Despite growing organizational adoption of CSR and TBL initiatives, a critical theoretical and empirical gap persists regarding how these organizational strategies translate into employee behavioral change and organizational outcomes in public service contexts. While organizations formally articulate CSR-TBL commitments encompassing social, environmental, and economic sustainability, the psychological and behavioral mechanisms through which individual employees perceive, internalize, and act upon these values remain inadequately theorized and tested in local government settings ([1]). This microfoundations gap is significant because the actual implementation of CSR-TBL initiatives, and consequently their impact on service quality and citizen satisfaction, fundamentally depends on employee-level cognition, motivation, and behavioral responses ([8]).

The study integrated several complementary theoretical perspectives. Social Exchange Theory and Psychological Contract Theory explain how employees reciprocate credible organizational fulfillment of obligations with attachment and discretionary effort ([9]). Social Identity Theory suggests that alignment with ethical and sustainability-oriented values strengthens identification and commitment ([25]; [11]). The Job Demands-Resources perspective and engagement research link organizational resources and meaningful work to vigor, dedication, and absorption ([17]; [34]), while Psychological Capital theory positions self-efficacy, hope, resilience, and optimism as personal capacities through which CSR perceptions influence outcomes ([2]; [3]; [7]). Public Service Motivation theory and resource-based considerations such as budget adequacy provide contextual conditions specific to government organizations ([28]; [30]). Prior TBL research further establishes the organizational-strategy pathway that this study extends to the employee level ([22]; [16]; [5]; [20]).

Drawing on these perspectives, twelve hypotheses were formulated. H1-H3 proposed that CSR-TBL organizational alignment has significant positive direct effects on organizational commitment (H1), employee engagement (H2), and psychological capital (H3). H4-H6 proposed that psychological contract fulfillment mediates the positive relationships between CSR-TBL alignment and organizational commitment (H4), employee engagement (H5), and psychological capital (H6). H7 proposed that organizational commitment mediates the relationship between employee engagement and task performance, and H8 that organizational commitment mediates the relationship between psychological capital and organizational citizenship behavior. H9 proposed that engagement and commitment sequentially mediate the relationship between CSR-TBL alignment and task performance. H10 proposed that organizational commitment and employee engagement function as parallel mediators between psychological contract fulfillment and task performance. Because the structural model also specifies a path from employee engagement to organizational commitment (H7 and H9), these two mediators are causally ordered within the full model; H10 therefore refers to the joint operation of the specific indirect pathways transmitted through each mediator, rather than to strictly independent parallel mediation. Finally, H11 proposed that public service motivation moderates the CSR-TBL alignment-commitment relationship, and H12 that organizational budget adequacy moderates the CSR-TBL alignment-psychological capital relationship. All hypotheses were tested at the 0.05 level of significance.

The study employed an exploratory quantitative research design to investigate employees’ perceptions of CSR-TBL integration and their relationships with organizational commitment, engagement, psychological capital, and performance within LGUs ([22]; [36]; [2]). Given the relatively underexplored nature of combining CSR-TBL constructs with employee-centered outcomes in the public sector, the exploratory approach enabled the examination of complex interrelationships and the refinement of theoretical linkages within the proposed model ([26]; [7]). PLS-SEM was adopted because it is well suited for analyzing complex frameworks with mediation and moderation effects in service and public-sector contexts ([3]; [24]).

The respondents were 319 employees of LGUs across Region VIII (Eastern Visayas), Philippines, drawn from multiple organizational levels, departments, and service areas ([35]; [14]). A convenience sampling method was employed to recruit participants who were available and willing to participate during the data collection period ([17]; [25]). The minimum sample size was determined using G*Power with an effect size of 0.15, a significance level of 0.05, and a power of 0.95, yielding a target of approximately 200 participants plus a 20% buffer; the achieved sample of 319 exceeded this requirement ([11]; [6]). By administrative jurisdiction, respondents were assigned in Leyte (38%), Samar/Western Samar (20%), Eastern Samar (13%), Southern Leyte (13%), Northern Samar (10%), Biliran (5%), and other districts (1%).

The sampling frame comprised the 136 municipal LGUs across the six provinces of Region VIII (Eastern Visayas), which were formally approached through their local chief executives and administrators; employees of the municipalities that granted authorization, covering all six provinces, participated in the study. Eligible participants were LGU personnel aged 18 years and above holding permanent (plantilla), casual, contractual, or job-order appointments, who had rendered at least six months of continuous service in their current LGU and were directly involved in administrative, technical, or frontline service functions; elected officials were excluded. Respondents were recruited through the human resource management offices of the participating LGUs, which disseminated the printed questionnaires and the digital survey link following the orientation sessions described in Section 3.4. Of the 350 questionnaires distributed, 319 valid questionnaires were retained after the data-cleansing procedures described in Section 3.5, yielding a usable response rate of 91.1%.

