Let's decompose the data and underline the main findings. Tax wedge, measured as a percentage of taxes and transfers paid in the share of total labor costs, is a suficient measure of the overall level of taxation. In the analysis of the effect of taxation on labor supply, the economic theory distinguishes between substitution and income effect. The former means that tax rate on labor supply would reduce the number of working hours and shift the individuals to allocate more time into leisure. The income effect, on the other hand, states that the effect of tax on labor supply would be neutral and would, hence, not have an effect on the relative allocation of resources between working hours and leisure. In general, there are two largely opposing views in economic policy regarding the relationship between taxation and labor supply. Conservative and liberal economists tend to emphasize the role of incentives. Higher tax rate would raise the labor cost and the corresponding decline in wages would be offset by the re-allocation of resources into leisure sector of the economy. Left-wing economists mostly disagree with the abovementioned proposition. Instead, they emphasize the role of elasticity of labor supply. According to New Keynesian view, short-run tax elasticity of labor supply is low in absolute terms, meaning that the amount of working hours does not respond significantly to relative changes in tax rates. The Keynesian economists thus emphasize the significance of income effect while the conservative and liberal economists tend to emphasize the substitution effect. The distinctions and policy effects of these two theoretical propositions remains a controversial issue of economic policy debate.

The country distribution data is extensively underlined in the data (link). For example, one-earner married couple at 100 percent of earnings distribution is taxed at 20 percent of the overall labor cost in the United States, 22 percent in Australia, 18 percent in Switzerland and 13 percent in New Zealand. On the other tax wedge statistics for Western Europe is a completely different picture. In Sweden, one-earner married couple at 100 percent of earnings distribution will earn only 57 percent of net earnings, indicating a 43 percent effective tax rate on labor supply. In Belgium, the effective tax rate on labor supply is 42.6 percent. Is the difference in effective tax rates statistically significant feature of productivity variation across OECD countries. The OECD recently composed a breakdown in key productivity statistics in developed countries (link). To estimate the relationship between productivity and tax burden, I collected data on multifactor productivity dynamics from Groningen Growth and Development Center (link) and data on tax wedge across OECD countries (link). I estimated the noted relationship for 16 developed OECD countries. Multifactor productivity is a dependent variable while tax wedge is an explanatory variable.The estimates are displayed on the graph below.

Multifactor productivity in OECD relative to the U.S. and total tax burden (% of labor cost)

Source: GGDC, OECD Factblog

The horizontal line on the graphs marks the U.S. level of multifactor productivity. As the graph shows, the underlying relationship between multifactor productivity (MFP)and tax wedge is fitted with quadratic equation. Total tax wedge explains about 34.5 percent of the variation in multifactor productivity index across OECD countries. The trend exerts a decreasing MFP as the tax wedge rises and, after reaching a local minimum, a slight increase alongside the proporional rise in tax wedge. The relevant question is at which rate of tax wedge the MFP reaches the minimum. Setting the first-order derivate to zero (dy/dx=0) yields 0.0016x-0.0488=0. Rearranging the equation yields x=30.5. The first-order derivative implies that MFP reaches the minimum at 30.5 percent tax wedge. The estimate sample-based tax elasticity of productivity ((dy/dx)(x/y)) is 1.78. The elasticity indicates that 1 percentage point increase in tax wedge would reduce the multifactor productivity in a sample country by 1.78 percent, ceteris paribus. In other words, the estimated slope coefficient suggests that a 1 percentage point increase in tax wedge would widen the MFP gap between the average country and the US by additional 0.05 index points.

In economic terms, MFP would decrease as long as the countries in the sample would exert less than 30.5 percent tax wedge. In the horizon after the local minimum (30.5 percent), MFP would initially increase but at a smaller rate than before the function would reaches the minimum point. At 20 percent tax wedge (close to Japanese level), the expected decrease in MFP would be 0.0168 index point. At 40 percent tax wedge, the expected increase in MFP would be 0.0152 index point. Economically, the relationship between MFP and tax wedge should not be interpreted as a mixed effect of tax wedge on MFP. In countries, where tax wedge exceeds 30.5 percent (Austria, Denmark, Sweden, Belgium etc.) , the multifactor productivity is close to the U.S. level. The only country with higher MFP level than the United States is Luxembourg. In these countries, the average number of annual hours worked per worker is lower than in the U.S. Hence, hourly productivity output is higher. The following graph illustrates the relationship between annual hours worked (per worker) and MFP level.

