2.1 Households

The representative household derives utility from consumption (C_t) and experiences disutility associated with labor (N_t) according to the following separable utility function:

where 𝔼 represents the expectation operator, β ∈ (0,1) is the discount factor, and h > 0 is the weight attached to the disutility from labor.

Each household supplies labor N_t to a firm and receives its after-tax wage bill (1-{\tau }_t^N)W_t{N}_t, where W_t is the real wage and {\tau }_t^N denotes the labor income tax. Households own all the initial corporate shares S_t, with the price per stock (equity wealth) given by q_t. The equity price describes the market valuation of assets outside the firm and is synonymous to the firm’s value. Ownership of the firm’s stocks entitles the household to earn an after-tax dividend per share of {\overline{D}}_t\equiv (1-{\tau }_t^N)D_t^a with {\tau }_t^D standing for the dividend tax and D_t^a the dividend net of corporate tax. At the beginning of the period, the household also lends B_t to the firm at an intraperiod gross rate of R_t.⁵ The household’s budget constraint is:

with T_t denoting lump-sum transfers from the government. For S_t > 0, and taking taxes, dividends, equity prices, loan interest rate, and the real wage as given, maximization of (1) subject to (2) yields the respective first-order conditions with respect to C_t, S_t+1, B_t, and N_t:

where Λ_t is the Lagrange multiplier on the household’s budget constraint or the marginal utility of consumption. Equation (4) is a typical stock Euler equation, which shows that the firm’s external value is equal to the present discounted value of the future share price and the dividend net of corporate and dividend taxation. Equation (5) dictates the interest rate on lending to the firm, which is zero in net terms due to the intratemporal nature of corporate debt in this model. Condition (6) determines the optimal labor supply that varies along the extensive margin as in Hansen (1985).

Using (4), we retrieve the usual definition of the gross after-tax return to the firm’s asset:

Iterating forward on (7) and using the transversality condition yields the discounted share price equation only as a function of the after-tax dividend:

2.2 Production and Investment Policy

A representative corporate firm hires labor N_t, owns the capital stock K_t, and combines these two inputs to produce output Y_t according to the following constant returns to scale (CRS) technology:

with α ∈ (0, 1) standing for the share of capital in production, and A_t denoting a common economy-wide technology shock. The law of motion for the technology shock is:

where A is the steady state level of technology, ρ^A is the degree of persistence, and ${\epsilon }_t^A$ is a mean zero random shock with a normal distribution and a constant standard deviation denoted by ρ^A.⁶ The firm accumulates capital according to:

where δ ∈ (0,1) is the capital depreciation rate, and I_t denotes investment. Following Lucas and Prescott (1971) and Christiano, Eichenbaum, and Rebelo (2011), we introduce a quadratic capital adjustment cost \Phi \left(\frac{I_t}{K_t}\right)=\frac{\gamma }{2}{(\frac{I_t}{K_t}-\delta )}^2{K}_t that is denominated in units of current capital. The parameter γ > 0 governs the magnitude of adjustment costs to capital accumulation, with the increasing and concave function \mathrm{\Phi }\left(\frac{I_t}{K_t}\right) satisfying: Φ′ (•) > 0, Φ″ (•) ≥ 0, Φ(δ) = 0, and Φ′(δ) = 0. Thus, the firm must pay an increasing and convex cost of net investment, measured by deviations of I_t from the amount of investment required to replace depreciated capital. Relative to the steady state where I/K = δ, positive or negative alterations in net investment result in a cost that makes capital depreciate faster. The functional form for Φ(•) is chosen such that the steady state equilibrium is unchanged.

