The Incomplete Leap: On the Transition From Union Registration to the First Collective Agreement

Ramos, Vincent Jerald; Suguitan, Edgar Antonio Article — Published Version The Incomplete Leap: On the Transition From Union Registration to the First Collective Agreement Industrial Relations Journal Provided in Cooperation with: John Wiley & Sons Suggested Citation: Ramos, Vincent Jerald; Suguitan, Edgar Antonio (2024) : The Incomplete Leap: On the Transition From Union Registration to the First Collective Agreement, Industrial Relations Journal, ISSN 1468-2338, Wiley, Hoboken, NJ, Vol. 56, Iss. 3, pp. 189-203, https://doi.org/10.1111/irj.12458 This Version is available at: https://hdl.handle.net/10419/323804 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/4.0/ Industrial Relations Journal ORIGINAL ARTICLE The Incomplete Leap: On the Transition From Union Registration to the First Collective Agreement Vincent Jerald Ramos 1,2,3 | Edgar Antonio Suguitan 4 1 Hertie School, Berlin, Germany | 2 Dynamics Research Training Group, Humboldt University Berlin, Berlin, Germany | 3 ESRC Centre for Population Change, University of Southampton, Southampton, UK | 4 Department of Economics and Political Science, University of the Philippines‐Baguio, Baguio, Philippines Correspondence: Vincent Jerald Ramos ([email protected]) Received: 20 August 2024 | Revised: 14 November 2024 | Accepted: 20 November 2024 Funding: This research benefited from funding from the Philippine Competition Commission's Long‐Term Research Program and the corresponding author has received funding from the German Research Foundation (DFG), grant number 390285477/GRK2458. Funding from the UKRI Economic Social and Research Council (ES/W002116/1), through the Centre for Population Change‐Connecting Generations project, is gratefully acknowledged. Keywords: collective bargaining | event history | union density | unionization ABSTRACT In decentralized systems where unionization and bargaining occur at the establishment level, what explains the (speedy) transition from union registration to the first collective agreement? While prior evidence in some developed countries (i.e., the United States) estimates this transition to be approximately a year, on average, little to no evidence exists in other contexts where unions play a less central role, as is the case in many developing countries. Addressing this gap, we analyze the Philippines where national unionization and collective bargaining coverage rates are relatively low. Using methods from survival analysis on novel register data of all new union registrations from 2016 to 2021, we descriptively demonstrate that: (i) median union membership density is low at 40%; (ii) only 20% of all new union registrations successfully register a contract within the first 5 years; and (iii) unions with higher densities, that are independent, and in the manufacturing sector are associated with elevated transition rates to a collective agreement. JEL Classification: J51, J52, C14 1 | Introduction In decentralized systems where bargaining occurs at the establishment level, unionization is a necessary but insufficient milestone towards workers' representation. While the implications of decentralized systems on an array of outcomes have been studied extensively (Katz 1993; Ortigueira 2013; Calmfors et al. 1988), the dynamics and transitions between union registration and the signing of the first collective agreement, especially in developing country contexts, have received little to no attention thus far. Contrary to the European experience with more established industrial relations institutions, unions in Southeast Asian countries, including the Philippines, exert weaker influence in setting wages and employment conditions, which was exacerbated during the COVID‐19 pandemic (Ford and Ward 2021). In such contexts, a puzzle arises—what are the ‘enabling conditions’for unions to secure a collective agreement? This paper descriptively fills this gap and contributes to our understanding of this transition. How many unions make the ‘successful leap’from unionization to a collective agreement? Which conditions facilitate (and accelerate) this transition? While bargaining dynamics are arguably heterogeneous and shaped by firm‐specific and other exogenous factors (i.e., negotiation styles), some structural explanations might be This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2024 The Author(s). Industrial Relations Journal published by Brian Towers (BRITOW) and John Wiley & Sons Ltd. 189 of 275Industrial Relations Journal, 2025; 56:189–203 https://doi.org/10.1111/irj.12458 relevant risk factors. For instance, unions with higher densities (e.g., the share of the workplace who are union members) might experience higher transition