umu.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • harvard1
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
GDP Growth Rate Nowcasting and Forecasting
Umeå University, Faculty of Science and Technology, Department of Physics.
2017 (English)Independent thesis Advanced level (professional degree), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

The main purpose of this project was to help Swedbank get a better understandingof how gross domestic product growth rate develops in the future froma data set of macroeconomic variables. Since GDP values are released long aftera quarter has ended Swedbank would like to have a model that could predictupcoming GDP from these data sets. This was solved by a combination ofgrowth rate predictions from a dynamic factor model, a vector autoregressivemodel and two machine learning models. The predictions were combined usinga weighting method called system averaging model where the model predictionwith least historical error receives the largest weight in the nal future prediction.In previous work a simple moving average model has been implementedto achieve this eect however there are several aws in a simple moving averagemodel. Most of these defects could in theory be avoided by using an exponentialweighting scheme instead. This resulted in the use of an exponentialweighting method that is used to calculate weights for future predictions. Themain conclusions from this project were that some predictions could get betterwhen removing bad performing models which had too large of a weight. Puttingtoo high weight on a single well performing model is also not optimal since thepredictions could get very unstable because of varying model performance. Theexponential weighting scheme worked well for some predictions however whenthe parameter , that controls how the weight is distributed between recent andhistorical errors, got too small a problem arose. Too few values were used toform the nal weights for the prediction and the estimate got unsteady results.

Place, publisher, year, edition, pages
2017.
National Category
Mathematical Analysis Computational Mathematics Probability Theory and Statistics Economics
Identifiers
URN: urn:nbn:se:umu:diva-132951OAI: oai:DiVA.org:umu-132951DiVA: diva2:1084527
External cooperation
Swedbank
Subject / course
Examensarbete i teknisk fysik
Educational program
Master of Science Programme in Engineering Physics
Available from: 2017-03-27 Created: 2017-03-25 Last updated: 2017-03-27Bibliographically approved

Open Access in DiVA

fulltext(4767 kB)29 downloads
File information
File name FULLTEXT01.pdfFile size 4767 kBChecksum SHA-512
43678c7e32b57cf2021a5b88cb3fdb0bad6ddb96b5c01c88478e16e7daa9bdd57e49d409bb76c1f9f47d3dba0d08b432b924db7b162fec1a3cbfb0aa1a4c8a15
Type fulltextMimetype application/pdf

By organisation
Department of Physics
Mathematical AnalysisComputational MathematicsProbability Theory and StatisticsEconomics

Search outside of DiVA

GoogleGoogle Scholar
Total: 29 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

Total: 157 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • harvard1
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf