adult height is essentially determined by events in the first three years of life, and height can have long-lasting negative impacts on health status (Strauss and Thomas, 2008). Using a 1958 British birth cohort who were followed prospectively into their adult years. Case et al., (2005) explore the long-lasting associations between childhood health on adult health, exam performance, employment, and measures of socioeconomic status (SES). They show that people who experienced low birth weight or chronic disease when they were children had worse health, poor exam performance, and lower working statuses, even after controlling for parental background, such as education and income. Almond (2006) use 1918 Influenza Pandemic, which was unexpected and short, to identify the in-uterus conditions effects on late health. Using roughly one-third of those born in early 1919 whose mothers contracted influenza while pregnant as treatment group, and those born in early 1918, who had essentially zero prenatal exposure to the 1918 pandemic as control group, he find that children of infected mothers were more likely to be disabled and experienced lower, as well as less educational attainment. Currie and Walker (2011) investigate the introduction of E-ZPass (electronic toll collection) greatly decrease traffic congestion and vehicle emissions near highway toll plazas. Their results suggest that traffic congestion contributes significantly to adverse health conditions among infants.
In 2018, over 67 thousand Americans died from a drug overdose (Hedegaard et al., 2020). Although the substantial rise in drug-related mortalities over the past decade has been largely driven by illicit opioid use, the origins of this epidemic are rooted in opioid prescribing patterns that began in the 1990s and persisted throughout the 2000s. While pharmaceutical companies, physicians, and patients have received increased scrutiny in recent years over their respective roles in the ongoing crisis, the actions and behaviors of private health insurers have gone largely overlooked. These firms play an integral role in coordinating care between large segments of the U.S. population and health care providers; as a result, health insurers are uniquely positioned to monitor and observe patterns of opioid prescribing and use. Furthermore, despite recent efforts at the federal level to address the opioid epidemic, many public health advocates contend that the most impactful change will occur through communal ventures. Because of their influence in local health care markets, developing a better understanding of how private health insurers’ incentives interact with enrollee opioid use is of first order concern.
intensive labor input. For example, highly rated nursing homes need higher staffing levels as well as appropriate medical equipment to ensure that their residents receive high quality of care. Second, in the field of corporate finance, the pecking-order model of financing hierarchy (Myers and Majluf, 1984) predicts that firm managers tend to have a preference ranking over financing sources, which typically starts with internal funds, followed by debt, and then equity. The creation of this preference is mainly due to the information asymmetry between managers and investors. Within the context of this study, increased revenues due to higher star ratings may help relieving nursing homes’ financial constraints. For example, those 5-star nursing homes with scores that are just above the threshold may use extra revenues to enhance staffing levels and improve amenities so that they can maintain a competitive advantage, which lead to an increase in costs. Similarly, from the behavioral economics perspective, nursing homes with scores that are just above the thresholds may have a tendency toward loss aversion. These nursing homes could have stronger motivations to invest in quality improvement and to maintain a higher rating status because they are more afraid of lowering their star ratings. Lastly, from the perspective of business strategy (Collis and Montgomery, 1997), successful firms typically focus on how to establish their competitive advantages to ensure long-term asset growth and prosperity but not necessarily short-term profits. For example, higher-rated nursing homes may consider strategies such as attracting more residents and improving payer mix to better position these facilities against competition long term.