rates to a collective bargaining agreement (CBA) relative to those with lower densities. Intuitively, higher densities offer unions stronger leverage in negotiations as the threat of action (e.g., strikes, work stoppages) is presumably larger. In neighboring Indonesia, firms are legally bound to bargain in good faith after a union is registered or after a certain threshold of internal support is met (Pralitasari, Feby, and Hamid 2023). As such, ‘size’(or density) is arguably critical when studying union success. In the Philippines, private sector unions are either independent and self‐governing or a local chapter of a larger national federation. Regardless, the bargaining process and the coverage of any resulting agreement are contained within establishments (Bitonio 2012). One could argue that federated unions may benefit from vertical and horizontal coordination and are thus more likely to succeed in bargaining (Brandl and Braakmann 2021; Organisation for Economic Co‐operation and Development 2019). On the contrary, independent unions likewise benefit from greater autonomy and flexibility as well as insulation from political ‘malignment’that is pervasive among Southeast Asian countries (Ford and Ward 2021). This paper focuses on a specific and underexplored outcome— the duration between union registration and the signing of the first collective agreement. In the United States, Combs (2021) estimates this transition to be a little over a year (465 days or approximately 1.27 years). Note that this is in a context where unions must be sizable enough to win an election and firms are de jure obligated to bargain in good faith. To the best of our knowledge, there is little to no empirical evidence addressing these questions in other contexts thus far, owing to the unavailability of extensive register data. Bridging this gap, we collect and compile novel register databases of all new union registrations and collective agreements in the Philippines from 2016 to 2021. This database contains limited but useful information about the firm and union names, broad industry categories (1‐digit level of the International Standard of Industrial Classification), geographic region, year of registration and number of members and total employees. This allows for a systematic and population‐level analysis of the transition (both selection into and duration) to the first agreement. Using event history regression models on this matched register data, we analyze how relative transition rates (or hazard ratios) are associated with three structural features—union density, union type and industry classification. These three features proxy for organizational and institutional factors that prior literature has demonstrated to shape firm‐union dynamics. First, union density measures the intra‐firm strength and leverage of unions with higher density ones benefiting from broader workforce support. Second, union type indicates whether a union is independent or is a local chapter of a larger federation. Finally, the type of industry denotes whether the union is in the manufacturing sector. Considering that unions are relatively more common in this sector, it is interesting to test whether manufacturing unions are more likely to secure a collective agreement. Previewing our findings, we highlight three notable results and contributions. First, to the best of our knowledge, this paper is the first to empirically utilize duration models in studying the transition from unionization to collective bargaining in any context. These methods are best suited for research questions where the outcome is the ‘time‐to‐event’or the duration between two states while accounting for right‐censoring. Second, we descriptively demonstrate extremely low CBA registrations— only 1 in 5 of the unions that have registered since 2016 have successfully registered a CBA as of 2021. Finally, our empirical analyses reveal that high‐density, independent and manufacturing sector unions have robustly elevated transition rates to the signing of the first CBA. Taken altogether, these results substantively suggest an ‘incomplete leap’towards workers' representation in the Philippines. The majority of new union registrations do not secure a CBA and those that were successful took years to negotiate. Unions without CBAs do not benefit from contractually binding wage increases and other nonwage entitlements. In effect, the gains of these ‘contractless unions’are inconsistent and limited at best or nonexistent at worst. Our paper emphasizes the need for industrial relations policies aimed at improving workers' representation to go beyond promoting unionization, as has been the practice in the Philippines. Instead, supporting unions throughout the negotiation process will ensure that unionized workers eventually reap tangible and enforceable gains. 