1. Top of page
  2. Abstract
  3. Introduction
  4. Unpaid Overtime: Consequences and Theoretical Explanations
  5. Overtime and Labour Flexibility Regulations in Greece
  6. Determinants of Unpaid Overtime in Greece
  7. Conclusions
  8. References

During the last decades an increase of unpaid overtime has been observed combined with the inversion of the working time decreasing trend. This article analyses the data from the Labour Force Survey to estimate the determinants of unpaid overtime in Greece and uses the results to test various theoretical explanations.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Unpaid Overtime: Consequences and Theoretical Explanations
  5. Overtime and Labour Flexibility Regulations in Greece
  6. Determinants of Unpaid Overtime in Greece
  7. Conclusions
  8. References

It has been observed that during the last three decades, there is a marked increase of unpaid overtime. This trend has been evident not only in less developed economies but in developed economies as well. Several surveys have shown (see e.g. Anger, 2005; Hetrick, 2000; Pannenberg, 2005) that unpaid overtime is on the ascendance in the UK, United States and Germany; thus reversing a downtrend existing from the first decades of the 20th century. More specifically, these surveys observe that unpaid overtime in these developed economies is equal to or has even surpassed paid overtime and its significance continues to increase. The 2007–08 global economic crisis (the first 21st century depression) and its subsequent years of dismal growth have aggravated further this trend. This contemporary trend has obvious and profound repercussions not only on workers' quality of life but also on the macroeconomy as a whole. First, they invigorate the phenomenon of increased working time, which is observed also in developed economies (and in Greece) over the same time period. Second, it implies a reduction of the real wage rate and hence to a deterioration of the workers' terms of reproduction. Both these two consequences affect crucially the way developed western economies functioned for the whole postwar period.

Apart from the above-mentioned consequences, the very existence of unpaid overtime, let alone its marked increase, posits crucial analytical questions to mainstream economic theory. If part of overtime is truly unpaid, then workers could only be obliged to provide it as they do not derive any utility from it. This poses a serious explanatory challenge to neoclassical economics because it denotes that labour unwillingly and without being paid for is offering part of its working time. This obviously violates the efficient price determination and income distribution. Moreover, it exhibits the existence of power relations within a free market economy which oblige labour to provide this unpaid work and, thus, benefit employers. If this argument holds, then it justifies other economic schools (and mainly the Marxist one) that argue that labour is subject to exploitation in a capitalist economy as it is not paid for the total of its contribution.

The gravity and the persistence of increased unpaid overtime cannot be ignored nor be relegated to an insignificant and fleeting phenomenon. For this reason, neoclassical theory has peaked up the theoretical gauntlet and offered a number of theoretical answers to this annoying phenomenon. All the offered answers argue that through different subtle mechanisms workers are actually being paid this unpaid overtime. Hence, they derive personal utility and are justly paid. At the same time, an extensive empirical literature has been accumulated, studying the factors determining unpaid overtime as this can facilitate both the understanding of the phenomenon and its policy response.

This work begins by reviewing critically the theoretical explanations offered regarding unpaid overtime. The next part describes in brief the institutional context of overtime and flexible work arrangements in Greece. The study of the Greek case is important not only for the historically unique economic crisis the country is going through, but also because it is a country where workers work long hours that are currently increasing and simultaneously part-time and flexible working arrangements are not well accepted and diffused throughout the economy, which can provide a point of comparison between part-timers and the rest of the workers, as will be analysed later. In the next part, factors determining unpaid overtime in Greece are estimated, with the use of an appropriate econometric model and micro data provided by the Greek Labour Force Survey.

The data used are for the year 2009, which is actually the first year of the Greek economic crisis (there was a negligible contradiction of the economy in 2008 as well). However, in 2009 the full force of the crisis had not emerged and the Greek economy continued its usual modus operandi, so by using the 2009 data probable transition effects can be avoided, as the full eruption of the crisis and the subsequent structural changes have affected severely the Greek labour market. These data can also provide the base for comparison of the situation before and after the crisis. The results for unpaid overtime are compared with those regarding paid overtime. The comparison of the determining factors of unpaid and paid overtime leads to very intriguing conclusions and reinforces the heterodox view that unpaid overtime stems from employers' power and pressure on workers. Thus, it is an unwilling and truly unpaid phenomenon that does not verify neoclassical explanations.

