adb economics working paper series

Transcription

1 Labor market regulations in the ontext o strutural transormation Priya Ranjan, Rana Hasan, and Erik Jan Eleazar NO. 543 April 018 adb eonomis working paper series ASIAN DEVELOPMENT BANK

2 ADB Eonomis Working Paper Series Labor Market Regulations in the Context o Strutural Transormation Priya Ranjan, Rana Hasan, and Erik Jan Eleazar No. 543 April 018 Priya Ranjan (pranjan@ui-edu) is a proessor at the University o Caliornia, Irvine; Rana Hasan (rhasan@adb.org) is a diretor; and Erik Jan Eleazar (eeleazar.onsultant@adb.org) is a onsultant at the Eonomi Researh and Regional Cooperation Department o the Asian Development Bank We would like to thank the seminar partiipants at LSU, ISI Delhi, Jadavpur University, and the partiipants at the Asian Development Bank Conerene on "Future o Jobs in Asia," held in Manila in 016, or valuable omments. ASIAN DEVELOPMENT BANK

3 Creative Commons Attribution 3.0 IGO liense (CC BY 3.0 IGO) 018 Asian Development Bank 6 ADB Avenue, Mandaluyong City, 1550 Metro Manila, Philippines Tel ; Fax Some rights reserved. Published in 018. ISSN (print), (eletroni) Publiation Stok No. WPS DOI: The views expressed in this publiation are those o the authors and do not neessarily relet the views and poliies o the Asian Development Bank (ADB) or its Board o Governors or the governments they represent. ADB does not guarantee the auray o the data inluded in this publiation and aepts no responsibility or any onsequene o their use. The mention o speii ompanies or produts o manuaturers does not imply that they are endorsed or reommended by ADB in preerene to others o a similar nature that are not mentioned. By making any designation o or reerene to a partiular territory or geographi area, or by using the term ountry in this doument, ADB does not intend to make any judgments as to the legal or other status o any territory or area. This work is available under the Creative Commons Attribution 3.0 IGO liense (CC BY 3.0 IGO) By using the ontent o this publiation, you agree to be bound by the terms o this liense. For attribution, translations, adaptations, and permissions, please read the provisions and terms o use at This CC liense does not apply to non-adb opyright materials in this publiation. I the material is attributed to another soure, please ontat the opyright owner or publisher o that soure or permission to reprodue it. ADB annot be held liable or any laims that arise as a result o your use o the material. Please ontat pubsmarketing@adb.org i you have questions or omments with respet to ontent, or i you wish to obtain opyright permission or your intended use that does not all within these terms, or or permission to use the ADB logo. Notes: In this publiation, $ reers to United States dollars. Corrigenda to ADB publiations may be ound at

4 I. CONTENTS TABLES AND FIGURES ABSTRACT iv v I. INTRODUCTION 1 II. RELATED LITERATURE 3 III. THE MODEL 5 IV. PLANNER'S PROBLEM 9 V. LABOR MARKET REGULATIONS 1 A. Severane Payments/Unemployment Insurane 1 B. Nonompliane, State Capaity, et. 14 C. Administrative Burden o Firing 15 D. Inrease in General Regulatory Burden 18 E. Minimum Wage Regulation 0 VI. EMPIRICAL EXERCISE 3 A. Data and Variables 3 B. Results 7 VII. CONCLUDING REMARKS 3 APPENDIX 33 REFERENCES 41

5 TABLES AND FIGURES TABLES 1 Summary Statistis 6 Regression Results, Dismissal Regulations 9 3 Regression Results, Redundany Compensation 30 4 Regression Results, Minimum Wage to GDP Ratio 31 FIGURES 1 Risk Aversion and Labor Market Outomes 11 Severane Payments and Labor Market Outomes 13 3 Administrative Burden o Firing and Labor Market Outomes 16 4 Regulatory Burden and Labor Market Outomes 19 5 Minimum Wage and Labor Market Outomes 6 Satter Plots: Share o Inormal Employment 7

6 ABSTRACT This paper onstruts a theoretial model to study labor market regulations in developing ountries within the ontext o strutural transormation. When workers are risk averse and the market or insurane against labor inome risk is missing, regulations that provide insurane to workers (suh as severane payments) are eiieny enhaning and promote strutural transormation. However, regulations that simply reate barriers to the dismissal o workers not only impede strutural transormation, they also end up reduing the welare o workers. The impliations o some other issues like general regulatory burden, weak state apaity, and minimum wage regulations are analyzed as well. The paper provides some empirial evidene broadly onsistent with the theoretial results using ross-ountry data. While dismissal regulations inrease the share o inormal employment, severane payments to workers do not. Key words: dismissal regulations, inormal employment, minimum wage, severane payments, strutural transormation JEL odes: J38, J46, O1, O17, O57