2 | Philippine Context As a former colony of the United States, the Philippine industrial relations landscape is heavily patterned after the American National Labor Relations Act of 1935 (Bitonio 2012)— unionization and bargaining are highly decentralized at the establishment level. This is contrary to other systems where ‘supra’trade unions or federations play a larger role, as is the case in Brazil for example (Maia 2019). In relatively centralized systems, horizontal coordination of unions occurs through federations (i.e., ‘DGB’in Germany). In theory, centralized bargaining systems function as if there is one ‘supra‐union’ (Dittrich 2008), whereas decentralized systems devolve negotiations at the firm or plant level, with firm‐based unions operating independently from one another (Dittrich 2008; Bitonio 2012). Usual analyses highlight some inefficiencies brought about by centralized bargaining, particularly on wages and productivity (Calmfors et al. 1988). Ortigueira (2013), Wallerstein and Western (2000) and Katz (1993) noted that centralized bargaining grew increasingly common in many industrialized societies between 1950 and 1980, but have since declined, with decentralized structures being favored. These systems are not mutually exclusive and, in practice, most landscapes are de facto ‘hybrid’. Indeed, in some countries, organized decentralized structures—where sector level agreements exist but only for broad frameworks for firms to follow— combine sectoral coordination with firm‐level flexibility for bargaining outcomes (Organisation for Economic Co‐operation and Development 2019) which could yield some efficiencies. Brandl and Braakmann (2021) find evidence that vertically coordinated or multilevel bargaining systems could lead to better 190 of 275 Industrial Relations Journal, 2025 performance and labor productivity in European Union (EU) countries. However, this is contingent on the extent of coordination between bargaining units across the different levels (firm, sector, country), and whether surrounding institutions can foster such coordination. It is perhaps more appropriate to think of centralized decentralized systems as a spectrum rather than a binary. 2.1 | Comparisons and Key Differences How does the Philippines' industrial relations landscape differ from its neighboring Asian countries? While unions in Asia generally possess weaker influence in the political sphere compared to those in Europe (Kuruvilla et al. 2002; Frenkel and Kuruvilla 2002; Kuruvilla and Venkataratnam 1996), certain intra‐regional features are notable. In Malaysia, government policies prohibit industrial or sector‐level unions thereby resulting in a ‘de facto’decentralization at the firm‐level as (Freeman 2010; Kuruvilla and Venkataratnam 1996). Meanwhile, transnational regulatory channels enhance and complement state regulations in Indonesia, consequently improving the enforcement of and compliance in minimum standards (Amengual and Chirot 2016; Caraway and Ford 2017). Finally, an interesting case is also that of Vietnam, where wildcat strikes—‘unauthorized’employee activity beyond failed bargaining—influence state policy and the regulation of labor standards and bargaining (Anner and Liu 2016). These features (e.g., transnational actors, wildcat strikes and restrictive government policies) are arguably less pronounced, if at all, in the Philippine context. Despite these differences, these developing Southeast Asian countries are characterized by common issues of fragmentation, declining union membership and support and large informal sectors (Asa 2024; McKay 2006; Viajar 2009; Kuruvilla et al. 2002; Kuruvilla and Venkataratnam 1996). As noted above, labor regulations in the Philippines were patterned after the US, considering the strong influence of postcolonial American political and economic institutions (Bitonio 2012; Hutchison 2016). At present, one key difference exists in terms of union recognition. In the Philippines, union elections are not held nor mandatory before collective bargaining. In the United States (Bhuller et al. 2022), union recognition consists of two main steps—first, workers sign union petitions and if they reach a majority of all employees, can seek voluntary recognition directly with the firm. Otherwise, they can file a petition to hold an election with the National Labor Relations Board (NLRB) if at least 30% of the workers sign or the firm declines to recognize. Second, after winning the elections, the firm must bargain in good faith over working conditions. 1 In any case, majority support is guaranteed before bargaining. In the Philippines, a union can register with the Bureau of Labor Relations with as little as 20% of employees signing up as members and could therefore bargain with the firm despite minority support. 