The Globalization of Farmland

Transcription

1 Pubic Discosure uthorized Poicy Research Working Paper 8456 WPS8456 Pubic Discosure uthorized Pubic Discosure uthorized The Gobaization of Farmand Theory and Empirica Evidence Rabah rezki Christian Bogmans Harris Seod Pubic Discosure uthorized Midde East and North frica Region Deveopment Research Group & Internationa Monetary Fund May 2018

2 Poicy Research Working Paper 8456 bstract This paper is the first to provide both theoretica and empirica evidence of farmand gobaization whereby internationa investors directy acquire arge tracts of agricutura and in other countries. theoretica framework expains the geography of farmand acquisitions as a function of cross-country differences in technoogy, endowments, trade costs, and and governance. n empirica test of the mode using goba data on transnationa deas shows that internationa farmand investments are on the aggregate ikey motivated by re-exports to investor countries rather than to word markets. This contrasts with traditiona foreign direct investment patterns where horizonta as opposed to vertica foreign direct investment dominates. This paper is a joint product of the Word Bank Midde East and North frica Region, the Deveopment Research Group, and the Internationa Monetary Fund. It is part of a arger effort by the Word Bank to provide open access to its research and make a contribution to deveopment poicy discussions around the word. Poicy Research Working Papers are aso posted on the Web at The authors may be contacted at rarezki@wordbank.org, cbogmans@imf.org and hseod@wordbank.org. The Poicy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about deveopment issues. n objective of the series is to get the findings out quicky, even if the presentations are ess than fuy poished. The papers carry the names of the authors and shoud be cited accordingy. The findings, interpretations, and concusions expressed in this paper are entirey those of the authors. They do not necessariy represent the views of the Internationa Bank for Reconstruction and Deveopment/Word Bank and its affiiated organizations, or those of the Executive Directors of the Word Bank or the governments they represent. Produced by the Research Support Team

3 The Gobaization of Farmand: Theory and Empirica Evidence * Rabah rezki, Christian Bogmans, and Harris Seod JEL cassification: F1; F2; Q17. Keywords: Large-Scae Land cquisitions, Food Independence, Patform FDI, Land Governance. * We are gratefu to Vanessa Diaz Monteongo, Ziqi Li and mjad Khan for exceent research assistance, to Brian Bankespoor for guidance on the construction and use of geographic data, and to Oivier Banchard, Jeffrey Franke, Davide Maacrino, kito Matsumoto, Maurice Obstfed, François de Soyres, and the participants of the EERE 2017 Conference in thens, the 2018 Land Governance and Poverty Conference hed at the Word Bank, and the IMF Research Department seminar for very usefu comments. Rabah rezki: Office of the Chief Economist Midde East and North frica Region, The Word Bank, th street NW, Washington, DC 20433, US. Emai: rarezki@wordbank.org. John F. Kennedy Schoo of Government, Harvard University, 79 John F. Kennedy St, Cambridge, M Christian Bogmans corresponding author: Research Department, Internationa Monetary Fund, 1900 Pennsyvania ve NW, Washington, DC 20431, US. Emai: cbogmans@imf.org. Harris Seod: Energy and Environment Team, Deveopment Research Group, The Word Bank, 1818 H Street NW, Washington, DC 20433, US. Emai: hseod@wordbank.org.

4 1 Introduction Transnationa acquisitions of and raise important questions regarding food security and economic deveopment. Looking back in history, the surge in and acquisitions over the past decade brings back memories from the coonia era. In the 15th century, Portuga and Spain, the then most advanced maritime powers, conquered a arge number of countries around the word and subsequenty secured contro over these countries and resources and their subjugated popuations. Subsequent coonia empires incuded the French, British, Dutch and Japanese, who aso imposed their rues on foreign territories and controed vast swathes of and. These invasions ed to a dramatic expansion of goba trade in natura resources and agricutura products, as we as in saves and indentured servants to work on pantations. Most recenty, foowing the spike in food prices, there has been a booming interest in the direct acquisition of farmand in deveoping and emerging markets, often invoving mutinationa companies and foreign governments. The recent expansion in the gobaization of farmand has ed to a poarized debate between those wecoming foreign investments, hoping they wi hep raise agricutura and yieds and aeviate poverty, and those who see the phenomenon as a and grab Financia Times, 2016 and Boomberg, The present paper is the first to provide both theoretica and empirica evidence of this new wave of investments, which marks a new trend towards the gobaization of farmand. The increased interest of internationa investors to acquire farmand is part of a broader set of deveopments that are changing the nature of commercia agricuture at the goba scae. These incude the increased importance of mutinationa companies and foreign direct investment in promoting sectora growth, and a more prominent roe for goba vaue chains in expanding food suppy Maertens and Swinnen, In this ongoing process of agricutura gobaization, the voume of internationa trade in agricutura commodities increased amost five-fod in a period of three decades, from about $200 biion in 1980 to amost $1.1 triion in 2010, the argest growth recorded by any sector in that period. Interestingy, the acquisition of arge tracts of and by private investors and soeverign weath funds coincides with the rising demand for food associated with rising incomes and growing popuations. This rising interest in farmand suggests that gobaization has entered a new phase, one that is defined by the integration of pristine and in deveoping countries into the word economy. ccording to Coier and Venabes 2012, this transition has paraes with China s process of economic deveopment. Like the increase in foreign investment into China driven by its abundance of abor, the sudden shift in the foreign acquisition of and in frica and esewhere may have been triggered by arge spatia differences in factor productivity. But there are aso significant differences between the gobaization of China s cheap abor and the gobaization of farmand. Based on an eary empirica account of the phenomenon, rezki et a provide evidence that investors tend to favor countries where oca popuations have 2

