Informal unemployment and education

Transcription

1 Kom and Larsen ZA Journa of Labor Economics 06 5:8 DO 0.86/s ZA Journa of Labor Economics ORGNAL ARTCLE Open Access nforma unempoyment and education Ann-Sofie Kom and Birte Larsen * *Correspondence: b.eco@cbs.dk Copenagen Business Scoo, rederiksberg, Denmark u ist of autor information is avaiabe at te end of te artice Abstract Tis paper deveops a four-sector equiibrium searc and matcing mode wit informa sector empoyment opportunities and educationa coice. We sow tat underground activities reduce educationa attainments if informa empoyment opportunities mainy are avaiabe for ow-educated workers. A more zeaous enforcement poicy wi in tis case improve educationa incentives as it reduces te attractiveness of remaining a ow-educated worker. However, unempoyment aso increases. Caracterizing te optima enforcement poicies, we find tat reativey more audits soud be targeted towards te sector empoying ow-educated workers; esewise, a too ow stock of educated workers is materiaized. JEL Cassification: H6,, J64 Keywords: Tax evasion, Te informa sector, Education, Matcing, Unempoyment ntroduction Researcers ave been puzzed by te fact tat observed tax evasion in ig-income countries, despite ow audit rates and fairy modest fines, is substantiay ower tan wat is predicted by teory. Andreoni et a. 998 argue tat tis discrepancy is most ikey expained by non-economic factors, suc as moraity, guit, and same. However, Keven et a. 0, wo conducted a arge fied experiment on individua tax fiers in Denmark, suggest tat tis discrepancy is expained by te degree of tird-party reporting. As incomes for individuas are not sef-reported, rater reported by a tird party suc as te empoyer, it is difficut, and tus costy, to evade taxes. Tese costs, bot due to tird-party reporting, or even moraity, guit, or same, tend to reduce te profitabiity of evading taxes and imit te size of te informa sector, atoug te expected punisment fees are ow reative to taxes. n tis paper, we argue tat tese types of costs may expain wy igy educated workers to a esser extent evade taxes and work informay tan ow-educated workers. f igy educated workers to a smaer extent work in industries wic ande cas payments and to a arger extent are subject to tird-party reporting, it wi be more difficut, and tus more costy, for tese workers to evade taxes. Tis is consistent wit data. Evidence indicates tat manua workers, or workers wit a ower eve of forma education, to a substantiay iger degree face informa empoyment opportunities compared to igy educated workers. Pedersen 003, using te same questionnaire design for Germany, Great Britain, Denmark, Norway, and Sweden, confirms tat skied bue coar workers carry out more informa market activities tan 06Te Autors. Open Access Tis artice is distributed under te terms of te Creative Commons Attribution 4.0 nternationa License ttp://creativecommons.org/icenses/by/4.0/, wic permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to te origina autors and te source, provide a ink to te Creative Commons icense, and indicate if canges were made.

2 Kom and Larsen ZA Journa of Labor Economics 06 5:8 Page of 36 oters. igure sows te extent of informa activities in te five countries by industry. Most informa work are carried out in te construction sector, foowed by te agricutura sector, otes, and restaurants. Tis pattern is aso confirmed for Denmark, by Hvidtfedt et a. 0, and for Germany, by Haigner et a. 0, using representative survey data. urtermore, performing ogistic regressions for te five countries, Pedersen 003 confirms tat te ikeiood of informa market activities fas wit te engt of education. n addition, Boeri and Garibadi 005 sow for tay tat mainy workers at te ower end of te ski distribution engage in informa activities. Te fact tat mainy ow-educated workers seem to work in te informa sector suggests tat te coice of educationa attainment is potentiay distorted. nforma empoyment opportunities foregone wit education may simpy reduce te incentives for workers to acquire education. Te aim of tis paper is to investigate te equiibrium impact of underground activities on abour market outcomes and educationa attainment in ig-income countries, as we as to caracterize te optima enforcement poicy. Atoug arser punisment poicies may correct for a too ow stock of educated workers, tota unempoyment may increase wit suc poicy. n fact, we ave itte guidance from researc to wat extent forma sector jobs repace jobs in te underground economy as tose jobs disappear wit stricter informa sector punisment. or tis purpose, we deveop a four-sector genera equiibrium mode featuring matcing frictions on te abour market. Unempoyed workers searc for jobs in bot a forma and an informa sector, and workers decide weter or not to acquire iger education based on teir abiity eves. Education is considered to be a once and for a investment in uman capita and takes pace as soon as te worker enters te abour market. n order to isoate te mecanisms and increase te transparency of te mode, we keep te differences between te forma and informa sectors at a minimum. Te ony dissimiarities between te sectors are tat taxes are not paid in te atter and tat productivity in te forma sector may be iger tan productivity in te informa sector. nstead of paying taxes, informa sector firms ave to pay a fine in case tey are it by an audit and detected as tax ceaters. n addition, firms in te informa sector are assumed to face conceament costs. n our mode, we et conceament costs capture costs associated wit ig. raction of informa sector work by industry. Pedersen 003

3 Kom and Larsen ZA Journa of Labor Economics 06 5:8 Page 3 of 36 conceaing taxabe income due to tird-party reporting or even moraity, guit, or same. Te costs reduce te profitabiity of evading taxes and imit te size of te informa sector atoug te expected punisment fees are ow reative to taxes. n ine wit Keven et a. 0, we aso et tese costs be iger te more income tat is idden from te tax autorities. We find tat underground activities reduce te incentives to acquire iger education if informa empoyment opportunities mainy are avaiabe to ow-educated workers. More zeaous enforcement poicies wi in tis case improve educationa incentives as tese reduce te attractiveness of remaining a ow-educated worker. However, if aso igy educated workers to a arge extent are exposed to informa empoyment opportunities, te incentives to acquire iger education may fa wit stricter enforcement poicies as underground work pays off better to workers wit ig productivity. Moreover, we find tat actua unempoyment most ikey increases, atoug te officia unempoyment fas. inay, caracterizing te optima enforcement poicies, we find tat reativey more audits soud be targeted towards te sector empoying ow-educated workers; esewise, te outcome is a too ow stock of educated workers. Te present paper extends te strand of tax evasion iterature wic departs from te assumption of imperfecty competitive abour markets by incorporating invountary unempoyment troug te incusion of searc frictions. 3 See, for exampe, ugazza and Jacques 004, Boeri and Garibadi 005, and Kom and Larsen 006 wo aso mode underground activities in ig-income countries. Tese studies focus on abour market outcomes and rey on asymmetries between te forma and te informa sector, suc as eterogeneity in moraity, in order to expain te co-existence of a forma and an informa sector. Tere are aso numerous studies based on searc teoretica frameworks investigating issues of informa empoyment from te point of view of ow- and midde-income countries. As one can argue tat te nature of te informa sector can be quite different in ow- and midde-income countries compared to ig-income countries, te modeing strategies usuay differ in tese set-ups. As pointed out by La Porta and Seifer 04, te informa sector in ow- and midde-income countries is usuay uge and contains sma, unproductive, and stagnant firms. Moreover, te informa sector in tis iterature is usuay seen as an unreguated sector. or an exampe, taking a Latin American perspective, see Abrect et a. 009 tat accounts for worker eterogeneity and considers te impact of payro taxes and severance pay on unempoyment in te presence of an informa sector. Te informa sector can be seen as an unreguated sector wic is not affected by payro taxes and oter forma poicies. 4 Te recent study by Megir et a. 05 takes a sigty different modeing approac in its focus on underground activities in Brazi as te paper considers on-tejob-searc and firm eterogeneity. Workers may searc for jobs bot in te forma and te informa sector, and searc frictions make it profitabe for firms to start bot types of jobs. Te paper is organized as foows. n Section, we provide an empirica background and motivation for te paper. n Section 3, te mode is set up. Section 4 offers a comparative statics anaysis of an increase in te reative punisment of informa activities. Section 5 considers optima poicy, and finay, Section 6 concudes.

4 Kom and Larsen ZA Journa of Labor Economics 06 5:8 Page 4 of 36 Background and motivation As individuas engaged in underground work do not wis to be identified, it is notoriousy difficut to coect accurate information about tese activities. or natura reasons, we terefore ave imited knowedge about te empirica reationsips between informa activities and oter economic outcomes. n tis section, we construct a cross-sectiona data set of 4 OECD countries to investigate te reationsips between factors affecting underground activities and educationa outcomes. A OECD countries are incuded in te sampe provided tat we ave data on te size of te informa sector and information on te ega and reguatory framework for te purpose of tax compiance coected by te Goba orum on Transparency and Excange of nformation for Tax Purposes OECD 0. Since te informa economy cannot directy be measured, one as to rey on indicators tat capture informa sector activities in order to estimate te size of te sector. Here we use te most recent estimates derived by Scneider et a. 00. nstead of using a metod wic assumes tat a singe factor or indicator can capture a activities in te informa sector, suc as te currency demand approac or te eectricity approac, tey estimate te size of te informa sector using a metod wic incudes mutipe causes and indicators of te informa sector. 5 igure provides a picture of ow arge te informa sector is in reation to GDP in te different countries. f, as we argue, informa empoyment opportunities are foregone wit iger education, we soud observe a ower stock of educated workers in countries were it is more profitabe to work in te underground economy. Tus, countries wit ess strict enforcement poicies or ower conceament costs reative to te tax burden soud ave a smaer stock of igy educated workers. To measure te costs of informa sector work in a country, we construct a variabe based on te ega and reguatory framework on te avaiabiity of, and access to, information of importance for tax compiance. Te data used is coected by te Goba orum on Transparency and Excange of nformation for Tax Purposes OECD 0. Te Goba orum as set out a arge number of standards in order to increase tax compiance, and troug ig. Te size of te sadow economy as a fraction of GDP for eac country in 007

5 Kom and Larsen ZA Journa of Labor Economics 06 5:8 Page 5 of 36 a process of peer reviewing, te orum assesses te ega and administrative framework in eac member country. More specificay, te peer-reviewing process provides information if te standards are in pace, in pace but tere is need for improvements, and not in pace. rom tis information, we construct an index capturing te costs of evading taxes. 6 Te cost of evading taxes needs to be reated to te cost of not evading taxes. 7 We et tese costs be captured by te tax wedge of tota abour costs to te empoyer reative to te corresponding net take-ome pay for te average singe worker witout cidren. Tis data from te OECD Taxing Wages database provides unique information on te income taxes paid by workers and te famiy benefits received in te form of cas transfers as we as te socia security contributions and payro taxes paid by teir empoyers, for eac of te OECD countries. igure 3 pots te percentage of te tota popuation, 5 64 years od, oding a tertiary education in 007 OECD 0 against our measure for te cost of evading taxes reative to not evading taxes. Consistent wit our ypotesis, we observe a positive correation between te measures; te ess attractive it is to work in te informa sector, te more workers wi coose a iger education. n ig. 4, te aim is to see if te reative cost of evading taxes is negativey correated wit te size of te informa sector. ndeed, we observe a negative correation between te size of te underground economy as a fraction of GDP and te percentage of te 5 64 years od of te popuation oding a tertiary education. Tus, economies were te informa sector is more extensive aso tend to be economies were a ower fraction of te popuation educates temseves. Ceary, as it is caenging to get an accurate measure of te size of te informa sector and, as we ave done ere, to get a measure of te costs of evading taxes, tis section ony serves to provide correations between te variabes in focus. To identify causa reationsips between, on te one and, tax and punisment poicies and, on te oter and, educationa outcomes is giving te avaiabe data an overweming task. Next, we buid an equiibrium mode to investigate tese reationsips, as we as to pin down te mecanisms. ig. 3 raction of 5 64 years od wit tertiary education as a function of te wedge between te informa and te forma sector for 007

