The role of supply chain integration in strengthening the performance of not-for-profit organizations: evidence from the food banking industry Open Access

Managing operations in a humanitarian context such as those faced by food banks is quite challenging as there is an increasing demand for services (Kotler and Andreasen, 1991), enhanced competition among food banks for privately contributed revenues (Schwartz, 1989), and significant shortages of professionally trained and experienced personnel (Wolf, 1984; Kovacs and Spens, 2011). Therefore, the management teams at food banks are seeking ways to improve their performance and leverage their supply chain partners more effectively.

Food banks are not-for-profit organizations that aim to alleviate the hunger problem in the society by working with donors on the upstream and member agencies on the downstream of their supply chains. These organizations operate in highly complex environments due to the uncertainties associated with supply and demand. In order to cope with the uncertainty and to be able to deliver food to the underserved communities, foodbanks need to utilize information and coordinate well with their partners so that the donations can be matched with demand in a timely manner. Therefore, effective and efficient supply chain management approaches become critical in this resource-constrained setting, where it is unknown which resources are available (Kovács and Spens, 2007). Based on what we know from the previous research on for-profit firms, and given the complexity of food bank supply chains along with the necessity of food banks to work with partner firms, it is natural to contend that supply chain integration is a key determinant of performance (Kim et al., 2018, p. 246). Exactly how the constituent parts of supply chain integration relate to each other in humanitarian organizations, however, is an open research question (Dubey and Altay, 2018). Three common integration dimensions that have been investigated in the supply chain management literature include internal integration, demand integration and supply integration (Schoenherr and Swink, 2012; Flynn et al., 2010; Frohlich and Westbrook, 2001). Demand integration and supply integration are also referred to as external integration in some studies (Stank et al., 2001). Collaboration between functions and information sharing processes within an organization determines the internal integration. Some indicators of internal integration in the literature include intra-firm functional team activities between functions, such as operations, marketing and logistics, to achieve the goals of the supply chain (Schoenherr and Swink, 2012; Ataseven and Nair, 2017). Moreover, demand integration relates to the collaboration activities and information sharing with key customers to meet their expectations and improve supply chain performance. Similarly, supply integration involves coordination activities as well as information sharing with key suppliers to achieve supply chain performance by enabling better planning, forecasting and process design (Schoenherr and Swink, 2012; Ataseven and Nair, 2017).

Thus, we study the three key dimensions of supply chain integration – supply integration, demand integration and internal integration and empirically examine how each is related to the others and how they impact food banks' ability to create basic programs that help in delivering food to the underserved communities. Basic programs are service offerings customized to meet the needs of different segments in the service area of a food bank. The supply chain management capabilities and practices of the organizations will be a determinant of the value creation activities (Ramdas, 2003), which are represented by the basic programs in this setting. To answer these pertinent issues related to food bank supply chain management, we created a unique dataset that combines secondary data containing contextual, operational, and financial information about food banks in the US with primary data collected from managers at 110 different food banks. Our theoretically developed research hypotheses are tested by means of seemingly unrelated regression (SUR) techniques and a Monte Carlo simulation-based mediation analysis.

We use organizational information processing theory to shed light on the relationships between supply chain integration and performance in this highly uncertain setting. Similar to Wong et al. (2011), we posit that the information processing ability of the organization and the information quality become especially critical in highly uncertain environments (Galbraith, 1973). Organizations deal with external as well as internal information since they interact with partners beyond their boundaries (Thompson, 1967; Wilburn, 2009; Wong et al., 2011). In order to increase their information processing capability to overcome the obstacles caused by highly uncertain contexts, organizations develop practices and managerial approaches with the ultimate aim of improving performance. Supply chain integration is one of these approaches and it has been studied before in commercial settings (Wong et al., 2011; Wolf, 2011). In this humanitarian context, there is an interplay between internal and the external integration dimensions that constitute the information processing mechanism, which drives performance through the basic programs offered to the underserved communities.

There are various conceptualizations and relationships among supply chain integration dimensions tested in the literature. Flynn et al. (2010) discuss a contingency and configuration approach to supply chain integration and performance relationships. Regarding the contingency approach, the authors state that an organization's environment determines the structures and processes, and there is no single best way for organization design. In terms of the general theoretical framework to be used in this humanitarian setting, we come up with an alternative precedence sequence between external and internal integration, and support this argument with the contingencies present in the non-profit environment considered for this study. In contrast to most of the previous research on supply chain integration that had been conducted within the context of for-profit supply chains, which shows that managing internal integration forms the foundation for integrating with suppliers and customers, we find that, for not-for-profit food banks, external integration should precede internal integration and that demand integration has a stronger influence on performance than supply integration. Thus, it appears that the heavy reliance of food banks on external partners requires an internal integration structure that is somewhat customized to these external relationships. Our findings also show that supply integration and demand integration do not directly influence the number of basic programs developed by food banks. Instead, internal integration plays an important mediating role in the development of basic programs. The number of basic programs, in turn, has a direct impact on the amount of food distributed per food insecure individual.

