Dimitris Georgarakos, Michael Haliassos and Giacomo Pasini
Household Debt and Social Interactions
Review of Financial Studies | Volume 27, Issue 5 (May 2014), 1404-1433

The role of standing relative to peers in influencing household economic behavior has been explored in many contexts, including consumption and labor supply, but less attention has been paid to how “catching up” or “keeping up” with peers (often referred to as “the Joneses”) is financed. In particular, almost no attention has been paid to whether perceptions of relative standing contribute to borrowing and to the potential for financial distress. Are people who perceive themselves as poorer than their social circle more likely to borrow and, if so, to borrow more relative to what is typically associated with their own resources and characteristics? Does such socially induced borrowing contribute to a worsening of indicators of potential financial distress, such as the debt-to-income and loan-to-value ratios?

The question of social influences on debt is distinct from that relating to consumption; concern with relative standing may lead to greater consumption, but it need not lead to a greater tendency to borrow or to run into financial distress for at least three reasons. First, households can increase labor supply, leaving room for an increase in both consumption and saving. Second, households may choose to reduce saving but may not be willing or able to raise borrowing in response to status concerns. Third, even if borrowing is undertaken, it may not significantly increase the potential for financial distress.

The Method

Investigating the influence of social interactions on debt behavior presents at least two major challenges. First, many households are willing to display their assets and consumption but prefer to leave any debts undisclosed. Thus, it makes sense to look for evidence that households adjust their debt behavior not to the debts of their peers per se but to their perception of relevant peer characteristics, like income. In the Manski terminology, instead of “endogenous effects”, one needs to focus on “exogenous (or contextual) effects”. A second challenge has to do with the scarcity of location information in household finance data. In view of privacy laws dictating anonymity of information, data collectors typically remove location details, but this step makes it impossible to identify a circle of “neighbors” or “colleagues.”

Pioneering papers that study the asset side focus on a special population group, a financial asset observed by peers, or on sociability as a factor facilitating the collection of asset-relevant information. Duflo demonstrate that individual participation of librarians in retirement investment plans is influenced by participation choices of colleagues in the same library. Hong show that more sociable individuals are more likely to own stocks.

To the best of our knowledge, our paper is the first to investigate the influence of social interactions and comparison effects on borrowing behavior. We exploit a unique feature of population-wide representative data from the Netherlands (Dutch National Bank Household Survey), namely, that the respondents report various characteristics of their peers, such as income, as these respondents actually perceive them. Thus, we can focus on whom respondents consider as peers and study their perceptions regarding their own financial position relative to them.

The Findings

We find that once we control for demographics, resources, region, time fixed effects, region-specific time trends, and other factors that typically determine borrowing needs, a higher average income in the social circle, as perceived by a household, increases a household's tendency to borrow. The estimated effects are sizeable for both collateralized and consumer debt: a 1,000 euro increase in the perceived monthly average household income of peers (corresponding to 0.85 of one standard deviation of peer income) is estimated to raise the unconditional likelihood of having collateralized (uncollateralized) loans by 10% (7%). Not only is this influence significant among those who perceive their income to be below average for their social circle, it also extends to the likelihood of future financial distress, indicated by the debt service ratio and the loan-to-value ratio. We verify the robustness of these results using several approaches, including instrumental variable estimation and placebo tests. Our aim is to rule out uninteresting alternative explanations of the peer-income-own-borrowing relation and address the potential for reverse causality or spurious correlation between the two. The former could arise, for example, if people who borrow are more likely to think of their peers as earning more than they do. The latter could be induced by similarities in unobserved characteristics with those of peers, which tend to induce both higher peer incomes and a greater tendency to borrow without any direct causal link between the two.

We also find that once the perceived peer income is controlled for, the tendency to have uncollateralize loans is partly related to the perceived spending ability of peers. Expectations about the next period's income are statistically significant but do not render peers' perceived incomes insignificant. This finding suggests that average peer income does not simply reveal prospects for the future income of the respondent (a mere “tunnel effect”) but also represents a comparison effect.

Policy Implications

Although our analysis does not rule out that much of this socially induced additional borrowing is repayable, at least ex ante, our finding that it tends to worsen indicators of potential financial distress suggests that repayment problems might still occur ex post, especially if borrowers and lenders fail to take proper account of all relevant risk factors. This paper provides a powerful additional rationale for promoting debt literacy on the part of households and scrutiny on the part of lenders: a potential for financial distress is not only generated by objective borrowing needs of households but can also be influenced by how they perceive themselves relative to their social circle. Moreover, this happens regardless of their income class and without requiring accuracy of the perceptions of relative standing.