Part-time versus Full-time employees : a productivity comparison

(1)

Part-time versus Full-time employees:

A productivity comparison

Name Mark Verweij

Student Number 10288082

Program Economics & Business

Track Finance & Organisation

Name Supervisor Silvia Dominguez Martinez

Date January 22h_{, 2016}

Abstract

The goal of this thesis is to analyze the difference between individual productivity of part- and full-time employees. Using a unique cross-sectional data-set including information on the function, age, gender, yearly contractual full time equivalents and yearly productivity of 123 individual UWV employees in the year 2015. A division is made between short and long (hourly) part-time jobs, to conclude if different levels of part-time jobs have a different productivity effect. The results suggest that there is strong evidence that part-time employees produce more per full time equivalent than their full-time counterparts. Findings also imply that there is a difference in productivity effect of heterogeneous levels of part-time jobs. The smaller (hourly) the part-time job, the more productive the employee. Gender does not have a significant effect and age has a small positive effect on individual productivity.

(2)

1 Introduction

The Dutch labor force has seen a growing number of part-time employees. The number of part-timers doubled between 1996 and 2009 to about 3 million, whilst the number of full-timers stagnated at around 4,5 million (Centraal Bureau voor de Statistiek, 2010).

And this is not only the case in the Netherlands; this structural change of the labor market is seen all over Europe and North America. Part time employment has risen sharply during the latest economic downturn. In an attempt to retain incumbent workers, many firms have preferred reducing employee working hours to layoffs (European Commission Directorate-General for Employment, Social Affairs and Equal Opportunities, 2010).

So part-time employment is growing. On the one hand the human capital theory predicts part-time workers to be less productive than full-timers (Ilmakunnas and Maliranta, 2005). On the other hand some papers suggest part-time employment to be beneficial for firm level productivity (Garnero et al., 2014). There are advantages making part-timers more productive; part-timers are less fatigued (Brewster et al., 1994) and they take on and complete relatively large workloads (Cataldi, Kampelmann and Rycx, 2012). There are disadvantages making them less productive; their skill improvement is argued to be slower (Felstead et al., 2000) and Part-time employment leads to coordination and cummunication difficulties (Lewis 2003).

So there is unclear picture if, and why part-time employees have different productivity levels. The previous literature has mainly tested the effect of shares of part-time employees within a company on total firm productivity. The difference in individual productivity between full- and part-timers has, to my knowledge, not yet been considered. This productivity difference could have a large impact on policy and future research. It could impact cooperation’s recruitment strategies and the way employee performance/productivity is researched.

This brings me to the research question; what is the difference in individual hourly productivity between part-time and full-time employees? This research question will be answered using two hypotheses: The first hypothesis is based on the information described in the papers of Garnero et al. (2014), Wotruba (1990) and Künn-Nelen, et al. (2013). These papers all point to the direction that part-time employees are more productive than full-time employees. Hypothesis one is: Individual part-time employees have higher hourly productivity than individual full-time employees. The second hypothesis is based on Garnero et al. (2014) who found that long part-time functions (25-35 hours) have a significant positive productivity effect, while small part-time functions did not.

Hypothesis two is: Individual part-time employees working larger part-time jobs, meaning more hours, are relatively more productive than individual part-time employees working smaller part-time jobs.

(4)

age, sex, yearly contractual full time equivalents and yearly productivity of 123 individual UWV employees in the year 2015. The data allows us to compare individual employee productivity with their contractual labor hours. This is a different and innovative approach since previous research, to my knowledge, mostly use data with information on different firm productivity and part-time employment shares. These studies compare different firms. This study compares the productivity of employees within one firm, possibly making the data more homogenous and fit for comparison.

The dependent variable in the equations is individual employee production. The independent variable is an employee’s contractual full time equivalent. Different levels of part-time employment are tested using dummy variables. And two control variables are used for age and gender.

The results in this paper are that individual part-time employees are more productive per full time equivalent than full-time employees. And that there are differences in productivity for different (hourly) levels op part-time employment.

This paper is organized as follows: The next section reviews related literature starting with the definition of part-time employment, the advantages and disadvantages of part-time employment and previous research regarding part-time employment and (firm) productivity. Section 3 gives the hypotheses. Section 4 discusses the data and descriptive statistics, section 5 the Model Specification, section 6 the results and, finally, section 7 describes the conclusion and discusses the limitations of the paper and suggestions for future research.

(5)

2 Related Literature

2.1 Full-time and part-time employment

This section gives a description and a definition of part-time employment.

Full- and part-time are measurements used to distinguish different types of employment contracts. A part-time contract is a form of employment that carries fewer labor hours than a full-time contract. Policies in favor of part-time work are often aimed to enable employees with multiple roles, such as parenting, to combine professional and private roles (Garnero et al., 2014).

National definitions of part-time work are based either on hourly thresholds, the job-assessment of an employee or on a combination of both (Van Bastelaer et al., 1997). There are large variations in the hourly threshold definitions of part-time employment. A national hours cut-off is used to differentiate between part-time and full-time work in Australia (35 hours), Austria (35), Canada (30), Finland (30), Hungary (36), Iceland (35), Japan (35), New Zealand (30), Norway (37), Sweden (35), Turkey (36) and the United States (35). For certain countries (Australia, Norway and Sweden) additional criteria are applied in specific situations (Van Bastelaer et al., 1997). The CBS, the Dutch institution of statistics, tells us that within the Netherlands a part-time contract is a contract that contains less than 35 hours (“Deeltijdwerk”, 2016).

Varying levels of part-time employment can be measured using full-time equivalents (FTE) or whole time equivalent (WTE). The Government Accountability Office (GAO) defines FTE as the number of total hours worked divided by the maximum number of compensable hours in a full-time schedule as defined by law.

The next two sections consider the advantages and disadvantages of part-time employment.

2.2 disadvantages of part-time employment

This section considers the disadvantages of part-time employment

The existence of fixed costs within firms, such as; administrative, recruitment and office maintenance costs lead to a disadvantage of part-time employment. These fixed costs do not increase proportionally with working hours (Montgomery, 1988), making part-time workers relatively more expensive in the hours they work.

Jobs have a start-up time. When an employee starts his shift, it takes a certain amount of time to reach optimal productivity. This means that relative productivity rises during a worker's shift. That implies that a part-time employee, when working short days, spends a relatively large part of his labor day reaching optimal productivity, making him less productive (Barzel, 1973).

