Financial incentives to work in the context of a complex reform package and growing wages: the Polish experience 2005–2011

We examine the consequences of changes in work incentives in Poland between 2005 and 2011 resulting from a complex tax and benefit reform package and substantial real wage growth. While marginal and participation tax rates (PTRs) in the majority of analysed cases fall as a result of the introduced reforms, the conclusions from looking at replacement rates (RRs) for the population eligible for means tested benefits are generally different. These suggest that despite significant tax giveaways incentives on the labour market weakened for families with children and for those eligible to safety net benefits. Yet despite these negative policy implications, we show that work incentives improved substantially over the period due to significant real wage growth. When analysing the effect of the reforms on financial incentives to work, we often find conflicting conclusions when using the PTRs and RRs to reflect financial attractiveness of employment. This is not necessarily surprising, but serves as a note of caution on the use of each of them independently.


Introduction
In the process of adjustment of tax and benefit policies governments continuously face the tradeoff between equity and efficiency with attempts to balance out the incentive effects of redistribution and poverty alleviation (see Blundell, 2001). Since reforms in different segments of the tax and benefit system are generally implemented separately and often without much coordination between the different components, very frequently changes in taxes are assessed in separation from benefit reforms. In the first case attention focuses on the consequences of tax policies for labour market incentives, while in the latter on the implications of benefit reforms for incomes of poorest households. The issues are further blurred if substantial reforms in different parts of the system accumulate over a number of years. Understanding the consequences of a sequence of reforms then requires a more complex approach to the entire reform package (see, e.g. Clark & Leicester, 2004;or Clark, Dilnot, Goodman, & Myck, 2002).
In this paper we consider the labour market consequences of the entire reform package introduced in Poland during the two terms of parliament from 2005 to 2011 and present its effects on tax rates and replacement rates (RRs). The analysed package of reforms significantly changed the structure of taxes on labour and means-tested benefits with the combined effect of the reforms on household incomes in the range of 1.7-2.2% of the GDP (Domitrz, Morawski, Myck, & Semeniuk, 2013). Among the principal reforms were the introduction of a very generous child tax credit (CTC) in 2007, reduction of the disability rate in social security insurance contributions (2007 and 2008), reduction of the number of tax rates in personal income tax (2009), and several changes implemented to the system of family and safety net benefits. The latter generally implied that more generous benefits have been targeted to a much smaller number of families and increasingly focused on those on lowest incomes.
The reforms took place at the time of rapid economic growth which was reflected in growing levels of employment and real wages. This provides for an interesting context for the analysis of the role of reforms and the broader economic environment on changes in financial incentives on the labour market. Poland, as one of few countries in the European Union managed to avoid falling into recession in the years we analyse. The cumulative GDP growth between the first quarter of 2005 and first quarter of 2012 has been as high as 32.8%. Even during the crisis years, between the first quarters of 2008 and 2011 the GDP grew by 6.7%.
This has been associated with strong dynamics of internal consumption, resulting from growing levels of disposable incomes associated with higher employment (see Figure A1), wages as well as changes in the tax and benefit system. 1 While dynamics of wages slowed down at the end of the period, wages continued growing in real terms. Employment levels among those aged 25-34 suffered after 2008, but among other groups either remained stable or increased further. In particular, employment rates of both men and women aged 45-54 and 55-59 grew after 2008. 2 What is striking is that the dynamics of employment rates after 2008 do not match the observed increases in unemployment (see Table A1). The combination of employment and unemployment rates suggest a virtually uninterrupted growth in participation among men and women of all age groups, a finding that points to the important role of the supply side on the labour market. In this context it is important to understand the role of both wages and the tax and benefit policies. 3 Between the beginning of 2005 and 2012 the mean wage increased in real terms by 19.7%, while the statutory minimum wage grew by 43.0%. 4 At the same time reductions in the level of taxation of earnings significantly affected the relationship between gross and net wages. As we show in this paper marginal tax rates and participation tax rates (PTRs) fell for almost all types of families and earnings ranges. However, RRs computed for the same scenarios often suggest a different pattern of changes to labour market incentives. If individuals react to the level of disposable income in and out of work in making their labour market decisions, this suggests weakening of incentives at the extensive margin for a substantial number of households despite falling tax rates. Only if we add real wage growth on top of the implemented reforms does the pattern reflected in changes to the RRs become consistent with the observed changes in participation. Our analysis thus points to the decisive role of the demand factors as determinants of the observed participation patterns in the analysed time frame.
