Drivers of Food Price Inflation in Turkey The World Bank Group © 2020 The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org Some rights reserved This work is a product of the staff of The World Bank. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because the World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Attribution — Please cite the work as follows: “World Bank. 2020. Drivers of Food Price Inflation in Turkey. © World Bank.” All queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. Cover photo: Simone D. McCourtie / World Bank  Content Acknowledgements................................................................................................... vi Executive Summary...................................................................................................vii Introduction................................................................................................................ 1 ... 3 I. Food Price Development – Variations Across Agricultural Products and Regions. II. Drivers of Food Price Inflation ............................................................................... 9 III. Regional Price Formation and Transmission: The Case of Table Tomatoes, Green Peppers and Onions................................................................................... 18 Overview of Price Dynamics for Table Tomatoes, Green Peppers and Dry Onions............ 18 An Overview of the Tomato, Green Pepper and Onion Sub-Sectors................................. 20 An Econometric Analysis of Price Formation and Transmission in Table Tomato, Green Pepper and Onion Markets. ................................................................................... 32 .................................................................................... 44 IV. Policy Recommendations. References................................................................................................................ 47 ........................................... 49 Appendix 1. Turkey Administrative Structure (NUTS-2). Appendix 2. Supply and Demand Balances for Tomatoes, Green Peppers, and Dry Onions..................................................................................................... 50 Appendix 3. Methodology for Price Co-movement Analysis..................................... 52 Appendix 4. Price Decomposition Results for Tomatoes, Green Peppers, and Dry Onions. .................................................................................................... 53 Appendix 5. Bivariate Error Correction and Network Centrality Methodology......... 54 Appendix 6. Domestic Market Linkages for Table Tomatoes, Green Peppers, .................................................................................................... 56 and Dry Onions. Appendix 7. Domestic Market Linkages for Table Tomatoes, Green Peppers, and Dry Onions – Spatial Analysis........................................................................ 57 Appendix 8. Production and Prices for Table Tomatoes, Green Peppers, and Dry Onions Across the Regions..................................................................... 59 Appendix 9. Price Transmission Methodology.......................................................... 62 ................................................... 63 Appendix 10. Price Transmission Analysis Results. © 2020 The World Bank Group iii Drivers of Food Price Inflation in Turkey List of Tables Table 1. ............................................................... 9 Supply and Demand Drivers of Food Prices. Table 2. ................................................ 11 Impact of Land Productivity on Food Price Inflation. Table 3: Relative Labor Productivity in Service and Industry Sectors, ..................................................................................................12 Selected Countries. Table 4. ............................................................................... 15 Tomato Cost Structure, TL/Ton. Table 5. Product Groups with the Highest MFN Tariff Rates Applied by Turkey, 2005–2018. ............................................................................................................. 16 Table 6. ..............21 Global Tomato Production (For Fresh Consumption and Processing), Tons. Table 7. Tomato Consumption in Turkey (Fresh and Processed)............................................23 Table 8. Global Chilies and Pepper Production, Tons.............................................................24 Table 9. ..................................26 Green Pepper Consumption in Turkey (Fresh and Processed). Table 10. Global Dry Onion Production, Tons..........................................................................27 ...........................................................................28 Table 11. Dry Onion Consumption in Turkey. Table 12. Summary of State Support Payments for Tomatoes, Green Peppers, and Onions.............................................................................................................. 31 Table 13. Pairwise (Consumer and Producer) Cointegration Regression and Tests ..................................................................................40 for Monthly Dry Onion Prices. Table 14. Framework for Public Policy Options to Address High Food Price Inflation and Volatility...........................................................................................................44 List of Figures Figure 1. Turkish, European Union and United States Food Price Inflation................................ 3 ......................................4 Figure 2. Monthly Food Inflation and USD-TL Exchange Rate Change. Figure 3. Comparison of Food and Core Price Inflation ............................................................ 5 Figure 4. Price Inflation and Volatility of Processed and Unprocessed Food.............................. 6 ................................................... 6 Figure 5. Price Dynamics of Unprocessed Food Components. Figure 6. Dynamics of Food Price Sub-Indices..........................................................................7 Figure 7. Food Price Volatility Dynamics . ................................................................................7 ........................................................10 Figure 8. Processed vs. Primary Food Trade, Billion USD. Figure 9. Correlation Between Consumer Prices and Productivity Levels for Tomatoes and Dry Onions....................................................................................................... 11 ........................................12 Figure 10. Producer Support Estimate as % of Gross Farm Receipts. ...... 13 Figure 11. GSSE to Agriculture Relative to the Aggregate Value of Agricultural Production. Figure 12. The Spatial Pattern of Soil Erosion Across Regions, Mg/Ha per Year........................ 13 Figure 13. Water Stress Assessment Across Regions................................................................14 Figure 14. Comparison of Agricultural PPI and Food Price Index...............................................14 iv © 2020 The World Bank Group  Figure 15. MFN Rates Applied to Vegetable Imports.................................................................17 Figure 16. Nominal Protection Coefficients Comparison ..........................................................17 ..................................................................... 19 Figure 17. Regional Differences in Tomato Prices. Figure 18. Regional Differences in Green Pepper Prices (Sivri Variety)...................................... 19 .................................................................20 Figure 19. Regional Differences in Dry Onion Prices. Figure 20. Regional Table Tomato Production, Production and Yields ......................................22 Figure 21. Comparison of Turkish Tomato Yields with the Yields in Comparator Countries . .....22 .........................24 Figure 22. Fresh Tomato Exports Compared to Total Fresh Vegetable Exports. Figure 23. Regional Green Pepper Production and Yields (All Pepper Varieties)........................25 Figure 24. Dynamics of Regional Green Pepper Yields (All Pepper Varieties).............................26 Figure 25. Share of Green Pepper Exports in Total Fresh Vegetables Exports ..........................27 Figure 26. Regional Dry Onion Production and Yields...............................................................27 ...................................................................28 Figure 27. Dynamics of Regional Dry Onion Yields. ..................................29 Figure 28. Share of Dry Onion Exports in Total Fresh Vegetable Exports. Figure 29. Share of Price Variance Explained by Seasonal and Cyclical Components, and a Trend............................................................................................................. 33 Figure 30. Seasonal Component of Green Pepper Prices..........................................................34 Figure 31. Seasonal Component of Table Tomato Prices........................................................... 35 ....................... 36 Figure 32. Table Tomato Price Structure During the Harvest and Lean Seasons. Figure 33. Green Pepper Price Structure During the Harvest and Lean Seasons.......................37 Figure 34. Dry Onion Price Structure During the Harvest and Lean Seasons............................. 38 Figure 35. Comparison of Producer and Consumer Prices for Onions.......................................41 Figure 36. Comparison of Producer and Consumer Prices for Table Tomatoes..........................42 Figure 37. The Structure of the Fresh Tomato Value Chain in Turkey........................................43 © 2020 The World Bank Group v Drivers of Food Price Inflation in Turkey Acknowledgements This report was prepared by a  team led by Çakmaklı, Selva Demiralp, Sevcan Yeşiltaş and Kateryna Schroeder (Agriculture Economist and Muhammed A. Yıldırım (Koç University) prepared Task Team Lead) and Mustafa Alver (Operations the background paper on the drivers of food price Officer and co-Task Team Lead), under the inflation at the national level. Michael Norton con- overall guidance of Frauke Jungbluth, Practice ducted spatial analysis. Aolin Gong, Sinan Hatik, Manager. The team was comprised of John Varun Kshirsagar, and Bora Surmeli provided valu- Baffes (Senior Agriculture Economist, EPGDR), able research support. Administrative assistance Alain Kabundi (Senior Economist, EPGDR), Ulrich was provided by Rosalie Trinidad and Funda Canli. Schmitt (Lead Agriculture Economist, SCAAG), Editorial support was provided by Luisa LaFleur David Tuchschneider (Senior Rural Development and graphic design by Andrey Fedorov. The au- Specialist, SCAAG), and Pinar Yasar (Senior thors are grateful to Erdem Atas, Christopher Brett, Economist, ECCTR). Laurent Debroux, Paavo Eliste, Madhur Gautam, David Knight, Julian Lampietti, Habib Nasser Rab, The background paper on spatial and vertical Luz Berania Diaz Rios, and Sergiy Zorya for the price transmission analysis was prepared by Barry thoughtful comments that helped to improve the Goodwin (North Carolina State University). Cem study. vi © 2020 The World Bank Group Executive Summary Executive Summary Food price inflation has increased persistently are squeezed between high input costs and low in recent years in Turkey with a widening diver- farm-gate prices. gence from international food price inflation. In Turkey, food price inflation has grown at an aver- Against this background, the study analyzes age rate of 11.5 percent since 2011 and reached the main inefficiencies in the Turkish agricultur- its peak in 2019 at an average rate of 30 percent. al sector through the lens of food price forma- Starting in 2012, Turkish food price inflation began tion and discusses the policy actions needed to to diverge from international food price inflation. strengthen the sector’s competitiveness. This is As world prices reached their peak in 2011 and done through a  series of econometric exercises started leveling off, increases in Turkish food prices aimed at empirically testing the factors that drive accelerated. food price inflation at the national level and a deep dive into the price formation and transmission for Three interlinked dynamics characterize food table tomatoes, green peppers, and dry onions, price development in Turkey, with a  magnitude the top three fresh vegetables produced in Turkey, not seen in other markets: (i) domestic food price across country regions to identify inefficiencies in inflation that is consistently above core inflation; these selected value chains that contribute to ris- (ii) unprocessed foods, and, particularly, fresh veg- ing prices. etables and fruits, that drive food price inflation; and (iii) high food price volatility. While none of The results of the analysis show that inefficien- these characteristics are unusual in and of them- cies in agricultural markets, augmented by mac- selves, their magnitude in Turkey when compared roeconomic factors, have put food prices on an to other countries, makes them highly unusual. upward trajectory. The depreciation of the Turkish lira and inflation expectations, demand-side pres- High food prices and price volatility bear welfare sure from a growing population, changing consum- implications for both consumers and agricultur- er preferences, and supply-side elements such as al producers. For consumers, food prices carry low productivity, constitute the mix of factors that considerable weight in their expenditure baskets— are driving food price inflation over the long run. up to 29 percent for the lowest income group in These factors exist alongside short-run supply and Turkey. Rising food price inflation reduces the pur- demand imbalances at the local level that are driv- chasing power of Turkish households, if all else re- en by seasonality as well as limited spatial and ver- mains equal. On the producer side, high food pric- tical market integration, and lead to increased price es can potentially create incentives for increased variability across the country. Particularly troubling productivity and raise farmers’ incomes and com- is the increasing divergence between producer and petitiveness. However, as this study shows, high- consumer prices that implies that producers do not er consumer prices largely do not pass through receive market price signals due to structural inef- to producers in Turkey. As input prices have been ficiencies along value chains and limited cross-re- growing at a  rate higher than that of food price gional linkages. Short-term positive price shocks inflation, Turkish agricultural producers find them- stemming from such inefficiencies can further exert selves in a difficult situation as their profit margins upward pressure on price levels over time. © 2020 The World Bank Group vii Drivers of Food Price Inflation in Turkey The report findings underscore the need to transparency and linkages are other priorities for move away from short-term policy responses policymakers to improve market efficiency and al- to a  broader policy strategy aimed at stabiliz- low market participants to respond to existing ar- ing prices through increased productivity and bitrage opportunities in a  timely manner. Limited improved market linkages. A  reevaluation of in- domestic market integration exacerbates seasonal vestment priorities and the feasibility of agricul- price fluctuations and results in Turkish farmers tural state support must be undertaken in light not benefitting from price increases. This reduces of price developments and must be accompanied their welfare and limits their incentives to invest in by stronger efforts to boost agricultural produc- productivity-enhancing technologies. The digitali- tivity growth in order to mitigate any longer-term zation of agriculture offers great promise to allevi- rise in food prices. In the short term, better ac- ate some of the friction that exists in Turkish value cess to credit and better extension services can chains and increase the efficiency of agricultural help improve productivity. In addition, incentives production and distribution. Ultimately, higher pro- to implement environmentally sustainable practic- ductivity growth and more efficient markets would es need to be introduced to decrease the current allow for not only stabilizing food prices, but also and future implications of natural resource deple- for making agricultural markets more resilient to tion. In the medium and long term, state expendi- any broad-based price pressures that may emerge tures in agriculture should be repurposed toward from additional external shocks, such as climate the provision of public goods. Improving market change variability or the COVID-19 aftermath. viii © 2020 The World Bank Group Introduction Introduction Turkish food price inflation has greatly outpaced share of their income on food. A median Turkish world price inflation since 2012. A  general up- household devoted, on average, 20.3 percent of its ward trend in global food prices began in 2006 and expenditures to food and non-alcoholic beverages reached its peak during 2011, after which prices (Turkstat, 2018). This share is higher for low-in- began to level off. Turkish food price levels, in con- come groups at 28.7  percent. Rising food price trast, kept steadily rising. Between 2012 and 2019 inflation reduces the purchasing power of Turkish average food prices in the international markets households, all else being equal, while increases in declined by 17  percent while they increased by price volatility translate into larger fluctuations in 130 percent in Turkey. purchasing power. Domestically, food price development in Turkey is High food prices and volatility also have import- characterized by three interlinked dynamics: do- ant consequences for the welfare of produc- mestic food price inflation that is consistently above ers. Higher agricultural prices can raise farmers’ core inflation; unprocessed foods, particularly, fresh incomes, improve competitiveness and stimulate fruits and vegetables, that drive food price inflation; investment for longer-term economic growth. and high food price inflation volatility. In almost However, the potentially positive impacts of high all years since 2003, the inflation rate of food and prices in the long term depend critically on whether non-alcoholic beverages exhibited a positive diver- appropriate policies and infrastructure are in place gence from core inflation. Between January  2003 to allow the producers to benefit from higher pric- and February 2020, yearly food price inflation av- es. High price volatility, on the other hand, bears eraged 10.6  percent compared to average core negative implications for producers as it creates inflation of 8.3  percent, and this divergence has uncertainty, and can negatively affects farmers’ increased in recent years. Unprocessed foods ex- decisions to invest in productivity improvements. hibit higher inflation levels and volatility than pro- cessed foods with the divergence far exceeding Rising food price inflation and volatility neces- those in comparator countries. Food price levels sitate an understanding of the mechanics of in Turkey have also been increasingly volatile over food price formation in Turkey. The objective of the past several years. Between 2003 and 2007, the study is to conduct an in-depth analysis of the average annual food price inflation volatility levels main inefficiencies in the Turkish agricultural sec- were equal to 0.87  percent (after accounting for tor through the lens of food price formation and seasonality and time trends), while in 2014–2018, formulate policy actions to strengthen the sector’s the comparable average amounted to 1.25 percent. competitiveness. This is done through a series of None of these dynamics are unusual in themselves, econometric exercises aimed at empirically testing what makes them so in the Turkish context is their the factors that may be driving food price infla- magnitude when compared to other countries, in- tion at the national level. In addition, a deep dive cluding the European Union and the United States. into the price formation and transmission for to- matoes, green peppers, and dry onions—the top High food price inflation can have a major impact three fresh vegetables produced in Turkey—across on low-income households that spend a  large the country’s regions was conducted to identify © 2020 The World Bank Group 1 Drivers of Food Price Inflation in Turkey inefficiencies in these selected value chains that level, grounded in econometric analyses and a lit- have contributed to rising price levels. erature review. Section Three includes an in-depth analysis of regional price formation and transmis- The report is structured as follows: Section One sion for tomatoes, green peppers, and dry onions. presents an overview of food price dynamics Section Four presents policy recommendations across different agricultural commodities and re- aimed at tackling the existing inefficiencies in gions. Section Two analyzes various factors that Turkey’s agricultural sector that are contributing drive food price inflation in Turkey at the national to rising food prices. 