Chapter 2



Risk Tolerance and its Effect on Majoring in ArtA Thesis Submitted in Partial Fulfillment of theRequirements of the Renée Crown University Honors Program atSyracuse UniversityAustin ChurchCandidate for Bachelor of Science in Economicsand Renée Crown University HonorsSpring 2020Honors Thesis in EconomicsThesis Advisor: _______________________ Dr. Perry Singleton, Associate ProfessorThesis Reader: _______________________ Dr. William Osborne III, Honors Adjunct ProfessorHonors Director: _______________________ Dr. Danielle Smith, Director ? (Austin Church 2020)AbstractThis paper looks into risk attitudes of college students and how those attitudes affect the likelihood of majoring in art in college. The theory of human capital investment provides the baseline for the analysis. The financial prospects of art graduates make a degree in the arts risky relative to other majors. Data from the National Longitudinal Survey of Youth 1997 was used to show that more risk tolerant individuals are more likely to major in art. The willingness to bend rules was used as a proxy for an individual's attitude towards risk; bending rules is a risky endeavor, and attitudes towards that risky activity serve to represent overall attitudes towards risk. To come to this conclusion, a logistic regression model was used to determine the effect of increased willingness to bend rules on the likelihood of majoring in art. Ultimately, the difference in the probability of majoring in art between a risk averse student and a risk tolerant student was shown to be roughly 3.5%. Risk-taking attitudes are also shown to have an effect on the likelihood of individuals majoring in the humanities. The effect of risk-taking attitudes on the likelihood of majoring in art also has implications for the structure of art classes and in art outreach efforts.Executive SummaryThis paper looks into the effect of students' risk-taking attitudes on the likelihood of studying art in college. There is a large body of literature on the economic decision to attend college or not; there has also been some research into what causes students to select their topic of study. The research reported on in this paper was conducted by first finding sufficient data on students; the National Longitudinal Survey of Youth 1997 provided this data. The survey interviewed a population of young people once a year, and their current college major and risk attitudes were among the topics asked about. The independent variable used was the students' likelihood to bend rules. This variable is not exactly equivalent to risk-taking attitudes; the likelihood to bend rules serves as a proxy for risk-taking attitudes because bending rules is a risky endeavor and attitudes towards this instance of risk is telling of one's overall attitude towards risk. The dependent variable was a binary variable that took the value of one if the student majored in art, and it took the value of zero if the student majored in something else. With the variables in place, the effect of risk-taking attitudes on studying art was measured through a logistic regression. Regression analysis is the use of a function to estimate causal relationships. After the regression was run, the coefficient on the independent variable was generated; with the coefficients in place, the resulting function can be seen as an estimated line of best fit that relates the independent to the dependent variable. A logistic regression is a form of regression that is non-linear, which provides a more accurate estimate if the effect in question is non-linear. The dependent variable is binary, either a student is majoring in art or they are not, so the equation gives the probability that a student studies art; changes in the independent variable change the probability of a student studying art. This effect is what the regression captured. This regression analysis supported the overall argument this paper is presenting. The argument is that art majors are some of the lowest earning majors on average (Carnevale et al. 12-18); this low level of earnings provides risk to the students who study art as their investment in their education is more likely to not pay off. For any given cost of college, majoring in art has a relatively higher likelihood of not recuperating that cost compared to higher earning majors. Students who are more willing to take on risk would be more likely to take on this risk and study art in college, holding interest and other relevant factors constant. The regression supports this. Other results that were uncovered through the research process were also listed. Among those results was the causal relationship between risk-taking attitudes and studying the humanities, and the inconclusive causal relationships between risk-taking attitudes and majoring in the social sciences, STEM, or professional fields. The standard errors here were too large to either confirm or deny the causal relationship. This result furthers economic understanding of what factors influence the subjects that are studied in college. Knowing this could help schools target students who are more likely to take to art and help art departments recruit students who would be more likely to excel in those departments. This result could also help existing art departments structure classes to maximize their effectiveness (especially at the elementary and high school levels).Table of ContentsAbstract……………………………………….……………….………….. iiiExecutive Summary………………………….