Without loss of generality, we assume that the total number of shares S_t is normalized to 1, with the firm having no access to issuing new stocks. To finance new capital investment and dividend distributions, the firm can use internal funds (retained earnings) or external debt financing from the household.⁷ In the case of the latter, the total investment loan is tied to the value of the collateralized capital stock. Particularly, for R_t = 1 and B_t = I_t we consider the following occasionally-binding borrowing constraint:

where 𝔼_t is the inside value of the firm’s capital stock (Tobin’s (1969) Q, derived below), and θ_t ∈ (0,1) is the proportion of capital used as collateral in order to obtain the investment loan, or alternatively the loan-to-value (LTV) ratio. The above collateral constraint can be motivated by a costly contract enforcement problem stating that if the firm cannot pay its debt, the creditor can take over the firm and seize its’ assets (Kiyotaki and Moore (1997) and Wang and Wen (2012)). As it is costly to liquidate capital after seizure, the lender retrieves only a fraction θ_t of the collateral asset value. Moreover, we allow for aggregate stochastic changes in the firm’s funding conditions and the associated degree of credit market frictions by assuming that θ_t varies according to the following AR(1) process:

with θ representing the steady state borrowing limit or the average stance of credit availability, ρ^θ the degree of persistence, and {ϵ}_t^{\theta } a mean zero random shock with a normal distribution and a constant standard deviation denoted by σ^θ. We interchangeably refer to this stochastic process as a financial, credit, LTV, or collateral shock, in the vein of Jermann and Quadrini (2012), Liu, Wang, and Zha (2013), and Perri and Quadrini (2018), among many others.

The firm’s before-tax dividend in period t is:

where corporate profits are defined as π_t = Y_t −W_t N_t. Denoting {\tau }_t^K as the corporate tax rate and using the intratemporal debt assumption with R_t = 1, the after-corporate tax dividend is:

Therefore, it is clear that investments and debt are expensed out of profits after corporate taxes are levied. With a dividend tax {\tau }_t^D, the firm maximizes the following present discounted value of the after-tax dividend payout ${\overline{D}}_t$:

subject to (9), (11), and (12). The term β^t (Λ_t /Λ₀) represents the firm’s stochastic discount factor and {\Lambda }_t=C_t^{-1} is the marginal utility of household’s consumption. The Lagrangian for the firm’s optimization problem is:

with λ_t denoting the Lagrange multiplier on the capital accumulation constraint (11), and φ_t the Lagrange multiplier on the borrowing constraint (12). The firm’s first-order conditions with respect to the choice of input factors (N_t, K_t+1) and investment (I_t) are:

The corresponding complementary slackness condition is:

Given the quadratic form of the capital adjustment cost, we rearrange equation (19) to obtain an explicit Q-theoretic investment function:

In a world with capital adjustment costs but without collateral constraints and dividend taxation, investment exceeds the depreciation rate when the value of newly installed capital, as measured by Tobin’s Q, is greater than 1. As is well known, Q_t represents the shadow price of capital that equals the Lagrange multiplier on the capital accumulation constraint. If γ > 0 and the marginal source of investment is new borrowing, the Qtheory equation implies that I_t is increasing in Q_t, and decreasing in the tightness of the borrowing constraint φ_t. Intuitively, investment is determined at the point where the firm is indifferent between investing in an additional unit of capital with price (1-{\Phi }_I^{\text{'}}(\frac{I_t}{K_t}\left)\right)Q_t after adjustment costs, and paying out dividends to the household with value 1 (1-{\tau }_t^D). The presence of an occasionally-binding collateral constraint (φ_t ≥ 0) raises the marginal cost of investment, leading the firm to accelerate dividend distributions in order to maintain the equality between the return to investment inside and outside the firm. Put differently, to achieve a higher level of investment, the shadow value of capital must increase in line with the marginal cost of investment.

Now, combine (18) with (19), and use (9), to derive the optimal physical capital Euler equation:

The left-hand side of (22) represents the current value of Q_t, whereas the right-hand side measures the discounted value sum of the future marginal product of capital net of corporate and dividend taxation, the reselling value of non-depreciated capital, future adjustment costs, and the option value of capital used as a collateral asset. Notably, for the credit-constrained firm, acquiring a marginal unit of investment via borrowing raises the anticipated value of capital and acts to relax the borrowing limit in the next period. The marginal benefit from a higher collateralized capital stock that can be used to secure future loans is represented by the term 𝔼_t(Q_t+1ϕ_t+1θ_t+1). The firm equates between the marginal cost and the expected marginal gains from investment, both of which are altered by the collateral constraint and dividend taxation.