2 © 2020 The World Bank Group I. Food Price Development – Variations Across Agricultural Products and Regions I. Food Price Development – Variations Across Agricultural Products and Regions The divergence between Turkish and interna- levels in the international market declined by tional food price levels has been increasing, 17  percent while in Turkey food prices increased largely driven by the depreciation of the Turkish by 130 percent. Akcelik et al. (2016) show that the lira (Figure 1A). During the world price spikes of upward trend in food prices in the European Union 2007–2008, international food prices increased by ended after 2009, whereas in Turkey food prices 68  percent on average, compared to 17  percent kept rising. Similarly, since about 2007, Turkish in Turkey. However, as international food prices inflation started exhibiting greater volatility com- returned to their pre-crisis levels by August 2009, pared to both the European Union and the United Turkish food prices continued to increase. The sit- States (Figure 1B). uation repeated itself during the 2011–2012 inter- national food prices surge. Once again, internation- A comparison of the growth rates for food price al food prices increased much faster than those in indices with the depreciation of the Turkish lira Turkey, but the subsequent price decline observed suggests that the exchange rate has been an im- in the international markets was not experienced portant driver of the divergence between Turkish in Turkey. From 2012 to 2019, average food price and world prices (Figures 2A and B). Ozmen Figure 1. Turkish, European Union and United States Food Price Inflation A) Turkish domestic and international food price indices, 2003–2019 B) Monthly change in food price index (de-trended and de-seasoned) 400 10 Food Price Index – 2006–2007 = 100 % 350 8 6 300 4 250 2 200 0 150 -2 100 -4 50 -6 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 World Turkey Turkey EU USA Source: FAOSTAT, 2019 Source: Cakmakli, C. et al., 2019 Note: Food price indices are calculated in national currencies. © 2020 The World Bank Group 3 Drivers of Food Price Inflation in Turkey Figure 2. Monthly Food Inflation and USD-TL Exchange Rate Change A) Food and non-alcoholic beverage index vs. USD/TL exchange rate B) Monthly change 600 7 25 USD/TL Exchnage Rate Food and Non-Alcoholic Bev. Index, 2003=100 % 500 6 20 5 15 400 4 10 300 3 5 200 2 0 100 1 -5 0 0 -10 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Food and Non-Alcoholic Beverage Index Food and Non-Alcoholic Beverage Index USD/TL Exchange Rate USD/TL Exchange Rate Source: Central Bank of Turkey, 2019 Source: Çakmaklı, C. et al. based on data from the Central Bank of Turkey, 2019 and Topaloglu (2017) analyze the pass-through Since 2003, Turkey’s food and non-alcoholic bev- of import prices and exchange rate into inflation erages index—the measure of overall food price for Turkey over the sample period of 2005–2015 inflation used in this report—increased much taking the heterogeneous nature of the CPI into faster and displayed higher volatility than core account. Their results show that the exchange inflation and the overall consumer price index rate pass-through to processed and unprocessed (CPI). From 2003 to 2019, food price inflation av- food prices has been one of the highest (23 and eraged 10.6 percent, which was higher than core 27 percent, respectively) among CPI components. inflation, at 8.3  percent, and headline inflation, Campa and Goldberg (2005) provide evidence at 9.4 percent (Figure 3A). In recent years the di- that countries with higher exchange rate volatili- vergence has increased. The food sub-index that ty have higher pass-through elasticities of import accounts for 93 percent of the food and non-al- prices. Two additional studies, Kara and Ögünç coholic beverages index, is the primary driver be- (2008, 2012) and Kara et al. (2017), also confirm hind the increasing divergence from core inflation. exchange rate spillovers as significant internation- Similarly, volatility in food price inflation exceeds al sources of inflation in Turkey. Exchange-rate volatility in the core inflation index (Figure 3B). pass-through into food prices can be channeled After de-trending and de-seasoning the indices, through both the imports of final food products core inflation does not exhibit much volatility over and of intermediate materials used for production. time, except for the wild swings that occurred in Between 2012 and 2018, Turkish imports of agri- September and October 2018 when Turkey expe- food products (HS1-24) increased by 19.6 percent, rienced a significant currency depreciation. Food reaching $12.8 billion in 2018. Cereals, oilseeds, price inflation, in contrast, exhibits a much high- and live animals accounted for 44.3 percent of to- er volatility and has been increasing over time. tal imports in 2018. The behavior of domestic food prices, when com- pared to non-food prices in the country, suggest At the same time, Turkey’s food prices have been that there are underlying domestic factors that increasing faster and displaying higher volatil- are exerting upward pressure on food price levels ity than overall prices in the domestic market. and volatility. 4 © 2020 The World Bank Group I. Food Price Development – Variations Across Agricultural Products and Regions Figure 3. Comparison of Food and Core Price Inflation A) Price levels B) Monthly changes 600 8 Index. 2003=100 % 500 6 4 400 2 300 0 200 -2 100 -4 0 -6 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Food and Non-Alcoholic Beverages Food and Non-Alcoholic Beverages Core Core CPI Source: Central Bank of Turkey, 2020 Source: Cakmakli, C. et al., 2019 Food price inflation levels and volatility can Within unprocessed foods, fresh fruits and vege- vary significantly across regions1 within tables are driving the volatility (Figure 5). A visu- Turkey. As of June 2019, the aggregate Turkish al comparison of the price index volatility of fresh food price index stood at 501. Regionally, the fruits and vegetables and other unprocessed foods highest level of food price inflation was observed leads to the conclusion that the variability in fresh in Izmir with the highest recorded index level of fruits and vegetables is driving the overall volatility 552, followed by Artvin, Giresun, Gümüşhane, of unprocessed foods. Hence, understanding the Ordu, Rize, and Trabzon at 534. The lowest food reasons behind the limited transmission of volatili- index was observed in Edirne, Kırklareli, and ty between processed and unprocessed foods may Tekirdağ at 479. lie in the analysis of the transmission of volatility between processed and unprocessed fruits and The unprocessed food price index exhib- vegetables. For fresh tomatoes and green peppers, its higher levels and volatility than the pro- which this study analyzes in depth, 31.5  percent cessed foods price index (Figure 4B). Between and 39  percent of production, respectively, goes January 2004 and June 2019, yearly unprocessed into processing, while the rest is consumed fresh food price inflation averaged 11.6 percent, com- or exported. Vegetables produced for processing pared to 9.9  percent for processed food. The are priced through contract farming arrangements divergence between the two has accelerated and are less variable, while the price of vegetables since 2017. In addition, monthly changes in the for fresh consumption is determined by market unprocessed food category exhibit greater vol- forces, resulting in higher price volatility. atility, even after controlling for trends and sea- sonality. If compared to the European Union, the Vegetables, fruits and meat push the food in- divergence in price levels between the two types dex to higher levels. Jointly, these three indices of food indices is significantly higher for Turkey constitute 45.8  percent of the food index (meat: (Figure 4A). 18.2 percent; vegetables: 18.1 percent; and fruits: 1 See Appendix 1 for details on Turkish regions (NUTS-2 classification). © 2020 The World Bank Group 5 Drivers of Food Price Inflation in Turkey Figure 4. Price Inflation and Volatility of Processed and Unprocessed Food A) Difference between processed vs. unprocessed foods’ price levels B) Price volatility 200 20 Index, 2003=100 % 15 150 10 100 5 0 50 -5 0 -10 -50 -15 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Turkey EU Processed Food Unprocessed Food Source: Central Bank of Turkey, 2020 Note: de-trended and de-seasoned Source: Cakmakli, C. et al., 2019, based on data from the Central Bank of Turkey, 2019 Figure 5. Price Dynamics of Unprocessed Food Components 1000 Indices, 2003 = 100 800 600 400 200 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Fresh Fruits and Vegetables Turkey: CPI: Fresh Fruits and Vegetables (NSA, 2003 = 100) Other Unprocessed Food Turkey: CPI: Otcher Unprocessed Food (NSA, 2003 = 100) Source: Central Bank of Turkey, 2019 9.5 percent). The lowest price levels are observed prices, in 2009 by meat prices, and in 2018–2019 for the sugar, jam, honey, chocolate, and confec- by animal products and vegetables. However, all tionery sub-index. On a  yearly basis, the major sub-item food prices went up quite sharply in the drivers of food inflation vary. In 2006–2007, food latter period due to a sharp depreciation in the ex- price volatility was driven by vegetable and fruit change rate (Figure 6). 6 © 2020 The World Bank Group I. Food Price Development – Variations Across Agricultural Products and Regions Figure 6. Dynamics of Food Price Sub-Indices 1000 Indices, 2003 = 100 800 600 400 200 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Bread, cereals Meat Milk, eggs Fruit Vegetables Sugar, jam Coffee, tea Source: Central Bank of Turkey, 2019 Prices for vegetables have been highly volatile averaged 10.8  percent, and overall food infla- when compared to other food groups. Between tion averaged 10.7  percent. For vegetables, the January 2004 and September 2019, yearly veg- highest year-on-year inflation levels were ob- etable price inflation averaged 14.0  percent served in April 2019 (96.3 percent), March 2019 while price inflation for both fruits and meat (90.2 percent), and January 2019 (80.6 percent). Figure 7. Food Price Volatility Dynamics A) Vegetables, fruit and meat sub-indices monthly changes B) Turkish and European vegetables sub-index 40 40 % % 30 30 20 20 10 10 0 0 -10 -10 -20 -20 -30 -30 -40 -40 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Vegetables Fruits Turkish Vegetables Price Index Meat CPI EU Vegetables Price Index Note: de-trended and de-seasoned. Note: de-trended and de-seasoned. Source: Central Bank of Turkey, 2020 Source: Cakmakli, C. et al., 2019, based on data from the Central Bank of Turkey, 2019 © 2020 The World Bank Group 7 Drivers of Food Price Inflation in Turkey The highest month-on-month changes in the increases in vegetable prices. The vegetable vegetable sub-index occurred in January  2017 sub-index consists of 34 products. Vegetables with (33.5 percent), January 2019 (29.7 percent), and the highest weights (2019 est.) include (weights in September  2018 (29.0  percent). For fruits, the the CPI basket are shown in parenthesis) tomatoes highest monthly changes occurred in May  2011 (0.82), potatoes (0.51), cucumbers (0.30), onions (60.7  percent), May  2006 (31.0  percent), and (0.23), and green peppers (0.19). Many vegetables May  2009 (19.2  percent). These numbers are have recorded very high annual changes, in some very high for monthly inflation levels. After 2013, cases surpassing 100 percent. These price chang- monthly volatility in fruit prices decreased while es, in return, affect the inflation rate in Turkey. For volatility in vegetable prices increased, indicat- instance, a  100-percent increase in tomato pric- ing that vegetable products are the main driver es translates to an approximate 0.815-percent of the variation observed in overall food infla- increase in the overall inflation level. Vegetable tion (Figure 7A). Vegetable prices in Turkey show prices also vary significantly from region to region a more pronounced volatility as compared to price and price dispersion increases over time. For ex- variations in the European Union (Figure 7B). The ample, in May  2019, tomato prices ranged from price swings first increased after 2009 and be- Turkish lira 4 (TL) to TL 5.4 per kilogram (kg), po- came more pronounced after 2016. tato prices ranged from TL 3.8 per kg to TL 5.1 per kg, cucumber prices ranged from TL  2.3  per  kg Fresh vegetables such as tomatoes, cucumbers, to TL  4.1  per  kg, and onion prices ranged from and dry onions have the highest weights in this TL  2.6  per kg to TL 4.4  per kg across different sub-index and therefore determine most of the regions (TURKSTAT, 2020). 8 © 2020 The World Bank Group II. Drivers of Food Price Inflation II. Drivers of Food Price Inflation Table 1. Supply and Demand Drivers Demand and supply dynamics in the agricul- of Food Prices ture sector determine the stability of food pric- es. Due to the nature of agricultural production, Supply Demand price stability requires that either demand or Research and development supply are elastic. Generally, for any given geo- Productivity levels Population growth graphic region, there are relatively few shifts in Climate change Income growth Low variance the demand function in the short run, although Environmental sustainability Dietary changes and exceptions exist stemming from speculative be- Agricultural and trade tastes havior or policy interventions. Market prices are policies (long-term) generally influenced by the supply function that Weather is shaped by long-term, less variable drivers or Seasonality of production Speculation Pests and diseases Panic or hoarding short-term dynamics due to, for example, the High variance Input costs Government trade and seasonality of production and weather patterns. Agricultural and trade inventory policies Table 1 summarizes key supply and demand driv- policies (short-term) Exports ers of food prices according to their predictabili- Imports ty, such as whether the drivers are low variance Source: Adapted from Timmer (2018) and easy to predict or high variance and difficult to predict. Low variance drivers evolve over time and affect food supply and demand in the long run. On the supply side, these center around fac- Variability in trade both on the import and export tors determining productivity in the sector, in- sides can further contribute to short-term fluctu- cluding investments in research and development ations in the supply of and demand for food. The (R&D), level of depletion of natural resources, cli- remainder of the section focuses on the key sup- mate change patterns, and an overall approach ply and demand drivers of food price formation in to agricultural and trade policies. Population and the context of Turkey’s rising food price inflation income growth as well as dietary changes and levels. tastes, such as increasing consumption of pro- teins and vegetables, determine long-term de- Long-run demand drivers mand-side pressures. On the demand side, growing population in High variance drivers of food prices are more of Turkey has been the key driver behind the a short-term nature and, hence, have the poten- long-run food demand growth. Turkey’s pop- tial to change price formation equations rapidly ulation has grown by 10.5 million people over and unexpectedly. These have mainly to do with the past ten years. Between 2012 and 2019, the inherent to agriculture production variabili- Turkey has also welcomed approximately 3.7 mil- ty due to weather conditions, seasonality, pests lion refugees, representing an increase in total and diseases, or a  human factor, such as short- population of 4.5 percent. Consequently, over- term policy interventions in the sector or trade, all demand for food has increased in Turkey. speculative behavior in the market or hoarding. Turkish consumers, following global trends, have © 2020 The World Bank Group 9 Drivers of Food Price Inflation in Turkey also increased their demand for healthier foods, Productivity levels such as fresh vegetables. Between 2010/11 and 2018/19, per capita consumption of vegetables Low productivity levels constrain food supply increased from 269 kg to 274 kg, and of fruits – growth. Despite increasing since 2010, per capita from 93 to 99 kg. food supply remains below the levels recorded in the mid-1990s, putting pressure on food prices. A second component of demand—agri-food ex- Both low agricultural productivity levels and high ports—rose by 15.9 percent between 2012 and levels of import protection contribute to such dy- 2018 and reached a total value of $17.7 billion. namics. Research shows that there is a strong link Turkey is a  net exporter of agricultural products between land productivity and food price inflation; with a strong net positive position for processed food price inflation is lower in provinces where foods (Figure 8). Between 2005 and 2011, the land productivity growth is higher (Table 2). value of agri-food exports grew by 11 percent an- nually while food price levels were relatively sta- Land productivity and the consumer price of to- ble. Between 2012 and 2019 export value growth matoes and dry onions in Turkey are negatively decreased to only 3.1  percent, likely driven by correlated (Figure 9). Provinces where productiv- currency depreciation with unprocessed exports ity is higher had lower producer prices. Differences growing faster than processed exports. However, in productivity could be associated with quality, in terms of volume, exports doubled between but the data do not allow for the identification of the 2010/11 and 2018/19 marketing years. The variations in quality of tomatoes across regions fastest growth in exports was observed for un- in Turkey. As discussed in the next section of the processed foods, such as oilseeds, pulses and ce- report, tomato, green pepper, and onion yields in reals; vegetable exports increased by 15 percent Turkey are much lower than in comparator coun- during this period. However, exports of tomatoes, tries—increases in yields can help balance vegeta- a vegetable playing an important role in food price ble price growth. inflation growth, increased only slightly between 2010/11 and 2018/19 from 1.1  million tons to Labor productivity is also low in the coun- 1.2 million tons (UN COMTRADE, 2020). try. The productivity gap between workers in Figure 8. Processed vs. Primary Food Trade, Billion USD A) Primary B) Processed 8 12 billion USD billion USD 10 6 8 4 6 2 4 0 2 -2 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Imports Exports Net Imports Exports Net Source: UN COMTRADE, 2020 10 © 2020 The World Bank Group II. Drivers of Food Price Inflation Table 2. Impact of Land Productivity agriculture and in other sectors in Turkey is on Food Price Inflation high, particularly given the level of Turkey’s eco- nomic development. As shown in Table 3, the (1) (4) ratio of labor productivity in agriculture relative -0.307*** -0.318*** to the service sector stood at only 34  percent Land productivity [0.039] [0.039] in Turkey, based on 2012–2016 averages. This is much less than in countries such as Russia Non-food consumer price 0.349** 0.338** and Ukraine as well as many European Union index (%) [0.132] [0.133] countries. Agriculture as a % of total -5.560 land area [6.343] In Turkey, agricultural producers do not always 2.191** 3.574* receive market price changes or competitive Constant [1.017] [1.988] pressures, which results in low incentives for Observations 312 312 them to invest in productivity growth. The ex- tent to which policies create disincentives for Notes: Standard errors in brackets, * p < 0.10, ** p < 0.05, *** p < 0.01. producers to change their production practices can be gauged by looking at the importance of Dependent variable: Food price inflation. Fixed-effects esti- mates using NUTS2 data MOFAL and TURKSTAT from 2005 to the mix of policy instruments used in agricul- 2016. The model includes time dummies. ture. In Turkey, the producer support estimate2 Source: World Bank (2018) as a  percentage of gross farm receipts is close to 15  percent (Figure 10), which is lower than in the European Union. But in terms of the in- strument mix, agricultural support in the country report. Such payments account for more than is skewed towards measures which may distort 62 percent of total support payments in agricul- the market—payments based on outputs and on ture. By comparison, the comparable share in the input use. These are designated as area-based European Union is 18 percent and 38 percent in payments as described in the next section of the the United States. Figure 9. Correlation Between Consumer Prices and Productivity Levels for Tomatoes and Dry Onions 7.5 9.5 LN (tomato prices) LN (onion price) 9.0 7.0 8.5 6.5 8.0 7.5 6.0 7.0 5.5 6.5 5.0 6.0 1.0 1.1 1.2 1.3 1.4 1.5 0.7 0.8 0.9 1.0 1.1 LN (tomato yields) LN (onion yields) Source: Authors 2 The producer support estimate is an indicator of the annual monetary value of gross transfers from consumers and taxpayers to agricultural producers, measured at the farm gate level, arising from policy measures that support agriculture, regardless of their nature, objectives, or impacts on farm production or income. © 2020 The World Bank Group 11 Drivers of Food Price Inflation in Turkey Table 3: Relative Labor Productivity in Service and Industry Sectors, Selected Countries Country/Region Agriculture/Service Agriculture/ Industry Per capita GDP Europe & Central Asia average 0.22 0.24 24,505 Greece 0.27 0.33 22,599 Turkey 0.34 0.36 13,249 Italy 0.51 0.63 34,135 France 0.57 0.65 41,522 Ukraine 0.59 0.67 3,034 Russian Federation 0.66 0.49 11,542 Spain 0.66 0.55 29,991 Hungary 0.74 0.79 14,090 Netherlands 0.85 0.63 50,872 Note: Based on average values, 2012–2016. Value-added are expressed in $ (2010=100). Source: WDI (2018). Figure 10. Producer Support Estimate as % of Gross Farm Receipts production on the General Services Support Expenditures3 (GSSE) (Figure 11). The main el- 25 ement is financing the development and main- % tenance of infrastructure, which accounts for approximately 75  percent of the GSSE. At the 20 same time, while expenditure for the agricultur- al knowledge and innovation system increased 15 in the last decade, its share in GSSE expenditure has remained around 5 percent during the 2016– 16.42 5.51 2018 period. 