Appendix 1. “Fiscal Devaluation”: What Is It—and Does It Work?
- International Monetary Fund. Fiscal Affairs Dept.
- Published Date:
- September 2011
The idea that tax reforms can mimic the effects of exchange rate devaluation is not new. For instance, the formation of the European Union prompted a recognition that moving from the taxation of goods on an origin basis (according to where they are produced) to a destination basis (according to where they are consumed) is essentially equivalent to an exchange rate devaluation, because imports are brought into tax, and exports are taken out. More generally, the potential for tax changes to affect both current and capital accounts has long been recognized.
What is new is the current focus on domestic tax changes as a potential response to difficulties in some euro area countries, in the specific form of a revenue-neutral shift from employers’ social contributions toward a value-added tax.1 Such a reform has come to be known as a “fiscal devaluation,” though there are other tax reforms that could equally well be called the same. It has been urged, in particular, as a way for Greece and Portugal to improve their competitiveness. This appendix reviews both theory and evidence in asking two questions: How would this work, and how large might the effects be?
How Might It Work?
With a fixed nominal exchange rate and a fixed nominal wage, a fiscal devaluation is expected to reduce the foreign currency price of exports and raise the relative consumer price of importables, thereby improving competitiveness. With a fixed money wage—more precisely, a fixed money wage net of employers’ social contributions (SCRs)2—a reduction in the rate of those contributions reduces unit labor costs and thus lowers producer prices, including those of exported goods and services. The higher VAT rate—a destination-based tax—bears on domestic consumption, but not on exports,3 so it offsets the impact on domestic consumer prices of the reduction in domestic producer prices, and it increases the consumer price of imports. Foreign demand for exports increases and domestic demand for imports falls; consequently, the current account improves—as it would with a depreciation of the real exchange rate. All this is not to say, of course, that tax policy is the best way to address the structural problems underlying wage rigidities: it is not. The point is rather that it can perhaps provide some temporary mitigation and smooth the impact of the more fundamental reforms required.
The effectiveness of this strategy requires rigidity in both the exchange rate and the nominal wage. With a flexible exchange rate, the increased demand for exports and reduced demand for imports will cause an appreciation of the nominal exchange rate that undoes the competitiveness impact of the tax shift. Even if the exchange rate is fixed, a fiscal devaluation will have no real effect if—or when—domestic wages adjust: as workers find their real wage reduced by the increased VAT rate, they (or their trade unions) will aim to increase their nominal wages, moving the real producer wage back toward the prereform equilibrium (a process that any wage indexation, of course, would accelerate). In the meantime, a fiscal devaluation would be expected to reduce unemployment, but because of the adjustment just described, with no longrun impact on product or labor markets. Box A1.1 elaborates.
Box A1.1Employment Effects of a Fiscal Devaluation
Denote by W the wage after the employer’s social contribution (at rate T), but before the VAT (at tax-inclusive rate τ): this is the wage that is assumed to be nominally rigid. The wage cost to the employer, on which labor demand depends, is thus W(1 + T), while the net wage received by the employee, on which labor supply depends, is W (1 − τ).1 With a flexible wage, the equilibrium in the figure below is at the wage W* and employment level L*, where L* =LD [W* (1 + T)] = LS[W*(1−τ)]. But with the wage fixed at W, there is initially unemployment of AB.
Suppose now there is a fiscal devaluation: a reduction in T combined with an increase in τ, calibrated to ensure that the initially employed continue to pay the same total tax, which, denoting postreform values by a prime, requires that
With the nominal wage fixed at W, employment expands to L′, closer to the initial full employment level. In the longer run, however—and in the absence of a minimum wage or other obstacles—the wage adjusts to clear the labor market. This will be at precisely the same level of full employment as before the reform, with a higher-wage presocial contribution of
ensuring equality of labor demand and supply at employment level L*. The fiscal devaluation thus accelerates the elimination of unemployment, mitigating the effect of nominal rigidities, but with no impact on the long-run equilibrium: the wage after the employers’ contribution increases just enough to offset the impact on workers of the higher VAT rate, leaving the real product wage unchanged.
