How can the transfer system affect the working

How can the transfer system affect the working behavior of low skilled men? Evidence from a dynamic structural life-cycle model Preliminary Version Peter Haan∗ , Victoria Prowse† and Arne Uhlendorff‡ February 29, 2008

Abstract In this paper we develop a dynamic structural life-cycle model of labor supply behavior which explicitly accounts for the effects of income taxation and the transfer system. In additional to including a detailed depiction of the tax and transfer system, the model recognizes the demand-side driven rationing risk that might prevent agents from realizing the labor supply state that, according to life-cycle utility maximization, is optimal. We use this framework to study the employment effects of transforming a traditional welfare state, as is currently in place in Germany, towards a more Anglo-American system in which a large proportion of transfers are paid to the working poor. Keywords: Life cycle labor supply, Involuntary unemployment, In-work credits.

∗

DIW Berlin, PSE Paris, [email protected] Department of Economics, University of Oxford, [email protected] ‡ IZA Bonn, [email protected] †

1

Introduction

In this paper we develop a dynamic structural life-cycle model of labor supply behavior which explicitly accounts for the effects of income taxation and the transfer system. In additional to including a detailed depiction of the tax and transfer system, the model recognizes the demand-side driven rationing risk that might prevent agents from realizing the labor supply state that, according to life-cycle utility maximization, is optimal. This allows us to distinguish between voluntary non-employment and involuntary unemployment. This framework is used to study the employment effects of transforming a traditional welfare state, as is currently in place in Germany, towards a more Anglo-American system in which a large proportion of transfers are paid to the working poor. Traditionally, governments have designed transfer systems and income support programs to provide assistance to the poor and thus to guarantee a degree of equity in society. However, over the last two decades several governments have started to use the transfer system in addition as a policy instrument to increase work incentives by subsidizing work, so called in-work credits. The most prominent examples of in-work credits are the Earned Income Tax Credit (EITC) in the US and the Working Tax Credit (WTC) in the UK. The idea of these programs, often referred to as “Making Work Pay” policies, is to target working lowincome households with an income supplement that is contingent on work. In today’s political discussion, in-work credits are seen as an important method of increasing work incentives for groups with high rates of non-employment. A large empirical literature has evaluated the effects of the EITC and the WTC on labor market behavior (for comprehensive surveys, see Hotz and Scholz (2003) and Blundell (2000)). These studies are either based on ex-post evaluation methods exploiting quasinatural experiments or use semi-structural estimation techniques to evaluate policy reforms from an ex-ante perspective (e.g. Blundell, Duncan, McCrae, and Meghir (2000) and Haan

1

and Uhlendorff (2007)). So far, most of the literature has concentrated on the effects of in-work credits on the labor market outcomes of women. The focus on women, and lone mothers in particular, has arisen in part because in-work credits were initially targeted at this group. Moreover, as well documented in the literature, e.g. Blundell and MaCurdy (1999), the labor supply responses of women tend to be much larger than the behavioral changes of men. Evidence on the labor market effects of in-work credits on men is therefore still scarce. However, given the recent political debate surrounding the shifting of resources towards “Making Work Pay” policies, we think it is important to draw attention to the labor market effects of in-work credits on the behavior of men. Thus, in this paper, we study the effect of in-work credits on the working behavior of men with a high probability of not working, namely men with little or no education. We analyze the working behavior of men using a dynamic structural life-cycle model of labor supply which accounts in detail for the effects of the tax and transfer system and distinguishes between voluntary and involuntary unemployment. This rather complex modeling strategy has two distinct advantages over more traditional approaches. First, observed labor supply choices result from solving an intertemporal decision problem featuring human capital accumulation, intertemporally non-separable preferences and an intertemporal budget constraint. Therefore, it is desirable to work with a life-cycle model in which the individual accounts for the effect of his current behavior on his future utility and working behavior. Second, both demand side driven involuntary unemployment and preference based voluntary non-work are relatively common among low skilled men, the population of interest in this study. Hence, a realistic model of the labor supply behavior of the low skilled must differentiate between these two reasons for non-work. The model we propose builds on a large body of literature analyzing labor supply behavior over the life-cycle. Blundell, MaCurdy, and Meghir (2008) classify the life-cycle labor supply

2

literature in two streams according to the channel through which dynamic effects enter the model. The first class of models account for saving and consumption and thus introduce dynamic effects through the intertemporal budget constraint. Preferences, however, are assumed to be separable over time. This literature goes back to Heckman and MaCurdy (1980) and MaCurdy (1981). The resulting theoretical model predicts that individuals will reduce labor supply early and late in the life-cycle while using the savings channel to maintain a constat marginal utility of consumption. Several studies have used this approach to estimate the labor supply effects of tax reforms over the life-cycle. One example is Ziliak and Kniesner (1999) who model the effects of progressive income taxation on life-cycle labor supply. The authors use their dynamic model and analyze income tax reforms occurring in the US during the 1980s and find larger positive labor supply effects than those found in evaluations based on static labor supply models. In the second class of life-cycle labor supply models, to which our approach belongs, the dynamics of labor supply enter via the dependence of current preferences, prices or constraints on previous labor supply behavior. Models in this category allow the current employment decision to affect future labor supply behavior due to habit formation or through effects on future budget constraints due to human capital accumulation or the dependence of benefit entitlement on the individual’s working history. These models therefore capture intertemporal dependencies directly by conditioning current labor supply incentives on past labor supply behavior. Dynamic labor supply models of this form are part of the large literature on dynamic programming which was initiated by the contributions of Wolpin (1984), Pakes (1986) and Rust (1987). To the best of our knowledge, the first study to use dynamic programming to estimate a life-cycle labor supply model was Eckstein and Wolpin (1989) who focused on the labor force participation of married women. The key feature of their model specification is that

3

accumulated experience is endogenous in the wage process and thus the labor supply decision affects future wages. Keane and Wolpin (1997) extended this model by treating both schooling and experience an endogenous choice variables. The authors also allow a very general specification of unobserved heterogeneity and stochastic effects which vary according to labor market choices. This methodology has strongly influenced the following literature and has been the reference model for numerous studies of life-cycle labor supply including Adda, Dustmann, Meghir, and Robin (2006), Berkovec and Stern (1991), Eckstein and Wolpin (1999) and Klaauw (1996). Belzil (2007) provides an extensive survey of this literature and reviews the methodological advances and developments. We build on some aspects of the contributions of Eckstein and Wolpin (1989) and Keane and Wolpin (1997) and address two central issues which we believe have not previously been included in a life-cycle model of labor supply. First, we model demand-side rationing of the labor supply choice. Blundell, Ham, and Meghir (1987), Bingley and Walker (1997) and Ham (1982). In general, in dynamic discrete choice models agents choose their current actions to maximize the discounted expected value of their lifetime utility. In our framework, we additionally allow for rationing risk that prevents agents from realizing their optimal labor supply choice and hence results in involuntary unemployment.