10 7.55 Another risk factor for productivity growth po- 5 9.29 tential is the depletion of natural resources, par- 4.67 ticularly in the context of climate change. Many 3.61 0 regions in Turkey suffer from high levels of soil EU Turkey USA erosion (Figure 12) and high levels of water stress Most distorting support in the PSE (Figure 13). Both have negative implications for Other support in PSE yields. For example, the correlation between yields and levels of soil erosion in Turkey (per estimates Source: OECD, 2020 in Figure 12) are -.30 for tomatoes, -.29 for on- ions, and -.21 for green peppers. There exists room for improvement in invest- Input prices ments in R&D and other public goods in agri- culture to further increase productivity growth Rising input costs have caused the producer potential. Currently, Turkey spends less than price index (PPI) of agricultural products to one percent of its aggregate value of agricultural increase faster than overall prices in Turkey. 3 The General Services Support Estimate (GSSE) is an indicator of the annual monetary value of gross transfers to services provided collectively to agriculture and arising from policy measures which support agriculture, regardless of their nature, objectives, and impacts on farm production, income, or consumption of farm products. 12 © 2020 The World Bank Group II. Drivers of Food Price Inflation Figure 11. GSSE to Agriculture Relative to the Aggregate Value of Agricultural Production 6 % 5 4 3 2 1 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Chile European Union New Zealand United States Turkey Source: OECD, 2020 Figure 12. The Spatial Pattern of Soil Erosion Across Regions, Mg/Ha per Year Soil Erosion 0–1 1–3 3–5 5–10 >10 Note: for methodology of soil erosion assessment, see Borrelli et al. (2013). https://www.nature.com/articles/s41467-017- 02142-7 Source: Authors, using Global Soil Erosion data from the Joint Research Centre of the European Commission From 2003 to 2017, the PPI of agricultural Prices of agricultural inputs have a strong pos- products increased on average 8.2  percent per itive effect on food price increases in Turkey. year as compared to the food and non-alcohol- Research by Eren et al. (2017), who utilized ic beverages index that increased by 7  percent a  panel vector autoregressive VAR model to in- (Figure 14). Monthly changes in the agricultural vestigate the impact of producer prices, quanti- PPI and the food price index are strongly cor- ty of production, and export and import quan- related. tities on consumer food prices, suggests that © 2020 The World Bank Group 13 Drivers of Food Price Inflation in Turkey Figure 13. Water Stress Assessment Across Regions4 Baseline Water Stress Low Water Stress High Water Stress Note: for more detailed methodology on water risk indicators, see Hofste et al. (2019).5 Source: Authors, using data from Hofste et al. (2019) Figure 14. Comparison of Agricultural PPI and Food Price Index A) Food CPI and agricultural PPI comparison B) Monthly changes (de-trended and de-seasoned) 400 15 % Price Index 2003=100 10 300 5 200 0 100 -5 0 -10 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 CPI- Food and non-alcoholic beverages CPI - Food Producer Price Index - Agricultural products PPI - Agriculture products Source: Central Bank of Turkey, 2017 consumer food prices are affected mostly4by period. Earlier research by Ciplak and Yucel producer prices and quantity of production sup- (2004) that applied VAR to examine the relation- plied to5the domestic market over the 1996–2016 ship between agricultural and consumer prices in 4 This data set shows the percentage of total crop production in areas facing different levels of water stress. Crop production data is overlaid with the baseline water stress indicator, a measure of demand and supply for water in a given area. 5 https://www.wri.org/publication/aqueduct-30 14 © 2020 The World Bank Group II. Drivers of Food Price Inflation Turkey over the 1994–2003 period also suggests be to increase agricultural productivity, as global that a  10-percent increase in agricultural prices practice shows. resulted in a 5.1-percent increase in food prices within six months. The Turkey Economic Monitor (October 2019) highlights that producer prices tend to be more The causality between rising prices of agricul- responsive to exchange rate movements com- tural inputs and rising food prices can be ex- pared to consumer prices. Currency deprecia- plained by growing demand for inputs and tion can lead to higher producer prices, which, their share in production costs. Between 2015 in turn, tend to be passed on to consumers to and 2017, the use of pesticides increased by recover the producers’ costs. On average, the 38.6 percent, from 39,000 tons to 54,000 tons. pass-through from producer prices to consum- The use of chemical fertilizers per hectare of land er prices in Turkey has increased in recent years increased by 24  percent during 2016–2017. In from 35 percent in 2003–2013 to 48 percent in the cost structure of tomatoes, fuel/electricity, 2013–2019. Overall, Turkey is import-dependent seedlings, fertilizer, and pesticides account for for most of the major agricultural inputs used in 48.5 percent of total production costs (Table 4). production—fertilizer, chemicals, animal feed, Interviews with farmers showed that there are fuel, and machinery. For example, between 2013 no problems in terms of availability of quality in- and 2018, fertilizer imports (HS  31) on average puts in the country (EBRD, 2018). It is rather the accounted for $1.2 billion, with an average trade issue of access that constitutes a  serious prob- deficit for fertilizer in the same period of $1.1 bil- lem for them. Input prices tend to change almost lion. Similarly, over the same period, the import immediately with exchange-rate changes and value of insecticides (HS  3808) was $313  mil- energy price fluctuations, confirming high levels lion with a trade deficit of $238 million. In 2017, of exchange-rate pass-through into the inputs Turkey imported 85 percent of its total chemical markets as discussed earlier. As input sellers are fertilizer consumption with 74 percent of total well-organized in the form of associations, they fertilizer imports being nitrogen-based fertilizers. pass on price increases to farmers, but not price Machinery and equipment in the agriculture sec- decreases. Overall, input prices globally tend to tor are also affected by exchange rate volatility grow faster than prices of agricultural and food due to their high imported input material content. products. And while collective action may help de- Finally, both currency depreciation and the rise in crease the level of input price pass-through from international oil prices fed into diesel and gasoline input dealers to farmers, a more robust strategy prices that, in turn, translated into higher expens- for farmers to deal with rising input prices would es for farmers. Table 4. Tomato Cost Structure, TL/Ton Cost element Average amount Share in total cost, % Energy (Fuel/Electricity) 19.01 10.55 Female labor 39.75 22.06 Male labor 15.22 8.45 Fertilizer cost 22.77 12.63 Seedling cost 35.91 19.93 Pesticides cost 9.69 5.38 Water price and amortization 7.98 4.43 Land rent 29.88 16.58 Total Production Cost 180.22 100.00 Source: EBRD, 2018 © 2020 The World Bank Group 15 Drivers of Food Price Inflation in Turkey Import protection policies Turkey’s agricultural markets are also among the most protected in the world. Turkey has Turkish agri-food products have historical- historically had high import tariffs on vegetables ly been the most protected group of goods in (Figure 15)12,13 The difference between the MFN the country, resulting in upward pressure on rates for vegetables imposed by Turkey and other the food supply (Table 5). Animal products6 regions around the world consistently increased have experienced the highest applied most-fa- between 2009 and 2018, when Turkey lowered its vored nation (MFN)7,8 tariff rates, ranging from applied MFN rate. For example, in 2013, Turkey’s 121.5 percent in 2010 to 28.5 percent in 2018. rate was 49 percentage points higher than the av- Vegetables9 are the second most protected cat- erage for the North American countries. While em- egory. In 2018, the applied MFN tariff rate on pirically, it is difficult to establish the causality be- vegetables accounted for 23.5  percent; for to- tween the price levels for selected food products matoes it was even higher at 48.6 percent. The and the level of import tariffs, fruits and vegetables applied MFN tariff rate for food products10 other that have high applied MFN rates also exhibit high- than for animal products and vegetables is low- er than average price levels in Turkey. As an exam- er but above MFN rates for other product (non- ple, tomatoes, onions, cucumbers, and potatoes, food) groups. Similar dynamics are observed in which drive price inflation for vegetables, have an the Turkish fruit markets. Ad valorem tariff11 rates average applied MFN rate of 36.8 percent, as com- are higher for fruit products compared to vege- pared to an average rate of 25.7 percent applied to tables. The highest applied tariff is for bananas, all vegetables (HS group 17). While Turkey is a net at 146 percent. Most other major fruits such as exporter of many agricultural products, import re- apples, melons, citrus fruits, and grapes have tar- strictions make the supply curve less elastic, lead- iff levels ranging from 54 percent to 86 percent. ing to more variable price responses. Furthermore, Nuts have a  relatively lower tariff rate among the price impact that stems from seasonal short- these products. ages of various agricultural products, such as Table 5. Product Groups with the Highest MFN Tariff Rates Applied by Turkey, 2005–2018 Product group 2005 2010 2013 2018 Animals 53.7 121.5 56.2 28.5 Vegetables 22.5 26.5 51.5 23.5 Food products 15.7 15.1 13.4 17.3 Footwear 11.6 11.0 11.4 12.0 All others <10 <10 <10 <10 Source: WITS, 2019 HS 01-05. 6 The applied MFN tariff is a normal non-discriminatory tariff charged by one World Trade Organization (WTO) member on 7 imports from other WTO members (excludes preferential tariffs under free trade agreements and other schemes or tariffs charged inside quotas). Most-favored-nation treatment (GATT Article I, GATS Article II and TRIPS Article 4), the principle of not discriminating 8 between one’s trading partners. HS 06-15. 9 10 HS 16-24. 11 Ad valorem tariff measures tariff rate charged as percentage of the price. 12 Data for 2012 and 2014 is not available for Turkey. 13 The Europe and Central Asian region includes Turkey. 16 © 2020 The World Bank Group II. Drivers of Food Price Inflation Figure 15. MFN Rates Applied to Vegetable Imports 60 50 40 30 20 10 0 2005 2006 2007 2008 2009 2010 2011 2013 2015 2016 2017 2018 Turkey East Asia & Pacific Europe & Central Asia Latin America & Caribbean Middle East & North Africa North America Source: WITS, 2019 Figure 16. Nominal Protection Coefficients vegetables or fruits, may be exacerbated by re- Comparison strictions on the import side. 1,5 Price Index 2003=100 Over the long run, import protection may have significant impact on the efficiency and competi- tiveness of the sector, as it shields domestic pro- ducers from the need to respond to competitive 1,0 pressures. The extent to which domestic producers are protected from international price fluctuations can be measured by a  nominal protection coeffi- cient (NPC) that is measured as the ratio between 0,5 the average price received by producers (measured at the farm gate), including net payments per unit of current output, and the border price (measured 0 at the farm gate). In Turkey, was 1.12 (2018 est.), 2014 2015 2016 2017 2018 which suggests that Turkish farmers, overall, re- Turkey EU New Zealand ceived prices that were 12 percent above interna- Kazakhstan Argentina tional market levels. While the NPC in Turkey has been gradually decreasing over time, it remains Source: OECD, 2020 higher than in comparator countries (Figure 16). © 2020 The World Bank Group 17 Drivers of Food Price Inflation in Turkey III. Regional Price Formation and Transmission: The Case of Table Tomatoes, Green Peppers and Onions Vegetable price index levels have been much Tomato price variation can be significant across higher than all the other food indices, particularly the regions of Turkey, with price levels and vol- after mid-2018, and have been driving the over- atility negatively correlated with production lev- all food price index up. To understand the driv- els. Tomato prices vary from region to region, rang- ers of price inflation for vegetables, an in-depth ing from 4 TL per kg to 5.1 TL per kg (May 2018– analysis of the price formation and transmission April  2019 average) (Figure 1716). Some of these for table tomatoes, green peppers, and dry on- price differences can be attributed to transportation ions14 (the most produced fresh vegetables in costs, however, a  visual analysis of prices across Turkey) has been undertaken across country re- regions does not support the notion that transport gions to identify inefficiencies in these selected prices are the sole reason for regional price differ- value chains, and how these have contributed to ences. Overall, prices increase from South to North rising price levels. and the peak values are observed in the Northeast and Northwest. The opposite pattern is observed Overview of Price Dynamics for Table when we look at tomato production, with South Tomatoes, Green Peppers and Dry Onions and Southwestern cities producing the bulk of to- matoes. A negative correlation is observed between Table tomatoes production and price levels (-0.17) and between production levels and price volatility (-0.30). It is Between 2005 and 2019 tomato prices have also informative to look at the price changes at the increased fivefold with growing monthly fluc- highest point in the last year (between May 2018 tuations in recent years (TURKSTAT, 2020). and April  2019) which occurred in January  2019. Tomato prices have increased from TL  1.5  per Overall, prices increased around 51 percent during kg in January 2005 to TL  4.8  per kg in May that month, but regional price increases were be- 2019, with a peak price of TL 6.7 in April 2019. tween 36.9  percent and 65.3  percent, with the Overall, the price growth has accelerated since highest increase in the tomato deficit regions of November 2017. Monthly fluctuations measuring Erzurum, Izmir and Gaziantep. year-to-year changes have also been significant for tomatoes during the analyzed period, ranging Green peppers from -50 to 117  percent,15 and have increased since November  2017 with an average monthly Green pepper prices have been increasing at change of 54 percent (between November 2017 a  slower rate compared to tomato prices, but and May  2019), as compared to 13  percent be- have been more volatile. Since 2005 green tween June 2008 and October 2017. pepper prices increased from 1.34  TL per kg in 14 The report focuses on table tomatoes, green peppers, and dry onions for fresh consumption. Jointly these vegetables account for 50 percent (or 12.8 million tons) of fresh vegetable production (2019 est.) in Turkey. 15 Not including the price spike in October 2010. 16 See Appendix 8 for additional details. 18 © 2020 The World Bank Group III. Regional Price Formation and Transmission: The Case of Table Tomatoes, Green Peppers and Onions Figure 17. Regional Differences in Tomato Prices Wholesale Markets <500k 500k–1m 5.23 5.00 1m–10m 5.11 5.06 4.89 4.53 5.14 10m–25m 5.13 25m–100m 4.90 4.76 4.78 4.80 >100m 4.51 4.16 4.42 4.70 Tomato Production (tons) 4.87 4.93 <200k 4.29 4.50 4.61 4.54 3.95 200k–400k 4.70 4.43 4.11 400k–600k 600k–800k 800k–1m >1m Highways 0 250 500 km Note: Numbers on the map reflect average tomato prices during May 2018–April 2019. Source: Authors, based on statistics from TURKSTAT January  2005 to 4.57  TL per kg in May  2019. per kg in Malatya to 7.31  TL per kg in Istanbul. However, prices experienced a  significant spike Unlike in the case of tomatoes, there is no sig- between January and April 2019, reaching on av- nificant correlation between production and price erage 11  TL per kg. In general, price growth has levels; however, there is a  strong positive cor- accelerated since about January 2017. relation (0.45) between consumption estimates and price levels. Just as in the case of tomatoes, There is also a significant price variation in green analysis of the price changes during March 2019, peppers across regions (Figure 18). Between when the highest prices were observed, show May 2018 and April 2019, average prices for green that price changes are uneven across the regions. peppers across the regions ranged from 5.78  TL In March  2019, prices on average increased by Figure 18. Regional Differences in Green Pepper Prices (Sivri Variety) Pepper Prices 7.1 7.2 7.1 6.1–6.5 8.1 6.8 6.7 7 6.5–7.0 6.7 7.3 7.8 7.8 7 7.0–7.5 6.7 7.5–8.0 6.4 6.6 6.1 6.1 8 8.0–8.1 6.5 6.6 6.2 7.1 6.5 7.2 Highways 6.7 6.7 0 250 500 km Note: Numbers on the map reflect average green pepper prices during May 2018–April 2019. Source: Authors, based on statistics from TURKSTAT © 2020 The World Bank Group 19 Drivers of Food Price Inflation in Turkey 32.3 percent, ranging from a 20-percent increase observed in Ağrı. Other regions in Eastern Anatolia in Hatay to a 48-percent increase in Malatya. also exhibited relatively lower increases. Dry onions Between 2005 and 2018, price dispersion17 across the regions has steadily increased for all three Dry onion prices have increased drastically since vegetables—from 0.18 to 0.61 for green peppers, June 2018. Between January 2005 and May 2019, 0.14 to 0.43 for tomatoes, and 0.05 to 0.34 for dry onion prices increased from TL 0.4 to TL 3.3 per kg. onions. Increasing regional food-price dispersion The peak in price levels was observed in April 2019 points to a deterioration in regional integration or at TL  5.9  per kg. The fluctuation in onion prices market segmentation stemming from inefficiencies significantly increased in 2018 with overall month- in food distribution supply chains, as is discussed in ly changes ranging from 50 percent to 80 percent. more detail in this section. Policies and strategies Overall, onion prices are much less volatile than aimed at linking farmers to local and national mar- those of table tomatoes and green peppers. kets and facilitating access to storage and distribu- tion systems may contribute to a reduction in food Regional variability in prices also exists for dry price dispersion across the regions of Turkey. onions, as in the case of tomatoes and green peppers (Figure 19). Between May  2018 and An Overview of the Tomato, Green Pepper April  2019, onion prices ranged from TL  2.6 to and Onion Sub-Sectors TL  4.4  per kg. Prices were higher in Western and Northeastern Anatolia. There is a  strong negative Tomatoes (For Fresh Consumption correlation between production volumes and prices and Processing) and between production volumes and price vola- tility, -0.38 and -0.44, respectively. In June  2018, Turkey is the fourth largest producer of tomatoes when the highest average price increase of close to in the world. In 2018, its production accounted 83 percent was observed, regional price increases for 12.2 million tons (Table 6), or 5 percent of total ranged from 61 percent to 105 percent. The largest world production, behind only China (61.6  million price increase was observed in the Balıkesir region tons), India (19.4 million tons), and the United States (104.6 percent), followed by the Aydın and Muğla (12.6 million tons). While the share of Turkey’s to- regions. The lowest price increase of 61 percent was mato production in total world production remained Figure 19. Regional Differences in Dry Onion Prices Onion Production (tons) 7.08 8.07 7.17 7.08 6.78 6.72 6.99 <50k 7.26 6.66 50k–100k 7.83 7.76 7.01 6.66 100k–150k 6.43 6.61 6.14 150k–200k 7.99 6.10 200k–250k 7.15 6.51 6.47 6.60 6.23 7.22 >250k 6.70 6.65 Highways 0 250 500 km Note: Numbers on the map reflect average dry onion prices during May 2018–April 2019. Source: Authors, based on statistics from TURKSTAT Price dispersion is calculated using the standard deviation of prices for each crop across 26 regions. 17 20 © 2020 The World Bank Group III. Regional Price Formation and Transmission: The Case of Table Tomatoes, Green Peppers and Onions Table 6. Global Tomato Production (For Fresh Consumption and Processing), Tons Countries 2009 % of total (2009) 2018 % of total (2018) China 45,365,542 23% 61,631,581 25% India 11,148,800 6% 19,377,000 8% USA 15,457,480 8% 12,612,139 5% Turkey 10,745,572 5% 12,150,000 5% Egypt 10,278,539 5% 6,624,733 3% Other 107,678,210 53% 131,492,588 54% World Total 200,674,143 100% 243,888,041 100% Source: FAOSTAT stable, in absolute terms, tomato production has have been increasing in recent years. An average increased by 13.1  percent, from 10.7  million tons farm size for vegetable producers, including toma- to 12.3 million tons. About 70 to 75 percent of to- toes, is 1 hectare (EBRD, 2018). tal tomato production in Turkey is consumed fresh and 25 to 30 percent of production is processed.18 Table tomatoes in Turkey are produced both in Approximately 80  percent of tomatoes produced open fields and in greenhouses. In 2019, green- for processing are used to produce tomato paste, house production of table tomatoes accounted for 15  percent is used for canned tomatoes, and the 4.1 million tons (or 46.1 percent of total table to- rest is used for ketchup, tomato juice and other mato production). The share of table tomatoes pro- products (Fidan, 2002; Sarisaçli, 2005). duced in greenhouses has increased by 25 percent between 2009 and 2019. Greenhouse production Tomato production (for fresh consumption and activities are mainly clustered in the coastal regions. processing) accounted for 34 percent (12.8 mil- Antalya is the largest greenhouse producer re- lion tons19) of all fresh vegetable production in gion—greenhouse production accounts for 94 per- the country. Table tomato production accounts cent of total production (2019  est.)