REVIEW OF ECONOMIC PERSPECTIVES NÁRODOHOSPODÁŘSKÝ OBZOR VOL. 16, ISSUE 2, 2016, pp. 121 134, DOI: 10.1515/revecp-2016-0009 Gender Wage Dfferences n the Czech Publc Sector: A Mcro-level Case Veronka Hedja 1 Abstract: The am of ths study s to estmate the unexplaned gender pay gap n ndvdual departments of a Czech hosptal, to fnd out whether ths gap vares between departments and, f so, to dentfy the possble causes of these dfferences. To estmate the unexplaned part of the gender pay gap, we use the average treatment effect on the treated (ATT), and to dentfy the causes behnd dfferences n the unexplaned gender pay gap we use a lnear regresson model. We fnd that the ATT vares sgnfcantly between departments. To explan these dfferences, we use selected characterstcs of the departments: the department's sze, the proporton of women n the department, and the gender of the departmental head. We come to the concluson that women's wages ncrease relatve to male wages as the proporton of female employees grows. On the other hand, the unexplaned gender wage gap s not proven to be lower n smaller or female-led departments. Key words: Gender pay gap, Gender, Healthcare, Publc sector, Wage dfferences JEL Classfcaton: I14, J16, J24, J71 Receved: 12 August 2015 / Accepted: 15 March 2016 / Sent for Publcaton: 16 June 2016 Introducton The exstence of wage dfferences between men and women s a well-known fact. Accordng to the Czech Statstcal Offce (2014), data from 2012 showed that women n the Czech Republc earned 78 percent of average male gross monthly earnngs, wth varaton across the dfferent sectors of the Czech economy. Usng the CZ-NACE classfcaton, these gender wage dfferences were found to be largest n groups K: Fnancal and nsurance actvtes, Q: Human health and socal care actvtes and J: Informaton and communcaton. Here, women earn less than 70 percent of the average male wage. Meanwhle, the lowest gender wage dfferences were dentfed n sector H: Transportaton and storage, where female earnngs amount to more than 95 percent of male earnngs. Ths data n tself gves no evdence as to the exstence of wage dscrmnaton aganst women. To a certan extent, gender wage dsparty can be explaned by observed dfferences n the characterstcs of men and women. It s the part of the gender pay gap (GPG) 1 Vysoké učení techncké v Brně, Fakulta podnkatelská, Kolejní 2906/4, 612 00 Brno, hedja@fbm.vutbr.cz. 2016 by the authors; lcensee Revew of Economc Perspectves / Národohospodářský obzor, Masaryk Unversty, Faculty of Economcs and Admnstraton, Brno, Czech Republc. Ths artcle s an open access artcle dstrbuted under the terms and condtons of the Creatve Commons Attrbuton 3.0 lcense, Attrbuton Non Commercal No Dervatves.
REVIEW OF ECONOMIC PERSPECTIVES that remans unexplaned despte these observable dfferences that s usually consdered a measurement of wage dscrmnaton aganst women. Ths s known as the remuneraton effect, the effect of dscrmnaton or most frequently smply as the unexplaned gender pay gap. A number of emprcal studes have been devoted to estmatng the unexplaned gender pay gap n the Czech Republc (Jurajda 2003, Mysíková 2012). These have shown that dfferences n the average characterstcs of men and women only account for a small part of the raw gender pay gap, and that constant characterstcs lke occupaton and sector are mportant factors n explanng dspartes between male and female wages. The unexplaned gender pay gap also vares accordng to sector publc and prvate. Jurajda (2003) analysed the gender pay gap n the Czech and Slovak prvate and publc sectors. He concluded that the unexplaned gender pay gaps dffered dramatcally between these sectors. In the prvate sector t represented more than 60 percent of the raw gender pay gap, whle n the publc sector t amounted to approxmately 40 percent of the raw gender pay gap. Hedja (2014) examned dfferences n the unexplaned gender pay gap between dfferent sectors of the Czech economy and concluded that the unexplaned gender pay gap was lower n the publc sector than n the prvate sector, although t was stll present n the publc sector. These dfferences may be explaned by dfferent wage settng mechansms n the prvate and publc sectors and the easer mplementaton of ant-dscrmnaton polces n the publc sector. However the exstng unexplaned gender pay gap cannot be entrely attrbuted to wage dscrmnaton aganst women; t only represents the upper lmt. Part of the unexplaned