Browsing by Subject "Bayesian estimation"
Now showing items 1-2 of 2
-
(2022)During the past decade, central banks have become even more central to modern economies than before. Their main goal is price stability, and they try to achieve it with various methods. The effectivity of the traditional methods, especially controlling the short-term interest rates, has become smaller due to the zero lower bound constraint, and new unconventional methods have been introduced. This thesis investigates the effects of the unconventional monetary policy actions of the European Central Bank. Structural vector autoregressive models are one of the main tools in studying the effects of the monetary policy. To identify a monetary policy shock, one traditionally needs to impose restrictions to the model. However, that requires restrictive assumptions about the dynamics of the model which is being studied. To overcome this issue, one can identify the shocks statistically by some properties of the data, without any additional restrictions. A key benefit of statistical identification is the possibility to test the plausibility of previously used sign or zero restrictions. In the empirical application of this thesis, a Bayesian vector autoregressive model identified statistically based on non-normality of the error terms is utilised to study the effects of the European Central Bank's unconventional monetary policy. The Bayesian estimation was performed with a Differential Evolution Markov Chain algorithm, allowing fast calculation. The empirical analysis resulted in two key findings. First, the impulse response functions implied by the model are in line with previous studies and the statistically identified model gives support to the previously used sign restrictions. Second, the model is sensitive to the sample period which suggests that the effects of the European Central Banks policy actions might have changed over time.
-
(2023)Over the past decade, central banks have engaged in strong monetary stimulus in response to the aftermath of the financial crisis that began in 2007. On one hand, expansionary monetary policy has enabled economies to return to a growth path, but on the other hand, it has also artificially inflated financial securities prices. While monetary policy has been widely studied from the perspective of real economic, the evidence of its effects on financial intermediates, namely the banking sector, is still limited. This topic has important policy implications, due to banks’ ability to generate adequate profits being relevant for the sustainability of the banking system and, as such, for its capacity to provide sufficient credit to the economy and protect depositors’ funds. The purpose of this study is to measure and reflect on the relationship between these two entities. In monetary policy research, the structured vector autoregressive (SVAR) analysis is widely used to interpret reactions of economic variables to structured shocks. Various identification strategies can be employed to ascertain structured shocks. One, quite robust identification approach is to utilize characteristics of the data and obtain a statistical identification approach. This identification method has advantages over others that seek to specify the model a priori based on restrictions on the shock dynamics. Imposing such restrictions in advance is not unambiguous. Usually, prior restrictions seek support from economic theory. Eventually, the results and reactions of variables to shocks can be interpreted using impulse responses. The empirical analysis of this study was conducted using a Bayesian SVAR model. The model was statistically identified, assuming non-Gaussian error terms. Estimation was performed using a Markov Chain algorithm, more specifically a Non-U-Turn Sampler (NUTS), which allows for efficient estimation of multiple chains simultaneously. The analysis partly supports previous dichotomic evidence. Based on the impulse responses, an unconventional monetary policy shock temporarily increased the value of the banking variable by approximately 1%. This supports previous research evidence that negative interest rate effects can be compensated for by changes in business models and by making non-interest-dependent businesses more profitable.
Now showing items 1-2 of 2