3. New Lessons for Macroeconomics and Finance Theory

The consensus antecedent to the GFC had been founded in the theoretical propositions of models that consider variants of financial frictions only on the borrowers’ side of the credit markets as in, for example, Bernanke and Gertler (1999, 2000, 2001), Filardo (2000), Cecchetti et al. (2000), Batini and Nelson (2000), Gilchrist and Leahy (2002), and Iacoviello (2005).

Surveys of this channel and early empirical results (mainly for the United States) include among others Gertler (1988), Bernanke (1993), Ramey (1993), Gertler and Gilchrist (1993), Kashyap and Stein (1994), Bernanke and Gertler (1995), Cecchetti (1995), Hubbard (1995), Bernanke et al. (1996), and Kashyap and Stein (1997).

Detken et al. (2010) build their argument on propositions similar to Assemacher-Weche and Gerlach (2010) and the verity of “Lucas critique”; the latter is a criticism of econometric models used for policy evaluation which do not qualify for optimal decision rules of economic agents varying systematically with policy alterations. The “Lucas critique” addresses the use of past data to estimate relevant statistical relationships to forecast the effects of novel policy implementation, since the estimated regression coefficients are not invariant, but will change along with agents’ decision rules in response to the new policy (Ljungqvist, 2008). In this vein, Detken et al. (2010) recognize the inability to defend the empirical evidence seemingly supporting the argument that “LATW monetary policy is too blunt an instrument” (since the negative effects to the real economy are of a greater magnitude than the effects on asset prices). They accept that “only in a policy regime in which the principle of LATW has explicitly been adopted could a test be conducted regarding the effectiveness of the signalling and herd-breaking channels” (p. 322).

See, for example, Ioannidou et al. (2009), Altunbas et al. (2010), Buch et al. (2010), Delis and Brissimis (2010), Delis and Kouretas (2011), Maddaloni and Peydró (2011), Delis et al. (2017), Paligorova and Santos (2012), and Dell’Ariccia et al. (2013).

Rajan (2005) identifies an institutional factor-driven search for yield that results in higher risk-seeking behavior by to maintain yields after rates on safer asset decline.

They contend that relatively small short-term interest rate changes can lead to large risk-taking effects from financial intermediaries.

In Borio and Zhu (2008), different valuation effects lead to riskier profiles, as, for example, the use of collateral that gains value from an expansive monetary policy.

As in Dell’Ariccia et al. (2010), cheaper short-term debt due to accommodative monetary policy raises levering incentives of financial intermediaries, in addition to asset risk incentives through interaction with banks’ limited liability.

See, for example, Smaghi (2009), Blanchard et al. (2010), Mishkin (2011a, 2011b), Eichengreen et al. (2011), Baldwin and Reichlin eds. (2013), and Smets (2014).

Ikeda (2022) recognizes that “the advancement of the literature on monetary policy in asset price bubbles lags far behind the literature on rational bubbles, … [and] monetary policy analyses in asset price bubbles in a New Keynesian framework—a standard workhorse of monetary policy analyses for many central banks—remain scarce” (p. 1570).

Evidence to this point is given by the extraordinary unconventional actions that central banks took during the GFC to support the functioning of financial markets (the US Troubled Asset Relief Program (TARP) is an early vivid example) (see Wood, 2015, and also Oda & Okina, 2001, for an early account on central bank actions to support the functioning of segmented markets at times of a crisis).

Woodford (2011a, 2011b) introduces banks’ leverage target levels, which give rise to a positively sloped loan supply curve that shifts procyclically with banks’ profitability and capital, as well as with changes in the policy rate. He contends that such a loan supply curve may arise due, for example, to intermediaries’ facing costs for originating and servicing loans, with marginal costs increasing with the volume of lending, or to regulatory limits or market-based constraints restraining leverage. Similarly, Woodford (2011a) introduces a risk-taking channel of monetary policy by applying a regime as in Stein (2012), where fire sales during a financial crisis distort the behavior of financial intermediation, into an otherwise traditional New Keynesian model of monetary policy.

Smets (2014) distinguishes two main channels for this effect to hold: (1) a need for a stronger involvement in distributional policies (see, for example, Brunnermeier & Sannikov, 2013) and a resort to quasi-fiscal operations (see, for example, Pill, 2013), which call for both a greater monetary policy accountability and a political involvement that both jeopardize central bank independence and give rise to political pressure, and (2) the presence of time inconsistency problems for monetary policy as the monetary authorities may end up providing more liquidity than needed for long-run price stability if the fundamental problems of debt overhang following a financial crisis are not addressed.

