Unemployment, labour force participation and employment in the north of
the Netherlands
A simulation of three different employment growth scenarios for 2010-2030
Master thesis for Economics August 2011
The main goal of this thesis is to present a more realistic simulation of the unemployment rate, employment growth rate and labour force participation rate in three different employment growth scenarios for the north of the Netherlands from 2010 up to 2030. Schudde et al. (2010) expect a negative unemployment rate after 2017. This thesis argues that this is not a realistic result. To improve these expectations, this thesis assumes an endogenous labour force participation rate. The expected future levels of the above variables in the different scenarios are calculated using the Blanchard and Katz simultaneous equation model (1992) with regional data. The main results from this simulation are that the expected value of the unemployment rate is not negative and that a scenario with a higher employment growth rate will lead to a higher participation rate and a lower unemployment growth rate.
Keywords: labour force participation, unemployment, employment growth, north of the Netherlands
Janneke Kaman j.kaman@student.rug.nl
2 1. Introduction
The north of the Netherlands contains the departments: Groningen, Friesland and Drenthe. This is a labour market with a relatively high unemployment rate and a relatively low labour force participation rate. Another feature of the labour market in north of the Netherlands is a lower educational level. There is a relatively low demand of individuals who have received higher education. Highly-educated individuals will withdraw from the region. The development of employment is an important indicator of the labour market (SEAN 2003). Vermeulen (2005) explains that regional labour markets can exist, because migration is costly for labour and capital.
A high unemployment rate is costly to the community. For instance, if there are many unemployed workers, more money is needed to pay for the social security system. Therefore, an economic policy aimed at increasing the labour force participation rate has several economic and social benefits. An important benefit of a higher labour force participation rate and a lower unemployment rate is that the costs of the social security system will be reduced.
It is therefore interesting to know how labour force participation and unemployment will develop in the future. For the northern part of the Netherlands this has been investigated in the paper of Schudde et al. (2010).
This paper investigates how employment and unemployment will develop in the period 2010-2030 in three different scenarios. These scenarios are: quick recovery1, double dip2 and the new common3. The quick recovery scenario contains an autonomous employment growth path of 1.75 percent. The second scenario, double dip, contains a second economic dip in 2011. After this dip, the economy will improve slowly to an autonomous employment growth path of 1.75 percent annually. The last scenario, new common, assumes that after 2009, there is a small improvement of the economy after 2009 and it expects a lower autonomous employment growth path, namely one percent per annum.
The study of Schudde et al. (2010) expects a negative unemployment rate after 2017 in the quick recovery scenario, which is an unrealistic result. The negative unemployment rate is possible, because Schudde et al. (2010) have assumed a fixed labour force participation rate. This labour force participation rate will remain at the level of the year 2009.
1 In Dutch: snel herstel 2 In Dutch: dubbele dip 3
3 In reality, labour force participation will decrease if unemployment increases. This is for example due to the discouraged worker effect. This means that disappointed job seekers leave the labour market. Because the disappointed job seeker is not longer part of the active working force, the individual drops out of the unemployment rate and thus the labour force participation rate will decrease. Several authors, for instance Fleisher and Rhodes (1976) and Elmeskov and Pichelmann (1993), describe this negative relationship between unemployment rate and labour force participation rate.
The study of Schudde et al. (2010) expects a negative unemployment rate after 2017, which is not a realistic expectation. The aim of this thesis is to simulate a more realistic expectation for unemployment rate in Groningen, Friesland and Drenthe with an endogenous labour force participation rate. This thesis will use the three scenarios described above of employment growth and investigates what the effects are of an endogenous labour force participation rate on unemployment and labour force participation. This thesis uses the model of Blanchard and Katz (1992) and argues that these simulations represent more realistic expectations.
This paper addresses the following question: What are more realistic developments of the labour force participation rate, the unemployment rate and the employment growth rate in these three different scenarios in the period 2010-2030? In this thesis, the labour force participation rate, the employment growth rate and the unemployment rate are endogenous variables.
4 Blanchard and Katz model and the study of Schudde et al. (2010). Finally, section seven outlines the conclusions of the thesis.
2. Developments in the labour force participation
2.1. History of labour force participation over the preceding three decades
Before simulating the three different scenarios in the near future, an overview of the development of labour force participation over the preceding three decades is presented in this section. In the 1980s, the Dutch labour force participation rate was lower in comparison with other EU countries. In the 1990s, the labour force participation rate increased. This was the effect of a changing economic policy. Before this change in economic policy, the key government target in this policy area was to diminishing unemployment. After the policy turnover, the target of economic policy changed to stimulating the labour force participation rate, because a higher labour force participation rate will lead to a lower unemployment rate and will also solve the hidden unemployment problem. The government wanted to stimulate the labour force participation of elderly, women and workers who receive an invalidity insurance benefit4. This policy was successfully and led to high employment growth in the 1990s. Nowadays, the labour force participation rate in the Netherlands is one of the highest in Europe.
