Conflicted financial advice: disclosure revisited

ABSTRACT

We examine the impact of disclosing an advisor’s conflict of interest in providing financial advice to a client in an experiment. We find that an advisor’s conflict of interest harms the client and that disclosing the conflict harms the advisor. Unlike earlier literature, we do not find that disclosure of the advisor’s conflict of interest results in moral licensing or strategic exaggeration behaviour by the advisor nor, relatedly, that disclosure disadvantages the client.

KEYWORDS:

• Conflict of interest
• disclosure
• financial advice
• experiment

JEL CLASSIFICATION:

• G28
• D14
• D18

Acknowledgement

We thank, without implicating, Nawaaz Khalfan for research assistance, Catherine de Fontenay for comments on earlier work and seminar and workshop participants at Monash University, NZAE 2016, ACE 2016, EEA-ESEM 2016 and the Australian Law & Economics 2016 conferences.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ See Chen and Richardson (2016) for a fuller discussion of the literature, our experiment and results.

² With this ordering, the impact of conflict may be biased by possible temporal effects such as learning or fatigue separate from conflict. We do not observe, however, any temporal differences across rounds within any treatment. Furthermore, any possible bias is not a severe problem for our purposes since our primary focus is on comparing disclosure versus no disclosure given conflict of interest rather than conflict versus no conflict whether with or without disclosure.

³ Taken from §961B(1) of the Corporations Act 2001 in Australia. The statement is similar to business principles advertised by certain financial firms; e.g. Goldman Sachs’ 14 principles begin with ‘Our clients’ interests always come first’ and end with ‘Integrity and honesty are at the heart of our business’. See http://www.goldmansachs.com/who-we-are/business-standards/business-principles/. Accessed on 17 November 2016.

⁴ Not all signed-up participants appeared on the day, resulting in more clients than advisors, so some advisor responses were re-used for a second client. In the ‘No Disclosure’ treatment, we have 34 unique A/C pairs (26 advisors and 34 clients); in the ‘Disclosure’ treatment, 31 unique A/C pairs (28 advisors and 31 clients.)

⁵ Fixed effects estimation would not allow estimating the disclosure impact β₂ in Equation 1 which is of primary interest. Since disclosure_i is fixed for any A/C pair i and invariant across j, this results in singularity with the unobserved effect for i fixed across j.

Additional information

Funding

This work was supported by the College of Business and Economics at the Australian National University under Grant [RSG 2014-15]. The experiment conducted for this study received research ethics approval from our university.