رشد بهره وری سازگار با محیط زیست حساس: یک تحلیل کلی با استفاده از شاخص ملکوئیست -لوئنبرگر

We examine conventional and environmentally sensitive total factor productivity (TFP) in 41 developed and developing countries over the period of 1971 to 1992. Due to the non-availability of reliable input and CO2 emissions price data, the study uses directional distance function to derive Malmquist–Luenberger (ML) productivity index. The index allows us to decompose the TFP into measures of technical and efficiency changes. DEA is used to compute the directional distance functions. We find that TFP index value is not different when we account for the CO2 emissions relative to the situation when they are freely disposable. But for the components of TFP change: technical and efficiency changes, the null hypothesis of whether the indexes are the same under two different scenarios cannot be accepted. Issues of catch-up and convergence, or in some cases possible divergence, in productivity are examined within a global framework. The paper also studies the impact of openness on conventional and environmentally sensitive measures of productivity.

Concerns about the impact of climate policy on ‘productivity’ or ‘economic growth’ have made countries hesitant about reducing CO2 emissions. Climate policy has different dimensions: economic, technological and ecological. The economic dimension offers solutions in terms of price signals and the technological dimension sees solutions in terms of appropriate technological development and adoption. The ecological dimension adopts a more holistic view of man–nature relationship and calls for ‘green accounting’ or ‘sustainable development’. This paper tries to present an extension of economic approach that includes aspects of technological development and adoption as well as green accounting. Productivity has acted as a significant engine of growth, allowing living standards in the world to advance rapidly throughout the twentieth century. However, its traditional measures do not account for production of harmful by-products such as CO2, which may lead to environmental damage. It is conventionally measured using index numbers, which require data on prices of all outputs and inputs and the price information for bad outputs does not exist. The distance function approach can help overcome such problems as it requires data only on quantities of inputs, outputs and pollutants. Unlike this study, others estimating productivity using the distance function approach have focused on desirable outputs only (e.g., Färe et al., 1994b and Lall et al., 2002). There are also cases that have used micro-economic in contrast to macro-economic data used by the present study while estimating the total factor productivity (TFP) in the presence of bad outputs (e.g., Yaisawarng and Klein, 1994, Ball et al., 1994, Chung et al., 1997 and Hailu and Veeman, 2000). A method of measuring TFP using distance function that is growing in popularity is the use of Malmquist indexes. However, incorporation of bad outputs into the Malmquist indexes can be problematic. As the Malmquist indexes are based on Shepherd distance functions, which are radial in nature, firms cannot be credited with the reduction of bad outputs. This does not allow for changes in technology reducing the amount of pollution generated while increasing production of good outputs. It does not capture any “de-coupling” of the production of good outputs with bad outputs. If there has been a de-coupling of pollution and production, then there may be computational problems using the Shepherd distance function (Chapple and Harris, 2003). There are several studies on the measurement of productivity changes in industries, which produce good and bad outputs simultaneously during the production process. Some of these studies have treated the bad outputs as inputs,1 while the others have treated these as synthetic output such as pollution abatement (e.g. Gollop and Roberts, 1983). Murty and Russell (2002) have pointed out that the treatment of bad outputs as inputs is not consistent with the materials balance approach. The approach adopted by Gollop and Robert to treat the reduction in bad output as good output creates a different non-linear transformation of the original variable in the absence of base constrained emission rates (Atkinson and Dorfman, 2002). To overcome this problem, Pittman (1983) proposed that good and bad outputs should be treated non-symmetrically. He suggested the maximal radial expansion of good outputs and contraction of bad outputs. Chung et al. (1997) have used the directional distance function to calculate production relationships involving good and bad outputs that treats good and bad outputs asymmetrically. This study follows Chung et al. (1997) and uses directional distance function to measure Malmquist–Luenberger (ML) productivity index and its components. The components of productivity index—technical and efficiency changes are analogous to the notions of technological innovation and adoption, respectively. The ML index credits producers for simultaneously increasing good outputs and reducing the production of bad outputs such as CO2. It also offers an alternative way of assigning weightage on the relative importance of the bad outputs which can be interpreted as if consumers have preferences for reducing bad outputs regardless of the actual damage resulting from these products (Färe et al., 2001). Although the ML index does not directly relate to changes in welfare level, it does provide a complete picture of productivity growth under environmental regulations of emissions that are of concern to society. The measures of productivity are often obtained under alternative assumptions about the disposability of CO2. That is, it could be either strongly or weakly disposable. While, strong disposability implies that a country can reduce CO2 emissions without incurring any abatement costs, weak disposability assumes diversion of resources from the production of good outputs. Thus the ML index encompasses green accounting while accounting for undesirable outputs.2 This paper uses non-parametric linear programming method to estimate directional distance function. Thus for each year the same ‘meta’ best practice frontier is constructed based on the data for 41 countries for the period 1971–1992. Each country is then compared to this best practice frontier to provide the performance scores. Productivity analysis helps to understand the level of economic prosperity, standard of living and the degree of competitiveness of a country, although it is not the only determinant of economic growth and welfare. Therefore, it is important to find which factors determine productivity growth in the countries in the presence of reduction in carbon emissions. Though there are various theories that explain productivity growth in countries, two are of particular interest.3 One, the convergence theory states that in low-income countries productivity tends to converge towards those of high income countries, (Baumol, 1986 and Baumol et al., 1989). The rationale behind the convergence hypothesis is the concept of diminishing returns to capital. In the developed countries the capital–labor ratio is found to be high in comparison to developing countries and therefore the marginal productivity of capital in them should be low. Two, the endogenous growth theory advocates that the difference in productivity between developed and developing countries remains constant or even diverges over time (Arrow, 1962). The foundation of endogenous growth theories lies in the concept of increasing returns to scale, that are generated from externalities associated with the acquisition of technical knowledge. However, there are institutions and policies that determine the development process of a country (Olson, 1996). This paper tries to extend this literature by empirically examining the causes of productivity changes while accounting for carbon emissions. The remainder of the paper is structured as follows: In Section 2, we discuss the theoretical approach of the paper. Section 3 discusses the data used in the study and its results. The data set is richer than the past examinations of efficiency and productivity analyses in that it includes energy as input. The addition allows for a more thorough assessment of the production processes that generate carbon emissions from the use of energy. The paper closes in Section 4 with some concluding remarks.