The development of the labour force participation rate is influenced by (1) a change in demography, (2) an increase in the level of education, (3) and a change in social opinion about labour force participation. The demographic change refers to an ageing population. This well-known phenomenon will lead to a lower labour force participation rate, because part of the population aged between 55-64 years exhibits a lower participation rate.
Another development that influences the participation rate, is an increase in the level of education; the labour force participation rate is higher under highly-educated individuals. Via a higher level of education, labour productivity also increases and thus leads to a higher income level. The higher earnings lead to higher opportunity costs. So, these individuals will have more incentives to participate in the labour market.
Finally, a change in the social opinion has increased labour force participation. Through the high unemployment in the 1970s, the costs of social security were high. On top of this development, the emancipation of women increased in this period just as the education
4
5 level. This is an important reason why there was an increase in labour force participation. In addition, economic policy started to stimulate labour force participation actively (WRR 2000).
There is explained which developments influences the participation rate. Next, it is useful to clarify the reasons why a high percentage in labour force participation is necessary to maintain the welfare state. A high percentage of the labour force participation rate has positive economic effects.
First, through more labour force participation, it is possible that labour costs will decline, as a result of an increasing demand for labour. A reduction in labour costs will lead to lower prices for goods and services but will also stimulate the demand of services which previously were performed within the households, such as child care. These services will hence shift from the informal market to the formal market. This development will create new employment and leads to an increase in GDP.
Secondly, if more workers pay employed persons' insurance contributions, the premiums of the contributions will decline. This leads to a better competitive position of the Netherlands, because the relative prices of goods and services will decrease. Through lower costs of labour, the quality of goods and services can increase. Furthermore, lower premiums lead to a larger basis for social security (WRR 1990).
2.2. Future expectations of labour force participation
The European Commission (2000) has developed ambitious targets for labour force participation within the European Union. In the Lisbon treaty concluded in 2000, it wants to realize an overall participation rate approximating 70 percent for the whole EU and for women a participation rate of 60 percent in 2010. For 2020, the Commission wants to realize an overall participation rate of 75 percent for the entire EU. The European Commission wants to realize these targets through a higher participation of women, of elderly people and a better integration of migrants in the labour market. In general, the labour force participation rate of these groups is below the average labour force participation rate.
6 Social policies have chosen earlier strongly influence the development of labour supply in the future. The two most important policy effects are: (1) The cancelling of the possibilities to retire prematurely, and (2) a revision of the prerequisites for invalidity insurance benefits. Through this change in policy, the number of individuals assumed to be incapable to work will diminish. These policy decisions therefore lead to an increase in labour force participation.
The growth of labour force participation among women springs from two causes. First, older cohorts of women are replaced by younger cohorts. Since younger cohorts of women have higher participation rates than older cohorts, participation increases. Secondly, there is a trend that younger cohorts participate more in the labour market than the same cohorts did in previous years. It is expected that this cultural trend will persist in the following years.
Demography influences the labour force participation rate when the relative size of population groups changes. There is an inverse U-shaped relationship between age and labour force participation rate. For example, ageing leads to a lower participation rate, because in general elderly individuals will participate less than younger individuals. Moreover, younger individuals will follow education when there is high rate of unemployment (Elhorst 2008).
Based on these developments, Euwals and Van Vuuren expect a labour force participation of 75 percent after 2014.
3. Three scenarios of employment growth
3.1. Demand shocks
This thesis uses the model of Blanchard and Katz (1992) to simulate the unemployment rate, the labour force participation rate and the employment growth rate simultaneously in the period 2010-2030. The advantage of such a simultaneous model is that the effect of every exogenous variable can be determined on every endogenous variable; so with a demand shock, it is possible to analyze what the effects are for the unemployment rate and the labour participation rate. According to Armstrong and Taylor (2000) a positive demand shock leads to an increase in employment. This increase in employment implies that new jobs must be filled by unemployed workers; therefore the unemployment rate will be reduced.
7 A second effect of higher employment is that the regional income of households will increase. An increase in households’ income means that there is more money available for consumption. The demand for consumption goods will thus also increase. Hence, the demand for output in the region will increase, which leads as well to an increase in employment.
Decressin and Fatás (1995) observe that a demand shock leads to a permanent increase in the regional employment for Europe and the US. They observe inward migration in the regions resulting in a higher employment in the long run. They find for Europe that in case of a demand shock, the shock is mainly absorbed by higher labour force participation during the first three years.