Unpaid Overtime: Consequences and Theoretical Explanations

  1. Top of page
  2. Abstract
  3. Introduction
  4. Unpaid Overtime: Consequences and Theoretical Explanations
  5. Overtime and Labour Flexibility Regulations in Greece
  6. Determinants of Unpaid Overtime in Greece
  7. Conclusions
  8. References

The increase of unpaid overtime has, apart from the purely theoretical ones, crucial practical economic consequences. First, it increases total working time. Given that since the mid-1980s overworking (i.e. the tendency of the overall work time to increase) has been observed in several developed economies (Bluestone and Rose, 2000; Leete and Schor, 1994; Schor, 1991), and in Greece, it is obvious that one of its channels is overtime and particularly unpaid overtime. Second, unpaid overtime leads to the reduction of the wage rate. It has been argued within the framework of the human capital theory that university graduates, who usually get administrative or hierarchically higher jobs, get increased salaries because they are paid for the human capital they acquired with their studies. Because unpaid overtime is predominantly observed in this category of employees, this implies that their wage rate is reduced when they incur unpaid overtime. This fact worsens the position of these employees and, at the same time, weakens the explanatory power of human capital theory. However, the existence and the increase of unpaid overtime has another more crucial theoretical consequence. According to neoclassical theory, each employee is paid for his whole marginal product of labour. Therefore, there is neither an unpaid amount of labour nor labour exploitation in the capitalist economy. On the contrary, according to Marxist economics each employee is paid for only a fraction of his working day. The rest is unpaid labour wherein product goes to the employer, so that firm's profit comes out from exactly this unpaid labour time. The existence of unpaid overtime reveals a part of this unpaid labour and, consequently, reinforces the Marxist argument.

The major theoretical explanations of unpaid overtime that have been proposed from the existing literature are presented and also criticized below.

Unpaid overtime as the result of uncertainty regarding a task's time of completion

This explanation (Bell and Hart, 1999a) argues that unpaid overtime exists because in certain cases the employer knows the wage rate that corresponds to an employee's qualifications but does not know exactly the time required for the completion of the task. This does not imply a kind of asymmetrical information because neither the employer nor the employee has an accurate knowledge of the time required for the task. Hence they agree that the task has a degree of uncertainty regarding its completion time. In these cases, work process can be altered during its execution without being able to foresee such alterations in advance. Moreover, it is not unusual in such cases that the supervision costs be very high and even prohibiting. Under these conditions, it is possible that employer and employee agree on a work time that can be very different from the actual one. If the agreed work time is less than the actual one (i.e. if the work time is underestimated), then the employee has to work unpaid in order to finish the task.

According to this approach, unpaid overtime is not, even in the end, paid. But the one guilty for this is not the employer, but the wrong work time estimation of the employee. This underpayment can be possibly covered by another task where, again purely accidentally, work time is overestimated and thus the employee is paid more than he should be.

This approach focuses on a very specific case and cannot account for the whole phenomenon of unpaid overtime which is spread well beyond the borders of a particular type of work task. Moreover, it cannot explain the particular upsurge of unpaid overtime during the recent decades.

Unpaid overtime as the outcome of an auction process

According to this explanation (Bell and Hart, 1999b), employers do not have a fair knowledge of neither the work time required for a task nor each employee's productivity. In such a case, employees know that employers will probably choose those workers that declare that they can do the job in less time (hence implying that they are more productive than the others). If the selection procedure is one of competitive underbidding (an auction), then less productive workers have a motive to declare a work time less than their actual one, so as to be selected. If they are selected and given that the employer has no motive to check their work time because his profits are not affected, they work more than agreed in order to complete the job. This additional work time is unpaid overtime.

This is also a very weak argument that can explain only a very limited number of cases of unpaid overtime, but it cannot explain the phenomenon as a whole and its upsurge during particular historical periods.

Unpaid overtime as a signalling device

This is a game-theoretical sui generis neoclassical approach. It assumes that typical general equilibrium conditions (in particular, perfect information) do not hold. In order to avoid market failure, some signalling mechanism must be established which will shed light and disclose accurate information (Anger, 2008). This mechanism is set as a dynamic game of incomplete information in which players ‘flash’ signals to one another.

This approach argues that the labour market is characterised by incomplete information on behalf of employers regarding workers' abilities. In such a case, the formation of the work contract is a dynamic game where signalling plays a crucial role. An individual worker might, by offering to work unpaid overtime, choose to send a signal to the employer that his personal qualities are superior to those of others. The signalling device approach maintains that the more competent workers are those who can easily work overtime and, in this way, they signal to their employers their superior working quality (Anger, 2008; Meyer and Wallette, 2005). Consequently, the surplus labour—performed through overtime—does not produce surplus product but it is a mere signal by the workers to the employers. The signalling device approach is in stark contrast to the auction approach because the former considers unpaid overtime as a proof of a worker's superior quality whereas the latter as one of inferior quality.

Several objections can be raised regarding the signalling device explanation. The first problem is that it rests on very restrictive assumptions. For example, its hypothesis that agents have a strategic plan for the game and that they exhibit a sequential rationality cannot stand in the face of reality. The assumption of perfect competition has also well-known problems particularly when applied for explaining a phenomenon like unpaid overtime and its particular upsurge during a specific historical period. Second, even the modelling of the signalling mechanism is open to criticism. For example, signalling could signify completely different characteristics of a worker than a supposed superior quality. It might signify that he is more flexible towards employers' demands because he feels insecure about his job. Third, this approach recognises only the informative and not the productive use of unpaid overtime. It can be counter-argued that even in the case that one worker has more physical and/or mental capabilities for working long hours, but these abilities play no productive role, then there is no use in acquiring useless (for the production) information. By using signalling in order to send information, a lot of precious effort is wasted inefficiently while trying to rectify inefficiency. Last, but not the least, if some workers work unpaid overtime—in order to signal their superior qualities—it is not certain that all of them would be rewarded accordingly. This means that those that tried but failed would not be compensated for their contribution.

