FINAL CONCLUSION AND RESULTS PROJECT: Estimation of the Marginal Propensity to Consume (MPC) DATA: 961 households with Income, Consumption, and Gender (head) METHOD: OLS Regression | Stata + R | MLE Distribution Fitting --- MAIN FINDINGS --- 1. ESTIMATED MPC = 0.5934 For every additional ₹1 of income, households spend approximately ₹0.59 on consumption and save the remaining ₹0.41. The estimate is highly statistically significant (t = 37.20, p < 0.001). The 95% confidence interval is [0.562, 0.625], which is narrow, indicating HIGH PRECISION of the estimate. 2. GENDER EFFECT = ₹4,912 (NOT SIGNIFICANT) Male-headed households consume about ₹4,912 more than female-headed households after controlling for income. However, this effect is statistically insignificant (t = 0.83, p = 0.405), consistent with the broader empirical literature which finds gender to be a weak predictor of consumption once income is controlled for. 3. MODEL FIT: R-squared = 0.5917 Income alone explains ~59% of the variation in household consumption, which is strong for cross-sectional microdata. 4. DISTRIBUTION FITTING: Both Log-normal and Gamma were fitted to income via MLE. The Log-normal distribution is preferred based on lower AIC, BIC, KS statistic, and Anderson-Darling statistic. This aligns with Gibrat's Law and the empirical income distribution literature. 5. ALL DIAGNOSTIC TESTS PASSED: - Breusch-Pagan (heteroskedasticity): p = 0.264 -> No heteroskedasticity - Shapiro-Wilk (normality of residuals): p = 0.221 -> Residuals are normal - Ramsey RESET (misspecification): p = 0.604 -> Model is well-specified OLS assumptions are satisfied. Results are reliable and unbiased. 6. LITERATURE COMPARISON: Our MPC of 0.593 is consistent with the empirical range of 0.35-0.90 reported in the literature (Parker et al., 2013; Boehm et al., 2025; Kosar & Melcangi, 2025). It falls within the standard range for households with moderate liquidity constraints. CONCLUSION: The modified Keynesian consumption function fits the data well. Income is the primary and statistically significant driver of consumption. The MPC of ~0.59 implies a fiscal multiplier greater than 1 (1/(1-MPC) ≈ 2.44), suggesting that income-support fiscal policies would have a substantial stimulative effect on aggregate demand in this household sample. Gender plays no significant independent role in consumption once income is accounted for, consistent with prior microeconomic evidence.

1. Findings and Conclusions After processing the raw data and using the ARIMA package to model ride-share data, I was able to make predictions for the 25 days beyond the current data set. 2. Qualitatively the data shows that as the weather gets warmer, the number of bike rentals increases, and over the course of two years, the number of rentals increases over the number of rentals from the previous year. 3. As the data terminates at the end of one cycle, I expect the number of rentals to increase to a level higher than it was a year before, which is what the models are predicting. 4. Therefore, the results were what I expected; the data appears to oscillate up and down over a 1-year period, with the overall data moving towards higher rental numbers.

This project utilises R, ggplot2, and the tidyverse to visualise vaccination percentages across European Union member states. The visualisation merges spatial data from the maps package with a custom dataset (EUvaccine.csv). Key Technical Features: Data Cleaning: Filtering missing values and joining spatial data using left_join. Geospatial Mapping: Using geom_polygon to render country borders and fill colours based on vaccination density. Image Composition: Utilising the cowplot library (ggdraw and draw_image) to overlay a custom image onto the map visualisation.

This is the Economic Trend Analysis Visualisation of Economics data from the ggplot2 package of RStudio. Using Economics data of the USA, I analyse the Us Unemployment trend from 1967 to 2015, predict the personal savings rate of US citizens, and using Time series forecasting, I forecast the unemployment trend in the US. After analysis I combined the seasonal pattern + Trend + Residuals, I derived the Trend composition and decomposed the unemployment time series.