is a full Pro­fes­sor of account­ing and con­trol­ling and the direc­tor of the Cen­ter of Account­ing Research at the Uni­ver­si­ty of Graz, Aus­tria. He holds a doc­tor­ate from the Uni­ver­si­ty of Tech­nol­o­gy, Vien­na, and an hon­orary doc­tor­ate from the Uni­ver­si­ty of Munich. His main research deals with finan­cial report­ing, man­age­ment account­ing, and cor­po­rate gov­er­nance.

Finan­cial Report­ing

Big data and financial reporting

By Alfred Wagen­hofer

Big data per­me­ates almost all areas of busi­ness. Finan­cial report­ing is no excep­tion. I dis­cuss select­ed aca­d­e­m­ic research that makes use of nov­el tech­nolo­gies and data.

Being cog­nizant what users can do – and actu­al­ly do – is use­ful when design­ing the con­tents of finan­cial report­ing. New infor­ma­tion tech­nolo­gies facil­i­tate gen­er­at­ing and access­ing a host of new data and pro­vide sophis­ti­cat­ed tools to ana­lyze and com­bine such data to find new asso­ci­a­tions and pat­terns. For users of finan­cial report­ing oppor­tu­ni­ties abound: investors use big data to make supe­ri­or invest­ment deci­sions, ana­lysts to improve fore­casts, audi­tors to assess audit risk, enforce­ment agen­cies to iden­ti­fy red flags, and researchers to find nov­el results. Here are two areas that recent account-
ing research has addressed.

Com­pa­ny per­for­mance tends to affect read­abil­i­ty

It seems intu­itive that the read­abil­i­ty of a finan­cial report has to do with the con­tent it pur­ports. One might sus­pect that read­abil­i­ty is low­er if a com­pa­ny faces poor per­for­mance because it tries to obfus­cate unfa­vor­able finan­cial infor­ma­tion. Alter­na­tive­ly, com­pa­nies that expe­ri­ence bad results may feel pres­sure to bet­ter explain that per­for­mance. Psy­cho­log­i­cal fac­tors, such as attri­bu­tion – jus­ti­fy­ing low per­for­mance by events out­side management’s con­trol –, can play a role. And impres­sion man­age­ment may be at work.

Sim­ple mea­sures of read­abil­i­ty are the length of the report or the por­tion of boil­er­plate dis­clo­sures, which adds clut­ter, or asym­met­ric report­ing about risks and chances. Lin­guis­tics has devel­oped sev­er­al mea­sures of read­abil­i­ty. A well-known mea­sure is the fog index that counts words per sen­tence and the per­cent­age of com­plex words. More advanced tools mea­sure cau­sa­tion, exclu­sive­ness, tone, and emo­tion through defined dic­tio­nar­ies.

Among oth­ers, research finds the fol­low­ing asso­ci­a­tions:

– Finan­cial reports of com­pa­nies with low earn­ings are hard­er to read, and com­pa­nies with easy-to-read reports have more per­sis­tent pos­i­tive earn­ings
(Li, 2008).

– Declin­ing per­for­mance and non-ver­i­fi­able infor­ma­tion, such as fore­casts, come with a more pos­i­tive and opti­mistic tone. Bet­ter cor­po­rate gov­er­nance reduces these asso­ci­a­tions (Mel­loni et al., 2016).

– Liq­uid­i­ty, insti­tu­tion­al own­er­ship, and ana­lyst fol­low­ing are high­er for com­pa­nies with bet­ter read­able finan­cial reports (Lang and Stice-Lawrence, 2015).

– Com­pa­nies with hard-to-read, com­plex finan­cial reports pro­vide more vol­un­tary dis­clo­sures to “guide through the fog” (Guay et al., 2016).

Per­son­al traits of exec­u­tives influ­ence report­ing qual­i­ty

Bor­row­ing from psy­chol­o­gy, one notion is that nar­cis­sis­tic exec­u­tives are more will­ing to exploit their pow­er and infor­ma­tion asym­me­try to engage in mis­re­port­ing. Anoth­er one is that over­con­fi­dent man­agers over­es­ti­mate their abil­i­ties and skills and assign an inflat­ed sub­jec­tive prob­a­bil­i­ty to future out­comes. Stud­ies find inno­v­a­tive ways to mea­sure per­son­al traits.

– A study uses the promi­nence of the pho­to­graph of the CEO and his or her cash pay rel­a­tive to sec­ond-high­est paid exec­u­tive and finds more nar­cis­sis­tic CEOs tend to over­state report­ed EPS (Olsen et al., 2014).

– Using the size of the CFO’s sig­na­ture in the annu­al report, a study finds that more nar­cis­sis­tic CFOs are asso­ci­at­ed with more earn­ings man­age­ment, less time­ly loss recog­ni­tion, weak­er inter­nal con­trols, and a high­er prob­a­bil­i­ty of restate­ments (Ham et al., 2015).

– Mea­sur­ing CEO over­con­fi­dence by their will­ing­ness to hold deep-in-the-mon­ey options close to matu­ri­ty, a study finds that over­con­fi­dence is pos­i­tive­ly relat­ed with earn­ings man­age­ment and bench­mark-beat­ing (Hsieh et al., 2014).

– Not exact­ly a trait, but nev­er­the­less excit­ing, a study uses CFOs’ golf­ing records as proxy for leisure con­sump­tion and finds that finan­cial reports of com­pa­nies that have CFOs with high­er golf­ing ranks are of low­er qual­i­ty (Big­ger­staff et al., 2016).

Big data as the new nor­mal

Not all stud­ies make a lot of sense. Often do they lack the­o­ret­i­cal back­ing, use ques­tion­able prox­ies, and fail to estab­lish causal­i­ty. They are typ­i­cal­ly large-sam­ple stud­ies with low explana­to­ry pow­er. Yet big data make it dra­mat­i­cal­ly eas­i­er to explore nov­el asso­ci­a­tions. Pre­par­ers of finan­cial reports should be sen­si­tized of uncon­ven­tion­al uses (and mis­us­es) of cor­po­rate finan­cial infor­ma­tion.