The Everyday Economic Effects of the 2025 Government Shutdown on American Consumers
Advisor: Rob Biemesderfer
There have been 4 major government shutdowns in the past 10 years. As the government shutdowns become more frequent, it is important to analyze the economic impacts they have on American consumers. Though studies have
observed a general decline in U.S. net growth due to shutdowns, there has been a lack of specific attention towards what individual households experience. I plan to use statistical analysis to determine how everyday fiscaln factors, including income and spending patterns, stock market volatility, U.S. GDP growth, consumer confidence, change in unemployment claims, and news or sentiment data, changed as a result of the 2025 government shutdown in particular. This spring, I will be interning at Datanyx, a data analysis company, and familiarizing myself with their tools will help me apply statistical models to my own research. By bringing awareness to unknown effects, I can assist lawmakers in making decisions about future shutdowns on behalf of their constituents.
Project Blog
Post 1: Introduction
Growing up with a mom who worked for the federal government, my family has always found the news of a government shutdown difficult to take. I can’t imagine what it must be like for airport workers who are forced to show up every day without pay. What feels most frustrating about this recurring issue is that it is so preventable. More can easily be done to either reduce the number of government shutdowns or minimize their likely damaging and effects on society.
Social justice has always been a passion of mine. After taking AP U.S. Government & Politics, AP Macroeconomics/Microeconomics, and summer enrichment courses that further added to my knowledge of how the government and economy function, I also gained real-world experience through internships with multiple congressmen and as a campaign volunteer. This background has deepened the value I place on policy in bettering the lives of constituents.
I have previously researched possible solutions to the increasing frequency and length of government shutdowns. Congress has failed to pass legislation with the intention of fulfilling any of these propositions. Additionally, negative incentives to Congress stand little chance of gaining approval. There needs to be a political will among lawmakers to implement one of the many options that can fix this problem. By analyzing data from the 2025 government shutdown and the economic impacts it has had on American consumers, my research can demonstrate the necessity and urgency of action against government shutdowns. Interning at Tekzima, a data analysis company, will allow me to gain beneficial skills to apply as I analyze various variables, including income and spending patterns, stock market volatility, U.S. GDP growth, consumer confidence, changes in unemployment claims, and news or sentiment data, through statistical analysis.
Week 2: Blog Week 2 - Learning Data Analysis and Collecting Data
This week, I began my internship. The coding work at my company mainly relies on Python. Since I mainly have experience with modified versions of C, I underwent some training in order to better understand their preferred form of code and how to implement it within their platform. Because their platform is a data analysis model, this included learning how to code for inputs in data sets and how to use them to perform analysis. I am also learning about the convenient ways their tool can be utilized and its many purposes. I was even able to discuss the possibility of using their tool to analyze the data I am collecting for my Senior Project to provide additional evidence, though my initial intentions were for my conclusions to be merely backed up by forms of statistical analysis, including a factor analysis with promax rotation, linear regressions on software called Gretl, and further calculations on Excel.
At the same time, I have begun my independent work by collecting data for my Senior Project. Through public resources, I have gained many datasets beneficial for determining possible economic patterns affected by the 2025 government shutdown. Sources such as the U.S. Bureau of Economic Analysis, the U.S. Bureau of Labor Statistics, and the U.S. Census Bureau have proven useful for my analysis of quantitative factors such as U.S. GDP growth, income and spending patterns, and changes in unemployment claims. Though stock market volatility, consumer confidence, and news or sentiment data are factors that are generally considered more qualitative, I have found many indexes and economic databases online that measure these factors quantitatively, allowing me to analyze them in a similar process.
Week 3: Blog Week 3- Expanding My Literature Review and Revising My Methodology
As I continue to expand my literature review, I have discovered many recent publications with more detail on the effects of the 2025 government shutdown. When I first began my research project, I anticipated this would occur since it was a very new event at the time and researchers would likely need more time to make their conclusions. However, one aspect the emerging literature I did not expect was analysis the two partial government shutdowns that have already occurred in 2026. Even though the main reason I chose to analyze the 2025 government shutdown in is that it was the most recent shutdown at the time, I will continue to focus on it because it is still the longest shutdown in U.S. history. Additionally, it will be much easier to analyze the effects of a government shutdown on consumers during a full governmental closure. It will be a much better representation of all the impacts can come with a government shutdown.