2 Historically, union interest, measured through the median union density (union members/employees) has been at a steady persistent decline. While there are certainly new union registrations every year, their relative sizes have been shrinking. Figure 1shows that the median union density is initially characterized by some data noise 3 after the Philippine independence of 1946, followed by a relatively stagnant phase until 1990. Since then, government efforts to ‘maintain industrial peace’occurred alongside pursuing flexibilization policies (Sale and Sale 2014). From 2016 to 2021, our period of study, median union density is around 40% but reached a historic low at 27% in 2021, owing to the COVID‐19 pandemic. As of 2022, union membership and collective bargaining coverage rates (as a share of total employment) in the Philippines are at historic lows at around 6.5% and 8.4% of total employment, respectively (Philippine Statistics Authority 2024). Finally, another challenge that confronts unions in the Philippines at present is the persistent ‘red‐tagging’, extra‐judicial killings and disappearances of union leaders and other organized workers. Based on recent reports from the Human Rights Watch (2024), 72 union leaders and members have been killed since 2016. They have also been publicly attacked under FIGURE 1 | Median union density by year of registration. Source: Own estimates using the union register data from the Bureau of Labor Relations. 191 of 275 accusations of being linked to communist groups as sympathizers or insurgents. These labels have been used to justify violence and crackdowns, putting unionists in active danger (Human Rights Watch 2024). Arguably, national federations or larger unions are more susceptible to these risks. 3 | Literature Review and Hypotheses This section begins with a recognition that unionization, in and of itself, is a necessary but insufficient condition towards workers' representation. Since firms inherently possess outsized power relative to workers (e.g., using threats of termination and plant shutdowns) (Kaufman 2012; Bronfenbrenner 2009), unions have historically countered this asymmetry through the signing of binding contracts, consequently improving earnings and working conditions (Flanagan 1990; Hirsch 2008; Kristal 2023). Drawing from existing empirical literature, we develop three hypotheses around the structural conditions that may facilitate the transition to collective agreement. 3.1 | Does ‘Size’Matter? As discussed above, a key peculiarity in the Philippine case is that unions can register and seek recognition upon obtaining at least 20% of the employee support. Hence, it is plausible and common to see relatively ‘small unions’(as a share of the firm's workforce) bargain with the firm. This poses a problem, as low‐ density unions certainly possess less leverage. Indeed, having more organized workers leads to larger political demonstrations and overall influence in public sentiment (Nguyen‐Van and Terraz 2021; Kallas, Park, and Aleks 2023). Within firms, higher union densities can strengthen how unions can formalize workplace governance as more workers participate in the union's activities and the bargaining process (Edralin 2016; Kaufman 2012). Prior work has argued that decentralization plays a role in the decline of overall and intra‐firm union density as employers take a more adversarial and hostile stance to limit the effects on their production process that their competitors might not be similarly exposed to (Cutcher‐Gershenfeld and Kochan 2004;Schnabel2013). Bronfenbrenner (2009) noted that adversarial employer tactics such as threats of shutdowns, surveillance and termination, among others, impede unionization. Consequently, lower‐density unions have a more limited bargaining power and a more protracted contract settlement (Schnabel 2013;Nguyen‐Van and Terraz 2021). In terms of contract settlement, Kallas, Park, and Aleks (2023) find empirical evidence that higher union density is positively associated with the probability of settling a first contract within the first 24 months of union formation. These results echo prior findings that low union density is associated with low and protruded growth in bargaining coverage, despite the reported number of workers who are members of unions (Bitonio 2012; Lamarche 2015). In the Philippine context, Hutchison (2016) points out that it's not just that union densities are low, but also that most establishments are small. Hence, the decentralized nature of bargaining per se constrains workers' associational power. Thus, we expect that low‐density unions have less bargaining leverage and find it more difficult to successfully reach an agreement. We deduce this expectation from the US experience whereby the size of the bargaining unit (initial signatories of the union cards) is associated with the election outcome of the union and their eventual bargaining success (Ferguson 2008; Bronfenbrenner 2009). As such, we test the hypothesis that: H1: (density). Unions with higher membership densities have an elevated transition to the first contract agreement. 