9th Oxford Workshop on Global Priorities Research

Talk Details

Monday 21 March 2021

9:30 - Orri Stefansson, “Longtermism and social risk-taking” 

A social planner who evaluates risky public policies in light of the other risks with which her society will be faced, should judge favourably some such policies even though she would deem them too risky when considered in isolation. I suggest that a longtermist would—or at least should should—evaluate risky policies in the former rather than the latter way; hence, longtermism supports social risk-taking. I consider two formal versions of this argument, discuss the conditions needed for the argument to be valid, and briefly compare these conditions to some risky policy options with which actual public decision-makers are faced.

10:30 - Florian Brandl, “A Framework for Donor Coordination: Some Results and Future Directions”

We consider a mechanism design framework where multiple agents can fund public projects via voluntary monetary contributions. An important application of this setting is donor coordination, which allows altruistic donors to coordinate their contributions to different causes. We aim to identify a mutually beneficial distribution of the individual contributions. In the preference aggregation problem that we study, agents with linear utility functions over projects report the amount of their contribution, and the mechanism determines a socially optimal distribution of the money. We identify economically relevant properties -- such as efficiency, incentive compatibility, and fairness -- and provide mechanisms satisfying them whenever they are compatible. The talk will focus on modeling assumptions limiting the applicability of the framework, some resulting open problems, and avenues for future work to address those.

11:20 - Rossa O’Keeffe-O’Donovan, “The Economics of the Long Term”

Should we write a paper introducing "longtermism" and/or key principles of global priorities research to the mainstream, academic economics community? If so, what should the paper focus on, what should be its key points? Should it define longtermism and/or the longtermist hypothesis, explain the terms, argue that they hold empirically? Should we aim for a short, high level paper, or a longer, more detailed analysis? 

I will present an overview of a working group that we ran at GPI to explore these questions. I'll then crowdsource ideas for what a possible paper should look like and if/when we should write it. The session won't follow the traditional academic talk formula - I'll aim to make it interactive as a way of generating ideas and stimulating discussions. Please bring a laptop!

12:20 - Jonathon Hazell, “Business Cycles and Long Run Welfare”

What are the consequences of recessions for long run welfare? Standard macroeconomic models assume that the consequences are negligible. The reason is that the effects of recessions are assumed to be temporary. So, in the long run, their effects vanish. However, empirically, recessions seem to have persistent or permanent effects on output, and therefore may affect long run welfare—that is, there may be hysteresis. This talk outlines some empirical strategies to estimate hysteresis in macroeconomic data, and to understand its consequences for welfare. 

12:40 - Will MacAskill, “Values and the long term”

In this talk, I investigate the Moral contingency thesis (MCT): For a wide range of significant moral beliefs and practices, and assuming no near-term extinction of Earth-originating intelligent life, and no takeover by radically misaligned superintelligence, the expected value of the very long-run future is highly sensitive to the moral beliefs and practices that are predominant over the coming few centuries. 

14:15 - Hayden Wilkinson, “Is the expected value of the future well-defined?”

One common justification for (axiological strong) longtermism goes like this: the correct axiology over outcomes is impartial and aggregative; and the correct extension of this axiology to cases of risk is expected value theory. Combined (and accompanied by some empirical assumptions), these two claims seem to entail longtermism. But this justification does not work if the (impartial, aggregated) value of the humanity's future is undefined. For instance, one's probability distribution over possible values of the future may have tails resembling those of the Pasadena game, or a Cauchy distribution. If so, expected value theory does not justify longtermism; instead, it justifies no practical judgements at all. In this paper, I argue that our evidence implies just such a probability distribution. Indeed, it implies a distribution that cannot be evaluated even by various extensions of expected value theory. I propose a novel further extension of expected value theory, which allows us to salvage the case for longtermism. I also consider how risk-averse decision theories might deal with such a case, and offer a surprising argument in favour of risk aversion in moral decision-making.

