Beyond the labour income tax wedge: the unemployment-reducing effect of tax progressivity

Abstract

In this paper, we argue that, for a given overall level of labour income taxation, a more progressive tax schedule increases employment. From a theoretical point of view, higher progressivity increases overall employment through a wage moderating effect and also because employment of low-paid workers is more elastic to wages. We test these theoretical predictions on a panel of 21 OECD countries over 1998–2008. Controlling for the burden of taxation at the average wage, our estimates suggest that a more progressive tax schedule reduces the unemployment rate and increases the employment rate. These findings are confirmed when we account for the potential endogeneity of both average taxation and progressivity. Overall, our results suggest that policy-makers should not only focus on the detrimental effects of tax progressivity on in-work effort, but also consider the employment-enhancing effects.

Appendices

Appendix 1: Theoretical model

We here consider the general model with an intensive labour supply margin. Let $\ell$ be the effort provided by employed individuals and the money-equivalent of this disutility.³² One can think of $\ell$ as hours of work or as the intensity of in-work effort. We assume that providing effort $\ell$ leads to a flow of output $f(\ell )$, with $f^\prime (.)>0>f^{\prime \prime }(.)$ and a flow of disutility $\ell$. Denoting w the total gross wage (and not the wage rate), Eq. (4a) becomes:

$$\begin{aligned} r\ E= w-T(w)-\ell + \delta \left( U-E\right) \end{aligned}$$

(10)

The case without the intensive margin is retrieved by setting the disutility $\ell =0$ and $f(\ell )=y$. Equations (2) and (10) imply, respectively:

$$\begin{aligned} J=\frac{f(\ell )-w}{r+\delta } \qquad \text {and} \qquad E-U=\frac{w-T(w)-\ell -r\ U}{r+\delta } \end{aligned}$$

(11)

Maximising the (log of) the Nash product with respect to the wage and in-work effort leads to:

$$\begin{aligned} \frac{\gamma \left( 1-T^\prime (w)\right) }{E-U}=\frac{1-\gamma }{J} \qquad \text {and} \qquad \frac{\gamma }{E-U}=\frac{\left( 1-\gamma \right) f^\prime (\ell )}{J} \end{aligned}$$

Using (6) and $\Psi =\Psi (w)$, the first of these two equations gives the sharing rule (5). The second of these equations implies the labour supply condition:

$$\begin{aligned} \left( 1-T^\prime (w)\right) f^\prime (\ell )=1 \end{aligned}$$

(12)

which determines effort $\ell$ as a decreasing function of the marginal tax rate, independently of any other variable. Combining the sharing rule (5) with (11) leads to:

$$\begin{aligned} \gamma \ \Psi \left( f(\ell )-w\right) =\left( 1-\gamma \right) \left( w- \frac{\ell +r\ U}{1-\tau }\right) \end{aligned}$$

Moreover, we get from (4b) and the sharing rule (5):

$$\begin{aligned} \frac{r\ U}{1-\tau }=\frac{b}{1-\tau }+ p(\theta ) \frac{\gamma \ \Psi }{1-\gamma } J = \frac{b}{1-\tau }+c\frac{\gamma \ \Psi }{1-\gamma } \ \theta \end{aligned}$$

(13)

where the second equality uses the free-entry condition (3). The (gross) wage equation is

$$\begin{aligned} w=\frac{\gamma \ \Psi }{1-\gamma +\gamma \ \Psi } \left( f(\ell )+c\ \theta \right) + \frac{1-\gamma }{1-\gamma +\gamma \ \Psi } \frac{b+\ell }{1-\tau } \end{aligned}$$

which gives (7a) in the case of exogenous labour supply by taking $\ell =0$ and $f(\ell )=y$. Equation (7b) follows directly. Combining this wage equation with the free-entry condition (3) leads to:

$$\begin{aligned} \left( 1 + \frac{\gamma \ \Psi }{1-\gamma }\right) \frac{r+\delta }{q\left( \theta \right) } +\frac{\gamma \ \Psi }{1-\gamma } \theta =\frac{1}{c}\left( f\left( \ell \right) -\frac{b+\ell }{1-\tau }\right) \end{aligned}$$

(14)

Equation (14) determines implicitly equilibrium tightness $\theta$ as a decreasing function of the average tax rate $\tau$ and of the CRIP $\Psi$, while the sign of the relationship with in-work effort $\ell$ is ambiguous. Using (1), it also determines the fraction $1-u$ of participants that are employed.

