Friends with money

The authors find evidence that personal connections lead to significantly lower interest rates being charged, and that companies with personal connections to their lenders appear to outperform companies that have no personal connection in terms of both improved credit ratings and share price performance. Other studies have explored the impact of personal relationships on business and investment decisions, and some evidence suggests that they may compromise a lending bank’s objectivity and ability to make a good decision. However, the evidence presented contradicts that point, and although the reasons are not clear, the findings are interesting in the context of whether personal connections help to facilitate more efficient capital markets.

Analysis

The authors address two main questions: firstly, whether there is a causal relationship between personally connected borrower and lenders and the lending deals that are made between them (i.e. do they give their ‘friends’ better deals), and secondly, whether these personal connections lead lending banks to make good or bad lending decisions.

Loan information was drawn from a standard market data source (Dealscan) and uses three key variables: deal size, interest rate charged including fees (‘all-in’ spread), and the number of covenants. The sample included syndicated loans made to over 5000 US public companies between 2000 and 2007, by around 2000 commercial banks. The study considers whether the borrowing company has connections to any of the lenders in the syndicate, not only the lead, on the basis that previous studies have shown many banks often syndicate as a group for many deals and the lead position is rotated between them.

Data on senior management was sourced from the data agency BoardEx. Personal connections between 65,000 individuals who had been either board members or executives at borrowing or lending institutions were evaluated using university connections (if they graduated from the same educational institution in the same year or one year removed) and past professional connections (if they worked for the same corporation or were members of the same board at the same time, at least five years previously and only if the connection did not involve the current borrowing or lending company). The time lag is an important control to ensure that the personal connections are distant enough from the loan that the authors can be confident the connections between lenders and borrowers have not come about as a result of the lending relationship, i.e. reverse causality is highly unlikely. For similar reasons connections through participation in social networks, as opposed to professional networks, are excluded, because accurate data on the timing and extent of participation is not widely available.

Results

The statistical analysis finds that the presence of at least one pre-existing personal relationship significantly reduces borrowing costs, and the impact increases with more relationships within the syndicate. The absolute impact is smaller for firms with better credit ratings as might be expected given that their loans costs are smaller, but the downward impact on borrowing costs increases as credit quality declines. At least one connection results in a better deal by about the same magnitude as a shift in credit rating, and on average 1.5 additional connections leads to 24 basis points reduction in loan spread, though the incremental value of each connection reduces with an increase in number of connections.

Personally connected syndicates also seem to lend more than average, are less likely to impose covenants, and where present the covenants are fewer. Loan covenants are considered only in terms of the number, not type (which can be highly variable) and the authors find that doubling the number of personal connections reduces the likelihood of covenants being attached to the loan.

These results are robust when controlling for specific firm, industry, loan and macroeconomic characteristics. The authors also consider in detail other possible explanations and sources of external influencing factors, such as geographical proximity (which could logically affect the number of personal connections), the presence of previous lender relationships, and the size and activity of the lenders, and find that the effect of personal connections is still strong.

To evaluate whether the personal connections are leading banks to make good or bad decisions, Engelberg et al use the development of credit ratings as a proxy for loan performance (since specific loan performance data is not usually publicly available). Subsequent to loans being made, the credit ratings of companies that have personal connections with their lending syndicate appear to improve. This relationship holds for all rating categories, although the magnitude varies, and the effect increases cumulatively up to 3 years after the loan. Subsequent stock returns also support the lending decisions, with connected firms outperforming unconnected firms over 1, 2, and 3 years after receiving a loan. Doubling the number of personal connections is associated with an increase in riskadjusted stock returns of almost 5%.

Conclusions

The authors conclude that personal connections between borrower and lender are leading to preferential terms for the borrower and an improved outcome for the lending bank and that this is especially influential for firms with poor or even no credit rating. The precise reasons for the outcome are difficult to discern since it is possible to formulate multiple theories, but the authors speculate that it could be due to better information flow or monitoring that is permitted by the common professional networks. They cite the example of microcredit organisations such as the Grameen Bank seems to support this theory, where the community network plays an important role in loan screening and monitoring.