wage dfference may be explaned by dfferences n the unobserved characterstcs of ndvduals (e.g. talent, work effort, loyalty) or by characterstcs unavalable from offcal data (e.g. actual job content, responsblty). The second ssue could be mnor and s related to the use of detaled data from just one frm or selected frms. Hedja and Musl (2012) used data from a Czech hosptal and conclude that the unexplaned gender pay gap s neglgble, as t amounts to only 2.6 percent. On the other hand, usng data from a prvate sector company Hedja and Musl (2011) concluded that the dfferent observable characterstcs of men and women could explan only a small part of the gender pay gap, f any: they found the unexplaned gender wage gap to be approxmately 20 percent to the dsadvantage of women. Recently, studes have begun to appear that focus on dentfyng the causes behnd varaton n unexplaned gender pay gaps. Arulampalam et al. (2007) and Chrstofdes at al. (2013) analyse gender wage dfferences n selected European countres. They use the selected countres' specfc polces amed at reconclng work and famly lfe and ther wage-settng nsttutons to explan dfferences n gender pay gaps between the countres. They dentfy the relatonshp between the unexplaned gender pay gap and country-specfc polces and nsttutons. Hedja (2014) searches for the causes of the dfferences n unexplaned gender pay gap wthn ndvdual sectors of the Czech economy. Usng a lnear regresson model wth the proporton of women, the proporton of female managers, the proporton of small frms and ownershp (publc or prvate) as explanatory varables, she ndcates that ownershp and the proporton of women n management contrbute sgnfcantly to varaton n the unexplaned gender wage gap. We have taken nspraton from these studes and use the maxmum mcro-level atttude. 122
Volume 16, Issue 2, 2016 Ths study s devoted to wage dfferences between men and women workng n a hosptal. It ams to dentfy the unexplaned gender pay gap n ndvdual departments and to examne whether ths vares and, f so, dentfy the possble causes behnd the dfferences. Lke Hedja and Musl (2012), we use data from a Czech hosptal; however n ths case we use a dfferent dataset (data from a larger hosptal). Unlke prevous studes, ths study ams not only to detect the exstence of potental wage dscrmnaton aganst women n the selected frm but also to dentfy the causes of exstng gender wage dfferences at the frm level. The frst secton of the paper descrbes the data used and methods appled. In order to estmate the unexplaned gender pay gap, we use the average treatment effect on the treated (ATT) and to dentfy the causes of dfferences n the unexplaned gender pay gap we use a lnear regresson model. The second secton reports the results obtaned and dscusses these, comparng them wth conclusons reached by prevous studes. We fnd that female wages ncrease relatve to male wages as the proporton of female employees grows, and wth female heads of department. Data and Methods We use data from a Czech hosptal that was wllng to provde us wth the necessary nformaton. The hosptal n queston s a large unversty hosptal, whch boasts over 2,000 beds and has more than 5,000 employees. Table 1 Raw gender pay gap by occupaton Occupaton Number of employees Proporton of women (%) Raw gender pay gap techncal and economc worker 499 78.2-0.293 worker 471 39.1-0.241 pharmacst 32 90.6 0.077 pharmaceutcal assstant 42 100.0 - other professonal 166 80.7-0.034 doctor 928 56.7-0.167 lower medcal worker 145 91.7-0.039 mdwfe 147 100.0 - orderly 633 83.7-0.107 paramedcal worker 718 95.1 0.170 nurse 1,639 97.1 0.092 medcal laboratory techncan 262 97.3 0.143 Total 5,682 81.7-0.157 Note: Raw gender pay gap s calculated as the dfference between the logarthm of women's average hourly wage and the logarthm of men's average hourly wage n the gven occupaton. Source: Data from hosptal, author s computatons. 123