Unpaid overtime as the derivative of team work

This approach attempts to explain unpaid overtime by focusing on tasks requiring team work. In firms where team work is used, the replacement of a team member can be very costly as the acquisition of the ability for team work is a lengthy process (Aoki, 1984). Therefore, redundancies tend to be avoided. However, because workers do not have the same productivity, unpaid overtime can sometimes be exercised as a form of periodic compensation for the limited performance of the less productive team members. Furthermore, in cases of illness or absenteeism, the team leaders feel obliged to cover the scrappy outcome so as the progress of team work will not fall behind. It is for these two reasons, according to this approach, that unpaid overtime arises.

A more general explanation has used the notion of the ‘post-Fordist’ workplace (van Echtelt et al., 2006; Perlow, 1999) which is ‘time greedy’ and may create through a process of small decisions a cumulative effect which was never the direct choice of any worker but it actually occurred. This cumulation good (or bad) can be the increased working time, or the increased overtime.

This is, nevertheless, a rather weak theoretical explanation based on exceptional empirical cases. But even in these cases, why does the tendency of the cumulative effect has to be towards overwork and not to the opposite direction? In such a case, the only realistically possible explanation is the existence of some kind of pressure, probably not the ‘Fordist’ kind of pressure, by the employers.

Unpaid overtime as expression of gift exchange

According to Akerlof (1982), particular firms can pay employees a higher wage than the one predominating in the market in order to motivate them to be more productive. This might imply that these workers work with greater intensity during their work time. On the other hand, this extra work effort might not take the form of increased intensity but the form of increased working time. This can be a Pareto optimum situation. In this case, it can be expected that unpaid overtime will increase with the increase of the hourly wage rate.

Unpaid overtime as a means for achieving the Pareto optimum

This is a purely neoclassical explanation. According to typical neoclassical theory, wages are endogenously determined through a utility maximising process where workers maximise their utility and employers their profits. This process results in a market clearing equilibrium, where each productive factor receives a just compensation for its contribution. This approach about unpaid overtime departs from the above typical neoclassical theory by assuming a heterogeneous labour market and that the partial equilibriums that are established in its different segments are not efficient. Therefore, this decentralised equilibrium that is finally achieved is not an efficient one. As culprits for creating these inefficient partial equilibriums are considered the state intervention and the trade unions' interference. It is argued that these two mechanisms regulate work time (by posing specific limits) and the cost of overtime which might hurt the interests and the preferences of both employees and employers. For example, when overtime remuneration is set through collective agreements or by the state, there might be a case where employees would be willing to work overtime even with a smaller payment. This obviously would be preferred by employers as well. In such a case, the Pareto optimum is obviously violated. In order to surpass legal obstacles, employers and employees might agree to either an increase of normal work time (if this is possible) or to overtime work that is only partially paid. In this way, the legal remuneration for overtime is actually reduced. [For a similar treatment, see Trejo (1991; 2001)]. Through this process, deviations from the Pareto optimum are being rectified by employees agreeing to work with a smaller overtime payment, and this is done by accepting to be paid for only a part of their overtime work (Bell and Hart, 1999a).

This very neat neoclassical explanation has critical theoretical and empirical deficiencies. First, it is almost impossible to be tested empirically and for this reason there is a marked lack of relevant studies. Second, it is very weak analytically. Workers' preferences cannot be considered as invariable and exogenous to the economic and social process of determining work time and wage. It is not coincidental that most employees do not prefer a different work time from that they actually have, and which is usually uniform for all and has been determined through collective social and political processes rather than through individual bargaining. This means that the social and political processes through which wage and work time are being determined affects workers' preference formation and this is not an exogenous process. Therefore, social and institutional elements, such as the creation of collective organisations by employers and employees, respectively, cannot be abstracted from the economic processes determining wage and work time. Consequently, the pure neoclassical approach that focuses on individual bargaining on the basis of personal preferences cannot grasp properly these social processes and, thus, fail to account adequately for unpaid overtime.

Unpaid overtime as deferred payment

According to this approach, unpaid overtime is compensated for in the future. More specifically, employees provide unpaid work hours in order to increase output, which in turn augments employer's profit. Then they can be compensated in the future with higher wages, other benefits (e.g. bonuses) and promotions (Bell and Freeman, 2001; Bratti and Staffolani, 2007; Campbell and Green, 2002). Therefore, Pannenberg (2005) maintains that the term ‘unpaid’ overtime is inappropriate because, although not paid on a quid pro quo basis and at current time, they are ultimately compensated for through some form of future deferred payment.

A number of criticisms apply also to this explanation. For example, it cannot be secured that all the workers are in the end compensated for their additional effort or that they receive a full and not a partial compensation. On the contrary, there is evidence that such compensation does not take place (Anger, 2005).

