Scribd Logo
Upload

Welcome to Scribd!

  • Volatile and Spatially Varied - The Geographically Differentiated Economic Outcomes of Resource-Based Development in Peru, 2001-2015 - Elsevier Enhanced Reader

    Uploaded by

    Daniel
    11 views13 pages

    Original Title

    Volatile and spatially varied_ The geographically differentiated economic outcomes of resource-based development in Peru, 2001–2015 _ Elsevier Enhanced Reader

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    11 views13 pages

    Volatile and Spatially Varied - The Geographically Differentiated Economic Outcomes of Resource-Based Development in Peru, 2001-2015 - Elsevier Enhanced Reader

    Uploaded by

    Daniel

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    of 13
    ‘The Extractive Industries and Society 6 (2019) 1143-1155 ELSEVIER The Extractive Industries and Society journal homepage: www.else\ Content lists available at ScienceDisest ccomilocate/exis Original article Volatile and spatially varied: The geographically differentiated economic outcomes of resource-based development in Peru, 2001-2015* José Carlos Orihuela’, Victor Gamarra Echenique” * Pol Unter Cais dl er, er pont Uno Cases dl er Le, Pr ower seat, “The macroscnomie impacts of resource-based development are diverse across national space. However, the more export dependent snd geographically peripheral the region, the more susceptible 1810 boom nd bust ‘ces, We typify resource based economic development in Peru inthe period 2001-2015, analyzing regional ‘export specialization, growth volatility, and de-indkstralization, thre resource curse symptoms. With the ‘commodity cycle: (i) expor speciation f not the same i all mineral eons (i regional growth volt rach higher at regional level than at national level: andi therein convincing eae of de indastrlizaion and the Duteh disease, because a world economy surge does noc operate asa national resource discovery, We sy ‘economic evolution within resurcesich Pera is volatile and spatially varied. At the national level, gol ‘opperslependent emu is nt as vlerable as other mineral dependent counties to extemal shocks. At the subnational level, growth volatility very high foresters of regions. Eeonomle geography sade ean con tribute co challenging poplar resaurce cure aeounts ofthe development economies trace: we should be asking where, when, and Wit there seurseorbesing, and what type of rater than searching fora definitive epioa! rom vttty zoo geoph ‘universal answer on the developmental effects of mineral abundance 1. Introduction Do natural resources curse the economic development prospects of nations? Or if not the prospects of nations, those of their resource-rich regions? Adding to the new economie geography on resource-based development (Arias et al, 2014; Rehner et al, 20145 Phelps et al., 2015), our evidence shows that development based on natural re sources is geographically uneven within countves. Similar claims have been made by economic geographers in relation to industrially so phistcated, non-resource-dependent economies (Main, 2000; Hayter, 2004; Boschma and Frenken, 2009; MacKinnon et al., 2009), but we find them applicable to developing, resource-dependent ones aswell Minerakabundant Peru, the Latin American economy with the best macroeconomic indicators during the last commodity supercycle, makes an appealing ease to inform the resource curse debate; indeed, Peru has become a recurrent case study within a growing local resource ‘curse literature (Cust and Poelhekke, 2015; Manzano and Gutierrez, 2019, Orihuela et al 20198), Studying the 2001-2015 commodity ‘eyele, we contribute tothe resource curse debate by documenting that ‘economic development within emblematic resource-rich Pena is volatile and spatially diverse ‘The resource-based development debate is commonly famed as 0 ‘curse of blessing” dichotomy, and subsnational variation was not previously explored (van der Ploeg, 20715 Frankel, 2012; Venables, 2016; Orihuela, 2018). This flaw can certainly be attributed to dis: ciplinary paradigms anid methods, as for the most part the economists who hape the debate have been searching fora general rule applicable to national economies (Sachs and Warmer, 1997; Lederman and Mahoney, 2007; van der Ploeg, 2011; Orihuela, 2018). Recent eco omic literature has opted for a “local resource curse” line of research, sven the wide availability of microdata that has recently opened up, Yet we sil observe little attention to sub-national variation. Thus, new quantitative methods based approaches exploit national household survey data to find out whether mining districts or local commu: nities—rather than mining countries—are cused or blessed (Caselli and Michaels, 2013; Alleott and Keninston, 2014; Cust and Pocthekke, 2015; van der Ploeg and Poclhekke, 2017; Manzano and Gutierrez, 2019). Then, the evidence for loeal resource eurse appears also incon: elusive, In the case of Pers, some studies find mixed outcomes or scattered and disappointingly modest benefits, given the significance of “We are very grateful to Pedro Francke and PRSCO workshop participant for useful comments and suggestions *coresponding author. ‘mal addresses ohuclaeapucp.edu pe QL. Ohuela, vitor gamareaeq@puep pe (V. Gamara Echenigque) hitps://dokong/10.1016/}xis2019.05.018 Received § Febrry 