Our study is different from the existing body of literature in that it investigates the interplay between supply chain integration dimensions and its effect on the programs created in the food banks to deliver food, which is operationalized as an objective measure reported by food bank organizations. While there is some research on supply chain integration in humanitarian organizations (Ataseven et al., 2018; Kim et al., 2018), this is the first study that combines primary (survey) data on humanitarian supply chain integration and objective data such as food distributed and financial data from IRS 990 forms to understand the unique coordination challenges in a humanitarian setting to the best of our knowledge. Intellectual capital (human, organizational and social capital) has been studied as the antecedent of supply chain integration in food banks empirically (Ataseven et al., 2018); on the other hand, the mechanism through which supply chain integration dimensions translate into performance in this humanitarian setting has not been addressed before. The findings from our empirical investigation present new insights that extend the theory and practice of supply chain integration within the context of humanitarian logistics. In the next section, we elaborate on the food banking context and explain the complexities that these humanitarian organizations face.

To help explain some of the unique supply chain integration challenges that food banks face, we first describe the primary functions that most food banks serve in the US market. Food banks are not-for-profit organizations that collect, organize, and deliver food to other member agencies – such as soup kitchens, food pantries and shelters – as well as directly to individuals in order to reduce the hunger problem in the society. Food banks can also be considered as a waste management system for the overall food industry since a significant percentage of the total grown, processed and manufactured food is not consumed because of expiration, overproduction, damage, marketing and other decisions. The result is that billions of pounds of food is wasted each year while almost one billion people worldwide do not have enough food to eat. Thus, the primary purpose of food banks is to gather surplus food that would otherwise be wasted and deliver it to the people who need it the most.

The supply network of food banks is quite complex, consisting of private organizations that provide support in the form of money and food on the supply side; and member agencies including food pantries, soup kitchens and shelters, and individuals who help in managing the distribution of food on the demand side. Food banks get donations from corporations, individuals and federal and state partnerships. Fundraising activities are important for food banks, and vary from one food bank to another due to the level of donated food supply. These funds often times are necessary for food banks to purchase a certain percentage of the food that they distribute. On the demand and distribution side, member agencies pay a certain amount of money to their respective food banks to purchase food to be distributed to people in need. For many working and non-working Americans, the support coming from food banks has become a major mode of sustenance.

While food banking is relatively new in some parts of the world, it has grown and progressed more in US, Canada and Europe (Riches, 2002). The first food bank in the US was established in 1967 in Phoenix, Arizona with the aim of matching the interests of the food industry regarding how to handle surplus food, and the goal of charity organizations to provide resources to communities in need (Riches, 2002). The idea then grew to other states, and countries such as Canada and UK. Over time, umbrella organizations (e.g. Feeding America, Global Foodbanking Network) have been established and food banks have become institutionalized. Food banks started to engage in partnerships with corporations that donate large amounts of food to these organizations. Governments also support food banks, not only in terms of grants, but also with policies such as the 1976 Tax Reform Act, which permitted corporate tax deductions of cost plus 50% of any appreciated value of the donated food (Daponte and Bade, 2006). Policies such as these have provided incentives to donors and supported the industry. On the distribution side, food banks distribute food on their own as well as work with several not-for-profit organizations to increase their ability to reach the underserved communities.

Compared to most firms in the food industry, the food banking sector has its own challenges and idiosyncrasies. While there are some similarities with the for-profit supply chains, the way food banks operate along with the resource constraints they face make it worthwhile to study their operations, understand the unique environment and provide with solutions to the issues that they encounter. Moreover, the performance of food banks is measured on the basis of the amount of food distributed to the communities in need, which is quite different from the performance measures used by commercial organizations. There are substantial benefits from the ability of food banks to more efficiently undertake their operations that are aimed at resolving the disparity between food production and consumption.

Ataseven et al. (2018) report some differences of food bank organizations that set them apart from commercial organizations as the uncertainty in food and monetary supply, greater variety of agencies that food banks need to work with to deliver the food, the constrained resources and non-profit objectives and the limited power of the final customer. Food banks are in a position to manage “food, friends and funds” the best way possible to deliver the services efficiently and effectively. Lack of standardization in the expiration dates, quality and quantity of the incoming food makes sorting process a bottleneck in this environment (Ataseven et al., 2018). Moreover, monetary and volunteer support is variable, which makes it more complex to manage operations in this context. In the downstream of these supply chains, there are agencies such as food pantries, soup kitchens and shelters that deliver the support to the underserved communities. In contrast to for-profit supply chains, humanitarian supply chains in general, and food bank supply chains in particular, are supply-driven, and customers (named as clients in this context) do not possess a lot of power (Tomasini and Van Wassenhove, 2009, p. 6). Food banks are resource-constrained organizations, which function in complex supply chains due to the variety of upstream and downstream partners they work with, and these organizations need to deploy their limited resources in the best way possible.

The non-profit performance metrics that food banks use, like other humanitarian organizations and unlike commercial companies, are dependent on the social goal of the organization. Food banks measure the amount of food distributed to underserved communities as a performance indicator. Moreover, wasted food as a result of not being able to match supply and demand in a timely manner in food banks is a more critical problem here since the goal of food bank organizations is making this coordination possible with less capacity cushion in the process. Next, we discuss how food bank performance is driven by the management of its supply chain.