Part-time employment may also lead to coordination difficulties (Lewis, 2001). Organizations that require specific skills and combine multiple part-time employees to complete one specific task, normally handled by one full-timer, experience

(6)

communication gaps. These gaps result from the inability to properly transfer a partially completed task from one part-timer to another (Bonamy & May, 1997). Edwards and Robinson (2004) analyzed questionnaires held among nurses operating within the United Kingdom. Respondents stated that communication problems, increased administrative costs, overhead expenditures associated with training and suboptimal service continuity are disadvantages that were more frequently present when employing part-time employees.

Skill improvement is also argued to be slower among part-time employees. Felstead et al. (2000), Walby and Olsen (2003) and Branine (2003), all show evidence that part-time employment is associated with a relative low skill level, or a stagnation in skill development. Part-time employees might be less committed to career goals. Parenting and other career interruptions could give less motivation to personally invest in their job specific skills. At the same time, organizations might be less willing to invest in part-time workers because of those career interruptions. The part-time nurses interviewed by Edwards and Robinson (2004) stated that they had less access to career specific training and felt like they were less likely to be promoted than their full-time colleagues.

2.3 advantages of part-time employment

This section considers the advantages of part-time employment

Brewster et al. (1994) claim that fatigue has a negative influence on the volume and quality of productivity. The longer the work day, the more fatigued an employee becomes. Part-time employees, when working less hours a day, will be less fatigued resulting in higher levels and better quality production. Part-time labor days are heterogeneous; part-timers might work full-time days as well. If part- and full-timers have identical labor days, productivity difference caused by long work day fatigue will not occur.

Part-time jobs allow employees a more natural sleep rhythm, making workers less fatigued and stressed (Pierce & Newstrom, 1983; Baltes et al., 1999). There is more evidence suggesting that part-time employment reduces stress. A study by Branine (2003) focused on hospital staff in Denmark, the UK and France found that part-time work is usually associated with low absenteeism and lower levels of stress.

Full-time employees moving to part-time employment often experience un-adjusted expectations from their management, working hours are reduced but, the workload stays practically the same (Edwards & Robinson, 2000; Lewis 2001, 2003). This translates in a part-time worker achieving (nearly) full-time productivity in a smaller number of hours. Cataldi, Kampelmann and Rycx (2012) found that part-time jobs with more than 25 weekly labor hours often have the same level of productivity as their full-time counterparts, making part-timers more efficient and productive.

A study by Künn-Nelen et al. shows that firms with a larger percentage of part-timers are more productive. Part-timers give employers the ability to react more flexibly to market changes and peak demand. Owen (1978) argues that organizations employ part-time labor to avoid overlapping shifts of full-time workers.

(7)

Edwards and Robinson (2000) found that the UK metropolitan police use part-time employment because it helps to retain experienced staff that would otherwise exit the labor force (especially women). That effect would mean an outflow of knowledge and experience that is important to help educate less experienced officers. Edwards and Robinson (2000) also found that part-time employment increased job satisfaction and commitment.

Some of the advantages of part-time employment are dependent on the heterogeneity of part- time employment. The repercussions of part-time arrangements on productivity are likely to differ substantially when an employee is absent during much of the workweek (short part-time jobs) or working almost full-time (long part-full-time jobs) (Garnero et al., 2014). It will also depend on the way an employee divides his part-time contract over the workweek (e.g. 5, 6 hour day or 4, 8 hour days).

2.4 previous research regarding part-time employment and

(firm) productivity

This section considers three previous researches regarding part-time employment and (firm) productivity.

Künn-Nelen et al. (2013) wanted to identify the relation between different shares of part-time employees and firm production. They used a unique matched employer–employee dataset regarding Dutch pharmacies (𝑛 = 236). Based on that data firms’ part-time and full-time labor shares were constructed. The number of prescription lines delivered to customers serves as a “hard” firm-level productivity measure. Künn-Nelen, et al. (2013) used a modified Cobb-Douglas function. Using FTEs instead of the number of workers to define the employment shares. They compared the productivity of the total full-time FTE share with that of the total part-time FTE share. They include age as one of their control variables control. Finding that a larger part-time employment share leads to greater firm productivity. But they argue that most of these effects were due to allocation differences between part- and full-time employees. Not because of a difference in individual productivity. Künn-Nelen et al. (2013) were able to make this distinction because they had data on the timing of labor demand, giving them an insight into when what sort of employees were deployed. Künn-Nelen, et al. (2013) for instance found that part-timers enabled full-timers to take lunch breaks so that a firm was able to remain open during those times.

Wotruba (1990) compared part- vs. full-time salespeople from four U.S. direct selling companies. The paper showed that part- timers were more productive, measured by earnings per labor hour, this measure was chosen because it directly reflected the number of sales made. This data was based on two questionnaires (𝑛 = 493). The difference in the mean performance between the full- and part-timers was substantial but only significant at 10.4% level, because of large variations around the means. One of the reasons for the productivity difference given by

(8)

Wotruba (1990) was that part-timers had the ability to achieve higher earnings per hour via commission selling. Making it possible for part-timers to achieve much higher hourly earnings.

Garnero et al. (2014) used matched employer-employee panel data on Belgian private-sector firms to estimate the relationship between wage/productivity differentials and the firm’s labor composition in terms of part-time work and gender (𝑛 = 5,171). Garnero et al. (2014) took the heterogeneity of part- time employment into account because the repercussions of part-time arrangements on wages and productivity are likely to differ when an employee is absent during much of the labor week (short part-time jobs) or working almost full-time (long part-time jobs). A modified Cobb-Douglas model is used for the estimations. The dependent variable is a firm’s hourly added value. The independent variables are different part-time shares within firms. In a second model a distinction between long and short part-time jobs is made. Dummy variables are used for observable firm characteristics such as age, gender composition and education levels. Findings suggest that male part-timers with a long (25-35 hour) part-time have higher levels of productivity than other employees. This finding is supported by the outcome of a paper by Cataldi, Kampelmann and Rycx (2012), finding that male and female part-timers working long (25-35) hour part-time jobs have relatively higher productivity.

Studies relying on the assumption that wages reflect productivity found part-timers to be less productive than full-part-timers (Hirsch 2005; Baffoe-Bonnie 2004; Aaronson & French 2004; Ermisch & Wright 1993). But there are studies that find that wage gaps can be attributed to the differences in the types of jobs and occupational segregation. After controlling for those characteristics the wage gaps are often no longer significant (Manning & Petrongolo (2008); Hardoy & Schone (2006); Rodgers (2004). Implying that the assumption that wages reflect productivity is not correct, making it an improper way of comparing full-time and part-time productivity differences.

(9)

3 Hypotheses

Based upon the previous literature and the formulated research question, the following two hypotheses are formulated.

