THE OTHER SIDE OF GREEN PORK: HOW HETEROGENEITY IN

ENVIRONMENTAL PREFERENCES AFFECTS THE DISTRIBUTIONAL AND

EFFICIENCY IMPLICATIONS OF CLIMATE POLICY

April 17th, 2013

Joel R. Landry

CONGRESS

MotivationGlobal climate change is largely recognized by most scientists and economists as requiring immediate and consequential action by policy-makers. Which level of government should bear primary responsibility for policy-making in this area is much less clear.

While the federal government came close to passing a comprehensive climate bill in 2009, this effort ultimately failed like all other attempts that came before it. State governments have begun to act unilaterally through the passage of Renewable Portfolio Standards as well as other regional trading initiatives.

MotivationWhy the federal government has so far failed to act as well as why some state governments have opted to act unilaterally, speaks to the considerable heterogeneity in climate beliefs that is especially prevalent in the US.

According to annual Gallup polls between 2008 and 2010 that asked respondents how seriously they viewed climate change as a problem on a scale from 1 (``very serious problem'') to 4 (``not a problem''), Democrat respondents consistently reported views that are on average a whole point lower—a single standard deviation—than Republican respondents.

Such widespread heterogeneity in climate beliefs or environmental preferences is a critical feature of US policy-making with respect to climate change and is hardly news. More important is how this heterogeneity interacts with policy-making at the federal and state levels in critically different ways and how this heterogeneity drives the distortionary effects of both types of policymaking.

Key QuestionsHow does heterogeneity in environmental preferences affect the climate policies chosen by state or decentralized policy-makers as well as federal or centralized policy-makers?Is state policy-making with respect to climate change more or less distortionary than federal policy-making, relative to a policy that maximizes aggregate Marshaling surplus? That is, which level of government should act to address climate change?How does heterogeneity in environmental preferences effect these relative distortions?How does heterogeneity in environmental preferences effect the distributional impacts of both federal and state policymaking?Can alternative policymaking regimes be established that minimize the distortionary impacts of state and/or federal policymaking.

Overview of Modeling Framework

Integrates a general equilibrium model of the US economy in which emissions are generated to models of state and federal decision-making.State decision-making reflects horizontal fiscal competition whereby states set local emissions caps (conditional on the caps selected by other states) where capital is mobile between states.

The result is a Nash equilibrium of local caps that when summed up are less stringent than the cap which maximizes aggregate surplus.

Federal decision-making reflects legislative bargaining whereby a winning electoral coalition of legislators simultaneously choose a national cap and an allocation of free emissions permits to legislators’ districts.

Resulting cap can be more lax or more stringent than the cap which maximizes aggregate surplus.

Overview of Empirical Approach

Given data on congressional district emissions, production quantities, and endowments, as well as district level estimates of permit allocations under WM and the WM cap level, use our model of federal policy-making to identify the vector of heterogeneous environmental preference parameters that characterize each district’s implied marginal emissions damages from climate change.

In effect, find the vector of preference parameters such that the observed coalition of legislators who voted for WM, given the observed cap and permit allocation is predicted by our federal policy-making model relative to a BAU baseline of no climate policy.

Given this vector, aggregate to the state level and recover the counterfactual state policy-making equilibrium.Solve for aggregate surplus maximizing emissions level, and compare.

Contribution to the LiteratureThis paper provides an empirically identified, welfare consistent comparison of the distortions that emerge from both federal and state policy-making.

This differs from calibrated simulation models (Parry 2003) and reduced form methods (Winner 2005, Devereux et al 2008) that attempt to estimate the distortionary costs of decentralized policymaking alone.Exploit an observed federal policy and electoral coalition to parameterize full model.Will perform two important validation checks:

Show that when we apply legislative bargaining model to the Senate (which has preferences that are aggregates of congressional districts contained within a state), that our model predicts WM would not pass.Show that the pattern of state policymakers that adopt aggressive climate policies largely reflects those states which have adopted RPS, or regional climate initiatives.

Contribution to the Literature

Extends legislative bargaining model of Volden and Wiseman (2007,2008) to the case of complementary policies, since, in the case of climate policy, the cap level co-determines the amount of permits available to freely allocate to districts.Show that the central result from Ogawa and Wildasin (2009), namely that the decentralized equilibrium can achieve the first best, is sensitive to the assumption that policymakers do not account for the impact of their policies on the equilibrium rate of return to capital.

Show that greater heterogeneity in environmental preferences increases the likelihood that the decentralized equilibrium that emerges will in fact be distortionary.

