Description
Résumé: Une augmentation du déficit des administrations publiques peut produire deux effets : elle peut stimuler la demande globale et l'emploi dans le court terme et provoquer une diminution de la production potentielle dans le long terme. Dans la présente étude, ces effets sont illustrés au moyen d'un modèle macroéconomique de simulation dynamique. Ce modèle n'est pas un outil de prévision, mais vise plutôt à combler le fossé entre les keynésiens et les économistes de l'offre et à faire le jour sur l'effet d'éviction d'une augmentation du déficit budgétaire sur l'investissement. Par souci de simplicité, les auteurs ont choisi un modèle d'économie fermée. Synthèse assez conventionnelle du schéma keynésien traditionnel de type IS-LM et du schéma néoclassique de croissance élaboré par Solow et Tobin, le modèle en question résulte de l'incorporation au modèle statique IS-LM d'une fonction de production et d'une fonction d'investissement néoclassiques ainsi que d'une courbe de Phillips avec anticipations inflationnistes. La courbe de Phillips avec anticipations adaptatives établit le lien entre le court terme keynésien, où les niveaux de production et d'emploi sont déterminés par la demande, et le long terme monétariste, où ils sont déterminés par l'offre. Les études sur le financement des déficits publics par voie d'emprunt font ressortir deux questions cruciales pour la compréhension des effets macroéconomiques des déficits publics. La première a trait au risque que l'État n'arrive pas à contenir l'aggravation de son déficit à cause de l'effet de capitalisation des charges d'intérêt; la seconde concerne la question de savoir si les ménages considèrent un accroissement de la dette publique comme une augmentation de la richesse. Pour tenir compte du risque de dérapage des déficits publics, les auteurs font l'hypothèse que l'État réagit aux variations des charges d'intérêt réelles en modifiant sa fiscalité et ses programmes de dépenses. Ils croient que cette règle de simulation, quoique arbitraire, éclaire mieux le processus d'accumulation de la dette que ne le ferait une règle autorisant le dérapage de la dette publique. Le modèle n'offre pas de réponse à la question de savoir si, dans leurs décisions de consommation et d'épargne, les ménages prennent en compte les impôts futurs visant le financement de la dette publique. Les auteurs ont préféré faire des simulations sous trois hypothèses différentes : l'hypothèse, soutenue par Barro, que les impôts futurs sont pleinement pris en considération; l'hypothèse, soutenue par Modigliani, qu'ils sont partiellement pris en considération; l'hypothèse qu'ils ne sont pas du tout pris en considération. Si les ménages considèrent que le montant de la dette publique est supérieur à la valeur actualisée des impôts futurs qui s'y rapportent, le financement du déficit peut avoir un important effet d'éviction sur la formation du capital. Un accroissement des dépenses publiques peut aussi entraîner un effet d'éviction (peu importe son mode de financement) si la propension marginale à consommer est inférieure à l'unité. Pour bien distinguer les deux effets d'éviction l'un de l'autre, les auteurs ont simulé une réduction de l'impôt et un accroissement des dépenses. Les résultats obtenus sont strictement théoriques, car basés sur un modèle assez imparfait d'une économie très simple. Sur la base de l'hypothèse de Barro mais d'une faible propension marginale à consommer (60 %) par rapport au revenu disponible, une hausse des dépenses équivalant à 1 % du PIB provoquerait une diminution de 2,5 % de la production potentielle à long terme. Par contre, sous l'hypothèse keynésienne que les ménages ne tiennent aucun compte des impôts futurs, une augmentation des dépenses de même ampleur entraînerait une réduction de 7 % de la production potentielle, dont 2,5 % serait liée à l'accroissement des dépenses, et 4,5 % au financement du déficit. Les simulations du modèle, qui s'appuient sur des coefficients établis à partir d'un vaste éventail de travaux empiriques sur l'économie canadienne, ont permis de définir le profil temporel des réactions de cette dernière à divers chocs de dépenses ou d'impôts effectués sous différentes conditions. Les résultats des simulations montrent que l'effet expansionniste keynésien d'une hausse du déficit public est de très courte durée et beaucoup plus faible que l'effet de contraction sur la production potentielle en longue période. Le multiplicateur des dépenses culmine aux alentours de 0,6 à 0,7 au cours de la deuxième année, devient négatif après cinq ans et varie, selon l'hypothèse retenue relativement aux impôts futurs, de -2,5 à -7 au voisinage de l'équilibre. Les taux d'intérêt réels augmentent d'environ 0,3 point de pourcentage sous l'hypothèse barrovienne et d'environ 1 % si les ménages ne tiennent aucun compte des taux futurs de l'impôt. L'étude fait ressortir que le coût de financement du déficit peut être très élevé dans le modèle en question; il pourrait l'être davantage encore si on laissait l'offre de travail diminuer par suite d'une baisse du salaire réel après impôts. Cependant, il faut noter que les résultats dépendent beaucoup de l'élasticité de la production par rapport au capital (part du capital) qui est probablement trop élevée comme le reconnaissent les auteurs. Les auteurs présentent aussi les résultats de simulations partielles afin d'illustrer la nature de la contrainte d'offre dans le modèle. Ceux-ci font état d'un multiplicateur keynésien « pur » d'environ 2 lorsque le salaire nominal est exogène et que la politique monétaire accompagne la politique budgétaire.
Abstract: An increase in a government deficit can have two effects: short-term stimulation of aggregate demand and employment, and long-term contraction of potential output. In this paper, these effects are illustrated using a dynamic, macroeconomic simulation model. The model is not a forecasting tool; it is intended to bridge the gap between Keynesian and supply-side economics and to improve our knowledge of the process by which an increase in the government deficit can crowd out investment in the economy. A closed-economy model was chosen for simplicity. The model is a fairly conventional synthesis of the traditional Keynesian IS-LM model and the Solow-Tobin neoclassical growth model. A neoclassical production function, a neoclassical investment equation, and an expectations-augmented Phillips curve are introduced into the static IS-LM model. The Phillips curve, with adaptive expectations, links the (Keynesian) demand-determined model of output and employment in the short run and the (monetarist) supply-determined model of output and employment in the longer run. The literature on debt financing identifies two issues that are critical to understanding the macroeconomics effects of government deficits. The first issue is the risk that the government may lose control over its mounting deficit as a result of the compounding effect of escalating interest payments; the other is the question of whether households regard government debt as net wealth. This model deals with the prospect of uncontrolled deficit expansion by assuming that the government reacts to changes in its inflation-adjusted interest bill by making offsetting changes in its tax and expenditures programs. The authors believe that, although arbitrary, this simulation rule yields greater insight into the process of debt accumulation than a rule that would permit uncontrolled expansion of government debt. The question of whether households make allowance in their consumption/saving decisions for the future taxes required to finance the public debt is not answered here. Rather, the authors simulate the model under three alternative assumptions: full allowance (as advocated by Professor Barro), partial allowance (as advocated by Professor Modigliani), and no allowance at all. If households see the value of government debt as exceeding the discounted value of their future tax liabilities arising from that debt, then deficit financing can cause significant crowding out of capital formation. Some crowding out can also result from an increase in government spending, regardless of how the additional expenditure is financed, if the marginal propensity to consume is less than one. To separate these two effects, simulations were run for both a tax cut and an increase in spending. The results of the study are strictly theoretical since they are based on a fairly crude model of a very simple economy. These results indicate that, under full allowance for future tax liabilities, but low marginal propensity to consume out of disposable income (60 per cent), an increase in spending equivalent to 1 per cent of GDP would cause a 2.5 per cent contraction in potential output over the long run. Under the Keynesian assumption of no allowance for future tax liabilities, the same increase in spending would cause a 7 per cent reduction in potential output: 2.5 per cent relating to the expenditure increase, and 4.5 per cent relating to the deficit financing. Simulations were done with the model, which used coefficients based on a wide range of empirical studies relating to Canada, to illustrate the time path of responses to various tax and expenditure shocks under different conditions. They showed that the duration of the Keynesian expansionary effect of an increase in a budget deficit is relatively short and that its magnitude is much smaller than the long-run contractionary effect on potential output. The expenditure multiplier in the model peaks at 0.6 to 0.7 in the second year, becomes negative after five years and, depending on the allowance households make for future taxes, grows to between -2.5 and -7 near equilibrium; real interest rates increase by about 0.3 percentage points in the case of full allowance for future taxes and by about 1 percentage point in the case of no allowance at all. The study suggests that the cost of deficit financing can be very large in this model; it would be larger still if the labour supply was allowed to contract as a result of lower after-tax real wages. But it should be noted that the results are critically dependent on an assumed production elasticity of capital (the capital share) that the authors recognize may be too high. The authors also present partial model simulations to illustrate the nature of the supply constraint in the model. They show that the pure Keynesian multiplier in their model is about 2, when money wages are exogenous and monetary policy is accommodative.
