Climate changes and new productive dynamics in the global wine sector

Lamonaca, Emilia; Santeramo, Fabio Gaetano; Seccia, Antonio Article Climate changes and new productive dynamics in the global wine sector Bio-based and Applied Economics (BAE) Provided in Cooperation with: Firenze University Press Suggested Citation: Lamonaca, Emilia; Santeramo, Fabio Gaetano; Seccia, Antonio (2021) : Climate changes and new productive dynamics in the global wine sector, Bio-based and Applied Economics (BAE), ISSN 2280-6172, Firenze University Press, Florence, Vol. 10, Iss. 2, pp. 123-135, https://doi.org/10.36253/bae-9676 This Version is available at: https://hdl.handle.net/10419/321727 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Bio-based and Applied Economics 10(2): 123-135, 2021 | e-ISSN 2280-6172 | DOI: 10.36253/bae-9676 Bio-based and Applied Economics BAE Copyright: © 2021 E. Lamonaca, F.G. Santeramo, A. Seccia. Open access, article published by Firenze University Press under CC-BY-4.0 License. Firenze University Press | www.fupress.com/bae Citation: E. Lamonaca, F.G. Santeramo, A. Seccia (2021). Climate changes and new productive dynamics in the global wine sector. Bio-based and Applied Economics 10(2): 123-135. doi: 10.36253/bae-9676 Received: September 5, 2020 Accepted: December 16, 2020 Published: October 28, 2021 Data Availability Statement: All relevant data are within the paper and its Supporting Information files. Competing Interests: The Author(s) declare(s) no conflict of interest. ORCID EL: 0000-0002-9242-9001 FGS: 0000-0002-9450-4618 AS: 0000-0003-4549-6479 Climate changes and new productive dynamics in the global wine sector Emilia Lamonaca*, Fabio Gaetano Santeramo, Antonio Seccia University of Foggia, Italy * Corresponding author. E-mail: emilia.lamo[email protected] Abstract. Climate change has the potential to impact the agricultural sector and the wine sector in particular. The impacts of climate change are likely to differ across producing regions of wine. Future climate scenarios may push some regions into climatic regimes favourable to grape growing and wine production, with potential changes in areas planted with vines. We examine which is the linkage between climate change and productivity levels in the global wine sector. Within the framework of agricultural supply response, we assume that grapevines acreage and yield are a function of climate change. We find that grapevines yield suffers from higher temperatures during summer, whereas precipitations have a varying impact on grapevines depending on the cycle of grapevines. Differently, acreage share of grapevines tends to be favoured by higher annual temperatures, whereas greater annual precipitations tend to be detrimental. The impacts vary between Old World Producers and New World Producers, also due to heterogeneity in climate between them. Keyword: climate change, acreage response, yield response, Old World producers, New World producers. JEL code: F18, Q11, Q54. 1. INTRODUCTION In both academic research and policymaking agenda there is growing awareness that climate change and the agri-food sector are closely related, and that those links deserve investigation and understanding to analyse the evolution of global agriculture, and to anticipate future challenges such as climate change adaption and mitigation (Falco et al., 2019; Santeramo et al., 2021). Agriculture, on which human welfare depends, is severely affected by climate change. Some adverse effects, already observed, are likely to intensify in the future, contributing to declines in agricultural production in many regions of the world, fluctuations in world market prices, growing levels of food insecurity (Reilly and Hohmann, 1993; Meressa and Navrud, 2020). Adaptation potential and adaptation capability to climate change may exacerbate differences between regions. In a globalised world, the macro-level impacts of climate change are driven by comparative advantage between regions (Bozzola et al., 2021). If impacts of climate change on productivity 124 Bio-based and Applied Economics 10(2): 123-135, 2021 | e-ISSN 2280-6172 | DOI: 10.36253/bae-9676 Emilia Lamonaca, Fabio Gaetano Santeramo, Antonio Seccia differ between regions, then adjustments through production patterns may dampen the adverse effects of climate change (Costinot et al., 2016; Gouel and Laborde, 2021). Although the agricultural sector is identified as the most sensitive and vulnerable sector to climate change (e.g., Deschenes and Greenstone, 2007), the effects of climate change on the wine sector and on different producing regions (i.e., Old World Producers, New World Producers) is still an open question. How do productivity levels react to changes in climate? Do climate change impacts on production patterns differ between Old World