CDEP-CGEG WORKING PAPER SERIES CDEP-CGEG WP No. 15 Firm’s Response and Unintended Health Consequences of Industrial Regulations Christopher Hansman, Jonas Hjort, Gianmarco Leon February 2015 FIRMS’ RESPONSE AND UNINTENDED HEALTH CONSEQUENCES ∗ OF INDUSTRIAL REGULATIONS Christopher Hansman Jonas Hjort Gianmarco Le´on Columbia University Columbia University Universitat Pompeu Fabra and BREAD and Barcelona GSE February 20, 2015 Abstract Regulations that constrain firms’ externalities in one dimension can distort incentives and worsen externalities in other dimensions. In Peru’s industrial fishing sector, the world’s largest, fishing boats catch anchovy that plants along the coast convert into fishmeal. Matching administrative daily data on plant production, ground-level air quality data, hospital admissions records, and survey data on individual health outcomes, we first show that fishmeal production worsens adult and child health through air pollution emitted by plants. We then analyze the industry’s response to a 2009 reform that split the Total Allowable Catch (TAC) into boat-specific, transferable quotas (ITQs) to preserve fish stocks and reduce overcapacity. As predicted by a two-sector model with heterogeneous plants, on average across locations, fishmeal production was spread out in time, for two reasons: (i) boats’ incentive to “race” for fish was removed, and (ii) inefficient plants decreased production and efficient plants expanded production (across time). The reform greatly exacerbated the industry’s impact on health, causing e.g. 55,000 additional hospital admissions for respiratory diseases. We show that the reason is that longer periods of moderate air pollution are worse for health than shorter periods of higher intensity exposure. Our findings demonstrate the risks of piecemeal regulatory design, and that the common policy trade-off between duration and intensity of pollution exposure can be critical for industry’s impact on health. JEL codes: D2, L5, L7, O1, I1, Q5 Keywords: Industrial regulations, firms, externalities, air pollution, health, fishing, Peru, ITQs ∗[email protected], [email protected], [email protected]. We thank Doug Almond, Michael Best, Francois Gerard, Janet Currie, Amir Jina, Namrate Kala, Amit Khandelwal, Ilyana Kuziemko, Rocco Macchiavello, Matthew J. Neidell, Anant Nyshadham, Andrea Prat, Wolfram Schlenker, Alessandro Tarozzi, Miguel Urquiola, EricVerhoogen,ReedWalkerandseminarparticipantsatBREAD,Columbia,CREi,theEconometricSociety,IZA, NEUDC, Norwegian School of Economics, Princeton, Stanford, UPF, World Bank DRG, and the 2014 Summer WorkshopinDevelopmentEconomicsinAscea,Italyforhelpfulcommentsandsuggestions. WearegratefultoJesse Eiseman, Miguel Figallo, Adrian Lerche and Leonor Lamas for excellent research assistance and field work. Cesar Casahuama´nkindlysharedaccesstothefishmealproductiondata. HjortacknowledgesfinancialsupportfromCIBER at Columbia University, and Leo´n from the Spanish Ministry of Economy and Competitiveness, through the Severo Ochoa Programme for Centres of Excellence in R&D (SEV-2011-0075) and grant ECO2011-25272. 1 1 Introduction Industries generate multiple types of externalities. This is especially so in developing countries, wheremarketfailuresarecommonplace(Laffont,2005). Althoughregulationsaretypicallydesigned with a particular externality in mind (e.g. overuse of a depletable natural resource), regulatory incentives also affect which firms survive and thrive and how firms organize production. If such industry responses in turn affect the extent of other sector externalities (e.g. pollution from the production process), regulations can have welfare consequences that differ from the planned partial equilibrium effects.1 In this paper we focus on the world’s largest industrial fishing sector, located in Peru (Paredes andGutierrez,2008). Theregulationsinplaceareaimedatmaintainingindustryprofitabilitywhile avoiding overfishing (see e.g. Hardin, 1968; Ostrom, 2008; Huang and Smith, 2014), but the plants that convert the raw fish into fishmeal also emit large amounts of air pollution.2 We first document thisadditionalexternalitybyidentifyingthefishmealindustry’scausaleffectonthehealthofPeru’s coastal population. We then show how the sector responded to the 2009 introduction of individual, transferable fishing quotas (ITQs) – the regulatory system most commonly prescribed to preserve fish stocks and maximize total sector profits.3 Finally, we show that the reform dramatically worsened the industry’s effects on health, and trace the increased impact to boats’ and plants’ reorganization of production in response to the 2009 ITQ reform. To estimate the causal effect of fishmeal production on health, we exploit government-imposed, semi-annual fishing ban periods in a difference in differences approach. We link administrative hos- pitaladmissionsdataandrepeatedcrosssectionsofhouseholdsurveystoadministrativedataonall production of fishmeal at the day×plant level. The results show that exposure to fishmeal produc- 1InthewordsofGreenstoneandJack(2015,p.10): “...differenttypesofmarketfailuresmayinteractincomplicated ways that can make an otherwise efficient correction (e.g., a Pigouvian tax) to one market failure suboptimal in the presence of others”. 