ECONOMETRICS CONFERENCE


International Econometrics Conference - Center for Economics and Econometrics, Bogazici University

21 JUNE 2014
IBRAHIM BODUR HALL, NATUK BİRKAN BUILDING

9:00-9:15      Registration

9:15-10:00    Keynote Speech: Rafaella Giacomini (University College London) "GMM with Latent Variables"

10:00-10:30   Coffee Break

10:30-12:30  Econometric Theory I

Mehmet Caner (North Carolina State University) “Uniform Confidence Intervals for High Dimensional Parameter Case”
Andreea Enache: (Toulouse School of Economics)  “Nonparametric estimation of a class of contract theory”
Rehim Kılıç: (Atlanta Fed) “Robust Inference in Smooth Transition Predictive Regressions”
Selin Öztürk: (Bilgi University) “Testing for structural breaks with local smoothers: A  simulation study”

12.30- 14:00   Lunch at Kennedy Lodge

14:00-16:00   Econometric Theory II

Burak Saltoğlu (Bogazici University) “An Anatomy of a Systemic Banking Crisis ”
Tsengos. T. (University of Guelph)  “Markov regime switching in mean and in fractional integration parameter”
Mirza Trokic: (Bilkent University) “Functional Coefficient Models for Nearly (Possibly  Weakly)  I(1) Processes”
Cavit Pakel: (Bilkent University) “Bias Reduction in Nonlinear and Dynamic Panels in the Presence of Cross-Section Dependence, with a GARCH Panel Application”

16:00-16:30pm  Coffee break

16:30-18:30pm  Applied Econometrics

Bill Barnett: (University of Kansas) “An Analytical and Numerical Search for Bifurcations in Open Economy New Keynesian Models”
Resul Aydemir (Istanbul Technical University) “Volatility Spillovers and Dynamic Interactions between exchange rates and interest rates in the fragile five”
Kurmas Akdoğan: (TCMB) “Asymmetric Behavior of Inflation around the Target in Inflation-Targeting Emerging Markets”
Nurullah Gür: (Istanbul Ticaret University) “Financial System, R&D Intensity and Comparative Advantage”

Abstracts:

Mirza Trokic (Bilkent University)
Title: “Functional Coefficient Models for Nearly (Possibly Weakly) I(1) Processes

The focus of this article is on nonlinear time varying coefficient models when the covariates and coefficient components are weakly, nearly, or possibly purely integrated time series processes. Local linear fitting is used to derive coefficient estimators along with their asymptotic distributions. The rates of convergence for the estimators is shown to differ based on whether stationary, weakly, nearly, or purely integrated covariates are being modelled. Similar conclusions also hold for the derived optimal bandwidth parameters.

Andreea Enache (Toulouse School of Economics)
Title: “Nonparametric estimation of a class of contract theory” (joint work with Jean-Pierre Florens, Toulouse School of Economics)

The structural econometric approach to models of principal-agent has been an area very little explored by researchers, maybe because of the difficulties in uniquely recovering the primitives of the model from the data. The novelty introduced by our paper is twofold. Firstly, we locally identify the static version of a classical adverse selection model. Although this has been done previously, our contribution comes from the fact we present also the estimation issues and we compute the speed of convergence for our estimator. Within the static approach we present two extensions of the model accounting for different semi-parametric specifications of the cost function of the agent. Secondly, we analyse in a dynamic setting an example of adverse selection problem in a principal-agent framework (we take the classical case of an uninformed regulator) in which the types of the agents evolve over time both in an independent and in a correlated manner.

Bill Barnett (University of Kansas)
An Analytical and Numerical Search for Bifurcations in Open Economy New Keynesian Models” (joint work with William A., Barnett University of Kansas and Center for Financial Stability, and Ünal Eryılmaz, OECD)

We explore bifurcation phenomena in the open-economy New Keynesian model developed by Gali and Monacelli (2005). We find that the open economy framework brings about more complex dynamics, along with a wider variety of qualitative behaviors and policy responses. Introducing parameters related to the open economy structure affects the values of bifurcation parameters and changes the location of bifurcation boundaries. As a result, the stratification of the confidence region, as previously seen in closed-economy New Keynesian models, remains an important research and policy risk to be considered in the context of the open-economy New Keynesian functional structures. In fact, econometrics and optimal policy design become more complex within an open economy. Dynamical inferences need to be qualified by the risk of bifurcation boundaries crossing the confidence regions. Policy design needs to take into consideration that a change in monetary policy can produce an unanticipated bifurcation, without adequate prior econometrics research.