—followed by for 8.8  million tons. While climatically tomatoes Zonguldak (56  percent) and Adana (41  percent). can be grown anywhere in Turkey, commercial Nearly all open-field table tomato production goes production is concentrated in the coastal regions20 to domestic markets and consumption; greenhouse of Adana and Antalya (Figure 20). Jointly they pro- production is mainly targeted towards exports. duce over 4 million tons of tomatoes (2019 est.). Tomato production usually starts with seed sowing Table tomato yields have been consistently in- and continues with sapling plantation. The most creasing in Turkey but remain significantly be- suitable time for seed sowing is the beginning of hind those observed in the United States and the spring with a  high concentration in March and European Union. The average table tomato yield April. Harvesting takes place around June and July. in Turkey is 5.3 tons/decare (2019 est.), but var- Alternative greenhouse production methods for ies greatly across the regions (Figure 21), ranging year-round production are developing, such as the from 2.3 tons/decare in the marginally producing high-altitude greenhouse production in Antalya, region of Mardin to 12.4  tons/decare in Antalya. Erdemli, and Mersin. Vegetable producers in Turkey Yields for tomatoes grown in greenhouses tend are usually small and medium-scale farmers, albe- to be higher than those grown in open fields. it large-scale investments in vegetable production The correlation between tomato yields and share This report focuses on tomatoes for fresh consumption. 18 2019 est. 19 Throughout the report the regions refer to statistical regions defined in accordance with the European Union’s Nomenclature 20 of territorial units for statistics (NUTS). © 2020 The World Bank Group 21 Drivers of Food Price Inflation in Turkey Figure 20. Regional Table Tomato Production, Production and Yields Wholesale Markets <500k 500k–1m 408.6 371.1 326.1 382.2 1m–10m 425.0 698.3 252.3 10m–25m 679.8 386.5 730.2 398.2 25m–100m 706.1 356.9 >100m 494.4 488.8 397.8 Tomato Production (tons) 723.2 493.9 564.1 <200k 900.9 394.8 291.6 444.6 228.2 200k–400k 1241.4 928.3 400k–600k 600k–800k 800k–1m >1m 0 250 500 km Highways Note: Numbers on the map reflect average regional yields, kg/decare (2019 est.). Source: Authors, based on statistics from TURKSTAT Figure 21. Comparison of Turkish Tomato Yields with the Yields in Comparator Countries 35 tons/da 30 25 20 15 10 7.2 6.7 6.6 6.9 7.1 7.2 7.3 7.5 6.2 5 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 Turkey China India US EU average Source: FAOSTAT of greenhouse production across the regions is in leading tomato producing countries, such as 0.64. Overall, tomato yields have been increasing the United States and countries in the European over time, but remain lower than those observed Union.21 Specifically, according to FAOSTAT,22 21 Some of the differences in yields may be attributable to differences in tomato varieties. 22 There is a certain discrepancy between the FAOSTAT and TurkStat numbers on tomato yields. This report uses the TurkStat estimates; FAOSTAT estimates are used solely for the purpose of international comparisons. 22 © 2020 The World Bank Group III. Regional Price Formation and Transmission: The Case of Table Tomatoes, Green Peppers and Onions Table 7. Tomato Consumption in Turkey Turkey’s yields (7.6  tons/decare, 2018 est.) are (Fresh and Processed)25 higher than those for China (6.5 tons/decare) and India (2.7  tons/decare), two of the largest world Year Per capita (kg) Total (MT) producers of tomatoes, but are significantly lower 2014 119.5 9,285,983 than yields in the United States (10.7 tons/decare) 2015 118.6 9,340,969 or yields in the largest tomato producing European Union countries (an average of 34.3 tons/decare23). 2016 116.3 9,284,769 2017 116.9 9,443,060 Several factors contribute to relatively low toma- 2018 109.9 9,013,786 to yields in Turkey. One of the reasons for lower Source: Turkstat, 2020 yields, as compared to the European Union, is the limited adoption of improved production techniques by growers. As the bulk of production occurs on small farms, sowing or planting seedlings, mainte- amounted to $322.4  million, which constitutes nance, and harvesting are generally done by hand, 31  percent of the value of total fresh vegetable and mechanization levels remain much lower than exports in the country (UN Comtrade, 2020). in the European Union. Country-wide producer or- However, both in absolute and relative terms, ganizations representing the interests of producers the value of fresh tomato exports in US  dollar in the subsector are largely nonexistent. At times, equivalent has been decreasing, primarily driv- the lack of unions and organized production along en by currency depreciation and the loss of the with limited stock capacity lead to price fluctuations Russian market post-2015 (Figure 22). In terms with producers having to sell below production of volume, exports of tomatoes (fresh and pro- costs (Erturk & Çirka, 2015). While drip irrigation cessed) remained relatively stable in the last five and fertigation methods are the norms in green- years (see Appendix 2), between the 2013/14 house tomato production, in open field production and 2018/19 marketing years, with Turkey ex- row irrigation is the most common irrigation meth- porting about 10  percent of its annual produc- od and the use of fertilizers is still based on farmer tion volume. In 2016–2018, the largest importers habits rather than soil analysis (Abak, 2016). of Turkish fresh tomatoes have been Romania, Belarus, Ukraine, and Saudi Arabia. In 2018, there Tomato consumption has been relatively sta- was some recovery in exports to Russia and these ble in Turkey over recent years with a  slight exports accounted for $30.5 million (or 10 percent decrease in 2018 (Table 7). In 2018, Turkey per of total fresh tomato export value). In compari- capita consumption amounted to 109.9  kg. This son, in 2015, prior to the Russian ban on imports is a slight decrease from the preceding years: be- of Turkish tomatoes, Turkey exported $258 million tween 2014 and 2017, average consumption was worth of tomatoes to Russia (or 71 percent of its close to 118 kg per capita. total fresh tomato export value). Turkey is a  net exporter of tomatoes, realizing Green peppers no sizeable imports. In 2019, Turkey exported $303  million worth of tomatoes and ranked as Turkey is the third largest producer of green pep- the fifth largest tomato exporter24 in the world pers in the world. In 2018, its production account- after Mexico ($2.16  billion), Spain ($1.03  billion), ed for 2.5 million tons (Table 8), or 5 percent of total Canada ($379 million), and Belgium ($306 million) world production, behind only China (18.2  million (UN  Comtrade, 2020). Between 2014 and 2018, tons) and Mexico (3.4 million tons). While the share the average value of Turkish fresh tomato exports of Turkey’s green pepper production in total world Average calculated for Portugal, Spain, France, Germany, Austria, Ireland, the United Kingdom, Denmark, Finland, Sweden, 23 Belgium and the Netherlands, as the largest tomato producers in the European Union. Historically, Turkey has been a top ten world exporter of table tomatoes. 24 There are no separate statistics available on fresh tomato consumption in Turkey. 25 © 2020 The World Bank Group 23 Drivers of Food Price Inflation in Turkey Figure 22. Fresh Tomato Exports Compared to Total Fresh Vegetable Exports 1200 million 1000 800 600 400 200 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 Tomato exports Fresh vegetables exports Source: UN Comtrade, 2020. Table 8. Global Chilies and Pepper Production, Tons Countries 2009 % of total (2009) 2018 % of total (2018) China 14,520,301 34% 18,214,018 33% Mexico 1,941,564 4% 3,379,289 6% Turkey 1,837,003 4% 2,554,974 5% Spain 932,191 2% 1,275,457 2% Egypt 792,836 2% 713,752 1% Other 23,256,601 54% 28,848,004 53% World Total 43,280,496 100% 54,985,494 100% Source: FAOSTAT, 2020 production increased by one percentage point be- green peppers (34 percent), bell peppers (14 per- tween 2009 and 2018, in absolute terms green cent), and banana peppers (5 percent) comprise the pepper production increased by 39.1 percent from rest of the production and are consumed fresh. The 1.8 to 2.5 million tons. About 55 percent of total largest production of green peppers in the country green pepper production in Turkey is consumed is concentrated in the regions of Adana, Antalya, fresh and 45  percent of production is processed. Balikesir, Bursa, and Manisa (Figure 23). Jointly they Approximately 80  percent of green peppers pro- produce 1.7 million tons of green peppers or 65 per- duced for processing is used to produce pepper cent of total green pepper production in the coun- paste and 20 percent is used for dried spices. try. Green pepper seedlings are planted towards the end of April. Harvesting starts at the end of June In 2019, Turkey’s green pepper production ac- and early July and continues until mid-October. counted for 5  percent (2.6  million tons) of all vegetable production in the country. Among the Just as in the case of tomatoes, green peppers four types of green peppers produced in Turkey, are produced both in open fields and in green- capia peppers make up 47 percent of total produc- houses, however, greenhouse production re- tion and are produced for processing whereas long mains limited. Currently, greenhouse production 24 © 2020 The World Bank Group III. Regional Price Formation and Transmission: The Case of Table Tomatoes, Green Peppers and Onions Figure 23. Regional Green Pepper Production and Yields (All Pepper Varieties) Pepper Production (tons) 1489.9 1358.8 1242.0 2354.0 <100k 1448.6 2923.0 999.8 1521.4 100k–200k 3070.0 3637.0 1205.1 2320.3 200k–300k 1163.6 2901.0 1992.2 300k–400k 1243.9 1591.0 2672.3 400k–5m 2595.5 2453.7 1485.3 2062.0 2035.9 2162.4 >500m 7247.8 5822.7 Highways 0 250 500 km Note: Numbers on the map reflect average regional yields, kg/decare (2019 est.). Source: Authors, based on statistics from TURKSTAT of green peppers accounts for 749,000  tons yields in Turkey (3.1  tons/decare, 2018  est.) are (or 29 percent of total green pepper production). higher than those in China (2.6 tons/decare) and Of the four types of green peppers, greenhouse Mexico (2.4 tons/decare), two of the largest world production of long green peppers accounts for producers of green peppers, but are lower than 41 percent of total production. The share of long yields in the United States (3.6  tons/decare) and green peppers produced in greenhouses has in- significantly lower than yields in the largest green creased by 6.9 percent (or 104,000 tons) between pepper producing European Union countries (an 2009 and 2019. Greenhouse long green pepper average 12.1 tons/decare28). production activities are mainly clustered in the coastal regions. Adana is the largest greenhouse Green pepper consumption trended slightly up- producer region and accounts for 70  percent of ward in Turkey over recent years. The consump- total long green pepper production (2019  est.), tion of green peppers has increased from 23.4 kg followed by Antalya (56  percent), and Zonguldak per capita in 2014 to 25.3 kg per capita in 2018 (22 percent). (Table 9). Green pepper yields remain significantly behind Turkey is a net exporter of green peppers, and those observed in the European Union. Average has no sizeable imports. In 2019, Turkey ex- green pepper yield (all varieties) in Turkey is ported $124 million worth of green peppers and 2.3 tons/decare (2019 est.) but varies across the ranked as the fifth largest green pepper export- regions (Figure 23), ranging from 0.9 tons/decare er29 in the world after Mexico ($1.37  billion), in the marginally producing region of Trabzon to Spain ($1.32  billion), Canada ($439  million), and 7.2  tons/decare in Antalya. Overall, green pep- the United States ($252  million) (UN  Comtrade per yields have been increasing over time, but 2020). Between 2014 and 2018, the average ex- remain lower than those observed in comparator port value of Turkish green peppers (HS070960) countries26 (Figure 24). According to FAOSTAT,27 amounted to $92.5  million, which constitutes Some of the differences in yields may be attributable to differences in green pepper varieties. 26 There is a certain discrepancy between the FAOSTAT and TurkStat numbers on green pepper yields. This report uses the 27 TurkStat estimates; FAOSTAT estimates are used solely for the purpose of international comparisons. Average calculated for Spain, the Netherlands, Italy, Romania, Greece, Hungary, Bulgaria, France, the United Kingdom and 28 Belgium. Historically, Turkey has been a top ten world exporter of green peppers. 29 © 2020 The World Bank Group 25 Drivers of Food Price Inflation in Turkey Figure 24. Dynamics of Regional Green Pepper Yields (All Pepper Varieties) 15 12 9 6 2.7 2.8 2.9 3.0 3.0 3.0 3.0 3.0 3.1 3 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 Turkey China Mexico US EU average Source: FAOSTAT Table 9. Green Pepper Consumption in Turkey (Fresh and Processed)30 2016, Russian imports plummeted to $2.2 million (2.46  percent) and in 2017, they fell further to just $760,000 (0.8 percent). Year Per capita (kg) Total (MT) 2014 23.4 1,818,086 Dry onions 2015 23.1 1,817,878 2016 23.4 1,865,358 Turkey is the sixth largest producer of dry onions 2017 26.2 2,113,574 in the world. In 2018, its production accounted for 1.9 million tons (Table 10), or 2 percent of to- 2018 25.3 2,072,161 tal world production, behind China (24.8  million Source: Turkstat tons), India (22.1  million tons), the United States (3.3. million tons), Egypt (2.9  million tons), and Iran (2.4 million tons). While the share of Turkey’s 9  percent of the value of total fresh vegetable onion production in total world production re- exports in the country, and has been increasing mained stable, in absolute terms onion produc- since 2010 (Figure 25) (UN Comtrade, 2020). In tion has slightly increased from 1.8 million tons to terms of volume, exports of green peppers have 1.9 million tons. The area used in onion production also been increasing over the last five years except has been consistently decreasing over time, from for a  slight decrease in 2017 (see Appendix 2). 788,000 decares in 2004 to 614,000 in 2019. In 2016–2018, the largest importers of Turkish green peppers have been Germany, Romania, In 2019, dry onion production accounted for the Netherlands and Bulgaria. In 2018, there 7.1 percent (2.2 million tons) of all vegetable pro- was some recovery in exports to Russia, reach- duction in the country. Dry onion production is con- ing $11.3  million (or 9.6  percent of total green centrated in the regions of Ankara, Samsun, Bursa, pepper export value). In 2015 Turkey exported Adana, and Hatay (Figure 26). Jointly they produce $11.1 million worth of green peppers to Russia (or nearly 1.8 million tons of onions (2019 est.). Onions 14 percent of total green pepper export value). In are produced in open fields in Turkey through three 30 There are no separate statistics available on fresh green pepper consumption in Turkey. 26 © 2020 The World Bank Group III. Regional Price Formation and Transmission: The Case of Table Tomatoes, Green Peppers and Onions Figure 25. Share of Green Pepper Exports in Total Fresh Vegetables Exports 1200 million 1000 800 600 400 200 96 118 69 78 75 82 80 78 90 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 Pepper exports Fresh vegetables exports Source: UN Comtrade, 2020 Table 10. Global Dry Onion Production, Tons Countries 2009 % of total (2009) 2018 % of total (2018) China 21,046,969 22% 24,775,344 20% India 12,158,800 13% 22,071,000 18% USA 3,429,100 4% 3,284,420 3% Egypt 2,128,580 2% 2,958,324 2% Iran 1,529,996 2% 2,406,718 2% Turkey 1,849,582 2% 1,930,695 2% Other 52,958,262 55% 63,750,477 52% World Total 95,485,707 100% 121,549,161 100% Source: FAOSTAT Figure 26. Regional Dry Onion Production and Yields Onion Production (tons) 1854.4 1705.4 758.9 1663.1 1190.0 <50k 2531.7 3393.6 0 50k–100k 1381.7 3916.9 4831.5 1280.7 100k–150k 3846.1 2262.7 1493.9 150k–200k 3351.4 2327.8 2106.7 200k–250k 3188.8 2576.4 1763.8 2640.0 3240.2 2780.2 >250k 1346.3 4404.1 Highways 0 250 500 km Note: Numbers on the map reflect average regional yields, kg/decare (2019 est.). Source: Authors, based on statistics from TURKSTAT © 2020 The World Bank Group 27 Drivers of Food Price Inflation in Turkey different methods: direct seed sowing, seed sowing (2.4  tons/decare) and India (1.9  tons/decare), for shallots and re-sowing these and saplings. Seed two of the largest world producers of onions, sowing produces yields in 5 to 6 months and is the but are lower than the United States (6.9  tons/ most common while shallot production (producing decare), Iran (4.3 tons/decare), and in the larg- shallots in the first year then re-sowing the shallots) est onion producing European Union countries and saplings are also practiced. The most suitable (4.2 tons/decare32) (Figure 27). Figure 27. Dynamics of Regional Dry Onion Yields33 8 tons/da 7 6 5 4 3.6 3.4 3.5 3.5 3.5 3.6 3.3 3.3 3 2.7 2 1 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 Turkey China India Iran US EU average Source: FAOSTAT, 2018 season for seed sowing is during autumn in general. Onion consumption has been increasing in Turkey For early maturing types, seed sowing begins from over the past few years, although there was mid-August and ends in mid-October. Harvesting a  slight decrease in 2018 (Table 11). Between differs according to the production method and 2014 and 2017, dry onion consumption gradually place of production, but direct seed sowing is har- vested within 5 to 6 months of sowing. Table 11. Dry Onion Consumption in Turkey Dry onion yields in Turkey are significantly Year Per capita (kg) Total (MT) lower than those in many of the large onion 2014 18.19 1,413,481 producing countries. The average onion yield in Turkey is 2.5 tons/decare (2019 est.) but var- 2015 19.49 1,534,975 ies across the regions (Figure 26), ranging from 2016 20.70 1,651,785 0.8  tons/decare in the marginally producing re- 2017 22.37 1,807,980 gion of Zonguldak to 4.8 tons/decare in Ankara. 2018 21.75 1,783,426 According to FAOSTAT,31 Turkey yields (3.6 tons/ Source: Turkstat decare, 2018 est.) are higher than those for China 31 There is a certain discrepancy between the FAOSTAT and TurkStat numbers on onion yields. This report uses the TURKSTAT estimates; FAOSTAT estimates are used solely for the purpose of international comparisons. 32 Average calculated for France, Germany, Greece, Italy, the Netherlands, Spain,and the United Kingdom. 33 Some of the difference in yields may be attributable to differences in onion varieties. 28 © 2020 The World Bank Group III. Regional Price Formation and Transmission: The Case of Table Tomatoes, Green Peppers and Onions increased from 18.19 to 22.37 kg per capita. In government lowered the import tariff on onions 2018, it slightly decreased, if compared to 2017, from the usual 49.5  percent to 0  percent, which to 21.75 kg per capita. resulted in a surge of onion imports that year. Turkey is a net exporter of dry onions. Between Agricultural and trade policy environment 2014 and 2018, the average value of Turkish onion for tomatoes, green peppers and onions (HS 70310) exports were equal to $29.8  million, which on average accounted for 2.4 percent of to- The agricultural support programs for crop pro- tal fresh vegetable exports over the same period duction can be clustered into four categories: (a) (Figure 28). In 2019, Turkey ranked as the ninth area-based payments, (b) biological and biotechni- largest dry onion exporter in the world after India cal support payments, (c) deficiency/compensato- ($367  million), Mexico ($356  million), the United ry payments, and (d) other agricultural subsidies. States ($288  million), Egypt ($244  million), Spain Area-based payments cover diesel, fertilizer, soil ($213 million), New Zealand ($115 million), Poland analysis, organic agriculture, good agricultural prac- ($104 million), and Pakistan ($67 million). Between tices, small-scale family farmer supports payments, 2010 and 2018, the largest importers of Turkish and specific subsidies for hazelnut production and onions were Russia (an average of 47 percent of olive gardens rehabilitation. Biological and biotech- total exports or $10.9 million) and Iraq (23 percent nical support payments aim to reduce the use of or $5.4 million). Georgia is ranked the third or the chemicals in crop production and to protect human fourth largest importer depending on the year and health and maintain the natural balance by dissem- accounted for about 7  percent of total exports ination of alternative techniques instead of chemi- (or $1.5 million) (UN COMTRADE, 2020). Turkey’s cals. Deficiency/compensatory payments are used imports of onions tend to be low (an average of to balance the supply of products. Compensatory $45,000 between 2014 and 2018), however, in payments are used to encourage farmers to pro- 2019, the country imported $33  million worth of duce alternative products by avoiding the produc- onions due to the lower than usual production tion of surplus crops, while deficiency payments are resulting from unfavorable weather conditions, used to encourage the production of supply defi- shrinkage in the production area, and product loss cit crops. Farmers also receive other agricultural due to disease and improper storage techniques. subsidies including certified seed and sapling sup- To counteract rising onion prices in 2019, the port payments, CATAK (Environmentally Protected Figure 28. Share of Dry Onion Exports in Total Fresh Vegetable Exports 1200 million 1000 800 600 400 200 16 22 21 27 37 18 13 41 17 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 Onions exports Fresh vegetables exports Source: FAOSTAT, 2018 © 2020 The World Bank Group 29 Drivers of Food Price Inflation in Turkey Agricultural Land) payments, agricultural extension considered area-based payments. The aim of or- and consultancy services support, and R&D project ganic and good agricultural practices support is support. In addition, since 2017, the Ministry of to expand sustainable agricultural practices while Agriculture and Forestry (MOAF) has implemented improving traceability and food safety. Farmers an agricultural basin-based model that focuses on must be registered with the FRS and the Organic 21 crops34 that are eligible for four different types Agriculture Information System (OAIS) to be eligi- of subsidies.35 The program aims to establish an ef- ble to receive payments as of 2019. Eligible to- ficient agricultural inventory by determining agricul- mato, green pepper, and onion producers receive tural basins, planning production based on demand, organic agricultural support at the highest rates supporting basin-based production, and providing for fruits and vegetables. efficient and rational usage of supports. Ziraat Bank is the main financial institution that provides sup- In addition to area-based payments, tomato, port payments to farmers. The total agricultural green pepper, and onion farmers can receive support program budget of MOAF was increased to biological and biotechnical support payments. TL 22 billion in 2020 from TL 12.9 billion in 2017. These support payments aim at scaling up the From 46 percent to 51 percent of each year’s bud- biological and biotechnical ways of production get was allocated for crop production supports be- and decreasing chemical residues stemming tween 2017 and 2020. from the use of pesticides for improved human health. Support payments differ in open fields and For tomatoes, green peppers, and onions, state greenhouses. A  separate Greenhouse Cultivation support has remained relatively consistent Registry System keeps track of greenhouse pro- in terms of composition since 2015, while the ducers, and greenhouse tomato and green pep- amounts of outlays have been slightly increas- per farmers must be registered with this system ing for most of the support types (Table 12). to benefit from the subsidies. Open field producers Tomatoes, green peppers, and onions benefit from must be registered for the regular FRS. some of the subsidies under the area-based pay- ments, biological and biotechnical support pay- Tomato, green pepper, and onion producers ments and other agricultural subsidies. However, are also eligible for Farm Accounting Database deficiency/compensatory payments are not used System Participation support. The payments are for vegetables. Under the area-based payments based on participation and aim to collect detailed system, tomato, pepper, and dry onion producers accounting information from farmers to monitor can benefit from six different payments, three of the country’s overall agricultural performance. which cover the cost of inputs such as diesel, fer- Farmers must be registered in the FRS or any oth- tilizer, and soil analysis. These subsidies constitute er relevant administrative registry of the Ministry the primary form of support for vegetable produc- in order to be eligible. Participation in the system tion in Turkey and cover up to 60 percent of the requires the farmers to periodically share account- cost of production expenses. To be eligible to re- ing information during the year through surveys. ceive the payments, farmers must be registered in Farmers registered in the database receive pay- the Farmer Registry System (FRS). In addition, fruit ments the year after their compliance is verified and vegetable producers who have less than 5 de- by the responsible unit in the Ministry. Payments cares of agricultural land can benefit from small- have been increasing since 2014. In 2019, eligible scale family farmer support. farmers received 600 TL per year. Farmers can apply for organic agriculture and Agricultural Extension and Consultancy Services good agricultural practices payments for toma- support is another subsidy available for tomato, toes, green peppers, and onions which are also green pepper, and onion producers. This type of 34 Wheat, barley, rye, paddy, corn, triticale, oats, lentils, chickpeas, haricot beans, cotton, soy, oily sunflower, canola, safflower, tea, hazelnut, olive oil, potato, onion (dry) and forage crops. 35 Diesel-fertilizer payments, certified seed use, deficiency payments, and forage crops payments. 30 © 2020 The World Bank Group III. Regional Price Formation and Transmission: The Case of Table Tomatoes, Green Peppers and Onions Table 12. Summary of State Support Payments for Tomatoes, Green Peppers, and Onions Support Programs Unit 2015 2016 2017 2018 2019 I. Area-Based Payments a. Soil Analysis Support TL/analysis 125 0 40 40 40 b. Diesel Support* TL/ha 33–79 110 51.9–373.6 68.1–454 80–660 c. Fertilizer Support* TL/ha 47.5–82.5 40 40 40 d. Small-scale Family Farmer Support TL/ha N/A 1000 1000 1000 1000 e. Organic Agriculture Support by Years 2019 Category Type (2) Unit 2015 (3) 2016 2017 2018 IC GC Category Type 1 TL/ha 700 1000 1000 1000 700 350 Category Type 2 TL/ha 700 700 700 400 200 Category Type 3 TL/ha 300 300 300 100 50 Category Type 4 TL/ha 100 100 100 No C4 f. Good Agricultural Practices Support by Years(4) Certification 2019 (5) Cultivation Type Unit 2015 2016 2017 2018 Type C1 C2 C3 C4 IC TL/ha 500 500 500 500 500 400 300 100 Total Package for Open Fields GC TL/ha 400 400 250 200 150 100 IC TL/ha 1500 1500 1500 1500 1500 No greenhouse Total Package for Greenhouse Cultivation GC TL/ha 750 cultivation II. Biological and Biotechnical Struggle Support Cultivation Type Support Type Unit 2015 2016 2017 2018 2019 Biological 350 350 350 500 500 Total Package for Open Fields Biotechnical 350 350 450 500 800 TL/ha Biological 3500 3500 3500 4000 4000 Total Package for Greenhouse Cultivation Biotechnical 1100 1100 1100 1200 1200 III. Differentiation/Compensatory Payments (are not used for fruits and vegetables) IV. Other Agricultural Subsidies a. Farm Accounting Database Participation TL/agribusiness 475 425 500 600 600 b. Agricultural Expansion and Consulting TL/year 600 20,000 35,000 38,000 46,000 * Support payments differ across product groups Notes: (1) Diesel support payments for onions accounted for 170 TL/ha in 2019. (2) There are four categories of agricultural commodities for the purposes of organic farmer support. Tomato and pepper are in the first category and dry onion is in the third category. (3) There was no category distribution in 2015. Crops were classified as fruits and vegetables and field crops. Pepper, tomato and dry onion are all in the F&V. (4) Onions do not qualify for greenhouse production support due to the open field nature of their production. (5) Peppers and tomatoes are in C1, but dry onions are in C2 in 2019. IC: Individual Contract. GC: Greenhouse Contract. Source: MOAF Support Programs Bulletins, 2015–2019. support aims to increase agricultural productivi- Other requirements for eligibility vary according ty through the dissemination of information. To to the type of farming. For farmers engaged in be eligible, farmers must be registered with FRS. rain-fed farming, the minimum requirement is © 2020 The World Bank Group 31 Drivers of Food Price Inflation in Turkey 100 decares of land under cultivation. For those of tomatoes, green peppers, and onions was not engaged in both rain-fed and irrigated farming, tested empirically, it is plausible to conclude that the minimum requirement is 100  decares with there is no strong direct causality between the at most 50  decares irrigated, while for those two. Such a causality can primarily stem from ei- engaged in irrigated farming only, the minimum ther the magnitude of support that can signifi- required land is 50  decares. The minimum re- cantly distort prices or uncertainty associated quirement for greenhouse production is two de- with frequent policy change. As shown above, cares of land under cultivation. The extension and neither factor is attributable to the current policy consulting services subsidy also covers organ- situation in the vegetable markets. The indirect ic farmers engaged in all of the above, and the impact of support payments on rising food prices, minimum land under cultivation required is half however, may come from the sub-optimal allo- of each category discussed. The services can be cation of resources in the vegetable sector and provided by agricultural consultants/consultan- limited incentives for farmers to increase their cy companies or associations authorized by the productivity. In terms of trade policy, high im- Ministry, producer organizations, and Chambers port protection rates reduce the elasticity of the of Agriculture. The services are officiated by con- vegetable supply curve. While Turkey is a  net tracts based on Ministry guidelines and signed be- exporter for all three vegetables, the price impact tween the provider and the farmers. Until 2016, that stems from seasonal shortages of supplies 600 TL per year was paid in two installments for of these vegetables may be exacerbated by the each consulting farmer to the authorized compa- restrictions on the import side. In addition, import ny/organization for up to eight agricultural con- tariffs limit incentives for producers to improve sultants. In 2016, there was a  significant jump their productivity and slow down the exit of in- in payments for extension services (Table 12). efficient producers from the sector, lowering the In 2019, 46,000 TL per year was designated to be sector’s overall competitiveness. paid in two installments for each consultant hired in one of the authorized consulting organizations An Econometric Analysis of Price for up to five consultants. Payments for 2019 are Formation and Transmission in Table planned to be made in 2020. The payments are Tomato, Green Pepper and Onion Markets made in full provided that the consultants are re- tained for 12 months. Determinants of price formation in fresh vegetable markets From the standpoint of trade policy, vegetables in Turkey enjoy high import protection rates. In One way to better understand the dynamics of 2018, the applied MFN tariff rate on vegetables price formation is to decompose it into separate was 23.5  percent, including 48.6  percent for to- cycles and trends of different frequencies. This matoes, 19.5  percent for green peppers, and allows for a more granular distinction between per- 49.5 percent for onions (WITS data). While the ap- manent and transitory price shocks.36 Specifically, plied MFN rates have been at the same level for the for the purposes of this analysis, price variances select vegetables over the last decade, the overall for the three vegetables of interest were decom- MFN rate for vegetables has increased since 2016. posed into three components: seasonal compo- In 2019, the Government of Turkey lowered import nent, cyclical component and a long-term trend as tariffs on onions to zero percent, triggering large per the following equation: import inflows to the country. pt ≡ Tt + Ct[1,k] + Ct[0,1], where Agricultural and trade policy have implications for vegetable price levels. Although the causal- Ct[0,1] captures short-term cyclical movements with ity between agricultural policies and the prices a periodicity of less than one year to capture the 36 Baffes and Kabundi, 2020, forthcoming. 32 © 2020 The World Bank Group III. Regional Price Formation and Transmission: The Case of Table Tomatoes, Green Peppers and Onions seasonality influence on food prices arising from with 25 percent. In the case of onions, the cyclical the harvest cycles.37 Tt captures a long-term trend component and the long-term trend account for that reflects general price level changes over time. 84 percent of price variability. Lastly, Ct[1,k] accounts for cyclical patterns in eco- nomic activity that separate seasonal drivers from The current production and trade structures can a long-term trend. For the purposes of this study, help explain the role of the seasonal component Ct[1,k] is aligned with a duration of a traditional busi- in the price formation of selected vegetables. ness cycle associated with economic activity with Both tomatoes and green peppers share a  simi- periodicity of 2–8  years, following NBER’s tradi- lar open-field growing season. However, most of tional definition (Burns and Mitchell 1946); hence, the green pepper production in Turkey is open- k = 8. field. On average 29  percent (749,000  tons) is produced in greenhouses, ranging from 14  per- Decomposition of the monthly prices for to- cent for capia peppers to 76 percent for banana matoes, green peppers, and dry onions into peppers. The share of greenhouse production for seasonal and cyclical components, and a  long- tomatoes is much higher at 46 percent. This ex- term trend over the period from January 2005 tends the overall growing season for tomatoes. to December 2019 unmasks the heterogeneity In addition, as alternative greenhouse produc- of their formation (Figure 29). Specifically, in tion methods for year-round production of to- the case of green peppers, the seasonal compo- matoes are developing, tomato production will nent, on average, accounts for over 43  percent become even less prone to seasonal price vari- of price variability across all regions, followed by ability. At the same time, unlike tomatoes and a long-term trend that drives 34 percent of price green peppers, dry onions can be stored for up to variability. For tomatoes, the long-term trend is 12 months, so seasonal variation in prices is less the main driver of price variability, accounting for likely as is confirmed by the analysis. On the trade 53  percent, followed by the cyclical component side, tomatoes are the most exported vegetable Figure 29. Share of Price Variance Explained by Seasonal and Cyclical Components, and a Trend 0.6 0.53 0.5 0.43 0.44 0.41 0.4 0.34 0.3 0.25 0.20 0.20 0.2 0.14 0.1 0 Seasonal Cyclical Trend Seasonal Cyclical Trend Seasonal Cyclical Trend Green Peppers Table Tomatoes Dry Onions Note: The results presented in the figure reflect an average across all the regions in the country. For a more detailed regional analysis, see Appendix 4. Source: Authors. Evidence in support of seasonal influences on food price variability is well documented in the literature. See Baffes et al. 37 (2015), Sahn et al. (1989), and Kaminski et al. (2014). © 2020 The World Bank Group 33 Drivers of Food Price Inflation in Turkey in Turkey – 31 percent of total vegetable exports, trucks further decreases the shelf-life of the fresh compared to 9  percent for green peppers and produce. 2.4 percent for dry onions. Such trade dynamics decrease the supply-side downward pressure on Like green peppers, tomato price dynamics tomato prices at harvest time as there is large also followed a  relatively consistent seasonal- foreign demand for tomatoes. ity pattern between 2005 and 2015 (Figure 31). However, peaks and troughs in tomato prices are A more detailed analysis of the price variance much less pronounced than in the case of green driven by the seasonal component offers ad- peppers, which can once again be largely explained ditional insights into seasonal influences on by the high share of greenhouse production and the prices of green peppers and tomatoes. large export share that smooth supply and demand In the case of green peppers, the seasonal pattern pressures. The analysis shows that the consistency is relatively consistent since 2005 (Figure 30)  – of the seasonality pattern gets disrupted in two in- with prices decreasing at harvest time and rising stances, which sheds light on the contributing fac- in the off-season. More so, over time, the am- tors to tomato price formation in recent years. The plitude of price variability caused by seasonali- first disruption occurred in late 2015/early 2016, ty has increased, with price increases becoming likely driven by Russian restrictions on Turkish im- more pronounced. While it is difficult to be sure ports, including tomatoes. The second disruption which factors are causing this dynamic, it is plau- occurred in the summer of 2018 when prices rose sible to assume that it may be driven by a grow- during the harvest season. A  similar situation oc- ing demand for fresh vegetables, including green curred in the summer of 2019. Turkey experienced peppers, during the off-season months. What is economic turmoil in mid-2018, with a significant ex- clear, however, is that the market is increasingly change rate depreciation and a subsequent increase unable to deal with balancing supply and demand in input costs that led to a rise in tomato prices. To for green peppers in the off-season due to a short control price spikes, the government through munic- production cycle, poor handling and packaging ipalities started to sell tomatoes and other vegeta- techniques, and inadequate storage capacity. As bles at a lower price in specific places in cities. While vegetable production is dominated by small-scale the volume of vegetables sold through these chan- farmers, they often lack appropriate on-farm stor- nels was limited, it did signal market uncertainty. age techniques and facilities to safeguard produc- Finally, in December 2018 and January 2019 floods tion. Lack of enough cold storage and refrigerated devastated greenhouses in several districts of the Figure 30. Seasonal Component of Green Pepper Prices 5 4 3 2 1 0 -1 -2 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Source: Authors 34 © 2020 The World Bank Group III. Regional Price Formation and Transmission: The Case of Table Tomatoes, Green Peppers and Onions Figure 31. Seasonal Component of Table Tomato Prices 1.5 1.2 0.9 0.6 0.3 0 -0.3 -0.6 -0.9 -1.2 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Source: Authors Antalya region, causing an additional supply shock. cohesion for seasonal and cyclical components As a result, tomato price variability significantly in- across the three vegetables. Several results creased in this period, as can be seen in Figure 31. emerge from the co-movement analysis.40 First, within-group correlations are high, especially for The effects of macroeconomic fundamentals on the cyclical component. Second, cross-commod- cyclical components differ across all three veg- ity correlations are low compared with the with- etables, ranging from 19 percent for green pep- in-group classification. Third, cyclical components pers to 62 percent for dry onions. The duration of score high in correlation with seasonal compo- a cyclical component in this analysis to a large ex- nents. The findings suggest that macro fundamen- tent mimics a business cycle. As such, it is expect- tals serve as a cohesion force for price movements ed that macroeconomic fundamentals play an im- across different commodities. portant role in price formation within such cycles. The results obtained from a linear regression con- Implications for policymakers. The analysis of firmed that macroeconomic fundamentals explain the impacts of the seasonal component on the a varying share of price variability within a cyclical price formation of green peppers and tomatoes component for the three vegetables. Specifically, suggests the possibility of significant gains for macroeconomic factors38 explain 62  percent of farmers from better post-harvest storage tech- price variance under a cyclical component for on- niques that would allow farmers to take advan- ions, 55 percent for tomatoes, and 19 percent for tage of seasonal price differentials. There are also green peppers. potential gains from lengthening the production season through investments in greenhouse pro- The analysis of price co-movement39 across duction. Policy interventions, such as creating frequencies highlights the difference in price an enabling environment for producers to access Macro fundamentals used in regressions include: for onions – core inflation, energy prices, real effective exchange 38 rate advanced economies, minimum wage, industrial production and trade activity (exports and imports of goods) for onions; core inflation, energy price, USD, minimum wage; for tomatoes – core inflation, energy prices, real effective exchange rate advanced economies, minimum wage, industrial production and trade activity (export and imports of goods); and for green peppers – core inflation, energy price, USD, minimum wage, industrial production and trade activity (exports and imports of goods). A dummy was added to reflect the import embargo imposed by Russia on Turkey in November 2015. 39 See Appendix 3 for a detailed methodology on price co-movement analysis. 40 For this analysis, monthly price data for the period from January 2005 to December 2019 was used. © 2020 The World Bank Group 35 Drivers of Food Price Inflation in Turkey finance and strengthen collective action in the Kshirsagar (2019). For each analyzed pair or regions, sector can help small-scale farmers access the a region that adjusts to the lagged spread between more advanced storage technologies, green- two cointegrated price series is the endogenous house production as well as enhance marketing (i.e. a  price follower) market with regards to that opportunities for smallholders, particularly into pair. If it does not adjust, it is considered exogenous the more lucrative formal retail markets. In ad- (i.e. price leader) and is considered to lead the price dition, training and educating farmers through formation. In the following step, a network central- extension services, including e-extension, can in- ity is estimated that allows for the identification of crease farmers’ knowledge on post-harvest han- the sources of demand and supply shock for each dling and storage of vegetables. vegetable of interest and, consequently, the chan- nels through which they influence domestic prices Origins of vegetable price shocks (see Appendix 5 for a more detailed methodology). Another important way to analyze the drivers of For tomatoes, the Tekirdag region is the most price variability across the regions is by identify- influential market during both harvest and lean ing the origin of price shocks in the country and seasons, pointing to the role of tomato exports their transmission across the regions. To do so, in Turkish tomato price formation (Figure 32).41 we first apply a bivariate error correction model to While Tekirdag in itself is a region with small levels determine the direction of price influence across of tomato production and relatively low average the regions, following the analysis framework from yields (Figure 17), it borders Greece and Bulgaria, Baffes (1991), Baffes et al. (2019), and Baffes and and provides access to the European Union and Figure 32. Table Tomato Price Structure During the Harvest and Lean Seasons Harvest season Lean season Note: The size of the circles indicates the relative importance of a market. The direction of the arrows should be interpreted as ‘influenced by’. For example, in the Istanbul-Aydin pair during the harvest season, the arrow points towards Istanbul. It means that the Aydin price is ‘influenced by’ the Istanbul price. Source: Authors. 