Firm Balance Sheet Liquidity, Monetary Policy Shocks, and Investment Dynamics
📄 Summarized from the full manuscript · Human-reviewed for faithfulness before publication
In brief
Do highly leveraged firms cut investment most after a rate hike, or is it firms with little cash on hand? Using Compustat data and identified monetary shocks, this paper finds it is the cash-poor firms: once liquid-asset holdings are controlled for, leverage loses its power to predict which firms pull back investment. A model with fixed costs of issuing long-term debt explains why -- it breaks the usual assumption that cash is just "negative debt," generating financially healthy but cash-poor firms that invest almost every extra dollar they receive. These firms hold about a third of aggregate capital, which the paper shows can triple the impact of cash-flow-based stimulus.
What this paper finds — and why it matters
This paper asks whether corporate leverage or corporate liquid-asset (“cash”) holdings better predict which firms’ investment is most sensitive to monetary policy, challenging the conventional treatment of cash as simply “negative debt.” Using local projections (following Jordà 2005) on a Compustat panel and monetary policy shocks identified from high-frequency fed funds futures surprises, the paper finds that both higher leverage and lower liquid-asset ratios independently predict weaker capital accumulation for two to three years after a contractionary shock (a 10-percentage-point higher leverage ratio or 10-point lower liquid-asset ratio predict roughly 0.2 and 0.4 percentage points of additional cumulative capital-growth slowdown, respectively), but once both variables are included jointly, leverage loses its statistical significance while liquid assets remain robustly predictive – evidence, the paper argues, that the cross-sectional negative correlation between leverage and cash holdings creates an omitted-variable bias in leverage-only specifications common in the earlier literature. To explain why liquidity, not leverage, is the more robust predictor, the paper builds a heterogeneous-firm general equilibrium model in the Khan-Thomas (2013) tradition, modified with a fixed cost of issuing long-term debt, which breaks the standard equivalence between cash and negative debt: because debt issuance is lumpy, many firms optimally choose to hold sizeable cash buffers between infrequent debt issuances rather than paying down debt to zero, so their liquid-asset position, not their current leverage, governs how exposed they are to a monetary tightening’s effect on borrowing costs and cash flow. Calibrated to match firm-level investment, debt-issuance, and lifecycle facts, the model reproduces the paper’s central empirical pattern – liquid assets subsume leverage’s explanatory power once both are included – something a conventional liquid-debt version of the same model (without fixed issuance costs) cannot do, since there every firm with nonzero debt remains continuously exposed to the current borrowing rate, keeping leverage a strong independent predictor. The fixed issuance costs also generate a class of relatively large, high-net-worth firms that are not borrowing-constrained yet optimally choose not to raise new debt or accumulate cash in a given period – exact corporate analogues of the “wealthy hand-to-mouth” households of Kaplan and Violante (2014) – with a marginal propensity to invest out of a transitory cash windfall of essentially one; in the calibrated model these “liquidity-constrained” firms hold roughly a third of aggregate capital, versus under 8% held by firms at a binding borrowing constraint. Because of this large mass of high-marginal-propensity-to-invest firms, the model’s response to an unexpected, deficit-financed government transfer to the corporate sector is roughly three times larger on impact than in a recalibrated model without issuance costs, illustrating that balance-sheet liquidity, not leverage, is the more relevant margin for the transmission of shocks and policies that operate through firms’ cash flows.
Summary of a classic paper, AI-assisted and human-reviewed. See the linked original for the authoritative claims and full conditions.
Note on source text: this paper is forthcoming/newly published in the Journal of Political Economy (DOI 10.1086/744130); the freely available full text used for this summary is the widely circulated June 2019 job-market-paper draft (CREI/UPF), since the final published text was not yet accessible. Core empirical and quantitative results are not expected to have changed in kind, but exact figures in the published version may differ from those quoted here.
Questions & answers
Q1. What is the paper’s central empirical puzzle, and how does it challenge the standard “financial accelerator” view of firm heterogeneity?
“The conventional macro-finance view regularly abstracts from the notion that firms’ decisions to accumulate liquid assets (‘cash’) are distinguishable from their management of debt. Cash is not ’negative debt’” (Introduction, p. 2). Where much of the literature (following Bernanke and Gertler 1989, Kiyotaki and Moore 1997, Bernanke, Gertler and Gilchrist 1999) uses leverage as the summary measure of financial distress, this paper asks whether firms’ separate liquid-asset holdings are a better predictor of investment sensitivity to monetary policy – and finds they are.