To highlight the link between the ‘traditional’ and ‘new’ views of dividend taxation through the investment credit limit, observe from (22) that even a constant dividend tax rate ({\tau }_t^D={\tau }_{t+1}^D={\tau }^D) produces asymmetric effects on the marginal cost and benefit of investment when ϕ_t > 0. Conversely, for ϕ_t = 0 and ${\tau }_t^D={\tau }_{t+1}^D={\tau }^D$, the dividend tax drops out from (22), leaving the capital-investment outcome unchanged as implied from the ‘new’ view. Intuitively, the collateral constraint multiplier drives a wedge between the frictionless valuation of capital outside the firm, $(1-{\tau }_t^D)$ and the Tobin’s Q augmented for the adjustment costs in the credit-constrained economy (see (19)). When the marginal source of funds is determined by new external financing, a permanently lower dividend tax raises Q_t and the return to investment, which, in turn, lifts I_t. This connection between investment financing via debt and dividend taxation is in the spirit of the ‘traditional’ view.⁸ Below we derive the conditions under which the borrowing constraint is binding or non-binding, and show how the representative corporate firm responds differently to dividend tax changes depending on the value of θ and the initial steady state dividend tax rate.

2.3 Government

The government sets the labor income, corporate, and dividend income tax rates. Total revenue from the various taxes is used to finance lump-sum transfers to households subject to the following balanced budget:

We examine a fiscal policy disturbance in the form of a dividend tax shock only, while holding the corporate and labor income tax rates constant at their steady state values: ${\tau }_t^K={\tau }^{K}and{\tau }_t^N={\tau }^N$ Indeed, the aim of our paper is to analyze the macroeconomic effects of dividend taxation in the presence of an investment credit limit.⁹

2.4 Competitive Equilibrium

In a competitive equilibrium, the markets for labor, capital, dividends, debt, and stocks clear with S_t = 1 for all t. For the goods market clearing condition, we combine (2), (9), (11), (15), and (23) to obtain the economy-wide resource constraint:

The optimal conditions associated with the household’s, firm’s, and government’s behavior characterize the competitive general equilibrium summarized by (24) and the following equations:

Definition 1 (Competitive Equilibrium) Given the initial level of the capital stock (K₀) and shock processes ${\{{\theta }_t,A_t\}}_{t=0}^{\infty }$, a competitive equilibrium for the economy with an occasionally-binding credit constraint (φ_t ≥ 0) is defined as a sequence of dividend tax policies ${\left\{{\tau }_t^D\right\}}_{t=0}^{\infty }$, price vectors ${\{q_t,Q_t\}}_{t=0}^{\infty }$, and private sector allocations ${\{Y_t,C_t,N_t,K_{t+1},I_t\}}_{t=0}^{\infty }$, that satisfy equations (24)-(30).

3.1 The Long-Run Effects of Collateral Constraints and Dividend Taxation

In the non-stochastic steady state, all variables are constant and denoted without the time subscript. A detailed calibration part is presented later in the section before the short-run analysis, but in order to produce some of the figures in this subsection that support the theoretical arguments, we set β = 0.97, δ = 0.12, A = 1, N = 0.3, and a = 0.30. We also fix τ^K = 0.35 and T^N = 0.28, which approximately correspond with average long-run effective U.S. corporate and labor income tax rates, respectively.¹⁰

Proposition 1 The dividend tax rate τ^D and the borrowing limit 9 determine whether an economy is subject to a constrained or a slack equilibrium. In particular:

(i) If

then there exists a unique steady state constrained equilibrium (denoted by superscript B for ‘binding’) with

(ii) If

then there exists a unique steady state unconstrained equilibrium (denoted by superscript NB for ‘non-binding’) with φ = 0.