6 Kom and Larsen ZA Journa of Labor Economics 06 5:8 Page 6 of 36 ig. 4 raction of 5 64 years od wit tertiary education as a function of te size of te sadow economy as a fraction of GDP in Temode Tis section deveops a four-sector genera equiibrium mode wit forma and informa sector empoyment opportunities and educationa coice. Workers differ in te abiity to acquire education. Abiities, e, are uniformy distributed between 0 and, e [0,],and te cost of iger education, ce, is decreasing in abiity. Tus, workers wit a ig eve of abiity wi find it more tan wortwie to invest in iger education, wereas workers wit ow abiity wi not. Workers not attaining iger education wi from now on be referred to as manua workers. Bot manua and igy educated workers aocate searc effort optimay between te forma and te informa sector. Once matced wit a firm, tey bargain over te wage. Te economy tus consists of four sectors: te forma and informa sectors for manua workers denoted, m and, m and te forma and informa sectors for igy educated workers denoted, and,. 3. Matcing Manua and igy educated workers searc for jobs in bot a forma and an informa sector. or simpicity, we assume tat ony unempoyed workers searc for jobs. Tis is a simpification, i.e. we do not acknowedge tat te connection to te abour market given by working in te forma or informa sector may bring about job opportunities not avaiabe wie unempoyed. Te matcing functions for te four categories of jobs are given by X j v j j γ u,werex j is te sectoria matcing rate, v j is te sectoria vacancy rate, and u is te unempoyment rate and j, and m,. Teratesare defined as te numbers reativey to te abour force of manua and igy educated workers, respectivey. Te exponents in te matcing function are set to be equa to af in order to simpify te wefare anaysis were we derive te optima tax and punisment system wen we ave imposed te traditiona Hosios condition. n tat case, we can disregard congestion externaities as te easticity of te expected duration of a vacancy is equa to te bargaining power of workers in a symmetric Nas bargaining situation. 8 Workers aocate searc effort optimay across te forma and te informa sector. A worker wit educationa eve wi direct units of searc for a forma sector job and units of searc for an informa sector job. Tus, workers wit different eves of education

7 Kom and Larsen ZA Journa of Labor Economics 06 5:8 Page 7 of 36 may differ in teir aocation of searc time across a forma and informa sector. Eac worker s tota searc intensity is, owever, exogenousy given and normaized to unity, i.e. +, m,. Teparameterγ < captures te effectiveness of searc fas wit searc effort, i.e. te first unit of searc in one sector is more effective tan te subsequent units of searc. Tis coud capture tat different searc metods are used wen searcing for a job in a market. Te more time tat is used in order to searc in a market, te ess efficient searc metods ave to be used. Te transition rates into informa and forma sector empoyment for a particuar worker i are λ i i γ and λ i i γ,were v / γ u and v / γ u are abour market tigtness, m,, measured in effective searc units. Te rates at wic vacant jobs become fied are q j j, j,, m,. 3. Vaue functions Let U, E,andE denote te expected present vaues of unempoyment and empoyment for manua and igy educated workers. Te vaue functions for worker i ten reads ru i R + λ i E U i + λ i E U i aui, m,, re i R + w i + s U E i aei, m,, rei R + w i + s U Ei ae i, m,, 3 were r is te exogenous discount rate, w j is te sector wage, and s is te exogenous separation rate. R is a ump sum transfer tat a individuas receive from te government wic refects tat te government as some positive revenue requirements. 9 Te parameter a is te rate by wic a worker is dying, and it captures tat tere is a constant fow of workers out of te abour market at eac instant of time. Anaogousy, tere is an equay sized fow of workers into te abour market eac time period as peope are born at te same rate. Tis keeps te popuation constant, normaized to unity, and enabes us to ook at te impact of various poicies on educationa attainment despite te fact tat education is an irreversibe investment. Let J j and V j j, represent te expected present vaues of an occupied job and a vacant job in te forma and informa sectors, respectivey. Te arbitrage equations for forma and informa sector jobs paying te wage w j i j, and a vacant job are ten rji y w i + z + s V Ji aji, m,, 4 rv q J V ky av, m,, 5 rj i y w i + pα + κ + s V J i aj i, m,, 6 rv q J V ky av, m,, 7 were z istepayrotaxrateandy j, j,, m,, is productivity. Te parameter p is te auditing rate wic captures te probabiity of being detected empoying a worker in te informa sectorand α is te associated firm punisment fee rate. Vacancy costs are indexed by factor k to te productivity in te sector and written ky j, j,, m,.0

8 Kom and Larsen ZA Journa of Labor Economics 06 5:8 Page 8 of 36 Te conceament costs, κ, m,, capture tat it is costy to ide income from te tax autorities. Te costs coud, for exampe, capture wat Keven et a. 0 refer to as tird-party reporting. Wen tere is tird-party reporting of income, suc as te firm reporting te wage payments directy to te tax autorities, tis as to be agreed upon aso by te worker, wic is costy. Tese conceament costs coud aso be oter direct costs associated wit conceaing evasion, as we as moraity costs associated wit evading taxes. f firms iring igy educated workers ave a arder time conceaing teir activities tan firms iring manua workers, ten κ >κ m. Tis is te case if, for exampe, tirdparty reporting is more common for igy educated workers, or as assumed in Keven et a. 0, te margina costs of evasion increase wit te amount of income evaded. Atoug tis is ikey to be te case, we do not a priori impose any restriction on te vaues of κ,, m. n order to improve te transparency of te mode, we disregard taxation, expected punisment, and conceament costs on te worker side. Tis is of no importance for te resuts. Te unempoyed worker i aocates searc between te two sectors, i,inordertomax- imize te vaue of unempoyment, ru i. A necessary condition for an interior soution is tat γ<, wic ods by assumption. Te first-order condition can be written as γ i γ E U i i E U, m,. 8 i Workers aocate teir searc between sectors to equaize te net margina returns to searc effort across te two sectors. 3.3 Wage determination Wen a worker and firm meet, tey bargain over te wage, w j i, taking economy-wide variabes as given. Te first-order conditions from te Nas bargaining wit equa bargaining power for workers and firms can be written as J E U + z, m,, 9 J E U + pα + κ, m,, 0 were we ave imposed symmetry and te free entry condition, V j 0, j,, m,. We can now derive an equation determining ow searc is aocated between te forma and te informa sectors in a symmetric equiibrium by substituting 9 and 0 into 8 and using J ky q γ γ and J ky q y y from 5 and 7 togeter wit free entry. Tis yieds ψ, m,, were ψ +pα+κ +z is te cost wedge between te informa sector and te forma sector. Wen workers aocate teir searc between te forma and te informa sectors in equiibrium, tey account for te wedge, ψ, and for te forma reative to te informa

