SRM refers to the collaboration between the firm and its suppliers to maintain the flow of needed materials, resources, parts, and information for the purpose of producing goods or providing services efficiently and effectively [6] [7] . SRM plays a vital role in shaping the FP and involves managing all interactions with third party firms that supply goods or services to a firm in order to maximize the value of these interactions [1] [2] . In practice, SRM requires creating closer and more collaborative relationships with key suppliers in order to reveal and realize new value and reduce risk of failure [6] .

SRM involves a set of practices undertaken by a firm to boost effective SCM. According to [16] (p. 606), “the ultimate goal of supplier management practices is to achieve better performance in the supply chain”. [5] described the latest evolution of SRM practices, which include mainlining communication with major key suppliers, establishing long-term contracts, and pursuing joint investments with suppliers. While, [17] identified four aspects of SRM, namely sharing information, developing collaborative approaches, joint decision making, and system coupling with key suppliers. [18] used many activities to represent SRM practices. These activities are sharing information with suppliers through a coordinated information system, working with key suppliers in product and process innovation, working with them for developing new technologies, and involving them in the implementation of firm’s logistics network. Additionally, [19] viewed SRM in terms of supplier selection, supplier development, and supplier performance review. Moreover, [20] represented SRM through supplier base reduction, effective buyer-supplier relationship, communication, cross-functional teams, and supplier involvement. Consequently, the literature depicts SRM practices from a variety of different perspectives with a common goal of ultimately improving FP. After reviewing and consolidating the literature, four distinctive practices were developed. They are supplier selection, supplier involvement, communication, and buyer-supplier relationship. Table 1 lists these practices along with their definitions and supporting literature.

Lean is based on the idea of improving efficiency of the process by eliminating waste [11] . Any activity that uses resources but does not create any additional value is a waste should be removed [27] . [28] defined lean as a set of practices focused on cost reduction by elimination of non-value-added activities from the firm’s operations. In brief, lean can be defined as achieving more outputs with less inputs. Although lean concept came from manufacturing sector, it is now

being applied widely in service sector [9] [29] [30] . Thus, lean can be applied in every business and every process. It is not only a tactic or a cost reduction program, but also a way of thinking and acting for an entire firm [11] [27] .

Regardless of establishing what lean is, it remains important to establish how to become lean. This can be achieved by a set of synergistic managerial practices called lean practices. These practices include but are not limited to SRM, continuous improvement, just-in-time (JIT), total quality management (TQM), work teams, employee involvement, and cellular manufacturing. Many scholarly works (e.g., [10] [11] [31] [32] [33] [34] ) offered a presentation of the most frequently identified lean practices in literature. Other works (e.g., [8] [10] [11] [28] [35] [36] ) supported the positive impact of these lean practices on FP. On the other hand, a considerable stream of literature (e.g., [9] [13] [30] ) has focused on synthesizing lean with other management practices.

This study synthesizes lean and SRM paradigms to constitute a lean-based SRM system for Jordanian manufacturing firms. LSRM is a systematic approach that integrates effective supplier relationship with lean practices to drive inefficiencies out of the firm. To apply LSRM paradigm, firms need to adopt a set of LSRM practices. LSRM practices can be defined as a set of SRM practices in a firm to boost its efficient SCM. They can also be defined as a set of supplier-related activities that are focusing on eliminating waste by reducing inventory and improving quality in the supply chain. Until now, a debate about determining the LSRM practices remains indecisive [11] , as there is no clear list of LSRM practices. Therefore, a wide literature on numerous publications has been reviewed in order to develop an exhaustive list of the main LSRM practices. Four practices were found among the most commonly used SRM practices in lean settings, namely supply flexibility (SF), JIT delivery (JD), information integration (II), and supplier partnership (SP). These practices with their descriptions and appearance in literature are listed in Table 2.