7 I. INTRODUCTION Strutural transormation is an integral proess o eonomi development whereby resoures move rom less produtive ativities to more produtive ativities. This an happen through intersetoral realloation o resoures in whih ase resoures move rom say, less produtive agriulture to more produtive manuaturing. Alternatively, strutural transormation an also take plae through intrasetoral realloation o resoures. An example o the latter would be a derease in the share o inormal/unorganized manuaturing and inrease in the share o ormal/organized manuaturing. In this paper s theoretial model we ous on the latter aspet o strutural transormation. The literature on inormality identiies two key auses o it. One has to do with the motive o evading taxation. That is, by remaining inormal, irms an avoid taxation. The other explanation is in terms o exessive regulation and labor market rigidities. The idea is that the regulation o labor markets in developing ountries an have elements, suh as restritions on iring or high minimum wages, that indue irms to avoid meeting these regulations by remaining inormal/unregistered. Again, this paper ouses on the latter without denying the importane o the ormer. Governments in developing ountries attempt to regulate labor markets along a number o dimensions suh as wages, hours o work, hiring and iring restritions, and onditions o work. When analyzing the desirability o any regulation, inluding those in the labor market, a natural question to ask is i the beneits o these regulations exeed the ost. This immediately suggests the need to develop a ramework whih will allow us to analyze the optimality o dierent labor market regulations. The standard ramework to study labor market regulations in developed ountries is a one-setor model that analyzes how a regulation aets the demand and supply or labor, and onsequently the aggregate output (e.g., Summers 1989). This may not be appropriate in a developing ountry ontext, however. An important eature o developing ountries is that a large ration o the workore is employed in inormal/unorganized irms and as mentioned earlier, it is ommonly believed that the exessive regulation o labor markets is one o the key ators responsible, preventing resoures rom moving to more produtive ormal irms. 1 The aim o this paper is to provide an analytial ramework to analyze labor market regulations in developing ountries and study how they aet the intrasetoral alloation o resoures between inormal and ormal irms. In addition, it also provides some suggestive empirial evidene broadly onsistent with the theoretial results. In most theoretial works dealing with labor market poliies, to be disussed later, workers are assumed to be risk neutral, in whih ase there is no role or soial protetion. Poliies suh as employment protetion or unemployment insurane are then viewed as distortionary by either raising the ost or employers or reduing the job searh inentives or workers. Suh models learly understate the beneits o labor market regulations beause i workers are risk neutral there is no need to insure them against labor market risk through soial protetion. Thereore, we onstrut a theoretial ramework with two key eatures: risk-averse workers and the existene o an inormal/unorganized setor. These two eatures allow us to study labor market regulations in a developing ountry ontext. Another eature o our model is job destrution by irms. This allows us to study labor market poliies relating to iring o workers. 1 National Commission on Employment in the Unorganized Setor ound that in , out o a total employment o million workers in the Indian eonomy, million were in the inormal setor (Kanbur 017). Aording to Ghani, Kerr, and O'Connell (013), the employment share o the unorganized setor in Indian manuaturing has remained at around 81% between 1989 and 005.

8 ADB Eonomis Working Paper Series No. 543 In our model, there is a mass o entrepreneurs with heterogeneous entrepreneurial ability who an set up a irm in the inormal setor using a simple tehnology with diminishing returns. Alternatively, they an set up a irm with a more sophistiated tehnology in the ormal setor. There is a mass o risk averse but homogeneous workers who an work in either setor. While workers have idential produtivities in the inormal setor, the produtivity in the ormal setor is math speii. Firms in the ormal setor ommit to a level o wage in advane (similar to the ompetitive searh literature) and hene ire workers with very low math-speii produtivity. This is a way to apture endogenous job destrution in a stati model and is similar in spirit to the ramework o Blanhard and Tirole (008). Workers who are ired in the ormal setor get utility rom home prodution/leisure. In equilibrium, workers are indierent between working in the ormal setor with a risk o unemployment or in the inormal setor with a guaranteed but low-paying job. Thereore, wages are neessarily higher in the ormal setor. To see the role played by risk aversion in driving the results, we irst derive results or the benhmark ase o risk-neutral workers. It is shown that i workers are risk neutral then the deentralized outome is prodution eiient. However, when risk-averse workers ae the risk o unemployment in the ormal setor and markets or insurane against labor inome risk are missing, the deentralized outome is not eiient. Moreover, employment in the ormal setor and aggregate output both are dereasing in the risk aversion o workers. Introduing poliies that provide insurane to risk-averse workers is eiieny enhaning. It is shown that i the government an mandate severane payments to ired workers, both the aggregate output and the amount o labor employed in the ormal setor inrease. Alternatively, an unemployment insurane program unded by a layo tax ahieves the same outome. While the two are equivalent theoretially, one or the other may be superior in pratie. Mandated severane payments may suer rom nonompliane, either willul or beause o the inability o a irm to pay it during rises. In this ase, an unemployment insurane unded by a layo tax may be preerable. The United States (US) unemployment insurane program with experiene-rated ontribution has this eature. The downside o state-provided unemployment insurane is that it may run into weak state apaity. In an extension, we apture weak state apaity by assuming that when government ollets money rom irms, only a ration is given bak to workers as beneits. The remaining ration is lost due to orruption, mismanagement, et. Clearly, leakage makes the poliy ineetive and i the leakage is suiiently high, no unemployment insurane may be better than poorly provided government unemployment insurane. Sine workers are aing a risk o unemployment, one may be tempted to onlude that a poliy that redues job destrution in the ormal setor would be welare improving. Employment protetion poliies that make it diiult or irms to ire workers would be one suh poliy. It turns out that suh poliies are ounter produtive. While they redue iring by irms in the ormal setor, they end up worsening the existing distortion and redue aggregate output. Thereore, while the provision o unemployment insurane an aid the proess o strutural transormation, iring restritions that reate administrative hurdles or iring impede the proess o strutural transormation. It is also shown that an inrease in regulatory burden that makes it ostly or irms to enter the ormal setor redues the size o the ormal setor in addition to reduing aggregate output. Finally, the model is extended to study the impliations o minimum wage regulations in this setting. A binding minimum wage in the ormal setor redues the size o the ormal setor thereby impeding strutural transormation. However, output and worker welare inrease initially beore dereasing. Sine the job destrution in the baseline model is suboptimally low, a binding minimum wage inreases job