With an endogenous labour supply margin, productivity is a function of effort and the disutility of effort plays a role similar to the value in unemployment b. For a given level of effort $\ell$, a rise in progressivity (a decrease in the CRIP $\Psi$) increases employment through the same wage moderating effect. Moreover, a rise in tax progressivity decreases effort $\ell$. This reduction has two opposite effects on the total surplus of a match. On the one hand, production decreases. On the other hand, the disutility of work also decreases. The net effect of effort $\ell$ on employment depends on how the term $f\left( \ell \right) -(b+\ell )/(1-\tau )$ in the right-hand side of (14) varies with $\ell$. Using (12), a reduction in effort $\ell$ increases (decreases) this term whenever taxation is regressive (progressive) i.e. when $\Psi >1$ (i.e. $\Psi <1$). Therefore, by continuity, and provided that the taxation is not too progressive so that the labour supply effect of tax progressivity remains dominated by the wage moderating effect, a rise in tax progressivity (a reduction in the CRIP $\Psi$) increases employment and decreases the unemployment rate.

Let us denote $X \equiv \frac{\gamma \ \Psi }{1-\gamma }$. From (14), the effect of the CRIP $\Psi$ on tightness happens only through a change in X. A rise in the CRIP $\Psi$ holding the average tax rate $\tau$ constant increases participation only if it increases the value in unemployment. Rewriting (13) as:

$$\begin{aligned} r\ U=b+c\ X\ \theta (1-\tau ) \end{aligned}$$

implies that a rise in the CRIP for a fixed $\tau$ increases participation if and only if $X\ \theta$ is increasing in X. Denoting the elasticity of the job filling rate by $\eta (\theta )\equiv -\theta q^\prime (\theta )/q^\prime (\theta )$ and differentiating (14) in $\theta$ and in $X=\frac{\gamma \ \Psi }{1-\gamma }$ gives:

$$\begin{aligned} \frac{d\theta }{\theta }=- \frac{{X\ \frac{r+\delta }{q(\theta )}+X\ \theta }}{\eta (\theta )(1+X)\ \frac{r+\delta }{q(\theta )}+X\ \theta } \ \frac{dX}{X} \Rightarrow \quad \frac{d(\theta \ X)}{\theta \ X}=\frac{{\left[ \eta (\theta )(1+X)-X\right] \frac{r+\delta }{q(\theta )}}}{\eta (\theta )(1+X)\ \frac{r+\delta }{q(\theta )}+X\ \theta } \ \frac{dX}{X} \end{aligned}$$

Hence, $X\ \theta$ is increasing in X, and thereby participation is decreasing in progressivity, if and only if $X<\eta (\theta )(1+X)$. Given the definition of X, the latter condition is equivalent to:

$$\begin{aligned} \frac{\gamma \ \Psi }{1-\gamma +\gamma \ \Psi } < \eta (\theta ) \end{aligned}$$

that is to effective bargaining power being lower than the efficient one prescribed by the Hosios (1990) condition (see Pissarides (2000)), i.e. by unemployment rate being inefficiently low.

Appendix 2: Data appendix

Main variables used in the analysis

UNR: unemployed persons divided by the labour force (harmonised; OECD economic outlook).

Urate: unemployed persons divided by the labour force (non-harmonised; OECD Employment Database based on National Labour Force Surveys).

Erate: persons in employment divided by the working age population (non-harmonised; OECD Employment Database based on data from National Labour Force Surveys).

Prate: persons in the labour force divided by the working age population (non-harmonised; OECD Employment Database based on data from National Labour Force Surveys).

ATR, 67, 100, 167 AW: average tax rates (ATR) including taxes, social security contributions (net of cash benefits received) for the average worker (single person, no child) at 67, 100 and 167 % of average earnings. These data are drawn from the OECD tax Database for the years 2000–2008 and extended back to 1997, using information from OECD Taxing Wages (historical model B). The two datasets are constructed using the same methodology, the only difference being that computations from Taxing Wages historical model B use as a benchmark the average production worker (APW) rather than the average worker (AW). We used the common support of the two datasets (i.e. the years 2000–2004) to rescale the ATRs from OECD Taxing Wages to the corresponding ATRs from the OECD Tax Database. This allowed us to exploit the yearly variation of OECD Taxing Wages’ ATRs between years 1999/2000, 1998/1999, and 1997/1998 to extend the relevant time series from the OECD Tax Database back to 1997.

Instruments

NoTrust: percentage of respondents that give answer 4 (i.e., “none at all’) to questions $E069\_8$ in WVS1-5, V212 in EVS4, V207 in EVS3, q553i in EVS2, v546 in EVS1 (how much confidence in civil service). The period is as follows:

1980–1989: coverage by EVS1/WVS1 but for CHE, covered by EVS2. Surveys carried in 1981 for AUS, BEL, DEU, DNK, ESP, FIN, FRA, GBR, IRE, JPN, NLD; 1982 for CAN, NOR, NOR, SWE, USA;

1990–1994: coverage by EVS2/WVS2. Surveys carried in 1990 for AUT, BEL, CAN, DEU, DNK, ESP, FIN, FRA, GBR, ITA, JPN, NLD, NOR, PRT, SWE, USA. Notice that we have two observations for ESP (in year 1990) corresponding to both WVS2 and EVS2 being carried that year.