Waxman negotiates with Greene (D—TX), receives 2% of permits for crude oil

refiners.WM climate bill passes House Energy and Commerce Committee on May 21st, 2009

Timeline of Major Events

Waxman takes over as chairman of House Energy and Commerce Committee.

Waxman negotiates with Boucher (D—VI), a coal ally, for an emissions reduction of 17% by 2020. Boucher obtains 35% of

cap as permits for electricity LDCs, 9% for natural gas LDCs, and 5% for merchant

coal.

Waxman negotiates with Doyle (D—PE) and other heavy manufacturing states, receive 15% of permits for vulnerable

industries (steel, aluminum, etc.).

Peterson (D—MN) and 45 midwestern reps threaten to send the bill to the House

Agric. Comm. for full mark-up unless agric. concessions were made, gets USDA

oversight of agric. offsets (not EPA), other concessions.

Bill goes up for House vote on June 26th, 2009.

Waxman cuts deals on House floor: Rush (D—IL) obtains promise of $1 billion in

clean energy jobs.

Waxman cuts deals on House floor: Grayson (D—FL) obtains promise of $50 million for center to study connections

between hurricanes and climate change.

Waxman cuts deals on House floor: Bean (D—IL) obtains language limiting sale of

over the counter derivatives.

Waxman cuts deals on House floor: Perlmutter (D—CO) secures loan

guarantees and incentives for energy efficiency and distributed generation.

WM passes the US House just after 7 PM on June 26th, 2009. 25% of votes came

from NY and CA.

Bill moves to Senate. McCain (R—AZ) and 10 Midwestern or coal state Democratic

senators comes our against WM.

Boxer (D—CA) and Kerry (D—MA) announce bill to move through Boxer’s

Environment and Public Works Committee. Mark-up is boycotted by Republicans, but clears committee anyway. By November

the bill is declared dead.

Kerry (D—MA) tries again with Leiberman (I—CT) and Graham (R—SC). Reid (D—NV)

moves vote to Spring 2010, after healthcare vote.

December 2009. Copenhagen (doesn’t) happens. Kyoto not expanded and

voluntary Copenhagen Protocol signed.

January 27th 2010. Obama State of the Union offers “grand bargain” on climate to

Republicans.February 26th, 2010. Graham declares

“cap and trade dead.”

Was WM consistent with Legislative Bargaining?

Was WM consistent with Legislative Bargaining?

Was WM consistent with Legislative Bargaining?

Economic Modelj=1,…J jurisdictions (states or congressional districts)Each jurisdiction has a representative consumer with utility, given by:

Has a supply function of capital given by:

Has a labor endowment with value denoted by:

Economic ModelEach jurisdiction has a private good producer with constant returns to scale technology given by:

Private good production produces emissions as an externality:

Maximizes profits according to:

Market Clearing

(federal only)

State PolicymakingEach jurisdiction is a state government.Each state government chooses a local emissions cap to maximize their aggregate surplus conditional on the local emissions caps selected by other state governments.This is a model of “horizontal fiscal competition.”Unlike Ogawa and Wildasin (2009) we do not:

Assume that state governments choose climate policy irrespective of their impact on the rate of return to capital.Assume that total capital supplied is fixed in the economy.

State PolicymakingWhen the rate of return to capital is exogeneous to state policy-makers, as it is assumed to be in Ogawa and Wildasin 2009, the shadow price of permits (alternately, local emissions tax) that results is given by:

Consequently, we can establish a result equivalent to their Proposition 1:

State PolicymakingConversely, when the rate of return to capital is endogenous to state policy-makers then the equilibrium price of permits (e.g. tax rate) that results is instead:

State PolicymakingThus the decentralized equilibrium that emerges in this case will be characterized by our alternate Proposition 2.

It logically follows that it would be useful to know whether state governments, upon starting at the aggregate surplus maximizing equilibrium (the decentralized equilibrium when the rate of return is exogeneous), would wish to deviate to our decentralized equilibrium.

State Policymaking (J=2)

State Policymaking (J=2)

More heterogeneity in environmental preferences, more likely to deviate!

Will States Deviate?

Federal PolicymakingEach jurisdiction (congressional district) sends a legislator.Legislators wish to maximize the aggregate surplus to their district (utility of representative consumer less cost of own capital supplied)A legislator is randomly selected to propose a policy consisting of a national emissions cap and an allocation of emissions permits to each jurisdiction.A policy passes if it attracts at least a simple majority of legislators voting for it, including the proposer.A legislator will vote for a policy if its aggregate surplus under the policy exceeds or is equal to its aggregate surplus under no climate policy.