9 Labor Market Regulations in the Context o Strutural Transormation 3 destrution whih is a soure o gain. However, this is oset by the distortionary eet o minimum wage on the proit o entrepreneurs. A binding minimum wage redues the proit in the ormal setor, and hene, the mass o entrepreneurs in the ormal setor is dereasing in minimum wage. The latter eet starts dominating soon, and it turns out that a binding minimum wage doesn't ahieve prodution eiieny in the model. In addition to the theoretial model, the paper also provides some empirial evidene on the relationship between labor market regulations and the share o inormal employment using rossountry data. The share o inormal employment is obtained rom the International Labour Organization s (ILO) Key Indiators o the Labor Market (KILM) database while the measures o labor market regulations are obtained rom the data set developed by the Centre or Business Researh o the University o Cambridge. The key empirial result is that striter dismissal regulations and higher minimum wages inrease inormality but more generous severane payments do not. These empirial results are broadly onsistent with the theoretial results. The remainder o the paper is organized as ollows. The next setion provides a brie survey o the related literature. Setion III sets up the baseline model and derives the deentralized equilibrium. Setion IV sets up the planner's problem to study the eiieny properties o the deentralized outome. Setion V studies labor market regulations whih is ollowed by empirial evidene in setion VI. Setion VII provides onluding remarks. II. RELATED LITERATURE Fugazza and Jaques (004) onstrut a theoretial model where the inormal setor doesn't omply with regulations and tax ontributions. Workers ae a subjetive ost o working in the inormal setor. There is direted searh by workers, and thereore, workers with low subjetive ost work in the inormal setor while those with high ost work in the ormal setor. In equilibrium, irms are indierent between opening vaanies in the two setors. The key result is that an inrease in tax or soial seurity ontributions inreases inormality while inreased enorement redues inormality. In Zenou (008) there are two setors: ormal with searh ritions and inormal with ompetitive wages. Workers are idential and reely mobile between the two setors. In the inormal setor, there is diminishing returns to labor whih makes the supply o labor to eah setor endogenous. In addition to the searh ritions, the ormal setor has an entry ost or vaany ost and unemployment beneits. He perorms omparative statis with respet to unemployment beneits, hiring osts, et. and inds the expeted results on the inidene o inormality. In de Paula and Sheinkman (007), agents hoose to beome a ormal setor entrepreneur or an inormal setor entrepreneur or a worker. Agents dier in their entrepreneurial ability. Firms in both setors hire apital and the ost o apital is lower or ormal setor irms but they also have to pay a tax. Inormal entrepreneurs an evade taxation but they are aught with a positive probability in whih ase the entire output is seized. The probability o getting aught depends on the size o apital. Thereore, irms requiring more apital are in the ormal setor while others are in the inormal setor. The highest ability entrepreneurs set up irms in the ormal setor, middle ability entrepreneurs in the inormal setor and those with lowest entrepreneurial ability deide to beome workers. The authors perorm omparative stati exerises with respet to the rate o taxation and the rate o interest.

10 4 ADB Eonomis Working Paper Series No. 543 Albreht, Navarro, and Vroman (009) introdue ex-ante worker heterogeneity with respet to worker produtivity in the ormal setor and only high-produtivity workers work in the ormal setor while others work in the inormal setor. There is exogenous job destrution in the inormal setor as in a standard Pissarides model but the job destrution is endogenous in the ormal setor in response to an idiosynrati produtivity shok. Firms have to pay a severane tax in the ormal setor in addition to a payroll tax. Severane tax inreases average employment duration in the ormal setor, redues overall unemployment, redues the number o ormal setor workers, and redues the number o workers who aept any type o oer. A payroll tax redues average employment duration in the ormal setor, redues the number o ormal setor workers, and inreases the size o the inormal setor and the number o workers aepting any type o oer. Total unemployment rises. Severane tax dereases average produtivity while the payroll tax inreases it, but under both poliies, net output alls. In Bosh and Esteban-Pretel (01) ormal employment ontrats involve a hiring ost, payroll tax, and a iring tax upon job destrution while inormal jobs do not inur these osts but are destroyed at a higher rate. Inormal jobs are also subjet to government monitoring. They are deteted with some probability, in whih ase the job is destroyed and a ine is paid. Inormal workers are allowed on the job searh or a ormal job. Inormal jobs arise beause or some low produtivity mathes it is optimal or irms and workers to establish an inormal relationship that irumvents ormal regulations. Ongoing ormal and inormal mathes draw new idiosynrati produtivity at dierential rates (the rate being higher in the inormal setor aounting or the higher job destrution there). In this setting, poliies that derease the ost o ormal jobs, or inrease the ost o inormality, raise the share o ormal employment while reduing unemployment. In Meghir, Narita, and Robin (015), irms in the ormal setor are subjet to orporate tax on proits, soial seurity ontributions or their workers, and severane payments upon laying them o, and have to pay a minimum wage. Workers in the ormal setor are subjet to inome tax and are eligible or unemployment insurane unded by taxes. Firms in the inormal setor are monitored and i aught, pay a ine. Firms dier in their produtivity, post a wage and hoose a setor in whih to post vaanies. Workers are homogeneous and meet with irms randomly and aept or rejet the wage oer. It is shown that tighter enorement leads to less inormality and higher output. As ar as the empirial evidene on the impat o labor regulations on inormality is onerned, Chatterjee and Kanbur (014) use irm-level data rom the Annual Survey o Industry (ASI) and National Sample Survey Organisation (NSSO) to study inormality in India. While the ASI data aptures irms registered under the Fatories At whih requires irms with more than 10 workers to be registered the NSSO data is a survey o unregistered irms. Any irm in the NSSO data with 10 or more workers is nonompliant with the Fatories At. There may also be some NSSO irms with nine workers who stay small and inormal preisely to evade regulations. The authors ind that the NSSO irms with employment o nine or more workers onstitute % o all irms and 11% o employment. To sum up, the existing literature either doesn't have job destrution by irms and hene annot talk about employment protetion-related poliies or i they allow or job destrution as in Meghir, Narita, and Robin (015), then workers are risk neutral in whih ase there is no need or employment protetion. That is, while several papers in the literature examine the role o employment protetion poliies on inormality and aggregate output, the rationale or employment protetion (risk The NSSO irms with 10 or more workers aount or 1.5% o all irms and 9.8% o employment. Thereore, the number o irms that are nonompliers is muh larger than the number that keep their size low to avoid regulations.