Human capital theory explanation of unpaid overtime

Another popular explanation of unpaid overtime is based on human capital theory. According to the latter, the education of a worker is considered as his personal investment that would lead to greater productivity and, hence, better wage. Therefore, an employee might choose to work unpaid overtime in order to enhance his work experience and, hence, his human capital, with the expectation that he will be compensated in the future with an increased wage (Booth et al., 2003; Campbell and Green, 2002). Consequently, this explanation argues that in fact unpaid overtime is paid through future higher remuneration.

This explanation of unpaid overtime faces serious empirical difficulties. Surveys have shown (Bell et al., 2000) that unpaid overtime is especially widespread among specialised and educated workers. This, in turn, leads to a reduction of their real hourly wage rate. This fact contradicts the basis of the human capital argument. Consequently, it weakens its ability to explain the unpaid overtime phenomenon.

Unpaid overtime as form of labour exploitation

Most of the previous explanations come from the neoclassical tradition and broadly argue that unpaid overtime is a catachrestic term as this work time is actually paid through some indirect mechanism. Thus, every productive factor, at least ultimately, takes what he deserves and there is no exploitation in the capitalist economy.

On the other side, there is a completely different and more direct explanation coming from Marxist economics. This explanation argues that this working time is really unpaid and it is additional evidence of the exploitation of labour in the capitalist system. More specifically, it is argued that unpaid overtime is the result of pressure by employers on employees—usually using the threat of dismissal—and is not a voluntary phenomenon. Through unpaid overtime, employers achieve an increase in output without at the same time incurring an increase in wage costs. This, in turn, augments their profits which is the operating motive of the capitalist system.

In a nutshell, Marxist economics argue that what is being sold in the labour market is workers' ability to work. It is paid what is required in order to reproduce this ability (by taking into account the technical but also the social requirements for this reproduction). Then this ability to work is applied in production in the form of work time. According to Marxist economics, only workers have creative abilities, that is, are able to create new wealth in addition to their cost. On the contrary, fixed capital has no creative abilities and thus cannot create new wealth. An employer pays for workers' remuneration and for the cost of fixed capital, and he expects his enterprise not only to cover these costs but also to provide her with a profit. This profit comes from the new wealth that only workers can create. Hence, work time is divided in a paid part (reflected in wage) and an unpaid part (reflected in profit). The size of the work time as well as its division between its paid and its unpaid segment is a dynamic element as it is affected by the struggle between employers and employees. An increase of the work time without an increase of the wage (or with a lesser wage increase) augments profits and workers' exploitation (Marxism's rate of surplus value). Unpaid overtime is a method for increasing total working time and at the same time increasing its unpaid segment. If in a historical period increasing unpaid overtime is observed, this implies that for some reason the power of the employers has been augmented and they can pressurise employees to concede it.

The Marxist explanation argues that from the mid-1980s and onwards, there is a secular tendency of total work time to rise (see e.g. Schor, 1991). Increasing unpaid overtime is part of this trend. The increased power of the employers and their ability to impose unpaid overtime is the result of the neo-conservative restructurings of the last decades and particularly the neo-liberal onslaught.

Overtime and Labour Flexibility Regulations in Greece

  1. Top of page
  2. Abstract
  3. Introduction
  4. Unpaid Overtime: Consequences and Theoretical Explanations
  5. Overtime and Labour Flexibility Regulations in Greece
  6. Determinants of Unpaid Overtime in Greece
  7. Conclusions
  8. References

According to the recent laws N.3385/2005, N.3863/2010 and N.4093/2012, working between 40 and 45 hours per week (for enterprises working five days a week) is considered overwork and is compensated with a premium of 20 per cent. Working more than 45 hours per week is considered overtime work and is rewarded with an overtime premium of 40 per cent. If overtime is not announced or exceeds the yearly upper limit (depending on the firm type), it is considered illegal and the premium rises to 80 per cent. So unpaid overtime is illegal in Greece. Nevertheless, many enterprises and especially the small and medium-sized ones have a capability of relaxing the labour laws in their favour (Burtless, 2001). Of course this is the case for unpaid overtime in the rest of Europe as well. This phenomenon is expected stronger when no union or any other form of workers representation exists.

Overtime (paid or unpaid) is the preferred way for the Greek employees and employers to deal with the variations of demand, compared with the flexible working arrangements (Mihail, 2003). Although part-time work was always present in the Greek informal sector, flexible work methods were legislated rather late (Laws N.1892/1990 and N.2639/1998) in Greek labour relations. This is perhaps one of the reasons that Greece is one of the EU countries with minimum impact of part-time work [less than 6 per cent of the total employment for the year 2009, according to our estimations using the Labour Force Survey (LFS) micro data]. But the main reason is, in our opinion, that part-time work has historically been attached with lower wage rates and benefits and the lack of career prospects (Mihail, 2003). If the above are combined with the, already before the crisis, low Greek wages can give an explanation for the low popularity of part-time jobs among the Greek workers. This conclusion is verified by the fact that approximately 46 per cent of the part-timers in the year 2009 and almost 60 per cent in 2012 (according to our estimations using the LFS micro data) would prefer a full-time job, but did not make it.