A structured questionnaire measured the nine constructs of the model through dimension-specific and global items adapted from validated scales and contextualized to LGUs: CSR-TBL Organizational Alignment (governance and ethics, stakeholder engagement, environmental responsibility, and social responsibility); Psychological Contract Fulfillment (authenticity, transparency, resource commitment, and stakeholder respect); Organizational Commitment (emotional attachment, value identification, willingness to contribute, and retention); Work Engagement (vigor, dedication, and absorption); Psychological Capital (self-efficacy, hope, resilience, and optimism); Task Performance (role execution, quality achievement, and service delivery); CSR-Related Organizational Citizenship Behavior (advocacy, sustainability participation, community service orientation, and ethical citizenship); Public Service Motivation; and Organizational Budget Adequacy. Items were rated on a five-point Likert scale.

The construct measures were adapted from established, previously validated instruments: CSR-TBL Organizational Alignment items were adapted from employee-focused CSR perception scales ([35]) and TBL operationalizations ([22]); Psychological Contract Fulfillment items were adapted from psychological contract measures applied in public-sector engagement research ([9]); Organizational Commitment items drew on the affective commitment tradition of [23]; Work Engagement items were adapted from the Utrecht Work Engagement Scale ([33]); Psychological Capital items were adapted from the Psychological Capital Questionnaire ([18]); Task Performance items were adapted from the Individual Work Performance Questionnaire ([15]); CSR-related Organizational Citizenship Behavior items followed established OCB operationalizations ([27]); and Public Service Motivation items were adapted from [29] scale. Adaptation followed a structured process: items were reworded to reflect LGU terminology, mandates, and service contexts; the draft instrument was reviewed for content and face validity by a panel of experts in public administration, human resource management, and research methodology; and it was pilot-tested with 30 LGU employees who were excluded from the final sample, with all scales attaining pilot Cronbach’s alpha coefficients above 0.70. The instrument was administered in English, the official working language of Philippine LGUs, with orientation sessions clarifying terminology. During measurement-model purification, items were retained if they exhibited indicator loadings of at least 0.70, contributed to construct AVE of at least 0.50, and showed no substantive cross-loadings, reducing the initial pool of 158 items to the 40 indicators reported in Table 1.

All nine constructs were specified as first-order reflective measurement models, consistent with their conceptualization as latent perceptions reflected in, rather than formed by, their indicators ([12]; [32]). Although CSR-TBL Organizational Alignment is conceptually multidimensional (governance and ethics, stakeholder engagement, environmental responsibility, and social responsibility), it was not estimated as a higher-order construct; following item purification, it was modeled as a unidimensional reflective construct whose retained indicators (CSR11-CSR15) capture employees’ overall perception of alignment across these dimensions. The reflective specification is appropriate because the indicators are expected to covary as manifestations of a common underlying perception, the removal of an indicator does not alter the conceptual domain of the construct, and the constructs satisfy the internal-consistency and convergent-validity criteria associated with reflective measurement reported in Table 1 ([12]). The same reasoning applies to the remaining multidimensional constructs, whose retained indicators function as reflective manifestations of the overall construct rather than as defining components of it.

A formal transmittal letter was addressed to regional directors and LGU executives to secure authorization, followed by orientation sessions with participating employees and administrators. Participation was voluntary; respondents were informed of the confidentiality of their responses and of their right to withdraw at any time, and written informed consent was obtained prior to participation. Data collection occurred over a three-week period using both printed questionnaires for geographically remote offices and digital forms for offices with reliable connectivity. Completed instruments were reviewed for completeness, anonymized, and coded before analysis.

Responses underwent a rigorous cleansing procedure covering completeness checks, sincerity screening through reverse-scored items and low-variability flags, duplicate removal, logical-consistency checks, and outlier examination. Frequency counts and percentages described the respondents, and weighted means summarized construct perceptions. PLS-SEM using WarpPLS assessed the measurement model (indicator loadings, Cronbach’s alpha, composite reliability, Average Variance Extracted, the Fornell-Larcker criterion, and the Heterotrait-Monotrait ratio) and the structural model (model fit and quality indices, R², full collinearity VIF, Q², effect sizes, and bootstrapped confidence intervals), consistent with recommended PLS-SEM reporting practice ([13]; [31]).

Females constituted 53% of the respondents, and the largest age group was 30 - 39 years old (43%). Nearly half held a bachelor’s degree (47%), while the largest government-service group had served for one to three years (25%) and the largest current-LGU tenure group had less than one year of service (30%). Most respondents held permanent or civil service appointments (79%), and specialist or professional employees in non-supervisory positions formed the largest organizational-level group (34%). Administration and finance was the most represented service area (25%), followed by community development (22%). The largest LGU-size categories were organizations with 101 - 250 employees (30%) and 251 - 500 employees (29%), and most respondents were assigned in rural areas (58%).

Respondents reported favorable perceptions across all nine constructs. CSR-TBL Organizational Alignment was rated Strongly Agree (M = 4.35, SD = 0.78), with its dimensions ranging from 4.29 to 4.43. Psychological Contract Fulfillment (M = 4.12, SD = 0.88) and Organizational Commitment (M = 4.12, SD = 0.85) were rated Agree. Employee Engagement received the lowest construct mean but remained favorable (M = 4.00, SD = 0.92), with vigor at 4.13, dedication at 3.91, and absorption at 3.81. Psychological Capital was rated Agree (M = 4.18, SD = 0.78). Task Performance (M = 4.17), Organizational Citizenship Behavior (M = 4.15), Public Service Motivation (M = 4.36), and Organizational Budget Adequacy (M = 4.27) were likewise favorably perceived, with Public Service Motivation emerging as the highest construct-level mean.

