My Research


2017.02.10 - Article In Submission:

Sequential Changepoint Detection in Factor Models for High-Dimensional Time Series


ABSTRACT: We develop a sequential procedure to detect changepoints in a static approximate factor model. Speci cally, we define a ratio of eigenvalues of the covariance matrix of N response variables. We compute this ratio each period using a rolling window over time, and declare a changepoint when its value breaches an alarm threshold. To substantiate our procedure, we investigate the asymptotic behaviour (as N goes to infinity) of our ratio, and prove that, for specific eigenvalues, the ratio will spike upwards when a changepoint is encountered but not otherwise. We prove that the ratio can be consistently estimated. We propose a bootstrap to obtain alarm thresholds. We present simulation results and an empirical application based on FTSE 100 data.


2017.02.01 - LSE PhD Upgrade Presentation:

Sequential Changepoint Detection in Factor Models (Click for Slides)


SUMMARY: LSE research students generally undergo a review process during the second year of their PhD. During this review, their work is assessed by a panel of faculty members who then make the decision of whether or not to upgrade students from MPhil to PhD registration status. For my upgrade meeting, I submitted the first chapter of my thesis for evaluation and I presented slides summarising the work I conducted during 2016. I'm happy to report that I passed!


2016.11.30 - ESRC Writing Competition:

Is yesterday's model capable of explaining today's data? (Click for Article)

SUMMARY: I entered the ESRC Writing Competition for 2016. The brief was to write 800 words on "how your research is making sense of society, and why your research matters" in a manner that is interesting and accessible to people outside your field of expertise. I didn't win, but I've uploaded my entry here.


2016.02.24 - LSE Time Series Group Presentation:

Structural Instabilities in Factor Models for Time Series (Click for Slides)


SUMMARY: Within the literature relating to the estimation of high-dimensional factor models in time series, there exists a debate about the importance of structural instabilities. Stock and Watson (among others) argue that (under certain conditions) we need not worry too much about "small to moderate" structural instabilities. On the other hand, more recently, several papers have argued that "large" structural breaks can arise in real-world applications, and that it is imperative we take these into account when estimating factor models. This presentation highlights the key arguments on both sides of the debate, using specific papers as examples.