Procyclical Fiscal Policy and Asset Market Incompleteness

Layer 1: Overview

Developing and emerging economies exhibit procyclical fiscal policy on both the spending and taxation sides: government expenditures expand in booms and contract in recessions, and tax rates fall in good times while rising in bad times. This is the mirror image of optimal countercyclical policy prescribed by standard theory and practiced in advanced economies. Understanding why developing countries pursue policies that amplify already-volatile business cycles is a long-standing puzzle in international macroeconomics.

This paper develops a small open economy model with Ramsey-optimal fiscal policy to argue that standard incomplete asset markets — without sovereign default risk, limited commitment, or high risk premia — are sufficient to explain procyclical fiscal policy on both the spending and the taxation sides. The authors proceed in three stages: a static two-state model that isolates a novel theoretical result; a calibrated infinite-horizon DSGE model that replicates the result and quantifies welfare costs; and a cross-country empirical section providing reduced-form support.

The paper covers 121 countries (99 developing, 22 OECD) using data on real government consumption, real GDP, and VAT rates updated from earlier studies. The average correlation between the cyclical components of real government spending and real GDP is 0.29 for developing countries versus -0.12 for OECD countries (both significant at the 1 and 5 percent levels, respectively). For tax policy, the average correlation between changes in the VAT rate and real GDP is -0.22 for developing countries (significant at the 1 percent level) versus -0.06 for industrial countries (insignificant at the 5 percent level), confirming procyclical tax behavior in non-OECD economies.

The core theoretical contribution is a novel result established in a static model: under financial autarky (extreme market incompleteness), government spending is always procyclical regardless of preference parameters, but tax rates can be procyclical, acyclical, or countercyclical depending on the relative magnitudes of the intertemporal elasticities of substitution for private versus public consumption (sigma_c and sigma_g). The key is the “consumption preference channel”: when sigma_c exceeds sigma_g, private consumption rises proportionally more than public consumption in good times, expanding the tax base by more than the increase in government spending, which allows the fiscal authority to reduce tax rates. The ratio of private to public consumption comoves positively with the business cycle when sigma_c > sigma_g — the empirically-relevant case — generating procyclical tax policy.

Under complete markets, both government spending and tax rates are acyclical regardless of preference parameters.

The DSGE model introduces an infinite-horizon setting with endogenous production and labor supply and access to a non-state-contingent international bond with a debt-elastic interest rate spread. This adds a “consumption smoothing channel” that works against procyclicality: when households can borrow to smooth consumption following adverse shocks, the tax base contracts less, reducing the pressure to raise taxes. However, when the model is calibrated to non-OECD countries — using a debt-elasticity parameter of phi = 0.125 (estimated from non-OECD panel data using EMBIG spreads and public debt) and TFP persistence of rho_A = 0.95 — the consumption preference channel dominates the consumption smoothing channel. The correlation between government spending and output exceeds 0.95 across all values of sigma_g examined (from 0.5 to 1.5) and across all considered debt elasticities. The cyclicality of tax rates flips sign as sigma_g crosses sigma_c, consistent with the static result.

A moment-matching exercise calibrated to non-OECD data selects sigma_g = 0.25, phi = 1, and rho_A = 0.95 as best-fit parameters. The model successfully replicates four targeted moments — standard deviations of output and private consumption, and the correlations of government spending and tax rates with output — and also matches the untargeted positive comovement of the private-to-public consumption ratio with GDP. The model accounts for only about one-tenth of observed government spending volatility and one-fifth of tax rate volatility, indicating additional non-Ramsey sources of fiscal variation exist.

Welfare costs of fiscal procyclicality are computed using a Lucas (1987) approach. With no financial frictions (phi approximately 0), welfare costs are approximately 0.015 percent of lifetime consumption. Increasing phi to the calibrated non-OECD value of 0.125 nearly doubles welfare costs to approximately 0.03 percent of lifetime consumption. More persistent TFP shocks (higher rho_A) amplify procyclicality further.

