I am an applied macroeconomist with research interests in the intersection of monetary economics, financial economics and time-series analysis. Methodologically, my work combines careful attention to the identification of causality, measurement, and economic theory. Two topics are of particular interest: first, the identification of monetary policy shocks in structural models; second, the incorporation of expectations into a structural VAR (SVAR). They reflect the application and theoretical endeavor in my research. Thematically, I have studied questions motivated by the major macroeconomic developments of the past two decades, including the 2000s housing cycle in the U.S., the role of the financial sector and other forces in the Great Recession, the monetary policy response to the zero lower bound, the enhanced transparency of the Federal Reserve’s communication.
In what follows, I summarize all of my published and working articles since 2020.
Identification of monetary policy shock in Structural VAR
In the aftermath of the Great Recession, monetary policy shock is particularly challenging to identify for three reasons. The information in the FOMC monetary policy announcement becomes increasingly complicated and multidimensional. Meanwhile, the conventional policy instrument, i.e., the federal funds rate, is uninformative at zero lower bound. Lastly, monetary policy turns out to be more informational and forward-looking. I take up these and related challenges in a series of articles.
Improving identification strategies is beneficial to gain insight in monetary policy transmission.
Shifting attention to the longer-end of the yield curve is a natural extension of the latest research on monetary economics, as short-term interest rates are insufficient to reflect the Fed’s actions on the entire yield curve. For instance, forward guidance and large-scale asset purchases (LSAP) programs explicitly aim to affect longer-term interest rates. High-frequency financial data can facilitate the identification of discretionary monetary policy shocks from the noisy fluctuation of a long-term rate. For example, futures and options on Treasury yields imply the real-time optimal expectations of interest rates and their risks given an efficient financial market. Price variations of those derivatives around monetary policy announcements render us inference of policy impacts. The working paper, “The Long-term Rate and Interest Rate Volatility in Monetary Policy Transmission” (Chen, 2020), utilizes high-frequency options data to measure the impact of monetary policy announcements on risk perception in the long-term bond market. Based on a Structural VAR with high-frequency identification, I identify monetary policy shock in the risk-taking channel by taking the fact that when the long-term bond market perceives higher volatility after an FOMC policy announcement, the long-term bond yield, particularly the term premium, escalates. This MP shock transmits to the real economy through the financial sector – i.e., the credit premium added on corporate lending rate surges, and real production suffers in the following year. In comparison, an unanticipated jump of the three-month ahead federal funds futures also drives up the long-term interest rate, but it stimulates muted responses in credit frictions and real activities. The results impose a question mark on the validity of the Keynesian-type interest rate channel, in which short-term interests do the heavy lifting in monetary policy transmission. Meanwhile, I collect evidence for the effectiveness of the risk-taking channel, in which risk perception of financial intermediaries plays a central role in policy propagation.
A critical unresolved issue in Chen (2020) is that the expectations formed in financial markets may be irrational. Then how to incorporate the theory-consistent rational expectations into a small-sized SVAR with only observables? The working paper, “Embedding Rational Expectations in a Structural VAR: Internal and External Instruments for Set Identification (Chen and Valcarcel, 2022a)”, proposes a novel approach that embeds Rational Expectations (RE) into a low-dimensional structural vector autoregression (SVAR). Rational expectations are imposed on all economic agents, such as households, firms, the monetary authority and the financial sector. We establish an instrumental variable procedure internal to the SVAR founded on a purely theoretical framework, which does not rely on any mapping strategy to a reduced form. Alternatively, a separate strategy considers data external to the SVAR to aid in identifying structural shocks on a purely empirical basis. We report clouds of responses from a RE-consistent theoretical model as well as regions of plausible responses from the empirical approach. We conclude that a Taylor Rule characterization of monetary policy shocks remains relevant when the theoretical RE-SVAR is properly augmented with information from fluctuations - or momentous events - in markets that have garnered increased attention since 2008, such as reserves and various money markets.
Another avenue for identification is to revisit the information contained in monetary aggregates and their components.
Money growth was once used as the monetary policy target but lost its appeal as a policy instrument in the 1990s. However, growing empirical evidence indicates that the decoupling of money and monetary policy may be a measurement problem rather than an identification issue. The work-in-process paper, “A Note on the Relative Stability of Money Demand”, revisits the controversy of unstable money demand function in the U.S. We find that if taking into account the substitution effect among monetary assets in the measurement of monetary quantity, we observe a stable long-run relationship between money demand and the costs of holding in the whole sample from 1967 to 2018 and different subsamples. Thus, it is necessary for the aggregation of monetary assets to reflect the substitution effect when a quantity-theoretical approach to monetary policy is of concern.