2019; Received in revised form 31 May 2019; Ascepted 21 May 2019 ‘Alle online 02 Sly 2019 2214790K/ © 2019 Elsevier Lid. All righ reserved 16. Onto. Gamer Ero the macroeconomic Doom (Zegarra et al, 2007; Tieel and Eseobal, 2015 Orihuela et al, 2019b), Others are more optimistic, but point out that positive outcomes appear to vary across space and over time Aragin and Rudd, 2013; Loayza and Rigolini, 2015; Orihuela and Gamarra, 2019). Should mixed outcomes—within and across cour. teies—not be the norm? How ean quantitative research promote a more fruitful scholarly exchange? We contend that documenting within ‘country spatial variation eas for rethink of conventional analytical frameworks and the bold statements that often go with them. Going beyond curse or blessing, ve call for complexity and variation to be ‘embraced in the study of natural resource based development: resource ‘curse phenomena are contingent on time and place, to begin with. ‘To contribute tothe field of resource curse studies witha more time: ‘and-space sensitive approach, we analyze the macroeconomic perfor. mance of Peruvian regions. We do so by dravving on the empirical work fof Richner et al. (2014) and Jarreau and Poneet (2012), finding an ‘economic geography approach that puts the hypothesis of uneven de velopment at the center ofthe discussion, What we show statistically is ‘2 picture of diverse subnational macracconomie dependence an more than a mining product, To begin with, Peru {8 not a mineral mono: ‘export economy: the not only trades copper but aso gold, zinc, and ‘other metals, although gold and copper have a higher commercial vo lume. Moreover, as commodity prices do not always covariate and re sources are not distributed uniformly throughout an intricate national terctory made up of ard coastline, Andean mountains andthe Amazon, it is problematic to expect subnational economic convergence. We find five clusters of regions when looking into the relationship between ‘export dependence and variance in regional economic growth. And (0 ‘add nuances to an already complex portrait, export specialization in tensified along with price surges in one group of regions, but not in ‘another. Finally, we find that in most—but not all—cases, and con sidering that the less dependent and wealthier regions are ll coastal, the more export dependent and geographically peripheral the region, the mote susceptible it is to boom-and-hst eles as documented by the ‘variance in regional economic growth rates, ‘The remainder ofthis article is organized as follows First, we define the main features of resource-based development in Peru. Then, we present a basie analytical framework forthe empirical study. Next, we ‘set out our methods and data, followed by the results and a discussion based on the quantitative analysis. Finally, we conclude by highlighting lessons from the case study and their implications for advancing an ‘evolutionary economic geography perspective of the resource curse. 2. The ease of resource-dependent Pers Peru is resource eh country, but one with a diversified resource: based economy. Natural resources are distributed unevenly across ‘ecological (Pacific Ocean, arid coast, Andes, and Amazonia) and pol Leal (Cwenty-five) regions. Moreover, the modern economic history of Peru is characterized by numerous boom-and-bust cycles, sizeable i legal economies of coca inthe Andes and gold and timber inthe Andean Amazon, and a national politial-economie system centralized in the ‘canstal megaeity of Lima. By “diversified resource-based economy” = mean, fist, that mining exports are important, but their collective historical significance varies. Mining exports have fluctuated between 10% and 70% over the last 125 years (see Fig. 1). Second, mining ex: ports mean diferent things a different points in history, fom silver and Ritates in the 19th century to copper, gold, lead, oil and natural gas, silver and 2ine in the 20th and 21st centuries. Third, these resource ‘economies are distributed throughout the national territory, in spatially lifferentiated ways. And while some early mining regions remain ac tive others die and new ones arise; gold eyeles, in particular, have both killed off and restore life to a subset of mineral regions (Thorp and Bertram, 1978; Thorp etal, 2012; Dargent et a, 2017). Such historia and spatial complexities qualify “resource-abun- danee/dependence”, to the extent that one may want to label mineral ‘The erat nts Soe 6 2019) 1149-1155 Fig 1. Mining and Oil Export in Chile and Peru (Percentages) 1900-2015. rich Peru a historial case of diversified dependence, in contrast to minerabrich Chile to the south, which isa historical case that more closely resembles mono-commodity dependence. For example, in Chile, mining meant copper in the mid-nineteenth century, nitrates in the 1870s 0 the 1920s, and since then copper again, alos all of which i, located in the northern regions. In that country, and unlike Per, Fig. shouts that ining exports have never accounted for less than 50% of al exports over the last 125 years. In tur, Table 1 presents the main commodities in the export basket and macroeconomic indicators. The structural reforms of the 1990, which liberalized the mining and oil sectors, met with a new, positive international-eeonomie outlook.” Put together, reforms, foreign capital