Bookmark and Share
  • Text +
  • Text -


Research Presentation Videos

Watch Jacob Fiksel '15 discuss his research project.

Analyzing Test Scores and Teacher Performance in Pakistani Schools

Rebecca Howland (2014); Mentor(s): Tahir Andrabi

Abstract: My work consisted of two distinct elements. The first was organizing raw data on student and teacher test scores from Pakistani schools. The second was done in collaboration with Developments in Literacy (DIL), a Pakistani education non-profit. The goal of this component was to analyze a set of baseline data for a new teaching intervention that DIL is in the process of implementing. The study contains two groups: a control group and a group that will receive the intervention. DIL will continue to test the teachers and compare the results of these tests in the two groups. This involved performing an item analysis (determining the quality of question on a multiple-choice test as well as how teachers performed on each question). In order to understand the areas in which teachers need to improve, I identified low correct response rates and matched the percentage with the specific type of question. An example of a question classification going from most general to most specific was English: Reading: Grammar: function of pronouns. I performed these analyses using Excel for data management and STATA for data analysis.
Funding Provided by: Aubrey H. and Eileen J. Seed Student Research Fund

Evaluating the effect of uncertainty reductions in the selection of wind projects

Kristen McCormack (2015); Additional Collaborator(s): Katie Dickinson (National Center for Atmospheric Research); Matthew Hendrickson (3TIER); Mentor(s): Bowman Cutter

Abstract: This project focuses on understanding the role of uncertainty in project selection decisions that are made during the development of a wind farm. In past decades, wind projects have underperformed pre-construction energy estimates by an average of about 10%. This has put a strain on various players in the wind industry, and has damaged the reputation of wind energy. Research to address this gap in performance has primarily considered ways to reduce the bias in pre-construction estimates. However, even with unbiased energy modeling, uncertainty may lead to significant overprediction as well as a tendency to pick suboptimal projects. By simulating selection decisions made by utilities during power purchase negotiations, the effect of pool size and uncertainty on suboptimal selection and overprediction were analyzed. The results of the model indicate that, with no change to model bias, reducing model uncertainty may result in a reduction in the overprediction of wind energy and may enable more optimal selection of projects. This model may be used by utilities and developers to adjust their own selection processes and to determine the value of investing in more precise modeling techniques.
Funding Provided by: Faucett Catalyst Fund

The Green Premium in New York City: Energy Efficiency and Certification in the New York Commercial Real Estate Market

Angela Gunn (2014); Mentor(s): W. Bowman Cutter

Abstract: LEED and Energy Star certification recognize environmentally friendly buildings, providing a free market incentive to “go green.” Such buildings can be advertised as both socially responsible and less expensive to operate, and existing literature demonstrates that LEED and Energy Star buildings in the United States have higher sale prices and rental rates than similar, non-certified buildings. It is believed that owners of highly efficient, uncertified buildings may not reap the same benefits because energy efficiency is difficult to signal to buyers and renters. This project explores two issues. First, I consider whether the “green premium”– the higher price renters and buyers pay for an efficient building– is primarily determined by third-party certification or actual efficiency. Second, I compare the green premium for commercial buildings in the New York City market to estimates of the national green premium. I use building-level information from the national LEED and Energy Star directories, New York City’s 2011 benchmarking data, and the CoStar real estate database to compare rent and sale prices of New York buildings with certification to those that are highly efficient but not certified, and to similar buildings that are less efficient. I seek to better understand the role of third-party certification groups in motivating commercial building owners to focus on energy efficiency, and to determine whether that efficiency is valued differently in New York City.
Funding Provided by: Faucett Catalyst Fund