Broersma and Van Dijk (2001) present the effects of a positive regional labour demand shock in the Dutch labour market. They find that in the north, unemployment will mainly absorb the shock in the short run. An increase in the labour force participation, however, will absorb the biggest part of the long run effects of this shock. This long run effect is even higher than the share of migration and unemployment. This can be explained by a relatively high unemployment rate in the northern part of the Netherlands. The underlying argument is that through relatively high unemployment rate, jobs can be filled relatively easily by unemployed workers. Broersma and Van Dijk (2001) also observe an adjustment speed of three to five years. This adjustment speed is high in comparison with other regions of the Netherlands. A higher adjustment speed means that the labour market is more flexible because the labour market needs less time to adjust to a demand shock.
3.2. The scenarios
8 Year Quick recovery Double dip New common
2009 -4.00% -4.00% -4.00% 2010 1.50% 1.50% 0.50% 2011 1.75% -3.00% 1.00% 2012 1.75% 1.00% 1.00% 2013 1.75% 1.50% 1.00% 2014 1.75% 1.75% 1.00%
Table 1: Assumptions for employment growth in three scenarios in Schudde et al. (2010)
Table 1 shows an expected negative employment growth of four percent in 2009. At the moment when this paper was published, this negative growth rate was commonly expected. However, the actual employment growth in 2009 was -0.8 percent for Groningen, Friesland had an employment growth of zero percent, and for Drenthe it was -0.2 percent.
3.3. Survey of the northern labour market
This section analyses the results of the study by Schudde et al. (2010). This paper assumes a steady decline in labour force in the period 2010-2020. The labour force will decline with four percent5 in the northern part of the Netherlands, in line with the expectations of the Statistics Netherlands6.
Schudde et al. (2010) compare the results of the three different scenarios with the situation of 2009. In 2009, the unemployment rate was approximately six percent and employment was about 732,000.
Figure 1: Development of the unemployment rate in % (left), right: employment (x1000). V represents the quick recovery scenario, W represents the double dip scenario and nieuwe normal represents the new common scenario.
Source: Schudde et al. (2010), p18.
5 The decline in labour force is from 780,000 to 750,000. This is 4%. 6
9 From figure 1, one can observe that in the first scenario, quick recovery, the unemployment rate will achieve the highest value of eleven percent in 2011. After this year, employment will increase and the unemployment rate will decrease and become zero after 2017. This is caused by the assumption of an autonomous employment growth rate of 1.75 percent. In addition, the unemployment rate is taken as the difference between labour force and employment. This means that the labour force participation rate is exogenous. In this scenario a labour shortage will arise.
In the second scenario, double dip, it takes more time to recover from the second dip in 2011. Initially the unemployment rate will increase to ten percent in 2012-2013. After that, one can observe a downward sloping trend and this leads to an unemployment rate of approximately six percent in 2017. In addition, it takes more time to achieve the autonomous employment growth path of 1.75 percent. This growth path will only be achieved in 2014.
Finally, the new common scenario will exhibit an employment decrease after 2009, with the rate reaching is at its lowest level in 2011. After this year, employment will rise very slowly and will stay below the level of 2009 until 2020. The unemployment attains its highest value of nine percent in 2011. After that, a downward sloping trend commences and the unemployment rate will be approximately six percent in 2016.
In general, one can observe in all three scenarios that when the economy recovers from the negative employment growth in 2009 or from the second dip in 2011, the unemployment rate will fall. The unemployment rate in the scenarios double dip and quick recovery will fall more rapidly in comparison with the new common scenario. This is caused by the relatively lower autonomous employment growth path in the new common scenario.
Schudde et al. (2010) also studied two demographic variations of the quick recovery scenario, namely a strong decline in the labour force and an extra migration of workers to the North. In the demographic scenario of strong decline, the labour force will decline stronger in the period 2010-2020. Via this stronger decline in the labour force, the unemployment rate will decline faster after 2010, so a labour shortage arises after 2014. Moreover, the unemployment rate will become zero earlier, namely in 2016.
10 the unemployment rate will return to its 2009 level in 2013. Finally, positive migration will also lead to employment growth, which will concentrate in economic centres.
In general, one can observe in the three scenarios, that a larger labour force will lead to a longer period with a higher unemployment rate. When there is a smaller labour force, the unemployment rate will be lower.
4. Data
4.1. Description of the data
The data on employment, unemployment and labour force participation for Groningen, Friesland and Drenthe is derived from Eurostat and covers the period 1999-2009. These data is on a regional level (NUTS2). For the Netherlands NUTS2 data corresponds to the twelve departments of the Netherlands.
The model in this thesis contains regional data instead of national data. The reason being that the Blanchard and Katz model has a regional basis. Blanchard and Katz (1992) introduce this model in a paper which is based on regional data. The second reason is to explain the regional differences within the country. National data cannot distinguish the different regional levels of employment, unemployment and labour force participation between the different regions.
