There are often large underlying flows between employment, unemployment and inactivity.
On average more people move from unemployment to employment than the other way. More people move from employment to inactivity than the opposite direction and more people move from inactivity to unemployment than the other way.
Over the past year over 500,000 each quarter have moved from unemployment into employment. Just over 400,000 per quarter have moved from employment to unemployment.
Comparing Q1 2013 with Q2 2013 around 1.4% of employed people moved into unemployment.
Comparing Q1 2013 with Q2 2013 around 22% of unemployed people moved into employment.
In each year over the past decade, the top two jobs that people either joined or left were sales assistant/retail cashier and elementary personal service occupations, such as bar staff or waiters.
Between Q1 2013 and Q2 2013, those with a degree were 2.6 times more likely to move from unemployment to employment than were those with no qualifications.
Between Q1 2013 and Q2 2013, those who had been unemployed for less than three months were 3.2 times more likely to move into employment than those who had been unemployed for over two years.
In June to August 2013 there were 31.3 million people aged 16 to 64 within the UK labour market, comprising of 28.9 million who were in employment and a further 2.5 million who were unemployed. There were also a further 9.0 million people who were outside of the labour market, known as economically inactive, which are individuals who are not looking for, or available to work. Reasons for economic inactivity include looking after the family or home, early retirement, study and being sick or disabled. All of the upcoming analysis includes only those aged 16-64.
People are constantly moving in and out of the labour market as well as within it, for example between employment and unemployment. Many people also move within employment as they move from one job straight into another. Although the net change may only show, for example, a small increase or decrease in employment, there are hundreds of thousands of people moving out of employment into unemployment and inactivity, as well as, in the opposite direction into employment from unemployment and inactivity. Therefore looking at the flows between each of these economic statuses is useful in understanding the dynamics of the UK workforce.
Information on the UK labour market comes from the Labour Force Survey (LFS) which is a quarterly survey in which households are interviewed for five consecutive quarters. This makes it possible to estimate the movements of individuals between quarters in employment, unemployment and inactivity. As a result, a longitudinal data set can be created over a two quarter period so that changes in labour market states and the movement of individuals between them can be analysed. However, this means that we can only look at the difference in labour market states between one point in time and three months later, and this will mean that some of the changes between labour market statuses, for example someone may move from employment to unemployment and then move back into employment within the three month period will be missed.
Comparing the first and second quarters of 2013, and looking at those aged 16 to 64, 27.8 million people stayed in employment between the two quarters, 1.5 million remained unemployed and a further 8.0 million remained inactive.
Looking at the flows, which are people moving from one state to another, and comparing Q1 2013 to Q2 2013, there were more people who moved from unemployment to employment (545,000) than in the other direction (404,000). This resulted in a net flow into employment from unemployment of 141,000 (545,000 minus 404,000).
The net flow into employment from unemployment was offset by a net flow from employment into inactivity of 33,000, resulting in a net increase in employment of 108,000 (141,000 minus 33,000).
As this data set only includes those individuals who are matched on age and date of birth from the previous quarter of the LFS there is the possibility for bias in the results (further explanation of this can be found in Appendix 1). Therefore the best estimate of employment levels will come from the cross sectional data. This analysis provides an insight into the people who are moving between employment, unemployment and inactivity.
The 141,000 net loss in unemployment from the flow to employment was offset by a net flow into unemployment from inactivity of 125,000, resulting in a net decrease in unemployment of 16,000 (125,000 minus 141,000).
So people move in and out of the three states all of the time but the general direction is a net flow from inactivity into unemployment as people who were not available or looking to work begin their job search, as well as, those people who have turned 16 or have left full time education.
This is followed by a net flow from unemployment to employment as people then find work and finally there will be a net flow from employment to inactivity as some people in work choose to leave the labour market, for example to look after the family or home, or to retire.
Since the end of 2001, which is the earliest period that a consistent time series is available in with to analyse labour market flows, there has been a constant net flow from unemployment to employment, except during the economic downturn seen across the UK in 2008/09. This would have been because some businesses responded to the poor economic conditions by laying off staff, as well as not recruiting staff to replace those workers that had left.
Since the end of 2009, the flow from unemployment to employment has been higher than the flow in the opposite direction, partly because the number of unemployed was so much higher than it was pre recession. Also the number for both flows has been consistently higher than that experienced before the economic downturn.
Looking at each year between April 2002 and April 2013, Sales Assistants and Retail Cashiers were the job groups within which most people either left or entered. Elementary Personal Service Occupations, such as hospital and hotel porters, bar staff and waiters/waitresses, had the second highest number of people either entering or leaving in each year over the same period.
