Ruben Hipp
Econometrics, Time Series, Empirical Finance
I am principal economist in the financial stability department of the Bank of Canada and my interests are in statistics, econometrics and finance. In my research, I work on financial stability and use Time Series Analysis, especially time varying parameters, structural VARs and regularization methods.
You can download a pdf version of my CV here.
Research
On Causal Networks of Financial Firms
Structural Identification via Non-parametric Heteroskedasticity Modeling
the latest version can be downloaded here, previously appeared as SWP
Abstract
We investigate the causal structure of financial systems by accounting for contemporaneous relationships. To identify structural parameters, we introduce a novel non-parametric approach which exploits that most financial data empirically exhibit heteroskedasticity. The identification works locally and, thus, allows structural matrices to vary smoothly with time. With this causality in hand, we derive a new measure for systemic relevance. An application on volatility spillovers in the US financial market demonstrates the importance of structural parameters in spillover analyses. Finally, we highlight that the COVID-19 period is mostly an aggregate crisis, with financial firms' spillovers edging slightly higher.
Estimating Large-Dimensional Connectedness Tables: The Great Moderation through the Lens of Sectoral Spillovers
joint with Felix Brunner, submitted (R&R)
the latest version can be downloaded here, previously appeared as SWP
Abstract
We estimate sectoral spillovers around the Great Moderation with the help of forecast error variance decomposition tables.
Obtaining such tables in high dimensions is challenging since they are functions of the estimated vector autoregressive coefficients and the residual covariance matrix. In a simulation study, we compare various regularization methods for both and conduct a comprehensive analysis of their performance. We show that standard estimators of large connectedness tables lead to biased results and high estimation uncertainty, which can both be mitigated by regularization. To explore possible causes for the Great Moderation, we apply a cross-validated estimator on sectoral spillovers of industrial production in the US from 1972 to 2007. We find that the spillover network has considerably weakened, which hints at structural change, e.g., through improved inventory management, as a critical explanation for the Great Moderation.
Estimated spillovers:
Connectedness networks for the respective periods. The force-directed graph drawing algorithm arranges the nodes. That is, two nodes appear closer in the graph if they have stronger connections to each other. Although we initialize the subgraphs on the same scale, the algorithm cannot guarantee that the graphs are comparable in size. The size of the node relates to the respective average weight of the sector in the IP index. The colors depict the out-connectedness. For the sake of the visualization, we cap the color scale at 1. The sectors with the highest out-connectedness are labeled.
Decomposing Systemic Risk: The roles of Contagion and Common Exposures
joint with Grzegorz Halaj
the latest version can be downloaded here
Abstract
We estimate the effect of contagion and common exposure through the lens of a structural model derived from the balance sheet identity. Banks' capital vary endogenously via value functions of assets and liabilities. Through a regression approach inspired by the structural VAR literature, we infer the interdependence of banks' financial conditions. Contagion, in this model, can occur through direct exposures, fire sales, and market-based sentiment, while common exposures result from portfolio overlaps. We apply this model to granular balance sheet and interbank exposure data of the Canadian banking market. First, we document contagion as a time-varying phenomenon, with the highest levels around the Great Financial Crisis (GFC) in 2008 and lower levels for the pandemic period. Second, we find that after the introduction of Basel III the relative importance of risks has changed, hinting that sources of systemic risk have changed structurally. Our new framework complements traditional stress-testing exercises focused on single institutions by providing a holistic view of risk transmission.
Local Stationarity within Network Dynamics
Statistical Inference for Financial Connectedness
joint with Carsten Jentsch
Abstract
In the setting of financial networks, institutions build links through stacks of contractual obligations, which are unobserved and need to be estimated for systemic risk inference. Diebold and Yilmaz (2014) introduced a popular connectedness measurement and successfully established a new standard in estimating these unobservables. By implicitly assuming slowly changing stacks of obligations, they implemented an overlapping rolling window VAR, which partly includes that networks evolve smoothly over time.
Although many papers are based on the concept of financial connectedness, its asymptotics lacks background theory. Exploiting framework of local stationarity, we can fill the gap of statistical inference and generalize the idea of financial connectedness. For this purpose, we propose a local linear Kernel estimator for VAR coefficients curves. We derive explicit expressions for the limiting bias and variance and prove a CLT. As the limiting distributions turn out to be complicated for practical applications, we propose a model-based bootstrap procedure that builds on our estimates. Simulations on random dynamic networks substantiate the performance and accuracy of this method. In an application on the U.S. Financial market, we apply our bootstrap method to set up confidence intervals. We also point out practical issues in the setting of financial connectedness and advise on how to handle bandwidth selection and estimation imprecision in periods of financial stress.