Table 1 presents the indicator loadings, reliability, and convergent-validity results after item purification. The 40 retained indicators produced loadings from 0.701 to 0.884. Cronbach’s alpha ranged from 0.710 to 0.913, composite reliability from 0.763 to 0.935, and AVE from 0.518 to 0.742, satisfying the conventional thresholds for indicator reliability, internal consistency, and convergent validity ([13]; [31]).

Table 1. Indicator loadings, reliability, and convergent validity.

Table 2 presents the Fornell-Larcker results. The square roots of AVE on the diagonal (0.720 to 0.861) exceeded all corresponding inter-construct correlations; the largest off-diagonal correlation was 0.696 between Organizational Commitment and Employee Engagement. Table 3 presents the HTMT ratios, which ranged from 0.050 to 0.770 and remained below the 0.90 criterion, with the highest ratio between Organizational Commitment and Employee Engagement. Both criteria therefore supported the discriminant validity of the nine constructs.

Table 2. Discriminant validity using the Fornell-Larcker criterion (square roots of AVE on the diagonal).

Table 3. Discriminant validity using the HTMT ratio of correlations.

Table 4 presents the model-fit and quality indices. The Average Path Coefficient (0.124, p = 0.002) and Average R-Squared (0.130, p < 0.001) were significant, collinearity indices remained well below their limits, and the Tenenhaus goodness-of-fit of 0.279 indicated medium overall model adequacy. The causality-quality ratios satisfied their respective minimum criteria.

Table 4. Model fit and quality indices.

Table 5 shows that explanatory and predictive performance was uneven across the endogenous constructs. Organizational Commitment had the strongest explained variance and predictive relevance (R² = 0.491; Q² = 0.461), followed by Task Performance, Employee Engagement, and Organizational Citizenship Behavior, while Psychological Contract Fulfillment and Psychological Capital showed weak explanatory and negative predictive results. Full collinearity VIF values (1.135 to 2.223) indicated no critical collinearity.

Table 5.Coefficient of determination, full collinearity VIF, and predictive relevance.

To permit a complete evaluation of the hypothesized mediation mechanisms,Table 6 first reports all estimated direct structural paths in the model, including the antecedent path from CSR-TBL Organizational Alignment to Psychological Contract Fulfillment, the paths from Psychological Contract Fulfillment to the proposed mediators, and the paths from each mediator to the performance outcomes.

Table 6. Direct structural path estimates of the full model.

Note. N = 319. Coefficients for the five hypothesized paths (CSR → OC, CSR → EE, CSR → PsyCap, CSR × PSM → OC, and CSR × OBA → PsyCap) are as reported in Table 7. The remaining estimates were re-derived from the final cleaned dataset (N = 319) using the same measurement and structural specification; standard errors are bootstrap estimates (5,000 resamples) and p-values are one-tailed, consistent with the reporting convention of the original estimation.

Table 7 presents the results of the twelve hypotheses. Only H10 was supported: Organizational Commitment and Employee Engagement jointly mediated the relationship between Psychological Contract Fulfillment and Task Performance (indirect β = 0.087, p = 0.001, 95% CI [0.034, 0.145]). The direct paths from CSR-TBL Organizational Alignment to Organizational Commitment, Employee Engagement, and Psychological Capital were not statistically supported, nor were the isolated mediation paths through Psychological Contract Fulfillment and Organizational Commitment, the sequential engagement-commitment pathway, or the moderation effects of Public Service Motivation and Organizational Budget Adequacy, both of which had negligible effect sizes.

Because the structural model includes a path from Employee Engagement to Organizational Commitment (H7 and H9), the two mediators specified in H10 are causally ordered rather than strictly independent. Accordingly, the aggregate indirect effect reported for H10 (β = 0.087, 95% CI [0.034, 0.145]) was decomposed into its three specific indirect components—PCF → OC → TP, PCF → EE → TP, and the three-path effect PCF → EE → OC → TP—reported in Table 8 with bootstrapped confidence intervals. The decomposition shows that the indirect effect was transmitted primarily through the engagement pathway (PCF → EE → TP: β = 0.091, 95% CI [0.043, 0.150]), whereas the commitment-only component (β = 0.003) and the three-path component (β = −0.011) were small and not statistically significant, permitting direct verification of the central mediation claim.

Table 7.Results of hypothesis testing.

Table 8.Decomposition of the H10 indirect effect into specific indirect paths.

Note. Component effects were re-derived from the final cleaned dataset (N = 319) using the same model specification, with 5,000 bootstrap resamples, percentile confidence intervals, and one-tailed p-values. The sum of the re-derived components (0.083) corresponds to the aggregate indirect effect reported for H10 (β = 0.087) within re-estimation rounding.