A vital question in identifying monetary policy shocks with monetary aggregates is to understand the role of monetary aggregates in the monetary authority’s decision rule. Are they a supplementary source of information besides the federal funds rate or are they directly indicative of monetary policy actions? The published paper, “Monetary Transmission in Money Markets: The Not-So-Elusive Missing Piece of the Puzzle (Chen and Valcarcel, 2021, Journal of Economic Dynamics and Control, lead article), investigates the effects of U.S. monetary policy shocks from alternative policy indicators for a modern sample encompassing 1988–2020. The choice of the Wu and Xia (2016) shadow federal funds rate leads to persistent price puzzles. These puzzles arise despite the inclusion of the usual suspect fixes such as commodity prices, federal funds futures and forward rate data. We find they occur at monthly and quarterly frequencies. We consider alternative indicators with the same broad monetary aggregates Keating et al. (2019) employed in their investigation of a historical sample. They provide a consistent resolution of the price puzzle and do not require the ad hoc inclusion of commodity prices or futures data. This price puzzle correction is not a feature of our time-varying approach as it obtains from constant parameter econometric estimation. Our analysis suggests monetary policy has transmitted substantial expansionary effects in money markets in the aftermath of the 2007 Financial Crisis and the decade that followed.
In the same strand of research considering alternative policy indicators, the Working paper, “An Explicit Structural Model of Forward-Lookingness in Money Rules and Taylor Rules (Chen and Valcarcel 2022b)”, proposes a novel structural identification that assumes a policy feedback rule, like the Taylor rule, is appropriate. Then we generate a cloud of impulse responses of macro variables associated with this feedback rule. By adjusting structural parameters in the feedback rule and observing simulated impulse responses, we manage to answer the question of how model uncertainty affects estimation. In 5625 different specifications of the policy feedback rule, we find that the Wu and Xia (2016) shadow rate, as the policy indicator, gives rise to a large set of puzzling responses of the price and real output. Substituting the policy indicator with Divisia M4 or M2 incurs a far smaller incidence of puzzle responses. Our results suggest that the superiority of Divisia monetary aggregates in indicating monetary policy is not a consequence of favorable structural assumptions but an outcome of the comprehensive information content about policy actions in money.
Other explorations in structural identification
Is there a limit on the federal debt level? It is an unresolved question from the supply side. Congress may constantly expand the statutory debt limit. To answer this question, we need additional insight into the demand side where economic agents hold the Treasuries as an all-weather safe asset. The debt limit may be the level at which investors would be reluctant to purchase more. The working paper, “Safety Premia, Treasury Duration and Safe Asset Aggregation (Chen and Valcarcel, 2022c)”, measures the optimal quantity of Treasuries considering the safe asset holder’s decision. We discusses the sources and user costs of safety services obtained from holding Treasuries as the representative safe asset. Furthermore, we aggregate the quantity of safe assets by tracking the utility generated by the U.S. Treasury securities and the expenditure on safety services. Rather than simply sum up the quantity of outstanding Treasuries in different durations, we impute heterogeneous user costs and aggregate those imperfectly substituted safe assets using the Divisia superlative indexing method. This Divisia safety service index reflects the Treasury debt valuation from a dynamic trading perspective, in which the holder trades safe assets at different durations as imperfect substitutes. At the same time, the simple-sum aggregate represents the “buy and hold” perspective, in which holders are indifferent to Treasuries at various maturities. We further provide evidence that the discrepancy between the Divisia and simple-sum aggregates captures the insurance service flows, with which the holder buys and sells safe assets at different durations to insure against idiosyncratic risks in liquidating non-safe assets. This insurance service has a critical implication on how economic agents deal with news on fundamentals with noise.
The working paper, “Housing Price Rigidity: Evidence from Asymmetric Impacts of Money Supply (Chen and Ming, 2022)”, finds that the non-linear relationship between monetary policy and housing prices—i.e., the housing price downward rigidity—suggests the resilience of housing price in response to an easing monetary environment at downtime. We use a novel measure of the money supply (i.e., Divisia M2) to indicate monetary policy actions and estimate a bivariate threshold vector error correction model (TVECM) on housing prices. This model comes with three innovations. First, the quantity theory of money is embedded in the cointegration equation by a structural restriction on the cointegrating coefficient. Second, with proper identification on the long-run relationship, the wealth and liquidity effects of monetary policy are isolated via the hierarchical design in VECM, such as the long-term cointegration and short-term error correction. Third, the degree of liquidity and the threshold for non-linearity is endogenously determined by the deviation from the cointegration relationship. The threshold variable is the residuals of the cointegration equation, measuring the deviation of housing appreciation from its long-term cointegrating relationship with money growth. A modification behavior is evident in the lower regime, indicating housing price increases to reflect a liquid monetary environment. This finding suggests the resilience of housing prices with the aid of expansionary monetary policy at a downtime. In the upper regime, error correction is not observed and the housing appreciation persists even in a tight credit availability. This result is consistent with the autonomous housing boom independent of the monetary development.