3.2 | Does ‘Union Type’Matter? Private sector unions in the Philippines are classified as either of two types—‘independent’or a ‘local chapter’of a larger federation. Could we expect that federation chapters are more likely to secure a collective agreement than independent unions (or vice‐versa)? We draw on prior literature on union coordination and networks to develop our theoretical expectations. On the one hand, federations possess advantages in terms of being formal channels of horizontal and vertical coordination. Federations are more actively and directly engaged in networks with external parties including political and socio‐civic organizations, and can more easily mobilize movements and protests (Kallas, Park, and Aleks 2023; Oskarsson 2003; Anner, Fischer‐ Daly, and Maffie 2021; Paolucci and Roche 2024b). In the case of Ireland, ‘confederations’became formal conduits for horizontal and vertical coordinated bargaining and informal inter‐ union mechanisms existed to coordinate strategies in negotiations (Paolucci and Roche 2024a). Greater mobilization efforts increase the pressure on firms to heed their demands (e.g., stronger effects of work stoppages) (Kaufman 2012). Kuruvilla et al. (2002) and Kuruvilla and Venkataratnam (1996) stated that in many Asian countries (i.e., Japan, Korea and the Philippines), there have been attempts to strengthen federations in pursuit of union revitalization, although the Philippines has notably lagged in this regard. Furthermore, federations and networks, in general, allow unions to connect with informal workers through socio‐civic organizations as they are often restricted from union membership (Hutchison 2016; Viajar 2009). Absent the channels of federation networks, unions might not be able to coordinate and consolidate the calls of those involuntarily left in temporary or informal work arrangements—‘temps’or ‘contractuals’ (Litwin and Shay 2022; Asa 2024). In the Philippines, informal workers can only form workers' associations with no bargaining capacity, but they are also allowed to affiliate with federations, hence the usage of networks by unions (Bitonio 2012). Indeed, in developing countries where informal employment is rampant (Organisation for Economic Co‐operation and Development 2019; Lamarche 2015), unions can utilize federation networks to broaden their organizational relevance and to the larger labor movement, as is the case in the Philippines and Indonesia (Hutchison 2016; Serrano 2018; McKay 2006; Caraway and Ford 2017). Being part of a larger federation also allows stronger coalition‐building through coordination, 192 of 275 Industrial Relations Journal, 2025 union‐to‐union ‘learning’and mutual support, which can be useful during the bargaining stage (Cardullo, Conti, and Sulis 2020; Ronnmar et al. 2023; Anner, Fischer‐Daly, and Maffie 2021; Gumbrell‐McCormick and Hyman 2019). The context of globalization has also underscored the need for workers to consolidate their calls to action through transnational networks, as these power resources can also be leveraged in contexts where the state offers weak labor regulations (Amengual and Chirot 2016; Locke, Amengual, and Mangla 2009; Hutchison 2016; Kuruvilla and Li 2021; Jackson, Kuruvilla, and Frege 2013). Union federations in Asia have, to some extent, utilized these networks to drive better union coordination, like, for example, being springboards or bases for specific benefits and bargaining outcomes (Buiser 2021; Kuruvilla et al. 2002; Frenkel and Kuruvilla 2002), however, there is still room for improved embeddedness of Asian unions compared to those in other developed countries like in Europe (Kuruvilla et al. 2002). On the other hand, independent unions, being stand‐alone entities are also advantageous, as their bargaining strategies and positions are more flexible, targeted and autonomous (Hendricks and Kahn 1982), with little influence by federations. In liberal market economies such as Ireland, the flexibility of independent and firm‐level unions is associated with expanded union membership and enthusiasm as well as securing more tangible benefits (Paolucci and Roche 2024b). Federations, as entities, are comparable to ‘second‐tier’unions (Bitonio 2012), which means that workers are subsumed under two layers of representation—the local federation chapter and the national federation