Thanasis Tsengos (University of Guelph)
Markov regime switching in mean and in fractional integration parameter” (joint work with Ege Yazgan and Harun Özkan, Bilgi University)

We propose a specific general Markov-regime switching estimation both in the long memory parameter d and the mean of a time series. Following Tsay(2009) we employ Durbin-Levinson-Viterbi(DLV) algorithm, which combines the Durbin-Levinson and Viterbi procedures, in two state Markov-switching parameter estimation. It is well-known that existence of mean break and long memory in time series can be easily confused with each other in most cases. Thus, we aim at observing the deviation and interaction of mean and d estimates for different cases. A Monte Carlo experiment reveals that the finite sample performance of the proposed algorithm for a simple mixture model of Markov-switching mean and d changes with respect to the fractional integrating parameters and the mean values for the two regimes.

Nurullah  Gür (İstanbul Ticaret University)
Financial System, R&D Intensity and Comparative Advantage”

In this paper, we test whether financial system affects export performance of R&D intensive industries. We consider four different dimensions of financial system: (a) financial development, (b) financial liberalization, (c) financial integration and (d) foreign banks. Our results show that financial development and financial integration increase exports relatively more in R&D intensive industries. These effects are highly robust. Financial liberalization and foreign banks do not have such effects. Our results also show that the positive effect of financial integration disappears when the quality of institutions and the level of financial development are low.

Resul Aydemir (Istanbul Technical University)
Volatility Spillovers and Dynamic Interactions between exchange rates and interest rates in the fragile five” (joint work withBülent Güloğlu , Istanbul University,   Ercan Sarıdoğan , Istanbul Technical University)

The bursting of the U.S. housing bubble caused the values of subprime mortgage-based securities to plummet, which in turn triggered the 2008 global financial crisis due to liquidity problems in the financial system. The crisis reached its peak when Lehman Brothers declared its bankruptcy on 15th of September, 2008. To avoid the risk of a financial collapse, the U.S. Federal Reserve (Fed) has taken steps to launch its quantitative easing programme (i.e., creating money and buying bonds and other financial assets from banks), which was ended on 18th of December, 2013 when the Fed announced its first tapering (It would reduce its purchases beginning in January 2014). These historical shocks are well known to have had huge impacts on the foreign exchange, money and credit markets, especially in such countries as Brazil, India, Indonesia, South Africa and Turkey, recently called as the Fragile Five. In this paper, we will address three issues for each country in the Fragile Five: a) how these shocks affect the size and persistence of the volatilities in exchange rates and interest rates, b) dynamic interactions between exchange rates and interest rates, c) dynamic interactions between the US interest rates and domestic interest rates. To that end, we will first estimate multivariate GARCH model and derive conditional variances and dynamic (time varying) conditional correlations with covariances. Then, we will analyze the effects of these historical shocks and also hypothetical shocks that may arise in the future on the volatilities of exchange rates and interest rates.  We will utilize volatility impulse response functions developed by Hafner and Herwartz (2006) to achieve that objective. 

Kurmas Akdoğan (Central Bank of the Republic of Turkey) Asymmetric Behaviour of Inflation around the Target in Inflation-Targeting Emerging Markets

We explore the asymmetric behaviour of inflation around the target level for inflation-targeting emerging markets. The first rationale behind this asymmetry is the asymmetric policy response of the central bank around the target. Central banks could have a stronger bias towards overshooting rather than undershooting the inflation target. Consequently, the policy response would be stronger once the inflation jumps above the target, compared to a negative deviation. Second rationale is the asymmetric inflation persistence. We suggest that recently developed Asymmetric Exponential Smooth Transition Autoregressive (AESTAR) model provides a convenient framework to capture the asymmetric behaviour of inflation driven by these two effects. We further conduct an out-of-sample forecasting exercise and show that the predictive power of AESTAR model for inflation is high, especially at long-horizons.

Selin Öztürk (Bilgi University)
Testing for structural breaks with local smoothers: A simulation study

Testing for parameter stability has been an active area of research in econometrics. Recently Chen and Hong (2012) (CH hereafter) developed a consistent test for smooth structural changes as well as abrupt structural breaks with known or unknown change points. Their test is based on a contrast of estimators approach in the contest of a smooth time-varying coefficient model with the contrasting estimators being the estimated regression function (fitted values) of a restricted constant parameter model and the unrestricted time-varying parameter model. Within the same framework by means of an extensive Monte Carlo simulation study based on the original design of HC we provide evidence on the performance of the CH test with other local-smoother based consistent tests in the literature notably the linearity test of Li, Huang, Li and Fu (2002) (hereafter LHLF) and the functional form test of Li and Wang (1998), (hereafter LW). The LHLF test adapted to the case of parameter stability performs favorably well to HC test. Also the test of Li and Wang does well even though it is designed to test against general nonlinear alternatives and not specifically against unknown structural breaks. The Monte Carlo comparison includes both the asymptotic and bootstrap versions of the different test statistics examined.

