Mátyás Farkas, European Central Bank
“I rely heavily on the experience and understanding of structural models I gained while working and researching at the IMFS”
Mátyás Farkas obtained his Ph.D. in Finance at the Graduate School of Economics, Finance, and Management at Goethe University in February 2020. Mátyás currently works in the ECB's Monetary Policy Strategy division as an economist. In this position, he provides input to the Governing Council’s regular monetary policy decisions and contributes to the ECB's strategy review. In September 2018, Mátyás joined the ECB’s Graduate Programme, where he is worked in the Monetary Policy Research, Market Infrastructure Management and the Forecasting and Policy Modelling divisions. From December 2014 until his start at the ECB, Mátyás was an integral member of the Macro Model Base Team. He became the backend developer of the Macroeconomic Model Data Base, contributing to the platform’s newly introduced compatibility under multiple operating systems and to its transmission into a web-based, platform-independent comparison tool and database. Mátyás also co-developed the real-time DSGE forecast comparison platform for the MMCN's forecast competition. His research focused on the role of expectation and monetary policy in DSGEs, on practical questions of DSGE estimation. The Hungarian-born researcher acquired his M.Sc. degree in Money and Finance from Goethe University in 2013. He pursued his undergraduate studies at Corvinus University Budapest and holds two Bachelor’s diplomas, one in Finance and Accounting and one in International Relations.
How would you describe your job to other people?
My job is, in layman's terms, to condense the insights of technical models into policy relevant simple messages. A key challenge of my work is to be able to anticipate and identify needs that will arise from the policy debate, and contribute with clear policy implications derived from complex models. In particular my work is to find a translation of the current and future economic and financial challenges into a policy response supported by economic analysis and models.
What do you like most about your job?
The vast spectrum of issues to be addressed. I also enjoy that my work is central to the monetary policy decision making: I am directly exposed to the policy debate and working with outstanding, experienced economists of the field which provides plenty of opportunity to develop. I not only need to keep up with methodological advances, but integrate them to my analysis anticipating stakeholder needs. This usually translates into a thrill of finding the balance of exhaustive research and feasibly and timeliness.
What was the main focus of your research at the IMFS?
My research at the IMFS had two foci, first, a theoretical one, studying on the role of expectations, risks and unconventional monetary policy; second, a practical one, developing methods to estimate DSGEs and assess their forecast performance. My theoretical research shows that expectations and sources of risks are linked (unconventional) monetary policy. As an example, I would mention that our paper with Michael Binder and Volker Wieland. In it we studied the impact of heterogeneous, higher-order beliefs on business cycle dynamics; highlighting that diverse beliefs can become an endogenous source of risk to the economy and showed that heterogeneous expectations models can be solved using rational expectations solution techniques.
My job market paper was also theoretical endeavour. In it I combined adaptive learning with heterogeneous expectations to study efficacy of forward guidance and with it central bank credibility: providing a solution to the forward guidance puzzle. I showed that a model with endogenous belief switching, a form of adaptive beliefs, can both solve and nest the forward guidance puzzle: delivering the insight that self-fulfilling beliefs can be the result of central bank action.
On the second, practical workstream, I have been a researching questions related to DSGE estimation: I have co-developed the real-time DSGE forecast comparison platform for the MMCN's forecast competition and co-authored the paper with Michael Binder, Zexi Sun, John Taylor, Volker Wieland and Maik Wolters “Forecasting the Great Recession in the United States: First Results from a Model Comparison Exercise”.
Furthermore, in a joint paper with Bálint Tatár titled “Bayesian Estimation of DSGE Models with Hamiltonian Monte Carlo”, published as IMFS Working Paper 144, we proposed a novel method to estimate DSGEs using the Hamiltonian Monte Carlo (HMC) method. We show that HMC is not only more efficient than the benchmark Random Walk Metropolis Hastings method, it enables to identify estimation inefficiencies and problems, like multimodality, at the level of individual structural parameters. We show that the original Smets Wouters model features a second mode where monetary policy has lower efficacy, one that standard estimation methods so far were unable to show.
How is your job at the ECB related to your work at the IMFS?
In day to day work I rely heavily on the experience and understanding of structural models I gained while working and researching at the IMFS. As an example, I could bring the insights from the model comparison of the MMB: the knowledge of a huge collection of DSGE models makes me well suited to address policy questions. In particular, my experience with robust monetary policy rules helps with aspects of the strategy reviews, like optimal policy designs and the role of backward-looking expectations. Furthermore, my software development skills of coding, testing, documenting, all acquired on the job at the IMFS, come very handy when we productify analysis supporting the policy decisions.
What did you enjoy most regarding your time at the IMFS?
I appreciate the intellectual freedom and the opportunities for self-development I experienced at the IMFS. The IMFS as a research institute excelled at providing possibilities to explore challenging questions. For example, entasking Ph.D. students with difficult tasks coupled with an outstanding team to support the work was and continues to be the IMFS’s recipe for success. I enjoyed the sprawling exchange with colleagues at the IMFS, the hours we brainstormed and worked through to test ideas and show proofs of concept.
Lastly, I also enjoyed the educational aspect of the work at the IMFS: I was expected to learn DSGEs and develop them at the same time, this fostered understanding beyond what is expected at on the job learning, but what is a prerequisite for a serious research contribution.