4.1.1. Employment growth rate
Employment is defined as the number of employed workers between 15-65 years. The employment rate is the percentage of employed workers in relation to the total population in a region. The employment growth rate has been derived by taking log of the relative difference between employment in the two consequent periods. Formally this yield to: log(et/et-1), where et measures the employment level in the current period and et-1 measures the employment level in the previous period.
4.1.2. Labour force participation rate
11 4.1.3. Unemployment rate
Unemployment is defined as the part of the labour force that is out of a job, but is available to start working and is actively searching for a job. The unemployment rate is the number of unemployed persons in the total labour force.
4.2. Data analysis
From the data of Eurostat, one can compute a negative employment growth rate for the three regions in 2009. A negative employment growth rate is remarkable. One can explain the negative value through the credit crisis in the concerning year. The same phenomenon was visible during the previous crisis, when all three regions exhibited a negative employment growth rate in 2004.
One can observe that the employment growth rate fluctuates over time but it can also be observed that there are differences in the employment growth rate between the regions. For example in Groningen and Drenthe, the employment growth rate is relatively high in 2000, whilst in Friesland the growth rate is relatively low in the concerning year.
When looking at the unemployment rate, one can derive that this rate responds to employment growth. When the employment growth is relatively low or negative in a specific year, the unemployment rate increases in the region the following year, and vice versa.
For Groningen and Drenthe, the unemployment rate is relatively high in 1999. After that, one can observe stabilization. The unemployment rate in Friesland is almost stable in the period 1999-2009.
12
Figure 2a: percentage regional unemployment rate 1999-2009
Figure 2b: percentage regional labour force participation rate 1999-2009
13 5. The Blanchard and Katz model
5.1. Sketching the model
The model of Blanchard and Katz contains three equations, which explain the interaction between employment growth, labour force participation and unemployment at a regional level. The model is defined by the following formulas:
u = β11u[-1] + β12p[-1] + β13e[-1] + β14e (1)
p = β21u[-1] + β22p[-1] + β23e[-1] + β24e (2)
e = β31u[-1] + β32p[-1] + β33e[-1] (3)
where u is unemployment rate, p is the log of the labour force participation rate and e is the employment growth rate. All these variables are endogenous. The model is recursive because the unemployment rate and the labour force participation rate in the current period are dependent on the current employment growth rate and the employment growth rate of the previous year. In the Blanchard and Katz model, employment growth is explained by unemployment, labour force participation, and employment growth of the previous year.
This thesis adds fixed effects to the model. These fixed effects are also used in the paper of Elhorst et al. (2006). The fixed effects are included for the unemployment rate and the labour force participation rate.
In the paper of Elhorst et al. (2006), there are also time-period fixed effects included in the estimation, but in this simulation this is left aside because time-period fixed effects are used to correct for economic expansion and retraction. In the assumptions of this thesis, one does not expect economic retraction and expansion. Therefore, time-period fixed effects are not applicable to this simulation.
The model is now defined by the following formulas:
u = β11u[-1] + β12p[-1] + β13e[-1] + β14e + xit + cu (4) p = β21u[-1] + β22p[-1] + β23e[-1] + β24e + yit + cp (5)
e = β31u[-1] + β32p[-1] + β33e[-1] + ce (6)
14 is the regional fixed effect for labour force participation in region i. The constant of each formula is given by c.
5.2. Description of the coefficients
The coefficients used in this paper are derived from the estimation results of Elhorst et al. (2006). These coefficients are based on data from 1986-2001 in 111 different regions in seven EU countries.
Endogenous variables Unemployment Labour force participation Employment growth u[-1] 0.763415 -0.360037 -0.023152 p[-1] -0.053578 0.75177 -0.02502 e -0.150609 0.408767 e[-1] -0.025125 -0.023151 -0.000694 constant -0.000558 -0.07177 0.002406 R-squared 0.997 0.975 0.154 Fixed effects Groningen -0.004797 -0.003677 Friesland -0.005842 -0.004481 Drenthe -0.00488 -0.001481
Table 2. Estimation results of the coefficients for the Blanchard and Katz model. Source: Elhorst et al. (2006), p 25.
Now, an explanation of the coefficients of table 2 is presented. First, one can observe in the unemployment rate equation, that the unemployment rate of the current period is highly correlated with the unemployment rate in the previous period. The value is respectively 0.76. The value of u[-1] in the unemployment rate equation means that if the unemployment rate in the previous period is one percent higher, the value of the current unemployment rate will be 0.76 percent higher.
In economics, lagged variables are considered a good estimation for current variables. For instance, if unemployment or labour force participation was high in the previous period, it is unlikely that the unemployment rate or labour force participation rate in the current period have changed to a relatively lower level.