FP is a set of managerial and critical methodologies that allow firm managers to attain one or more pre-selected goals [21] . Until now, there is neither consensus nor agreement among scholars and practitioners about the definition of FP and its elements [21] [30] [46] . Thus, there are several definitions and indicators used to measure it [6] [21] [30] [47] [48] .

The relationships between the considered LSRM practices and their influence on FP are presented in the following subsections.

In order to explore the association between LSRM practices and FP, a proposed research model that combines all of the aforementioned hypotheses is developed. This model, shown in Figure 1, represents a structural equation model (SEM) that needs to be studied. LSRM practices and FP constitute the construct latent variables of this model.

The model proposes that LSRM practices have a positive relationship with FP. It also includes five main elements; the four LSRM practices, namely SF, JD, II, and SP (i.e., independent variables) and FP (i.e., dependent variable). Furthermore, there are a set of arrows that depict the interactions between these five elements. Note that arrows having two piles indicate that the relationship between the two construct latent variables is in two directions. Each arrow is also represented by its associated hypothesis presented in the previous section. In summary, the model investigates the relationships between LSRM practices themselves in one side and their impacts on FP from the other side. Besides that, this model differs in several ways from other models in literature. Unlike earlier studies that focused on examining the relationship between supply chain integration

and FP (e.g., [6] [17] [26] [71] ), this study looks more closely at the SRM, which is part of supply chain integration. In examining FP, some researchers focus on either financial performance (e.g., [3] [72] ) or non-financial performance of the firm (e.g., [4] [17] [24] [71] ). This study takes both of them into account [73] . Consequently, this study presents the first research model that illustrates the effect of LSRM practices on each other and explains their effect on FP.

Based on the literature review, relevant questionnaire items were drafted in order to measure the construct latent variables of the research model. To improve the understanding of its content, the resulting draft was reviewed by five academic professors in the field of logistics and SCM at the German Jordanian University, and seven executive managers from different manufacturing firms. Then, some modifications were made according to their notes and recommendations. The items were ensured to be written in such a way that reduces the likelihood of misunderstanding the overall practice by the respondent. Thereafter, the finalized version of the questionnaire was valid to be used.

In its final form, the questionnaire comprised two main sections; the first section included questions about the firms’ type and size, while the second section of the questionnaire contained 50 follow-on accurate and specific items about the implementation level of each LSRM practice and FP. The respondents were asked to evaluate each item using five-point Likert scale anchored from 1 = not adopted to 5 = fully adopted, and centered at 3 = partially adopted. The next subsections present the construct latent variables of the model and how they were measured in the questionnaire with supporting references.

The population of this research consisted of all manufacturing firms in Amman, the capital city of Jordan, that are listed in the Amman Chamber of Industry (ACI). According to [80] , there are 1200 manufacturing firms that are located in

Amman and they are classified into 10 industrial sectors, namely therapeutics, plastic and rubber products, chemicals, engineering and electrical industries, furniture, construction, printing and paper, food and supply, garments, and mining. As recommended by [81] , the sample size that is suitable for this population is 292. Our questionnaire was sent to 300 Jordanian manufacturers who were willing to participate in this study, after being contacted by email and phone. The selected sample was taken from all the typical industrial classifications distinguished by [80] to ensure a full representation of manufacturing sector in Jordan [9] . At the end, a total of 162 valid responses were received, making a response rate of 54%. This percentage goes in line with several other studies that were conducted within the area of lean management and used a similar distribution methodology (e.g. [9] [30] [82] ). Moreover, the authors of this study administrated and aggressively followed up the questionnaire distribution process themselves. Their involvement contributed to the delivery of the questionnaire to as many respondents as possible and helped in clarifying any ambiguity concerning the definitions or any other issues related to the questionnaire. The characteristics of the selected sample are shown in Table 8.

Managers in the top or middle levels who have a direct contact with suppliers and those with responsibilities related to SCM activities were asked to participate in the questionnaire. Hence, respondents included general managers, supply chain managers, operations managers, plant managers, purchasing managers, and others. It is worth mentioning that a pretest pilot study was conducted before distributing the questionnaire to the targeted firms in order to test the questionnaire items for their clarity, comprehensiveness, and acceptability. As suggested by [83] , a random sample of 25 firms was selected for this pilot study. Eighteen responses with a response rate of 72% were received. The received responses were carefully studied and analyzed, and the required adjustments were made.