Q2. How does the empirical identification strategy work?
The paper estimates panel local projections (Jordà 2005) of firms’ fixed-capital growth on the interaction of within-firm-demeaned financial position (leverage or the liquid-asset ratio) with a high-frequency-identified monetary policy shock, controlling for firm and sector-by-quarter fixed effects (Section 2.3, eq. 1). Demeaning financial position within each firm, rather than using its raw level, “ensure[s] that our results are not driven by permanent heterogeneity in responsiveness across firms” – a choice explicitly motivated by the model of Section 3, in which firms are ex ante homogeneous (Section 2.5).
Q3. What are the headline magnitudes of the separate leverage and liquidity effects?
“A 10 percentage point higher leverage ratio or a 10 pp lower ratio of liquid assets to total assets predict approximately 0.2 and 0.4 pp slower cumulative growth of capital during the two to three years after a one standard deviation monetary policy contraction” (Introduction, p. 3). The leverage-based differences “become negative starting about 4 quarters after the shock and statistically significant 7 quarters after,” while liquidity-based differences peak around three years after the shock and are somewhat larger in magnitude (Section 2.4).
Q4. What happens when leverage and liquid assets are controlled for jointly, and how does the paper interpret this?
“When simultaneously controlling for liquid asset holdings, the relevance of leverage in explaining differences in firms’ capital accumulation responses over the medium run disappears… [while] there are no significant changes in the explanatory power of liquid assets” (Section 2.4). The paper’s interpretation: “the negative correlation between leverage and liquid asset holdings in the cross-section of firms leads to an omitted variable bias in the leverage regression… cash holdings more consistently predict heterogeneous investment responses to monetary policy shocks over the horizon under consideration” (Introduction, p. 2).
Q5. What is the key model innovation, and what problem does it solve?
The model embeds heterogeneous firms with collateral constraints (in the Khan and Thomas 2013 tradition) but adds a fixed cost of issuing long-term debt, which “breaks the equivalence between cash and negative debt on a firm’s balance sheet” (Section 4.2.3). Because issuing new debt is lumpy and costly, firms do not continuously adjust debt to keep net worth (cash minus debt) at a single optimal level; instead they let cash and debt drift between infrequent issuance episodes, so a firm’s current liquid-asset ratio – not its current leverage – becomes the more informative state variable for predicting its exposure to a change in borrowing costs.
Q6. Why can’t a conventional (liquid-debt) heterogeneous-firm model reproduce the paper’s central empirical pattern?
Re-solving the model with liquid long-term debt and no issuance costs, “controlling for liquid asset holdings, high leverage retains considerable negative predictive power for firms’ capital accumulation after a contractionary monetary policy shock, with very high statistical significance,” unlike in the baseline (Section 4.2.3). The reason: “because debt is liquid, every firm with nonzero debt is continuously ‘attached to’ the debt market, actively responding to borrowing rates, in turn making leverage a strong predictor of sensitivity to borrowing cost fluctuations” – exactly the mechanism the fixed issuance cost is designed to break.
Q7. How well does the calibrated model reproduce the empirical estimates in a monetary policy shock experiment?
Simulating a 25-basis-point contractionary policy shock through the calibrated general-equilibrium model (with empirically estimated paths for the risk-free rate and the borrowing spread), the joint regression on model-simulated data shows leverage loses significance while a 10-point-lower liquid-asset ratio still predicts about a 0.15 pp stronger contraction in capital during the first year, “slightly below half the quantitative effect seen in the empirical estimates” (Section 4.2.2). The paper notes the model’s cross-firm differences build up more slowly than the interest-rate shock paths themselves, “echo[ing]… Bernanke and Gertler (1989)”: financial frictions and net-worth dynamics add persistence not just to the aggregate economy’s response but to cross-sectional differences in firm behavior as well.
Q8. What are “liquidity-constrained” (LC) firms, and how do they relate to Kaplan and Violante’s (2014) “wealthy hand-to-mouth” households?
LC firms are defined as those “who are currently not issuing new long-term debt but are simultaneously not acquiring any cash nor paying dividends” – by construction not borrowing-constrained, “yet they invest all available sources of liquid funds… into capital” (Section 5.2). The paper draws the parallel explicitly: such firms “appear as firms with strong balance sheets, yet might at the margin exhibit a marginal propensity to invest of virtually unity – an exact analogue of the wealthy hand-to-mouth consumers of Kaplan and Violante (2014),” and cites corroborating firm-level evidence from Zwick and Mahon (2017), who find investment by low-liquid-asset firms “more than twice as responsive” to cash-flow-generating tax incentives as that of high-cash firms.
