The Risk-Shifting Hypothesis

In our paper, The Risk-Shifting Hypothesis: Evidence from Sub-Prime Organizations, which was recently presented at Harvard, we provide evidence consistent with risk-shifting in the lending behavior of a large subprime mortgage originator (New Century Financial Corporation) starting in 2004. This change follows the monetary policy tightening implemented by the Fed in the spring of 2004, which resulted in an adverse shock to the large portfolio of loans New Century was holding for investment. New Century reacted to this shock by massively resorting to deferred amortization loan contracts (“interest-only” loans).

Financial institutions, when in distress, may take excessive risk. Because they do not bear the losses in case of failure, shareholders of distressed banks have a natural preference for risky lending, fueling asset bubbles, banking crises and prolonged recessions. Our goal in this paper is to provide direct evidence of risk shifting in a large financial institution. To this end, we use the internal records of a major subprime originator, New Century Financial Corporation (NC). NC is a good candidate to study risk-shifting: in 2004, its payout ratio went up from 5% to 90%, consistent with textbook asset substitution. Against this background, our project-level (loan-level) data allows us to accurately characterize the distortion in risk preferences induced by financial distress.

We start with a simple illustrative model that characterizes project choice in financial distress. In our model, financial distress distorts shareholders’ preferences towards investments that payoff more in the states of nature where the firm does not default. A simple calibration exercise shows that this distortion can be sizable even under a relatively low probability of bankruptcy: projects with large negative NPV can be selected by shareholders provided they have a sufficiently strong covariance with the firm’s survival. In other words, the firm needs not be near bankruptcy to risk-shift. We apply the basic insights of our model to New Century (henceforth NC), the second largest subprime mortgage originator in the U.S. during from 2004 to 2007. We first document that NC was at risk financially as early as 2004. Since 2003, a significant fraction (about 20%) of NC’s originations was kept on its balance sheet as long-term investment. As a result, its leverage increased dramatically from 2002 to 2004, up to 90% in 2004. The monetary tightening implemented by the Fed at this time impaired NC’s franchise value, mostly for two reasons: (1) a massive balance sheet mismatch, as many loans paid fixed rates but were financed with flexible rates (2) an increase in the repayment risk of flexible rate mortgages made to subprime households whose monthly repayments would predictably increase. Regulatory shocks, competition, as well as the progressive saturation of the real estate market, further increased the pressure on New Century’s shareholders. An indirect indication of financial distress is the remarkable increase in payout ratio, from 6% until 2003 to 95% in 2005Q1.

Because of the large loan portfolio held on its balance sheet, NC’s survival was inherently tied to real estate prices. In this context, our model delivers three precise testable predictions. First, when closer to financial distress, NC should originate loans whose repayments are more sensitive to house price growth. The economic intuition is direct: because NC would be bankrupt in case of a collapse in real estate prices, loans with a high exposure to real estate prices were more attractive to NC shareholders. This explains an important change in NC’s issuance policy in 2004, namely the issuance of “interest-only” loans, which grew massively from 2 to 20% of total originations in 2004. Due to their delayed amortization feature, these loans exhibit a massive increase in due repayments 24 months after origination. As a result, many borrowers need a refinancing when the interest-only period expires. However, refinancing is only possible if the borrower has built enough equity in the house, which for an “interest-only” loan happens only through price increases (as borrowers do not repay the principal). Therefore, the repayment of interest-only loans depends more strongly on real estate prices than for other types of loans. This feature of interest-only loans is evident when we analyze NC’s internal records on repayments. Supporting this risk-shifting interpretation of the switch to interest-only loans, we find a revealing modification of wording (within an otherwise similar paragraph) in the 2004 10k compared to the 2003 one: In its description of criteria used to approve interest-only loan applications, New Century erased in the 2004 10k an important restriction that used to be applied before that year in the evaluation of the repayment ability of borrowers.

NC was not isolated in this increased use of deferred amortization loans. Using information from 10K filings, we find that originators with more “skin in the game” (i.e. originators who held more loans as long-term investment in 2003) originated larger amounts of deferred amortization loans in 2005. In this small cross-section, exposure to the 2004-2006 monetary shock thus correlates well with the later origination of “real estate price contingent” loans. This is consistent with our risk-shifting interpretation.

Our model also predicts that NC should originate more loans in regions where real estate prices are more correlated with the return of its “legacy” assets. Again, the intuition is that NC was doomed to be bankrupt if its legacy assets’ returns turned out to be low. NC shareholders were thus to benefit more from loans with payoffs highly correlated with its legacy assets’ return. To boost this correlation should originate relatively more loans in regions where real estate prices correlate strongly with the returns of its legacy assets. In the data, we proxy for the returns on NC’s legacy assets with a weighted-average real estate index, where the weight for each MSA is the share of NC’s portfolio held in that MSA. We then define an MSA exposure to NC’s legacy assets as the point estimate of a regression of real estate inflation at the MSA level on this proxy for NC’s legacy assets’ return and call it β^NC. As predicted by theory, this coefficient is a strong predictor of the geographic dispersion of loan origination.

Finally, our model predicts that NC should originate more price-sensitive loans in regions whose prices are the most correlated with its legacy assets. This is a direct consequence of our first two predictions. NC survives if its legacy assets do well. This implies high real estate prices in regions with a high correlation with NC’s legacy assets. This in turn implies that price-sensitive loans in these regions perform well. Thus, these loans in these regions have superior returns for shareholders who are only interested in returns contingent on survival. In the data, NC did originate more interest-only loans in regions with a higher β^NC.

Our paper may have implications for monetary policy. In response to a heating real estate market, policy makers thought in 2004 that increasing interest rates was the appropriate response. Our paper suggests this decision had unintended consequences: by pushing mortgage originators closer to financial distress, the monetary tightening led mortgage originators to increase risk-taking. In the case of New Century – and probably of other originators who held large amounts of loans on their balance sheet – risk-taking meant enhancing exposure to real estate price risk. This may well have fuelled the real estate bubble and eventually accentuate the burst of this bubble.

The full paper is available for download here.