Working Papers

Job Market Paper: An Asset-Price Centric New Keynesian Model

Latest version: February 2026

Abstract Main Results PDF BibTeX

This paper develops a New Keynesian model in which asset wealth effects drive consumption dynamics. Due to information frictions, agents cannot perfectly distinguish aggregate from idiosyncratic asset-price movements, leading to wealth illusion. The marginal propensity to consume (MPC) out of capital gains and the MPC out of income jointly determine equilibrium consumption. With MPCs calibrated directly to their empirical estimates, the model closely replicates U.S. consumption during the 1998--2018 boom-bust cycle. The framework offers a unified mechanism for the transmission of discount-rate shocks, in which realized asset prices serve as sufficient statistics for consumption responses. Applied to monetary policy, it reconciles the gap between micro and macro estimates of the elasticity of intertemporal substitution (EIS). Second, it provides a tractable approach to monetary policy that targets asset valuations. Asset prices determine aggregate demand and balance-sheet policies are modeled via an Asset-Price Taylor rule. Third, a heterogeneous-agent extension reveals that inequality is a key determinant of macroeconomic volatility, in contrast to standard HANK models. Policies reducing inequality also stabilize the macroeconomy.

BibTeX Citation

@article{gong2025asset,
  title={An Asset-Price Centric New Keynesian Model},
  author={Gong, Zheng},
  year={2025},
  month={October},
  note={Working Paper}
}

Main Results

×

Notation: \(MPC^S\), \(MPC^H\) are MPCs out of stock and housing capital gains; \(MPC^Y\) is the MPC out of income. \(A^{S,*}/C^*\), \(A^{H,*}/C^*\) are wealth–consumption ratios. \(\hat P_t^S\), \(\hat P_t^{H}\), \(\hat C_t\) are % deviations of stock prices, house prices, and consumption. Model-implied consumption is given by

\[ \hat C_t = \frac{MPC^S}{1-MPC^Y}\frac{A^{S,*}}{C^*}\hat P_t^S + \frac{MPC^H}{1-MPC^Y}\frac{A^{H,*}}{C^*}\hat P_t^{H}. \]

Notes: Correlation between data and model-predicted consumption: 0.802. Left: S&P 500 and Case-Shiller U.S. National Home Price Index. Right: Actual per capita real consumption (solid blue) and APNK model prediction (dashed red). Data sources: Shiller (stock/housing prices); FRED (PCEC, HNOCEA, BOGZ1LM653064155Q, HNOREMV). All series are converted to per capita real terms.

Notes: The AP-HANK model's response to the Great-Recession asset-price shock. Units: (i) asset price index and (ii) aggregate consumption response: percentage deviations from 1998 levels; (iii) Liquid wealth bottom-50% share and (iv) consumption bottom-20% share: percentage points. Vertical lines mark the approximate boom, crisis, and recovery phase boundaries. The high-liquidity scenario is where the adjustment probability of illiquid wealth is 0.999. The wealth tax scenario is discussed in Section 5.3 where 1% annual tax is applied to illiquid wealth over $3 million.

Notes: Unconditional standard deviations of output and consumption from long simulations for calibrations with different levels of aggregate MPC out of income. MPCs are reached by varying the liquid-asset return, variance of the transitory income shock, and illiquid-asset adjustment probability while re-calibrating the subjective discount factor to clear the market.

❮ ❯

Figure 1: Asset-price shock IRFs (placeholder)

When and How Does Household Heterogeneity Matter for Aggregate Fluctuations?

Latest version: September 2025

Major update coming soon!

Abstract Main Results PDF Code BibTeX Dec 2024 ver

I establish a distributionally neutral benchmark for aggregate shock transmission in incomplete‑market heterogeneous‑agent (HA) economies, where all agents are equally exposed to the shock. In this benchmark, aggregates satisfy the equilibrium conditions of a fictitious representative‑agent (RA) economy. Leveraging this result, I develop a tractable framework to identify and quantify redistribution mechanisms that drive the divergence between HA and RA outcomes. The framework (i) uncovers the mapping from deep structural parameters to redistribution and (quantitatively) to general‑equilibrium dynamics; (ii) clarifies the roles of fiscal policy and investment; (iii) provides rescalable sufficient statistics portable across shock types; and (iv) identifies new redistribution channels in two‑asset HANK and overlapping generation models.

BibTeX Citation

@techreport{gong2025does,
  title     = {When and How Does Household Heterogeneity Matter for Aggregate Fluctuations?},
  author    = {Gong, Zheng},
  year      = {2025},
  number    = {No. crctr$224\_2025\_624$},
  institution = {University of Bonn and University of Mannheim, Germany},
  note      = {Discussion Paper}
}

Main Results

×

Notes: “The Power of Forward Guidance Revisited” McKay, Nakamura and Steinsson (2016) consider an HA economy’s response to a one-time real-rate cut in Quarter 20. The left panels decompose their HANK model’s output response into RANK and redistribution effects. The right panel further decomposes the redistribution effects into the contribution of interest rate, (labor) income, and tax exposure channels"

Notes: “The Intertemporal Keynesian Cross” Auclert, Rognlie and Straub (2024) discuss fiscal multipliers in a two-account quantitative model. The left panels decompose their model’s consumption responses to a government spending shock (1% of the steady-state output) into RANK and redistribution effects. The right panel decomposes the redistribution effects into the contribution of redistribution channels.