This is the reason why the initiation of labour flexibility laws provoked strong reactions by the labour unions. Finally, the imposed law states that flexible forms of work could be implemented only under the agreement of the labour unions wherever they exist (Mouriki, 2001). This is a major reason why flexible arrangements including part-time work remained scarce in traditional branches of the secondary sector of the economy, while they were spreading in the tertiary sector, where small and medium enterprises and the lack of labour unions are the norm. The above conditions resulted in a very limited unionisation in atypical employment activities (Gialis and Karnavou, 2008).

On the other hand, shift work is mostly used in large enterprises of the secondary and tertiary sector and is rather uncommon in small enterprises. These establishments are strongholds of the Greek labour unions and also, because of their size, find it rather difficult to avoid the labour laws, so it can be argued that shift workers and workers in big enterprises are relatively better (legally and union) protected compared with their co-workers in other firms.

Determinants of Unpaid Overtime in Greece

  1. Top of page
  2. Abstract
  3. Introduction
  4. Unpaid Overtime: Consequences and Theoretical Explanations
  5. Overtime and Labour Flexibility Regulations in Greece
  6. Determinants of Unpaid Overtime in Greece
  7. Conclusions
  8. References

The increase of paid and unpaid overtime in developed economies drew the attention of many researchers and there is an extensive literature on this issue, mainly examining the labour market conditions for large economies such as Britain, Germany and the United States. A significant part of this literature examines the factors that increase the probability for a worker to provide paid and unpaid overtime. The detection of the determinants of unpaid overtime is important for practical and theoretical reasons as well. The results can be used in practice in order to construct appropriate policy instruments than can affect this phenomenon.

According to the literature (which was referred in the theoretical part earlier), a main determinant of unpaid overtime is related to the complexity of the modern post-Fordist work environment. The possibility of a career boost is also increasing the probability of unpaid overtime as well as the supervisory position in the job hierarchy. Education has a positive impact on unpaid overtime, too. There is also a large difference among white collar workers, who tend to provide much more unpaid overtime, and their blue collar counterparts, who on the contrary provide more paid overtime. Women have slightly smaller probability of unpaid overtime compared with men and the existence of children reduces slightly this probability. Age also reduces the probability, too.

As for the determinants of paid overtime, approximately the same factors have been investigated. Marriage is found to increase the probability of paid overtime, as well as the work in small enterprises. The usual working hours were found to have opposite results: positive effect on paid overtime in Britain and the United States while negative effects in Germany.

In the next section the determining factors of paid and unpaid overtime will be estimated for Greece with the help of the Logit model and data from the recent LFS of the Eurostat.

Methodological issues

In order to estimate the determinants of paid and unpaid overtime, two binary logistic regression models (logit) will be used, one for each case.The fact of working paid and unpaid overtime or not can be described by the following probability function (cumulative probabilities distribution function) which gives the (paid or unpaid) overtime probability

  • display math(1)

If P is the probability of an outcome to occur, then 1-P is the probability not to occur as shown in the following equation

  • display math(2)

where z = β0 + β1x1 + … βixi

From (1) and (2) we take:

  • display math(3)

where P(Yi = 1)/(1 − P(Yi = 1)) is the odds ratio, the ratio being modified in every modification of the Xi independent variables.

The logit modification of the overtime probability P(Yi = 1), can be denoted as (Cramer, 2004; Gujarati, 2003; Pindyck and Rubinfield, 1981):

  • display math(4)

where Log[P(Yi = 1)/(1 − P(Yi = 1))] is the neperian logarithm of the odds ratio. In this way, the odds ratio logarithm becomes a linear function of the Xi independent variables and the model changes to the classic linear (towards the parameters) regression model, which can be estimated and provide the values for the βi parameters.

Model 1: estimation of the determining factors of unpaid overtime

The first model is focusing on unpaid overtime. The equation to be estimated is the following

  • display math(5)

Many, but not all, of the variables that entered the model are mostly used by the existing literature. These are variables describing personal characteristics of the employee, variables describing the labour relationship and some other that describe structural aspects of the enterprises. Finally, variables describing workers preferences are examined, too. They are analytically described in Table 1.

Table 1. Variables used in econometric models for unpaid overtime
Dependent variable  
Supply of unpaid overtime (HWOVERPU)Binary0 = No unpaid overtime
1 = Performed unpaid overtime
Independent variables  
SEXBinary0 = Female
1 = Male
AGEContinuousAge in years
NATIONALBinary0 = Greek nationality
1 = Foreign nationality
SUPVISORBinary0 = Non supervisor
1 = Supervisor
FTPTBinary0 = Full timer
1 = Part timer
SFIFTWKBinary0 = No sift work
1 = Sift work
HWUSUALContinuousUsual weekly working time (hours)
WISHMOREBinary0 = Doesn't wish more working time
1 = Wish more working time
HATLEVELContinuousEducational level
ISCO 1DBinary0 = Manual labour
1 = Non-manual labour
STARTIMEContinuousMonths on last employer
HHNBO014ContinuousNumber of children
SIZEFIRM2ContinuousNumber of employees in the firm
DUMNACEPRBinary0 = Rest sectors
1 = Primary sector
DUMNACESEBinary0 = Rest sectors
1 = Secondary sector
DUMNACETRBinary0 = Rest sectors
1 = Transportation—construction sector
DUMNACETHBinary0 = Rest sectors
1 = Tertiary sector

The above variables originate from the LFS but have been processed to suffice the aims of the research.

