kth.sePublications KTH
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • 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
Univariate Forecasting for REITs with Deep Learning: A Comparative Analysis with an ARIMA Model
KTH, School of Architecture and the Built Environment (ABE), Real Estate and Construction Management, Real Estate Economics and Finance.
KTH, School of Architecture and the Built Environment (ABE), Real Estate and Construction Management, Real Estate Economics and Finance.ORCID iD: 0000-0003-4849-0726
(English)Manuscript (preprint) (Other academic)
Abstract [en]

This study aims to investigate whether the newly developed deep learning-based algorithms, specifically Long-Short Term Memory (LSTM), outperform traditional algorithms in forecasting Real Estate Investment Trust (REIT) returns. The empirical analysis conducted in this research compares the forecasting performance of LSTM and Autoregressive Integrated Moving Average (ARIMA) models using out-of-sample data. The results demonstrate that in general, the LSTM model does not exhibit superior performance over the ARIMA model for forecasting REIT returns. While the LSTM model showed some improvement over the ARIMA model for shorter forecast horizons, it did not demonstrate a significant advantage in the majority of forecast scenarios, including both recursive multi-step forecasts and rolling forecasts. The comparative evaluation reveals that neither the LSTM nor ARIMA model demonstrated satisfactory performance in predicting REIT returns out-of-sample for longer forecast horizons. This outcome aligns with the efficient market hypothesis, suggesting that REIT returns may exhibit a random walk behavior. While this observation does not exclude other potential factors contributing to the models' performance, it supports the notion of the presence of market efficiency in the REIT sector. The error rates obtained by both models were comparable, indicating the absence of a significant advantage for LSTM over ARIMA, as well as the challenges in accurately predicting REIT returns using these approaches. These findings emphasize the need for careful consideration when employing advanced deep learning techniques, such as LSTM, in the context of REIT return forecasting and financial time series. While LSTM has shown promise in various domains, its performance in the context of financial time series forecasting, particularly with a univariate regression approach using daily data, may be influenced by multiple factors. Potential reasons for the observed limitations of our LSTM model, within this specific framework, include the presence of significant noise in the daily data and the suitability of the LSTM model for financial time series compared to other problem domains. However, it is important to acknowledge that there could be additional factors that impact the performance of LSTM models in financial time series forecasting, warranting further investigation and exploration. This research contributes to the understanding of the applicability of deep learning algorithms in the context of REIT return forecasting and encourages further exploration of alternative methodologies for improved forecasting accuracy in this domain. 

Keywords [en]
Forecasting, Equity REITs, deep learning, LSTM, ARIMA
National Category
Economics and Business
Identifiers
URN: urn:nbn:se:kth:diva-337259OAI: oai:DiVA.org:kth-337259DiVA, id: diva2:1801032
Available from: 2023-09-28 Created: 2023-09-28 Last updated: 2023-12-01
In thesis
1. US Equity REIT Returns and Digitalization
Open this publication in new window or tab >>US Equity REIT Returns and Digitalization
2023 (English)Licentiate thesis, comprehensive summary (Other academic)
Abstract [en]

This licentiate thesis is a collection of two essays that utilize time-series econometric methods in real estate finance. The first essay applies econometric modelling on Real Estate Investment Trust (REIT) index returns, focusing on estimating the effect of the quantitative easing (QE) and quantitative tightening (QT) programmes on U.S. equity REIT index returns, while controlling for several other important macro-financial factors. The second essay instead focuses on forecasting U.S. equity REIT index returns empirically, where the performance of a traditional econometric model (ARIMA) is compared to a modern state-of-the-art deep learning-based model (LSTM).

Digitalization, which encompasses a broad range of technological advancements, is the main factor that we study for its impact on REIT investments. One perspective on the impact of digitalization on REITs is its effect on inflation. Digitalization has the potential to increase productivity and reduce costs, which could help to keep inflation low. This, in turn, has in the recent decades provided a supportive environment for REIT investments through lower interest rates, which we partly investigate in the first essay.

Another perspective is that digitalization has not only led, but is also expected to lead, to significant innovations in the field of artificial intelligence (AI) and machine learning (ML), including deep learning (DL), which is a subset of ML. Many researchers and market practitioners are currently working to develop models that can use large amounts of data to make better forecasts and investment decisions. If successful, these models could significantly improve the performance of REIT investments. Can DL models be trained to make better forecasts for making investments? This is a question we ask ourselves in the second essay.

The study of digitalization and its effects on inflation has been a growing area of interest in recent years, with researchers exploring the potential impact of technological advancements on macroeconomic trends, which founded the base to our studies. However, recent developments in the global economy have shifted the focus of this research, as inflation levels have unexpectedly risen from what was previously believed to be a low and stable environment. As a result, the setting and framework for our research on digitalization and inflation have been significantly altered, as we have tried to adapt to this changing landscape.