\[ \begin{align*} -\omega(z^t)&=\overbrace{\underbrace{(b^{*}(z^{t-1})-B^*)(R^{ra}_{t}-R^*)}_{\text{interest rate exposure}}+\underbrace{(\tau^{*}(z^t)-r^*B^*) -(\bar\tau^{ra}(z^t)-r_t^{ra}B^*)}_{\text{tax exposure}}}^{\text{net interest rate exposure}} \nonumber \\ &+\overbrace{\underbrace{(\hat y^{L,ra}(z^t) - \hat Y_t^{L,ra})y^{L,*}(z^t)}_{\text{labor income exposure}}+\underbrace{(\hat D_t^{ra}- \hat Y_t^{ra})D^{*}\left(v^*(z^{t-1})-\frac{y^{L,*}(z^t)}{Y^{L,*}}\right)}_{\text{income portfolio exposure}}}^{\text{income exposure $(\hat y^{ra}(z^t) - \hat Y^{ra}_t) y^{*}(z^t)$}} \nonumber \\ &+\underbrace{\hat C_t^{ra}(y^*(z^t)-c^*(z^t))-\hat P_t^{ra}P^*(v^*(z^t)-v^*(z^{t-1}))}_{\text{saving flow exposure}}\nonumber\\ & + \underbrace{( b^*(z^t)-b^{ra}(z^t))-R^{ra}_t( b^*(z^{t-1})-b^{ra}(z^{t-1})) + (\bar\tau^{ra}(z^t)-\tau^{ra}(z^t))}_{\text{liquidity (bond)}}. \end{align*} \]

Notes: Individual-level redistribution induced by an aggregate shock to the HA economy, defined with individual characteristics and the RA economy's response to the shock.

The contribution of redistribution (channels) to total consumption responses

Literature	Redistribution	Net Interest Rate Exposure		Income Exposure		Liquidity
		Interest Rate	Tax	Portfolio	Labor	Bond Supply	Illiquid Assets
Werning (2015)	0	0	0	0	0	0	N.A.
McKay, Nakamura and Steinsson (2016)	-99%	25%	-44%	N.A.	-80%	0	N.A.
Bilbiie (2020)	33%	0	0	N.A.	33%	0	N.A.
Auclert, Rognlie and Straub (2024)	143%	11%	-4%	-7%	0	160%	-17%
Wolf (2021); Wolf (2023); Angeletos, Lian and Wolf (2023)	100%	0	0	N.A.	0	100%	N.A.

Notes: The contribution of redistribution (channels) to total consumption responses. In the quantitative models, the total consumption responses are calculated as the sum of consumption responses over the period from 0 to 300. The effects of “saving flow exposure” are negligible and not shown.

❮ ❯

MNS2016

Fiscal Constraints on Monetary Policy: How Debt Limits Monetary Effectiveness

with Christian Bayer and Keith Kuester

Latest version: February 2026

Draft available upon request

Abstract Slides

Standard fiscal theory of the price level (FTPL) emphasizes the necessity of long-run fiscal backing for monetary stabilization. We demonstrate that when public debt provides liquidity services, the timing of this backing is a first-order determinant of monetary transmission. By influencing the natural rate of interest, the trajectory of the debt stock modulates the economy's response to rate shocks. While front-loaded fiscal consolidations amplify monetary contractions by depressing the natural rate, delayed adjustments raise the natural rate, dampening the policy effect. Using a tractable bond-in-utility framework, we show that the standard FTPL is a limiting case of infinitely delayed adjustment. Quantitative analysis in an incomplete-markets OLG model reveals that a twenty-year delay in fiscal adjustment is sufficient to fully neutralize monetary policy. Consequently, effective stabilization requires active coordination between monetary policy and the timing of fiscal adjustment.

Tacit Collusion of Partial Cross Ownership Under Cournot Competition

Latest draft: February 2020

Abstract PDF BibTeX

Partial cross-ownership (PCO) among firms affects their incentives to engage in collusion. I analyze the collusion behavior in an n-firm industry featuring a cross-ownership network, under Cournot competition. Increasing PCO can hinder tacit collusion under the uniform output distribution scheme. However, this scheme is not always feasible for collusion. I examine different subgame perfect equilibria and conclude that tacit collusion is more likely to be facilitated when PCO increases.

BibTeX Citation

@article{gong2020tacit,
title={Tacit Collusion of Partial Cross Ownership Under Cournot Competition},
author={Gong, Zheng},
year={2020},
month={February},
note={Working Paper},
url={https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3306203}
}

---

​Research Notes

"An Example of Two-side Commitments in Bayesian Persuasion"​, 2021

I give an example in which one player's optimal commitment is also her response to another player's optimal commitment in Bayesian persuasion. I conjecture that the Sender's optimal commitment is a dual problem of the Receiver's optimal commitment (and vice versa) when there is a positive probability that the players cannot respond optimally. I also discuss the interpretation of this restriction in realistic settings.

"Inequality and Monetary Policy in a Lucas Island Model"​, 2021
I study how the heterogeneity in marginal propensities to earn (MPE) affects output's response to money supply shocks in a Lucas Island model. The simplicity of the Lucas Island model allows me to obtain an analytical solution and solely focus on the role of MPE. Relative to the benchmark case in which wealth inequality is absent, the output’s response is ambiguous. However, when incorporating a realistic correlation between unemployment status and nominal wealth, the output response is amplified. Inflation-induced redistribution from wealthy to poor households increases aggregate labor supply, as the labor supply of poor (and unemployed) households remains unresponsive to redistribution. This creates a positive correlation between MPE and redistribution, driving the amplification.