Producers and New World Producers? As suggested by Mozell and Thach (2014), the narrow climatic zones for growing grapes may be severely affected both by short-term climate variability and longterm climate change. A vast majority of earlier studies on the impacts of climate change have analysed the effects on domestic markets, leaving underinvestigated the effects on world production (Reilly and Hohmann, 1993). In the wine-related literature, previous studies reveal that the impacts of climate change are likely to differ across producing regions of wine. Jones et al. (2005) suggest that, currently, Old World Producers (i.e., European regions) benefit of better growing season temperatures than New World Producers. However, future climate scenarios may push some regions into climatic regimes favourable to grape growing and wine production (Lamonaca and Santeramo, 2021). All in all, there is the potential for relevant changes in areas planted with vines due to changes in climate (Moriondo et al., 2013; Seccia and Santeramo, 2018). Projected scenarios of future climate change at the global and wine region scale are likely to impact the wine market. In particular, spatial changes in viable grape growing regions, and opening new regions to viticulture would determine new productive scenarios in the wine sector at the global level. Given this background, our contribution aims at understanding how productive patterns allow different producing regions (e.g., Old World Producers, New World Producers) to respond to changes in climate. Specifically, we examine the linkage between climate change and productivity levels in the global wine sector. In this regard, Rosenzweig and Parry (1994) argue that doubling of the atmospheric carbon dioxide concentration would lead to only a small decrease in global agricultural production. In addition, Reilly and Hohmann (1993) suggest that interregional adjustments in production buffer the severity of climate change impacts both at global and domestic level. From a methodological perspective, the study of agricultural supply response has traditionally decomposed it in terms of acreage and yield responses (e.g., Haile et al., 2016; Kim and Moschini, 2018). Our contribution examines how climate change affects acreage and yield response for grapevines. To this aim, we assume that land allocations are consistent with the choices of a representative farmer who maximises expected profit. We posit that cropland can be allocated between grapevines and all other crops. Because these two allocation choices exhaust the set of possible land allocations, total county cropland is assumed to be fixed. Thus, the decision problem can be stated as that of choosing acreage. We assume that the acreage shares are a function of expected per acre revenue, given by the product between the output price and expected yield, and of climate change. Investigating both the responsiveness of grapevine acreage and yield to climate change allows us to conclude on the global supply response. While our cross-countries analysis is informative on the production patterns in the wine sector at a global scale, it cannot conclude on the effects of climate change at the micro-level (e.g., grape growers, wine producers). Indeed, a country-level analysis does not capture differences within countries in terms of both grapevine yield and climate variability, particularly in geographically heterogeneous countries such as the United States, Canada, Russia, China (Kahn et al., 2019). 2. ESTIMATING THE RELATIONSHIP BETWEEN CLIMATE CHANGE AND GRAPEVINES PRODUCTION 2.1 Yield response equation Following Kim and Moschini (2018), we postulate a simple linear equation for yield response. In detail, the expected grapevines yield of county i at time t(yit) is modelled as: yit = α + αi + βTt + γ’Xit,s + εit (1) where αi are country-specific intercepts; Tt is a linear trend variable and β the related parameter; the vector Xit,s includes climate variables specific for county i, time t, and season s (i.e. 30-years rolling average seasonal temperatures and precipitations, Tempit,s and Precit,s), we also posit a quadratic relationship between climate and yields (i.e. Temp2it,s and Prec2it,s); γ’ is the vector of parameter of interest1; α and εit are a constant and the error term. Following the climate literature (e.g., 1 It is worth noting that the parameter captures the climate sensitivity of grapevine yield without considering the implicit adaptation to climate change, differently from analyses based on the Ricardian model of climate change (e.g., Mendelsohn et al., 1994). 