Sales Assistants and Retail Cashiers and Personal Service Occupations have a large number of people working within them compared to some other occupations, which may explain why there is more movement in and out of them. It is also worth noting that the majority of the Sales Assistants and Retail Cashiers and Elementary Personal Service Occupations job roles entered or left were part time roles.
Looking at the movements in relation to the number of people within each occupation group, and where there is a sufficient sample, between April 2012 and April 2013, the highest rate of movement from employment to unemployment was within Elementary Process Plant Occupations, which include roles such as industrial cleaning process occupations and packers, bottlers, canners and fillers.
Analysing job roles entered or left by men and women separately the results show that for men Sales Assistants and Retail Cashiers was the top job role left for 6 of the 11 years between April 2002 and April 2013 and top for job roles entered in 10 of the 11 years. The top job role that was left in 2008 and 2009 were for those employed in Construction Trades.
For women, Sales Assistants and Retail Cashiers were top for the flows from employment into unemployment and the reverse direction from unemployment to employment. In 2011 and 2012 Teaching Professions was in the top three job roles left by women.
It is possible to look at the number of people who flow from one of the economic status in relation to the number of people within that status, to give an estimate of the chance of someone moving. In April to June 2013, there were around 28.8 million people aged 16-64 employed with 404,000 moving from employment to unemployment, so the chance of being employed in one time period and unemployed in the next was about 1.4 in 100. This is known as the ‘hazard of unemployment.’ This means that about 1.4% of people in work became unemployed between the first and second quarters of 2013 - for example, being sacked, being made redundant, or having their contract ended. There will also be further movements within employment as individuals may leave one job and enter another one in a three month period.
Looking back over the period from 2001, the unemployment hazard rate peaked at 1.9% between the first and second quarters of 2009 with the hazard rate starting to rise from the second quarter of 2008 when the recession started. Since the end of the recession the hazard has fallen back to around 1.4% but this is still higher than the levels seen prior to the recession of around 1.2%. This could in part be due to more people taking temporary or short term jobs in recent years. So people within work are now more likely to move into unemployment than they were in the period before the recession.
In the latest period covering the first and second quarters of 2013 those aged 16-24 saw an unemployment hazard rate of 3.3% compared with 1.1% for those aged 40-54 and 1.0% for those aged 55-64. This could be due to the fact that older workers have gained more job experience over their careers and may have been in their current position for a considerable time, as well as, being more likely to leave the labour market to retire rather than to become unemployed.
There is also a churn of younger workers who may have been doing jobs that were not very secure, for example temporary jobs that some may fit in around their study. This churn will also include those young workers who may be trying out a variety of jobs to try and establish the right occupation for themselves.
The youngest age group saw the largest increase in the unemployment hazard rate as a result of the recession, with a peak of 4.2% between the first and second quarters of 2009. The hazard rates across the period for the other age groups of 25-39, 40-54 and 55-64 have followed a similar pattern to each other, and all four age groups show an increase in the unemployment hazard rate as a result of the recession.
It is also worth noting that the hazard rate for those aged 16-24 has now returned to around its pre recession level, whereas the hazard rates for the other three age groups are still above their pre recession levels.
The data used for all age and gender labour market flows analysis uses four quarter rolling averages since seasonally adjusted figures are not available. By using four quarter rolling averages it removes much of the noise in the data and yet still maintains the trend of the series.
The recession had a larger impact on the likelihood of men moving from employment to unemployment compared with women. For the period from the first quarter 2003 to the first quarter of 2008, the period before the recession, the average unemployment hazard rate was 1.4% for men and 1.1% for women.
However, after the recession, covering the period from the second quarter of 2008 to the second quarter of 2013 the average unemployment hazard rate for men had risen to 1.9% compared with 1.4% for women. Therefore the gap between the likelihoods of men and women moving from employment to unemployment has increased from around 0.3 percentage points to 0.5 percentage points.
The unemployment hazard rate for men saw a sharper rise as a result of the recession compared to the rate for women. The unemployment hazard rate for men peaked at 2.1% over the first two quarters of 2009, an increase of around 0.7 percentage points compared with its pre recession average. The unemployment hazard rate for women peaked at 1.4% in the first quarter of 2009, which was an increase of around 0.3 percentage points from its pre recession level. Overall, the unemployment hazard rates for men and women have followed a similar path to one another over the past decade.