The context of the Polish reforms and the changing economic environment provides an interesting case for the comparison of conclusions concerning financial attractiveness of work using different approaches to the measurement of financial incentives, that is, tax rates and RRs. As Adam and Browne (2010) point out 'the RR measures the strength of work incentives whereas the PTR measures the effect of the tax and benefit system on work incentives'. The two measures thus differ by concept and construction. What we demonstrate in the paper is that they may suggest conflicting conclusions as far as the effect of policy is concerned and thus should not be treated interchangeably.
The rest of the paper is organised as follows. We first present a description of changes in social security contributions, personal income tax and benefits regulations that were implemented in Poland between 2005 and 2011 (Section 2). This is followed by the description of the consequences of these reforms on family incomes and the approach taken to measuring labour market financial incentives (Section 3). In Section 4 we evaluate the effect of the reforms on the magnitude of changes in financial incentives to work at the intensive and extensive margins, and compare them to the changes in financial attractiveness of work resulting from real wage growth. Section 5 concludes the paper. Poland: 2005-2011 2.1. The system and reforms: social insurance and income taxation Employment income taxation in Poland consists of three components: social security contributions (SSC), personal income tax and health insurance contributions (Tables A3-A4).

Main tax and benefit reforms in
Nominally the contributions to social security are divided between the employee and the employer. Contributions to retirement pensions are paid at 19.52% split equally between the employee and the employer. Similarly, up to 2007 the rates of disability insurance used to be equally divided into employee and employer contributions and amounted to 13.0%. As we show in Table A3 these rates of contributions have been dramatically reduced in 2007 and 2008. On the employee side the disability contribution rate was lowered from 6.5% to 1.5%, while on the employer's side the rate was reduced from 6.5% to 4.5%. In total therefore the disability rate was reduced from 13% to 6%. Except for sickness and accident insurance (respectively at 2.45% and 1.93%) all contributions are paid only up to an upper threshold which is equivalent to 30 times the expected average monthly gross wage. The value of this threshold has been regularly increased in real terms reflecting the growth of earnings (see Table A3).
The gross wage net of SSCs constitutes the tax base for the computation of personal income tax. Every employee is eligible to claim a revenue credit and each tax payer is granted a universal tax credit. Until the end of 2008 the Polish system had three tax rates -19%, 30% and 40%. This changed as of January 2009 with an introduction of a two rate system at 18% and 32% with the second rate applicable from annual taxable income of 85,526 PLN (21,435 euro). 5 Over the period from 2005 to 2011 both the income tax thresholds as well as revenue costs and the universal tax credit have remained unchanged in nominal terms (see Table A4). The Polish tax system operates as a joint system for married couples and lone parents, and as of 2007 includes a generous CTC for every dependent child which has become one of the major tax deductions in the Polish tax system  annually costing the government approximately 6.0bn PLN (0.5% of the GDP). Finally the health insurance contribution in 2005 was 8.5% of gross earnings net of social security contributions, and was raised in two steps to 9.0% in 2007. Of this 7.75pp is deductible from income tax.

The system and reforms: social transfers
The system of social transfers in Poland consists of three major types of financial means-tested transfers: family benefits (FB), housing benefit and social assistance. Universal transfers are paid to disabled children (nursing allowance), to the elderly aged over 75 (nursing supplement or nursing allowance), and from January 2006 as child birth benefits.
In 2005 5.2 mln children received the family allowancethe main element of FB, which are by far the most significant means-tested benefit system in Poland. Family benefits are paid to families with children conditional on their net income being below a specified threshold, the value of which was frozen in nominal terms at 504.00 PLN (126.30 euro) per month per person since 2004 with a higher level applicable to families with disabled children (583.00 PLN, 146.10 euro). Family benefits consist of the family allowance on top of which there exist several supplements for such groups of families as lone parents or families with more than two children. Up until September 2006 the value of the family allowance was differentiated on the basis of the number of children in the family. The system introduced in 2006 instead differentiates benefit values by the age of children. In September 2006, and again three years later the rates of benefits were substantially increased. As a result, the average family allowance increased in real terms by nearly 60% between 2005 and 2011. However, despite these increases the overall spending on FB was reduced due to the falling number of eligible families. This resulted from improved financial situation of many families, but was partly a consequence of freezing of the eligibility threshold the value of which fell in real terms by 20%. This element alone reduced the number of children eligible to family allowance by 820,000 (Myck, Morawski, Domitrz, & Chrusciel, 2011), that is, about 18% of the number of children receiving the benefits in 2005. The values of family allowance and its main supplements in 2005 and 2011 are given in Table A5.