125 Climate changes and new productive dynamics in the global wine sector Bio-based and Applied Economics 10(2): 123-135, 2021 | e-ISSN 2280-6172 | DOI: 10.36253/bae-9676 Kurukulasuriya et al., 2011; Massetti et al., 2016), we use a four-season model, assuming that seasonal differences in temperatures and precipitations are likely to impact grapevines productivity. However, we exclude climate normals of the winter season which is characterised by the dormancy of grapevines; in fact, the annual growth cycle of grapevines begins with bud break in the spring season and culminate in leaf fall in the autumn season. We explore the relationship between grapevines yield and climate variables to estimate the potential effects of climate change using either ordinary least squares (OLS) or quantile regression (QR). The model in equation (1) is estimated in an OLS fashion on the whole sample and on subsamples of Old World Producers and New World Producers. The properties of QR have motivated its application in the context of agriculture and weather, mostly focusing on the impact of climate change on various crop yield distributions (Conradt et a., 2015). The QR facilitates a thorough analysis of the differential impact of climate change across the yield distribution; a QR approach is useful in such situations and for considering asymmetry and heterogeneity in climatic impacts (Barnwal and Kotani, 2013). 2.2 Acreage response equation Total county cropland (A) is assumed to be fixed and land allocations are presumed to be consistent with the choices of a representative farmer who maximises expected profit. We posit that agricultural land can be devoted to two alternative uses, grapevines and all other crops. The decision problem can be stated as that of choosing acreage shares sk ≡ Ak ⁄ A, where Ak is the acreage allocated to the k-th use (k = 1 for grapevines and k = 2 for all other crops). Because A is fixed, increased land allocation to any one crop is equivalent to an increase in its share sk, maintaining the land constraint s1 + s2 = 12. Empirically, observed acreage share of grapevines in county i at time t(sit) is modelled as: sit = λ + λi + θTt + φsit-1 + ψrit + ω’Zit + νit (2) 2 Due to a land constraint, a representative farmer may decide to allocate more (less) acreage to grapevine reducing (increasing) the share of acreage devoted to other crops to maximise expected profits. This may be a sort of implicit adaptation to climate conditions. For instance, due to warmer temperatures, acreages devoted to grapevine in Italy may increase to the detriment of acreage intended to other production (e.g., apple tree, pear tree). As suggested in Ricardian literature in climate change economics (e.g., Timmins, 2006; Kurukulasuriya et al., 2011; Bozzola et al., 2018). where the set of conditioning variables includes country-specific trend effects, λi; a time trend, Tt, capturing exogenous technological progress; expected per acre revenue, rit; past acreage shares, sit-1, climate variables, Zit, which may directly affect planting decisions (i.e. 30-years rolling average annual temperatures and precipitations, Tempit and Precit, and their squares, Temp2it and Prec2it). The term λ is a set constant terms; θ, φ, and ψ are parameters to be estimated, ω’ is the vector of climate-specific parameters; νit is the error term. The term sit-1 allows us to account for the behaviour of producers that adjust their acreage when they realise that the desired acreage differs from the acreage realised in the previous year; it captures the dynamic effects on acreage allocation (Santeramo, 2014). Following Kim and Moschini (2018), we interact own output price and expected yields estimated in equation (1), to obtain the expected per acre revenue (i.e., rit = pit ∙ yit). Since our study is a country-level analysis, consistent with Hendricks et al. (2014) we assume that the country-level expected prices are exogenous: this assumption allows us to deal with potential endogeneity of prices. In order to compute the expected per acre revenue variables for the acreage response equations, we rely on the OLS estimate of equation (1). We follow an approach similar to Haile et al. (2016) and Kim and Moschini (2018) and estimate the model in equation (2) using a system generalised method-ofmoments (GMM) estimator, based on a one-step estimation with robust standard errors. In fact, applying OLS estimation to a dynamic panel data regression model, such as in equation (2), results in a dynamic panel bias because of the correlation of the lagged dependent variable with the country-fixed effects (Nickell, 1981). Since current acreage is a function of the fixed effects (λi), lagged acreage is also a function of these country-fixed effects. This violates the strict exogeneity assumption, thus the OLS estimator is upward biased and inconsistent. A solution to this issue consists in transforming the data and removing the fixed effects. However, under the within-group transformation, the lagged dependent