Proof 1. See Appendix ■

Figure 1 provides a visual representation of Proposition 1. The threshold between the constrained and the unconstrained equilibria lies in the region of empirically-plausible values of τ^D and θ.¹¹ The Lagrange multiplier φ is decreasing in the fraction of the value of capital that can be borrowed against, as a rise in θ makes the borrowing constraint less binding. Without dividend taxation, we must set θ^B < δ for the collateral constraint to bind. Introducing dividend taxation breaks down this relationship by lowering the market valuation of capital, and reducing the value of the collateralized capital stock, both of which result in the tightening of the borrowing constraint. Put differently, a hike in the dividend tax rate and/or a fall in the LTV ratio can move the long-run unconstrained equilibrium regime to a constrained one. The two regions create two different steady states that yield distinct values of the capital stock, Tobin’s Q, and equity prices. This is formally expressed in the following proposition.

Constrained (white) and unconstrained (grey) equilibrium regions.

Citation: IMF Working Papers 2022, 127; 10.5089/9798400214721.001.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Constrained (white) and unconstrained (grey) equilibrium regions.

Citation: IMF Working Papers 2022, 127; 10.5089/9798400214721.001.A001

• Download Figure
• Download figure as PowerPoint slide

Constrained (white) and unconstrained (grey) equilibrium regions.

Citation: IMF Working Papers 2022, 127; 10.5089/9798400214721.001.A001

• Download Figure
• Download figure as PowerPoint slide

Proposition 2 The steady state values of the capital stock, Tobin’s Q, and equity prices depend on the value of φ and therefore on whether the economy faces a constrained or an unconstrained credit regime. Specifically:

(i) If φ > 0 (i.e., binding region), the capital stock, Tobin’s Q, and equity prices are given by:

(ii) If φ = 0 (i.e., slack region), capital stock, Tobin’s Q, and equity prices are determined by:

Proof 2. See Appendix ■

The credit constraint φ acts to raise the firm’s marginal cost and Q by lifting borrowing costs, and driving a wedge between the internal and external valuations of capital. In order to maintain the same level of wealth, the shareholder requires an equity premium as reflected by the effective augmented rate of return on stocks 1 + φ (1-τ^D)^-1 (β^-1 – 1), that is increasing in the tightness of the borrowing constraint. In the binding steady state environment, a higher φ raises the spread between the frictionless share return, equal to the household’s rate of time preference (β^-1 – 1), and the stock return in the credit-constrained economy. As a result, the firm reduces the capital stock and investment, and increases dividend distributions when financial frictions become more prevalent; i.e., {\left(\frac{K}{N}\right)}^B<{\left(\frac{K}{N}\right)}^{NB} for φ > 0.

In the frictionless economy, the wedge between the physical capital stock and its market valuation is determined by the dividend tax only as seen from (39). A cut in τ^D raises the stock price proportionally and increases the value of the household’s wealth. The household is willing to hold more wealth as long as the rate of return is equal to the time preference rate. As a consequence, share prices rise, while the capital stock, investment, and output remain the same. This conforms with the ‘new’ view of dividend taxation, wherein a change in the dividend tax rate impacts the firm’s sources and uses of funds symmetrically, as also shown by McGrattan and Prescott (2005) and Santoro and Wei (2011).

Nevertheless, when the collateral constraint binds, a change in τ^D alters the effective rate of return on stocks required by the household, thereby resulting in a direct impact on the firm’s capital and investment decisions. Here, the physical capital Euler equation (22) and its steady state representation in (34) are distorted by the combination of φ > 0 and τ^D. A dividend tax cut that, ceteris paribus, raises asset prices, reduces the cost of capital [1 + φ(1 – τ^D)^-1](β^-1 – 1), and stimulates K and consequently I. Intuitively, the tax relief limits the tightness of the borrowing constraint and facilitates additional lending for investment purposes. Furthermore, the upward pressure on Q stemming from a positive φ is offset by any decrease in τ^D such that the book value of capital remains unchanged at δ/θ^B following a tax reform in the binding long-run equilibrium (observe (35)). Equity prices, on the other hand, rise in response to the tax reduction due to the positive relationship between q and K (see (36)).