Federal PolicymakingA proposing legislator will seek to maximize permits to their own district, which means they will allocate as few permits as is possible to other districts to get their policy passed.A coalition will be a simply majority of size JM if proposing legislator must provide positive permits to any coalition members.A coalition can be super-majoritarian (J>x>JM) or even unanimous (=J) if x or J legislators, respectively, require zero permits to vote yes.Since permits always have positive value to proposer, the proposer has incentive to allocate permits exactly equal to the total cap level.

Federal PolicymakingAdditional remarks:

Ceteris paribus, coalitions of climate believers are more likely to vote for a climate bill than climate skeptics because the positive utility they receive from reducing emissions through the cap is greater.Of course may still require permits in order to secure their vote, but the number of permits I would need to give them will be fewer than a climate skeptic.Thus proposing legislators will have to offer fewer permits to coalitions of climate believers, which will increase the number of permits the proposer will be able to acquire.

Federal PolicymakingUnder certain parameter restrictions where we permit only heterogeneity in environmental preference parameters, a randomly selected proposer z will choose a federal cap (attracting the votes of coalition Jz) such that the equilibrium ex post permit price (alternately, emissions tax) is given by:

Federal PolicymakingTo understand this further, note that:

The permit price (emissions tax) that maximizes aggregate surplus is given by:

And, the permit price that maximizes the value of permits is given by:

Federal PolicymakingThus the federal policy reflects two objectives, corresponding to each component:

The Direct ComponentCap internalizes emissions damages of those within the coalition.

This reflects majoritarian bias.Hence, this term from PAS

Is not the same as this term from Pz:

Federal PolicymakingThus the federal policy reflects two objectives, corresponding to each component:

The Redistribution ComponentCap yields permits which have value to legislators which they wish to sequester to their districts.All else equal, more permits are better than less.

Not surprisingly, thus, this component of the tax reflects PRM, the maximum value of permits that are possible.

Note the Direct Component is weighted by and the Redistribution component by .

Federal PolicymakingThese weights reflect the schizophrenia of federal policymaking with respect to climate change.

The result is that while the Direct Component may imply a cap that is set too laxly, the Redistribution Component will imply a cap that is set too stringently.

Recall that this weight is given by:

Thus as the number of legislators included in the electoral coalition, JM, approaches J, than Pz approaches PAS. If we have two masses of legislators where one mass clearly prefers no climate policy (“skeptics”) and the other prefers a climate policy (“believers”) then the coalition is not likely to be super-majoritarian or unanimous and thus federal policymaking is likely to be distortionary.

Federal Policymaking

Federal Policymaking

Federal policymaking can lead to greater deviations from aggregate surplus maximizing policy than state

policymaking.

In the PipelineImportance of heterogeneity in federal model is not easy to see when J=2 since each proposing legislator just builds a coalition of just themselves heterogeneity is most interesting when it changes the coalitions that form.

Still trying to figure out best way to report the output from the federal model which is voluminous.

Finish debugging efficient algorithm that solves federal modelFor some parameter combinations with J>2, I am not getting solutions largely because the algorithm terminates when it searches in imaginary space. So need to code random restarts across multiple starting vectors.Combinatorial solution method appears to have no issues, but will not be able to solve for anything greater than J=15.Have compared results of algorithms for J=2 and J=3 and are identical to 0.06%, not sure if I should be troubled by this or not.

In the PipelineFinish code that identifies preference parameter vector using federal model and observed WM policy and coalition.

Do this for a “simplified” version of the model first with 32 quasi districts.

Quasi districts formed by first clustering congressional districts by ADAGE region and vote (gets the vote count right by region), then by how alike they are given observed profits, capital used, and value of capital supplied.Still too large for a first pass? Might want to re-cluster at more aggregate level to reduce dimensionality of search/permit easier debugging.

Do the full model with 435 congressional districts.

For state policymaking model, will aggregate congressional districts within a state to parameterize the model.

With 50 states, we have 50 FOCs which characterize the 50 reactions functions. If anyone has ideas how one can parallelize (grid over random restarts?) or solve this more rapidly (nested search?) would love to chat.

Thanks!Comments, better way to frame arguments, cast results, questions, concerns will all be accepted with open arms.If you’d like to grab coffee or lunch off-line to discuss this project, it would be a privilege to get your thoughts, so hit me up.