1995–1999: coverage by EVS3/WVS3. Surveys carried in 1995 for AUS, ESP, JPN, USA; 1996 for CHE, FIN, NOR, SWE; 1997 for DEU; 1998 for GBR, BEL, GBR, NZL; 1999 for AUT, BEL, DEU, DNK, ESP, FRA, GBR, GRC, IRE, ITA, NLD, PRT, SWE, USA. Notice that we have two observations for ESP (1995 and 1999), DEU (1997 and 1999), GBR (1998, 1999) and USA (1999), corresponding to both WVS3 and EVS3 being carried in those countries.

2000–2004: coverage by WVS4 but for FIN and NZL, covered by EVS3 and WVS5, respectively. This period is generally not covered by any EVS wave, thus the majority of European countries is not surveyed. Surveys carried in 2000 for CAN, ESP, FIN, JPN; 2004 for NZL.

2005–2008: coverage by EVS4/WVS5. Surveys carried in 2005 for AUS, FIN, ITA, JPN; 2006 for CAN, DEU, FRA, GBR, NLD, SWE, USA; 2007 for CHE, ESP; 2008 for AUS, CHE, DEU, DNK, ESP, FRA, GRC, IRE, NLD, NOR, PRT. Notice that we have two observations for AUS (2005 and 2008), CHE (2007 and 2008), DEU (2006, 2008), ESP (2007, 2008), FRA (2006, 2008), NLD (2006, 2008), POL (2005, 2008), corresponding to both WVS5 and EVS4 being carried.

Observations were averaged out by country and period thus obtaining an unbalanced panel of 21 countries covering the period 1990–2008 in five years averages. Missing observations were obtained by linear interpolation. The initial observation covering the period 1980–1989 has not been used in the empirical analysis, but provided the basis to obtain the observation for the period 1990–1994 by linear interpolation for countries where observations were missing for this period.

Leftism: Difference between the shares in legislative seats of left-wing and centrist parties minus the share in legislative seats of right-wing parties. Authors’ calculations using data from “Electoral, Legislative, and Government Strength of Political Parties by Ideological Group in Capitalist Democracies, 1950-2006: A Database”, by Duane Swank. See http://www.marquette.edu/polisci/faculty_swank.shtml.

Taxconsol: Cumulated sum of documented tax increases drawn by historical sources and records based on the methodology developed by Romer and Romer (2010), drawn by Devries et al. (2011) (see http://www.imf.org/external/pubs/cat/longres.aspx?sk=24892.0).

For estimates presented in Table 5 (Columns 2a and 2b), observations for the missing countries (Greece, Norway, New Zealand, Switzerland) have been reconstructed applying the methodology by Guichard et alii 2007 to the cyclically adjusted current receipts, general government, as a percentage of potential GDP (OECD Economic Outlook).

Pcdebt: General government gross financial liabilities, as a percentage of GDP (OECD Economic Outlook).

Other variables used in the analysis

Unemployment rates by skill level: Unemployed persons divided by the labour force by age group and educational attainment. Low-skilled workers completed up to secondary education, while high-skilled workers completed tertiary education or more. Young workers have an age comprised between 15 and 24 years. Old workers have an age comprised between 25 and 54 years. These are authors’ calculation based on World Bank’s Development Indicators.

EPL: Unweighted sum of the OECD synthetic index of employment protection legislation (OECD Employment Outlook).

UnionDensity: Union density (% of unionised workers; OECD Employment Outlook).

UBRR: Average unemployment benefit replacement rates (average of replacement rates across various earnings levels, family situations and durations of unemployment; OECD Benefits and Wages Database).

Wcoord: Coordination of wage bargaining (classification is based on Kenworthy’s 5-point classification of wage-setting coordination scores; ICTWSS Database).

Irate: Long-term interest rate on government bonds (OECD Economic Outlook).

Outgap: Percentage deviation of output from trend (OECD Economic Outlook).

Inflchange: Change in the inflation between two consecutive years (authors’ calculation using data from the OECD Economic Outlook).

Openness: The trade-to-GDP ratio is the sum of exports and imports of goods and services relative to GDP (OECD International Trade Indicators database).

Appendix 3: First-stage results

See Table 6.

Table 6 First-stage results