3.1 Hypothesis 1

Previous research on individual productivity differences between full-time and part-time employees within a single cooperation is scarce. The findings in the papers of Garnero et al. (2014), Wotruba, (1990) and Künn-Nelen, et al. (2013) provide evidence that part-time employees are more productive than full-time employees in the hours they work.

𝐻_! Individual part-time employees have higher hourly productivity than individual full-time employees.

3.1 Hypothesis 2

Garnero et al., (2014) take the heterogeneity of part- time employment into account. They found that males working long part-time functions (25-35 hours) have a significant positive effect on firm level productivity, while small part-time functions do not. Cataldi, Kampelmann & Rycx (2012) find that part-time jobs with more than 25 weekly labor hours often have similar levels of productivity as their full-time colleagues.

𝐻! Individual part-time employees working larger part-time jobs, meaning more

hours, are relatively more productive than individual part-time employees working smaller part-time jobs.

(10)

4 Data & descriptive statistics

We use a unique cross-sectional employee productivity dataset. This dataset is provided by the UWV. UWV (Employee Insurance Agency) is an autonomous administrative authority (ZBO) and is commissioned by the Ministry of Social Affairs and Employment (SZW) to implement employee insurances and provide labor market data and -services. Their core tasks include (“about UWV”, n.d.):

● employment – helping clients remain employed or find employment, in close cooperation with the municipalities;

● social medical affairs – evaluating illness and labor incapacity according to clear criteria;

● benefits – ensuring that benefits are provided quickly and correctly if work is not possible, or not immediately possible;

● data management – ensuring that the client needs to provide the government with data on employment and benefits only once.

This dataset gives us all the information needed to identify a relation between productivity differences of part-time and full-time. The dataset includes information on the function, age, sex, yearly contractual FTE’s and yearly productivity of 123 UWV employees in the year 2015. The information is on two distinct functions within the UWV; Labor specialist and Insurance Physician. Below shows a frequency table specifying the characteristics of the employees:

Table 1 Frequency table

Specification of the division of employees based on gender, part-‐/full-‐time employment and age

Source: UWV dataset

Frequency table

N Male Female Average age

Part-‐time 74 34 40 53 Part-‐time Cohort 1 33 9 24 55.2 Part-‐time Cohort 2 41 25 16 50.8 Full-‐time 49 35 14 52.78 Total 123 69 54 52.8 Notes: Cohort 1 = 0<FTE<0.84, Cohort 2 = 0.84<FTE<1, Full-‐time = 1<FTE<1.05.

It is clear from this table that the percentage of males working long part-time and full-time jobs is high compared to the percentage of females. It seems that woman prefer to work the shorter part-time jobs, this might be the case because woman, compared to man, more often take on the parenting role (Centraal Bureau voor de Statistiek,

(11)

2010).

Below is a table showing the descriptive statistics of the sample and variables used.

Table 2 Descriptive statistics

The descriptive statistics for the full sample and it’s variables Source: UWV dataset

Summary Statistics

Variable Obs. Mean Std. Dev. Min Max

Efficiency 123 0.54 0.28 0 1.36 Actual FTE 123 0.86 0.19 0.31 1.05 Cohort 1 33 0.27 0.44 0 0.839 Cohort 2 41 0.33 0.47 0.84 1 Cohort fulltime 49 0.40 0.49 1 1.05 Gender* 123 0.56 0.50 0 1 Age 123 52.8 9.80 30.4 67.5

Notes: Cohort1 = 0<FTE<084, cohort 2 = 0.84<FTE<1 Cohort fulltime= 1<FTE<1.05 *56% of employees are male .

4.1 Dependent variable

The dependent variable in the model is yearly individual employee efficiency. Efficiency is defined as the realized production divided by the target production. The UWV measures employee productivity using a points per product system. The number of points gathered over the year is the employee’s individual realized productivity. The yearly number of points gathered by an UWV employee is determined as follows:

The employees have different products that can be completed within their function. These products are worth a certain number of points dependent completion time of a product. The base-product for the points-system is the ’decision to end latency employee WIA’. This product is worth one point. All the other products take less time, so their value is a fraction of that the base product:

𝑝𝑜𝑖𝑛𝑡 𝑣𝑎𝑙𝑢𝑒 𝑝𝑟𝑜𝑑𝑢𝑐𝑡 = 𝑚𝑖𝑛𝑢𝑡𝑒𝑠 𝑡𝑜 𝑐𝑜𝑚𝑝𝑙𝑒𝑡𝑒_! / 𝑚𝑖𝑛𝑢𝑡𝑒𝑠 𝑡𝑜 𝑐𝑜𝑚𝑝𝑙𝑒𝑡𝑒 𝑎𝑖05 Where ai05 is the product-code for the ’decision to end latency employee WIA’ and 𝑖 is the other product code. Appendix 1.1 gives an overview of the products and their point-value.

The UWV assumes that a full-time Insurance Physician has 1223 productive hours a year. An Insurance Physician should be able to complete a production of 371 points in those 1223 hours (see table 3). Meaning that 1 point is achieved in around 3.33 hours. The 371 points is a full-time employee’s target. The realized

(12)

number of points divided by the target is the efficiency. This is used as the dependent variable because different employee efficiency levels can be compared. Efficiency is a relative measure, it does not matter how many hours an employee works, the measure depends on an employee’s ability to be productive enough to reach his/her targets. The targets are based on an employee’s number of labor hours. Making it a proper variable to use for comparing productivity levels of employees with different contractual FTEs. The direct productivity measure, yearly number of points obtained by an employee, would be unfit for comparison because it is not a relative measure. Employees working more hours have more time to reach higher productivity than employees working fewer hours, comparing points would not give an indication of the hourly productivity differences.

4.2 Independent variable

Our independent variable is the effective contractual full-time equivalent of UWV employees. Centraal Bureau voor de statistiek. (z.j.) concludes that, in the Netherlands, a contract containing less than 35 hours a week is a part-time contract. The UWV defines a part-time contract as one containing less than 38 hours a week. 38 hours a week is one full time equivalent (FTE). Someone working 30 hours therefore works 30 ℎ𝑜𝑢𝑟𝑠/38 ℎ𝑜𝑢𝑟𝑠 = 0,79 𝐹𝑇𝐸. The dataset also contains employees working 1,05 FTE. These employees actually work the same weekly number of hours as 1 FTE employees, which is 40 hours, the only difference is that these employees have 13 days less paid leave. Below in table 3 is an hourly specification of what a yearly full-time equivalent consists of, and what the direct (effective) number of labor hours are after subtracting certain posts that shorten the effective labor hours (e.g. Absenteeism, paid leave, personal hygiene, etc. Appendix 1.11 for more detail.):

Table 3 Hourly specification

A specification of the total working hours of a 1 and 0.5 Insurance Physician, minus deductions. In order to specify the direct number of labor hours used for production Source: UWV dataset

Hourly specification

1 FTE 0,5 FTE

Total labor hours (52*x40**) 2.080 1040

Deductions:

Leave, holidays 152 76

Sick, paid & special leave 442 221 Additional deduction 363 300

Total direct hours 1.223 559

(13)

This specification shows the division of hours in a one FTE and a 0.5 FTE contract. As stated earlier there are a lot of subtractions to the total amount of hours (52 𝑤𝑒𝑒𝑘𝑠 ∗ 40 ℎ𝑜𝑢𝑟𝑠 = 2080), these subtractions, are not directly used for production (e.g. training, education, HRM interviews and training new employees).