Seamless integration of the processes along the supply chain has been shown to enhance the competitive capabilities of firms (Frohlich and Westbrook, 2001). In general, the broader the arc of integration an organization has, the more successful it becomes. Examination of supply chain integration in the not-for-profit sector has not kept pace with that in the for-profit enterprises. The food banking sector presents an important context since the uncertainty of incoming food and managing an increasingly complex demand necessitates appropriate levels of internal and external integration (Knox and Gruar, 2007). The unique nature of this context requires a closer examination of the interplay among supply integration, demand integration and internal integration to enrich our understanding of supply chain integration in the humanitarian context.

In the previous literature on supply chain management, authors have argued that effective coordination of each partner organization's internal processes forms the initial step to integrate the inter-organizational supply chain activities (Tracey, 2004). There are several studies that conceptualize internal integration as an antecedent of external integration (Tracey, 2004; Braunscheidel and Suresh, 2009; Horn et al., 2014). Internal attitudes and procedures need to be aligned before the inclusion of partners in the integration process (Tracey, 2004). Internal cohesion of processes will encourage external parties to join the integrated processes. Overall, evidence from the literature suggests that a reduction of internal barriers precedes the removal of barriers to external integration (Frohlich and Westbrook, 2002; Braunscheidel and Suresh, 2009).

While this perspective seems to hold true for private organizations that develop supply chain relationships with each other, this pattern of relationship between internal and external relationship may not hold in various contexts. Considering the food banking context, in terms of its most basic function, food banks obtain food from donors such as farms, manufacturers, distributors, retail stores, consumers and other sources, and make it available to those in need through a community agency network. Hence, the ability to forge linkages with supply sources and distribution partners is foundational in terms of the operations of food banks. As compared to for-profit organizations where internal integration is a challenging task, food banks are typically organized with a management layer supported by volunteers who help in the day-to-day operations. The resource constraints and the nature of the business do not lend themselves to a multi-departmental structure. Yet, the ability of the management staff and volunteers to coordinate efforts within a food bank is dependent on the level of integration with supply sources and distribution partners (Ataseven et al., 2018).

This theoretical reasoning accords well with some studies that have shown that inter-organizational collaboration is an antecedent to intra-organizational collaboration (Stank et al., 2001; Sanders, 2007). The literature on supply integration in humanitarian organizations emphasizes the importance of information exchange between the donors (e.g. manufacturers, farms, retailers) and food bank organizations in terms of quantities that can be shipped and time of deliveries; since the coordination of receiving and warehousing of the perishable supplies requires a careful plan executed by the management, staff and volunteers to make the food delivery possible (Ataseven et al., 2018). Hence, we hypothesize that:

Supply chain players such as suppliers, manufacturers, distributors, clients, competitors and other organizations that do not belong to these categories are involved in external integration (Kim et al., 2018). On the downstream of the supply chain, the ability of food banks to coordinate the distribution tasks such as dispatching, shipping and outbound inventory management are dependent on a clear understanding of their relationships with external partners such as soup kitchens, food pantries, churches etc. Previous literature on supply chain integration in humanitarian organizations point out the role of demand integration in terms of understanding the distribution capacities (e.g. amount of food), the geographic reach and transportation capabilities developed through the relationships of the humanitarian organizations with their agencies (Ataseven et al., 2018). Matching the supply of food, that would otherwise be wasted, to the demand of people that are in need requires demand integration that serves as the basis for the needed internal integration. Accordingly, we hypothesize that:

Ramdas (2003) argues that an organization's variety creation can happen in varying levels. The degree of customization will depend on customer needs, market valuation and external and internal supply chain capabilities. We argue that the level of basic programs created by a food bank depends on the interplay between the internal and external supply chain capabilities of the organization, which are characterized by the supply chain integration dimensions within a food bank. Food banks cater to the needs of their constituents by developing new programs. The basic programs developed and run by food banks are customized services for different groups of clients. Some examples include programs targeted for seniors and kids. For instance, many food banks have “Backpack” programs that provide a rotating menu of simple to prepare but nutritious food for low-income neighborhood children. Mid-Ohio Food Bank launched “Urban Farms of Central Ohio” program in 2012 as a specialized produce distribution program that aims to increase access of underserved communities to local produce, and to educate the community about fresh produce and a healthy diet. The same food bank has a “Mobile Markets” program to deliver fresh produce, dairy and bakery items to low income senior residences and daycare centers. These programs run by food banks are, in a sense, similar to the variety of products and services that a commercial firm offers to its customers. This is akin to new product/service development and the ability to offer a wide variety of product offerings requires integrated effort along with suppliers, customers, as well as within the food banks. As mentioned before, the impact of supply chain integration on performance has been emphasized in the literature (Frohlich and Westbrook, 2001; Ragatz et al., 1997). In particular, previous studies have pointed out that supply chain integration plays an important role in managing a large mix of products and services (e.g. Frohlich and Westbrook, 2001; Braunscheidel and Suresh, 2009). Littler et al. (1995) argues that frequent inter-organizational communication, building trust, and ensuring that all parties act as expected are some of the key success factors for new product development. Supply integration, demand integration and internal integration provide the ability to undertake coordinated program development efforts, and enable seamless exchange of information and ideas within food banks as well as between food banks and their supply chain partners.

