ESTIMATING AND FORECASTING TRADE FLOWS BY PANEL DATA ANALYSIS AND NEURAL NETWORKS

Elif NUROĞLU

Article Details

ESTIMATING AND FORECASTING TRADE FLOWS BY PANEL DATA ANALYSIS AND NEURAL NETWORKS

Authors:

Elif NUROĞLU

Publisher:

Istanbul University

Main Topics:

Management

Journal:

İstanbul Üniversitesi İktisat Fakültesi Mecmuası 2014 Vol.64 Issue 1, pp.85-111

Document Type:

Article

Publication Date:

6/2014

Document Language:

English

Document ID (Org No.):

2000510557

Frequency:

Bi-Annual

eISSN:

2602-3954

pISSN:

2602-4152

Abstract:

 

This paper aims to investigate bilateral trade flows among EU15countries from 1964 to 2003 with their determinants by using panel data analysis and neural network modeling. When we compare explanatory power of both models, it appears that neural networks can explain larger variation in bilateral exports compared to the panel data analysis. Moreover, in comparing out-of-sample forecasting performances of panel model and neural networks, it is seen that neural networks produce much lower MSE which makes them superior to the panel model. One of the main relative benefits of the neural network model is nonlinearity, as it uses sigmoid functions instead of linear functions as building blocks. This partly explains its success in our study. Another advantage of neural networks is that they make no a priori assumptions about the population distribution and the relationship between explanatory variables and the dependent variable.

 


This paper aims to investigate bilateral trade flows among EU15countriesfrom 1964 to 2003 with their determinants by using panel data analysisand neural network modeling. When we compare explanatory power ofboth models, it appears that neural networks can explain larger variation inbilateral exports compared to the panel data analysis. Moreover, in comparingout-of-sample forecasting performances of panel model and neural networks,it is seen that neural networks produce much lower MSE which makesthem superior to the panel model. One of the main relative benefits of theneural network model is nonlinearity, as it uses sigmoid functions insteadof linear functions as building blocks. This partly explains its success in ourstudy. Another advantage of neural networks is that they make no a prioriassumptions about the population distribution and the relationship betweenexplanatory variables and the dependent variable.Keywords: gravity model, panel data, neural networks, EU15, bilateral tradeTİCARET AKIMLARININ PANEL VERİ ANALİZİVE YAPAY SİNİR AĞLARI İLE TAHMİN VEÖNGÖRÜSÜÖZETBu çalışmada 15 Avrupa Birliği ülkesi arasında 1964’ten 2003’e kadargerçekleşen ticaret akımları ve bunları etkileyen faktörler panel veri analizive yapay sinir ağları modellemesi kullanılarak incelenecektir. Her iki modelinaçıklama gücü karşılaştırıldığında yapay sinir ağlarının karşılıklı ticaretipanel veri analizine göre daha iyi açıkladığı görülmüştür. Ayrıca, örneklemdışı tahmin performansları karşılaştırıldığında da yapay sinir ağlarının panelveri analizine göre çok daha düşük ortalama karesel hata verdiği tespit edilmiştir.Yapay sinir ağlarının en önemli avantajı doğrusal olmamaları, yaniyapı taşlarının doğrusal fonksiyonlar değil de sigmoid fonksiyonlardan oluşmasıdır.Bu onların çalışmamızdaki başarısını kısmen açıklar. Yapay sinirağlarının diğer bir avantajı da nüfus dağılımı ile bağımlı ve bağımsız değişkenlerarasındaki ilişki hakkında apriyori varsayımlarda bulunmamalarıdır.Anahtar Kelimeler: çekim modeli, panel veri, yapay sinir ağları, AB15,iki taraflı ticaret

.

Article Details

ESTIMATING AND FORECASTING TRADE FLOWS BY PANEL DATA ANALYSIS AND NEURAL NETWORKS

Authors: Elif NUROĞLU

Publisher: Istanbul University

Main Topics: Management

Journal: İstanbul Üniversitesi İktisat Fakültesi Mecmuası 2014 Vol.64 Issue 1, pp.85-111

Document Type: Article

Publication Date: 6/2014

Document Language: English

Document ID (Org No.): 2000510557

Frequency: Bi-Annual

eISSN: 2602-3954

pISSN: 2602-4152

Issues

Authors:

Elif NUROĞLU

Publisher:

Istanbul University

Main Topics:

Management

Journal:

İstanbul Üniversitesi İktisat Fakültesi Mecmuası 2014 Vol.64 Issue 1, pp.85-111

Document Type:

Article

Publication Date:

6/2014

Document Language:

English

Document ID (Org No.):

2000510557

Frequency:

Bi-Annual

eISSN:

2602-3954

pISSN:

2602-4152