14:55 - Marcus Pivato, “Autonomy and Metapreferences”

The standard model of rational choice in economics treats the preferences of the agent as exogenous.  This raises interesting philosophical problems:  if a person cannot *choose* her own preferences, then she is not really "autonomous" ---she is condemned to slavishly maximize a preference order that has been "imposed" upon her from the outside.  This calls into question whether a purely welfarist-consequentialist approach to normative economics (based on preference-maximization) really captures everything that is morally relevant.  A further issue (less relevant to consequentialists, but more relevant to deontologists) is *morally responsibility*:  it an agent's preferences are exogenous, then arguably we cannot hold her morally responsible for her choices (good or bad), because these choices are simply the result of maximizing an (unchosen) preference order.  

These issues will become increasingly important in the future, when it may become technologically possible to deliberately modify people's preferences (e.g. via neurosurgery), and to create artificial persons (either electronic or biological) with "designed" preferences.  Indeed, it may be that *most* people in the far future have artificially engineered preferences. 

But suppose that a person *could* choose her own preferences.  On what basis would she make such a choice?  Presumably, on the basis of "second order" preferences.  But how does she choose these second-order preferences?  This leads to an obvious infinite regress.  Furthermore, what does rational choice mean when the agent must simultaneously optimize with respect to first-order, second-order, and higher-order preferences?   What happens when her higher-order preferences come into conflict with her lower-order preferences?  In this talk, I will introduce some mathematical models of such "metapreferences", and discuss possible solutions to these problems.  This is a preliminary report on work in progress.

15:35 - Charlotte Siegmann, “Speaking for the Billions to Come — Estimating the Effect of Future Representatives in Politics”

Political decisions regularly neglect the interests of the future. A variety of institutions and mechanisms have been proposed to alleviate this problem. I argue institutional design should be reliably tested to predict whether it contributes to the alleviation of presentism.  I discuss the results of two trial experiments with UK and US citizens. In the first, I identify how being or not being a future representative changes one’s political opinions and examine priming vs. identity treatments and the role of peer pressure and moral licensing. In the second, I explore the voting behaviour for institutions representing future generations. I briefly outline directions for further research.

17:05 - Andreas Schmidt, “Moral uncertainty: a new methodology for bioethics?”

Bioethics combines academic research with practical work that guides ethical decision-making in health policy, biomedical research, and clinical practice. To capture these ‘dual goals of bioethics’, bioethicists have proposed their own methodological frameworks. We first outline shortcomings of existing frameworks such as High Theory, Anti-Theory, and Principlism. We then propose the Moral Uncertainty Framework: bioethics as an overall practice should improve our credences in moral theories and their implications to bioethical practice and use the best available framework for making decisions under moral uncertainty to guide bioethical practice. Unlike existing frameworks, this framework better captures both the epistemic and practical goals of bioethics, allows for practical judgements under moral disagreement and uncertainty, assigns a clearer role to moral theory, and has a clearer path towards all-things-considered rather than just pro tanto judgements. As a result, bioethical practice should hopefully lead to both better theorising and ethical decisions. While we focus on bioethics, much of what we say should also apply to applied ethics and the ethics of public policy more generally. 

17:45 - Maya Eden, “Endogenous fertility, social discounting and social risk aversion”

Population models differ in their predictions about how the current number of births affects future population levels. Adopting a longtermist perspective, I discuss the implications of these models for social discounting and social risk aversion. In the Malthusian model, long-run population levels are insensitive to the number of births today. But in a model with constant or increasing returns to labor, each loss of a life today has a permanent effect on the number of future beings. This is very preliminary work.