Duration analysis easily accommodates time varying covariates, and so is ideal for this context where an individual’s current BMI plays a significant role in determining their BMI (and their obesity status) in the next period. Within the realm of standard duration analysis lays three separate and distinct parameterization strategies. The first, nonparametric analysis, does not require any assumption about the distribution of failure times (i.e. the baseline hazard), and does not include covariates. This modeling technique essentially lets the data speak for itself, and models the time to occurrence based solely on the passage of time. The second modeling approach is to utilize, semiparametric analysis, which, does not require an assumption about the baseline hazard, but parameterizes the effect of covariates. Finally, the third modeling approach, parametric estimation, requires an assumption about the shape of the baseline hazard, and includes covariates in the analysis. In general,
There is a fast growing body of recent literature on interconnection of commodity markets or the role of financialization in markets co-movements. Saghaian (2010) presents empirical results using vector autoregression (VAR) and Granger causality supplemented by a directed graph theory modeling approach to identify the links and plausible contemporaneous causal structures between energy and commodities in the grain sector (wheat, soybean and corn). Although Saghaian (2010) finds strong correlation among oil and food prices with monthly data from 1996 to 2008, there is mixed evidence of a causal link from oil to the other three commodities. B¨ uy¨ uk¸sahin and Robe (2017) model dynamic correlations between equity market and commodity in grains and livestock sector, and find that world business cycle shocks have a substantial and long-lasting impact on the food markets co-movements with equity market, while changes in the intensity of financial speculation have a short-lived and not significant impact on cross-market return linkages using various specifications of structural vector autoregression (SVAR). Tang and Xiong (2012) find increasing correlation since 2004, but they model dynamics of correlations by rolling-window for all pairwise combinations of commodities one after another, which is inefficient as they do not explicitly take all information into account and not necessarily robust to the structural change in correlations. Adams and Gl¨ uck (2015) consider structural breaks in correlations but their sample only include eight commodities and also do not provide a joint estimation of dependence structure in futures returns. Most of these studies, however, only focus on specific commodities or just use low frequency data (monthly or weekly), and one may want to know if these empirical results are still robust if relative high frequency information of more futures markets is used.
Table 1.9 presents the estimates for / 7 , / 8 , / 9 , and / ; in equation (1.4). Based on the p-values for the t-test (row (f) in each panel), we are unable to reject the null hypothesis that there is no differential impact of G2 LBW (or IUGR) on male children born to the high SES group in all panels across all models. In contrast, we find some evidence that females born to the high SES groups are less affected by the intergenerational correlation in LBW (or IUGR). Out of 16 coefficients (row (c) in each panel), three coefficients in panel A and one coefficient in panel B and panel C are statistically significant at the 5% level (two of these five are significant at the 1% level). In panel A, except for SGA (5th percentile), females born to LBW (or IUGR) mothers in a county with a low unemployment rate are 2.26-2.50 percentage points less likely to be LBW (or IUGR). This difference represents a decrease of around 30% as compared to the base-line incidence of LBW (or IUGR) in females. The evidence from town-level income and parental education is weaker. However, we only find a significant differential impact on females born into towns with high average income in the model with 2SD < mean (in panel B) and those born in counties with high parental education in the model with FT LBW (in panel C). 31 We find no differential impact on males and females born in counties that experienced the most improvement in SES (in panel D). Thus, our results weakly support the findings in the literature: children born into favorable socioeconomic conditions suffer less as a result of poor maternal health (Currie and Moretti 2007; Bhalotra and Rawlings 2013). Moreover, our findings indicate
The regression discontinuity research design is compelling because it overcomes the omitted variable bias problem given a modest set of assumptions. Moreover, some of the assumptions are partially testable. Interpretation of the effects of Medicare eligibility on outcomes as causal requires two assumptions (Lee and Lemieux, 2009). First, the conditional expectation functions of the potential outcomes must be continuous with respect to age across the Medicare eligibil- ity threshold, which is analogous to saying that in the absence of treatment at age 65 outcomes would trend smoothly. There are two standard test for violation of the continuity assumption. First, there should be no changes in other variables across the threshold if this assumption holds. I test for discontinuities in educa- tion, race, labor market participation, health status, and Social Security receipt and present the results in Figure A3 and Table A2. These estimates are all statis- tically insignificant except for a decline in the fraction of the sample that is white, which is likely due to sampling error given that none of the other background characteristics change at the threshold. Second, I plot the density function of the running variable (spousal age) and test for a discontinuity at the age-65 threshold