REVIEW OF ECONOMIC PERSPECTIVES The data provded s admnstratve n character, and dates back to 2010. Employees wth long-term llnesses or who were on maternty or parental leave or workng on a short-term contract were excluded from the sample. The followng characterstcs of the hosptal's employees were measured: gender, age, hghest level of educaton, occupaton, number of years worked at the hosptal, workng tme (where 1 ndcates full tme), department, hours worked per year, overtme hours, days of sck leave, days of annual leave and annual gross wage. The gross hourly wage s calculated as a rato of the annual gross wage (ncludng bonuses) and the sum of worked hours (ncludng overtme) and hours of annual leave (days of annual leave*8*workng tme). Our dataset enables us to dentfy 12 groups of employees workng n more than 60 departments: worker, orderly, techncal and economc worker, lower medcal worker, medcal laboratory techncan, paramedcal worker, nurse, mdwfe, pharmacst, pharmaceutcal assstant, other professonal, doctor. The number of employees and proporton of women n these ndvdual occupatons are shown n table 1. Female employees domnate n most of the occupatons, wth the excepton of workers. In seven cases women represent more than 90 percent of the employees n the gven occupaton (lower medcal worker, medcal laboratory techncan, paramedcal worker, nurse, mdwfe, pharmacst, pharmaceutcal assstant). In order to work wth a relatvely large and sample of men and women, and one that would be homogenous n terms of job content, we selected to work only wth orderles and doctors. These groups of employees are present n a large number of departments, whle they also provde us wth an nterestng comparson between two occupatons at the top and bottom of the herarchy n terms of educaton and sklls requred. Table 2 Observed characterstcs of male and female orderles Characterstc Mean men Mean women Age (year) 39.74 43.57 Educaton (%) - lower secondary 14.56 26.1 - upper secondary wthout certfcate 0.00 0.34 - upper secondary wth certfcate 42.72 53.04 - upper secondary - school dploma 37.86 19.93 - tertary - hgher professonal 0.97 0 - tertary bachelors level 0.97 0 - tertary - masters and doctoral level 2.91 0.68 Full tme 0.98 0.98 Years at the hosptal 8.29 8.34 Sckness (days) 5.93 10.73 Overtme (hours) 54.04 29.21 Female manager (%) 41.75 15.20 N 103 296 Source: Data from hosptal. 124
Volume 16, Issue 2, 2016 As we had set out to examne dfferences n the unexplaned gender pay gap between ndvdual departments, we excluded departments whch had only male or only female orderles and doctors. Our fnal data set covers 1279 employees (399 orderles and 880 doctors) workng n 45 departments. The average characterstcs of these orderles and doctors (dvded by gender) are shown n tables 2 and 3. Table 3 Observed characterstcs of male and female doctors Characterstc Mean men Mean women Age (year) 40.33 38.10 Educaton (%) - tertary - masters and doctoral level 100 100 Full tme 0.63 0.66 Years at the hosptal 11.57 10.74 Sckness (days) 5.03 14.30 Overtme (hours) 169.30 136.71 Female manager (%) 13.49 21.91 N 378 502 Source: Data from hosptal. To dentfy the unexplaned gender pay gap and fnd out whether t vares between departments, we estmate the unexplaned gender pay gap at a department level usng an estmaton of the average treatment effect on the treated (ATT). To receve a maxmum mcro-level vew, we estmate these separately for our two selected occupatons: orderly and doctor. The ATT s the average beneft resultng from beng treated. In our case, the ATT s the mean effect for women n the form of a lower wage resultng from beng a woman. We count the ATT for each department usng the followng calculaton formula for ATT: ATT E y ( 1) y (0) T 1. (1) where T s the bnary treatment ndcator, T = 1 denotes treatment and T = 0 otherwse, y(1) s the potental outcome wth treatment and y(0) s the potental outcome wthout treatment. In our case, beng treated means beng a woman. We can rewrte the ATT as: y ( 1) T 1 E y (0) T 1. ATT E (2) where ATT represents the gender pay gap that cannot be explaned by the dfferent characterstcs of men and women. The term E y ( 1) T 1 s the sample average of the logarthm of women's gross wages and ( 0) 1 E y T s the sample average of the logarthm of what women's gross wages would be f they were men. From our sample, we know the frst term on the rght hand sde of equaton 2, the sample average of the logarthm of women's hourly gross wages. The second term, the aver- 125