Supply integration takes place between the food banking organization and its food, fund and disaster partners including farms, manufacturers, distributors, volunteer organizations and retail stores. This integration is the sum of all activities that help in better planning of operations at the food bank in terms of the quantities, timing and delivery of incoming food. The extent of information technology usage, such as radio frequency identification systems, to communicate and integrate with suppliers differs in each food bank. In addition to these differences, the management and coordination of processes in terms of receiving, warehousing and inventory change from one organization to another, showing variation in supply integration in this setting (Ataseven et al., 2018).

Supply integration enhances the understanding about the abilities of the supply base. Suppliers' awareness of the partner organization's internal processes and goals enables the suppliers' planning and strengthens service and product development (Ragatz et al., 1997; Handfield et al., 1999; Petersen et al., 2005; Koufteros et al., 2005). Food banks often work with leading organizations to develop programs meant to target a specific community in need. For instance, local food banks work closely with General Mills to develop programs (see https://www.outnumberhunger.com). Similarly, the Central Virginia Food Bank developed a new program that required a change in focus from canned goods to fresh foods by working with local farms that provided 200,000 pounds of produce ^[1]. Feeding America, the largest network of food banks, works with several organizations that are part of its partnership ^[2] teams. Moreover, like General Mills, Target has partnered with many local food banks including The Greater Boston Food Bank by funding and supporting “Backpack” programs and “Meals for Minds” mobile food pantries that focus on children so that they do not have to worry about their next meal and can concentrate more on their education. Morgan Stanley also provides support to food banks in their “Fill the Plate” program to improve child health. This further substantiates the integrated efforts by corporations and food banks in developing basic programs. Given that theory and practical evidence suggest that integration with suppliers forms an important basis for developing basic programs, we hypothesize that:

As indicated before, there are certain activities such as dispatching, shipping and outbound inventory management that a food bank needs to complete on the downstream of the supply chain (Ataseven et al., 2018). These activities are coordinated with agencies such as pantries, shelters and soup kitchens. The food bank organizations need to exchange information with these partners regarding the quantities of donations, the range of the service area and the level of outreach to underserved communities and transportation and delivery arrangements (Ataseven et al., 2018). All of the efforts that make the delivery of food possible in the food banking industry and the level of coordination downstream define the demand integration in this setting. Demand integration plays an important role in developing a product concept based on the underlying needs of the target population (Iansiti and Clark, 1994). Integrating the operations with downstream partners is critical to get information on demand patterns and customer requirements. The lack of demand side integration leads to inefficiencies in the system such as poor customer service and higher amounts of waste. Especially in service operations, the characteristics of products/services such as customer participation, heterogeneity and perishability add to the complexity of activities, and increase the need to have demand integration in place to come up with the right scope of services for the clients (Frohlich and Westbrook, 2002). In this respect, coordination with the external partners of a food bank makes it possible to deal with the complexity in the system and to broaden the service range. For instance, in North Texas the Tarrant Area Food Bank created the “Backpacks for Kids” program to provide nourishment on weekends for children at high risk of hunger in conjunction with elementary and middle schools of the area ^[3]. Similarly, the “Hunger Intervention Program” in the state of Washington is a collaborative initiative involving food banks, schools and community partners ^[4]. These examples illustrate that basic programs require inputs from both upstream and downstream partners in food bank supply chains. Accordingly, we hypothesize that:

Internal integration relates to collaboration among various departments of an organization to meet customer requirements (Flynn et al., 2010). It focuses on joint planning, information sharing, and cross-functional teams to ensure that the order fulfillment process is seamlessly executed within an organization. Studies have attributed the ability of firms to provide superior customer service and respond to changing environmental conditions to their level of internal integration (Van Hoek, 1998; Beamon, 1999). Dröge et al. (2004) find that internal integration is positively related to time-to-product, time-to-market, and customer responsiveness. These time-based performance measures, specifically the time-to-product, is crucial for offering a wide range of products. Collaboration among staff members responsible for external relations, operations, program development and food distribution allows food banks to create more programs that cater to the needs of various people. In addition, the ability to integrate volunteer activities within the overall plan also plays an important role in program development. Based on this reasoning, we hypothesize that:

Dröge et al. (2004) consider external integration to relate to strategic design issues and internal integration to relate to tactical process issues. The impact of high-level strategic design activities comprising of supplier partnership and customer relationship on performance depends on the ability of organizations to coordinate and execute internal processes. Management scholars have long advocated that the impact of strategic planning on performance is contingent on the ability of organizations to tactically manage the associated tasks (Anthony, 1965; Steiner, 1969). Steiner (1969) asserts that strategic planning relates to, “…the process of determining the major objectives of the organization and the policies and strategies that will govern the acquisition, use, and disposition of resources to achieve those objectives” (p. 34) and tactical planning is “the detailed deployment of resources to achieve strategic plans” (p. 37). Given the nature of food bank operations, which are heavily reliant on the relationships with suppliers and distribution agencies, the strategic design and planning relates to supply and demand integration. However, similar to a project life cycle (Slevin and Pinto, 1987), food banks go through a phased approach for developing basic programs. Slevin and Pinto (1987) define project life cycle as the framework that describes the resource requirements of stages in the process, and it includes four main phases of conceptualization, planning, execution and termination. While the conceptualization and planning of various programs are dependent on external integration with suppliers and distribution partners, the impact of supply integration and demand integration on the number of basic programs is dependent on the ability of food banks to develop cross-functional integration between the management staff responsible for food distribution, external integration, operations and the development of basic programs. In essence, similar to the assertion in Germain and Iyer (2006), the ability of food banks to internally integrate in terms of inter-functional relationships, trade-offs and processes is expected to play an important mediating role to translate external or inter-organizational integration into superior performance. Accordingly, we hypothesize that:

Product variety is often thought to provide a competitive edge in for-profit organizations by offering products or services tailored to specific market segments and producing higher sales volumes (Berry and Cooper, 1999). Broader product lines enable firms to meet customer demand more closely and increase the reach to customers. Kekre and Srinivasan (1990), for example, have attributed high market shares to the ability of firms to offer broad product lines. Extending this argument, a larger number of basic programs developed by food banks should increase the “market share” by enabling the food to be distributed to a large portion of the underserved community. For example, introducing kids cafes programs serves elementary school children and increases the food bank's percentage served of the population in need to receive food. Similarly, senior brown bag programs help meet the needs of senior community members and increases the reach to this segment. Moreover, various programs attract media and donor attention, thereby increasing the probability of incoming financial support. In this respect, the amount of food distributed per food insecure individual will be dependent on the number of basic programs run, as the programs will be structured to meet different client needs and lead to a higher amount of food distributed by the food banks. Accordingly, we hypothesize that:

The overall research model is presented in Figure 1.

To undertake this research investigation, data on the population served and amount of food distributed annually was gathered from Feeding America's website. Feeding America, formerly known as America's Second Harvest, is the largest food banking organization in the US serving about 37 million people annually in 50 states via its large network of food banks and thousands of member agencies. Feeding America relies on monetary and food donations from government agencies, food industries, institutions and individuals. The funds are used for food bank operations and resources. The food donated to and procured by Feeding America is delivered to regional food banks to be stored in the warehouses until they are delivered to the member agencies and individuals.

At the time of this study, the Feeding America network comprised 202 food banks located throughout the United States. The operational information for these food banks was collected from their websites and this data was then merged with the financial information that was collected from the IRS 990 forms of the food banks. Food banks report their financial information in the IRS 990 forms to ensure transparency of operations and funding to their stakeholders. We were able to obtain the financial information for 120 different food banks from their websites. These food banks were part of the initial dataset that was created for this research.

In addition to this secondary data, we collected primary survey data from food banks about the supply chain integration practices and several key organizational variables by means of an online survey that was administered to US-based food banks who were members of the Feeding America network. The scales for the supply chain integration constructs are adapted from previous literature in supply chain management (Schoenherr and Swink, 2012). We conducted a case study with a local food bank executive to understand the context better and clarify the meaning of supply, demand and internal integration in food banks before we collected survey data. The questionnaire was sent to the chief operating officer (COO) of a food bank to ensure face validity of items contained in the survey. He was asked to go through the survey, indicate if there was any ambiguity, and record the time that it took to complete the survey. The questionnaire was modified in light of the feedback received from the food bank COO to ensure readability and clarity of the survey instrument, and to reflect the contextual differences in the food bank organizations. The first wave of the web survey was distributed in September 2012 where an e-mail introducing the study and a web link for the survey questionnaire was sent to food bank executives. In total, 36 food banks executives completed the survey in this first wave.

A second wave of data collection combining the same web survey with phone calls and face-to-face meetings with managers of some of the non-respondent food banks was completed in May 2013. The potential respondents were asked to complete the survey to be eligible to enter a drawing to win a gift donation to the winning food bank. Moreover, the food bank executives were also told that they would be provided with an executive report once the project was completed. These incentives were provided to increase the response rate, which is known to be problematic in organizational survey research (Baruch and Holtom, 2008). The second wave resulted in 74 additional responses. The total sample size, combining the first and second wave of data collection, was 110, with an overall response rate of 54.5%. This response rate is much higher than the average observed response rates from other survey based studies and surpasses the 20% level recommended in the literature (Malhotra and Grover, 1998).

The first wave and the second wave respondents were compared using “Age” and “Warehouse Size” as well as a randomly selected construct measurement item to test whether non-response bias was a problem in the sample (Armstrong and Overton, 1977). The t-test results indicated no statistically significant differences between the first wave and second wave responses at the 0.05 level (Difference in “Age”: 95% CI: [−2.60, 2.47], difference in “Warehouse Size”: 95% CI: [−26,061.98, 22,154.95]), difference in “Supply Integration Item (SUP1)”: 95% CI: [−0.273, 0.591]. Thus, the results support that non-response bias is not present in the data.

The secondary dataset was matched and merged with the survey data to create a dataset of 71 food banks with complete information on the various aspects that are considered in this study. This represents just over 35% of the population of food banks that are part of the Feeding America network.