2The fishmeal plants are located at the ports. When a boat docks, its anchovy catch is transported into a plant throughconveyorbeltsandimmediatelyconvertedintofishmeal(thevalueoftheresultingfishmealrapidlydecreases asthefishdecays)byburningorsteamingthefish. Localsarguethattheindustryisresponsibleforhealthproblems stemmingfromairpollutionemittedintheprocessofconvertingthefishintofishmeal. Ina2008article,TheEcologist magazinereportedthat“Whenwevisited oneheavily afflicted community[inthe fishmeal town of Chimbote], more than a dozen women and children gathered [...] to vent their anger at the fishmeal plants. They claim the plants that loom over their houses are responsible for asthma, bronchial and skin problems, particularly in children. ‘We know the factories are responsible for these [problems], because when they operate the illnesses get worse’, says one youngwoman[...] Anothersayswhentheplantsareoperatingthepollutionissothickyoucannotphysicallyremain on the street. Footage [...] seen by The Ecologist illustrates typical conditions when fishmeal plants are operational: billowing black smoke drifts through the streets, obscuring vision and choking passers-by [...] Pupils at a Chimbote school [...] also complain of health problems. ‘It causes fungal growths, breathlessness, we cannot breathe’, says one boy.” (The Ecologist, 2008). Such complaints are supported by case studies (Cerda and Aliaga, 1999; Sueiro, 2010; MINAM, 2010, 2011), and local doctors: “Dr Ramon de la Cruz, dean of Chimbote’s Colegio Medico del Consejo Regional XIX, told The Ecologist: ‘All these respiratory problems are caused by contamination from the fishing industryinChimbote’[...] Cruzstatesthatthereisadirectcorrelationbetweentheonsetoffishmealproductionand illness in children in Chimbote.” (The Ecologist, 2008). 3See e.g. Boyce (2004, p.1): “In fishery management, an optimal instrument, individual transferable quotas (ITQs), exists”. Costello, Gaines and Lynham (2008) find that the introduction of ITQs in general tends to slow declines in fish stocks, and Natividad (2014) finds efficiency gains in the Peruvian fishing sector after the 2009 ITQ reform. 2 tion in the last 30 or 90 days worsens health, increasing respiratory (and total) hospital admissions, increasing reported health issues and medical expenditures among adults, and increasing reported health issues and coughs among children ≤ 5 and children ≤ 1. The estimated health effects among adults are not concentrated among the five percent of our sample in fishmeal locations that re- port to work in the fishing industry, nor driven by seasonal migration or labor market responses. Estimates from regressions where we instrument for ground-level concentration of PM10, PM2.5, NO and SO with fishmeal production suggest that the industry’s impact on health is explained 2 2 in large part by air pollution emitted by plants. We then investigate how the fishmeal industry responded to the 2009 ITQ reform, and how the response changed the industry’s impact on health. Before the reform, the North/Central region operated under a market-wide “total allowable catch” (TAC) system and semi-annual fishing ban periods, while a small region in the South was largely unregulated (due to fears that Chilean fishing activity would offset any environmental or industrial benefits of regulation). The 2009 reform introduced boat-specific, transferable quotas (ITQs) country-wide and extended fishing ban periods to the South. While there was only a small decrease in total production of fishmeal, the reform benefited fish stocks and increased total sector profits.4 To illustrate how we expect firms’ response to the introduction of individual property rights to affect the geographical and temporal distribution of fishmeal production, we present a simple two-sector model with heterogeneous plants. The model predicts that production will be spread out in time after the reform: the introduction of ITQs removes boats’ incentive to rapidly capture as much as possible of the TAC, and