The housing benefit and social assistance benefits are the principal means-tested safety net benefits in Poland. In 2005 about 0.76 mln households received housing benefits, and although there have been no major changes in the rules determining its payments, the number of recipients has been regularly falling (down to 0.40 mln in 2011, see Table A7). This may have been partly a result of improved economic situation of the households, although a stricter approach to recipients may be a more important reason behind this. As in the case of FB the eligibility thresholds determining receipt of housing benefits have also been kept at constant nominal level.
Social assistance in Poland consists of permanent and temporary assistance. Permanent benefits are granted to individuals incapable to work because of advanced age or disability (who are not eligible to social security pensions). Temporary social assistance is given in case of the combination of low income and such conditions as long-term illness, disability, unemployment, etc. Permanent social assistance is solely based on income and health or age conditions, while eligibility to temporary assistance includes an informal wealth-test in the form of a subjective assessment of the overall living standard by a representative of the local social assistance office. The amount of social assistance is computed as a difference between the net household income per person and a legislated minimum income, but only a fixed proportion of this difference is guaranteed by the central budget. This proportion was increased in October 2006 (from 30% to 35% for one-person households and from 20% to 25% for 2+ households), and then again in in October 2008 (both up to 50%). The decision to supplement the guaranteed amount lies with the local authorities. On average only a small fraction of the remaining difference is supplemented. Between 2005 and 2011 the legislated minimum income levels were reduced in real terms by 7% for one-person households and by 13% for larger households (see Table A6), while at the same time the increased guaranteed proportion of payments implied real increases of guaranteed amounts. Thus while the values of temporary social assistance between 2005 and 2011 increased in real terms, the number of recipients fell by 32% (see Table A7).

Measuring the effects of policies and wage growth on financial incentives to work
Financial incentives to work in 2005 and 2011 are presented in this paper for a selected number of stylised households. 6 This approach has a number of features which make it suitable for the illustrative purpose of our paper. In particular the analysis can clearly specify particular labour market scenarios for the chosen households and we can single out incentives for the chosen measures of gross wages. 7 In order for the parameters of the two systems to be comparable and to reflect the consequences of changes in prices over the period on the real value of disposable incomes, the elements of the 2005 system expressed in nominal terms have been uprated to 2011 prices using the CPI index with cumulative inflation over the period at 19.6%. As noted among others by Sutherland, Hancock, Hills, and Zantomio (2008b) there may be different approaches to system indexation with different interpretation of system 'neutrality' (see also Sutherland, Evans, Hanckock, Hills, & Zantomio, 2008a). From the point of view of modelling of labour market incentives it seems appropriate, though, to index all elements of the system with CPI since we focus on changes in the real value of the resulting disposable incomes.
In all of these cases we analyse family budget constraints conditional on overall labour earnings but also look at incentives conditional on the hourly wage level by the degree of work intensity. Given the specificity of the Polish benefit system in all four cases, we discuss the effects of reforms on labour market incentives assuming two benefit eligibility scenarios. In one, the families are never eligible to receive social assistance or housing benefit (regardless of their income), while in the other they are eligible to these benefits if their income levels are low enough in the analysed circumstances. 8 These family types constitute a significant proportion of families of working age in Poland. In the Polish Household Budgets' Survey data from 2009 about 18.13% of working age families were couples with children, 36.77% were couples without children, and single adult families with and without children constituted respectively 7.42% and 37.67%. Interestingly, despite generally low employment rates in Poland, the employment rate among adults in couples with two children in 2009 was 78.6%. This points to the importance of looking at the changes in financial incentives for second earners in families with children which we consider in Section 4.4.

