Working Paper Series no. 674: Rent creation and sharing: new measures and impacts on TFP

Extensive empirical literature based on cross-country-industry panel data has been devoted in recent years to the analysis of the impact of competition on productivity. Many of the papers concerned draw on the OECD's anti-competitive Non-Manufacturing Regulation (NMR) indicators to estimate the productivity impact of the lack of competition in product markets. A few papers have also used the OECD's Employment Protection Legislation (EPL) indicators to gauge the productivity impact of the lack of competition or flexibility in labor markets. In the present study, we propose new measures of rent creation or (notional) mark-up and workers’ share of rents on cross-country-industry panel data to estimate both types of productivity impacts and use the OECD regulations indicators as relevant instrument to take care of endogeneity and to make sure that the resulting estimates assess the proper impacts of regulations without being biased by other confounding effects.

In Blanchard and Giavazzi's (2003) theoretical approach, the creation of rents results from product market regulations, while workers’ rent-sharing is influenced by labor market regulations. The analysis of Spector (2004), which is also theoretical, leads to the same conclusions: a decrease in the barriers to entry reduces the rent to share between capital and labor and thus impacts real wages negatively. These models have received empirical corroboration on a cross-country-industry panel by Askenazy, Cette and Maarek (2018), who use value-added prices and the value-added labor shares as indicators of rent and rent-sharing. The empirical investigation of Azmat, Manning and Van Reenen (2012) has also shown, on a cross-country-industry panel, that different components of the OECD NMR indicator have contrasting impacts on labor shares. They find a positive influence of the ‘state control’ component (a combination of sub-indicators on government ownership, control and interference in the running of the industry) and a negative influence of barriers to entry on labor shares. These findings may reflect the fact that barriers to entry tend mainly to augment total rents, while `state control’ increases workers’ bargaining power and their share of rents.

Our present paper follows on from Cette, Lopez and Mairesse (2016a), not only because we exploit the same cross-country-industry panel covering 14 OECD countries and 19 industries over the 1985-2005 period, but also because we follow a similar, but direct and a priori preferable approach. In our previous paper, we first used industry production prices and wages as very crude indicators of mark-ups and workers’ rent-sharing, and we then instrumented these indicators with the NMR and EPL indicators to evaluate the impacts of product and labor market regulation on productivity. This is what we do here, but on the basis of the new measures of mark-up rates and workers’ rent shares that we propose.

We first explain the new measures of mark-up and workers’ rent-sharing at the country-industry level, which are largely inspired by the firm panel data analyses of Dobbelaere and Mairesse (2013, 2015, 2017). We then assess the relationship of these new measures with the OECD NMR and EPL regulation indicators. Finally, using the OECD indicators as instrumental variables, we assess the impacts on total factor productivity (TFP) of mark-up and workers’ share of rents. While standard measures of firms’ market power, such as the Lerner index, assume perfect labor markets, our new measures relax this assumption by taking into account that workers may appropriate part of the rent created in a given industry. To give a preview of our results, we find that our new measures lead to a more in-depth understanding of the impacts of regulations that overall corroborates Blanchard and Giavazzi's (2003) theoretical conclusions, but with interesting differences.

Concerning the relationship of the OECD indicators with our measures, we find that: (i) NMR is positively linked with the mark-up and workers’ share of rents, a result which is consistent with Jean and Nicoletti's (2015) estimates of increasing  inter-industry wage differentials; (ii) EPL has a negative impact on hours worked per output unit, which offsets an increase in workers’ rent per hour and leads to a non-significant impact of EPL on workers’ share of rents; iii) EPL has a positive impact on the wage per hour of low- and medium-skilled workers positively, but a negative one on their hours worked per output unit, particularly for low-skilled workers, thus also on their share of rent, whereas highly-skilled workers tend to be much less affected. Regarding the impacts on TFP, we find that lack of competition on the product market and of flexibility on the labor market have a detrimental impact on TFP, which is consistent with Cette, Lopez and Mairesse (2016a) and the previous literature.

Using our estimation results to calibrate an illustrative out-of-sample policy simulation, we show that a decrease in mark-up rates and workers’ bargaining power that would result from a switch of countries' NMR to the lowest NMR levels might lead to an average increase in TFP of 3.7% in the long run. But the adoption of these levels of lowest regulation would require very large-scale product market structural reforms in some countries, such as France and Italy. The implementation of such reforms cannot be considered politically or socially realistic in the short to medium term.

The original findings of this paper give more content to the channels of the impact of market (both product and labor) regulation on the rent-building and rent-sharing processes. They also confirm that the impact of regulations on productivity occurs via lack of competition and labor market flexibility. A limitation of our investigation concerns the distinction between short- and long-run effects. Our estimation results using static specifications may be interpreted as long-run estimates, but short-run changes are important in understanding the mechanisms for and political feasibility of reforms. The use of a dynamic approach, such as an Error Correction Model, would be interesting but hard to implement because of the lack of time variability in our regulation indicators, particularly the EPL indicator. We are probably at the limit of the use of a cross-country-industry panel dataset, and some corroboration may be needed, perhaps from firm-level data.