The three aspects of supply chain integration were captured by means of survey questions. Specific survey items that were used to gauge internal integration, supply integration and demand integration are presented in Table 1. These items were adapted from the existing supply chain integration literature (Schoenherr and Swink, 2012). Given that supply chain integration constructs are correlated, discriminant validity tests are conducted to evaluate the degree to which these latent constructs are distinct from each other. We assessed the discriminant validity of the constructs by comparing two confirmatory factor analysis models, one of which the correlation between the latent variables is set equal to 1, and another where the correlations are free. When we compare the two models, a significantly lower χ2 value for the unconstrained model relative to the constrained model result indicates discriminant validity (O'Leary-Kelly and Vokurka, 1998). The χ2 value is 460.23 with 119 degrees of freedom for the constrained model. The χ2 difference is significant indicating better fit for the unconstrained model. The χ2 difference test indicated that the unconstrained model explains the data better, thus, we established discriminant validity. The reliabilities of the three supply chain integration scales are assessed using Cronbach's alpha. All of the reliabilities of the scales used in the study are above the recommended cutoff value of 0.70 (Nunnally, 1978). The reliability measures for these constructs are presented in Table 1. Moreover, we analyzed convergent validity using factor loadings on the latent constructs (Hair et al., 1998). Since all the item loadings on their corresponding constructs are greater than 0.5, we conclude that convergent validity is established. Furthermore, we observe that all the estimated path coefficients on the respective latent factor are greater than twice their standard error, which supports the significant relationships between path coefficients and the latent constructs (Anderson and Gerbing, 1988).

Harman's single factor test (Podsakoff et al., 2003) is conducted to test if there is a common method bias in the data. When all the measurement items of the variables are entered in an exploratory factor analysis, a single significant factor is observed if substantial common method variance is present in the dataset (Podsakoff and Organ, 1986). Our exploratory factor analysis for supply chain integration constructs with 17 measurement items using principal component factor analysis and Promax rotation with Kaiser normalization yielded four factors that have an eigenvalue greater than 1. The theory suggests three dimensions of supply chain integration and a three-factor solution was supported by a Scree plot as well. While the first factor accounted for 41% of the variance in the data, the three retained factors cumulatively accounted for 61% of the variance. Therefore, we moved forward with a three-factor solution. Also, these results suggest that common method variance is not an issue in the dataset. We conducted confirmatory factory analysis (CFA) to establish the validity of supply chain integration constructs. The CFA for supply chain integration constructs yielded CFI = 0.89, TLI = 0.87, χ²/df ratio of less than 2 (216.813/116), and RMSEA = 0.09. Given the sample size for the study, these values provide reasonable evidence of validity of the constructs (Hu and Bentler, 1998). The standardized factor loadings are presented in Table 1.

The information regarding the basic programs of the food banks was collected from Feeding America's website. The websites of food banks provide information about various programs such as Fresh Produce, Mobile Pantry, Food Stamps, Senior Brown Bag Distribution, etc. The total number of basic programs for each food bank is used to operationalize this variable. On average, the food banks in the sample had 12.51 basic programs.

Feeding America's website provides information regarding food distributed by food banks (in lbs.) and the number of food insecure people that live within the service area of each food bank. The annual average food distributed by food banks in the sample was 15.2 million pounds and the average number of food insecure people in the service area was 289,639. We divided the food distributed by a food bank by the number of food insecure people in the service area of the food bank to operationalize this variable.

The support from corporations and individuals in the form of in-kind donations constitute the largest portion of income of food banks. The fundraising efforts help food banks raise this money that is subsequently put into the food distribution programs. Hence, the amount of fundraising activities of a food bank, measured by fundraising expenses, is expected to impact the amount of food distributed by food banks. Accordingly, we obtain information regarding fundraising expenses from the IRS 990 forms and consider it as a control variable. On average, the food banks in the sample spent $740,668 in fundraising. Organizational size has been shown to influence performance, since large organizations have more resources through which the performance could be strengthened (Tsai, 2001). To control for this effect, we include “total assets” of food banks in the model. The information regarding the total assets of food banks was obtained from the IRS 990 forms. The average value of total assets in the sample is $12.1 million. Yiu et al. (2005) report that the longer the history of an organization, the greater the organization's embeddedness in its environment. Moreover, older organizations are more experienced in their areas of operations (Lukas et al., 1996). Accordingly, we control for the “age” of food banks as reported in their survey responses. The mean age of the food banks in the sample is 30 years.

We present the descriptive statistics and inter-item correlations of the variables employed in the model in Table 2.

Seemingly unrelated regression (SUR) is an econometric analysis method that allows for simultaneously running a system of regression equations and accounts for correlated error terms across the variables (Autry and Golicic, 2010). Zellner (1962) introduced this method as an efficient estimation of generalized least-squares models, where the variables that are independent in one equation can be a dependent variable in another equation in the system (Autry and Golicic, 2010). As SUR has the power to account for contemporaneous cross-equation error correlations, it has advantages over other approaches such as path modeling. It has been shown to be an appropriate methodology when multiple equations are simultaneously tested and when there is a chance that variables in the models are related to each other (Autry and Golicic, 2010). Autry et al. (2010) reports that SUR is an effective method for estimating models depicting mediating and/or moderating conditions using cross-sectional data. This technique is also known to alleviate endogeneity concerns (Autry and Golicic, 2010), since possible correlation between error terms are accounted for and the focal variables can be modeled as both independent and dependent within the model (Greene, 2003, p. 340). There are several other methods, such as structural equation modeling (SEM), that are used to estimate similar models. However, the ratio of the number of responses to the number of parameters being estimated would be quite low in the case of SEM, as each measurement item coefficient would need to be estimated. Although a set of 71 responses out of 202 food banks targeted is quite comprehensive, it is not sufficient to estimate a complete SEM. Thus, due to this reason along with the advantages discussed above, SUR was deemed as the appropriate methodology for this study.

