On the Usefulness of Combining Strike and Protest Research: Some Insights from the Spanish Case (2000-2016)1

During the past half century, the study of social contention has been characterized by a division between research on labor movements and studies on other social movements. This division also left its mark on the study of modes of action: while labor scholars mainly examined strikes, social movement scholars have increasingly come to focus on street protests. This article is a contribution to bridging the gap between the two research areas both on theoretical and empirical levels. On a theoretical level, I discuss the usefulness of combining economic and political models of contention from the two research areas. On an empirical level, I use official data provided by Spanish ministries to examine and relate the workers’ use of strikes and street protests between 2000 and 2016 in Spain. Examining strikes and street protests jointly does not only provide a fuller picture, it also helps to discern contrasts and thus the specificities of each mode of action. 1 This work was supported by the Swiss National Science Foundation grant P1LAP1_178066. I am grateful to Romain Carnac, Ahmed Chapi, Gilles Descloux, Olivier Fillieule and Philippe Gottraux for their invaluable and constructive comments on earlier drafts of this paper. I would also like to express my gratitude to the anonymous reviewers as well as to the editors of this special issue for their help and guidance in preparing the manuscript for publication. 22 N. POHL ON THE USEFULNESS OF COMBINING STRIKE AND PROTEST ... EMPIRIA. Revista de Metodología de Ciencias Sociales. N.o 52 septiembre-diciembre, 2021, pp. 21-52. ISSN: 1139-5737, DOI/ empiria.52.2021.31364


INTRODUCTION
Workers go on strike and paralyze production. They call for protests and take the streets. What is obvious to any trade union activist seems less evident within academia, where scientific specialization has resulted in a strongalthough weakening-division between strike research and protest analysis (Yon 2016). Scholars of the former research area mainly examine class divisions and workplace related conflicts, while scholars of the latter generally focus on street protests-and most often on those promoted by non-labor organizations. Since the turn of the millennium, an increasing number of scholars has come to challenge this separation in the study of labor and non-labor actors and their respective (ascribed) modes of action (Fantasia y Stepan-Norris 2004;Giraud 2009;Grote y Wagemann 2019;Yon 2016). Extending the study of labor contention to modes of action other than strikes seems particularly relevant with respect to the Spanish case, where workers and unions remain core players in the production of social contention in the streets (Luque y González 2017). 2 The present article is a contribution to further decompartmentalizing strike research and protest analysis. The article builds on previous studies on levels of labor contention in the recent history of Spain (Luque 2012;Luque y González 2017;Pohl 2018). However, it is novel in two ways. First, it broadens the study of labor contention by systematically including labor related street protests (i.e. those street protests called by workers, works committees or unions). The aim here is to provide a better understanding of how strike patterns relate to evolutions of labor protests in the streets. Second, it establishes a theoretical dialogue between strike research and protest analysis. Both research areas have in fact extensively dealt with factors expected to favor or hinder the emergence of contentious collective actions. Yet, the analyses have remained largely confined to the respective research object -either strikes or street protests. In this article, I compare the ways in which the two research areas conceptualize economic and political factors expected to affect the emergence contentious collective actions.
To be more precise, I deal with the Business Cycle Model and Grievance Theory regarding the economic dimension and the Political Exchange Model and the Political Opportunity Perspective for the political dimension. I discuss where the respective models coincide and where they diverge and subsequently examine their applicability to the Spanish case between 2000 and 2016.
The study draws on yearly aggregate data on strikes and labor related street protests between 2000 and 2016 in Spain. The data on strikes is provided by the Spanish Ministry of Employment and Social Security and the data on street protests by the Spanish Ministry of the Interior. The operationalization of the data is inspired by Shorter and Tilly's notion of "shapes of strikes" (Shorter y Tilly 1971;1974), which allows one to examine the data in relative terms, notably by calculating frequencies of strikes and street protests among a the working population.
The study has the great merit of examining relationally two modes of action that have often been kept separate. However, the analysis also has some limits, which have to be clarified from the outset. Relying on yearly aggregate data makes it virtually impossible to produce a fine-grained process-oriented analysis. Only very broad year-to-year changes can be examined. To this, one has to add the examined theoretical perspectives, which seek to establish general relationships between seemingly 'external' (economic or political) factors and the likelihood of contentious collective actions. As a consequence of the aggregate nature of the data and the focus on rather macrological theories, the multiple actors as N well as their interactions-or the precise micro-and mesological processes in the emergence of contentious collective actions-fade into the background. 3 The article is structured as follows: In the first section, I present a brief overview of the principal economic and political models used in strike research and protest analysis. The second section deals with the official strike and protest data as well as with its operationalization. The core section of the article deals with the applicability of the different theoretical models to labor related contention in Spain between 2000 and 2016. In a concluding section, I discuss how empirically and theoretically combining strike and protest research can produce further insights into the patterns of social contention-in Spain and beyond.