In the meantime, I have also been working on revising my methodology for my next deliverable. I am unfamiliar with some of the analysis approaches I plan on using because they have been recommended to me by advisors after hearing about my topic. Since I have no personal experience with these methods, I need to do further research on them in order to explain their use in my process and to know how to implement them when the time comes. Thus, I have been learning about a factor analysis with promax rotation my teacher recommended which might be useful to identify which factors are most closely related to one another. I will continue searching for different sophisticated statistical analysis methods to see which ones will work best for my research.
Week 4: Blog Week 4- Continuing My Search For Data
This week, I have continued to search for my data regarding different affected economic factors of the 2025 government shutdown. I initially sought to include as many variables as possible, including income and spending patterns, stock market volatility, U.S. GDP growth, consumer confidence, changes in unemployment claims, and news or sentiment data. However, some sources I planned on using, such as the U.S. Census Bureau, have yet to be updated for the 2025 fiscal year. Additionally, there are some variables that I have found data on, but they are in the form of graphs or listed within paragraphs. Since these forms will be difficult to utilize, especially through statistical analysis and data analysis methods, I am choosing to only accept CSV tables and possibly XLSX files for my research. As a result, I may not be able to include all of the factors I initially intended to. I have decided that I will continue my search for data next week as well, before I begin my analysis stage, but if I do not come across the proper format for variables, as I have already experienced with income patterns, I may choose to leave these factors out so I can move on and begin my analysis.
Week 5: Blog Week 5- Experimenting and Learning New Methods
I began performing statistical analysis on my data this week. It was harder than I had anticipated because even before the analysis could begin, much formatting had to be done in order for the data to be digestible by the platforms I was using. I thought since I had downloaded my data in CSV form, it would naturally be converted into an applicable form, but I found myself making multiple tweaks to my tables before my platforms would accept them, which will make this analysis phase of my project slightly more lengthy and tedious than expected.
As for my analysis, I tried out two different methods this week. The first was using an economic software called Gretl to perform linear regressions. It is intended primarily for users with a programming background, so it was slightly difficult for me because I have taken simple coding and computer classes in middle school, but I lack more intricate experience. I began experimenting with the different options and playing around with the data so I could understand how to use it. Some aspects of it were intuitive, but I require further research on other aspects before I can fully begin my regressions and note discontinuity.
My second method was using Datanyx, a data analysis tool developed by Tekzima, the company I am interning for. I initially thought I wouldn’t be able to use this approach because the tool is mainly designed for much larger pools of data. However, after conversing with colleagues about my project, I saw that it would still be a worthwhile tool to use. It has been a busy week for the company, but I was still able to spend a bit of time learning about how to use Datanyx from them and experiment with different models. I hope to meet with them more next week to see in more detail how I can implement it into my research.
Week 6: Continuing Analysis and Beginning to Make Interpretations
This week, I continued to perform linear regressions with my data. Similar to my hardships last week, I continued to face formatting issues that would not allow my calculations to run smoothly, but after making multiple further changes to my data, I got the hang of it. Though the software I am using can be complex, from the tutorials I’ve watched, I’ve been able to implement a regression simple enough for beginners. I am analyzing my dataset as a time series because I am observing my variables, such as consumer spending and unemployment patterns, on a monthly basis and treating October and November of 2025 as a shock point, since that is the time immediately following the initiation of the 2025 government shutdown. Thus, I am not just using the single event of the shutdown as a regressor, but rather specifically using the period of time following the shutdown as a regressor against my multitude of dependent variables.
So far, I have found extremely low p-values, meaning that there is a statistically significant discontinuity in the trend of my variables following the beginning of the shutdown. This is not enough information to establish a causal relationship between my dependent variables, as there are hundreds of confounding factors I can think of, such as seasonal jobs or expenses, that could influence my regression. However, I do think that my calculations still illustrate a strong correlation and can be used to predict the outcomes of future shutdowns. Gretl also has a feature that has allowed to create visual plots of my data with a line showcasing when the shutdown started. I hope to incorporate these graphs into my final paper and presentation so that readers and audience members can easily visualize the conclusions that I make as I explain them.
Week 7: Blog- Beginning My Final Product and Complexifying My Analysis
I began working on my first draft of my final product this week. While I only made minor changes to my introduction and methodology sections so far, I was able to add a bit more information to my literature review through the additional sources I’ve read through in the past few weeks. One trend I noticed in recent sources published about the effects of the 2025 shutdown is that they are focusing on even more avenues through which the shutdown has impacted the economy, such as its impacts on public health, travel, business, and benefits. Reading these texts made me realize the massive scope of the 2025 shutdown and how the variables I am using to measure how households have been economically impacted are only a fraction of how many exist, which I plan to mention in the limitations section of my conclusion.