1

The second central issue addressed in this paper concerns the effects of the tax and transfer system on life-cycle employment. In standard life-cycle models, the rewards to work are taken to be the gross wage. Thus these models capture neither progressive income taxation nor governmental transfers. Given the importance of the tax and transfer system in all developed countries, we argue that a detailed depiction of the tax and transfer system is necessary to describe choice specific rewards and thus to accurately capture work incentives. Rust and Phelan (1997), Blau and Gilleskie (2006) and Heyma (2004) argue in the same way 1

For similar approaches in the context of static labor supply models see e.g.

4

when analyzing the effect of social security on retirement behavior, while Yamada (2007) includes progressive income taxation when analyzing the life-cycle employment behavior of Japanese women. However, all of these paper model only selected parts of the transfer system. For example, given their application, Rust and Phelan (1997), Blau and Gilleskie (2006) and Heyma (2004) focus only on specific programs for the elderly while not implementing income taxation or transfer programs relevant to the whole population, while Yamada (2007) abstracts from many of the details of the Japanese taxation system. In contrast, in this paper we argue that, the for the purpose of evaluating of welfare reforms, it is necessary to model in detail the whole tax and transfer system. Indeed, due to means testing, all parts of the tax and transfer system are linked and interact. Consequently evaluating the effect of a change to one aspect of the tax system requires the entire tax system to be modeled. In order to obtain the precise work incentives provided by the tax and transfer system we draw on a detailed tax microsimulation model. We use the above described model to evaluate the the life-cycle employment effects of introducing work-contingent transfer programs in Germany. The empirical analysis is based on panel data from the German Socio Economic Panel (SOEP) for the years 2000 - 2006. In the empirical analysis we focus on low- and no-educated men of working age (25-59 years), a group exhibiting high levels of both non-employment and involuntary unemployment. An analysis of the effects of welfare reforms on this group of Germany men is interesting for several reasons. The current German tax and transfer system can be characterized as a traditional welfare state in which the central purpose of the transfer system is to provide support for the non-working poor. However, due partly to the pressure of relatively high non-employment rates, there exists an ongoing debate about reforming the welfare state towards the AngloAmerican system which emphasizes support for the working poor. Moreover, since it is the low- and no-educated are most dependent on the transfer system much of the political debate

5

has been concerned with policies affecting this group of individuals. While recent reforms have somewhat improved labor supply incentives for this group of individuals, the general design of the transfer system was hardly affected. Thus, in the political debate, there is still demand for the introduction of in-work credits of the form currently in place in the US and the UK.

1.1

An Overview of the Model

This section describes a discrete dynamic life-cycle model of individual labor supply in the presence of labor market constraints which might prevent an individual from realizing his desired hours of work. Utilities are a function of labor market state specific net incomes and thus the model explicitly accounts for the effects of the tax and transfer system on work incentives. Individuals are assumed to be rational forward-looking agents implying that every year each man acts so as to maximize his discounted expected lifetime utility. Moreover, when studying the male labor supply behavior, we simplify the utility maximization problem of the household to the individual decision process of the man and treat the working behavior and fertility of the female spouse, if present, as exogenous with respect to the man’s current behavior. We focus on men of prime working age, defined as 25-59 years. By excluding men aged under 25 years we avoid the complexities of modeling educational choices (Keane and Wolpin, 1997). The sample is further restricted to men with low levels of educational attainment. The low educated are particularly threatened by involuntary unemployment over the whole life-cycle and this group is, in general, characterized by relatively low labor market attachment. The model proceeds as follows. At ages t = 25, ..., 59 years individual i may seek a job with overtime hours (o), defined as 44 weekly working hours, full-time hours (f ), defined as 38.5 weekly working hours, or may choose to be non-employed (n). Hence the individual’s

6

preferred labor market state j ∗ ∈ {o, f, n}. This discrete distribution of hours is motivated by the empirical distribution of working hours which is discussed in Section 2. Following Blundell, Ham, and Meghir (1987), we distinguish preference based non-employment from demand side driven involuntary unemployment (u). In our model, an individual becomes involuntarily unemployed if, irrespective of his previous employment behavior, he is unable to find or keep a job with his preferred hours of work. This definition of involuntary unemployment is consistent with several sources of involuntary unemployment including frictional unemployment, minimum wage legislation or unionized wage setting. Having recognized the possibility of demand side rationing, the individual’s observed labor market state is denoted by j ∈ {o, f, n, u}. Individual i’s probability of being unrationed and thus obtaining or keeping a job with his preferred hours of work is given by Γi,t . The probability of rationing depends on individual and household specific characteristics, regional labor market indicators and the individual’s previous labor market state. Furthermore, for men previously not working the probability of being unrationed is a function of the job arrival rate while for men in previously in employment the lay-off rate affects the likelihood of rationing. In our framework is not possible to distinguish the between the job arrival rate and the separation rate. However, in the empirical specification, we attempt to capture variation in job arrival and separation rates by allowing the effects of regional labor market indicators to be different for the those working over-time, those holding full-time jobs, the involuntary unemployed and the non-employed. In each labor market state j = o, f, n, u individuals receive a flow utility Ui,j,t which is a function of leisure and net income in state j, the individual’s demographic characteristics, including household structure variables, and their previous labor market state. These utilities determine current desired labor supply and consequently job search behavior. Incomes out of work are determined by non-labor income and the transfer system. Income in over-