Abstract [sv]

Denna licentiatuppsats är en samling av två forskningsartiklar som använder tidsserieekonometriska metoder inom finansiell ekonomi med fokus på fastighetsaktier. Den första artikeln tillämpar ekonometriska metoder på tidsseriedata för amerikanska börsnoterade fastighetsfonder, Real Estate Investment Trusts (REITs), med fokus på att uppskatta effekten av icke-konventionella penningpolitiska aktiviteter (kvantitativa lättnader och kvantitativ åtstramning) på avkastningsserierna, samtidigt som vi kontrollerar för andra viktiga makroekonomiska och finansiella variabler. Den andra artikeln fokuserar istället på att bygga modeller för prognoser av avkastningen på avkastningsserierna empiriskt, där prognosfelen för en traditionell ekonometrisk modell (ARIMA) jämförs med en modern djupinlärningsbaserad modell (LSTM).

Digitalisering, som omfattar ett brett spektrum av tekniska framsteg, är den viktigaste faktorn som vi studerar för dess inverkan på REIT-investeringar. Ett perspektiv på digitaliseringens inverkan på REITs är dess effekt på inflationen. Digitalisering har potential att öka produktiviteten och minska kostnaderna, vilket kan bidra till att hålla inflationen låg. Detta har i sin tur under de senaste decennierna varit fördelaktigt för REIT-investeringar genom lägre räntor, vilket vi delvis undersöker i den första uppsatsen.

Ett annat perspektiv är att digitaliseringen inte bara har lett, utan också förväntas leda, till betydande innovationer inom området artificiell intelligens (AI) och maskininlärning (ML), inklusive djupinlärning (DL), som är en delmängd av ML. Många forskare och professionella aktörer arbetar just nu med att utveckla modeller som kan använda stora mängder data för att göra bättre prognoser och investeringsbeslut. Om de lyckas kan dessa modeller förbättra resultatet för REITinvesteringar avsevärt. Kan DL-modeller tränas för att förbättra möjligheterna till att göra mer tillförlitliga prognoser och därmed öka chanserna till att göra mer lönsamma investeringar? Det är en fråga vi ställer oss i den andra artikeln.

Digitalisering och dess effekter på inflationen har varit ett starkt växande fält inom såväl forskning som praktisk tillämpning de senaste åren, med forskare som undersöker den potentiella inverkan av tekniska framsteg på makroekonomiska trender, vilket har legat till grund för våra studier. Den senaste tidens utveckling i den globala ekonomin har dock flyttat fokus för denna forskning, eftersom inflationsnivåerna oväntat har stigit från vad som tidigare ansågs vara en låg och stabil miljö. Som ett resultat har miljön och ramarna för vår forskning om digitalisering och inflation ändrats avsevärt, eftersom vi har försökt anpassa oss till detta föränderliga landskap.

Place, publisher, year, edition, pages
Stockholm: Kungliga Tekniska högskolan, 2023. p. 39
Series
TRITA-ABE-DLT ; 2352
Keywords
REITs, quantitative easing, quantitative tightening, deep learning, LSTM, REITs, kvantitativa lättnader, kvantitativ åtstramning, djupinlärning, LSTM
National Category
Economics and Business
Research subject
Real Estate and Construction Management
Identifiers
urn:nbn:se:kth:diva-340300 (URN)978-91-8040-791-5 (ISBN)
Presentation
2023-12-18, H1, Teknikringen 33, KTH Campus, public video conference link https://kth-se.zoom.us/j/64745915450, Stockholm, 09:00 (English)
Opponent
Supervisors
Note

QC 20231201

Available from: 2023-12-01 Created: 2023-12-01 Last updated: 2024-01-24Bibliographically approved

Open Access in DiVA

fulltext(1220 kB)374 downloads
File information
File name FULLTEXT02.pdfFile size 1220 kBChecksum SHA-512
896f09e361b4bb5279eb28987cb32cf6305f8723195029d229e6e5fc0d72adbaedd50840e2b424716e41892282d9f81b91a03bcd34dc04539eec01f170fefad2
Type fulltextMimetype application/pdf

Authority records

Axelsson, BirgerSong, Han-Suck

Search in DiVA

By author/editor
Axelsson, BirgerSong, Han-Suck
By organisation
Real Estate Economics and Finance
Economics and Business

Search outside of DiVA

GoogleGoogle Scholar
Total: 449 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

urn-nbn

Altmetric score

urn-nbn
Total: 818 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • 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