During the period between 2008 and early 2022, developed countries experienced low and relatively stable inflation rates and monetary policy rates. However, the end of the pandemic was marked by a sharp increase in inflation rates above central banks' target rate. Most of these central banks responded with rapid increases in policy rates intended to reduce inflation and to keep inflation expectations from becoming unanchored. Yet, the prolonged period of low rates followed by a fast increase in interest rates would reduce the market prices of many financial assets and create strains within the financial system. With this backdrop, the Atlanta Fed's Center for Financial Innovation and Stability and the Center for the Economic Analysis 
of Risk at Georgia State University organized the conference Interest Rate Variability and the Financial Sector during October 19–20, 2023, as an opportunity to bring together economists, regulators, and industry representatives to discuss the latest research on the effects of interest rates on the financial system.
This post reviews some of the highlights from the conference, which included sessions related to the exposure of banks and other financial institutions to interest rates, the implications of such exposure for the effectiveness of monetary policy, and the evidence on the effects of interest rate variability on financial stability.
Financial institutions' exposure to interest rate variability
The first day of the conference focused on the effect of higher interest rates on financial firms' balance sheets, with some additional considerations about the consequences for their liquidity. Along with causing a reduction in longer-duration asset values, an increase in interest rates also causes an increase in the value of longer-duration liabilities and might increase or decrease a bank's derivatives portfolio. Yet, the accounting values of most bank assets and liabilities are not marked to market in response to changes in interest rates. This practice creates the potential for large declines in bank asset values, which may or may not be offset by gains in the derivative portfolio and by gains (decreases in the present value) of liabilities. Moreover, although the rate paid on many liabilities of financial firms (for example, bank deposits) is relatively insensitive to changes in market rates, the holders of these liabilities have the right to withdraw their funds on demand or over a short horizon. Thus, financial firms' ability to offset losses on their assets with gains on their liabilities depends in part on the stability of their funding.
Measuring bank interest rate risk
The paper Monetary Tightening and U.S. Bank Fragility in 2023: Mark-to-Market Losses and Uninsured Depositor Runs? by Jiang, Matvos, Piskorski, and Seru sought to measure individual bank exposure to the sort of deposit run that triggered the collapse of Silicon Valley Bank. Their first goal is to assess the reduction in bank capital that would occur if banks' assets were marked to market to reflect the increase in rates from March 2022 to March 2023. The second goal is to ask whether the bank could be subject to a self-fulfilling run—that is, where the mark-to-market value of the assets is insufficient to cover the face value of the insured liabilities after a run by the uninsured depositors. The paper calculates the number of exposed banks for varying percentages of uninsured depositors running, taking into account that in many cases not all uninsured depositors would run. The results indicate that if 50 percent of the uninsured depositors ran, then 186 banks would be insolvent (with assets amounting to $300 billion) and—in the extreme case of 100 percent withdrawal—more than 1,600 banks would be insolvent (with total assets of $4.9 trillion).
The paper Interest Rate Risk in the U.S. Banking Sector by Abdymomunov, Gerlach, and Sakurai considered two other ways of measuring banks' exposure to interest rate risk: net interest margin (NIM) and economic value of equity (EVE) during the period between 1997 and 2021. These alternative measures of interest rate risk are widely used by banks. NIM is interest earnings minus interest expense, and it reflects gains and losses from interest rate changes spread through the life of the assets and liabilities rather than recognized immediately when rates change. EVE is the present value of assets minus the present value of liabilities. The paper finds that banks' interest rate sensitivity increased over the sample period and that smaller banks were more sensitive to higher rates.