This is a very partial view and would apply even in a closed economy. The benefits are likely to be greater in an open economy—the “pure” devaluation aspect—to the extent that demand shifts towards domestic tradables.1 For simplicity, the analysis abstracts here from personal income taxation and employee’s social contributions. Allowing for these, τ would become τ = (1 – τVAT)(1– τPIT)(1– τSCE).
|(1) Non-euro||(2) Euro||(3) Non-euro||(4) Euro|
|Net exports, lagged one year||−0.310***||−0.065||−0.126||0.046|
|Revenue from social security contributions by employers, change||−0.654**||−2.242*|
|Revenue from social security contributions by employers, lagged one year||−0.198||−0.420|
|VAT revenue, change||−0.683**||−1.799***|
|VAT revenue, lagged one year||−0.053||−0.209|
|Average tax rate on labor, change||0.161||0.286**|
|Average tax rate on labor, lagged one year||0.028||0.007|
|VAT rate, change||−0.207||0.471***|
|VAT rate, lagged one year||0.197**||−0.013|
|Total tax revenue, change||−1.907***||−0.923*|
|Total tax revenue, lagged one year||−0.183**||−0.047|
|Number of observations||407||114||130||105|
The fact that the effects of a fiscal devaluation may largely be temporary does not mean that they are irrelevant. This is particularly true when the economy, because of nominal downward rigidities in nominal wages, is initially in a disequilibrium position, with an overvalued real exchange rate and involuntary unemployment. In these conditions, a fiscal devaluation can speed up the adjustment in the labor market, which may otherwise take a long time to implement. The end result—the point at which the real exchange rate and the unemployment rate converge in the long run—may not be much affected by the fiscal devaluation (in this sense the effects would be temporary), but the speed of convergence can be much faster. This faster speed of adjustment is critical in countries where doubts may otherwise arise about the sustainability of the adjustment process under a pegged exchange rate. The analogy between a “fiscal devaluation” in the sense above and nominal currency depreciation is imperfect. For instance, the proportionate impact on export prices of a nominal depreciation would be the same for all commodities, but that of a fiscal devaluation will be greater for more labor-intensive products. And if nontradables tend to be more labor intensive than tradables, this will mitigate the shift of resources into the latter.
Beyond the Basics
While the basic theory is elegant, important complexities arise in practice:
A fiscal devaluation reduces the value to the consumer of nonlabor income, whether from transfers or capital income, affecting the labor market and income distribution. The analysis in Box A1.1 ignores any impact of reform on incomes of those outside the labor market. To the extent, however, that out-of-work benefits to the unemployed are not uprated to reflect the increased VAT, most labor market models suggest a long-term fall in structural unemployment. Moreover, an increase in the real consumer wage in the long run, because of the shift in the tax burden from labor to nonlabor income, will boost labor supply incentives. But there will also be a reduction in the real value of pensions, for instance. To the extent that some benefits are uprated in recognition of adverse equity effects, this will dilute the revenue raised by the increase in the VAT rate and so allow only a smaller reduction in SCRs and hence a smaller gain in employment and labor supply.
The precise effects on the labor market and income distribution will depend on how the reform is designed, with a case for focusing SCR cuts on lower wage levels. Increasing the VAT—whether by raising the standard rate or raising reduced rates—is generally slightly regressive. At the same time, the distributional effect of reducing SCRs will depend on whether or not the upper limit on such contributions that some countries impose is also reduced. More generally, there may be a case for focusing the SCRs cut at lower wage levels, as there is reason to suppose employment is more sensitive to tax considerations in the lower part of the wage distribution. Net tax relief for low-paid workers improves both equity and efficiency by improving incentives for labor market participation where distortions are largest.
However, labor market distortions induced by SCRs may be smaller than the analysis above supposes if SCRs are perceived as having an actuarial link with benefits. Unlike personal income taxes, social contributions often carry some benefit entitlement. The link may be more apparent than real, but there is evidence that it is the perception that matters for labor supply responses (Disney, 2004). Cutting this link may thus exacerbate labor market distortions. Moreover, the responsibility for social funds in many countries is shared between the government and organizations of employers and employees, raising further practical complexities.
