Research Proposal – Financial Literacy and Mortgage Decision Making Among First Home Buyers

Introduction and Background

Entry into home ownership is the largest financial decision most Australian households will make, and it is being made under increasingly difficult conditions. Capital city dwelling prices have grown faster than household incomes for more than two decades, lengthening the period required to accumulate a deposit and pushing first home buyers into higher loan-to-valuation ratios and longer loan terms. Australian Bureau of Statistics lending data show that the average loan size taken by this group has risen substantially over the past five years, leaving new entrants with larger debts relative to income than any previous cohort (ABS 2025).

The interest rate environment has magnified the stakes. After the tightening cycle that began in 2022, repayments on a typical variable rate loan rose sharply, and the Reserve Bank of Australia has repeatedly identified recent high debt-to-income borrowers with limited savings buffers as the group most exposed to repayment difficulty (RBA 2024). The Australian Prudential Regulation Authority requires authorised deposit-taking institutions to assess serviceability at least three percentage points above the product rate, a control designed to absorb exactly this risk (APRA 2024). Yet a buffer applied by a lender does not equip a borrower to evaluate the product they are signing.

That evaluation task is genuinely difficult. A prospective borrower must weigh variable, fixed and split rate structures, introductory discounts, comparison rates, offset and redraw facilities, package fees and lenders mortgage insurance, all against an uncertain future income path. ASIC found that many borrowers considered only a few products and relied heavily on a single information source before committing (ASIC 2019). The proposed study asks how well first home buyers understand these trade-offs, and what happens to their choices when that understanding is incomplete.

Problem Statement

Three gaps motivate this research. First, financial literacy in Australia is usually measured with general instruments covering compound interest, inflation and risk diversification, which predict broad financial outcomes but say little about whether a borrower can compare a two-year discounted variable loan against a flat-rate alternative. Domain-specific mortgage literacy has not been systematically measured in an Australian first home buyer population.

Second, the behavioural literature demonstrates that even numerate consumers make systematic errors in mortgage markets, particularly where introductory pricing exploits present bias and where advertised rates operate as anchors (Agarwal, Ben-David & Yao 2017). Whether these effects survive in the Australian setting, where the comparison rate is a mandated disclosure, is untested.

Third, roughly seven in ten new residential loans in Australia are now originated through a mortgage broker. ASIC has documented that broker-originated loans differed systematically from lender-direct loans on leverage and interest-only structure (ASIC 2017), and the best interests duty introduced in 2021 was intended to address the resulting conflicts. No Australian study has tested whether the broker channel compensates for low borrower literacy or instead amplifies its consequences. The responsible lending obligations under the National Consumer Credit Protection Act 2009 (Cth) place the assessment burden on the credit provider, but they do not determine which of several suitable products a borrower selects; product choice, and therefore lifetime cost, remains with the consumer.

Aim and Research Questions

The aim of the proposed study is to determine how mortgage-specific financial literacy and behavioural biases jointly shape the loan feature choices of recent Australian first home buyers, and whether the channel through which the loan is acquired moderates that relationship. Three research questions follow:

1. RQ1: How does mortgage-specific financial literacy vary across recent Australian first home buyers, and which household characteristics predict it?
2. RQ2: To what extent do present bias and anchoring influence trade-offs between introductory discounts, rate structure and total cost of credit, after controlling for literacy and household constraints?
3. RQ3: Does the acquisition channel, broker or direct lender, moderate the relationship between literacy, bias and the loan features ultimately selected?

Literature Review

Measuring financial literacy

The dominant measurement tradition derives from Lusardi and Mitchell (2014), whose three-item instrument established that financial illiteracy is widespread and concentrated among younger, lower-income and female respondents. Longitudinal Australian evidence from the Household, Income and Labour Dynamics in Australia (HILDA) survey replicates this pattern and links lower literacy to weaker wealth accumulation (Wilkins, Vera-Toscano & Botha 2024). The instrument’s brevity is also its limitation: performance on generic numeracy items correlates only loosely with the applied competence required to compare credit products, so a domain-specific measure is required.

Behavioural biases in credit choice

Present bias, formalised by O’Donoghue and Rabin (1999), describes the systematic overweighting of immediate payoffs relative to deferred costs. In mortgage markets it predicts that borrowers will favour a visible short-term saving, such as a two-year discounted rate, over a lower cost across the full loan term. Anchoring, first demonstrated by Tversky and Kahneman (1974), predicts that the first rate encountered will constrain subsequent judgement even when it is uninformative, and Agarwal, Ben-David and Yao (2017) show that less sophisticated borrowers make measurable and costly errors in exactly this way. Australia provides a useful test case because the mandated comparison rate is intended to counteract anchoring, yet ASIC (2019) reports that borrowers rarely used it as the basis for comparison.