In the latest period covering the first and second quarters of 2013, 545,000 people aged 16-64 moved from unemployment to employment. This equates to an employment hazard rate of 21.8%. This means that for every 100 people who were unemployed, around 22 moved into employment over a three month period.
Even though the gross numbers for those moving from unemployment to employment and vice versa are very similar, the employment hazard rate is much higher than the unemployment hazard rate since the number of people employed is much larger than the number unemployed.
The employment hazard rate peaked in the second quarter of 2002 at 31.8%. The average employment hazard rate for the period from the first quarter of 2002 up to the first quarter of 2008, the period before the recession, was 30.4%. However, from the second quarter of 2008 onwards, which includes the economic downturn, the employment hazard rate average has dropped to 22.8%.
An important factor is that since the recession there has been an increase in the number of people who are unemployed and since the employment hazard rate is calculated by dividing the number of people who have moved from unemployment to employment by the initial stock of those unemployed, if the stock of unemployed increases and the number moving into employment remains constant, then the hazard rate will decrease. This could help to explain the decrease in the employment hazard rate since the recession. Since the end of 2008 the number of people moving from unemployment to employment has been increasing, but not by enough to prevent the hazard rate from falling.
In the latest period, covering the first and second quarters of 2013, those aged 55-64 had an employment hazard rate of 18.7%. This compares with 23.9% for those aged 16-24, 23.7% for those aged 25-39 and 21.8% for those aged 40-54. Throughout the period from the beginning of 2003 the oldest age group has always had the lowest employment hazard rate compared with the other three age groups.
Overall it seems that those in the oldest age group are least likely to move from employment to unemployment, but if they do end up unemployed they are less likely to return to employment. All age groups have seen a decrease since the recession. The youngest age group, for example, dropped from around 30% in the period before the recession to 22.6% between the final quarter of 2009 and the first quarter of 2010.
The employment hazard rates for men and women have followed fairly similar patterns to one another, and both peaked around 28-29% in the quarter before the recession. The employment hazard rate for men dropped further and faster than that for women since the recession.
The male employment hazard rate dropped to 20.4% over the period from the final quarter of 2009 to the first quarter of 2010, whereas the female employment hazard reached its lowest level of 21.1% between the third and fourth quarter of 2011, and between the fourth quarter of 2011 and the first quarter of 2012. Since then however, the hazard rates for both men and women have improved, but they are still someway below their pre recession levels.
Since the start of the recession in April 2008 more people have moved from inactivity into unemployment. These people have chosen to re-enter the labour market with the aim of finding a job, although it will also include those entering the labour market for the first time.
In the latest period covering the first and second quarters of 2013, 523,000 people moved from inactivity into unemployment. In the period before the recession started the number of people who moved from inactivity into unemployment was 403,000. The increase in the number of people moving from inactivity into unemployment began before the start of the recession but it has increased more quickly since 2008.
One of the reasons for this could be due to the increasing financial pressures put on households as a result of the recession, so that people cannot afford remain out of employment in order to look after the family or home and instead need to bring in an income. Also the increased availability of flexible working hour jobs may mean that people who want to work but have not been able to due to having to look after the family or home can now look to find a job with flexible working hours.
In addition a number of recent welfare reforms may have had an impact, for example, the replacement of Incapacity Benefit with Employment and Support Allowance as well as the changes to conditionality for Lone Parent Income Support. Lastly, students who may have gone straight into employment are now more likely to move into unemployment from inactivity.
In the latest period, covering the first and second quarters of 2013, the number of those aged 16-24 who moved from inactivity into unemployment was 262,000. This is larger than the pre recession levels of around 200,000.
The number of those in the youngest age group who have moved from inactivity into unemployment has been rising over the last decade and increased at a faster rate during the recession. The number of those moving from inactivity into unemployment is a lot lower for the other three age groups and the oldest two age groups of 40-54 and 55-64 did not see any impact as a result of the recession.
Since the age groups have differing numbers of years within them it is also worth looking at the hazard rates for those moving between inactivity and unemployment. The hazard rates for this flow between inactivity and unemployment show very similar results to the gross flows numbers. The youngest age group of 16-24 had a hazard Rate of 10.7% in the latest period covering the first two quarters of 2013. This is much higher than the second highest group, those aged between 25-39, who had a hazard rate of 6.7%.
Those in the youngest age group of 16-24 and the oldest age group of 55-64 see the largest number of people moving from employment into inactivity. In the latest period, covering the first two quarters of 2013, there were 160,000 people aged 16-24 and 121,000 people aged 55-64 who moved from employment into inactivity.