RELATING THEORIES ON INDUSTRIAL CONFLICT WITH PROTEST THEORY
Scholars of industrial conflict and social movement scholars have extensively examined ebbs and flows of social contention. These evolutions have often been associated with changes in the economic or political context. In this section, I present the main theoretical models that have emerged in the two research areas.
In strike research, the Business Cycle Model is the principal model relating the state of the economy with the occurrence of strikes. While the proxies for measuring the state of the economy have varied over time (Knowles 1952;Rees 1952;Rist 1912;Skeels 1971), the principal argument has essentially remained the same: the numbers of strikes are expected to increase in periods of economic boom and to decrease in times of economic downturn. Rees (1952: 381) explains why strikes are expected to be more frequent in periods of economic prosperity: Businesses risk losing market shares and thus become more vulnerable to work stoppages. Furthermore, as a consequence of increasing demand and decreasing supply of manpower, businesses face more difficulties in hiring new staff. The workers, for their part, can rapidly take on a new job if they are dismissed. Under such conditions, workers are likely to protest if their wages do not keep pace with economic growth. The author also explains why this process is expected to be reversed in periods of economic contraction. When the economy goes down, workers participating in strikes (and union activism more broadly) run greater risks of being laid off and simultaneously face greater difficulties in finding a new job. For the employers, on the other side, it becomes easier to substitute laid-off workers. Employers are also expected to be less vulnerable to strikes because of full inventories or because lost production can rapidly be made up. In such a situation, resorting to strike action is expected to be less effective and to entail greater risks for the workers, which is ultimately expected to lead to a decrease in numbers of strikes. In social movement studies, economic factors for contentious collective actions have played a more marginal role. The most explicit reference to economic cycles can be found in the Relative Deprivation Hypothesis (Davies 1962;Gurr 1970), which expects rebellions to occur when actual need satisfaction ('what is') increasingly diverges from expected need satisfaction ('what should be'). After a long period of disregard, the economic dimensions of social contention have regained scholarly attention in the aftermath of the Great Recession. Under the umbrella of Grievance Theory, scholars started to reconsider 'economic deprivation and grievances as drivers of protest' (Kriesi, Häusermann y Lorenzini 2020). At a micro level, scholars found a positive relationship between the individuals' perception of deprivation and their actual protest participation (Rüdig y Karyotis 2014). At a macro level, increasing (general) levels of deprivation have been shown to be positively related with protest activity (Kriesi, Häusermann y Lorenzini 2020;Quaranta 2015). In this perspective, economic downturns and the resulting levels of deprivation are seen as environmental cues for the politicization of the deprivation suffered at the individual level: 'a visibly deteriorated economic context is understood as providing fertile ground for the subjective understanding of one's deprivation as something that can be challenged and redressed collectively and politically' (Grasso y Giugni 2016: 675-676).
Comparing the Business Cycle Model and Grievance Theory, one finds contradictory assumptions regarding the relationship between the state of the economy levels of contention. The Business Cycle Model suggests increasing levels of contention when the economy does well, while Grievance Theory expects levels of contention to be higher when the economy goes down, and the sharper the downturn, the more likely protests become.
Regarding the political dimension of contention, the predominant model in strike research is the Political Exchange Model (Hibbs 1978;Pizzorno 1978). Introduced by Pizzorno (1978: 279), the notion of political exchange refers to the government's provision of goods (notably labor-friendly policies) in exchange for union support. The power of unions and workers in the political sphere is here not seen as a function of demand and supply (contrary to the Business-Cycle Model). Instead, it is expected to stem from the unions' capacity to withdraw the social consensus and to threaten the established social order. According to Pizzorno (1978: 292), conflict will replace political exchange 'when the unions are either unable or unwilling to exchange moderation for power, and either unable or unwilling to prevent new "entries" and the formation of new collective identities,' which are capable of challenging the political order without the backing of established unions.
Later scholars have mainly taken up former hypothesis, according to which the degree to which unions moderate their tactics largely depends on the bonds they have established with governments (Hibbs 1978;Korpi y Shalev 1979). In countries where bonds between government and unions are strong, the latter are expected to avoid the costs of strikes and to intervene in the political sphere, where they have good chances to attain their goals. Such strong bonds were found to be more likely in countries that combined high levels of state interven- tion with durable social democratic rule-e.g. Norway or Sweden. In contrast, in countries where these bonds are weak or absent-e.g. Canada, Ireland or the United States-unions have little possibilities to engage in a political exchange and conflicts are mainly fought at the company level and notably through strikes (Hibbs 1978;Korpi y Shalev 1979). The probability of an ongoing political exchange is affected by other political factors-most notably the presence or absence of a sizable Communist party (Hibbs 1976). Where Communist parties are strong they are expected to hamper or undo the social democrats' efforts in engaging in a political exchange with unions.
In social movement studies, political factors for contentious collective actions have mainly been examined in terms of political opportunities or political opportunity structures, which are expected to favor or hinder the emergence of social movements and contentious collective actions (Kitschelt 1986;Kriesi et al. 1995;Tarrow 1989). Synthesizing previous conceptualizations, McAdam (1996: 27) identifies four broad dimensions of political opportunities for social movements: (1) A polity's degree of openness; (2) the relative stability of alignments between different elite groups within a given political system; (3) the availability of influential allies (4) the government's use of repression against social movements.
Studies of single countries have come to focus on the more 'configurational' opportunities and notably on the degree of unity on the left, the left's presence (or absence) in government and the left's (majority or minority) role within the government (Kriesi 1995;2004). Regarding the unity of the left, the action space of labor movements is expected to remain large where the left is divided (Kriesi 1995: 181). Social democrats have to compete for hegemony on the left and therefore remain receptive to labor demands. In contrast, the action space of labor movements is expected to be narrower where the left is unified, since social democrats shift their attention to issues related to new challengers.
Regarding the participation in government, left-wing protest mobilization is expected to be higher under right-wing than under left-wing governments (Koopmans y Rucht 1995; Kriesi 1995). 4 Under right-wing rule, social democrats are expected to ally with social movements in mobilizations against the incumbent government. If the political left is in power, the opposite occurs: Social movements lose a powerful ally and their mobilization capacity is reduced. The social movements' mobilization capacity is expected to be further reduced by the social movements' anticipation of reforms, which should (at least) partially fulfill their demands. While protest levels are expected to be higher under right-wing than under left-wing governments, Koopmans and Rucht (1995) expect protest levels to be highest under mixed (i.e. coalition) governments. The right in government is seen to make external pressure 'necessary,' while the left in government makes protest 'opportune,' in the sense that protest will meet a receptive environment. What is more, the political left might even actively seek support among social movements in order to exert pressure on the coalition partner (Kriesi 1995: 182). The political exchange model and the political opportunity perspective converge on two arguments: a stable social democratic government is expected to reduce the levels of labor related contention, while the presence of a strong Communist party can potentially undo the pacifying effects of social democratic rule. The political opportunity perspective further examines the potential effects of mixed coalition governments and suggests that levels of left-wing contention could be even higher under a mixed than under a conservative government.
What strike and protest patterns can one expect when applying the above described economic and political models to the Spanish case? To be able to formulate some working hypotheses (see section 4), I now turn to the examined data and to their operationalization.