11 Labor Market Regulations in the Context o Strutural Transormation 5 aversion with missing insurane market) is missing in these studies. Hene, it is diiult to arry out a normative exerise involving employment protetion poliies. III. THE MODEL There are two setors in the eonomy: a ormal setor with a sophistiated prodution untion and an inormal setor with a simpler prodution untion. The number o workers is given by L. Workers an work in the ormal setor or inormal setor. Workers are risk averse and their utility untion is given by U( ), with U( ) 0 and U( ) 0. There is a unit mass o entrepreneurs with the distribution o entrepreneurial ability given by G( ). Entrepreneurs have to deide whether to set up a irm in the ormal setor or inormal setor. Firms in the ormal setor ae a ixed ost o F. A part o this ixed ost ould arise rom meeting the regulatory requirements o the state (the so-alled ost o ormalization). The other part o the ixed ost should be thought o as the ost o using a more sophistiated tehnology. Thereore, in our setting, inormality an arise both due to regulatory burden and the unproitability o modern tehnology or low ability entrepreneurs. The prodution untion in the inormal setor is given by Z B L (1) 1 i( ) i( ) ; 0 1 B where B' aptures the general produtivity o the inormal setor and is given by B where 1 (1 ). The normalization o the onstant produtivity term in the above ashion redues notational lutter. is the entrepreneurial ability. 0 1 aptures the diminishing returns to labor and allows or a determinate irm size. The wage in the inormal setor is w i whih is determined ompetitively. Denote the proit o an entrepreneur in the inormal setor by i( ). The optimal hoie o labor is given by 1 B L ( ) w () 1 1 i i Thereore, the amount o labor hired by an entrepreneur with ability is given by 1 1 Li( ) Bw (3) i 1 The proit o an inormal setor entrepreneur with ability is given by ( ) B L( ) wl( ) Bw (4) 1 1 i i i i Clearly, i( ) is inreasing in and dereasing in w i. The ormal setor prodution untion is more ompliated. Here the produtivity o a worker is math speii. Denote the math-speii produtivity o a worker by whih is drawn rom a

12 6 ADB Eonomis Working Paper Series No. 543 uniorm distribution over [0.1]. One the math-speii produtivity o a worker is revealed, the irm an deide whether to retain the worker or ire them. I irms use a uto rule whereby they retain workers with produtivity above and ire others, then the average produtivity o retained workers is 1. I they hire L workers then they retain (1 ) L o them, and hene the amount o labor in eiieny units that is used in prodution is The prodution untion is e 1 L L (5) 1 1 Z( ) A L ( ) (6) A. is the manager- where A is the ommon produtivity term and we use the normalization speii produtivity term. 1 A A ormal setor irm announes a ontrat ( w, ) whih must satisy the ollowing partiipation onstraint o workers. (1 ) U( w ) U( z) U( w ) (7) i That is, a worker expets to get a wage o w with probability 1 and z (ash equivalent o home prodution or leisure) with probability (in the event that the worker is ired). The expeted utility rom partiipating in the ormal setor must at least be as high as the utility rom working in the inormal setor U( w i ). We are assuming that the ired workers annot work in the inormal setor. It is as i workers at the beginning o eah period in our stati model make a deision whether to work in the inormal setor or look or a job in the ormal setor with the possibility o being ired and staying unemployed. 3 Thereore, a irm in the ormal setor solves the ollowing maximization problem: ( ) (1 ) ( ) st..(1 ) U( w ) U( z) U( w ) 1 1 Max A L w L L ( ), w, i 3 The assumption is similar to the assumption in Zenou (008) where workers an either work in the inormal setor or searh or a job in the ormal setor. In his dynami model, the value o unemployment in the ormal setor equals the value o a job in the inormal setor. Also, or the problem to be meaningul, it must be the ase that w i z So, the workers who are ired in the ormal setor suer a loss and thereore must be ompensated through a higher ormal setor wage.