……………….………….. ivChapter 1: Introduction……………………………………………… 1Chapter 2: Background……………………………………………… 2Human Capital Investment …………………………………... 3Literature Review……………………………………………... 5Chapter 3: Empirical Strategy……………………………………… 6Chapter 4: Data ……………………………………………… 8Chapter 5: Results ……………………………………………… 9Robustness……………………………………………... 10Other Results……………………………………………... 11Chapter 6: Conclusion ……………………………………………… 12Works Cited.……………………………………………………………… 14Appendices………………………………………………………………… 15 Chapter 1: IntroductionOne's educational goals are influenced in many ways. Personal interest, future employment prospects, and familial background are just some of the relevant factors. Knowing what factors influence these education decisions is important, since further education is expensive and has a dramatic effect on lifetime earnings. The specific factor that was investigated in this paper is risk aversion; this paper focused on art as a specific major of interest. The question this paper aims to answer is as follows: what is the effect of risk-taking attitudes on the likelihood of studying art in college? The theory of human capital investment was spelled out as was how major choice fits into this broader framework. An overview of existing literature on the subject was also conducted to provide a more complete grounding for the results. A logit discrete choice model was used to uncover the effect of risk-taking attitudes on the likelihood of majoring in art. Attitudes towards bending rules were used as a proxy for overall attitudes towards risk. All data was collected from the National Longitudinal Survey of Youth 97 which was conducted by the US Bureau of Labor Statistics. The independent variable was crafted from students’ responses to a survey question about their attitudes towards bending rules. The individuals surveyed self-reported their majors (if in college) and the major variables were created from that data. The effect uncovered is that more risk loving students are more likely to study art in college, and more risk averse students are less likely to study art in college. This effect was also found to stand up to robustness checks which further support the veracity of the relationship. It was also shown that risk-taking attitudes do have an effect on the likelihood of majoring in the humanities. The effect of risk-taking attitudes on majoring in the social sciences, STEM fields, and professionally oriented majors was shown to be inconclusive. This research also provides ideas for further research that would build upon this result. One idea for future research in this area is whether risk-taking attitudes affect the percentage of college grads who followed their art major with a career in art. Another idea is to see if risk attitudes affect the likelihood of pursuing an art career without formal education instead of pursuing a formal education in art.Chapter 2: BackgroundThe choice of whether to obtain higher education or not is a worthy topic of research in its own right; however, there is another choice for the people who do decide to pursue higher education: what to study while attending. Before getting into the theory behind this important choice, an overview of the United States higher education system will be useful in clarifying terms and will provide a sturdy ground to grow the theory out of. Students who attend college first obtain an undergraduate degree, the two broad categories of which are associate and bachelor’s degrees. Associate degrees are typically two-year programs, and bachelor’s degrees are usually four-year programs. A bachelor’s degree is required for most post-graduate degrees, which are masters and doctorate degrees. The focus of this paper is undergraduate studies, so postgraduate degrees will not be discussed further. An important distinction that needs to be fleshed out is the difference between a college and a university. These terms are commonly used interchangeably, but there is a distinction in the United States. While both are institutions that provide undergraduate degrees, universities often contain multiple colleges and they also provide more extensive postgraduate programs. In America, the distinction is not drastic, but there are differences (Wellman). The distinction is not vital to the results, so "college" will be used to refer to both colleges and universities in this paper unless specified otherwise. While at college, students are required to select a major; a major is a field that the student specializes in. Some schools may call their specialization requirements different things, concentration is a popular alternative name. Harvard University is an example of an institution that uses concentrations instead of majors; however, the overall demand for a specialization is the same. Different majors have different requirements that must be satisfied in addition to the overall school requirements. These requirements are typically a certain number of courses in the field, with some courses being mandatory. Majors can be broadly categorized into a few categories: humanities, social sciences, professional, creative, and STEM, which is an acronym for Science, Technology, Engineering, and Mathematics. Humanities majors are sometimes called the liberal arts; a few examples of majors that fall into this category are philosophy, history, English, and languages/linguistics. Social Sciences are majors that deal with social interactions/the study of society; a few examples of such