Purpose: Within the two-tier bargaining system, the role of complementary collective bargaining is somewhat controversial. In this paper, the authors analyse collective agreements from a triple perspective: scanning the contents of firm-level complementary collective agreements (CCAs); identifying the factors that determine the probability of signing a CCA and analysing the relationship between the latter and firm performance with a focus on the role of different negotiated topics. Design/methodology/approach: The empirical procedure is based on 2 main linked sources: longitudinal balance sheet data and a cross-sectional dataset of a representative sample of Italian firms with at least 15 employees, including some retrospective information. The innovative dataset derives from integrating multiple sources. The main empirical approaches include Generalized Method of Moments (GMM) estimations, multivariate regressions, as well as instrumental variable (IV) estimations to overcome simultaneity issues. Findings: With respect to the probability of signing a CCA, on the firms' side, the authors find a positive role of the degree of firm capitalisation and affiliation with an employers' association and a negative role of family firms compared to non-family firms; on the workers' side, a positive role of the workers' unionisation rate and a positive but differentiated weight of workers' union representations and industrial conflicts. With regard to firm performance, the authors' estimates suggest that signing a CCA is associated with an average increase of 3% in total factor productivity (TFP) and 7.8% in labour productivity. By investigating the contents of the complementarity agreements, the authors show that bargaining a wider range of topics implies advantages that are not homogenous, benefitting more efficient firms. Moreover, the authors find a specific positive and significant role for three main interacting issues: economic incentives, organisation and employment. Research limitations/implications: The cross-sectional structure of the data on bargaining practices prevents detecting causal relationships due to either potential common driver(s) of both the target variables (firm performance) and bargaining practices (simultaneity bias) and unobservable time-invariant firm-level characteristics (heterogeneity bias). Practical implications: According to the authors' results, policymakers should operate along four fiscal channels to spur the efficiency of firms, via CCA. First, tax incentives stimulate higher firm capitalisation, as this seems to be a CCA-favouring factor. Second, deduction in taxable income for union members, which should led to higher membership rates, hence raising the likelihood of obtaining a CCA. Third, incentives aimed at directly promoting the greater diffusion of CCAs as a source of improved performance. Fourth, fiscal tools aimed at favouring the negotiation of either specific contents or "bundles" of contents, which the authors' estimates show as an additional performance-enhancing tool of CCA practices. Originality/value: The conceptualisation of the contents of CCA as organisational investments and the whole probability function of signing a CCA are quite innovative. Moreover, the econometric strategy takes account of several potential sources of bias when estimating the relevant coefficients at each stage, which is currently not fully considered in the literature. Finally, this is the first study to shed light on both the diverse outcomes associated with different negotiated topics (in terms of quantity and quality) and the distinction between short and medium-long term effects. [ABSTRACT FROM AUTHOR]