Bad outputs are ignored by the traditional measures of productivity and hence have limited use with regards to policy evaluation. However there are environmental regulations and resources are diverted from traditional productive activities to pollution abatement. As a result, these traditional measures of productivity found that environmental regulations have an adverse effect on productivity. These findings ignore the key feature of environmental regulations that diverting the resources in abatement activities leads to reduction in environmental bad outputs. The traditional measures of productivity ignore the reduction in bad outputs due to abatement activities since typically no prices are available for the undesirable outputs such as CO2 emissions, except for the situations when tradable permits are used to restrict the emissions. This study presents an extended view of TFP growth measured through ML index using directional distance function. The index throws insight into the sources of productivity growth to estimate an adjusted rate of TFP growth while accounting for CO2 emissions minimization activities. Through an asymmetrical treatment of good and bad outputs, the TFP index is decomposed into efficiency and technical changes. This index provides a common dialog of different perspective of climate change debate by expanding the basic economic concept of productivity to identify the combined role of technological innovation and adoption, and green accounting. The ML index is calculated using the non-parametric directional distance function for a group of 41 countries consisting of 21 Annex-I countries and 20 Non-Annex-I countries during the 1971 to 1992 period. On average for either of the group of countries, the value of standard Malmquist is not different from ML index. But for the components of TFP, technical and technical efficiency changes the null hypothesis of whether the different indexes are same when emissions are ignored and when they are accounted for cannot be accepted for either of the groups of countries. Out of 41 countries only six, Iceland, Hong Kong, Japan, Luxembourg and Netherlands, Switzerland were innovators. None of the developing country was shifting the frontier under either scenario. Subsequent regression analyses find that the environmentally sensitive measure of productivity is higher in those countries, which are having the higher GDP per capita. The value of ML index is negatively associated with technical efficiency and capital labor ratio, implying presence of convergence hypothesis. Moreover, it also finds that the energy intensity of production is negatively related to the environmentally sensitive measure of productivity. However, the conventional measure of productivity remains unaffected by the composition of output growth. The openness of a country increases its TFP whether it is measured by the standard Malmquist index or ML index. Beyond measuring of environmentally sensitive productivity growth, the present analysis demonstrates the richness of the technique that allows for investigation of important research questions on the underlying processes that influence productivity growth. Notwithstanding the striking feature of the techniques used here, data limitations involved in estimation remain an important factor. It is, therefore, necessary to be cautious while applying these results to policy formulation.