What Determines the Variation in Doctors’ Wages?: A Study of Swedish Physicians

(1)

Thesis Work D

Author: Rebecka Cederholm Advisor: Kathleen Cannings Spring Semester 2007

What Determines the Variation in Doctors’ Wages?

- A Study of Swedish Physicians

(2)

Abstract

During the Fall of 2001 a survey created by Kathleen Cannings and sponsored by the Swedish Medical Association was sent out to a random sample of 1 out of every 12 medical doctors in Sweden. Using this data, linear regressions have been estimates to study the effects of variables such as age, gender, and unionization on the wages of Swedish doctors. The results indicate that variables such as age, tenure, and union bargaining all have a positive impact on wages. The relative wage advantage was around 7 percent for both age and union bargaining, while the effect of tenure was lower by about 5 percent. These three variables have more positive effects for male doctors, which suggest that gender discrimination is still a reality in the medical field. During 2001, male physicians could expect an almost 6 percent higher average wage than their female colleges.

Keywords: Wages, Swedish medical doctors, wage bargaining, gender discrimination.

(3)

1. Introduction

The medical profession has a high status in most societies. This can at times be reflected in physicians’ wages. Medical school is becoming increasingly difficult to get into when the grades necessary are raised. Therefore, most doctors can gain prestige not only through their positions as “healers and helpers” but also for their skill and dedication.

Yet on a more individual level, what is it that affects a doctor’s wage? In Sweden, gender equality is of utmost importance in society as a whole, including the discussions around wages. Strong women have led the fight for equality, and laws have been passed to diminish the gender discrimination that still infiltrates many professions. The question is then if these discussions and laws have made a difference in the medical field, or if men still earn more than women. To speculate further one could also examine if older or more experienced doctors earn more and the consequential reasons for this.

The purpose of this paper is to study the different effects of several variables on the wages of doctors in Sweden. These variables which will be studied include: age, gender, marital status, medical degrees, the percentage of male doctors at the individual’s workplace, if the individual has a mentor at their current workplace, the number of hours worked overtime, satisfaction at the individual’s workplace, the number of years worked at the current workplace (tenure), and the number of patients met each week.

This thesis starts by explaining and discussing the variables and model chosen for these regressions. In order to estimate the determinants of doctor’s wages I have chosen to use an ordinary least squared (OLS) regression model with the log of monthly wage (before tax) as my dependent variable. Data from a survey of 1021 Swedish doctors from 2001, written by Kathleen Cannings and sponsored by the Swedish Medical Association, will be used. The statistical program of choice is SAS. Thereafter earlier literature will be studied and discussed. Thirdly, the first regression model including all the observations, both male and female doctors, will be run and the results analyzed. After that the regression will be run separately for men and women. Here the focus will be on the gender differences in the effects of the independent variables on doctors’ wages. Finally a concluding summary will be made.

(5)

2. The Data

During the Fall of 2001 a survey sponsored by the Swedish Medical Association was sent out to a random sample of 1 out of every 12 medical doctors in Sweden, a total of 1956 individuals. The goal of the study was to investigate working conditions and situations of doctors. As with most surveys, not all were answered, and not all that were answered were usable. The final data used in this paper includes 1021 medical doctors, 430 women and 591 men.¹ Table 1 below includes summary statistics for the included variables.

Table 1. Summary Statistics for Swedish Doctors.

Variable Mean Std Dev Minimum Maximum

AGE (years) MALE WAGE (kronor)

MARRIED DOCEXAMINA

MENTOR UNION SATISFIED MALEPROC(%) OVERTIME(hrs/week)

TENURE(years) NRPATIENTS

46.156 0.579 40677.84 0.848 0.178 0.401 0.513 0.639 57.865 5.420 8.191 43.283

9.785 0.494 12751.66 0.359 0.382 0.490 0.500 0.481 23.673 6.473 8.201 25.913

25 0 15000 0 0 0 0 0 0 0 0 0

65 1 245000 1 1 1 1 1 100 60 40 200

1 “Enkät till svenska läkare” (2001), by Kathleen Cannings, sponsored by the Swedish Medical Association, grant from Rådet för Arbetslivsforskning.

(6)

2.1 Variables

• AGE – The individual’s age.

• AGESQ – Age squared.

• WAGE – Monthly salary before tax.

• LNWAGE – The logarithm of monthly salary before tax.

• MALE – Dummy variable where 1 if male, 0 otherwise.

• MARRIED – Dummy variable which takes the value 1 if the individual is married or living in a common law marriage, 0 otherwise.

• DOCEXAMINA – Dummy variable which takes the value 1 if the individual has a medical degree and a doctorate, 0 if only a medical degree.

• MENTOR – Dummy variable which takes the value 1 if the individual has had a mentor in the organization.

• UNION – Dummy variable which takes the value 1 if the latest wage negotiation was made through a union.

• SATISFIED – Dummy variable which takes the value 1 if the individual feels satisfied with their current job/workplace.

• MALEPROC – The percentage of male doctors at the workplace.

• OVERTIME – Average number of hours worked overtime per week.

• TENURE – Number of years worked at the current job site.

• TENURESQ – Number of years worked at the current job site squared.