There are a few reasons why people aged 16-24 may move from employment into inactivity, which include leaving to return to education or that they were working between term semesters of study so at one interview period they were employed but the next they are students, or that they simply have a job while they are still in full time education.
Many of those in the oldest age groups of 55-64 are leaving to retire, although the numbers of people in this age group leaving employment for inactivity has dropped since the recession which may imply that those workers aged 55-64 are having to stay in employment as they cannot afford to retire. The change to the state pension age may also have had an impact on this since women have to retire later.
We can also look at the hazard rate for the flow between employment and inactivity since the age groups have a differing number of years in them. The results using hazard rates show a very similar pattern to those using total numbers for each flow.
For all the data included in this report please click this link:
The chance of moving from employment to unemployment, or in the opposite direction is influenced by a number of factors. For example some jobs are less secure than others, individuals with more experience may be valued more and having a higher educational attainment may open up more opportunities for people. Many factors will also be linked, for example older people will on average have more experience than younger people and this may be a key part in explaining why they have a lower hazard of unemployment.
To control for many of the factors a statistical technique, known as a logistic regression, can be used, which will produce an odds ratio. This technique aims, for example, to work out the likelihood of moving from unemployment to employment depending on a certain characteristic, such as educational attainment or unemployment duration, while keeping all other characteristics the same. The resultant odds ratio will show, for example, that you are ‘x’ times higher for you to move from employment to unemployment if you have no qualifications compared with someone with a degree.
A more in-depth description of the methodology used in this analysis can be found the Appendix 2.
The next section will show the results of this analysis by looking at a variety of topics including qualification, job skill, employment length, unemployment duration and the impact of being in either the public or private sector. It is worth noting that the following reported results only show major differences between the likelihood of different groups and is to show the general overview of the types of factors that can influence an individual’s movement around the labour market.
With more and more people applying for every job and the increased costs of pursuing further education, the impact of qualifications on the likelihood of entering or leaving a job is very important individual characteristic.
Looking over the latest available period covering April-June 2012 to April-June 2013 someone with a degree was 2.6 times more likely to move out of unemployment into employment compared with someone who had no qualifications. This means that the chance of someone with a degree moving from unemployment to employment is 2.6 times larger than the odds of someone with no qualifications making the same flow, with all other factors held constant.
Someone with a degree is also 1.7 times more likely compared with someone with GCSE or equivalent levels of qualifications to move from unemployment into employment. This shows that the chance of exiting unemployment for employment increases the higher one’s educational attainment.
The likelihood was similar prior to the economic recession, covering the period from April-June 2002 to April-June 2008, whereby someone with a degree qualification was 2.9 times more likely to move from unemployment to employment compared with someone with no qualifications.
People without qualifications were 2.6 times more likely to move from inactivity into unemployment compared with someone who had a degree between April-June 2012 to April-June 2013. This could be because those with degree level qualifications move straight from inactivity into employment.
Individuals with a lower educational attainment were also more likely to lose their job and move from employment into unemployment. In the latest year, someone with no qualifications was 1.6 times more likely than someone with a degree to lose their job.
These results will be linked to the skill of the job the individual was doing as those with low levels of qualifications tend to be in low skilled jobs which tend to have the highest churn rate of staff meaning that more people are entering and leaving these types of jobs than higher skill jobs.
Following on from the impact of qualifications is the effect of the skill level of the job people are in or are entering. There is a higher churn rate among lower skilled jobs meaning that there is a larger staff turnover compared with high skilled jobs. This may be because lower skilled jobs tend to be less secure and companies tend to invest less money and training in low skilled jobs compared with high skilled ones.
Between April-June 2012 and April-June 2013, one was 1.4 times more likely to leave a low skilled job compared with a high skilled job.
The skill level of the job had more of an impact pre recession than between April-June 2002 and April-June 2008 as one was 1.7 times more likely to move from employment to unemployment if one was in a low skill job than a high skilled job. This compares with 1.5 times more likely for the recession years from April-June 2008 to April-June 2013.
However, one was also more likely to move into a low skill job from unemployment. Between April-June 2012 and April-June 2013, around two thirds of people who moved from unemployment to employment moved into either a low skill or lower middle skill job. People are more likely to move into a low skilled job from unemployment, but people with lower qualifications are less likely to move out of unemployment into work. This suggests the higher return to work rates of those with better qualifications is because of many of them are taking low skilled jobs, for example students.