5 insecure property rights, echoing the concerns of many goba observers. Surprisingy, the ogic underpinning these investments thus seems to run counter not ony to the standard findings in the foreign direct investment iterature, but aso to the parae story of China opening up to the rest of the word foowing institutiona improvement Song et a., 2011 and di Giovanni et a., In this paper, we present a muti-country mode of internationa trade with two sectors of production agricuture and manufacturing and two factors of production abor and and. We assume that investors can use a home-country specific technoogy to engage in mutinationa production abroad. gricutura varieties are both host-country and investor-country specific à a rmington The rationae behind this doube rmington assumption is that whereas differences in agro-ecoogica conditions ead to differentiated agricutura products around the gobe, agricutura products aso differ given the diversity of agricutura technoogies used by different investors or because of their different branding. In each country, there is an exogenous and fixed and price, which infuences the quantity of and suppied to the commercia agricutura sector. Using this framework, we derive a biatera gravity-type specification of demand for and. We first do this in a mode with investor differentiation that has both horizonta and vertica FDI motives, and then for an aternative mode without investor differentiation that ony has the vertica motive. Under both motives, biatera investment in and is shown to depend on endowments and technoogy, as we as access to word markets or its inverse, remoteness. In the first mode, we find that host-country remoteness negativey affects investment as investors prefer to produce cose to export markets. In the second mode, however, due to the absence of investor differentiation, host country-remoteness impies ess investor competition and therefore encourages investment. To test the predictions of the mode, we combine data from the Land Matrix, an onine database of arge-scae and acquisitions, with country-eve data on and endowments, and governance and agricutura productivity per worker. We proxy for remoteness, that is, for the inverse of market potentia, in three different ways. Furthermore, foowing Siva and Tenreyro 2006, Fay 2015, nderson and Yotov 2016 and others, we estimate our gravity-type specification using Poisson Pseudo Maximum Likeihood PPML, to dea with heteroskedasticity and a arge proportion of zero observations. Our nove resut is that host-country remoteness a ack of host country access to agricutura consumer markets has been a positive and sizeabe determinant of arge-scae biatera and acquisitions. Whereas an increase in the host country s suitabe and endowment by one standard deviation from the mean is found to increase the number of biatera deas by 86.1%, an increase in the remoteness of the host-country by one standard deviation from the mean is expected to increase the numbers of deas per country pair by as much as 115.8%, depending on the remoteness proxy that is considered. In sum, host-country remoteness appears to be a reevant determinant 3

6 of investment, and its importance is comparabe to or even exceeds that of agricutura and endowments in expaining arge-scae and acquisitions. This finding is fuy consistent with the food independence motive underpinning such investments. Our finding that host-country remoteness pays a positive roe in driving biatera investment may refect a reative scarcity of and at the goba scae. Indeed, in our mode, under ow suppy but high demand for and, investors are competing with one another and ook to invest in remote countries to shied themseves from competition. To a certain extent, our finding contradicts the intuition formuated by Coier and Venabes 2012 in the case of frica that and markets are and-abundant, investor-scarce, with demand very sma reative to potentia suppy. The two observations can be reconcied by noticing that athough and in genera may be abundant, what matters is the amount of avaiabe and suitabe and, which is avaiabe in smaer quantities. The remainder of this paper is as foows. Section 2 presents a muti-country mode of trade and and acquisitions. Section 3 uses the mode to derive a gravity stye equation for domestic and transnationa demand for and. Section 4 presents the econometric specifications for our cross-country anaysis. Section 5 describes the data used. Section 6 estimates the determinants of transnationa and deas using biatera regressions, and focusses on the roe of remoteness. Section 7 concudes. 2 Modeing internationa trade and and acquisitions 2.1 The genera setting In this section we present a muti-country mode that accounts for internationa investments in and and for trade of agricutura products. The word consists of N countries. Each country produces an agricutura good and a manufactured good M using abor L and and T. In each country i = 1...N, factors of production are avaiabe in quantities L i and T i. though abor can be used in both the manufacturing and agricutura sector, and is used for the production of agricutura goods ony. Labor is domesticay mobie between the agricutura and the manufacturing sector within each country and capita is internationay mobie. In-country and mutinationa production The manufacturing and the agricutura sectors produce goods that are differentiated by the country of origin. For the agricutura sector, this is consistent with trade being motivated by the consumption of different varieties of agricutura products see Costinot et a., Because agricutura potentia varies across countries, producers grow different crop varieties of the same crop or different crops across countries. Our mode, however, offers an extra ayer of differentiation for agricutura goods as investors from a country i can aso engage in production of an agricutura variety in another country, something we refer to as mutinationa production. Differentiation according to both the country of 4

7 production and the origin of the investor is a reaistic feature of the mode as consumers may for instance differentiate between coffee sod by a nationa producer of a Latin merican country i.e., a fair trade producer and coffee produced in the same country by a mutinationa firm. We wi detai in the next subsection how these two types of differentiations affect consumer utiites. For the time being, simpy note that on the production side, the agricutura technoogies Z i used by investor country i differ for each host country and are given by a vector of Tota Factor Productivity TFP terms, Z i {Z i, = 1...N}. With these notations, Z is the technoogy for own-country production of the agricutura variety by country. For simpicity, we abstract from mutipe crop choices by assuming that an investor country produces ony one type of differentiated crop in a given country. 1 so note that, as standard in the iterature, trade is subject to iceberg costs: in order to se one unit of good to consumers in importing country n, a firm producing in country must ship t n 1 units of its product. 2 Patform and food-independence motivated investments in and Our representation of mutinationa production in the agricutura sector revoves around the transfer of technoogy from an investor country to a host country see Ramondo and Rodríguez-Care, s in the Ramondo and Rodríguez-Care framework, investors can engage in mutinationa production and use a host specific variety of their home-country technoogy to produce in cose proximity to other markets thereby saving on trade costs t n as in Brainard 1997, Markusen and Venabes 1998, Hepman et a and others or in order to benefit from ower production costs in the host country. In the main version of our mode, both features of the the Ramondo and Rodríguez- Care framework are present and producers invest abroad with the purpose of re-exporting to word markets, a pattern of investment to which we, foowing Ekhom et a. 2003, refer as patform FDI. Hence, in the patform FDI version of our mode, investors are competing with other investors for i access to resources i.e., abor and and in the host countries, as we as ii for access to agricutura consumer markets in importing countries. In addition, we aso consider an aternative mode in which country i invests in country in order to benefit from ower oca production costs whie reimporting the agricutura good to its own domestic market ony see section 3.2. s we wi show in more detai in section 3.2, this pattern is obtained by shutting down the host-country market access channe. We wi interpret investment that foows this vertica FDI pattern as food-independence driven. In the rest of this section, we present the genera structure of our framework, which is common to both the patform FDI and the food independence modes. 3 1 We aso abstract from considering mutinationa production in the manufacturing sector, an issue which is outside the scope of this paper. 2 For simpicity and without oss of generaity, we do not differentiate the trade costs of the manufacturing and the agricutura goods. Introducing sector-specific trade costs woud ony make notations burdensome without changing the intuition of the mode. 3 Whie FDI typicay invoves the investment of physica capita, here investors bring knowedge capita, i.e., agricutura technoogy, and hire oca workers to work the and they rent. 5