DATA SOURCES AND OPERATIONALIZATION
To examine the applicability of the different theoretical models to the Spanish case, I relate economic and political changes (as independent variables) with the evolutions of strikes and street protests (as dependent variables). For the state of the economy, I mainly rely on employment and unemployment data provided by the Spanish National Statistics Institute and the OECD. For the political dimension, I examine the degree of unity of the left, the presence or absence of the left in government and the (majority or minority) role within the government.
As dependent variables I examine strikes and labor-related street protests (i.e. street protests called by workers, works committees or unions). The data on strikes is provided by the Spanish Ministry of Employment and Social Security. To collect the data, the ministry applies a definition proposed by the International Labour Organization (1993: 2), which considers a strike as "a temporary work stoppage effected by one or more groups of workers with a view to enforcing or resisting demands or expressing grievances, or supporting other workers in their demands or grievances." All strikes are taken into account irrespective of whether or not they had been previously announced, as long as they exceed the duration of one hour. The data includes information about numbers of strikes, numbers of participants and numbers of working days lost due to strike actions. These variables are broken down according to different subcategories (e.g. regions, economic sectors, strike motives, strike outcomes). There are, however, some important gaps in the data. While the data includes information for the nationwide general strike in 2002, no information is reported for the three later general strikes (one in 2010 and two in 2012) or for one nationwide strike of the public sector in 2010. The data on the three general strikes could be completed by drawing on works by other scholars (González y Luque 2014; Luque y González 2016).
Regarding street protests, I use data provided by the Spanish Ministry of the Interior. This ministry publishes yearly aggregate data on the exercise of funda- mental rights such as the right of assembly and demonstration. The data considered here include rallies and marches in the public sphere. Regarding these two modes of action, the ministry distinguishes between different types of organizers, which makes it possible to isolate street protests called by workers, works committees or unions (here referred to as 'labor-related street protests' or simply 'labor protests'). For the period between 2000 and 2008, the ministry furthermore distinguished between announced, unannounced and forbidden street protests and also provided data on the total numbers of participants for each protest type.
In 2009, the numbers of participants were dropped for all types of protests and as of 2010, the unannounced demonstrations were also omitted. Over the time span under study, the ministry has thus become more selective regarding the data to be published on street protests, which leaves me with announced street protests as the only consistent value for the entire time span. The operationalization of strike and protest activity is inspired by Shorter and Tilly's concept of 'shapes of strikes ' (1971; 1974), which is intended to go beyond the absolute numbers provided by the ministries. Initially applied to strikes, the notion distinguishes between three vectors: First, frequency, which is calculated by the total number of events over the population under study. Second, size, which is defined as the total number of participants over the total number of events. Third, duration, which is specified as the aggregate amount of time that individuals spent on the events over the total number of participants. Based on these three vectors, two more dimensions can be defined: magnitude, which results from the multiplication of size and duration; and volume, which is equal to a population's total time spent on the events.
The following equation illustrates the 'shapes of strikes' (Shorter y Tilly 1971;1974):

