Thin markets and the search for capital

Textbook economics assumes “thick markets”, meaning that there are lots of buyers and sellers and the process by which they encounter each other is well-oiled enough not to be worth bothering with.

A demand curve is derived by considering all the potential buyers in a market, each with their own budgets and desires, and asking how many of them would purchase the item at a given price. The question of how buyers and sellers find each other does not arise.

Financial markets are usually thick. There is a lot of money sloshing around and a lot of people being paid to find sensible things to do with it. If a financial asset is priced to sell, it will find buyers without difficulty.

Anyone trying to raise capital in an emerging or frontier market might not recognise this picture. A business trying to borrow $50m might be paying a competitive rate but all the commercial banks in that market might collectively only be willing to lend $30m. This affects how we think about additionality – it is not just about finding investments that a private investor would reject, it is about understanding the market and where the gaps are. In this situation, a DFI would be additional by supplying the other $20m, alongside private investors.

In an extreme case, where there is a fundamental shortage of capital relative to investment opportunities in a market, we needn’t worry so much about “crowding out” – if a DFI beats a commercial investor to a deal, that investor would then allocate their money to another deal that would have failed to find capital, instead. [1] Any investment by a DFI would increase the overall volume of investment in such markets.

In more developed markets that situation would be hard to explain. The supply of money is “elastic”, meaning it should respond quickly to where profitable opportunities are to be found. An inelastic supply is more plausible in emerging and frontier markets where the domestic financial sector is small, and the market is of limited interest to global investors. But even in underdeveloped financial sectors, over time we should usually expect the supply of capital to respond to a surplus of good investment opportunities.

A thin market does not imply capital scarcity, however. A market could be thin, meaning that there are few buyers and sellers, but supply and demand could balance. We cannot conclude that capital is scarce and DFIs do not need to worry about additionality merely because we observe low levels of activity in a market. It will often make more sense to think about the balance of supply and demand for certain categories of investment.

Perhaps there might be a shortage of growth equity, for example, but not of short-term debt.

Economics has a way of handling situations where the process of sellers finding buyers is a task in itself, called “search and matching” models.[2] These models are useful because they can take us along the continuum from a thin market, where only few offers materialise in a reasonable period, to a thick market where matches are almost instantaneous.

These models are mainly used in labour economics, where the process of advertising a vacancy, interviewing applicants and making a job offer requires time and resources, and where, from the worker’s perspective, searching for a job is also time consuming. The basics are something like this:

• Both employers and job seekers only embark on a search if they think it’s going to be worthwhile, relative to their alternatives. Searching is costly. If parties receive no satisfactory offers, at some point they will give up searching.
• When either party receives an offer, they decide whether to accept it by comparing the value of that match to the value of continuing to search.
• Different employers and job seekers have different characteristics and want different things, so the values they place on potential matches and on continuing to search are specific to them.
• There is a “matching function” that determines how quickly potential matches are made, and the probability of finding a successful match, which handles things such as how many vacancies and job seekers there are (how thick or thin the market is).

These models have been applied to other settings, including investment.[3] The remainder of this blog will discuss how they might work when applied to development finance.

• Even in a thin market with a shortage of capital there will still be some competition for individual deals. Investors will search through many of the available opportunities before making offers and businesses may receive multiple offers.
• Although it would be possible to model a situation in which the probability of a DFI being additional could be close to 0 or close to 1, there would be many situations in which additionality is somewhere in between.[4] The model could tell us, if a DFI declines a given deal, what are the chances the business would go on to successfully match with a commercial investor? That probability will change depending on characteristics of the firm in question and of other investment opportunities, and how thick the market is.
• Investment opportunities have some objective characteristics, such as the sector, geography, and track record of the sponsor, but each investor would form their own view of the risks and the private and social returns on offer. It would probably be easiest to model those subjective valuations as some function of observable characteristics plus a random element. Businesses also have subjective ideas about their own value, and some will fail to make a match because their self-valuation is too far from that of investors.
• Because investors form subjective valuations of potential matches, the theory could naturally capture how DFIs differ from commercial investors. Investments could offer social returns (impact) alongside private returns which some investors could incorporate that into matching decisions.