Second, a malpractice incidence can severely damage a physician’s reputation, and as Dra- nove et al. (2012) have shown, such reputational damages are associated with economically significant costs. Direct monetary costs arise relatively seldom from a malpractice claim, as most physicians are fully insured against malpractice risks (Danzon 2000, Zeiler et al. 2007). For this reason, physicians should care more about the probability of being sued than awards. One goal of liability for medical malpractice is to align the interests of physicians and other healthcare providers with those of patients: by punishing healthcare professionals for providing too little care, liability is supposed to reduce adverse health outcomes. However, as we know since at least from Kessler and McClellan (1996), liability can also induce physicians to provide too much care. This is referred to as defensive medicine, which, in the economics literature, is defined as care that physicians order to avoid lawsuits but for which cost ex- ceeds expected benefits. The empirical evidence suggests that physicians practice defensive medicine by increasing treatment intensity for heart attack patients (Kessler and McClellan 1996, Avraham and Schanzenbach 2015) and ordering more imaging services (Baicker et al. 2007). The evidence regarding the rates of Cesarean sections, whose excessive use is of- ten attributed to liability pressure, is less conclusive: while Dubay et al. (1999) and Shurtz (2013) find that physicians perform more Cesarean sections following an increase in liability pressure, Currie and MacLeod (2008) and Amaral-Garcia et al. (2015) find the opposite.
With regard to the context of the thesis, two of the three studies present evidence from Benin, a country that has been studied in comparatively lesser detail than many of its African neighbours, but provides an interesting case study in development. A small Francophone West African country of around 11 million inhabitants, Benin is bordered by Togo, Burkina Faso, Niger and Nigeria. Benin has seen rapid development since the fall of the communist regime in the late 1980’s. Its economy is dominated by the service and agricultural sectors, with cotton representing the largest export. The country is, today, ruled by a comparatively stable democracy by African standards, and has seen stable growth levels in recent years. Yet, Benin is still faced with many challenges: poverty remains high and the gains from development have not been evenly shared geographically. Chapter 3 of this thesis studies an example of such uneven development in depth; whilst national primary school enrolment rates have seen almost unparalleled improvements since 1990, many regions have not shared in this progress and still lag behind. Furthermore, national completion rates have fallen somewhat, reflecting some of the difficulties that accompany such rapid progress. The country has also experienced rapid population growth since 1990, with the population having more than doubled in just 20 years. This has meant that even stable high growth rates of over 5% in recent years have not been sufficient to reduce poverty levels; the most recently available data suggests that over one third of Benin’s citizens still live in poverty – the poverty headcount ratio in 2011 was 36.2% (WDI, 2016). Furthermore, life expectancy remains low and child mortality high, with around 100 deaths per 1000 births, as of 2015.
This table lists summary statistics of the innovation factor, Fama-French three factors (from Kenneth French’s website), and Carhart momentum factor (constructed according to Carhart (1997). The five factors are denoted by RDCA (innovation factor), M KT (market factor), SM L (size factor), HM L (value factor), and M OM (momentum factor), respectively. In June of each year t, NYSE, Amex, and NASDAQ stocks are divided into three size groups using the breakpoints for the bottom 30%(Low), middle 40%(Medium), and top 30%(High) of the ranked values of market equity (price times shares outstanding from CRSP) in June for NYSE stocks. In each June, I also independently break NYSE, Amex, and NASDAQ stocks into three book-to-market groups based on the breakpoints for the bottom 30%, middle 40%, and top 30% of the ranked values of book-to-market ratio for NYSE stocks. Book-to-market ratio is calculated as the book value of equity in year t − 1 divided by the market value of equity in December of year t − 1. Aslo independently, in each June, I sort NYSE, Amex, and NASDAQ stocks into three innovation groups based on the breakpoints for the bottom 30%, middle 40%, and top 30% of the ranked value of RDCA, which is defined as R&D capital scale by total assets. 27 portfolios are formed from the intersections of the three size groups, three book-to-market groups, and three innovation groups. Monthly value-weighted returns on the 27 portfolios are calculated from July of year t to June of year t + 1, and the portfolios are rebalanced in June of each year. Thus, every month there are nine low R&D portfolios and nine high R&D portfolios. The innovation factor is defined as the difference, each month, between the simple average of the returns on the nine high R&D portfolios and the simple average of the returns on the nine low R&D portfolios. Panel A lists some basic statistics of the five factors. SKEW and KU RT refer to skewness and kurtosis, respectively. Panel B lists the correlation matrix of these factors. Panel C reports the coefficients of the regressions of innovation factor on traditional factors using two different asset pricing models: Fama-French three-factor model, and Carhart (1997) four-factor model. The sample period is from July 1963 to December 2009.