7

time and full-time jobs is derived from individual’s gross market wage, the hours of work associated with over-time and full-time jobs and the tax and transfer system. Through the gross wage, the distribution of in-work incomes is conditional on individual characteristics that affect wages. In our specification, consumption is assumed to equal current net income. As stated by Blundell, MaCurdy, and Meghir (2008) dynamic programming models of labor supply largely ignore households’ saving and borrowing decisions. Rust and Phelan (1997) discuss this assumption in some detail and provide arguments in favor of equating saving with consumption, the main justification being the lack of reliable information on consumption, savings and assets in longitudinal data. Moreover, as we employ a sample of low educated men ignoring the saving decision is less severe than in many other applications. The individual’s decision problem can be expressed in terms of the value function V (si,t , Yi,t−1 ) which equals the discounted expected value of the individual’s utility from time t onwards assuming that in each year from time t onwards the individual makes his job search decision so as to maximize the discounted expected value of his future utility. The value function depends on the individual’s previous employment state, Yi,t−1 = (Yi,o,t−1 , Yi,f,t−1 , Yi,n,t−1 , Yi,u,t ) where Yi,j,t for j = o, f, n, u are indicators of the agent i being in employment state j at time t, and the state variables si,t which consist of all other variables entering the contemporaneous utilities and probability of successful job search at time t. The individual is assumed to know the current value of si,t but, at time t, may not know the values of all or some elements of si,t+1 . However, the distribution of si,t+1 is known to the individual at time t and is assumed to depend only on si,t and Yi,t . The value function for this problem takes the following form  o u  Γi,t Vi,t (si,t , Yi,t−1 ) + (1 − Γi,t )Vi,t (si,t , Yi,t−1 )   f u (s , Y Vi,t (si,t , Yi,t−1 ) = max   Γi,t Vi,t (si,t , Yi,t−1 ) + (1 − Γi,t )Vi,t i,t i,t−1 )   n (s , Y Vi,t i,t i,t−1 )

    ,   

(1)

j where Vi,t (si,t , Yi,t−1 ) for j = o, f, n, u are employment state specific value functions with the

8

following recursive structure o Vi,t (si,t , Yi,t−1 ) = Ui,o,t + δEt [Vi,t+1 |si,t , Yi,t = (1, 0, 0, 0)],

(2a)

f Vi,t (si,t , Yi,t−1 ) = Ui,f,t + δEt [Vi,t+1 |si,t , Yi,t = (0, 1, 0, 0)],

(2b)

n Vi,t (si,t , Yi,t−1 ) = Ui,n,t + δEt [Vi,t+1 |si,t , Yi,t = (0, 0, 1, 0)],

(2c)

u Vi,t (si,t , Yi,t−1 ) = Ui,u,t + δEt [Vi,t+1 |si,t , Yi,t = (0, 0, 0, 1)].

(2d)

In the above δ is the discount factor which is assumed to be 0.95. Given these definitions, the first and second arguments of the right hand side of equation (1) represent the individual’s discounted expected lifetime utility if he seeks a job with respectively over-time hours or full-time hours at time t and from time t + 1 onwards makes optimal labor supply decisions. Likewise, the last agrement of the right hand side of equation (1) is the man’s discounted expected lifetime utility if he chooses to be non-employed today and from time t + 1 onwards makes optimal job search decisions. Equations (1) and (2a)-(2d) implicitly define the individual’s optimal job search decision at each time t = 25, ..., 59. For the purpose of subsequent analysis the individual’s decision problem is restated in terms of the two following quantities ∆o,f

f o = Vi,t (si,t , Yi,t−1 ) − Vi,t (si,t , Yi,t−1 ),

o ∆o,n = Vi,t (si,t , Yi,t−1 ) −

n (s , Y Vi,t 1 − Γi,t u i,t i,t−1 ) + Vi,t (si,t , Yi,t−1 ). Γi,t Γi,t

(3a) (3b)

An individual will search for a job with over-time hours at time t if and only if ∆o,f ≥ 0 and ∆o,n ≥ 0. Similarly, the individual will search for a job with full-time hours at time t if and only if ∆o,f < 0 and ∆o,n − ∆o,f ≥ 0, and will choose to be non-employed at time t if and only if ∆o,n − ∆o,f < 0 and ∆o,n < 0.

1.2

Discussion of the Model

Although only four labor market states are distinguished, the model is sufficiently general so as to allow an analysis of labor supply behavior at both the extensive (participation decision) 9

and intensive (working hours decision) margins. Moreover, this model extends the previous literature in two important respects. First, the possibility of involuntary unemployment is recognized and the rationing process is modeled jointly with the discrete choice model of labor supply. We emphasize the importance of the demand side constraints as the population of interest, namely low skilled men, is characterized by a high risk of involuntary unemployment. Second, in contrast to Eckstein and Wolpin (1989) and the literature thereafter, we make the labor supply decision dependent on net household income rather than on gross earnings. Hence, we account in detail for income taxation and the transfer system. This is central to understanding how government policy influences labor supply behavior and creates inefficiencies in the market by distorting prices, and how policies can be designed to make work pay. When modeling the necessary details of tax and transfer system we apply a microsimulation model which requires not only information about the individual but also information concerning the household, including the number of children, marital status and the working behavior of the partner. This detailed modeling at the household level is necessary for our research questions because in nearly all developed countries most transfer schemes, as well as income taxation in several countries, depend on household level demographic variables. These extensions, however, lead to several caveats of our modeling approach. Most importantly, we cannot estimate earnings and labor supply behavior jointly as in Eckstein and Wolpin (1989). This is because the microsimulation is too involved to be included in the overall dynamic likelihood estimation. Specifically, incorporating the microsimulation model into the dynamic programming problem implies a number of state variables that is computationally prohibitive. Instead we will develop a multi-step estimation procedure, discussed below, which is similar to the two-step estimation method used by Rust and Phelan (1997).2 2

Yamada (2007) follows a different approach which highlights the trade-off between the level of detail

included when modeling that tax and transfer system and the estimation procedure. He models only selected features of the taxation system and working with this relatively simple structure it is possible to jointly estimate equations describing earnings and labor supply.