Q9. How large is the aggregate importance of these high-MPI firms, and what does the fiscal-transfer experiment show?
In the baseline calibration, only 7.6% of aggregate capital is held by firms at a binding borrowing constraint, while “a third of aggregate capital is in the hands of firms” at the liquidity constraint (Table 3, Section 5.2). Simulating an unexpected, lump-sum-tax-financed government transfer to the corporate sector (a shock used by Bernanke, Gertler and Gilchrist 1999 to illustrate their financial accelerator), the paper finds “the impact effects on aggregate output, labor, and investment are about three times larger in the case with issuance costs” compared to the recalibrated no-issuance-cost model, “since a significant fraction of capital is held by firms with high marginal propensities to invest” (Section 5.2).
Q10. How does this paper’s leverage/liquidity distinction relate to Ottonello and Winberry’s (2020) default-risk findings, and what does the paper conclude for policy?
The paper notes in Appendix B.2 that Ottonello and Winberry (2019/2020), using the same within-firm-demeaning approach on leverage alone, find “significantly weaker responses in capital stocks for up to a year after a monetary shock” that “become insignificant afterwards,” whereas this paper’s liquid-asset-based measure remains predictive over a longer, two-to-three-year horizon (Section 2.5). The conclusion draws out a policy implication directly: “the efficacy of fiscal policy and investment subsidies could be increased by targeting firms with the highest liquidity needs, instead of blanket transfers or targeting those with high leverage” (Section 6), while noting further empirical work is needed – including accounting for unused credit lines alongside cash holdings – to operationalize such targeting.
Key terms in this paper
Definitions below follow the paper's own usage.
- Fixed debt issuance costs
- The paper's central model device (Section 3): firms issuing new long-term debt pay a fixed cost, which "breaks the equivalence between cash and negative debt on a firm's balance sheet." Because issuance is lumpy, firms optimally let cash and debt drift between infrequent issuance dates rather than always holding zero net debt, so a firm's current liquid-asset ratio -- not its leverage -- best summarizes how exposed it is to fluctuations in the current cost of external finance. The paper shows a conventional model with liquid debt and no issuance costs cannot generate this disconnect: "because debt is liquid... every firm with nonzero debt is continuously 'attached to' the debt market," making leverage a strong predictor there too (Section 4.2.3).
- Liquidity-constrained (LC) vs. borrowing-constrained (BC) firms
- Two categories of firms in the calibrated model (Section 5.2): firms "at a binding borrowing constraint" (BC), who are actively raising debt up to their collateral limit, versus firms "at a liquidity constraint" (LC), who are not currently issuing new debt but also are not acquiring cash or paying dividends -- so by construction they are not borrowing-constrained yet invest every marginal dollar of liquid funds. In the baseline calibration, only 7.6% of aggregate capital is held by firms at the BC, while fully "a third of aggregate capital is in the hands of firms" at the LC.
- Marginal propensity to invest (MPI) out of cash windfalls
- The paper's explicit corporate analogue (Section 5.2) of Kaplan and Violante's (2014) "wealthy hand-to-mouth" households: LC firms "appear as firms with strong balance sheets, yet might at the margin exhibit a marginal propensity to invest of virtually unity," because fixed issuance costs make it optimal for them to hold cash buffers well below their collateral limits rather than pay down debt or accumulate more cash, leaving them fully exposed to transitory "cash windfalls" -- good productivity draws, government transfers, or temporary cost declines -- at the margin.
- Within-firm demeaning of financial position
- The paper's specification choice (Section 2.3-2.4) of interacting monetary shocks with within-firm-demeaned financial position, x_{jt-1}-E_j[x_{jt}], rather than its raw level, "to ensure that our results are not driven by permanent heterogeneity in responsiveness across firms" -- motivated by the model's assumption that firms are ex ante homogeneous. The paper shows this choice is not innocuous: Ottonello and Winberry (2019) demeaning leverage the same way "predicts significantly weaker responses... for up to a year" while this paper's longer investment-response horizon and its liquid-asset variable together yield the paper's distinct conclusion that liquidity, not leverage, dominates over 2-3 years.