The empirical section provides cross-country evidence. Capital controls (measured by Fernandez et al.’s 2016 de jure indices across 32 transaction types in 10 asset classes over 1995-2015) are larger in non-OECD countries by an order of magnitude, and the null of equal completeness is statistically rejected. The estimated debt-spread elasticity for non-OECD countries using public debt is phi = 0.125 (significant at the 1 percent level), versus 0.002 for OECD countries (insignificant). GDP volatility measured by the standard deviation of HP-filtered real GDP is 3.28 for non-OECD countries versus 1.47 for OECD countries, a difference of more than twofold.

The policy implication is that completing markets — through sovereign wealth funds, contingent credit lines with international financial institutions, or structural fiscal rules that force saving in good times — could reduce procyclicality and yield welfare gains estimated at up to twice the Lucas-type cost attributable to current friction levels.

Layer 2: Deep Dive

What is the main theoretical result, and how does it advance beyond the prior literature?

The paper establishes that incomplete markets (modeled as financial autarky or an upward-sloping supply of funds) are necessary and sufficient to generate procyclical government spending, but are only necessary — not sufficient — for procyclical tax rates. The direction of tax cyclicality depends on the relative intertemporal elasticity of substitution of private consumption (sigma_c) versus public consumption (sigma_g): procyclical if sigma_c > sigma_g, acyclical if equal, countercyclical if sigma_c < sigma_g. This overturns the widespread impression from Cuadra et al. (2010) that incomplete markets cannot generate procyclical tax rates. Prior work invoked sovereign default risk or limited commitment; this paper shows those additional ingredients are unnecessary.

What is the consumption preference channel and why is it empirically relevant?

The consumption preference channel works as follows: when households have a stronger preference for private over public consumption (sigma_c > sigma_g), private consumption rises proportionally more than government spending in good times. The wider tax base allows the government to reduce tax rates while still financing higher spending, generating procyclical tax policy. Empirically, the ratio of private to public consumption comoves positively with output in non-OECD countries — the model matches this as an untargeted moment — so the procyclical case (sigma_c > sigma_g) is the empirically relevant one. The model’s best-fit calibration selects sigma_g = 0.25 against sigma_c = 1.

What is the consumption smoothing channel and when does it dominate?

In the DSGE model, households can issue non-state-contingent bonds, partially smoothing consumption against shocks. A negative TFP shock therefore causes a smaller fall in consumption (the tax base), reducing the fiscal authority’s need to raise taxes procyclically. This consumption smoothing channel works against tax procyclicality. It dominates when the debt-elastic spread is low (cheap borrowing) and TFP shocks are transitory (low rho_A). For the calibrated non-OECD parameterization — phi = 0.125 and rho_A = 0.95 — the supply of funds is steep enough and shocks persistent enough that the consumption preference channel dominates, and procyclical tax policy results.

What role does TFP persistence play?

Higher TFP persistence amplifies business cycle volatility and deepens the procyclicality of fiscal policy. When a negative TFP shock is more persistent (rho_A rises from 0.42 as in Mendoza 1991 toward 1.0), consumption falls more sharply and for longer, shrinking the tax base substantially. This forces the fiscal authority to raise taxes more aggressively in recessions, increasing procyclicality. The half-life of a TFP shock with rho_A = 0.95 is close to seven quarters, versus less than a quarter at rho_A = 0.42. Aguiar and Gopinath (2007) motivate the use of high persistence as a distinguishing feature of emerging market business cycles.

How are the two types of financial frictions — market incompleteness and debt-elastic spreads — distinguished?

Asset market incompleteness refers to the dimension of available financial instruments (financial autarky: none; incomplete: risk-free bond; complete: full set of state-contingent claims). The debt-elastic spread (governed by phi_c and phi_g) captures the steepness of the supply of external funds, which can be high even when access to a bond market exists. The authors note these are not isomorphic: Fernandez and Gulan (2015) provide microfoundations for the debt elasticity in an environment with defaultable private debt and asymmetric information, holding market incompleteness constant. Both frictions independently amplify business cycles and procyclicality, but the paper treats them separately in both calibration and empirical proxies.

What are the three propositions from the static model?

Proposition 1: Government spending is acyclical under complete markets and strictly procyclical under financial autarky, regardless of the values of sigma_c and sigma_g. Proposition 2: Tax rates are acyclical under complete markets. Under financial autarky, tax rates are acyclical if sigma_c = sigma_g, countercyclical (positive correlation with output) if sigma_c < sigma_g, and procyclical (negative correlation with output) if sigma_c > sigma_g. Proposition 3: Under financial autarky, the procyclicality of government spending increases with output volatility. If taxes are procyclical (sigma_c > sigma_g), tax procyclicality also increases with output volatility. Under complete markets, output volatility has no effect on fiscal cyclicality.

What is the moment-matching exercise and what does it conclude?

The exercise calibrates four parameters — TFP volatility (sigma_A), TFP persistence (rho_A), the government consumption elasticity (sigma_g), and the debt-spread elasticity (phi) — to minimize a quadratic loss function over the four targeted moments: standard deviations of income and private consumption, and correlations of taxes and government spending with real GDP, using non-OECD country data with balanced panels of more than ten consecutive annual observations. The best-fit parameters are sigma_g = 0.25, phi = 1, and rho_A = 0.95. The model matches the sign and approximate magnitude of the four targeted moments and also replicates the untargeted positive comovement of the private-to-public consumption ratio with output. It accounts for only about one-tenth of observed government spending volatility and one-fifth of tax volatility, suggesting other sources of fiscal variation beyond Ramsey dynamics.

How are welfare costs calculated and what are the magnitudes?

Welfare costs are computed in the Lucas (1987) tradition: they equal the permanent share of steady-state consumption that households in a frictionless economy (no shocks) would need to forgo to achieve the same lifetime utility as households in the economy with TFP shocks and varying degrees of fiscal procyclicality induced by different values of phi. Using 100,000 simulated quarters with sigma_g = 0.5, sigma_c = 1, sigma_A = 0.0129, and rho_A = 0.95, welfare costs rise from approximately 0.015 percent of lifetime consumption when phi is near zero to approximately 0.03 percent at the calibrated non-OECD value of phi = 0.125 — nearly doubling as procyclicality increases. The paper acknowledges that higher phi also imposes other costs beyond procyclicality per se.

What empirical proxies are used and what do they show?

Asset market incompleteness is proxied by four indices from Fernandez et al. (2016) covering de jure restrictions on capital inflows and outflows across 32 transaction types and 10 asset classes for 1995-2015: overall inflow restrictions (kai), outflow restrictions (kao), bond inflow restrictions, and bond outflow restrictions. Each index ranges from 0 to 1. All four indices are higher for non-OECD countries than OECD by an order of magnitude, with the null of equality statistically rejected. For debt-spread elasticity, the paper estimates the model’s functional form (spread regressed on an exponential function of debt-to-output) using panel fixed effects, with spreads proxied by EMBIG for non-OECD, T-bill spreads over German Bunds for EU-OECD, and UIP-implied spreads for other OECD. Using public debt, the elasticity for non-OECD is phi = 0.125 (significant at 1 percent) versus 0.002 for OECD (insignificant). GDP volatility (standard deviation of HP-filtered real GDP) is 3.28 for non-OECD versus 1.47 for OECD.

How does this paper relate to Cuadra et al. (2010) and Riascos and Vegh (2003)?

Riascos and Vegh (2003) showed in a calibrated model that incomplete markets can explain procyclical government spending, but their model faced government borrowing at the risk-free rate across all states, which Cuadra et al. argued prevented the model from generating negative output-tax rate correlations. Cuadra et al. (2010) incorporated both incomplete markets and sovereign default risk, showing that their combination yields procyclical fiscal policy on both spending and revenue sides. This paper argues that Cuadra et al.’s assessment left the mistaken impression that incomplete markets per se are insufficient for procyclical taxes. The current paper shows this impression is wrong: standard incomplete markets without default risk yield procyclical tax rates when the empirically-validated condition sigma_c > sigma_g holds.

What are the policy implications and their scope conditions?