Conditions for the VAT to be trade-neutral are stringent. A uniform VAT applied at the same rate to all consumption items has no impact on relative consumer prices. Leaving aside possible income effects, it would thus have no effect on demand or hence on trade. But nonneutralities arise if—as is the case almost everywhere, especially in Europe—VAT rates differ sharply across commodities. Feldstein and Krugman (1990) argue, for instance, that tradables are generally taxed more heavily under the VAT than nontradables; a higher standard VAT rate then reduces the relative consumer price of nontradables, encouraging substitution out of tradables.4 It is unclear how realistic the presumption is, but in the EU, nine member states currently apply reduced rates to nontradable labor—intensive services. A fiscal devaluation would favor nontradables not only through the VAT effect but also through labor intensity considerations. In other cases, however, reduced rates apply to tradables, such as zero-rated food in the United Kingdom. The net direction of the impact of an increase in the VAT rate on net exports is then unclear. Nonetheless, these considerations suggest some case for focusing on raising the reduced rates. This, though, may call for compensating measures to protect low-income households, reducing the affordable reduction in the SCR.
Compliance and timing issues also need consideration. A higher VAT rate may exacerbate tax evasion and avoidance, especially where the standard rate is already high: in Greece and Portugal the standard rate is already 23 percent. This is another reason to focus any VAT increase on raising reduced VAT rates on specific products. On the other hand, the reduction in SCR rates would be expected to improve compliance, though on balance VAT noncompliance is likely the greater problem. If the higher VAT rate is announced in advance, consumers may bring forward consumption, with a temporary adverse impact on the current account. There is ample evidence of such effects.5
The effects of a fiscal devaluation will be smaller if such a devaluation is undertaken by several countries—but the final effect could still be beneficial. As with a competitive nominal devaluation, the real impact on any country is reduced if the same reform is undertaken elsewhere. Fiscal devaluation can be seen in this light as a form of tax competition—the outcome of which is often mutually damaging. In the current context, however, a shift from SCR to the VAT may be a structural improvement, given evidence that that the tax wedge on labor harms growth more than do taxes on consumption (Arnold, 2008).
What Is the Evidence?
Simulations suggest that fiscal devaluations have small but positive effects on the current account.6 For example:
The Bank of Portugal (2011) uses a general equilibrium model (PESSOA) to simulate a shift from SCR to VAT equivalent to 1 percent of GDP. In the first year, this boosts total exports by 0.5 percent and improves the trade balance by 0.6 percent of GDP. After three years, the effect on the trade balance has disappeared. Yet there is a sustained modest increase in GDP and employment, caused by an expansion of labor supply associated with a shift in the tax burden from labor to nonlabor incomes.
The European Commission (EC) uses the QUEST model to simulate a similar shift in Portugal (In’t Veld, 2011). Rescaling to a shift equivalent to 1 percent of GDP, net exports increase in the short run by 0.11 percent of GDP. The effect gradually disappears, and the long-term effect on net exports is negligible. In the long term, the reform boosts employment and GDP by almost 1 percent after five years. This assumes no compensation of social benefit recipients for the higher VAT; with compensation, the expansion of GDP and employment is 0.3 percent, although this increases to 0.6 percent if the elasticity of labor supply is doubled.
The European Central Bank (ECB) (2011) applies its EAGLE model to simulate a fiscal devaluation in Portugal. Taking the version with a trade elasticity of 2.5, it reports an improvement in the trade balance in the second year of almost 0.2 percent of GDP (for a tax shift of 1 percent of GDP). The effect disappears after four years. The model also suggests more persistent gains in employment and GDP in the order of 0.3 percent, as in the other models.
For France, the Ministry of Economy, Finance, and Industry (2007) finds that a 1.5-point increase in the VAT rate used to finance a general cut in social contributions raises employment by 30,000. That rises to 300,000 if focused on the lower paid.
These results, of course, are driven by the model specifications and calibrations. Is there any evidence that fiscal devaluations work in practice?7
New empirical evidence on the impact of fiscal devaluations suggests that the impact is consistent with simulation predictions (Table A1.1). Using an unbalanced panel of annual observations for 30 OECD countries between 1965 and 2009, de Mooij and Keen (2011) regress the change in net exports (in percent of GDP) against (changes in and first lag of) a series of controls8 and, the focus of interest, variables relating in turn to (a) the revenue from the SCR and the VAT and (b) rates of labor taxation and VAT. To allow for heterogeneity of response arising from fixed exchange rates, separate estimates are reported for observations within and outside the euro area.