9 Kom and Larsen ZA Journa of Labor Economics 06 5:8 Page 9 of 36 sectoria tigtness, /, as we as for reative productivity, y /y. t foows tat reativey more searc wi be directed towards te forma sector if expected punisment pus conceament costs are iger tan te tax payments, i.e. if ψ >, if forma sector tigtness exceeds informa sector tigtness i.e. / >, and/or if productivity in te forma sector is iger tan productivity in te informa sector, y /y >, and vice versa wen ψ <, / <, and y /y <. By use of Eqs. 7 and 3 in Eqs. 9 and 0, equiibrium producer wages, ω j,aregivenby ω w + z y + k ω w + pα + κ y + γ γ k m,,, m,. 3 An increase in tigtness, j, makes it easier for an unempoyed worker to find a job and at te same time arder for a firm to fi a vacancy. Tis improves te worker s reative bargaining position, resuting in iger wage demands. An increase in searc wi instead increase te firm s reative bargaining position. Tis is te case as firms wi ten find it easier to matc wit a new worker in case of no agreement. Te improved bargaining position for firms moderates wage pressure. 3.4 Labour market tigtness Labour market tigtness for te forma sector and te informa sector are determined by Eqs. 4,5, 6, and 7 using te free entry condition and te wage Eqs. 33 and 34: k r + s + a k γ,,, 4 i k r + s + a k γ,,. 5 i By use of te equiibrium searc aocation equation in 3, were i γ y γ ψ y, in 5, it becomes cear tat te wedge, ψ, and productivity differences, i y /y, are te crucia factors determining te size of te forma sector in reation to te informa sector. n case productivity is te same in te forma and informa sectors, ence, y /y, ten wen ψ >, and tus expected punisment pus conceament costs are iger tan payro taxes, informa sector producer wages are iger tan forma sector producer wages. n tis case, it is reativey more attractive for firms to enter te forma sector, wic makes forma sector tigtness exceed informa sector tigtness. Hence, we obtain tat > and <, m, if y /y ψ > andviceversa wen y /y ψ <. Notice tat te forma sector exceeds te informa sector > and <, m, bot if te wedge is equa to, ψ, and te forma sector is more productive tan te informa sector, y /y >, as we as if te forma and informa sectors are equay productive and te wedge is arger tan, ψ >. As te forma sector exceeds te informa sector in size in most ig-income countries, it is most reaistic to consider te case wen y /y ψ >. Tis impies considering te situation wen te expected punisment rate pus conceament costs exceed te tax rate, i.e. pα + κ > z, wen bot te forma and informa sectors are equay productive, wic

10 Kom and Larsen ZA Journa of Labor Economics 06 5:8 Page 0 of 36 does not seem unreaistic given a broad interpretation of conceament costs. n fact, as discussed in te introduction, positive conceament costs κ > 0suctatpα + κ > z coud potentiay expain te puzze of wy we observe a reativey sma informa sector atoug we, at te same time, observe rater ow audit rates and fairy modest fines, i.e. pα <z. n addition, wen te productivity in te forma sector exceeds tat of te informa sector, te forma sector is even more ikey to exceed te informa sector in size. However, we do not a priori impose any restrictions on te size of ψ, pα, orκ wen deriving te resuts in tis paper. Wen discussing resuts tat depend on te size of ψ, owever, we focus te discussion on wat we beieve is te most reaistic case. n ig. 5, we can use Eqs. 4 and 5 to derive reative tigtness as a function of searc intensity and iustrate tis equation in a, / diagram togeter wit Eq.. Bot equations ave a negative sope, and te former curve wi be fatter tan te atter around te equiibrium insuring a stabe equiibrium. Wen te wedge increases, ψ >ψ or y /y increases, ten te searc intensity decreases for given reative abour market tigtness, / ; tis reduction in searc intensity increases / and tereby unti a new equiibrium is reaced. n ig. 5, we ave eft out subscript to ease exposition. 3.5 Education Wen workers decide weter to acquire iger education or remain as manua workers, tey compare te vaue of unempoyment as an educated worker and te associated costs of iger education to te vaue of unempoyment as a manua worker. Workers tat find it optima to acquire iger education view tis as a once and for a investment in uman capita, and it takes pace as soon as te worker enters te abour market. As in most studies, we assume tat education is costy but it takes no time. 3 Te cost of iger education depends on individua abiity, e i [ 0, ], and is given by ce i,werec e i <0 and c e i > 0. 4 Te margina worker as an abiity eve, ê, wic makes im or er just indifferent between acquiring iger education and remaining as a manua worker. We write te condition determining te abiity eve of te margina worker as r + a U cê r + a U m 6 ig. 5 Tigtness in te forma sector reative to tigtness in te informa sector and searc intensity