variable remains correlated with the error term, and therefore the fixed-effects estimator is downward biased and inconsistent. To overcome these problems, the GMM is a more efficient estimator that allows the estimate of a dynamic panel difference model using lagged endogenous and other exogenous variables as instruments. In particular, the system GMM technique transforms the instruments themselves in order to make them exogenous to the fixed effects (Roodman, 2009). ˆ ˆˆ ˆ 126 Bio-based and Applied Economics 10(2): 123-135, 2021 | e-ISSN 2280-6172 | DOI: 10.36253/bae-9676 Emilia Lamonaca, Fabio Gaetano Santeramo, Antonio Seccia 3. DATA SOURCES AND SAMPLE DESCRIPTION The empirical analysis relies on a rich dataset of historical temperature and precipitation data (from 1961 to 2015) and historical trade flows data (from 1996 to 20153) for 14 countries. The selected countries are Argentina, Australia, Brazil, Canada, China, France, Germany, Italy, New Zealand, Russian Federation, South Africa, Spain, the United Kingdom, the United States. They account for more than two-third of the volume of wine production (70% in 2016, Global Wine Markets, 1860 to 2016 database). This group of countries includes both Old Works Producers and New World Producers and countries belonging to Northern or Southern Hemisphere4. Table 1 provides descriptive statistics for key variables, also distinguishing between Old World Producers and New World Producers. Historical country-specific monthly average temperature and precipitation data have been collected from the Climate Change Knowledge Portal World Bank (World Bank, 2018). Annual and seasonal climatologies (i.e., rolling 30-years averages5) of temperature (in °C) and precipitations (mm) have been constructed using historical weather data. As for seasonal climatologies, monthly data have been clustered into three-month seasons: December (of the previous year) through February as winter, March 3 The longer time period used for climate data allows to build climatologies (i.e. 30-years averages) of temperature and precipitations: in 1996 (the starting point of the final dataset) climate normal is based on a real 30-years average. 4 The list of countries by group is presented in Appendix A.1. 5 Differently from other studies that aggregated to data by weighting each information at the grid level by the amount of agricultural area the grid contains (e.g., Gammans et al., 2017), we use simple average of climate data aggregated at the country level. through May as spring, June through August as summer, and September through November as autumn. These seasonal definitions have been adjusted for the fact that seasons in the Southern and Northern Hemispheres occur at exactly the opposite months of the year. The annual 30-years average temperature is 10.37 ºC (table 1). Within this group, annual average temperatures are about 1 ºC higher for Old World Producers than for New World Producers, reflecting the fact that New World Producers are mostly located to lower latitudes (figure 1). The difference in average temperatures between Old World Producers and New World Producers tends to be higher during winter (3.97 °C of Old World Producers and 0.77 °C of New World Producers; table 1). The annual 30-years average precipitation is 68.55 mm and is about 5 mm greater in Old World Producers Figure 1. List of countries. Source: elaboration on Anderson and Nelgen (2015). Notes: Old World Producers in blue, New World Producers in red. Table 1. Descriptive statistics for key variables. Variable Unit All producers Old World Producers New World Producers Acreage ha 303,640 (±347,791) 560,850 (±435,259) 160,745 (±162,051) Share of acreage -0.01 (±0.02) 0.02 (±0.00) 0.001 (±0.001) Yield t/ha 10.50 (±4.59) 3.96 (±1.22) 12.09 (±1.13) Price USD/t 779.27 (±448.80) 528.60 (±40.70) 708.32 (±396.59) 30-years average temperature (annual) °C 10.37 (±8.51) 10.86 (±1.87) 10.10 (±10.52) 30-years average temperature (spring) °C 9.90 (±9.08) 9.70 (±1.54) 10.01 (±11.28) 30-years average temperature (summer) °C 18.76 (±4.76) 18.26 (±2.54) 19.04 (±5.61) 30-years average temperature (autumn) °C 10.92 (±8.21) 11.57 (±2.03) 10.55 (±10.12) 30-years average precipitation (annual) mm 68.55 (±36.13) 71.89 (±17.46) 66.69 (±43.09) 30-years average precipitation (spring) mm 62.35 (±34.87) 67.18 (±11.14) 59.66 (±42.50) 30-years average precipitation (summer) mm 82.17 (±44.21) 61.95 (±19.52) 93.40 (±49.81) 30-years average precipitation (autumn) mm 74.56 (±44.14) 82.93 (±24.25) 69.91 (±51.47) Note: Average values and standard deviation in parentheses. 127 Climate changes and new productive dynamics in the global wine sector Bio-based and Applied Economics 10(2): 123-135, 2021 | e-ISSN 2280-6172 | DOI: 10.36253/bae-9676 than in New World Producers. However, seasonal differences are observed: during summer, the level of precipitations is much lower in Old World Producers than in New World Producers (table 1). In our sample, we observe a 6% increase in median values of 30-years average temperature over twenty years (figure 2). As suggested in Jones et al. (2005), Old World Producers benefit of better growing seasons as compared to New World Producers. It should be kept in mind, however, that the strength of seasonality varies significantly across the globe, with seasons being more homogenous around the Equator. Country-specific annual data on areas planted with vines (in ha) and yields of areas planted with vines (in t/ha), collected from the FAOSTAT database, are described in table 1. The FAOSTAT database also provides country-level annual acres for agricultural land. Total agricultural land includes two components: i.e., cropland (arable land and land under permanent crops) and land under permanent meadows and pastures. In the methodological framework, we assume that agricultural land can be devoted to two alternative uses, grapevines and all other crops. The latter category should capture all acres that could have been not planted to grapevines. Hence, we obtain the category all other uses as the difference between total agricultural land and acres planted with vines. In our model, we also use countryspecific annual price data for grapes (USD/t), collected from the FAOSTAT database. In order to obtain the reduced per acre revenue, we interact own output price and expected yields estimated in equation (1). Within our sample, despite the expansion of areas planted with vines in New World Producers during the last decades, acres intended to grape growing are, on average, more than three times larger in Old World Producers (561 thousands ha with respect to 161 thousands ha, table 1). However, grapevines yields are much larger for New World Producers (12.09 t/ha) than for Old World Producers (3.96 t/ha). Yields are often not normally distributed but are negatively skewed (e.g., Swinton and King, 1991). This is also what we find in the distribution of grapevines yield in our sample (figure 3). A distribution of yield different from a normal distribution may be associated with the frequent occurrence of outliers; for instance, yield realisations may not follow the pattern described by the majority of yield observations (Conradt et al., 2015). It is worth noting that countries with grapevines yields within 25th percentile are Canada, Spain, France, United Kingdom, New Zealand, Russian Federation, whereas countries with yields of grape within 75th percentile are Argentina, Australia, Brazil, China, Germany, United States, South Africa. 4. RESULTS AND DISCUSSION 4.1 Yield response The estimation results for the yield response, based on equation (1), are reported in tables 2 (OLS estimates)6 and 3 (QR estimates). The results in table 2 show that the higher the average temperatures in producing countries during summer, the lower the grapevines yield. Greater precipitations are beneficial for yield during the early growing season (i.e., spring), but detrimental during the 6 In a sensitivity analysis, we analyse the effects of annual climatic variables on grapevine yields. The results, reported in table A.2 in the Appendix, highlight differences between Old World Producers and New World Producers. While higher annual average temperatures are detrimental (up a certain threshold) for Old World Producers, New World Producers benefit of greater annual average temperatures and precipitations. 55.0 55.2 55.4 55.6 55.8 56.0 56.2 56.4 56.6 56.8 57.0 10.5 10.6 10.7 10.8 10.9 11.0 11.1 11.2 11.3 11.4 11.5 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Precipitation (mm) Temperature ( °C) Temperature Precipitation Figure 2. Median 30-years temperatures and precipitations in 19972015. Source: elaboration on data from CRU of University of East Anglia. Note: data refer to the sample of 14 major producers of wine. Figure 3. Distribution and descriptive statistics for grapevines yield. Min: 1.22 Max: 19.50 Median: 10.57 Mean: 10.50 Std. Dev.: 4.59 Skewness: -0.15 Kurtosis: 2.25 128 Bio-based and Applied Economics 10(2): 123-135, 2021 | e-ISSN 2280-6172 | DOI: 10.36253/bae-9676 Emilia Lamonaca, Fabio Gaetano Santeramo, Antonio Seccia late growing season and the harvest time (i.e. summer and autumn). The relationship between summer climate and yields is nonlinear7. The overall effects are mostly driven by the impacts of climate change on grapevines yields of New World Producers. Differently, grapevines yield of Old World Producers seem not affected by climate change. The results are consistent with evidence from vine-related literature. In fact, Merloni et al. (2018) report that higher temperatures can have a negative impact on