without the glut of fish early in the season less efficient plants are predicted to decrease production or exit the industry, and more efficient plants to spread their productionacrosstime. Thesepredictionsfindsupportinthedata.5 Theindustryconsolidationand evening out of production due to the reform were applauded by environmentalists and the fishmeal sector (see e.g. International Sustainability Unit, 2011), but may have worsened the industry’s impact on health if dispersed production increases the total amount of pollution emitted and/or the health impact of a given amount of emitted pollution. When comparing fishmeal locations to non-fishmeal locations before and after the reform came into effect, we find that the introduction of ITQs on average dramatically worsened the impact of fishmeal production on adult and child health, leading to for example 55,000 additional respiratory hospital admissions. Cost/benefit calculations that are suggestive but conservative indicate that the monetized cost of the reform’s impact on health surpassed the increase in sector profits and value of increased fish stocks. Geographical heterogeneity in the estimated reform effects supports the hypothesis that in- 4Theincreaseinfishstockswaslikelyduetolowerjuvenilefishcapturepost-reform. Thesystemusedtodetermine the total allowable catch for a given season did not change in 2009. There were likely several reasons why profits increased after the reform; for example, a reduction in plant overcapacity. 5Boats in the North/Central region spread out fishing in time as the ITQ reform came into effect. (Boats in the previously unregulated southern region fished for fewer days of the year after the reform due to the introduction of banperiods.) FishmealproductiondaysincreasedintheNorth/Centralregionandinlocationswithefficientplants. Production days decreased in the South and in locations with inefficient plants. 3 troduction of ITQs affected health through changes in the time profile of production. For the North/Central region, where production was spread out in time as a result of the reform, the esti- mated reform effects are negative (adverse), large and significant. For the smaller southern region, where fishmeal production days decreased with the reform, the estimates are insignificant or signif- icantly positive, for example we see a reduction in coughs among children. Similarly, the estimated reform effects are significantly more negative for locations with efficient plants, when compared to locations with inefficient plants. Finally, we investigate why spreading plant production over time exacerbates the impact on health. The time profile of pollution shows a post-reform change that corresponds to the change in the time profile of production, with less temporal variation after the reform. In general, the concentration of air pollution appears to increase linearly in the level of plant production, and none of the measured air pollutants increase in concentration after the reform. It thus appears that the explanation why dispersing air polluting plant production across time leads to a deterioration in health lies in the shape of the health production function (rather than in more pollution being generated with dispersed production).6 We conclude (a) that the introduction of individual property rights aimed at preserving fish stocks and sector profits in Peru exacerbated the fishmeal industry’s impact on health because changes in incentives and industry dynamics led production to be spread out in time in most locations; and (b) that the two are linked because longer periods of exposure to moderate levels of air pollution are worse for health than higher intensity, shorter periods of exposure. Overall our findings highlight the risks of piecemeal regulatory design. This paper contributes both to the literature on how to design regulation of industrial external- ities, with particular relevance for developing countries, and to the literature on health effects of air pollution. The primary focus in the former has been on comparing (i) the magnitude of decreases in the targeted type of externalities (e.g. pollution or overextraction of a resource – see Costello, Gaines and Lynham (2008) for the case of ITQs for fish) to (ii) the economic costs of compliance.7 A handful of pioneering papers explore unforeseen effects of regulations on the targeted type of externalities (e.g. due to plant substitution between different pollutants)8 or the economic costs of 6We use the term “health production function” to mean either the three-dimensional relationship relating health at a given point in time to both the duration of exposure to air pollution and the intensity of exposure, or one of thetwounderlyingtwo-dimensionalrelationships(theso-called“dose-response”and“duration-response”functions), depending on context. 