The types of jobs in these classifications include Sales & Retail Assistants and Elementary Personal Services Occupations such as hospital and hotel porters, bar staff and waiters/waitresses, and elementary administration occupations such as postal workers and mail sorters.
Employment length is an important characteristic to consider as it may have an impact upon job security as it would be expected that the longer one had been in one’s job the less likely one is to lose it.
Between April-June 2012 and April-June 2013, one was 8.5 times more likely to move from employment into unemployment if one had been employed for less than six months compared with someone who has been employed for between 5 and 10 years.
Over the same period one was 3.0 times more likely to move from employment into inactivity if you had been employed for less than six months compared with someone who has been employed for over 10 years.
The length that one had been employed has had more of an impact since the 2008/09 recession. On average over the period from April-June 2008 to April-June 2013 one was 7.0 times more likely to move from employment into unemployment if one had been employed for less than six months compared to someone who has been employed for between 5 and 10 years.
This compares with being 6.2 times more likely if one had been employed for less than six months compared with someone who has been employed for between five and 10 years over the period between April-June 2002 and April-June 2008.
This shows that more recently employment length has had an impact on the level of an individual’s job security and gives credence to the idea that the last person employed will be the first one who leaves for unemployment.
Similar to the impact of employment length, unemployment duration will have an impact on the chances of finding a job, as those who tend to be long term unemployed are those with the lowest skills and will also have a tendency to lack motivation to find a job if they have been unemployed for a long time.
Over the year from April-June 2012 to April-June 2013, if one had been unemployed for less than three months one was 3.2 times more likely to move from unemployment into employment compared with someone who has been unemployed for over two years, and 1.9 times more likely compared with someone who has been unemployed for between six and 12 months.
The impact of the duration of unemployment on the chances of moving from unemployment into employment has changed since the 2008/09 downturn. Over the period from April-June 2002 to April-June 2008, one was 7.3 times more likely to make this flow if one had been unemployed for less than three months compared with someone who had been unemployed for over two years. This compares to 4.3 times more likely for the period covering April-June 2008 to April-June 2013.
Details of the policy governing the release of new data are available by visiting www.statisticsauthority.gov.uk/assessment/code-of-practice/index.html or from the Media Relations Office email: email@example.com
These National Statistics are produced to high professional standards and released according to the arrangements approved by the UK Statistics Authority.
Information on the UK labour market comes from the Labour Force Survey which is a quarterly survey in which households are re-interviewed for five consecutive quarters. This is done in waves so that every one of the five waves 20% of the interviewees leave the Survey and a new wave enters.
However, for both the two quarter, the dataset used for this analysis, and the five quarter longitudinal dataset matching of individuals, so that their labour market activity can be analysed, takes place using a unique identification variable for each quarter, with the age and date of birth of each case matched. As a result of this, only those cases that are matched are kept so the analysis of labour market flows can only be looked at for 80% of the households that took part in the survey. It is also worth noting that this methodology excludes imputed cases (i.e. information brought forward from a previous period).
Estimating gross labour market flows is subject to a couple of methodological issues, firstly, there are biases that arise from non-response. This is because different groups of people have different likelihood of dropping out of the survey between waves, which means that they cannot be linked and their labour market activity cannot be analysed. There are many reasons why you lose survey members after the first interview, for example, they have moved house or it is not possible to contact them, or they simply refuse to continue to take part. Since these people do not represent the typical population losing them from the survey can introduce bias in the results.
Non-response tends to be higher for young adults, those that are single, those who live in privately rented accommodation and those unemployed. This bias is minimised by a multiple stage weighting method that initially constrains the weights by housing tenure to the cross-sectional datasets, then constrains the weights to the population totals and finally calibrates the weights using age, sex, region and economic status.
Secondly, bias also arises from response errors in the data which produce false flows between economic states. Every survey ever undertaken is susceptible to response bias as respondents may give, deliberately or unintentional, information that is not accurate. This could be due to many reasons, such as mis-understanding the question that they have been asked, or a lack of knowledge or because they want, for some reason, to give incorrect information.
International research does suggest that for questions related to economic activity status, the errors are not systematic and so when looking at the cross-sectional datasets, they cancel out. However, when linking individuals, such errors will lead to changes of economic activity status which will be false and therefore exaggerate any potential bias upwards and produce an over estimation of the gross flows numbers. Gomes (2011), states that the upward bias in the LFS longitudinal data is more prevalent for the flows between unemployment and inactivity.