Measures of the financial incentives to work
The effect of changes in the tax and benefit rules on families' disposable incomes are analysed from two perspectives. First, to highlight the key aspects of the reform package we present the direct effect of reforms on the values of households' disposable incomes. This is followed by the analysis of the changes in financial incentive to work on the intensive and extensive margins. The effective marginal tax rates (EMTR) are used for the first purpose and the PTRs as well as the RRs for the second. 9 The EMTR measures a fraction of any additional gross wage increase that is taxed away. Formally, this corresponds to the first derivative of the budget constraint at a given point, but given the nonlinearities in the system our results are obtained numerically by comparing the difference in values of disposable incomes resulting from a small change in gross income from work. This takes the form: Y (v) and Y (v + Dv) represent the value of disposable income when total labour cost is equal to (ω) and (v + Dv), respectively. 10 λ stands for the tax and benefit function. The PTR takes into account the difference between net incomes in and out of work measured relative to gross earnings in work, and is defined as follows: . for a single person: where v(w, h) is total labour cost at h hours, and Y (v(w, h), l) and Y (0, l) are total net incomes in and out of work respectively given the tax and benefit function λ. . for a one-earner couple: where similarly v(w 1 , h 1 ) is the total labour cost of the first earner, and Y (v(w 1 , h 1 ), 0, l) and Y (0, 0, l) are total net incomes in and out of work respectively, given the tax and benefit function λ. . for the second earner in couples keeping the earnings of the first earner constant (v(w 1 , h 1 )), the PTR may be calculated as: where v 2 (w 2 , h 2 ) are total gross earnings of the second earner, and RRs are computed as rates of net incomes in and out of work: . for a single person: . for the first-earner in couple: . for the second-earner in a couple: In all of the above cases the definition of income out of work covers only social benefit incomes, such as FB, housing benefit and social assistance. We assume that families do not have any other sources of income, except for lone parents who receive alimony payments at the median value observed in the data. It is also important to note that we assume in all of the analysed cases that there are no disabled people in the households, in which case the families could be eligible for specific types of benefits or for higher amounts, and that households do not receive unemployment insurance benefits. 12 While often changes in the PTRs and RRs show similar patterns of work incentives as noted by Adam et al. (2006) 'both these measures (e.g. RR and PTR) attempt to capture the incentive to work at all, but they are different, and as a result of this, these measures behave differently following different sorts of changes in income'. One of the intuitive features of the PTR is that for a given change in net incomes in and out of work the measure increases in the value of gross earnings, which implies that if one has to work longer hours to gain the same income over and above that received out of work, then work is less attractive. The PTR also allows meaningful analysis of changes in the rates of tax in cases when income out of work is zero. By construction of the RRs this always results in the ratio being zero, regardless of the level of income in work. On the other hand the PTRs have the rather counterintuitive feature that they are constant if for a given level of gross earnings, disposable incomes in and out of work grow by the same amount. However, in particular given the focus of this paper, the disadvantage of the PTR is that it cannot be used to compare the tax and benefit system for different wage distributions as the rates cannot distinguish between higher gross wages resulting from higher work intensity (i.e. longer hours) or higher productivity.
On the other hand RRs remain constant if incomes in and out of work grow by the same proportion, and they are useful in interpreting changes in work incentives at different hourly wage rates for a given level of work intensity. Generally speaking, for a given level of earnings, the PTR will fall if the absolute change in disposable income in work, Y (v), is higher than the absolute change in the disposable income out of work, Y (0). The RR in turn will fall if the proportional change in Y (v) is greater than the proportional change in Y (0).

Reforms and their effects on components of income
The reforms implemented between 2005 and 2011 affected all major elements of the tax and benefit system. Since these elements are strongly interrelated the changes in social security had their implications for income tax, and all reforms to the taxation of earnings affected benefit eligibility. In Figure 1 we present the combined set of consequences of the reforms for two of the four family types -a lone parent with one child and a couple with two kids. The figures, conditional on an hourly wage corresponding to the minimum wage (Figure 1(a) and 1(c)) or the mean wage (Figure 1(b) and 1(d)), show how four elements of the tax and benefit system affected incomes. In the figures we distinguish three elements of the implications to changes in the taxation of earnings and changes to FB. With respect to earnings taxation three elements are presented separately: . changes in 'net earnings' related to reforms of employee SSCs, health insurance and income tax thresholds and rates; . the effect of the CTCidentified separately out of the income tax reforms; . changes to the employer side of SSC ('SSCemployer'); Changes to FB reflect, on the one hand, the increases in the value of payments for the families who continue to receive them, but on the other the loss of benefits resulting from the reduction in the real value of eligibility thresholds and increases in net earnings following the SSC and income tax reforms.