Keynote by Mark Flannery
Mark J. Flannery focused his keynote talk on the effect of higher interest rates on US banks' capital. His presentation, based on a paper written with Sorin M. Sorescu and titled Partial Effects of Fed Tightening on U.S. Banks' Capital , measures the banks' asset losses during 2022. Their paper finds that the unrealized losses on banks' bonds and loans was $1.1 trillion before taxes, compared with banks' book value of equity of $2.2 trillion. Based on these results, Flannery determines that if unbooked losses were recognized, half of US banks holding half of total assets would fail to meet their required minimum capital ratios at the end of 2022. Flannery then observed that the "800-pound Gorilla" was whether the gains on lower-rate deposits would be sufficient to cover the asset losses. He concluded by observing that although there is reason to think deposits are, on average and over time, sufficiently sticky to adequately hedge the losses, this is not true in all time periods, especially the more recent periods.
Bank risk hedging
Although the interest rate losses on assets in 2022 are relatively easy to measure, the possibility exists that these losses were hedged. Greenwald, Schulhofer-Wohl, and Younger presented a deeper dive into the potential for deposits serving as a hedge in their paper Dynamic Deposit Betas: Implications for Monetary Policy and Financial Stability 
. The paper focused on what bankers call the "deposit beta;" that is, the percentage change in the deposit rate for a given change in the federal funds rate. The deposit beta is often assumed to be a fixed percentage resulting in a linear relationship between the funds rate and the average rate paid on deposits. The paper finds that the relationship is not linear, with deposit rates increasing at a faster pace at higher rates. This result has a positive implication and suggests that monetary policy becomes more effective as rates increase. However, given the above discussion, it has the negative implication that interest rate gains on deposits may be less effective in offsetting the asset losses than simpler models would suggest.
The analysis suggests banks may have taken substantial mark to market interest rate losses on their asset portfolio that are not fully offset by gains on their deposits. The paper Do Banks Hedge Using Interest Rate Swaps? by McPhail, Schnabl, and Tuckman considers the possibility that banks are hedging their exposure using interest rate derivatives. This paper examines the use of interest rate swaps by the top 250 Federal Deposit Insurance Corporation (FDIC)–insured banks from the third quarter of 2017 to the fourth quarter of 2019. The paper finds that banks are very active in the swap market, but mostly as dealers for other participants. Banks' net exposure to interest rate swaps (which accounts for offsetting positions) is small relative to their overall exposure. The paper concludes that, as a prevailing practice, banks don't rely on swaps to hedge interest rate exposure of their assets.
Measuring risk exposures with equity data
The papers discussed above use financial accounting data to measure interest rate risk. An alternative source of data to measure that risk is the market price of the firm's common equity stock. One advantage of equity price data is that it is available at a higher frequency (daily as opposed to quarterly for most accounting data), which allows for the analysis of the higher-order moments of the distribution of risk exposure. The paper Bank Convexity Risk 
by Cremers, Gandhi, and Yang uses equity price data to estimate the exposure of banks to interest rate volatility as proxied by the implied volatility on the options contract on the Treasury five-year futures. The results suggest that banks are exposed to the volatility of rates, and that banks with greater exposure tend to have more volatile balance sheets in terms of their quantities of bank deposits, credit, and liquid assets.
Another advantage of using equity data is that it captures investors' views about the risk exposure and management of publicly traded financial firms. The paper Measuring Interest Rate Risk Management by Financial Institutions by Brunetti, Foley-Fisher, and Verani examines interest rate risk management of insurance companies. The paper finds that insurance companies generally have effective interest rate risk management but that, among insurers, life insurance companies are less sensitive to interest rate risk than property and casualty insurance companies.
Fixed-income mutual funds
Banks are not the only type of intermediary that invests in debt securities funded by demandable obligations. Bond and money market funds also follow this strategy. Thus, a natural question is how this balance sheet structure relates to their returns and liquidity. The paper Hidden Duration: Interest Rate Derivatives in Fixed Income Funds by Choi, Kim, and Randall examines the extent to which fixed income mutual funds hedge their interest rate exposure using derivatives over the period between July 2019 and September 2022. The paper finds that that although fixed-income funds widely use interest rate derivatives (IRDs), there is variation across funds and time periods. The funds that speculated on rates using IRDs outperformed in declining-rate environments but suffered worse performance during the rate hikes of 2022. Moreover, funds that speculate with IRDs tended to double down on that strategy after taking losses on their derivatives. Finally, government bond funds and investment-grade bond funds experienced more outflows after negative fund returns.