The mechanism implies that reducing financial frictions — either by completing asset markets or by flattening the supply of external funds — would moderate fiscal procyclicality and generate Lucas-type welfare gains. Concrete instruments include: sovereign wealth funds that allow self-insurance in good times; contingent credit lines with international financial institutions that provide access to funds in bad times; and structural fiscal rules (as in Chile’s structural balance rule) that force saving in booms, effectively completing markets through institutional commitment. The scope condition is that these gains are relevant for non-OECD countries characterized by high capital controls, steep debt-elastic spreads, and volatile output — not for OECD economies where markets are already more complete and fiscal policy is acyclical or countercyclical.

What are the main limitations acknowledged by the paper?

The model is deliberately parsimonious and accounts for only about one-tenth of observed government spending volatility and one-fifth of tax rate volatility. Additional shocks beyond TFP and world interest rate variation — including political economy forces, commodity price cycles, and demand shocks — are clearly relevant. The model also only accounts for a fraction of the private consumption-output correlation, suggesting missing amplification mechanisms. The paper does not structurally identify the model from micro-data and relies on moment matching over a grid rather than formal estimation. The welfare cost calculation attributes all welfare loss to fiscal procyclicality, but higher phi also raises the cost of debt in ways unrelated to fiscal cyclicality.

What is the role of political economy explanations, and does this paper displace them?

The paper presents the financial frictions explanation as complementary to rather than a replacement for political economy explanations (such as Tornell and Lane 1999’s voracity effect or Alesina et al. 2008’s Leviathan-starving hypothesis). The paper’s claim is narrower: from an applied theory perspective, incomplete markets alone are sufficient to generate the stylized facts, so additional ingredients such as sovereign risk or limited commitment are not required to explain the basic puzzle. Whether political economy or financial frictions are quantitatively more important in explaining the cross-country variation in fiscal cyclicality remains an open question.

Key Concepts

Procyclical fiscal policy: In this paper’s usage, government spending is procyclical when it rises in good times and falls in bad times (positive correlation with output), and tax policy is procyclical when tax rates fall in good times and rise in bad times (negative correlation between tax rates and output). The paper stresses that the ratio g/y is not an appropriate cyclicality measure because y is endogenous.

Consumption preference channel: The mechanism by which households’ relative preference for private over public consumption (sigma_c > sigma_g) causes private consumption to expand proportionally more than government spending in good times, widening the tax base relative to spending needs and allowing the fiscal authority to cut tax rates procyclically.

Consumption smoothing channel: The countervailing mechanism present in the DSGE model: when households can borrow at relatively low cost to smooth consumption, adverse TFP shocks cause a smaller fall in the tax base, reducing the government’s need to raise taxes in recessions. This channel works against tax procyclicality and is weaker when the debt-elastic spread is steep.

Debt-elastic interest rate spread (phi): A country-specific premium on external borrowing that increases with the stock of debt, following the Schmitt-Grohe and Uribe (2003) formulation. In this paper, phi governs the slope of the supply of external funds and proxies for the severity of financial frictions distinct from the dimension of market incompleteness. Non-OECD countries are estimated to have phi = 0.125, compared to 0.002 for OECD.

Financial autarky: The polar case in which neither households nor the government can buy or sell financial securities internationally; all financial transactions must be within the country, so the domestic interest rate adjusts endogenously to clear markets. In the model, this case delivers the strongest procyclicality, equivalent to very high phi.

Ramsey optimal fiscal policy: The paper solves for the fiscal policy (tax rates and government spending) that maximizes household welfare subject to the government’s budget constraint and private sector implementability conditions. This is used rather than an ad-hoc fiscal rule, so procyclicality is an optimal response to frictions rather than a policy failure.

Lucas-type welfare cost: Measured here as the permanent fraction of steady-state consumption that a household in a shock-free economy would forgo to achieve the same lifetime utility as a household in the stochastic economy with TFP shocks and a given level of debt-elastic financial friction. The paper reports that this cost nearly doubles as phi rises from near zero to the calibrated non-OECD value of 0.125.