11 Kom and Larsen ZA Journa of Labor Economics 06 5:8 Page of 36 By using Eqs. 3, it is cear tat workers proceed to iger education if te expected income gain of education exceeds teir cost of education. However, as wages are endogenous, we can use Eqs. and 6 togeter wit te first-order conditions for wages and Eqs. 5, 7, and togeter wit te free entry condition. Tis gives te foowing rewriting of condition 6: cê k y + z o y m o m, 7 were o / γ,, m. Equation 7 gives ê as a function of te endogenous variabes and, m,. Workerswite ê coose not to acquire education, wereas workers wit e > ê acquire education. Hence, ê and ê constitute te manua and educated abour forces, respectivey. Te rigt-and side of Eq. 7 is te expected income gain of attaining education. Tis gain needs to be positive in order for, at east some, workers to proceed to iger education. Te fact tat productivity is iger for igy educated workers, wic gives rise to an educationa wage premium, provides incentives for iger education. However, iger education may potentiay aso be associated wit osses in expected income. or exampe, if conceament costs are iger for igy educated workers, i.e. κ > κ m, reativey more attractive informa empoyment opportunities for manua workers wi be foregone in case of iger education. Tis reduces te incentives for education. 5 Ceary, in order to study te non-trivia case were at east some workers proceed to iger education, it is necessary to assume tat tere is a net gain in te expected income of iger education. Tus, we need to assume tat productivity differences between manua and igy educated workers are sufficienty ig, i.e. y /y m > o m/o.moreover, to guarantee a non-trivia interior soution were at east some, but not a, individuas coose to acquire education, te individua wit te igest abiity faces a very ow cost of education, more specificay c 0, and te individua wit te owest abiity faces a very ig cost of education, i.e. im e 0 ce. See te Appendix for te proof of te existence of ê 0,. 3.6 Empoyment and unempoyment Te equations determining te empoyment rates in te forma sector and te informa sector, n, n, and te unempoyment rates, u, m,, are given by te fow equiibrium equations and te abour force identity. 6 Te officia unempoyment rate u o is given by u o u + n. Soving for te empoyment and unempoyment rates yieds λ λ n s + a + λ + λ, n s + a + λ + λ,, m, 8 s + a u s + a + λ + λ, u o s + a + λ s + a + λ + λ,, m. 9 A comparison of te unempoyment rates for manua and igy educated workers requires assumptions about te size of te conceament costs. f conceament costs are iger for educated workers, i.e. κ >κ m, te officia unempoyment rate is aways ower for igy educated workers tan for manua workers, i.e. u o < uo m. Tis is aso wat is observed in data. However, if furtermore, y /y ψ >,, m, and ence te informa sector is smaer tan te forma sector, te actua unempoyment rate is

12 Kom and Larsen ZA Journa of Labor Economics 06 5:8 Page of 36 iger for te igy educated workers, u > u m, i.e. in tis case, manua workers ave a ower actua unempoyment rate tan igy educated workers. Te foowing Proposition summarizes te resuts. Proposition Te officia unempoyment rate is ower for igy educated workers, u o < uo m, if tey face iger conceament costs, κ >κ m. Te actua unempoyment rate is iger ower for igy educated workers, u > u m u < u m,ifteyfaceigerconceament costs κ > κ m and tese conceament costs are ig ow enoug to induce y /y ψ > y /y ψ <,, m. or proofs of a te Propositions, see te Appendix. Te actua and te officia tota number of unempoyed workers are given by U TOT êu m + ê u, U o TOT êuo m + ê u o. 4 Comparative statics Tis section is concerned wit te impact of more severe punisment of informa activities on abour market performance and educationa attainment. We ony consider fuy financed canges in enforcement poicies. Hence, canges in te audit rate and te punisment fees are aways foowed by adjustments in te tax rate so as to baance te government budget constraint given by ên m w m z +ên m w m pα + ên w z+ ên w pα R. Rewriting tis budget constraint in terms of producer wages using ωm w + z and ω w + pα + κ, m, yieds zên m w m + z + pαên m w m + pα + κ m + z ê n w + z + pα ê n m w + pα + κ R 0 were R isteexogenousrevenuerequirement. rom 40, it foows tat an increase in te audit rate or te punisment fee, p or α,or an increase in te tax rate, z, wi, for a given tax base, aways increase government revenues. Te tax base may, owever, fa and tereby reduce revenues. f we assume tat we are ocated on te positivey soped side of te Laffer curves, te anaysis is straigtforward. Suc an assumption impies tat te direct effect of taxation and punisment on government revenues wi aways dominate te impact on revenues since te tax base may be reduced. An increase in te audit or punisment rate ten aways cas for a reduction in te tax rate in order to regain a baanced government budget. A fuy financed increase in te punisment of te informa sector ten induces ψ to increase bot because pα increases and because z fas. Atoug te most ikey scenario is wen iger punisment rates ca for tax reductions in order to fufi te government budget, te resuts obtained in tis section for te impact of iger reative punisments of informa activities on producer wages, tigtness, searc, empoyment, and unempoyment rates do not depend on tis assumption. Te reason is tat tese variabes are ony affected by te wedge, ψ, and not directy by z and pα. However, wic wi become cear, as educationa attainment coud be discouraged by a direct increase in taxation wic, in turn, may ave a compositiona effect on tota

