Forecasting Formal Employment in Cities Citation Lora, Eduardo. “Forecasting Formal Employment in Cities.” CID Research Fellow and Graduate Student Working Paper Series 2019.114, Harvard University, Cambridge, MA, May 2019. Published Version https://www.hks.harvard.edu/centers/cid/publications/fellow-graduate-student-working-papers Permanent link https://nrs.harvard.edu/URN-3:HUL.INSTREPOS:37366838 Terms of Use This article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http:// nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of-use#LAA Share Your Story The Harvard community has made this article openly available. Please share how this access benefits you. Submit a story . Accessibility
42
Embed
Forecasting Formal Employment in Cities Share Your Story
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Forecasting Formal Employment in Cities
CitationLora, Eduardo. “Forecasting Formal Employment in Cities.” CID Research Fellow and Graduate Student Working Paper Series 2019.114, Harvard University, Cambridge, MA, May 2019.
Published Versionhttps://www.hks.harvard.edu/centers/cid/publications/fellow-graduate-student-working-papers
Terms of UseThis article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http://nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of-use#LAA
Share Your StoryThe Harvard community has made this article openly available.Please share how this access benefits you. Submit a story .
1.IntroductionUnitedNationsSustainableDevelopmentGoal8is“Promotesustained,inclusiveandsustainableeconomicgrowth,fullandproductiveemploymentanddecentworkforall”. More specifically, target 8.3 seeks to “[b]y 2030, achieve full and productiveemploymentanddecentwork forallwomenandmen, including foryoungpeopleand personswith disabilities, and equal pay forwork of equal value”. This paperassesses how achievable this target is for Colombia, based on a novel theory offormalemploymentcreationincitiesandtwocomplementaryforecastingmethods:standardregressionsandmachinelearning.
Cities are necessary for economic growth to take place through a process ofdiversification and innovation that leads to productive employment and decentwork for largersharesof thepopulation.However,urbanization isnotasufficientcondition for industrialization and productive employment: the expected relationbetweenurbanization, industrializationandemploymentqualityisabsentinmanypartsof theworld (Gollin, JedwabandVollrath,2016).Urbanizationpatterns,andnot just urbanization rates or macroeconomic factors (such as natural resourceabundance) may shed light on the role of cities in economic growth and formalemploymentcreationassuggestedbytwostrongstylizedfacts(O’Cleryetal2018):(1) formal occupation rates aremore variable across citieswithin countries thanacross developing countries (Figure 1), and (2) larger cities create proportionallymoreformalemployment(Figure2).
Figure1.Boxplotsforthedistributionofformaloccupationratesinasetof56developingcountries(leftplot)andcitiesinBrazil,Colombia,MexicoandtheUS.Weobservealargervarianceinformalityrates across cities within countries than across countries, suggesting that the study of thedeterminants of formality across cities is a relevant area of study in connection with SustainableDevelopmentGoal8(“fullandproductiveemploymentanddecentworkforall”).Source:O’Cleryetal(2018).
3
Figure 2. Formality rates increasewithworking age population for cities across Brazil, Colombia,Mexico and the US: larger cities have disproportionatelymore workers in the formal sector thansmallercities,apatternthatisstatisticallysignificantinthefourcountriesasshownatthebottomofthefigures.Source:O’Cleryetal(2018).
2.TheoreticalframeworkOneofthecentralissuesineconomicdevelopmenttheoryisthereasonforthesizeandpersistenceofinformallaborindevelopingeconomies.Sinceformalfirmshaveaccess to capital and technology that make themmore productive than small orfamily businesses, what explains that large chunks of the labor force are notoccupiedintheformalsectorwherelaborconditionsarebetterthanintheinformalsector?Economic theoryhasprovidedseveralexplanations. Indualisticmodelsofinformality, the self-employed and their family businesses are fundamentallydifferent from formal firms in the type of human capital they use –mainlyuneducated and unproductive entrepreneurs and managers–, and in what theyproduce –mainly low-quality products for low-income customers. The formal andinformalsectorscoexistbecausetheyaredifferent(Lewis1954,HarrisandTodaro1970, Rauch 1991). An alternative view is that of De Soto (1989, 2000), whoconsiders that informal firms are an untapped reservoir of productive resourcesheld back by government regulations. Relatedly, Levy (2008) sees informal
4