41 See Appendix 7 for more detailed results. 36 © 2020 The World Bank Group III. Regional Price Formation and Transmission: The Case of Table Tomatoes, Green Peppers and Onions Eastern European markets that serve as the key with access to the largest importers, one can in- importers of Turkish fresh tomatoes. All the ma- fer that the domestic pricing mechanism for these jor road routes used to transport Turkish cargo to two commodities is largely influenced by export this region start out in Bulgaria and continue to demand pressure from these countries.42 Romania, Moldova, Ukraine, Belarus and Russia. For Russia, another route involves ferries travel- For green peppers, regional price linkages reflect ling straight to Novorossiysk, but it is not popular the highly seasonal nature of production. During due to a limited ferry schedule and low capacity. the harvest season, Balikesir and Bursa serve as the As mentioned previously, in 2016–2018, Romania, most influential markets (Figure 33).43 Both regions Belarus, and Ukraine were the largest importers of are among the largest producers of green peppers, Turkish fresh tomatoes, and prior to 2016 Russia accounting for 18.4  percent of total production, was the largest importer. Exports remain consis- pointing to the downward supply pressure on prices tently high during the entire year, which explains during harvest. In contrast, during the lean season, the importance of the Tekirdag region in price deficit markets, such as Agri, Kayseri, and Kirikkale formation in both lean and harvest seasons. As become important for price formation, exerting an such, between 2015 and 2018, Turkey on aver- upward pressure on the prices of peppers. At the age exported $25.0 million worth of tomatoes per same time, the largest greenhouse pepper produc- month, ranging from $16.4 million during the sum- ing regions, Adana and Antalya, provide an annual mer and early fall months to $33.6 million during supply of peppers, counterbalancing the existing the late fall to early spring months. Given that the deficit in the lean season. The findings reconfirm an regions responsible for influencing most domestic earlier conclusion that there are significant gains for tomato markets are located close to the border farmers from applying better post-harvest storage Figure 33. Green Pepper Price Structure During the Harvest and Lean Seasons Harvest season Lean season Source: Authors. This association is conjectural. To formally test it we would need similar price series for the corresponding markets in the 42 neighboring countries. (See Baffes and Kshirsagar, 2020 for an example.) See Appendices 7 and 8 for more detailed results. 43 © 2020 The World Bank Group 37 Drivers of Food Price Inflation in Turkey techniques and extending the production season with more than 90 percent of imports taking place through greenhouse production to take advantage from October to March. of seasonal price differentials. Implications for policymakers. The network Both cross-regional supply and demand imbal- centrality analysis highlights the important role ances and export demand play a role in the price played by border markets in price formation for formation of dry onions. In the harvest season, onion and tomato markets, as well as the import- regions driving the price formation are a  surplus ant role of seasonal demand and supply imbal- region: Ankara, which is one of the largest pro- ances across selected regions for green peppers. ducers of onions in the country; and two deficit Apart from the need to reduce the effects of regions: Erzurum and Agri (Figure 34).44 Both re- seasonality of production on market prices that gions produce very limited volumes of onions and is discussed in more detail in the previous sec- have the lowest yields in the country. In addition, tion, an important implication of the analysis of Agri borders Georgia, which historically has been demand and supply shocks presented here is the one of the three largest importers of Turkish on- need to strengthen agricultural market informa- ions, exerting additional demand pressure on the tion systems in the country. Such systems should region. Historically, more than 60 percent of onion offer just-in-time provision of supply, demand, trade with Georgia took place between May and and price information in order to enhance mar- July. In the lean season, price formation is primar- ket transparency and point to existing cross-re- ily driven by the region of Van that borders Iraq, gional supply and demand imbalances. In turn, the second largest importer of Turkish onions. this would lead to a  more efficient allocation of Between 2010 and 2018, Iraq on average imported resources in vegetable markets. Creating an en- $5.4 million worth of onions from Turkey annually abling environment for agricultural e-commerce Figure 34. Dry Onion Price Structure During the Harvest and Lean Seasons Harvest season Lean season Source: Authors. 44 See Appendices 7 and 8 for more detailed results. 38 © 2020 The World Bank Group III. Regional Price Formation and Transmission: The Case of Table Tomatoes, Green Peppers and Onions is another avenue for the government to lower peppers, and dry onions in Turkey. Spatial price transaction costs that exist in vegetable markets transmission was tested across major wholesale by facilitating price discovery, helping match buy- markets46 in the country and across several surplus ers and sellers, and reducing the cost and time of and deficit markets for the analyzed vegetables. each transaction. In addition, vertical price transmission between consumer and producer prices was tested for on- Transmission of prices across regions ion prices.47 Consumer prices48 were regressed on producer prices at each location. In both cases, Transmission of prices for commodities in mar- the Engel-Granger cointegration procedure was kets separated by time, market form, and space used to estimate the long run cointegration rela- are an important indicator of overall market ef- tionship and the implied price transmission elas- ficiency and performance. The conceptual un- ticities. In addition, for spatial price transmission, derpinnings of any price transmission model rely Johansen cointegration tests were conducted for on arbitrage conditions which dictate that prices each of the relevant pair-wise price comparisons that wander too far apart trigger spatial activities (see Appendix 8 for more details on methodology). that act to draw prices together (i.e., by buying Finally, for the spatial price transmission analysis, where the commodity is cheap and selling where multivariate VAR models containing the entire set it is demanded). Thus, arbitrage ensures that price of consumer prices for each individual commodity differentials (i.e., the difference between two pric- were analyzed to generate orthogonalized impulse es at the same point in time) will be disciplined so response functions to test how each analyzed as to not wander arbitrarily in excess of transport market responds to an exogenous shock (one or processing costs. Adherence to perfect price standard deviation of the VAR error terms). transmission is often termed as the “Law of One Price” (LOP)45 that implies perfect transmission The results of the spatial price transmission of shocks and price and exchange rate transmis- analysis suggest that wholesale and consumer sion elasticities equal to one. Elasticity of less than prices appear to move in a similar pattern in the one suggests some barriers to the transmission of long run.49 The long-run price transmission was price shocks. This may reflect policies, market in- tested between two central markets, Istanbul and frastructure (i.e. storage, distribution), and short- Ankara,50 and a  wide variety of “satellite” mar- comings in transportation networks. kets across the country. For the wholesale price transmission tests, such satellite markets includ- Three types of analyses were conducted to test ed Adana, Antalya, Aydin, Manisa, Mersin, and market efficiency for table tomatoes, green Samsun. These are the regions that have the 45 A typical specification of the LOP, given in logarithmic terms, is pti = α0 + β1ptj + β2πij t + εt where pti is the logarithmic transformation of the price in market i and πij t is the exchange rate for market j in terms of market i’s currency. For prices quoted in the same currency, the logarithmic exchange rate is zero. Perfect market integration and adherence to the LOP is implied when α0 = 0 and β1 = 1, reflecting the arbitrage condition of pti = ptj. This condition, however, abstracts from trade and transportation costs, which may impose significant differences in regional market prices. For this analysis, daily price data for the period from January 1, 2018 to December 31, 2019 was used. 46 Similar analyses were not conducted for table tomatoes and green peppers due to missing data in the corresponding price 47 series. For this analysis, monthly price data for the period from January 2005 to December 2019 was used. 48 Spatial trade is, by definition, a dynamic process since commodity exchanges across different markets likely involves delivery 49 lags. This may suggest that deviations from a parity equilibrium exist but should not be persistent in the long run. The choice of these two central markets, against which all satellite market prices are compared, is somewhat arbitrary. 50 However, these two cities are the largest in Turkey and thus are likely to be important in the distribution of commodities to consumers. Ankara also plays an important role as a large producer of onions and tomatoes. © 2020 The World Bank Group 39 Drivers of Food Price Inflation in Turkey largest wholesale markets in the country. For the represent reasonably strong evidence of well-inte- consumer price transmission tests, the satellite grated markets. Exogenous shocks to most mar- markets included Antalya, Bursa, Istanbul, Izmir, kets trigger statistically significant reactions in the Konya, Samsun, Trabzon, and Van, representing other markets. In nearly every case, market shocks a mix of surplus and deficit markets for tomatoes, appear to trigger reactions in other markets. In all green peppers, and onions located throughout the cases for wholesale prices, the adjustments to ex- country. The results presented in Appendix 7 show ogenous shocks are rapid and are only significant that long-run price transmission elasticities, de- for the first few days after the shock. noted as slope parameters, are always statistically significant and generally range from 0.60 to 1.0 For consumer prices, differences in transmission for wholesale prices and 0.85 to 1.0 for consumer patterns across the regions point to the existence prices, suggesting that the markets are well-inte- of market inefficiencies.51 For monthly consumer grated. Additional Johansen trace cointegration prices, for most of the analyzed regions exoge- test results further confirm that the logarithmic nous shocks again usually evoke adjustments that prices are cointegrated with some exceptions. are complete on average after 0.5–1  month (see Appendix 7 for detailed results). However, some An analysis of the speed and magnitude of the markets do not appear to be well-integrated in transmission of short-term shocks confirms high that exogenous shocks do not tend to affect other levels of market integration across wholesale local markets. This is the case for the Izmir and markets. Findings show that all analyzed markets Trabzon tomato markets, Van pepper markets, and respond to shocks in the central markets. For daily Konya and Samsun onion markets. onion prices, Antalya, Manisa, and Mersin appear to have prominent interactions with the other markets While there is a relatively good, albeit not uni- in that exogenous shocks to these markets tend to form, transmission in consumer prices across re- result in statistically significant responses in most gions, the strength of the relationship is much of the other markets. In the case of daily pepper weaker between consumer and producer prices. prices, strong price leadership roles are exhibited by Data limitations resulted in a  formal price trans- Adana, Istanbul, Mersin, and Samsun. In the case of mission analysis to be conducted only for onions. daily wholesale tomato prices, the impulses again The results presented in Table 13 show that price Table 13. Pairwise (Consumer and Producer) Cointegration Regression and Tests for Monthly Dry Onion Prices Intercept Slope Market R Square Johansen Trace Estimate Std. Err. Estimate52 Std. Err. Izmir 0.4021 0.0316 0.7440 0.0780 0.46 4.72 Bursa 0.5195 0.0567 0.5874 0.1722 0.10 8.06 Ankara 0.4811 0.0371 0.5355 0.0426 0.60 6.03 Konya 0.3970 0.0360 1.1290 0.1118 0.49 10.07 Isparta 0.2338 0.0211 1.2197 0.0854 0.66 5.74 Samsun 0.6239 0.0338 1.1161 0.0649 0.74 6.07 Trabzon -0.0834 0.0722 0.6763 0.1361 0.19 2.56 Van 0.5690 0.0321 0.6964 0.0886 0.37 5.96 Source: Authors 51 Results for wholesale and consumer price transmission cannot be directly compared due to differences in the length of the analyzed series. See footnote above. 52 Greater than one price transmission elasticity points to a certain degree of “overshooting” in adjustment to price shocks at the producer level. 40 © 2020 The World Bank Group III. Regional Price Formation and Transmission: The Case of Table Tomatoes, Green Peppers and Onions transmission elasticity, measured by a slope esti- prices accelerated in late 2017 with a  particular- mate, remains relatively low for most of the ana- ly sharp difference observed since the summer of lyzed markets. This is also evidenced in the much 2018. The findings suggest that while food price lower R-square values, when compared to the re- growth has accelerated, price increases have not sults of the spatial price transmission. Adjustments been passed through to producers, limiting their to equilibrium shocks are also much slower than incentives to improve productivity. in the case of consumer prices. Here, the half- lives of price adjustments generally last around An increasing wedge between producer and con- 3–4 months, which suggests that one-half of the sumer prices for both onions and tomatoes in deviation from equilibrium is eliminated over a 3- recent years is at least partially driven by supply to 4-month period. shocks. The Aydin and Antalya regions were hit by severe floods in the summer of 2018, which had This is also reflected in the visual analysis of the a devastating impact on tomato greenhouse pro- onion prices over time (Figure 35). After 2012, duction and led to price spikes that were absorbed producer prices for dry onions have been increas- by consumers. In the same year, production vol- ingly de-linked from consumer prices. This ten- umes decreased for onions, driven by unfavorable dency increased in 2018, implying that onion pro- weather conditions, product loss due to diseases, ducers do not benefit from high price signals paid and losses in warehouses as well as shrinkages in by consumers. With an increase in input prices as the production area due to decreasing profit mar- shown earlier, the profit margins of the onion pro- gins for producers. This led to a sharp increase in ducers are squeezed between high input costs and consumer prices for onions that was alleviated in low farm-gate prices. early 2019, when the government reduced import tariffs to zero, allowing onion imports. In both cas- Similar dynamics are observed for tomatoes es, the price increases were absorbed by consum- (Figure 36). Consumer prices have been grow- ers, but not transferred to producers. ing at a  much higher rate than producer prices. In the Aydin region, a  wedge between consum- The current structure of the agri-food supply er and producer prices emerged in 2014 and has chains in Turkey is characterized by various progressively increased since then, particularly factors that may be limiting vertical price since 2016. In the Antalya region a divergence in transmission efficiency. A  deeper look into the Figure 35. Comparison of Producer and Consumer Prices for Onions 6 5 4 3 2 1 0 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Consumer prices Producer prices Source: TURKSTAT, 2019 © 2020 The World Bank Group 41 Drivers of Food Price Inflation in Turkey Figure 36. Comparison of Producer and Consumer Prices for Table Tomatoes Aydin region Antalya region 8 8 TL/kg TL/kg 7 7 6 6 5 5 4 4 3 3 2 2 1 1 0 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Consumer prices Producer prices Consumer prices Producer prices Note: Complete producer price data is only available for Aydin and Antalya where year-round tomato production in greenhouses takes place. Source: TURKSTAT, 2019 structure of the tomato value chain (Figure 37) wholesalers, and retailers in the Izmir region under offers insights into where such inefficiencies may three scenarios of marketing flows: 1) producer– exist. Under Turkish Law, there are three types of wholesale commissioner–retailer; 2) producer– wholesale activity: commissioner, trader, and pro- trader (registered/informal); and 3) producer–re- ducer groups. Commissioners mediate between tailer. Under the first scenario, profit margins for farmer and retailer for a commission (up to 8 per- farmers, commissioners, and retailers were 7, 8, cent of the farm gate price) without owning the and 13  percent, respectively. When the commis- product (EBRD, 2018). Traders, on the other hand, sioner is replaced with an unregistered collector buy the products from farmers and sell them to (trader), the price received by the farmer does not retailers. In general, the role of the local trader or change, but the trader is able to sell tomatoes to commissioner is critical to the producer: they ad- retailers at a  cheaper price, since s/he does not vance all liquidity needs in the crop cycle through pay tax. This increases the profit margin of the re- cash, provide inputs, support the organization of tailer by 2 percent. In the third flow, supermarkets manual labor, and offer technical crop and weather collect tomato harvests from the farmers directly. advice, which binds the farmers to directly sell to Under this scenario, farmers still receive the same their specific buyer. Producer groups are gener- price as in the previous scenarios, but supermarket ally not active in the fresh vegetable or fruit sec- profit margins increase to 21 percent. The analy- tors. Two thirds of the fresh tomatoes produced sis presented here indicates that domestic tomato in Turkey are sold through traders and commis- price formation, in the case of consumer prices, sioners while the remainder are sold by farmers is largely being driven by export markets. While directly to retail chains. profit margins were not estimated for a marketing flow scenario “producer–trader–exporter”, given Tomato farmers have very limited bargaining the price divergence in consumer and producer power when dealing with both wholesalers and prices for tomatoes, it is plausible to assume that retailers. A study conducted by EBRD (2018) an- farmers have a  limited bargaining power in this alyzed profit margins received by tomato farmers, scenario as well. 42 © 2020 The World Bank Group III. Regional Price Formation and Transmission: The Case of Table Tomatoes, Green Peppers and Onions Figure 37. The Structure of the Fresh Tomato Value Chain in Turkey Domestic consumer District Other stores Local Chain District Greengrocer Producer (Restaurant, Supermarket Supermarket Bazaar catering, etc.) Bazaar FFV Wholesale Market Trader (outside of FFV Trader in FFV Wholesale Market) Comissionaire Market Fresh (Table) Tomato Producer (Open field) Source: EBRD, 2018 Implications for policymakers. Divergence be- prices for their produce. Limited cross-region- tween producer and consumer prices shows that al price transmission further points to deficient producers do not receive the existing market market linkages resulting in an inefficient alloca- price signals due to structural inefficiencies along tion of resources. Policies and strategies aimed the value chains and limited cross-regional link- at linking farmers to local and national markets ages. Hence, any policy aimed at increasing pro- through improved collective action, better market ductivity associated with the production cycle, transparency and digital marketplaces, to name that is extension, access to credit, etc., will have a  few, will result in higher farm-gate prices and only a  limited impact on farmers’ incentives to greater incentives for farmers to adopt produc- improve productivity if they do not receive higher tivity enhancing measures. © 2020 The World Bank Group 43 Drivers of Food Price Inflation in Turkey IV. Policy Recommendations Food price inflation in Turkey is complex and factor in higher inflation expectations, becoming driven by various interacting and interdepen- a  self-fulfilling prophecy. In this regard, the most dent factors. The depreciation of the Turkish lira crucial elements of the monetary policy response and inflation expectations, demand-side pressures to contain food prices in the short run should fo- of a growing population, changing consumer pref- cus on strengthening