Hacettepe University Graduate School Of Social Sciences Department of Economics REAL BUSINESS CYCLE MODELS IN EMERGING ECONOMIES Oğuz Kaan KARAKOYUN Ph. D. Dissertation Ankara, 2024 REAL BUSINESS CYCLE MODELS IN EMERGING ECONOMIES Oğuz Kaan KARAKOYUN Hacettepe University Graduate School Of Social Sciences Department of Economics Ph. D. Dissertation Ankara, 2024 ACCEPTANCE AND APPROVAL The jury finds that Oğuz Kaan Karakoyun has on the date of 03/06/2024 successfully passed the defense examination and approves his Ph. D. Dissertation titled “Real Business Cycle Models In Emerging Economies”. Assoc. Prof. Dr., Şen Bilin NEYAPTI (Jury President) Assoc. Prof. Dr., Mustafa Aykut ATTAR (Main Adviser) Assist. Prof. Dr., Ömer Kağan PARMAKSIZ Prof. Dr., Bahar BAYRAKTAR SAĞLAM Prof. Dr., Ayşe Yasemin YALTA I agree that the signatures above belong to the faculty members listed. Uğur ÖMÜRGÖNÜLŞEN Graduate School Director YAYIMLAMA VE FİKRİ MÜLKİYET HAKLARI BEYANI Enstitü tarafından onaylanan lisansüstü tezimin/raporumun tamamını veya herhangi bir kısmını, basılı (kağıt) ve elektronik formatta arşivleme ve aşağıda verilen koşullarla kullanıma açma iznini Hacettepe Üniversitesine verdiğimi bildiririm. Bu izinle Üniversiteye verilen kullanım hakları dışındaki tüm fikri mülkiyet haklarım bende kalacak, tezimin tamamının ya da bir bölümünün gelecekteki çalışmalarda (makale, kitap, lisans ve patent vb.) kullanım hakları bana ait olacaktır. Tezin kendi orijinal çalışmam olduğunu, başkalarının haklarını ihlal etmediğimi ve tezimin tek yetkili sahibi olduğumu beyan ve taahhüt ederim. Tezimde yer alan telif hakkı bulunan ve sahiplerinden yazılı izin alınarak kullanılması zorunlu metinlerin yazılı izin alınarak kullandığımı ve istenildiğinde suretlerini Üniversiteye teslim etmeyi taahhüt ederim. Yükseköğretim Kurulu tarafından yayınlanan “Lisansüstü Tezlerin Elektronik Ortamda Toplanması, Düzenlenmesi ve Erişime Açılmasına İlişkin Yönerge” kapsamında tezim aşağıda belirtilen koşullar haricince YÖK Ulusal Tez Merkezi / H.Ü. Kütüphaneleri Açık Erişim Sisteminde erişime açılır. o Enstitü / Fakülte yönetim kurulu kararı ile tezimin erişime açılması mezuniyet tarihimden itibaren 2 yıl ertelenmiştir. (1) o Enstitü / Fakülte yönetim kurulunun gerekçeli kararı ile tezimin erişime açılması mezuniyet tarihimden itibaren ... ay ertelenmiştir. (2) o Tezimle ilgili gizlilik kararı verilmiştir. (3) 03/06/2024 Oğuz Kaan KARAKOYUN i 1“Lisansüstü Tezlerin Elektronik Ortamda Toplanması, Düzenlenmesi ve Erişime Açılmasına İlişkin Yönerge” (1) Madde 6. 1. Lisansüstü tezle ilgili patent başvurusu yapılması veya patent alma sürecinin devam etmesi durumunda, tez danışmanının önerisi ve enstitü anabilim dalının uygun görüşü üzerine enstitü veya fakülte yönetim kurulu iki yıl süre ile tezin erişime açılmasının ertelenmesine karar verebilir. (2) Madde 6. 2. Yeni teknik, materyal ve metotların kullanıldığı, henüz makaleye dönüşmemiş veya patent gibi yöntemlerle korunmamış ve internetten paylaşılması durumunda 3. şahıslara veya kurumlara haksız kazanç imkanı oluşturabilecek bilgi ve bulguları içeren tezler hakkında tez danışmanının önerisi ve enstitü anabilim dalının uygun görüşü üzerine enstitü veya fakülte yönetim kurulunun gerekçeli kararı ile altı ayı aşmamak üzere tezin erişime açılması engellenebilir. (3) Madde 7. 1. Ulusal çıkarları veya güvenliği ilgilendiren, emniyet, istihbarat, savunma ve güvenlik, sağlık vb. konulara ilişkin lisansüstü tezlerle ilgili gizlilik kararı, tezin yapıldığı kurum tarafından verilir *. Kurum ve kuruluşlarla yapılan işbirliği protokolü çerçevesinde hazırlanan lisansüstü tezlere ilişkin gizlilik kararı ise, ilgili kurum ve kuruluşun önerisi ile enstitü veya fakültenin uygun görüşü üzerine üniversite yönetim kurulu tarafından verilir. Gizlilik kararı verilen tezler Yükseköğretim Kuruluna bildirilir. Madde 7.2. Gizlilik kararı verilen tezler gizlilik süresince enstitü veya fakülte tarafından gizlilik kuralları çerçevesinde muhafaza edilir, gizlilik kararının kaldırılması halinde Tez Otomasyon Sistemine yüklenir * Tez danışmanının önerisi ve enstitü anabilim dalının uygun görüşü üzerine enstitü veya fakülte yönetim kurulu tarafından karar verilir. ETİK BEYAN Bu çalışmadaki bütün bilgi ve belgeleri akademik kurallar çerçevesinde elde ettiğimi, görsel, işitsel ve yazılı tüm bilgi ve sonuçları bilimsel ahlak kurallarına uygun olarak sunduğumu, kullandığım verilerde herhangi bir tahrifat yapmadığımı, yararlandığım kaynaklara bilimsel normlara uygun olarak atıfta bulunduğumu, tezimin kaynak gösterilen durumlar dışında özgün olduğunu, Tez Danışmanının Doç. Dr., Mustafa Aykut ATTAR danışmanlığında tarafımdan üretildiğini ve Hacettepe Üniversitesi Sosyal Bilimler Enstitüsü Tez Yazım Yönergesine göre yazıldığını beyan ederim. Oğuz Kaan KARAKOYUN iv ACKNOWLEDGEMENTS I would like to thank the members of my thesis committee, Assoc. Prof. Dr. Şen Bilin Neyaptı and Prof. Dr. Bahar Bayraktar Sağlam, for their valuable comments during the progress of this dissertation. I am also grateful to Prof. Dr. Ayşe Yasemin Yalta and Asst. Prof. Dr. Ömer Kağan Parmaksız for accepting to be in my dissertation examination committee and for their insightful comments. I would like to express my deepest gratitude to my advisor Assoc. Prof. Dr. Mustafa Aykut Attar for his guidance, understanding, encouragement, and endless support throughout my Ph.D. journey. I will never forget the time you dedicated to giving me feedback, the time you spent with me day and night, and the courage you provided as you pushed me to explore new ideas and research questions. After starting my professional career, I experienced numerous instances of motivational exhaustion. You always assisted me in regaining my composure and confidence. I am deeply grateful for your continuous encouragement, which has been similar to that of a member of my family, in addition to your role as an advisor. I would also like to express my gratitude to Nusret Doru for the discussions we have had, for the diverse academic perspectives you have provided me with, and for consistently putting forth his best effort when I required it. I am grateful to Dr. Emre Şahin for his unwavering support in all of my problems, irrespective of our closeness or distance. Some people are no different than your family. They make every effort to provide you with the assistance you require. I would like to extend my appreciation to Sinem Yıldırım, Egehan Eğitim, Günal Tarhan, Hilal Kaçmaz, Efe Yavuzer, and Cem Sevim for their unwavering support throughout my doctoral journey, which spanned multiple cities and countries. I am aware that they are unable to make any further sacrifices when I consider the ones they have already made. It’s a chance for me to have you in my life. During my five-year tenure in central banking, I acquired a wealth of knowledge. I had the opportunity to observe nearly every aspect of the central banking system, financial markets, and general economy. However, the friendships I formed were more valuable than the knowledge I acquired. I am grateful to all of my friends, with a particular mention to Mirsad Tugay Sipahioğlu, Ömer Eren Çatalçam, and Emre Karacabey, whose names I am unable to disclose. This journey would be v considerably more stressful and difficult without your assistance. I am appreciative of your assistance in making the hours we spent working together more tolerable. I trust that we will be able to preserve this friendship until the end of our lives. I am grateful to my closest friends Bekir Emre Körpe, Burak Türkileri, Burak Demirel, Caner Odabaş, Fehim Aksoy, Halil İbrahim Kuru, Hüseyin Gündoğan, Şeyma Betül Sercan, and Zeynep Yılmaz. I’m able to complete this journey because of your support, which enabled me to maintain my mental endurance. I cannot express my gratitude to my family enough. Their encouragement is beyond words. Since the very first day, they have maintained their confidence in me. During the most challenging times, they were always there for me. I am deeply thankful to Aslıhan Beyazoğlu, Aziz Talas, Bahittin Karakoyun, Enes Beyazoğlu, Fatma Yılmaz, Hanife Karakoyun, Sevgi Talas, Sevil Mete, and Simge Cankaya. Last but not least, I want to express my deepest gratitude and appreciation to my beloved wife, Asst. Prof. Dr. Ece Çiğdem Karakoyun. We grew up together. We dreamed together. We work hard together to make these dreams come true. Long distances sometimes separated us, and other difficulties arose. However, we have overcome all of our problems together and will continue to do so in the future. I could not have gotten to this point without you. You have always believed in me more than anyone else. You were always there for me, even during your darkest days. I can’t express how grateful I am to have you with me in our story so far. We are now on a new page in our story. Let us walk towards our dreams together. My dear wife, I love you with all of my heart. I wholeheartedly dedicate this thesis to you. I am grateful for your patience, support, encouragement, smiles, compassion, and faith. vi ABSTRACT KARAKOYUN, Oğuz Kaan. Real Business Cycle Models in Emerging Economies, Ph. D. Dissertation, Ankara, 2024. Emerging Market Economies (EMEs) exhibit different economic dynamics compared to developed markets. While several EMEs have strong growth due to their trade surplus, most of them rely on imported inputs for their production processes and face a significant amount of foreign debt. This thesis seeks to address two crucial research questions concerning the distinctive characteristics of EMEs. The primary objective is to investigate the origins and transmission mechanisms behind economic fluctuations in an EME characterized by a trade deficit and substantial foreign debt. The other objective is to identify the fundamental attributes of EMEs that rely on imported inputs, taking into account different sectors. Türkiye is an appropriate subject for the study because it has consistent trade deficits and a significant amount of foreign debt. In Chapter 1, we employ both a dynamic stochastic general equilibrium (DSGE) model and the Bayesian estimation technique using the Turkish data. The results indicate that Türkiye exhibits a higher degree of sensitivity to growth shocks. Furthermore, analyzing the fluctuations of the trade balance to output ratio reveals that country premium and domestic spending shock processes, both in the medium and long terms, account for a substantial portion of its fluctuations. The most significant finding is the model’s ability to accurately capture the fluctuations in Türkiye’s crisis periods (1994, 2001, and 2009). Over time, the production of final goods has become more reliant on imported inputs. The reasoning behind this situation led to the creation of a conceptual framework in Chapter 2, which includes imported inputs as a factor of production and distinguishes between different sectors. To the best of our knowledge, there is a scarcity of literature that combines imported inputs in production processes and sector differentiation within a theoretical model. The model’s findings indicate that there is an inverse relationship between non-tradable goods sector and the macroe- conomic variables associated with international trade. Moreover, country premium shock holds the highest significance in elucidating the macroeconomic fluctuations. Keywords Emerging Market Economy, Business Cycles, Dynamic Stochastic General Equilib- rium, Bayesian Estimation vii ÖZET KARAKOYUN, Oğuz Kaan. Yükselen Ekonomilerde Reel İş Çevrimi Modelleri, Doktora Tezi, Ankara, 2024. Yükselen Piyasa Ekonomileri (EME), gelişmiş piyasalara kıyasla daha farklı ekono- mik dinamikler sergilemektedir. Bazı EME’ler ticaret fazlaları nedeniyle güçlü bir büyüme kaydederken, çoğu üretim süreçlerinde ithal girdilere bağımlı ve önemli düzeyde dış borçla karşı karşıyadır. Bu tez, EME’lerin ayırt edici özelliklerini dikkate alarak iki önemli araştırma sorusunu ele almayı amaçlamaktadır. Temel amaç, ticaret açığı ve önemli miktarda dış borçla karakterize edilen bir EME’deki ekonomik dalgalanmaların kökenlerini