VoxEU Column Frontiers of economic research

Distorted financial advice

Most households lack the skills to make financial decisions and therefore rely on the advice of experts. But do experts provide unbiased recommendations or do they distort their advice to serve their own needs? This column provides evidence that they do distort their advice, using the mortgage market in Italy. If households receive no biased advice, the relative cost of different types of mortgages should be the only variable affecting household choices. But the findings show that characteristics of the lender matter over and above relative prices. This supports the view that banks affect household choices not only through a price channel, but also through an advice channel.

What financial advice do experts provide?

When it comes to financial products, households have limited ability to make a choice and they must rely on the advice of experts, and the more complex the decision, the more this is the case. Often, the expert they rely on for advice is also the supplier of the financial product, so that a conflict of interest may arise and with it the question: Does the expert provide recommendations in the customer’s interest or in his own?

Casual evidence and personal experiences suggest that banks and intermediaries tend to distort their customer choices towards high-fee products or products they have on their shelves and want to get rid of. For instance, one of the authors of this column once asked his own bank for advice on how to invest a sum of money. He was recommended to put all in a bond issued by Cirio – a company that soon after would have gone bankrupt. Cirio was exposed to the bank and the bank was also Cirio’s advisor in the bond placement. Later it was discovered that the money from the placement was meant to repay first the loan to the bank. No doubt, the advice was meant to benefit the bank not the customer.

Needless to say, personal experiences, while suggestive, are hard to generalise. They may just reflect the rotten apple that is in every barrel (the reflexive defence whenever misconduct surfaces in the midst of some organisation or industry – such as the subprime mortgage market or the car industry after the Volkswagen scandal). Yet, systematic evidence is difficult to pile up.

One approach that has been followed is to compare the performance of advised and non-advised accounts. Applying this approach to German banks, Hackethal et al. (2012) find that advised accounts typically perform worse than non-advised accounts, which is consistent with large-scale, systematically distorted advice. But this evidence is also consistent with a less cynical interpretation. The customers who choose to be advised are likely to be those with lower financial skills. Advice can help bridge the gap but cannot make miracles. A bit like doctors, they do provide care to sick patients, but even after treatment the patient’s health, not surprisingly, will not be as good as that of a healthy person. This is not because the treatment was not good, but because doctors’ patients have, on average, more precarious health.

A second approach tries to circumvent this problem using field experiments. For instance, Mullainathan et al. (2009) sent trained auditors posing as customers to intermediaries to get advice. They document that customers’ biases are augmented by professional advice, indicating potential suppliers’ distortions in households’ financial decisions. In a similar study of the Indian life insurance market, Anagol et al. (2012) find that agents motivated by commission recommend strictly dominated, expensive products between 60% and 90% of the time. The main problem with these experiments is that one cannot be sure that the audited advisors would offer the same biased advice in the kind of long-term, repeated client relationships that one finds in the real world. Customers of banks will return to the bank’s premises over and over and one may argue that this is enough to discourage abuses. But does it?

Studying financial advice: New approach

To complement and extend the above approaches and overcome some of their deficiencies, we propose a data-based methodology (Foa et al. 2015). Different from the previous approaches, this methodology does not need to observe whether advice is sought or not to establish that it is present and is distorted. This is important because intermediaries may advise customers even when the latter do not solicit any advice. Our idea is that if households are wary, the relative price of different financial products should be a sufficient statistic for their choice. This is the only aspect of the supplier that should matter. To wit, once I have chosen to go to a certain grocery, to decide how many apples and how many peaches of a given type I want to buy all that I need to know is the price of the two goods posted on the shelf. Other features of the grocery – for example, whether there is still room in the grocery refrigerators – should not matter. But if the observed choice of the customer, once the price of the two goods has been taken into account, depends systematically on the room available in the grocery refrigerator, then there is a chance that the grocery owner is advising his customers and distorting there choices; in particular, if customers tend to buy more peaches than apples at groceries with a lower refrigerator capacity. In fact, peaches are more perishable than apples, and thus groceries have incentives to sell them first – and to advise customers to buy them – when they lack refrigerator space.

We apply this idea to the mortgage market and, in particular, to households’ choice between fixed-rate and adjustable-rate mortgages. Once they have decided they want to obtain a mortgage and have chosen a bank, their choice should depend on the relative cost of the two types of mortgages but not on other characteristics of the bank, such as the cost at which the bank can raise long-term funding. If the latter matters for the household choice, this would be evidence for the presence of biased financial advice. In particular, if households borrowing from a bank that faces a higher cost of raising long-term funding systematically tend to choose adjustable rate mortgages more often than households that borrow from another bank that can obtain long-term funds at a lower cost – even if the relative price of fixed and adjustable mortgages is the same – this could be evidence of biased advice. In other words, if some households are naive and rely on advice, the bank can push them toward the mortgage for which it has a cost advantage, thus retaining a competitive advantage as the higher costs are not passed through to prices. The price is not a sufficient statistics for the household choice which depends directly on the bank identity.

To detect distorted financial advice, we use data on a sample of 1.6 million mortgages originated in Italy by 175 banks between 2004 and 2010. Our test detects the presence of biased financial advice in the Italian mortgage market.

A new test for biased financial advice

A first look at household mortgage choices confirms that banks' identities matter a great deal for household choices in our data. Banks’ identity (i.e. bank fixed effects), taken in isolation, explains around 10% of the total variance of household choice in our sample, and remains significant when relative prices of the two types of mortgages and household characteristics are added to the model. Relative prices also matter, consistent with the role assigned to prices by theory. Prices and banks’ identity jointly explain roughly half of the variance in household choices, and the coefficient in front of relative prices is negative and statistically significant.

The importance of fixed effects, though suggestive of our mechanism, does not prove it. The same finding may be generated by sorting. Households with a preference for adjustable-rate mortgages may be self-selecting into some types of banks, so that banks' identity would have predictive power for household choices even in the absence of biased advices. To rule out sorting, we rely on time-varying bank-specific variables, and in particular such that affect the relative cost for a bank of issuing an adjustable versus a fixed mortgage rate. If the results are driven by sorting, changes in the incentive for banks to issue one mortgage rather than the other, should not affect household choices at the same bank, while they should if advice is the driver of the results. As time-varying factors, we choose the bank bond spread (the difference between long- and short-term funding for the bank), access to securitisation, and the share of deposits in total funding. Banks with lower bond spreads, easier access to securitisation, and a high share of deposit funding should find it cheaper to originate fixed-rate mortgages than banks with higher bond premiums, no access to securitisation, and a low share of deposit funding. The latter should be more likely, ceteris paribus, to advise adjustable rate mortgage and their customers to choose them, controlling for the relative cost of adjustable and fixed rate mortgage.

Our results are unambiguous.

  • Time-varying factors matter for household choices over and above prices (they are statistically and economically significant). For example, a 1% increase in the bank bond spread lowers the probability of choosing a fixed mortgage rate by 2.8 percentage point, which is roughly 10% of the effect of prices.

While prices are the main determinant of household choices, the role of distorted advice is non-negligible.

Interestingly, and consistent with the idea that the first victims of distorted financial advice are the least financially literate consumers, the effect of time-varying factors on households choices is stronger for unsophisticated households. Sophisticated households seem to be able to anticipate the bank’s conflict of interest and disregard the advice. Furthermore, we find that banks distort advice particularly when they find it costly to re-price mortgages; that is, biased advice may be tempting as a way to save on price adjustment costs.

Concluding remarks

We use a novel methodology to detect the presence of biased financial advice from banks to households choosing a mortgage, based on the idea that, in case of advice, supply factors matter for choices over and above prices. Relying on our test, biased advice in the Italian mortgage market is sizeable and widespread, and is thus a feature of the working of the industry, not merely the reflection of a rotten apple in an otherwise flawless basket.

Disclaimer: The views expressed in this column are those of the authors and do not necessarily reflect those of their respective institutions


Anagol, S, S Cole and S Sarkar (2012), “Understanding the Advice of Commissions-Motivated Agents: Evidence from the Indian Life Insurance Market”, Harvard Business School Working Paper, No. 12-055.

Foà, G, L Gambacorta, L Guiso and P E Mistrulli (2015), “The Supply Side of Household Finance”, CEPR Discussion Paper Series, 10714.

Hackethal, A, M Haliassos and T Jappelli (2012), “Financial Advisors: A Case of Babysitters?”, Journal of Banking and Finance, 36(2), 509-24.

Mullainathan, S, M Nöth, and A Schoar (2012), “The Market for Financial Advice: An Audit Study”, NBER Working Paper, 17929.

2,310 Reads