A starting point
Understanding Iraq’s economy in the years after 2003.
01 IRAQ ECONOMIC OBSERVATORY
Iraq’s economic outlook.
Iraq’s Economic Outlook: Building an Economic Analysis System with Iraq as a Case Study
I grew up in Iraq, and I want to understand how its oil wealth relates to pressure on public finances and the difficulty of building a more diverse economy. This is the Python application and research I am developing for my bachelor's thesis at ASE.
Economic Informatics · ASE Bucharest
My intended defence: July 2027
Economic analysis
Reading the evidence, following the sources and asking better questions.
A Python system
Bringing data, methods and an interactive application together.
Possible futures
Exploring forecasts, uncertainty and clearly stated scenarios.
02 / REAL SOURCES AND CLEAR WINDOWSTHE DATA COMES FIRST
I start with
the new Iraq.
2003 marks the US-led invasion and the fall of Saddam Hussein's regime. I chose this as the starting point for understanding Iraq's economy in the period that followed.
Annual history
Growth, real income per person and manufacturing run through 2025. Oil rents keep their available 2003–2021 window.
World Bank WDI ↗Monthly fiscal detail
The continuous model uses 2016–June 2026. I preserve cumulative accounts, derive monthly increments and keep the earlier archive separate for review.
CBI statistical bulletins ↗A published annual outlook
The IMF April 2026 projections through 2031 provide a separate baseline for sensitivity calculations.
IMF World Economic Outlook ↗January 2016–June 2026 in the continuous fiscal sample used by the saved monthly experiment.
A mixed catalogue of source tables and series with different periods, definitions and units. The count is not a combined dataset.
Earlier CBI records remain candidates for source review. They are kept outside the accepted model sample and are not used to extend it.
NATIONAL ECONOMIC PORTRAITDATED DATA / CLEAR UNITS
Iraq at a glance.
The national picture, with the date and source kept beside every number.
Loading the saved research preview…
IRAQ / NATIONAL INDICATORS
Iraq
World Bank WDI · saved October 2026. Published national series can include estimates. Each card uses its own latest available year.
01 / WHAT I WANT TO UNDERSTANDDATA / ECONOMICS / CODE
I want to understand
Iraq better.
What interests me is how Iraq has developed after the US-led invasion in 2003 and the fall of Saddam Hussein's regime. I want to understand why having valuable oil resources has not made development straightforward, look at the evidence carefully, and build a tool that helps me explore the questions.
Iraq’s Economic Outlook: Building an Economic Analysis System with Iraq as a Case Study
How has Iraq developed since 2003?
I chose 2003 because it marks the invasion and the fall of Saddam Hussein's regime. I want to follow the economy after that change, using growth, real GDP per person and manufacturing.
What puts pressure on the budget?
I am examining recorded revenue and spending month by month, while keeping oil-related income and accounting adjustments visible.
Can a forecast improve on a simple baseline?
I compare five methods on the same months and inspect their errors and uncertainty before choosing a final model.
How much do the assumptions change the outlook?
I can change an oil-price or spending assumption and calculate the resulting pressure around a published annual baseline.
I want to understand what the evidence shows. A country comparison can help me spot a pattern, but it cannot prove what caused it.
RESULTS DASHBOARDHISTORY / OUTLOOK / MODELS / SCENARIOS
The results, so far.
A quick route into the evidence, saved model results and assumptions I can change.
INTERNATIONAL SOURCE PROJECTIONS
The IMF outlook for Iraq
Choose a source release or compare revisions. These are IMF projections, separate from my models and the April scenario baseline.
Solid lines show historical or estimated IMF values; dashed lines show source projections. These are separate from my saved monthly revenue forecasts and user-selected budget scenarios.
ANNUAL GDP MODELMETHOD / BENCHMARKS / LIMITS
A model I can evaluate.
Separate from the monthly revenue experiment and the IMF source outlook.
03 / THE APPLICATION I AM BUILDINGWORKING PROTOTYPE
IRAQ ECONOMIC OBSERVATORY
Try the work
for yourself.