ve aktarım mekanizmalarını araştırmaktır. Diğer amaç ise farklı sektörleri dikkate alarak, üretimleri ithal girdilere dayalı olan EME’lerin temel özelliklerini belirlemektir. Türkiye, istikrarlı ticaret açıkları ve önemli miktarda dış borcu olması nedeniyle bu çalışma için uygun bir öznedir. Bölüm 1’de hem dinamik stokastik genel denge (DSGE) modelini hem de Türkiye verilerini kullanan Bayesyen tahmin tekniğini kullanıyoruz. Sonuçlar, Türkiye’nin büyüme şoklarına karşı yüksek düzeyde has- sasiyet gösterdiğine işaret etmektedir. Ayrıca ticaret dengesinin çıktıya oranındaki dalgalanmalar incelendiğinde, hem orta hem de uzun vadede ülke primi ve yerel har- cama şoku süreçlerinin dalgalanmaların önemli bir kısmını oluşturduğu görülmekte- dir. En önemli bulgu, modelin Türkiye’nin 1994, 2001 ve 2009 kriz dönemlerindeki dalgalanmalarını doğru bir şekilde yakalayabilmesidir. Zamanla, nihai mal üretimi ithal girdilere daha bağımlı hale gelmektedir. Bu du- rumun ardındaki mantık, Bölüm 2’de ithal girdileri üretim faktörü olarak içeren ve farklı sektörler arasında ayrım yapan kavramsal bir çerçevenin oluşturulmasına yol açmıştır. Bildiğimiz kadarıyla, üretim süreçlerindeki ithal girdileri ve sektör farklılıklarını teorik bir model çerçevesinde birleştiren literatür eksikliği bulunmak- tadır. Modelin bulguları, ticarete konu olmayan mallar sektörü ile uluslararası ticare- tle ilişkili makroekonomik değişkenler arasında ters yönlü bir ilişki olduğunu göster- mektedir. Ayrıca, makroekonomik değişkenlerdeki dalgalanmaların aydınlatılmasın- da ülke primi şoku en açıklayıcı şok sürecidir. Anahtar Sözcükler Yükselen Piyasa Ekonomisi, İş Çevrimleri, Dinamik Stokastik Genel Denge, Bayesyen Tahminleme TABLE OF CONTENTS ACCEPTANCE AND APPROVAL . . . . . . . . . . . . . . . . . i YAYIMLAMA VE FİKRİ MÜLKİYET HAKLARI BEYANI . . . . ii ETİK BEYAN . . . . . . . . . . . . . . . . . . . . . . . . . . . iii ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . iv ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi ÖZET . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii LIST OF ABBREVIATIONS . . . . . . . . . . . . . . . . . . . . xiv INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . 1 CHAPTER 1: UNDERSTANDING REAL BUSINESS CYCLES IN TÜRKİYE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 1.1. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . 10 1.2. MODEL . . . . . . . . . . . . . . . . . . . . . . . . . . 15 1.3. DATA AND CALIBRATION . . . . . . . . . . . . . . . 18 1.4. ESTIMATION . . . . . . . . . . . . . . . . . . . . . . . 20 1.5. MAIN RESULTS . . . . . . . . . . . . . . . . . . . . . 26 1.5.1. Second Moments . . . . . . . . . . . . . . . . . . . . 26 1.5.2. Impulse Response Functions . . . . . . . . . . . . . . . 27 1.5.3. Conditional Variance Decomposition . . . . . . . . . . . . 32 1.6. DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . 35 1.6.1. Türkiye’s Crisis Experiences . . . . . . . . . . . . . . . 35 1.6.2. Türkiye And Emerging Markets Comparison . . . . . . . . 41 1.6.2.1. Data Facts . . . . . . . . . . . . . . . . . . . . 41 1.6.2.2. Model Comparisons . . . . . . . . . . . . . . . . 47 1.6.3. Domestic and Foreign Shock Effects . . . . . . . . . . . . 49 1.7. CONCLUSION . . . . . . . . . . . . . . . . . . . . . . 51 CHAPTER 2: NON-TRADABLE GOODS AND IMPORTED INPUTS IN AN EMERGING MARKET ECONOMY . . . . . . . 54 2.1. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . 54 2.2. MODEL ECONOMY . . . . . . . . . . . . . . . . . . . 61 2.2.1. Environment . . . . . . . . . . . . . . . . . . . . . . 61 2.2.2. Decision Problems . . . . . . . . . . . . . . . . . . . . 63 2.2.2.1. Households . . . . . . . . . . . . . . . . . . . . 63 2.2.2.2. Firms . . . . . . . . . . . . . . . . . . . . . . 67 2.2.2.2.1. Final Goods Producer . . . . . . . . . . . . . 67 2.2.2.2.2. Non-tradable Goods Producer . . . . . . . . . 69 2.2.2.2.3. Tradable Goods Producer . . . . . . . . . . . 71 2.2.3. Rest of the World . . . . . . . . . . . . . . . . . . . . 73 2.2.4. Market Clearing Conditions . . . . . . . . . . . . . . . . 74 2.3. CALIBRATION . . . . . . . . . . . . . . . . . . . . . . 76 2.4. MAIN RESULTS . . . . . . . . . . . . . . . . . . . . . 79 2.4.1. Steady State Solutions . . . . . . . . . . . . . . . . . . 79 2.4.2. Second Moments Analysis . . . . . . . . . . . . . . . . 81 2.4.3. Impulse Response Functions . . . . . . . . . . . . . . . 84 2.4.3.1. Final Goods Technology Shock. . . . . . . . . . . . 85 2.4.3.2. Non-Tradable Goods Technology Shock . . . . . . . . 86 2.4.3.3. Tradable Goods Technology Shock . . . . . . . . . . 87 2.4.3.4. Intertemporal Preference Shock . . . . . . . . . . . 88 2.4.3.5. Country Premium Shock . . . . . . . . . . . . . . 89 2.4.3.6. Foreign Price Shock . . . . . . . . . . . . . . . . 90 2.4.3.7. Imported Input Price in Tradable Goods Sector Shock. . 92 2.4.3.8. Imported Input Price in Non-Tradable Goods Sector Shock . . . . . . . . . . . . . . . . . . . . . . . . . . 93 2.4.4. Variance Decomposition Analysis . . . . . . . . . . . . . 95 2.5. ROBUSTNESS . . . . . . . . . . . . . . . . . . . . . . 96 2.6. CONCLUSION . . . . . . . . . . . . . . . . . . . . . .101 CONCLUSION. . . . . . . . . . . . . . . . . . . . . . . . . . .105 BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . . . . .110 APPENDIX 1 . . . . . . . . . . . . . . . . . . . . . . . . . . .116 1A. STEADY STATE CONDITIONS . . . . . . . . . . . . . .116 1B. IDENTIFICATION TEST RESULTS . . . . . . . . . . . .118 1C. MULTIVARIATE MCMC CONVERGENCE DIAGNOSTICS . . . . . . . . . . . . . . . . . . . . . . . .119 1D. UNIVARIATE MCMC CONVERGENCE DIAGNOSTICS .120 APPENDIX 2 . . . . . . . . . . . . . . . . . . . . . . . . . . .123 2A. LIST OF COUNTRIES AND FOREIGN VALUE-ADDED SHARE OF EXPORTS . . . . . . . . . . . . . . . . . . . . .123 2B. STEADY STATE CONDITIONS . . . . . . . . . . . . . .124 APPENDIX 3: ETHICS COMMISSION FORM . . . . . . . . . .129 APPENDIX 4: ORIGINALITY REPORT . . . . . . . . . . . . .131 xi TABLES 1 The Calibrated Parameters from the Relevant Literature . . 18 2 The Calibrated Parameters Using the Turkish Data . . . . . . 19 3 Summary Statistics of the Selected Variables (in Percent) . . 21 4 Prior and Posterior Distributions . . . . . . . . . . . . . . . . . 24 5 Second Moments of Data and Model (in Percent) . . . . . . . 27 6 Conditional Variance Decomposition . . . . . . . . . . . . . . . 33 7 Historical Growth Experiences for Türkiye, Argentina, and Mexico over Different Time Periods (in Percent) . . . . . . . . 45 8 Summary Information of the Studies Being Compared . . . . 47 9 Comparisons of Different Model Results . . . . . . . . . . . . . 48 10 Calibrated Parameters of the Model . . . . . . . . . . . . . . . . 77 11 Comparison between Steady State Solutions and Turkish Data 81 12 Model Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 13 Variance Decomposition of the Model (in Percent) . . . . . . 95 14 Foreign Value-Added Share of Exports in 1995 and 2020 . . . 123 xii FIGURES 1 Historical Data of Türkiye: 1951-2019 (in Percent) . . . . . . 22 2 BIRFs of the Selected Variables to the Technology Shock (at) 28 3 BIRFs of the Selected Variables to the Growth Shock (gt) . . 29 4 BIRFs of the Selected Variables to the Preference Shock . . 30 5 BIRFs of the Selected Variables to the Contry Premium Shock 31 6 BIRFs of the Selected Variables to the Domestic Spending Shock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 7 The Trajectory of Output Growth in Türkiye During the Years of Crisis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 8 The Trajectory of the Trade Balance to Output Ratio of Türkiye During the Years of Crisis . . . . . . . . . . . . . . . . 37 9 The Trajectory of the Technology Shock (at) During the Years of Crisis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 10 The Trajectory of the Growth Shock (gt) During the Years of Crisis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 11 The Trajectory of the Country Premium Shock (µt) During the Years of Crisis . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 12 Log-linear Quadratic Trends and Business Cycles for Ar- gentina, Mexico, and Türkiye . . . . . . . . . . . . . . . . . . . . 42 13 The Ratio of Türkiye’s Real GDP Per Capita to That of Argentina and Mexico . . . . . . . . . . . . . . . . . . . . . . . . 44 14 Import Content of Exports for the Selected Countries . . . . 57 15 Illustration of the Interrelationships of the Agents . . . . . . . 62 16 IRFs of the Variables to the Final Goods Technology Shock . 85 17 IRFs of the Variables to the Non-tradable Goods Technology Shock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 18 IRFs of the Variables to the Tradable Goods Technology Shock 88 19 IRFs of the Variables to Intertemporal Preference Shock . . 89 20 IRFs of the Variables to Country Premium Shock . . . . . . . 90 21 IRFs of the Variables to Foreign Price Shock . . . . . . . . . . 91 22 IRFs of the Variables to Imported Input Price in Tradable Goods Sector Shock . . . . . . . . . . . . . . . . . . . . . . . . . . 92 23 IRFs of the Variables to Imported Input Price in Non-Tradable Goods Sector Shock . . . . . . . . . . . . . . . . . . . . . . . . . . 94 24 IRFs under Different Interest Rate Definitions: Country Premium Shock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 25 IRFs under Different Interest Rate Definitions: Foreign Price Shock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 26 IRFs under Different Interest Rate Definitions: Final Goods Technology Shock . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 27 IRFs under Different Shock Persistency Parameters: Coun- try Premium Shock . . . . . . . . . . . . . . . . . . . . . . . . . . 100 28 IRFs under Different Shock Persistency Parameters: Foreign Price Shock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 xiii 29 IRFs under Different Shock Persistency Parameters: Final Goods Technology Shock . . . . . . . . . . . . . . . . . . . . . . . 101 30 Identification Test Results . . . . . . . . . . . . . . . . . . . . . . 118 31 Multivariate MCMC Convergence Diagnostics . . . . . . . . . 119 32 Univariate MCMC Convergence Diagnostics (a) . . . . . . . . 120 33 Univariate MCMC Convergence Diagnostics (b) . . . . . . . . 121 34 Univariate MCMC Convergence Diagnostics (c) . . . . . . . . 121 35 Univariate MCMC Convergence Diagnostics (d) . . . . . . . . 122 36 Univariate MCMC Convergence Diagnostics (e) . . . . . . . . 122 xiv LIST OF ABBREVIATIONS AR Autoregressive Process BC Business Cycles BIRF Bayesian Impulse Response Functions BML Bayesian Maximum Likelihood Estimating CAY Current Account to Output Ratio CDS Credit Default Swap DSGE Dynamic Stochastic General Equilibrium EME Emerging Market Economy EME-RBC Emerging Market Economies - Real Business Cycles EMEs Emerging Market Economies FOCs First Order Conditions GDP Gross Domestic Product GHH Greenwood Hercowitz Huffman Preference GMM Generalized Method of Moments GNI Gross National Income HP Hodrick-Prescott IMF the International Monetary Fund IRF Impulse Response Functions LOP Law of One Price MCMC Markov Chain Monte Carlo ML Maximum Likelihood OECD the Organisation for Economic Co-operation and Development OLS Ordinary Least Squares PPP Purchasing Power Parity PWT Penn World Table RBC Real Business Cycles SOE Small Open Economy SS Steady State TBY Trade Balance to Output Ratio TUIK Turkish Statistical Institute US the United States VAR Vector Autoregression WB World Bank WBD World Bank Database 1 INTRODUCTION “The good thing about science is that it’s true whether or not you believe in it.” - Neil deGrasse Tyson Macroeconomic research has traditionally placed significant emphasis on the analysis of business cycles, which are the recurrent fluctuations in economic activity marked by periods of expansion and contraction. Gaining a comprehensive understanding of the fundamental factors that cause these variations is of the utmost significance for policymakers, economists, and market participants alike. This understanding of- fers valuable insights into the mechanisms behind economic growth, unemployment, inflation, and financial stability. The Real Business Cycle (RBC) theory, a major framework in macroeconomics, provides a theoretical perspective for analyzing the origins and transmission mechanisms of business cycles.1 Emerging Market Economies (EMEs) are essential contributors to the global econ- omy since they stimulate development, foster innovation, and provide investment opportunities. Nevertheless, these countries frequently have distinctive obstacles such as trade deficits and foreign debt, which can have a substantial influence on their economic performance and stability. In this thesis, Chapter 1 seeks to analyze the fluctuations in economic activity in EMEs that are characterized by both nega- tive trade balances and external debt, with a specific focus on the case of Türkiye. In essence, the primary question driving Chapter 1’s investigation is: What are the mechanisms underlying the fluctuations in economic activity in EMEs that exhibit both a trade deficit and foreign debt, with a specific focus on the case of Türkiye? This research question aims to explore the complex interplay between trade imbal- ances, foreign debt, and economic fluctuations in the Turkish economy. Türkiye is a suitable subject for the study because it is a notable EME with per- sistent trade deficits and a significant amount of foreign debt. Through an analysis of Türkiye’s experience, our objective is to acquire valuable knowledge on general trends and dynamics that could be applicable to other EMEs encountering compa- rable difficulties. 