About
Angavu
Macro.
Angavu Macro is an independent quantitative research project for constructing, comparing and evaluating macroeconomic forecasts—and the evidence behind them.
About Angavu Macro
I begin with transparent benchmarks—random walk, IMA(1,1) and direct autoregression—and test whether ARIMA and inflation-specific challengers improve out-of-sample forecasting performance. Inflation-specific specifications are evaluated only when the necessary information is available at the forecast origin. Broader multivariate and shrinkage-based approaches remain directions for future research.
Model selection is based on validation mean squared forecast error (MSFE) for a specified target, forecast horizon and estimation regime. Forecast-bias and scale diagnostics, calibration, model stability and independent holdout evaluation, where available, provide additional evidence. Stability and review requirements can also constrain model adoption; they do not silently replace the MSFE selection rule.
What interests me is not simply which model records the lowest error in a particular historical sample. It is whether that advantage persists, when it breaks down, and what can be learned from disagreement between competing forecasts. Understanding how forecast performance changes across economic conditions, estimation windows and volatility remains part of the broader research agenda.
Angavu Macro treats each published forecast as a research record. It preserves the forecast origin, information cutoff, model identity, competing estimates, evaluation evidence and provenance. Historical forecasts can be compared across origins and, where outcomes are available, assessed against realized data. Contemporaneous releases remain distinguishable from later reconstructions.
The current empirical work focuses on U.S. macroeconomic indicators, including inflation, economic activity, unemployment and monetary policy rates. The project remains under active development and does not assume that more elaborate models necessarily produce better forecasts.
Intended use
Angavu Macro is an independent applied quantitative research project under active development. Its forecasts, model comparisons and analytical outputs are intended for economic research, quantitative analysis and methodological evaluation.
The project is designed to help researchers and quantitative analysts examine competing forecasts, assess historical performance and understand the evidence and limitations behind published estimates.
Angavu Macro does not provide investment advice, trading recommendations, individualized financial advice or recommendations to undertake any particular financial transaction.
Forecasts are conditional on the underlying data, models, estimation windows and available information. Historical forecasting performance does not guarantee future accuracy.
About the Author
Bihemo Kimasa
Contact: bkimasa@gmail.com
I'm a macroeconomist and labour economist with a Ph.D. in Economics from the University of Konstanz, an M.Sc. in Quantitative Economics from the University of Kiel, and a B.Sc. in Electrical Engineering and Computer Science from Jacobs University Bremen.
My academic interests span macroeconomics, labour and product markets, firm dynamics, monetary economics and quantitative economic modelling.
Doctoral research
I completed my doctorate at the University of Konstanz in 2021, under the supervision of Professor Leo Kaas, among others.
My dissertation, Essays on Firm Dynamics and Labor Market Flows, investigated how firm dynamics interact with labour-market flows and the allocation of workers across occupations and firms.
The dissertation comprises three research essays. The first examines occupational job and worker flows using German establishment data. The second develops a quantitative general-equilibrium model to investigate occupational heterogeneity, labour-market reallocation and misallocation.
The third essay, co-authored with Leo Kaas, studies firm dynamics in frictional labour and product markets. It combines firm-level evidence with a model of heterogeneous firms competing for workers and customers, examining how productivity, demand and uncertainty affect prices, output and employment.
Together, these studies reflect my broader interest in understanding how microeconomic heterogeneity and changing economic conditions shape aggregate outcomes.
Monetary economics and quantitative training
Before my doctorate, I studied Quantitative Economics at the University of Kiel, with training in macroeconomics, monetary economics, advanced statistics and econometrics, financial theory, financial markets and applied empirical methods.
My M.Sc. thesis, Import Price Shocks and Imported Inflation in a Small Open Economy, developed a microfounded New Keynesian model to study the transmission of external price shocks and the monetary-policy responses available to a small open economy.
The model brings together three sources of external inflationary pressure: imported intermediate goods, including energy; imported consumption goods; and foreign inflation. Using a calibration informed by the Chilean economy, I examined how these shocks affect domestic inflation, the output gap and the real exchange rate, and compared the implications of domestic-inflation and CPI-inflation targeting.
The results highlighted the importance of the source of inflationary pressure when evaluating monetary-policy trade-offs.
The thesis received the 2013 Erich Schneider Prize at the University of Kiel, awarded for outstanding academic theses.
My earlier training in electrical engineering and computer science also informs my interest in computational modelling and reproducible quantitative research.
Research approach and AI assistance
Angavu Macro is an independent applied research project. Its development has been substantially accelerated by AI assistance, particularly OpenAI's ChatGPT, including GPT-5.5 and later models.
I direct the research, determine the methodological questions and priorities, and retain responsibility for the interpretation, development decisions and limitations of the work. AI has assisted with implementation, analysis workflows and documentation, but that assistance should not be confused with independent scientific validation.
The project is evolving, and its models, empirical results and software remain subject to testing, revision and critical evaluation.
Selected methodological references
The following works are part of the research foundations documented in Angavu's bibliography.
- Rossi, B. (2021). Forecasting in the Presence of Instabilities: How We Know Whether Models Predict Well and How to Improve Them. Journal of Economic Literature, 59(4), 1135–1190.
- Giacomini, R., & Rossi, B. (2010). Forecast Comparisons in Unstable Environments. Journal of Applied Econometrics, 25(4), 595–620.
- Stock, J. H., & Watson, M. W. (2007). Why Has U.S. Inflation Become Harder to Forecast?. Journal of Money, Credit and Banking, 39(s1), 3–33.
- Franses, P. H. (2020). IMA(1,1) as a New Benchmark for Forecast Evaluation. Applied Economics Letters, 27(17), 1419–1423.
- Mincer, J., & Zarnowitz, V. (1969). The Evaluation of Economic Forecasts. In Economic Forecasts and Expectations, NBER.
- Coibion, O., Gorodnichenko, Y., & Kamdar, R. (2018). The Formation of Expectations, Inflation, and the Phillips Curve. Journal of Economic Literature, 56(4), 1447–1491.