«Arbresh MAMUDI, State University of Tetovo, Geoff PUGH, Staffordshire University Business School ABSTRACT The prevalent view on the stabilization ...»
Arbresh MAMUDI, State University of Tetovo,
Geoff PUGH, Staffordshire University Business School
The prevalent view on the stabilization role of fiscal policy has evolved over time.
The theory on the effectiveness of fiscal policy on output stabilization is controversial and
indeterminate. The empirical evidence is huge reporting heterogeneous results: the range of empirical findings varies from negative to higher than one.We apply MRA analysis to a set of 65 empirical studies on fiscal policy estimated by either single equation approaches (SEE hereinafter) or Vector autoregression (VAR) models. In order to identify the systematic influence of certain country/structural characteristics and study characteristics on estimates of the multiplier, we classify studies with respect to the model class, type of fiscal impulses, method of calculation of multipliers, time-series properties of the data, and several other control variables. In addition, we augment the MRA model with primary data accounting for structural characteristics of the countries investigated by this literature. A novel approach of this study will be to augment MRA with primary data on labour market variables to see if there is any systematic variation to be attributed to labour market institutions.
Our main findings are as follows. Fiscal multipliers differ in terms of model classes.
VAR models yield significantly higher multipliers than do SEE approaches. In line with Keynesian theory, expenditure multipliers are higher than tax shock multipliers, with public consumption being the most effective shock across several specifications of our model.
Permanent shocks yield higher multipliers and multipliers estimated for a longer horizon are higher than impact multipliers. Quarterly data yield lower multipliers compared to annual data. Fiscal policy in transition countries appears to be more effective than in advanced economies, contrary to a priori expectations, although the results are not stable across different specifications. In VAR analysis, different identification strategies give rise to significantly different multipliers. In line with theoretical predictions, the reported multipliers are smaller if estimated for open economies and economies in expansion and bigger for relatively more closed economies and economies experiencing a financial crisis. Structural characteristics from primary data appear to explain less of the variation of multiplier estimates; the openness channel seems to be an important determinant of the multipliers while, from the labour market characteristics, the replacement rate appears to significantly affect the value of the multiplier. The results on the publication selection test suggest that the literature is affected by publication bias.
In conclusion, our findings suggest that the heterogeneity of the reported multipliers mostly arises from the study characteristics, however, the degree of openness and labour market characteristics appear to be important factors that should be considered while estimating fiscal multipliers.
Keywords: fiscal multipliers, fiscal policy, meta-analysis Introduction The prevalent view on the stabilization role of fiscal policy has evolved over time.
The recent crisis and the zero bound for interest rates renewed interest in the importance of fiscal policy in fighting the recession. The empirical literature on the size of the fiscal multipliers, in the post crisis period, is growing fast. However, the results are far from consensus. Estimates of multipliers are all over the map, the range of empirical findings varies from negative to higher than one. A critical evaluation on the development of the theory of fiscal policy and employment suggests that the effectiveness of fiscal policy with respect to output stabilization is controversial and indeterminate. The inconclusive theory and a variety of empirically estimated fiscal multipliers make the application of meta-regression analysis (MRA hereinafter) a suitable set of tools to review the literature, to explain the heterogeneity of the empirical results and to investigate possible publication selection bias in the literature.
We apply MRA analysis to a set of 65 empirical studies on fiscal policy estimated by a single equation approach (SEE hereinafter) and Vector autoregression (VAR) models. In order to identify the systematic influence of certain country/structural characteristic and study characteristic on estimates of the multiplier, we classify studies with respect to the model class, type of fiscal impulses, method of calculation of multipliers, time-series properties of the data, and several other control variables. In addition, we augment the MRA model with primary data accounting for structural characteristics of the countries. A novel approach of this study will be to augment MRA with primary data on labour market variables to see if there is any systematic variation to be attributed to labour market institutions.
The results are consistent with theoretical predictions. Fiscal multipliers differ in terms of model classes. VAR models yield significantly higher multipliers than do SEE approaches. In line with Keynesian theory, expenditure multipliers are higher than tax shocks multipliers, with public consumption being the most effective shock across several specifications of our model. Data characteristics are also significant in explaining the variation of estimated multipliers. Multipliers calculated on quarterly data are lower and the longer is the horizon of calculation the higher is the multiplier value. The duration of the fiscal shock also proves to be a significant determinant of multiplier values suggesting that permanent shocks yield higher multipliers. The reported multipliers in transition countries appear to be higher compared to advanced economies, contrary to a priori expectations;
however, the results are not stable across different specifications. Studies controlling for structural characteristics of the economy report significantly different multipliers than conventional studies (which don’t control for these characteristics). As expected, the reported multipliers are smaller if estimated for open economies and economies in expansion and bigger for relatively more closed economies and economies experiencing a financial crisis.
The subsample analysis of VAR models suggest that the differences in identification strategies are significant, while different types of VAR models are insignificant factors for determining the value of multipliers. Regarding the labour market characteristics, only replacement rates appear to be a significant determinant of the multiplier effect. In addition, although the degree of openness is an important factor, contrary to theoretical predictions, variables accounting for the indebtedness of the economy, financial development and monetary policy appear to be insignificant factors in determining multipliers.