The UWV does not properly correct the direct number of labor hours when going from one FTE to another fraction of one FTE. They simply take the direct number of hours from 1 FTE (1223) and multiply that with the FTE fraction (e.g. 1123 ℎ𝑜𝑢𝑟𝑠 ∗ 0.5 𝐹𝑇𝐸 = 612 ℎ𝑜𝑢𝑟𝑠). This is not correct. As you can see in the hourly specification in appendix 1.11, the Italic posts are proportionally subtracted when 1 FTE (2080 hours) turns to 0,5 FTE (1040) hours (e.g. personal hygiene; 100 hours get subtracted from 1 FTE and 50 hours from 0,5 FTE). But the underlined posts are not proportionally subtracted. This is because every employee gets the same number of hours subtracted for those posts, no matter the FTE level. This is problematic because the number of direct labor hours used for production is overestimated for part-time employees. Using the uncorrected FTEs in the analysis would make part-time employees look less productive; they actually have less direct (production) hours than according to the uncorrected FTEs, so their productivity is achieved in fewer hours.

The FTEs need to be corrected in order to make a fair comparison between part- and full-time efficiency. So the dataset provided by the UWV is corrected using the following formula:

((𝐹𝑇𝐸_! ∗ 𝑑𝑖𝑟𝑒𝑐𝑡 ℎ𝑜𝑢𝑟𝑠 𝐹𝑇𝐸_!) − ((1 − 𝐹𝑇𝐸_!) ∗ 237))/𝑑𝑖𝑟𝑒𝑐𝑡 ℎ𝑜𝑢𝑟𝑠 𝐹𝑇𝐸_!

Where 𝐹𝑇𝐸_! is the FTE level, 𝑑𝑖𝑟𝑒𝑐𝑡 ℎ𝑜𝑢𝑟𝑠 𝐹𝑇𝐸_!is 1223, and 237 is the sum of the underlined, not-proportionally subtracted posts. This way the extra amount of the underlined posts, that weren't subtracted when simply multiplying the direct number with the FTE fraction, is subtracted. The calculation and specification shown above are applicable to the function of Insurance Physician. The calculation for Labor Specialist is shown in appendix 1.12.

In order to check the impact on productivity of different levels/cohorts of part-time employment, two different levels of part-part-time employment are constructed (Garnero et al. 2014). These levels are based on the, according to the UWV, three most common cohorts of part-timers. That division is as follows cohort 1: employees working 0.5 FTE or less, cohort 2: between 0.5 and 0.84 FTE and cohort 3 between 0.84 and smaller than 1 FTE. This means that 1 to 1.05 FTE is considered full-time. The first cohort, employees working 0,5 FTE or less, only has 8 observations. To small a number of observations to have sufficient statistical power according to

Stock and Watson (2007). Garnero et al. (2014) and Cataldi et al. (2012) use two cohorts, short (25 hours and less) and long (between 25 and 35 hours) part-time jobs. But that division would have resulted in the short cohort only containing 10 observations, which still would not have sufficient statistical power (Stock and Watson, 2007). That is why UWV cohorts 1 and 2 are combined. The new cohort,

(14)

cohort 1 has 33 observations and is specified as: 0<FTE<0.84. Three binary variables remain; cohort 1, cohort2 and cohort fulltime. Each cohort’s binary value equals 1 when the FTE value falls within the cohort (0 when it does not). In the analysis the full-time cohort will be used as the reference category.

4.3 Control variables

Individual productivity is influenced by a lot of factors. A literature review Ailabouni (2010) gave us an overview of the 42 most important factors (see appendix 1.2). Six of those factors are included in the dataset; age, gender, overtime, lunch breaks and working hours. The last three factors are included in the FTE specification of each employee, as can be seen in the hourly specification shown in appendix 1.11 and 1.12. The other two, age and gender, are included separately.

Age: Künn-Nelen et al. (2013), Garnero et al. (2014) and Wotruba (1990) use age in their analysis. Aubert et al. (2003) finds that individual productivity increases until an age of 40, after that age, the individual productivity remains stable. Workers aged 40 and higher are roughly 5% more productive than workers aged 35-39, while workers below 30 are 15 to 20% less productive (Aubert et al., 2003). So age might have an effect on individual productivity and that effect is tested in this paper’s analysis. Age is a linear variable within the empirical model. Gender: Both Künn-Nelen et al. (2013) and Garnero et al. (2014) use gender in their analysis. Garnero et al. (2014) finds that only man working long (25+ hours) part-time jobs have a positive productivity effect. The productivity differences between part- and full-timers could rely on the reasons why individuals choose certain work schedules; if men and women have different motives, then that might result in a gender bias in the productivity differences (Anxo et al., 2002). Gender might have an effect on individual productivity and that effect is tested in this paper’s analysis. Gender is a binary variable within the empirical model; the dummy equals 1 when gender is male (0 when female).

Most of the other factors in Ailabouni (2010) his overview are out of the scope of this paper because they are not included in the dataset. Some of those factors might not have such a large influence because they are similar for the employees in our dataset, such as all the Environmental Factors. Some of the factors will be similar within functions, such as levels of education, past training (all employees within function receive same amount of training hours) and wages (reward schemes).

(15)

5 Model Specification

Previous literature on productivity effects of part-time employees normally base their empirical model on a Cobb-Douglas function (Hellerstein, et al. (1999), Hellerstein et al. (2004), Aubert and Crépon (2003), Garnero et al. (2014), Künn-Nelen et al. (2013)). These studies compare multiple firms, sometimes over multiple industries and multiple years. They research the effect of labor force composition, the effect of a percentage of part-time workers in the total labor force, on total firm productivity. They do not research the productivity difference of individual part- and full-time employees. The before mentioned studies use panel data in their analysis.

This paper’s model uses a cross-sectional dataset. The dependent variable in the equation is employee 𝑖′𝑠 yearly efficiency; this is the number of points obtained by that employee divided by his/her target number of points. The independent variable is employee 𝑖′𝑠 contractual FTE; this is a certain fraction of the number of labor hours in a full-time contract. The models discussed in the next section will be estimated using ordinary least squares regressions.

