Headlines about private credit ‘cockroaches’ and Business Development Company investors running for the hills are a regular occurrence. Just how worried should we be about this burgeoning part of the credit landscape?
In this piece, we focus on the relationship between the traditional financial sector (banks and insurers) and ‘private credit’ in the US. What we mean by private credit here is direct lending to US middle market companies (the term can be used more broadly to include, for example, investment grade lending).
We argue that private credit is best understood not as a repeat of past shadow‑banking excesses, but as a leveraged extension of the traditional financial system – one that concentrates risk differently and therefore demands a different lens from investors and regulators alike.
Mapping today’s private credit ecosystem
Middle market companies typically have revenues of up to $1 billion. It is called direct lending because the funds go directly from investors to the companies (ie with no banks involved). However, the actual lending decision – which companies get the cash – is typically trusted to private credit fund managers or Business Development Companies (BDCs) in about a 50/50 split. The investment in these vehicles is via both equity and debt (again about half and half). The banking sector provides most of the debt funding (about three quarters).1
Direct middle market lending is leveraged lending. The companies involved are 6x-8x leveraged with an average credit rating of B- to CCC+. The sector makes up about a third of the US leveraged lending market and is about $1.4 trillion in size.2 (We believe this is a good figure, but it is close to impossible to get precise data because it is ‘private’.) The other two thirds are the leveraged loan market and the high yield (HY) bond market, both of a similar size. Figure 1 is a simplified version of this setup.
Figure 1: The middle market lending setup
Source: Columbia Threadneedle Investments’ analysis, March 2026.
At the beginning of this century, banks in the West exploited capital rules by creating an alphabet soup of credit products. Vehicles such as structured investment vehicles (SIVs), collateralised debt obligations (CDOs) of residential mortgage-backed securities (RMBS) and asset-backed commercial paper (ABCP) conduits were often opaque, off-balance sheet and highly leveraged. These entities grew rapidly, inflating prices in credit and housing markets. We all know how that ended with the global financial crisis (GFC) of 2008.
Fast-forward just five years from then and bankers in China were taking a similar approach. Corporate debt levels exploded through ‘shadow banking’ products such as local government funding vehicles (LGFVs), asset management products (AMPs) and wealth management products (WMPs). Essentially, this is another off-balance sheet debt frenzy, just with different acronyms. The effects of that debt binge are still playing out in Chinese commercial real estate today.
Regulatory constraints, unintended consequences
Have the lessons of the past reshaped credit intermediation? In some important respects the answer is no. Following the GFC, banks did face tighter oversight on leverage, most notably through the 2013 leverage lending guidelines and the subsequent Shared National Credit exams, which effectively capped the amount of leverage banks could extend to borrowers (around 6x). However, companies that wanted larger, more flexible and often cheaper financing began turning to lenders that were not restrained by these supervisory limits. At the same time, investors spent years searching for yield in a very low rate environment. This fuelled rapid expansion in non-depositary financial Institutions (NDFIs) such as private credit funds and Business Development Companies willing to provide the leverage banks could not. Banks responded by increasing their exposure to NDFIs, allowing them to remain involved in the private credit market while staying within supervisory boundaries and maintaining valuable client relationships.
The regulatory environment of the past decade has pushed banks towards lending to NDFIs by giving these exposures far more favorable capital treatment than traditional commercial loans. For example, under Basel III, loans to NDFIs often receive materially lower risk weights – 20%-30% versus 100% for traditional commercial and industrial (C&I) loans. This is because banks sit in a second-loss position and the spreads on these loans are usually higher, driven by supply and demand and complexity of the transaction. Together, these factors raise the economic appeal of NDFI lending and help explain why banks have continued to expand their presence in this channel, with little incentive to slow the growth unless capital rules change.
More recently, the easing of capital requirements and moderation of stress capital buffers for banks has only reinforced this trend. With more balance sheet capacity to deploy, banks have found NDFI lending to be the most attractive outlet relative to originating the same loans directly. This dynamic has reshaped the credit landscape, with NDFI loans rising from around $200 billion a decade ago to around $1.5 trillion in Q4 2025, which represents around 11% of total U.S. bank loans.
Politicians have short memories
This rapid growth has occurred against a backdrop of weakening supervisory oversight. The 2018 rollback of parts of the Dodd-Frank Act raised the threshold for enhanced oversight of banks from $50 billion to $100 billion, which meant category 4-5 regional banks with assets under those levels faced far less scrutiny. That lack of oversight was a driving factor in the regional bank failures we saw in Q1 2023, such as the Silicon Valley Bank.
The trend continued in October 2025 when Vice Chair of Supervision at the US Federal Reserve (Fed), Michelle Bowman, announced plans to cut around 30% of the Fed’s supervision and regulation staff. For context, a 30% cut from the current 499 authorised positions (as per the latest budget of the Federal Reserve System3) would bring headcount down to around 350, the lowest since 2011. All of this points to a clear decline in supervisory intensity since 2018, creating blind spots at a time when bank exposure to NDFIs has become far more significant.
These blind spots matter because NDFI lending has become increasingly complex. Banks now extend subscription lines, net asset value (NAV) loans, and warehouse facilities, each with different collateral structures and risk profiles. These are exposures that require time, expertise and consistent examination, yet the resources devoted to them appear to be moving in the opposite direction.