REVIEW OF ECONOMIC PERSPECTIVES age of the logarthm of what women's hourly gross wages would be f they were men, needs to be estmated. There are several possble ways of makng such an estmate - for more detals see Wooldrdge (2002). We estmate ths usng coeffcents of the male wage functon y T. X u. 0 0 where y s the logarthm of the male gross hourly wage, β 0 s the vector of the coeffcents of the wage functon, X s the vector of the chosen observed characterstcs of men and u s a dsturbance term. In most departments the number of male employees (orderles or doctors) s too small for us to be able to estmate the coeffcents of the male wage functon for each department separately and so we nstead estmate the male wage functon for the whole sample of male orderles/doctors usng department as one of the explanatory varables. Apart from department (a dummy varable denotng the department where the employee works), the other explanatory varables used are: age (the age of the employee n years); educaton level (a dummy varable for the hghest level the employee has attaned); years n the frm (the number of years the employee has worked n the hosptal), squared years n the frm, full tme (a dummy varable for full tme postons), sckness (the number of days the employee has spent on sck leave) and overtme hours (the number of overtme hours the employee works). The dummy varables used for the educaton level dstngush seven stages of educaton: lower secondary educaton, upper secondary vocatonal educaton wthout certfcate, upper secondary vocatonal educaton wth certfcate, upper secondary educaton wth school dploma, tertary hgher professonal educaton, bachelors level tertary educaton and masters and doctoral level tertary educaton. The explanatory varable for educaton level s only used n the wage functon for orderles (because all doctors have the same level of educaton). The motvaton for for the ncluson of the last two explanatory varables (sck leave and overtme hours) s the fact that more overtme hours and a smaller number of sck leave days are both assocated wth a hgher average hourly wage. The ATT s estmated as the dfference between the average of the logarthm of women's gross hourly wages and the average of the predcted values of female wages computed usng estmated coeffcents for the male wage functon. y ( 1) T 1 E 0. X T 1. ATT E (4) Where. 1 E s the mean of the predcted wages (the logarthm of the gross 0 X T hourly wage) for every woman n the sample. The estmated ATT represents the unexplaned gender pay gap. The negatve sgn ndcates that women receve relatvely lower wages compared to men. The results obtaned are the same as those found when usng the Oaxaca-Blnder decomposton wth men's wages as the equlbrum wage. We estmate the ATT separately for each ndvdual department at the hosptal. To dentfy the possble causes of varaton n the estmated ATT, we construct a lnear regresson model where we use the calculated ATT for the ndvdual departments as the de- (3) 126
Volume 16, Issue 2, 2016 pendent varable. To explan varaton n the unexplaned gender pay gap between departments, we use smlar control varables to those used n Hedja (2014), but at the departmental level: these are the gender composton of the department, the department sze and a dummy for a female head of department. ATT j (5) 1. proportonofwomen j 2. femalemanager j 3. sze j u j. Where j denotes the department, proporton of women s the proporton of female orderles/doctors workng n the department, female manager s a dummy varable for the department havng a female head, sze denotes the number of employees n the department, and u j s the dsturbance term. The unexplaned gender pay gap may be affected to some extent by the proporton of women n the department: the unexplaned pay gap between male and female employees n the same poston could be lower n departments wth ether a very low or very hgh proporton of female representatves. Women may have more male characterstcs n departments wth a low proporton of women; they mght then be perceved as men and ths mght be reflected n ther salary. The same could be vald for men workng as orderles/doctors n departments wth a domnant share of female orderles/doctors. We use a dummy for the proporton of female orderles/doctors, whch dstngushes three groups: departments wth 0-20 percent female orderles/doctors, departments wth 40-80 percent female orderles/doctors, and departments wth more than 80 percent female orderles/doctors. The unexplaned gender pay gap could be lower n female-led departments. Some emprcal studes have confrmed that the presence of a female manager leads to a decrease n gender pay gap (Hultn and Szulkn 1999 and 2003, Cardoso and Wnter-Ebmer 2010, Hedja 2015). These conclusons are n accordance wth the socal dentty theory that states that ndvduals tend to favour members of ther own group over members of other groups (Tajfel 