majors are economics, political science, and sociology. Professional majors are majors that are designed to prepare the student for a specific career; a few examples of majors in this category are business and nursing. Creative majors are majors that focus on creative expression; a few examples of this type of major are the arts and architecture. STEM majors are majors that fall under the four fields in the acronym. There can be some overlap between these major groups, but these categories are helpful for a rough picture. This paper will be focusing on art majors (which falls under the creative category) and what factors influence students to choose those majors in college. Human Capital InvestmentBefore diving into the college major decision, an overview of the economic theory regarding attending college will be helpful. The predominant theory regarding the choice of going to college for further education treats this decision as a financial investment. A person continues their education if they believe the investment is a good one. A good investment in this instance is when the benefit of the additional education outweighs the monetary and opportunity costs of that education. The opportunity costs of education are the lost wages that a person forgoes when they enter an education program. Further education enhances an individual's position in the labor market; this superior position allows the worker to command higher wages. An important consequence of this is that the education must carry value; education that brings lower wage premiums is less likely to make the expected utility of the investment positive. Other important factors regarding the investment decision are the cost of borrowing money (interest rate), the opportunity cost of entering the labor market with their current level of education, and expected length of career. The intrinsic value of education to an individual also plays a role, but these effects will not be delved into in this paper (Borjas 229-276). This model is making the reasonable assumption that students are considering the costs and benefits of college when they make the college and major choice decisions.Not all of those factors have the same influence on the decision of which major to select once in college. One factor that is important to this decision is the value of the major. Certain majors have wildly variant average starting salaries. The majors that are associated with lower incomes still have students enrolled in them. So, while this factor is important, it cannot be the sole explanation. Personal interest is likely a large influence, but that is very hard to measure; it will not be the focus of this paper. Since expected return plays a role in the overall college decision, choosing a major to maximize this return is important. The majors that are lower earning on average are riskier than other, more lucrative, majors. This measure of risk is the focus of this paper. When looking at art majors, willingness to accept risk is what will be shown to have an impact on the likelihood of selecting this major. Literature ReviewA prior study by De Paola and Gioia looked into risk aversion and major choice and found that risk aversion does have an effect on what majors are selected. This paper looked at general fields and found that risk averse people are more likely to study any other field instead of the social sciences. This research, which sampled students at a midsized Italian University, shows there is a link between risk aversion and college major choice (de Paola and Gioia 1-19); this link will be built upon in this paper.In a study by Lisa Dickson, both race and gender were shown to affect college major choice. She found a gap of 16 percentage points in the probability of white women studying engineering and computer science compared to white males; white males are 16 percent more likely to major in engineering or computer science than white women, even after adding control variables. This study also found that women are more likely to major in the humanities and other majors relative to the social sciences (Dickson 1-17). Her paper shows that race and gender have an effect on the probability of studying different majors. Research by Wiswall and Zafar looked into the determinants of college major choice. They found that the expected earnings of each major affect the likelihood of selecting that major. Interestingly, they also found that the individuals' beliefs about their future earning in each major also has an effect. The individuals' beliefs about the likelihood of graduation in a given major is also a causal factor. Personal taste is also identified as a causal factor (Wiswall and Zafar 791-824). Their research helps provide a richer understanding of all of the factors that influence college major choice. It also provides empirical evidence that students do take expected earnings into effect when selecting a major. This lends support to the idea that students have some level of understanding of the risk involved in each major. Risk tolerance has also been shown to have an effect on college major choice. Belzil and Leonardi found that risk aversion is a deterrent to pursuing higher education. This means that more risk averse students are less likely to enter college (Belzil and Leonardi 35-70). This paper looks into whether risk aversion has a further effect on the likelihood of studying art in college. In a Georgetown study on the economic value of college majors, art