The broker channel and responsible lending

Brokers may reduce search costs and improve matching, or may introduce conflicts through commission structures that reward larger and longer loans. ASIC (2017) found higher average leverage among broker-originated loans, a finding that prompted the best interests duty. This literature establishes the channel as a plausible moderator rather than a mere control variable, yet no Australian study has modelled it as such alongside measured literacy and elicited bias.

Conceptual Framework

Figure 1 presents the proposed conceptual model. Mortgage-specific financial literacy, behavioural biases and household constraints are treated as antecedents that operate on search and comparison behaviour, which in turn shapes the loan features selected and, ultimately, exposure to mortgage stress. The acquisition channel is modelled as a moderator of the path from search behaviour to feature choice, on the reasoning that a broker substitutes for a borrower’s own search effort. The regulatory context described above constrains the choice set without determining the selection made within it.

Methodology

Design and sampling

The study will adopt an explanatory sequential mixed methods design in which a quantitative survey is followed by qualitative interviews that interpret the statistical results. The survey will recruit approximately 500 Australians who settled a first home loan within the preceding 24 months, drawn from an accredited online research panel with quotas on state, age band and household income so that the achieved sample approximates the national first home buyer profile reported by the ABS (2025). Eligibility will be confirmed by screening on purchase date, prior ownership and loan status.

Discrete choice experiment

A discrete choice experiment will be embedded in the survey to observe trade-offs rather than rely on stated preference alone. Six attributes will be varied: rate structure, introductory discount size and duration, comparison rate, offset facility, annual package fee and loan term. Each respondent will complete twelve choice tasks presenting three hypothetical loans plus an opt-out, generated from a Bayesian D-efficient design (Hensher, Rose & Greene 2015). Anchoring will be manipulated by randomising the headline rate shown before the tasks; present bias will be elicited separately through an incentivised multiple price list over payments at one and twelve months. The minimum sample for a design of this size follows the conventional rule of thumb:

• n >= 500c / (t x a), where c is the largest number of levels on any attribute, t is the number of choice tasks and a is the number of alternatives per task
• n >= (500 x 4) / (12 x 3) = 2,000 / 36 = 55.6, therefore a minimum of 56 respondents

The proposed sample of 500 far exceeds this minimum, a surplus that supports latent class estimation and subgroup comparison by acquisition channel.

Illustrative decision problem

The significance of present bias here can be shown with a worked example on a loan of $650,000. Product A offers 5.74 per cent for 24 months and then reverts to 6.49 per cent; Product B offers a flat 6.09 per cent.

• Introductory saving = (6.09% – 5.74%) x $650,000 x 2 years = 0.0035 x 650,000 x 2 = $4,550
• Annual cost after reversion = (6.49% – 6.09%) x $650,000 = 0.0040 x 650,000 = $2,600
• Break-even = $4,550 / $2,600 = 1.75 years after reversion

A borrower who does not refinance is therefore worse off from roughly the fourth year of the loan, a product risk that the APRA serviceability buffer, which tests capacity to repay rather than the wisdom of the product chosen, does not address.

Qualitative phase and analysis plan

Twelve semi-structured interviews of approximately 60 minutes will be conducted with survey respondents purposively selected for maximum variation on literacy score and acquisition channel. Interviews will explore how decisions were reached, what information was consulted and how advice was weighed, and will be analysed using reflexive thematic analysis (Braun & Clarke 2021). Quantitatively, RQ1 will be addressed through regression of the literacy index on household characteristics; RQ2 through mixed logit models of the choice data with interaction terms for elicited bias, reported as willingness to pay in dollars; and RQ3 through a channel interaction term supported by multi-group comparison. Integration will occur through a joint display aligning each research question with its quantitative result and qualitative explanation.

Ethical Considerations

The project will not commence before approval by the university Human Research Ethics Committee and will be conducted in accordance with the National Statement on Ethical Conduct in Human Research (NHMRC 2023). Financial information is sensitive and the topic may distress participants experiencing repayment difficulty, so the survey will avoid account identifiers, permit item non-response throughout, and conclude by signposting the ASIC MoneySmart resources and the National Debt Helpline. Participants will be told explicitly that the research provides no personal financial advice and that the hypothetical products are not offers of credit. Consent will be obtained separately for the survey and any interview, recordings transcribed with identifiers removed, and data stored on encrypted university infrastructure for five years. Reimbursement will be modest so that it is not coercive.

