Analysts’ forecasts timeliness and accuracy post-XBRL

• Date: 04 March 2019
• Published date: 04 March 2019
• Pages: 151-188
• DOI: https://doi.org/10.1108/IJAIM-05-2017-0061
• Author: Sherwood Lane Lambert,Kevin Krieger,Nathan Mauck
• Subject Matter: Accounting & Finance,Accounting/accountancy,Accounting methods/systems

Analysts’forecasts timeliness and

accuracy post-XBRL

Sherwood Lane Lambert and Kevin Krieger

Department of Accounting and Finance, University of West Florida, Pensacola,

Florida, USA, and

Nathan Mauck

University of Missouri-Kansas City, Kansas City, Missouri, USA

Abstract

Purpose –To the authors’knowledge, this paper is the first to use Detail I/B/E/S to study directly the

timeliness of security analysts’next-year earnings-per-share (EPS) estimates relative to the SEC filings of

annual (10-K) and quarterly (10-Q) financial statements. Although the authors do not prove a causal

relationship, theyprovide evidence that the average time from firms’filings of 10-Ks and 10-Qs to the release

of analysts’annualEPS forecasts during short timeframes (for example,15-day timeframe from a 10-K’s SEC

file date) subsequentto the 10-K and 10-Q filing dates significantly shortened withXBRL implementation and

then remainedrelatively constant following implementation.

Design/methodology/approach –Using filing dates hand-collected from the SEC website for 10-Ks

during 2009-2011 and filing dates for 10-Ks and 10-Qs during 2003-2014 input from Compustat along with

analysts’estimated values for next year EPS, actual estimated next year EPS realized and estimate

announcement dates in Detail I/B/E/S, the authors study the days from 10-K and 10-Q file dates to

announcementdates and the per cent errors for individualestimates during per- and post-XBRL eras.

Findings –The authors find that analysts are announcing next-year EPS forecasts significantly more

frequently and in significantlyshorter time in zero to 15 days immediately following 10-K and 10-Q file dates

post-XBRL as compared to pre-XBRL. However, the authors do not find a significant change in forecast

accuracypost-XBRL as compared to pre-XBRL.

Research limitations/implications –Because this studyuses short timeframes immediately following

the events (filingsof 10-Ks and 10-Qs), therelationship between 10-Ks and 10-Qs with and without XBRL and

improved forecast timeliness is strengthened. However, even this strengthened difference-in-difference

methodology does not establish causality. Future research may determine whether XBRL or other factors

cause the improvedforecast timeliness the authors’evidence.

Practical implications –This improved efficiency may become critical if financialstatement reporting

expands as a resultof new innovations such as Big Data and continuousreporting. In the future, users may be

able to electronically connect to financial statement data that firms are maintainingon a perpetual basis on

the SEC website and continuouslymonitor and analyze the financial statement data dynamicallyin real time.

If so, then unquestionably,XBRL will have played a criticalrole in bringing about this future innovation.

Originality/value –Whereas previous studies have utilizedSummary IBES data to assess the impact of

XBRL on analyst forecasts, the authors use Detail IBES to study the effects of XBRL adoptiondirectly by

measuringdays from 10-K and 10-Q file dates in Compustat to each estimate’s announcementdate recorded in

IBES and by computing the per cent error using each estimate’s VALUE and ACTUAL recorded in Detail

IBES. Theauthors are the first to evidence a significant shorteningin average days and an increase in per cent

of 30-daycounts in the zero- to 15-day timeframe immediatelyfollowing the fillings of 10-K s and 10-Qs.

Keywords Analysts, Forecast accuracy, Detail I/B/E/SFE/S,

eXtensible Business Reporting Language (XBRL), Forecast timeliness

Paper type Research paper

A draft version of this paper won the “Best Paper of the Conference”award at the 2016 American

Accounting Association (AAA) Southeast Regional Conference.

Analysts’

forecasts

timeliness

151

Received11 May 2017

Revised15 January 2018

19March 2018

Accepted2 May 2018

InternationalJournal of

Accounting& Information

Management

Vol.27 No. 1, 2019

pp. 151-188

© Emerald Publishing Limited

1834-7649

DOI 10.1108/IJAIM-05-2017-0061

The current issue and full text archive of this journal is available on Emerald Insight at:

www.emeraldinsight.com/1834-7649.htm

1. Introduction

This paper investigates the impact of eXtensible Business Reporting Language (XBRL) on

the timeliness and accuracy of financial analysts’next-year EPS forecasts. Whereas

previous studies have utilizedSummary Institutional Broker Estimate System (I/B/E/S)data

to assess the impact of XBRL on analyst forecasts,this study measures the number of days

from 10-K and 10-Q filing dates in Compustat to each estimate’s announcement date

recorded in Detail I/B/E/S. Also, we compute the per cent error using each estimate’s

VALUE and ACTUAL recorded in Detail I/B/E/S. We are the first to evidence a significant

shortening of average daysand an increase in per cent of 30-day counts in the zero to 15-day

timeframe immediately following the fillings of 10-Ks and 10-Qs. Each forecast record in

Detail I/B/E/S contains the date when the analyst announced his or her forecast. This

information is not providedin Summary I/B/E/S[1]. Thus, to our knowledge, we are the first

study to directly evaluatethe effect of XBRL on the timeliness of analysts’forecasts. We find

that the timeliness of analysts’next-yearEPS forecasts made within short timeframes of 15

days or less after the filing of the prior-year10-Ks and forecast-year 10-Qs has significantly

improved during post-XBRL. Using the filing dates (which we hand-collect from the

Security and Exchange Commission (SEC) website), we measure the number of days

between the 10-K or 10-Q filing date and the forecast announcement date in Detail I/B/E/S

for each forecast. Hence, we can use Detail I/B/E/S to precisely measure the time in days

between when individual analystsannounce their forecasts and the filing dates of 10-Ks and

10-Qs. Our timeliness findings are importantbecause they are consistent with analysts now

having the capability to more efficiently input and process machine-readable financial

statement data usingcomputerized analytical routines.

Our findings may be particularly important as the future unfolds. They are consistent

with analysts now having the automated capability to more efficiently cope with future

innovations in financial statement reporting, which were often impractical with EDGAR

(such as reporting information more frequently than quarterly, or annually, and reporting

larger quantities and more types of information). Our consideration of potential

improvement in analyst forecast accuracy, post-XBRL, shows no dispositive evidence of

marked improvement.

The remainder of this paper proceeds as follows: Section 2 reviews prior research.

Section 3 presents our hypotheses, describes our methodology and provides further detail

regarding our data. Section 4 presents our findings. Section 5 provides a summary of our

results, describes potential areas for future research and concludes. Appendix provides

similar results to Section4, but with a specialized subsample mandatingconsistency of firm-

analyst identity overtime.

2. Background and prior research

On January 30, 2009, the SEC established requirements for companies to file their financial

statements using XBRL and mandateda phased implementation of XBRL during 2009-2011.

In fiscal year 2009, “large accelerated filers”,with a total market value of at least US$700m,

began using XBRL to file their financial statements with the SEC. “Acceleratedfilers”with

an aggregate market value of between US$75mand US$700m started using XBRL in fiscal

year 2010 for filing their financial statements with the SEC. The remainder of reporting

companies, called “non-accelerated filers”, began using XBRL in fiscal year 2011[2]. Hence,

starting with fiscal year 2011, and forevery year thereafter, XBRL financial statements are

available for virtuallyall SEC-reporting companies. The SEC believed that requiringissuers

to file their financial statements using XBRL’s interactive data format would enable

investors, analysts and the SEC to captureand analyze information more quickly and at less

IJAIM

27,1

152

cost than via a static format, unreadableby machine (Kernan, 2008;SEC, 2009). Some initial

research regarding the impact of XBRL, however,has been mixed. For example, Dong et al.

(2016) find agreement that more firm-specific information may come quickly to light under

XBRL. Meanwhile, Blankespoor et al. (2014) indicate that, contrary to initial SEC claims,

XBRL adoption further advantages larger investors who might be more readily able to

absorb big data.

During the mid-1990s, the SEC had mandated that companies file their financial

statements using the electronic data gathering, analysis, and retrieval (EDGAR) system.

With EDGAR, companies file their financial statementsusing an electronic document such

as portable document format (PDF) or hyper-text markuplanguage (HTML). At the time of

its development, EDGAR was seen as a revolutionaryimprovement for data users. EDGAR,

however, lacks many of the features provided by the development of XBRL. For example,

EDGAR does not provide machine-readable data that can be input directly to computer

programs. Thus, while an improvementover paper reports, EDGAR still requires inefficient

human interaction to locate, copy and paste data from EDGAR into machine-readable files

for input to computerizedroutines.

Analysts evaluate companies’past performances to forecast future performances, such

as next-year earnings and long-term growth projections. In assessing these past

performances, analysts use data contained in financial statements, as well as aggregated

financial statement data such as Standardand Poor’s (S&P) Compustat. Substantial time is

required to convert financial statement data into aggregate data. It is reasonable to posit

that analysts who use data taken directly from financial statements have benefited from

being able to input this data directly into their computerized analyses. In addition to

potentially providing machine-readable data more quickly, XBRLuses a specific taxonomy

that more accurately defines data reported in financial statements than the less-precise

terminology used in EDGAR and paper reports. XBRL’s taxonomy might be expected to

improve the accuracy of those using financial statement data. For example, Chychyla and

Kogan (2014) describe how XBRL has specifically aided in alleviatingrecording errors, like

those frequently seen in Compustat.Expanding data analysis may provide the potential for

improving the accuracy of analysts’forecasts. Improved timeliness is one of the primary

benefits considered possible with XBRL. Through the use of the internet, financial

information can be made available, not just at the end of the fiscal year or quarter, but

practically in real time. XBRL could improve firms’abilities to provide real-time data by

eliminating the need to re-key data, thus improving the speed of data acquisition (Baldwin

et al.,2006).

Much of the early work regarding XBRL is well summarized in a review article by

Perdana et al. (2015). Several recent studies have begun to consider how XBRL may impact

the frequency and accuracy of analysts’earningsforecasts. Ly (2012) studies whether XBRL

affects the numbers and dispersion of analyst earnings forecasts and finds evidence

supporting an increase in the numbers of estimatesand a decrease in the dispersion of their

earnings forecasts followingXBRL adoption. Liu and O’Farrell (2013) studies firms from six

nations (Belgium, Italy, Japan, Singapore, Spain and South Korea) that completed

mandatory adoption of XBRL by 2009 and finds that XBRL adoption, in general, has a

positive impact on forecast accuracy, but differing accounting values (professionalism,

uniformity, conservatismand secrecy) among the studied nations moderate XBRL’s impact.

In particular, nations with high professionalism values may experience greater resistance

against regulatory mandate for XBRL adoption. Liu et al. (2014) uses a sample of Phase I

and Phase II US filers in 2009 and 2010 and finds a significant positive associationbetween

mandatory XBRL adoption and both analyst coverage and earnings forecast accuracy;

Analysts’

forecasts

timeliness

153