leadership. This additional layer might be disadvantageous and might negatively influence the bargaining process. One example is how federations might be embroiled in national controversies or ‘red‐tagged’as sympathetic to certain political minorities (e.g., left‐wing militant groups), in which case their local chapters might suffer from marginalization and a weaker bargaining leverage. Indeed, Bitonio (2012) reports that CBA coverage by independents is higher than that of federation affiliates or chapters. In the Philippines, federations are supra‐level networks composed of at least 10 local chapters or workplace‐level unions. Federation chapters accounted for 60% of all CBAs in 2010−11 (Bitonio 2012). Note that in the Philippines, it is possible that a federation chapter and an independent union both operate within the same establishment (Bitonio 2012). While federations engage with external parties and participate in social movements, lobbying, protests and other political demonstrations atthenationallevel(Hirsch2008; Oskarsson 2003), the locus of bargaining remains at the workplace level. This ‘network advantage’might aid the bargaining process and expedite the first collective agreement. Buiser (2021) noted that unions can better coordinate when they have access to networks offered by federations or transnational organizations (i.e., unions for seafarers in the Philippines benefitting from transnational networks of seafarers). Indeed, Kallas, Park, and Aleks (2023) find empirical evidence in the United States that coalition building and public rallies are associated with a higher likelihood of settling first contracts within plants. There is no conclusive evidence from the literature on whether federation chapters or independent unions have an accelerated transition to the signing of a first contract and the preceding theoretical considerations may point to either direction. If the ‘network advantage’of federation chapters outweighs the ‘flexibility advantage’of independent unions, one can expect the former as more likely to secure a contract. Along these lines, we hypothesize that federation chapters have access to more information and network resources that can aid in the bargaining process such that: H2: (union type). Local federation chapters have an elevated transition to the first contract agreement compared to independent unions. 3.3 | Cross‐Sectoral Differences We know from prior literature that unionization and collective bargaining outcomes differ across sectors. Given the historical embeddedness of labor movements in the industrialization of developed countries, it is unsurprising that unions are more strongly present in the manufacturing sector to this day (Martin and Brady 2007). Scheuer (1997) notes that the sectors with ‘collectivist legacies’like manufacturing, construction and transport have higher union membership and bargaining coverage than the distributive trades of the services sector. As such, manufacturing is seen as the ‘traditional union stronghold’(Schnabel 2013,259). Relatedly, the global decline in unionization is heavily attributed to structural transformation—the shift in employment composition from manufacturing to services, where unions grapple to expand (Schnabel 2013;Hirsch2008). Among developing countries, industrialization is expected to increase unionization and bargaining coverage rates if their development trajectories follow that of today's industrialized and high‐income societies (Martin and Brady 2007). However, the economic expansion that many developing countries have been experiencing, including the Philippines, has skipped the massive expansion in manufacturing in favor of directed growth in services (Brady 2007; Serrano 2018). In the Philippines, manufacturing remains the traditional stronghold for unions, although it has also experienced declining membership rates over time (Bitonio 2012; Serrano 2018). Kuruvilla et al. (2002) noted that strong anti‐union behavior is pronounced in certain sectors, particularly in the large export‐oriented electronics industry where firms deem unions as deterrents from investors. In particular, both the Philippines and Indonesia have experienced strong institutional barriers that inhibit union coordination within export‐oriented sectors (e.g., marginalization of workers engaging in strike activity) (Caraway and Ford 2017; Hutchison 2016; Viajar 2009; Kuruvilla et al. 2002). Erickson et al. (2003) noted that in the service sectors (e.g., transport, utilities and hospitality), firms have demonstrated a greater willingness to shut down operations, whether temporarily or permanently. This fundamentally challenges and hampers union success in the services sector. Our theoretical expectation is rather intuitive. If union activity is more robustly present in one sector, the chances of ‘union success’in terms of securing a collective agreement might also 193 of 275 be higher there. Recognizing the historically‐oriented and strong presence of unions in the manufacturing sector which may facilitate the bargaining process, we test the hypothesis that: H3: (sector). Manufacturing sector unions have an elevated transition to the first contract agreement relative to unions in nonmanufacturing sectors. 