Multicollinearity of the dataset is measured by the coefficient of correlation. According to [84] , multicollinearity indicates the degree by which questionnaire items measure the same entity. The coefficient of correlation ranges from −1 to 1, and a value of 0.9 or above for the inter-item correlations indicates the possibility that two or more items measure the same entity. Zero correlation means that there is no relation between the items. Table 9 shows the correlation matrix for all construct latent variables of the research model. It is worth mentioning that the correlation analysis was conducted between the mean of answers for each construct latent variable. Since there is no value above the threshold of 0.9, it is confirmed that there are no multicollinearity type problems in our model.

Validity of the research instrument can be confirmed by ensuring both content and face validity [79] . To ensure content validity of the research instrument, questionnaire items were drafted based on multiple sources of data (i.e., various scholarly works mentioned in the literature review). To ensure face validity, the resulting draft was reviewed by a panel of judges of academic professors and executive managers as mentioned earlier. Consequently, research instrument validity was assumed.

**Correlation is significant at the 0.01 level (two-tailed).

Reliability includes internal consistency of the construct latent variables [9] . It is measured using the Cronbach’s alpha (α) coefficient, which gives a reflection of the degree to which different items complement each other and measure the same overall concept [85] . The closer Cronbach’s α coefficient to 1.0, the greater internal consistency of the items in the questionnaire [85] [87] . Although values above 0.7 are often considered to be acceptable [86] [87] , a Cronbach’s α value of 0.8 is mainly the target [87] . A value of 0.6 or less indicates unsatisfactory consistency reliability [84] . Table 10 shows that Cronbach’s α values for all construct latent variables and the overall LSRM practices are more than 0.7. Consequently, research instrument reliability was also assumed.

Model fitness refers to the degree to which the SEM matches the observed data. For the purpose of this study, we used Chi-square (χ²) test, relative chi-square ratio (χ²-ratio), root mean square error of approximation (RMSEA), comparative fit index (CFI), and t-test to measure the model fitness.

While, a high value of χ² means that there is a poor fit between the observed and expected frequencies, a small value of χ² indicates a good model fitness. Due to the effect of sample size, Chi-square test is not always a clear assessment tool of the model fitness. Thus, various kinds of fitness indices that are independent of the sample size have been developed. According to [88] , the χ²-ratio is computed through dividing the χ² value by the corresponding degrees of freedom (DF). If the value of χ²-ratio is less than 5, the proposed model will be accepted [88] . Table 11 shows that the χ²-ratio for all hypotheses is less than 5, which means that the proposed model can be accepted.

RMSEA can also be used to measure the model fitness. RMSEA value of 0.05 or less means that there is a close fitness of the model in relation to the DF, and the models with RMSEA value of 0.10 or more have poor fitness [87] . Table 11 shows that all RMSEA values of the proposed model are less than 0.05. Thus, the results of RMSEA test support our model fitness.

The fitness index that is least affected by sample size comparing with other indices is CFI [89] . It ranges from zero for a poor fitness to 1.0 for a good fitness. However, a value of 0.9 is considered as an indicator for good fitness. This index

is derived from comparing the hypothesized model with the independent model, and it provides a measure of the complete co-variation [90] . Table 11 shows that all CFI values are very close to 0.9, which means that our proposed model has good fitness.

The t-test was used to measure the statistical significance of the relationship between SEM variables at a specific level of significance (α = 0.01). In this test, the t-value computed by a specific equation or using a statistical software (referred to as t₀) is compared with the value of t indicated in the t-distribution table (referred to as t_α,DF in one-tailed test and t_α/2,DF in two-tailed test) [87] [91] . In our two-tailed test, if |t₀| > t_α/2,DF, this indicates that there is a significant relationship between the two variables [91] . Since the absolute value of all t-values presented in Table 11 exceeds the values indicated in the t-distribution table, all developed hypotheses are supported.