1982, Tajfel and Turner 1979). Applng ths theory to remuneraton, female heads of department may evaluate female employees better than male employees. Hence, we use a dummy for a female head of department as one of our control varables. When examnng the mpact of the manager s gender on wage dscrmnaton aganst women, t s essental to take nto consderaton managers' wage-settng powers. In the Czech publc health sector, wages are regulated by Government Regulaton No. 564/2006 on the salares of employees n publc servce and admnstraton. Ths regulaton defnes wage classes and grades, assgns workers to these, and sets the base gross wage for each class and grade. There s no regulated maxmum for bonuses, and at the hosptal we are studyng, bonuses are not even controlled by any nternal regulatons. Ths means that an employee's actual gross wage, ncludng bonuses, depends on the bonuses granted by the head of department, who s solely lmted n ther allocaton by the sze of budget avalable n that department. In other words, n ths case, heads of department have relatve flexblty n wage formaton. Unexplaned gender pay dfferences could also be lower n small departments than n larger ones, because employees tend to know each other better n small departments and may also be nclned to dsclose nformaton about ther wages. In such cases, t would be harder for managers to dscrmnate between women's and men's wages whle man- 127
REVIEW OF ECONOMIC PERSPECTIVES tanng a good workng envronment and good relatonshps n the department. Ths s why we use a dummy for department sze (number of employees). To estmate the coeffcents of the model we used the OLS estmator wth robust standard errors. Results and Dscusson Usng equaton 4 we estmate the average treatment effect on the treated for each department at the hosptal. We estmate these separately for orderles and doctors. The results are shown n table 4, where besde the ATT we also report the raw gender pay gap, calculated as the dfference between the logarthm of the average hourly wage of female orderles/doctors and the logarthm of the average hourly wage of male orderles/doctors. The ATT results represent the unexplaned gender pay gap. The negatve sgn ndcates that women receve relatvely lower wages than men. Table 4 Raw gender pay gap and average treatment effect on the treated orderles doctors orderles doctors Dep Raw ATT Raw ATT Dep Raw ATT Raw ATT GPG GPG GPG GPG 1-0.307-0.233 - - 24 - - 0.120 0.138 2 0.136 0.048-0.222-0.057 25 0.156 0.092-0.168-0.015 3-0.079-0.028-0.219-0.023 26 - - -0.443 0.031 4 - - -0.100 0.040 27-0.085-0.065 - - 5 - - 0.028 0.097 28-0.011-0.067 - - 6 0.092 0.039 0.000-0.051 29 - - 0.200 0.074 7 0.216 0.182 0.226 0.390 30 - - 0.166 0.134 8-0.071-0.154-0.305-0.113 31 0.166 0.138-0.319-0.168 9 - - -0.156-0.111 32 - - -0.151-0.207 10 0.243 0.205-0.108-0.060 33 - - 0.033 0.057 11 - - -0.234-0.132 34 - - 0.538 0.092 12 0.061-0.022-0.129-0.149 35 - - -0.214 0.019 13-0.100 0.010-0.374 0.008 36 - - 0.110 0.103 14 - - -0.196-0.049 37 - - -0.011 0.004 15 - - -0.213 0.018 38 - - -0.266 0.052 16 - - 0.068 0.037 39 - - -0.345-0.044 17 - - -0.001 0.014 40 - - -0.123-0.188 18 - - 0.484 0.641 41 - - 0.019 0.128 19 0.161 0.194 0.027 0.016 42 - - -0.234-0.066 20 - - -0.097 0.094 43-0.197-0.179-0.290-0.152 21 - - -0.040-0.120 44 - - 0.052 0.015 22-0.079-0.072 - - 45 0.297 0.331 - - 23 0.201-0.156 0.001 0.177 Source: Author's computatons. The department-specfc ATT results are lower (n absolute value) than the raw gender pay gap n most cases; ths means that dfferences n characterstcs between men and women explan part of the raw gender wage dfference. Nevertheless, part of the raw wage dfference between male and female employees remans unexplaned n some 128
Volume 16, Issue 2, 2016 departments, both for orderles and doctors, and ths phenomenon s qute vared n extent. Furthermore, n most cases, the ATT result vares for orderles and doctors workng n the same departments. Where orderles are concerned, the estmated ATT ranges from -0.233 for department 1 to 0.331 for department 45. Ths ndcates that male orderles earn about 23 percent more than female orderles n department 1, whle n department 45 female orderles earn about 33 percent more than ther male counterparts; these dfferences cannot be explaned by the known observed characterstcs of the men and women concerned (ther age, educaton level, number of years at the hosptal, sckness leave, overtme hours