8 2.2 Preferences and consumption Each country n has a representative agent who consumes a varieties of the agricutura and manufacturing goods. We assume that the upper-tier utiity function is Cobb-Dougas: u n = Q n γ Q M n 1 γ 1 where Q n and Q M n are the quantities of the composite agricutura good and composite manufacturing good consumed in country n see equations 2 and 3 beow for the exact formuas, and γ ]0, 1[ and 1 γ represent the expenditure shares on agricutura and manufactured goods respectivey. To account for the idea that consumers differentiate between goods based on the country of origin, we specify agricutura consumption and manufacturing consumption in country n as CES aggregates over the N number of discrete varieties that are differentiated by country of production, with and Q n = N =1 ε q ε 1 ε 1 ε n 2 Q M n = N =1 εm q M ε M 1 ε M 1 ε M n 3 where qn respectivey qm n is the consumption of agricutura goods respectivey manufactured goods produced in country and consumed by the representative consumer in country n, and ε > 1 respectivey ε M > 1 is the easticity of substitution between agricutura varieties respectivey manufacturing varieties produced in different countries. Because our mode aows for differentiated mutinationa production of the agricutura good, we further specify the consumption in country n of the agricutura goods produced in country as a CES aggregate over the N number of discrete varieties from different investor countries, with q n = N i=1 σ q σ 1 σ 1 σ ni 4 where qni represent country n s consumption of the agricutura variety that is produced in country by investors from country i, and σ > 1 is the easticity of substitution between agricutura varieties produced by different investors. Observe that if σ, then the agricutura varieties are not differentiated by investor country. 6

9 In this doube rmington setting for the consumption of agricutura goods, observe that 2 accounts for product differentiation by country of origin where the good is produced, whereas 4 accounts for product differentiation by investor country by whom the good is produced. 2.3 Production Firms in the manufacturing sector produce a manufacturing good under constant returns to scae using abor L M ony: y M = B L M 5 where we have assumed that in order to produce one unit of the manufactured good a producer in country requires 1/B units of abor. Production of agricutura commodities by an investor from country i in host country requires the input of and Ti and abor L i, which are acquired ocay in host country, under a constant-return to scae Cobb Dougas production technoogy: L yi α = Z i T 1 α i i 6 α 1 α where α and 1 α are the output easticities with respect to abor and and. To account for adjustment costs in the export of agricutura technoogy or adaptation to foreign contexts, we assume that agricutura TFP may decay with distance between the investor and the host country in a mutipicative way. 4 We thus have Z i = Z i F t i 7 where Z i is the intrinsic investor productivity and F is a decreasing function of transport costs t i. 2.4 Equiibrium ssuming perfect competition for each variety, the producer price of each manufacturing variety equas its unit cost. Based on the manufacturing production function 5, this impies that the price for the manufactured good in country is: p M = 1 B w 8 4 Our assumption is based on the ideas of Diamond 1999, who hypothesized that the diffusion of agricutura knowedge and technoogy is inhibited by geography. 7

10 where w is the return to abor in country. Simiary, under perfect competition, the unit cost function associated with the agricutura production function 6 of investor i in country is: where f is the price of agricutura and in country. c i = w α f 1 α Z i 9 Consumers maximize utiity by choosing how much to consume of each manufacturing variety and each agricutura variety. in three stages. We can sove the utiity maximization probem of consumers First, at the top-tier, consumers maximize 1 by choosing agricutura and manufacturing consumption subject to the budget constraint, I n = P n Q n + P M n Q M n, where I n, P n and P M n are defined as nationa income and the respective agricutura and the manufacturing price indices in country n. We obtain expenditures D n and D M n on agricutura and manufactured goods consumed in country n respectivey: and D n = γi n 10 D M n = 1 γ I n 11 Second, consumers maximize agricutura utiity 2 and manufacturing utiity 3 by aocating tota expenditures across the agricutura and manufacturing varieties from different countries. Let p n be the consumer price in country n of the agricutura varieties produced in country. Then Xn and XM n, the respective consumptions by country n of the agricutura and manufacturing varieties produced in country, are given by the foowing Marshaian demand functions: p Xn 1 ε = n Pn γi n 12 where X M n = tn p M P M n 1 εm 1 γi n 13 and P M n P n N h=1 N tnh p M h h=1 p nh 1 ε 1 1 ε 14 1 εm 1 1 ε M 15 8

11 are the respective consumer price indexes of the composite agricutura good and the composite manufacturing good consumed in country n, aso commony referred to as the inward mutiatera resistance indices see nderson and van Wincoop, Third, consumers in country n maximize agricutura subutiity 4 by choosing the eve of consumption of the various investor-country differentiated varieties produced in country, taking tota expenditures on varieties from country as given. The resuting Marshaian demand function, Xni, that is, country n consumption of the variety produced in country by investor country i, is: X ni = tn c i p n 1 σ p n P n 1 ε γi n 16 where N p n = tn c j j=1 1 σ 1 1 σ 17 is the CES price index of the various agricutura varieties produced by different investors in country and consumed in country n. Observe that the producer factory-gate price index for the agricutura sector in country is p N j=1 c j 1 σ 1/1 σ = p n t n 18 s can be seen from equations 16 and 13, imports decrease with distance from exporters and with producer price. Next, et us define tota revenues of sector s {, M} in producer country as the sum of imports from a countries n [1,.., N] of goods produced in incuding own imports, so that V N n=1 X n and V M N n=1 XM n. With these notations, we aso define goba nomina outputs as V N =1 V and V M N =1 V M for the agricutura and manufacturing sectors respectivey. Using 12 and 13, substituting for X n and XM n gives: in the formuas for V and V M V = p Ω 1 ε V 19 V M = p M Ω M 1 ε M V M 20 where and Ω [ n=n n=1 tn P n 1 ε γi n V ] 1/1 ε 21 9

12 Ω M [ n=n n=1 tn P M n 1 εm 1 γi n V M ] 1/1 εm 22 are outward mutiatera resistance indices in the agricutura and manufacturing sectors respectivey. They are aso the inverse of the agricutura and manufacturing market potentias of country, that is, the weighted sum of the market sizes of a trade partners of country. 5 Note here that revenues or nomina outputs by sector decrease with factory gate prices and increase with market potentia. Rearranging 19 to substitute for the power transform of factory-gate prices, p 1 ε, into equation 14, we can write the inward mutiatera resistance terms Pn as: P n = h=n h=1 t nh Ω h 1 ε V h V 1/1 ε To cose the mode, we must now specify how the amount of agricutura and T i=n i=1 T i is determined. For simpicity, we assume that the government eases and to agricutura investors at a fixed price, f = f, that is determined outside of our mode, and suppies a tota quantity of and that meets investors demand at this price. Hence, in our framework f is exogenous and T is endogenous, and some and, in the amount of T T, is not used for commercia agricuture. 6 Next, we note that nationa income in each country is given by the sum of abor earnings and government revenue from suppying and to the agricutura sector: 7 23 I = w L + f T 24 Before defining the equiibrium, et us now define aggregate demands for abour and and in each country. From equations 16 and 13, we obtain the agricutura revenues of investors from country i in country : Vi N n=1 X ni = c 1 σ p i p Ω 1 ε V. Next, recognizing that by virtue of the Cobb-Dougas production technoogy in 17 a fraction 1 α of agricutura revenues are paid out to agricutura and, we obtain a formua for and demanded by agricutura investor i in country, T i = 1 αv 1 α i = f c i p 1 σ p Ω 1 ε V 25 f 5 Note that these measures of market potentia attach more weight to nearby markets t smaer t n and markets with higher prices arger Pn. s 6 Note that it is straightforward to consider an aternative setting in which the government sets the suppy of and to be aocated to agricutura use, T = T, and ets the price f adjust. ternativey, one coud consider a setting in which both f and T are endogenous, resuting from an upward-soping and suppy curve, but this woud make the mode much more compicated. 7 Our setting with perfect competition in a markets impies zero profits. 10