1See Stockman (1988). 2 Chapter 1 provides significant advancements in macroeconomic modeling, specifi- cally within the framework of EMEs. Initially, Chapter 1 introduces a modified Dynamic Stochastic General Equilibrium (DSGE) model, specifically tailored to ac- curately depict the various economic dynamics of Türkiye. This study improves the representation of Türkiye’s economic conditions by incorporating trade deficits and foreign debt into the model, leading to a more comprehensive and precise analysis compared to previous models. Furthermore, Chapter 1 presents innovative and current estimation results, address- ing a significant deficiency in the current body of research. Currently, there is a dearth of research that employs data from Türkiye to compute a DSGE model. Our study findings demonstrate that growth shocks are a crucial factor in driving eco- nomic fluctuations in Türkiye. These findings offer policymakers useful information that can guide their decision-making process. The focus on Türkiye, along with the concurrent analysis of trade imbalances and foreign debt, indicates significant advancement in the use and assessment of DSGE models for EMEs. Over time, growing globalization and technological advancements have altered coun- tries’ industrial structures. The production of final goods is increasingly dependent on both domestic resources and imported inputs. This rationale led to the establish- ment of a theoretical framework in Chapter 2, which incorporates imported inputs as a production factor and sector differentiation. As far as we know, there is a dearth of existing literature that incorporates both imported inputs into production processes and sector differentiation within a theoretical model. Integrating imported goods into production processes is essential due to the sub- stantial amount of imported inputs in EMEs. By incorporating imported inputs into the model, the domestic economy forms a connection with the global economy, not only through the final goods but also through the use of inputs. Thus, we il- lustrate the influence of fluctuations in global prices and interest rates on domestic production through different transmission mechanisms. Moreover, it is feasible to accurately examine the shock dynamics that may occur in imported input prices. Furthermore, this method enables the expansion and improved understanding of the various channels through which changes in the external environment, such as shifts in interest rates and price levels, can impact the domestic economy. Chapter 2 of the thesis aims to examine the economic dynamics of EMEs affected by imported inputs and sector differentiation, with a particular emphasis on the 3 non-tradable and tradable goods sectors. Chapter 2 also conducts an examination of the non-tradable and tradable goods sectors, comparing their characteristics and performance. Chapter 2 aims to explore EMEs’ economic structure, investigating how fluctuations across different sectors and the utilization of imported inputs influ- ence their performance. The study attempts to analyze the impact of non-tradable and tradable goods sectors on different parts of the economy. Integrating imported inputs into the production function of an EME’s DSGE model significantly improves the model’s realism and analytical capabilities. EMEs often have robust ties to global value chains, relying on imported intermediate goods and raw materials to conduct their production processes. By incorporating imported inputs, the model achieves a more accurate depiction of the actual price framework and interdependencies within the economy, highlighting the impacts of fluctuations in exchange rates and disturbances in the global supply chain on domestic output. This inclusion also facilitates a comprehensive analysis of trade policies, such as tariffs and trade agreements, and their influence on economic activity. Moreover, it enhances our comprehension of how exogenous disruptions, such as changes in worldwide demand and fluctuations in commodity prices, propagate throughout the economy. The inclusion of imported inputs enhances the evaluation of monetary and fiscal policies by highlighting their impact on production costs and inflation. To summarize, this modification improves the accuracy of macroeconomic predictions and the effectiveness of policy simulations, boosting the DSGE model’s value as a tool for policymakers in EMEs. Prior to discussing the approaches employed in both Chapters 1 and 2, it would be beneficial to provide some background information on the RBC theory. The funda- mental premise of the RBC theory is that fluctuations in overall economic activity are predominantly caused by external disturbances to technology or productivity, rather than by changes in monetary policy or other nominal factors.2 The RBC theory, created by Finn Kydland and Edward Prescott in the early 1980s, diverges from classic Keynesian methods of macroeconomic research. The RBC theory fo- cuses on the influence of aggregate demand and nominal rigidities on stimulating business cycles.3 The RBC theory differs from traditional Keynesian macroeconomic models by its 2See King et al. (1988) and Hansen (1985). 3See Gali (1999), McCollum (1999) and Kydland and Prescott (1990). 4 reliance on microeconomic principles rather than ad hoc aggregate behavioral as- sumptions.4 The RBC models presume agents to possess forward-looking skills and rationality,5 enabling them to make optimal decisions that aim to maximize their utility or profit, while taking into account budgetary and technological constraints. This microeconomic foundation allows for a thorough examination of the conse- quences of different policy regimes and institutional arrangements. Another notable feature of the RBC theory is the postulation of adaptable prices and wages. In the RBC models, prices and wages exhibit high flexibility, promptly adapting to variations in supply and demand conditions, thereby guaranteeing the equilibrium of markets and the efficiency of resource allocation.6 This assumption is in direct opposition to the Keynesian models, which emphasize the significant impact of nominal rigidities, such as rigid prices and wages, on causing fluctuations in the business cycles. The work of Lucas (1977) is a significant and influential contribution to the field of macroeconomic theory. This work presents fundamental ideas such as rational expectations and endogenous fluctuations, which fundamentally alter economists’ understanding and examination of the dynamics of business cycles. These concepts form the foundation of the RBC theory. Kydland and Prescott (1982) focus on the concept of “time to build,” which denotes the delay between making investment decisions and finalizing capital goods, as a pi- oneering work of the RBC theory. They demonstrate that their model is capable of reproducing important empirical patterns observed in business cycles. These pat- terns include the enduring nature of output fluctuations, as well as the relationship between investment and output. Additional novel studies on the topic include those by Christiano et al. (1999), Eichenbaum (1995), Gaĺı (1999), and Long and Plosser (1983). In contrast to the traditional approach, Hodrick and Prescott (1997) employ a method that divides the time series into two distinct components: cycle and trend. The Hodrick-Prescott (HP) filter, or other detrending techniques, have gained wide- spread recognition for the analysis of time series data in business cycle studies since their 1997 publication. 4See Tobin (1992). 5See Gottschalk (2005). 6See Ball et al. (1988). 5 The origins of the RBC theory can be traced back to the analysis of complex indus- trialized economies, but its application is not limited to these particular contexts. There has been an increasing interest in utilizing RBC models for EMEs because of their dependence on international trade and capital movements. These economies encounter different problems and prospects in handling changes in the business cy- cles, such as susceptibility to exogenous shocks, restricted policy independence, and structural vulnerabilities.7 Applying the RBC theory to EMEs involves numerous complicated factors to con- sider. This statement encourages a thorough analysis of the degree to which the fundamental assumptions of the RBC theory continue to hold true, especially con- sidering the existence of international connections and trade openness. Furthermore, the inclusion of global factors, such as fluctuations in currency rates and movements of capital, in the RBC models prompts important inquiries on the consequences for comprehending the complexities of business cycle dynamics. Moreover, policymakers in EMEs encounter the difficult responsibility of managing the compromises between the goals of maintaining stability within the country and the limitations imposed by external factors. In order to achieve economic resilience and sustainability, it is imperative to efficiently manage the competing priorities at hand. This thesis, from a different point of view, aims to address these issues by employing two distinct models across two separate chapters. A trade deficit arises when a country’s imports exceed its exports, leading to a net outflow of goods and services. Türkiye’s ongoing trade deficits have generated wor- ries about the long-term sustainability of its economic development. Chapter 1 will examine the role of trade deficits in the fluctuation of Türkiye’s business cycles, im- pacting important economic variables including the Gross Domestic Product (GDP) growth, investment growth, and consumption growth. Foreign debt refers to the total financial obligations that a country has to repay to international lenders. Türkiye accumulates a substantial amount of foreign debt over time,8 raising concerns about its ability to repay the loan and its susceptibility to external shocks. Chapter 1 will analyze the effects of foreign debt on Türkiye’s business cycles, investigating how debt dynamics interact with other macroeconomic factors to influence economic performance. 7See Neumeyer and Perri (2005), Uribe and Yue (2006), Aguiar and Gopinath (2007) and Garcia-Cicco et al. (2010). 8Türkiye’s external debt to GDP ratio was 51% in 2022, according to the World Bank dataset. 