and commodity prices led to a new export eyele in 1993-2015. We now turn to key contemporary features of economic geography, focusing on how export diversity benefits Peru and the spatial form this diversity takes. As we have noted, and as ean be seen in Fig. 2, Perw exports a number of minerals but gold and copper are by far the most Important ones, accounting for 24% and 19% of exports over the last five yeas, respectively.” During the 2008-2009 crisis, while gold prices kept escalating, copper prices fell drastically (see Figs. 2 and 3). The gold price trend changed after 2012, this time moving along with copper and greatly impacting Peru's exports (see Fis, 3). The following two maps (Fig. ) show the geographical diversity of export special tation, and therefore mineral dependence, within Pers. Regional mi eral exports vary in size and composition. From south to north, copper 's prevalent inthe southemmost regions of Moquegua and Tacna, which border Chile. Mineral deposits in Arequipa and Cusco are copper- and gold-rich.* The coastal regions south and north of Lima are mineral rieh but diversified: lea fas large-scale iron mining and export agriculture; "While most mining takes place in the Andes, most ol activity located in Amazonia, Amazonia i also the ste of (eal and egal) gold mining, logging and, more recent, oil palm plantations. Finally, the Pacific Ocean yielded fun export in the 19th entary and shmeal expors since the mid 20th fener. "etween World War Il and the Debt Crisis, Pen’ resource dependence ‘moved fom fisheries inthe Pacific Ocean to olin Arszonia nn the northern Continental shelf and copper i the southern regions of Maquegua and Taena (the Perwian side ofthe Aacima Desert, which also host copper in Chil). ‘There was no foreign investment in mining troughs the 19805 *The county is ranked among the biggest mineral prosucers in the word it ts the thintlargest producer of copper, the sath of gld, and the second of sive. * West of Cusco, small and poor Apuriae is siver dependent an exceptional cae. In te, sented ona sity ifesent geology, the ental Andes ate lead Fie, but copper and zine ae also significant. To the eat, the peripheral Andean ‘Amazonian regons of Pun and Madge de Digs ae goi-ieh and eonstite extreme casts of export dependency. As these two frontier reglons coastitte the epicenter of egal gold mining, expors Figures are incomplete (legal logging and coa are theater noticeable resource exporslinke activities over there and ehewhere in the Eastern Andes or Andean Arson). 16. Onto. Gamer Ero Groth, Exports Growth and Leong Exports in Peru 1945-2018. Source: Authors own eleuation, based on BCRP, ORLAD y WD, ‘er the Andes clas prod Copper, gol Sd and tn entra ‘copper loge and Tacs pole i : 1 : i i eh foe (ace Ocen-Anee copper nonthem Coney eh Mou Pie Oo ‘ner copper Modan and Tae) aor Rate Grom ean 1 Rate Gre a Nove: @) 1961-72, ‘The erat nts Soe 6 2019) 1149-1155 lg. 2, Peru's Export Performance, 2001-2018. coppertich Ancash, in tutn, is more diversified because of is fishes industry. In the northern Andes, goldvich Cajamarca and La Libertad constitute another geologial cluster, this time gold-based.” Pura has rot welcomed new mining (in its large-scale form, but illegal aetvity takes place regardless, but benefits from cil. Finally the northern coastal regions are more diversified because oftheir sea access and ir rigation infrastructure." In short, few regions are not mineral depen- dent, but dependence means quite different things, aeross space and over time, Between 2001 and 2015, see Fig. 4, the pie charts showing export structures changed in size and composition. The most significant in- ereases to0k place in Arequipa, Puno, La Libertad and Madre de Dios, beach of which became important gold exporters; and the central regione of Hudnweo, Pasco and Junin, as well ¢ Ancash, which became more copper-dependent. In turn, more drastic changes in composition took place in Ayacucho and Apurimae, which went from not exporting any ‘metals in 2001 to exporting gold and silver by 2015. Clearly, Peru's fscanomie geography appears to be even more complex than that of resource-rich Chile, a country subject to various ease studies in the recent literature, including those from which we draw our methods (Arias et al, 2014; Rehner et al, 2014; Phelps et al, 2015). With a larger territory (1.275 million km2), 60% of which pertains 0 the Amazon basin, and a larger number of political regions, Peru has sreater diversity in terms of ecological and politcal regions and a more Varied poo! of natural resources, minerals in patiular. 