13 From equation 25 it then foows that the aggregate demand for and in host country the sum over a investors demand for and in country, T = N i=1 T i can be written a T = p Ω 1 ε V T. 1 α f Recognizing that a fraction α of agricutura revenues are paid out to agricutura abor and that a manufacturing revenues are paid out to manufacturing abor, the aggregate demand for abor the sum over a agricutura investors demand for abor in country, L = i=n i=1 L i, pus the demand for abor by manufacturing producers L M can be written as L = α w p Ω 1 ε V + 1 w p M Ω M 1 ε M V M. Note that a terms on the right-hand side of the aggregate abor demand and the aggregate and demand equations can be written as functions of the vectors w = w 1,..., w N and T = T1,..., TN, such that we can think of these equations as a system of 2N equations in w and T. 8 Soving this system of equations in the unknowns w and T can aso be interpreted as finding the zeros of an excess demand system Gw, T = G L w, T, G T w, T, where G L = G L 1, G L 2,..., G L N and G T = G T 1, G T 2,..., G T N are respectivey the N-dimensiona excess abor demand and excess and demand vectors: G L w, T = α p w Ω 1 ε V + 1 p M Ω M 1 ε M V M L 26 w G T w, T = 1 α T p Ω 1 ε V f 27 Finay, note that we derive the country s baanced trade BT condition by adding up the excess demand for abor and and conditions and rearranging terms: I = p Ω 1 ε V + p M Ω M 1 εm V M 28 We are now ready to define the mode s equiibrium. Definition 1. n equiibrium is a vector w, T R 2n ++ such that G L = 0 and G T = 0 for = 1,..., N, with c i, p, P n and Ω satisfying equations 9, 18, 14 and 21 respectivey, and p M, P M n and Ω M satisfying equations 8, 15 and 22 respectivey. To prove that an equiibrium as in Definition 1 exists, we extend Theorem 2 in varez and Lucas 2007 which is an existence theorem for an exchange economy to our rmington economy with 8 Our reasoning is as foows. First, note that unit costs, prices and nationa incomes can be written as functions of wages and agricutura and suppies. Unit costs c i depend on w, so that prices p and p n depend on w, impying that Pn depends on w. Simiary, pm depends on w and P M depends on w. Nationa income I depends on w and T. Second, it foows that V,V, V M, V M, Ω and Ω M are functions of the vectors w and T. 11

14 two sectors and two factors of production. We assume that a countries own a sufficient quantitity of and so that an interior equiibrium aways exists, i.e., T < T for a = 1,..., N. Proposition 1. There is at east one vector of wages and quantities of agricutura and, w, T R 2n ++, such that the markets for agricutura and and abor cear in every country: G= 0. Proof. See ppendix C. Now that we have estabished the foundations of our framework, we next derive testabe hypotheses that can be used in conjunction with data on fows of cross-border and acquisitions, and avaiabiity and and governance, among other determinants. 3 The determinants of arge-scae and acquisitions: Patform FDI or food independence? In this section, we use our theoretica framework to characterize patterns of and investments and derive gravity-stye equations for transnationa investment in and between any given investorhost country pair. We consider two aternative modes rooted in the same genera framework; in the first one presented in Section 3.1, investors produce abroad with the purpose of exporting to other countries as in patform FDI, whereas in the second one presented in Section 3.2, and acquisitions are driven by a food independence motive and the resuting pattern of investment resembes vertica FDI, i.e., investors from a given country wi produce abroad and se their entire production to their own domestic market. s we wi show in subsection 3.2, the second version of the mode is obtained making different assumptions which resut in shutting down the horizonta, market-seeking FDI mechanism, eaving ony the resource-seeking FDI mechanism. By comparing the gravity equations from the two different modes, we are abe to formuate a number of hypotheses that aow us to ook for evidence of food-independence and patformdriven FDI. We present the biatera investment equations associated with reexport patform or food-independence patterns in the two subsections beow. s we wi show, host-country remoteness pays a different roe in the two modes as a determinant of FDI patterns: whereas host-country remoteness deters biatera investment under the patform FDI mode, it attracts biatera investment in the food independence mode. 3.1 Land acquisitions as patform-driven investments We describe here the additiona assumption made to obtain the patform FDI mode. Foowing Ramondo and Rodríguez-Care 2013, we assume ε = σ and finite. 9 We can thus rewrite 9 This simpifying assumption equates the easticity of substitution between agricutura varieties produced in different countries and the easticity of substitution between agricutura varieties produced by different investors. We discuss the impication of this assumption for the empirica anaysis in section

15 equation 25, the amount of and in country acquired by producers from country i, as: T i = 1 α c 1 σ i f p 1 σ p Ω 1 ε c V = 1 α i Ω 1 σ V 29 f Equation 29 predicts the quantity of and acquired in a biatera reationship between any given host country and investor country i under the patform FDI motive. Proposition 2. Under the patform FDI motive, the quantity of and eased by investor country i in host country is decreasing in host-country remoteness from agricutura consumer markets Ω, decreasing in the return to abor w and and price f in host-country via c i, and increasing in the agricutura productivity of the investor-country, Z i via c i. Proof. The resuts foow directy from inspection of equations 9 and 29. The most interesting resut summarized in Proposition 2 is the reationship between biatera investment and host-country remoteness or its inverse, market potentia. Under the patform FDI motive, investors tend to invest ess in countries that are remote from arge agricutura consumer markets, as it impies a ower aggregate demand for their product. In the next subsection, we turn to another version of our mode in which the reationship between investment and remoteness is reversed. 3.2 Land acquisitions for food independence In the mode presented in the previous subsection, and investment is driven by a patform FDI motive. To see this, we observe from equation 29 that and investment is increasing in market potentia, reminiscent of horizonta export market-seeking FDI, and decreasing in the oca cost of production, which is characteristic of vertica import resource-seeking FDI. However, by changing our assumptions, we can shut down the horizonta export market-seeking mechanism and restrict the pattern of and investment to refect a vertica FDI motive ony. Intuitivey, vertica FDI occurs when each investor country has an absoute advantage in producing abroad and serving its home market over a the other investor countries. Mathematicay, this can be warranted under two conditions. The first condition is that varieties are no onger differentiated by the country of the investor i.e., when σ, which ensures that in every host country there is ony one investor country that produces i.e., the most efficient one. Without imposing a further condition, however, the pattern of agricutura production woud be indeterminate. The second condition we impose is that each investor country hods an absoute agricutura TFP advantage to produce abroad and reimport to its own country over a other investors Denoting Z ni and t ni the agricutura TFP and biatera trade cost of investor i in host country exporting to consumer country n, the condition requires t nn /Z nn <t ni /Z ni, for a n,, i n. 13