Several important conclusions follow from the analysis of these changes. First of all, it is notable that single earner families working at the minimum wage do not earn enough to take full advantage of the CTC even at full time or more than full time hours. In the case of a married single-earner couple with two children only full time work at the mean or higher wage generates enough taxable income for the family to take full advantage of the CTC. In the case of this family type we can also see the significant implications of eligibility criteria for FB. This affects the family in the range of hours between 31 and 42 per week, and since the entire benefits are withdrawn once the family crosses the eligibility threshold, the implied loss is as high as 111.79 PLN per month. For higher income families this loss was (partly or fully) compensated by the CTC. Overall Figure 1 demonstrates how significant the reforms were for family incomes. The actual gains naturally depend on the assumptions concerning the incidence of social security and tax changes. If we take the short term assumption of legislated incidence (i.e. assume that employees gain the full extent of changes to employee social security and income taxes, but not employer SSCs), then families with a full time working adult gained from 52.19 PLN per month in the case of a lone parent receiving the minimum wage to 107.74 PLN in the case of a married couple with two children on mean earnings. If on top of that we include the potential gains from cuts of employer SSCs (as suggested for example in Gruber, 1997; see also Fullerton & Metcalf, 2002), these grow respectively to 84.58 PLN and 140.13 PLN per month.

Reforms, earnings and budget constraints
The analysis of labour market incentivesboth at the intensive and the extensive margin draws on the relationship between gross earnings and disposable incomes, that is, on the family budget constraints. Before turning to the analysis of labour market incentives in Section 4, below we focus on the implications of the reforms on the entire budget constraints of the selected stylised families. What we show is on the one hand the degree of changes in disposable income conditional on the intensity of work resulting from the reforms. On the other hand, to put the reforms in a broader perspective, we analyse the effect of changes in real gross earnings identified separately from policy changes. As we demonstrate, overall dynamics of wages between 2005 and 2011 have implied a significant shift in the level of financial incentives to work, which was generally stronger than the effect of the reforms. When combined, however, these two parallel developments very significantly changed the attractiveness of work. In the analysis below we assume the legislated incidence of SSCs, which means that changes in the employer part of the contributions are not reflected in the budget constraint. The justification of this is to avoid double counting in the analysis of wage effects on disposable incomes, since any shift of the reduced tax burden towards wages will already be reflected in the wage growth. One has to remember though, that part of the observed dynamics of wages after 2008 may reflect a policy change with regard to the disability insurance contributions (see Section 2). Figure 2 demonstrates the role of tax and benefit reforms for the four stylised families assuming that the earner in the family receives the minimum wage. The budget constraints are drafted conditional on the number of hours worked per week for the uprated 2005 system and the system from 2011. The first outcome that stands out in Figure 2 is the fact that nearly in all cases along the entire distribution of hours worked, the 2011 budget line is above that from 2005. The only exception is the case of lone parents working at the minimum wage who due to the withdrawal of FB at about 20 hours of work in the 2011 system have lower incomes up to about 30 hours of work per week. As we noted earlier, minimum wage earners cannot take advantage of the CTC even if they work full time, so the loss of FB is not compensated through the new element of the tax system. In the scenario in which families receive neither the Housing Benefit nor Social Assistance, at full time hours singles without children who earn the minimum wage have gained about 52.12 PLN per month (5.30%), lone parents gain 52.19 PLN (3.68%), couples without children 42.00 PLN (4.02%) and couples with two children 107.74 PLN (9.32%). Important changes can be seen in the level of income for families who do receive the safety net benefits, in particular in the case of couples. Those without children could see their incomes rise by 32.34 PLN (2.46%), while income of those with two kids grows by 107.74 PLN (7.13%). Because of the family structure couples with two children in which the earner receives the minimum wage have not been affected by the restrictions in the access to family benefits. Among those who receive safety net benefits, couples with children where one of the parents works at the minimum wage have seen their full time incomes rise by 107.74 PLN (9.32%), compared to an increase among couples without children by only 32.34 PLN (2.46%) per month. As we shall see below eligibility for safety net benefits has significant implications on the resulting changes in the incentives to work for all of the families considered.