• NRPATIENTS – Average number of patients seen per week.

As can be seen in the table above, the wage distribution is skewed to the left, with a mean only 25000 kronor over the minimum wage. The maximum, on the other hand, is over 200000 kronor over the mean. Unfortunately the data does not distinguish between those who work part- or fulltime, which can be seen in the very low minimum wage. 15000 kronor cannot possibly be the fulltime wage of a Swedish doctor. Also, it is more common for women, at least after childbirth, to work part time, which may affect the data at hand. Only 17.8 percent of the doctors in the dataset had both a medical degree and a doctorate. This can be a possible explanation for the skewdness in the wages, since logically those with doctorates will have a significantly higher salary.

The age variable, on the other hand, is very evenly distributed, with a mean almost precisely in the middle of the 40-year span. Evenly distributed data can lead to a better overview for my results. When it comes to the gender variable, the dataset is also fairly evenly distributed, with 58 percent of the sample being male. In 2001 about 40 percent of the Swedish physicians were women according to the Swedish Medical Association. Given that one of the goals of my thesis is to compare differences between the genders, using similar numbers of observations for men and women will make the results more credible.

(7)

A very large percentage of the sample, 85 percent, was either married or living together with their partners. Since the medical field sometimes has the bad reputation of demanding that individuals put work before family, this may be a surprisingly large percentage. At the same time doctors have a high occupation status, which is an attractive trait for a partner in many social circles. The data shows that 40 percent of doctors had mentors in the organization where they worked and 51.3 percent had bargained their last wage audit with the help of their union.

In the original data, the variable SATISFIED was measured on a six-point scale, where 1 was the value chosen if the doctor was unsatisfied, and 6 if he/she was extremely satisfied with the current workplace. In order to easily include this variable in my regression, I created a dummy variable where the points 1 to 3 became 0, unsatisfied, and points 4 to 6 became 1, satisfied. Although this is a simplification of the data, I still find it interesting to include it in the model. About 64 percent were satisfied with their current workplace. This means that a significantly large portion of Swedish doctors in 2001 were not satisfied with their jobs. The mean number of overtime hours was 5.42, with a few extreme cases of doctors working 40 to 60 hours overtime a week. Most have worked around 8 years at their current workplace.

According to this dataset, doctors meet an average of 43 patients a week.

According to the human capital model, in order for it to be worthwhile for an individual to move to a new job and/or location, the expected future income must outweigh the cost of leaving their current residence. Both monetary and non-monetary costs have to be included in this analysis. This means that employers in potentially less attractive areas of the country have to raise wages and benefits in order to bring workers into those areas.² Thus one can imagine that areas like Luleå and Malmö have to produce attractive opportunities compared to Stockholm, the capital city. Statistics from Saco in 2003 show just this; that several of the rural parts of northern Sweden have some of the highest wages for doctors. The average wage for a doctor in Stockholm was 5000 kronor lower than in Norrbotten.³

Other variables that would have been interesting to study but where the number of missing values was very high, include number of children, medical specialization, how much one is

2 Björklund et. al. (2000), p. 200 f.

3 ”Stor lönespridning bland svenska läkare” (2003) p. 1001.

(8)

paid for overtime, and how much one feels that the union represents one’s interests. I believe that medical specialty and position, for example chief surgeon positions, have a significant impact on one’s salary. It would also have been interesting to include other, more psychological variables, such as if the individual feels that the schedule they have is feasible, if they feel physically and emotionally tired after work, or how well they find that their collogues share the same values, etc. All of this and more was covered by the survey, but the percentage that answered was quite low, and would have significantly lessened the number of observations. Therefore I have chosen to exclude these variables to attain a better overview of doctors’ situation in Sweden. Furthermore, the cumulative years worked was dropped due to its strong correlation with age. To maximize the information available age and tenure were instead included in the regressions.

Separate summary statistics for female and male doctors follow in the tables below.

Table 2. Summary Statistics for Female Doctors.

AGE(years) WAGE(kronor)

MENTOR UNION SATISFIED MALEPROC(%) OVERTIME(hrs) TENURE(years)

NRPATIENTS

44.119 37011.39 0.771 0.106 0.429 0.502 0.618 49.027 4.709 6.292 41.231

9.717 9608.22 0.421 0.308 0.496 0.501 0.486 22.475 5.552 6.675 23.787

25 15000 0 0 0 0 0 0 0 0 0

64 77500 1 1 1 1 1 95 60 35 200

(9)

Table3. Summary Statistics for Male Doctors.

AGE(years) WAGE(kronor)

MENTOR UNION SATISFIED MALEPROC(%) OVERTIME(hrs/week)

TENURE(years) NRPATIENTS

47.637 43345.47 0.905 0.230 0.380 0.520 0.654 64.423 5.914 9.558 44.733

9.573 14037.87 0.294 0.421 0.486 0.500 0.476 22.388 7.004 8.901 27.246

25 17000 0 0 0 0 0 0 0 0 0

65 245000 1 1 1 1 1 100 60 40 200

The average age of the male doctors is slightly above that of the female doctors, 48 compared to 44. The mean wage is substantially higher for men, around 6000 kronor above the wages for women. 90 percent of male doctors are married compared to 77 percent of female doctors and twice as many men as women have achieved their doctorate.

