Evaluation of Türkiye’s Domestic Prices and Exchange Rate Relationship with State Space Models
Fikriye Ceren Bostancı, Selçuk KoçIn this article, an economic analysis of Türkiye is discussed to introduce State Space Models. State Space Models offer a preferred modelling method as they have flexible structures. As an economic application of this study, an analysis is conducted with Türkiye’s monthly Consumer Price Index (CPI), USD/TRY and EUR/TRY for the period 2005-2024. In the first stage, a modelling approach with only explanatory variables is adopted. Subsequently, in addition to the models with only explanatory variables, a dummy variable was added to be valid from January 2023 onwards. Deterministic and stochastic features were considered in the analysis, and these different model structures were estimated. As a result, it is concluded that the best model structure is the Stochastic Level and Explanatory Variable-Dummy Variable model for the USD/TRY and EUR/TRY. The results show that the effects of USD/TRY and EUR/TRY on the CPI are similar and positive. However, the CPI is found to be relatively less sensitive to fluctuations in the USD/TRY and EUR/TRY. In addition, the use of a dummy variable in the model allowed the comparison of domestic prices between the two periods. Accordingly, it is inferred that domestic prices in the first month of 2023 and thereafter are lower than those in the previous periods. These findings have important implications for a better understanding of the relationship between exchange rates and domestic prices in Türkiye and contribute to the literature in terms of demonstrating the use of State Space Models in an economic application.
Durum Uzayı Modelleri ile Türkiye’nin Yurtiçi Fiyatları ve Döviz Kuru İlişkisinin Değerlendirilmesi
Fikriye Ceren Bostancı, Selçuk KoçBu çalışmada, Durum Uzayı Modellerinin (State Space Models) tanıtılması amacıyla bir Türkiye iktisadi analizi ele alınmaktadır. Durum Uzayı Modelleri, esnek yapılara ve veri kaybını önleyen bir tahmin yöntemine sahip olması sebebiyle tercih edilen bir modelleme yöntemi sunmaktadır. Bu çalışmanın iktisadi uygulaması olarak, Türkiye’nin 2005-2024 dönemine ait aylık Tüketici Fiyat Endeksi (TÜFE), Dolar/TL ve Euro/TL değişkenleri ile bir analiz yapılmıştır. Analizin ilk aşamasında, yalnızca açıklayıcı değişken içeren bir modelleme yaklaşımı benimsenmiş ve TÜFE, Dolar/TL, Euro/TL gibi ekonomik göstergeler bireysel olarak ele alınarak analiz edilmiştir. Ardından, daha karmaşık bir modelleme sürecine geçilmiştir; bu süreçte yalnızca açıklayıcı değişken içeren modellerin yanı sıra, 2023 yılının Ocak ayından itibaren geçerli olacak bir kukla değişken eklenmiştir. Kukla değişken eklenerek yapılan ikinci modellemede, belirli dönemlerde gerçekleşen yapısal kırılmalar ve politika değişikliklerinin veriler üzerindeki etkisi daha kapsamlı bir şekilde incelenmiştir. Analizde deterministik ve stokastik özellikler de dikkate alınmış ve bu farklı model yapıları Kalman Filtresi (1960) yöntemi kullanılarak tahmin edilmiştir. Bu analizler sonucunda, en iyi model yapısının Dolar/TL ve Euro/TL değişkenleri için Stokastik Seviye ve Açıklayıcı Değişken-Kukla Değişken modeli olduğu sonucuna ulaşılmıştır. Elde edilen sonuçlar, Dolar/TL ve Euro/TL değişkenlerinin TÜFE üzerindeki etkisinin benzer oranda ve pozitif yönde olduğunu göstermektedir. Bununla birlikte TÜFE’nin Dolar/TL ve Euro/TL değişkenlerindeki dalgalanmalara nispeten daha az hassasiyet gösterdiği; başka bir deyişle, Dolar/TL ve Euro/TL değişimlerine göre daha düşük esneklik sergilediği tespit edilmiştir. Ayrıca modelde kukla değişken kullanılması iki dönem arasında yurt içi fiyatlarının karşılaştırılmasına olanak tanımıştır. Buna göre 2023’ün ilk ayı ve sonrasındaki yurtiçi fiyatlarının önceki dönemlere göre daha düşük olduğu çıkarımı yapılmıştır. Bu bulgular, Türkiye’de döviz kuru ve yurt içi fiyatları arasındaki ilişkiyi daha iyi anlamak için önemli çıkarımlar sunmaktadır ve Durum Uzayı Modellerinin iktisadi bir uygulamada kullanımında göstermesi bakımında literatüre katkı sunmaktadır.
