The Network Effect: Strong as Ever

By Steve Reider

The network effect is the phenomenon by which large branch networks capture a disproportionate share of market deposits.  For example, a four-branch network captures more than twice the deposit volume of a two-branch network; an eight-branch network captures more than twice the deposit volume of a four-branch network.  Viewed another way, average deposit size per branch increases as a function of the number of branches, as each incremental branch a financial institution adds provides a lift to – and derives benefit from – all its preexisting branches in the market.

The network effect explains why financial institutions have historically pursued concentrated deployment strategies, e.g., 10 branches in one market versus five branches in each of two markets, as the broader representation leveraged the network effect. However, the erosion of in-branch transaction volume – average per-branch volumes are down 35% over the past five years – raises the question of whether the network effect persists; whether consumers still disproportionately place deposits with more widespread providers.

To confirm the extent of the network effect, Bancography reprised research it conducted in 2004 and 2010 that measured the correlation between bank rankings within the 50 largest U.S. metropolitan areas on two measures: deposits per branch and number of branches. The study measured the correlation between the two variables using Spearman’s rank correlation coefficient and controlling for main office biases and other outlier effects.  The Spearman’s statistic measures the degree to which the rankings of each variable (as opposed to the values) align; this helps minimize the effect of outliers.  If the network effect holds, we would expect a high correlation.  If perfectly correlated, the institution with the greatest number of branches in a market would show the highest average deposits per branch; the owner of the next largest branch network would show the second highest deposits per branch, and so on.

The 2004 and 2010 iterations of the study both found significant positive correlations, with the more recent study finding a positive relationship between network size and average branch size statistically significant at the .05 level in 44 of the top 50 metros and at the .01 level in 25 of those metros (lower significance levels equal stronger correlations). However, the years since 2010 have brought a wave of consolidation that reduced the number of U.S. banks from 7,830 (June 2010) to 6,058 (June 2016).  Further, that consolidation, plus the rise of online and banking channels, led many institutions to pare their branch networks.  Accordingly, it becomes reasonable to question whether recent developments changed the dynamics of the network effect – specifically, do large branch networks still reap disproportionate balance gains?

The study results confirm that not only do large networks remain beneficial, but that the extent of the benefit large networks realize has increased. In this most recent iteration of the study, 41 of the top 50 metros showed correlations significant at the .01 level and 47 at the .05 level, and the average correlation between deposits per branch and branch count across the top 50 metros increased from 42% to 58%.

Several factors likely contributed to the increased correlations, i.e., the greater market power enjoyed by those providers with the largest branch networks. First, the financial crisis winnowed a number of smaller institutions that led with rate-based appeals and accrued sizable deposit bases even with smaller branch networks (this group would reduce correlation levels with higher deposits per branch than their branch-count ranking would predict).  Second, a middle tier of community banks maintained lending levels (and thus deposit funding needs), even as the largest banks reduced leverage ratios while focusing more on problem loan resolution than asset growth.  Third, large banks bore the brunt of the industry’s reputational damage during the financial crisis; and in that timeframe some consumers opted to migrate to smaller banks, in effect rewarding fiscal responsibility over location convenience.  However, in recent years the reduction in rate-driven strategies, the reassertion of large network banks in lending (that drove a renewed emphasis on deposit growth), and the fading memories of the financial crisis appear to have increased the willingness of consumers to select based primarily on location convenience.

Finally, note the majority of the industry’s branch consolidations over the past five years arose from large institutions. And in lowering the denominator in the deposits per branch calculation, these banks boosted their ranking on that measure to a greater extent than they lowered their ranking on the branch-count measure.  Banks with market leading outlet share positions could rationally contract their branch networks down to one more branch than the second-place provider.

The above analysis notwithstanding, keep in mind that correlation is not causation. For example, if larger banks hold greater technological resources, they will likely be able to introduce more sophisticated electronic offerings at an earlier stage (e.g., mobile banking).  So perhaps it is the presence of those broader service offerings that allow larger institutions to realize disproportionately greater per-branch balance gains, and the correlation with branch-network size is simply a byproduct.

Still, the preponderance of statistical evidence suggests larger branch networks as more efficient from a balances-per-branch standpoint, yielding distinct implications:

  • For banks operating in multiple markets, consider exiting those where the bank holds lower-tier market positions and instead concentrate capital and noninterest investments in established strongholds.
  • For smaller banks operating in a single market, keep in mind that the network effect holds not only at the metropolitan area level but also at the corridor level. Thus, rather than scattering branches across an entire metro area, strive to build a cohesive network in one or two corridors while ceding others entirely, as this will yield increased per-branch deposits.
  • For banks maintaining market-leading branch networks, consider the branch-count differential between your institution’s network and that of the next-largest competitor. If consumers choose based on the largest branch network, then is there benefit to having the largest network by four branches versus the largest network by two branches?  Or can the bank contract two branches and remain secure that it still can enjoy the benefits of the largest network in the market?

Although the network effect holds in markets across the country, note the strength of the effect varies. Several markets with strong legacies of thrifts and mutual savings banks show lower correlations, as institutions in that sector of the industry tend more toward rate-based, branch-light operating models.  Further, several Sunbelt metros with low overall branch concentration levels show lower correlations, perhaps reflecting that no one provider has yet amassed a dominant branch position in those markets.  Still, overall the effect remains demonstrable and prominent across the nation’s major metro areas, suggesting that, even in an age of electronic banking and reduced branch service visits, the presence of a broad branch network creates brand awareness and consumer confidence that foster outsized deposit gains.

Steve Reider is president of Bancography, based in Birmingham, Ala. Bancography provides consulting services, software tools, and marketing research to financial institutions.