13 Labor Market Regulations in the Context o Strutural Transormation 7 Denoting the Lagrangian multiplier assoiated with the onstraint by, the irst-order onditions are L ( ): A L ( ) w (1 ) (8) w : 1 L 1 U w 0 (9) : A LF U wuz wl (10) Using ( 9 ) write ( 10 ) as ( ) A L w (11) where U( w ) U( z). U( w ) Thereore, or a ormal setor irm with produtivity, the three endogenous variables L, w, and are determined by the ollowing three equations: A L ( ) w (1 ) ( ) A L w (1) (13) or a given w i. Note that (1) and (13) imply (1 ) U( w ) U( z) U( w ) (14) i 1 w w (15) Equations (14) and (15) determine w and independent o the irm produtivity parameter. That is, eah ormal setor irm oers the same w, ontrat. The amount o labor hired by a ormal setor irm is 1 1 L ( ) A w (1 ) (16)

14 8 ADB Eonomis Working Paper Series No. 543 Thus, in the model, a high-produtivity irm is larger but pays the same wage and has the same iring threshold. Denote the proit gross o the ixed ost o operations o a ormal setor irm by ( ) whih is given as ollows: 1 1 A w ( ) (17) The net return rom entrepreneurship in the ormal setor is ( ) F. The marginal entrepreneur with entrepreneurial ability is indierent between setting up a irm in the ormal setor or in the inormal setor, and this indierene ondition is given by ( ) i( ) A Bwi F w (18) Entrepreneurs with open irms in the inormal setor, while the rest open irms in the ormal setor. We assume that the parameters are suh that the ollowing ondition is satisied in equilibrium: A 1 1 B w wi 1 1 (19) The ondition above ensures that in the absene o the ixed ost F, all entrepreneurs would wish to set up irms in the ormal setor and is satisied i A is suiiently higher than B. In this ase, it is the ixed osts o setting up business in the ormal setor that prevents some low ability entrepreneurs to enter the ormal setor. As mentioned earlier, a part o the ixed ost is tehnologial and a part ould be regulation related. In this ase, there will be some inormality due to tehnologial reasons and the rest an be attributed to regulations. The total employment in the inormal setor is given by i I L L( ) dg( ) (0) where Li ( ) is given by (3). Similarly, the employment in the ormal setor is given by F L L ( ) dg( ) (1)

15 Labor Market Regulations in the Context o Strutural Transormation 9 where L ( ) is given in (16). Sine the total amount o labor in the eonomy is L, the ollowing must be true: I F () L L L I F The model above has the ollowing endogenous variables: w,, wi,, L, L. They will be untions o the exogenous variables: A, BF,, and L. Equations (14), (15), (18), (0), (1), and () determine the six endogenous variables o interest. I L and L F give us the sizes o the two setors in terms o employment. w and w i are the wages in the two setors, and is the unemployment rate in the ormal setor. Sine there is a possibility o unemployment in the ormal setor, the wage in the ormal setor must be higher than that in the inormal setor: w w. i I Denote the aggregate output in the inormal setor by Z and the aggregate output in the F ormal setor by Z. Net output or the eonomy, whih is our measure o prodution eiieny, is given by I F F Y Z Z L z (1 G( )) F (3) F where the term Lz is the home prodution o ired workers and the last term is the ixed ost o ormal setor irms. In the omparative statis below, our variables o interest will be the net output, Y, the share F o the ormal setor in labor ore, L, and the welare o workers, U( w ). i IV. PLANNER'S PROBLEM To show how the risk aversion o workers ombined with a missing market or insurane against labor inome risk reates prodution ineiieny in the eonomy, we set up the problem or a planner who is interested in maximizing the aggregate output (3). The planner will have to alloate both labor and entrepreneurs to the two setors to maximize aggregate output. The planner reognizes the limited span o ontrol o entrepreneurs (diminishing returns to labor) and also the ixed ost, F, that eah entrepreneur has to inur to start a business in the ormal setor. The planner undertakes the ollowing maximization exerise. 1 1 ( ) 1 B Li dg A L dg Max ( ), L ( ), Li( ), I F z L dg G F LL L ( ) ( ) ( ) ( ) ( ) ( ) ( ) (1 ( )) ( )

16 10 ADB Eonomis Working Paper Series No. 543 The details o the planner's maximization exerise are given in the appendix. What is notable is that the solution to the planner's problem orresponds to the deentralized outome when workers are risk neutral. That is, in the risk-neutral worker ase, the alloation o resoures between the ormal and inormal setors in the planner's problem orresponds to the deentralized outome. Thereore, in the baseline model, the distortion in the deentralized ase arises due to the risk aversion o workers. While we have proved analytially that when workers are risk averse, the deentralized outome is prodution ineiient, we have not answered the question o the diretion o ineiieny: Are there too many workers in the ormal setor or too ew workers? We veriy numerially that with risk-averse workers, the aggregate net output is lower than in the planner's problem and there are too ew workers in the ormal setor. More preisely, we veriy that the aggregate net output is dereasing in the degree o risk aversion as is the share o workore in the ormal setor. Additionally, the ration o entrepreneurs operating in the ormal setor is smaller as is the welare o workers. Finally, when workers are risk averse, irms hoose a muh lower ompared to what the planner hooses. These results are veriied in Figure 1. In Figure 1, as well as in all other numerial examples in the paper, we assume a onstant relative risk aversion utility untion 1 x 1 o the ollowing orm: U( x) where is the risk aversion parameter. In Figure 1, the planner's 1 outome orresponds to the risk-neutral outome (where 0). Figures 1a 1 show that the ollowing variables o interest are dereasing in the degree o risk aversion: aggregate net output, share o workore in the ormal setor, share o entrepreneurs in the ormal setor, rate o job destrution, rate o unemployment, and worker welare. Intuitively, when workers are risk averse and markets or insurane are missing, irms have to pay workers a higher wage to meet their reservation utility given that they may be ired. As a result, the level o hosen by irms in a deentralized equilibrium is less than what is optimal rom the prodution eiieny point o view. This also redues the proitability o irms in the ormal setor leading to ewer entrepreneurs and workers in the ormal setor. So ar, we have not allowed irms to oer severane payments voluntarily. We have restrited the ontrats oered by the irms in the ormal setor to a ( w, ) pair. What i irms an oer severane payments voluntarily or i the government mandates severane payments? We veriy below that prodution eiieny an be attained through severane payments.