10

A further limitation of our approach concerns the data used for the analysis. The required information on the household level demographics and sources of non-labor income prevents us from drawing on the administrative data for Germany which has been used by Adda, Dustmann, Meghir, and Robin (2006). Instead, we use panel data from the Socio Economic Panel Study (SOEP) which include the required family and income information. However, the structure of the SOEP is such that individuals are observed only in certain years in their working life. Therefore, as described below, the approach of Heckman (1981b) is used to control for selection effects in the initial observations.

1.3

Empirical Specification

For the purpose of the empirical analysis, the probability of being not rationed at time t is given by Γi,t = Λ(ηzi,t +ηo wi,t Yi,o,t−1 +ηf wi,t Yi,f,t−1 +ηn wi,t Yi,n,t−1 +ηu wi,t Yi,u,t−1 +λYi,t−1 +ci,s ), (4) where Λ denotes the logistic distribution function. The probability of being unrationed is conditioned on observed individual characteristics zi,t and measures of local labor market conditions wi , t. We allow for different effects of local labor market conditions on the probability of being rationed depending on the individual’s previous labor market state. ci,s represents unobserved time-invariant individual specific random effects with a distribution described below.

3

The following specification of the contemporaneous utility functions is adopted Ui,j,t = γj Yi,t−1 + θj mi,j,t Yi,t−1 + βj xi,t + µj + ci,j + εi,j,t for j = o, f, n, u.

(5)

The first term in the above represents the effect of the individual’s previous employment state on his current flow utility which is unrelated to net income. The second term in the 3

Potentially, mobility between the different regions might cause an endogeneity problem when estimating

the rationing risk. However, over the observed period, only 135 of the 2437 households moved between different regions and only 12 of the movers changed their employment status when moving. Thus, mobility should not cause any inconsistency in the results.

11

above expression denotes the effect of the individual’s income CRRA in state j, mi,j,t , on the individual’s state specific flow utility at time t. The effect of income is allowed to vary according to the individual’s previous employment state reflecting, for example, a higher marginal utility of income among individual’s current in employment then among individual previous out of work. The third term in Equation (5) reflects the effect of individual and household characteristics, xi,t , on state specific flow utilities at time t. The alternative specific coefficients on individual characteristics, previous employment state and income allow the effects of these variables on state specific flow utilities to vary according to the chosen employment state. This reflects the interaction of leisure with these variables. This is more flexible than the alterative method of interacting an arbitrary function of leisure all of the explanatory variables and then imposing a common coefficient on the interacted variables across employment states. The random effects ci,j for j = o, f, n, u allow individuals to have systematic differences in the unobserved components of their flow utilities, and are necessary to establish the extent to which persistence in labor market outcomes is due to the effect of previous employment outcomes rather than persistent unobserved individual characteristics (see Heckman (1981a), Hyslop (1999)). The last component of the flow utilities is the preference shock εi,j,t . This term captures the time-varying component of the individual’s unobserved preferences. To facilitate subsequent derivations, the following definition for j = o, f, n, u is made

qi,j,t = γj Yi,t−1 + θj mi,j,t Yi,t−1 + βj xi,t + µj + ci,j + δEt [Vi,t+1 |si,t , Yi,t ].

(6)

For the final period T, ET [Vi,T +1 |si,T , Yi,t ] = 0, so the future expectation for j = o, f, n, u are zero. Denote the probability of the individual realizing his desired employment state j∗ = o, f, n at time t, conditional on si,t , Yi,t−1 and ci by Ωi,j∗,t (si,t , Yi,t−1 , ci ). Combining Equations (3a)- (3b) with the assumption that εi,j,t are independent over time, individuals 12

and employment states and to have a type I extreme value distribution and normalizing εi,u,t = εi,n,t , the probabilities for each desired employment state can be expressed as the following multinomial logit probabilities.

  Ωi,o,t (si,t , Yi,t−1 , ci ) = Pr  

 o ≥ Vi,t o Vi,t

≥

n Vi,t Γi,t

f Vi,t

−

1−Γi,t u Γi,t Vi,t

>

o Vi,t

  Ωi,f,t (si,t , Yi,t−1 , ci ) = Pr  

(7a)

 f Vi,t f Vi,t

≥

n Vi,t Γi,t


40. All numbers are in %. Source: SOEP, wave 2000-2006.

3

Estimation Results [TO BE COMPLETED]

3.1

Performance of the Model

In-Sample Prediction of the Model In addition to the working behavior of the full sample, we simulate the employment effect for four different sub-groups distinguished by East- and West-Germany, and by skill level. We analyze these sub-groups by simulating the life-cycle employment behavior of a a large number of men who at age 25 years are single with no children.17 The men’s wages and the level of unemployment in their local labor markets at age 25 years are taken to be the average values of these variables among the relevant group of sampled men at age 25 years. Over time the men’s characteristics, including marital status and children, and the values of wages and unemployment evolve according to the transition probabilities observed in the data.

[Figures 2 - 5 about here]

We present the population shares of the four working states over the life cycle, and, in addition the overall participation rate which is the sum of full-time and over-time work. The distribution of the employment states over the life-cycle is quite similar for the four sub-populations. Participation rates are relatively high until the age of 45 and after that they decline. The decreasing participation rate for men older 50 is explained by the strong 17

Each simulation is conducted using a sample size of 12000.

26

increase in voluntary non-employment. This is either early retirement or if unemployed the older men are not willing to actively search for a job for their remaining working life and are therefore classified as being voluntary non-employed. Except for the medium skilled west Germans overall participation is slightly inverse U-shaped. The lower participation rates in the early working years is due to involuntary unemployment. In these sub-population the share of involuntary unemployment is highest for young age groups and then monotonically decreasing. For the medium skilled West-Germans involuntary unemployment is constantly low at a rate lower than 5%. The main difference between the subpopulation can be seen in the level of overall participation. Whereas for the low-skilled men in East Germany participation is over the whole working age lower than 70%, medium skilled West-Germans have rates over 90% over a large range of the life cycle. Only after the age of 50 it declines, yet at slower rates than for the former group which faces participation rate lower than 50% in their 50ies. Life-cycle participation rates for the two other groups are similar and are somewhere between the described extremes.