This compact browser preview highlights selected saved results. Open the hosted Python application on its own page to try the public version.
Try the Python applicationOpen the application in its own workspace.
MY RESEARCH WORKBENCHSaved source snapshots: October 2026
Loading the saved research preview…
The longer economic picture
I keep the history from 2003 and compare the yearly figures as they were published.
How to read this view
- Real GDP per person: economic output divided by population, adjusted to 2015 prices. It is not household income.
- Real GDP growth: the annual percentage change in economic output after price changes.
- Manufacturing share: manufacturing value added as a percentage of GDP.
- Oil rents: the estimated value of oil production after production costs, as a percentage of GDP. This is not government oil revenue.
Each point is a published annual observation. Missing years stay missing. Hover or tap a line to inspect the country, year and saved value.
Select country lines to focus on them. Select several to compare, or show only your selection.
Source-backed context in this period
These dated events provide context. Their timing does not establish the cause of a change in the chart.
Does a more complex model help?
I compare five methods on the same 24 monthly total-revenue increments, from July 2024 to June 2026, in nominal trillion IQD. The models do not forecast the monthly budget balance. A three-month horizon predicts the single third target month, not a three-month sum. MAE is the average size of the errors; RMSE gives bigger mistakes more weight.
How to read the forecast comparison
- MAE: average absolute prediction error. Lower is better.
- RMSE: prediction error with more weight given to large mistakes. Lower is better.
- Trillion IQD: one trillion Iraqi dinars, at the prices of that period.
- 95% prediction interval: the model’s uncertainty range for a future observation, under its assumptions.
The methods share 19 origins in each evaluation stage. Selection uses earlier validation MAE; held-out results are reported separately. Multi-month errors can be dependent. The chart uses a saved June 2026 origin.
WHAT I FOUND SO FAR AR(1) has the lowest MAE. ARIMA(1,1,1) has the lowest RMSE. I still need to check the model assumptions and remaining patterns in the errors before making my final choice.
The saved origin ends in June 2026; it is not a fresh forecast. Shaded ranges are conditional model intervals. Simple benchmarks have no supplied intervals, which does not mean zero uncertainty.
View forecast score table
| Method | MAE | RMSE | Targets |
|---|---|---|---|
| Last month | 3.155 | 5.222 | 24/24 |
| Same month last year | 3.669 | 5.539 | 24/24 |
| AR(1)LOWEST MAE | 2.815 | 4.551 | 24/24 |
| ARIMA(1,1,1)LOWEST RMSE | 2.956 | 4.418 | 24/24 |
| SARIMA(1,0,0)(0,1,1)[12] | 3.528 | 5.000 | 24/24 |
What if oil prices fall?
I use the IMF's April 2026 yearly projections as a starting point. Change the assumptions to see how the budget balance changes.
How to read the budget scenario
- Baseline: the balance calculated from the IMF’s published annual revenue, spending and GDP projections.
- Scenario: the balance after your price assumption changes the selected share of revenue.
- Balance as % of GDP: revenue minus spending, divided by fixed baseline GDP. Negative values show a deficit; positive values show a surplus.
- Percentage points: the difference between the two balance percentages, not a percentage change.
The sensitive share is an assumption. Price and quantity multiply together; other revenue and spending have separate controls. GDP and the exchange rate remain fixed.
The assumptions describe an accounting scenario, not its likelihood. The balancing threshold solves the same equation; it does not predict a market oil price.
Selected annual money flow
Four completed CBI years, 2022–2025. The 2024 and 2025 totals use December cumulative accounts. These preliminary, revisable figures stay separate from the IMF perimeter and projections.
The published spending breakdown
IMF Article IV 2025, Table 2. Select one level of the hierarchy; parent totals are not additional spending.
What makes a country comparable?
I compare countries using their 2003–2007 oil rents, real GDP per person and manufacturing. Change how much each feature matters to see how the ranking changes. My calculation uses log GDP per person to reduce the effect of very large income differences.