13 Kom and Larsen ZA Journa of Labor Economics 06 5:8 Page 3 of 36 unempoyment, te repercussions troug te government budget constraint wi be of importance for tese variabes. To iustrate tis, we discuss te potentia scenario were te government revenue fas as arser punisment of te informa sector is impemented, and te government needs to increase te tax rate in order to baance te budget. 7 n tis case, wen tere is a simutaneous increase in z and pα, tere wi be ess sizeabe adjustments in te abour market outcome variabes for exampe, producer wages, ω j, and empoyment rates, nj as tese variabes are ony affected by te wedge, ψ, wic is not atered as muc wen bot z and pα increase. Te tax base adjustment of importance in tis case is ten te number of educated workers. Te stock of educated workers is affected by te reform bot because te wedge is atered but aso directy as z enters into 37 for given wedges. Wit te effect working troug te wedge being smaer in tis case, te iger tax rate is reducing te incentives to acquire iger education troug te direct effect. Tis tax base adjustment ten reduces tax revenues. However, as ong as te direct impact on R dominates te negative effect on te tax base troug ess education, te increase in z wi baance te government budget. Tis scenario wi not ater te resuts in te abour market anaysis considering te effect of more ars punisment of te informa sector on producer wages, tigtness, searc, empoyment, and unempoyment rates. Te reason is, as said, tat tese variabes are ony affected by te wedge, ψ. Te required increase in z in te above considered case ony impies tat ψ increases by ess tan if z was reduced, and te effect on te variabes wi be ess sizeabe. n fact, even if z increases to suc an extent tat ψ actuay fas, te resuts wi od. 8 Moreover, it is of no importance for te resuts wic side of te Laffer curves we are ocated on. However, te repercussions troug te government budget constraint wi be of importance for educationa attainment and tus for te composition of unempoyment. n te budget constraint in 0, potentia auditing costs are eft out. To incude auditing costs wi not affect any of te Propositions we derive in tis section. However, it affects te wefare anaysis as it tends to favour costess taxation and punisment fees at te expense of auditing. Te impications for te case of auditing costs is sown in te Appendix. 4. Sector aocation Atoug te resuts of fuy financed punisment of informa activities in Propositions and 3 od irrespective of ow te government budget restriction is affected, to stress te intuition, we present te resuts based on te standard case wen an increase in p or α increases reative punisment ψ. 9 Te effects on te aocation of searc and empoyment across te forma and te informa sector are summarized in te foowing Proposition. Proposition A fuy financed increase in te reative punisment of te informa sector, ψ, wi reaocate searc intensity and empoyment towards te forma sector, i.e. fas, n increases, and n fas. More zeaous enforcement wi make informa work ess attractive, inducing unempoyed workers to reaocate teir searc effort towards te forma sector. n addition, wen searc is reaocated towards te forma sector, te wage bargaining position

14 Kom and Larsen ZA Journa of Labor Economics 06 5:8 Page 4 of 36 strengtens for firms in te forma sector wereas it fas for firms in te informa sector. Te ower producer wages in te forma sector stimuate forma firms to open vacancies, wie at te same time, informa firms are discouraged to open new vacancies as tey now face iger producer wages. As a consequence tat bot vacancies and searc effort are reaocated towards te forma sector, te forma sector empoyment rate increases at te expense of informa empoyment. Tese mecanisms can expain te empirica findings in Ameida and Carneiro 0 wo use data on inspections carried out in Brazi. 4. Unempoyment rates As became cear in Proposition, empoyment in te forma sector increases at te expense of empoyment in te informa sector foowing more severe punisment of te informa sector. Wie tis is somewat expected, it is a priori not cear wat woud appen to te unempoyment rates. We ave te foowing resuts: Proposition 3 A fuy financed increase in te reative punisment of te informa sector, ψ, wi aways cause te officia unempoyment rate u o to fa, wereas te actua unempoyment rate u increasesif y /y ψ > fas if y /y ψ <. Te actua unempoyment rates increase wit more severe punisment of informa work if y /y ψ >. Te reason for tis is tat te arge conceament costs discourage workers from searcing, and firms from opening vacancies, in te informa sector. ncreased punisment of te informa sector wi encourage furter reaocation of searc and workers away from te informa sector, were reativey efficient searc metods are used, towards te forma sector. Tota searc efficiency ten fas, inducing unempoyment to increase. Te fact tat searc becomes ess efficient wen reaocated towards te forma sector aso as an impact on unempoyment working troug wage formation and tigtness. As searc is reaocated towards te forma sector, te wage demand is moderated in te forma sector and exaggerated in te informa sector. As te efficiency of searc in te forma sector increases by ess tan te efficiency of searc in te informa sector is reduced, te informa sector wage pus wi dominate te forma sector wage moderation. Tus, te incentives to open up a vacancy in te forma sector subsides te disincentives to open up a vacancy in te informa sector; te forma sector tigtness wi increase by ess tan te informa sector tigtness fas wen y /y ψ >. Te opposite ods if y /y ψ <. n tis case, too muc searc and too many firms are aocated into te informa sector as tere is a reative cost advantage of producing underground. Tota searc efficiency woud ten improve wen te government tries to combat te informa sector. Te officia unempoyment rate aways fas wit more ars punisment of informa activities as workers to a arger extent become formay empoyed. n tis unempoyment measure, workers in te informa sector were counted as unempoyed to start wit. 4.3 Education rom 7, it is cear tat more severe reative punisment of te informa sector affects te number of educated workers as suc poicy increases ψ. Tis effect is furter reinforced if te tax rate is reduced in order to assure a baanced government budget as te increase in ψ is reinforced by a reduction in z. However,areducedpayrotaxratewi

