Carbon Prices, Policy Uncertainty and Green Investments

How to cite

Dolls, Mathias; Link, Sebastian; Liski, Matti; Wolf, Gerome. 2026. "Carbon Prices, Policy Uncertainty and Green Investments." AEA RCT Registry. AEARCTR-0017864. https://doi.org/10.1257/rct.17864-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Mathias Dolls* , Sebastian Link , Matti Liski , Gerome Wolf

* Primary investigator

Registry ID
AEARCTR-0017864
DOI
https://doi.org/10.1257/rct.17864-1.0
First registered
Last updated
Status
On going
Trial dates
2026-02-05 – 2026-02-20
Countries
Private
JEL codes
D22, D25, D81, H23, H32, C90
Registry record
http://www.socialscienceregistry.org/trials/17864

Abstract

Achieving the Paris Agreement's climate goals requires a surge in private investment -- on the order of trillions of dollars annually -- yet firms often underinvest in the low-carbon transition. Carbon pricing is the central policy lever, implemented either through carbon taxes or cap-and-trade systems. In the German context, relevant instruments include Germany's Brennstoffemissionshandelsgesetz (BEHG, since 2021) and the European Emissions Trading System (ETS, since 2005), with a second ETS (ETS-2) covering buildings, road transport, and small industrial plants planned to start in 2028. Despite these frameworks, firms may delay or scale back decarbonization because `green investments' are partly irreversible, front-loaded, and characterized by long payback horizons. Uncertainty about future carbon prices---stemming from potential policy reversals, market volatility, and uncertain future carbon prices---may be especially pivotal.

This project studies how both expected levels of future carbon prices and uncertainty around those prices influence firms' decarbonization investments. Methodologically, we will embed a conjoint experiment among managers in German firms in the ifo Business Survey to identify attribute-level trade-offs. The findings will shed light on how the nature of uncertainty shapes firms' timing decisions - when managers prefer to wait and when they prefer to move early - which is directly relevant for the design of policies that aim to provide a more predictable policy environment. In the remainder of this pre-analysis plan, we describe the experimental strategy, our hypotheses and the survey design.

Experimental design

We implement a conjoint experiment which will be included as an extra module in a large survey of German firms (described below). In what follows, we refer to the module as the survey. Respondents take the survey online and on the first screen, they are informed about its expected length (roughly 10 minutes) and aim (to better understand how companies take potential future carbon prices into account in their investment decisions). Respondents are told that they will be presented with various scenarios to this end.

At the beginning of the survey, we elicit a) the investment amount firms usually spend for a typical investment project and b) the payback horizon of such a typical investment project. We ask respondents to think of a major investment project that their company has most recently carried out, for example a tangible investment (machinery, buildings, vehicle fleet) or an intangible investment (research and development, software, licenses). These `typical' investment projects, in particular their payback horizons, serve as reference points later on in the survey.

In the main part of the experiment, respondents are asked to imagine that their firm is considering an investment project which would reduce its carbon emissions and increase its energy efficiency. We will refer to this project as a ``green investment'' or ``decarbonization investment''. They are informed that its investment amount and useful life are comparable to their firm's typical investment project (elicited at the beginning of the survey) and that the economic viability of the green investment largely depends on the future trajectory of the carbon price.

Next, respondents are told that they will be asked to make investment decisions under five scenarios of future carbon prices, each of which specifies the expected average carbon price over the useful life of the green investment and an uncertainty range around this average. Respondents are informed that the average carbon price over the green investment's useful life can take any value within the uncertainty range with equal probability. In addition, we tell respondents that, in each scenario, they will be informed about the expected payback period of the green investment and the associated uncertainty range implied by the carbon price scenario.

The five scenarios for the expected average carbon price over the useful life of the green investment and its uncertainty range - the two attributes of our conjoint experiment - are randomly selected without replacement, so that respondents do not see any scenario twice. The order in which the two attributes are shown is randomized across respondents, but fixed within respondents.

The expected future carbon price can take on the following attribute levels: [50 EUR, 75 EUR, 100 EUR, 125 EUR, 150 EUR, 175 EUR, 200 EUR]. The uncertainty range can take on the following attribute levels (expressed as percentage deviation from the expected carbon price): [+/- 0\%, +/- 25\%, +/- 50\%, +/- 75\%, +/- 100\%]. For the ease of interpretation we additionally show the uncertainty range for the expected carbon price in monetary units.

The expected payback period of the green investment and its uncertainty range resulting from the respective carbon price scenario will be shown below the two attributes.

For each scenario, respondents are asked how likely it is that their firm would carry out the green investment under the stated conditions on a categorical scale ranging from ``very unlikely'' to ``very likely''. This question constitutes the outcome question in our experiment. As described below, our research design allows us to estimate average marginal component effects for the two attributes in our conjoint experiment. In other words, we can measure how changing one attribute level to another causally affects the probability that the green investment is carried out in a given scenario, on average over all levels of the other attribute in our design.

In the post-experimental part of the survey, we elicit several dimensions of firms' climate- and investment-related environment and decision making. These questions, described below, are used for heterogeneity analyses.

Design details

Randomization method
randomization in online survey
Randomization unit
firm
Clusters
Our conjoint experiment will be embedded as a module in the February 2026 wave of the ifo Business Survey (IBS), a well-established monthly panel of German firms. The IBS is representative (across industries and size) of the universe of private sector firms in Germany (Hiersemenzel et al. 2022) and is typically completed by the CEO or the owner of the firm (Hennrich et al. 2023). The module, shown after respondents have completed the core questions of the IBS, will be presented to manufacturing firms with more than 25 employees for whom carbon prices are most relevant in their cost structure and which have the capacity to undertake substantial investments. These are typically firms operating in energy- and emissions-intensive sectors facing strong incentives to react to energy cost pressures as well as climate policy risk in light of irreversible decarbonization investment decisions. Participation in our experiment is voluntary. Given that our module is only open for manufacturing firms with more than 25 employees and given that participation is on a voluntary basis, we expect that up to 1.000 respondents will participate in the experiment.
Observations
Each respondent (firm) makes five investment decisions, i.e., 5 * number of clusters
Arms
No treatment arms
Primary outcomes
Our main question of interest elicits whether the firm would carry out the green investment under the presented conditions. The question reads: Please imagine that a typical investment project of your company pays off after approximately x years. How likely is it that your company would choose the investment project to reduce carbon emissions and increase energy efficiency under the stated conditions? (0) "very unlikely" to (10) "very likely".

Cited by

No citation data available for this registration yet.

Keywords