The model specification starts with a baseline-model that is extended with control variables to test 𝐻_!. Then the different cohorts for the two levels of part-time employment will be added to test 𝐻_!.

The baseline-model, model 1, is defined as the following: (1) 𝐸𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑐𝑦 = 𝑏!+ 𝑏!𝐹𝑇𝐸! + 𝜀!

Model 1 test the effect of the level of an employee’s FTEs on that employee’s efficiency. Model 2 introduces the control variables age and gender. Age is a linear variable, and Male a binary variable for gender (value 1 for male, 0 for female):

Model 1 & 2 allow us to test the effect of the level of contractual FTEs on the efficiency of employees. The two dummy variables are added to control for the possible age and gender effects. Model 3 replaces the continuous FTEs with the full-time binary dummy variable (value 1 for full-full-time, 0 for part-full-time) to test the average efficiency differences between part- and full-timers.

Model 4 replaces continuous FTEs with the two cohorts for the two different levels of part-time employment. In order to test if the average efficiency per contractual FTE differs for different levels of part-time employment compared to full-time employment.

Model 5 replaces one part-time cohort with the full-time cohort in order to test if the two part-time cohorts have different average efficiency levels compared to each other. These last three models allow us to test the second hypothesis.

Robust standard errors are used in all the regressions to control for heteroskedasticity.

(16)

6 Results

Table 3 Regression analysis

Analysis of the full sample Source: UWV dataset Efficiency 1 2 3 4 5 Variables Actual FTE -‐0.358** (0.148) -‐0.358** (0.164) Cohort 1 0.170*** (0.071) Cohort 2 0.033 (0.056) -‐0.138** (0.069) Cohort FT -‐0.087*** (0.0514) -‐0.161*** (0.063) Gender 0.033 (0.055) 0.009 (0.053) 0.043 (0.056) Age 0.004* (0.002) 0.005 ** (0.002) 0.004 (0.002) N 123 123 123 123 123 R2 _0.059 _0.084 _0.053 _0.084 _0.058

Notes: */**/*** denotes significance at a 10%/5%/1% level, resp., cohort1 = 0<FTE<0.84, cohort 2 = 0.84<FTE<1, cohort FTE= 1<FTE<1.05, value in brackets are the robust Std. Err.

6.1 Results analysis

Column 1 shows the results for the baseline model (equation 1), without any control variables, estimating the effect of the contractual FTEs on the efficiency. The coefficient that estimates the effect of contractual FTEs is statistically significant at 1% and has a value of -0.358. This implies a strong negative relationship; en employee working 1 extra full time equivalent is 35.8% less efficient. This result is similar to the findings of Cataldi, Kampelmann and Rycx (2012) that found that certain part-timers had production levels similar to full-time employees but within fewer hours.

(17)

Column 2 shows the result for the baseline model with the added control variables age and gender (model 2). The coefficient that estimates the effect of actual contractual FTE’s remains the same (-0.358) and is statistically significant at 1%. The coefficient for estimating the effect of gender is statistically insignificant. But age has a positive effect of 0.004 and significant at 10%. Implying that if an employee ages one year, he will become 0.4% more efficient. Aubert et al. (2003) finds that individual productivity increases until an age of 40, after that age, the individual productivity remains stable. The analysis in this paper does not specify such exact age effects, but it does seem that aging does add to individual efficiency and so to individual productivity. Gender is insignificant.

Column 3 shows the results of the model that replaces continuous FTEs with a full-time employment dummy variable (full-time equals value 1 and part-time 0). The coefficient estimating the effect of full-time employment is -0.087 and is statistically significant at 1 percent. On average a full-time employee is 8.7% less efficient than a part-time employee. Age is significant at 5% and has a positive effect of 0.5%. Gender remains insignificant.

Column 4 shows results of the model that replaces continuous FTEs with 2 dummy variables for short and long part-time jobs (model 4). The coefficients that estimate the difference between the average efficiency of the levels of part-time compared to full-time employment are positive. Only cohort 1 is statistically significant at 1%. Age and gender are statistically insignificant. The coefficient estimating the effect of cohort 1 is 0.170 implying that on average, an employee working a short part-time job (FTE< 084) is 17% percent more efficient/productive than a full-timer. Garnero et al. (2014) find the opposite; in their analysis the share of male long part-time workers is the only significant coefficient that has a positive productivity effect.

Column 5 shows the difference between cohort 1’s average efficiency compared to that of cohort 2 and cohort full-time. The coefficient estimating the difference between cohort 1 and 2 is significant at 5% and has a value of -0.138 implying that employees working longer part-time jobs (0.84<FTE<1) are on average 13.8% less efficient/productive than employees working shorter part-time jobs (FTE<0.84). The coefficient estimating the difference between cohort 1 and cohort full-time is significant at 1% and has a value of -0.161 implying that on average, an employee working a short part-time job (FTE< 084) is 16.1% percent more efficient/productive than a full-timer.

Cohort 1 and 2 are also tested to see if their mean efficiency is significantly different using an F-test. This test, checking if cohort 1&2 are equal, is rejected at the 5% level by the F-test since the P-value is smaller than .005 (Prob > F = 0.046). Implying that in 95% of the observations the mean efficiency of cohort 1 and 2 is significantly different.

(18)

6.2 Additional analysis

Gender does not have a significant relation to efficiency/productivity in any of the models that are analyzed. This is interesting since Garnero et al. (2014) only found significant evidence that a labor force share of male workers in long part-time jobs (between 25-35 hours) increase firm productivity. Implying that gender does have a productivity effect within different size part-time jobs.

Appendix 2 shows a table where efficiency is regressed against interaction terms between cohorts 3 and gender:

𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑐𝑦 = 𝑏_!+ 𝑏_!𝑀𝑎𝑙𝑒 + 𝑏_!𝑐𝑜ℎ𝑜𝑟𝑡3 + 𝑏_!𝑀𝑎𝑙𝑒×𝐶𝑜ℎ𝑜𝑟𝑡3 + 𝜀_!

This is done in order to test whether gender has a significant effect on efficiency within part-time employment. The interaction term estimating the gender effect within part-time employment is statistically insignificant.