The budget lines presented in Figure 3 show the constraints for two of the four family types under the assumption of the mean wage. Apart from reflecting the exercise presented for the minimum wage in Figure 2, in this case we demonstrate also the effect of changes in the real value of the wage from 2005 to 2011. This is done by drafting the levels of disposable incomes corresponding to: Looking at full time work we can see that the total real difference in the disposable income between 2005 and 2011 amounts to 544.12 PLN for a single adult without children (Figure 3(a)) and to 541.78 PLN for a couple with two children (Figure 3(b)). Of these amounts a significant proportion relates to the changes in the tax and benefit system introduced over the period (difference between '2011(wage '11)  This means that at the level of full time average wages the tempo of real earnings growth has contributed about twice as much to the growth of real disposable incomes compared to the implemented reductions in the level of taxation. This is an important context within which one has to consider the implications of the reforms which we discuss below, and it demonstrates that despite a slow-down in the rate of growth of real earnings at the end of the period, real earnings growth led to substantial increases in the financial gain from work. At this point it is important to note though that, as we mentioned above, depending on the incidence of employer's social security contributions, some of the real earnings growth could have been driven by the reduction in these contributions implemented in 2008, however even if the entire reduction of two percentage points were translated into higher gross earnings, this would still correspond to a small fraction of the total real earnings growth between 2005 and 2011.
The discussion in the sections below focuses on the most important aspects of the policy effects in terms of labour market incentives, both on the intensive and the extensive margin. We discuss the implications of the reforms and of changes in gross wages for labour market incentives of first earners in families.
The analysis draws on the budget constraints computed for all of the family types under different wage and system scenarios. In Section 4.2 we discuss the consequences of reforms on incentives on the intensive margin, following which we look at changes in financial gains to work on the extensive margin for first earners in the considered families (Section 4.3) and for second earners in the couple with two children (Section 4.4).

Reforms and incentives on the intensive margin
The reforms to social security and labour taxation have had a very significant effect on the level of the labour tax wedge in Poland  . As a result of the entire package of reforms introduced between 2005 and 2011 the effective marginal rate of taxes on earnings (EMTRs) has been reduced on almost the entire range of earnings. Two examples of the MTR schedule under the two systems are presented in Figure 4. Figure 4(a) shows the EMTR schedule for a single person without children, i.e family (1 + 0), while in Figure 4(b) we demonstrate the schedule for one-earner families with two children (family (2 + 2)). 13 As we can see, the only  (1). The full EMTR on disposable income including withdrawal of the Housing Benefit and Social Assistance is presented in Figure A2 in the appendix. range of earnings over which the marginal rate is higher in the 2011 system is from 7250 PLN to 8895 PLN for family (1 + 0) and from 7245 PLN to 8400 PLN for the family with two children. This increase in the marginal rate results from the fact of real increases in the upper earnings limit for social security over which contributions to retirement and disability pensions stop being paid. This threshold increased regularly with the government's expected growth of nominal earnings and grew in real terms by 16% between 2005 and 2011.
The figures show several other interesting features of the reform package. The schedule for family (1 + 0) reflects very strongly the effect of the changes in the personal income tax schedule, in particular the effect of abolishing the 30% rate and setting the threshold for the new higher rate (32%) at the old higher rate level. This significantly reduces the overall EMTR on the range of earnings between 4695 PLN and 7240 PLN from 53.39% to 41.18%. On the other hand families with children saw their EMTR fall substantially as a result of the CTC. In our example the EMTR for family (2 + 2) falls further to 33.71% over the range from 1460 PLN to 3370 PLN, which roughly covers the range between the minimum and mean earnings.
The full EMTR schedule including the implications of benefit withdrawal looks significantly more complicated (see Figure A2 in the appendix) with a number of rates over 100% due to benefit thresholds beyond which benefits stop being paid. The figures show also the consequences of benefit increases for those receiving Social Assistance and the Housing Benefit. As a result of an increase in the generosity of payments in Social Assistance and due to its withdrawal as work income grows, the full EMTR increases for lowest levels of earnings from 61.37% to 70.13% for family (1 + 0) and from 55.85% to 70.13% for family (2 + 2).