17 Labor Market Regulations in the Context o Strutural Transormation 11 Figure 1: Risk Aversion and Labor Market Outomes Aggregate output Share o enterpreneurs in ormal setors Figure 1a: Risk aversion and aggregate output Risk aversion Figure 1: Risk aversion and share entrepreneur in ormal setor Risk aversion Figure 1e: Risk aversion and unemployment Share o labor ore in ormal setor Job destrution Figure 1b: Risk aversion and share labor ore in ormal setor Risk aversion Figure 1d: Risk Aversion and job destrution Risk aversion Figure 1: Risk aversion and worker welare Unemployment Risk aversion Worker welare Risk aversion A=8, B=1, L=1, γ= /3, z=.5, F=.75 Soure: Authors alulations.

18 1 ADB Eonomis Working Paper Series No. 543 V. LABOR MARKET REGULATIONS A. Severane Payments/Unemployment Insurane Suppose irms an oer severane payments voluntarily. This takes the orm o a payment, s, to eah ired worker. So, irms oer a ontrat ( w, s, ) where s is the severane payment promised to eah ired worker. Now, the irms in the ormal setor will do the ollowing maximization exerise: 1 1 Max A L( ) w (1 ) L( ) L( ) s L, w,, s st..(1 ) U( w ) U( sz) U( w ) i The equations or this ase are derived in the appendix. It is shown that the optimal hoie o severane payments by irms is haraterized by w s z. That is, irms ully insure workers against unemployment risk. This level o severane payments also ensures prodution eiieny. Intuitively, the key distortion in the baseline model disussed earlier is the lak o insurane or risk-averse workers. Sine severane payments involve a transer rom risk-neutral irms to risk-averse workers, they are eiieny enhaning. Sine irms have an inentive to oer severane payments voluntarily, there is no need or a mandate. However, i irms suer rom a redibility problem, a promise to pay severane payments in the uture in exhange or lower wages now may not be redible. I there are redibility problems assoiated with a irm's promise, then a government mandate an have a role. I the government mandates severane payments at the appropriate level, prodution eiieny will be restored. Figure shows the impat o a mandated severane payment on the labor market outomes o interest. The relationship between severane payments and the labor market outomes o interest is nonlinear. Aggregate net output, share o labor in the ormal setor, and worker welare inrease with severane payments and then they start dereasing beyond a point. The turning point omes at the level o severane payments that ully insures workers and is also the prodution-eiient level o severane payments. Any severane payments in exess o this is ounterprodutive. A nonlinear relationship obtains between unemployment and the level o severane payments as well. Note that the impat on unemployment is a omposite eet o hanges in the share o labor in the ormal setor and the rate o job destrution by the irms. As shown in Figure d, the rate o job destrution inreases irst and then dereases with severane payments as does the rate o unemployment shown in Figure e. Figure learly shows the eiieny-enhaning role o severane payments. A small severane payment inreases aggregate output and promotes strutural transormation. It inreases worker welare even though the unemployment rate inreases. Figure shows the impat o severane payments on the ration o entrepreneurs in the ormal setor. Sine this orresponds to the number o irms in the ormal setor, this an be thought o as the impat o severane payments on the extensive margin o employment. One advantage o our large irm model is that we an study the impliations o a poliy on irm-level employment as well, that is, we an study the impat o a poliy on the intensive margin as well. Figure g shows the impat o severane payments on the intensive

19 Labor Market Regulations in the Context o Strutural Transormation 13 margin (irm level) o employment. Figure g ompares irm-level employment in the absene o any poliy with the irm-level employment when optimal severane payments are in plae. The solid blak aptures the no poliy intervention ase while the dashed line aptures the optimal severane payments ase. The line to the let o the point o disontinuity aptures the employment o inormal irms while the line to the right aptures the employment o ormal setor irms. The point o disontinuity is lower with optimal severane payment whih simply aptures the extensive margin eet. That is, there are more ormal setor irms with optimal severane payments. For a given level o produtivity, the employment o ormal setor irms is higher with optimal severane payments than without it. For inormal setor irms, the opposite is true. That is, they have higher employment with no poliy than with optimal severane payments. The reason is that the inormal wage is higher in the optimal severane payments ase than without it. Thereore, optimal severane payments inrease ormal setor employment both at the extensive and intensive margins. Figure : Severane Payments and Labor Market Outomes Aggregate output Share o entrepreneurs in ormal setor Figure a: Severane payments and aggregate output Serverane payments Figure : Severane payments and share o entrepreneurs in ormal setor Serverane payments Share o labor ore in ormal setor Job destrution Figure b: Severane payments and share o labor ore in ormal setor Serverane payments Figure d: Severane payments and job destrution Serverane payments A=8, B=1, L=1, γ= /3, Z=.5, ρ=1, F=.75 ontinued on next page