16 This can be the case, for instance, when the production or suppy technoogy is broady considered to incorporate marketing and ogistics costs, which wi typicay be ower to serve the home market. 11 Taken together, these two assumptions de facto shut down competition among investors and ensure that each consumer market is served by an investor from the same country. In the ppendix we show that this set of assumptions resuts in the foowing specification for biatera investment: T i = ti Z i Ω P i where the foowing terms are redefined, with Ω and Ṽ w αf 1 α 1 ε V. Ω 1 ε γi i T V 30 N i=1 ti 1 ε Z i γii P i V, P i N =1 ti Z i Ω 1 ε Ṽ V Proposition 3. Under the food independence motive, the quantity of and eased by investor country i in host country is increasing in host-country remoteness from agricutura consumers Ω, investor-country remoteness from agricutura producers P i, agricutura productivity of the investor-country Z i, expenditures on agricutura goods γi i, agricutura and T, and decreasing in biatera trade costs t i. Proof. The resuts foow directy from inspection of equations 30. Equation 30 refects that, conditiona on the amount of agricutura and that is made avaiabe to investors in equiibrium, country pairs with remote hosts and remote investors feature more deas. This is simiar to the roe of importer-country and exporter-country remoteness in the determination of biatera fows in trade modes, as in nderson and van Wincoop To see this, consider the roe of host-country remoteness for instance. things equa, if the host country s access to consumer markets in the rest of the word is weaker by virtue of the higher prices of its agricutura product in those markets, then the investor country wi be abe to acquire more and in the host country. Hence, under the food independence motive, we see that the market potentia of the destination country is now a negative determinant of biatera and acquisitions: in equiibrium, investor countries that are far away from agricutura producers P i high invest more in nearby countries t i ow that are far away from agricutura consumers Ω sma. In contrast, under the patform FDI configuration, investor-country remoteness from agricutura producers had no impact on and acquisitions and host-country remoteness payed a negative roe. 11 See Matsuyama 2007, who anayzes the genera equiibrium consequences of making technoogies of suppying goods dependent on whether the destination is home or abroad. 14

17 3.3 Differentiating between food-independence and patform-driven investments in and In the previous subsections we used the two versions of our mode to derive two different gravitystye equations that predict biatera patterns of and investment. These equations are presented in Tabe 1. Tabe 1: Gravity equations for and-acquisition Biatera Patform = 1 αc i Ω 1 σ Food independence T i T i = f V t 1 ε i Z i γi it Ω Pi V From the overview of equations in Tabe 1, we observe that there are both commonaities as we as differences between the gravity equation under the patform FDI and food independence modes. For exampe, from Tabe 1, we earn that if biatera investment in and is decreasing in host-country remoteness and if biatera investment does not depend on investor-country GDP nor on investor-country remoteness from agricutura producers, then these findings woud be suggestive of and deas as a type of patform FDI. 12 If biatera investment, however, is increasing in investor-country GDP and investor-country remoteness from agricutura producers, and if host-country remoteness from agricutura consumers stimuates investments, then this woud be suggestive of food-independence-motivated FDI. It shoud be noted that the remoteness terms are specified in sighty different ways in each mode as can be seen by comparing the t n and t i Z i terms in equations 21 and 30 respectivey. The reason for this is that under patform FDI, investors from different countries do not compete directy, as their home-country specific technoogies are used to produce an investor-specific and country-specific variety, whie in the food independence version of our mode they do compete directy as investors a produce the same country-specific variety, and as such their agricutura TFP Z i has a direct impact on their competitiveness in acquiring and reative to other investors. Hence the occurrence of the term t i Z i in the host-country remoteness term of the food independence mode instead of t n in the host-country remoteness term of the patform FDI mode 30. However, we show in ppendix C that because Z i can be written as a mutipicative function of investor-country agricutura productivity and a biatera distance term, then the remoteness terms Ω and P i in the food independence mode wi in genera be magnified versions of the simper remoteness terms Ω and P i in the patform-fdi mode. This is because distance 12 Note that the absence of a roe for investor-country GDP and for investor-country remoteness is consistent with absence of imports. 15

18 enters twice in the remoteness formuas in the food independence mode as opposed to ony once in the remoteness formuas in the patform-fdi mode. In economic terms, this is because in the food independence mode, in addition to discouraging agricutura trade, remoteness aso deters investment by inhibiting the transfer of technoogy. Observe, moreover, that if we further assume F t i = 1/φt i in the distance decay function 7, the empirica proxy that we wi use to measure remoteness wi be exacty the same for the two modes see section 4.2 beow and the ppendix. For these reasons, we wi ony consider empirica proxies of the remoteness term Ω as defined in the patform-fdi mode. In the next section, we wi use Tabe 1 to generate an empirica specification and hypotheses that are directy inked to our mode. By testing these hypotheses, we are abe to seek empirica support for the two different motives potentiay driving arge-scae and acquisitions. 4 Econometric pproach 4.1 Specification In view of our theoretica framework, we are interested in two sets of hypotheses that expain transnationa investments. First, we want to test the extent to which endowment of and, popuation size and agricutura productivity in host and investor countries affect biatera and deas. Second, we want to test the roe of remoteness in stimuating or discouraging and deas, which wi aso shed ight on the motives for those investments whether patform of food-independence driven. We present our theory-inspired proxies for the host-country and investor-country remoteness terms in subsection 5.1. To ensure our empirica resuts are robust, we rey on different approaches from the trade iterature to proxy for remoteness. Based on the biatera and investment equations presented in Tabe 1, we propose the foowing empirica specification for biatera investment in and: m=4 Ti = γ 0 + γ 1,m ndist i,m + γ 2 COL ij + γ 3 nt + γ 4 nw + γ 5 nq 31 m=1 +γ 6 nz i + γ 7 nl i + γ 8 nω + γ 9 np i + ε i where Ti represents investments of country i in country i.e., the number of deas or tota and covered under those deas over the time period considered for detais of the reevant time periods considered in the empirica anaysis, see section 5.1. In this biatera gravity equation, DIST i,m is the physica distance between investor country i and host country if it fas in interva m, and zero otherwise. Foowing, Eaton and Kortum 2002 and nderson et a. 2015, this construction aows to account for non-inearity by decomposing the effect of physica distance into four intervas m = 1, 2, 3 and 4, which correspond 16