The first two equations estimated quite a large coefficient for the effect of actual FTEs on efficiency. Looking at the data revealed that some observations stood out, e.g. employees working full-time jobs that had obtained a very small number of, or zero points. Observations like these result in outliers, observation with extremely low levels of efficiency. The exact reasons why observations like these exist are only know to UWV management. Frequently these observations were the result of prolonged illness and employees getting assigned to special projects. Management knows which observations have special circumstances but that data is unavailable. This paper defines outliers as employees that efficiency smaller than one standard deviation from the mean. The mean is 54% and de SD is 28% making al efficiency observations smaller than 26% outliers. This results in 24 outliers in total. The outlier definition is based on the definition given in a paper by Abouleish et al. (2000). Appendix 2.1 shows the table with the results of the regressions done in the section 6.1 with the trimmed data excluding the outliers.

The continuous coefficients estimating the effect of actual FTEs on individual employee efficiency (model 1&2) are statistically significant at 5%. The estimates have a similar effect as before the data trimming, except that the effects are smaller. They go from a negative 35.8% effect to a negative 31.3% effect on efficiency per added FTE (without dummy variables). And from a negative 35.8% effect to a negative 30% effect on efficiency per added FTE when dummy’s for gender and age are added. The age effect is similar, as prior to the data trimming, gender remains statistically insignificant. The same holds for model 3, the dummy variable for full-time employment has the same sort of effect, but smaller than before the data trimming. The effect goes from an average a full-time employee being 8.7% less efficient, than a part-time employee, to being 7.5% less efficient. Age remains significant and gender remains insignificant. The trimmed data influences the estimations made by model 4 in the same way, cohort 2 remains statistically insignificant and the effect of cohort 1 (significant at 1%) is smaller, going from 17% to 15.2%. Age does become significant, now having a positive 0.4% (per extra year)

(19)

effect on efficiency. Model 5 was again influenced in the same way; cohort 2 and cohort fulltime remain statistically significant. The differences between average efficiency decreased from -13.6% (cohort 2) and -16.1% (cohort full-time) to -12.6% and 15.1%.

The F-test used to test if the mean efficiency of cohort 1 and 2 are significantly different again implied that in 95% of the observations the mean efficiency of cohort 1 and 2 are significantly different (Prob > F = 0.0458).

The outliers, the extreme observation of unproductivity/inefficiency had an effect on the estimators, but they did not change the negative relation between actual FTE and efficiency and the finding that on average cohort 1 part-time employees are more efficient than full-time employees.

The data trimming did decrease the number of observations from 123 to 99 this smaller sample size could lead to a decreased precision in estimating the coefficients. However, the R-squareds of the 4 models increased after the data trimming. This is an indication that the data is a better fit to the empirical model. The impact of the results on the two hypotheses is discussed in the three paragraphs.

Efficiency is measured as an employee’s realized productivity divided by that employee’s target productivity. The target productivity is based on the employee’s labor hours (FTE). So a more efficient employee has a higher hourly productivity.

𝐻! States that individual part-time employees have higher hourly productivity

than individual full-time employees. Looking at regression 1, 2 & 3, in the first three columns of table 3, statistically significant evidence is presented that indicates that 𝐻! can be accepted. Employees working higher levels of full time equivalents appear

to be less efficient/ productive. And part-time employees are, on average, more efficient/productive than their full-time counterparts.

𝐻! States that individual part-time employees working larger part-time jobs are

relatively more productive per labor hour than individual part-time employees working smaller part-time jobs. The results of regression 4 in the 4th column of table 3 imply something else. According to the results the employees working the smaller part-time jobs (FTE<0.84) are more efficient/productive than full-time employees. Employees working long part-time jobs do not have a statistically significant different level of production compared to full-time employees. The results of regression 5 in the 5th column support that conclusion. This would indicate that employees working smaller part-time jobs are the most efficient/productive; implying that 𝐻! cannot be

(20)

7 Conclusion and discussion

This paper analyses the difference in individual productivity between part- and full-time employees. Two hypotheses are tested: 𝐻_!, Individual part-time employees have higher hourly productivity than individual full-time employees and 𝐻_!, individual part-time employees working larger part-time jobs are relatively more productive per labor hour than individual part-time employees working smaller part-time jobs.

Strong evidence supporting the first hypotheses is found, a negative relation between the individual employee efficiency and their actual full time equivalent. Künn-Nelen et al. (2013) found that a larger part-time employment share leads to greater firm productivity. Garnero et al. (2014) found that positive productivity effects are driven by male part-timers working more than 25 hours. These studies both research the effect of part-time employment shares on firm productivity. Not the individual difference between part- and full-time employees. Wotruba (1990) found that individual part-time salespeople were more productive than full-time salespeople. The magnitude of the relations presented in these three papers is hard to compare to the findings in my paper, because of the different empirical models used. Künn-Nelen et al. (2013) find that a 10% increase of the part-time share is associated with 4.8% increase of firm productivity compared to the same increase in the full-time share. Garnero et al. (2014) finds that a 10% increase of the part-time share is associated with an 8% increase in firm’s productivity, compared to the same increase in the full-time share. This effect was only present when the increased part-time share consisted of males working between 25-35 hours.

This paper finds evidence that supports 𝐻_!, on average part-timers are 8.7% more productive than full-timers. And working 1 extra full time equivalent makes an employee, on average, 35.8% less productive.

The evidence in this paper does not support the second hypothesis. The results of the regressions comparing the productivity of different levels of part-time employment show that the employees working the smallest part-time are, on average, most productive. Garnero et al. (2014) find that the positive productivity effect of long part-time jobs only appears when the part-time workers are male. Cataldi et al. (2014) only find a positive productivity effect when the part-timers work held long part-time jobs.

The control variable gender is statistically insignificant, the interaction term estimating the gender effect within part-time employment is statistically insignificant. Age has a positive effect on individual productivity.

Based on the results acquired in this paper, the answer to the research question, ‘What is the difference in individual hourly productivity between part-time and full-time employees?’, is that part-time employees have a higher hourly productivity level compared to full-time employees. The smaller the part-time job, the higher the hourly productivity level.

(21)

part-time employees? The literature review might give away the answer. The existence of fixed costs within firms, such as; administrative, recruitment and office maintenance costs lead to a disadvantage of part-time employment. These fixed costs do not increase proportionally with working hours (Montgomery, 1988). Making part-time workers are relatively more expensive in the hours they work. Then there is the problem of managing the extra employees. Part-time employment increases the number of employees needed. A larger number of employees increases coordination and communication problems and inefficiencies. Making more supervisors and managers necessary to keep processes streamlined, increasing the costs (Owen, 1978).

There are various limitations to the research described in this thesis. The full time equivalents in the data that is used in the estimations are employees contractual FTEs. Some of employees might have deferred from their contractual FTEs, because of health problems, personal issues, special assignments or other special circumstances. This would in turn mean that an employee having a full-time contract might not have had the number of production points associated with that contract. Stimpfelet al. (2012) and McPhillips et al. (2007) found that the longer working hours are related to health issues such as burnouts. Full-timers experiencing health issues more often, could lead to full-timers being more likely to defer from their contracts. Making their production/efficiency an unfit reflection of their contractual FTE.