Strike volume = strike frequency x strike size x strike duration
This conceptualization has the advantage of making visible various aspects of contentious collective actions, which absolute numbers do not reflect. The identical magnitude or volume can in fact vary in its shape. For instance, as shown in Figure 1, high strike size combined with low strike duration can produce the same strike magnitude (or strike volume) as its counterpart. Because the Spanish Ministry of the Interior provides numbers of protesters only until 2009 and no information about the duration of protests, this study will mainly compare the respective frequencies of strikes and labor protests.
The official data examined here are the most comprehensive about strikes and street protests in Spain. However, these official data sets are far from depicting an objective reality of social conflict in Spain. One reason for caution are the political purposes of official statistics (Camard 2002). Whether and how public ministries keep official records of certain social phenomena is in fact a result of political decisions. Regarding the Spanish case, the data on both strikes and street protests became more selective as of 2009. It seems reasonable to assume that political interests guided the decisions to omit information on the more contentious actions such as general strikes or unannounced street protests.
Technical aspects of the data collection procedure also lead to (sometimes unintended) biases (Franzosi 1989). In Spain, for instance, there is no obligation for companies to communicate the number of workers involved in strike action. For this reason, public administrations determine the number by consulting the company management, entrepreneurs' associations, unions, or by producing their own surveys on the participation in strike action. It can be assumed that numbers provided by the employers' side tend to minimize the impact (and thus the size) of strikes while the unions' side is expected to boost these numbers. Since two out of three external sources are from the employers' side, there is a risk that the numbers provided by the ministry tend to underestimate the impact of strikes. The vagueness regarding the collection of this information suggests handling with care the number of participants in strike action and the derived number of working days lost due to strikes.

ECONOMIC AND POLITICAL DIMENSIONS OF LABOR CONTENTION IN SPAIN
Having presented the theories and the data to be examined, I now proceed to the actual analysis. In the following two sections, I separate the economic and the political dimensions of labor related contention. In order to relate the theoretical models with the Spanish case, I begin each section with an overview of the relevant (economic or political) evolutions in Spain, which allow me to formulate some working hypothesis. In both sections, I examine strikes and labor related street protests separately at first and then present a relational reading of the two modes of action. In a final section, I propose a short synthesis combining both economic and political factors in the analysis of strikes and labor related street protests.

Economic dimensions of strikes and labor protests in Spain
For the time period under study (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016), economists dealing with the Spanish case distinguish between a pre-crisis period (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007), a crisis period (2008-2013) and a period of stagnation or 'moderate recovery' as of 2014 (Banco de España 2017). Spanish labor market statistics reflect these three periods (see Table 1). The first eight-year period presents a relatively steady increase in the total number of persons employed. The unemployment rates at first oscillated between 10.6 percent and 11.5 percent, but showed a declining trend after 2003. This trend was reversed, still slightly in 2008 and sharply in 2009. After having reached peak values in 2 013, unemployment rates decreased again as of 2014 and the total number of persons employed increased accordingly. The Business Cycle Model predicts a positive relationship between economic boom and strike activity. Thus, when applying the model to the above-described economic evolution, one can formulate the following hypothesis regarding the levels of labor related contentious collective actions in Spain: H1: The level of labor contention is expected to be high during the precrisis period (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007), low during the crisis period (2008-2013) and intermediate during the period of moderate economic recovery (2014)(2015)(2016).
Grievance Theory, expecting higher levels of contention during times of widespread relative deprivation, suggests a different hypothesis for the Spanish case: H2: Regardless of the levels prior to the crisis, labor contention is expected to increase with the outbreak of the Great Recession and to lower again after 2014.