Our model therefore produces distortionary steady state effects of dividend taxation without having to assume internally growing firms over the life-cycle and/or heterogenous firms facing different finance regimes as in Korinek and Stiglitz (2009), Gourio and Miao (2010), and Erosa and González (2019). The steady state values of δ, θ, and τ^D determine whether the representative firm is subject to a binding or non-binding credit constraint, which, in turn, dictates to what extent dividend tax adjustments affect the economic activity. The above discussion and the steady state implications of dividend tax reforms are summarized in the subsequent proposition.

Proposition 3 A cut (hike) in the dividend tax rate increases (lowers) the stock of capital and welfare when the economy is credit-constrained, conforming the ‘traditional’ view of dividend taxation. In an unconstrained economy, dividend taxes are irrelevant for the marginal investment decisions and welfare, as hypothesized by the ‘new’ view of dividend taxation.

Figure 2 visualizes the changes in the steady state values of the capital-to-labor ratio, Tobin’s Q, equity prices, and welfare when the tax rate is varied between 0 and 50 percent under three distinct borrowing scenarios.

Steady state values of the capital-to-labor ratio, Tobin’s Q, equity prices, and welfare when the dividend tax rate is varied between 0 and 50 percent under three different borrowing regimes.

Citation: IMF Working Papers 2022, 127; 10.5089/9798400214721.001.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Steady state values of the capital-to-labor ratio, Tobin’s Q, equity prices, and welfare when the dividend tax rate is varied between 0 and 50 percent under three different borrowing regimes.

Citation: IMF Working Papers 2022, 127; 10.5089/9798400214721.001.A001

• Download Figure
• Download figure as PowerPoint slide

Steady state values of the capital-to-labor ratio, Tobin’s Q, equity prices, and welfare when the dividend tax rate is varied between 0 and 50 percent under three different borrowing regimes.

Citation: IMF Working Papers 2022, 127; 10.5089/9798400214721.001.A001

• Download Figure
• Download figure as PowerPoint slide

In the constrained equilibrium (θ = 0.10), a fall in the dividend tax elevates the capital stock and share prices, but leaves Tobin’s Q unchanged (see the first row of Figure 2). As capital is the main driver of output and welfare in neoclassical production economies, tax reductions are thus welfare enhancing in the binding regime.¹² By contrast, in the slack equilibrium (θ = 0.25), where the firm finances investment via retained earnings, a tax on dividends only influences Q and q, leaving capital, investment, and welfare unchanged (see the third row of Figure 2). For the intermediate case (θ = 0.16) and as observed from the second row of Figure 2, the economy finds itself in a constrained equilibrium when the dividend tax rate is greater than 25 percent. If tax cuts occur from those initially relatively higher tax rates, K/N and steady state welfare increase until reaching their levels in the slack regime and remain unchanged thereafter (as also postulated from Propositions 1 and 2). In all credit regimes and from a qualitative perspective, the capital-to-labor ratio and welfare respond in an identical fashion to dividend tax adjustments.

3.2 Quantitative Analysis and Short-Run Model Dynamics

3.2.1 Calibration

The time period in the model is one year. The baseline parameterization used to analyze the short-run dynamics is summarized in Table 1. We largely follow the calibration of general equilibrium models that are calibrated at an annual frequency, and which include investment frictions and/or various forms of taxation (e.g., Gomes (2001), McGrattan and Prescott (2005), Gourio and Miao (2010, 2011), and Anagnostopoulos, Cárceles-Poveda, and Lin (2012)). Specifically, some parameter values are borrowed from the literature whereas others are calibrated or estimated in order to approximately match some model moments with those observed in the U.S. data.