There are signs that changes in both the SCR and—surprisingly—the VAT have significant trade effects, noticeably more so in the euro area. A decrease in SCR revenue is associated with a short-run increase in net exports, as expected. Moreover, the coefficients are much larger in the euro countries (column 2 of the table), and the hypothesis that both coefficients are zero is more decisively rejected. The results using tax revenue may however be subject to serious endogeneity issues: a consumption boom, for instance, might be associated with both higher VAT revenues and lower net exports, giving rise to a misleading negative correlation. In this respect, the results reported in columns (3) and (4), using rates rather than revenue (at the cost of a smaller sample size), may be preferred. Now the signs of significant tax effects are less convincing for non—euro countries, but both variables are strongly significant in the euro case. The labor tax variable now refers to the total tax wedge on wage income—not just social contributions—so the positive coefficient now found likely reflects a wider range of effects than those operating through employers’ labor costs.
The results—while very preliminary—suggest that the trade effects of fiscal devaluation, while not trivial, should not be overestimated. Taking the results in column (2) of Table A1.1, for instance, the effect of a revenue shift of one point of GDP from SCR to VAT is an increase in net exports of 0.443 (= 2.242 - 1.799) points of GDP. This would be permanent (the coefficient on the lagged dependent variable being insignificant), though the induced reduction in unemployment (one of the control variables) might be expected to dampen the effects over time. To perform a similar calculation from the rates equation, estimates of the base of each tax are needed. Taking Portugal, for example, a shift of the same magnitude might imply an increase in net exports—permanent, subject to the same caveat as above—of around 0.3 percent of GDP.9 These effects are similar to those from the simulation models: slightly larger than those of the EC and the ECB, and slightly smaller than those of the Bank of Portugal model.
These results are tentative and, as noted, econometric issues remain. Nonetheless, they reinforce the sense from both theory and simulations that fiscal devaluations can have significant effects—but large shifts are likely needed for effects to be substantial. Moreover, initial conditions may matter, in a way not captured by either simulation models or empirical estimates. As noted, in countries that combine a highly rigid labor market with a fixed exchange rate, reducing involuntary unemployment through a reduction in real wages (and boosting net exports through a real depreciation) may take a long time. A fiscal devaluation might generate especially larger benefits from the acceleration of adjustment—an effect that is not likely to be captured by the estimates above.
Calmfors (1998) long ago recongnized the potential to undertake such a reform for countries adopting the euro.
The focus on SCRs in this context reflects the view that the relevant rigidity comes from contracts specific in terms of payment after SCRs. If it was just the net wage received by the worker that was fixed, a cut in the employee’s contribution or personal income tax would do just as well.
Destination-based taxes other than the VAT could also play a role. Excises, for instance, have precisely the same trade-neutrality property as the VAT, being charged on imports but remitted on exports. Recurrent taxes on residential property seen as widely underused in many countries—may have similar appeal. These other possibilities, important in practice, are not pursued here.
To the extent that nontradables tend to be labor-intensive services, this will amplify the effect though the lower SCR.
See for example Keen and others (2011).
There are several notable fiscal devaluations (Denmark in 1988, Sweden in 1993, and Germany in 1997), but it is difficult to identify any causal relationship from them: Calmfors (1998) discusses.
Franco (2011) estimates a number of VAR equations with data from Portugal and then simulates the impact of an SCR reduction and an offsetting increase in the VAT. Scaled to a shift of 1 percent of GDP, the simulations suggest that this reform would result in a reduction of imports by 3.5 percent of GDP and a similar increase in exports. Net exports would thus expand by 7 percent of GDP—much more than in other simulation models.
Including, for instance, unemployment and growth, the government balance, the old-age dependency ratio, and controls for unobserved time and country variation.
In Portugal, the base of the labor income taxation implied by the OECD data is around 33 percent of GDP, while the implied VAT base (the ratio of revenue to GDP divided by the standard rate) is 41 percent. So revenue neutrality requires that a reduction of one point of GDP in revenue from labor taxation, calling for a cut in the average labor tax rate of 3 percentage points, be combined with an increase of about 2.4 points in the standard rate of VAT. The point estimates then imply an increase in net exports of - (0.286 x 3) + (0.471 x 2.4) = 0.27 percent of GDP.