The paper Similar Investors by Georg, Pierret, and Steffen addresses liquidity management by banks and money market funds. The failure of Silicon Valley Bank highlighted the risk to banks and their depositors from situations where the depositor outflows are likely to be highly correlated. Such correlations may arise from prime money market funds—large investors in unsecured bank obligations such as certificates of deposit and commercial paper. The paper measures similarity by looking at the extent to which pairs of money funds are investing similar proportions of their funds in the same set of security issuers (financial and nonfinancial firms). The paper finds that if several money funds are investing in the same bank at the same date, the funds that are more similar tend to subsequently reduce their exposure to that bank. However, this response only applies to money fund investments in the obligations of financial firms and not for nonfinancial firms.
Banks and monetary policy
The session on banks and monetary policy had three papers that analyzed banks' responses to monetary policy actions. The paper Monetary Policy Transmission through Bank Securities Portfolios by Greenwald, Krainer, and Paul examines the extent to which unbooked losses in banks securities portfolio affected the supply of new loans during the period between the first quarter of 2021 and the first quarter of 2023. The sample consists of 30 banking organizations that are subject to the Federal Reserve's stress test. The extent to which banks' regulatory capital is affected by the gains or losses in their securities portfolio depends upon the bank's stated reason for holding the security. The gains or losses on securities held as part of the trading portfolio flow through to bank capital. The change in securities' values held in the available for sale (AFS) portfolio generally does not affect a bank's capital, with an important exception that the changes do flow through to capital at the very largest banks. Finally, the changes in the held to maturity (HTM) portfolio have no effect on bank capital. This paper finds that losses in the AFS portfolio result in reduced credit, especially the unhedged losses at the largest firms. Moreover, this reduction in credit has a real effect on small firms' investment. Overall, these results suggest that monetary policy would be more effective if all banks had to recognize securities losses in their AFS portfolio.
Cao, Dubuis, and Liaudinskas find that firms' relationships with their bank also influence the way in which monetary policy affects bank lending in their paper Relationship Lending and Monetary Policy Pass-Through . This paper uses Norwegian administrative tax and bank supervisory records to track firms' relationships with their banks. The empirical results suggest that starting at low policy rate levels, the effect of further rate cuts is smaller for firms with long-term bank relationships, and the paper provides a theoretical model that rationalizes these results. Overall, the paper argues that the lengths of bank-firm lending relationships are an important variable for aggregate monetary policy transmission.
The last paper of the session was Banks' Risks Exposures and the Zero Lower Bound by Schneider, which examines banks' risk-taking behaviors as the central bank's policy rate approaches the zero lower bound (ZLB). The paper begins with the observation that rate movements toward the ZLB reduce returns from maturity transformation and reduce the spread between deposits and the rate on reserves. The paper shows that risk-averse banks may hedge this deterioration in their investment opportunities by taking more risk as rates decline even though this means they will incur losses when rates increase.
Monetary policy and financial stability: Historical evidence
The papers in this session study the links between monetary policy and financial stability over the longer run, leveraging data of historical crises for developed countries since the late 1800s. In brief, they argue that a fast increase in monetary policy rates following a long period of low rates materially increases crisis risk.
