India has a multitude of environmental regulations but a history of poor enforcement. Between 1996 and 2004, India's Supreme Court required 17 cities to enact Action Plans to reduce air pollution through a variety of command-and-control (CAC) environmental regulations. We compare the impacts of these regulations with the impact of changes in coal prices on establishment-level pollution abatement, coal consumption, and productivity growth. We find that higher coal prices reduced coal use within establishments, with price elasticities similar to those found in the US. In addition, higher coal prices are associated with lower pollution emissions at the district level. CAC regulations did not affect within-establishment pollution control investment or coal use, but did impact the extensive margin, increasing the share of large establishments investing in pollution control and reducing the entry of new establishments. For reducing SO2 emissions, our results suggest that higher coal prices were more effective in improving environmental outcomes than command and control measures.

In a voluntary emissions-reductions system, regulators must evaluate and sign off on firms' claims of what they would do absent credits. This paper uses the behavior of non-participants to ex-post evaluate these claims. We focus on carbon offset projects in industrial energy efficiency, co-generation, input substitution, and fuel switching that are supplied by firms in India to the international emissions trading market through the Clean Development Mechanism (CDM). We identify the firms involved in over 600 CDM projects in a comprehensive dataset of Indian manufacturing firms. We first look for signs of strategic selection into program participation. After controlling for firm size and industry, there is no evidence that applicants are more likely to have decreasing emissions trends pre-application. We then evaluate behavior ex-post. We find that participants indeed reduce emissions intensity relative to similar non-participant firms, but in a way that is moderated by a greater expansion of output. Looking across project types, the largest emission reductions come from projects that improve energy efficiency and export excess energy to the grid. Fuel switching and input blending projects are more questionable recipients of credits because non-participants engage in these activities at similar rates.

Due to increasing coal use and related emissions, in 2006 the Chinese government mandated that the top 1000 energy-using firms adopt strict targets for reducing industrial energy consumption. This paper exploits the structure of the Top 1000 program to investigate the costs and benefits of an energy efficiency program in a rapidly industrialising economy. We match the Chinese Annual Census of Industrial Enterprises of over 200,000 firms per year with Top 1000 participation data and city-level air quality and economic outcomes. We find that although participating firms' productivity grows slower than that of non-participating firms, the difference in growth rates is very small (less than 1%). Although we cannot rule out that overstated compliance lowered program cost, we find significant improvements in city-level air quality associated with the program, suggesting that firms did indeed respond.

Using 19 years of annual data from over 30,000 firms, I show that India's 1991 trade liberalization and market reforms lead to within-firm improvements in energy efficiency and reallocation of market share to more energy-efficient firms. This shift mitigated post-liberalization growth in energy use and associated greenhouse gas emissions. Two mechanisms led to increases in emissions intensity: increasing generator use post-liberalization and decreasing tariffs on intermediate material inputs, which helped firms that use inputs inefficiently to retain market share. But reductions in tariffs on intermediate capital inputs led to a 23% improvement in within-firm fuel efficiency.  And liberalization of FDI and licensing requirements led to reallocations of market share to more fuel-efficient firms, further reducing emissions.

Since 2000, the Australian government has offered an up-front subsidy based on system size to encourage adoption of solar photovoltaic technologies by households. State and territory governments also offer production-based subsidies in the form of solar feed-in tariffs. Over 1 in 10 households have since installed solar panels; in Queensland, over 1 in 5 have solar panels. We use data from the Australian Clean Energy Regulator to estimate the impact of the subsidies on residential uptake, using postcode pairs along state and solar insolation boundaries to control for potentially endogenous policy-setting of subsidy levels. We find that a $100 increase in up-front rebates led to a 17% increase in installations. This response differs by income bracket, with middle-income postcodes being the most sensitive. There is also evidence of anticipatory behaviour by households, installation rates surge before downwards rate adjustments. Breaking up the period studied into three phases, we see that the up-front subsidy increased in effectiveness over time. Feed-in tariffs, on the other hand, were more effective in the early stages, and with no evidence of an effect in the later phases, when the bulk of installations occurred. We predict that 23% of installations would not have occurred absent up-front subsidies. The combined production of these marginal installations displaces approximately 0.02% of annual emissions from the power generation sector in Australia.

We document how imperfect information generates heterogeneous effects in information treatments with personalized high-frequency feedback and peer comparisons. In our field experiment in retail electricity, we find high and low energy users symmetrically underestimate and overestimate their relative energy use pre-treatment. Responses to personalized feedback, however, are asymmetric. Households that overestimate their relative use and low users both respond by consuming more. These boomerang effects provide further evidence that peer-comparison information programs, even those coupled with injunctive norms, are not guaranteed to lead to increases in pro-social behavior.