While 43 percent of the women have mentors at their workplace, only 38 percent of the men have the same. A slightly higher percentage of male doctors have had their last wage bargain done through the union, 52 percent compared to 50 percent for the female doctors. Continuing the overview, 62 percent of women were satisfied with their jobs, the percentage slightly higher for men, where it is at 65 percent. Men work on average about one hour more overtime per week than women, and have on average had a three year longer tenure at their current workplace. The average number of patients seen per week is again slightly higher for men, who see 45 patients versus 41 for women.

(10)

3. The Model

The variables used are as mentioned: age, gender, marital status, earned degree, the percentage of male doctors at the individual’s workplace, if the individual has a mentor at their current workplace, the number of hours worked overtime, satisfaction at the individual’s workplace, the number of years worked at the current workplace, and the number of patients met with each week. The variables AGE and TENURE have also been squared in order to study if the marginal effect is linear or not. In this order they are presented in the following equation:

i i i

i

i i

i

i i

NRPATIENTS TENURESQ

TENURE

SATISFIED UNION

MENTOR OVERTIME

MALEPROC

DOCEXAMINA MARRIED

MALE AGESQ

AGE LNWAGE

ε β

β β

β

β β

+ +

+

+ +

+

=

13 12

11

10 9

8 7

7

6 5

4 3

2 1

I have chosen to take the logarithm of the dependent variable, the monthly salary before tax, in order to be able to analyze the results of the regression in percent form. In this way, the estimates of the independent variables can be directly assessed onto the dependent. Since studying the relative impact is the purpose of this paper, I found this to be the better alternative.

The OLS model minimizes the sum of all the residuals squared in order to optimize the results. An unbiased and consistent model can be derived fairly simply with this regression model.⁴All the analyzed results are under the assumption of ceteris paribus, all else equal.

When studying the effect of one of the independent variables on wages, for example overtime, one has to keep in mind that this is when all other explanatory variables are kept constant.

I will now turn to a more thorough discussion of the variables and relevant theory from earlier literature, beginning with human capital variables and continuing with labor condition variables.

4 Wooldridge (2006), s. 33 f.

(11)

4. Human Capital

4.1 Age and Tenure

People often debate whether or not age and experience affects one’s salary. In Sweden all doctors’ salaries, with the exception of the Allmän Tjänst (AT) doctors, who are similar to interns, are individually bargained and set.⁵ Hence this could suggest that the individuals’

age and experience should not make such a big difference within the country since the wages are so personally based. Yet reality shows that this is not the case. The age structure in the medical field today is such that men are on average older and that there are more men that are chiefs and surgeon generals. Therefore their wages tend to be higher than those of their female coworkers.⁶ According to my summary statistics the average age for male doctors in 2001 was 4 years over the average for female doctors.

Even though history shows great differences in experience levels of men and women, a converging trend can be seen. By 1998 the difference in average experience had declined to a point were women were only 3 years behind their male counterparts. Tenure’s gender gap has also increased significantly, to 0.5 years in 2002.⁷ The data used in this thesis shows a 3-year higher tenure average for men.

A possible explanation for the increase in income, due to labor market experience, is that on- the-job training increases an employee’s productivity and that productivity increase leads to higher earnings.⁸

4.2 Gender

Many studies have been made investigating gender discrimination in wages, and this continues to be a hot topic in discussions on wage determinants. One reason for this may be that even though continuous work is being done, a gap between salaries for men and women in the same job can still be seen. Blau et al. conclude that gender discrimination may be the explanatory variable that accounts for up to 40 percent of the difference in men and women’s

5 ”Läkare: Utbildning och arbetsmarknad” (2007).

6 E-mail 2007-05-09 from Karin Rhenman, Sveriges läkarförbund, Stockholm.

7 Blau-Ferber-Winkler (2006), p. 185 f.

8 Blau-Ferber-Winkler (2006), p. 189.

(12)

income⁹. Another issue may be that more women tend to work overtime than men, especially after childbirth.

One core reason for wage discrimination against women is the gender assumptions that filter through our society. Employers strive to act in a way that is most economically rational when hiring employees and setting wages. When they find statistics that show women as more unreliable employees, with more sick days, higher rates of quitting, and so on, this can easily be used as a reason not to promote, hire, or raise wages of their female employees.¹⁰ According to Francine D. Blau, this problem is not quite as large in Sweden as in many other countries. Sweden has a more beneficial family leave policy than the United States, for example, and therefore women can stay better attached to the labor market and may face less discrimination in this area.¹¹ In other countries where the family leave policy is much more restricted, it can become very difficult to combine the role of motherhood with the need to stay connected to the labor force.

In 2000, statistics from the Swedish Medical Association showed a 12.6 percent higher wage for male doctors. According to the study women not only received a lower entrance wage, but were also more seldom promoted than men. Yet in 2001 the laws for equality were made stricter, with a goal of erasing gender discrimination in wages within three years.¹² Unfortunately this goal was not reached, though the situation has improved. A study of surgeon generals’ wages in Sweden from 2005 showed that though the difference between the sexes still exists, the percentage was now between 1.3-5.3 percent, depending on which area of the country.¹³ If these numbers fit the rest of the medical field as well, the results are surprisingly positive. Yet my hypothesis is that the data from 2001 will still show wage discrimination between the genders.