Extensive margin: reforms and incentives for the first earner
Results concerning financial incentives to work on the extensive margin for the first earner in the family are presented in Tables 1 and 2, and for the second earner in the couple with two children in Table 3. For illustration we also present examples of PTR and RR schedules for single adults without children and for couples with two children in Figure 5. These schedules reflect important differences in how labour market incentives are captured by the two measures, and in particular how their sensitivity to the level of out of work income differs. RRs are generally much more sensitive to the level of income at zero hours of work. This is natural, given the definition of the two  (2) and (3) measures and their first derivatives with respect to Y (0) (namely: , h)). Since generally gross incomes will be higher than disposable incomes in work, the slope of the RR with respect to income out of work will be higher. 14 In Figure 5 this is reflected in the schedules representing incentives for first earners in the (2 + 2) family ( Figures 5(b) and 5(d)). The fact that in these cases out of work income has been significantly increased as a result of changes in the value of family and safety net benefits (see Figure 2), the RRs suggest weakening of labour market incentives at the extensive margin even for high levels of earnings. Both in the case when the (2 + 2) family is assumed to receive the Housing Benefit and Social Assistance and in the alternative scenario of no eligibility, RRs are higher for the 2011 system compared to the uprated system from 2005. For the family which is eligible to safety net benefits the RR at full time minimum wage earnings grows from 0.550 to 0.648, and at mean earnings from 0.317 to 0.373 (for details see Table 2). The change in the value of benefits  (5) and (6)  also dominates changes in net labour incomes in the PTR measure for low and medium paid families eligible to safety net benefits (see Table 1). For these families, more or less up to the level of mean wage the PTRs are higher in the 2011 system compared to the 2005 system. At full time minimum wage the PTR grows from 0.594 to 0.654 between the two systems, while at full time mean wage the tax rates are almost identical (0.560 in 2005 and 0.557 in 2011). On the other hand, the PTRs are significantly lower for the (2 + 2) family which does not receive safety net benefits. In this scenario the PTR at full time minimum wage falls from 0.378 to 0.334, and for mean wage earners from 0.428 to 0.379. Figure 5 shows that at higher earnings PTRs generally fall between 2011 and 2005 while RRs either stay at the same level or increase as in the case of the couple with two children (not eligible for HB and SA - Figure 5(d)).

Extensive margin: reforms and incentives for the second earner
In the analysis of financial incentives to work on the extensive margin for the second earner we present results for different combinations of earnings focusing only on the couple with two kids (family 2 + 2). Within this family type we distinguish couples on low-, medium-, and high wage combinations. We assume that the earnings of the first person are fixed at full time minimum, median or 75th percentile earnings, and examine PTRs and RRs for the minimum and mean earnings of the second partner assuming half-time and full-time work. Results of PTR and RR calculations for this family type are presented in Table 3. As one could expect second earner RRs under all system/wage combinations grow with the earnings of the first earner confirming the negative effect of first earner's income on the financial incentives to work for the second earner. For example in the case of second earner's full time work at the mean wage under the 2011 system RRs grow from 0.360 through 0.505, to 0.597 depending on whether the first earner receives the minimum, median or 75th percentile wage. It is notable that if we look at changes in the system between 2005 and 2011 in all except two combinations of first and second earner scenarios RRs suggest weakening of financial incentives to work for the second earner. In some cases these changes in incentives are substantial. If the first earner receives median earnings, for example, and the second earner is on the minimum wage, the RR grows from 0.793 to 0.843. The only exceptions are for the combination of earnings at the minimum wage of both partners when the second earner works part time (RR is reduced from 0.701 to 0.689 between the 2005 and 2011 systems), and the full time earnings on the mean wage of the second earner when the first earner receives the minimum wage. This may be explained by high increases in the level of the minimum wage.
These negative effects of the 2005-2011 package on work incentives measured by the RRs suggest a different pattern of changes compared to that reflected in the PTRs, where for all earnings combinations the rates are reduced except when a second earner has a full-time median wage and a first earner has a minimum wage ( Table 3).