Constructs and Measures

Table 1 summarises the constructs, their measurement approach and the research question each serves.

Table 1: Constructs, operational definitions and measurement approach

Project Timeline

Table 2 sets out the twelve-month schedule by phase, activity and week; phases overlap deliberately, with instrument development proceeding while ethics approval is pending and interviews beginning before quantitative analysis concludes.

Table 2: Twelve-month project timeline by phase, activity and week

Significance and Expected Contribution

The study offers three contributions. Theoretically, it integrates domain-specific literacy measurement with elicited behavioural parameters in a single model, moving beyond the common practice of treating literacy and bias as competing explanations. Methodologically, pairing a discrete choice experiment with a randomised anchor manipulation allows the effect of disclosure to be observed rather than inferred.

Practically, the findings speak to live Australian policy questions. If borrowers with adequate general literacy still misprice introductory discounts, the case strengthens for reforming how comparison rates are presented at the point of decision rather than for expanding generic education. If the broker channel compensates for low literacy, that supports the best interests duty examined by ASIC (2017); if it amplifies bias, the remuneration question reopens. Estimated willingness-to-pay figures would also give ASIC MoneySmart and school-based capability programs quantified examples of the cost of common errors.

Limitations

Four limitations are acknowledged in advance. Hypothetical bias is inherent to stated preference methods; the design mitigates this through realistic attribute levels, an opt-out alternative and validation against the loan each respondent actually holds. Recall error is possible across a 24-month window, so key loan details will be verified against documentation where participants consent. Panel recruitment excludes aspiring buyers unable to purchase, so the results describe successful entrants rather than the affordability barrier itself. Finally, the cross-sectional design means associations between literacy, bias and stress cannot be interpreted causally; a follow-up wave at 24 months is proposed as future work.

Conclusion

Australian first home buyers are entering the market with historically large debts, under prudential settings that protect lenders against default but not borrowers against choosing an expensive product. This proposal sets out a mixed methods design capable of separating what borrowers do not know from what they know but discount, and of testing whether the broker channel repairs or reinforces that gap. The expected outputs, a validated mortgage literacy measure, willingness-to-pay estimates for common loan features and an evidence-based account of the broker moderation effect, are directly usable by Australian regulators, lenders and financial capability educators working on housing affordability.

References

Agarwal, S, Ben-David, I & Yao, V 2017, ‘Systematic mistakes in the mortgage market and lack of financial sophistication’, Journal of Financial Economics, vol. 123, no. 1, pp. 42-58.

Australian Bureau of Statistics 2025, Lending indicators, ABS, Canberra.

Australian Prudential Regulation Authority 2024, Quarterly authorised deposit-taking institution property exposures, APRA, Sydney.

Australian Securities and Investments Commission 2017, Review of mortgage broker remuneration, Report 516, ASIC, Sydney.

Australian Securities and Investments Commission 2019, Looking for a mortgage: consumer experiences and behaviour in home loan lending, Report 628, ASIC, Sydney.

Braun, V & Clarke, V 2021, Thematic analysis: a practical guide, Sage, London.

Hensher, DA, Rose, JM & Greene, WH 2015, Applied choice analysis, 2nd edn, Cambridge University Press, Cambridge.

Lusardi, A & Mitchell, OS 2014, ‘The economic importance of financial literacy: theory and evidence’, Journal of Economic Literature, vol. 52, no. 1, pp. 5-44.

National Consumer Credit Protection Act 2009 (Cth), Commonwealth of Australia, Canberra.

National Health and Medical Research Council 2023, National statement on ethical conduct in human research, NHMRC, Canberra.

O’Donoghue, T & Rabin, M 1999, ‘Doing it now or later’, American Economic Review, vol. 89, no. 1, pp. 103-124.

Reserve Bank of Australia 2024, Financial stability review, RBA, Sydney.

Tversky, A & Kahneman, D 1974, ‘Judgment under uncertainty: heuristics and biases’, Science, vol. 185, no. 4157, pp. 1124-1131.

Wilkins, R, Vera-Toscano, E & Botha, F 2024, The Household, Income and Labour Dynamics in Australia survey: selected findings from waves 1 to 22, Melbourne Institute of Applied Economic and Social Research, University of Melbourne.