7Gray and Shadbegian (1993) show that firm TFP drops after the introduction of environmental regulations. Greenstone (2002) finds that air pollution regulations reduce employment and the growth of capital stock. List et al. (2003) find that environmental regulations inhibit the formation of polluting plants. Greenstone, List and Syverson(2012)findthattheintroductionofenvironmentalregulationsintheU.S.ledtoadeclineinTFP.However, regulationsofsomeformsofairpollutionhavebeenfoundtoincreaseplantTFP(BermanandBui,2001;Greenstone, List and Syverson, 2012). 8Sigman (1996); Greenstone (2003); Gibson (2015)explore plant substitutionbetweenregulated andunregulated pollutants. Becker and Henderson (2000) find that, in the U.S., environmental regulations favoring small firms led to a shift in industry structure towards single-plant firms, which in turns contributed to environmental degradation. Field,GlennersterandHussam(2011)demonstratethatalarge-scaleeffortinBangladeshtoconvincehouseholdsto abandon deep wells believed to be contaminated by arsenic in favor of surface wells increased child mortality from diarrheal diseases because the surface wells were contaminated with fecal bacteria. Knittel and Sandler (2011) show 4 compliance (e.g. due to effects on market power).9 This paper is to our knowledge the first to focus instead on (iii) the impact on another type of externality that was ignored in the design of the regulation. We document that piecemeal regulatory design targeting a subset of externalities can dramatically worsen outcomes in the ignored dimensions, yielding ambiguous total welfare effects of otherwise laudable reforms.10 We also contribute to a nascent literature on how to regulate industrial pollution in developing countries (Laffont, 2005; Estache and Wren-Lewis, 2009; Burgess et al., 2012; Duflo et al., 2013; GreenstoneandJack,2015;Dufloetal.,2014;GreenstoneandHanna,2014;Jia,2014). Greenstone and Hanna (2014) note that pollutant concentrations are extremely high in many developing coun- tries,11 imposing significant health costs and highlighting the need for effective regulation. Several innovative recent papers illustrate the need to take institutional capacity and the prevailing incen- tivestructuresintoaccountwhendesigningregulation(Burgessetal.,2012;Dufloetal.,2013,2014; Jia, 2014). The evidence in this paper underscores the need for regulatory design that addresses all relevant externalities simultaneously. Comprehensive regulatory design is particularly important in developingcountrieswhereconcurrentmarketfailuresarecommonplace(seee.g.Field, Glennerster and Hussam, 2011).12 Air pollution has been shown to adversely affect the health of adults (see e.g. Brook RD et al., 2010; Moretti and Neidell, 2011; Schlenker and Walker, 2011; Chen et al., 2013; Currie et al., 2014), and children (see e.g. Chay and Greenstone, 2003; Case, Fertig and Paxson, 2005; Chay and Greenstone, 2005; World Health Organization, 2006; Jayachandran, 2006; Currie and Almond, 2011; Currie and Walker, 2011; Gutierrez, 2013; Roy et al., 2012; Currie et al., 2014, 2015), espe- cially respiratory and pulmonary health outcomes. An existing literature analyzes how the impact depends on the intensity and duration of exposure, but the evidence is primarily correlational. Our analysis especially complements two of a handful of papers that provide convincing causal evidence on the dose-response and the duration-response functions. Chen et al. (2013) find much bigger effects on health and mortality of sustained exposure to high levels of air pollution than (the effects found elsewhere of) short-term exposure. Chay and Greenstone (2003) find non-linearity in the response of infant mortality to reductions in air pollution in the U.S. that are consistent that the benefits of carbon pricing are understated because such regulations reduce not only greenhouse gases but also local pollution. Davis, Fuchs and Gertler (2014) show that the expected benefits of buy-back programs are overstated because households may use newer appliances more. 9Ryan (2012) and Fowlie, Reguant and Ryan (2014) find that allocative inefficiencies due to changes in market power in the U.S. cement market counteract the social benefits of carbon abatement regulations. 