This study examines the relationship between domestic prices and exchange rates in the Türkiye economy and uses State Space Models and the Kalman Filter method to analyse this interaction. State Space Models are highly flexible models used in economic analysis and prevent data loss. This feature makes state-space models very useful in time-series analysis. In this framework, the Kalman filter is used in the estimation of the state space models. State Space Models and the Kalman Filter make important contributions in revealing economic relationships that are difficult to measure.
In this article, first, the theory of the economic issue is discussed and then the general representation of State Space Models is analysed. Then, the structures of the State Space Models according to different specifications are discussed separately and the model structures are explained through equations. Then, the Kalman Filter method, which is the estimation method of State Space Models, developed by electrical engineer Rudolf Emil Kalman in 1960 for use in control engineering, is explained. The feature of the Kalman Filter that makes it very useful in the use of economic series is that it can be applied to both stationary and unit rooted series. Due to these useful features, the Kalman Filter was adapted to the economic series by Andrew Harvey in 1989 for use in econometric analyses.
This study analyzes the relationship between Türkiye’s Consumer Price Index (CPI) and the USD/TRY and EUR/TRY exchange rates separately over a period extending from 2005 to 2024. Studying this long period provides a detailed analysis of the relationship between the CPI and USD/TRY and EUR/TRY. The analysis is initially carried out using State Space Models with only explanatory variables and investigates the direction and degree to which the USD/TRY and EUR/TRY exchange rates are driving the CPI. In addition to this modelling, a dummy variable representing the first month of 2023 and the following months was added to the model, and it was investigated how the dummy and explanatory variables explained the CPI variable. The purpose of adding the dummy variable to the model is to include the structural breaks and policy changes in the Turkish economy in the model. For both models, adding a dummy variable to the model made a positive contribution to the model.
In order to analyse the State Space Models with Explanatory Variables and State Space Models with Explanatory Dummy Variables mentioned in the above paragraph in a more holistic sense, the deterministic and stochastic properties of the explanatory and dummy variables are included separately in the model. In this way, the possibility of correctly selecting the model type that gives the best model result has increased. It was determined that the model that gave the best result was the Stochastic Level Explanatory Variable-Dummy Variable Model, and this model was accepted as the final model.
According to the model results, it is concluded that the USD/TRY and EUR/TRY variables positively affect the CPI variables. More specifically, a 1% change in the USD/TRY increases the CPI by approximately 0.1752%, while a 1% change in the EUR/TRY increases the CPI by approximately 0.1768%. The values are very close to each other, but it is observed that the EUR/TRY affects the CPI more than the USD/TRY. This is thought to be due to the fact that Türkiye imports mostly from European countries. These values also provide the elasticities. Accordingly, it can be interpreted that the CPI is inelastic against USD/TRY and EUR/TRY. This shows that domestic prices react little to the fluctuations in USD/TRY and EUR/TRY. Since there is also a dummy variable in the model, it is possible to compare the domestic prices of the two periods. Accordingly, in the CPI-USD/TRY model, Türkiye’s domestic prices are 0.04378% lower for January 2023 and later than the other dates, while in the CPI-EUR/TRY model, Türkiye’s domestic prices are 0.04375% lower for January 2023 and later than the other dates. It is noteworthy that these values are quite close to each other. This means that the policy changes that have started to be experienced since the first month of 2023 have a decreasing effect on domestic prices.
At the end of the article, it is underlined that Türkiye, like every developing country, is quite sensitive to external shocks. To become more resilient to these external shocks, policies that encourage domestic production and exports are suggested.
