Hi,
My name is Matthias Schmidtblaicher. I work at QuantCo, where we develop algorithms for pricing, forecasting, claims management, and medical diagnosis.
I received my PhD in economics in 2020 from the EUI, where I was awarded the Vilfredo Pareto prize for best dissertation. In it, under the supervision of Peter Hansen, I examined inference about present value relationships and developed models that measure vaccine compliance.
Please follow this link to access my CV. It also contains my contact information. For information on working at QuantCo, please check our Notion site.
Infrequently, I add notes about results in statistics or economics that catch my attention here. Below, you can also find information about my somewhat dated academic activities.
2022/03/06: Benign overfit in ridgeless regression.
2021/10/24: Predicted probabilities are inherently unfair.
2020/08/11: Problems with tests for conditional expected accuracy.
Abstract: Increased vaccine hesitancy presents challenges to public health and undermines efforts to eradicate diseases such as measles, rubella, and polio. The decline is partly attributed to misconceptions that are shared on social media, such as the debunked association between vaccines and autism. Perhaps, more damaging to vaccine uptake are cases where trusted mainstream media run stories that exaggerate the risks associated with vaccines. It is important to understand the underlying causes of vaccine refusal, because these may be prevented, or countered, in a timely manner by educational campaigns. In this paper, we develop a dynamic model of vaccine compliance that can help pinpoint events that disrupted vaccine compliance. We apply the framework to Danish HPV vaccine data, which experienced a sharp decline in compliance following the broadcast of a controversial TV documentary, and we show that media coverage significantly predicts vaccine uptake.
Abstract: Immunization programs’ resilience to shocks is central to their success, but little empirical evidence documents resilience in action. We sought to characterize the decline of HPV vaccination in Denmark after negative media coverage and recovery during a national information campaign. We conducted a population-based retrospective cohort study of all girls born in Denmark from 1997 to 2006 (N = 328,779), aged 12–15. The outcome measure was HPV vaccine uptake (first dose), as reported to the Danish national health registry from 2009 to 2019, when HPV vaccine was freely available to girls in primary care clinics in Denmark. Events that created 4 natural time periods for study were HPV vaccine reaching the uptake of other vaccines in the national program (2009), some negative media coverage of HPV vaccination (2013), extensive negative media coverage (2015), and a national information campaign about the vaccine’s safety and effectiveness (2017–2019). In the period with some negative media coverage, HPV vaccine uptake fell to 83.6% (95% CI:78.0%–89.7%) of baseline uptake. In the period with extensive negative media coverage, uptake fell even further to 49.6% (95% CI:44.5%–55.2%) of baseline uptake. After the information campaign, HPV vaccine uptake recovered to its baseline level (109.2%, 95% CI:90.1%–132.4%) due in part to catch-up doses. Despite the recovery, an estimated 26,000 fewer girls initiated the vaccine than if uptake had not declined. The experience in Denmark offers one of the first opportunities to document how a nation grappled with negative media coverage of HPV vaccination and the steadying impact of action by national authorities.
Abstract: I analyse some commonly used bubble detection tests that have been derived from simplified present value relationships. I apply the tests to a flexible, estimated present value model and study their size and power properties. I find that, under the null hypothesis of price being equal to fundamental value, no time series based test allows for correct inference under conventional critical values. Comparing tests under the alternative hypothesis, tests for periodically explosive growth minimize maximum regret.