4.3 Marriage

According to the results of Hemström (1998) marriage has no effect on either men or women’s wages. The sample used in their study was of highly educated individuals, who often prioritize their work very much, which is given as a possible reason why one may see

9 Blau-Ferber-Winkler (2006), p. 249.

10 Figart-Mutari-Power (2002), p. 212.

11 Blau (1998), p. 22.

12 Olin (2001), p. 4950.

13 Gunnarsdotter (2006), p. 823.

(13)

different results here than in society as a whole.¹⁴ Doctors fit into this category, as they have to spend many years studying before achieving their degree. Even though Hemström’s result is encouraging, I expect that marital status can influence wages.

5. Labor Conditions

5.1 Mentor

Do doctors need mentors? According to Caroline Doherty, the medical industry is often infiltrated by a “sink or swim” mentality, where doctors should help and not need help themselves. If there is talk of mentorship, it is almost always an older, experienced doctor taking a younger one under his or her wing, so to speak. Doherty emphasizes the need for peer support even within the medical profession, and leads workshops to prepare doctors in this area. Learning to listen and help the mentee come to their own conclusions and solutions to problems can be hard for many doctors, who are used to having all the answers.¹⁵ Hopefully mentorship has a positive effect on wages.

According to certain research, there are indicators that the chances are higher that men receive career development benefits from sponsorship and networks that already exist and can be found in male-dominated worksites. Though this is positive for the male doctors, it is a negative effect for gender equality. Another problem with mentorship is that some doctors find that the supporting and the assessing roles can clash.¹⁶

5.2 Unions

In this paper the union specification that is used asks the interviewees if their latest wage bargain was done with the help of a union or individually. As an overview a short background on unions and their roles in wage bargaining is in order.

Union membership in Sweden has grown rapidly over the last half of the 20^th century, in 1950 51.3 percent of the workforce was unionized, and this percentage grew to 83 percent by 1989.

As early as 1965 public sector unions were given the same collective bargaining rights as

14 Hemström, Maria (1998), p. 153.

15 Doherty (2004), p. 6 ff.

16 Conner et al. (2000), p. 747 ff.

(14)

unions of the private sector.¹⁷ The goal of most wage bargaining is solidaristic wage policy, that the same work should give the same pay without regard to the employers’ profitability.

Though not all problems with wage inequality are solved with this policy, it is a definite step in the right direction.¹⁸

As in any bargaining situation, there must be another party and issue for the union to bargain about. With wages for example, unions must find excess funds in the companies that they feel would be more correctly spent on higher salaries. Furthermore, in order for the unions to have power to demand change, they have to have control over the labor supply available. The more elastic the demand is for wage increases, the lower the actual wages will be set. In Sweden, even workers who are not part of the union can benefit from the union’s bargaining, and therefore we do not expect a significant wage difference between the organized workers and those who are not. Still, empirical studies showed a 4 percent difference in 1981, where the organized workers could expect the higher wage.¹⁹

In the United States, there has been an overall negative trend in unionization, though the decline has been much greater for men. This has lead to women covering a larger percentage of union workers than before, and thus the wage gap has diminished as well. In the past some research suggested that women have a slightly higher wage premium connected to unions than men, though most studies show the gains as relatively even between the genders.

Recently, certain researchers found a trend suggesting a slight advantage for men.²⁰

More than 90 percent of Swedish doctors belong to the medical union, The Swedish Medical Association (SMA). Collective agreements are made through the union; this includes issues such as working hours, wages, pensions and sick leave. SMA members are registered both at a local branch and a national branch, the latter dividing members into groups depending on professional position.²¹

Most Swedish doctors today are employed in regions or county councils and are therefore covered by the Wage Agreement of 2005, made by the Swedish county councils,

17 Martin (1992), p. 40 ff.

18 Martin (1992), p. 54.

19 Björklund et. al. (2000), p. 228 ff.

20 Blau-Ferber-Winkler (2006), p. 281 f.

21 ”The Swedish Medical Association” (2007).

(15)

municipalities, and the SMA. According to this contract, the annual wage audit can be done in one of two ways. The first is through negotiation between the SMA’s local coalition and the county council or region, which is the more traditional approach. The alternative is a wage setting conversation between the chief and employee. A possible third option is a combination between the two. Most use the traditional wage bargaining form, but there are almost always also a designated meeting between the doctors and the person who decides their salaries.²²

Stig-Eric Åström, president in Östergötland’s medical organization believes that many who try to bargain for higher salaries without the help of the union are not sufficiently prepared.

One goal of the SMA is to encourage individuals who are preparing to wage bargain to get in contact with their union.²³ Results mentioned in a study by Grankvist and Regnér suggest that individual bargaining has a positive effect on wages, especially for men.²⁴ Though many resent discussions promote individual bargaining, I tend to believe that unions on average still have more wage power than one person does.