• Search and matching changes how we think of “market pricing” and “concessionality”. In a thick frictionless market, sellers will not deviate from the market price because if they are too expensive no one will buy from them and if they are too cheap then they are needlessly forgoing income. In a thin market there aren’t enough comparable assets to establish “a market price”. In a thin market, a seller might receive a handful of quite different offers, and from a buyer’s perspective, similar sellers might be asking for different prices. That blurs the boundary between pricing on commercial terms and being concessional.
• An concessional impact investor might be willing to accept an expected return they perceive to be below a commercial hurdle rate but still find themselves occasionally outbid by a commercial investor with a different subjective view of the business. However, the presence of DFIs or impact investors who are willing to accept subjectively sub-commercial returns, when they see high social returns, would make it much easier for some businesses to find capital (those who most commercial investors would reject).
• If the theory is expanded to include characteristics of financial products that firms care about, such as loan tenor and terms, it would naturally capture how financial sector development improves matching efficiency (by offering a greater range of financial products), which will result in more investment, because more searches will be successful and more businesses will find it worthwhile to embark on a search. If there is a surplus of good investment opportunities, investors will quickly make good matches, which should result in more investors entering the market.
• The costs of search could be decomposed into a fixed cost of search per period and a match-specific transaction cost. Some matches may require more work to conclude from the investor and/or the business than others, so when either side encounters a potential match, they must also decide whether to incur transaction costs. DFIs can be additional by being more willing to bear transaction costs where they perceive high social returns.
• Both parties would need to have defined outside options – for example, rather than trying to match with businesses raising capital in private markets, investors could buy listed corporate bonds, and firms could use internal cash flows rather than find external investors (or simply shelve their expansion plans). Even in cases where asset managers have capital allocated for fixed investment strategies (so buying bonds is not an option), the model will need to define when they would prefer to return capital to their investors rather than invest it.

It would generally helpful if everyone had a realistic mental model of how development finance works. The idea that markets are thin and raising capital involves a costly search process helps explain why additionality is better thought of as a probability than as a binary.[5] This matters for how researchers and commentators think about additionality, which is often approached as something that can be measured at a transaction level without consideration of market-level additionality.

Search frictions help us see that the question of whether a business can raise money from a commercial investor in a thin market is about more than whether the perceived risk-adjusted return (and transaction costs) satisfy commercial hurdles. Those characteristics are still central to the decision of whether an investor will accept or reject a potential match, but firms may also sometimes fail to raise capital because the search process is too inefficient or because there is a mismatch between market-level supply and demand. Search frictions point towards another way in which DFIs can mobilise private investment, by helping firms and investors find each other.

Theory would be more useful if it could suggest some ways to distinguish low levels of activity from capital scarcity and matching frictions in practice. DFIs already try to gauge additionality using background information on the extent of investment activity in the local market. Perhaps theory might suggest other indicators to look for. I do not know how easy it would be to obtain information about average search durations and failure rates in capital raising, for example, but that might help DFIs calibrate where they set the bar for additionality decisions.

Search and matching models paint a more realistic picture of investment in emerging and frontier markets, but they don’t capture everything that matters. For example, if we want to understand the behaviour of banks and what that implies for which small firms can access credit and which cannot, the process of how firms and local banks find each other is probably not the most pressing problem and we need to look at other aspects of the system.

[1] We might worry a bit because not all deals are the same

[2] Here is an academic survey

[3] Here is an example of a search and matching model applied to venture capital.

[4] See “Our approach to contribution” for an explanation of how BII assess the likelihood of additionality and incorporates that into investment decisions.

[5] We have previously written about taking a probabilistic approach to additionality, simply because it’s often impossible to know whether an investment is additional in the act of making it, so it’s helpful to consider what makes additionality more or less likely. In these models it should be possible to compute a probability (by running a Montecarlo simulation) that a given (modelled!) investment would result in an increase in the volume of investment in a market.