Forecasting India's GDP over 2015-2019 to assess the performance - - PowerPoint PPT Presentation

Forecasting India's GDP

• ver 2015-2019

to assess the performance

• f Modi Government

FCAS A Group 7

PG ID Student Name

61910640 Shashank Suryae 61910066 Amol Kamath 61910411 Soumya Joshi 61910805 Mithilesh Parvataneni 61911003 Gary Rohan Singh 61910474 Rashi Choudhari

Business Problem

This project is financed by the Department of Expenditure, Ministry of Finance, India with the aim of assessing the impact made by the Government of India on the GDP during the years 2014-2019 or its years in power Specifically, the impact made by the various schemes that were introduced by the Govt. of India as a panacea for the problems faced by Indian industries. Challenges that we faced while building this forecasting model : 1. Unavailability of reliable data for certain GDP components 2. Only annualized data is available resulting in small data sets for modeling 3. Discounting external global factors and their impact on the Indian GDP

Business Problem

Client Goals 1. The client will use this project to compare the country’s actual GDP against the forecasted GDP 2. To pitch the incumbent party’s successful solution in improving the economical situation of the country Forecasting Impact A model that under-forecasts the GDP implies that the Government was successful in making India better as compared to the usual growth pattern and paints the Government in a favorable light while a model that over-forecasts does the exact opposite

Forecasting Goal

• To accurately forecast GDP over a forecasting horizon of 5 years (2014-2019), the GDP was

broken down to its base components – Household consumption, Government consumption, Gross Capital Formation, Net of Exports

• Each of these components was separately evaluated and forecasted using a variety of different

forecasting methods and models

• These were then combined to generate the total GDP forecast
• These forecasted GDP values were then compared with the actual values and the impact of the

Government schemes was assessed

• In addition to the above GDP components, employment was also

forecasted and used to get a GDP estimate for comparison purposes

Data Description

The World Bank - https://data.worldbank.org/country/india was used as the data source for this forecasting exercise.

Data components – Annualized

• GDP (current US$) – 1960 - 2017
• Households and NPISHs final consumption expenditure (Current US$) – 1960 - 2017
• General government final consumption expenditure (Current US$) – 1960 - 2017
• Gross capital formation (Current US$) – 1960 - 2017
• Net of Exports (Current US$) – 1960 - 2017
• Employment to population ratio – 1991 - 2017

Forecasting Methods & Outputs

Direct GDP Forecast Household Expenditure Governmen t Expenditure Net Exports Investments Employmen t

Regression (Quadratic with seasonality) ARIMA model with lag-1 Regression (Multiplicative Seasonality) Exponential (optimized) Holt-winters (additive) Double Exponential

Year Direct GDP Forecast (in Bn $) Household Expenditure Government Expenditure Net Exports Investments Employment

2015 2020.8395 1285.6876 220.487

• 60.8936

714.4857773 690003216.4 2016 2239.4323 1395.3661 238.619

• 60.8936

735.300555 694336320.5 2017 2463.8382 1515.2553 257.124

• 60.8936

756.1153326 698669424.5 2018 2518.4852 1646.3734 260.753

• 60.8936

776.9301103 703002528.5 2019 2774.5451 1789.8465 293.107

• 60.8936

797.7448879 707335632.6

Forecasting Methods & Outputs

Year Direct GDP Forecast (in Bn $) GDP from 4 components UCL from 4 components LCL from 4 components GDP forecast from Employment Actual GDP Figures (World Bank) 2015 2020.8395 2159.767 2324.009 1995.525 1805.946 2102.390809 2016 2239.4323 2308.392 2472.634 2144.150 1856.053 2274.229711 2017 2463.8382 2467.601 2631.843 2303.359 1906.16 2600.818244 2018 2518.4852 2623.163 2787.405 2458.921 1956.267 2823 (P) 2019 2774.5451 2819.805 2984.047 2655.563 2006.374 3035 (P)

Data Evaluation

Partitioning

• Training (1965 - 2009)
• Validation (2010 - 14)
• Forecast (5 years)

Metrics

• RMSE, MAPE for

reviewing the errors

• Residual plots to check if

trend & seasonality was captured

Benchmark

• Naïve model was used

as the benchmark for every time series data against which other methods were compared

Comparison

• For every dataset, we

compared the metrics and chose the model with lowest errors & appropriate residual plot

Conclusion

Conclusion

• The GDP has outperformed the mean prediction in all 3 cases
• It touches the 95% UCL of GDP for 4 components - Statistical errors are defined when

collectively measured in the set

• This implies that the government has done a good job of changing the trajectory of GDP
• GDP forecasted through employment is significantly lower - indicating an increase in

productivity amongst the employed workforce

• This report can be used as a work of proof to showcase the achievements of the current

government and by the ruling party for the upcoming elections