5.3 Satisfaction

Richard Freeman is one of the foremost spokespersons for the economic importance of job satisfaction. Satisfaction can be classified as a subjective variable that is hard to measure due to its psychological aspect. Yet Freeman finds the study of job satisfaction to bring interesting insight into economic analysis that earlier was left unexplored. For example, one study found that satisfaction seemed to be a stronger reason for quitting one’s job than wages.

Furthermore it reported that on average 90 percent are satisfied with their current job, with differs greatly from my data, where only 64 percent report to be satisfied.²⁵

Though worker satisfaction is important to the overall well-being of organizations, little has been found to argue that it has an effect on wages. Yet more understanding of economic movement and labor-market behavior can be found by studying worker satisfaction.²⁶ Discussions have focused instead on the reverse connection, wages’ effect on satisfaction.

Economists have found that common belief within companies’ management is that employee income can effect the companies’ productivity. High salaries should lead to a strong feeling of

22 E-mail 2007-05-09 from Karin Rhenman, Sveriges läkarförbund, Stockholm.

23 Gunnarsdotter (2006), p. 823.

24 Granqvist – Regnér (2004), p.27.

25 Freeman (1978), p. 135 ff.

26 Hamermesh (1999), p. 3.

(16)

satisfaction for the employees, who then may attract an even more productive workforce, and this will finally lead to a higher level of profitability for the company.²⁷

5.4 Number of Patients

According to the American Medical Association the number of patients seen per week differs between male and female doctors. While female doctors who are working full-time on average see 87 patients, male doctors see on average 102 patients.²⁸ The situation is quite different here in Sweden according to my statistics, where the average amount of patients seen per week is 41 and 45 for female and male doctors respectively. It is my hope that the number of patient visits does not affect wages. If it does, doctors could simply try to meet more people in order to increase their income, and issues such as being thorough may have less importance. Yet seeing many patients can be seen as a sign of effectiveness, and therefore bring on a higher salary.

6. Results and Analysis

Tests for hetroskedasticity show that there will be no significant effect on the results, and there are no problems with multicolliniarity in my models. When the term statistically significant is used, the parameter estimate is correct with an error of max 1 percent, unless another significance level in indicated. Thus, one can with 99 percent certainty, claim that the parameter estimate of the explanatory variable is different from zero. This is the case throughout the remainder of this thesis. In all of the three regressions covered the intercept parameter is statistically significant, which proves that the mean of my dependent variable differs from 0.

6.1 Overall Regression

In the model including the entire dataset, 781 observations were used in the regression, the rest having missing values in one or more of the included explanatory variables. The adjusted R-squared value was 0.5897, which means that 58.97 percent of the variation in these doctors’

wages is explained in the model. The results of the regression can be found in Table 4 below.

27 Adams (1975), p. 361.

28 Nowlan (2006).

(17)

Table 4. Linear Regression with Log of Wage as the Dependent Variable.

Effects on Doctors’ Wages Variable Parameter

Estimate

Std Error Intercept 8.2908*** 0.155 AGE(years) 0.0785*** 0.007 AGESQ -0.0007*** 0.0000

8

MALE 0.0566*** 0.014 MARRIED 0.0016 0.018 DOCEXAMINA 0.0759*** 0.018 MENTOR -0.0455*** 0.014 UNION 0.0732*** 0.013 SATISFIED 0.0207 0.013 MALEPROC(%) 0.0006** 0.0003 OVERTIME(hrs/week) 0.0011 0.001

TENURE(years) 0.0157*** 0.003 TENURESQ -0.0004*** 0.0001 NRPATIENTS 0.0004 0.0003 Note: *** and** indicate statistical significance at a 1 and 5 percent level respectively.

6.1.1 Human Capital Variables

According to the results, a doctor’s age in 2001 had a considerable impact on what salary they had. For every year older an individual was, their wage was also 7.85 percent higher. The AGESQ variable was significant as well, though in a negative direction. This points toward that the increased income that comes with increased age declines; it is not linear but instead a

(18)

diminishing marginal return. This is a considerable increase, and follows the information provided by the Swedish Medical Association.

Tenure significantly explained variation in wages, but not a very large part of it. If a doctor had worked an additional year at his or her current job, their salary will be 1.57 percent above those with one year less of tenure. One possible reason for this could be as earlier mentioned, that on the job training leads to higher productivity that is rewarded with higher salary. As with age, the squared variable of tenure is significant and negative, thus pointing toward a decline in the positive effects of tenure.

Gender also had a significant impact on doctors’ wages. The male physicians could count on a 5.66 percent higher salary than their female colleagues. This is substantially lower than the 12.6 percent found in the statistics from SMA in 2000. Of course their data is much more extensive than the data available for this study, yet it is noteworthy that only one year later the wage discrimination seems to be much less. This could be due to the enforcement of stricter equality laws. Further analysis of gender differences follows in the upcoming models.

According to the findings of the model, a doctor’s marital status had no significant impact whatsoever on his or her wage. This is quite interesting, since one would think that marriage would change individuals’ priorities, which may lead to choosing to work less, and this in turn could result in a negative effect on their wage. Yet at the same time it concurs with the earlier studies by Hemström, that marital status does not affect wages.