The first paper, Monetary Policy, Inflation, and Crises: Evidence from History and Administrative Data by Jiménez, Kuvshinov, Peydró and Richter, argues that the path of monetary policy rates matters before a financial crisis, especially if this path is U-shaped. In particular, interest rate hikes increase the likelihood of crises if they were low for a long period before the hike. The reason why this U-shaped pattern induces crisis dynamics is because low-for-long monetary policy rates boost credit and asset prices, generating high-growth periods ("red zones"). The combination of low-for-long rates and "red zones" imply that when rates increase again, credit risk rises and nonperforming loans hit intermediary balance sheets, increasing the likelihood of banking crises. Furthermore, the U-shaped monetary policy rate path is an empirical regularity before financial crises but not before noncrisis recessions.
The second paper in the session, Loose Monetary Policy and Financial Instability by Grimm, Jorda, Schularik and Taylor, put forth a similar argument: that long periods of expansionary monetary policy combined with a sudden reversal of interest rates can increase crisis risk. Using macrofinancial data and a banking crisis chronology for 18 countries during 1870–2020, the paper estimates that a decline in monetary policy rates by 1 percentage point over a five-year period raises the likelihood of a financial crisis by 5.5 percentage points in the next five to seven years and by 15.5 percentage points in the following seven to nine years. Furthermore, crisis dynamics are generated by credit booms, house price booms, and their interactions ("red zones").
The core message of the papers in this session was that short-run gains from accommodative monetary policy (such as recovery from a crisis and a boost in the credit-to-GDP ratio) need to be carefully weighed against the medium-term increased risk of financial crises. Thus, while calibrating these effects, policymakers could consider "leaning against the wind"—that is, deviating from the central bank's inflation target to account for financial stability purposes. Not doing so may lead to significant financial stability costs.
Consequences of "leaning against the wind"
Goldberg and López-Salido kicked off the last session of the conference with the paper "Sowing the Wind" Monetary Policy , which studies whether, when, and how central banks should lean against the wind. The paper also highlights the mechanisms by which leaning against the wind can worsen both the boom and the bust. It models an economy where boom periods are characterized by financial speculation and leveraged risk taking, followed by busts when aggregate demand and asset prices decline as the share of optimist investors drops. The paper argues that the central bank should "lean against the wind" with tighter monetary policy only a little at the start of the speculative boom and more aggressively over time as speculation continues. It should avoid leaning unexpectedly or systematically across cycles, because such behavior could worsen recessions through announcement and anticipation effects. The paper shows that these effects reduce welfare as they induce significant interest rate and asset-price volatility. Thus, "leaning monetary policy may sow the wind."
The conference concluded with a paper titled Interest Rate Hikes, Collateral Deterioration and Search for Yield: Evidence from Shadow Banks by Casu and Gallo. This paper studies the asset-backed commercial paper (ABCP) market before the 2007 financial crisis to understand the links between monetary policy tightening and risk taking of nonbanks. It focuses on a particular type of nonbanks—ABCP conduits—that perform maturity transformation similar to banks but have no deposit franchise and are funded by highly runnable liabilities. The paper uses new data on collateral at the conduit level, between January 2003 and March 2007, to document an increase in net issuance of ABCP while collateral worsened. Thus, it emphasizes a mechanism by which interest rates hikes before the 2007 crisis were associated with increased risk taking of nonbanks amid higher funding costs and increasing demand for "safe assets" (such as triple-A rated notes). Overall, the paper emphasizes that strong demand for "safe assets" can lead to search for yield outside the formal banking system and thus affect the stability of the broader financial sector.
Conclusion
The studies presented at the 2023 CEAR-CenFiS conference addressed research topics related to the consequence of interest rate fluctuations for financial stability. Rising interest rates and falling asset prices have tested the resilience of the financial sector in the recent period. Many of the papers discussed the interest rate exposures of banks and other financial institutions and the interactions that such exposures generate between monetary policy and financial stability. Such issues will continue to be top of mind for policy makers in the near future.
By Larry Wall, research center executive director of the Center for Financial Innovation and Stability,
Camelia Minoiu, research economist and adviser, and
Yuritzy Ramos, a research analyst, all in the Atlanta Fed's Research Department