Securitization without risk transfer

Abstract

We analyze asset-backed commercial paper conduits, which experienced a shadow-banking run and played a central role in the early phase of the financial crisis of 2007–2009. We document that commercial banks set up conduits to securitize assets worth $1.3 trillion while insuring the newly securitized assets using explicit guarantees. We show that regulatory arbitrage was an important motive behind setting up conduits. In particular, the guarantees were structured so as to reduce regulatory capital requirements, more so by banks with less capital, and while still providing recourse to bank balance sheets for outside investors. Consistent with such recourse, we find that conduits provided little risk transfer during the run, as losses from conduits remained with banks instead of outside investors and banks with more exposure to conduits had lower stock returns.

Introduction

Securitization was traditionally meant to transfer risks from the banking sector to outside investors and thereby disperse financial risks across the economy. Because the risks were meant to be transferred, securitization allowed banks to reduce regulatory capital. However, in the period leading up to the financial crisis of 2007–2009, banks increasingly devised securitization methods that allowed them to retain risks on their balance sheets and yet receive a reduction in regulatory capital, a practice that eventually contributed to the largest banking crisis since the Great Depression. In this paper, we analyze one form of securitization, namely, asset-backed commercial paper (ABCP) conduits (henceforth, conduits), as an example of how banks exposed themselves to such risks.

Conduits are special purpose vehicles managed by large commercial banks. Conduits purchase medium- to long-term assets, which they finance by issuing short-term asset-backed commercial paper. Given this structure, conduits are similar to regular banks in many ways and form an integral part of financial intermediation that has over time come to be called “shadow banking.” Put simply, shadow banking is that part of the intermediation sector that performs several functions that traditionally are associated with commercial and investment banks but runs in the shadow of the regulated banks, in the sense that it is off-balance sheet and less regulated.¹

As shown in Panel A of Fig. 1, ABCP outstanding grew from $650 billion in January 2004 to $1.3 trillion in July 2007. ² At that time, ABCP was the largest money market instrument in the United States. For comparison, the second largest instrument was Treasury bills with about $940 billion outstanding. However, the rise in ABCP came to an abrupt end in August 2007. On August 9, 2007, the French bank BNP Paribas halted withdrawals from three funds invested in mortgage-backed securities and suspended calculation of net asset values. Even though defaults on mortgages had been rising throughout 2007, the suspension of withdrawals by BNP Paribas had a profound effect on the market.³

As shown in Panel B of Fig. 1, the interest rate spread of overnight ABCP over the federal funds rate increased from 10 basis points to 150 basis points within one day of the BNP Paribas announcement. Subsequently, the market experienced the modern-day equivalent of a bank run that originated in shadow banking, and ABCP outstanding dropped from $1.3 trillion in July 2007 to $833 billion in December 2007.⁴ Apparently, investors in ABCP, primarily money market funds, became concerned about the credit quality and liquidation values of collateral backing ABCP and stopped refinancing maturing ABCP.

Our main conclusion in this paper is that, somewhat surprisingly, this crisis in the ABCP market did not result (for the most part) in losses incurred by those actually invested in ABCP. Instead, the crisis had a profoundly negative effect on commercial banks because banks had (in large part) insured outside investors in ABCP by providing explicit guarantees to conduits, which required banks to pay off maturing ABCP at par. Effectively, banks had used conduits to securitize assets without transferring the risks to outside investors, contrary to the common understanding of securitization as a method for risk transfer. We argue that banks instead used conduits for regulatory arbitrage.

We first show and describe the structure of the guarantees that effectively created recourse from conduits back to bank balance sheets. For the most part, these guarantees were explicit legal commitments to repurchase maturing ABCP in the event that conduits could not roll over their paper, not a voluntary form of implicit recourse. The guarantees could be structured as liquidity guarantees, a contract design that would reduce their regulatory capital requirements to at most a tenth of the capital required to back on-balance sheet assets (especially after this regulation was confirmed as a permanent exemption by regulators in the United States in July 2004; see Fig. 2). Such liquidity guarantees would cover most assets' credit and liquidity risks and effectively absorb all losses of outside investors. For comparison, banks also had the option to use weaker guarantees that did not cover all of the assets' liquidity and credit risks or use stronger guarantees that had strict capital requirements.