Model estimation and results

In Table 2 that follows, the results of the estimation of regression equation (5) are presented. They were extracted with the help of the Stata econometric program, StataCorp LP, 4905 Lakeway Drive, College Station, Texas, USA.

Table 2. Results of the first Logit model for unpaid overtime
 Odds ratioaBRobust std. errorbp > z
  1. a

    Odds ratio.

  2. b

    Robust standard errors.


Robust standard errors were used in order to avoid the effects of heteroscedasticity, which is usual in that kind of samples.

Most of the examined variables are statistically important, which is important for the validity of the model. Where the same explanatory variables were used (sex, age, supervisory position), the results are consistent with the typical findings of the existing literature, that is, for other European countries and the United States.

As for the results, men have 11 per cent bigger probability of conducting unpaid overtime than women, and age is decreasing the probability of unpaid work. These results are consistent with the results of the studies for the other countries mentioned above. Nationality or the wish for more hours has no statistically important effects on providing unpaid overtime. On the contrary, usual working time is increasing the odds for unpaid overtime.

An interesting result comes from the comparison of the correlation of part-timers and shift-workers to unpaid overtime. Being a part-timer almost doubles the probability of working for free, while shift-workers have 16 per cent less probability of unpaid overtime.

The results also indicate that white collar workers have 18 per cent more probabilities of unpaid overtime, compared with their blue collar counterparts, a result that verifies those of previous studies. This result reinforces the interpretation of the complex duties presented above.

Education level is also found to increase the likelihood of unpaid overtime (by 6 per cent for each educational level), and this result seems to confirm the argument presented above, that unpaid overtime diminishes the explanatory power of the human capital theory.

Each child is found to reduce by 9 per cent the probability of working unpaid overtime, while the size of the firm does not have any effect at all. Finally, the secondary sector exhibits 25 per cent less probabilities of unpaid overtime compared with the tertiary sector.

Model 2. Estimation of the determining factors of paid overtime

The second model is focusing on paid overtime. The estimated equation is the following

  • display math(6)

As one can ascertain, the same variables were used as in the first model, for comparability reasons. So the explanatory variables are the same as in Table 1, with the only difference on the dependent variable, which is now paid overtime.

Model estimation and results

The estimation results are shown in Table 3.

Table 3. Results of the second Logit model for paid overtime
 Odds ratioaBRobust std. errorbp > z
  1. a

    Odds ratio.

  2. b

    Robust standard errors.

ISCO 1D0.82−

Again, most of the variables used are statistically significant. Being a supervisor increases the probability of paid overtime by 92 per cent (which is much smaller rate of increase compared with the case of unpaid overtime). The wish of the workers to work more increases the probability of paid overtime by 30 per cent, which indicates that one part of the paid overtime is conducted by the will of the workers. Contrary to unpaid overtime, blue collar workers have bigger probabilities than white collar colleagues to work paid overtime. The same contradiction holds for the comparison between secondary and tertiary sectors. A worker in the secondary sector is more probable to work paid overtime than his counterpart in the tertiary sector. Nationality is important in this case, because being a foreigner reduces the probability of paid overtime. Every child increases by 1 per cent the probability of paid overtime.

The results are becoming more interesting if a comparison is made between the determining factors of paid and unpaid overtime. For convenience, the results are gathered in one table (Table 4).

Table 4. Comparison of the results for unpaid and paid overtime
Model 1 and Model 2 results
Explanatory variablesUnpaid overtimePaid overtime
Odds ratioOdds ratio
  1. *Statistically important at 10 per cent level of significance.

ISCO 1D1.18*0.82*

Comparing the results

One main interest is in the comparison among the part-timers and the rest of workers and especially shift-workers. The reason for this is that part-timers can be used as a proxy for all other workers that are more vulnerable (Pollert and Charlwood, 2009) and unprotected. Especially in Greece, as argued above, part-time workers are even weaker against the will of their employers. On the other end of the scale, shift-workers are traditionally located in big enterprises of the secondary and tertiary sectors, which have to work with shifts. They are part of the traditional Greek work force, which is (until the crisis) protected by law and by relatively strong trade unions. So, shift workers can be used as a proxy for well protecting their rights workers. Under the above description, the comparison between these different groups of workers can be quite enlightening. As one can observe in Table 4, while both categories exhibit increased probability of paid overtime (with shift workers more increased than part-timers), when it comes to unpaid overtime, part-timers double their probabilities compared with the median worker, while shift workers have less probabilities compared with the general population. This is a clear difference among the part-timers and the rest of the workers and especially the shift-workers, which gives a strong argument for the explanation that unpaid overtime is not voluntarily provided but is the result of pressure upon the workers from their employers. If they are explained that way, unpaid hours are indeed unpaid.