The The results indicate that while generally the rates of tax on participation have fallen for second earners, the RRs suggest that the relative value of income out of work and in work has increased, reducing financial attractiveness of work for second earners with children. These effects are consistent with the introduction of the CTC which affects mainly the net incomes of the first earner, and changes in the value of the withdrawal threshold of FB in particular at relatively low earnings. Looking at the two different measures of financial incentives, especially in the case of the second earner, leads to different conclusions concerning the effect of the reform package.

Extensive margin: wages and incentives to work
The conclusions on the role of systemic changes in the period 2005-2011 concerning the financial incentives differ in many instances depending on whether we use the PTRs or RRs to measure the effects. In this section we consider the role of changes in earnings on the financial attractiveness of work using the full and part time RRs of first earners. In this case the relevant comparison is between RRs computed under the 2011 system for 2011 wages ( Table 2) and for the indexed 2005 minimum and mean wages under the same system presented in Table 4. In interpreting the results we need to remember that if the 2pp reduction in employer social security contributions affected gross wages, part of the wage effect is driven by policy.
As one could expect for cases where the comparison can be meaningfully conducted (i.e. where out of work income is not zero) RRs are higher when we use the lower 2005 wages, regardless of whether we apply them to the 2011 or the uprated 2005 system. In some cases, for example for family (1 + 0) at part time work and receiving safety net benefits the wage effects on RRs are low, but in most analysed examples the wage effects are substantial and either dominate the negative effects of system changes or significantly reduce them. For example, taking the (2 + 2) family where the first earner works full time (last columns of Tables 2 and 4)

Conclusion
Between 2005 and 2011 the Polish system of taxes and benefits saw a series of significant reforms. The rates of SSCs and the income tax were reduced and a generous CTC was introduced to further lower the level of income tax. At the same time the government raised health insurance contributions and froze income tax thresholds which worked against the tax cuts. On balance, though, taxes on labour have been substantially reduced. This resulted in significant reductions of the EMTRs, in particular for higher earners. At the same time values of FB and social assistance were raised, though availability of these benefits to families on low labour incomes was reduced through freezing of eligibility criteria.
During the analysed period there was a substantial increase in labour force participation of men and women among all age groups. The higher labour supply levels were reflected in growing levels of employment andat the time of the economic slowdownin higher numbers of registered unemployed. As we showed in this paper growing wages and not tax and benefit reforms were a more likely cause behind these changes.
Since the analysed period was also a time of rapid growth of real earnings, this provides for an interesting setting to analyse changes in labour market incentives resulting on the one hand, from policy reforms and on the other, from wage developments. This context has been used in this paper, to identify the consequences of the reforms and wage growth for labour market incentives  (5) and (6) and to compare the way how two main measures of financial incentives to work on the extensive margin, namely PTRs and RRs, reflect these changes. We demonstrated that very often conclusions concerning changes in the financial incentives to work on the extensive margin differ when we refer to the PTRs or RRs. This is particularly the case when we consider families which experienced significant changes to their out of work income, and in the case of second earners in couples. Given the different nature of the two measures, we should not expect that they would always indicate the same direction of change, yet the degree to which conclusions concerning the effect of policies and wages on work incentives differ is substantial. The PTRs usually suggest significant increases in the financial attractiveness of work as a result of the 2005-2011 Polish reform package, while the RRs in nearly all cases of households with children and those eligible to safety net benefits indicate worsening of financial incentives to work. This demonstrates that depending on the nature of the reforms the two measures may imply conflicting conclusions concerning changes in the financial incentives to work, and suggest that one ought to be cautious in interpreting their implicationsin particular when complex policy packages are compared. Falling PTRs may often not be reflected in lower values of RRs.
In parallel with the objectives of poverty relief, work incentives are key concerns determining tax and benefit policy of most governments. By international standards work incentives in Poland are relatively strong for single earner households and are much weaker for second earners (OECD, 2011). With growing concerns over low fertility and its consequences for Poland's demographics the governments have in the last few years implemented a number of measures to increase the financial attractiveness of work for low income families. It seems however, that the new government's approach (after the elections of October 2015) may give less importance to the consequences of policy on employment incentives. The experience of the analysed period in this paper suggests that this, in particular if combined with a slow-down in the dynamics of wages, could have important negative implications for the level of employment in the Polish economy.