19 to distances beow 3, 000 kiometers, between 3, 000 km incuded and 7, 000 km, between 7, 000 km incuded and 10, 000 km, and equa to or greater than 10, 000 km. COL ij is a dummy variabe for historica coonia ties between investor country and host country. Observe that both physica distance and coonia ties are proxies for trade costs between country i and country. T is a measure of host-country and that is both suitabe and avaiabe see section 5.1 for detais regarding definition and construction, w is host-country GDP per capita proxying for abor costs, q is the same host-country index that measures the recognition of preexisting and rights and associated eve of tenure security in rezki et a see section 5.1 of the present paper, Z i is investor-country agricuture vaue-added per worker proxying for agricutura TFP, L i is investor-country popuation size proxying for expenditures on agricutura consumption, Ω and P i are measures of host-country remoteness from agricutura markets and investor-country remoteness from agricutura producers respectivey see section 4.2, and ε i is the error term. Observe that we propose a singe specification for equation 31 but that the expected sign of some of the regression coefficients wi be mode specific. so observe that the host-country agricutura productivity is absent from the biatera regression specification. From an estimation perspective, omitting host-country agricutura productivity is justified because it is ikey correated with other host-country regressors such as GDP per capita which we incude in our specification. From a theory perspective, it is aso warranted. Indeed, in the food independence version of the mode, it is the investor who brings its productivy to the host-country, and in the patform-fdi mode, the irreevance of host-country agricutura productivity is a direct consequence of our assumption that ε = σ foowing Ramondo and Rodríguez-Care In ight of Propositions 2 and 3, our expectations are as foows: popuous countries acquire more and γ 7 > 0, a the more if they are agricuturay productive γ 6 > 0. Investors tend to target countries with abundant suitabe agricutura and γ 3 > 0 as in rezki et a and Lay and Note They are aso more interested to acquire and in countries with cheaper abor costs γ 4 < 0. In ine with the finding of rezki et a. 2015, we expect investments to be made more difficut when preexisting and rights are protected γ 5 < 0 and when distance between the investor and the host country is arge γ 1,m < 0 for a m. Finay, observe that hostcountry remoteness from agricutura markets can pay in different ways, depending on whether the motive for investment is food independence γ 8 > 0 or patform FDI γ 8 < 0. If the motive is food independence, we aso expect more investments by investor countries that are remote from food producing countries, i.e., that have what nderson and van Wincoop 2003 refer to as high outward mutiatera resistance γ 9 > To see the atter point, note that host-country productivity, producer price, and unit cost can be written as Z j=n 1 σ 1 1 σ 1 j=1 Z, p = 1 Z w α ij f 1 α and c i = 1 w Z α i f 1 α respectivey, so that the biatera 1 α 1 1 σ investment equation can be written as Ti w α Z = i f 1 α 1 ε 1 ε σ Ω Z 1 ε V f. ssuming ε = σ and given that Z i = Z i F t i, it is easy to see Ti is a function of Z i but not Z. 17

20 4.2 Estimation We estimate equation 31 using Poisson Pseudo-Maximum Likeihood PPML, which provides a consistent estimator for our biatera gravity equation that is robust to measurement error, heteroskedasticity and the incusion of zeros see Siva and Tenreyro, n important chaenge in estimating our gravity-stye equations is how to propery construct a measure of remoteness. nderson and van Wincoop 2003 have shown that if gravity equations do not account for these terms, the other coefficients of interest such as e.g. trade costs may become biased Badwin and Tagioni, To address this issue, we aternativey consider four different approaches to measure or account for remoteness. We present them sequentiay beow. First, we use inear approximations to the remoteness terms a a Baier and Bergstrand Namey, we account for host-country s remoteness with the GDP-weighted average distance of that country to a other countries, which accounts for agricutura export potentia. We measure investor country i s remoteness as the agricutura GDP-weighted average distance to a host countries, which accounts of agricutura import potentia. Hence, we use: Ω = P i = N n=1,n N =1, i t θn n 32 t ξ i 33 where t n t i is the physica distance between consumer country n investor country i and host country, θ n is the importer s share of goba GDP, and ξ is the exporter s share of goba agricutura production. Note that both remoteness measures eiminate the first-order endogeneity between GDP and biatera investment by removing the interna distance component t θ. The inear approximations suggested by Baier and Bergstrand 2009 are a particuary attractive soution to the puzze of measuring host-country and investor-country remoteness, as it aows us to compute the remoteness measures directy using observabe information on distance, GDP and agricutura GDP. disadvantage of this approach, however, is that one woud preferaby use a direct measures of trade cost t n instead of physica distance to cacuate these remoteness terms. Second, another approach therefore improves upon the proxy suggested by Baier and Bergstrand 2009 by using estimates of biatera agricutura trade costs instead of distance. To come up with trade-cost estimates, we use a two-stage structura gravity approach recenty proposed by nderson and Yotov 2016 and nderson et a In the first step of the two-stage 14 Based on Monte Caro anaysis, Baier and Bergstrand 2009 show that their reduced form approach to remoteness which we use in our paper resuts in regression coefficients that are virtuay identica to those estimated in the structura approach of nderson and van Wincoop

21 procedure we estimate a dynamic pane version of the biatera agricutura trade equation 12 in mutipicative form using PPML. This estimation aows us to obtain estimates of the biatera fixed effects ˆµ n for the country pairs for which agricutura imports are observed: X n,t = exp [π,t + χ n,t + µ n + β 1 RT n,t ] ε n,t 34 where π,t, χ n,t and µ n respectivey represent the producer/exporter time-varying fixed effect, the consumer/importer time-varying fixed effect, and the pair fixed effect, and RT n,t is a dummy variabe that indicates the presence of a regiona trade agreement between countries and n at time t the incusion of which is necessary to obtain estimates of the biatera fixed costs and ε n,t is a remainder error term. Since we do not observe agricutura trade fows for a possibe country-pair combinations, we add an additiona step to our procedure. In that second stage, we use the estimates of the pair fixed effects ˆµ n as a dependent variabe and regress them on a set of standard determinants of trade costs and importer and exporter fixed effects. This can be written: [ exp [ˆµ n ] = exp π + χ n + m=4 m=1 η 1,m ndist n,m + η 2 CONT IG n + η 3 COMLNG n + η 4 COL n ] ε n where CONT IG n and COMLNG n are dummy variabes for respectivey a common border and common anguage between countries n and. The predicted vaues from this second stage regression, that is, ˆt 1 ε n = exp [ˆπ + ˆχ n + ˆη 1 n DIST n + ˆη 2 CONT IG n + ˆη 3 COMLNG n + ˆη 4 COL n ], are then used to compete the fu biatera trade cost matrix by substituting for the predicted vaues for the country-pairs that are missing. 15 To recover the trade cost estimate ˆt n from the exponentiated term ˆt 1 ε n, we need to pick a vaue for the agricutura trade easticity of substitution ε. We foow Tombe 2015 who finds an agricutura trade easticity of substitution of This vaue sits amost exacty in between the vaue of 5.4 and 2.82 that Costinot et a obtain for respectivey the trade easticity of substitution at the crop eve and the easticity of substitution across different crops, and thus appears to be a reasonabe pick. rmed with a biatera trade cost matrix, we then cacuate the agricutura mutiatera resistance indices, Ω and P i. s before, we rey on the approximations of Baier and Bergstrand 2009, eqs , but use our trade cost estimates instead of physica distance to substitute for biatera trade cost terms. 15 Note that to estimate the two-stage regression mode expressed by eqs , we rey on the Stata command ppm pane sg, deveoped by Thomas Zykin, which is the ony PPML command in Stata that aows for fast estimation of pane specifications with a arge number of fixed effects see Zykin et a for an appication and a technica companion to this appication