Investigating publication selection bias, the funnel plot analysis as well as the formal FAT-PET test, which are applied by using sample size as a proxy for precision, suggests that the literature on fiscal multipliers is infected by publication bias. Controlling for publication bias is thus an important feature of this MRA.
In summary, our findings suggest that the heterogeneity of the reported multipliers mostly arises from the study characteristics: data settings and methods of estimation, while the effect of structural characteristics is mixed. However, the degree of openness and labour market characteristics appear to be important factors that should be considered while estimating fiscal multipliers.
The rest of this paper is organized as follows: In the first section a narrative review of the fiscal multipliers literature is provided. The next section describes the preparation for MRA, the data collection process and some descriptive statistics of the variables. Section 3 provides an introduction to MRA methods and the results of the analysis are provided in Section 4 along with some robustness checks. The end of the chapter provides conclusions from the MRA.
2.1 Literature review
2.1.1 Theoretical overview Theoretical predictions about the transmission mechanism of fiscal policy in an economy are controversial and differ not only about the size of the effects of fiscal policy on macroeconomic variables but also about the direction of these effects. While most theoretical models agree on the positive effect of an expansionary fiscal shock on output, the disagreement is about crowding in vs. crowding out effects on private consumption, investment and net exports. The effectiveness of fiscal policy as a stabilization tool depends on the assumptions about the type of the economy, type of agents, the degree of openness and the exchange rate regime of the economy.
In a new classical model, with flexible prices, assuming the economy is in equilibrium and there are no spare or non-used capacities there is no role for fiscal policy. In a Keynesian model, assuming a demand deficient economy with sticky prices, and non-forward looking agents a stimulus to demand can have multiplier effects. Higher government spending increases income, employment and consumption. Aggregate demand determines output, hence, fiscal expansion causes higher output. Fiscal policy is expected to be more effective in a closed economy compared to an open economy. In a closed economy model a fiscal expansion boosts private consumption leading to an increase in aggregate demand, output and money demand, hence, if money supply is fixed, this results in an increase of interest rates, which in turns partially crowds out private investment. Consequently, output, total investment and consumption increase. In the Keynesian model, the use of fiscal stimulus, in an open economy under a flexible exchange rate regime, will increase real interest rates and, given price stickiness, the real exchange rate will appreciate, which will lead to a loss of international competitiveness and trade balance deterioration as a result. The expansionary effect of fiscal stimulus may be entirely offset by the reduction in net exports; hence there is no fiscal multiplier. In contrast, under a fixed exchange regime, monetary policy will react in order to prevent the interest rate increase, hence exchange rate appreciation, therefore, amplifying the effect of the fiscal multiplier. In a new classical view, a fiscal shock does not change international relative prices in the absence of home bias; hence the real exchange rate is not affected by the shock. If home bias is assumed, the relative price of domestically produced goods in terms of foreign produced goods increases, which leads to an appreciation of the real exchange rate (Ravn et al., 2007).
The type of agent assumption is crucial in analysing the effectiveness of fiscal policy.
Most of the studies employ DSGE models, which incorporates forward looking agents and rational expectations. RBC models, assuming flexible prices and perfect competition suggest that fiscal policy is ineffective in stimulating the economy. Their argument against the effectiveness of fiscal policy is ‘Ricardian equivalence’ (Barro, 1979) based on the permanent income hypothesis, the rational expectation assumption and non-liquidity constrained consumers. This hypothesis states that debt-financed government spending may have no impact on consumer spending, because the public will save the tax cut in order to pay for future tax increases that will be initiated to pay off the debt. New classical economists also argue that higher spending financed by lump-sum taxes induces a negative wealth effect, leading to a decrease in private consumption, a contemporaneous increase in labour supply and, therefore, a decrease in the marginal productivity of labour and in real wages, which results in higher output and employment. However, the multiplier is less than one due to the ‘crowding out’ effect of private consumption (Baxter & King, 1993). A simple NK- DSGE model, adding microfoundations and some market failure in the models of general equilibrium analysis of inter-temporal optimisation by rational economic agents, misses the Keynesian positive effect of fiscal expansion on consumption. Its predictions are in line with those of a RBC model: an increase in output and decrease in consumption, but the model predicts an increase in the real wage. However, several modification in DSGE by altering preferences, adopting non-separable utility or deep habits in consumption, introducing nonRicardian consumers and allowing for spending reversals may provide different predictions about the effectiveness of fiscal policy. If a model adopts a non-separable utility function in consumption and leisure, the negative wealth effect of the fiscal expansion raises hours worked, which decreases leisure, therefore, the marginal utility of consumption increases.
Hence, households will work more and consume more mitigating the negative wealth effect, leading to an increase in output, employment and consumption. Also, introducing habit persistence in consumption in a model where monopolistic competition is assumed in the goods market, will result in predicting no crowding out effect of fiscal policy in consumption.