10In fact our evidence suggests, consistent with theory (see e.g. Clark, 1980; Boyce, 2004; Weninger, 2008) and empiricalevidence(seee.g.Natividad,2014),thatfortheparticularcaseofindividualtransferablequotas(ITQs)for openaccessresources,thereisnotrade-offbetweendecreasesinthetargetedtypeofexternalitiesandeconomiccosts as the direct economic effects of ITQs are positive and thus to add to rather than substract from the environmental benefits. 11See also Hanna and Oliva (2014); Ebenstein (2012); Chen et al. (2013); Rau, Reyes and Urzua (2013); von der Goltz and Barnwal (2014). 12This paper can also be seen as contributing to the literature on the economics of property rights. See e.g. GoldsteinandUdry(2008)andseveralpapersbyEricaField,especiallyField(2007),forexamplesofthebenefitsof securing property rights in developing countries, and Costello, Gaines and Lynham (2008) for benefits in the case of open access resources such as fish. 5 with concavity in the dose-response function (although the authors point out that there are other possibleexplanationsoftheirfindings). Weaddnewevidenceontheshapeofthehealthproduction function by documenting that exposure to longer periods of moderate air pollution are worse for health than higher intensity, shorter periods of exposure. In their excellent survey of the literature on air pollution and health, Pope III et al. (2011) conclude that there is no existing evidence on the effect of simultaneous changes in duration and intensity of exposure – a trade-off commonly faced by policymakers. The paper is organized as follows. In Section 2 we discuss background on the setting, why fishmeal production may affect health, and the 2009 ITQ reform. In Section 3 we present the data, and in Section 4 we go through our empirical strategy. Evidence on how and why fishmeal productionaffectshealthispresentedinSection5. Section6analyzes,theoreticallyandempirically, the industry’s responses to the 2009 ITQ reform, and Section 7 how and why its impact on health changed as a consequence. Section 8 discusses the total costs and benefits of the reform and regulatory design, and Section 9 concludes. 2 Background 2.1 Peru’s fishmeal sector Peru’sanchovyfisheryaccountsforaround10percentofglobalfishcapture(ParedesandGutierrez, 2008). While“artisan”fishingboatsarepresentinmosttownsalongthecoast,theindustrialfishing sectorisconcentratedinaround22townswithasuitableport. Onlyindustrialboatscanlegallysell fish for “indirect consumption”, primarily meaning for the production of fishmeal. The country’s fishmeal plants, all located at ports, produce about a third of the global supply of fishmeal. In 2008, the sector consisted of 1,194 active industrial fishing boats and 110 fishmeal plants, and seven large companies accounted for around 70 percent of the country’s fishmeal production. The industrial fishing sector is very capital intensive. Paredes and Gutierrez (2008) estimate thattherewereonlyabout26,500jobsinthesectorin2008. (ThePeruvianMinistryofProduction’s Statistics Office gives similar numbers, estimating 19,853 workers in “extraction of the resource” for the industrial sector and 9,335 in processing in 2007). The figures in Christensen et al. (2014) suggest that the share of those jobs that were in fishmeal plants was about a quarter, indicating that the 110 plants operating in 2008 employed on average about 60 workers each. Five percent of our adult sample in fishmeal locations (≤ five kilometers from a fishmeal port) reports to work in “fishing”.13 Industrial fishing and fishmeal production only takes place certain periods of the year. In Table I, we present summary statistics from individual survey data (described below) comparing fishmeal and non-fishmeal locations in and out of the production season. The share of sampled adults that 13Fivepercentofadultscorrespondstoeightpercentofthosewhohaveajob. Someindividualsthatworkoutside oftheindustrialfishingsectormayalsoreporttoworkinfishing(e.g. artisanfishermen)sothatthetrueproportion who works in the fishmeal sector is likely even lower. 6 were born in a different district than where they were surveyed is lower in fishmeal locations in season (37 percent) than out of season (40 percent). The numbers thus indicate that there is little seasonal work migration to the fishmeal locations. The reason is likely that jobs in the industrial fishing sector are quite stable.14 Many fishmeal firms keep the (relatively high-skill) plant workers on payroll also outside of the production season.15 In the APOYO (2008) survey of industrial fishing workers, 40 percent report not working at all outside of the production seasons; a large proportion of the remainder work as artisan fishermen outside of the industrial seasons. The summary statistics in Table I show negligible changes in average incomes in fishmeal loca- tions during the production season. While the industrial fishing sector does have linkages to the local economies, economic activity thus appears to vary less with the production cycle than one might expect. The summary statistics also show little change in mean household demographics, so- cioeconomicstatusandlabormarketoutcomesduringproductionseasons. Differencesinhousehold characteristics between fishmeal and non-fishmeal locations are also modest, although households in fishmeal locations appear to be of slightly higher socioeconomic status. 