Peer effects, personal characteristics and asset allocation

Both academic researchers and practitioners know that behavioural biases can drive household financial decisions away from standard finance models, and that people don't always follow the professional advice they receive. Peer effects have been found to influence financial decision making, but it can be hard to know the relative importance of different influences. For example, does an individual listen more to their friend or their colleague? And are these social connections more important than personal characteristics and financial advice in an individuals' financial decision making?

My colleagues and I have been trying to shed some light on these questions by considering the influence of household, workplace and neighbourhood peer effects on an individuals' asset allocation decisions. We also compared the importance of peer effects to personal characteristics (age, gender, wealth and income) and financial advice. Our results reveal the most important factors (in order) are household peer effects, personal characteristics and workplace peer effects. Neighbourhood peer effects and financial advice play a less important role.

The study used a large sample of 42,187 individual KiwiSaver accounts, the investors lived in 28,380 households in 462 different neighbourhoods and worked in 14,392 unique workplaces. Of these, almost 7000 received financial advice. As one of the very few OECD countries that have negative average household saving rates for the 15-year span from 2001 to 2015, the findings can help policymakers to understand the important role of behavioural biases. In this way, scarce resources can be allocated more efficiently to support New Zealanders to better prepare for retirement.

Individuals' asset allocations were measured by their percentage KiwiSaver investments in the four major asset classes: cash, bonds, property and shares. Household peers were identified by matching identical residential street address. Co-worker peer groups are generated by matching workplace names and locations. Neighbourhoods were generated by grouping individuals by their postal code.

If your colleague holds more of an asset class, you probably will too

Different peer effects were studied separately first. We found a 1 percent increase in equity holdings by household peers would result in the individual investor holding 0.34 percent more equity than they would otherwise. The equivalent percentages for cash, bonds, and property ranged from 0.22 percent to 0.37 percent.

For workplace peer effect, a 1 percent increase in equity holdings invested by a co-worker increased the amount of equity assets held by an individual by 0.39 percent. The cash, bonds, and property percentages ranged from 0.32 percent to 0.43 percent.

In the above two scenarios, we assume the individual investors are only affected by their household members or co-workers. We then considered a scenario where an individual investor is influenced by his household and workplace peers, as well as his neighbours at the same time. And we found, on average, a 1 percent increase in a specific asset class holding by household peers will result in a 0.29 percent increase of the individual's holding in the same asset class, the corresponding changes for workplace and neighbourhood peer effects were 0.291 percent and 0.16 percent respectively.

All the above analysis has controlled for investors' personal characteristics and financial advice. Our results on these factors are largely consistent with the previous findings. Older and female investors hold more cash and bonds and less property and shares, and wealthier investors invest more in property and shares. Individuals receiving financial advice tend to hold more property and shares.

What's more important? Personality or peers?

We then compared the statistical relative importance of all these factors. We are able to explain 25 percent of the variations in individuals' asset allocation decisions by combining household and workplace effects with personal characteristics. This increases to 26.8 percent if we include neighbourhood peer effects and financial advice. Individually, the most important factors are household peer effects and personal characteristics, which explain 15.5 percent and 14.3 percent of variations respectively, with workplace peer effects explaining 5.1 percent of the variations.

We also found personal characteristics are more important when investors are younger, female or receive no financial advice and household peer effects are more important when investors are younger or older. It also seems workplace peer effects are stronger when individuals just start working and choose default funds pre-selected by their employers, and financial advice plays a more important role for older, wealthier and male investors.

The influence of household and workplace peer effects decreases when an individual lives in a larger household, or works in a place with more co-workers. The same holds true for neighbourhood peer effects but, interestingly, all peer effects influence male investors more than female investors.

Overall, this research contributes, in a general sense, to the "nature versus nurture" discussion in financial literature. It clearly shows the significance of nurture factors and the role that peer groups play in financial decision-making, highlighting the crucial role that financial education can play.