Out-of-Sample Prediction of the Model [TO BE COMPLETED]

3.2

Labor Supply Elasticities

In order to understand and quantify the labor supply behavior over the life-cycle we derive labor supply elasticities. Again, we focus on the four sub-populations defined above. In the dynamic discrete choice model of labor supply is it not possible to analytically calculate labor supply elasticities. Instead we derive elasticities numerically by increasing gross wages by 10%. Based on the structural estimates, the microsimulation model and the estimated future net-household income under the scenario with higher wages we derive the induced life-

27

cycle labor supply. The post-reform expected net-household income is estimated using the same first-order Markov specification as above, however conditioned on the updated statespecific gross-earnings. We assume that the the labor demand restrictions are not affected by the transfer reform. In this sense our analysis is partial since we do not model potential wage and labor demand-effects of the reform. In the following figures we present the elasticities with respect to the participation rate and weekly working hours. [Figures 6 and 7: about here] Elasticities with respect to participation and working hours are very similar in size and have a similar shape over the life cycle. The working hour elasticity is slightly lower since we find that the average probability of over-time work is reduced by wage increase. In other words working household tend to shift on the intensive margin from over-time to full-time. This can be explained by the concavity of the income preferences. In the labor supply estimation we allow for a flexible specification of the concavity of income which is state specific. We find a stronger curvature for over-time than for full time work which explains this phenomenon.18 For all groups we find the same age profile of the elasticities. The relative employment effects markedly increase with age. This is in particular true for the last years of the working life excluding the very final years. As mentioned above, since our model does not include a proper retirement module, the interpretation of the behavior around retirement is impossible. The life-cycle elasticities differ for several reasons. First involuntary unemployment matters. Ceteris paribus, the higher the rationing risk, the lower the realized employment effects of increased work incentives. This effect is particular important for the young East-Germans and low-skilled West Germans. As shown in Figures 2 - 5 , for these groups the rationing risk is highest in the beginning of the life-path and then decreasing with age. Moreover, compositional changes over the life path, such as marriage or children might have an influence 18

Simulations where only the wages in full- or over-time are increased support this finding (Results not

reported).

28

on the responses to work incentives. The strongest influence however have the estimate age effects and the general employment pattern over the life cycle. The latter effect can be seen as a level effect that increases the relative employment effect as overall participation is markedly declining at the end of the work-life. Even in absolute terms the effects of the Employment Bonus are highest around the age of 55 and this is strongly related to the estimated age-profile of the preference structure. The other important reason is the state dependence in working behavior. As mentioned above, we find a significantly positive dependence of the working behavior over time. This implies increasing participation in earlier ages will ceteris paribus lead to higher participation in older ages. Hence, elasticities at the end of the working life can be decomposed in this stated-dependence effect of the early working history and of the incentive effect at the current age.

4

The life cycle Employment Effects of In-Work Transfers: The Employment Bonus

The German welfare system can be characterized as a traditional welfare system with relatively generous out-of-work transfers that are withdrawn at high rates when people start working. In the political discussion this has often been criticized and the low working incentives have been identified as one central reason for the relative high unemployment. This is in particular true for the low skilled as the German transfer system generates the lowest incentives for low-wage group. Drawing on the international experience, mainly from the US and the UK, there is an ongoing debate about changing the German welfare system towards shifting more transfers to the working poor and thus increasing work incentives. One prominent example in the political discussion is the UK Working Tax Credit (WTC). Conditional on a minimal amount of working hours, workers receive a transfer that is means tested against the gross family earnings. Blundell (2000) and more recently Brewer and Browne (2006)

29

discuss the British experience with in-work credits and survey the labor market effects of recent changes in the design of the WTC. Haan and Myck (2007) analyze the hypothetical work incentives and static labor supply effects of the introduction of the WTC into the German system. In line with the British experience they find positive effects for first earners in couples and single households which are counteracted by strong negative effects for the secondary earner. The negative effects are related to the conditioning on family rather than individual earnings. A reform which avoids the negative secondary earner effects is the Employment Bonus, as implemented in Belgium, e.g. see Bargain et al. 2006 . This transfer program is conditioned on individual full-time equivalized earnings and differs therefore from the WTC in several important points. First, it is not conditional on a minimal amount of weekly working hours but increases proportionally with working hours. Second, the entitlement is based on individual rather than on household earnings. That implies that the Employment Bonus avoids the well understood negative secondary earner effects of programs similar to the WTC. Lastly, as the Employment Bonus depends on equivalized earnings, individuals with earnings above a defined threshold, do not receive any transfer, thus this program is strongly targeted at the low-wage people. In the following we focus on the Employment Bonus and analyze the effect on work incentives and the life-cycle labor supply effects when introducing an Employment Bonus. The Bonus is introduced as a subsidy of the social security contributions which are of important size in Germany. The contributions for an employee amount to about 20% of gross earnings. Full subsidy of 140 Euro per months - expressed in full-time equivalent income - is paid up to a full-time equivalent gross earnings of 1,210 Euro/month (about 7 Euro gross hourly wage). Above this threshold, it is phased out at a taper rate of 17.8% and is fully exhausted at a fulltime equivalent income of 2,000 Euro (about 12 Euro gross hourly wage). We introduce the

30

Employment Bonus on top of the legislation of the tax and transfer system as implemented in the year 2005. Therefore, for low working hours and very low wages the Bonus will be means-tested against the out-of work transfers and do therefore not affect the work incentives. To be consistent with the assumption of the life-cycle labor supply model we only introduce the Employment Bonus for men; women face the fiscal system of the year 2005 and therefore they do not adjust their labor supply.

4.1

Work incentives of the Employment Bonus

In order to understand the effects on the work incentives we present budget-lines for stylized households. We focus on a very low-wage (7.5 Euro per hour) and a medium low wage (10 Euro per hour) single men without children.