15 Kom and Larsen ZA Journa of Labor Economics 06 5:8 Page 5 of 36 aso ave a direct effect on te stock of educated workers. More specificay, a reduction in te tax rate, z, for a given wedge, wi increase te number of educated workers. Tis foows as taxation is more armfu to ig income earners, and consequenty, a tax reduction wi improve te income reativey more for ig income earners. However, before considering repercussions working troug te budget constraint, et us first consider te impact of a more zeaous enforcement poicy on education, for a given tax rate. We ave te foowing resuts: Proposition 4 An increase in te audit rate, p, or in te punisment rate, α,wicten increases ψ, wi increase reduce te number of educated workers if te reative productivity of education is in te foowing range y /y m [ o m /o, gκ, κ m o m /o ] y /y m gκ, κ m o m /o, were gκ, κ m > if κ >κ m and y /y y m /y m. Proof. We know from above tat te existence of an interior soution of ê requires tat y /y m > o m /o. Differentiating te educationa equation wit respect to expected punisment reveas tat te impact on education is determined by te sign of y m do m /d pα y do /d pα wic is equa to te sign of y /y m gκ, κ m o m /o, were te term gκ, κ m is arger tan for κ >κ m and g κ, κ m A γ y γ ψ y A m γ y m m m y ψ γ m m + ψ + ψ m > for κ >κ m and y y y m y, m were A + o / /ψ + y /y o. See te Appendix for te fu proof. Q.E.D. Te impact of a more zeaous enforcement poicy on educationa attainment depends on ow attractive underground work is to manua and educated workers. Wen conceament costs are iger for igy educated workers, more zeaous enforcement poicies tend to induce more workers to educate temseves. Tis foows as κ >κ m impies tat manua workers to a arger extent face informa abour market opportunities. Terefore, more zeaous enforcement poicies, wic make it ess attractive to work in te informa sector, wi be more armfu to manua workers. Tis effect may, owever, be counteracted by te fact tat igy educated workers ave iger productivity and terefore earn iger wages. As aso informa activities are igy productive for tese workers, tis impies tat more ars punisment, in tis perspective, is more armfu for te igy educated worker. Tus, even if igy educated workers face ess informa empoyment opportunities, tese opportunities are more profitabe. Tis reduces educationa incentives. Wic of te two effects dominate wi tus depend on ow sizeabe te differences in informa empoyment opportunities and productivity are. f underground empoyment opportunities in an economy foremost are avaiabe to manua workers, more ars punisment of underground activities wi pus more workers into education, tus increasing te stock of educated workers in te economy. However, if tese empoyment opportunities to a arge extent aso are avaiabe for igy educated workers, arder punisment wi arm igy educated workers more as tese opportunities are more profitabe to productive workers. Tis eads to ess workers educating temseves.

16 Kom and Larsen ZA Journa of Labor Economics 06 5:8 Page 6 of 36 Note tat Proposition 4 ony provides te sufficient conditions for wen te educationa stock increases and wen it fas wit more ars punisment of te informa sector witout considering te financing of te reform. Provided tat we are ocated on te positivey soped side of te Laffer curve, we can concude te foowing: Proposition 5 f an increase in te audit rate, p, or in te punisment rate, α,increases te number of educated workers as given by Proposition 4, te financing of te reform wi furter reinforce te increase in te stock of educated workers if z needs to fa so as to baance te government budget. Tis simpy foows as taxation as a direct effect is more armfu for ig income earners, and consequenty, a tax reduction, in order to maintain a baanced government budget, wi be more beneficia for ig income earners, tus encouraging educationa attainments. 4.4 Unempoyment rom Propositions 3, 4, and 5, it foows tat more severe punisment of te informa sector potentiay increases te tota number of unempoyed workers. f te forma sector is arger tan te informa sector, te unempoyment rates for bot manua and igy educated workers are augmented. Moreover, if informa empoyment opportunities to a significanty arger extent are avaiabe for manua workers, more workers wi attain iger education wen informa activities are punised more severey. Tis tends to increase tota unempoyment as te actua unempoyment rate, incuding informa work, is iger for igy educated workers. Aso, reca tat tis reaocation effect is reinforced if we are ocated on te positivey soped side of te Laffer curve. Tus, in tis case, tota unempoyment increases bot because te unempoyment rates for a workers increase and because workers are reaocated towards te sector were te unempoyment rate is igest. More generay, te Proposition summarizes te resut: Proposition 6 A fuy financed increase in te audit rate, p, or in te punisment fee, α, increasesdecreases te number of unempoyed workers if te reative productivity of [ ] education is in te foowing range y om y m o, g κ, κ m o m o were g κ, κ m > if κ > κ m, were te financing of te reform furter reinforces te reaocation effect if z needs to fa so as to baance te government budget. 5 Wefare Tis section is concerned wit wefare anaysis and te optima design of punisment poicies. As sown above, increasing te punisment fees or te audit rates affect te number of educated workers as we as te aocation of searc and jobs across te forma and informa sectors. Tis is essentia wen considering te impact on wefare. or simpicity, we ere et y y,, m. Moreover, as te Hosios condition ods by assumption, as we ave assumed tat te easticity of te expected duration of a vacancy is equa to te bargaining power of workers in a Nas bargaining situation, we can disregard congestion externaities on te abour market. Moreover, we do not need to be concerned about inefficiencies in terms of too

17 Kom and Larsen ZA Journa of Labor Economics 06 5:8 Page 7 of 36 ow or too ig educationa attainments due to te odup probem as te abour markets for workers wit ig and ow education are separated. Tis enabes us to focus on oter, ess we-known, distortions in tis section. Ceary, owever, if, for exampe, te Hosios condition does not od, te tax and punisment poicies coud potentiay be used to correct for congestion externaities. Te standard socia wefare measure, anaogous to te one described in, for exampe, Pissarides 000 under no discounting, is used and can be written as W êw m + ê W de, W m u m y m u m ky m m, W u y u ky ce, 3 were γ + γ, m,. Te wefare measure consists of aggregate production minus tota vacancy costs, i.e. note tat u k v + v k, m,, and minus te aggregate costs of education. Wit te assumption of risk neutra individuas, we ignore distributiona issues, and ence, wages wi not feature in te wefare function. See te Appendix for te derivation of tis wefare measure. Let us first derive te sociay optima coice of tigtness, searc, and stock of educated workers by maximizing te wefare function in 3 wit respect to m, m,,, m,,andê. Te sociay optima soution is soved from te foowing seven conditions: 0 γ γ 0,, m,, 4 sk + y y m [ k γ ] 0, m,, 5 k γ c ê 0. 6 We can now compare te sociay optima soution wit te market outcome. rom, 4, and 5, it foows tat te market soution for searc and tigtness coincides wit te sociay optima aocation if te imposed tax and punisment poicy are suc tat ψ m ψ. Tis concusion is intuitive as any poicy tat induces a deviation of te ψ, m, from unity impies a favourabe treatment of te forma or te informa sector wic, in turn, induces a distortion in te sectoria aocation of searc and tigtness between te forma and informa sectors. or exampe, if searc to a arger extent is aocated to te forma or informa sector instead of te oter, te searc is inefficienty used as ess efficient searc metods in tat sector need to be used. Moreover, as discussed in reation to Proposition 3, a favourabe treatment of eiter te forma or te informa sector induces too many firms to open vacancies in tat sector witout accounting for te externaity tey impose on oters. n fact, unempoyment is minimized wen te aocation of searc and tigtness across te forma and informa sectors is equa, and so is vacancy costs. Tus, wefare is maximized wen searc and tigtness are aocated equay across te forma and te informa sector.