The estimated strength (i.e., standardized regression weights) of each relationship between all construct latent variables are presented in Table 11 and shown in Figure 2. Improving or harming any one of the four LSRM practices will affect each other, and will have an impact on FP according to the estimated strength of each relationship among them. This implies that none of them should be ignored or given less attention. It is clear that SF practice has the highest impact on FP with a regression weight of 0.89, followed by II practice with a regression weight of 0.83, then JD with a regression weight of 0.81, and SP practice has the lowest impact on FP with a regression weight of 0.64.

Table 12 shows a descriptive analysis comprises finding the mean ( x ¯ ), variance (S²), and implementation index for each construct latent variable. It also shows that the selected sample of Jordanian manufacturing firms implements the considered LSRM practices by 50.58%. The highest implementation level is in SF practice with a percentage of 65.11%, while the lowest implementation level is in JD practice with a percentage of 32.31%.

Figure 3 draws the relationship between the impacts of LSRM practices on FP using the estimated strength of the relationships listed in Table 11 versus the implementation index of these practices listed in Table 12. The horizontal dashed line represents the average of the implementation index values of the overall LSRM practices, which is 50.58%, and the vertical dashed line represents the average impact of these practices on FP, which is 0.79. These two dashed lines divide the diagram in terms of impact on FP (either high or low) and

implementation index (either high or low) into four quarters: high impact on FP—high implementation level, low impact on FP—high implementation level, low impact on FP—low implementation level, and high impact on FP—low implementation level.

From the classification shown in Figure 3, it can be concluded that the greatest opportunity for improving Jordanian firms’ performance should be focused on aspects related to JD and II practices due to their high impact on FP and low implementation level by Jordanian manufacturing firms. However, SF practice should not have high priority by Jordanian manufacturing firms because its implementation level is the highest among other practices (i.e., 65.11%). In addition, Due to the fact that Jordanian firms maintain good relationships and deep connections with their suppliers [15] , SP practice has the second highest implementation index (i.e., 63.27%) after SF practice. SP practice should have the lowest priority by Jordanian manufacturing firms because its estimated impact on FP is the lowest among other practices (i.e., 0.64).

Although t-test can be used for hypotheses testing, measuring the p-value for each relationship can also be used to test hypotheses. If p-value is less than α then there is a significant relationship between the studied variables [91] . At the 0.01 level of significance, Table 11 shows that p-value for each relationship is less than α. This leads to the same results of the t-test that support all of the developed hypotheses.

ANOVA is one of the popular techniques of hypotheses testing [87] . It is a method of decomposing the total variability in a set of observations, as measured by the sum of the squares of these observations from their average, into component sums of squares that are associated with specific defined sources of variation [91] . Table 13 shows a two-tailed one way ANOVA test of the developed model at the 0.01 level of significance. It is clear that f₀-values exceed the value of f_0.005,v1,v2 for each LSRM practice. This means that all LSRM practices have positive and significant relationships with FP.

The findings of the study at hand corroborate with previous studies on the positive and significant impact of the ordinary SRM practices on FP (e.g., [1] [3] [6] [18] [49] [64] ). The study findings are also consistent with previous studies that explored the role of adopting SRM practices in lean settings and contexts (e.g., [38] [53] ). Moreover, our findings go in line with the previous results of several national (e.g., [79] [92] ) and international studies (e.g., [8] [28] [35] [36] ) that revealed a positive and significant relationship between lean practices and FP. Taken together, the general assumption in literature is that either lean practices affect FP or SRM practices affect FP, without investigating the effect of both of them combined on FP. This study improves our understanding of how integrating both SRM and lean paradigms could be translated into improved FP. Consequently, the finding that LSRM practices have a positive and significant relationship with FP is an important contribution to the logistics and SCM literature.