پیش بینی منحنی عملکرد و نقش اطلاعات اقتصاد کلان در ترکیه
|کد مقاله||سال انتشار||تعداد صفحات مقاله انگلیسی||ترجمه فارسی|
|6005||2013||7 صفحه PDF||سفارش دهید|
نسخه انگلیسی مقاله همین الان قابل دانلود است.
هزینه ترجمه مقاله بر اساس تعداد کلمات مقاله انگلیسی محاسبه می شود.
این مقاله تقریباً شامل 6150 کلمه می باشد.
هزینه ترجمه مقاله توسط مترجمان با تجربه، طبق جدول زیر محاسبه می شود:
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Economic Modelling, Volume 33, July 2013, Pages 1–7
In this study we investigate the yield curve forecasting performance of Dynamic Nelson–Siegel Model (DNS), affine term structure VAR model (ATSM VAR) and principal component model (PC) in Turkey. We also investigate the role of macroeconomic variables in forecasting the yield curve. We have reached numbers of important results: 1—Macroeconomic variables are very useful in forecasting the yield curve. 2—The forecasting performances of the models depend on the period under review. 3—Considering the structural break which associates with change in monetary policy leads models to produce better forecasts than the random walk. 4—The role of exchange rate should not be ruled out in forecasting the yield curve in an emerging market like Turkey.
Forecasting the term structure of interest rates has long been of interest to financial economists, central bankers and portfolio managers since it plays a crucial role in pricing financial assets and their derivatives, managing financial risk, allocating, portfolios, structuring fiscal debt, conducting monetary policy, and valuing capital goods (Christensen et al., 2011). As the term structure of interest rates carries important information about the monetary policy and the market risk factors, numbers of theoretic and empirical researches for forecasting the yield curve are being conducted. However, as argued in Exterkate (2008), forecasting the term structure of interest rates is not an easy task and many attempts to outperform a simple random walk in forecasting the yield curve have failed. The literature on the modeling of the yield curve is mainly dominated by the no-arbitrage affine term structure models (ATSM). This literature is started by Vasicek (1977) and Cox et al. (1985). Duffie and Kan (1996) characterize and Dai and Singleton (2000) classify these models. Vasicek (1977) and Cox et al. (1985) propose a single factor model, an instantaneous short rate that drives the market, however, it produces poor yield curve forecasts (Duffee, 2002). Chen and Scott (1993) argue that one factor is not appropriate to characterize the entire yield curve and propose multifactor generalization of the CIR (Cox, Ingersoll and Ross) model. On the other hand, Duffie and Kan (1996) characterize the exponential term structure models which are a class of models that the yields are an affine function of the latent state variables. Following Duffie and Kan (1996), these types of affine models become particularly popular (Christensen et al., 2011). Dai and Singleton (2000) analyze the affine term structure models and show that the yield curve movements can be reduced to three factors. In a seminal paper, Ang and Piazzesi (2003) describe the joint dynamics of the term structure of interest rates and macroeconomic variables by an ATSM VAR. By assuming that the state vector follows a Gaussian VAR, they show that imposing the no-arbitrage restrictions and incorporating macroeconomic variables increase the forecasting performance of the VAR. As they bring in the picture the role of macroeconomic variables in dynamics of the yield curve movement, their study fills such a very important gap that the ATSM literature does not mention the role of macroeconomic variables so far. For 1-month ahead forecast horizon, their model shows better performance than the random walk but with a small gain. An alternative approach is proposed by Nelson and Siegel (1987). This approach uses statistical techniques to explain the movement in the yield curve and becomes very popular among practitioners and central banks. Diebold and Li (2006) extend the yield curve model of Nelson and Siegel (1987) to the dynamic form and show that the Dynamic Nelson–Siegel Model (DNS) compared to the many benchmark models including the ATSM and the random walk, produces superior out of sample forecasts especially for one year ahead forecast horizon. Diebold et al. (2006) extend the model of Diebold and Li (2006) by incorporating macroeconomic variables. While Diebold and Li (2006) use a two step approach, Diebold et al. (2006) propose a one step approach using the state space framework. They argue that a one step approach should improve out of sample forecasts, however, they did not provide and forecast result. Yu and Zivot (2011) investigate out of sample forecasting performance of the one step and two step DNS models and find that a one step approach does not improve forecasting performance. Instead, a two step approach provides more accurate forecasts. To assess the relative importance of no-arbitrage restrictions versus large information sets in forecasting the yield curve, Favero et al. (2012) investigate the forecasting performance of the DNS, the ATSM with a small number of macroeconomic variables and the ATSM with a large set of macroeconomic variables. Following the literature, instead of using all of macroeconomic variables individually, they extract common factors. They find that macro factors are very useful in forecasting the medium and the long rates and the financial factors are very useful in forecasting the short rates. They also show that the models with macroeconomic variables have superior forecasting performance than those of the random walk for most of the cases considered. In this study, we investigate forecasting performance of the DNS, the ATSM VAR and principal component (PC) models in Turkey. By incorporating a set of macroeconomic variables, we analyze the role of macroeconomic variables in forecasting the yield curve. Since the Turkish economy has experienced a monetary policy change in 2002 accompanied by a political change around 2002, we regard this date as a potential date of a structural break. To take the structural break into account we divide the samples into pre-2002 and post-2002 periods. In our case, the change in monetary policy is associated with the implementation of an Inflation Targeting (IT) regime. The rest of the paper consists of six parts. The first part describes the data set. The second part provides a general framework for forecast. The third part presents models and estimation techniques. The fourth part provides forecasting procedure. The fifth part presents the empirical findings and the last part concludes.
نتیجه گیری انگلیسی
In this study we investigate the yield curve forecasting performance of three specifications namely Dynamic Nelson–Siegel Model (DNS), no-arbitrage affine term structure VAR (ATSM) model and principal component (PC) analysis in Turkey. To take into account the structural break, we divide our sample as pre-2002 and post-2000 period. For the entire period and two sub-periods, firstly we investigate the forecasts of the yield-only models, i.e. we use only extracted yield factors for forecasting, and then we incorporate two sets of macroeconomic information; the first set contains inflation rate, output gap and policy rate and the second set contains inflation, output gap and exchange rate. The out of sample forecasting experiment shows that the forecasting performances of the models are time varying. In the entire period, the yield-only models cannot beat the random walk but in the pre-2002 sub-period the yield-only models produce superior forecasts than the random walk. For the post-2002 sub-period, the yield-only PC model can beat the random walk for medium and long forecasting horizons. We find that, the macro factors are very useful in forecasting the yield curve. For the entire period, the first macroeconomic information set significantly improves the forecasting performance of the models. For the pre-2002 period, replacing the policy rate with the exchange rate produces better forecasts. However, for the post-2002 period both the macroeconomic information sets rate produce similar results.