In general, non-normality of the error terms and heteroscedasticity occur together in the data (Kutner et al., 2005, p. 132). We performed the Shapiro–Wilk test and did the necessary logarithmic and square root transformations to ensure that the variables satisfy the normality requirements (Kutner et al., 2005, p. 132). SUR models assume that the error terms are homoscedastic. We tested for the assumption of errors with constant variance (homoscedasticity) via the Breusch–Pagan test (Kutner et al., 2005, p. 118). The result of this test indicated that the error variances are not constant. Cameron and Trivedi (2009, p. 160) propose that bootstrapping can be used in conjunction with SUR when the error terms are heteroscedastic. This method allows us to get robust standard errors, and in the case that the error terms are homoscedastic, the results converge to the default standard errors (Cameron and Trivedi, 2009). We used this methodology for estimation by using the default bootstrap option.

We also checked whether multicollinearity was a problem in the data by examining the variance inflation factors (VIF). A VIF value in excess of 10 is generally considered as an indication of multicollinearity being an issue influencing the least squares estimates (Kutner et al., 2005, p. 409). The VIF values were all below 3 indicating that multicollinearity is not a concern in this study. The system of equations that are simultaneously estimated using SUR methodology are specified as:

Consistent with recent recommendations for conducting mediation analyses (Rungtusanatham et al., 2014; Malhotra et al., 2014), we conduct a significance test for the indirect effect and report both the size and confidence interval for the indirect effect. Also, as per the recommendations, the Monte Carlo simulation method with bootstrapping is used to test the mediation effect (Rungtusanatham et al., 2014). The RWeb code was used to get the sampling distribution using a Monte Carlo simulation (Selig and Preacher, 2008) since Monte Carlo simulation has been highlighted as a preferred technique by Malhotra et al. (2014).

As shown in Table 3, supply integration is positively associated with internal integration, thereby supporting hypothesis H1 (β = 0.222; p < 0.05). The results also show that demand integration is positively associated with internal integration, lending support for hypothesis H2 (β = 0.501; p < 0.05). We do not find support for the hypotheses H3 and H4; that supply integration and demand integration are positively associated with the number of basic programs in food banks (β = −0.076 and β = 0.115, respectively). However, internal integration is positively associated with the number of basic programs (β = 0.481; p < 0.05) and the number of basic programs is positively associated with the amount of food distributed per food insecure people (β = 0.389; p < 0.05), lending support for H5 and H7, respectively.

The results of our mediation analysis are presented in Table 4. The significance tests of the indirect effects are used to test for the mediation hypotheses. While the indirect effect of supply integration (β = 0.107) is non-significant, we find that the indirect effect of demand integration (β = 0.241; p < 0.05) is significant. In addition, as per the recommendation for the Monte Carlo method for assessing mediation effects (Malhotra et al., 2014; Rungtusanatham et al., 2014), the 95% confidence intervals for the indirect effects of supply integration [0.001, 0.269] and demand integration [0.030, 0.527] do not include zero, thus partially confirming the mediation hypotheses H6a and H6b.

The control variables - fundraising expenses, total assets, and age of food banks were not found to be significantly related to the food distributed by food banks (β = −0.110, β = −0.015 and β = 0.173, respectively).

This study responds to a call for further research in slow onset disasters that has received limited attention in the humanitarian logistics and supply chain management area (Kunz and Reiner, 2012). The impact of supply chain integration on performance has been well studied in the literature (Ataseven and Nair, 2017; Mackelprang et al., 2014). There are certain glitches in supply chains that result from disconnects, which create variation and waste in the system. Lee et al. (1997) discuss the bullwhip effect, which takes place due to lack of information sharing between the partners along the supply chain and results in excessive inventory, poor forecasts, costly damage control and inadequate service. Therefore, utilization of quality information becomes critical in all organizations, and more so in not-for-profit organizations due to the scarce resources that they need to depend on. Supply chain integration has been a point of interest as a countermeasure for inefficiencies in the system, such as the bullwhip effect. Koufteros et al. (2005) shed light on the new product development process in relation to internal and external integration with the contingency approach. They claim that when complexity and uncertainty in the environment escalate, organizations search for ways to increase integration and knowledge sharing to process information more effectively to come up with new products and ultimately achieve operational and financial performance. This brings the information processing theory into play in terms of why integrating the supply chain is needed and how it can help the organizations deal with the complexity in the environment. This idea has been investigated within the context of commercial supply chains (Wong et al., 2011).