An additional finding towards this explanation is that while the wish to work more (for those workers who have this wish) is positively affecting the probability of paid overtime, it has no significance in the case of unpaid overtime. So, part of paid overtime can be explained by the workers' will, but the same does not hold for unpaid overtime. This means that people who wish to work more do not provide more unpaid hours compared with the others. If they saw unpaid overtime as a means for gaining future income, or a promotion that would give them increased future income, one would expect that those who wish to work more (to earn more money) would provide more unpaid hours, which is not the case.

The explanation of unpaid overtime as a result of employers' pressure is further supported by the finding that seniority leads to less unpaid overtime, as senior workers are less vulnerable to employers' pressures (the large redundancy payments is one of the reasons) than newly hired ones. The above results reinforce the explanation that unpaid overtime is not a voluntary offer from the part of the workers but reflect the increasing relative strength of the employers in the new era of labour market deregulation and high unemployment.

Another outcome of the comparison between paid and unpaid overtime is related to the effect of children. Every child rises by 1 per cent the probability of paid overtime. This can be interpreted as the result of two opposite forces. On the one hand, the need for more money, and on the other hand, the need for more free time to spend with the children. When it comes to unpaid overtime, the existence of an additional child reduces by 6 per cent the probability of conducting them. It is more than obvious that those who have more children prefer paid to unpaid overtime and this would not have happened if they were expecting a compensation of any kind for the extra unpaid hours.

The comparison of manual to mental labour verifies the basic outcome of all available studies that white collar workers perform relatively more unpaid and less paid overtime compared with their blue collar counterparts. This finding is of course reinforcing the argument that the increasing complexity of work tasks is one of the reasons for the recent rise in unpaid overtime. It is also a verification of the validity of the procedure used in this research, as it is consistent with a typical finding of the literature. This finding, combined with the previous finding of a positive relationship between unpaid overtime and educational level makes quite problematic the theory of human capital, as mentioned in the theoretical analysis above.

Finally, a very useful result is that the usual working time is positively correlated and in almost the same degree, both with paid and unpaid overtime. This is a clear indication that a reduction in usual working time would decrease the probabilities of both paid and unpaid overtime. This means that a policy of working time reduction would not be offset by the increase in overtime (both paid and unpaid), so such a policy can be effective in Greece. It has to be reminded after all that Greek workers are among the hardest working in the European Union (EU) in terms of working time and that working time was increasing in Greece at least until the break out of the economic crisis.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Unpaid Overtime: Consequences and Theoretical Explanations
  5. Overtime and Labour Flexibility Regulations in Greece
  6. Determinants of Unpaid Overtime in Greece
  7. Conclusions
  8. References

Unpaid overtime is rising in most of the developed economies, so it has attracted the interest of many researchers, who are attempting to find the determining factors of the phenomenon, which can also help to provide the necessary theoretical explanations. For the conventional neoclassical economic theory, there should be no unpaid working time, so the explanation given is that this time is not really unpaid but it is paid in a different way. A variety of explanations were provided inside this framework to justify the non-existence of unpaid working time. On the other side, heterodox and Marxists paradigms argue that unpaid overtime is as natural as unpaid working time itself: it constitutes the foundation of capitalists' profits. Under this explanation, increasing unpaid overtime is the proof that capital is attempting to increase the appropriation of absolute surplus-value under the contemporary conditions of prolonged economic crisis.

This article is an attempt to give an answer to the above question, that is, if the unpaid overtime is really unpaid or not. It estimates the determining factors of paid and unpaid overtime in Greece, using the data from the European LFS. The comparison of the determinants of paid and unpaid overtime provides a better view of the attitude of different segments of the working class against both paid and unpaid overtime. The main result is that there is a substantially different attitude towards paid and unpaid overtime by these different segments, providing strong evidence that unpaid overtime is not voluntarily offered, thus it is really unpaid. An additional result is that supervisors and highly educated workers provide more unpaid overtime, enforcing the explanation of the complexity of work tasks and at the same time weakening the explanatory power of human capital theory.

It has to be noticed that the ‘complexity of work tasks’ explanation of unpaid overtime is not in contradiction to the ‘rising exploitation’ explanation: they both describe parallel developments of the modern workplace, developments with common denominator the increase of working time in general and unpaid working time in particular. A third conclusion of the article is that real working time is positively correlated both to paid and unpaid overtime, which means that a working time reduction policy will not be offset by an increase in overtime, thus can be effective.