was shown to have one of the lowest average starting salaries. The median salary of college graduates in art in the 21-24 year old age range was $28,000. This figure rose to $49,000 in the 25-49 year age range. Both of these figures rank near the bottom of all college majors. Additionally, the 25th percentile of art earnings in the later group is around $30,000; this is slightly below the median salary of high school graduates. The data show how majoring in art is a financially risky proposition. Art majors also have some of the lowest rates of graduate degree attainment, so additional wage premiums from those degrees are also less risk reducing than for most majors (Carnevale et al. 12-18).Chapter 3: Emprical StrategyA proxy will be used in order to uncover the effect of risk tolerance on the choice to major in art in college. This proxy is the subjects' attitude towards bending rules. This proxy is effective because bending the rules without outright disregarding them is a risky endeavor; while bending the rules is not a direct disobedience to the rules, it does show a willingness to walk the line on what is acceptable, and that behavior is risky. The more likely the student is to tolerate that risk, the more likely they are willing to tolerate other risks. The variable is a self-reporting of the subjects' own attitudes. The question that generated the variable is "Even if I knew how to get around the rules without breaking them, I would not do it." The survey respondents then selected their answer on a scale from 1-7, with 1 being disagree strongly and 7 being agree strongly. An answer of disagree strongly means that they would act in a way to get around the rules without breaking them if the opportunity arose. So, a person who answers 1 is most likely to bend rules, and a person who answered 7 is least likely to bend rules. This variable is the independent variable and will be denoted as X. The variable of interest is FineArt, a binary variable that takes a value of one if the student majored in the arts, and it takes the value zero if the student did not major in the arts. This variable was generated through the following process. The raw data had the students report their current major during multiple periods of each year from 1997 to 2017. A new variable was created that started out as all missing observations, and then each term of each year was replaced with the most recent major declared in the surveys. This was done in order to have one data point that represented the final major that each student reported. The major that is associated with each college going individual in the sample is the last major the individual reported to the interviewer. With the majors variable in place, the dummy variable for arts majors was left to be created. The way the majors were entered into the dataset changed in 2010, so there was a cross reference between the majors before the switch and after the switch; this cross reference was done by having a separate major variable for the two periods. The fine arts major was coded differently in the two time periods (before 2010 and after 2010). The fine arts variable was crafted by looking at the major variable and coding a value of one if the student had fine arts selected before and nothing after, or nothing in the before period and fine arts in the after period, or fine arts selected in both periods, or a non-fine arts major in the first period and fine arts in the later period. The regression equation will look like the following:Pry=1X=1/(1+e-(α+βX))This type of regression is a logit non-linear probability model. The coefficient measures how much a unit change in the independent variable changes the z-score of the standard logistic distribution function. The z-score here is equal to α+βX. For example, β = 0.5 means a unit change in X increases the z-value by 0.5. It will be shown later that this model stands up to robustness checks of two other probability models. Chapter 4: DataThe data used in this paper come from the National Longitudinal Survey of Youth 1997. This survey was ran by the US Bureau of Labor Statistics and it looks at students born between 1980 and 1984; there were yearly interviews with the subjects of the survey starting in 1997 and ending in 2017/2018. The survey started with 8,984 children in 1997, and there were 6,734 responses in the latest round of interviews. There were two subsamples that comprised the overall sample. The main subsample was a cross-sectional sample comprised of 6,748 children who were meant to serve as a representative sample of the United States population during round one of the surveys. The second subsample was a smaller sample of 2,236 children that oversampled black and Hispanic or Latino people that were born in the same period as the first subsample. The data covered a wide range of topics, from parental and environmental information to attitudes, health, and crime information. The interviews for the survey were conducted through a computer-assisted personal interview instrument that was administered by an interviewer with a laptop. The interviews were conducted in person if possible; in person interviews were preferred. An important result of this is that the answers to all the survey questions are self-reported; this self-reporting allows the people surveyed to decide their attitudes towards certain questions directly. In total, there were 4,955 students in the sample that had data relating to college major choice. The