businesses as entrenched firms that survive in spite of their low productivity byavoidingtaxesandregulations.Lastly,inlaborsearchmodelsthattakeintoaccountthe costs and benefits of labor regulations, informal employment is not theconsequence of exclusion, but the result of labor market frictions betweenheterogeneousworkersandfirms(Albrecht,NavarroandVroman2009;BoschandMaloney2010;Ulyssea2010;Meghir,NaritaandRobin2015).While empirical evidence has been provided in support to each of theseexplanations of informality, none of them recognizes the two stylized factsmentioned in the introduction, namely that formal occupation rates across cities(within a given country) have a larger variance than across countries, and thatformal occupation rates are directly and significantly associatedwith city size. Inotherwords,noneofthemainstreamtheoriescanexplaintheroleofcitiesinformalemployment creation. Furthermore, some of the main variables put forward bythose theories to explain the presence of informality –such as social securityregimes and labor hiring and firing legislation—have little or no variance acrosscitieswithineachcountry.In view of these shortcomings, this paper adopts the theoretical frameworkdevelopedbyO’Cleryetal(2019),whichdiffersfromprevioustheoriesinanumberofways.First,itfocusesoncitiesratherthancountries,becausecitiesaretheactuallocationswhereworkers and their employers interact. Second, it emphasizes skilldiversity–whichiscentralinurbaneconomics—ratherthanskilllevels,educationalattainment or managerial capabilities. Third, it assumes that firms evolve bytinkering with skills because many feasible technologies cannot be known inadvance, but need to be discovered. Formal employment creation in cities resultsfrom this evolutionary process. In larger cities, firms have better access to thediverseskillstheyneedtoproducemoresophisticatedgoods.Themaincomponentsofthemodelcanbesummarizedasfollows(forthecompletemodelseeO’Cleryetal2018):Citysizeandskilldiversityaretakenasexogenous.Eachfirmislocatedinacity,butsellstothewholenationalmarketunderperfectcompetition.Theoutputoffirm𝑟,whichbelongstoindustry𝑗attime𝑡,isgivenbyaCESproductionfunctionwhoseonlyproductionfactorsaretypesoflabordifferentiatedbyskill,whereskills𝑘arehardtosubstituteforoneanother:
Since𝜃! 𝑘 ≥ 1foreveryskill,industrycomplexityislargerinindustrieswhichcombinealargersubsetofsophisticatedskills.Finally, firms transition from less to more sophisticated industries following aprobabilisticrulesuchthattheconditionalexpectedvalueofafirm´scomplexityoneperiodaheadis:
𝐸! 𝐶!!!! 𝑗! 𝑟 = 𝑗 = 𝑝 𝐶!
! + 1− 𝑝 𝛽! 𝑗, 𝑗!!!∈! 𝐶!!!
!"#$%&'()* !"#$%#&'(
(4)
Duetothedefinitionof𝛽! 𝑗, 𝑗! ,thecomplexitypotentialofagivenfirmdependson(i)theindustrytowhichthefirmcurrentlybelongs,(ii)thedistancebetweensuchindustry and those industries to which the firm could migrate, (iii) the relativeabundanceinthelocallabormarketofthosenewskillswhichthefirmmusthireinordertocarryoutanindustrytransition,and(iv)thesizeoflaborforceincity.
Noticethatcomplexitypotentialisaweightedaverageoftheindustrycomplexityofthemissingsectorswithweightsgivenbytheskillsimilaritybetweenthosesectorsandtheonesalreadypresent.Inorder tooperationalize equation (6), dataareneededon industrycomplexity,missingsectors,andskillsimilaritybetweenallpairsof industries.Sinceskillsaretacit knowledge and therefore unobservable, industry complexity and complexity
6
potentialmustbecomputedindirectly.Tothatend,O’Cleryetal(2019)makeuseofthe methodologies developed by Hidalgo and Hausmann (2009) and Neffke andHenning(2013). Inessence, industrycomplexity isameasureof therangeofskillsneededinanindustry,whichisobtainedfromthenumberofindustriespresentinthecitiesthathavetheindustry(iethoseindustriesthathaverevealedcomparativeadvantage greater than 1 in city, based on formal employment shares) and thenumber of cities that have the industry (ie those cities where the industry hasrevealed comparative advantage greater than1).Skill similarity between apair ofindustriesismeasuredbytherelativeintensityofthelaborflowsbetweenthetwoindustries, and missing industries in city are those with revealed comparativeadvantage lower than 1 (Appendix 1 provides further details on computationmethods).
3. DataandempiricaldefinitionsLike in O’Clery et al (2019), I use data for Colombian cities larger than 50,000inhabitants.MydefinitionofcitiesrestsonthemethodologyproposedbyDuranton(2015)todefinemetropolitanareas.Itconsistsofaddingiterativelyamunicipalitytoametropolitanareaif thereisashareofworkers,aboveagiventhreshold,thatcommute from themunicipality to themetropolitan area. Assuming a 10 percentthreshold,themethodologygenerates19metropolitanareasthatconsistoftwoormore municipalities (comprising a total of 115 municipalities). Since another 43individualmunicipalities havepopulations above50,000 inhabitants, a total of 62citiesisobtained.
The main data source for the 62 cities is the social security administrative datacollected by the Health and Social Security Ministry, known as PILA (PlanillaIntegradadeLiquidaciónLaboral). PILA contains informationbyworker and firmon days of work, sector of activity and municipality.2To aggregate these data, Icounttheshareof theyeart thateachworkereffectivelycontributedtothesocialsecuritysystemthrough firmspercitycper industry j (𝑒𝑚𝑝!,!).This is the formalemployment for a given sector (or for the aggregate of all sectors within a city).Sectors are defined at the 4-digit industry level of the International StandardIndustrialClassification(ISIC,revision3.0).