external buffers, reducing erences, as well as supply-side elements, such as market anxieties and managing inflationary expec- low productivity, constitute the mix of factors that tations. This would help reduce exchange rate vol- drive food price inflation in the long run. These fac- atility and inflationary pressures. tors work alongside short-run supply and demand imbalances at the local level and increase price Agricultural and trade policies also play an im- variability across the country. Short-term positive portant role in addressing the issue of food price shocks can further impose upward pressure price inflation: policies should aim to strength- on price levels over time, if structural inefficiencies en long-run productivity growth and reduce prevent such shocks from returning to their initial structural inefficiencies and seasonality of equilibrium. production to tackle short-term price variabil- ity. The various policy options which exist are Macroeconomic factors play an important role summarized in Table 14 and categorized across in driving price inflation, including food price low and high variance drivers of price inflation, inflation, in the country, hence, anchoring in- as discussed in section two of the analysis. In flationary expectations and reducing currency the medium- to long-run the focus should be on volatility are of utmost importance to temper bringing productivity and efficiency gains through price level growth. Unanchored expectations can the liberalization of imports and investments in contribute to dollarization and capital outflows (or agricultural R&D, logistics and distribution. In a slowdown in capital inflows); this can fuel infla- the short term, policy responses should focus tionary pressures through exchange rate pass- on reducing short-term production shocks, in- through. Moreover, unanchored expectations can cluding the seasonal variability of production, as also fuel a wage-price spiral as wage adjustments well as strengthening farmers’ skills for improved Table 14. Framework for Public Policy Options to Address High Food Price Inflation and Volatility Areas for short-term policy response Areas for medium-term policy response Knowledge and skills Investment in R&D Low Variance Drivers Environmentally sustainable practices Climate change adaptation and mitigation Access to credit Trade policy Seasonality of production High Variance Drivers Market transparency Investments in logistics and distribution Market linkages Source: Authors 44 © 2020 The World Bank Group IV. Policy Recommendations productivity and improving cross-regional market Turkey’s agricultural markets are among the most and supply chain linkages. protected in the world. In the short run, import re- strictions make the supply curve less elastic, leading Strengthening long-run productivity to more variable price responses. Over the long run, import protection may have significant impacts on Increasing productivity and the resilience of food the efficiency and competitiveness of the sector. production in the country should be the priority Turkey's trade policy should focus on the gradual of any policy response targeted at controlling lowering of import tariffs to alleviate the short- and food price inflation and volatility. As shown in long-term implications of trade protection. This sections two and three of the report, Turkey lags process, however, should be aided by supporting behind comparator countries in terms of land and farmers to increase their productivity and thereby, labor productivity; yields for table tomatoes, green competitiveness in international markets. peppers, and onions remain well below their po- tential. For domestic supply to keep up with grow- Reducing price variability associated ing demand pressures, higher productivity growth with seasonality needs to be achieved. Productivity gains can be achieved by increasing labor, land, and physical The analyses of price decomposition and price capital productivity under current uses, and by re- shock origins highlighted the importance of sea- allocating productive assets within the sector. In sonality in price formation for the analyzed veg- the short term, better access to credit and better etables, particularly green peppers. Significant extension services can help improve productivity. gains can be made in the short run by supporting In addition, incentives to implement environmen- farmers with better post-harvest storage tech- tally sustainable practices need to be introduced niques that would allow farmers to take better ad- to decrease the existing and future implications of vantage of seasonal price differentials. There are natural resource depletion. also potential gains from lengthening the produc- tion season through investments in greenhouse In the medium term, state expenditures in production. Policy interventions, such as creating agriculture should be repurposed toward the an enabling environment for producers to access provision of public goods. Such public goods finance and strengthening collective action in the include R&D; pest-and-disease control; strong sector, can help small-scale farmers access more public and private food safety standards; and advanced storage technologies and greenhouse an enabling environment for private investment. production methods as well as enhance marketing Currently, Turkey spends 78 percent of total sup- opportunities for smallholders, particularly into the port in agriculture53 on market price support and more lucrative formal retail markets. In addition, payments based on input use, which may have training and educating farmers through extension negative impacts on production (OECD, 2020). services, including e-extension, can increase farm- Spending on GSSE, on the other hand, accounts ers’ knowledge of post-harvest handling and stor- for only 15 percent of total agricultural spending age of vegetables. and less than one percent of the aggregate value of agricultural production. Within this allocation, Improving market integration to facilitate price development and maintenance of infrastructure pass-through to producers accounts for approximately 75 percent, while ex- penditure for agricultural knowledge and innova- Low domestic market integration exacerbates tion systems averages only 5 percent. seasonal price fluctuations, as the price trans- mission and price network analyses, highlighted In addition, medium-term policy should focus on in section three of the report, showed. In  addi- import liberalization to drive competitiveness and tion, if domestic markets are not well integrated, efficiency gains in Turkey’s agricultural sector. farmers may not benefit from price increases. Total support estimate. 53 © 2020 The World Bank Group 45 Drivers of Food Price Inflation in Turkey This reduces the welfare of farmers and limits their Agricultural value chains can be made more incentives to invest in productivity-enhancing efficient by simplifying regulations, eliminating technologies. Several policy interventions can help entry restrictions, and allowing for more improve market transparency and linkages. competition. The lack of competition and need for improving regulations at the wholesale and retail Farmer organizations can play a  critical role in levels may be contributing to high food prices and facilitating farmers’ access to markets. They can limiting the pass-through of price signals to farm- contribute to increasing farm productivity and sup- ers. In addition to regulatory changes, policy should ply by reducing production costs, helping meet mar- focus on making price formation more transparent ket standards, adding value, and integrating small at the wholesale market level by monitoring (a) producers into value chains. Relevant international transactions between commissioners and traders experience includes the histories of Land O’Lakes ‘inside-the-zone’ and ‘outside-the zone’; (b) prices and Ocean Spray, large agribusinesses in the United paid to farmers by commissioners and traders; (c) States that are organized as cooperatives. prices paid to commissioners and traders by retail- ers; and (d) volumes of trade and types of products Currently, in Turkey, there exists room for im- sold between wholesale zones in different cities. provement when it comes to the capacity of farmer organizations, including their access to Digitalization of agriculture can serve as a  tool finance. There is a need for strengthening sup- for alleviating some of the frictions that exist in porting policy and enabling frameworks in order to the Turkish value chains and increasing the ef- increase the effectiveness of farmer organizations. ficiency of agricultural production. Digitalization A  recent study (World Bank, 2018) offered guid- can provide stakeholders along the value chain ance on how to facilitate the creation and func- with better access to information about input and tionality of farmer organizations by simplifying the product markets; reduce reliance on intermedi- relevant legal and regulatory frameworks. Key el- aries; and better align production with demand. ements for regulatory and policy reforms include: Digital advisory services can improve the knowl- (a) improving the legal/regulatory framework for edge of agricultural producers by offering infor- farmer organizations modelled after global best mation on production and post-harvest methods, practices and based on the principles of self-gov- on-farm storage techniques, use of new technol- ernance and entrepreneurship; (b) introducing an ogy, fertilizers and agro-chemicals, standards, and incentives framework that links access to finance financial management. Accurate and timely mar- with institutional development, including bench- ket information through data collection, data an- marks towards professionalization, accountabili- alytics, and communication platforms offer great ty, and market orientation; (c) providing technical potential for more equitable market access for assistance for capacity development for farm- farmers. New digital platforms and applications er organizations to achieve professionalization can more efficiently link producers to consumers. (through matching grants or direct support); (d) Turkey has great preconditions for advancing the establishing an independent regulatory agency for digitalization of agriculture, however, additional in- farmer organizations; and (e) developing a knowl- vestments are needed (World Bank, 2020). edge management and training system for farmer organizations. Policies, strategies, and investments aimed at linking farmers to local and national markets Enhanced agricultural market and price moni- and facilitating access to processing, storage, toring can reduce information asymmetries and and distribution systems can help reduce region- improve market efficiency. For farmers, market al food price dispersion across Turkey. The gov- information can improve their awareness of mar- ernment can address infrastructure bottlenecks by ket opportunities and options and strengthen their creating an enabling environment for private in- bargaining power. For traders, market information vestment in processing, cold storage, and delivery can help them identify markets with good arbi- systems to ensure quality and safety and stimulate trage opportunities. public and private partnerships. 46 © 2020 The World Bank Group References References Abak, K. (2016). Türkiye’de Domatesin Dünü, Meyve-Sebze Fiyatları Üzerindeki Etkisinin Bugünü Ve Yarini. Retrieved from www.turktob. İncelenmesi”(2015) (in Turkish), CBT Research org.tr Notes in Economics No. 15/08. Akcelik, Fatih, Canan Yuksel Yucel et al. 2016. Bor et al. 2014. “Asymmetry in farm-retail price “Turkiye’de Gida Fiyatlari: Uluslararasi Bir Karsila- transmission in the Turkish fluid milk market.” stirma,” Technical Report, Research and Monetary New Medit, vol. 13, no. 2, 2014, p. 2+. Accessed Policy Department, Central Bank of the Republic 19 Apr. 2020. of Turkey 2016. Burns, A. F. and W. C. Mitchell. 1946. Measuring Aysoy, Cevriye, Duygu Halim Kirli, and Semih Business Cycles, National Bureau of Economic Tumen, “How does a shorter supply chain affect Research. NBER, New York. pricing of fresh food? Evidence from a natural experiment,” Food policy, 2015, 57, 104–113. Çakmaklı, C., Demiralp, S., Yeşiltaş, S. and M. Yıldırım. 2019. “An Investigation of Food Bai, Jushan and Pierre Perron, “Estimating and Inflation in Turkey”. Background paper prepared Testing Linear Models with Multiple Structural for the World Bank report “Drivers of Food Price Changes,” Econometrica, 1998, 66 (1), 47–78. Inflation in Turkey:” Baffes, J. 1991. “Some further evidence on the Campa, Jose Manuel and Linda S. Goldberg, law of one price: The law of one price still holds.” “Exchange rate pass-through into import prices,” American Journal of Agricultural Economics, Review of Economics and Statistics, 2005, 87 (4), 73(4), 1264–1273. 679–690. Baffes, J. and A. Kabundi. 2020 (forthcom- Ciplak, Ugur, Eray M Yucel et al., “Trade ing). “On the Cyclicality and Comovement of Protection Measures, Agricultural and Food Commodity Prices: Order within Chaos?” Prices,” Technical Report 2004. Baffes, J. and V. Kshirsagar. 2020. “Shocks to Croux, C., M. Forni, and L. Reichlin. Food Market Systems: A Network Approach,” 2001. “A Measure of Comovement for Agricultural Economics, 2020, 51:11–129. Economic Variables, Theory and Empirics.” Review of Economics and Statistics, 83 (2): Baffes, J., V. Kshirsagar, and D. Mitchell. 2019. 232–241. “What drives local food prices? Evidence from the Tanzanian maize market.” The World Bank Eren, Okan, Suleyman Hilmi Kal, Mustafa Utku Economic Review, 33(1), 160–184. Ozmen et al., “Turkiye’de Gida Enflasyonunun Belirleyicileri,” Technical Report, Research and B. Balkan, S. H. Kal, S. Tümen 2015. “Akaryakıt Monetary Policy Department, Central Bank of the Fiyatlarının Nakliye Maliyetleri Kanalı ile Taze Republic of Turkey 2017. © 2020 The World Bank Group 47 Drivers of Food Price Inflation in Turkey Erturk, Y., & Çirka, M. (2015). TÜRKİYE VE Nakamura, E., and J. Steinsson. 2008. “Five KUZEY DOĞU ANADOLU BÖLGESİ (KDAB)’NDE Facts about Prices: A Reevaluation of Menu Cost DOMATES ÜRETİMİ VE PAZARLAMASI. Yüzüncü Yıl Models,” Quarterly Journal of Economics, 123(4), Üniversitesi Tarım Bilimleri Dergisi, 25(1), 84–97. 1415-1464. https://doi.org/10.29133/yyutbd.236256 Nakamura, Emi and Zerom, Dawit, Accounting European Bank for Reconstruction and for Incomplete Pass-Through (2010). Review of Development (EBRD). 2018. Impact Analysis of Economic Studies, Vol. 77, No. 3, pp. 1192–1230, Support Measures for Fresh Fruits and Vegetables 2010. Available at SSRN: https://ssrn.com/ab- in Turkey. stract=1949311 FAO. 2017. Building Agricultural Market Ozertan, G., Saghaian, and H. Tekguc, 2015 Information Systems: A Literature Review. http:// “Turkey’s Red Meat Sector Transfer Price www.fao.org/3/a-i7151e.pdf Dynamics and Market Power.” Economics Business and Finance. Igan, D., A. Kabundi, F. Nadal De Simone, M. Pinheiro, and N. Tamirisa. 2011. “Housing, Credit, Ozmen, Mustafa Utku, Meltem Topaloglu et al., and Real Activity Cycles: Characteristics and “Disaggregated evidence for exchange rate and Comovement.” Journal of Housing Economics, 20: import price pass-through in the light of identifi- 210–231. cation issues, aggregation bias and heterogene- ity,” Technical Report 2017. Kara, A Hakan, Fethi Ogunc, Cagri Sarikaya et al., “Inflation dynamics in Turkey: A historical Timmer, P. 2008. Causes of High Food Prices. accounting,” Technical Report, Research and ADB Economics Working paper Series No. 128. Monetary Policy Department, Central Bank of the Republic of Turkey 2017. World Bank, 2018. Leveraging Producer Organizations to Improve the Efficiency of Agri- Kara, Hakan and Fethi Ogunc, “Inflation targeting Food Value Chains in Turkey. and exchange rate pass-through: the Turkish ex- perience,” Emerging Markets Finance and Trade, World Bank. 2020 (forthcoming). Turkey – Digital 2008, 44 (6), 52–66. Agriculture Profile. __ and __, “Pass-through from exchange rates and import prices to domestic prices,”Iktisat Isletme ve Finans, 2012, 27, 9–28. 48 © 2020 The World Bank Group Appendix 1. Turkey Administrative Structure (NUTS-2) Appendix 1. Turkey Administrative Structure (NUTS-2) Tekirdağ Istanbul Zonguldak Kastamonu Samsun Trabzon Kocaeli Ağrı Balıkesir Bursa Ankara Erzurum Kayseri Manisa Malatya Kırıkkale Van Izmir Konya Hatay Şanlıurfa Mardin Aydın Gaziantep Antalya Adana 0 250 500 km © 2020 The World Bank Group 49 Drivers of Food Price Inflation in Turkey Appendix 2. Supply and Demand Balances for Tomatoes, Green Peppers, and Dry Onions Table A.2.1. Supply and demand balances for tomatoes Unit 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 2017/18 2018/19 2019/20 Production Ton 10,052,000 11,003,433 11,350,000 11,850,000 11,850,000 12,615,000 12,750,000 12,150,000 12,841,990 Harvest losses Ton 351,820 385,120 397,250 414,750 414,750 441,525 446,250 425,250 449,470 Supply=Use Ton 9,712,652 10,630,086 10,963,759 11,446,552 11,444,975 12,184,425 12,314,993 11,759,556 12,409,949 Supply Usable production Ton 9,700,180 10,618,313 10,952,750 11,435,250 11,435,250 12,173,475 12,303,750 11,724,750 12,392,520 Imports Ton 12,472 11,773 11,009 11,302 9,725 10,950 11,243 34,806 17,492 EU 27/28 Ton 4,090 5,092 6,736 4,658 7,060 7,710 8,915 9,661 9,327 Use Domestic use Ton 8,672,133 9,513,286 9,848,760 10,187,265 10,317,759 10,989,375 11,109,482 10,604,454 11,189,964 Human consumption Ton 7,804,920 8,561,957 8,863,884 9,168,539 9,285,983 9,340,969 9,443,060 9,013,786 9,511,470 Losses Ton 867,213 951,329 984,876 1,018,727 1,031,776 1,648,406 1,666,422 1,590,668 1,678,495 Exports Ton 1,040,519 1,116,800 1,114,999 1,259,287 1,127,216 1,195,050 1,205,511 1,155,102 1,219,985 EU 27/28 Ton 320,241 341,368 336,278 406,910 374,034 447,103 506,454 390,548 471,786 Change in stocks Ton  –  –  –  –  –  –  – – – Human consumption per capita Kg 105.87 114.6 117.2 119.6 119.5 118.6 116.9 109.9 114.4 Degree of self – sufficiency % 111.85 111.6 111.2 112.3 110.8 110.8 110.7 110.6 110.7 Table A.2.2. Supply and demand balances for green peppers Unit 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 2017/18 2018/19 2019/20 Production Ton 1,986,700 1,975,269 2,042,360 2,232,308 2,232,308 2,307,456 2,608,172 2,554,974 2,625,669 Harvest losses Ton 37,699 37,379 38,679 42,137 42,137 43,500 43,500 47,975 49,122 Supply=Use Ton 1,950,293 1,939,972 2,005,283 2,190,959 2,191,030 2,265,494 2,567,204 2,508,960 2,578,343 Supply Usable production Ton 1,949,001 1,937,890 2,003,681 2,190,171 2,190,171 2,263,956 2,564,672 2,506,999 2,576,547 Imports Ton 1,292 2,082 1,602 788 859 1,538 2,532 1,961 1,796 EU 27/28 Ton 395 74 47 121 134 224 793 850 340 Use Domestic use Ton 1,772,422 1,795,117 1,845,886 2,020,095 2,019,864 2,072,620 2,348,415 2,302,402 2,342,505 Human consumption Ton 1,595,180 1,615,605 1,661,298 1,818,086 1,817,878 1,865,358 2,113,574 2,072,161 2,108,255 Losses Ton 177,242 179,512 184,589 202,010 201,986 207,262 234,842 230,240 234,251 Exports Ton 177,871 144,855 159,397 170,864 171,166 192,874 218,789 206,558 235,838 EU 27/28 Ton 131,336 102,992 108,567 116,746 108,643 126,989 141,177 132,118 147,566 Change in stocks Ton  –  –  –  –  –  –  – – – Human consumption per capita Kg 21.64 21.6 22 23.7 23.4 23.7 26.2 25.3 25.4 Degree of self – sufficiency % 109.96 108 108.5 108.4 108.4 109.2 109.2 108.9 110.0 50 © 2020 The World Bank Group Appendix 2. Supply and Demand Balances for Tomatoes, Green Peppers, and Dry Onions Table A.2.3. Supply and demand balances for dry onions Unit 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 2017/18 2018/19 2019/20 Production Ton 1,900,000 2,141,373 1,735,857 1,790,000 1,790,000 1,879,189 2,175,911 1,930,695 2,200,000 Harvest losses Ton 79,800 89,938 72,906 75,180 75,180 78,926 91,388 81,089 92,400 Supply=Use Ton 1,827,830 2,051,822 1,663,003 1,715,992 1,726,095 1,800,466 2,084,695 1,977,573 2,108,174 Supply Usable production Ton 1,820,200 2,051,435 1,662,951 1,714,820 1,714,820 1,800,263 2,084,523 1,849,606 2,107,600 Imports Ton 7,630 387 52 1,172 11,275 203 172 127,967 574 EU 27/28 Ton 0 – – – 1,313 – 0 0 0 Use Domestic use Ton 1,726,189 1,908,142 1,551,875 1,513,156 1,641,045 1,763,017 1,927,428 1,899,486 1,864,166 Human consumption Ton 1,614,800 1,786,288 1,445,388 1,413,481 1,534,975 1,651,785 1,807,980 1,783,426 1,746,414 Seed use Ton 25,079 26,448 28,893 24,018 24,018 23,082 23,077 21,085 24,544 Losses Ton 86,309 95,407 77,594 75,658 82,052 88,151 96,371 94,974 93,208 Exports Ton 101,641 143,680 111,128 202,836 85,050 37,449 157,267 78,087 244,008 EU 27/28 Ton 12,926 3,235 3,588 31,017 8,973 10,912 1,535 25,256 22,676 Change in stocks Ton  –  –  –  –  –  –  – – – Human consumption per capita Kg 21.90 23.9 19.1 18.4 19.8 21.0 22.4 21.7 21.0 Degree of self – sufficiency % 105.45 107.5 107.2 113.3 104.5 102.1 108.2 97.4 113.1 © 2020 The World Bank Group 51 Drivers of Food Price Inflation in Turkey Appendix 3. Methodology for Price Co-movement Analysis To measure the price co-movement among prices of tomatoes, green peppers and dry onions, the fol- lowing measure of dynamic correlation between two prices, pi and p j, at frequency λ has been applied as follows (Croux, Forni, and Reichlin 2001): Cpi p j (λ ) ρ p p (λ ) = i j . Spi (λ ) Sp j ( λ ) Cpi p j (λC)pi p j (λ ) ρ) ρ pi p j ( λ = (λ ) = pi p j j . . Cpi p j (λ ) , is the co-spectrum between p and i Cp ; Spi)(λ )S i j (λ Sand pp j( i( λλ))Sp j ( λ ) are spectral-density functions of pi and ρ p p (λ ) = i j p j, and -π ≤ . λ ≤ π, where -1 ≤ ρ i j (λ ) = ≤ 1. p p . Spi (λ ) Sp j ( λ ) pp Spi (λ ) Sp j ( λ ) Equation (1) can be expressed within a given frequency interval, say Λ+ = [λ1, λ2], 0 ≤ λ1 < λ2 ≤ π, in the multivariate framework as follows: Σ in=1 Σ m j =1 wpi wp j ρpi p j ( Λ + ) cohpi p j ( Λ + ) = Σ in=1 Σ m j =1 wpi wp j where pti and ptj represent n- and m-vectors of time series, pti = (pt1, pt2...ptn)' and ptj = (pt1, pt2...ptm n Σ) m ' and i =1 Σ j =1 wpi wp j ρpi p j Σ n m Σ and ρpi p j ( Λ + vectors wpi wp j denote ) of non-normalized m weights associated with p i and p j, respectively. cohpi p j ( Λ + ) = Like the dy- i =1 j =1 Σ i =1 Σ j =1 wpi wp j ρpi p j ( Λ + ) n t t Σ in=1 Σ m j =1 w wp j hpi p j ( Λ + ) = nnamicm correlation, -1 ≤ cohpi p j ( Λ + ) = ≤ 1, with the standard interpretation. This allows for estimating price pi Σ i =1 Σ j =1 wpi wp j Σ n Σ m w w co-movement at different frequencies using j =1 pi or i =1 equal p j different weights. Co-movement within the fre- quency domain is increasingly being used. See, for example, Igan et al. (2011), who analyzed co-move- ment of business cycles in house prices and several macroeconomic variables of 18 advanced econo- mies during 1981–2006 as well Schuler, Hiebert, and Peltonen (2015) who undertook a cross-country co-movement analysis on the financial medium-term cycles. 