A Bayesian DSGE Model of Financial Business Cycles

Tianyou Gu (2015); Mentor(s): Michael Kuehlwein

Abstract: The paper develops and estimates a dynamic stochastic general equilibrium (DSGE) model where credit market frictions and stock investment exacerbate the business cycles. In this framework, adverse shocks to the housing price or repayment rate force financial intermediaries to deleverage. As a consequence, credit crunch and counter-cyclical external finance premium drive both borrowers away, thus dragging down spending, investment, output and real asset prices. The novelty of the model is that it goes further to capture the fact that financial shocks are further amplified since both consumption and credit supply would drop as shareholders become less wealthy following the plunge of stock prices caused by deteriorating company fundamentals. In this way, any shock in productivity sector, investment sector, or bank intermediation sector is contagious, just as in the real world in which technology shock, investment speculation, or credit frictions shock could lead to boom or depression. The model is estimated with Bayesian techniques using data for the U.S. over the last 30 years. And the empirical analysis delivers the following results. First, decreased bank capital has a substantial impact on the real economy. Second, stock market plays an indispensable accelerator role in driving the bubbles and bursts. Third, the estimated model implies that nominal contract channel and housing collateral channel also contribute a lot to the feedback loop.
Funding Provided by: Pomona Alumni SURP Fund; Pomona College Department of Economics

Hours Misreporting in the Current Population Survey

Jacob Fiksel (2015); Mentor(s): Fernando Lozano; Michael Steinberger

Abstract: The Current Population Survey (CPS) is the most commonly used survey for investigating wages in the US economy. To estimate hourly wages, researchers must divide weekly earnings by hours worked—but the CPS asks for two different measures of hours per week: the hours the respondent usually works, and the hours worked last week. We document and explore the discordance in how respondents report these two measures of weekly hours over time. If misreporting of usual hours has changed, then this will automatically lead to spurious changes in estimated hourly wage measures. We take advantage of the longitudinal CPS format to suggest what we believe to be more accurate measure of hours worked: the mode response of hours worked last week for each respondent. Using a DiNardo, Fortin, and Lemieux (1996) decomposition technique, we control for changes in the misreporting of hours throughout the distribution of hours worked since 1996. We show that misreporting is an issue for wage measurements in a given year, but does not appear to be a contributing factor for the documented changes in wage inequality during the period.
Funding Provided by: Pomona College SURP

Heterogeneity in Conditional Cash Transfer Effects: Evidence from a Randomized Experiment in Malawi

Johnny Huynh (2014); Mentor(s): Michael Steinberger

Abstract: A growing body of literature documents the effectiveness of conditional cash transfers (CCTs) on increasing investments in children’s human capital. While cash transfers can increase children’s average rates of school attendance and test scores, how are these gains are distributed across treated students? How much do parents take into account latent capabilities of their children when deciding how to respond to CCT programs? We explore the distributional effects of a randomized cash transfer experiment for adolescent schoolgirls in Malawi first documented in Baird, McIntosh and Ozler (2011). We present a theoretical model showing that the positive mean effects of the CCT program should be concentrated among a specific subset of potential participants. To test our theory, we use a set of observable test scores to identify latent academic ability and estimate quantile treatment effects for the CCT program. Our evidence shows the impact of the CCT program on test scores is only significant on the third quartile of latent ability. As predicted, unconditional transfers do not show a similar distributional heterogeneity. We also explore the effect of the CCT on school enrollment and attendance, with inconclusive results. Our findings have strong implications for public policy, suggesting the primary beneficiaries of CCTs are higher-ability children, while having little effect on lower ability, often poorer, children.
Funding Provided by: Faucett Catalyst Fund

Does aggregate loan growth predict future performance of the financial sector?

Nathan Shekita (2015); Mentor(s): Michelle Zemel

Abstract: In this project, we examine bank lending and the effect this has on the overall market. We attempt to create a measure that can be used to predict an economic crisis with sufficient accuracy. Data was obtained from the Wharton Research Data Service, as well as the Federal Reserve Economic Data (FRED). Using logistic regression, we find that statistically, our measure does a better job at predicting crises than those defined from other literature. The Hodrick-Prescott filter was used to detrend and deseasonalize the data. We use both backward and forward looking techniques to assure the legitimacy of our findings. We also attempt to explain the reasoning as to why our measure works, as well as its possible applications.
Funding Provided by: Faucett Catalyst Fund

Research at Pomona