4 | Data and Methods 4.1 | Data and Analytical Sample Enterprise‐based unions (or workplace‐level unions) account for a vast majority of all unions in the Philippines. As of 2024, there are around 18,000 private‐sector unions, accounting for about 90% of all unions, with the rest being in the public sector (Bureau of Labor Relations 2024). Note that union registration in the Philippines does not expire, hence, it is likely that many of these unions may well be inactive. Of all unionized workers, only 18% are currently covered by a collective agreement. This large gap between union membership and bargaining coverage indicates many unions do not or cannot successfully form bargaining agreements at the enterprise‐or establishment‐level. To further probe the transition between union registration and collective agreement, we use two novel register data of new union registrations and CBAs between 2016 and 2021 from the BLR (n= 858). The union registration data set contains basic information about the union and the firm, the number of members and employees, as well as general information about the industry and geographic area. Meanwhile, the CBA data set contains agreements registered between 2016 and 2021, regardless of when their union was registered. Therefore, we merge these datasets using both firm and union names, resulting in a union‐level data set with an identifier of whether a CBA was eventually signed. This unique data set is a novelty in the literature as we are able to estimate ‘population‐level’ transitions rather than transitions among a select sample of unions. Our key variables of interest are: (1) whether or not a union has secured a CBA; (2) if yes, how long this transition took (in years); (3) union type (federation chapter or independent union); (4) density (high‐density vs. low‐density); and (5) sector group (manufacturing vs. nonmanufacturing). For additional context, we define union density as the share of union members relative to the total number of employees. As a stratifying variable for the transition rates, we define high density as unions with membership densities higher than 50% and low density for those at and below 50%. 4.2 | Event History Analysis As we are interested in the duration between two distinct states, we utilize empirical tools from event history analyses to estimate hazard ratios. This method accounts for right‐censoring in our data set, where the event has not yet occurred at the time of observation but could occur in the future. In our study, we only cover the unions that registered between 2016 and 2021. Therefore, a union that registered in 2020 might not have yet secured a CBA by the end of 2021, but it doesn't mean that it can't secure one at a later point. Event history models account for this ‘temporal intensity’. When right‐censoring is not addressed, as will be the case when conventional logistic regression models are used, this may bias the estimation of the ‘true’ hazard (Keiding and Andersen 2015). 4.2.1 | Survival Curves As a first‐pass and visual test for differences in transition rates, we employ the standard nonparametric estimation defined by Equation 1(Kaplan and Meier 1958), which gives the survival probabilities over time.  ≤ ∼ ∼                          s tD YX ()= 1−() Xt i i i (1) The observation times are denoted by X˜ i ,{i=1,…,n}, where each X˜ i captures the time of survival X i (with a CBA) or censorship (without a CBA). For each observation, D i is a dummy that indicates whether the union has a CBA. Meanwhile, Y(X˜ i ) captures all the censored unions at time tthat may still negotiate a CBA (e.g., ‘at‐risk’population) (Kaplan and Meier 1958; Kleinbaum and Klein 2012). When comparing survival curves, hypothesis testing is conducted using log‐rank tests (Peto and Peto 1972; Prentice and Marek 1979). 4 We stratify the survival curves by union density, union type and broad sector group, our three main variables of interest. 4.2.2 | Hazard Ratios As an extension to visually presenting the Kaplan−Meier survival curves, we employ a standard Cox proportional hazards (Cox PH) model that controls for additional confounders, namely the geographic location (island group) and the scope of the bargaining unit. Our estimands of interest are hazard ratios, the exponentiated coefficients from the regression model, and could be interpreted as the ‘hazard’(e.g., having a CBA) for one group relative to a reference group. Consider the model:  () ht h t βX()= ()×exp ii0(2) where the β i coefficients are transformed to hazard ratios. 