[Figure 8 about here]

Depending on the housing benefits a single household receives out-of work benefits of nearly 600 Euro per month. Due to the high withdrawal rate of the benefits when starting to work the work incentives are relatively low for low-wage men. For both stylized low-wage households, only when working more than 25 hours the benefits are completely withdrawn. After that the Employment Bonus markedly increases work incentives for both households. Since the Bonus is conditioned on the full-time equivalent, the higher working incentives remain even with high working hours. The dependence on the hourly wages becomes obvious. The household with very low wages receives close the maximum subsidy whereas for the medium wage person parts of the subsidies are withdrawn. With higher hourly wages the subsidy of the Bonus would be completely withdrawn. For couple households the work incentives are very similar and this distinguishes the Employment Bonus from the WTC or the US Earn Income Tax Credit (EITC). For a first earner - a household where the female spouse is not working - household out-of work benefits are

31

higher in particular in a household with children and therefore the effects of the Employment Bonus start to be present only at higher working hours. Other than that the same implications as for the single household are true. For a secondary earner - a household where the female spouse is working full time - the Employment Bonus has a positive working effect even at low working hours as this household is not eligible for out-of work transfers.

4.2

Effects on life cycle employment

Using the same numerical simulation method as for the gross elasticities we derive the lifecycle labor supply effects induced by the Employment Bonus. In the following figures we present the relative change in the participation rate and of the weekly working hours. In general the findings are very similar as for the labor supply elasticities. The pattern of the behavioral response are the same for the Employment Bonus. The differences in the employment rates can be therefore related to the different induce work incentives. Again we find that the relative participation and working hours effects are very similar in size and have a similar shape over the life cycle and that the working hour effect is slightly lower. This is explained by the concavity effect. The employment effects are largest for the sub-group with low wages, namely low educated and households living in East Germany. This is due to the incentive structure of the Employment Bonus which is reduced at higher wages. Most of the medium low-skilled west Germans are not at eligible for the employment bonus and thus their optimal life path remains unchanged.

[Figures 9 and 10 about here]

For all groups we find the same age profile. The relative employment effects markedly increase with age. This is in particular true for the last years of the working life excluding the very final years. As mentioned above, since our model does not include a proper retirement

32

module, the interpretation of the behavior around retirement is impossible. The life-cycle effects of the Employment Bonus differs for several reasons. The induced work incentives should be highest in the beginning of the working life and at the end since we find an inverse U-shaped wage profile. This incentive effect is counteracted or pronounced for other reasons. First involuntary unemployment matters. Ceteris paribus, the higher the rationing risk, the lower the realized employment effects of the Employment Bonus. This effect is particular important for the young East-Germans and low-skilled West Germans. Moreover, compositional changes over the life path, such as marriage or children might have an influence on the responses to work incentives. The strongest influence however have the estimate age effects and the general employment pattern over the life cycle. The latter effect can be seen as a level effect that increases the relative employment effect as overall participation is markedly declining at the end of the work-life. Even in absolute terms the effects of the Employment Bonus are highest around the age of 55 and this is strongly related to the estimated age-profile of the preference structure. The other important reason is the state dependence in working behavior. As mentioned above, we find a significantly positive dependence of the working behavior over time. This implies increasing participation in earlier ages will ceteris paribus lead to higher participation in older ages. Hence, the effects of the Employment Bonus at the end of the working life can be decomposed in this stated-dependence effect of the early working history and of the incentive effect at the current age.

Future Expectations versus Myopia In order to show the necessity of the estimation of a structural life-cycle model with forwardlooking agents we contrast the presented life-cycle effects of the Employment Bonus with the employment effects over the life-cycle derived in a myopic model where agents have no expectations. We focus only on the participation effects of the Employment Bonus since the main findings to not differ for the other subgroups and are similar for the changes in working

33

hours.

[Figure 11]

The employment effects derived in the a model with myopic agents are markedly smaller than those derived in the model with forward-looking households. This holds true for the whole life-cycle. This results underlines the importance of modelling future expectations. In this world, agents are aware that labor supply participation at young ages will positively affect the earnings in future years. Therefor early in the life-cycle agents react stronger to changes in work incentives. This effect carries through the whole working life and is even emphasized by the positive state dependence effects.

5

Conclusion

In this paper we have developed a dynamic structural life-cycle model of labor supply behavior which explicitly accounts for the effects of income taxation and the transfer system. In addition, the model recognizes the demand-side driven rationing risk that might prevent agents from realizing the labor supply state that, according to life-cycle utility maximization, is optimal. The life-cycle model is applied to evaluate the the life-cycle employment effects of introducing work-contingent transfer programs in Germany. The empirical analysis is based on panel data from the German Socio Economic Panel (SOEP) for the years 2000 - 2006. In the empirical analysis we focus on low- and no-educated men of working age (25-59 years), a group exhibiting high levels of both non-employment and involuntary unemployment. We find that an Employment Bonus which subsidies social security contributions conditional on individualized full-time equivalized gross-earnings has a strong positive labor-supply effect in particular towards the end of the working life.

34

References Adda, J., C. Dustmann, C. Meghir, and J.-M. Robin (2006): “Career Progression and Formal versus On-the-Job Training,” IZA Working Paper, 2260. Bargain, O., M. Caliendo, P. Haan, and K. Orsini (2006): “’Making Work Pay’ in a Rationed Labour Market,” IZA Discussion-Paper, 2033. Belzil, C. (2007): “The Returns to Schooling in Structural Dynamic Models: A Survey,” European Economic Review. Berkovec, J., and S. Stern (1991): “Job Exit Behavior of Older Men,” Econometrica, 59(1), 189–210, available at http://ideas.repec.org/a/ecm/emetrp/v59y1991i1p189210.html. Bingley, P., and I. Walker (1997): “The Labour Supply, Unemployment and Participation of Lone Mothers in In-Work Transfer Programmes,” Economic Journal, 107. Blau, D., and D. Gilleskie (2006): “Health insurance and retirement of married couples,” Journal of Applied Econometrics, 21, 935–953. Blundell, R. (2000): “Work incentives and ’in-work’ benefit reforms: a review,” Oxford Review of Economic Policy, 16(1), 27. Blundell, R., A. Duncan, J. McCrae, and C. Meghir (2000): “The Labour Market Impact of the Working Families Tax Credit,” Fiscal Studies, 21(1), 75–104. Blundell, R., J. Ham, and C. Meghir (1987): “Unemployment and female labour supply,” Economic Journal, 97, 44–64. Blundell, R., and T. MaCurdy (1999): “Labor Supply: A Review of Alternative Approaches,” in Handbook of Labor Economics Volume 3A, ed. by O. Ashenfelter, and D. Card, chap. 27, pp. 1559–1695. Amsterdam: North Holland. 35