^{Table 1}

– Benchmark Calibration

^{Table 1}

– Benchmark Calibration

Parameter	Value	Description
β	0.97	Discount factor
h	1.98	Labor disutility parameter
δ	0.12	Capital depreciation rate
α	0.30	Capital share in production
A	1.00	Average productivity parameter
θB	0.16	Borrowing limit in the benchmark binding equilibrium
θNB	(0.185,1)	Borrowing limit range in the non-binding equilibrium
τN	0.28	Labor income tax rate
τK	0.35	Corporate tax rate
τD	0.35	Dividend tax rate
γ	12.0	Capital adjustment cost parameter
ρθ	0.76	Degree of persistence – Financial shock
σϑ	0.088	Standard deviation – Financial shock

View Table

^{Table 1}

– Benchmark Calibration

Parameter	Value	Description
β	0.97	Discount factor
h	1.98	Labor disutility parameter
δ	0.12	Capital depreciation rate
α	0.30	Capital share in production
A	1.00	Average productivity parameter
θB	0.16	Borrowing limit in the benchmark binding equilibrium
θNB	(0.185,1)	Borrowing limit range in the non-binding equilibrium
τN	0.28	Labor income tax rate
τK	0.35	Corporate tax rate
τD	0.35	Dividend tax rate
γ	12.0	Capital adjustment cost parameter
ρθ	0.76	Degree of persistence – Financial shock
σϑ	0.088	Standard deviation – Financial shock

From Proposition 1, the region under which the collateral constraint is binding in steady state depends crucially on empirically-plausible ranges for the borrowing limit, the dividend tax rate, and the depreciation rate. Setting δ = 0.12 to match the average nonresidential investment rate found in the data, and using condition (31), we fix θ^B = 0.16 and τ^D = 0.35 such that the benchmark economy confronts a constrained steady state equilibrium with φ = 0.10 (see also (32) and Figure 1).¹³ Examining the time series of the investment rate and Tobin’s Q from 1980 to 2020, and using our steady state propositions, we find that θ over the sample term ranges from a minimum value of 0.10 to a maximum value of 0.25 with an average of 0.16.¹⁴ These estimates lie within range of Jermann and Quadrini (2012) and Covas and Den Haan (2011). The steady state dividend tax rate of 35 percent is fairly consistent with the U.S. average dividend income tax rate that includes both federal and state taxes.

Our choice of parameters imply a steady state investment to GDP ratio of 15.03 percent, an average dividend to GDP ratio of 4.47 percent, and Q^B = 0.75. All these statistics are pretty close to their long-run data counterparts. Notice that with the initial dividend tax rate fixed at 0.35, the frictionless steady state model with φ = 0 yields Q^NB = 0.65. Therefore, the binding collateral constraint in steady state helps to obtain a higher and more data-consistent estimate for Q.¹⁵ Nevertheless, to examine the dynamic responses following temporary dividend tax changes, we compare the case where the initial position of the economy is in a binding steady state equilibrium to the situation in which the long-run collateral constraint is slack. In the latter and for δ = 0.12 and τ^D = 0.35, we can choose any value θ^NB ≥ 0.185 so that φ = 0 in line with condition (33).

Turning to the adjustment cost parameter, estimates of γ vary significantly in the literature, where its magnitude is typically calibrated using equation (21) in order to match the elasticity of the investment rate with respect to Tobin’s Q with its data equivalent. This investment Q regression elasticity may range from a mean of 0.0423, as computed by Hayashi (1982), to 0.23, as calibrated by Jermann (1998). In a model like ours, the investment Q elasticity is equal to the inverse of the adjustment cost coe¢cient.¹⁶ We pick γ = 12, corresponding with an elasticity of 0.083, as an intermediate compromise value. Including capital adjustment costs helps to better relate the model-implied volatility and hump-shape movements in aggregate investment and output with the data.

Finally, using standard Bayesian estimation techniques, we calibrate the persistence parameter and standard deviation of the financial shock to reproduce the standard deviation of the nonres-idential investment rate found in the data for the period 1980–2020.¹⁷ The annually estimated occasionally-binding model implies choosing ρ^θ = 0.76 and σ^θ = 0.088.¹⁸ Table 2 reports the standard deviations and autocorrelations of key variables in our model and how they fare with their respective equivalents in the data.

^{Table 2}

– Comparing Model and Data: Business Cycle Moments

Model standard deviations are computed from 10,000 simulations keeping θ_t stochastic. All statistics are annualized, detrended, and measured in percentage terms. Data statistics are computed from detrended annual series (1980–2020) extracted from the FRED database.