2.2 The fishmeal production process, pollution and health Fishmeal is more valuable the fresher the fish when processed. Fishing boats therefore go out for at most one day at a time. Each fishmeal plant has its own docking station at the port. After docking at the station of the plant that has purchased the fish, the fish is offloaded and transported into the plant through a conveyor belt. Inside the plant the fish is weighed and cleaned. After cleaning, the fish is dried and converted into fishmeal by either exposure to direct heat or steaming. Fishmeal is storable for 6 – 12 months (but fishmeal companies report that they rarely store for long). Air pollution may occur in the form of chemical pollutants (such as carbon dioxide (CO ) and 2 nitrogen dioxide (NO )) from the plants’ heavy use of fossil fuels, in the form of noxious gases (e.g. 2 sulfur dioxide (SO ) and hydrogen sulfide (H S)) released as fish decompose, and in the form of 2 2 microscopic natural particles (PM10 or PM2.5) released during the drying and burning processes. Case studies have found high levels of air pollution near fishmeal ports during production periods, as discussed in detail in the appendix. As also discussed at greater length in the appendix, air pollution in the form of particulate matter, chemical pollutants and gases associated with fishmeal production has been shown to cause a range of health problems in adults and children, especially respiratory disease episodes. Peru’s fishmeal plants are also alleged to pollute the ocean by releasing “stickwater” onto the beaches or into the ocean (see e.g. Rivas, Enriquez and Nolazco, 2008; Elliott et al., 2012). Stickwater can cause skin- and gastrointestinal diseases and conjunctivitis in humans (a) through direct exposure and (b) indirectly, by stimulating the growth of pathogens in the ocean, which 14 In a country-wide survey of workers in the sector conducted by the consulting firm APOYO in May 2007, 87 percent report having worked for the same company or fishing boat owner throughout their career, on average for about 14 years (APOYO, 2008). Many workers are unionized and 44 percent have permanent work contracts. 15Earnings from work on industrial fishing boats can be high during the industrial fishing season so that boat- workers’ earnings may fluctuate more over the year (see e.g. Pereda, 2008). 7 can enter seafood and thus, ultimately, humans (Pruss, 1998; Fleming and Walsh, 2006; Garcia- Sifuentesa et al., 2009). 2.3 Regulations and the 2009 ITQ reform TheregulationsimposedonthePeruvianfishmealindustryareaimedatpreservingfishstockswhile maintaining industry profitability. In the North/Central marine ecosystem (down to the −16◦S parallel), semi-annual fishing bans were in place for industrial boats during the periods when the anchovies reproduce throughout our data period (the exact start and end dates of the ban periods vary by year). In addition, before the 2009 reform, industrial boats in the North/Central region operated under a sector-wide “Total Allowable Catch” (TAC). The size of the aggregate quota varied by season and was set by the government agency in charge of publishing official estimates of fish stocks (IMARPE). In the much smaller southern marine ecosystem, fishing was allowed throughout the year and no aggregate quota was in place before the 2009 ITQ reform. In 2008, officials estimated excess capacity in the industrial fishing sector (the industrial fleet and fishmeal plants) of 35–45 percent (Tveteras et al., 2011). Concerned about excess capacity and declining fish stocks, the goverment announced a new law introducing a system of individual, transferable quotas (ITQs) in the industrial fishing sector on June 30th, 2008. The ITQ law came into effect in the North/Central region on April 20th, 2009 and in the South on July 7th, 2009. In the South, the new ITQ system also meant that an aggregate quota and fishing ban periods were introduced for the first time. AllindustrialboatswereincludedinthenewITQsystem. Individualboatquotaswerespecified as a share of the regions’s aggregate quota for the relevant season. The quota-share was based on historical catches and a boat’s hull capacity. The quotas could be transferred between boats, subject to certain rules.16 Quotas could not be transferred between the North/Central region and the South. 