Is this regulatory arbitrage all over again?
These developments are a clear example of regulatory arbitrage at work. By channeling credit to NDFIs, banks exploit gaps between traditional oversight and the shadow banking sector. While these arrangements boost profitability, they introduce opaque credit risks that may surface under stress.
The capital advantage is material. Let’s look at some numbers for bank lending to private credit funds, which roughly account for a quarter of all NDFI lending. The bank makes a loan to the fund. The fund, in turn, makes loans to middle market companies. This sort of bank lending attracts a 20% risk weighting. From the bank’s perspective, a $100 loan requires $2 of capital (50x leveraged). Conversely, when a bank lends directly to this sort of company, the exposure will attract a 100% risk weight. For a loan of $100, the bank will hold approximately $10 of capital (10x leverage).5
However, that is not the full story. In the event of difficulty, the fund equity investors typically take the first losses, so the fund would need to suffer around 50% losses on those loans for the bank to lose money. Cumulative five-year defaults on B-/CCC+ rated loans are around 25%. Even with low recovery rates, that builds in significant protection for the bank. AA-rated tranches of collateralised loan obligations (CLOs) have similar structural protections and losses are extremely rare.
Nothing to worry about then? The picture is more nuanced. Although private credit funds typically hold portfolios of more than 300 loans, that diversification has limits. Around 20% of loans in private credit funds are to software companies. Recent developments in artificial intelligence (AI) have raised questions about the business models of many of these highly leveraged names. Much of the exposure matures in 2030 and beyond, leaving plenty of time for risks to grow. If AI leads to problems at software companies, some of the losses are likely to come back to the banking sector via this channel.
Where do the risks sit?
Pinpointing this is difficult. Concentration risks are obscured by limited and uneven bank disclosure. Visibility of credit quality and concentration risks remains limited. Banks have argued that disclosure around private credit exposures is comparable to that provided for commercial real estate (CRE) and commercial and industrial (C&I) lending. However, the private credit boom largely occured during a period of benign credit conditions. As Warren Buffett has observed, “It’s only when the tide goes out that we discover who’s swimming naked.”
When loans are growing this fast, investors need more information to get comfortable with underwriting. The problem is compounded by significant risk transfers (SRTs), whereby the risks have been transferred elsewhere in the financial system, often to the same private credit investors to whom the banks are lending. In times of market stress, credit markets pull the funding from areas where disclosure is poor. The funding only comes back when disclosure improves, and the market can appropriately price the risk.
Has competition overtaken caution in alternatives?
We believe it has, with rapid asset growth and excess capital putting increasing pressure on underwriting standards. For the past five years, AUM at the large US alternative asset managers has been growing at around 15% per annum. There is a huge amount of dry powder – more than $1 trillion in private equity and $400 billion in private debt – as it becomes increasingly popular for investors to make allocations to alternative managers.6 Between 2004 and 2007 lending standards deteriorated precipitously as lenders competed for assets. In private markets, competition may lead to a loosening of underwriting standards; covenant-light and PIK structures become more popular. Many might argue that the larger alternative asset managers have good track records of allocating capital. They do, but so did banks before the GFC.
Has sentiment turned?
We believe it probably has. Private credit funds and BDCs are now facing redemptions. If all redemption requests had been honored, it would have amounted to a net outflow from the asset class of around 3% in the first quarter of this year. However, these products are gated, meaning only a certain percentage of redemption requests need to be met to prevent forced selling. Gating is a sensible feature of these funds. Illiquid credit is not designed to be sold to realise cash.
The macroeconomic context
When comparing the current explosion of NDFI lending to the US pre-GFC and China in 2015 it is important to look for the buildup of leverage at the economy level. In the early 2000s, US household debt-to-GDP increased by about 30 percentage points to around 100%. In China in the five years to 2017, non-financial corporate debt-to-GDP increased by about 50 percentage points to 200%.
Crucially, that is not happening here. This sort of direct lending has been growing fast, but it has been taking share from the loan and HY bond market. The overall leveraged finance market (HY bonds plus leveraged loans plus directly lending) has not been growing relative to nominal GDP. What is more, the US corporate sector broadly does not have a leverage problem. All this lacks that classic hallmark of a credit bubble.
It is important to note, however, that in both the US pre-GFC and in China in 2015 the government sector was not overly indebted, with government debt-to-GDP at around 60% as the crisis struck. This allowed the government to fiscally stimulate, pulling the economy out of trouble. Now, however, that would be much more difficult. Government debt levels are already high and, in the US, the foot is already to the floor fiscally.
How material are the risks?
Deregulation, rapid credit growth, lack of transparency, sector concentration and leverage are rarely a good combination for bondholders. However, we should not overstate the risks here – it’s not all doom and gloom. Overall, the US economy is in decent shape.
While there will inevitably be some blowback to banks and insurers, first losses will mainly fall to equity investors in private credit funds and BDCs. Banks will have less capital next year than they did last year, but that is still better than it has been for decades. Earnings are strong, most management teams are savvy, allowance for credit losses remain conservative and forward-looking loss estimates at the overall bank level are benign.
Across global portfolios, we are taking a more cautious stance toward US bank‑related credit risk, with a bias toward higher‑quality exposures and positioning for an environment in which credit spreads are likely to widen.