Another limitation was the fact that there was no data on the way a part-time employee distributed their labor hours among the weekly labor days. The effects of different FTEs might be different when an employee works three eight hour days, or five four and a half hour days.

A lot of the factors influencing individual productivity presented by Ailabouni (2010) are not controlled for in this paper. I cannot be sure what the effect of extra variables would be, although Cataldi, Kampelmann & Rycx (2012) found that most of the worker characteristics used in their model were insignificant.

The estimates would probably also be more precise if the sample size was larger than a 123 observations. Using a larger sample of individual employee production data and using a larger number of (different) organizations, Might make the result more generalizable.

Then there is the possibility of endogenous variables. Omitted variables can be the cause for endogeneity, creating a more comprehensive model, including more worker characteristics for example might decrease the possibility of endogeneity. The R-squares of the regressions are quite small, below 20%. So the data does not fit the model to well. So large parts of variability between the variables have not been accounted for. Creating a more comprehensive model with more variables could create a better fit. But more variables always increase the R-squares, so the adjusted R-squared should be used if a more comprehensive model is used.

It is worth noting that the difficulty level of the assignments distributed among employees at the UWV is random. Meaning that part-time employees do not get easier cases than full-timers. So the productivity differences found in this paper are

(22)

not influenced by a nonrandom distribution of cases among employees.

A challenge for future research in this area is to analyze larger datasets. Including more individual employees, more employee characteristics and different organizations and industries. This specific data is not widely available, but it would make the results more usable for policy making. It could also minimize the number omitted variables and therefor decrease the probability of endogeneity. Future research could also check for endogeneity of a variable using an IVV regression.

(23)

References

Aaronson, D., & French, E. (2004). The Effect of Part‐Time Work on Wages: Evidence from the Social Security Rules. Journal of Labor Economics, 22(2), 329-252.

Abouleish, A. E., Zornow, M. H., Levy, R. S., Abate, J., & Prough, D. S. (2000). Measurement of individual clinical productivity in an academic anesthesiology department. ANESTHESIOLOGY-PHILADELPHIA THEN HAGERSTOWN-,93(6), 1509-1516.

About UWV. (n.d.). Retrieved from

http://www.uwv.nl/overuwv/english/about-us-executive-board-organization/detail/about-us/core-task-and-ambitions

Ailabouni, N. (2010). Factors affecting employee productivity in the UAE construction

industry (Doctoral dissertation, University of Brighton).

Anxo, D., Flood, L., & Kocoglu, Y. (2002). Offre de travail et répartition des activités domestiques et parentales au sein du couple: une comparaison entre la France et la Suède. Economie et statistique, 352(1), 127-150.

Aubert, P., & Crépon, B. (2003). Age, wage and productivity: firm-level evidence.

French version: Aubert, P. and B. Crépon, 95-119.

Baffoe-Bonnie, J. (2004). Interindustry part-time and full-time wage differentials: regional and national analysis. Applied Economics, 36(2), 107-118.

Baltes, B. B., Briggs, T. E., Huff, J. W., Wright, J. A., & Neuman, G. A. (1999). Flexible and compressed workweek schedules: A meta-analysis of their effects on work-related criteria. Journal of Applied Psychology, 84(4), 496. Barzel, Y. (1973). The determination of daily hours and wages. The Quarterly

Journal of Economics, 220-238.

Van Bastelaer, A., Lemaître, G., & Marianna, P. (1997). The definition of part-time work for the purpose of international comparisons.

Bonamy, J., & May, N. (1997). Service and employment relationships. Service

Industries Journal, 17(4), 544-563.

Branine, M. (2003). Part-time work and jobsharing in health care: is the NHS a family-friendly employer?. Journal of health organization and management,17(1), 53-68.

(24)

Tregaskis, C., Brewster L., Mayne A. H. O. (1998). Flexible working in Europe: the evidence and the implications. European Journal of Work and Organizational

Psychology, 7(1), 61-78.

Cataldi, A., Kampelmann, S., & Rycx, F. (2012). Does it pay to be productive? The case of age groups. International journal of manpower, 33(3), 264-283.

Centraal Bureau voor de statistiek. (z.j.). Deeltijdwerk [Persbericht]. Geraadpleegd

van

http://www.cbs.nl/nl-NL/menu/methoden/toelichtingen/alfabet/d/deeltijdwerk.html

Centraal Bureau voor de Statistiek. (2012). Sociaal economische trends, 1e kwartaal 2012. Meeste werknemers tevreden met aantal werkuren. Retrieved from

http://www.cbs.nl/NR/rdonlyres/65E42E80-882C-4B11-AEA0-BFA49FB0BA6F/0/2012k1v4p20art.pdf

Edwards, C., & Robinson, O. (1999). Managing part‐timers in the police service: a study of inflexibility. Human Resource Management Journal, 9(4), 5-18.

Ermisch, J. F., & Wright, R. E. (1993). Wage offers and full-time and part-time employment by British women. Journal of Human Resources, 111-133.

European Commission Directorate-General for Employment, Social Affairs and Equal Opportunities. (2010). Employment in Europe 2010 (ISSN: 1016-5444). Retrieved from http://ec.europa.eu/social/main.jsp?catId=119&langId=en

Felstead, A., Ashton, D., & Green, F. (2000). Are Britain's workplace skills becoming more unequal?. Cambridge Journal of Economics, 24(6), 709-727.

Garnero, A., Kampelmann, S., & Rycx, F. (2014). Part-Time Work, Wages, and Productivity Evidence from Belgian Matched Panel Data. Industrial & labor

relations review, 67(3), 926-954.

Hardoy, I., & Schøne, P. (2006). The Part‐Time Wage Gap in Norway: How Large is It Really?. British Journal of Industrial Relations, 44(2), 263-282.

Hellerstein, J. K., & Neumark, D. (2007). Production function and wage equation estimation with heterogeneous labor: Evidence from a new matched employer-employee dataset. In Hard-to-measure goods and services: Essays

in honor of Zvi Griliches (pp. 31-71). University of Chicago Press.

Hellerstein, J. K., Neumark, D., & Troske, K. R. (1996). Wages, productivity, and

worker characteristics: Evidence from plant-level production functions and wage equations (No. w5626). National Bureau of Economic Research.

(25)

Hirsch, B. T. (2005). Why do part-time workers earn less? The role of worker and job skills. Industrial & Labor Relations Review, 58(4), 525-551.