3, Analytical framework ‘Though counties ave not fa, the economic geography of Peru in particular, most of the resource cuse literature approaches them as if they were space-free entities. While the term *resouree curse” Is com monly attributed (0 geographer Richard Auly (1994 and 2001), who argued thatthe underperformance of resource-rich countries is due to weak domestic linkages of mining, there is very litle economic geo- graphy research on the “curse. However, the case of Chile has attracted some attention (Arias et al, 2014; Rehner etal, 2014; Phelps et al. 2015), AS Hayter et al. (2003, 17) pat it, economic geography “has no Interest in what is perceived as ‘old fashioned! economic geography Thus, contemporary research on the so-called global commodity chain, tlobal value change, and global production change overlooks extractive industries. In patitlar, resource peripheries feature nominally or are ‘ignored altogether in the burgeoning evolutionary economic geography GBarnes et a., 2001; Hayter et al, 2003; Bridge, 2008; Patchell and Unlike small-scale god mining inthe south ofthe country, both legal and igs, there are large open pt developments inthe nother Andes, "Of the remaining regions, some have neither mineral nor ol endowments, bt are subject nonetheless to other kinds of reeoure-crsephenomen ‘notably San Mtn ith aca. Amazonian Loreto and aya are wood and Fish exporters, snd the cost Tubes exports Fish and shes, 16. Onto. Gamer Ero = / 25 An VU ih VN ped Hayter, 2013; Martin and Sunley, 2015). ‘Turning to development economics, resouree-curse postulations ean be traced back to Adam Smith, who singled out mining as an economic ‘activity that reduced the wealth of nations (The Wealth of Nations, Book IV). The resource curse thesis per se gained popularity following ‘Sachs and Warner's (1995) erosscountey econometric nalsi, Which linked dependency on resource exports to poor economic grovth rates. ‘The association between natural resource abundance (dependence) and ‘poor economie development became highly contested, however. Con ‘hisions differ subject to sampled data, variable definitions, and model specification (Lederman and Maloney, 2007; Brunnschweller and Bulte, 20055 Wick and Bulte, 2009; van der Ploeg, 2011) Partly as a response to these problems with the cross-country ‘econometric literature, and partly a result ofthe mierodata that is now ‘available, « loeal or subnational resoutce curse literature ie currently ‘emerging among economists and political scientists (van der Ploeg, ‘2011; Ross, 2015). Within-country economic studies, however, also point to both curses and blessings related to natural resource abun dance and/or dependence. The empirical assessment of resource curse hypotheses is subject to dependent variable choice as well asthe te pporal and spatial scope of analysis (van der Ploeg, 20115 Cust and Poeliekke, 2015; Orihuela, 2018). The same applies to resource curse ‘economic studies on Peru: the relationship between natural resources ‘and economic development is neither simple nor wniversal (rags and Rudd, 2013; Ticct and Escobal, 2015; Loayza and Rigolini, 2016; ‘Orinela eta, 20190); besides, there is als confiet (Bebington eta, 2008; Arellano-Yaguas, 2011; Paredes, 2016; Gruber and Orihucla, 2017; Orihuela et al, 20198). 8, Prices of Principal Mining Products and Exchange Rate, 2000-2015. ‘The erat nts Soe 6 2019) 1149-1155 oo yh 7 \ Went me While empirically disputed, in theory there is mare than one pos- sible type of “resource curse”. There are various possible classifications, including those of Van der Ploeg (2011), Aragén etal (2015), Cust and Poethekke (2015), Gilberthorpe and Papyrakis (2015), Van der Ploeg and Poolheike (2017) and Orihucla (2018). Resouree eurse varieties Inchude economic, political and environmental phenomena, which ean certainly be subject to interplay. Our research deals fundamentally with the harmful volatility and the Dutch disease hypotheses, two of the ‘most poplar types of resource curse and postulated fo be behind low national growth rates and political turmoil ‘Volatility arguments are diverse. Van de Ploeg (2011) and Aragé tal (2015) explain that commodity price volatility could lead to ur certainty and therefore reduce investment in physical or human eaptal as well as expenditure on other kinds of learning and teehnological Innovation, which in turn may reduce productivity and economic south inthe long tun. There is also the argument that i leads to poor Financial market development. As regards the short run, price volatility fan produce balance of payments crises and/or may fuel poltical conflict through various effects and mechanisms (Prankel, 2010; Van de Ploeg, 2011; Orihuela, 2018), Cross-country research indicates that its foreiga debt rather than resource abundance what causes low economic growth rates, debt exposure being related to commodity price volatility (Manzano and Rigobsn, 2001; Hausmann and Rigob6n, 2003; Van der Ploce, 2071), In turn, economie historians point to the centrality of volatility in shaping the political economy of development in Latin America (Prebisch, 1950; Thorp, 1998; Bértola and Ocampo, 2012). Looking into the subnational level, in the context of poor and devel. ‘ping national economies, regional specialization fs key for subnational 16. Onto. Gamer Ero Pana A: 2001 ‘The erat nts Soe 6 2019) 1149-1155 Panel 2018 Fig. 4, Reglonal Exports Stictre of Per, ‘growth volatility (Auty, 2001; Arias etal, 20145 Rehner et al, 2014). Dutch disease theories, the most popular of the resource curse theses, link resource abundance to bad economic performance: eX: ‘change rate appreciation crowds out altemative export sectors, mane faeturing in particular (Corden and Neary, 1982; Gorden, 1984 Krugman, 1987), More widely, Dutch disease theories expect resouree booms to generate rise in demand of non-tradable goods, which in tum leads to a reduction in the allocation of labor and eapital to the manufacturing setor In shor, when a national economy experiences a resource boom, either because of terms of trade improvement or re source discovery, the manufacturing sector, or any other traded sectors such as agricultare, will ikely shrink and the non-trade sector will ‘expand (Sachs and Warner, 1995; Van der Ploeg, 2011). According to ‘Sachs and Warner (1995), within a competitive markets perspective, & shrinkage in manufactures is only a problem when manufactures ereate ‘externalities in production, which the authors calla speculative and yet, unproven claim. Theoretical disputes aside, the empirical evidence on the Dutch disease is mixed (Sal24-Martin and Subramanian, 2003; Van der Ploeg, 2011; Aragon etal, 2015). Acknowledging the complexity of potential resource curse me- cchanisms and equilibria, our research assesses export specialization, volatility and de-indusrialization symptoms across political regions in Peru, Our general hypothesis is that these phenomena are likely to be present, yet in diverse forms because commodity dependence is not uniform and economic geography matters. On commodity dependence, mineral economies can be very different, beginning withthe mineral {hat are involved: mono-dependence and basketdependence are not the same thing. On economic geography, mining regions can be highly diverse in terms of resource endowments and ecological conditions (as «liscussed in the previous section, integration with global value bain, ‘domestic agglomeration and learning economies, as well as institutional ‘context, to name the ustal factors of economic developanent. In partis ‘ular, the institutional dimension, which we believe goes beyond con ventional corruption and poliiel resource eurse aeeounts (Robinson Cal, 2006 Brollo eta, 2010; Caselli and Michaels, 2013; Boutilier, 2017), is commonly called upon to explain why some resource abun- ‘dant economies succeed while others fail (Gan det Ploeg, 201 Orthucla, 2018). Because resource curse effects take place conditional to temporal and place-elated politieal-economie phenomena, mixed ‘outcomes and uneven subnational development are likely co preva 4. Methods and data ‘To assess the hypothesis of spatially differentiated resource-based, conomie development in Peru, we dravs on the methodologies of Rehner et al, (2014) and Jarreat and Poncet (2012). Throughout the discussion, by “the case of Chile” we refer to ehner tal, (2014). We explore three resource-curse related phenomens, using & set of quan- Uiative approximations. Specifialy, we test three hypotheses: (a) the specialization hypothesis, through an analysis of regional export spe clalization of mineral-rich regions and their recent evolution; (b) the macroeconomic volatility hypothesis, through an analysis of the re lationship between export specialization and GDP growth ona regional basis; and (e) the Dutch disease or de-industialization hypothesis, through an analysis of regional export sectors, their growth and growth variability. As such, we proceed as follows: () Regional export specialization and its recent evolution: using the methodology of Recher et al. 2014), we estimate the Herfindahl- Hirschmann Index (HHD, which isan index that is within the range ‘of 1/n and 1.0, where ‘nis the number of products in the Peruvian ‘export basket. Because ofthe high export specialization of some of the regions, some inthis HHT are lose to 1 (Rhodes, 1993; Calkins, 1983), while other regions are close (0 0 (such as the region of Lima, which is the most divesitied of all) (©) Export specialization and regional economic growth volatility: in this case, we relate the export specialization of Peru's regions (using the HD and the regional GDP growth rate, taking averages from 16. Onto. Gamer Ero ‘oth indicators for the period 2001-2015.” We classify the regions in clusters, using the non-hierarchical cluster technique, the ‘k ‘means’ methodology, and classifying the variables mentioned above using their own value and their standardized value (zscore® stan dardization).” In addition, drawing on Jarreau and Poncet (2012), ‘we earty out regression analysis for alternative robustness cheeks, Regressions assess the yearly growth rate of each region and the volatility of growth (average). We use panel data fixed effects for the first variable and pooled regression for the second one.” (© Export growth and exploratory Dutch disease analysis: for this ‘purpose, we look at the export growth rates of the main resource ‘products the volatility ofthe international prices of those products, the relationship berween these variables at different time periods for 2001-2015, and finally, some stylized facts about mining and ‘manufacture GDP and exports, We Focus on gol and copper exports due to their greater significance forthe national economy. We do so bby comparing export growth rates beoween sectors and yeats using Tess.” We obtained our database from the Ministry of Trade and Tourim's Integrated Foreign Trade Information System (Siena Integrado de Informacion de Comercio Exterior, SIICEX), which eompiles data on re: ¢gional exports in FOB prices for the period of 2001-2015." In turn, we ‘obtained the information about the international prices of the resources from Bloomberg. Finally, we obtained the information on regional GDP ‘and the US dollar-Peruvian sol conversion from the National Institute of ‘Statistics (INED and the Gentral Bank of Peru (BCRP), respectively. The ‘series on regional GDP isin real terms, with 2007 asthe base year. All such series were caleulated using the same methodology (INET, 2015). 5. Results and discussion: macroecone diverse ‘effects are spatially (@) Growing export specialization in some cases, diminishing in others At frst sight, we ean see that coastal regions have the lowest HHL (6ee Table 2). Lima (which has some mining) and the port city of Callao (hich today, along with Lima, constitutes a single megacity of ten million people) have the lowest export specialization indicators, fl loved by other coastal and non-mining regions. Taking the group of * Although we ws the average HH, and tha afew regions have HAT tha varies considerably, we conser the average as central measure to carryout ‘ur analysis. The tse of heii HE ane the Final Hin the hstr analysis ‘des at significantly change the composition ofthese. "This vale f obined by subtracting the mean ofthe obseriations an di ding bythe standard deviation, othe mean and the standard deviation ofthe resting variables are 0 and 1, respectively. "To determine the optimal numberof luster, we use the stoppng-ule value methotology (Cains ad ara, 1974). This methodology computes an Index for eat luster soluon, whee args (or smal vale) indieate more lstnguished clustering, The technique asses nimber of chasers tha provides best simile intcluster characteristics and best different characterises be tween cases "ease panel data for growth ates becuse our main dependent variable, regional growth rate, available for each year of analysis In the ease ofthe ‘olatiliy of growth, we use as main dependent variable the variance of eo omic growth 2001-2010), this we have same vals foreach regio i all yeas "Specifically, we compare gold, copper and overall mining sector export growth ates individually, withthe non-sning exports, at diferent intervals of Time, This allows us to explore whether mining sectors have better export performance diring boom, eis snd overall ime period, based on economic ‘growth performance. "Although the database had very smal evel of disaggregation, there moval oenteme voles di not fet the indices, We climinated al tem that ha an export value offs than USD 1,000. ‘The erat nts Soe 6 2019) 1149-1155 ‘mining regions alone, the average HHI value for non-landlocked mining regions excluding Lima was 0.208 in 2001 and 0.200 in 2015; while the average HHI value for landlocked mining regions was 0.573 in 2001, and 0.485 in 2015." Some of the coastal regions, however, have a relatively high HHI vole, as they are highly dependent on resource exports Export specialization does not always increase along with mineral ries. tn particular, minerabvieh Taena and Moquegua (southern re sions next to Chile and rich in cooper deposits) defy the export-spe- Claliztion hypothesis. The mining boom cycle brought about a de- crease in HHI values, much more substantial in the case of the less dependent Taena'”. The dry coastal region of lea is another ease of high complexity. Compared to Moquegua and Taena, mining Is less im porant. Instead of copper, ea has one large ion mine in the hands of a Single Chinese enterprise, which of course exports the ion to China, in addition to small-scale informal gold mining. In common with nor- ‘mining, coastal Par in the north, fea has one ofthe lowest HHI scores, after Lins and Callao! Mesnwhile, in contrast with the likes of lea and Piura (or Moquegua and Taena), southern landlocked mining re sions have become more export specialized ‘© Variety of region types and regional growth votatlty higher than chile’s We auddress the relationship between export specialization and economic growth volatility through eluster analysis, using average HHI and variance of economie growth as inputs. We document two im: Porcant differences compared with regional economic growth in Chile: @ a more complex typology is needed to depict the relationship be "ween export specialization and variance of regional economic growth; and (i regions in Peru have undergone much greater economic grovith volatility, three times more on average, with a sub-set of cases of ex treme dependence/volaility. While the basic economic geography of Chiles composed of the mineral-rich northern regions, the agricultural- and manufacturing based Central Valley, and the water- and forestrich southern regions, Pru has mining all over the Andes, some on the coast and some in the Amazon, while export agriculture takes place funds- mentally along the dry irrigated coast. However, iis important to say that regions with high growth volatility are not directly linked to the specialization and dependence on exports of basic products. Extreme Volatility is presented in mining regions as well as non-mining regions. Also, some regions that are resource-dependent may not have mnich volatility of growth as other non-dependent ones (specifically those that are gold dependent) Intemational trade dependence has both pros and cons. Amid 2 good International context, exports an therefore aggregate income Increase. However, those regions that are highly specialized inthe ex Dr of rane materials are more sosceptible to extemal shocks, which ‘makes economic growth for such regions highly volatile and ultimately unsustainable, There have been no visible efforts to cope with this de- velopmental challenge. Regional governments now administer larger bbudlgts than ever. Inthe last two decades, fiscal decentralization and the implementation of a scheme for fiscal mining-rent transfers to "SNomandlocked regions are Ancash, Arequipa, ca, La Libertad Lambayeque, Moguegu, Pita, Tacna and Tambes (ee Table 2) "Landlocked regions ate Amazonas, Apuriae, Ayscucho, Cajamea, Cusco, Huaneaveles, Huse, Juni, Loreto, Madre de Dis, Paseo, Puno, Sn Marin and Ussyali "© Ther proximity to both Cle and Bolivia are Features tat characterize the regional development af theie two regions. For a fure historical analy, i has also co be considered chat large stale mining started in Moquegua sind “Taco decades before the pos-1980s mining cycle (se Tle 1 above. "(es important to mention that fe has along taion of grape and psco production (agroindustry), which was Increase in recent years to meet the ‘pear in world demand. 16. Onto. Gamer Ero ‘Table 2 Export Speciation ~ HHL citferentiated by Region, ‘The erat nts Soe 6 2019) 1149-1155 = = a ae = — Boose 2 i Grow a r i crows 9a Da ene a ‘a erage 200.2015) Fig. 5. Expt Specialization and GDP Growth Sabity, 901-2015, regional governments and municipalities, the so-called canon minero” (rellano-Yaguas, 2011; Gruber and Orihuela, 2017), have increased the input of subnational polities in regional development. A remedy for ameliorating high regional growth volatility would thus be eounter- ‘yclical fiscal policy caried out by regional governments. There have been no such institutional developments, however; quite the contrary, regional government expenditure is pro-yelial. At the national level, a stabilization fund was implemented a decade ago, although much smaller than that established by Ministry of Finance economists in Chile (Grankel, et a., 20135 Orihuela, 2012), ‘The analysis of export specialization (average HHI) and the varia: Lion of regional GDP growrh rates are presented based on the standard deviation of the regional GDP, se Fig. 5. Compared to the Chilean case, Peruvian regions appear hyper-voltile. Unlike the Chilean ease (whieh only made three clusters), k = 5 clusters had the best performance."” Cluster 1 regions have low HHI and relatively lower variation in eco nomic groneth. In this eustr are the coastal regions of Lima and Callao, Piura, and Lambayeque, as well as Amazonian Loreto and Ueayall AS Fig. 6 shows, no Andes-centered regions are located in this cluster. The average HHT ranges from values close to 0.050 to mean values of 0.350, Where the mean is 0.167. What these politial regions have in common fs that they are not eminently mining regions, This is evidence that * Guster anal toppng rules are used to determine the number of clus tes. A stopping rile vale (also alle an inden is computed for each easter solution Large values (or smal, depending on te particular stopping rule) indicate more distinct clustering. We use Calis and Harabase's (1974) stop ing rue, which consists ofa pseudo index, where larger values indicate more Aisin chserng. 16. Onto. Gamer Ero Hy Group 5 Group 4 @ Group3 DB Group 2 CGroup 4 ‘The erat nts Soe 6 2019) 1149-1155 Fig. 6. Spatial Distebution of Exports Clusters in Per, 2016, regions of low export specialization and without many mining exports tend to have a more stable and less volatile economic growth. In turn, ‘Cluster 2 presents low HHI but higher GDP volatility than Cluster 1. Here we have the mining regions of Junin, Husnuco, Tea, Arequipa and Tacha, Junin, Arequipa and Hudnuce have a diversified export basket, yet are dependent on mining resowees: gold, silver, cope lead and Zinc. ln the ease of Tacna and ca, mining has a relatively lower sig nificance among regional exports. 16. Onto. Gamer Ero ‘Table 3 Descriptive Statistics of Clusters ‘Source: Author awn calculation, bas on SIICEX and INEI data Exports and Real GDP messured in milion do ‘Average HAL Var of growth Exons Real GDP Bxpen/GP ser? 0259 00041 taka 2707 soo uter3 0582 0.0020 eno 248 ae laters 6367 0.0007 saa ha 75% ie a ee een ee ee oe Cie 4 (but not as much as Cluster 2) and much higher export specialization. ‘Within this cluster are the gold export regions of Cajamarea La Libertad ‘and Puno, ia addition to Huancavelica, Amazonas and San Martin, While deprived Huancavelica is a region that exports several metal, ‘Amazonas and San Martin have no mining, Thus, in the period of ‘analysis, there was relatively stable economic growth when the region's main export product vas gold, as well as non-mining produets. In other ‘words, gold looks less likely to become a curse for short-term regional ‘economic grovth (for long term grow, curse/blessing will depend on ‘where gold rents are being invested). Gold prices are not as volatile as those of copper. Fg. 3 shows that gold prices continued to increase even ‘during the world crisis in 2009, but alike the others metals the price stated falling in 2012, ‘The next two clusters are those with high GDP volatility; the regions Involved are quite diverse. Cluster 4 groups together HHI values of between 0.200 and 0,600, mediumm-level export specialization, along with high regional economic growth variance. The regions in this cluster are Aneash, Cusco and Ayacucho in the Andes, and coastal “Tumbes, onthe border with Ecuador. Its noteworthy that Ancash and Cusco exports are both highly dependent on copper (7S and 85 percent, respectively) while the Cusco economy is quite dependent on tourism In turn, Ayacucho constitutes an exceptional case. At the beginning oF the period, the region was not dependent on any mineral resource; its basket of metals was subject to variation over the years. Finally, ‘Tumbes’ story is completely different because its exports come from fisheries. In tur, Cluster 5 Is made up of eases with extremely high ‘growth volatility along with high export specialization: Apurimac, Madre de Dios, Pasco, and Moquegua. Although their exports are con. ‘entrated on mining prodts, this constitutes a rather heterogenous set ‘of regions in terms of natural resource endownents. There is one case of ‘ copper-dependent region, Moguegua, and one gold dependent region, ‘Table 4 ‘Growth Rate and Volatility of Growth Regressions, 2001-2015. ‘The erat nts Soe 6 2019) 1149-1155 Maaire de Dios. The other two regions have diversified baskets, ‘As Fig. 6 shows, there is no clear spatial pattern of clusters. In turn, “Table 3 presents descriptive statites ofthe clusters. As we ean see, the contribution of exports to GDP may also bea key factor in its volatility, particulary fr ister 5, the most volatile, The extreme dependence on exports (602%) makes the GDP highly dependent on international prices. ‘As robustness checks, we carry out regression analysis, drawing on Jagreau and Poneet (2012). The baseline specification consists in linking growth rates (or their volatility) and exporis concentraton/ Sophistication by using a growth determinants regression, controlling for human capital, inital real GDP per capita, and regional investment rate, The results are shown in Table 4, First, we show that, on average, regions with more concentrated export baskets (high HHD are asso clated with higher growth rates volatility (columns 4, 5 and 6). This s breeause the relationship i postive in Fig. 5 (AF we trace a linear rela tion, the scope will be positive). Thus, being located in the coastal re tion, the most diversified, reduces significantly the volality of growth, Consistent with our cluster analysis. Moreover, a higher investment rate may be associated with lower volatility, but the relationship isnot ae strong as the aforementioned determinants, The results indicate that seographical characteristics predict growth volatility; in tum, the 'grouth rate seems not to be affected by the HHI, but by human eapital formation (positively) and inital GDP per-capita (negatively, as in a ‘conventional Solow growth model). ‘Volatile mineral exports and no evidence on Dutch disease In Table 5, we compare exports growth rate of gold, copper and total mining versus the average of other sectors (non-mining sector). While Pig, 2 shows that other exports kept growing along with copper and gold exports in 2001-2018, Table § breaks the information down by relevant time periods: 2001-2008 (expansion), 2008-9 eriss), 2009-2012 (recovery) and 2012-2016 (end of eyele). The 2008.9 crisis, in particular, had different outcomes for different export sectors. This, fundamentally, is because while the prices of copper, zine and petro- leum fll sharply as global demand shrunk, gold prices continued in «easing (se Fig. 9 above). Copper (ike oll) fs much more volatile than gold, asthe latter works as an asset of last resort. Thus, the 2008.9 Change in international conditions had a different impact on mono- dependent Chile than it did on basket-dependent Peru. However, de- terioration in international conditions does not always take the same form. While the gold price kept climbing during the crisis of 2008-2009, the 2012-2015 dowsntuim meant the same for copper and gold prices, and gold exports fll more sharply (see Table 5). Dependent Varian ‘Geow Rate ‘Vouatity of Gow © @ o o © o Ta ea GDP per aia 0026 anna 0.028 oosree bara ‘oma (0002) (022) (oon (oon (0007 (0.06) ‘get 005" ‘os 0s 062+ acre fowae™ (0003) (0005) (ooo (0.09 (0.0%) (0.08) Invest te otra “aoe “04 erro “oon coaoe (oa) (001) oot) o.002) (0.002) (002) Human opts apr zie 0% “pons ‘os (oor) (019) (0054) (0053) (04) cus oi oie prc “aare (0.005) (0.008) (01s) 012) ‘ons eon dues “0.001 0.002} onmamt aoa oan 024s! =osises (0025 (0.025) (068) (0H Omervatons 0 0 260 360 "Notes Columns 1, 2and 3 are panel daa ted effcs regressions, while colamas 4, 5 and 6 re pooled egresion. °" Sigaicant atthe 1 percent level. Signicant atthe 5 percent level» Significant atthe 10 pereent level 16. Onto. Gamer Ero ‘Table S Export Growth Rate - Differentiated by Period and Sector ‘The erat nts Soe 6 2019) 1149-1155 ‘Source: Authors own calculation based on data fom SIICEX (2016) Column Signifeance show ees of exports growth rates of Gold, Copper and Total Mining versus ‘exports growth atesof her Seto. Other Sectors refers onon-mining exports. NS means significant at 10%. reins significant. ** means sgniffean at 1%" means significant 59% © Vacable ‘Sector N ‘Sandia ae, Avera Stenicance ‘get growth ate 20012015 0) cout 16 ast va oar Comper a a7 was tm “etl Mining 258 as m0 Dogon tie Sor ais a3 i129 S "xpont rowth ate 2001201166) ois 7” 80 mo 01s Compe 7” 058 sat 0802 ‘ott Mining mt aes 200 core Othe Sr 154 a2 v7.00 . "xpon rowth ate 2001-2008 5) ola i 3% 79 doa ‘oa Niniag 1% oer 2468 saa tie Sr 2 a3 ner ‘ "xpon roth ate 2008-209) ois in ar 237 oni tie Sr a 3s

    You might also like