یک تحقیق تجربی از مالیات بر حقوق و بیکاری در اقتصادهای باز و بسته

We investigate experimentally the economic effects of wage taxation to finance unemployment benefits for a closed economy and an international economy. The main findings are the following. (i) There is clear evidence of a vicious circle in the dynamic interaction between the wage tax and unemployment. (ii) Employment is boosted by budget deficits but subsequent tax rate adjustments to balance the budget lead to employment levels substantially lower than theoretically predicted. (iii) A sales risk for producers due to price uncertainty on output markets appears to cause a downward pressure on factor employment. For labor the wage tax exacerbates this adverse effect.

For more than two decades now unemployment has figured prominently on the political and economic agenda of many industrialized countries. Although a great number of theoretical and empirical studies exists and many proposals have been made, there is no consensus yet on how to solve the unemployment puzzle (Oswald, 1997).1 Welfare state arrangements and the accompanying tax burdens are generally considered as an important factor fostering unemployment. Using European economies as an example, Snower (2000) points out that these arrangements were established under economic fair-weather conditions. Once the weather started to deteriorate they more and more resembled quicksand. Rising unemployment led to higher social transfers, producing higher taxes on a shrinking tax base, which in turn negatively affected employment: “And so the cycle continues” (Snower, 2000, p. 44). Indeed, quite a few studies point at a positive feedback effect of general taxes and labor taxation (the tax-wedge) on unemployment. Good examples for this group of studies are Tullio (1987) and Daveri and Tabellini (2000). The former author presents an analysis of long-run consequences of tax and public debt increases in Europe and concludes that “tax-financed growth in government expenditure (…)(…) has caused unemployment” (Tullio, 1987, p. 769). In a similar vein, the latter authors provide empirical evidence that in Europe high labor costs are a main cause for high unemployment rates and that these high labor costs are mainly due to high taxes on labor. Whereas the mentioned studies clearly suggest the possibility of a vicious circle others contest its significance as a cause of unemployment. According to Oswald (1997): “Contrary to conventional wisdom, high unemployment does not appear to be primarily the result of overly generous benefits, trade union power, taxes, or wage inflexibility” (p. 14); and, more specifically: “Payroll taxes alone are uncorrelated with the pattern of unemployment across nations” (p. 8). Consistent with this view, OECD (1990, Annex 6.A) does not find any significant long-run influence of the total tax wedge on labor costs and by implication on employment. Similarly, Gruber (1997) reports that a pay-roll tax reduction of 25 percentage points, which took place in Chile in the early eighties of the last century, had only negligible effects on employment. Thus, notwithstanding a great number of empirical investigations,2 the literature seems still undecided regarding the actual vicious nature of labor taxation. Consequently, Nickell and Layard (1999), summarize very cautiously that “the balance of the evidence suggests that there is probably some overall adverse tax effect on unemployment and labor input.” ( Nickell and Layard, 1999, p. 3060; emphasis added). This may not be surprising in view of the many difficulties researchers face in the field concerning the adequate specification of an econometric model dealing with the complexity of full-scale economies and the serious lack of data (see e.g. Nickell and Layard, 1999; Nickell et al., 2005). From this perspective, it seems worthwhile to make use of a relatively novel, complementary approach: Laboratory experimentation, which is also empirical but allows one to get a more focused and controlled view of specific economic mechanisms, like the nexus between labor taxation and unemployment.3 Through the investigation of relatively simple economies—which are nevertheless real in the basic sense that real people make real choices with real consequences—additional and less noisy information may be obtained about this nexus and the behavioral adjustment processes that the vicious circle argument refers to. Of course, the abstraction exploited by this research method implies a trade-off with external validity that cannot be completely avoided.4 However, note that theoretical models in fact face a similar problem. Furthermore, as we will show below, our experimental design and the results obtained capture features that are also observed in naturally occurring economies. In our view, a particularly important contribution of laboratory experiments to macroeconomics can be that it may shed a new light on the question of the microfoundation of macroeconomics. Especially, behavioral aspects like the significance of norms or cognitive features such as loss aversion, hyperbolic discounting and bounded rationality in expectation formation can be fruitfully investigated with the help of laboratory experimentation (see e.g. Akerlof, 2002; Adam, 2005). Furthermore, it can generate interesting new empirical as well as theoretical research questions. This we also hope to show with our study. This study is based on a research project regarding the economic performance of tax systems commissioned by the Dutch Ministry of Social Affairs and Employment, following a motion carried by the Second Chamber of the Dutch parliament.5 To the best of our knowledge, it is for the first time that policymakers explicitly asked for laboratory experiments to support macro-economic policymaking. Our paper fits into a still relatively small but growing stream of ‘design’ studies, as Alvin Roth labels them in his Fisher Schultz Lecture (Roth, 2002). From a broader perspective, our study is part of an emerging research field showing the usefulness of macro-economic experiments as a complementary research tool next to the more traditional methods of theoretical and field empirical analysis.6 More specifically, we experimentally investigate the economic performance of economies operating under a wage tax financed unemployment benefit system, as it is common in many developed countries. Two different economic environments are investigated: A closed economy, and an international economy with a relatively small ‘home country’ and a large ‘foreign country’. The latter allows us to study a large and a small open economy simultaneously. In both cases there are input markets for capital and labor, and output markets for the goods produced by two production sectors. In the open economies environment, there are international markets for capital and one commodity. In each environment two tax regimes are imposed in sequence. During the first part of each experimental session the wage tax rate is held constant. This allows us to investigate whether the economies stabilize and, if so, at what level of the different economic variables. To analyze the dynamic interaction between the wage tax and unemployment as well as other indicators of economic performance, in the second part of each session, the tax rate is adjusted to the previous period's deficit in the budget of the tax–benefit system. In this paper, we are mainly interested in investigating whether wage tax financed unemployment benefits may produce a vicious circle boosting unemployment and deteriorating the performance of an economy as a whole. We therefore abstract in this first approach from other factors conducive to unemployment, like efficiency wages or institutions fostering insider–outsider effects. Furthermore, we will not distinguish between voluntary and involuntary unemployment because, for the empirical questions at hand, this distinction is “fruitless” (Layard et al., 1991, p. 91).7 Moreover, from a budgetary point of view it does not really matter much whether in the end the benefits have to be paid for voluntarily or involuntarily unemployed units of labor. To facilitate equilibration and efficiency we implement competitive double auctions for all markets. This trading mechanism has been shown to be very effective in fostering trade and equilibration in experimental markets (see e.g. Davis and Holt, 1993). In this way, we give the theory of competitive markets, that we will use as benchmark, its best chance to perform well. Moreover, finding a vicious circle in such a competitive environment would only strengthen the significance of the result.8 Our main findings are the following: First, for all economies we find clear evidence of a vicious circle in the dynamic interaction between the wage tax and unemployment. Second, employment seems to be boosted by deficits in the budget of the tax–benefit system. However, subsequent tax rate adjustments in order to balance the budget, as typically required by the wage tax financed unemployment benefit system, lead to employment levels that substantially fall short of the predictions obtained from the general equilibrium benchmark model. Third, and related to the previous point, there appears to be a downward pressure on the employment of production factors caused by a (disequilibrium) sales risk for producers. Due to uncertainty about output prices, and hence revenues, producers are reluctant to employ inputs. For labor this downward pressure on employment is exacerbated by the wage tax. Our results provide support for the hypothesis of a ‘risk-compensated price mechanism’ where the reluctance of producers to employ inputs is accompanied by (in comparison with the equilibrium predictions) too low input prices and too high output prices. This mechanism is not accounted for in general equilibrium models. Reliance on such models in policymaking would, therefore, lead to unexpected unemployment and disappointing economic performance. This may also help explain why one seems to have been taken by surprise by the vicious circle of wage taxation and unemployment. The organization of the remainder of the paper is as follows. Section 2 presents the experimental design and theoretical predictions. Results are given and discussed in Section 3. Section 4 concludes.