22 Third, an aternative approach to come up with a proxy for remoteness buids on Franke and Romer It consists in running a PPML regression of the biatera agricutura trade equation 12 using the exogenous regressors DIST n, CONT IG n, COMLNG n and COL n. It is then possibe to use the predicted vaues of this regression, that is, agricutura trade as predicted by the beforementioned geographica determinants, as a proxy for the inverse of remoteness. Fourth, another approach suggested by Redding and Venabes 2004 and Feenstra 2004 does not buid a proxy for remoteness but consists in running as a benchmark PPML regressions with investor-country and host-country fixed effects that fuy absorb the remoteness terms. However, accounting for remoteness in this way precudes the researcher from obtaining coefficient estimates of the effects of remoteness. 5 Data 5.1 Sources To measure biatera investments in and, we use the Land Matrix as of June 2016, which is an onine database that contains extensive information on and deas that have undergone ground verification by NGOs affiiated with the Internationa Land Coaition nseeuw et a., 2012 and provides the most compete dataset avaiabe. 16 The 2016 version that we use is much improved compared to the 2011 version previousy used in rezki et a The data incudes information on 2,152 transnationa deas negotiated between 2000 and 2016, incuding the origin country of the investor, and the tota area covered by the investment. For the few and deas that invove mutipe ocations or mutipe investor countries, we spit them into severa subprojects, which eaves us with a sampe of 2,601 biatera and deas. Removing the projects outside agricuture or biofues, such as those associated with tourism, industry and other renewabe energy, we are eft with a tota of 2,122 biatera deas. With this information we cacuate the cumuative number of projects and cumuative quantity of and associated with these deas by investor-country / host-country pair. We sum a deas by country pair over an unrestricted period and a restricted period Our main empirica anaysis focusses on the restricted period. The reason for this is twofod. First, monitoring of arge-scae and deas did not consistenty take pace before the onset of the goba economic crisis of s retroactive information on deas in eary years may be hard to come by, it is therefore possibe 16 In this dataset, a dea is defined as an intended, concuded, or faied attempt to acquire and through purchase, ease, or concession that meets the foowing criteria: It 1 entais a transfer or rights to use, contro, or ownership of and through sae, ease, or concession; 2 occurred after the year 2000; 3 covers an area of 200 hectares or more; and 4 impies the potentia conversion of and from smahoder production, oca community use, or important ecosystem service provision to commercia use. The vast majority of the deas in the database 80% are isted based on two or more independent sources, e.g., research papers, poicy reports and government sources, and ony 6% of the deas are based on media reports ony Note et a

23 that the seective incusion of eary years increases the ikeihood of mismeasurement. Second, there is reason to beieve that the intense scrutinity of these deas by civi society was most intense in the aftermath of the 2008 goba crisis so that the incusion of ater years aso increases the possibiity of mismeasurement. s we discuss in section 6.2, resuts are nevertheess simiar across the restricted and unrestricted periods. To measure the determinants of and acquisitions at the nationa eve, we compie data from a variety of sources, incuding: the physica distance between countries and a dummy variabe for a former coonia reationship from the GeoDist database Mayer and Zignago, 2011; a and governance index at the nationa eve which measures the eve of recognition and associated tenure security of preexisting rights in rura areas see rezki et a., 2015; 17 a dummy variabe for regiona trade agreements for the period that is borrowed from Mario Larch s Regiona Trade greements Database from Egger and Larch 2008 and which incudes a 468 mutiatera and biatera trade agreements as notified to the Word Trade Organization for the ast 66 years from 1950 to 2015; and a set of variabes from the 2009 Word Deveopment Indicators database Word Bank that incudes agricutura vaue added per worker in constant 2010 US $, GDP per capita in current US $ and popuation size. Furthermore, for gross food production current US $ and biatera food trade current US $, we use the CEPII TradeProd database a trade and production dataset originay constructed by de Sousa et a that covers 26 sectors using the Internationa Standard Industria Cassification ISIC revision 2, and 151 importing and exporting countries. For our purpose of estimating trade costs as expained in the previous section, we sum the trade fow data of the categories food ISIC 311 and beverages ISIC 313 for each year in the period Because proper estimation of requires the avaiabiity of intranationa trade fow data, we use the CEPII TradeProd dataset which aready contains these fows for many countries and impute missing vaues by using the nderson and Yotov 2010 method. 19 Finay, we construct measures of avaiabe suitabe and, using the Goba gro-ecoogica Zones GEZ data jointy deveoped by the Food and gricuture Organization FO and the Internationa Institute for ppied System naysis IIS For each of five crops wheat, maize, oi pam, sugarcane, soybean we cacuate the maximum of the suitabiity indexes which has been rescaed and is comprised between 0 and 100 under rainfed and baseine cimate conditions in each 5-arc minute 10km by 10km grid ce. We then cacuate the tota area of a the ces for which this maximum for at east one crop is greater than 70 which corresponds to high or very high suitabiity under rainfed conditions, 17 The index is the first component of a principa component anaysis on and governance variabes contained in the 2009 Institutiona Profies Database de Crombrugghe et a We aso experimented with using agricutura trade fow data from the UN Comtrade database and found comparabe estimates of trade costs. 19 To do this, we first determine for each country the ratio of aggregate intranationa trade over tota goods expenditures for a sectors for which the data are avaiabe. Second, we mutipy tota expenditures on food and beverages with this ratio and take this as our estimate of intranationa trade for those countries for which the data are missing. 21