5 Hazard ratios will show the difference in transition rates for independent, high‐density and manufacturing unions, relative to their reference groups. One important assumption of the Cox PH model is that of proportional hazards, which means that the relative hazard functions do not change over time (Keiding and Andersen 2015; Grambsch and Therneau 1994). We demonstrate an empirical test for this in the Appendix. 6 Although stemming from epidemiological literature, the Kaplan−Meier survival function and Cox proportional hazards model have been utilized to understand the duration of certain 194 of 275 Industrial Relations Journal, 2025 events in the social sciences (Quiroz Flores 2022). These models have been used to estimate labor market and industrial relations phenomena such as unemployment duration after college graduation in Vietnam (Trang et al. 2024)and the effects of early conflict on the duration of bargaining relationships and the probability of dissolution of bargaining relationships for private sector firms in Ontario (Weinberg 2018). We build on this strand of the literature that uses event history models to analyze durational outcomes given the presence of censoring. 5 | Results and Discussion 5.1 | Descriptive Results We first show the trends in union registration and collective agreement over time. Figure 2shows that indeed, a vast majority of new unions have been unable to secure a collective agreement. Of the 858 unions in our analytical sample, only 16% (or 139) had managed to successfully register a CBA. Table 1shows the composition of all union registrations, by our key explanatory variables of interest—union density, type and sector. Due to missing values, the total number of observations for union density is slightly lower. These are further disaggregated by the CBA indicator (with CBA or no CBA), with a corresponding relative percentage count for the CBA indicator. This is to descriptively show how many events and censored cases we have across different levels of our explanatory variables. To aid the interpretation of this table, among all new union registrations in our data set with information about union density, 59% are classified as ‘low density’(50% and below membership). Among low‐density unions, only 39% have secured a CBA compared to 61% of high‐density unions. Focusing more closely on union density, it is interesting to note that there is a difference in the distribution for those with and without CBAs. For the entire sample, we find that the median union density is around 40%. Stratifying by those with and without CBAs, Figure 3shows that those without CBAs have a more right‐skewed distribution, indicating that more unions without CBAs have relatively lower densities compared to those FIGURE 2 | Union and CBA registrations. CBA, collective bargaining agreement. TABLE 1 | Summary statistics. Explanatory variable Frequency % With CBA Rel. % No CBA Rel. % Union density 769 119 650 Low density 452 58.8 46 38.7 406 62.5 High density 317 41.2 73 61.3 244 37.5 Union type 858 139 719 Federation chapter 644 75.1 75 54.0 569 79.1 Independent union 214 24.9 64 46.0 150 20.9 Sector 858 139 719 Nonmanufacturing 432 50.3 48 34.5 384 53.4 Manufacturing 426 49.7 91 65.5 335 46.6 Abbreviation: CBA, collective bargaining agreement. 195 of 275 that have one. 7 Interestingly, unions that settled a CBA had a median union density of 56%, and those without were at 36%, providing the first descriptive support for one of our hypotheses that (high) density accelerates first contract settlement. 5.2 | Survival Curves Figure 4shows the Kaplan−Meier survival curves for the entire data set (Panel A) as well as disaggregated curves by density, type and broad sector group (Panels B−D). Formal hypotheses testing using log‐rank tests indicate significant differences in survival probabilities by union density, union type and broad sector group (see Table A1 in the Appendix). Note the curves all depict a consistently ‘low transition rate’where none has reached s(t)<0.5. This is expected as we only observe 139 CBAs from 858 new union registrations. Consequently, there is no ‘median’time‐to‐event. Our data suggests that within the first 5 years after union registration, most unions (80%) do not have a CBA. Even high‐ density, independent and manufacturing sector unions do not FIGURE 3 | Union densities. FIGURE 4 | Kaplan−Meier survival curves. 196 of 275 Industrial Relations Journal, 2025 FIGURE A1 | Proportional hazards test (fully‐specified model). FIGURE A2 | Interaction models. 203 of 275