and workng tme). The estmated ATT was postve n nne of the eghteen departments examned that employ at least one male orderly. Ths ndcates that n half the examned departments, female orderles earn more than ther male co-workers wth smlar observed characterstcs. The stuaton for doctors s very smlar: the ATT ranges from -0.207 for department 32 to 0.641 for department 18, and s postve n twenty three out of forty departments (more than half). Table 5 Lnear regresson model - orderles Model (1) (2) (3) (4) (5) Sze - 50-100 employees -0.320*** (0.056) - 100-150 employees -0.312*** (0.077) - 150 + employees -0.381*** (0.038) Proporton of women - 0-20 percent female orderles - 80 + percent female orderles Proporton of women (orderles) - - -0.378*** (0.030) -0.349*** (0.093) -0.398*** (0.046) -0.215** (0.093) 0.087 (0.058) -0.417*** (0.069) -0.349*** (0.097) -0.398*** (0.048) -0.292 (0.164) 0.087 (0.061) -0.405*** (0.049) -0.364*** (0.048) -0.431*** (0.042) - - - 0.443*** (0.089) Female manager - - 0.077 (0.105) Constant 0.331*** (0.000) 0.331*** (0.000) 0.331*** (0.000) -0.466*** (0.089) -0.390*** (0.057) -0.442*** (0.047) - - - - 0.542*** (0.124) - 0.084 (0.084) 0.036 (0.060) -0.030 (0.083) R 2 0.303 0.534 0.557 0.556 0.584 N 18 18 18 18 18 Note: ***sgnfcant at the 1 per cent level, **sgnfcant at the 5 per cent level, *sgnfcant at the 10 per cent level, robust standard errors n brackets Source: Author's computatons. 129
REVIEW OF ECONOMIC PERSPECTIVES If we focus only on departments that employ both doctors and orderles, we observe that the ATT s only smlar for doctors and orderles n four such departments and that n many cases, orderles take postve values of ATT and doctors negatve values. Ths fact ether ndcates the exstence of wage dscrmnaton aganst male orderles and female doctors, or an unobservable hgher qualty among female orderles and male doctors n comparson to ther co-workers of the opposte gender. We use a lnear regresson model (equaton 5) to dentfy the causes of the dfferences n unexplaned gender pay gap between the departments. Tables 5 and 6 present the results. We test the sze of the department, the proporton of women n the department and the gender of the head of department as factors that mght explan dspartes n the departmental ATT results. Table 6 Lnear regresson model - doctors Model (1) (2) (3) (4) (5) Sze - 50-100 employees 0.006 (0.050) - 100-150 employees 0.017 (0.071) - 150 + employees -0.124*** (0.044) Proporton of women - 0-20 percent female doctors - 80 + percent female doctors Proporton of women (doctors) - 0.040 (0.053) 0.047 (0.069) -0.069 (0.056) -0.001 (0.086) - 0.279** (0.130) 0.046 (0.054) 0.059 (0.074) -0.044 (0.066) 0.016 (0.090) 0.268** (0.113) 0.000 (0.051) 0.054 (0.091) -0.126** (0.055) - - - 0.267 (0.200) Female manager - - 0.064 (0.584) Constant 0.029 (0.027) -0.027 (0.044) -0.052 (0.055) - 0.007 (0.053) 0.062 (0.096) -0.101* (0.059) - - - 0.229 (0.168) - 0.063 (0.070) -0.134 (0.136) -0.136 (0.136) R 2 0.099 0.390 0.419 0.203 0.230 N 40 40 40 40 40 Note: ***sgnfcant at the 1 per cent level, **sgnfcant at the 5 per cent level, *sgnfcant at the 10 per cent level, robust standard errors n brackets Source: Author's computatons. 130
Volume 16, Issue 2, 2016 Departmental sze s measured by the department's total number of employees, n four sze categores. 2 The results for orderles and doctors are very smlar. The dsparty n orderles' earnngs to the dsadvantage to female orderles ncreases sgnfcantly (.e. the ATT decreases) n all three sze categores compared wth the lowest sze departments (whch have fewer than 50 employees), although the effect of departmental sze does not ncrease as departments grow beyond 100 employees. For doctors, a statstcally sgnfcant ncrease n unexplaned gender pay gap (decreased ATT) s only detected n departments wth more than 150 employees (the largest sze category). To examne the effect of the proporton of female orderles/doctors n the department on varaton n ATT among departments, we use dummy varables for three groups of departments: those wth less than 20 percent female orderles/doctors, those wth 20 to 80 percent female orderles/doctors and those wth more than 80 percent female orderles/doctors. 