24 after having excuded cutivated and, forests, and protected and. The variabe is expressed in 1,000 hectares Descriptive statistics Tabe 4 provides an overview of and that is avaiabe and suitabe for agricutura production for different regions in the word. It reveas that itte and remains in the Midde East and North frica and Southeast sia, whie Sub-Saharan frica and Latin merica sti have significant endowments of unexpoited and. Forests constitute amost 63 percent of a avaiabe, nonprotected and. Note that in our regressions, we use a definition of suitabe avaiabe and that excudes forests and protected areas. Tabe 5 presents basic characteristics of deas using data from the Land Matrix. projects in the Land Matrix database were initiated between 2000 and 2016, athough amost 80% of the deas were negotiated between 2006 and s of June 2016, the Land Matrix has information on 2,152 transnationa deas, with a cumuative size of 58.4 miion hectares, affecting 88 host countries wordwide. This roughy corresponds to an area the size of France or Ukraine, or 13 times the size of the Netherands. s one can te from Tabe 5, Sub-Saharan frica 884 deas and East sia 611 deas have been the most important target regions for investment, foowed by Latin merica 368 deas. Few deas have been recorded outside of these regions. In addition, the size distribution of these and deas is very uneven, with ony 52 deas covering a staggering 40.9 percent of a and traded. pproximatey 66 percent of a deas are smaer than 10,000 hectares or 100 km2. Dea impementation has been sow, both at the extensive and intensive margin: to date, 51.9 percent of the projects have been impemented, but ony 37.6 percent of the miion hectares of and associated with these impemented deas has been confirmed under production. We note that the fraction of deas not impemented is significanty smaer in Sub-Saharan frica 37 percent. ccording to our cacuations, 76.5 percent of a recorded deas have been inked to agricutura and biofue reated projects, see Tabe 5. This fact fits the narrative that arge scae and acquisitions have been party motivated by private sector expectations of higher food and biofue prices and government concerns over food independence. 20 Using this approach, we estimate that the tota quantity of suitabe avaiabe and defined here as uncutivated, non-forest, non-protected and suitabe for agricuture outside urbanized areas is about 403 miion hectares for the whoe word. 22

25 Tabe 2: Descriptive Statistics - Large-Scae Land cquisitions Dataset Variabes N mean sd min max project b 79, dist b og 50, coony b 50, suitabe nf np og 29, suitabiity index1 63,063 3,748 2, ,194 and governance 34, agri vaue added o og 37, popuation o og 45, gdp cap og 42, ogpi BB 42, ogω BB 42, =N og ˆ =1, i Xi 63, =N og 63, =1, n Xn ogpi Y 42, ogω Y 42, Notes: project b measures the number of deas concuded per investorhost country pair between , dist b og measures the og of physica distance in km between investor and host country, coony b represents a dummy variabe that equas 1 if investor and host country have been in a coonia reationship and equas 0 when they have not, suitabe nf np og represents the og of the quantity of nonforest, non-protected and in ha that is suitabe for agricuture, and governance is an index of tenure security enjoyed by existing and users, agri vaue added o og is the og of vaue added per worker in the agricutura sector of the investor country in US doars, popuation o og is the og of popuation size, gdp cap og is the og of GDP per capita in current US doar, ogpi BB is the eve of investorcountry remoteness a a Baier and Bergstrand 2009, ogω BB is the og of the eve of host-country remoteness a a Baier and Bergstrand =N 2009, og ˆ =1, i Xi is the og of predicted eve of investor-country n=n agricutura exports a a Franke and Romer 1999, og ˆ n=1,n Xn is the og of the predicted eve of host-country agricutura exports a a Franke and Romer 1999, ogpi Y is the og of the eve of investor-country remoteness a a nderson and Yotov 2016, ogω Y is the og of the eve of host-country remoteness a a nderson and Yotov

26 Tabe 3: Descriptive Statistics - Large Scae-Land cquistions Dataset VRIBLES N mean sd min max project b 79, dist b og 50, coony b 50, suitabe nf np og 29, and governance 27, agri va o og 37, popuation o og 45, gdp cap og 42, ogpi BB 42, ogω BB 42, =N og ˆ =1, i Xi 63, n=n og ˆ 63, n=1,n Xn ogpi Y 42, ogω Y 42, Tabe 4: vaiabiity of Suitabe Land for gricuture a in Mn ha, by Region. MEN SS LC EC EP S NM WNE vaiabe uncutivated, non-forest, non-protected and vaiabe uncutivated, forest, non-protected and Tota avaiabe uncutivated, non-protected and Notes: Region abbreviations refer to Midde East and North frica MEN, Sub-Saharan frica SS, Latin merica and the Caribbean LC, East sia and the Pacific EP, Europe and Centra sia EC, South sia S and Western Europe WNE. Tabe 5: Key Characteristics of Land Projects by Region, , by Region. MEN SS LC EC EP S NM Size of projects Host countries # N/ Tota area mn ha N/ Projects >= 1 mn ha # N/ Projects >= 250k ha and < 1mn ha # N/ Projects >=10k ha and <250k ha # N/ Projects<10k ha # 1, N/ Tota # projects 2, N/ Intended use percent gricuture N/ Biofues N/ Forestry, Industry and Other N/ Green N/ Unknown N/ Notes: Region abbreviations refer to Midde East and North frica MEN, Sub-Saharan frica SS, Latin merica and the Caribbean LC, East sia and the Pacific EP, Europe and Centra sia ECand South sia S. 24

27 Figure 1: Number of deas by target country 6 Resuts 6.1 Main resuts Tabe 6 presents the resuts from our biatera regressions. With PPML, the coefficient estimates of regressors that are expressed in ogs and eves can be interpreted as easticities and semieasticities respectivey. In coumn 1 we present the resuts of our baseine regression that incudes a the regressors as in eq. 31, except for the remoteness terms. In ine with our hypotheses, we find that countries with more avaiabe suitabe and are a more frequent target in and deas, whie the roe of and governance is not significant. Let us consider the roe of and endowments first. The coefficient on the variabe of interest suggest that a things ese equa the number of biatera deas increases by 3.5 percent for a 10 percent increase in the quantity of avaiabe and non-cutivated, non-forested and non-protected in the host country. This easticity is reativey stabe and varies between percent once we contro in various ways for mutiatera resistance in coumns 2-7. s for the roe of and governance, whie the coefficient enters with a negative sign in most coumns as in rezki et a. 2015, it is not found to be statisticay significant in any of the regressions, which contrasts with the previous findings in rezki et a in which Land Matrix data from 2011 was used. In addition to our baseine empirica specification being different to rezki et a. 2015, the non-significance of the and governance variabe coud indicate that and grabs have become more difficut in the recent period, possiby due to more intense civi society monitoring and increased transparency 25