52 © 2020 The World Bank Group Appendix 4. Price Decomposition Results for Tomatoes, Green Peppers, and Dry Onions Appendix 4. Price Decomposition Results for Tomatoes, Green Peppers, and Dry Onions Table A.4.1. Share of price variance explained by seasonal and cyclical components, a long-term trend, and cross-regional dynamics Green Peppers Table Tomatoes Dry Onions Region Ct[0,1] Ct[0,8] Tt Ct[0,1] Ct[0,8] Tt Ct[0,1] Ct[0,8] Tt Istanbul 0.38 0.20 0.39 0.19 0.24 0.56 0.15 0.40 0.44 Tekirdağ 0.44 0.20 0.34 0.21 0.24 0.54 0.15 0.38 0.47 Balıkesir 0.36 0.20 0.42 0.22 0.23 0.53 0.12 0.40 0.47 Izmir 0.40 0.21 0.37 0.19 0.26 0.54 0.16 0.45 0.38 Aydın 0.42 0.19 0.37 0.22 0.28 0.49 0.15 0.45 0.38 Manisa 0.45 0.20 0.33 0.20 0.26 0.52 0.13 0.43 0.42 Bursa 0.36 0.19 0.42 0.22 0.26 0.51 0.13 0.40 0.45 Kocaeli 0.46 0.20 0.32 0.18 0.25 0.55 0.15 0.43 0.40 Ankara 0.42 0.15 0.40 0.21 0.22 0.57 0.16 0.39 0.43 Konya 0.47 0.22 0.29 0.20 0.25 0.53 0.16 0.42 0.41 Antalya 0.37 0.19 0.41 0.18 0.25 0.56 0.13 0.42 0.45 Adana 0.45 0.21 0.30 0.21 0.27 0.49 0.14 0.43 0.43 Hatay 0.42 0.19 0.34 0.22 0.26 0.49 0.15 0.39 0.45 Kırıkkale 0.46 0.23 0.29 0.19 0.26 0.53 0.14 0.44 0.41 Kayseri 0.42 0.20 0.35 0.20 0.26 0.53 0.13 0.41 0.44 Zonguldak 0.42 0.19 0.37 0.18 0.25 0.56 0.15 0.41 0.42 Kastamonu 0.38 0.19 0.40 0.17 0.23 0.59 0.13 0.41 0.44 Samsun 0.44 0.19 0.34 0.23 0.25 0.51 0.14 0.42 0.43 Trabzon 0.44 0.19 0.35 0.20 0.25 0.54 0.14 0.42 0.43 Erzurum 0.50 0.18 0.29 0.22 0.27 0.49 0.13 0.38 0.47 Ağrı 0.44 0.19 0.34 0.20 0.27 0.51 0.14 0.38 0.46 Malatya 0.51 0.21 0.26 0.21 0.24 0.53 0.16 0.39 0.43 Van 0.48 0.21 0.28 0.17 0.26 0.54 0.13 0.33 0.53 Gaziantep 0.43 0.21 0.30 0.21 0.27 0.50 0.15 0.40 0.43 Şanlıurfa 0.43 0.22 0.32 0.19 0.26 0.53 0.15 0.40 0.44 Mardin 0.49 0.21 0.28 0.20 0.26 0.53 0.16 0.37 0.46 Notes: Tt, C t , and C [1–8] denote the trend, cyclical component, and seasonal component. The averages may not add up to [0–1] t 100 since the seasonal component moves from 2 months to 12 months. Hence, the remainder accounts for short-term variation of less than 2 months. © 2020 The World Bank Group 53 Drivers of Food Price Inflation in Turkey Appendix 5. Bivariate Error Correction and Network Centrality Methodology The bivariate vector error correction model (Engle and Granger 1987) can be specified as follows. Let pti and ptj be nominal (logarithmic) prices at time t for the relevant commodity (tomatoes, peppers or onions) for a pair of markets, i and j. Then, the following error-correction specification for each pair of markets is defined as: Δpti = μi + αij(ptj–1 – pti–1) + γ1 i Δptj–1 + γ2 i Δpti–1 + BiFt(•) + uti (1) Δptj = μi + α ji(pti–1 – ptj–1) + γ1 j Δptj–1 + γ2 j Δpti–1 + B jFt(•) + utj (2) where μi, αij, γ1 i , γ2 i , and γ3 i in equation (1) denote parameters to be estimated; and uti denotes an inde- pendently and identically distributed error term. Δ represents the first difference operator. Similar defi- nitions apply for equation (2). The term BiFt(•) is defined as follows:  2π t   2π t   4π t   4π t  B i ft ( i) = β1sin   + β2cos   + β3cos   + β 4cos    12   12   12   12  βs denotes parameters to be estimated; s = 1, 2, 3, 4 refer to the seasonality parameters (captured by the sin and cos functions). The network analysis component uses the adjustment coefficients from equations (1) and (2) as inputs to a network model, specifically computing the PageRank and betweenness measures—the former is discussed in detail while the latter is just reported in the results. The intention behind the PageRank measure can be understood by considering a naïve hypothetical measure of market linkages. Suppose we set the linkage between markets A and B to 1 if market A influences market B and zero otherwise. Then, we can construct a simple measure of systemic influence based on the number of markets that a given market influences. This hypothetical measure, however, suffers from three shortcomings. First, because this hypothetical measure takes only the value of one (if markets are linked) or zero (if markets are not linked), it does not consider the strength of the linkage. Second, the hypothetical measure assigns the same weight regardless of whether connections to this market influence other markets. To see this, consider four markets: A, B, C, and D. Then assume that market A influences market C and market B influences market D; market C does not influence any other market, while market D influences other markets. A naïve approach would assign the same importance to markets A and B. However, centrality measures (including the PageRank measure we employ here) account for the importance of the markets in the context of the network and, thus, assign a higher value to market B and a lower value to market A. Third, the importance of a market should be adjusted downward if a market influences a market that is also influenced by many other markets. In contrast to the naïve hypothetical measure, the contri- bution to the PageRank will be higher if a market influences a market with fewer other influences. In the context of the above four-market example (with the markets now playing a  different role), as- sume that A influences C and B influences D. Further, C is influenced by many other markets, but D is 54 © 2020 The World Bank Group Appendix 5. Bivariate Error Correction and Network Centrality Methodology only influenced by B, while both C and D have the same PageRank. Then, B will be assigned a higher PageRank than A. The first shortcoming is addressed by allowing the strength of each linkage to vary according to the size of the parameter estimate. The adjustment coefficients of the error-correction model, αij (0 < αij < 1), take the value of the corresponding parameter estimate if it is significantly different from zero at a 1 per- cent level of significance and zero elsewhere. The second and third shortcomings are addressed by using the PageRank measure. First, we begin with the matrix of all adjustment coefficient estimates of the error-correction model. Then, we construct a  matrix,, the elements of which (denoted as αij) are related to the adjustment coefficient matrix as follows. If αij is significantly different from zero at the 1 percent level of significance (which implies that market i is influenced by market j), the corresponding element of A, takes the value of αij. If αij is not significantly different from zero in the original matrix, then it takes a value of zero in A. Then, matrix A is adjusted as follows: If a row has at least one non-zero αij, the elements of that row are normalized to add up to 1. If all αij s in a row are zeros, then αij = 1/n, where n denotes the number of columns (cor- responding to number of markets). This new matrix, denoted by T, is the stochastic transition matrix. To ensure convergence, we follow Brin and Page (1998) and add (to the matrix T) a matrix in which every row adds up to 1 and every cell in the row has the same value. A dampening factor equal to 0.85 gives the weight (of the convex combination of the two matrices) assigned to the transition matrix T. Let S be the matrix created by this convex combination. The PageRank vector, denoted as PR, is estimated when convergence of the following equation is reached, PR(k+1) = PRkS, (3) where k denotes the number of iterations. We begin with an initial PR value that is the same for every market. The stochastic transition matrix T, and therefore the market link structure estimated using equa- tions (1) and (2), determines convergence to the final PR vector. The PR vector provides a synthesis of the full matrix of adjustment coefficients which reflects the under- lying network structure of the market system. Market systems with markets that are either not connect- ed or fully connected with each other will have every market receiving the same PR value while market systems with a few dominant markets will have a very unequal PR distribution. © 2020 The World Bank Group 55 Drivers of Food Price Inflation in Turkey Appendix 6. Domestic Market Linkages for Table Tomatoes, Green Peppers, and Dry Onions Table Tomatoes Green Peppers Dry Onions Regions Harvest Lean Harvest Lean Harvest Lean Adana 0.01 0.01 0.03 0.08 0.02 0.02 Agri 0.01 0.06 0.05 0.13 0.12 0.05 Ankara 0.06 0.05 0.04 0.01 0.12 0.07 Antalya 0.06 0.02 0.03 0.08 0.03 0.03 Aydin 0.01 0.01 0.03 0.01 0.02 0.02 Balikesir 0.06 0.01 0.15 0.05 0.02 0.02 Bursa 0.01 0.01 0.16 0.04 0.03 0.02 Erzurum 0.01 0.01 0.04 0.02 0.10 0.14 Gaziantep 0.01 0.01 0.05 0.01 0.02 0.03 Hatay 0.01 0.01 0.04 0.04 0.03 0.02 Istanbul 0.07 0.13 0.05 0.03 0.04 0.03 Izmir 0.01 0.01 0.05 0.06 0.03 0.03 Kastamonu 0.05 0.05 0.03 0.03 0.03 0.02 Kayseri 0.01 0.01 0.03 0.09 0.06 0.04 Kirikkale 0.01 0.01 0.02 0.06 0.03 0.03 Kocaeli 0.06 0.08 0.02 0.02 0.02 0.02 Konya 0.01 0.01 0.02 0.02 0.02 0.02 Malatya 0.01 0.01 0.02 0.06 0.02 0.02 Manisa 0.12 0.01 0.02 0.01 0.03 0.03 Mardin 0.01 0.01 0.02 0.02 0.02 0.02 Samsun 0.01 0.01 0.02 0.01 0.03 0.02 Sanliurfa 0.01 0.01 0.02 0.04 0.02 0.03 Tekirdag 0.32 0.31 0.02 0.01 0.02 0.02 Trabzon 0.01 0.03 0.02 0.02 0.03 0.02 Van 0.01 0.05 0.02 0.02 0.04 0.20 Zonguldak 0.10 0.13 0.02 0.02 0.03 0.02 Note: Numbers in the columns represent market i’s centrality in terms of its influence on other markets. 56 © 2020 The World Bank Group Appendix 7. Domestic Market Linkages for Table Tomatoes, Green Peppers, and Dry Onions – Spatial Analysis Appendix 7. Domestic Market Linkages for Table Tomatoes, Green Peppers, and Dry Onions – Spatial Analysis Figure A.7.1. Table tomatoes, Harvest season Tekirdağ Istanbul Zonguldak Kastamonu Trabzon Legend Kocaeli Samsun Ağrı Bursa Erzurum 0.01–0.10 Balıkesir Ankara Kayseri 0.11–0.15 Manisa Malatya >0.15 Kırıkkale Van Izmir Konya Aydın Hatay Gaziantep Şanlıurfa Mardin Antalya Adana Figure A.7.2. Table tomatoes, Lean season Tekirdağ Istanbul Zonguldak Kastamonu Trabzon Legend Kocaeli Samsun Ağrı Bursa Erzurum 0.01–0.10 Balıkesir Ankara Kayseri 0.11–0.15 Manisa Malatya >0.15 Kırıkkale Van Izmir Konya Aydın Hatay Gaziantep Şanlıurfa Mardin Antalya Adana Figure A.7.3. Green peppers, Harvest season Tekirdağ Istanbul Zonguldak Kastamonu Samsun Trabzon Legend Kocaeli Ağrı Bursa Erzurum 0.01–0.10 Balıkesir Ankara Kayseri 0.11–0.15 Manisa Malatya >0.15 Kırıkkale Van Izmir Konya Aydın Hatay Gaziantep Şanlıurfa Mardin Antalya Adana © 2020 The World Bank Group 57 Drivers of Food Price Inflation in Turkey Figure A.7.4. Green peppers, Lean season Tekirdağ Istanbul Zonguldak Kastamonu Trabzon Legend Kocaeli Samsun Ağrı Bursa Erzurum 0.01–0.10 Balıkesir Ankara Kayseri 0.11–0.15 Manisa Malatya >0.15 Kırıkkale Van Izmir Konya Aydın Hatay Gaziantep Şanlıurfa Mardin Antalya Adana Figure A.7.5. Dry onions, Harvest season Tekirdağ Istanbul Zonguldak Kastamonu Trabzon Legend Kocaeli Samsun Ağrı Bursa Erzurum 0.01–0.10 Balıkesir Ankara Kayseri 0.11–0.15 Manisa Malatya >0.15 Kırıkkale Van Izmir Konya Aydın Hatay Gaziantep Şanlıurfa Mardin Antalya Adana Figure A.7.6. Dry onions, Lean season Tekirdağ Istanbul Zonguldak Kastamonu Trabzon Legend Kocaeli Samsun Ağrı Bursa Erzurum 0.01–0.10 Balıkesir Ankara Kayseri 0.11–0.15 Manisa Malatya >0.15 Kırıkkale Van Izmir Konya Aydın Hatay Gaziantep Şanlıurfa Mardin Antalya Adana 58 © 2020 The World Bank Group Appendix 8. Production and Prices for Table Tomatoes, Green Peppers, and Dry Onions Across the Regions Appendix 8. Production and Prices for Table Tomatoes, Green Peppers, and Dry Onions Across the Regions Table A.8.1. Table tomatoes Price, TL/kg, Production, tons, Surplus or deficit Region Population, 2018 2005–201955 Volatility56 2005–2018 average region54 average Adana 4,045,211 991,193 Surplus 1.62 0.98 Ağrı 1,110,464 40,284 Deficit 2.17 1.13 Ankara 5,546,531 146,044 Deficit 2.00 1.13 Antalya 3,111,486 2,299,369 Surplus 1.95 1.05 Aydın 3,082,078 715,168 Surplus 1.89 1.02 Balıkesir 1,749,095 397,691 Surplus 2.05 1.15 Bursa 4,090,182 436,484 Surplus 2.02 1.09 Erzurum 1,073,727 77,852 Deficit 1.98 1.10 Gaziantep 2,806,058 45,976 Deficit 1.68 1.06 Hatay 3,271,927 239,961 Deficit 1.60 0.93 Istanbul 15,254,231 19,913 Deficit 2.23 1.12 Izmir 4,330,317 239,009 Deficit 1.95 1.17 Kastamonu 768,033 64,222 Deficit 2.09 1.16 Kayseri 2,429,092 43,328 Deficit 1.80 1.05 Kırıkkale 1,576,159 124,578 Surplus 1.77 1.01 Kocaeli 3,900,884 78,689 Deficit 2.06 1.10 Konya 2,454,474 182,279 Deficit 1.73 1.02 Malatya 1,739,325 68,751 Deficit 1.83 1.13 Manisa 3,088,210 272,534 Deficit 1.74 0.94 Mardin 2,254,061 39,473 Deficit 1.80 1.06 Samsun 2,785,880 819,801 Surplus 1.86 1.06 Şanlıurfa 3,763,301 215,857 Deficit 1.71 1.07 Tekirdağ 1,804,880 34,354 Deficit 2.13 1.19 Trabzon 2,648,868 14,136 Deficit 2.18 1.16 Van 2,143,427 75,936 Deficit 1.99 1.10 Zonguldak 1,039,320 20,407 Deficit 2.17 1.12 Estimated based on assumptions of national average consumption levels of 109.9 kg/person per year (2018 est.). 54 January 2005-May 2019. 55 Std. deviation. 56 © 2020 The World Bank Group 59 Drivers of Food Price Inflation in Turkey Table A.8.2. Green peppers Price, TL/kg, Production, tons, Surplus or deficit Region Population, 2018 2005–201958 average, Volatility59 2005–2018 average region57 sivri variety Adana 4,045,211 270,598 Surplus 2.91 1.79 Ağrı 1,110,464 1,169 Deficit 3.19 1.91 Ankara 5,546,531 8,764 Deficit 3.15 1.92 Antalya 3,111,486 319,844 Surplus 2.87 1.74 Aydın 3,082,078 59,560 Deficit 3.08 1.88 Balıkesir 1,749,095 206,415 Surplus 3.37 1.93 Bursa 4,090,182 182,539 Surplus 3.37 1.96 Erzurum 1,073,727 5,030 Deficit 3.02 1.89 Gaziantep 2,806,058 91,833 Surplus 2.88 2.20 Hatay 3,271,927 113,644 Surplus 2.89 1.77 Istanbul 15,254,231 2,270 Deficit 3.50 2.02 Izmir 4,330,317 116,752 Surplus 3.29 2.06 Kastamonu 768,033 12,454 Deficit 3.00 1.84 Kayseri 2,429,092 1,442 Deficit 2.81 1.79 Kırıkkale 1,576,159 6,922 Deficit 2.73 1.85 Kocaeli 3,900,884 21,698 Deficit 2.89 1.89 Konya 2,454,474 26,830 Deficit 2.72 1.87 Malatya 1,739,325 28,484 Deficit 2.54 1.81 Manisa 3,088,210 195,232 Surplus 2.73 1.79 Mardin 2,254,061 10,183 Deficit 2.64 1.89 Samsun 2,785,880 264,453 Surplus 2.93 1.89 Şanlıurfa 3,763,301 131,514 Surplus 2.69 1.86 Tekirdağ 1,804,880 13,116 Deficit 3.06 1.97 Trabzon 2,648,868 3,620 Deficit 3.05 1.93 Van 2,143,427 7,720 Deficit 2.65 1.72 Zonguldak 1,039,320 5,074 Deficit 3.11 1.88 57 Estimated based on assumptions of national average consumption levels of 25.3 kg/person per year (2018 est.) 58 January 2005-May 2019. 59 Std. deviation. 60 © 2020 The World Bank Group Appendix 8. Production and Prices for Table Tomatoes, Green Peppers, and Dry Onions Across the Regions Table A.8.3. Dry onions Price, TL/kg, Production, tons, Surplus or deficit Region Population, 2018 2005–201961 Volatility62 2005–2018 average region60 average Adana 4,045,211 178,273 Surplus 1.05 0.77 Ağrı 1,110,464 0 Deficit 1.33 0.85 Ankara 5,546,531 523,295 Surplus 1.13 0.76 Antalya 3,111,486 21,410 Deficit 1.32 0.95 Aydın 3,082,078 19,788 Deficit 1.24 0.93 Balıkesir 1,749,095 13,795 Deficit 1.46 1.02 Bursa 4,090,182 284,468 Surplus 1.38 0.95 Erzurum 1,073,727 289 Deficit 1.24 0.86 Gaziantep 2,806,058 41,059 Deficit 1.12 0.84 Hatay 3,271,927 217,583 Surplus 1.13 0.77 Istanbul 15,254,231 716 Deficit 1.34 0.89 Izmir 4,330,317 4,496 Deficit 1.29 0.96 Kastamonu 768,033 497 Deficit 1.22 0.88 Kayseri 2,429,092 28,409 Deficit 1.20 0.87 Kırıkkale 1,576,159 24,366 Deficit 1.13 0.80 Kocaeli 3,900,884 13,924 Deficit 1.30 0.87 Konya 2,454,474 97,732 Surplus 1.12 0.80 Malatya 1,739,325 6,304 Deficit 1.19 0.88 Manisa 3,088,210 58,542 Deficit 1.21 0.88 Mardin 2,254,061 946 Deficit 1.28 0.87 Samsun 2,785,880 572,915 Surplus 1.12 0.83 Şanlıurfa 3,763,301 20,422 Deficit 1.15 0.83 Tekirdağ 1,804,880 23,498 Deficit 1.38 0.91 Trabzon 2,648,868 270 Deficit 1.23 0.92 Van 2,143,427 22,446 Deficit 1.39 0.88 Zonguldak 1,039,320 468 Deficit 1.31 0.94 Estimated based on assumptions of national average consumption levels of 21.75 kg/person per year (2018 est.). 60 January 2005-May 2019. 61 Std. deviation. 62 © 2020 The World Bank Group 61 Drivers of Food Price Inflation in Turkey Appendix 9. Price Transmission Methodology Johansen’s Maximum Likelihood (ML) test (Johansen 1988) is commonly used to test for the presence of co-integrating vectors. To obtain the test results, we first specify the general VAR(k) model, where k is the number of lags: Pt = A0 + A1Pt–1 + ... + AkPt–k +ut t = 1, … , T, (1) where Pt is an n x 1 vector of prices, and A is the matrix of the coefficients to be estimated. This equa- tion is further converted into the following vector error correction model: ∆Pt = Π 0 + ΠPt −1 + Σik=−11Γi ∆ Pt −1 + θt (2) where Δ denotes first difference, Π0 = A0, Γi represents the dynamic effects, and Π captures the long-run effects of the analyzed series. The goal of the Johansen ML test is to estimate the rank of the Π matrix, which represents the number of co-integrating relationships. The residual-based test for cointegration, Engle-Granger (1987) procedure, consists of two steps. First, the long run relationship between the pairs of export log-prices is estimated as seen in the example of the relationship between Russian and US wheat prices: PtM1 = β 0 + β1 PtM2 + ε t (3) t M1 t β = β 0P+ Pwhere M1 ,1 β =PtM2 0 + βt1 P ε are prices t M2 + ε in t two selected markets. β0 is a constant, β1 stands for the price transmission elasticity, and εt is the error term. Second, unit-root tests (ADF, PP, and KPSS) are used to check if the residuals are stationary. Their stationarity would imply that analyzed series are cointegrated, i.e. they move together in the long-run. If two series are cointegrated, then the OLS estimators in (3) are super- consistent and can be used to characterize the series’ behavior. The major difference between the Johansen ML and Engle-Granger methods is that they require differ- ent model assumptions. The first one requires a normality assumption, while the latter one is insensitive to the distribution assumption. Therefore, one of the benefits of using the Engle-Granger method is in its relative efficiency over the Johansen ML test if normality does not hold. As to the benefits of using the Johansen ML method, it allows for obtaining more than one co-integrating relationship. The tests were used jointly to assess the robustness of the results. 62 © 2020 The World Bank Group Appendix 10. Price Transmission Analysis Results Appendix 10. Price Transmission Analysis Results Pairwise (Satellite to Istanbul) Cointegration Regression and Tests for Daily Wholesale Prices Satellite Intercept Slope R Johansen Market Estimate Std. Err. Estimate Std. Err. Square Trace63 Dry Onions Adana -0.1762 0.0161 0.8104*** 0.0195 0.70 10.19** Antalya 0.7543 0.0227 0.6496*** 0.0275 0.43 14.03** Manisa 0.5671 0.0164 0.7184*** 0.0198 0.64 15.11** Mersin 0.1037 0.0209 0.8357*** 0.0253 0.60 14.91** Aydin 0.9324 0.0165 0.6679*** 0.0200 0.61 13.50** Samsun 0.4398 0.0190 0.6044*** 0.0230 0.49 13.81** Green Peppers Adana -0.8659 0.0223 1.1243*** 0.0194 0.82 4.29 Antalya 0.4817 0.0191 0.6945*** 0.0166 0.71 7.44 Mersin -0.6098 0.0193 1.0870*** 0.0167 0.85 4.59 Aydin 0.2934 0.0394 0.7762*** 0.0342 0.41 13.35** Samsun 0.1188 0.0274 0.7948*** 0.0237 0.61 10.54** Table Tomatoes Adana -0.5265 0.0145 1.0349*** 0.0196 0.79 6.63 Amasya -0.1890 0.0156 1.0098*** 0.0212 0.76 6.90 Antalya 0.2951 0.0146 0.8197*** 0.0198 0.70 7.14 Manisa -0.4365 0.0142 1.0306*** 0.0193 0.80 6.47 Mersin 0.1970 0.0210 0.5991*** 0.0285 0.38 13.50** Aydin 0.1504 0.0208 0.7023*** 0.0282 0.46 10.79** Samsun 0.4372 0.0196 0.6348*** 0.0266 0.44 13.64** Note: * p < 0.10, ** p < 0.05, *** p < 0.01. The 0.05 critical value for the Johansen trace test is 7.50. 63 © 2020 The World Bank Group 63 Drivers of Food Price Inflation in Turkey Pairwise (Satellite to Ankara) Cointegration Regression and Tests for Monthly Consumer Prices Satellite Intercept Slope R Johansen Market Estimate Std. Err. Estimate Std. Err. Square Trace64 Table Tomatoes Istanbul 0.2163 0.0070 0.8582*** 0.0092 0.98 9.30** Izmir -0.0297 0.0100 0.9928*** 0.0130 0.97 10.74** Bursa 0.0838 0.0087 0.8976*** 0.0113 0.97 11.14** Konya -0.1507 0.0086 0.9980*** 0.0111 0.98 9.81** Antalya 0.0409 0.0097 0.9094*** 0.0126 0.97 9.24** Samsun -0.0438 0.0094 0.9572*** 0.0123 0.97 12.00** Trabzon 0.1744 0.0079 0.8842*** 0.0103 0.98 11.71** Van 0.0456 0.0100 0.9272*** 0.0131 0.97 10.75** Green Peppers Istanbul 0.2476 0.0149 0.8839*** 0.0125 0.97 18.49** Izmir 0.1344 0.0176 0.9192*** 0.0148 0.96 19.72** Bursa 0.2317 0.0139 0.8704*** 0.0117 0.97 13.57** Konya -0.1502 0.0354 0.9770*** 0.0296 0.88 15.45** Antalya 0.0000 0.0176 0.9239*** 0.0147 0.96 26.52** Samsun -0.0518 0.0290 0.9708*** 0.0242 0.92 18.91** Trabzon 0.0449 0.0270 0.9295*** 0.0226 0.92 17.50** Van -0.1013 0.0338 0.9253*** 0.0283 0.88 17.67** Dry Onions Istanbul 0.1800 0.0060 0.9504*** 0.0113 0.98 6.25 Izmir 0.1255 0.0073 0.9603*** 0.0137 0.97 8.10** Bursa 0.2072 0.0074 0.9505*** 0.0139 0.96 5.91 Konya -0.0061 0.0061 0.9864*** 0.0114 0.98 8.60** Antalya 0.1547 0.0085 0.9666*** 0.0160 0.96 6.40 Samsun -0.0267 0.0075 1.0268*** 0.0140 0.97 7.07 Trabzon 0.0677 0.0072 1.0220*** 0.0135 0.97 6.23 Van 0.2224 0.0107 0.9244*** 0.0200 0.93 4.81 Note: * p < 0.10, ** p < 0.05, *** p < 0.01. The 0.05 critical value for the Johansen trace test is 7.50. 64 64 © 2020 The World Bank Group