With increased availability of data in various fields, researchers often need to combine efficient empirical methods with innovative analytical modeling techniques to make data-driven decisions and gain managerial insights from the large-scale raw data. In light of this, my thesis combines empirical methods and analytical modeling to study several data-related problems in the fields of financial economics and healthcare management. The first two parts of the thesis focus on two topics in financial economics: the role of dynamic information in asset pricing and the link between index-based investment and intraday stock dynamics. The last two parts of the thesis study the ICU admission decisions and cardiac surgery scheduling using data from different hospital units.
The first part of the thesis focuses on the role of information in financial market. As a fundamental topic in asset pricing, information is known to play an important role in determining asset prices and market volatility. In most of the existing literature, the information environment, i.e., the amount of knowable information, is assumed to be fixed and independent of investor's choice. However, in a dynamic market, the level of available information can vary substantially due to changes in technology and regulations. On the other hand, rational news producers may respond to investors' demand for information. Such effects are commonly seen in the reality, but are less studied in the literature. To bridge this gap, we develop a model of investor information choices and asset prices where the availability of information about fundamentals is time-varying. A competitive research sector produces more information when more investors are willing to pay for that research. This feedback, from investor willingness to pay for information to more information production, generates two regimes in equilibrium, one having high prices and low volatility, the other the opposite. Information dynamics move the market between regimes, creating large price drops even with no change in fundamentals. In our calibration, the model suggests an important role for information dynamics in financial crises.
In the second part of this thesis, we investigate how the growth of index-based investing impacts the intraday stock dynamics using a large high-frequency dataset, which consists of 1-second level trade data for all S&P 500 constituents from 2004 to 2018 (500GB). We estimate intraday trading volume, volatility, correlation, and beta using estimators that are statistically efficient under market microstructure noise and observation asynchronicity. We find the intraday patterns indeed change substantially over time. For example, in the recent decade, the trading volume and correlation significantly increase at the end of trading session; the betas of different stocks start dispersed in the morning, but generally move towards one during the day. Besides, the daily dispersion in trading volume is high at the market open and low near the market close. These intraday patterns demonstrate the implication of the growth of index-based strategies and the active-open, passive-close intraday trading profile. We theoretically support our interpretation via a market impact model with time-varying liquidity provision from both single-stock and index-fund investors.
In the third part of the thesis, we study the intensive care units (ICUs) admission decisions in a large hospital system. In the case of ICUs, which provide the highest level of care for the most severe patients, it is known that admission rates of some patients decrease as occupancy increases. It is also known that, for at least some conditions, ICU admission is not just a function of patients’ illness, and that a significant proportion of the variation in ICU admission rates is due to hospital, not patient, factors. To understand such variation, we employ two years of data from patients admitted to 21 Kaiser Permanente Northern California ICUs from the ED. We quantify the variation in ICU admission from the ED under varying degrees of ICU and ED occupancy. We find that substantial heterogeneity in admission rates is present, and that it cannot be explained either by patient factors or occupancy levels alone. We use a structural model to understand the extent that intertemporal externalities could account for some of this variation. Using counterfactual simulations, we find that, if hospitals had more information regarding their behaviors, and if it were possible to alter hospital admission processes to incorporate such information, hospitals could reduce their ICU congestion in a safe way.
The last part of the thesis focuses on the impact of system workload on service time and quality in the context of cardiac surgeries. Using a detailed data set of more than 5,600 cardiac surgeries in a large hospital, we quantify how surgeon's daily workload level (e.g., number of surgeries) affects surgery duration and patient outcomes. To handle the endogeneity of surgeon's daily workload, we construct instrument variables using hospital operational factors, including the block schedule of surgeons. We find high daily workload of surgeons is associated with longer incision times and worse patient outcomes. Specifically, increased daily workload of surgeons leads to longer post-surgery length-of-stay in ICU and hospital, as well as higher likelihoods of reoperation and readmission for their patients. These results highlight the potential negative impact of surgeon's fatigue under long working hours. We then develop a surgery scheduling model that incorporates the effects of surgeon's daily workload levels.
| Identifer | oai:union.ndltd.org:columbia.edu/oai:academiccommons.columbia.edu:10.7916/d8-42a1-0h25 |
| Date | January 2021 |
| Creators | Shen, Yiwen |
| Source Sets | Columbia University |
| Language | English |
| Detected Language | English |
| Type | Theses |
Page generated in 0.0022 seconds