Those doctors that had acquired their doctorate as well as their medical degree could in 2001 count on a 7.59 percent higher wage than those with only a medical degree. I find it quite surprising that spending years acquiring a doctorate gives a similar wage increase as something as simple as being a year older. Further analysis of this result can be difficult since specializations play a big part and this is not something I study in this thesis.

6.1.2 Labor Condition Variables

One noteworthy result is that if the individual had a mentor in the organization, this in turn had a negative impact on wages. Those doctors who had mentors on average received 4.55 percent lower wages than those who did not. This is not only surprising but also worrisome, when what many see as a positive contact decreases one’s salary. Since mentors have shown

(19)

to have a good impact on their mentees satisfaction and well-being, the wages should reflect and encourage this and not the other way around. Something I would find interesting to study is if being a mentor, instead of having one, has a positive or negative effect on a doctor’s wages.

In contrast to the findings of Granqvist and Regnér, using the union to help in wage bargaining increased salaries in this data by 7.32 percent, which is quite a substantial difference. This result solidifies the opinions and goals of all unions; that using their services result in higher benefits. At the same time it may be quite a surprising result for all in favor of individual bargaining without the help of unions. Yet one must keep in mind that since the Swedish Medical Association sponsored this survey, the data might be slightly skewed, those who have very high wages and dislike their union may have opted not to answer the survey.

The number of patients seen per week had no significant impact on a doctor’s wage. This is a positive result, showing that increasing the number of patients met with is not a sure way of increasing one’s salary. Yet overtime, which can be seen as a signal of an individual’s loyalty and dedication is not rewarded significantly. Satisfaction with one’s workplace also has no significant effect on wage variation, which may not be that surprising. A higher wage may increase an individual’s satisfaction, but the reverse connection seems unlikely. Thus there may be a slight causality problem with this variable.

6.2 Gender Regression

When the model was run on women, 313 observations were used out of the total sample of women, which was 430. Even here the reason that not all the observations could be used was that there were missing values in one or more of the variables included in the model. The explanatory percentage is slightly lower than in the all included case, the model explains 52.98 percent of variation in female doctors’ wages. The regression on male doctors read 591 observations and used 486 of these. This model explains 58.49 percent of the variation according to the adjusted R-squared.

Even though many of the results in the following regressions are statistically significant, when comparing two models one must also secure the differences by making sure that the confidence intervals do not overlap. The confidence intervals used are at the 95 percent level.

(20)

Unfortunately none of the differences found between the models are significantly secured, but since the results are significant when looked at separately, they still point towards a differing trend between the genders. The results of the gender regressions follow in the two tables below.

Table 6. Linear Regression for Female Doctors with Log of Wage as the Dependent Variable.

Effects on Female Physicians’ Wages Variable Parameter

Estimate

Std Error

95% Confidence Limits Intercept 8.7930*** 0.251 8.29978 9.28624 AGE(years) 0.0577*** 0.012 0.03444 0.08095 AGESQ -0.0005*** 0.0001 -0.00075 -0.00022 MARRIED -0.0201 0.026 -0.07206 0.03178 DOCEXAMINA 0.0982** 0.042 0.01534 0.18110 MENTOR -0.0245 0.023 -0.07036 0.02141 UNION 0.0471** 0.022 0.00395 0.09028 SATISFIED 0.0091 0.022 -0.03513 0.05336 MALEPROC(%) 0.00004 0.0005 -0.00096 0.00104 OVERTIME(hours) 0.0011 0.002 -0.00287 0.00502 TENURE(years) 0.0201*** 0.005 0.01017 0.03005 TENURESQ -0.0005*** 0.0002 -0.00092 -0.00014 NRPATIENTS 0.0007 0.0005 -0.00020 0.00162 Note: *** and ** indicate statistical significance at a 1 and 5 percent level respectively.

(21)

Table 7. Linear Regression for Male Doctors with Log of Wage as the Dependent Variable.

Effects on Male Physicians’ Wages Variable Parameter

Estimate

Std Error

95% Confidence Limits Intercept 7.9227*** 0.207 7.51664 8.32873 AGE(years) 0.0958*** 0.009 0.07768 0.11389 AGESQ -0.0009*** 0.0001 -0.00107 -0.00067 MARRIED 0.0235 0.026 -0.02757 0.07463 DOCEXAMINA 0.0677*** 0.020 0.02808 0.10725 MENTOR -0.0522*** 0.018 -0.08661 -0.01791

UNION 0.0949*** 0.016 0.06295 0.12682 SATISFIED 0.0283* 0.017 -0.00462 0.06114 MALEPROC(%) 0.0007** 0.0004 0.00004 0.00143 OVERTIME(hrs/week) 0.0008 0.001 -0.00162 0.00322 TENURE(years) 0.0116*** 0.004 0.00453 0.01868 TENURESQ -0.0003** 0.0001 -0.00056 -0.00005 NRPATIENTS 0.0001 0.0003 -0.00047 0.00074

Note: ***,** and * indicate statistical significance at a 1,5, and 10 percent level respectively.