gender split in the data was pretty even, with 3,714 students being male (50.34%) and 3,664 students being female (49.66%). Chapter 5: ResultsWith the variables prepared and the regression model specified, a preliminary investigation into the effect of a positive attitude towards bending rules on majoring in art was performed. A two-way line graph was generated between the density of fine arts majors and the willingness to bend rules. This graph (Figure 1) shows a distinct trend: as the willingness to bend rules decreases, so does the likelihood of majoring in art. This preliminary result provided the impetus for running the regression proper. After running the logit regression, the results were significant at the 95% confidence level. The results of this regression and two other robustness regressions are contained in Table 1. The coefficient of interest (β) is -0.1372 and the constant term (α) is -2.617. So, the likelihood of majoring in arts starts with a z-value of -2.7542 and Pry=1X=1=1/(1+e2.7542)is the model evaluated at X=1 with the estimated coefficients. This gives a probability of such a student majoring in art of 0.0598 or 5.98%. As an individual's likelihood of bending rules decreases, the likelihood of that person majoring in art also decreases. The likelihood bottoms out at 2.71% for students with the highest disdain for bending rules. The overall change in the likelihood of studying art predicted by the logit model is 3.27%. These results show that the difference in the probability of studying art between a student who has a disdain for bending rules and a student who has an affinity towards bending rules is approximately 3.3%. Since the attitude towards bending rules is being viewed as a proxy for an attitude towards risk, the gap should be similar between a risk loving and a risk averse person. The higher likelihood of risk loving students majoring in art makes sense because majoring in art is financially risky relative to other majors.These results signify a link between risk-taking attitudes and the likelihood of majoring in art in college. There is an important caveat to this data that bears explanation. Many artists never study art in college; they start producing and selling their art without formal college education. These artists are not included in this paper. While it is likely that these results would hold for those artists also, further research is needed to confirm this point.RobustnessThe presence of this effect is also supported by the linear probability model (LPM). The linear probability model predicts the change in probability that the binary dependent variable takes the positive value through a linear means; the predicted probability is the y value once the linear function is evaluated at a given value of the independent variable. The LPM has a coefficient of -0.0053 which means that a unit change in the bend rules variable means a 0.53% change in the likelihood of studying art in college. The t-value of the coefficient is -3.40 which is well above the -1.96 threshold for the 95% confidence level; this means that this effect is statistically significant. The model predicts a 3.71% total change. This is similar to the logit model's prediction, and this supports the veracity of the models. This effect is also supported by the probit model. This model differs from the logit model in that it uses the standard normal distribution function. The expression α+βX gives the z-score associated with the independent variable. When X=1, β = -0.0614 and α = -1.498. The initial z-score is -1.559. This z-score means that the probability of such a student studying art is 6.06%. When X=7, the z-score is -1.928. This z-score corresponds to the probability of such a student studying art of 2.68%. The total change in the probability of studying art is estimated to be 3.38%. This total change is very similar to the total change for the other two regression models and provides further proof of the robustness of the results. Other ResultsAt first, the dependent variable of interest was a group of creative majors. The two majors that were selected in this group were arts and architecture. The group was restricted to only arts majors since the financial risk associated with majoring in architecture is significantly lower than majoring in art; therefore, architecture majors do not fit in the same group as art majors in the model. Before the restriction, gender was also used as an independent variable; gender was statistically significant. Men were more likely to major in one of these creative majors than females. Once architecture was removed and art was looked at alone, gender's effect was no longer statistically significant. Letting the gender variable be denoted as G, this regression equation was the following: Pry=1X=1/(1+e-(α+βX+?G))The standard error for gender's effect on majoring in art (Table 2) was very large, so gender still may have a role in determining the likelihood of majoring in art, but further analysis is needed to confirm this one way or the other. Another result worth mentioning is the effect of risk-taking attitudes on majoring in the humanities. This was done by taking a group of majors that fall under the umbrella of humanities and running the same analysis as was performed for art. The group of humanities majors was comprised of the following majors: foreign languages, education, English, history, and philosophy/theology. The result was statistically significant at the 