^{Table 2}

– Comparing Model and Data: Business Cycle Moments

	$\sigma \left(\frac{I_t}{K_t}\right)$	$corr(\frac{I_t}{K_t},\frac{I_{t-1}}{K_{t-1}})$	$\sigma \left(\frac{I_t}{Y_t}\right)$	σ(Yt)	$\sigma \left(\frac{{\overline{D}}_t}{Y_t}\right)$
Model	0.07	0.689	0.069	2.169	0.232
Data	0.07	0.783	0.068	2.029	0.310

Model standard deviations are computed from 10,000 simulations keeping θ_t stochastic. All statistics are annualized, detrended, and measured in percentage terms. Data statistics are computed from detrended annual series (1980–2020) extracted from the FRED database.

View Table

^{Table 2}

– Comparing Model and Data: Business Cycle Moments

	$\sigma \left(\frac{I_t}{K_t}\right)$	$corr(\frac{I_t}{K_t},\frac{I_{t-1}}{K_{t-1}})$	$\sigma \left(\frac{I_t}{Y_t}\right)$	σ(Yt)	$\sigma \left(\frac{{\overline{D}}_t}{Y_t}\right)$
Model	0.07	0.689	0.069	2.169	0.232
Data	0.07	0.783	0.068	2.029	0.310

Model standard deviations are computed from 10,000 simulations keeping θ_t stochastic. All statistics are annualized, detrended, and measured in percentage terms. Data statistics are computed from detrended annual series (1980–2020) extracted from the FRED database.

3.2.2 Temporary Dividend Tax Shocks

In the first counterfactual policy experiment, we compare the behavior of the unconstrained economy model with the occasionally credit-constrained model following a dividend tax rate cut of 10 percentage points; from an initial 35 percent to 25 percent. The tax adjustment in both scenarios occurs in period 1, is assumed to be temporary, and lasts for 8 periods. After the 8 periods, ${\tau }_t^D$ reverts to its previous long-run level. Suppose the tax policies are unanticipated initially, but once they occur, the agents have perfect foresight about their future paths. Despite largely being extended in early 2013, the original 2003 U.S. dividend tax relief was set to expire in 2012, highlighting the temporary yet persistent nature of such fiscal reform. The experiments conducted in this section allow us to study the strikingly different implications of transitional dividend tax reforms caused by the investment credit constraint and the economy’s initial steady state credit position.

The dynamics of the key variables following the temporary tax cut are shown in Figure 3. In the unconstrained regime, a transitional dividend tax shock generates a collapse in investment and therefore in capital accumulation and output. These results are largely in line with the findings of Gourio and Miao (2011), who also show that firms distribute large dividends and cut back on capital investment in response to a temporary dividend tax cut. Furthermore, from equations (27) and (28) with φ_t = 0, Tobin’s Q initially rises upon the impact of the tax reduction placing some upward pressure on I_t in period 1. However, Q_t starts to decrease until period 8 and then slowly converges to its steady state because ${\tau }_t^D$ rises back permanently to its original level at the start of period 9. Investment follows an opposite path to Q_t as the effect of increasing dividends in response to the tax cut dominates the otherwise positive relationship between investment and its shadow price. Due to an intertemporal substitution effect, consumption slightly increases in the first period, declines from periods 2 to 8, and starts to revert to its long-run level as soon as the tax shock ends. Given condition (25) and with capital predetermined at the start date of the tax reform, employment shrinks, which, in turn, amplifies the decrease in output.¹⁹ In view of the expected tax policy reversal from period 9, the firm responds by sharply cutting dividends and accelerating investment in period 8: This leads to a slower rate of decline in Q_t in the subsequent year. To summarize, lower temporary dividend taxes have an overall strong short-run contractionary impact on the real economy.

Dynamic responses following an unexpected temporary dividend tax cut in the occasionally-binding and permanently slack models. Except for the borrowing limit multiplier that is calculated in levels, all other variables are measured in percentage deviations from the different steady states corresponding with the different credit regimes.