Strike activity contradicting the Business Cycle Model?
The different vectors of strikes have evolved unequally over time (see Table  2). Furthermore, some vectors -notably strike size -vary greatly depending on whether or not general strikes are taken into account. In order to facilitate the subsequent interpretation, I first depict the evolution of the shape of strikes by omitting the general strikes and then show how the picture changes when they are included.   (7220) (1301) 1327 (2058)  Strike frequency-or the number of strikes per 100,000 employees-seems to be the only variable to show a specific crisis related pattern. It decreases from 5.9 in 2000 to 4.3 in 2005 and remains relatively low during the remaining precrisis period. The values increase with the outbreak of the economic crisis-still moderately in 2008 and sharply in 2009. Except for the year 2011, when the value drops to 5.1, strike frequency remains above 6 and reaches its peak in 2013, with 7.1 strikes per 100,000 employees. Frequency then decreases to 5.4 in 2014 and further drops to 4.2 in both 2015 and 2016-the lowest levels of the entire time span under study. These tendencies are also reflected in the mean values for the three periods under study: An intermediate value of 5 strikes per 100,000 employees for the pre-crisis period, a mean of 6 for the crisis period and a low of 4.6 for the post-crisis period.
For strike size-i.e. the number of strikers per strike-a general trend towards smaller strikes can be observed. The millennium starts with a strike size close to 3000 strikers per strike-a value that remains unparalleled during the remaining time span under study. Already during the pre-crisis period, the values decrease and reach levels between 490 and 790 workers per strike in the years 2004 to 2007. Values remain similar during the first two years of the crisis, but then drop to values between 280 and 460 workers per strike. In the post-crisis period, finally, strike size remains low and does not exceed 290 workers per strike.
Strike duration is the vector with the most inconclusive pattern, which is also reflected in the mean strike durations of the three periods: In all three periods, strikers struck for a mean duration of 2.6 days (if general strikes are left aside). All but two values range between 1.1 and 2.9 days per striker. Only in 2004 and 2012, the values reach higher levels, i.e. 8 and 4 days, respectively. While strikes have generally lasted longer in 2012, the value for 2004 results from a single, extraordinarily long-lasting strike in the agricultural sector in Andalusia. This single strike accounts for more than two thirds of the working days lost in that year. 5 Since strike magnitude results from multiplying strike size and strike duration, the tendency towards smaller strike size combined with the indistinct pattern of strike duration also translates into a tendency towards smaller strike magnitude. The same decreasing tendency can also be observed for strike volumes-despite the addition of strike frequency as a further multiplication factor. The fact that strike volumes decrease with the crisis period indicates that the increase in strike frequency during that same period could not compensate for the decrease in strike size.
The increases in the periods with general strikes. The fact that three out of four general strikes occurred during the second period under study also undoes the abovedescribed tendency towards smaller strikes. With a mean of 2,043 workers per strike, the second period now exceeds the mean of the first period (1,852 workers per strike) and the drop between the second and the third period is much greater. Strike duration decreases in both periods with general strikes, since all general strikes lasted only one day (i.e. less than the average strike duration when omitting general strikes). Here again, the effect is greater in the second period, where three general strikes occurred. Mean strike duration decreases from 2.5 days per striker in the first period to 1.9 during the crisis period and increases to 2.6 in the post-crisis period.
As a consequence of the higher values of strike size, means of both strike magnitude and strike volume also increase during the first two periods. Regarding strike size, the gap between the first two periods under study gets smaller (with 3,271 working days lost per strike during the first period and 2,752 during the second) and would actually be reversed if the outlier case in 2004 were left aside. The mean strike volume of the second period (166 working days lost per 1,000 employees) almost equalizes the pre-crisis period (169 working days lost per 1,000 employees). This points at the impact of the increase in strike frequency during the second period. Including the general strikes not only substantially increases the mean strike magnitudes and volumes of the first two periods, it also amplifies the drop between the second and the third period, since no general strike occurred during the last three years under study.
Based on these numbers, some tentative conclusions can be drawn regarding the relationship between the state of the economy and strike activity. One finds that there is a distinct crisis-related pattern only for strike frequency. Not only is the average strike frequency highest during the crisis period (6.0 strikes per 100,000 employees); that same six-year period also includes the four highest strike frequencies of the seventeen years under study. When including general strikes, the crisis period is furthermore characterized by an increase in strike size, but lower strike duration. At first sight, these numbers thus seem to contradict the Business Cycle Model, which predicts a decrease in strike activity-and notably strike frequency-during a crisis period. The increase in strike frequency and strike size during the crisis rather tends to provide support for the Grievance Theory.

A proliferation of labor-related street protests during the economic crisis
Is there a similar crisis related pattern for labor related street protests? As mentioned above, the study of street protests is limited to the frequency of announced street protests (the only consistent value for the entire time period under study). Rallies and marches organized by workers, works committees or trade unions account for a considerable share of the total number of announced street protests in Spain. As shown in Table 3 highest during the crisis period, when every other announced street protest was called by workers, works committees or trade unions. Frequency of labor related street protests, i.e. the number of labor related street protests per 100,000 employees, varied considerably over the time span under study. During the pre-crisis period, the frequency of announced labor protests oscillated between 8 and 13 protests per 100,000 employees. In 2008, protest frequency reached a level of 26-twice as high as the highest level registe-red during the pre-crisis period. Apart from the year 2011, when the frequency of labor related street protests experienced a drop, the values continued to increase, reaching their highest level (186) in 2012. Beginning in 2013, protest frequency started to decrease, with the biggest drop experienced between 2013 and 2014. However, protest frequencies remained high after 2014.
When comparing the three periods en bloc, one finds the lowest mean frequency of labor related protests for the pre-crisis period, for which a mean of 10 labor related street protests were counted for 100,000 employees. This value increased more than nine-fold during the crisis period (94) and reached an even slightly higher level during the post-crisis period (97). Similar to the observation for strikes, the strong increase in protest frequency during the crisis period also appears to be in stark opposition to the Business Cycle Model.