I also continued to work on the analysis of my data. To establish a stronger relationship between the shutdown and the variables I am measuring, I realized I need a stronger independent variable. Currently, I have only been using time, though I am aware of many potential confounding variables that may cause my dependent variables to exhibit a discontinuity during the period from October 1 to November 12 of 2025. I thought I would just have to accept this and mention this within the limitations section of my research paper, but I now believe I can find additional data on direct impacts of the shutdown, such as weekly federal employee furloughs, delayed or withheld federal payroll amounts, the number of agencies with lapsed appropriations, and the freezing of contractor payments. By regressing my dependent variables, which are more of secondary impacts of the shutdown against these immediate effects, I can establish a better relationship between the 2025 government shutdown and the major ways American households have been economically affected. It will take time for me to collect more data, but I believe once that is done, I can speedily conduct my regression analysis and complete the analysis and conclusion portions of my final product draft.
Week 8: Blog- Furthering My Analysis and Making Conclusions
As I continued to run regressions and calculations this week, I noticed that I was often unable to obtain statistically significant results. After doing some testing to figure out the problem, I realized my degrees of freedom were too low because I was using such a small sample size. Thus, I made the tedious and long decision to expand my data set. Instead of merely starting one year before the beginning of the shutdown, I started my range from January 2022. Although it would have been beneficial to expand my range even further, I believe the outlier of the 2020 pandemic, which prolonged into 2021, would skew the trend of my data. I included this as a limitation in the conclusion of my final product draft.
Additionally, since time has passed since I first began collecting data, the data for the entire first quarter of 2026 has been released, so I was able to incorporate that into my dataset as well, although I had originally only had January of 2026. By making these changes, I was able to produce results that were more meaningful and complex. I found that many variables, based on whether they affect people in the short term or long term, would similarly usually only have a significant p-value for D, a variable showcasing the instance of the shutdown, or D_t, representing the period following the shutdown. While personal income and unemployment claims were instantly changed, variables like consumer confidence and news or sentiment data took longer to gain traction as the effects of the shutdown spread. I was able to talk in detail about this observation in the analysis section of my final product draft.
Week 9: Improving My Final Product
This week was mainly spent improving the first draft of my final product that I submitted last week. Rather than focusing on the limitations of each regression based on the variables I used in each one, like last week, I focused more on identifying limitations of the approach as a whole this time, such as the April 2025 Liberation Day tariff increase, which could have influenced the pre-shutdown trend that I was comparing the post-shutdown trend to. Additionally, I added all my sources, from which I obtained my datasets, to my bibliography, as they are also important references.
I also realized I didn’t properly explain the variables I created to support the regression analysis, including t, D, D_t, and t_centered. To fix this, I created a section of my paper to define each variable and made a chart explaining what the characteristics of these variables contribute to my final interpretation. For coefficients, this usually meant explaining how they show a positive correlation when they are positive and significant, and for p-values, this meant detailing how they are considered significant when less than 0.05. I also explained the process of how I created my graphs and what they signified, as I realized that could be confusing for someone not as familiar with regression analysis.
To provide more context for my analysis, I also provided the basic universal simple regression equation and showed how adding other variables, through my multivariate regressions, affects that equation. I also added how the coefficient and p-value of shutdown-intensity variables should be interpreted in the table with the other independent variables.
Week 10: Blog- Improving My Final Product
This week’s work consisted of lots of tweaking to the first draft of my final product from last week. I realized I hadn’t updated certain sections of my methodology to be more detailed, so I wrote a step-by-step explanation of my process, since my previous instructions were quite vague. Since I went through a lot of changes from my original plans, I was able to talk about what modifications I made and why as well. I also edited my literature review to include more context on the 2025 government shutdown, in particular, and more information about how my approach is distinct from the existing literature.
Additionally, I dedicated a lot of time to improving my analysis section. I realized I had previously made a lot of statements that established a cause-and-effect relationship between certain variables. However, since my regressions do not include an experiment, I cannot make such claims. Thus, I fixed my wording to highlight the correlative relationship of my factors so readers do not get confused about what I may be trying to insinuate. For my conclusion, I was able to add more events that I was not aware could act as confounding variables before, such as the trade negotiations between the U.S. and China that also occurred in October of 2025.
As for my introduction, I addressed possible biases due to my background by explaining how I would be open-minded to whatever results my project produced. The research is useful regardless of what the outcome is, so I made sure to keep my language very neutral about what I expect the effects of the shutdown to be.