Theformalemploymentrateincitycinyeart(𝐹!,!)isdefinedasformalemploymentdivided by the city-wide population 15 years old or older (𝑝𝑜𝑝!,! , estimated byDANE):
𝐹!,! = 𝑒𝑚𝑝!,! 𝑝𝑜𝑝!,! (7)
2The datasets have information on age and gender, which we do not use. Unfortunately, it provides no information on education, which prevents us from testing our model predictions vis-à-vis the findings of previous works discussed in the introduction.
7
The (simple) average formal occupation rate in cities was only 20.3 percent ofworkingagepopulationin2015,witharelativelylargestandarddeviation(between11.1percentpoints).Importantchangesinurbanformaloccupationratesoccurredbetween2008and2015:theaggregateformaloccupationrateforthe62citieswentupfrom21.1percentto31.2percent,witha(simple)averageincreaseacrosscitiesof8.1percentpointsandastandarddeviationof5.4points.Formaloccupationwasfacilitated by a rate of GDP growth of 4.1 percent, and probably also by theeliminationinMayof2013ofpayrolltaxesrepresenting5percentofthewagebill(Kugler,KuglerandHerrera-Prada2017).
Bosch, Mariano, and William F. Maloney. 2010. “Comparative analysis of labor marketdynamicsusingMarkovprocesses:Anapplicationtoinformality”.LabourEconomics,17(4):621-31.
Gollin, D., Jedwab, R. & Vollrath D., “Urbanizationwith andwithout Industrialization”,Journal of Economic Growth (2016) 21: 35. https://doi-org.ezp-prod1.hul.harvard.edu/10.1007/s10887-015-9121-4
Hidalgo,CésarandRicardoHausmann2009.“TheBuildingBlocksofEconomicComplexity”,Proceedings of the National Academy of Sciences,106(26):10570-5.DOI:10.1073/pnas.0900943106.
Kugler,Adriana,MauriceD.KuglerandLuisO.Herrera-Prada.2017."DoPayrollTaxBreaksStimulate Formality? Evidence from Colombia’s Reform,"Economia, Journal of theLatinAmericanandCaribbeanEconomicAssociation,Fall2017:3-40.
Meghir, Costas, Renata Narita, and Jean-Marc Robin. 2015. “Wages and Informality inDevelopingCountries”.AmericanEconomicReview,105:1509-46.
Neffke, Frank and Martin Henning. 2013. “Skill Relatedness and Firm Diversification”,StrategicManagementJournal,34(3):297-316
O’Clery, N., Chaparro, J.C., Gómez-Liévano, A., *Lora, E. 2019. “Skill Diversity and theEvolutionofFormalEmploymentinCities”,submittedtoResearchPolicy.
Rauch, James E. 1991. “Modeling the Informal Sector Formally.” Journal of DevelopmentEconomics35(1):33–47.
Ulyssea, Gabriel. 2010. “Regulation of entry, labor market institutions and the informalsector”.JournalofDevelopmentEconomics,91(1):87-99.
24
Appendix1-CalculationMethodsforIndustryComplexityThisappendixexplainsthemethodsforcalculatingtheindustrycomplexityvariableintroducedattheendofSection2.ItisadaptedfromHidalgoandHausmann(2009)andNeffke andHenning (2013). The actual calculations used formal employmentdataofallindustriesproducingeithergoodsorservices(ISIC-AC,Rev.3,at4digits,usingsocialsecuritydatafromPILA).Intheequationsbelow,thesub-indexcindicatescitiesandthesub-indexpindicatesindustries.Whilenotimesub-indexisusedhere,allcalculationsareappliedforeachyearseparately(2008-2015).CalculationofRevealedComparativeAdvantagesThe computation starts with data for employment by industry, city and year,organizedinmatrixform:
DiversityandUbiquityCalculationsThe RCA matrix is transformed in a binary matrix depending on whether aparticularvalueislargerthan1ornot:
25
𝑀!" =1 𝑅𝐶𝐴!" ≥ 1
0 𝑅𝐶𝐴!" < 1
Thismatrixindicatestheindustriesthatarerelativelylargeineachcity.Thismatrixis thenused to compute theDiversity indicator at the city level, and theUbiquityindicatorat the industry level–that is, thecountof thenumberof industrieswithrelatively large employment for each city, and the count of the cities that have agivenindustrywitharelativelyhighintensity:
𝑘!,! = 𝑀!"!
𝑘!,! = 𝑀!"!
IndustryEconomicComplexityThe complexity of an industry can be measured by its ubiquity weighed by thediversity of the localities that have revealed comparative advantage in suchindustry.Extendingthisexerciseadinfinitum,correctingdiversitywithubiquityandvice-versawith consecutive iterations, is called themethodofreflections. It canbeexpressedasfollows:
𝑴𝑪×𝒌 = 𝝀𝒌Where𝒌isaneigenvectorof𝑴𝑪.Thesecondlargesteigenvectorof𝑴!istakenastheIndustryComplexityIndex.TheIndex iscalculatedonemployment levelsper industry/citycombination, includingonly industrieswith at least 50 formal employees in an averagemonth, and onlycitieswithatleast10industrieswith50ormoreformalemployees.