3 Model 2 n table 5 shows the results for orderles and model 2 n table 6 shows the results for doctors. The ATT s lower n departments wth less than 20 percent female orderles, whch ndcates a hgher unexplaned gender pay gap n these cases compared wth departments consstng of 20-80 percent female orderles. In the case of departments wth more than 80 percent female orderles the effect s not sgnfcant. The stuaton s dfferent for doctors: the ATT ncreases sgnfcantly n departments wth more than 80 percent female doctors whch ndcates that n these departments there s lower potental wage dscrmnaton aganst women. There s no clear evdence that gender wage dfferences are lower n departments wth an extremely hgh or low proporton of female employees. However, n both cases (and especally for doctors) the ATT tends to ncrease wth an ncrease n the proporton of female employees (orderles/doctors). Hence, we replace the prevously-used dummy varable wth the actual proporton of female orderles/doctors (model 4 n tables 5 and 6). The results show that a 10 percentage pont ncrease n the proporton of female orderles decreases the unexplaned gender pay gap (ncreases the ATT) by 4.4 percentage ponts, whle a 10 percentage pont ncrease n the proporton of female doctors decreases the unexplaned gender pay gap (ncreases the ATT) by 2.7 percentage ponts, although ths latter result s not statstcally sgnfcant. These results are dfferent from those reported n Hedja (2014) on the effect of the proporton of women on the unexplaned gender pay gap wthn ndvdual sectors of the Czech economy, where the effect was shown to be neglgble. Ths could be due to the fact that the proporton of women was not studed at the ndvdual frm or department level but only at the level of whole ndustres. Fnally we report the results of the model estmaton usng a dummy varable for a female head of department (models 3 and 5 n tables 5 and 6). Havng a female head of department ncreases the ATT by approxmately 8 percentage ponts for orderles and 2 We also tested the number of orderles/doctors n the department as an explanatory varable but ths dd not prove to be a statstcally sgnfcant factor n explanng dfferences n the ATT. 3 We also tred usng the proporton of female employees (n all postons) n the department as an explanatory varable, but found that ths was not statstcally sgnfcant. 131
REVIEW OF ECONOMIC PERSPECTIVES by 6 percentage ponts for doctors. These fndngs are consstent wth Hultn and Szulkn (1999 and 2003), Cohen and Huffman (2007), Cardoso and Wnter-Ebmer (2010) and Hedja (2015), all of whom concluded that the gender wage gap s reduced wth an ncrease n female management. However, n our case, these results are not found to be statstcally sgnfcant. Concluson The am of ths study was to dentfy the unexplaned gender pay gap n ndvdual departments of a Czech hosptal, to fnd out whether ths vares across the departments, and f so, to dentfy the possble causes behnd these dfferences. To receve a maxmum mcro-level vew we focused on just two occupatons wthn the hosptal: orderles and doctors. Frstly, we estmated the unexplaned gender pay gap between male and female orderles and doctors at the selected hosptal usng an estmaton of the ATT. We found that the estmated ATT vared sgnfcantly between departments, rangng from -0.233 to 0.331 for orderles and -0.207 to 0.641 for doctors. The ATT results proved to be postve n approxmately half the departments (for both occupatons), whch ndcates that female orderles/doctors earn more than ther male coworkers wth smlar observed characterstcs n half the departments examned, and vce versa. We then went on to use a lnear regresson model to dentfy the possble causes of the dfferences n ATT between the departments. We used department sze, the proporton of female orderles/doctors n the department and the gender of the head of department as control varables. Our results showed that the unexplaned gender pay gap decreased (the ATT ncreased) wth an ncrease n the proporton of female employees. A 10 percentage pont ncrease n the proporton of female orderles decreased the unexplaned gender pay gap (ncreased the ATT) by approxmately 5 percentage ponts. The effect was slghtly weaker for doctors: a 10 percentage pont ncrease n the proporton of female doctors n a gven department decreased the unexplaned gender pay gap by just 2 percentage ponts, and ths result was not statstcally sgnfcant. The unexplaned gender pay gap also declned when the head of department was female: a female head of department leads to an ncrease n women's wages n relaton to those of ther male coworkers by 8 percent for orderles and 6 percent for doctors. Nevertheless, these results were also not statstcally sgnfcant. Fnally, our results partly confrm the dea that wage dfferences between men and women ncrease wth growth n departmental. We dentfed a statstcally sgnfcant effect (for both orderles and doctors) n departments wth more than 150 employees. Fnally, we can conclude that our fndngs were very smlar for both orderles and doctors: n both cases, larger departments, hgher proportons of female employees, and havng a female head of department all resulted n decreasng the unexplaned gender pay gap. However, the effect of female managers was statstcally nsgnfcant n both cases, and the effect of the proporton of female employees was weaker and not statstcally sgnfcant for doctors. Our study used detaled data from a sngle hosptal to nvestgate gender pay gaps at departmental level. Ths enabled us to dentfy the effect of the head of department and 132
Volume 16, Issue 2, 2016 to mnmze bas from dfferent job contents n smlar occupatons. Usng data for a select group of employees only orderles and doctors enabled us to mnmze the effects on unexplaned gender pay gap that mght result from dfferences between partcular hosptal occupatons, and enabled us to make a comparson of the results between an occupaton wth low educaton and sklls requrements, and one wth extremely hgh requrements. On the other hand, usng data from only one hosptal (one company) reduces the explanatory power of our results, whch may not be vald across the Czech healthcare ndustry as a whole. Fundng: Ths paper was supported by nternal grant No. FP-S-15-2877 enttled The Selected Aspects of Fnancal Management n Internatonal Envronment at the Faculty of Busness and Management, Brno Unversty of Technology. Dsclosure statement: No potental conflct of nterest was reported by the author. References ARULAMPALAM, W., BOOTH, A. & BRYAN, M. (2007) Is there a glass celng over Europe? Explorng the gender pay gap across the wage dstrbuton. Industral & Labor Relatons Revew. 62(2). p. 163 186. DOI: 10.1177/001979390706000201 CARDOSO, A. R. & WINTER-EBMER, R. (2010) Female-Led Frms and Gender Wage Polces. Industral and Labour Relatons Revew. 64(1). p. 143-163. CHRISTOFIDES, L. N., POLYCARPOU, A. & VRACHIMIS, K. (2013)Gender wage gaps, stcky floors and glass celngs n Europe. Labour Economcs. 21(Aprl 2013). p. 86 102. COHEN, P. & HUFFMAN, M. (2007) Workng for the Woman? Female Managers and the Gender Wage Gap. Amercan Socologcal Revew. 72(5). p. 681-704. DOI: 10.1177/000312240707200502 Czech Statstcal Offce (2014) [onlne] Zaostřeno na ženy a muže 2013. [ct.20.9.2014]. Avalable on: http://www.czso.cz/csu/2013edcnplan.nsf/publ/1413-13-r_2013 HEDIJA, V. & MUSIL, P. (2012) Gender wage dfferences n the selected Czech publc sector company. Acta Unverstats Agrcultura et Slvculturae Mendelanae Brunenss. 60(7). p. 81-88. HEDIJA, V. & MUSIL, P. (2011) Gender pay gap applcaton n the specfc enterprse. Národohospodářský obzor/revew of Economc Perspectves. 11(4). p. 223-236. HEDIJA, V. (2014) Gender Pay Gap n Dfferent Sectors of Czech Economy. In 32nd Internatonal Conference MME 2014. Olomouc: Palacky Unversty. p. 275-280. HEDIJA, V. (2015) The Effect of Female Managers on Gender Wage Dfferences. Prague Economc Papers. 24(1). p. 38-59. 133
REVIEW OF ECONOMIC PERSPECTIVES HULTIN, M. & SZULKIN, R. (1999) Wages and Unequal Access to Organzatonal Power: An Emprcal Test of Gender Dscrmnaton. Admnstratve Scence Quarterly. 44(3). p. 453-472. DOI: 10.2307/2666958 HULTIN, M. & SZULKIN, R. (2003) Mechansms of Inequalty Unequal Access to Organzatonal Power and the Gender Wage Gap. European Socologcal Revew. 19(2). p. 143-159. DOI: 10.1093/esr/19.2.143 JURAJDA, S. (2003) Gender Wage Gap and Segregaton n Enterprses and the Publc Sector n Late Transton Countres. Journal of Comparatve Economcs. 31 (2). p. 199-222. DOI: 10.1016/S0147-5967(03)00040-4 MYSÍKOVÁ, M. (2012) Gender Wage Gap n the Czech Republc and Central European Countres. Prague Economc Papers. 2012/3(3). p. 328-346. TAJFEL, H. & TURNER, J. C. (1979) An Integratve Theory of Intergroup Conflct. In The Socal Psychology of Intergroup Relatons. Monterey, CA: Brooks-Cole. p. 33-47. TAJFEL, H. (1982) Socal Psychology of Intergroup Relatons. Annual Revew of Psychology. 33(1). p. 1 39. DOI: 10.1146/annurev.ps.33.020182.000245 WOOLDRIDGE, J. (2002) Econometrc Analyss of Cross Secton and Panel Data. Cambrdge: MIT Press. 134