6.2.1 Human Capital Variables

Age is statistically significant in both gender regressions; an extra year gives men on average 9.58 percent and women 5.77 percent higher wages. Though this difference is not statistically secured at the 95 percent confidence limits, it is still an important result. The squared age variable is also significant for both men and women though the parameter is slightly more negative for men, suggesting that the decline in marginal return of tenure is more rapid for men than for women. It is interesting to note that the results point toward that age has a higher positive effect for male doctors. A possible reason for this may be that men gain more

“status” from age than women do, and this in turn is reflected in their wages.

Being married or living with your partner is statistically insignificant even when looking at the genders separately. When applied to gender analysis this is a positive result, not only that

(22)

marital status does not have a significant impact on an individual’s wage, but also that there are seemingly no differences between the sexes.

Having a doctorate was only significant at the 5 percent level for women in this model, and increased their wages by 9.82 percent in comparison with those with only a medical degree.

For me, on the other hand, the significance of the parameter estimate was still at the 1 percent level, yet their wage effect was only 6.77 percent. This is quite an interesting result, suggesting that it pays off more for women to get a doctorate then men. A possible explanation for this may be that women have to work harder than men to achieve higher status and credibility in the medical field and thus increase their salary.

6.2.2 Labor Condition Variables

If a female doctor has a mentor in her organization it had no significant effect on her salary.

Male doctors on the other hand, could expect a 5.22 percent lower salary if they were mentored. This contradicts the research done by Conner et al., where they found that men were more likely to receive benefits from being connected to networks such as mentorship.

Here instead it has a strong negative effect on the wages for these individuals. Unfortunately this can point towards a sort of gender discrimination as well, men should be strong and able to handle themselves and are seen as weak if they have mentorship. Women on the contrary may have mentors, for as the “weaker” sex they need more help.

When wage bargaining with the help of one’s union, male doctors seem to have a stronger advantage, which concurs with recent research mentioned by Blau et al. The significance level of the union in the regression on males was higher, significant at the 1 percent level while the model was only significant at the 5 percent level for the female doctors. The men could also count on 9.49 percent higher salaries than those who bargained individually, while the union active female doctors only could count on a 4.71 percent greater wage. One initial thought on this result would be that men were more forceful in wage bargaining, and would be more likely to receive a high increase in salary when choosing to run solo without the union’s help.

Yet this result suggests the opposite, that women are awarded higher for individual wage bargaining.

Job satisfaction impacted male doctors’ wages at the 10 percent significance level, while it was still insignificant for female physicians. If men are satisfied with their current job their

(23)

wage may on average be 2.83 percent higher than those who are unsatisfied. It is interesting to find this result that disappears when all the observations are included in the model. This result points towards men’s satisfaction being more tied to their income compared to women in the medical field. A further discussion: is it due to a “maternal” instinct to care and heal that women care less about their wage? Or is the difference more due to the possibility that women’s satisfaction is more tied to variables such as work environment and other more psychological variables than to their salary?

As in the case above, the percentage of males at the individual’s workplace is significant for men but not for women. A 1 percent more male dominated workplace increases the male doctors’ wages with 0.07 percent. This is quite a small wage difference, but it is significant none the less, though only at the 5 percent level. Moreover it is very interesting that women’s wages are not affected by how many of the doctors at their workplace are men.

Tenure had a statistically significant impact on wages for both the sexes. Women have a 2.01 percent higher wage for every year of tenure, while the percentage for men is 1.16. Though this difference is not statistically secured, it still indicates a slight variation between how important tenure is for the wages of men versus women physicians. As with age, the coefficient of the squared tenure is negative, indicating a diminishing marginal return. This variable is significant at the 1 percent level for women and at the 5 percent level for men. The number of patients met with per week and the hours of overtime are insignificant even when the genders are looked at separately.

(24)

7. Concluding Discussion

Though much of the variation in doctors’ wages is left unexplored, this thesis sheds light on several aspects that seem to be of current interest to the wage study. Many of the results follow other research while some differ.

No reasons could be found that the number of hours worked overtime per week, the number of patients met with per week, or the doctors’ marital status, have any impact on the variance of a physicians wage. Furthermore, job satisfaction was an insignificant factor, except for when the genders were studied separately, where men’s satisfaction impacted their salary positively. Yet one has to remember the possible causality, since it is difficult to measure which variable affects the other.

Focusing on the gender aspects, a wage difference can still be seen between male and female medical doctors. Though several years have passed since 2001, it is unlikely that the 5.66 percent wage difference has completely disappeared. Furthermore, results show that mentorship seems to be insignificant for women and negatively significant for men, a subject I hope to do more research on in the future.

Variables such as age, tenure, and union bargaining were statistically significant throughout, and all have a positive impact on wages. This follows what I believe to be the common conception in our society. The relative wage advantage is around 7 percent for both age and union bargaining, while the effect of tenure is lower at almost 2 percent. All these three variables also seem to have more positive effects for male doctors, which suggest that gender discrimination is still a reality in the medical field.

I find that exploring what can affect a person’s income is not only interesting but also enlightening. We live in a changing society, and some aspects that were of great importance many years ago might be completely irrelevant today. At the same time, some things seem to never change in either strength or importance. Hopefully we will see gender discrimination completely disappear in the medical field in the near future.

