Prognostička točnost, nepristranost i racionalnost investicijskih istraživanja u Europi

doctoral thesis

• Title (english): Forecast accuracy, unbiasedness and rationality of investment research in Europe
• Author: Zoran Lukić; MBZ: 342682
• Mentor: Josip Arnerić (mentor); MBZ: 274256
• Committee member: Davor Zoričić (predsjednik povjerenstva); MBZ: 258113
• Granter: University of Zagreb; Faculty of Economics and Business; Zagreb
• Defense date and country: 2021-12-13, Croatia
• Scientific / art field, discipline and subdiscipline: SOCIAL SCIENCES; Economics
• Universal decimal classification (UDC): 330 - Economics. Economic science

Abstract

Prognoze investicijskih istraživanja o financijskim pokazateljima poduzeća značajno utječu na tržišta kapitala te su važno sredstvo potpore odlučivanju u financijskom sektoru. Nasuprot tome, dosadašnja istraživanja pretežito zaključuju da su ove prognoze nedostatno točne te da nisu nepristrane ili racionalne. Nalazi se većinom temelje na objedinjenim tržišnim podatcima za jedan pokazatelj i prognostički horizont, uglavnom za tržište SAD, te standardnim mjerama prognostičkih pogrešaka. U ovome je radu, umjesto standardnih mjera pogrešaka, predložen pojam točnosti linearnog trenda prognoza kao kriterij prihvatljivosti primjeren dugoročnim ulagateljima. Pojmovi nepristranosti i racionalnosti razmatrani su sukladno teoriji racionalnih očekivanja. Razrađena je postojeća zamisao uklanjanja procijenjene povijesne pristranosti iz budućih prognoza, u svrhu poboljšanja točnosti. Ključni su pojmovi matematički formalizirani, zajedno s dokazima potrebnih karakterizacija i svojstava, uključujući sukladnost točnosti trenda s nekim standardnim mjerama pogrešaka, kao i hijerarhijski odnos između točnosti trenda, nepristranosti i racionalnosti. Prisutnost hipoteziranih svojstava ispitana je na dosad najopsežnijem uzorku tržišno konsenzusnih prognoza europskih investicijskih istraživanja, za 233 društva s najvećim likvidnim tržišnim kapitalizacijama, tri glavna financijska pokazatelja te četiri prognostička horizonta, u razdoblju 2005. – 2017. godine. Rezultati ispitivanja, provedenih na razini pojedinih društava, upućuju da prevladavajući udio prognoza pokazuje točan trend (ukupno 64 – 67%), nepristranost (ukupno 75 – 77%) i slabu racionalnost (ukupno 67 – 71%). Postupak uklanjanja pristranosti umjereno je uspješan za kratkoročne horizonte, gdje smanjuje prosječnu pogrešku MAPE za ukupno 22%, pri čemu su pogreške smanjene kod 63% prognoza, dok za dugoročne horizonte postupak nije uspješan. Objašnjenje snažne prisutnosti poželjnih svojstava točnoga trenda, nepristranosti i slabe racionalnosti u prognozama investicijskih istraživanja jest pretežita uporaba financijskih modela društava u prognostičkom postupku, koji se, sukladno teoriji kompleksnih sustava, temelje na velikom broju prediktora i veza između varijabli te na stručnoj prosudbi analitičara.

Abstract (english)

Investment research forecasts of corporate financial indicators have a significant impact on capital markets and are an important decision support tools in financial sector. In contrast, previous research largely concludes that these forecasts are insufficiently accurate, and are not unbiased or rational. The findings are mostly based on composite market data for a single indicator and forecast horizon, mainly for the US market, and standard forecast error measures. In this study, instead of standard error measures, the concept of forecast linear trend accuracy is proposed as an acceptability criterion appropriate for long-term investors. The concepts of unbiasedness and rationality are considered according to the rational expectations theory. The existing idea of eliminating the estimated historical bias from future forecasts, in order to improve accuracy, is elaborated. The key terms are mathematically formalized, along with proofs of the required characterizations and properties, including the consistency of trend accuracy with certain standard error measures, as well as the hierarchical relation between trend accuracy, unbiasedness and rationality. The presence of hypothesised properties was tested on the most comprehensive sample of market consensus forecasts of European investment research to date, for 233 companies with the largest liquid market capitalizations, three main financial indicators and four forecast horizons, in the period 2005–2017. The results of tests, conducted at the level of individual companies, indicate that the prevailing share of forecasts exhibits accurate trend (64–67% in total), unbiasedness (75–77% in total) and weak rationality (67–71% in total). The bias elimination procedure is moderately successful for short-term horizons, where it reduces the average error MAPE by a total of 22%, with errors reduced in 63% of forecasts, while for long-term horizons the procedure is not successful. The explanation for the strong presence of desirable properties of accurate trend, unbiasedness and weak rationality in investment research forecasts is the predominant use of corporate financial models in the forecasting process, which are, in accordance with the complex systems theory, based on a large number of predictors and relations between variables as well as the expert judgment of analysts.

• Keywords: financijske prognoze; investicijska istraživanja; prognostička točnost; racionalna očekivanja; nepristranost; financijski modeli društava; točnost trenda
• Keywords (english): financial forecasts; investment research; forecast accuracy; rational expectations; unbiasedness; corporate financial models; trend accuracy
• Language: croatian
• URN:NBN: urn:nbn:hr:148:648318
• Promotion: 2022
• Study programme: Title: Economics and Business Economics; Study programme type: university; Study level: postgraduate; Academic / professional title: doktor/doktorica znanosti, područje društvenih znanosti, polje ekonomija (doktor/doktorica znanosti, područje društvenih znanosti, polje ekonomija)
• Type of resource: Text
• Extent: 249 str./p.
• File origin: Born digital
• Access conditions: Access restricted to students and staff of home institution
• Created on: 2022-02-06 12:36:28