grapevines yield and quality. An increase in extreme high temperatures in summer may have adverse consequences on grapevines phenology (Briche et al., 2014). In addition, Ramos et al. (2008) suggest that seasonal distribution of precipitation matter, with larger rainfall levels being crucial for grapevines at the beginning of the growing season (i.e., spring) whereas more stable precipitations are desirable from flowering to ripening (i.e., summer and autumn). The OLS approach is applied when the dependent variable is normally distributed, whereas QR is employed when the variable is not normally distributed (see figure 3). The QR (median) is more robust to outliers than mean regression (OLS)8. Furthermore, QR provides a clearer understanding of the data by assessing the effects of explanatory variables on the location and the scale parameters of the model (Conradt et a., 2015). The results of the QR reported in table 3 mostly confirm the non-linear relationship between grapevines yields and average temperatures in producing countries during summer. No substantial differences are observed across different quantiles of the distribution of grape7 The results are robust also controlling for different combinations of fixed effects: the results are reported in tables A.3 and A.4 in the Appendix. We further detect a non-linear relationship between grapevine yield and summer precipitation controlling for time fixed effects (common to all countries) and country-specific fixed effects. Differently, we cannot conclude on the relationship between grapevine yield and detrended climate variables obtained from the yearly weather deviation from the long-run climate (30-year rolling average), as recently proposed by Khan et al. (2019). The result is not surprising: while detrended climate variables capture short-run changes in climate conditions (i.e., weather shocks), 30-year rolling average temperatures and precipitations inform on long-run changes in climate conditions: It is unlikely that weather shocks on a year-by-year basis affect the responsiveness of the viticultural sector, but long-run changes in climate capture structural changes in the sector and are more likely to influence production decisions of a multi-year crop. A comparison between shortand longrun analyses is reported in table A.5 in the Appendix. 8 We conduct a multidimensional outlier detection analysis based on the ‘bacon’ algorithm, which identifies outliers based on the Mahalanobis distances (Billor et al., 2000, Weber, 2010). The algorithm allows the identification and removal of observations characterised by implausibly large or low entries of key variables. The results of the model estimated without outliers, reported in tables A.6 and A.7 in the Appendix, confirm the main results, although the effect of temperatures and precipitations on grapevine yields tend to be lower. vines yields. Differently, the results reveal that lower yield realisations (i.e., within 25th percentile) tend to be most affected by greater precipitations during the harvest time (i.e., autumn). It is worth noting that countries with grapevines yields within 25th percentile are mostly cool climate wine regions such as Canada and Russian Federation. Cool regions tend to have also higher rainfall levels and yields tend to be lower on average, rising production costs (Anderson, 2017). Table 2. Estimation results for grapevines yields, OLS. Variables Dependent variable: yield All producers Old World Producers New World Producers Temperature (spring) 1.4440 -9.5441 -1.4800 (1.7044) (12.7571) (2.1761) Temperaturesquared (spring) -0.3044*** 0.3965 -0.2577** (0.0747) (0.5755) (0.1209) Temperature (summer) -16.3650** -22.5187 -1.8786 (7.1026) (14.6183) (11.2236) Temperaturesquared (summer) 0.4258** 0.4752 0.3047 (0.1955) (0.3634) (0.3264) Temperature (autumn) 0.6543 -0.6787 -0.5129 (1.9410) (12.3068) (2.3252) Temperaturesquared (autumn) 0.0761 -0.0685 0.1321 (0.0888) (0.4882) (0.1181) Precipitation (spring) 0.5227* 0.4326 0.8057* (0.2795) (0.7043) (0.4339) Precipitationsquared (spring) -0.0041*** -0.0035 -0.0052*** (0.0015) (0.0048) (0.0019) Precipitation (summer) -0.3230* -0.0678 -0.0427 (0.1906) (0.3849) (0.3922) Precipitationsquared (summer) 0.0013 -0.0001 0.0005 (0.0009) (0.0022) (0.0013) Precipitation (autumn) -0.3507** -0.3838 -0.4272 (0.1601) (0.4282) (0.3758) Precipitationsquared (autumn) 0.0019** 0.0019 0.0019 (0.0008) (0.0020) (0.0017) Time trend 0.1392*** 0.3459* 0.0109 (0.0477) (0.1756) (0.1007) Observations 280 100 180 R-squared 0.9314 0.9656 0.8930 Notes: OLS estimate of equation (1) on the whole sample (All producers) and subsamples of Old World Producers and New World Producers. All specifications include country-specific constants. Robust standard errors are in parentheses. *** Significant at the 1 percent level. ** Significant at the 5 percent level. * Significant at the 10 percent level. 129 Climate changes and new productive dynamics in the global wine sector Bio-based and Applied Economics 10(2): 123-135, 2021 | e-ISSN 2280-6172 | DOI: 10.36253/bae-9676 4.2 Acreage response Table 4 presents the estimation results under the acreage models. All dynamic models (All Producers, Old World Producers and New World Producers) are based on a one-step GMM estimator. The ArellanoBond test for autocorrelation is used to test for serial correlation in levels. The test results indicate that the null hypothesis of no second-order autocorrelation in residuals cannot be rejected, indicating the consistency of the system GMM estimators. According to the Sargan test results, we fail to reject the null hypothesis of instrument exogeneity: the system GMM estimators are robust but weakened by many instruments. We fail to find a significant acres-price relationship, which could imply that many grapevines’ producers do not form their price expectations on the basis of information on expected per acre revenues. More importantly, the estimation results reveal that higher annual temperatures in producing countries are beneficial for grapevines acreage share. This is true for both Old and New World Producers, despite the effects are much larger in Old World Producers. As suggested in Ruml et al. (2012), among the many climatic factors affecting wine production, temperature appears to be most important. Differently, severe rainfall levels is significantly associated with less grapevines share. The negative effects of greater annual precipitations is entirely associated with New World Producers, whereas the Old World Producers seem not affected by changes in the rainfall levels. Table 3. Estimation results for grapevines yields, quantile regression. Variables Dependent variable: yield 25th percentile 50th percentile 75th percentile Temperature (spring) 0.8721 0.7711 1.3070 (1.9059) (1.8734) (2.3574) Temperaturesquared (spring) -0.1418* -0.2405*** -0.3368*** (0.0756) (0.0812) (0.1073) Temperature (summer) -22.4737*** -27.0681*** -23.1306*** (4.5501) (7.0368) (7.0902) Temperaturesquared (summer) 0.5454*** 0.7064*** 0.6102*** (0.1219) (0.1864) (0.1763) Temperature (autumn) 3.0239 1.9043 2.2129 (2.1210) (1.2873) (2.4223) Temperaturesquared (autumn) -0.1279 -0.0515 0.0525 (0.0813) (0.0611) (0.0998) Precipitation (spring) 0.2402 0.6707** 0.4740 (0.2974) (0.2899) (0.2913) Precipitationsquared (spring) -0.0024 -0.0048*** -0.0035* (0.0018) (0.0017) (0.0018) Precipitation (summer) -0.2866 -0.0272 -0.1956 (0.2024) (0.1155) (0.1925) Precipitationsquared (summer) 0.0014 -0.0001 0.0011 (0.0012) (0.0006) (0.0011) Precipitation (autumn) -0.3157* -0.1921 -0.1535 (0.1691) (0.1477) (0.1627) Precipitationsquared (autumn) 0.0019** 0.0011* 0.0010 (0.0008) (0.0006) (0.0007) Time trend 0.1523*** 0.1796*** 0.1024* (0.0574) (0.0534) (0.0545) Observations 280 280 280 Notes: QR estimate of equation (1) on the whole sample. All specifications include country-specific constants. Robust standard errors are in parentheses. *** Significant at the 1 percent level. ** Significant at the 5 percent level. * Significant at the 10 percent level. Table 4. Estimation results for grapevines acreage, Old World Producers and New World Producers. Variables Dependent variable: acreage share All Producers Old World Producers New World Producers Lagged acreage share 0.995*** 0.795*** 0.953*** (0.001) (0.046) (0.012) Expected per acre revenue -0.00003 -0.163 -0.0003 (0.00003) (0.109) (0.001) Temperature (annual) 0.107*** 38.983* 0.131*** (0.019) (22.496) (0.020) Temperature-squared (annual) -0.006*** -0.134 -0.008*** (0.001) (1.574) (0.001) Precipitation (annual) -0.107*** 18.447 -0.122*** (0.033) (11.384) (0.028) Precipitation-squared (annual) 0.001*** -0.120 0.001*** (0.0002) (0.081) (0.0001) Test for AR(1): p-value 0.096 0.106 0.239 Test for AR(2): p-value 0.238 0.326 0.266 Sargan test: p-value 0.134 0.592 0.926 Number of instruments 149 47 123 Notes: One-step generalised method-of-moments (GMM) estimate of equation (2) on the whole sample and on subsamples of Old World Producers and New World Producers. All specifications include a constant and a time trend. Coefficients and standard errors estimated are of the order of 10-6 for ‘expected per acre revenue’ and of 10-4 for climate variable. Observations are 198 for all producers, 47 for Old World Producers and 151 for New World Producers. Robust standard errors are in parentheses. *** Significant at the 1 percent level. ** Significant at the 5 percent level. * Significant at the 10 percent level. 