The above results provide further support to the arguments that working time decrease has halted in the modern western economies and an opposite trend of working time increase made its appearance. This opposite trend does not seem to be the outcome of workers' will, but the result of an increasing relative strength of the employers in a rapidly changing workplace environment.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Unpaid Overtime: Consequences and Theoretical Explanations
  5. Overtime and Labour Flexibility Regulations in Greece
  6. Determinants of Unpaid Overtime in Greece
  7. Conclusions
  8. References
  • Akerlof, G. (1982), ‘Labor Contracts as a Partial Gift Exchange’, Quarterly Journal of Economics, 97, 4, 543569.
  • Anger, S. (2005), ‘Working Time as an Investment?- the Effects of Unpaid Overtime on Wages, Promotions, and Layoffs’, German Institute for Economic Research (DIW Berlin) Discussion Papers.
  • Anger, S. (2008), ‘Overtime Work as a Signalling Device’, Scottish Journal of Political Economy, 55, 2, 167189.
  • Aoki, M. (1984), The Co-operative Game Theory of the Firm (Oxford, Clarendon Press).
  • Bell, D. and R. Hart (1999a), ‘Unpaid Work’, Economica, 66, 262, 271290.
    Direct Link:
  • Bell, D. and R. Hart (1999b), ‘Overtime Working in an Unregulated Labour Market’, IZA Discussion Paper No 44.
  • Bell, D., R. Hart, O. Hubler and W. Schwerdt (2000), ‘Paid and Unpaid Overtime Work in Germany and the UK’, IZA Discussion Paper, 133.
  • Bell, L. and R. Freeman (2001), ‘The Incentive for Working Hard: Explaining Hours Worked Differences in the Us and Germany’, Labour Economics, 8, 2, 181202.
  • Bluestone, B. and S. Rose (2000), ‘The Enigma of Working Time Trends’, in L. Golden and D. Figart (eds), Working Time: International Trends (London, Routledge) pp. 2137.
  • Booth, A., M. Francesconi and J. Frank (2003), ‘A Sticky Floors Model of Promotion, Pay and Gender’, European Economic Review, 47, 2, 295322.
  • Bratti, M. and S. Staffolani (2007), ‘Effort-based Career Opportunities and Working Time’, International Journal of Manpower, 28, 6, 489512.
  • Burtless, G. (2001), ‘The Greek Labour Market’, in R. Bryant , N. Garganas and G. Tavlas (eds), Greece's Economic Performance and Prospects (Athens, Bank of Greece/The Brookings Institutions) pp. 453493.
  • Campbell, D. and F. Green (2002), ‘The Long-term Pay-off from Working Longer Hours’, Department of Economics Discussion Paper, University of Kent, 02,05.
  • Cramer, J. S. (2004), ‘The Early Origins of the Logit Model’, Studies in History and Philosophy of Biological and Biomedical Science, 35, 4, 613626.
  • van Echtelt, P. E., A. C. Glebbeek and S. M. Lindenberg (2006), ‘The New Lumpiness of Work: Explaining the Mismatch between Actual and Preferred Working Hours’, Work, Employment and Society, 20, 3, 493512.
  • Gialis, S. and E. Karnavou (2008), ‘Dimensions of Atypical Forms of Employment in Thessaloniki, Greece’, International Journal of Urban and Regional Research, 32, 4, 882902.
  • Gujarati, D. N. (2003), Basic Econometrics (New York, McGraw-Hill).
  • Hetrick, R. (2000), ‘Analyzing the Recent upward Surge in Overtime Hours’, Monthly Labour Review, February.
  • Leete, L. and J. Schor (1994), ‘Assessing the Time Squeeze Hypothesis: Estimates of Market and Non Market Hours in the United States, 1969–1989’, Industrial Relations, 33, 1, 2543.
  • Meyer, A. and M. Wallette (2005), ‘Absence of Absenteeism and Overtime Work Signalling Factors for Temporary Workers?’, Department of Economics, Lund University, Working Paper Series, 15.
  • Mihail, D. (2003), ‘Atypical Working in Corporate Greece’, Employee Relations, 25, 5, 470489.
  • Mouriki, A. (2001), Labour Relations and Social Dialogue in Greece: The Difficult Path from State Custody to Independence, The Social Portrait of Greece (Athens, EKKE) pp. 145152.
  • Pannenberg, M. (2005), ‘Long Term Effects of Unpaid Overtime, Evidence from West Germany’, Scottish Journal of Political Economy, 52, 2, 177193.
  • Perlow, L. A. (1999), ‘The Time Famine: Towards a Sociology of Working Time’, Administrative Science Quarterly, 44, 1, 5781.
  • Pindyck, R. S. and D. L. Rubinfield (1981), Econometric Models and Economic Forecasts, 2nd edn (New York, McGraw-Hill).
  • Pollert, A. and A. Charlwood (2009), ‘The Vulnerable Worker in Britain and Problems at Work’, Work, Employment and Society, 23, 2, 343362.
  • Schor, J. (1991), The Overworked American: The Unexpected Decline of Leisure (New York, Basic Books).
  • Trejo, S. (1991), ‘Compensating Differentials and Overtime Pay Regulations’, American Economic Review, 81, 4, 719740.
  • Trejo, S. (2001), ‘Does the Statutory Overtime Premium Discourage Long Workweeks?’, IZA Discussion Paper No. 373.