FFT-method expert validity
Recognising unfair loan advice
Why is it difficult to provide decision support for problems of uncertainty?
How to construct a decision tree for a consumer problem - the FFT method of expert feature validity
A. What do you need?
B. How do you proceed?
- Aikman, D., Galesic, M., Gigerenzer, G., Kapadia, S., Katsikopoulos, K. V., Kothiyal, A., ... & Neumann, T. (2014). Taking uncertainty seriously: Simplicity versus complexity in financial regulation. Bank of England Financial Stability Paper, 28.
- Green, L., & Mehr, D. R. (1997). What alters physicians' decisions to admit to the coronary care unit?. Journal of Family Practice, 45(3), 219–226.
- Jablonskis, E., & Czienskowski, U. (2017). Decision trees online. http://www.adaptivetoolbox.net/Library/Trees/TreesHome#/
- Jenny, M. A., Pachur, T., Williams, S. L., Becker, E., & Margraf, J. (2013). Simple rules for detecting depression. Journal of Applied Research in Memory and Cognition, 2(3), 149–157.
- Luan, S., Schooler, L. J., & Gigerenzer, G. (2011). A signal-detection analysis of fast-and-frugal trees. Psychological Review, 118(2), 316.
- Martignon, L., Katsikopoulos, K. V., & Woike, J. K. (2008). Categorization with limited resources: A family of simple heuristics. Journal of Mathematical Psychology, 52(6), 352–361.
Taking out a loan should be well considered. After all, you make a financial commitment that will bind you for several years, if not decades. It makes sense to get support in the form of a loan advisor for such a serious decision. But even if the title "loan advisor" may suggest a certain independence - unfortunately often the opposite is the case. As an intermediary of a loan, advisors have an interest in profit, since they receive commissions, for example. So you have to check for yourself to what extent a loan consultation is actually fair: Does the consultation enable you to weigh up possible advantages and disadvantages? Our decision tree as a digital checklist helps you to recognise whether you are receiving fair advice.
When do I need this graphic?
This decision tree should support you when it comes to a personal consultation with a person sitting in front of you. It is not about online consultations or telephone consultations. The decision tree helps when it comes to consumer loans (in the range of 5,000 - 20,000 euros), for example, for a used car. It is not suitable for real estate financing or small loans.
You can use the decision tree after a consultation. You should never make the loan decision at the end of the first consultation. If you check the decision tree before a consultation, you could try to influence the conversation so that an exchange takes place on the important aspects of income and expenses.
The decision tree is particularly relevant for certain groups:
1. Older people, if they are less critical of consultants, because in previous decades a partly false expectation was manifested in the image of the neutral expert.
2 Younger people who act impulsively or prematurely because they think too little about the future, since they were not confronted with the negative consequences of financial actions before.
3. Insecure people who see the offer of consultants as a straw that must be taken.
4. People in financial distress, who are increasingly dependent on credit, and therefore often on loans with worse conditions.
Fairness is based on the concept of fair lending. It was modelled here as the median of the unweighted average of five main criteria (development, discretion, household analysis, payment protection insurance, needs).
You can also check the a consultation you have experienced against other quality-related criteria. Please note, however, that no advice or checklist is ever perfect. With each additional feature that you check, the risk of an incorrect assessment of the information offered increases.
Further features are:
- Loan amount is discussed (e.g. purpose)
- Very short conversations (NEGATIVE NOTE)
- Handing out of consultation forms
- Handing over of budget analyses
Where is the data coming from?
Cases – Which served as a basis?
Protocol data from 91 consultations served as the basis for training and test data, provided by the institut für finanzdienstleistungen e.V. (www.iff-hamburg.de) in agreement with the citizens' movement Finanzwende (www.finanzwende.de)
Target assessment – How were the consultations pre-assessed?
Fairness is based on the concept of fair lending (Reifner et al., 2013; Ulbricht et al., 2019). It was modelled here as the median of the unweighted average of five main criteria. These main criteria are development, discretion, budget analysis, payment protection insurance and needs.
Potential features – Which features were considered?
66 features were specified in the data set; in light of the small number of consultation cases, only the strongest relationships were considered.
Selection of features and modelling
The aim of the pre-selection of features was to limit the number of potential features for the prediction model. The feature selection was performed under two aspects: Testability through laypersons and statistical significance. Twelve features were used for modeling.
What is the quality of the data?
The data set was randomly divided into training data sets (two thirds) and test data sets (one third).
The model is of the following quality:
A cross validation of the identified decision tree resulted in the following quality measures: balanced accuracy = 0.85; correct detection of predominantly unfair consultations (share of 45% in the test set) with 0.99. This means that the decision tree detects 99 out of 100 of such unfair consultations..
The correct confirmation of fair consultations is 71 out of 100.
Potential for development
Empirical evaluation with consumers
• Phillips, N. D., Neth, H., Woike, J. K., & Gaissmaier, W. (2017). FFTrees: A toolbox to create, visualize, and evaluate fast-and-frugal decision trees. Judgment and Decision making, 12(4), 344-368.
• Reifner, U., Klinger, H., Knobloch, M., & Tiffe, A. (2013). „Fairness und Verantwortung im Konsumentenkredit – ein Bewertungsprojekt“, institut für finanzdienstleistungen e.V., https://www.iff-hamburg.de/wp-content/uploads/2013/12/Bericht_Fairness_20131118_FO1UR.pdf. Letzter Zugriff 25.11.2019.
• Ulbricht, D., Feigl, M., Freistedt, U., Peters, S., & Schacht, G. (2019). Faire Kreditvergabe - Schlussbericht. institut für finanzdienstleistungen e.V., https://www.iff-hamburg.de/wp-content/uploads/2019/02/Faire_Kreditvergabe_Layout.pdf. Letzter Zugriff 25.11.2019.
Last update: 27 November 2019.