Average annual hours worked per worker and multifactor productivity relative to the US

As the graph illustrates, there is a negative relationship between average annual hours worked and MFP. The estimated regression coefficient suggests that an increase in annual hours worked by 100 hours would reduce MFP level (compared to the U.S) by 0.06 index points. As expected, the implied elasticity of productivity is negative. Taking the average values ov both variables, an increase in average annual hours worked (per worker) by 1 percent would reduce the MFP relative to the U.S level by 1.16 percent, ceteris paribus.

The conclusion is that there is a significant and negative impact of tax wedge on multifactor productivity. Taking the differences in annual hours worked into account, tax wedge alone explains almost 35 percent of the differential in multifactor productivity across the OECD. The findings suggest that there is a relatively strong substitution effect in labor supply. The finding is underlined by the fact that higher tax wedge would correspondingly reduce annual hours worked. However, the interpretation should be taken with a grain of warning. As previous empirical studies have shown, there is a widespread disparity in the distribution of working hours in formal and household sector of the economy. A sizeable proportion of working hours in household sector of the economy is not officially measured. Consequently, a blick of distortion of the real relationship between labor supply and tax wedge, in its broadest sense, is a major impediment to the measurement of labor supply dynamics. These estimates will, in a large part, determine the future research of the effect of taxation on labor productivity.

The horizontal line on the graphs marks the U.S. level of multifactor productivity. As the graph shows, the underlying relationship between multifactor productivity (MFP)and tax wedge is fitted with quadratic equation. Total tax wedge explains about 34.5 percent of the variation in multifactor productivity index across OECD countries. The trend exerts a decreasing MFP as the tax wedge rises and, after reaching a local minimum, a slight increase alongside the proporional rise in tax wedge. The relevant question is at which rate of tax wedge the MFP reaches the minimum. Setting the first-order derivate to zero (dy/dx=0) yields 0.0016x-0.0488=0. Rearranging the equation yields x=30.5. The first-order derivative implies that MFP reaches the minimum at 30.5 percent tax wedge. The estimate sample-based tax elasticity of productivity ((dy/dx)(x/y)) is 1.78. The elasticity indicates that 1 percentage point increase in tax wedge would reduce the multifactor productivity in a sample country by 1.78 percent, ceteris paribus. In other words, the estimated slope coefficient suggests that a 1 percentage point increase in tax wedge would widen the MFP gap between the average country and the US by additional 0.05 index points.

In economic terms, MFP would decrease as long as the countries in the sample would exert less than 30.5 percent tax wedge. In the horizon after the local minimum (30.5 percent), MFP would initially increase but at a smaller rate than before the function would reaches the minimum point. At 20 percent tax wedge (close to Japanese level), the expected decrease in MFP would be 0.0168 index point. At 40 percent tax wedge, the expected increase in MFP would be 0.0152 index point. Economically, the relationship between MFP and tax wedge should not be interpreted as a mixed effect of tax wedge on MFP. In countries, where tax wedge exceeds 30.5 percent (Austria, Denmark, Sweden, Belgium etc.) , the multifactor productivity is close to the U.S. level. The only country with higher MFP level than the United States is Luxembourg. In these countries, the average number of annual hours worked per worker is lower than in the U.S. Hence, hourly productivity output is higher. The following graph illustrates the relationship between annual hours worked (per worker) and MFP level.

Average annual hours worked per worker and multifactor productivity relative to the US

As the graph illustrates, there is a negative relationship between average annual hours worked and MFP. The estimated regression coefficient suggests that an increase in annual hours worked by 100 hours would reduce MFP level (compared to the U.S) by 0.06 index points. As expected, the implied elasticity of productivity is negative. Taking the average values ov both variables, an increase in average annual hours worked (per worker) by 1 percent would reduce the MFP relative to the U.S level by 1.16 percent, ceteris paribus.

The conclusion is that there is a significant and negative impact of tax wedge on multifactor productivity. Taking the differences in annual hours worked into account, tax wedge alone explains almost 35 percent of the differential in multifactor productivity across the OECD. The findings suggest that there is a relatively strong substitution effect in labor supply. The finding is underlined by the fact that higher tax wedge would correspondingly reduce annual hours worked. However, the interpretation should be taken with a grain of warning. As previous empirical studies have shown, there is a widespread disparity in the distribution of working hours in formal and household sector of the economy. A sizeable proportion of working hours in household sector of the economy is not officially measured. Consequently, a blick of distortion of the real relationship between labor supply and tax wedge, in its broadest sense, is a major impediment to the measurement of labor supply dynamics. These estimates will, in a large part, determine the future research of the effect of taxation on labor productivity.