130 Bio-based and Applied Economics 10(2): 123-135, 2021 | e-ISSN 2280-6172 | DOI: 10.36253/bae-9676 Emilia Lamonaca, Fabio Gaetano Santeramo, Antonio Seccia 5. CONCLUDING REMARKS Climate change has the potential to impact the agricultural sector and the wine sector in particular (Mozell and Thach, 2014). Most of the previous studies analysing the impact of climate change on agriculture do not consider the effects of climate change on world production, markets and trade patterns (Reilly and Hohmann, 1993). Our analysis allowed us to understand if climate change is able to affect productivity levels of grapevines. Overall, we found that grapevines yield suffers from higher temperatures during summer, whereas precipitations have a varying impact on grapevines depending on the cycle of grapevines. In particular, we observed that greater precipitations are beneficial during the early growing season (spring), but detrimental during the late growing season and the harvest time (summer and autumn). Differently, acreage share of grapevines tends to be favoured by higher annual temperatures, whereas greater annual precipitations tend to be detrimental. The impacts however vary between Old World Producers and New World Producers, also due to heterogeneity in climate between them: the effects of temperatures are less pronounced for New World Producers, whereas precipitations have no effects for Old World Producers. As suggested in previous studies (e.g., Jones et al., 2005), Old World Producers benefit of better growing season, but climate change may push New World Producers into more favourable climatic regimes. The opening of new regions, benefiting of better climatic regimes, to viticulture would determine new productive scenarios and, as a result, new trade dynamics (Macedo et al., 2019). New productive scenarios are likely to favour the production of varietal wines from autochthonous grapes whose quality is strongly related to microclimatic and pedological conditions (Seccia et al., 2017). In addition, changes in trade regulations, that have largely influenced the agri-food market, are modifying also global trade of wine (Santeramo et al., 2019; Seccia et al., 2019). Such dynamics should not be neglected. Future research should be intended to examine how climate change could affect global trade of wine and to understand how importers and exporters could react to new trade dynamics, due to climate change, in terms of trade regulations. ACKNOWLEDGMENT The research has been supported by a Research Grant funded by the International Organisation of Vine and Wine (OIV). The authors are grateful to Dr. Martina Bozzola for collection and organisation of climate data, to Tatiana Svinartchuk, Tony Battaglene, and to the seminar audiences at the 9th AIEAA Conference and the EGU General Assembly 2021 for helpful comments. REFERENCES Anderson, K. (2017). How might climate changes and preference changes affect the competitiveness of the world’s wine regions? Wine Economics and Policy 6(1): 23-27. Anderson, K., and Nelgen, S. (2015). Global Wine Markets, 1961 to 2009: A Statistical Compendium. University of Adelaide Press. Barnwal, P., and Kotani, K. (2013). Climatic impacts across agricultural crop yield distributions: An application of quantile regression on rice crops in Andhra Pradesh, India. Ecological Economics 87: 95-109. Billor, N., Hadi, A.S. and Velleman, P.F. (2000). BACON: Blocked adaptive computationally efficient outlier nominators. Computational Statistics & Data Analysis 34: 279-298. Bozzola, M., Lamonaca, E., and Santeramo, F.G., (2021). On the impact of climate change on global agri-food trade. Working Paper. Bozzola, M., Massetti, E., Mendelsohn, R., and Capitanio, F. (2018). A Ricardian analysis of the impact of climate change on Italian agriculture. European Review of Agricultural Economics 45(1): 57-79. Briche, E., Beltrando, G., Somot, S., and Quénol, H. (2014). Critical analysis of simulated daily temperature data from the ARPEGE-climate model: application to climate change in the Champagne wine-producing region. Climatic Change 123(2): 241-254. Conradt, S., Finger, R., and Bokusheva, R. (2015). Tailored to the extremes: Quantile regression for index‐ based insurance contract design. Agricultural Economics 46(4): 537-547. Costinot, A., Donaldson, D., and Smith, C. (2016). Evolving comparative advantage and the impact of climate change in agricultural markets: Evidence from 1.7 million fields around the world. Journal of Political Economy 124(1): 205-248. Deschenes, O., and Greenstone, M. (2007). The Economic Impacts of Climate Change: Evidence from Agricultural Output and Random Fluctuations in Weather. The American Economic Review 97(1): 354–85 Falco, C., Galeotti, M., and Olper, A. (2019). Climate change and migration: Is agriculture the main channel? Global Environmental Change 59: 101995. Gammans, M., Mérel, P., and Ortiz-Bobea, A. (2017). Negative impacts of climate change on cereal yields: