Corporate Networks and Cost of Capital: A Study of Indonesia
This article is a collaboration between Wesley Barnes, founder of Brightriver Capital, a private equity fund manager, and Ben Charoenweng, PhD, a professor at National University of Singapore and co-founder, Head of Research of Chicago Global, a public equity systematic fund manager.
Introduction
Market transparency is a major challenge in emerging markets investing. We often have only superficial information on companies, their owners, and related parties. Specialists have traditionally accrued anecdotal knowledge about market players. In private equity, it is common to engage third parties for investigative background checks. Whether public or private, prospective investors need to look beyond the prospectuses, pitchbooks, and pro forma financials to truly understand where stakeholder interests are misaligned with those of the investors themselves.
In emerging markets, the alignment of shareholder interests often determines the ultimate success of an investment. Red flags that may arise in the social and professional connections of corporates may lead an investor to forgo an investment altogether. Identifying red flags is thus an essential part of any evaluation of securities risk, although many factors influence an investors’ conviction on whether to invest.
The quantitative measurement of social and professional connections among firms or corporations remains technically challenging. Investors rely on heuristics to offset (often successfully, sometimes not) the lack of a large dataset for empirical analysis of counter-parties. Connecting these heuristics directly to valuation has also been elusive, despite the growth of automated KYC and network visualization tools.
Our research fills that gap by using a network dataset covering a large sample of publicly-listed Indonesian companies, covering both associated individuals and firms, compiled using Brightriver Capital’s proprietary Lantern Graph tool. We compare network statistics from this data set to cost of capital information from Refinitiv to assess to which features of their respective networks influence the cost of capital, and by extension, company valuation.
Our research resulted in four key findings that highlight the importance and utility of network analysis for public and private equity investments.
- More network connections correlate with a lower weighted average cost of capital.
- Higher importance in a network correlates with a lower cost of equity capital.
- Lower network quality (based on red flags) and higher density correlates to higher costs of capital.
- Nearly all of impacts described above are driven by the cost of equity component of WACC.
The first two results align with common perceptions that well-known and sizable market participants are less risky and earn premiums in valuation. The third result highlights how even indirect connections to blacklisted or PEP entities can result in a valuation discount (by way of a higher cost of equity capital). The final result confirms the primacy of equity investors and their role in assessing risks and valuation.
Veteran emerging markets investors would not find such results surprising. In our research, network analysis provides a quantitative tool for more insightful due diligence and an expanded assessment of risk and valuation for public and private investors.
The Dataset
We use two sets of data on the network of individuals and entities and their relationships. All network measures, such as centrality measures, are constructed based on the entire network comprising both individuals and entities. Network measures include both information from publicly-listed as well as private companies as of May 2020. We compare this cross-section to the cost of capital data (the main outcome variable) from Refinitiv dated March 2021. We consider the weighted average cost of capital (“WACC”) and also split it into the equity and debt components separately.
Our final sample covers 505 publicly-listed Indonesian firms for which cost of capital estimates are available. We exclude financials and utilities from our analyses.
Network Scale
An individual entity’s network structure tells us much about an individual itself. Network structure relates to an individuals influence, information advantages, or susceptibility to peer group influence. Among the various metrics used in academic research, centrality metrics evaluate the prominence and influence of an individual entity within a larger network.
While there are several different types of centrality (each with different implications), we researched each entities’ “degree centrality,” which measures the number of individuals and companies to which an entity is connected relative to possible connections. While general, this metric has the benefit of straightforward calculation and understandability.
We find that overall, the higher the degree centrality, the lower the equity cost of capital. Our analysis controls for firm fundamentals, including total assets, asset tangibility, leverage ratio, and return-on-assets. Statistical tables appear at the end of this paper, and the results are robust.
The impact on the cost of capital and implications for valuation are meaningful. Doubling degree centrality decreases the WACC by 0.386%, equivalent to over 6% of the market average WACC of 6.10%.
Network Density
Network density refers to the number of shared connections in a network and is another key area of network research. Higher network density has been linked in research to shared behaviors among group members, a characteristic known as “homophily”. More commonly the concept is captured by the old adage “birds of a feather flock together”. In network science, it determines the speed and diffusion of information and behaviors.
Our research uses a metric called the Local Clustering Coefficient (“LCC”). The LCC measures how close a group is to having each member connected to each other. The higher the LCC, the denser the network.
Interestingly, the more closely-knit the community on this measure, the higher the cost of capital in our sample. While the impact of the relationship is not as material as centrality, it may represent the more correlated fundamentals of a firm with others in a densely connected community.
In a sense, it can be considered as “market beta” or systematic risk exposure within that community, leading to a higher risk for any entity within that tight community. Firms more embedded in a network will likely have more correlated fundamentals with their community and see fewer opportunities arising from network brokerage.
Doubling the clustering coefficient correlates with an increase in the WACC by 0.064%, equivalent to 1% of the average market WACC.
The statistical results for both degree centrality and clustering are included below.
Network Structure, the Cost of Capital and Valuation
As investors, we can integrate network structure measures into our assessment of valuation as part of our assessments of valuation discounts or premiums.
Below, we illustrate the combined effects of network size and density on the cost of capital and valuation. As we see the cost of capital rise, the prospective valuation (based on an analytical decomposition of EV/EBITDA) should fall. In extreme cases, the premium or discount to the average market valuation is substantial.
Network Quality
Theories around homophily point to shared behaviors in closely associated networks. Therefore, it is worthwhile to understand the quality of a counterparty’s network. For our research, we look at indicators of poor network quality, or “red flags”.
We define red flags as connections to individuals or firms who show up on local or international blacklists, are politically exposed (so called “PEPs”), or that are in substantial litigation. Avoiding these kinds of networks may improve an investor’s investment outcomes, particularly in multi-year engagements with a higher duration, such as a private equity transaction.
The market has provided various research tools, both project-based research and data platforms, for such diligence purposes. These tools identify red flags for an investment target and their direct circle of immediate connections. They are less effective at penetrating a target’s broad network connections without adding research time and cost. Our dataset allows us to take a broader look and analyze the impact of these relationships quantitatively.
Over half of the firms (52.8%) in our sample are connected to at least one blacklisted and PEP entity. We see a strong relationship between network quality measures and WACC. The more black-listed entities in a network, the higher we would expect to their cost of capital to be. Doubling the number of black-listed connections increases WACC by 0.061%. The economic effects of doubling PEPs and litigation in the network are less meaningful, 0.037% and 0.044% respectively, all when controlling for the total number of connections. Statistical tables are included below.
These results empirically support the standard business intuition that the observed quality of a network impacts invest-ability. In practice, these quality measures appear together in the same network (or even in relation to the same entity). While we did not measure the combined impact of these three areas directly, we would expect that the economic effects to be stronger on a combined basis, and especially so in higher density network structures.
All About the Equity
The impact of network quality may appear surprisingly muted given its implications. However, the results change dramatically when separating the key components of the WACC, the cost of equity and the cost of debt. We studied whether the relation between the cost of capital impact of different quality measures is driven by its equity or debt components. Our results document that cost of capital impacts are almost all driven by equity, and not debt.
Quantitatively, the research shows the doubling blacklisted connections increases the cost of equity by 17.4 bps (nearly 3 times the impact on the cost of capital above), with greater increases for PEP connections and litigation, which rise to 15.9 bps and 14.2 bps respectively. Statistical tables are included below.
The cost of debt capital is less sensitive to network structure and quality, and effectively dampens the overall impact on the cost of capital. We believe this likely due to the characteristics of local debt markets in Indonesia and the nature of the instruments. These results point to the primacy of equity investors in driving valuations through their assessments of company features, including network quality.
Conclusions
Investor decisions on valuation and the relative attractiveness of an investment are multi-faceted. They are crafted around a thesis that integrates both quantitative and qualitative measures of financial performance and risk. The relationship between network structure and the cost of capital presented in this research note represents an important area to add these assessments.
Traditional counterparty network analysis has focused on qualitative elements based on readily available biographical information or information gathered through structured background checks. Here, we studied the cross-sectional relation between measures of a firm’s network structure with its average cost of capital taking a quantitative approach.
It may seem intuitive to investors to prefer to invest in larger organizations that are more central to the market, and thus such organizations can garner higher relative valuations. We found quantitative support for this intuition in that larger networks (as proxied by network centrality) corresponded to a lower cost of capital.
Conversely, we found that closer-knit (more “embedded”) networks, as measured by a common clustering metric, correspond to an increased cost of capital. This may be the result of two features of high density networks. First, tightly connected organizations may feature more correlated fundamentals with their community. Second, there will be fewer opportunities for entities within the network to create value through intermediation. From an investor perspective, a highly embedded network also raises the possibility that negative characteristics shared among individuals in the network may also influence an investment target.
The research also incorporates the quality of the network by examining red-flag components as a relevant feature of evaluation. It demonstrates a higher cost of capital for “problematic” networks, defined by those with many blacklisted entities, PEPs, or legal actions. The results would not surprise seasoned investors, but the fact that it matters not just for the target, but also other entities in the target’s network, can be missed in traditional diligences.
In private markets, our research points to a novel, network analytic approach to identify new companies and individuals in a market for investment, enhance due diligence, and refine investment pricing. Importantly, early identification of positive and negative qualities of network structure can improve our ability to select the best partners, given that we will be working with them for several years before realising an investment.
The quality of shareholders and stakeholders is also a concern for investment in publicly listed firms. Future work remains on how time-series changes of network size and density impact financial value, but such research could ultimately lead to the robust identification of mispriced securities as part of a quantitative investment strategy. Even now, our research demonstrates the clear connection between network features and quality and equity valuations.
Supplement: Statistical Tables
Degree Centrality and Clustering Coefficients
Network Quality Measures
Equity vs Debt Costs of Capital