Borio (2012, 2013) stress that the adverse outcome of that form of inconsistency applies also to prudential, as well as fiscal policies, while Smets (2014) formally addresses the concept with respect to the macroprudential authority’s setting its policy, taking the monetary policy reaction into account in a context similar to Barrett et al. (2008).

Adam S. Posen (2023, April 15) hosting the annual “Macro Week” in Peterson Institute for International Economics (PIIE) in Washington (a series of speeches and onstage discussions by central bankers and finance officials) reiterates the above statement claimed by most central banking officials (especially European ones) during the annual Macro Week.

For a brief elaboration on the “Lucas critique,” see footnote 28.

It is contended (Mishkin, 2011a; Smets, 2014) that a lexicographic ordering may protect the central bank from an inflationary bias stemming from its involvement with financial stability; such an ordering implies that the central bank assigns primary focus to the price stability objective than the financial stability one (while also ensuring the attainment of financial stability concerns). Nevertheless, as Barrett et al. (2008) propose, modeling the trade-off between price and financial stability such an inflation bias may be addressed by rule-based LATW monetary policy. Smets (2014) additionally asserts that “such a credible mandate of the monetary authorities will also give the right incentives for the macroprudential policymakers to lean against the build-up of leverage and growing imbalances and not rely on inflation to solve their problems” (p. 291).

See, for example, De Bandt et al. (2009), Brunnermeier et al. (2009), and European Systemic Risk Board (2011) for accounts on macroprudential policy objectives and definitions of systemic risk.

See, for example, Buiter (2009), as well as Svensson (2015a, 2015b) who argues that the Swedish experience of macroprudential concerns having adversely affected the conduct of monetary policy is an example of this problem.

Using aggregate data, Taylor (2007) has argued that excessively low policy rates led to the housing bubble, while contrary arguments are provided in Posen (2009), Bernanke (2010), Bean et al. (2010), and Turner (2010).

Summers (2014, p. 69) views the regime of low real interest rates combined with low inflation (at the time) leading to low nominal interest rates, to stir risk-seeking activities by investors and greater reliance on Ponzi finance (leading to increased financial instability). These risk-seeking incentives may arise from behavioral aspects like money illusion, as a result of which the managers believe that low nominal rates indicate that real returns are low, encouraging them to purchase riskier assets to obtain a higher target return (Mishkin 2011a, p. 99). Typically, incentives to search for yield arise either from contractual arrangements or from fixed rate commitments. The former compensate asset managers for returns above a minimum level, and when nominal interest rates are low, high-risk investments may be the ones that lead to high compensation. The latter (often present in insurance companies) force the firm to seek out higher-yielding, riskier investments.

With the exception that this mechanism stems from lender-originated financial frictions (rather than from borrowers), it is closely related to the financial accelerator of Bernanke and Gertler (1999, 2000) and Bernanke et al. (1999).

In spring 2013, US Federal Reserve Chairman Ben Bernanke gave an initial statement that he expected QE asset purchases to be tapered off in a staggered manner to eventually wean the economy off the extra support, which was referred to as the “taper tantrum” illustrating the impact of effective central bank communication. The statement prompted strong reactions (“tantrums”) in financial markets not just in the United States but also worldwide, which were probably unintentional and undesired. The US market fell by 4% following the announcement, setting off an international chain reaction. A policy statement that should have ideally been factored in started causing global markets to react negatively as investors, digitally apt to disseminate information, overreacted promptly. Several advanced economies experienced a sharp reversal of capital flows and currency depreciation at the time, while several emerging economies (like India, Indonesia, Brazil, and Turkey) experienced a sharp decline in stock markets and higher sovereign yields. Global economies cushioned the shock in 2013, while the Fed tapered its asset purchases in January 2014.

For research supporting the risk-taking channel of monetary policy both antecedent to the GFC and after its emergence see, for example, Jiménez et al. (2008), Ioannidou et al. (2009), Adrian and Shin (2009, 2010b); Maddaloni and Peydró (2011), Buch et al. (2014), Jimenez et al. (2014), Heider et al. (2019), Bubeck et al. (2020), and ECB (2021).