Furthermore, one can observe the negative relationship between the unemployment and participation rate. If the labour force participation rate has increased with one percent in the previous period, one can observe from table 2 that the current unemployment rate level will decrease by 0.05 percent.
15 unemployment rate, respectively, -0.15. The underlying reason for these negative effects is that a new created job must be fulfilled through an unemployed worker or a non-participant.
Secondly, in the labour force participation equation, one can observe some similarities with the unemployment rate equation. In addition, the labour force participation rate of the current period is, highly correlated with the participation rate in the previous period. The value is respectively 0.75.
The value of u[-1] for the labour force participation rate (-0.36) means that if the unemployment rate in the previous period becomes one percent higher, labour force participation will decrease by 0.36 percent. This corresponds with economic theory; if the unemployment rate becomes higher, workers withdraw from the labour market due to the discouraged worker effect.
The values of e[-1] is only -0.023 for the labour force participation. This relation however is probably insignificant. The T-value in the paper of Elhorst et al. (2006) is 1.69. Employment growth rate in the previous period, therefore, only has a small effect on the current period.
Finally, it describes the coefficient of the employment growth equation. From table 2, one can observe a negative relationship between the employment growth rate and the lagged unemployment rate, the lagged labour force participation rate and the lagged employment growth rate. This relation, however, is probably insignificant7, thus the influence of the employment growth rate on the other endogenous variables is uncertain.
5.3. Computing the model
If the model is simulated with above coefficients, one obtains the following results for 2030. First, there is an expected labour force participation rate of 70.0 percent, the long run unemployment rate will be around 4.9 percent, and finally the employment growth rate will be around one percent. This can also observed in figure 3a-c.
7
16
Figure 3a: expected unemployment rate with the original coefficients of Elhorst et al. (2006)
Figure 3b: expected labour force participation rate with the coefficients of Elhorst et al. (2006)
17 As explained earlier in section 2.2, the above simulation does not result in the long run labour force participation rate of 75 percent expected by the European Commission (2000) and Euwals and Van Vuuren (2005). There are three reasons for the lower labour force participation rate in this simulation.
First, the lower value of participation can be explained by the difference in data samples between this thesis and the paper of Elhorst et al. (2006). The paper of Elhorst et al. (2006) contains data from 1986-2001, whilst here data from 1999 up to 2009 is used.
Secondly, there is a difference in the regions which are included. This thesis describes the north of the Netherlands, whilst the data used by Elhorst et al. (2006) contain regions from 111 different regions in seven EU countries.
Finally, as explained earlier, in the 1980s the labour force participation rate was lower, due to national economic policy aimed at increasing participation, the labour force participation rate in the Netherlands has increased. Nowadays the labour force participation rate of the Netherlands is one of the highest in Europe whilst, in the beginning of the 1980s, the participation rate of the Netherlands was one of the lowest.
In accordance with the above reasons, the labour force participation rate does not yield the expected result of 75 percent.
However, one can adjust the constant of the labour force participation rate in the labour force participation rate equation of the Blanchard and Katz model, to a level that will result in a 75 percent participation rate. The structure of the model will remain the same. This small adjustment, however, shows the strength and the possibilities of the model.
18
Figure 4a: expected unemployment rate in the basic model 1999-2030
Figure 4b: expected labour force participation rate in the basic model 1999-2030
19 Next, the model is simulated again with the adjusted constant for labour force participation. This simulation will be called the basic model for the remainder of this thesis.
One can observe that this simulation results in the expected labour force participation rate of 75 percent. Furthermore, one can derive the negative relationship between the labour force participation rate and the unemployment rate. The long run unemployment rate in this simulation hence ends up around 3.5 percent. This can also be derived from figure 4a-c.
When comparing the unemployment rate and the employment growth with the different values of the constant for the labour force participation rate, one can derive the following results. Through the above adjustment, the labour force participation rate becomes five percent higher and the unemployment rate decreases from 4.9 percent to 3.5 percent in the long run. This is approximately 1.4 percent lower. The adjustment of the constant in the labour force participation equation also leads to a small reduction in the employment growth rate. The employment growth ends up around 0.9 percent each year. This is a difference of 0.1 percent with the expected employment growth rate, simulated with the original coefficients in Elhorst et al. (2006).
6. Simulation and analysis of the three scenarios in the Blanchard and Katz model
In this chapter, a demand shock is simulated via an adjustment of the constant of employment growth. In the research of Elhorst et al. (2006), this coefficient is -0.002406. By adjusting the constant, one can simulate employment growth paths, which correspond to the different scenarios in the study of Schudde et al. (2010).