Abstract: An increase in a government deficit can have two effects: short-term stimulation of aggregate demand and employment, and long-term contraction of potential output. In this paper, these effects are illustrated using a dynamic, macroeconomic simulation model. The model is not a forecasting tool; it is intended to bridge the gap between Keynesian and supply-side economics and to improve our knowledge of the process by which an increase in the government deficit can crowd out investment in the economy. A closed-economy model was chosen for simplicity. The model is a fairly conventional synthesis of the traditional Keynesian IS-LM model and the Solow-Tobin neoclassical growth model. A neoclassical production function, a neoclassical investment equation, and an expectations-augmented Phillips curve are introduced into the static IS-LM model. The Phillips curve, with adaptive expectations, links the (Keynesian) demand-determined model of output and employment in the short run and the (monetarist) supply-determined model of output and employment in the longer run. The literature on debt financing identifies two issues that are critical to understanding the macroeconomics effects of government deficits. The first issue is the risk that the government may lose control over its mounting deficit as a result of the compounding effect of escalating interest payments; the other is the question of whether households regard government debt as net wealth. This model deals with the prospect of uncontrolled deficit expansion by assuming that the government reacts to changes in its inflation-adjusted interest bill by making offsetting changes in its tax and expenditures programs. The authors believe that, although arbitrary, this simulation rule yields greater insight into the process of debt accumulation than a rule that would permit uncontrolled expansion of government debt. The question of whether households make allowance in their consumption/saving decisions for the future taxes required to finance the public debt is not answered here. Rather, the authors simulate the model under three alternative assumptions: full allowance (as advocated by Professor Barro), partial allowance (as advocated by Professor Modigliani), and no allowance at all. If households see the value of government debt as exceeding the discounted value of their future tax liabilities arising from that debt, then deficit financing can cause significant crowding out of capital formation. Some crowding out can also result from an increase in government spending, regardless of how the additional expenditure is financed, if the marginal propensity to consume is less than one. To separate these two effects, simulations were run for both a tax cut and an increase in spending. The results of the study are strictly theoretical since they are based on a fairly crude model of a very simple economy. These results indicate that, under full allowance for future tax liabilities, but low marginal propensity to consume out of disposable income (60 per cent), an increase in spending equivalent to 1 per cent of GDP would cause a 2.5 per cent contraction in potential output over the long run. Under the Keynesian assumption of no allowance for future tax liabilities, the same increase in spending would cause a 7 per cent reduction in potential output: 2.5 per cent relating to the expenditure increase, and 4.5 per cent relating to the deficit financing. Simulations were done with the model, which used coefficients based on a wide range of empirical studies relating to Canada, to illustrate the time path of responses to various tax and expenditure shocks under different conditions. They showed that the duration of the Keynesian expansionary effect of an increase in a budget deficit is relatively short and that its magnitude is much smaller than the long-run contractionary effect on potential output. The expenditure multiplier in the model peaks at 0.6 to 0.7 in the second year, becomes negative after five years and, depending on the allowance households make for future taxes, grows to between -2.5 and -7 near equilibrium; real interest rates increase by about 0.3 percentage points in the case of full allowance for future taxes and by about 1 percentage point in the case of no allowance at all. The study suggests that the cost of deficit financing can be very large in this model; it would be larger still if the labour supply was allowed to contract as a result of lower after-tax real wages. But it should be noted that the results are critically dependent on an assumed production elasticity of capital (the capital share) that the authors recognize may be too high. The authors also present partial model simulations to illustrate the nature of the supply constraint in the model. They show that the pure Keynesian multiplier in their model is about 2, when money wages are exogenous and monetary policy is accommodative.