18 Kom and Larsen ZA Journa of Labor Economics 06 5:8 Page 8 of 36 Now et us compare te sociay optima stock of educated workers wit te educationa outcome induced by te market. As te market outcome in terms of sectoria aocation of searc and tigtness coincided wit te sociay optima one wen te government ets te market face ψ m ψ, we evauate aso te private outcome of education under tese conditions. Tis yieds te foowing market outcome of te stock of educated workers: y y m k + z γ c ê 0. 7 t immediatey foows tat a tax and punisment poicy wic impies tat ψ m ψ wi not provide incentives to te market to generate a sociay optima stock of educated workers. Comparing 6 and 7, in fact, reveas tat te market outcome induces too few workers to educate temseves if forma and informa sector jobs face uniform treatment in terms of ψ m ψ. Tis foows as taxes, captured by +z in 7, it igy educated workers more severey tan manua workers, wic reduces te incentives of education. rom tis, we can concude tat wefare woud increase if more workers cose to educate temseves wen ψ m ψ. Tis discussion brings us to te government s expicit coice of tax and punisment poicy. How soud te government punis informa work in order to maximize wefare? 5. Optima punisment poicy Te wefare anaysis above indicates tat it may be optima to punis tax-evading activities carried out by manua workers more severey tan tose carried out by igy educated workers. or exampe, if conceament costs are iger for igy educated workers, a punisment poicy wit ψ m ψ is ony possibe if te manua workers to a arger extent tan igy educated workers face punisment of informa activities. Tat is, pα as to be set reativey iger for manua workers if κ m <κ in order to induce ψ m ψ. Tis raises te question of weter it is possibe or not to target te punisment fees and audit rates towards te sector empoying manua vs igy educated workers. Wie governments potentiay coud, and in fact do, 3 target teir audits to specific sectors, i.e. aowing for p m to differ from p, tis is ess ikey te case for te fee rates. To find te sociay optima coice of audit rates for te sector empoying manua workers and te sector empoying igy educated workers, te wefare function in 3 is maximized by te coice of p m and p subject to te market reactions given by, 4, 5, 7, and 9 and te government budget restriction in 0. Tis yieds te foowing first-order conditions: ê m dψ m + d e 0, 8 dp m dψ m dp m d e dp m ê dψ + d e, 9 dp dψ dp d e dp

19 Kom and Larsen ZA Journa of Labor Economics 06 5:8 Page 9 of 36 [ were dψ d j j, d j dψ + d ] d dψ, j m,. Evauating te first-order conditions at te eves of p m and p ensuring tat ψ m ψ turns out to be very convenient and gives ψm dp m d ê d ê > 0, 30 dp m ψ dp d ê d ê < 0. 3 dp Provided tat we are ocated on te positivey soped side of te Laffer curves, we can concude tat Proposition 7 Wefare is maximized wen te sector empoying manua workers is audited to a arger extent tan te sector empoying igy educated workers, i.e. p m > p so as to get ψ < <ψ m if κ κ m. Proof. Evauate te first-order conditions 8 and 9 at ψ m ψ. rom te sociay optima aocation of searc and tigtness, ψ impiestat d 0, m,.ten dp m ψm d e d e dp m > 0and dp ψ d e d ê > 0 from 6 and 7 and d e d ê dp d d d e dp < 0 as dp m > 0, < 0 from 7. Tus, wefare improves by reaocation of audits towards te manua sector. f κ κ m, p m p at ψ m ψ, wefare improves by setting p m > p.fκ >κ m, te resuts are reinforced as p m > p aready wen ψ m ψ, and wefare improves by furter increasing p m and reducing p. Q.E.D. Te resut in Proposition 7 foows straigtforwardy from te first-order conditions wen evauated at te p m and p wic induces ψ m ψ. Te first term on te rigtand side of Eqs. 8 and 9 ten disappears as te distortions in searc and aocation of tigtness across te forma and te informa sector are fuy eiminated. n tis case, tere are no oter distortions present except tat too few workers ave cosen to educate temseves. Reca tat tis is a consequence tat taxation arms ig income earners reativey more. Tis distortion can, owever, be corrected for by increasing te audits in te manua sector and reducing tem in te sector for igy educated workers, wic is captured by te rigt-and side in 30 and 3. As informa sector work for manua workers becomes ess attractive wen te government increases te number of audits, manua workers are encouraged to acquire iger education. Simiary, ess audits in te igy educated sector furter encourages workers to acquire iger education. f conceament costs are iger in te sector empoying igy educated workers, i.e. κ > κ m, tere are even furter incentives for te government to focus teir audits on te manua sector. Tis foows as ig conceament costs work as a sef-reguating punisment of informa sector activities. Tus, if conceament costs are iger in te sector empoying igy educated workers, tis sector wi be in ess need of audits as conceament costs wi do part of te job of imiting te size of te informa sector. Moreover it foows tat