6.1. The scenarios
20
Figure 5a: expected results of the unemployment rate in the three scenarios until 2030
Figure 5b: expected results of the participation rate in the three scenarios until 2030
21 6.1.1. Quick recovery scenario
This thesis uses the possibility in the Blanchard and Katz model to adjust a coefficient. In order to simulate the quick recovery scenario, one adjusts the constant of the employment growth in such a way, that the employment growth rate in the long run converges to an equilibrium of 1.75 percent. When the value of the constant of the employment growth is higher, the employment growth rate becomes higher as well. The constant of the employment growth rate is thus adjusted to 0.01154 instead of 0.002406. Thus the value of the constant in the quick recovery scenario is higher than the value in the basic model.
One can observe that the unemployment rate is decreasing very fast in the period 2010 up to 2014. In the long run, however, unemployment will also decrease at a lower speed. The long run equilibrium for Groningen is 2.6 percent, for Friesland 1.9 percent and for Drenthe this unemployment rate is 2.2 percent. Compared with the other two scenarios, this scenario ends with the lowest unemployment rate. In contrast with the results from Schudde et al. (2010), this simulation does not result in a negative unemployment rate.
As a result of a negative employment growth and a high unemployment rate in 2009, the labour force participation has decreased to 76.3 percent in 2010. After 2010, the labour force participation rate will improve. This is due to a positive employment growth rate and via a decrease in the unemployment rate. This simulation shows that a lower unemployment rate will lead to an increase in the labour force participation rate.
6.1.2. Double dip scenario
To simulate the double dip scenario, a negative shock is added to the model for the employment growth rate in 2011. Schudde et al. (2010) expect a negative employment growth of three percent in 2011. Hence the equation of the employment growth in 2011 is adjusted to:
e = β31u[-1] + β32p[-1] + β33e[-1] + ce + DD (7)
where DD is the double dip, the negative shock in employment growth of -3 percent in 2011. For all other years the value is zero. The meaning of all other variables is the same as in equation (6).
22 the quick recovery scenario. The reason for this is that in both scenarios in the long run the employment growth is 1.75 percent. Notice that the constant of employment growth rate is higher in comparison with the basic model.
When analyzing the results of the model, it is immediately apparent from the simulations that the unemployment rate is higher in the years after 2010 in comparison with the quick recovery scenario. Moreover, the effect on the labour force participation rate in the following years is larger in comparison with the quick recovery scenario. This higher unemployment rate can be explained through the negative employment shock in 2011. As a result of a negative employment growth shock will have a negative effect on unemployment and labour force participation in the following years. Especially, the labour force participation is affected through this negative shock. In 2017, the economy has recovered from the second dip. In this year, the labour force participation rate will be back at the expected level of 75 percent.
In the long run, the participation rate is 0.1-0.2 percent lower and the unemployment rate is 0.1-0.2 percent higher when compared with the quick recovery scenario. For the three regions the labour force participation rate will be around 77.3 percent and the unemployment rate around 2.3 percent, with an employment growth of 1.75 percent annually.
6.1.3. New common scenario
In this scenario, the constant of the employment growth is adjusted to 0.0036 instead of 0.0024, which is used in the basic model. The adjustment is smaller in comparison with the quick recovery and double dip scenario and the value of the employment growth constant in this scenario is much lower in comparison with the other two scenarios. In the short run, the labour force participation rate will decline to around 74 percent in 2014. After this year, the participation rate will slowly increase. In the long run, the participation rate becomes 75.4 percent and the unemployment rate becomes 3.3 percent. Through this adjustment, in the long run the employment growth becomes one percent.
23 6.2. Comparison of the long term results of the scenarios
Before describing the long term results of the three scenarios, there is presented an overview of the outcomes in table 3.
Scenario Unemployment rate Labour force participation
Employment growth rate
Basic model 3.5% 75.0% 0.9%
Quick recovery 2.2% 77.4% 1.75%
Double dip 2.3% 77.3% 1.75%
New common 3.3% 75.4% 1.00%
Table 3: the value for the endogenous variables in the long run, 2030.
Table 3 presents the long run values for each scenario. From table 3, one can derive that the quick recovery scenario has the largest positive long run effects. In the three scenarios, the labour force participation rate is higher than in the basic model. This is due to the fact that the employment growth in the scenarios is higher than in the basic model. As explained earlier, higher participation means a lower unemployment rate. Therefore, it is not surprising that the unemployment rates are lower than in the basic model.
The values of the endogenous variables in the quick recovery scenario and the double dip scenario are almost the same in 2030. The employment growth rate is identical, because an employment growth rate of 1.75 percent is assumed in both scenarios. In the long run, the unemployment rate in the double dip scenario is 0.1 percent higher than in the quick recovery scenario. For the labour force participation rate, the opposite holds. Here, the participation rate is 0.1 percent higher in the quick recovery scenario. In the long run, the variables of the Blanchard and Katz model go back to their equilibrium, thus it is reasonable that these outcomes are almost identical in the long term.