Cavit Pakel (Bilkent University)
Bias Reduction in Nonlinear and Dynamic Panels in the Presence of Cross-Section Dependence, with a GARCH Panel Application

Nonlinear dynamic panel data models which contain individual-specific parameters are known to suffer from the incidental parameter issue. Although detailed analysis of this bias exists for panels with time-series dependence, the case of cross-section dependence has not yet been investigated. In this paper, I investigate the bias of the integrated likelihood method, which nests many common estimation methods as specific cases, allowing for dependence across both dimensions in a large-N large-T setting. Theoretical analysis reveals that the presence of weak time-series dependence does not lead to an extra bias term. However, cross-section dependence leads to a new type of bias the magnitude of which is determined by the strength of dependence. In particular, the bias is negligible under weak and cluster-type dependence, but not under strong dependence. Analytical expressions for the bias terms are also derived. These findings are utilised to propose a new estimation approach for modelling volatility in small samples using a panel of financial returns. Monte Carlo analysis reveals that the proposed method successfully fits GARCH with little bias and little or no increase in variance using 150-200 time-series observations, compared to around 1,000 observations required for successful GARCH estimation by standard methods. Finally, I consider two empirical illustrations; an analysis of monthly hedge fund volatility characteristics and a test of predictive ability using daily stock volatility forecasts.

Rehim Kılıç (Atlanta Fed-Emory University)
Robust Inference in Smooth Transition Predictive Regressions

This paper extends the traditional linear predictive regression with local-to-unit root regressors to a smooth transition predictive regression (STPR) framework that allows for regime-specific predictability and develops inference procedures for testing predictability within these models which are robust to the presence of local-to-unit root predictors and endogeneity of regressors. In order to circumvent the unidentified parameters problem that arise under the null of no-predictability, $t-$ test for the slope coefficient in the STPR model is optimized over the Cartesian product of the spaces for unidentified parameters; and to address the difficulties due to endogeneity of persistent regressors, the paper extends the instrumental variables (IVX) method proposed in cointegration testing literature. Limit distribution of the proposed  test using the IVX methodology is shown to be nuisance parameter-free. Simulations show that proposed test have excellent size and power properties. An application to stock return predictability reveals presence of regime-specific predictability and the usefulness of the proposed testing approach.

Mehmet Caner (North Carolina State University)
Uniform Confidence Intervals for High Dimensional Parameter case” (joint with Anders Bredahl Kock, Aarhus University)

The paper develops confidence intervals uniformly over parameter space when there are high dimensions, p>n. This extends the landmark paper of Buhlmann et al (2013) from the de-sparsified lasso to conservative lasso. Since lasso type estimators are not uniformly consistent, but very useful in shrinkage and model selection, these new estimators use their property. The new estimators augment the lasso type estimators with a relaxed inverse of the variance covariance matrix. We extend some of the results in Buhlmann et al (2013) paper, such as a new way of getting relaxed inverse. Our method is a weighted lasso with the aim of preventing false negatives (wrong choice of zeros, wrong shrinking of nonzeros to zeros). Compared to adaptive lasso in hi dimensional setting, this can get a full set of confidence intervals for each p parameters.

Burak Saltoğlu (Boğazici University)
An Anatomy of a Systemic Banking Crisis (joint with Taylan Yenilmez, Tinbergen Institute)

We have analyzed a systemic liquidity crisis by using a unique money market data set in which the coded identity of the counter parties of every trade is known. On the contrary to the recent findings, we could not observe a positive relationship between interconnectivity and systemic risk prior to the financial crisis. We have concluded that our conicting findings can be related to the degree of market concentration on the borrowing side of the funding market. Therefore, by using various statistical tools, we have investigated the heterogeneity between the borrowing and lending sides. The asymmetric treatment of heterogeneity for borrowing and lending sides, has revealed useful information in monitoring the dynamics of systemic risk. The high level of concentration in the borrowing side compared to the lending side has led to a lower interconnectivity but a higher systemic risk prior to the crisis. Our time series investigation has shown that, borrowing and lending centrality differential and reciprocity measures have made statistically significant contribution to the missing link between systemic risk and interconnectivity relationship. Based on our findings, we conclude more complementary tools should be used to generalize the relationship between systemic risk and interconnectivity.

News & Announcements


Click for news & announcements archive


Events


Click for events archive