Lewis, S. (2001). Restructuring workplace cultures: the ultimate work-family challenge?. Women in management Review, 16(1), 21-29.

Künn-Nelen, A., De Grip, A., & Fouarge, D. (2013). Is part-time employment beneficial for firm productivity?. Industrial & Labor Relations Review, 66(5), 1172-1191.

Manning, A., & Petrongolo, B. (2008). The Part‐Time Pay Penalty for Women in Britain*. The Economic Journal, 118(526), F28-F51.

Montgomery, M. (1988). Hours of Part‐Time and Full‐Time Workers at the Same Firm. Industrial Relations: A Journal of Economy and Society, 27(3), 394-406. Masayuki, M. (2010). Working Hours of Part-timers and the Measurement of

Firm-level Productivity (No. 10015).

McPhillips, H. A., Stanton, B., Zuckerman, B., & Stapleton, F. B. (2007). Role of a pediatric department chair: factors leading to satisfaction and burnout. The Journal of

pediatrics, 151(4), 425-430.

Owen, J. D. (1979). Working hours: an economic analysis. Working hours: an

economic analysis.

Pierce, J. L., & Newstrom, J. W. (1983). The design of flexible work schedules and employee responses: Relationships and process. Journal of Occupational

Behaviour.

Rodgers, J. R. (2004). Hourly wages of full-time and part-time employees in Australia.

Stock, J. H., & Watson, M. W. (2007). Introduction to Econometrics.: Pearson Education Inc. New York.

Stimpfel, A. W., Sloane, D. M., & Aiken, L. H. (2012). The longer the shifts for hospital nurses, the higher the levels of burnout and patient dissatisfaction.Health

Affairs, 31(11), 2501-2509.

Walby, S., & Olsen, W. (2003, July). The UK gender wage gap and gendered work histories. In conference of the British Household Panel Survey, Institute for

(26)

Wotruba, T. R. (1990). Full-time vs. part-time salespeople: A comparison on job satisfaction, performance, and turnover in direct selling. International Journal

of Research in Marketing, 7(2), 97-108.

(27)

Appendix

Appendix 1

(28)

Appendix 1.11

A specification of the total working hours of a 1 and 0.5 Insurance Physician, minus deductions. In order to specify the direct number of labor hours used for production Source: UWV dataset

1 FTE 0,5 FTE

Total labor hours (52*x40**) 2.080 (hours) 1040 (hours)

Deductions:

HBV leave (2 hours a week) 104 52 Sick leave (absenteeism 5,9%) 114 57 Paid leave 216 108 Holidays (6 days) 48 24 Special leave (1,5 days) 12 6

In office hours** 1586 hours 793 hours

Personal hygiene 100 50 Consultation coaching 30 30 Employment consultation 30 30 HRM-‐interviews 6 6 Cross-‐checking+casuistry 40 40 Protocols/medi prudence 28 28 Reregestration 28 28 Message/circulars 35 35

Training new employee 26 13

Process/system training 40 40

Total direct hours 1.223 hours 559 hours

Notes: =* weeks, ** hours, *** These are the actual office hours that are left after absenteeism, Italic posts are subtracted proportionally, underlinded posts are subtracted imporportionally

(29)

Appendix 1.12

Calculation of FTE’s including the number of direct/effective labor hours for Labor Specialists.

((𝐹𝑇𝐸_!∗ 𝑑𝑖𝑟𝑒𝑐𝑡 ℎ𝑜𝑢𝑟𝑠 𝐹𝑇𝐸_!) − ((1 − 𝐹𝑇𝐸_!) ∗ 201))/𝑑𝑖𝑟𝑒𝑐𝑡 ℎ𝑜𝑢𝑟𝑠 𝐹𝑇𝐸_!))

Where 𝐹𝑇𝐸_! is the FTE level, 𝑑𝑖𝑟𝑒𝑐𝑡 ℎ𝑜𝑢𝑟𝑠 𝐹𝑇𝐸_!is 1265, and 201 is the sum of the udnerlined posts. This way the extra amount of the underlined posts gets subtracted from the direct hours. See specification below

A specification of the total working hours of a 1 and 0.5 Labor Specialist minus deductions. In order to specify the direct number of labor hours used for production Source: UWV dataset

1 FTE 0,5 FTE

Total labor hours (52*x40**) 2.080 (hours) 1040 (hours)

Deductions:

HBV leave (2 hours a week) 104 52 Sick leave (absenteeism 5,9%) 114 57 Paid leave 200 100 Holidays (6 days) 48 24 Special leave (1,5 days) 12 6

In office hours** 1602 hours 801 hours

Personal hygiene 100 50 Consultation coaching 30 30 Employment consultation 30 30 HRM-‐interviews 6 6 Cross-‐checking+casuistry 40 40 Reregestration 20 20 Message/circulars 35 35

Training new employee 36 19

Process/system training 40 40

Total direct hours 1.265 hours 531 hours

Notes: =* weeks, ** hours, *** These are the actual office hours that are left after absenteeism, Italic posts are subtracted proportionally, underlinded posts are subtracted imporportionally

(30)

Appendix 1.2

Ailabouni (2010) his overview of the 42 most important factors influencing individual productivity.

(31)

Appendix 2

Table 1 Regression analysis including interaction terms

Analysis of the full sample Source: UWV dataset

Points per FTE 1

Variables Gender -‐0.022 (0.063) Cohort 2 -‐0.102 (0.081) Cohort 2*Gender 0.101 (0.110) N 123 R2 _0,0274

Notes: */**/*** denotes significance at a 10%/5%/1% level, resp., cohort 2 = 0.84<FTE<1, value in brackets are the robust Std. Err.

(32)

Appendix 2.1

Table 3 Regression analysis with trimmed data excluding outliers

Analysis of the full sample Source: UWV dataset Efficiency 1 2 3 4 5 Variables Actual FTE -‐0.313** (0.129) -‐0.300** (0.137) Cohort 1 0.152*** (0.059) Cohort 2 0.030 (0.049) -‐0.126** (0.060) Cohort FT -‐0.075*** (0.045) -‐0.151*** (0.054) Gender 0.016 (0.048) 0.009 (0.048) 0.026 (0.049) Age 0.004** (0.002) 0.005 ** (0.002) 0.004* (0.002) N 99 99 99 99 99 R2 _0.0761 _0.1140 _0.081 _0.122 0.083

Notes: */**/*** denotes significance at a 10%/5%/1% level, resp., cohort1 = 0<FTE<0.84, cohort 2 = 0.84<FTE<1, cohort FTE= 1<FTE<1.05, value in brackets are the robust Std. Err.

Part-time versus Full-time employees : a productivity comparison