A shift towards more defensive and isolated strikes as well as towards less costly modes of action
So far, the findings of increasing strike and protest frequencies during the crisis period seem to contradict the Business Cycle Model. However, one has to take into account that strikes and street protests did not move in parallel: After the outbreak of the Great Recession, frequency of labor protests increased more strongly than strike frequence. Unlike the levels of strikes, levels of street protests also remained high during the post-crisis period.
Looking at strike motives helps to understand the comparatively modest increase in strike frequency during the crisis period. The Spanish Ministry of Employment and Social Security distinguishes between 'collective bargainingrelated', 'other work-related' and 'not strictly work-related' strikes-the latter referring to union-related strikes, solidarity strikes, or strikes against policy reforms. Figure 2 shows that the frequency of these three types of strikes evolved unequally over time. A general (but unsteady) downward trend can be observed for the frequency of collective-bargaining-related strikes. The frequency of other work related strikes sharply increased in 2009, remained high until 2013 and showed again lower values as from 2014. The frequency of not strictly workrelated strikes remained rather low throughout the entire period-2014 being the only year that stands out with a peak value almost twice as high as the second highest value (measured in 2012). It becomes clear from Figure 2 that the general increase in strike frequency observed for the crisis period can largely be attributed to the increase in work (but not collective bargaining) related strikes.  The work (but not collective bargaining) related strikes can be further divided into 'defensive, ' 'offensive' and 'other' strikes. 6 Defensive strikes aim at preventing or undoing layoffs or deterioration of working conditions. Here, they include strikes against non-payment of wages, industrial restructuring, staff downsizing plans, the company's non-compliance with labor and employment laws as well as strikes against disciplinary measures adopted by the company (including dismissals of single workers). Offensive strikes strive for improvements of employment conditions. The data provides two motives that can be assigned to those offensive motives: 'improvements independent from collective agreements' and 'improvements for staffs not covered by collective agreements.' Finally, 'other' strike motives include all those that cannot easily be categorized into either offensive or defensive strikes. The following strike motives were categorized as 'other': workflow-related strikes, strikes due to labor-related accidents, and 'other' strike motives.
The analysis of the data shows that the share of defensive strikes approximately doubled in the course of the economic crisis (see Figure 3). 7 This increase in defensive strike motives was accompanied by a decrease in "offensive" strike 6 I do not consider the collective bargaining related strikes because they can be both offensive and defensive in nature. 7 Unfortunately, the categories used by the Spanish Ministry of Employment and Social Security in order to distinguish strike motives were modified in 2014. For this reason, conclusions regarding the defensive or offensive nature of strikes can only be drawn for the period 2000 to 2013. motives. At their peak, offensive strikes accounted for up fourteen percent of work (but not collective bargaining) related strikes. After 2008, they remained below the five percent level. However, the relative increase in defensive strikes went mostly at the expense of 'other' strikes. Note: Collective bargaining is excluded because it can be both offensive and defensive in nature. As from 2014, the data no longer allows to categorize the strike into defensive and offensive strikes.
Source: Spanish Ministry of Employment and Social Security; own calculations.
These findings suggest that the underlying assumption of the Business Cycle Model might still apply. The decreasing share of offensive and 'other' strikes indeed indicates that the workers' strike propensity decreased with the outbreak of the crisis. However, the workers' decreased propensity to go on strike seems to have been overcompensated by an increasing number of circumstances in which defensive strike actions became a means of last resort-most notably in cases of non-payment of wages or mass layoffs.
Another way to look at the unequal evolutions of strike and protest frequencies is to put them in a direct relationship with one another. The unequal increases suggest that there has been an overall shift in the way workers' expressed their discontent (see Figure 4). During the pre-crisis period, labor related street protests were roughly twice as frequent as strikes. In 2008, protests were already more than four times as frequent than strikes and this ratio continued to increase until 2012, when it reached a ratio of about 31 to 1. Although the relative share of strikes reached again slightly higher levels as of 2013, strikes remained (relatively seen) far less used than prior to the crisis. It becomes visible that the crisis was accompanied by a clear shift away from strikes towards labor protests in the streets. The Business Cycle Model provides a possible explanation for the workers' shift away from strikes to street protests. As the previous analysis of strike motives indicates, strikes increasingly became a means of last resort. It actually seems that the workers-weakened in their bargaining position-increasingly resorted to less costly actions such as street protests. Rather than contradictory, the Business Cycle Model and Grievance Theory thus seem to complement one another. The latter provides a possible explanation for the overall increase in labor related contention during the economic crisis. The former provides a credible explanation for the workers' shift away from strikes to labor related street protests.
The economic perspectives have helped to better understand some dimensions of industrial conflict-notably frequencies of both strikes and labor related protests-between 2000 and 2016 in Spain. Yet, there remain some unresolved issues. The general strikes (and notably their impact on strike size) are one such issue. Because of their very nature, these strikes call for a different reading. As Gall (2012: 682) puts it, '[t]he primary purpose of the mass strike is not to impose economic costs upon the employer (as a stand-alone strike with a private employer primarily attempts to do) but rather impose a political cost on the government.' In order to understand the occurrence of general strikes, one thus has to expand the focus and include the political dimensions of labor contention. The high frequency of labor-related street protests during the post-crisis period is another unresolved issue. Why did strike frequencies decline again to pre-crisis levels while the frequencies of labor related street protests remained high-on average even higher than during the crisis period? A closer look at the political dimensions of labor related contentious collective actions might provide some further insights.