We test for regulatory arbitrage using a novel panel data set on the universe of conduits from January 2001 to December 2009. First, we analyze guarantees provided by, and the type of, financial institutions that manage (“sponsor”) conduits. We find that the majority of guarantees were structured as capital-reducing liquidity guarantees and that the majority of conduits were sponsored by commercial banks (which among financial institutions are subject to the most stringent capital requirements). Also, the growth of ABCP stalled in 2001 after regulators considered increasing capital requirements for conduit guarantees (following the failure of Enron, which had employed conduit-style structures to create off-balance sheet leverage) (Wall Street Journal, 2001) and picked up again, especially the issuance of liquidity-guaranteed paper by commercial banks, after a decision against a significant increase was made in 2004 (see Fig. 2).⁵

Second, we examine whether more capital-constrained commercial banks were more likely to set up conduits. Using the sample of commercial banks with more than $50 billion in assets from 2001 to 2006, we find that liquidity-guaranteed ABCP was issued more frequently by commercial banks with low economic capital, measured by their book value of equity relative to assets. We use panel regressions to confirm that this result is robust to controlling for time trends, bank characteristics, and bank fixed effects. Interestingly, we find a much weaker relation between the issuance of liquidity-guaranteed ABCP and the bank's regulatory capital, measured as the Tier 1 regulatory capital relative to risk-weighted assets. And, we find no relationship between a bank's capital position and the issuance of nonliquidity guaranteed ABCP, which had no associated relief from a regulatory capital standpoint. These results are highly suggestive of regulatory arbitrage. In particular, the use of liquidity-guaranteed conduits allowed banks to reduce their economic capital ratio, while maintaining a stable regulatory capital ratio.

Third, we examine the effect of guarantees on conduits' ability to roll over maturing ABCP during the shadow-banking run. The regulatory arbitrage hypothesis suggests that banks did not transfer risks to outside investors. We test for risk transfer using variation in the strength of guarantees and examine whether conduits with weaker guarantees had higher spreads, and were less likely to roll over ABCP, once the run took hold in August 2007. Using conduit-level data on daily spreads and weekly issuances, we find that, starting on August 9, 2007 conduits with weaker guarantees (namely, conduits with extendible notes and structured investment vehicles or SIVs) experienced a substantial decrease in their ability to roll over maturing ABCP and a significant widening of spreads. Consistent with the lack of risk transfer, we find that conduits with stronger guarantees (namely, liquidity guarantees and credit guarantees) experienced a smaller decrease in issuances and a smaller rise in spreads.

Fourth, we analyze the extent of realized risk transfer by taking the perspective of an investor that was holding ABCP at the start of the run and studying whether the investor suffered losses by not rolling over maturing ABCP. The regulatory arbitrage hypothesis suggests that losses primarily remained with the sponsoring financial institution (henceforth, sponsor), not with outside investors. Using hand-collected data from Moody's Investors Service press releases, we identify all conduits that defaulted on ABCP in the period from January 2007 to December 2008. We find that all outside investors covered by liquidity guarantees were repaid in full. We find that investors in conduits with weaker guarantees suffered small losses. In total, only 2.5% of ABCP outstanding as of July 2007 entered default in the period from July 2007 to December 2008. Hence, most losses on conduit assets remained with the sponsoring banks. Assuming loss rates of 5–15%, we estimate that commercial banks suffered losses of $68 billion to $204 billion on conduit assets.

Finally, we examine the impact of a bank's exposure to conduits on bank stock returns. The regulatory arbitrage hypothesis suggests that banks were negatively affected by the run because they had insured conduits against losses. We focus our analysis on the narrow event window around the start of the financial crisis on August 9, 2007 to identify conduit exposure separately from the impact of other bank observables. We find that an increase in conduit exposure (measured as the ratio of ABCP to bank equity) from 0% to 100% (e.g., comparing Wells Fargo and Citibank) reduced the cumulative equity return by 1.1% during a 3-day window. The estimate increases to 2.3% when we expand the event window to one month. The result is robust to controlling for a large set of observable bank characteristics, and we find no effects prior to the run.

In summary, our results show that commercial banks used conduits to invest in long-term assets without holding capital against these assets. This evidence suggests that banks' investment decisions are at least partly motivated by activity aimed at circumventing regulatory constraints. Moreover, because these investments reflect significant maturity mismatch and default only in a severe economic downturn, banks are taking on rollover risk that is highly correlated within the financial sector. Hence, our analysis shows that regulatory arbitrage activity, if successful, can create significant concentrations of systemic risk in the financial sector. Regulatory arbitrage activity could result in a shadow banking sector that is intimately tied to the regulated banking sector, instead of transferring risks away from the latter.

The remainder of this paper is organized as follows. Section 2 discusses the institutional background. Section 3 presents our theoretical framework. Section 4 describes the data and discusses our empirical results. Section 5 analyzes the incentives of banks to set up conduits. Section 6 reviews the related literature. Section 7 concludes.