How to read the country comparison
Iraq is the reference, with a distance of zero. Every loaded comparison country is listed. The features come from the 2003–2007 period.
- Weights: how much each feature matters after the features are put on a common scale. Zero excludes that feature.
- Income: log real GDP per person, which reduces the effect of very large income differences.
- Distance: a unitless similarity measure. A lower value means a closer match on the selected features.
The ranking describes similarity within this country set. It does not establish why economies developed differently.
Country paths and oil-rent exposure
Select the countries you want to compare. The view starts with all twelve.
Each line keeps its original unit. Select a tag or line to focus a country; the legend colors stay consistent between the list and plot.
Oil-rent exposure and growth volatility
Each point summarizes the available 2003–2021 annual observations for one country: mean oil rents on the horizontal axis and sample standard deviation of GDP growth on the vertical axis. This descriptive comparison does not identify a cause.
COUNTRY / REFERENCEDISTANCE · UNITLESS
A lower distance means a closer match on these features, among the countries I have loaded. This does not prove why their economies developed differently. I may add Bahrain after checking its data.
This is a small taste of what the full application aims to offer. Use the link above to try the current public version of my Python application.
MEASUREMENTS AND CONTEXTDEFINITIONS / DATED EVIDENCE
What the evidence can tell me.
Informal work, unpaid obligations and measurement methods have different sources and definitions.
Informal work: the 2021 survey
Oil channels in the full Python application
05 / WHAT I HAVE BUILTA PROGRAM AND A RESEARCH PROCESS
More than
the final chart.
I want to be able to follow a value from its source to the result on screen. That means preserving definitions, checking the calculations and making the analysis reproducible.
- 01
A documented data pipeline
Source snapshots, units, gaps, fingerprints and cumulative-to-monthly conversion.
- 02
Explainable computations
Country distances, chronological forecast tests, visible uncertainty and scenario assumptions.
- 03
A working interface
Controls, source-quality views and exportable results that I can demonstrate and improve.
06 / LOOK THROUGH THE WORKMY CURRENT MATERIALS
The work is
taking shape.
01 / PRESENTATION
My thesis pitch
Browse the presentation I prepared for my thesis proposal.
VIEW MY PITCH ↓ 02 / RESEARCHThe plan behind it
My question, data, proposed methods and the work I still need to complete.
Read my research plan ↗ 03 / PYTHON APPLICATIONThe application I am building
Open the public application in its full-size workspace.
TRY THE APPLICATION ↗THE POWERPOINT · INSIDE THE PAGE
Browse my
thesis pitch.
Use previous and next, or the arrow keys, to browse the slides.
Original proposal slides · English01 / 19
MY THESIS PITCH
Iraq, oil and the budget
My original proposal for a Python application and research into Iraq’s economy.
KEEP OR PRESENT THE ORIGINAL DECK
07 / KEEPING THE WORK MOVINGMY ROUTE TO July 2027
What I'm
working on.
My approved thesis title
My thesis title is approved: Iraq’s Economic Outlook: Building an Economic Analysis System with Iraq as a Case Study. Professor Uscatu is coordinating the thesis, and I look forward to refining the work with my professors.
A place to keep the work together
I have organized the program, data, research and writing into one private repository, with this page as a way to share the project as it develops.
Moving the fiscal experiment to monthly data
After checking the archived CBI tables, I prepared a continuous 126-month fiscal core and tested five forecasting methods on 24 shared targets.
Starting with evidence
I checked the available datasets before choosing the direction, prepared the annual development history and started the first Python prototype.
I am starting with Iraq.
I want the program to go further.
Eventually, I want this program to be useful for any country, with a clear way to import different datasets. For now, I am building and testing it around the Iraqi data I understand and care about. Once that works well, I want to make the import process, country choices and analysis easier to adapt. Different datasets will still need their own checks and suitable methods; that wider version is something I want to develop after the core thesis.
THE REASON I KEEP COMING BACK TO THIS
A place I know.
An economy I want
to understand.
I want this project to help me ask better questions, become a better programmer and learn statistics I can keep building on.
Talk to me about the project