If monetary policy is used to lean against credit bubbles and therefore aiming at stabilizing the financial sector, in addition to the real sector, a main objection refers to the issue of asking one policy instrument to do two “jobs,” thus violating the Tinbergen (1939) principle. Ensuring stability in the financial sector is closely related to stabilizing the overall economy, as financial instability can lead to economic instability and inflation. However, since the nature of financial instability is distinct from the dynamics of inflation and economic activity, it can be considered a separate policy goal under the Tinbergen principle. This is because macroprudential supervision (a distinct tool) can be used to stabilize the financial sector. It thus considered preferable to utilize macroprudential supervision for maintaining financial stability while leaving monetary policy to concentrate on maintaining price and output stability (Lagarde, 2023).

In a seminal account, Borio and Lowe (2002) argue on the buildup of financial imbalances under low inflation regimes, in that sustained rapid credit growth combined with large increases in asset prices appears to increase the probability of an episode of financial instability. Adrian and Shin (2010a) show that marked-to-market leverage is strongly procyclical. They demonstrate that changes in dealer repos (the primary margin of adjustment for the aggregate balance sheets of intermediaries) forecast changes in financial market risk as measured by the innovations in the Chicago Board Options Exchange Volatility Index (VIX), thus showing that aggregate liquidity can be seen as the rate of change of the aggregate balance sheet of the financial intermediaries. Similarly, several methodologies rely on specific asset price processes using time series price data. Asset price process may be assumed to follow a simple diffusion model as in Jarrow et al. (2011a, 2011b), according to whom to detect bubbles, a Feller’s test for explosions is applied, which is estimated using a kernel method. A stochastic disorder model is specified in Shiryaev et al. (2014, 2015) that constitutes a more complex diffusion process where the drift and volatility parameters jump at change points. Phillips et al. (2015a, 2015b) develop bubble tests that look for explosive dynamics in long time series of asset prices using variants of an augmented Dickey-Fuller test, under an autoregressive framework.

Another strand of research exploits option-price cross-sectional data, which contain useful information about the state-price distribution (SPD) of the underlying asset which is related to the risk-neutral measure; the risk-neutral distribution and density function are identified by the partial derivative of a European option price with respect to the strike price (Breeden & Litzenberger, 1978). For nonparametric methods that recover the SPD from option data, see, for example, Aït-Sahalia and Lo (1998), Aït-Sahalia and Duarte (2003), Yatchew and Härdle (2006), Fan and Mancini (2009), Kitsul and Wright (2013), Song and Xiu (2016), Jarrow and Kwok (2021), and Lu and Qu (2021). In principle, estimates of the asset price bubble are obtained by deducing the fundamental asset value (proxied by the mean of the SPD).

Agur (2018) demonstrates in a game theoretic context that the coordination problem vanishes only when the unconstrained player (monetary authority) is so similar to the constrained player (macroprudential authority—in terms of the imprecision of macroprudential policy) that he prefers the same equilibrium.

For surveys, see Claessens et al. (2013), Galati and Moessner (2014), and Kokores (2022).

Following Svensson (2015a, 2015b), at the initial period, the central bank moves to an interest rate increase that reduces financial vulnerabilities and leads to higher unemployment for 3–4 years. Once unemployment reaches its steady-state level, the second period begins by defining the probability whether the economy is hit by a crisis, which is lower provided that interest rates were raised at the initial period. If a crisis emerges, the economy faces significantly higher unemployment, for a period equal to, or longer than, the initial period. In the end, to estimate the effect on welfare, unemployment is squared over both periods. In fact, it is the deviations of unemployment from the natural rate, which can be assumed to be zero to simplify computations, that is actually squared.

Blanchard and Galí (2007) referred to “a divine coincidence” demonstrated by many models, namely, the fact that to bring inflation to target coincided with bringing output to target as well; most models introduced wage rigidities or cost-push shocks and thus have exhibited some (rather not substantial) trade-offs (see Blanchard, 2006, for a discussion, or Goodfriend, 2004). Bayoumi et al. (2014) argue on the indications that the trade-off may be weakening.

By averaging two model variants, namely, one being a proxy of a large, closed economy, with another of a small open economy, we yield that effects peak after 1–2 years, while unemployment returns to steady-state peak after 3–4 years.

For a survey, see Altavilla and Ciccarelli (2009). Other models used by central banks—each with distinct particular features—are rather tailored to small open economies; an example is the model used by Sweden’s Riksbank, labeled the “Ramses” model, that estimates unemployment to react to monetary policy shocks in a more sluggish manner.