0.15	0.18	−0.23	0.14	0.46	0.34	0
Log of cost of feed	0.39	0.16	0.14	−0.16	−0.38	−0.13	0.18	0.04	−0.10	0.28	−0.25	−0.18	0.05	−0.63	0
Log of cost of litter	−0.12	0.53	0.13	0.19	−0.09	0.27	0.08	−0.21	0.53	−0.08	0.26	0.14	−0.30	−0.24	0
Log of cost of hired labor	0.44	−0.04	0.06	0.06	−0.20	0.08	0.12	−0.21	−0.14	0.52	0.26	0.16	−0.34	0.46	0
Log of cost of veterinary services	0.34	0.25	−0.03	−0.42	0.08	0.13	−0.02	0.15	0.14	−0.29	0.09	−0.63	−0.11	0.29	0
Log of cost of electricity	0.14	−0.10	−0.12	0.54	0.11	0.64	0.06	−0.04	−0.24	−0.07	−0.29	−0.27	−0.09	−0.10	0
Log of cost of transportation	0.30	0.01	−0.25	−0.32	0.38	0.18	−0.09	−0.24	0.34	0.12	−0.49	0.36	0.08	0.01	0

Robustness check

Table A5.

Impact of socio-demographic factors on poultry profitability by using linear regression

Variables	Coef.	St. Err.	t-value	p-value	[95 % Conf.	Interval]	Sig.
Log of age	0.2	1.11	0.18	0.857	−1.99	2.39
Log of education	−0.39	0.41	−0.95	0.345	−1.19	0.42
Log of family size	−1.34	1.03	−1.30	0.194	−3.37	0.69
Log of experience	0.73	0.43	1.69	0.09	−0.121	1.57	∗
Constant	11.77	4.51	2.61	0.01	2.87	20.67	∗∗∗
Mean dependent var		11.10	SD dependent var			4.04
R-squared		0.03	Number of obs			200
F-test		1.34	Prob > F			0.26
Akaike crit. (AIC)		1129.94	Bayesian crit. (BIC)			1146.43

∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1.

Table A6.

Impact of poultry production factors by using linear regression

Variables	Coef.	St. Err.	t-value	p-value	[95 % Conf	Interval]	Sig
Log of farm size	−0.16	0.28	−0.63	0.53	−0.73	0.38
Log of cost of tools and equipment	−0.96	0.48	−2.00	0.05	−1.91	−0.01	∗∗
Log of cost of chicks	−1.33	0.37	−3.63	0	−2.05	−0.62	∗∗∗
Log of cost of feed	0.79	0.51	1.57	0.12	−0.20	1.79
Log of cost of litter	−1.41	0.88	−1.61	0.11	−3.13	0.32
Log of cost veterinary services	1.46	0.56	2.61	0.01	0.36	2.56	∗∗∗
Log of cost of electricity	−0.11	0.45	−0.25	0.80	−1.00	0.77
Constant	23.41	9.93	2.36	0.02	3.82	43.00	∗∗

Mean dependent var	11.10	SD dependent var	4.04
R-squared	0.22	Number of obs	200
F-test	7.70	Prob > F	0.000
Akaike crit. (AIC)	1091.93	Bayesian crit. (BIC)	1118.31

∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1.

Table A7.

Impact of marketing factors on poultry profitability by using linear regression

Log of gross margin	Coef.	St.Err.	t-value	p-value	[95 % Conf	Interval]	Sig
Log of cost of hired labor	0.16	0.06	2.46	0.02	0.03	0.28	∗∗
Log of cost of transportation	0.11	0.09	1.16	0.25	−0.07	0.29
Constant	9.99	0.72	13.81	0.00	8.56	11.42	∗∗∗

Mean dependent var	11.10	SD dependent var	4.04
R-squared	0.04	Number of obs	200
F-test	3.85	Prob > F	0.02
Akaike crit. (AIC)	1123.72	Bayesian crit. (BIC)	1133.61

∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1.

Graph concerning the comparison of standardized coefficient of OLS and SR models

Fig. A1.

Comparison of standardized coefficient of OLS and SR models.

Graph representing the scree plot of eigenvalues after PCA.

Fig. A2.

Scree plot of eigenvalues after PCA with homoskedastic bootstrap.