Writing a dissertation is a journey, a journey that requires inspiration, perseverance, and deep commitment. For me, this journey could not have been completed without guidance and help from a large group of people. I wish to express my deepest gratitude to all of those who helped to make this journey a wonderful experience in my life and helped me contribute to the scholarly literature in economics.
as it rarely changes, and when it does, laws tend to change as well. Our setting bypasses these standard challenges. The Dodd-Frank Act caused the SEC to transfer oversight of “mid-sized” in- vestment advisers ($25-100 million in assets under management) from the SEC to state regulators, except for advisers located in Wyoming and New York. 29 This decision was announced on July 21, 2011 and in effect by January 1, 2012. The impetus for the shift was exogenous to mid-size adviser behavior. The intention was to free up SEC resources so that it could increase oversight of hedge funds and private equity firms. The size threshold was chosen out of convenience, as it would reverse a component of the 1996 National Securities Market Investment Act (NSMIA). NS- MIA had assigned mid-size advisers to the SEC as part of a broader effort to unify state-securities regulations. Just over 38% of all existing SEC-registered firms were affected by this re-juridiction. Using a differences-in-differences design, we study how a shift from SEC to state-regulator oversight affects the probability of a customer complaint. Complaints are a good measure of ad- viser misbehavior and better than available alternatives. Complaints are publicly disclosed and observable for every professional in the industry. In addition, complaints are preference-adjusted; regardless of preferences, complaints reveal the perception of substandard adviser advice. The most obvious alternative, studying investment returns, is sparsely available, and would yield un- suitable comparisons across clients of different preferences. Studying regulatory sanctions may instead represent change in regulator behavior, not adviser behavior. We construct a survivorship- bias-free panel dataset at the representative-year level. We narrow the time period to the three years before and after the implementation of Dodd-Frank (2009-2014). We assemble this data using a variety of regulatory disclosures made by the SEC.
We define regimes according to the size and direction of the variance of the residuals in the reduced-form model, with the different regimes affecting the coefficients in distinct ways. Interest rates and credit spreads are in regime I when both shocks are above one standard deviation over the mean. Interest rates and credit spreads are in regime II when both shocks are below one standard deviation under the mean. Finally, interest rates and credit spreads are in regime III when both shocks are within one standard deviation of the mean. Regimes I and II both capture high volatility regions of the distribution, with regime I pertaining to the upper tail and regime II to the lower tail of the distribution, while regime III captures the lower volatility region of the distribution. There are therefore two possible subsets associated with these three regimes, denoted from here on as [regime I&III] and [regime II&III]. Adopting this regime segregation method allows the capturing of the asymmetric effects of shocks on the interest rate-credit spread relation. Additionally, the different standard deviations of interest rates and credit spreads provide favorable conditions for an estimation through heteroskedasticity since the variances of interest rate shocks and credit spread shocks are not proportional. If the identification through heteroskedasticity approach performs well, results from an estimation based on Regime I&III should be very similar to those of an estimation based on Regime II&III. The estimates from these two subsets are shown in Table 1.3.