6 Ultimately, comprehending the intricacies of economic fluctuations in EMEs such as Türkiye is of utmost importance for policymakers, investors, and researchers who aim to navigate the complexities of global economic patterns. The primary objective of Chapter 1 is to provide insight into the difficulties and potential advantages that EMEs encounter in a world that is becoming more interconnected. This will be achieved by analyzing the relationship between trade deficits, foreign debt, and economic fluctuations. Chapter 1 uses a DSGE model as a methodology to focus on the Turkish economy. The existing body of literature encompasses a multitude of empirical and theoretical investigations pertaining to EMEs, including Argentina and Mexico. Specifically, there is a scarcity of studies that utilize the estimation approach with Turkish data. This is a novel contribution to the limited body of research on this topic. Chapter 1 builds the model based on the work of Garcia-Cicco et al. (2010). The country premium shock in the model has been altered, in contrast to the initial study on the Argentinian economy. The Turkish economy’s past trade deficit prompts an adjustment in the level of long-term external borrowing to equalize the trade balance. In summary, we have adapted the model to accommodate the coexistence of external debt and the trade deficit. The majority of the model parameters are calibrated using relevant literature and parameter values derived from the original study. Using annual Turkish data, we calibrate the other parameters, such as the proportion of government expenditures in the GDP and the ratio of long-term external debt to GDP. The calibration process utilizes two datasets, the Penn World Table (PWT) and the World Bank Dataset (WBD), from 1950 to 2019. We then proceed with the model estimation process. We employ the Bayesian estimation method (BML) as the chosen approach to estimate the non-observable parameters from the data, including the persistency and volatility parameters of the shock processes. The estimation results align with the previous investigations documented in the existing literature. Chapter 1 also examines the growth experiences of selected EMEs, namely Argentina and Mexico, and Türkiye during the past seven decades. Based on the data, it is evident that Türkiye experiences significant volatility when it experiences relatively high growth. This condition implies that the growth shocks have a greater impact 7 on Türkiye than on Argentina. Chapter 1 examines this particular implication. Additionally, we analyze the model findings using Impulse Response Function (IRF) analysis and the conditional variance decomposition. We identify the role of growth and technology shocks in explaining fluctuations in output growth. Furthermore, we find that both growth and country premium shock processes have a substantial role in explaining the fluctuations of the trade balance to output ratio. To perform a more comprehensive inquiry, we examine the impact of the domestic crises that occurred in Türkiye in 1994 and 2001, as well as the global financial crisis of 2008, on the country. In Chapter 1, the estimation approach reveals the connection between the shock variables, which are the technology, growth, and country premium shocks, and the selected variables, which are output growth and the trade balance to output ratio, during the years of crisis of Türkiye. We closely observe the three-year period before and after these three crises, which reach their peak intensity in the designated year zero. The shock variables derived from the model accurately portray Türkiye’s eco- nomic conditions during years of crisis. These findings represent the most significant contribution of Chapter 1 to the literature. In Chapter 2, the main objective is to introduce a benchmark model that incorpo- rates imported inputs and sector differentiation while also accurately representing the distinctive structure of EMEs. Chapter 2 constructs a DSGE model, which includes households contributing labor and capital to various sectors and firms pro- ducing final goods using labor, capital, and imported inputs. Furthermore, borrow- ing from external sources and importing inputs creates a connection with the rest of the world. Hence, there is an opportunity to analyze the impacts of fluctuations in exchange rates that link the domestic economy with the global markets. Sector differentiation refers to the existence of different sectors within the economy, each with its own set of characteristics and dynamics. Within EMEs, the non- tradable goods sector generally includes industries that largely serve the domestic market, including retail, construction, and services. Conversely, the tradable goods sector encompasses industries engaged in global trade, such as agriculture and man- ufacturing. Comprehending the complex nature of these industries is crucial for assessing their impact on economic growth. Imported inputs are essential in the production process of both the non-tradable 8 and tradable goods sectors in EMEs. The inputs might consist of raw materials, intermediate goods, and capital equipment, which are crucial for sustaining compet- itiveness and efficiency. Nevertheless, EMEs’ dependence on imported inputs also makes them vulnerable to other risks, including supply chain disruptions, volatility in exchange rates, and trade barriers. These risks have the potential to impact their economic stability. In contrast to contemporary works, the model incorporates exogenous foreign price shocks. This shock process refers to sudden changes that might happen in the prices of tradable goods in international marketplaces. It is crucial for EMEs that lack the authority to set global prices to analyze the impacts of this shock process. Furthermore, the model includes exogenous shocks to the price levels of imported inputs, which distinguishes it from the existing literature. We calibrate the model’s parameters following the relevant literature, and we also compute the model’s steady state solutions. We compare the obtained results with data from Türkiye as an indicator of an EME. Finally, we apply IRFs and variance decomposition analyses to examine the model results. We dedicate the final section of Chapter 2 to performing robustness tests. Given the impact of external circumstances on the domestic economy, we employ various sce- narios to determine the foreign interest rate. We analyze the IRFs for each scenario, where the interest rate is defined as debt-elastic, income-elastic, and exogenously given. The findings remain mostly unaffected by changes in interest rate definitions. Ultimately, we validate the model outcomes by arbitrarily selecting shock persistency parameters across multiple levels. Changing the persistency parameters has no impact on the model outcomes, despite alterations in the response values of the chosen variables to the shock processes and the duration of the shock processes’ influence. To summarize, this thesis seeks to address the existing gap in the literature regard- ing two significant issues. Initially, we analyze the origins and aspects of economic fluctuations in Türkiye or a similar EME with characteristics resembling those of Türkiye, such as a negative balance of trade and indebtedness to foreign entities. Next, we examine the impact of utilizing imported inputs in the production pro- cesses of an EME with sector differentiation. In addition to these two significant foundational contributions, we present novel shock processes, such as foreign price and imported input prices shock processes. 9 The thesis is organized in the following manner. Chapter 1 presents a DSGE model that has been constructed in alignment with the economic structure of Türkiye. Chapter 1 introduces the model parameter calibration and the estimation approaches. The findings will be analyzed using the second moments, IRFs, and the condi- tional variance decomposition analyses. We also compare the growth experiences of Türkiye and some selected EMEs such as Argentina and Mexico over the past 70 years by dividing them into sub-periods. Finally, we analyze the fluctuations in output growth and the trade balance to output ratio throughout the crisis periods of 1994, 2001, and 2008, as well as the shock processes derived from the model. Chapter 2 presents a DSGE model that incorporates both non-tradable and tradable goods sectors. Furthermore, the production of these sectors involves the utilization of imported inputs. Next, we focus on the calibration of the parameters and the subsequent analyses of the obtained results. Furthermore, Chapter 2 incorporates a range of rigorous tests to assess the robustness of the findings. Ultimately, the final chapter is the concluding remarks of the thesis. 10 CHAPTER 1 UNDERSTANDING REAL BUSINESS CYCLES IN TÜRKİYE 1.1. INTRODUCTION Emerging market economies (EMEs) are vital to the global economy, but their dis- tinct features present difficulties for conventional macroeconomic analysis. Their openness to engage in international trade makes them vulnerable to variations in worldwide demand, unexpected changes in trade conditions, and the movement of capital, all of which can have a substantial impact on domestic economic activity.9 Furthermore, their limited capacity to implement autonomous monetary and fiscal measures renders them susceptible to external disturbances and requires a reassess- ment of conventional macroeconomic frameworks.10 The literature started analyzing EMEs in the early 1990s and later compared them to developed countries. Mendoza (1991), Correia et al. (1995), and Agenor et al. (2000) are notable examples of pioneering studies in the field. Mendoza’s (1991) model accurately reflects the observed positive relationship between savings and investment, despite the complete mobility of financial capital and the countercyclical fluctuations in external trade. The findings indicate that fluctuations in “terms-of- trade” have a significant impact, explaining approximately 50% of the observed fluctuations in GDP. Correia et al. (1995) demonstrate that a simple type of time-separable preferences can accurately explain the cyclical fluctuations in the components of the national income account identity, as well as the countercyclical nature of the trade balance. Agenor et al. (2000) demonstrate numerous similarities in macroeconomic fluctua- tions between EMEs and advanced economies, such as the correlation of real wages with business cycles and the inverse relationship between government expenditures and economic fluctuations. However, there are also notable distinctions, such as the inverse relationship between the velocity of monetary aggregates and macroeco- nomic fluctuations of output. 9See Kose (2002). 10See Bauducco and Caprioli (2014) and Gali and Monacelli (2005). 11 Researchers are conducting a significant number of investigations to examine the origins of fluctuations in EMEs. Changes in global interest rate or global price level directly impact EMEs. Blankenau et al. (2001) claim that real interest rate shocks have the ability to account for approximately 33% of the fluctuations in output and over 50% of the fluctuations in net foreign assets and net exports. Kose (2002) also finds that changes in global prices are a major contributor to fluctuations in economic activity in EMEs. In general, early studies about EMEs lack financial rigidity. Furthermore, they have neglected to demonstrate numerous stylized phenomena evident in the data, such as the significant fluctuations in consumption and the negative correlation between GDP with the trade balance. Since the early 2000s, there has been a substantial increase in the number of successful studies reflecting the features of actual data. Neumeyer and Perri (2005) assert that global interest rates display a countercyclical pattern and serve as a leading indicator of the economic cycle. Nevertheless, Uribe and Yue (2006) have criticized Neumeyer and Perri’s study’s assumption that the United States (US) interest rate and the country spread follow a first-order and bivariate autoregressive process. Their primary finding indicates that the US inter- est rate shocks contribute to approximately 20% of the fluctuations in the overall economic activity of EMEs. According to Aguiar and Gopinath’s (2007) assertion, the main cause of fluctuations in EMEs is shocks to long-term economic growth, as opposed to short-term variations around a stable trend. In the literature, this argument is commonly known as “the cycle is the trend”. However, Garcia-Cicco et al. (2010) construct an extended model that incorporates financial frictions to demonstrate that the long-term Argentinian data does not support Aguiar and Gopinath’s (2007) findings. Besides, their findings corroborate Uribe and Yue’s (2006) research. Chang and Fernandez (2013) argue that trend shocks are insignificant in elucidating macroeconomic fluctuations. Their model explains overall changes by emphasizing the significant impact of financial frictions on typical transitory productivity shocks, whereas trend shocks have a minimal effect. Boz et al. (2012) corroborate the findings of Chang and Fernandez (2013), but their theory is based on learning effects rather than the financial frictions. Both studies use data from Mexico, but for different time periods. 