The Quantification of the Effects of Structural Reforms in OECD Countries

Abstract

This chapter presents a new supply side framework that quantifies the impact of structural reforms on per capita income in OECD countries. It presents the overall macroeconomic impacts of reforms by aggregating over the effects on physical capital, employment and productivity through a production function. It is found that product market regulation has the largest overall single policy impact 5 years after the reforms. But the combined impact of all labour market policies is considerably larger than that of product market regulation. It is also shown that policy impacts can differ at different horizons.

The chapter benefitted from useful comments and suggestions from Andrea Bassanini, Gilbert Cette, Alain de Serres, Sean Dougherty, Falilou Fall, Andrea Garnero, Alexander Hijzen, Catherine L. Mann, Fabrice Murtin and Jean-Luc Schneider. A short version was published in OECD Economic Studies 2016(1), 91–108 and a longer version appeared as OECD Economics Department working paper No. 1354. The views expressed in the paper are those of the author and do not necessarily reflect the opinions of the OECD or any other institution the authors are affiliated with. The underlying work to this chapter has also been published in the OECD Economic Journal, 4(1), 91–108, 2016, under the title “The quantification of structural reforms in OECD countries: A new framework”.

Appendix: Calculating Total Policy Impacts on Per Capita Income

3.1.1 Theoretical Considerations

In the new framework, similarly to previous frameworks, structural policies affect per capita income through the supply side components. The appropriate aggregation across the components is straightforward in a standard neo-classical model with a Cobb-Douglas aggregate production of the following form:

$$ Y={K}^{\alpha }{(hL)}^{1-\alpha },\kern1em 0<\alpha <1 $$

(3.1)

with h denoting labour-augmenting (Harrod-neutral) technological progress. Note that the empirical construction of the MFP measure that is used for the estimations relies on the formulation in Eq. (3.1). ¹³ However, under the assumption of constant returns to scale, Eq. (3.1) can be rewritten in the following way:

$$ Y= MFP\left({K}^{\alpha }{L}^{1-\alpha}\right) $$

(3.2)

where there is a very close link between multi-factor productivity (MFP) and h:MFP = h ^1 − α. Introducing per capita measures and after\vadjust{\pagebreak} some rearrangements, per capita income can be expressed as a function of MFP, the capital-output ratio ($\raisebox{1ex}{$K$}\!\left/ \!\raisebox{-1ex}{$Y$}\right.$) and the employment rate ($\raisebox{1ex}{$L$}\!\left/ \!\raisebox{-1ex}{${N}_{wa}$}\right.$):

$$ \ln \left(\frac{Y}{N_{pop}}\right)=\frac{1}{1-\alpha}\ln (MFP)+\frac{\alpha }{1-\alpha}\ln \left(\frac{K}{Y}\right)+\ln \left(\frac{L}{N_{wa}}\right)+\ln \left(\frac{N_{wa}}{N_{pop}}\right) $$

(3.3)

where N _pop and N _wa stand for total population and working age population, respectively.

The advantage of this formulation is that in a standard setting, all components are separable and independent from each other. Specifically, the capital-output ratio does not depend on either productivity or employment, neither is the employment rate influenced by productivity or capital.¹⁴

For simulating the effects of changes in policies, the above equation will be used in growth rates:

$$ \Delta \mathrm{ln}\left(\frac{Y}{N_{pop}}\right){=}\frac{1}{1-\alpha}\Delta \mathrm{ln}(MFP){+}\frac{\alpha }{1-\alpha}\Delta \mathrm{ln}\left(\frac{K}{Y}\right){+}\Delta \mathrm{ln}\left(\frac{L}{N_{wa}}\right){+}\Delta \mathrm{ln}\left(\frac{N_{wa}}{N_p}\right)\kern0.75em $$

(3.4)

where Δ captures differences over time, which can be interpreted as percentage changes. As mentioned above, MFP in our empirical framework uses the Harrod-neutral specification. Hence Eq. (3.4) can be rewritten as follows:

$$ \Delta \mathrm{ln}\left(\frac{Y}{N_{pop}}\right)=\Delta \mathrm{ln}(h)+\frac{\alpha }{1-\alpha}\Delta \mathrm{ln}\left(\frac{K}{Y}\right)+\Delta \mathrm{ln}\left(\frac{L}{N_{wa}}\right)+\Delta \mathrm{ln}\left(\frac{N_{wa}}{N_p}\right) $$

(3.5)

Similar to the calculation of MFP a standard value for capital elasticity is set in the simulations (α = 0.33). The last term capturing the share of working age population will be assumed to be unchanged over the simulation horizon. Alternatively, demographic projections by the United Nations could be used over the projection horizon (long-term scenarios project of the OECD, see Johansson et al. 2013).

3.1.2 Practical Considerations

MFP and capital deepening are measured in logarithms, while the employment rate is measured in percentage points (between 0 and 100). The simulation framework requires that the reform impacts are expressed in log-points for each supply side component, Percentage point changes in the employment rate are thus transformed into log-points by dividing the changes in the employment rate by the latest observed employment rate for the working age population $\raisebox{1ex}{$L$}\!\left/ \!\raisebox{-1ex}{${N}_{wa}$}\right.$ (which was 67% in 2013, averaged across all countries in the sample):

$$ \Delta \mathrm{ln}\left(\raisebox{1ex}{$L$}\!\left/ \!\raisebox{-1ex}{${N}_{wa}$}\right.\right)=\frac{\Delta \left(\raisebox{1ex}{$L$}\!\left/ \!\raisebox{-1ex}{${N}_{wa}$}\right.\right)}{\raisebox{1ex}{$L$}\!\left/ \!\raisebox{-1ex}{${N}_{wa}$}\right.} $$

Another issue about aggregation is how to obtain the aggregate employment effect from the demographic and skill groups of the population. Policy effects for these groups are aggregated using the groups’ weight in the working age population. For the illustrative simulations presented in this paper, the population structure of the average OECD country is used in the latest available year.