In the first chapter “Titles For Me But Not For Thee: Transitional Gains Trap of Property Rights Extension in Colombia” I apply Tullock’s “transitional gains trap” to the formalization of property titles in Latin America to understand public choice problems in institutional reform. In a country where land is governed by formal and informal institutions, policies to extend property rights will not be supported by voters holding legal title because it will devalue their properties. To test that prediction, I use data from Colombia where a peace deal to end a 50-year conflict with FARC (Revolutionary Armed Forces of Colombia) rebels was reached in 2016 and submitted to a public referendum. The deal included formalization of property titles across the nation as well as an end to the conflict. Using municipal-level data on voting and property ownership and controlling for the conflict’s history, I find that potential losses to formal property holders pushed median voter preferences toward opposition. A 1 % increase in legally titled land increases the dissenting vote share by three percentage points. These results are relevant to institutional reforms anywhere with corrupted property rights.
I would like to thank my advisors, Dimitri Vayanos and Georgy Chabakauri for all the invaluable guidance and support I received. Their comments and suggestions improved this thesis a lot. Their personal example helped me to develop as a scholar. I also benefited from numerous discussions with permanent and visiting fac- ulty at the London School of Economics. I would like to thank Amil Dasgupta, Daniel Ferreira, Dong Lou, Peter Kondor, Albert (Pete) Kyle, Semyon Malamud, Igor Makarov, Ian Martin, Rohit Rahi, Anna Obizhaeva, Martin Oehmke, Marzena Rostek, Gyuri Venter, Kathy Yuan and Kostas Zachariadis. Peter Kondor and Igor Makarov provided vital help during my job market year. Marzena Rostek provided invaluable enthusiasm, support and feedback for the first chapter of the thesis.
In our data the average foreign peer share is at 10.5% of students in a major- university, with 60 students being foreigners on average and a standard deviation of the share of 10.8 percentage points, see table 4.3. The average EU student share is 4.1% and non-EU 6.3%, with standard deviations of 4 and 8.3 percentage points. When we introduce our set of fix effects they, obviously, reduced the amount of vari- ation available. The standard deviation for foreign peer shares goes from 10.8 to 2.3 percentage points when looking at levels and residuals from fix effects regressions. For EU student shares standard deviations go from 4 to 1.3 percentage points and for non-EU from 8.3 to 2. This reduction on variation is desirable because it implies that we only exploit within group variation. In figure 4.9, we show that international students are mostly represented (about 20% of the whole undergraduate population) in Business studies, Engineering and Economics; followed by Maths and Computer science and Technologies. For EU students the most selected subjects are Economics, Business, Engineering and Architecture, figure 4.10. The non-homogeneous distribu- tion of foreign students across majors suggest that there may be substantial differences in terms of incentives and therefore selection across majors and universities. 31 This makes cross major-university variation not really useful in terms of identification of the effect of foreign students on natives, as we discussed in section 4.3.
The role of history in determining current expectations is probably greater in college football than in any other major team sport. The best programs attract the best recruits and the best coaches, leading to the persistence of success: the best teams today are largely the same as the best teams from thirty years ago. In college basketball, individual players can have a much greater impact on outcomes than in football; and salary caps and new-player drafts in professional sports reduce the persistence of success in those outlets. Historical win percent should then be a strong indicator of current expectations, and the higher the current expectations, the more likely it should be that a coach is fired, holding current performance constant. But “current performance” appears to be measured only by performance in the most recent three seasons, with the impact decaying quickly over these three years. This reflects two facts. Coaches’ abilities may change, so that the most recent performance should be the most relevant. In addition, earlier
It would not have been possible to complete this long process without many people. As I write this acknowledgements, I realize how lucky I am to receive their support. To begin with, I wish to express my sincere gratitude to my adviser, Dan Bernhardt, who is very patient and always willing to help. I have no idea how he can manage his time to help so many students. He is absolutely an invaluable asset to all grad students here. I am so grateful that Marty Perry came to the department. Without his encouragement, I must have given up this journey. I also like to thank the other committee members, George Deltas and Steven Williams, for their support and comments. I am indebted to the economics department. Many thanks to Carol Banks and Tera Martin Roy for being always attentive to my questions and requests that are usually made at short notice.