24 Comparing the expected results of this thesis with Broersma and Van Dijk (2001) observed an adjustment speed of three to five years after a demand shock. This corresponds with these simulations.
Decressin and Fatás (1995) observed for Europe that in the first three years, there are adjustments in the labour force participation rate in response of a labour demand shock. In these simulations, one can derive that the effects on the labour force participation are in line with this analysis.
6.3. Similarities and differences between the results of the Blanchard and Katz model and the survey of the northern labour market.
6.3.1. Similarities
Comparing the results of this thesis to those of Schudde et al. (2010), one can observe several similarities and differences. This section describes the similarities.
The expected long run results of the three scenarios exhibit several similarities. In both studies, when compared with the other two scenarios, the quick recovery scenario expects the lowest value of the unemployment rate in 2017. The quick recovery scenario is the most positive scenario for the regional labour market in both studies. Thus, a situation which is the most positive for the labour market will result in the lowest values of the unemployment rate.
Secondly, one can observe similarities in the expected results of the endogenous variables. In both studies, the unemployment rate is high after 2009. This is the effect of a negative employment growth in 2009. When the economy has recovered from this negative employment growth, in both studies the unemployment rate will decrease.
Finally, the development of employment growth is the same in both studies, but this could be expected, because in both studies a certain path of employment growth is assumed in the three scenarios.
6.3.2. Differences
Besides the similarities between the two studies, there are also differences between the expectations in unemployment, labour force participation and employment growth.
25 the study of Schudde et al. (2010). In this study the double dip scenario yields the highest value of the unemployment rate. The differences between the studies can be explained by the fact that in this thesis the labour force participation rate is endogenous. Therefore, it is possible that workers withdraw from the labour market when they see that the unemployment rate will increase. This means an increase in the hidden unemployment.
Secondly, there is a difference between both studies in the development of the unemployment rate after 2017. In this thesis the unemployment rate is depends as well on the labour force participation rate, which leads to an equilibrium value of the unemployment rate in the long run. A negative expected value of the unemployment rate is therefore impossible in this thesis. In the study of Schudde et al. (2010), the downward sloping trend of the unemployment does not stabilize but is a linear downward sloping line. This is caused by the combination of an autonomous employment growth path and a labour force participation rate which is constant over time.
Finally, one can observe differences in the expectations of the labour force participation rate between both studies. In this thesis, one can analyze that the labour force participation will be adjusted by the developments in employment growth and the unemployment rate. After 2009 the participation rate decreases, but in the following years the participation will recover, and in the long run it will stabilize around an equilibrium value. In the study of Schudde et al. (2010), the labour force participation is assumed exogenous, so it is constant over time.
7. Conclusion
This thesis presents a simulation of the unemployment rate, employment rate and labour force participation rate in three different scenarios in the north of the Netherlands from 2010 up to 2030.
First, it represents an explanation of the developments in the labour force participation rate in the Netherlands. The participation rate was low in comparison with other countries in Europe in the 1980s. In the 1990s, the labour force participation rate has increased. This is the effect of a policy turnover in the Netherlands. As explained in section 2.1, this increase in the labour force participation rate is also important to maintain the welfare state.
26 participation rate increases to 75 percent in respectively 2020 and 2014. The expected percentage is based on economic policy and demographic developments.
Schudde et al. (2010), expect that unemployment will disappear in the quick recovery scenario after 2017. This thesis has clarified why this is not a realistic expectation. The assumption behind this unrealistic expectation is an exogenous labour force participation rate. Schudde et al. (2010) assume a fixed participation rate identical to the rate of 2009. A more realistic assumption is that the labour force participation will decrease if unemployment increases. In this thesis the labour force participation rate is treated as an endogenous variable. This thesis uses the model of Blanchard and Katz (1992) to present more realistic expectations. In this model the rates of unemployment, labour force participation and employment growth are endogenous variables.
Furthermore, this thesis argues that the quick recovery scenario ultimately yields the lowest unemployment rate, namely 2.2 percent, and the highest labour force participation rate, namely 77.4 percent, in 2030. The new common scenario gives the highest unemployment rate, and the lowest participation rate of respectively 3.3 percent and 75.4 percent. This is caused by the assumption of a lower employment growth in the new common scenario as in the other two scenarios.
Next, this thesis argues that a negative shock in the employment growth rate, such as in the double dip scenario, has negative effects in the short run and in the long run. In the short run, the labour force participation rate immediately reduces, and unemployment will also remain higher. In the long run, the labour force participation rate is lower as compared with a situation without a shock and the unemployment in the long run is higher through the negative shock.
A higher employment growth will lead to a higher labour force participation rate and will thus reduce the unemployment rate in the long run. The underlying mechanism is that, if employment is created, these new jobs must be filled by unemployed workers.
