Economic Activity of the Disabled in Poland in 2010

Economic Activity of the Disabled in Poland in 2010 The disabled people in Poland are that part of labour force that has not been appreciated enough. Despite the fact that in the recent years the number of the disabled Poles who found employment has risen, their employment rate is still rather low. The majority of them (83% in 2010) are absent on the job market. The aim of the paper is to investigate how gender, place of residence, education, age and disability severeness affect the economic inactivity of the disabled and what impact their gender, place of residence and disability severity had on the likelihood of the reason for the inactivity. The author used the Polish Central Statistical Office data concerning the 4th quarter of 2010. The data were analysed by means of the logistic regression model for the dependent dichotomous variable as well as the multinomial logistic regression model. The estimated parameters helped to determine the inactivity risk quotient in relation to economic activity. They also permitted to calculate the probability of the disabled people's economic inactivity due to a particular reason.


Introduction
According to the 2002 National Census the number of the disabled persons in Poland reached almost 5.5 million. The disability of around 4.5 million was legally certified and 4.3 million were 15 plus years of age 1 . The number of the latter was gradually falling to reach 3393 thousand in 2010, which equalled 10.7% of the population aged 15 plus 2 . In the recent years the number of the employed disabled Poles has been growing. In 2010 their professional activity rate amounted 17.4%. The aim of this article was to find out how such variables as gender, the place of residence, age, education and the level of disability affected the economic activity of the disabled people in Poland in 2010. The hypothesis was made that the above mentioned variables in various ways determined the economic activity odds in the examined period. The analysis was conducted by means of the logistic regression model. Thanks to the estimated parameters the author could determine the ratios of the professional activity odds versus the professional passivity ones as well as the ratios of the disabled persons' employment odds versus their unemployment rates according to the examined people's selected features.

Statistical data used in the analysis
The analysed figures came from the publication by the Central Statistical Office (GUS) Aktywność ekonomiczna ludności Polski. IV kwartał 2010 (The Economic Activity of the Polish Population. The 4th Quarter of 2010). For the sake of the study the authors decided to use the term of 'economically active population' to describe people who were professionally active or inactive. The professionally active population included all employed (aged 15 plus) and unemployed Poles, while the group who was professionally inactive contained those who did not work and those who were neither employed nor sought jobs; the job seekers who were not willing to get employed; those who did not seek employment because they were promised a job and waiting to start it for over three months. The group of the unemployed consisted of people aged 16 plus with a certified disability degree or incapacity for work. The data concerned their gender, age, education, the place of residence and the disability degree.
The studied population's disability was divided into three groups. The 1 st (severe) disability degree concerns people from the 1 st disability group or holders of the certificate of full incapacity for work or independent existence. The 2 nd (moderate) disability degree relates to people granted the 2 nd disability group or the holders of the certificate of full incapacity for work. The 3 rd (light) disability degree related to the people belonging to the 3 rd disability group or the holders of either the certificate of limited capacity for work or the certificate of incapacity for farm work. In total the analysis covered 3393 thousand people and their population structure is presented in Table 1. Source: own study based on data from the Central Statistical Office (GUS).
The categories of education, age and the disability degree were divided into groups following the GUS classification. The examined population was split: according to their gender and place of residence -into two groups; according to their education level -into five groups; according to their age -into six groups; and, finally, according to their disability degree -into three groups. 82.6% of the economically active were professionally inactive, while 13.6% of the professionally active disabled persons were unemployed.

The logistic regression model
In order to analyse the data the author used the logistic regression model 3 . The logistic function expressing the incidence probability takes the form: and adopts the values from 0 to 1 4 .
In case of the dychotomic dependent variable the model can be written as follows 5 : where: Y -a dychotomic dependent variable, The expression p p − 1 describes the odds (or risk) of a specific event to happen, where p = P(Y = 1) is the success (or risk) probability. The expression ln is written down as logit(p) and is used in the logit model notation: To construe the logit model parameters we use their transformed form of exp(α i ) which is called the OR (odds ratio) 6 . In order to estimate the model parameters the author used the STATISTICA programme.

The analysis of the disabled persons' economic activity
The author examined the impact of such factors as gender, the place of residence, education, age and a disability degree on the disabled persons' professional activity and employment opportunities. To encode the explanatory variables binary system was used, which enabled the author to compare selected groups of individual features (represented by the number 1) with a selected group (represented by 0) 7 . Gender was represented by men, the place of residence -by rural areas, the education groups -by junior high school graduates, the age groups are represented by people aged 15-24 and, last but not least, out of the disability degree groups only the 1st degree was taken into consideration.
The first stage of the analysis included the construction of models where the dychotomic dependent variable (economic activity) was encoded in the following way: professional activity was attributed with 1, while professional inactivity -with 0. They were called the logit models of professional activity. The author studied the effect the determinants have on the professional activity odds of the disabled. On the second stage of the analysis the author estimated models where the dychotomic dependent variable (professional activity) was encoded in the following way: the employed were represented by 1, the unemployed -by 0. The models were called the employment logit models. It allowed the author to examine the impact of the determinants on the disabled persons' employment.
On the Figures 3-5 the education, age and disability degree groups are marked as in Table 4.  The Figure 1 shows that in 2010 the disabled women were less likely than men to be professionally active (by 35%). On the other hand their employment opportunities were higher by 7% than the men's. The Figure 2 presents the disabled persons' professional activity odds and employment odds ratios depending on their place of residence. It appears that the disabled residents of urban areas had 18% more chance to be professionally active than those living in the rural areas.
In contrast, the employment odds of the former were lower by 27% than the latter.   In both types of models the people who reached merely the junior high school level had the smallest odds (Fig. 3). Obviously, they could not be professionally active since they were still continuing their education, therefore they were treated as professionally inactive. As far as their employment is concerned, they did not have any professional qualifications, so they were not competitive on the job market. In both models the odds increased along with the education.
There was an exception from that tendency in the case of people with secondary education (in relation to professional activity), which may result from the fact that they were proceeding further with their education. Professional activity Employment S2/S1 S3/S1 S4/S1 S5/S1 S6/S1 On Figure 4 the odds ratios show that the lowest employment odds affected people older than 65, which was clearly the result with their retirement. Also the disabled persons aged 15-24 were professionally active to a small extent (probably due to further education) as well as those aged 60-64 (this age bracket includes a large proportion of retired females). The most likely to be professionally active were people aged 25-34, while the disabled 60-64-year olds were the most likely to be employed, followed by the group of 34-45-year olds. The least odds to be employed were typical of the youngest group.  When examining professional activity odds according to the disability degree, the author found out that the people with the 3 rd disability degree had the highest odds (6.02) in comparison to the 1 st degree holders. The 1st disability degree holders had the highest chance to be employed ( Fig. 5).

Conclusions
The author's analysis confirmed the hypothesis that in 2010 gender, the place of residence, education, age and a disability degree were the determinants of professional activity and employment of the disabled Poles. 25-34-year old male urban residents with university education and the disability of the 3 rd degree were most likely to be professionally active.
The least likely were 60+ female residents of rural areas with junior high school education and the 1 st degree disability. Surprisingly, the second type of the constructed models indicated a contrary to the common opinion influence of gender and of the place of residence on the chance of professionally active people to be employed. The obtained odds ratios showed that in the group of the professionally active the most likely to be employed were 60-64-year old women living in rural areas, having tertiary education and suffering from disability of the 1st degree. Find out more on the ways of encoding the explanatory variables in Hosmer, Lemeshow (2000).