Blundell, R., T. MaCurdy, and C. Meghir (2008): “Labor Supply Models: Unobserved Heterogeneity, Nonparticipation and Dynamics,” in Handbook of Econometrics VI. Elsevier. Brewer, M., and J. Browne (2006): “The effect of the working families tax credit on labour market participation,” IFS Briefing Notes. Eckstein, Z., and K. Wolpin (1989): “Dynamic Labour Force Participation of Married Women and Endogenous Wage Growth,” Review of Economic Studies, 56, 375–390. Eckstein, Z., and K. Wolpin (1999): “Why Youths Drop Out of High School: The Impact of Preferences, Opportunities, and Abilities,” Econometrica, 67(6), 1295–1339. Haan, P., and M. Myck (2007): “Apply with caution: introducing UK-style in-work support in Germany,” Fiscal Studies, forthcoming. Haan, P., and A. Uhlendorff (2007): “Intertemporal Labor Supply and Involuntary Unemployment,” Institute for the Study of Labor (IZA), (2888). Haisken De-New, J., and J. Frick (2005): Desktop Compendium to The German SocioEconomic Panel Study (SOEP). DIW, Berlin. Ham, J. (1982): “Estimation of a Labour Supply Model with Censoring Due to Unemployment and Underemployment,” The Review of Economic Studies, 49(3), 335–354. Heckman, J. (1981a): “Heterogeneity and State Dependence,” in Studies in Labor Markets, ed. by S. Rosen, pp. 91–139. Chicago Press, Chicago, IL. (1981b): “The Incedental Parameter Problem and the Problem of Initial Conditions in Estimating a Discrete Time-Discrete Data Stochastic Process,” in Structural Analysis of Discrete Data with Econometric Applications, ed. by C. Manski, and D. McFadden, pp. 179–195. MIT Press, Cambridge, MA. 36

Heckman, J., and T. MaCurdy (1980): “A Life-cycle Model of Femal Labor Supply,” Review of Economic Studies, 47, 47–74. Heckman, J., and B. Singer (1984): “A Method for Minimizing the Distributional Assumptions in Econometric Models for Duration Data,” Econometrica, 52, 271–320. Heyma, A. (2004): “A structural dynamic analysis of retirement behaviour in the Netherlands,” Journal of Applied Econometrics, 19(6), 739–759. Hotz, V. J., and J. K. Scholz (2003): “The Earned Income Tax Credit,” in MeansTested Transfer Programs in the United States, ed. by R. A. Moffitt, pp. 141–198. Chicago: University of Chicago Press. Hyslop, D. (1999): “State Dependence, Serial Correlation and Heterogeneity in Intertemporal Labor Force Participation of Married Women,” Econometrica, 67(6), 1255–1294. Keane, M. P., and K. I. Wolpin (1997): “The Career Decisions of Young Men,” Journal of Political Economy, 105(3), 473–522. Klaauw, W. (1996): “Female Labour Supply and Marital Status Decisions: A Life-Cycle Model,” The Review of Economic Studies, 63(2), 199–235. MaCurdy, T. (1981): “An Empirical Model of Labor Supply in a Life Cycle Setting,” Journal of Political Economy, 89, 1059–85. Pakes, A. (1986): “Patents as Options: Some Estimates of the Value of Holding European Patent Stocks,” Econometrica, 45(4), 755–84. Rust, J. (1987): “Optimal Replacement of GMC Bus Engines: An Empirical Model of Harold Zurcher,” Econometrica, 55(5), 999–1033. (1997): “Using Randomization to Break the Curse of Dimensionality,” Econometrica, 65(3), 487–516. 37

Rust, J., and C. Phelan (1997): “How Social Security and Medicare Affect Retirement Behavior in a World of Incomplete Markets,” Econometrica, 65(4), 781–832. Steiner, V., P. Haan, and K. Wrohlich (2005): “Dokumentation des Steuer-TransferMikrosimulationsmodells 1999-2002,” Data Documentation 9. Wolpin, K. (1984): “An Estimable Dynamic Stochastic Model of Fertility and Child Mortality,” The Journal of Political Economy, 92(5), 852–874. Yamada, K. (2007): “Marital and occupational choices of women: a dynamic model of intra-household allocations with human capital accumulation,” Mimeo. Ziliak, J. P., and T. J. Kniesner (1999): “Estimating Life Cycle Labor Supply Tax Effects,” Journal of Political Economy, 107(2), 326–359.

38

Appendix I: The Value Function Substituting equation (6) into equations (1) and adding and subtracting Γ(zi,t+1 , Yi,t , ci,s )εi,n,t+1 from line 3 gives of this expression gives

Vi,t+1 (si,t+1 , Yi,t ) 

= 

  Γ(zi,t+1 , Yi,t , ci,s )(qi,o,t+1 + εi,o,t+1 ) + (1 − Γ(zi,t+1 , Yi,t , ci,s ))(qi,u,t+1 + εi,n,t+1 )     max   Γ(zi,t+1 , Yi,t , ci,s )(qi,f,t+1 + εi,f,t+1 ) + (1 − Γ(zi,t+1 , Yi,t , ci,s ))(qi,u,t+1 + εi,n,t+1 )      qi,n,t+1 + Γ(zi,t+1 , Yi,t , ci,s )εi,n,t+1 + (1 − Γ(zi,t+1 , Yi,t , ci,s ))εi,n,t+1

      .       (22)