12 A significant amount of research in the literature is dedicated to studying EMEs and their business cycles. Scholars have conducted numerous important studies, particularly focusing on countries such as Argentina and Mexico. Currently, Türkiye, often regarded as one of the most important EMEs, lacks a comprehensive study of its business cycles. Researchers have conducted studies such as Alp et al. (2012), Alper (2002), Cebi (2012), Tastan and Asik (2014), and Yuksel (2013) to analyze the business cycles in relation to the Turkish economy. This chapter distinguishes itself from previous research by presenting an extended theoretical framework, utilizing long-term data for the estimations, and providing a more comprehensive explanation of economic fluctuations by incorporating financial frictions into widely accepted models. This chapter focuses on identifying the key factors that cause fluctuations in eco- nomic activity in EMEs, which are characterized by both trade deficits and foreign debt. To investigate this issue, we perform a case study on Türkiye. The objective of this research question is to investigate the intricate relationship between trade im- balances, foreign debt, and economic fluctuations in the Turkish economy. Türkiye is an appropriate subject for the study due to its status as an EME characterized by consistent trade deficits and a substantial level of foreign debt. Our goal in ex- amining Türkiye’s experience is to gain important insights into overall patterns and dynamics that may be relevant to other EMEs facing similar challenges. The objective of this study is to examine the long-term business cycles in Türkiye by employing the Financial Frictions Model established by Garcia-Cicco et al. (2010) using the DSGE methodology. The inclusion of financial frictions offers a more compelling explanation for the factors that drive economic activity and enables an examination of their significance. Financial frictions are significant factors in the occurrence of fluctuations. Moreover, the financial frictions model provides a more comprehensive explanation for the fluctuations when compared to the baseline models. The constructed model redefines the connection between domestic and world interest rates, resulting in a negative trade balance. Section 1.2. presents the details of the model. This chapter makes significant contributions to the field of macroeconomic modeling, particularly in the context of EMEs. Firstly, it presents a modified DSGE model that is specifically designed to accurately represent the different economic dynamics of Türkiye. This research enhances the portrayal of Türkiye’s economic environment 13 by including both trade deficits and foreign debt in the model, resulting in a more comprehensive and accurate analysis compared to prior models. Furthermore, this chapter provides novel and up-to-date estimation findings, filling a notable gap in the existing research. As far as we know, there is a lack of re- search that calculates a DSGE model using data from Türkiye. The results of our study indicate that growth shocks play a vital role in causing economic fluctuations in Türkiye. These findings provide policymakers with valuable insights that can inform their decision-making process. The emphasis on Türkiye, combined with the simultaneous examination of trade deficits and foreign debt, signifies notable progress in the implementation and estimation of DSGE models for EMEs. In this study, we conduct a historical comparison of the growth rates of Türkiye and selected EMEs, Argentina and Mexico. We obtain intriguing findings in the compar- ison by utilizing data spanning a substantial duration of around 70 years. While the 70-year average output growth and volatility of output growth exhibit similarities, we identify notable distinctions across different time periods. As an illustration, Argentina, which had a comparatively higher rate of economic growth during the 1990s, experienced a period of significant economic decline due to the crisis it faced from 1998 to 2002. When comparing the volatilities of the economic boom and crisis periods of Argentina, they exhibit significant similarities. Given the disparities in growth rates among Argentina, Mexico, and Türkiye, our study specifically examines the impacts of Türkiye’s domestic economic crises in 1994 and 2001, as well as the global financial crisis of 2008. During the crisis years, we examined the correlation between the shock variables derived from the model and several macroeconomic variables. We discover that negative growth shocks are a significant factor in Türkiye’s economic collapse. Furthermore, the research revealed that during times of crisis, the foreign trade deficit considerably decreased due to the negative growth shocks. Furthermore, the estimation process determines the model parameters using exten- sive historical yearly data specifically collected from Türkiye from 1950 to 2019. The data is sourced from the Penn World Table (PWT) dataset. The variables selected for estimation are the output growth (gY ), the consumption growth (gC), the in- vestment growth (gI), and the trade balance to output ratio (tby). We estimate the model parameters using the Bayesian estimation method. The estimation results align with the relevant literature. The estimated parameters, derived from Turkish 14 data, provide a novel contribution to the existing literature. In addition, we con- duct the Impulse Response Functions (IRFs) analysis and the conditional variance decomposition analysis based on the parameters obtained through the Bayesian es- timation technique. A shock process is persistent if its effects take a longer time to dissipate. We discover that the persistency of growth shocks in Türkiye is quite significant compared to Argentina and Mexico. Accordingly, growth shocks have a greater impact on nearly all macroeconomic indicators, especially in the long-term. Furthermore, there is a similarity between the persistency parameter of technology shock in Argentina, Mexico, and Türkiye. The model results corroborate the findings of Aguiar and Gopinath’s (2007) investigation. The study’s key findings indicate that the model accurately represents the data in relation to the second moments. The order of the volatilities of the selected variables is σtby < σgI < σgC < σgY in accordance with the Turkish data. In addition, the relative volatilities obtained from the model are consistent with the data. Moreover, the correlation coefficients of the variables exhibit compatibility between the data and the model results. While the correlation coefficients of gY with gC and gI are positive, the correlation coefficient of gY with tby is negative. The findings of the IRFs align with the prevailing macroeconomic expected out- comes. The positive technology and the positive growth shocks have a positive impact on gY , gC , and gI ; nevertheless, the consequences of these shocks vary. Like Garcia-Cicco et al.’s (2010) research, the country premium shock significantly affects the fluctuations of tby. Nevertheless, the country premium shock does not exert a substantial influence on gI . Technology shocks have a more significant impact on the output growth and con- sumption growth variables in the short run through conditional variance decomposi- tion, while growth shocks have a stronger effect in the medium and long terms. The Turkish data, unlike the Argentinian study, suggests that the intertemporal prefer- ence shock does not significantly influence the volatilities of the variables. However, in terms of the fluctuations of tby, the domestic spending shock appears to be espe- cially significant in the context of Türkiye. One of the primary factors contributing to this phenomenon is the comparatively greater government expenditures to GDP ratio observed in Türkiye compared to Argentina. 15 The following section of the study offers an in-depth analysis of the employed model. In Section 1.3. and 1.4., the analysis includes the utilization of the data, the process of calibration, and the outcomes of estimation. Section 1.5. investigates the results of IRFs in relation to several shock processes. Also, in Section 1.5., the results obtained in the original investigation are compared with the results obtained in this study specifically for Türkiye via conditional variance decomposition analysis. Section 1.6. includes discussions that specifically focus on the model’s findings and data. We initially focus on the periods of crisis in Türkiye. This section presents a comparative analysis of the data and the shock variables in different crisis periods. Section 1.6. also analyzes the historical growth rates of Türkiye and selected EMEs, Argentina and Mexico, by categorizing them into several sub-periods. In this section, we will also discuss what the model teaches us about Türkiye’s economic dynamics. Finally, the concluding remarks are presented in Section 1.7. 1.2. MODEL This section presents the constructed model. In the current economic environment, only final goods are available. The production of these goods is contingent upon the availability of both labor and capital. The production function is the following expression: Yt = atK α t (Xtht) 1−α (1) where the capital’s share of output α ∈ (0, 1). Yt, Kt, and ht are output, capital, and labor respectively. Furthermore, at and Xt are productivity processes. While the variable at represents the temporary technological shock, the variable Xt represents the permanent effects of the technological shock on productivity. To facilitate illustration and calibration, different notations are employed to represent shocks to the “level of productivity” (at) and the “growth of productivity” (Xt). Throughout the entire chapter, we represent variables with a trend in equilibrium with uppercase letters, and variables without a trend in equilibrium with lowercase letters. In the context of natural logarithms, it is presumed that at corresponds to a first- order autoregressive process (AR(1)): 16 ln at = ρa ln at−1 + εat ; εat ∼ N(0, σ2 a) (2) Xt is nonstationary. Let gt = Xt Xt−1 represents Xt’s gross growth rate. The real- izations of gt are often referred to as “growth” shocks due to the fact that they represent the stochastic trend in productivity. gt follows an AR(1) process: ln (gt ḡ ) = ρg ln (gt−1 ḡ ) + εgt ; εgt ∼ N(0, σ2 g) (3) where ḡ represents the deterministic gross productivity growth rate. The parameters ρa, ρg ∈ [0, 1) are the persistency parameters of the shock processes. Also, the parameters σa, σg > 0 are the volatilities of the shock processes. The household’s budget constraint is as follows: Yt + Dt+1 1 + rt = Ct + St + It +Dt + φ 2 (Kt+1 Kt − ḡ )2 Kt (4) where rt represents the domestic interest rate and the variable Dt+1 represents the amount of external debt obtained during period t. Also, Ct, It, and St are consump- tion, investment, and domestic spending, respectively. The last expression is the capital adjustment cost. φ is the capital adjustment cost parameter. To clarify, St