Information on the UK labour market comes from the Labour Force Survey (LFS) which is a quarterly survey, where households that take part are interviewed for five consecutive quarters. This makes it possible to analyse the movements of individuals between the three main labour market statuses of employment, unemployment and inactivity.
As a result of households being re-interviewed in successive quarters, a longitudinal data set can be created over a two quarter period so that changes in labour market states and the movement of individuals between them can be analysed. This means that we can only look at the difference in labour market states between one point in time and three months later.
However, the limitation with this method is that the 2 quarter longitudinal data may miss some of the changes between labour market statuses, for example someone may move from employment to unemployment and then move back into employment over a three month period. Therefore it can be argued the actual number of flows between the labour market statuses may be higher than the ones reported.
To analyse the impact of some of the different factors that may affect the likelihood of an individual undertaking a labour market flow a statistical technique known as a logistic regression can be used.
Logistic regression is less restrictive than ordinary sum of squares regression since it does not require normally distributed data, but like ordinary regression, logistic regression provides a co-efficient which measures each independent variable’s partial contribution to variations in the dependent variable.
The basic logistic regression analysis starts with logit transformations of the dependant variable through the utilisation of maximum likelihood estimation. This is done with the odds ratio, which in this analysis has been used to show the likelihood of a particular labour market flow occurring for an individual given a specific personal characteristic, such as educational attainment or employment length.
The odds ratio for an event occurring, for example moving from unemployment to employment if you have a degree compared to someone with no qualifications, is represented as the probability of the event outcome divided by 1 minus the probability of the event outcome. Since we are comparing one group with another, for example the chances of moving into employment for those with a degree and for those with no qualifications then the odds ratio is equal to the probability of moving into employment if you have a degree divided by the probability of moving into employment if you have no qualifications.
The odds ratio can be described as follows;
Oddsi = [ p_i/〖1-p〗_i ] = e β0 + β1X1+…. βnXn
where pi is the probability of an event i occurring, e is an exponential function and βo + β1X1 + … represents the regression model.
Therefore the odds ratio of an event occurring can be denoted as equalling the exponential of the regression model, or for each individual event, the exponential of its co-efficient, β. Within Stata it is possible to run the logistic regression to either give you the co-efficient, β, which you can then take the exponential of (this is done using the command logit) or to give the odds ratio rather than the co-efficient (this is done using the command logistic).
Each particular labour market flow has a different set of independent variables on the right hand side of the regression equation. This is because some factors, such as employment length, are no applicable when looking at flows from unemployment to inactivity. Below is a list of all of the different factors that have been included in some or all of the regression equations for the different labour market flows.
Looking at our analysis for those who have moved from employment into unemployment (EU) between April 2012 and April 2013 the regression equation is as follows;
EU = β0 + β1basic information + β2full time + β3job skill + β4employment length + β5education + β6public
where basic information includes the variables age, age2 and sex, full time denotes whether an individual’s job was full or part time, job skill denotes the skill level of the job ranging from high skill through upper and lower middle skill to lower skill. Employment length is the length of time an individual has been employed, with a reference category of less than 6 months. Education denotes the educational attainment of an individual ranging from degree or equivalent qualifications through to having no qualifications at all. Lastly public denotes whether the job role was in the public or the private sector.
The right hand side of the regression equation is the same for looking at the flow from employment into inactivity (EN) as well as for the flow in the opposite direction from inactivity to employment (NE).
This is because for each factor, such as job skill, there is a variable for the first quarter and the second quarter within the longitudinal data set. This means that for the flows out of employment the job skill variable for the first quarter will be used, whereas for the flow from inactivity into employment the job variable for the second quarter is used in the regression equation.
As noted that some of the above variables are not applicable for looking at flows out of unemployment, for example into employment (UE) the regression equation will be as follows;
UE = β0 + β1basic information + β2full time + β3education + β4unemployment duration
where the basic information, full time and education variables are the same as above. Since full or part time, employment length and public or private sector variables are not applicable for this flow, they have been replaced with unemployment duration which denotes the length of time an individual has been unemployed, ranging from less than 3 months to over 2 years.
For the regression analyses that looks at the pre and post recession years separately (one data set covered the period from April 2002 to April 2008 and the other data set covered the period April 2008 to April 2013) a variable ‘idall’ is included whereby each category is a different year, with the reference category being 2002 and 2008 in the respective data sets. Therefore, for example, for the flow from employment to unemployment (EU) the regression equation was as follows;
EU = β0 + β1basic information + β2full time + β3job skill + β4employment length + β5education + β6public + β7idall