Comparing these results with the results of Schudde et al. (2010), one can observe some remarkable differences in the results. First, in this thesis the labour force participation rate is an endogenous variable, therefore this variable reacts to the developments in the unemployment rate and employment growth. In addition, in this thesis it is impossible that the unemployment rate disappears in the long run, because the variables of the model of Blanchard and Katz (1992) always go to an equilibrium value.
28 References
Armstrong, H. and J. Taylor (2000), Regional Economics and Policy, third edition. Blackwell, Oxford.
Blanchard, O.J. and L.F. Katz (1992), Regional evolutions. Brooking Papers on Economic Activity, 1, pp. 1-75.
Blien, U. and F. den Butter (2003), Labour participation and unemployment: a comparison of developments and institutions in Germany and the Netherlands. WRR Working documents, W133. The Hague.
Broersma, L. and J. van Dijk (2001), Regional labour market dynamics in the Netherlands. Groningen.
Broersma, L. (2008), Sociaal-economische schets Noord-Nederland. Feiten over arbeidsmarkt, innovatie en mobiliteit. Groningen.
Decressin, J. and A. Fatás (1995), Regional labor market dynamics in Europe. European Economic Review 39, pp. 1627-1655
Elhorst, J.P., L. Broersma, A.S. Zeilstra and J. Oosterhaven (2006), Regionale arbeidsdynamiek. Werkloosheid, centraal loonoverleg en infrastructuur. Groningen
Elhorst, J.P. (2008), A spatiotemporal analysis of aggregate labour force behaviour by sex and age across the European Union. Journal of Geographical systems vol. 10, pp. 167-190.
Elmeskov, J. and K. Pichelmann (1993), Unemployment and labour force participation. Economics department working papers, No. 130, pp. 1-55.
European Commission (2000), Employment in Europe 2000. Brussels/Luxembourg.
Euwals, R. and D. van Vuuren (2005), Arbeidsaanbod tot 2050: een beleidsneutraal scenario. CPB Memorandum, Den Haag.
29 Groot, A.J. de and J.P. Elhorst (2009), Labour market effects of flexicurity from a regional perspective. Tijdschrift voor Economische en Sociale Geografie, vol. 101, No. 4, pp. 392-408.
Martin, R. (1997), Regional unemployment disparities and their dynamics. Regional Studies, vol. 31, 237-252.
Overman, H.G. and D. Puga (2002), Unemployment clusters across Europe’s regions and countries. Economics Policy, vol. 17, pp. 117-147.
Raad voor werk en inkomen, (2010), Arbeidsmarkt analyse 2010. July 2010
Schudde, B., J.D. Gardenier, L. Broersma and J. van Dijk (2010), Noordelijke Arbeidsmarkt Verkenning 2010, CAB, Groningen
Sociaal Economische Adviesraad Noord Nederland, SEAN, (2003), Naar een lerende regio in Noord Nederland. Groningen.
UWV Werkbedrijf (2010), Arbeidsmarktprognose 2010-2011. Met een doorkijk naar 2015, Amsterdam 3 juni 2010.
Vermeulen, W. (2005), Regional disparities in a small country? An assessment of the regional dimension to the Dutch labour market on basis of regional unemployment and participation differentials. The Hague.
WRR (1990), Een werkend perspectief; arbeidsparticipatie in de jaren ’90. The Hague, SDU uitgeverij 1990.
30 Appendix
Appendix 1a. Estimation results of the coefficients for the Blanchard and Katz model. Source: Elhorst et al. (2006), p 25.
Endogenous variables Unemployment Labour force participation Employment growth u[-1] 0.763415 -0.360037 -0.023152 p[-1] -0.053578 0.75177 -0,02502 e -0.150609 0.408767 e[-1] -0.025125 -0.023151 -0.000694 constant -0.000558 -0.05183* 0.002406 R-squared 0.997 0.975 0.154 Fixed effects Groningen -0.004797 -0.003677 Friesland -0.005842 -0.004481 Drenthe -0.00488 -0.001481
* is adjusted. Originally value: -0.07177
Appendix 1b. The constant of employment growth rate in the different scenarios
basic model 0.002406
quick recovery 0.011536
double dip 0.011510
new common 0.003627
Notice for Appendix 2-7:
31 Appendix 2: expected results with the coefficients of Elhorst et al. (2006)
Groningen Friesland Drenthe
year u e p year u e p year u e p
32 Appendix 3: expected results for the basic model
Groningen Friesland Drenthe
year u e p year u e p year u e p
33 Appendix 4: expected results for the quick recovery scenario
Groningen Friesland Drenthe
year u e p year u e p year u e p
34 Appendix 5: expected results for the double dip scenario
Groningen Friesland Drenthe
year u e p year u e p year u e p
35 Appendix 6: expected results for the new common scenario
Groningen Friesland Drenthe
year u e p year u e p year u e p