The results of the current study provide important insights into the way supply chain integration impacts performance in a not-for-profit setting. While empirical investigations pertaining to supply chain integration have almost entirely focused on for-profit enterprises, our findings show some new results pertaining to the relationship between supply chain integration and performance in a humanitarian context. This study extends the existing research by presenting the interplay between external and internal integration. Specifically, we show that, unlike the for-profit sector, where managing internal integration forms the important foundation for integrating with suppliers and customers (Tracey, 2004; Braunscheidel and Suresh, 2009; Horn et al., 2014), the reliance of food banks on external partners necessitates external integration to precede internal integration. This is a novel finding and the implications of this finding should be understood within the context of food bank operations. Moreover, given the complexities of the humanitarian context presented in this paper, we propose that the interplay between the integration dimensions act as a structural mechanism (Koufteros et al., 2005) to create new programs and serve the communities in need by attaining the social goal of delivering food. When one considers the types of services that food banks offer to communities in need, it is clear that coordination and accurate information are especially crucial for the perishable items to arrive at their destination in a timely manner. This enables the donations to reach the clients on time without being wasted. Wong et al. (2011) argue that the lack of supply chain integration will increase the likelihood of distorted supply and demand information, which in turn causes poor delivery reliability. Additionally, there is empirical evidence for the importance of suppliers and customers as information sources, especially in complex environments such as joint product development, task coordination, problem solving and cost reduction (Koufteros et al., 2005; Lee et al., 1997; Stank et al., 1999; Flynn and Flynn, 1999; Wong et al., 2011). Within the food banking context, knowing the capabilities of suppliers and the needs of clients enables food banks to coordinate their own operations internally in order to develop the programs so as to achieve the food delivery targets.

The exchange between food banks and their supply chain partners could manifest as organizing joint promotional events, developing contractual service arrangements, and crafting more evolved forms of relationships. These activities require an understanding of mission and sharing of human resources among organizations that are part of food banks' supply chains. Strategic interactions within this advanced stage of collaboration among food banks and their supply chain partners may extend beyond financial resources to include sharing of skilled manpower and best practices. In an integrated supply chain, food banks and their partners are part of a significant coordination effort in which reciprocal exchanges help foster such activities as team building, leadership development, project management, financial planning, marketing management and human resource management. These form the foundation for internal integration to take shape. Also, corporate executives serve on the board of food banks, providing guidance and strategic directions for internal integration (Austin, 2000). Similarly, a food bank's close collaboration with food distribution agencies also provides directions for structuring internal activities as they often provide manpower for executing internal activities. In sum, with the help of supply and demand integration, food banks mobilize and combine multiple resources and distinctive capabilities to foster internal integration. From a food bank's perspective, this knowledge sharing mechanism is a crucial enabler for structuring internal activities.

As mentioned earlier, basic programs of food banks are similar to the variety of products and services in a commercial setting. While there are mixed results regarding the relationship between product variety and performance in the literature (Ramdas, 2003; Patel and Jayaram, 2014), we hypothesized a positive relationship between the number of basic programs and food distribution. Our reasoning is that by offering a program catering to kids, a food bank works with schools and reaches out to this segment to make this program possible. Likewise, senior brown bag programs are catered for older populations and they take place in coordination with volunteers. Therefore, the supply chain integration materializes the basic programs, which ultimately become the mechanism by which the food is distributed. Consistent with the theory suggesting that a wider product portfolio helps in increasing market share and customer satisfaction (Kahn, 1998), our results show that the number of basic programs developed by food banks is positively related to the food distributed per food insecure individual in the target population. A higher number of programs cater to the specific needs of various segments of the underprivileged population. Similar to the market segmentation strategy, these programs explicitly recognize the existence of various segments of the population that put demand on food banks' resources and allocate resources among these segments (Frank et al., 1972; Mahajan and Jain, 1978). This organization-wide responsiveness to market needs, which is widely considered to be an important tenet of marketing orientation (Kohli and Jaworski, 1990), improves the performance of food banks as manifested in terms of its ability to distribute food to food insecure individuals.

Our study presents some insights in terms of managing humanitarian supply chains. First, we provide empirical support for the precedence of external integration to internal integration in food banks. This finding implies that food bank managers should focus on external integration initiatives in order to gain momentum in their system and then work on internal integration to form basic programs to deliver food to underserved communities. Second, the number of basic programs is positively associated with the amount of food distributed in the service area of the food bank. Therefore, in order to increase the outreach, food banks should figure out new and innovative programs to deliver food to various segments of the food deprived populations. In fact, some food banks do not only deliver food in their service areas, but also provide clothes, stationary items for kids and education programs to the communities in need. This shows that food banks are already striving to come up with creative outreach programs, and they should be encouraged to continue this way to relieve the hunger issue as well as to meet some other social goals to improve the welfare of the populations in their service areas.

There are few limitations of this study that provide directions for future research. While our study theorizes a causal link from external integration to internal integration, we do not consider the specific time period when these integration efforts were in effect within the food banks surveyed. To more clearly discern the causal chain, a longitudinal study of integration events and their impact on a food bank's performance would offer valuable insights. In this study, we theoretically argue that complementary assets, knowledge routines and governance structures resulting from internal integration within food banks is more effective when internal integration is built on supply and demand integration. An explicit consideration of knowledge assets and governance structures in food banks as a result of the interplay between external and internal integration would help in advancing this theory. Our study considers the number of basic programs by treating each program to be identical in terms of resource requirements. Consideration of the actual resources needed to develop each program and weighing them accordingly will strengthen the validity of the framework presented in this study. Our study focuses on food distributed per food insecure individuals within the area of food bank operation. It may be worthwhile to capture perceptual satisfaction measures from the population served by food banks to obtain alternative measures of food bank performance. Finally, since the data was collected over the 2012–2013 time period, a more recent dataset might be worthwhile to analyze the supply chain integration in food bank organizations.