Monetary Policy and Financial Conditions

Monetary Policy and Financial Conditions

August 18, 2026

• Policy Hub: Macroblog
• Publication
• Banking
• Financial Markets
• Macroeconomy
• Monetary Policy

Financial conditions play a central role in the transmission of monetary policy, and broad measures of financial conditions are often used as proxies for characterizing the stance of monetary policy (see, for instance, here and here). Sometimes, however, financial conditions indices appear somewhat inconsistent with the policy rate path set by the central bank. One such example is the dynamics of US financial conditions around the 2022-23 tightening cycle by the Fed. Figure 1 plots the effective fed funds rate (EFFR) and the Federal Reserve Board's measure of financial conditions (the financial conditions impulse on growth index, commonly called FCI-G), excluding the contribution by the policy (FFR) rate, from January 2021 to May 2026. The remaining components of FCI-G are the 10-year Treasury yield, the 30-year fixed mortgage rate, the triple-B corporate bond yield, the Dow Jones total stock market index, the Zillow house price index, and the nominal broad dollar index. Higher values of FCI-G indicate tighter financial conditions for economic activity (as measured by real GDP growth) over the following year. (The Goldman Sachs financial conditions index, which allows the analysis in this Macroblog post to be performed at a higher (weekly) frequency, yields similar results.)

Figure 1 shows that financial conditions started to tighten rapidly even ahead of the Fed's hiking cycle and peaked at the end of 2022. For the remainder of the Fed's tightening cycle and until September 2024, financial conditions eased markedly. Financial conditions continued to ease in the most recent period, with the FCI-G index in May 2026 at levels when the hiking cycle commenced in March 2022. This seeming discrepancy can be reconciled in part by the possibility that the "neutral" rate-the real policy interest rate that is neither stimulating nor restricting economic activity and with inflation anchored at the central bank's target-might have increased so that the current policy rate no longer exerts sufficient control over financial conditions. Another possibility is that another factor, beyond the policy rate, might be driving the observed dynamics of financial conditions. This factor could be related, for instance, to spillovers from global financial conditions (see here) or to the central bank's balance sheet. In this blog post, I entertain this latter possibility and show evidence consistent with the central bank's balance sheet occupying a prominent role in explaining changes in financial conditions.

Figure 2 plots a summary measure for January 2021 to May 2026 of the Fed's liquidity provision-the inverse of the share of banks' reserve in total Fed liabilities-along with the FCI-G index shown in figure 1. An increase in this balance sheet measure indicates tightening of the aggregate liquidity, while declines are associated with more abundant liquidity.

Unlike the policy rate depicted in figure 1, the Fed's liquidity measure is much more aligned with both the tightening (late 2021 through late 2022) and easing (early 2023 through late 2025) of the financial conditions index. But why is this a reasonable summary measure of the Fed's balance sheet dynamics?

Importantly, this measure captures the effects of both the size and composition of (the liability side of) Fed's balance sheet. For example, keeping the composition constant, a reduction in Fed security holdings would lead to a one-to-one reduction in reserves (pure size effect). On the other hand, keeping the size constant, a reduction in other liabilities (Treasury general account, overnight reverse repo program, or ONRRP) would lead to a one-to-one increase in reserves (pure composition effect). All other combinations of size and composition effects fall in between these two extremes.

It is instructive to elaborate a bit further on the compositional changes as their effect on liquidity is more complex and less obvious. Take, for example, the sharp drop in the blue line in figure 2 that occurred between February 2023 and March 2024, which is when the Fed was pursuing a runoff of its security holdings (Treasuries and mortgage backed securities) in its SOMA portfolio. But instead of this balance sheet runoff (quantitative tightening) leading to a commensurate reduction in reserve balances by banks, the banks' reserves at the Fed actually increased (by about $600 billion) over the February 2023-March 2024 period. The main reason for this was that the reduction in the SOMA portfolio did not drain "active" reserves held by banks but drained "dormant" reserves held by nonbank money market funds in the Fed's ONRRP as issuance decisions by the Treasury made holding short-term debt more appealing by offering a higher rate of return than ONRRP did. In fact, the RRP declined by around $1.5 trillion during this period, while the Fed's security holdings decreased by about $1 trillion. This boosted the level of bank reserves, which was further supported by the creation of the Bank Term Funding Program (BTFP) liquidity facility as an emergency response to the regional banking crisis in March 2023. These elevated bank reserves likely contributed to an easing of financial conditions through a portfolio-rebalancing channel, by which banks rotated out of reserves into other assets, resulting in a further increase of credit and liquidity in the financial system.

The discussion in this blog post so far suggests that the monetary policy stance, which drives financial conditions, might be a function of both the policy rate and the balance sheet proxy of its size and composition. In what follows, I will construct a simple index of monetary policy stance that better reflects the policy mix of these two components that might reinforce or, conversely, negate each other. In constructing this summary measure, the two components (EFFR and reserves as share of total liabilities) are standardized and aggregated with assigned weights of two-thirds on the policy rate and one-third on the balance sheet component. (This weighting is consistent with the Fed's framework for implementing monetary policy in which the policy rate serves as the primary tool for adjusting the monetary policy stance.) I experimented with more statistically sophisticated methods for selecting weights, but I found that these fixed weights provide a reasonable, conservative baseline for characterizing the overall monetary policy stance.

Figure 3 presents the constructed monetary policy index (MPI), along with the 10-year/2-year Treasury yield spread, which is commonly used as a proxy for current and future financial and economic conditions, over the longer period of January 2010 to May 2026. The MPI shows the monetary policy stance gradually tightening over this period, peaking in August 2023. Since then, the MPI declined from 1.85 to 0.92 in May 2026, a level still slightly above the previous peak from the middle of 2019. The (negative) correlation of MPI with the yield spread is surprisingly strong, suggesting that a flattened or inverted yield curve accompanies the tightening of the monetary policy stance. The likely explanation is that tightening monetary policy conditions and liquidity results in a "risk-off" market sentiment with flight-to-safety flows directed at the long end of the yield curve.

Next, I jointly model the MPI, FCI-G, and the 10-year/2-year spread for the period of January 2010 to May 2026. Statistically, these three variables are characterized by a similar degree of persistence and strong co-movement. In an estimated VAR(3), the lagged MPI helps predict both FCI-G and the spread, even after controlling for the lags of the other variables. Figure 4 reports the impulse response functions (IRFs) to a unit MPI shock from local projections (based on three lags of each variable and robust standard errors). In the figure, an MPI tightening shock is associated with a flattening or inversion of the yield curve and an increase in the financial conditions index.

The IRFs in figure 4 show strong responses-statistically significant, despite the small sample size and large number of estimated parameters-of the yield spread and FCI. The IRF for the spread peaks at −1.62 percent after 13 months, while the IRF for the FCI-G peaks at 1.34 standard deviations after 16 months. After the peak, the responses die down gradually, although the effect on FCI-G remains large and statistically significant after 24 months. The reported evidence on the strong co-movement of the constructed monetary policy index with measures of financial conditions highlights the need for further refining this signal for the purposes of better monitoring the transmission of the Fed's policy actions to the financial markets and the real economy.