Back to search

Athens University Economics and Business (AUEB)

Essays on the predictive ability of the yield curve in the aftermath of the global financial crisis: a nonlinear approach

Abstract

dc:description

Several studies have established the predictive power of the yield curve, ie: the difference between long and short term bond rates and the role of asymmetries in the term structure of interest rates, in terms of real economic activity, for the U.S. and various European countries. In this thesis, we reexamine the dynamic relationship between interest rates of different maturities for two countries, USA and Germany. We study the power of the Expectation Hypothesis before and after the European economic crisis of 2008 using monthly data on , euro rates for Germany and the United States over the period January 1 1999 through August 1 2015. We find that the expectations hypothesis of the term structure (EHT) implies cointegration between interest rates of different maturities. The interest rate data with five maturities move together which is in line with the predictions of the expectations hypothesis theory. As we have shown, that the interest rate series have long run equilibrium, we investigate the short run dynamic adjustments. Relying on recent advances in econometric methodology, we allow for the underlying market linkages to be subject to regime switching effects. We found strong evidence of the presence of the nonlinearity and asymmetry in the monthly USA and Germany interest rates.Next we explored the importance of the three factors, spread, 3m interest rate and curvature, of the yield curve and also the importance of other financial indicators such as the monthly growth rate of Chicago Board Options Exchange Market Volatility Index, the OECD Composite Leading Indicators Index and the Money Supply (as measured by RM2), as leading indicators of the economic activity in forecasting future industrial production growth. Overall, when using the data series from 1985‐2000 and 1985-2007 for USA and 1999-2009 for European countries, in‐sample results suggest in many cases the yield spread, the 3M T-Bill, the curvature and the OECD index are indeed important and have significant predictive power when forecasting industrial production growth over a six month time horizon but the yield spread is not significant over a one year time horizon for the same periods.We also analyze the asymmetry in the relationship between the financial indicators, as they defined above and economic activity. In particular, linear and nonlinear (STAR) models present significant predicting industrial production respectively from 1985 to 2015. The nonlinear models appear to outperform the linear ones in the in-sample estimations as the ratio between regression standard errors evidences. Recursive out-of-sample forecasts are also implemented comparing again the linear and the nonlinear models. The Diebold-Mariano forecast accuracy tests,namely, the DM test, exhibits a significance difference between the two competing models.On the one hand, in sample forecasts using the STAR models produces more accurate values of future industrial activity of the industrial activity. On the other hand out of sample forecasts using linear models produce more significant forecasts than STAR models.The idea of evolutionary programming and genetic algorithms is inspired by natural genetic variation and selection. In financial applications, we can utilize genetic algorithms to optimize trading rules or output growth and their parameters. In our work, we investigated the forecasting performance of the TAR family nonlinear models (SETAR, STAR, LSTAR etc.) vis-à-vis well-known linear benchmarks for four Eurozone countries. Two northern countries, Germany and France and two southern countries, Italy and Spain. Moreover, we introduced a novel hybrid parameter estimation approach based on evolutionary genetic algorithms and Least Squares as opposed to well-known method of Generalized Least Squares or simple OLS, commonly used for TAR model specification and forecasting. Based on econometric and profitability criteria we confirmed the argument by Stock and Watson (2003) that forecastability is also dependent upon the optimality of the estimated parameters in-and out-of-sample. Additionally, we revealed that the hybrid evolutionary optimization method generates better forecasts for the examined output growth based on the optimal parameters produced.Our predicting results may have important implications for numerous reasons. Business firms may be helped in deciding how much capacity will be needed to meet future demand. Government agencies will be helped when forecasting budgetary surpluses or deficits and Central Banks in deciding the stance of current monetary policy. If robust forecasting is feasible as shown in our study, investors could hedge and diversify their assets, and policy-makers could dynamically adjust their policies.

Degree

thesis:*
Grantor dc:publisher
Athens University Economics and Business (AUEB)
Year dc:date
2016

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Avdoulas, Christos
  • Αβδούλας, Χρήστος

Subjects

dc:subject × 14

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/40275

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Avdoulas, Christos; Αβδούλας, Χρήστος. Essays on the predictive ability of the yield curve in the aftermath of the global financial crisis: a nonlinear approach. Athens University Economics and Business (AUEB), 2016. http://hdl.handle.net/10442/hedi/40275