stands for unanticipated government expenditures. In the context of AR(1), the variable St signifies the exogenous stochastic domestic spending shock process: ln (st s̄ ) = ρs ln (st−1 s̄ ) + εst ; εst ∼ N(0, σ2 s) (5) where st = St Xt−1 and the parameter s̄ denotes the long-term share of public spending to GDP. The parameter ρs ∈ [0, 1) and σs > 0 are the persistency parameter and the volatility of the shock process respectively. The following expression shows the law of motion of capital: Kt+1 = It + (1− δ)Kt (6) where δ ∈ (0, 1) is depreciation rate. The household desires to maximize the following lifetime utility function by choosing consumption, labor, capital, and debt stock: 17 max {Ct,ht,Kt+1,Dt+1} E0 ∞∑ t=0 βtvt (Ct − θω−1Xt−1h ω t )1−γ − 1 1− γ (7) where β, θ, ω, and γ are the discount factor, labor coefficient in the utility function, exponent of labor in the utility function, and intertemporal elasticity of substitution, respectively. In Equation 7, vt is the intertemporal preference shock process. It is also exogenous and follows AR(1): ln vt = ρv ln vt−1 + εvt ; εvt ∼ N(0, σ2 v) (8) where the parameter ρv ∈ [0, 1) and σv > 0 are the persistency parameter and the volatility of the shock process respectively. We assume that this EME faces a debt elastic interest rate premium to be consistent with the related lirature (as in Aguiar and Gopinath (2007)): rt = r∗ + eµt−1 − 1 + ψ ( e Dt Xt−1 −d̄ − 1 ) (9) where r∗, µ, ψ and d̄ are the global interest rate, country premium shock, the debt elasticity of the interest rate and the long-term debt level respectively. The original study defines the d̄ parameter as the long-term ratio of trade balance to GDP. At the long-term equilibrium level, this assumption guarantees that the country can maintain external debt when experiencing a foreign trade surplus, and have a creditor from the outside world (negative external debt) when facing a foreign trade deficit. In the economic model of Türkiye, which features a trade deficit and external debt simultaneously, d̄ represents the long-term equilibrium external debt level and it can be shown as d̄ = d̃Y ss, where d̃ and Y ss represent the long-term debt to GDP ratio and the steady state level of output, as described in the Schmitt-Grohe and Uribe’s (2003) research. Furthermore, the country premium shock process, µt, is exogenous and follows AR(1) as well: lnµt = ρµ lnµt−1 + εµt ; εµt ∼ N(0, σ2 µ) (10) where the parameter ρµ ∈ [0, 1) and σµ > 0 are the persistency parameter and the volatility of the shock process respectively. 18 Eventually, the household maximizes Equation 7 subject to Equations 1, 4, and 6 by assuming as given the initial conditions, D0 and K0 and the shock processes, at, gt, st, vt, and µt. Also, no-ponzi condition must hold to obtain the deterministic steady state as follows: lim T→+∞ Et Dt+T∏T i=0(1 + ri) ≤ 0 (11) By log-linearizing the resource constraints and the first order conditions around the deterministic steady state, we numerically solve the normalized model. The uniqueness of the equilibrium is established based on the presumption of the first- order approximation. Furthermore, the steady state conditions of the model are presented in Appendix 1. 1.3. DATA AND CALIBRATION The unit of time is the year in this study. While some parameters are calibrated from the relevant literature, others are calibrated using data from Türkiye. Table 1 shows the calibrated parameters from the literature. In the relevant literature, the capital share in the production function (α) is assumed to be 0.32. Furthermore, a value of θ = 2.24 guarantees that households dedicate 20% of their time to labor in the long-term. The value of ω = 1.6 ensures that the labor supply elasticity ( 1 ω−1 ) equals 1.7, which is commonly used in the literature (as in Mendoza (1991)). Table 1: The Calibrated Parameters from the Relevant Literature Description Symbol Value Capital elasticity of the production α 0.32 Labor coefficient in utility θ 2.24 Exponent of labor in utility ω 1.60 Intertemporal elasticity of substitution γ 0.30 Capital depreciation rate δ 12.55% Discount factor β 0.98 Following the research of Garcia-Cicco et al. (2010), the capital depreciation rate is determined at 3% quarterly, and 12.55% annually. The intertemporal elasticity of 19 substitution is assumed to be 0.3 for Türkiye, in line with the findings of Havranek et al. (2015). Finally, we set the discount factor at 0.995 quarterly, and 0.98 annually, in accordance with Chang et al. (2015) and Brzoza-Brzezina et al. (2013). We decide to use the PWT dataset, widely recognized as a highly significant and reliable source for the research of Türkiye’s historical data. The PWT data is a compilation of national-accounts data created and managed by scientists at the University of California, Davis, and the Groningen Growth Development Centre of the University of Groningen. The PWT dataset’s purpose is to quantify the real GDP across different countries and track its changes over time. The updates incorporate additional countries, currently totaling 183, as well as statistics spanning from 1950 to 2019. In addition, the updates include information on capital, productivity, employment, and some other significant macroeconomic variables. We also utilize the World Bank Dataset (WBD) to adjust other parameters missing from the PWT dataset. The WBD spans from 1970 to 2022. We calibrate various model parameters by using the steady state equilibrium solu- tions of the model and the long-term annual data from Türkiye. PWT data provides the real GDP at constant PPPs (purchasing power parity) in 2017 US$. The dataset also provides the ratios of government spending, investment, consumption, capital stock, and other macroeconomic variables to output, allowing for easy computation of the required variables in the same unit as the real GDP. By simultaneously employing the data with the steady state solutions of the model, we use (i) external debt stock (D) to GDP (Y ) ratio to obtain the parameter d̃; and (ii) the government spending (G) to GDP ratio to determine the parameter s̄. The calibrated parameters using the data are displayed in Table 2. Table 2: The Calibrated Parameters Using the Turkish Data Description Symbol Data Value External debt to output ratio d̃ D Y 38.14% Share of public spending in GDP s̄ G Y 18.64% Note: The author generates the results using annual data for Türkiye from 1950 to 2022. Y , D, and G represent GDP, external debt stock, and government spending data for Türkiye. 20 1.4. ESTIMATION We use the Bayesian maximum likelihood estimation method (BML) and the data from Türkiye to estimate variables that are unobservable and difficult to accurately calibrate in the model. The literature prefers the maximum likelihood (ML) method for model estimation due to its advantages over other methods. It is possible to estimate certain model equations with the generalized method of moments (GMM) and the impulse-response matching methods. The ML estimation, on the other hand, is a system-based estimation method. All model equations are estimated collectively. Furthermore, this method offers the essential metrics for comparing models. The BML approach offers some additional benefits compared to the ML estimating method. The classical ML estimator treats the parameters as fixed but unknown values. It calculates the ML estimator by maximizing the likelihood function using the observed values of the variables. The BML technique assumes that the parame- ters are stochastic variables for which there exists a priori information. Finally, the BML method allows for the inclusion of measurement errors while also considering the potential dangers of model identification. The estimation is based on Turkish data spanning 1950 to 2019. We derive the real GDP data on a per capita basis by dividing it by the total population, which eliminates the effect of population size. The scale effect is also eliminated by ap- plying the natural logarithm to the obtained numbers. The output growth rate is calculated based on the obtained values. We also apply this strategy to investment and consumption data. We divide net exports by GDP for each year to calculate the trade balance to GDP ratio. We use the growth of investment (gI), the growth of consumption (gC), the growth of output (gY ), and the trade balance to output ratio (tby) data for the estimation. Table 3 provides the summary statistics for these variables. During our analysis of comprehensive Turkish data, we saw the 2001 economic crisis as a significant milestone in the country’s history. The 2001 Turkish economic crisis was a severe financial crisis that resulted in the devaluation of the Turkish lira and an enormous drop in the stock market. This crisis was caused by long-standing political and economic problems that had been suffering Türkiye for several years. 21 Table 3 displays the lowest and highest values of each selected variable, along with the average and standard deviation, across two different time periods: the pre-2001 crisis period and the post-2001 crisis period. The last two rows show the average and standard deviation over the entire period. Türkiye’s economic growth throughout the period leading up to the crisis, which encompassed the 1980s and 1990s, relied significantly on foreign investment. The fiscal capabilities of the Turkish government and banking sectors were inadequate to support significant economic growth. The government, already struggling with substantial budget deficits, partially sustained them by issuing large volumes of high- interest bonds to Turkish banks. The government was able to prevent a temporary default on the bonds due to the continuous increase in inflation. As a result, Turkish banks started to primarily participate in these high-yield bonds.11 Table 3: Summary Statistics of the Selected Variables (in Percent) Period Statistics gY gC gI tby Min -12.01 -16.11 -43.78 -5.96 1951-2001 Max 18.68 21.20 41.94 -0.41 Avg 2.54 2.20 2.59 -1.81 Std 6.29 6.78 15.63 1.31 Min -3.96 -3.67 -27.96 -10.21 2002-2019 Max 12.16 11.71 31.89 -2.71 Avg 4.25 3.15 7.93 -6.84 Std 5.04 4.26 15.27 1.96 1951-2019 Avg 2.99 2.45 3.98 -3.12 Std 6.03 6.24 15.71 2.67 Note: Min, Max, Avg, and Std represents minimum, maximum, average and standard devi- ation, respectively. The outcomes are computed by the author using Turkish data from 1950 to 2019 from the PWT dataset. Furthermore, the presence of political uncertainty in Türkiye would likely lead for- eign countries to exercise significant caution when considering any investment ini- tiatives. As a result of the crisis in 2001, the investment growth had a contraction of 43.78%. According to the data, this rate is the most significant decline in the 11See Özatay and Sak (2002). 