Citation: IMF Working Papers 2022, 127; 10.5089/9798400214721.001.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Dynamic responses following an unexpected temporary dividend tax cut in the occasionally-binding and permanently slack models. Except for the borrowing limit multiplier that is calculated in levels, all other variables are measured in percentage deviations from the different steady states corresponding with the different credit regimes.

Citation: IMF Working Papers 2022, 127; 10.5089/9798400214721.001.A001

• Download Figure
• Download figure as PowerPoint slide

Dynamic responses following an unexpected temporary dividend tax cut in the occasionally-binding and permanently slack models. Except for the borrowing limit multiplier that is calculated in levels, all other variables are measured in percentage deviations from the different steady states corresponding with the different credit regimes.

Citation: IMF Working Papers 2022, 127; 10.5089/9798400214721.001.A001

• Download Figure
• Download figure as PowerPoint slide

On the other hand, under the occasionally-binding credit regime, the same dividend tax cut results in an investment, capital, and output rise. The aforementioned large dividend payout prevailing in the unconstrained model is counteracted by the relaxation in the tightness of the collateral constraint (φ_t) that is directly impacted by the fall in ${\tau }_t^D$. Intuitively, the lower dividend tax raises Q_t and increases the value of the firm’s collateralized capital stock. With an initially binding steady state collateral constraint, the firm can therefore engage in additional borrowing and raise its investment in physical capital. The temporary 10 percentage point tax cut renders a slack credit constraint during periods 1 to 8 followed by an immediate jump to the originally positive long-run level of φ_t in period 9. The firm takes advantage of the interim relaxed credit environment to make further investments and to limit dividend payouts in the first period. However, from periods 2 to 8 and as the capital stock is expected to improve, which allows the firm to borrow against future earnings, dividend distributions increase while investment declines. Once the tax relief expires, both these variables slowly return to their steady states. Notice also that while consumption falls below its steady state in the first period due to intertemporal substitution, this variable overall exhibits lumpiness and remains above its long-run level throughout most of the duration of the tax reform and beyond. Overall, easing the tightness of the investment credit limit in relation to the binding steady state results in dividend taxes inducing short-run expansionary effects on the real economic activity.²⁰

Unlike our paper, Gourio and Miao (2011) in their extended model with debt financing do not predict that investment rises in the period when the dividend tax cut occurs. In fact, their model suggests that the transitional dynamics of real variables with and without debt are very similar. When the debt limit applies directly to investment in physical capital like in our framework, the short-term macroeconomic effects of temporary dividend tax reforms become more consistent with the ‘traditional’ view of dividend taxation. Output, labor, and investment also exhibit a procyclical relationship on impact regardless of the economy’s original position.

Finally, in relation to the earlier dividend tax literature without financial frictions or with frictions that are always binding, the model presented here captures the non-linearities and asymmetries associated with the occasionally-binding nature of the investment credit constraint. Figure 4 shows the dynamic responses following a 10 percentage point tax cut and hike relative to the binding steady state equilibrium. The tax rise tightens the credit constraint and brings about a magnified decline in investment, labor, and output when compared to the dynamics following the same size tax cut that temporarily switches the credit regime from binding to slack. Albeit quantitatively modest, our model does capture the potentially asymmetric dynamic responses of financial and real variables following unexpected temporary dividend tax changes (see also Boissel and Matray (2022)).

Dynamic responses following an unexpected temporary dividend tax cut and hike relative to the common binding steady state regime. Notice that the solid line is the same as the solid line in Figure 3.

Citation: IMF Working Papers 2022, 127; 10.5089/9798400214721.001.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Dynamic responses following an unexpected temporary dividend tax cut and hike relative to the common binding steady state regime. Notice that the solid line is the same as the solid line in Figure 3.

Citation: IMF Working Papers 2022, 127; 10.5089/9798400214721.001.A001

• Download Figure
• Download figure as PowerPoint slide

Dynamic responses following an unexpected temporary dividend tax cut and hike relative to the common binding steady state regime. Notice that the solid line is the same as the solid line in Figure 3.

Citation: IMF Working Papers 2022, 127; 10.5089/9798400214721.001.A001

• Download Figure
• Download figure as PowerPoint slide