(25)

References

Adams, Roy J. (1975), “Wage Determination: Reconciling Theory and Practice”, American Journal of Economics and Sociology, Vol. 34, Issue 4.

Björklund, Anders – Edin, Per-Anders – Holmlund, Bertil – Wadensjö, Eskil (2000), Arbetsmarknaden. Stockholm.

Blau, Francine D. – Ferber, Marianne A. – Winkler, Anne E. (2006), The Economics of Women, Men, and Work. Upper Saddle River.

Blau, Francine D. (1998), “The Gender Pay Gap” in Inga Persson-Christina Jonung (ed.), Women’s Work and Wages. London-New York.

Conner, M. P. – Bynoe, A. G. – Redfern, N. – Pokora, J. – Clark, J. (2000) “Developing senior doctors as mentors: a form of continuing professional development.

Report of an initiative to develop a network of senior doctors as mentors: 1994- 99”, Medical Education, Vol. 34. Blackwell Science Ltd.

Doherty, Caroline (2004), “Introducing mentoring to doctors: Challenging the sink or swim culture”, Development and Learning in Organizations, Vol. 18, No. 1.

”Enkät till svenska läkare” (2001), by Kathleen Cannings, sponsored by Sveriges Läkarförbund, grant from Rådet för Arbetslivsforskning.

Figart, Deborah M. – Mutari, Ellen – Power, Marilyn (2002), Living Wages, Equal Wages:

Gender and labor market policies in the United States. London-New York.

Freeman, Richard B. (1978), “Job Satisfaction as an Economic Variable”, The American Economic Review, Vol. 68, No. 2.

Granqvist, Lena – Håkan, Regnér (2004), Den nya lönebildningen-En forskningsöversikt och analys av lönebildningen för akademiker. SACO, Stockholm

Gunnarsdotter, Sara (2006), “Facket bra väg till högre lön”, Läkartidningen, Nr. 11, Vol. 103.

Gunnarsdotter, Sara (2006), ”Lönegap mellan könen består”, Läkartidningen, Nr. 11, Vol.

103.

Hamermesh, Daniel S. (1999) “The Changing Distribution of Job Satifaction”, NBER Working Paper Series, Working Paper 7332.

Hemström, Maria (1998), “Gender Differences in Pay Among Young Professionals in Sweden” in Inga Persson-Christina Jonung (ed.), Women’s Work and Wages.

London-New York.

“Läkare: Utbildning och arbetsmarknad” (2007), Sveriges Läkarförbund, www.slf.se

(26)

Martin, Andrew (1992), Wage Bargaining and Swedish Politics: The Political Implications of the End of Central Negotiations. Stockholm.

Nowlan, Mary H. (2006), “Women doctors, their ranks growing, transform medicine”, The Boston Globe, October 2. The New York Times Company.

Olin, Elisabet (2001), “Inom tre år ska lönerna vara jämlika”, Läkartidningen, Nr. 24, Vol. 98.

Rhenman, Karin (2007-05-09), [Sveriges läkarförbund, Stockholm], Ämne, [e-mail to Rebecka Cederholm {rece7070@student.uu.se}].

”Stor lönespridning bland svenska läkare” (2003), Läkartidningen, Nr. 12, Vol. 100.

”The Swedish Medical Association” (2007-04-28), Sveriges Läkarförbund. Available online:

[http://www.slf.se/templates/ArticleSLF.aspx?id=2033].

(27)

Appendix: Correlation Matrix

Pearson Correlation Coefficients

AGE WAGE MALE MARRIED DOCEXAMINA MENTOR AGE 1.00000 0.54880 0.17767 0.02478 0.24564 -0.30693 WAGE 1.00000 0.24538 0.06071 0.23213 -0.20742

MALE 1.00000 0.18414 0.15971 -0.04953

MARRIED 1.00000 0.07264 0.07326

DOCEXAMINA 1.00000 0.06105

MENTOR 1.00000

Pearson Correlation Coefficients

UNION SATISFIED MALEPROC OVRTIME TENURE NRPATIENTS AGE -0.08698 0.04491 -0.05247 0.15756 0.58251 0.02687 WAGE 0.05626 0.07924 0.08944 0.16845 0.38386 0.06518 MALE 0.01743 0.03650 0.32177 0.09158 0.19658 0.06660 MARRIED 0.01404 0.04294 0.06086 -0.02709 0.09499 0.04932 DOCEXAMINA -0.04768 0.05125 0.03289 0.14289 0.28948 -0.17546 MENTOR 0.01675 0.13182 0.03286 -0.06295 -0.14318 -0.07899 UNION 1.00000 0.05949 0.01590 0.00600 -0.19110 0.05144 SATISFIED 1.00000 0.01162 -0.06722 0.04288 0.02169 MALEPROC 1.00000 0.05425 0.03467 0.02045

OVERTIME 1.00000 0.11562 0.07818

TENURE 1.00000 -0.02762

NRPATIENTS 1.00000

What Determines the Variation in Doctors’ Wages?: A Study of Swedish Physicians