20 14 ADB Eonomis Working Paper Series No. 543 Figure : ontinued Figure e: Severane payments and unemployment Figure : Severane payments and worker welare Unemployment Servie payments Worker welare Servie payments Figure g: Firm level employment Employment Firm produtivity A=8, B=1, L=1, γ= /3, Z=.5, ρ=1, F=.75 Soure: Authors alulations. While we have talked about mandated severane payments as a poliy intervention, it is equivalent to an unemployment insurane program unded by a layo tax. So, suppose the government imposes a layo tax o s and transers the proeeds to unemployed workers. The outome is equivalent to the ase when s is the level o mandated severane payments. While the two are equivalent theoretially, one or the other may be superior in pratie. Mandated severane payments may suer rom nonompliane, either willul or beause o the inability o a irm to pay it during rises. In this ase, an unemployment insurane unded by a layo tax may be preerable. The US unemployment insurane program with experiene-rated ontribution has this eature. B. Nonompliane, State Capaity, et. While the theoretial model shows the equivalene between severane payments and unemployment insurane unded by a layo tax, mandated severane payments may suer rom nonompliane problems as happened during the inanial risis in the Republi o Korea. In this ase, a governmentprovided insurane unded by a layo tax may be superior. The downside o state-provided unemployment insurane is that it may run into weak state apaity. One way to model weak state

21 Labor Market Regulations in the Context o Strutural Transormation 15 apaity would be to assume that when government ollets money rom irms, only a ration is given bak to workers as beneits. The remaining ration 1 is lost due to orruption, mismanagement, et. 1 will apture strong state apaity. In this ase the maximization exerise or irms is modiied as ollows: 1 1 Max A L( ) w (1 ) L( ) L( ) s L, w, st..(1 ) U( w ) U( sz) U( w ) i The above aptures the at that while the layo tax is s, a ired worker atually gets s s. Clearly, this would redue the eetiveness o unemployment insurane. In the appendix, we veriy that i irms oer severane payments voluntarily, the privately optimal level o severane payments (or unemployment insurane unded by an equivalent layo tax) ails to restore prodution eiieny. The privately optimal level o s doesn't provide ull insurane: w s z We veriy numerially that i the level o is too low, that is the leakage is suiiently high, then irms do not want to oer any severane payments voluntarily. Additionally, the outputmaximizing level o severane payments is zero in this ase. That is, even a poliy maker doesn't want to mandate severane payments or provide unemployment insurane through a layo tax in this ase. Next, we study iring restritions in the model and show how the results dier rom severane payments. C. Administrative Burden o Firing To see how we an study iring restritions in the model, suppose that irms in the ormal setor have iring osts whih are in the nature o administrative burden denoted by b. It is important to note that the administrative burden is not a transer to workers. The administrative burden o iring b, should be distinguished rom the severane payments, s. The latter are a transer to workers while the ormer simply makes it ostlier or irms to ire workers. It should be thought o as the ost o omplying with the iring regulations. Now irms do the ollowing maximization exerise: 1 1 Max A L w (1 ) Lh bl L, w, st. (1 ) U( w ) U( z) U( w ) i The key equations or this ase are derived in the appendix. Figure 3 provides results or how the variables o interest hange in response to a hange in b. Note rom Figure 3a that the administrative burden o iring results is a derease in aggregate net output. This happens beause it

22 16 ADB Eonomis Working Paper Series No. 543 auses more resoures to low to the inormal setor thereby impeding the proess o strutural transormation. As shown in Figure 3b, the share o workers in the ormal setor dereases as the administrative burden o iring inreases as does the share o entrepreneurs in the ormal setor shown in Figure 3. Figure 3d shows that the rate o job destrution dereases as does the rate o unemployment shown in Figure 3e. The derease in unemployment is a onsequene o the derease in iring,, shown in Figure 3d, as well as the derease in the share o workers working in the ormal setor, shown in Figure 3b, whih is the only setor that has unemployment. Interestingly, an inrease in the burden o iring also dereases the welare o workers as shown in Figure 3. Thereore, not only is this poliy ineiient rom the point o view o prodution eiieny, but the general equilibrium eet is suh that the welare o workers dereases as well. Thereore, it ails to provide protetion to workers in addition to impeding strutural transormation. Figure 3: Administrative Burden o Firing and Labor Market Outomes Aggregate output Share o enterpreneurs in ormal setors Figure 3a: Firing tax and aggregate output Firing tax Figure 3: Firing tax and share o entrepreneurs in ormal setor Firing tax Share o labor ore in ormal setor Job destrution Figure 3b: Firing tax and share o labor in ormal setor Firing tax Figure 3d: Firing tax and job destrution Firing tax A=8, B=1, L=1, γ= /3, Z=.5, ρ=1, F=.75 ontinued on next page