The above can be separated as follows

Vi,t+1 (si,t+1 , Yi,t ) = 



  Γ(zi,t+1 , Yi,t , ci,s )(qi,o,t+1 + εi,o,t+1 ) + (1 − Γ(zi,t+1 , Yi,t , ci,s ))qi,u,t+1     max   Γ(zi,t+1 , Yi,t , ci,s )(qi,f,t+1 + εi,f,t+1 ) + (1 − Γ(zi,t+1 , Yi,t , ci,s ))qi,u,t+1      qi,n,t+1 + Γ(zi,t+1 , Yi,t , ci,s )εi,n,t+1 

            



  (1 − Γ(zi,t+1 , Yi,t , ci,s ))εi,n,t+1     + max   (1 − Γ(zi,t+1 , Yi,t , ci,s ))εi,n,t+1      (1 − Γ(zi,t+1 , Yi,t , ci,s ))εi,n,t+1

39

            

(23)



   Γ(zi,t+1 , Yi,t , ci,s )(qi,o,t+1 + εi,o,t+1 ) + (1 − Γ(zi,t+1 , Yi,t , ci,s ))qi,u,t+1     = max   Γ(zi,t+1 , Yi,t , ci,s )(qi,f,t+1 + εi,f,t+1 ) + (1 − Γ(zi,t+1 , Yi,t , ci,s ))qi,u,t+1      qi,n,t+1 + Γ(zi,t+1 , Yi,t , ci,s )εi,n,t+1

            

+(1 − Γ(zi,t+1 , Yi,t , ci,s ))εi,n,t+1 .

(24)

Factoring Γ(zi,t+1 , Yi,t , ci,s ) from the first part of Equation (24) yields 



 (1−Γ(zi,t+1 ,Yi,t ,ci,s ))  qi,o,t+1 + εi,o,t+1 + Γ(zi,t+1 ,Yi,t ,ci,s ) qi,u,t+1     Vi,t+1 (si,t+1 , Yi,t ) = Γ(zi,t+1 , Yi,t , ci,s ) max   qi,f,t+1 + εi,f,t+1 + (1−Γ(zi,t+1 ,Yi,t ,ci,s )) qi,u,t+1 Γ(zi,t+1 ,Yi,t ,ci,s )      qi,n,t+1 Γ(zi,t+1 ,Yi,t ,ci,s ) + εi,n,t+1

            

+(1 − Γ(zi,t+1 , Yi,t , ci,s ))εi,n,t+1 .(25) ud Taking the expectation of both sides of Equation (25) conditional on (spi,t+1 , suc i,t+1 , si,t+1 , Yi,t )

yields Equation (15).

40

0

.05

Density

.1

.15

Figure 1: Working behavior of men

0

20

40

60

hours

Notes: Weekly working hours are reported contractual hours plus reported paid over-time. Men are aged between 25 and 59. The distribution is censored at 60 hours per week which excludes about 2% of the relevant population. Source: SOEP, 2000 - 2005.

Figures

41

Figure 2: Life-cycle employment behavior: Low education East Germany

80

Over-time Full-time 70

Voluntary Non-employment Involuntary Unemplyoment Participation

60

in %

50

40

30

20

10

0 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59

Age

Notes: Life-cycle employment behavior is simulated for men aged 25-30.

Figure 3: Life-cycle employment behavior: Medium education East Germany

90

Over-time Full-time

80

Voluntary Non-employment Involuntary Unemplyoment

70

Participation 60

in %

50

40

30

20

10

0 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59

Age

Notes: Life-cycle employment behavior is simulated for men aged 25-30.

42

Figure 4: Life-cycle employment behavior: Low education West Germany

90

80

Over-time 70

Full-time Voluntary Non-employment

60

Involuntary Unemplyoment Participation

in %

50

40

30

20

10

0 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59

Age

Notes: Life-cycle employment behavior is simulated for men aged 25-30.

Figure 5: Life-cycle employment behavior: Medium education West Germany

100

90

Over-time 80

Full-time Voluntary Non-employment

70

in %

60

Involuntary Unemplyoment Participation

50

40

30

20

10

0 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59

Age

Notes: Life-cycle employment behavior is simulated for men aged 25-30.

43

Figure 6: Life-cycle gross-wage elasticities: Participation

0.5 East Germany Low Education

0.45

East Germany Medium Education Wesr Germany Low Education

0.4

West Germany Medium Education 0.35

Elastictiy

0.3 0.25 0.2

0.15 0.1

0.05 0 26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

Age

Notes: Gross wage elasticities are numerically derived by increasing gross hourly wages by 10%.

Figure 7: Life-cycle gross-wage elasticities: Working Hours

0.5 0.45

East Germany Low Education East Germany Medium Education

0.4

Wesr Germany Low Education West Germany Medium Education

0.35

Elasticity

0.3 0.25 0.2 0.15 0.1 0.05 0 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 Age

Notes: Gross wage elasticities are numerically derived by increasing gross hourly wages by 10%.

44

Figure 8: Budget constraint single man

Budget constraint single household without children 1400

1300

Monthly Net Household Income

1200

1100

1000

900

800

700

Status Quo - Hourly wage 7.5 Euro Employment Bonus - Hourly wage 7.5 Euro

600

Status Quo - Hourly wage 10 Euro Employment Bonus - Hourly wage 10 Euro

500

400 0

5

10

15

20

25

30

35

40

45

weekly working hours

Notes: Budget line for a stylized single man without children. Source: STSM.

Figure 9: Employment Bonus - Relative increase in participation

3.5

3

2.5

East Germany Low Education East Germany Medium Education Wesr Germany Low Education West Germany Medium Education

in %

2

1.5

1

0.5

0 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59

Age

Notes: Relative increase in participation rate.

45

Figure 10: Employment Bonus - Relative increase in working hours

3 East Germany Low Education East Germany Medium Education

2.5

Wesr Germany Low Education West Germany Medium Education

in %

2

1.5

1

0.5

0 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 Age

Notes: Relative increase in weekly working hours.

Figure 11: Expectations versus myopia- Participation effects for low skilled East Germans

3.5

3

Future Expectation Myopic

2.5

in %

2

1.5

1

0.5

0 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59

Age

Notes: Relative participation effects.

46