3 Data We combine five different types of data: hospital admissions data, individual- and household-level survey data, administrative regulatory data, administrative production and transaction registries, and data on pollution. Hospital admissions data. InformationonhospitaladmissionswasprovidedbythePeruvian MinistryofHealthandconsistsofcountsofallpatientsadmittedtoanypublichealthfacility(health posts, health centers, and hospitals) between 2007 and 2011. The data is at the facility×month level and gives information on the cause for admission (using the International Classification of Diseases system). 16Firms or individuals that owned several boats/quotas were free to allocate their total quota to a subset of those boats. Quota-owners could also rent their quota out to others, for up to three years at a time. 8 Individual- and household-level survey data. The nationally representative Encuesta Nacional de Hogares (ENAHO) is the Peruvian version of the Living Standards Measurement Study (LSMS). Since 2004 surveying has taken place throughout the year, and the order in which sampling clusters are surveyed is randomly determined. A subset of clusters are re-surveyed every year. Information on the “centro poblado” where each respondent is interviewed is recorded.17 In our analysis, we use the GPS coordinates of the centro poblado’s centroid. Adult women and men are interviewed. The survey focuses on labor market participation, income and expenditures, self-reported health outcomes, etc., as in other LSMSs. We also use the nationally representative Encuesta Demografica y de Salud Familiar (ENDES), whichisthePeruvianversionofaDemographicandHealthSurvey(DHS).Thesamplingframework is similar to ENAHO, with surveying taking place throughout the year since 2004. A subset of clusters are re-surveyed every year.18 GPS coordinates for sample clusters are recorded. Women between 15 and 49 years old are interviewed, and information on their children (five years old and under) recorded. The survey is comparable to other DHS surveys, focusing on self-reported and measured health outcomes. For both surveys, we use the years 2007–2011. Administrative regulatory data. We coded the dates of all fishing seasons from 2007 to 2011 and the size of each season’s aggregate quota from the government gazette El Peruano. Administrative production and transaction registries. The registry of all transactions between industrial fishing boats and fishmeal plants (including “within-firm” transactions) from 2007 to 2011 was provided by the Peruvian Ministry of Production. All offloads by industrial boats are included, i.e., all (legal) input into fishmeal production. Information on the date of the transaction, and the boat, plant and amount of fish involved (though not the price), is included. We also have access to the ministry’s records of fishmeal plants’ production/output, recorded at the monthly level, from 2007 to 2011. Pollution data. Air quality measurement stations are found only in the Lima area. Informa- tion on the daily concentration, from 2007 to 2010, of four air pollutants at each of five stations in Lima was provided by the environmental division (DIGESA) of the Ministry of Health. The measured air pollutants – PM10, PM2.5, NO and SO – have been shown to correlate with factory 2 2 production in many contexts and are commonly used in the health literature.19 Wealsouseweeklymeasurementsoftheconcentrationofcoliformsatpublicbeachesthroughout the country. This information is recorded by DIGESA to inform beachgoers about beach/water safety, and is reported as a binary variable. 17Centros poblados are villages in rural areas and neighborhoods in urban areas. After the sample restrictions we impose, 2096 sampling clusters with on average 77 households each are present in our sample. 710 centros poblados are present, with on average 228 households each. 18From 2004 to 2007, a fixed set of 1131 clusters was used, the survey order of which was randomized (as was the trimesterofsurveying). Thedefinitionofclusterschangedsomewhatin2008whenPeru’sstatisticalbureauupdated the sampling frame with the 2007 national census. Furthermore, 2008 was unusual in that only 722 clusters were surveyed. From 2009 to 2011, 1132 clusters were used, including a panel of 566 clusters surveyed every year. 19We lack data on the concentration level of H S and CO , which may also respond to fishmeal production. 2 2 9