23 Labor Market Regulations in the Context o Strutural Transormation 17 Figure 3: ontinued Figure 3e: Firing tax and unemployment Figure 3: Firing tax and and worker welare Unemployment Firing tax Worker welare Firing tax Figure 3g: Firm level employment Employment Firm produtivity A=8, B=1, L=1, γ= /3, Z=.5, ρ=1, F=.75 Soure: Authors alulations. Analogous to Figure g disussed earlier, Figure 3g shows the impat o the administrative burden o iring on the intensive margin (irm level) o employment. The point o disontinuity is higher with the burden o iring whih aptures the extensive margin eet shown in Figure 3. That is, there are less ormal setor irms i iring workers is onerous. Moreover, or a given level o produtivity, the employment o ormal setor irms is lower with iring burden than without it. For inormal setor irms, the opposite is true. That is, they have higher employment with iring burden than no poliy beause the inormal setor wage is lower with iring burden. Thereore, the administrative burden o iring dereases ormal setor employment both at the extensive and intensive margins. Anedotal evidene suggests that it is diiult or ormal setor irms to ire workers in many developing ountries inluding India and Sri Lanka. That is, b is high enough that 0. The result in Figure 3 suggests that this poliy is learly ineiient and more importantly it ails to raise the welare o workers who are the intended beneiiaries o suh poliies.

24 18 ADB Eonomis Working Paper Series No. 543 To sum up, while the administrative burden o iring that are not a transer to workers are ounterprodutive, severane payments or unemployment insurane unded by a layo tax have an important insurane role whih promotes strutural transormation. D. Inrease in General Regulatory Burden Any regulation that redues the inentive to enter the ormal setor slows down strutural transormation. The eets o a general regulatory burden suh as the ease o doing business an be aptured through F. An inrease in F redues the attrativeness o the ormal setor and thereore, hinders strutural transormation. A derease in F an apture the ease o doing business or it ould apture a tehnologial hange that allows irms to operate at a smaller sale, making smaller irms viable in the ormal setor. Figure 4 aptures the impat o a hange in F on several variables o interest. It is shown that the aggregate output (Figure 4a), the share o workers in the ormal setor (Figure 4b), and the share o entrepreneurs in the ormal setor (Figure 4) derease with F. It turns out that the wage that irms pay in the ormal setor, w, is dereasing in F. The iring rate,, inreases with F as shown in Figure 4d. A onsequene o an inrease in the iring rate and a derease in the share o workore in the ormal setor in response to an inrease in F is that the rate o unemployment beomes nonmonotoni as shown in Figure 4e. Finally, the welare o workers dereases with F as is veriied in Figure 4. Thereore, a redution in regulatory burden not only promotes strutural transormation and thereby inreases aggregate net output, it also inreases worker welare. That is, workers in both setors gain. The model that we have developed in the paper an also study the impliations o regulations/poliies that beome appliable or irms above a ertain size. For example, the Fatories At in India requires irms (using power) with more than 10 workers to register. Similarly, the Industrial Disputes At in India requires irms employing more than 50 workers to provide severane payments and irms employing more than 100 workers to seek permission rom the government beore iring any worker. In Sri Lanka, the iring restritions apply to irms with 15 or more workers. To see the impliations o suh regulations in the ontext o our model suppose the irm size above whih the iring regulation is binding is ˆL. The qualitative results are going to be similar to the ones desribed above. One dierene will be that now some ormal setor irms will have an inentive to redue their irm size to ˆL to avoid paying iring taxes. That is, the size distribution o ormal setor irms will have a hole around ˆL. 4 4 Abidoye, Orazem, and Vodopive (014) ind evidene that some irms in Sri Lanka try to keep the irm size below the threshold where iring restritions start binding.

25 Labor Market Regulations in the Context o Strutural Transormation 19 Figure 4: Regulatory Burden and Labor Market Outomes Aggregate output Figure 4a: Regulatory burden and aggregate output Share o labor ore in ormal setor Figure 4b: Regulatory burden and share o labor ore in ormal setor Regulatory burden Regulatory burden Figure 4: Regulatory burden and share o entrepreneurs in ormal setor Figure 4d: Regulatory burden and job destrution Share o entrepreneur in ormal setor Regulatory burden Job destrution Regulatory burden Figure 4e: Regulatory burden and unemployment Figure 4: Regulatory burden and unemployment Unemployment Worker welare Regulatory burden Regulatory burden A=8, B=1, L=1, γ= /3, Z=.5, ρ=1, F=.75 Soure: Authors alulations.

26 0 ADB Eonomis Working Paper Series No. 543 E. Minimum Wage Regulation We an also study the impliations o a minimum wage regulation in the model. It must be stated at the outset that the welare analysis beomes more ompliated in this ase beause we already have a distortion arising rom risk aversion and now we are introduing another distortion in the orm o a minimum wage. Suppose that the government imposes a binding minimum wage in the ormal setor and assume that the minimum wage regulation is not binding in the inormal setor. Denote the minimum wage by w. Firms in the ormal setor take this as given and perorm the ollowing maximization exerise: 1 1 MaxA L w(1 ) L L, st..(1 ) U( w) U( z) U( w ) i The irst-order onditions are L : A L w(1 ), (4) : A L UwUz wl (5) The above two equations along with (1 ) U( w) U( z) U( wi) determine L,, and or a given w and w i. Essentially, or all irms is given by U( w) U( wi ) U( w) U( z) (6) The variables or the inormal setor are determined as beore whih allow us to determine w i and along with and the share o workore in eah setor. The relevant equations are obtained simply by replaing w by w in equations (8) (31) in the appendix. 5 For the above problem to make sense, it must be the ase that w w i, that is why we assumed that the minimum wage doesn't bind in the inormal setor, and thereore, it doesn't diretly aet the inormal setor.