22 history of the country. Simultaneously, there was a decrease of 3.57% in the output growth and a decrease of 5.35% in the consumption growth. When comparing the 50-year long-term averages prior to the 2001 crisis with the period following 2001, significant disparities become apparent. The average rate of the output growth over the span of about 50 years was 2.54%, whereas the average rate in the period after 2001 was 4.25%. Additionally, there is a slight rise in the average rate of the consumption growth. Moreover, the most important rate increases are those related to the gI and tby. The average rate of gI increases roughly threefold, while tby increases almost fourfold in the post-2001 era. Türkiye, a country with a historical trade deficit, has experienced a notable escalation in its trade deficit since 2002. Upon analyzing the volatilities over these specific time periods, it is evident that there is no substantial change in the volatility of investment growth. The decline in volatility is observed in both the output growth and the consumption growth. However, tby demonstrates a rise in volatility from 1.31 to 1.96. Figure 1: Historical Data of Türkiye: 1951-2019 (in Percent) Note: g y, g c, g i and tby represent the output growth, the consumption growth, the investment growth, and the trade balance to output ratio, respectively. The x- and y-axes show the year and the percentage values, respectively. Figure 1 shows the values of these variables from 1951 to 2019. The figure clearly demonstrates the immediate occurrence of volatility in investment growth (the blue 23 line). The significant increase in the trade deficit after 2001 is readily apparent (the black line). Another notable issue in the figure is the close connection between the output growth and the consumption growth (the red and green lines). Table 4 shows the key statistics about prior and posterior distributions. For the prior distributions of shock persistency parameters ρg, ρa, ρv, ρµ, and ρs, the beta distribution is chosen because it is commonly used in the relevant literature. Addi- tionally, the beta distribution ranges from 0 to 1 in accordance with the definition of the persistency parameters. The inverse gamma distribution is commonly used for modeling the volatilities of shock processes, σg, σa, σv, σµ, and σs. Also, the inverse gamma distribution is defined only for positive real values, just like the volatility parameters. Initially, we select a broad range and substantial variance to calculate the mean values of the prior distributions. The estimation procedure involves iteratively substituting the estimated values with mean values to improve accuracy. Given the lack of precise predictions for the characteristics of the productivity growth, the interest rate debt elasticity, and the capital adjustment cost parameters, we prefer a uniform distribution for these parameters. This economy experiences a trade deficit if its productivity growth rate exceeds 1.03. The model must have a value below around 1.10 to satisfy the Blanchard-Kahn conditions.12 Thus, the value of ḡ is selected to fall between the range of 1.03 to 1.10. In addition, the capital adjustment cost and the interest rate debt elasticity parameters are selected from a significantly broader range, φ ∈ [0, 50] and ψ ∈ [0, 20] than the ranges of the other parameter values. We use a Markov Chain Monte Carlo (MCMC) chain with two million iterations to determine the posterior statistics. We eliminate the first million iterations during this process. In addition, the upper bound of the prior distributions for the standard deviations of measurement errors pertaining to the observed variables (σmeY , σmeC , σmeI , and σmetby ) is set at 25% of the variance of the relevant historical data. 12Blanchard and Kahn (1980) established certain criteria for a solution’s existence and unique- ness, which can be easily verified by examining eigenvalues calculated at the model’s equilibrium. The model’s solution is unique when the number of unstable eigenvectors in the system (the num- ber of eigenvalues greater than 1) matches the number of forward-looking (control) variables. 24 Table 4: Prior and Posterior Distributions Prior Distribution Posterior Distribution Parameter Distribution Mean Std Mean 5% 95% g Uniform 1.065 0.02 1.0899 1.0889 1.0910 ρg Beta 0.9 0.02 0.9807 0.9792 0.9819 ρa Beta 0.8 0.02 0.8353 0.8092 0.8615 ρv Beta 0.3 0.02 0.3007 0.2678 0.3337 ρµ Beta 0.6 0.02 0.6018 0.5692 0.6350 ρs Beta 0.4 0.02 0.4429 0.4086 0.4772 σg Inv. Gamma 0.03 2 0.0413 0.0349 0.0475 σa Inv. Gamma 0.03 2 0.0354 0.0300 0.0406 σv Inv. Gamma 0.50 2 0.3086 0.1278 0.4975 σµ Inv. Gamma 0.03 2 0.0792 0.0647 0.0931 σs Inv. Gamma 0.05 2 0.1908 0.1623 0.2180 φ Uniform 25 14.43 24.451 18.158 30.437 ψ Uniform 10 5.77 6.989 2.079 12.431 σmeY Uniform 0.008 0.0043 0.0141 0.0129 0.0151 σmeC Uniform 0.008 0.0045 0.0128 0.0089 0.0156 σmeI Uniform 0.020 0.0113 0.0381 0.0367 0.0393 σmetby Uniform 0.003 0.0019 0.0024 0.0001 0.0047 Note: Std represents the standard deviations of the distributions. The lower and upper bounda- ries of Highest Posterior Density (HPD) intervals are 5% and 95%, respectively. We perform various diagnostic tests prior to analyzing the estimation results. The identification test is the first diagnostic test. We utilize the identification test to verify the accurate identification of all intended parameters for the estimation pro- cess.13 Figure 30 in Appendix 1 shows the identification test results. The range of identification strength is from 0 to ∞. Using a logarithmic scale, the domain is now restricted to the interval (−∞,∞). Values ranging from 0 to 1 exhibit negative logarithmic values in terms of identification strength. A value of 0 in the level is shown as −∞ on the graph, which is why no bar is displayed since −∞ 13See Qu and Tkachenko (2012) and Iskrev (2010). 25 cannot be plotted. Therefore, since all parameters intended for estimation in the model are non-zero, we can infer that they are all identifiable. More precisely, ḡ is the most strongly identified parameter. Following ḡ, ρg, and ρa are more strongly identified than the other parameters. The figure also illustrates the degree to which the prior distribution mean (blue bars) and variance (yellow bars) influence the parameter determination. Brooks and Gelman (1998) created MCMC univariate and multivariate diagnostic tests. When the number of Monte Carlo chain exceeds one, we can calculate these tests. These tests can be calculated in this study because the estimation procedure relies on five Monte Carlo chains. Figure 31 in Appendix 1 displays the multivariate MCMC convergence diagnostics. Figures 32 to Figure 36 in Appendix 1 also display the univariate MCMC convergence diagnostics. In Figure 31, the first figure (interval) uses the Brooks and Gelman convergence test for the 80% interval, the second figure (m2) for the variance, and the third figure (m3) for the third moment. The blue line reflects combined samples from all chains, while the red line represents values from an individual chain. We anticipate the two lines to stabilize horizontally and approach each other if the chains converge. Thus, the estimated model successfully passes the multivariate and univariate MCMC convergence tests. Finally, the Metropolis-Hastings (MH) algorithm employs the selection of favorable candidates from the simulation-generated distributions around the mode to form the posterior distributions. At this stage, the acceptance ratio determines the appropri- ate candidates. The common consensus in the literature is that this ratio should be within the range of 20% to 40%, with the ideal target being around 33%.14 In the model estimation, the acceptance ratios for each chain are as follows: 32.20% for the first chain, 32.38% for the second chain, 32.30% for the third chain, 32.23% for the fourth chain, and 32.33% for the fifth chain. The estimation results are generally consistent with the findings in the existing literature. All parameters yielded a mean value that is not equal to zero. Upon examination of the persistency parameters, it is evident that the parameters of the technology and the growth shocks (ρa and ρg) exhibit more persistent processes compared to the other shock processes. 14See Roberts et al. (1997) and Neal (2011). 26 It is important to note that the persistency of intertemporal preference and the domestic spending shock processes (ρv and ρs) are quite smaller than the other shock processes. The estimated parameters φ and ψ are also compatible with the existing literature. Lastly, the posterior means of the measurement errors are close to the minimum levels, and they account for less than 2% of the volatilities of the observed variables. 1.5. MAIN RESULTS In this section, the second moments of the constructed model and Turkish data are compared. Additionally, we present the impulse response functions (IRFs) of the selected variables to the shock processes and the conditional variance decomposition table. 1.5.1. Second Moments Table 5 shows some statistics about the model and data. The selected variables are investment growth (gI), consumption growth (gC), output growth (gY ), and the trade balance to GDP ratio (tby), as previously stated. The dataset consists of yearly data spanning from 1950 to 2019. Table 5 also displays the relative volatilities of the selected variables. The statistics show that the main characteristics of the model and data are compatible, namely the magnitude and ordering of second moments, comparable relative volatilities, and the signs of the correlation coefficients. While the model demonstrates a slightly higher level of volatility in the selected variables compared to the observations derived from the data, the corresponding amounts of volatility in the variables are consistent between the model and the data. In short, σtby < σY < σC < σI for both the data and the model, where σ denotes the standard deviation of any variable. Furthermore, the volatility of gC relative to gY is consistent across the data and the model. The relative volatility of gI to gY is higher, as the data suggest. Ultimately, the data and model both indicate that the relative volatility of tby to gY is less than 1. However, the model suggests that this volatility is much lower. Moreover, the model and data exhibit consistent relationships between the selected 27 variables and the output growth. For instance, while the correlation coefficient between tby and gY (corr(tby, gY )) equals to −14.01% in the model, it equals to −25.99% in the data. We also observe comparable correlation coefficients between the output growth and both the consumption and investment growth. Furthermore, the correlation coefficients between the trade balance to output ratio and the other variables within the model align with the actual data. Both the model and the data exhibit negative correlation coefficients between tby and the other variables. Thus, the estimation findings accurately reflect the main characteristics of the data. Table 5: Second Moments of Data and Model (in Percent) Statistics Variables Relative Volatility gY gC gI tby σC σY σI σY σtby σY Standard Deviations Model 23.41 24.12 30.46 7.08 1.03 1.30 0.30 Data 6.03 6.24 15.71 2.67 1.03 2.61 0.44 Correlation with gY Model 94.30 81.17 -14.01 Data 60.64 73.09 -25.99 Correlation with tby Model -14.01 -3.82 -22.57 Data -25.99 -19.18 -24.22 Note: gY , gI , gC , and tby are output growth, investment growth, consumption growth, and the trade balance to GDP ratio, respectively. The three columns on the right display the relative vo- latilities. The outcomes are computed by the author. 1.5.2. Impulse Response Functions The IRFs analysis have used to closely examine the impact of shocks on the variables. The IRFs depict the expected path of endogenous variables when subjected to a shock of one standard deviation at time t = 0. Figure 2 to Figure 6 show the Bayesian Impulse Response Functions (BIRFs) of the selected variables to the shock processes. 28 The solid black lines in BIRF figures depict the path of variables. The solid red lines represent the deterministic steady state. The y-axis represents the percentage deviation from the deterministic steady state level and the x-axis is the period of time following the occurrence of a shock, which is set to 10 periods of time. tb y, g invest, g c, and g y represent the trade balance to GDP ratio, investment growth, consumption growth, and output growth, respectively. Lastly, the figures exhibit a 90% confidence interval. Figure 2 displays the BIRFs of the selected variables to the positive technology shock. When the shock occurs, consumption, output, and investment growth in- crease by around 8%, 6%, and 6%, respectively. The positive technological shock has a favorable impact on these variables, causing them to rise as expected. Follow- ing the initial increases, each variable eventually reaches the steady state level. Figure 2: BIRFs of the Selected Variables to the Technology Shock (at) Note: The solid black and red lines display the path of variables and the deterministic steady state values. The x- and y-axes represent the period of time following the occurrence of a shock and the percentage deviation from the deterministic steady state level. The positive technology shock has a positive impact on tby during the initial phase. A positive technological shock, characterized by increased productivity using the same resources, results in a lower level of foreign borrowing. Over time, the re- quirement for external borrowing increases as a result of enhanced production, and the ratio decreases below its long-term equilibrium level, but the change in tby is negligible. It persists at a level lower than its long-term level for an extended period of time. Eventually, the ratio converges to its steady state level. Figure 3 displays the BIRFs of the selected variables to the positive growth shock. 29 Growth and technological shocks have comparable effects on the variables. With the occurrence of a growth shock, gY and the gI increase by around 5%, and 20%, respectively. In response to the growth shock, gI outpaces gY . The growth shock impacts gC with a time lag. During the shock, consumption growth initially drops below the long-term equilibrium level of 2% due to a huge increase in investment growth. Nevertheless, it then rises instantly and stays above the long-term equilibrium level over an extended period of time. Nevertheless, the growth shock slightly decreases tby by around 2%. The reason for this phenomenon is an increase in demand for accessing external resources as a result of the accelerated growth rates. Following the initial movements, each variable eventually reaches the steady state level at a variety of time periods. Figure 3: BIRFs of the Selected Variables to the Growth Shock (gt) Note: The solid black and red lines display the path of variables and the deterministic steady state values. The x- and y-axes represent the period of time following the occurrence of a shock and the percentage deviation from the deterministic steady state level. Figure 4 shows the BIRFs of the selected variables to the positive intertemporal preference shock. An intertemporal preference shock refers to a sudden shift that alters the connection between present and future consumption by affecting the sub- jective discount rate. Positive intertemporal preference shocks elevate the value of present consumption and utility, leading to an increase in the consumption growth. In order to attain this level of spending, households boost their borrowing volume, leading to a reduction 30 in tby. Figure 4: BIRFs of the Selected Variables to the Preference Shock Note: The solid black and red lines display the path of variables and the deterministic steady state values. The x- and y-axes represent the period of time following the occurrence of a shock and the percentage deviation from the deterministic steady state level. Moreover, households offset the rise in consumption by decreasing their investments when the shock happens. A decline in investments results in a subsequent decrease in the growth rate of output during the subsequent period (referred to as period 2). Lastly, the impact of the preference shock is minimal when examining the values on the y-axis. The BIRFs of the selected variables to the negative country premium shock are displayed in Figure 5. The negative country risk premium shock can be assessed by observing an escalation in the country’s Credit Default Swap (CDS) rates, resulting in an increase in borrowing costs. When a country experiences a negative country premium shock, its ability to borrow money from foreign sources decreases and as a result, its trade balance improves. Conversely, households experiencing a decline in resources decrease their consump- tion, investment, and thus, their ability to produce. As a result, the shock has a negative effect on the growth of investment, the growth of consumption and the out- put growth. Following the first reaction, the variables partially regain their stability and ultimately converge towards their long-term equilibrium levels. 31 Figure 5: BIRFs of the Selected Variables to the Contry Premium Shock Note: The solid black and red lines display the path of variables and the deterministic steady state values. The x- and y-axes represent the period of time following the occurrence of a shock and the percentage deviation from the deterministic steady state level. Figure 6: BIRFs of the Selected Variables to the Domestic Spending Shock Note: The solid black and red lines display the path of variables and the deterministic steady state values. The x- and y-axes represent the period of time following the occurrence of a shock and the percentage deviation from the deterministic steady state level. Lastly, Figure 6 shows the BIRFs of the selected variables to the positive domes- tic spending shock. Based on the observed deviation amounts, it is evident that 32 domestic spending shock has a negligible impact on the fluctuations of gY and gC . However, the domestic spending shock has a particularly significant impact on in- vestments. The increase in domestic spending crowds out investments and causes investment growth to decline by around 4%. After four periods, the investment growth fully recovers and stabilizes at the long-term equilibrium level. The decline in gI , gC and the output growth leads to a moderate increase in the requirement for external borrowing, while also resulting in a deterioration in tby during the initial phase. Subsequently, there is a decline in external borrowing and the ratio remains above its long-term equilibrium level for a prolonged duration. 1.5.3. Conditional Variance Decomposition We share the conditional variance decomposition tables of the model across various time horizons to discuss the impact of shock processes on the selected variables of Türkiye in different time periods. Variance decomposition analysis quantifies the extent to which exogenous shocks account for the predicted error variance of a specific variable. This strategy al- lows for the separation of the significance and impact of various types of shocks on macroeconomic fluctuations. Essentially, this method divides the impact of each shock process on business cycle fluctuations into ratios. The conditional variance decomposition findings from model simulations are shown in Table 6, considering the short-term (1 period), the medium-term (5 periods), and the long-term (10 periods) effects. The original study for Argentina shows that growth shocks alone are insufficient to explain fluctuations in macroeconomic variables. However, the results of our study in Türkiye challenge this argument. This investigation confirms the widely debated topic of “the cycle is the trend” in the literature. Upon analyzing the short-term consequences, it appears that 61.1% of the volatility in the output growth is attributed to the technology shock, while 38.9% is attributed to the growth shock. The technological shock accounts for nearly all the fluctuations in gC . The growth shock primarily accounts for the volatility in gI . tby indicates that 28.9% of the volatility is attributed to the growth shock, 53.6% to the country premium shock, and 14.2% to the domestic spending shock in the short-term. 33 Table 6: Conditional Variance Decomposition gY gC gI tby Period 1 Technology Shock 61.1 88.3 7.7 2.9 Growth Shock 38.9 5.7 81.9 28.9 Intertemporal Preference Shock 0.0 1.5 0.1 0.4 Country Premium Shock 0.0 2.4 6.1 53.6 Domestic Spending Shock 0.0 2.2 4.3 14.2 Period 5 Technology Shock 32.7 49.0 7.4 4.1 Growth Shock 66.2 46.5 82.7 28.9 Intertemporal Preference Shock 0.0 1.5 0.1 0.5 Country Premium Shock 0.5 1.4 5.8 43.2 Domestic Spending Shock 0.6 1.6 4.1 23.3 Period 10 Technology Shock 22.6 34.1 6.8 10.6 Growth Shock 76.6 62.8 83.9 25.9 Intertemporal Preference Shock 0.0 1.0 0.1 0.5 Country Premium Shock 0.4 1.0 5.4 36.8 Domestic Spending Shock 0.5 1.1 3.9 26.2 Note: gY , gI , gC , and tby are the output growth, the investment growth, the consumption growth, and the trade balance to output ratio, respectively. The author computes the outcomes. The table expresses the values as percentages. When analyzing the medium-term consequences, the ability of the technological shock to account for fluctuations in the consumption growth and the output growth is notably diminished. The technological shock’s explanatory power on the volatili- ties in gI and tby remains modest in the medium-term. The growth shock has the greatest explanatory power for the fluctuations of most of the selected variables in the medium term. More precisely, the growth shock accounts for 66.2% of the output growth volatility, 46.5% of the consumption growth volatility, 82.7% of the investment growth volatility, and, finally, 28.9% of the volatility in tby, 34 respectively. In the medium term, the explanatory power of the intertemporal preference shock is negligible. The country premium shock remains a significant factor in explaining a substantial portion of the volatility in tby. Upon examination, the domestic spending shock accounts for 23.3% of the volatility in tby, but is deemed insignificant in explaining other selected variables. On the other hand, the growth shocks are responsible for most of the volatility in gY , gC , and gI over the long term, with percentages of 76.6%, 62.8%, and 83.9%, respectively. This discovery aligns with Aguiar and Gopinath’s (2007) theory that emphasizes “the cycle is the trend”, a concept well explored in the literature. In an EME such as Türkiye, the fluctuations in most variables can be mostly attributed to the growth shocks. The technology shock explains 22.6% of gY volatility, 34.1% of gC volatility, 6.8% of gI volatility, and 10.6% of tby volatility in the long term. Also, the impact of intertemporal preference, country premium, and domestic spending shock processes on the volatility in gI , gC , and gY are negligible. Besides, 36.8% of the volatility in tby is attributed to the country premium shock, whereas 26.2% is attributed to the domestic spending shock in the long term. The results indicate that the explanatory ability of the country premium shock is less than what the Argentinian study suggests. Also, the explanatory power of the domestic spending shock is greater in the context of tby volatility compared to the Argentinian sample. The primary explanation for this situation is the higher percentage of long-term government spending in GDP in Türkiye. To sum up, several key factors contribute to the diminishing explanatory power of shock processes, such as technology, over time. Firstly, the nature of the shock process itself is inherently temporary, meaning its effects are not long-lasting. Addi- tionally, the persistency parameter associated with technology shocks is lower com- pared to growth shocks, further contributing to their decreasing explanatory power. Finally, agents with forward-looking expectations tend to react less to technological shocks that they are aware are temporary in nature. 35 1.6. DISCUSSION This section includes observations that specifically focus on the model’s findings and data. We will initially focus on Türkiye’s significant historical periods of crisis. For this analysis, we have selected the economic crises that occurred in the years 1994, 2001, and 2009. Next, we will provide a concise comparison between Türkiye and a few selected EMEs. Lastly, we will incorporate a concise discussion section regarding the proportion of fluctuations in Türkiye that originate from internal factors versus foreign factors. 1.6.1. Türkiye’s Crisis Experiences Based on the analysis of long term data, it is determined that numerous shock processes have a significant impact on Türkiye. In this subsection, we will examine the dynamics of the output growth (gY ) and the trade balance to output ratio (tby) variables during crisis periods using the shock variables derived from the employed model. There is widespread recognition that Türkiye experienced severe domestic crises in 1994 and 2001. Following that, the global financial crisis that commenced in 2008 had an impact on numerous EMEs, including the Turkish economy. This subsection of the study provides a concise analysis of the causes and consequences of these crises. We then conduct an analysis of the model’s functioning during times of crisis, specifically focusing on the variables of gY and tby. A combination of facto