The Impact of Mortgage Broker Use on Borrower Confusion and Preferences

Standard mixed logit estimates show participants prefer: major lenders; lower interest rates; variable rates; lower fees; principal and interest repayments and the ability to make extra payments.

Home loans are important; the market is complex.

Many alternatives:
-Aust: 4000+ loans from 140 lenders -US: 4300 financial institutions originate many types of loans Many attributes: -Variable rates, fixed rates, interest only, principal and interest, equity redraw, ancillary features etc.

Borrowers go to mortgage brokers. Do they help?
-Brokers recommend more expensive, high fee products -Linked to higher default rates (Alexander et al. 2002;Elul et al. 2010, Jiang et al. 2014) -New mortgages from brokers: 65% Australia; 71% UK; 33% in US.

Brokers can help with:
-Search -Process -Marginal borrowers -Education (maybe) Survey 2: Collect absolute ratings of attribute confusion/ importance, and conduct choice task for attribute valuation.
-Borrowers assign less importance to relatively more confusing mortgage attributes => mis-perceptions?
-Broker use causes higher attribute confusion => less client education -Broker-users are willing to pay more for attributes that raise the duration of loans, and therefore broker commissions Findings: Consulting brokers impacts borrowers' perceptions and valuations of attributes.

Survey 1: Mortgage attribute relative confusion and importance
Page 10 Subjects ranked mortgage attributes by confusion and importance.
The most confusing attributes are rated as less important.1. Objective knowledge of attributes is: • Higher after taking out a mortgage; • Lower for broker-users.

Subjective confusion is:
• Lower after taking a mortgage; • Higher for borrowers who consulted brokers; • Inversely related to the relative importance assigned to the attribute.
Results: Brokers educate clients less than loan officers.

Controlling for selection into broker use by IV confirms a causal relation.
Task 2: Which mortgage would you be most (least) likely to choose?
Standard mixed logit estimates show participants prefer: major lenders; lower interest rates; variable rates; lower fees; principal and interest repayments and the ability to make extra payments.
1. Estimate mixed logit model estimate the willingness-to-pay (WTP) for each attribute at the individual-attribute level 2. Restrict the marginal utility of the interest rate to a constant (1) 3. Estimates show WTP in interest rates points (relevant 'price') 4. Measure impact of broker use, confusion, and financial literacy on preferences 5. Method is causal mediation analysis with single IV.
We estimate individual-level attribute preferences in terms of interest rate points from DCE data.
-Three stage estimation: 1.Total effect of broker use on attribute WTP (2SLS) Broker use

Attribute confusion
Attribute WTP Instrumental variable for broker selection is prevalence of broker users in participants' postcode.
Use prevalence from separate survey (n=1600).For the $ amount impact, we assume a 20-year $500,000 mortgage with a fixed rate of 3.5% p.a.
Page 21 Brokers impact preferences directly and via confusion.
-Three stage estimation: 1. Total effect of broker use 2. Effect of broker use on subjective attribute confusion (2SLS) Broker use

Attribute confusion
Attribute WTP Brokers impact preferences directly and via confusion.
-Three stage estimation: 1. Total effect of broker use on attribute WTP 2. Effect of broker use on subjective confusion 3. Indirect effect of broker use on attribute willingness to pay (2SLS) Broker use

Attribute confusion
Attribute WTP

Results show:
-More confused are WTP less for longer terms.BUT Brokers steer clients to higher WTP for longer terms.-More confused are comfortable with higher fees.-More confused WTP less for a principal and interest (v.IO) loan -Highly financially literate participants are WTP less for loans from major lenders.
Task 2 findings: Broker-users pay more for attributes that can raise broker fees.
• WTP among broker clients aligns with broker fees: • Directly steer clients to longer loan terms LMI most confusing; interest rates, prepayment, fees most important -Interest rates, repayment flexibility and fees are most important -Borrowers want a 'good deal' -Loan to value ratio is most confusing -LVR is critical factor in mortgage stress, refinancing options, chance of foreclosure -Inverse relation between confusion and importance Task 1: Borrowers place more importance on loan costs and less importance on relatively confusing attributes.Do brokers educate clients about confusing attributes?

1,881 mortgage borrowers (past, current or future intended) 25-65 yrs old Household income >$45K p.a. Fielded 2019 in Australia Task 1 -Confusion Learn about 13 common mortgage attributes Choose most and least confusing/important Compute scaled most-least scores Identify broker effects Task 2 -Valuation Discrete mortgage choices from 4 sets 3 options with 7 attributes Choose most and least appealing mortgage Analysis Mixed logit estimated on DCE responses Survey 1: Measure relative attribute confusion and importance; choice task for estimation of preferences. Online Sample: 1,200 mortgage borrowers (past, current or future intended) 25-65 yrs old Household income >$45K p.a. Fielded 2023 in Australia Task 1 -Confusion Learn about 13 common mortgage attributes Choose rate them as 'not', 'somewhat' or 'very' confusing / important Identify broker effects Task 2 -Valuation Discrete mortgage choices from 8 sets 3 options with 7 attributes Choose most and least appealing mortgage Analysis Mixed logit Estimate in willingness-to-pay space Interest points are numeraire Causal mediation analysis with single IV
-Want a 'good deal' -Low interest rates and fees -Broker does 'shopping'

confused, more satisfied and more confident about future loans =>
• Indirectly, via confusion, lead borrowers away from P&I • Failure to educate • Broker clients are more set up for future engagement of broker • Brokers should act in clients' 'best interests' => hard to evaluate • Address 'conflicted