POLITICAL DIMENSIONS OF STRIKES AND LABOR PROTESTS IN THE STREETS
After almost 40 years of fascist dictatorship and the so-called transition to democracy in the late 1970's, Spain was characterized by a de-facto two-party system during most of its democratic period. Since the election of the social democratic Spanish Socialist Workers' Party (PSOE) into government in 1982, the social democrats alternated one-party rule with the conservative People's Party (PP). 8 The social democrats governed between 1982-1996 and again between 2004-2011, and the conservatives between 1996-2004 and again between 2011 and the end of the time frame under study. However, the two parties had absolute majorities only between 1982 and 1986 (PSOE) and between 2000 and2004 (PP). During the other mandates, the governing parties depended on "pivotal" parties capable of determining the majority in parliament. The ex-communist United Left (IU) and different regionalist and nationalist parties such as the Basque Nationalist Party (PNV), or the Republican Left of Catalonia (ERC) and Convergence and Union (CiU) in Catalonia repeatedly played such roles (Orriols 2013: 362).
Although the one-party rule was upheld until 2016 (the end of the time frame under study), the Spanish political landscape and the conditions for government formation were radically transformed by the December 2015 General Elections. Two political parties-We Can (Podemos) and Citizens (Ciudadanos, Cs)broke the two-party system open, gaining 21 and 14 percent of the votes, respectively. In the aftermath of the election, no candidate for prime minister was capable of securing sufficient support for his election by the Congress of Deputies. The incumbent prime minister Mariano Rajoy (PP) provisionally remained in office and new General Elections were held in June 2016. In these elections, the Popular Party recovered some of its losses from 2015 and was ultimately able to secure another government mandate due to the support of Citizens and the abstention of the majority of the social democrats. The union landscape, for its part, remained rather stable during the time frame under study. Over the entire period, the two major unions-the Workers' Commissions (Comisiones Obreras, CCOO) and the General Union of Workers (Unión General de Trabajadores, UGT)-constantly accumulated more than two thirds of the votes at trade union elections (Beneyto 2018). According to the Political Exchange Model and the Political Opportunity Perspective, changes in the party in power can be expected to affect the major unions' propensity to resort to contentious collective actions. In both models, the frequency and size of labor related contentious collective actions are expected to reach higher levels under a conservative government.
Regarding the presence or absence of a strong Communist left, the principal player to consider at a country scale is the United Left (Izquierda Unida, IU), which was the third political force in parliament during much of the 1990s. During the 2000s, however, the party suffered strong electoral losses and was no longer able to form a parliamentary group on its own during two legislative terms (2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011). In 2011, IU could temporarily reverse the downward trend.

A tendency towards more demonstrative strikes under conservative governments
In both the political exchange model and the political opportunity perspective, one can expect strike frequency and strike size-and ultimately also strike magnitude and strike volumes-to be greater under conservative governments. A preliminary look at the data seems to support this hypothesis (see Table 4). Mean strike frequency is higher under conservative governments (mean of 5.4 strikes per 100,000 workers compared to 5.2 strikes per 100,000 workers under social democratic regime). Yet, the difference is only slight.  Inclusion of general strikes amplifies the differences between social democratic and conservative governments. Of the four general strikes, three took place under conservative administrations (one in 2002 and two in 2012) and one under social democratic government (in 2010). Under conservative administrations, strikes counted about three times as many participants as under the social democratic administration. At the same time, they were nearly a day shorter. Means of both strike magnitude and strike volume under conservative administrations are also about one and a half times higher than in the social democratic period. This indicates that the increase in strike size under conservative administrations more than outweighs the decrease in strike duration.
Under conservative administrations, strikes have thus taken a more demonstrative character (i.e. short but massive). This is generally in line with the political perspectives examined in this paper. The stronger bonds between major unions and social democrats (Political Exchange Model) or the major unions' anticipation of reforms (Political Opportunity Perspective) seem to reduce massive strike action during periods of social democratic government.
More surprising is the proximity of the strike frequencies between conservative and social democratic regimes. Under the PSOE, a mean of 5.2 strikes per 100,000 workers occurred, while the mean value for the conservatives (5.4 strikes per 100,000 workers) is only slightly higher. This invites one to have a closer look at the evolution of strike frequency. Figure 5 shows that both incumbent parties faced the highest levels of strike frequency during the crisis period. The parties also experienced the lowest strike frequencies either prior to the crisis (PSOE) or after the crisis period (PP). These parallels account for the similar strike frequencies between the two parties.  Changes in the political party were thus accompanied by stronger changes in strike size than in strike frequency. 11 There is one year, however, where strike frequency seems to have been affected by a political exchange. Figure 5 shows in fact a considerable dip in both absolute and relative numbers of strikes for 2011. This dip cannot be explained in economic terms: unemployment levels were still on rise and the numbers of companies affected by mass layoffs also reached a new peak in 2011. From an economic perspective, there is thus little reason to expect a sudden (and intermittent) drop in strike frequency. A political exchange between major unions and the social democratic government might provide a more plausible explanation. In 2011, the social democratic government faced in fact a delicate situation: it was seeking reelection while being confronted with strong challenges by the 15-M movement in the streets. One explanation for the drop in strike frequency could thus be that the major unions deliberately sought to reduce the numbers strikes in order to lower the pressure on the social democratic party and to favor its reelection.

Higher levels of labor protests under conservative governments
A look at the data on street protests shows that labor related street protests were both more numerous and more frequent under conservative governments (see Table 5). The ratio is about 2.5 for the absolute numbers and 2.7 for protest frequencies. Compared to strike frequency, where only a slight variation could be observed between the social democratic and the conservative period, labor related protest frequencies thus appear to vary more widely depending on the political party in power. The findings for labor related street protests are thus in line with the hypothesis of heightened levels of labor related contention under conservative governments.
A closer look at the evolution of protest frequencies shows why protest frequencies vary more widely than strike frequencies. Similar to strike frequencies, labor related street protests experienced a steep increase with the outbreak of the economic crisis (see Figure 6). Yet, contrary to strike frequency, these protest frequency experienced another sharp increase after Mariano Rajoy (PP) was elected prime minister in December 2011. Furthermore, although protest frequency declined during the post-crisis period (a similar trend to strike frequency), it remained comparatively high (unlike strike frequency). These differences account for the greater variation in protest frequencies than in strike frequencies depending on the party in power. The graph for protest frequency shows another similarity with that of strike frequency: the dip in 2011. This further supports the hypothesis that the major unions might have actively slowed the public expression of labor related contention in order to reduce the pressure on the incumbent social democratic government during the election year.
In sum, the findings obtained in this section suggest that labor related contention had a more demonstrative character under conservative governments: Strikes were more massive, but shorter and labor related street protests more frequent under conservative governments.

Relating political and economic models
The economic and political analyses of labor related contentious collective actions have produced important insights. The findings show that the economic crisis was accompanied by increases in both strike and labor related protest frequencies. They furthermore reveal that strikes were more massive but shorter and labor related protests more frequent under conservative governments. Combining the two perspectives shows that labor contention peaked during the crisis years with a conservative government (see Table 6).  The second highest frequency values are reported for the crisis period with a social democratic government. This suggests that frequencies are more related to economic than to political factors. A different picture emerges regarding strike size, where the value for the pre-crisis period with a conservative government exceed the crisis-value of the social democratic government. Overall, the findings indicate that the economic and political perspectives are most meaningful if they are combined.

CONCLUSION
In this article, I presented a relational analysis of strikes and labor-related street protests between 2000 and 2016 in Spain. To this end, I examined aggregate data provided by Spanish ministries and combined theories on industrial conflict with theories on social movements. More specifically, I discussed the respective economic and political models of contention and examined the extent to which they contribute to a better understanding of labor contention in the recent history of Spain.
The findings show that one can make both an economic and a political reading of the evolution of strikes and labor protests in Spain. With respect to the economic reading, the results indicate that the economic crisis (2008)(2009)(2010)(2011)(2012)(2013) was accompanied by an increase in the frequency of labor-related contention. This finding can be seen as being in line with Grievance Theory in social movement studies, since the theory expects higher levels of contention in times of economic distress. Yet, at the same time, this increase was unequal for strikes and labor protests. Mean strike frequency rose by one fifth during the crisis period. Mean labor protest frequency, however, grew more than ninefold. Thus, in relative terms, one finds that workers reduced their use of strikes in favor of street protests. This in turn can be seen as being in line with the Business Cycle Model in research on industrial conflict. Regarding the political dimension of labor contention, I could show that strike size and strike duration considerably varied with changes in government. In periods of conservative governments, strikes tended to be shorter but more massive. In other words, the strikes took a more demonstrative character (Gall 2012;Gallas, Nowak y Wilde 2012). Finally, imbricating economic and political perspectives shows that strike size and the frequencies of both strikes and labor protests were highest during the years of crisis under conservative government. One conclusion becomes evident: looking at strikes (or street protests) alone does not only lead to a partial view of labor contention, it can also result in interpretations that are entirely wrong. Future studies could extend the here presented macrological analysis of labor contention in Spain by including meso-and micrological processes. Thereby, greater emphasis could be given to (de-)mobilizing effects of shifts and changes happening within the "space of social movements" (Mathieu 2021)-notably the emergence of the 15-M or the Mareas (Spanish for "tides). Including these processes would further enrich the analysis and potentially qualify some of the here presented findings.
The present analysis is a strong call for a greater empirical and theoretical dialogue between research on industrial conflict and studies of social movements. The two research areas might sometimes conflict with one another, but, above all, they have proven to be mutually enriching. Relating the study of strikes with the study of labor protests thus seems an effective means to better understand the ebbs and flows of the respective modes of action.