95% confidence level (Table 3); the overall shift in the likelihood of majoring in the humanities due to risk-taking attitudes was roughly 4.25%. The effect of risk-taking attitudes on studying the humanities has a different direction than the effect on majoring in art; students who are more risk tolerant are less likely to major in the humanities. The following major groups were also explored: social sciences, STEM, and professional. The social sciences group contained the following majors: area/ethnic studies, psychology, and majors coded as ‘social sciences’ within the data. The STEM majors were the following: biology, engineering, mathematics, and physical sciences. Lastly, the professional group was comprised of the following majors: agriculture, business, communications, computer science, and health related majors. There was no statistically significant effect of risk-taking attitudes on selecting a major in any one of these categories. In all three cases, the estimated coefficients were small with a large standard error (Tables 4-6). These results do not rule out the possibility that the effect is present, but the size of the standard errors do not allow determination either way. Chapter 6: ConclusionAfter looking into the National Longitudinal Survey of Youth 97, a link between the willingness to bend rules and the likelihood of majoring in art in college was uncovered. The willingness to bend rules can be seen as a proxy for risk-taking attitudes. This use of a proxy means that this link can be seen as a relationship between risk-taking attitudes and the likelihood of majoring in art in college. More risk loving individuals are more likely to major in art than risk averse individuals, holding everything else constant. This finding does not analyze everything about the selection of major choice and even the choice to major in art; further research on this topic would be worthwhile. One potential further topic of research would be if familial income has an effect on how sensitive students are to the risk that is associated with majoring in art; it is possible that higher familial incomes may weaken the causal power of risk-taking attitudes on the likelihood of majoring in art. Another potential further topic of research is whether risk-taking attitudes affect whether art majors pursue a career in art. Many students' careers post-graduation are not directly related to their major, so seeing the proportion of art graduates that pursue an art career and whether risk tolerance has an effect on the decision to start a career in art would be valuable. A third potential research topic involves the proportion of artists who pursue art in college versus the people who forgo formal education to begin creating and selling art directly. Art is not a field that has college as a prerequisite, so seeing whether risk tolerance affects the decision to pursue art without a formal education or attain formal education would also be fruitful. Another topic that could be looked into is if the different entrance requirements to certain majors influence the likelihood of those majors being chosen. Some majors have higher entrance requirements than art does. Seeing if these entrance requirements play a causal role in students’ major choice would be valuable. This research also has the potential to manifest itself in future policy decisions. Knowing what factors go into making major choice decisions would help high schools give better education and advice in preparation of that choice. Additionally, this information could help educators better structure art courses to be more informative and engaging for students, especially the ones who are more likely to study art in college. With this in place, schools could have a better idea of what a good structure would be. This could lower costs of running such classes. Finally, this information can help college art departments in their outreach programs; knowing more about prior art majors can allow them to target potential art majors more efficiently and more cost effectively. Works CitedBelzil, Christian, and Marco Leonardi. “Risk Aversion and Schooling Decisions.” Annals of Economics and Statistics, 2013, pp. 35–70. JSTOR.Borjas, George J. “Human Capital.” Labor Economics, by George J. Borjas, McGraw-Hill Education, 2016, pp. 229–276.Carnevale, Anthony P, et al. “The Economic Value of College Majors.” Georgetown University Center on Education and the Workforce in the McCourt School of Public Policy, 2015, pp. 12–18., cew.georgetown.edu/cew-reports/valueofcollegemajors.De Paola, Maria, and Francesca Gioia. “Risk Aversion and Major Choice: Evidence From Italian Students.” pp. 1–19. Research Gate, 2011, profile/Maria_De_Paola2/publication/254399124_RISK_AVERSION_AND_MAJOR_CHOICE_EVIDENCE_FROM_ITALIAN_STUDENTS/links/02e7e53ce5ad6e80f2000000/RISK-AVERSION-AND-MAJOR-CHOICE-EVIDENCE-FROM-ITALIAN-STUDENTS.pdf.Dickson, Lisa. “Race and Gender Differences in College Major Choice.” The Annals of the American Academy of Political and Social Science, 2010, pp. 1–17.Wellman, Mitchell. “What's the Difference between a 'College' and a 'University'?” USA Today, 2017, story/college/2017/03/01/whats-the-difference- between-a-college-and-a-university/37428407/.Wiswall, Matthew, and Basit Zafar. “Determinants of College Major Choice: Identification Using an Information Experiment.” The Review of Economic Studies, 2015, pp. 791–824. JSTOR.Appendices ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download