For the closed as well as the small and the large country of the international economy, we have found evidence for negative economic effects of wage taxation as a means of financing unemployment benefits. Our results provide empirical support for a vicious circle in the dynamic interaction between the wage tax and the unemployment rate. Furthermore, it turned out that employment is boosted by allowing a deficit in the budget of the tax–benefit system. Keeping the tax rate constant at the level predicted by the competitive equilibrium model, convergence towards the competitive equilibrium is observed for many variables. However, this development is accompanied by economically significant budget deficits. Once the wage tax is forced to adjust in the direction of a budget balance the employment level, as well as real GDP and other indicators of economic performance, gradually tend to stabilize at a level that substantially falls short of the equilibrium prediction. An important observation in this context is that there appears to be downward pressure on the employment of production factors, which is not accounted for in existing general equilibrium models. The uncertainty risk or ambiguity averse producers are facing when buying inputs before precisely knowing what they will make for their outputs can explain this phenomenon. Because of the uncertainty producers restrict the purchase of inputs, which restricts outputs. Consequently, there is a tendency for input prices to be lower, and for output prices to be higher than the equilibrium predictions. It is consistent with this so-called risk-compensated price mechanism that a wage tax exacerbates these effects. Our findings suggest that policymakers relying on the outcomes of (theoretical) models neglecting this mechanism would underestimate the negative effects of wage taxation. According to the European Commission (1994), plans for an alternative, employment friendlier fiscal structure deserve greater attention and serious study. Our study suggests that shifting taxation from labor, and more generally any kind of inputs, to outputs is worthwhile to be investigated. Another interesting issue for future research would be to investigate the effects of adjusting the unemployment benefit instead of the wage tax to balance the budget (cf. Rochetau, 1999). From a theoretical perspective, the contribution of our experimental study is twofold. Firstly, it suggests that a better understanding of the determinants of unemployment can be obtained by allowing for output price uncertainty and risk or ambiguity aversion in economic models. In this respect, this paper provides support for some existing partial equilibrium models bearing this out. By neglecting these issues, present general equilibrium models seem to foster a too rosy view of the economic effects of wage taxation. Secondly, and more generally, the results of our experiment suggest a new perspective on the micro foundation of macroeconomics. In simple market experiments it was already shown that behavioral aspects should not be disregarded (see e.g. Myagkov and Plott, 1997 on loss aversion, Ball et al., 2001 on status). Our experiment adds that behavioral factors can also be relevant for the outcome in more complex market environments. This indicates that the significance of other features (like norms, cognitive limitations, emotions, and bounded rationality) in complex settings can be fruitfully investigated with the help of laboratory experimentation. Further empirical regularities found by these experiments can generate new theoretical research questions and lead to new and better micro- and macro-economic general equilibrium models. From a broader methodological point of view, the technology developed for running macro-economic experiments opens up the possibility to study many other important issues in the lab, like public debt or the impact of labor market institutions, for example. In light of the findings obtained so far, macro-economic experiments seem to offer an interesting and challenging research tool which complements the more traditional theoretical and field empirical analysis.