Tuesday 22 March 2021

9:15 - Kevin Kuruc, “Optimal Population Size: Evidence from a Malthusian Semi-Endogenous Growth Model”

The relationship between human population sizes and economic well-being has been debated for centuries. We revisit this question in a modern, forward-looking setting using a semi-endogenous growth model with renewable resources. People create ideas but also deplete natural resources. Conditional on a stable population, the model produces an analytical steady-state, allowing us to study the per capita income-maximizing population level. Four main findings arise. First, a sufficient statistic for the (negative) elasticity of the Malthusian channel is the income share of the fixed factor. Second, because the aforementioned income share is small, the local, all-things-considered, elasticity of long-run per-capita income with respect to population is likely positive. Third, this all-things-considered elasticity only falls to zero if non-fixed factors and the fixed factor are complementary; this is not empirically supported, implying that the optimal population is unbounded in the model. Finally, and related to the findings of Jones (2020), because the knowledge accumulation from an increased population takes many years to occur, an impatient planner will choose a smaller population size than the steady-state optimum.

10:15 - Andreas Mogensen, “The Hinge of History Hypothesis: Reply to MacAskill”

Some believe that the current era is uniquely important with respect to how well the rest of human history goes. Following Parfit, call this the Hinge of History Hypothesis. Recently, MacAskill has argued that our era is actually very unlikely to be especially influential in the way asserted by the Hinge of History Hypothesis. In this talk, I respond to MacAskill, pointing to important unresolved ambiguities in his proposed definition of what it means for a time to be influential and criticizing the two arguments used to cast doubt on the claim that the current era is a uniquely important moment in human history. 

10:55 - Silvia Milano, “Civic Governance of AI Services: the case of Digital advertising”

12:00 - David Thorstad, “Existential risk pessimism and the time of perils”

Recent authors have argued that it is overwhelmingly important to mitigate existential risks: risks which threaten to curtail the survival or development of humanity. This position is often supported by pessimistically high estimates of current levels of existential risk. In this paper, I extend a model by Toby Ord and Thomas Adamczewski to do two things. First, I argue, across a range of modeling assumptions pessimism tends to hamper rather than strengthen the case for existential risk mitigation. Second, I show that pessimism is unlikely to ground the overwhelming importance of existential risk mitigation unless it is coupled with an empirical hypothesis: the time of perils hypothesis. However, I argue, the time of perils hypothesis is probably false. I conclude that existential risk pessimism may tell against the overwhelming importance of existential risk mitigation.

12:40 - Johan Gustafsson, “Being Effective about Altruism Entails Utilitarianism”

Even though most effective altruists are utilitarian, effective altruism is often taken to be a broad church that allows for non-utilitarian moral views. In this talk, however, I argue that very minimal assumptions about effectiveness and altruism jointly entail, through value-pump arguments, all premises of Harsanyi’s social-aggregation argument for utilitarianism. So effectiveness and altruism entail utilitarianism. Value-pump arguments are often dismissed with claims that agents can avoid value-pump if they have foresight or if they are resolute. These objections don’t work. The foresight objection can be rebutted with value pumps that also works against agents who rely on foresight. And the resoluteness objection is either incoherent or conflicts directly with the ideal of effectiveness. Hence effective altruists are utilitarian.

15:00 - Antony Millner, “Non-dogmatic Climate Policy”

Disagreements about normative aspects of social time preferences have led to estimates of the Social Cost of Carbon (SCC) that differ by orders of magnitude.  We investigate how disagreements about the SCC change if planners are non-dogmatic, i.e., they admit the possibility of a change in their normative views, and internalise the preferences of future selves.  Although non-dogmatic planners may disagree about all the contentious aspects of social time preferences, disagreements about the SCC reduce dramatically.  Admitting the possibility of a change in views once every 40 years results in a five-fold reduction in the range of recommended SCCs.

15:40 - Dean Spears, “A Larger World Population Yields Benefits that Exceed Climate Damages”

The global human population is projected to peak within about one century and then to shrink. A shrinking population generates fewer emissions but also has economic costs not previously considered in integrated climate-economy models: fewer innovators and an unfavorable dependency ratio. Here we use DICE, a leading climate-economy model to contrast the UN-projected Depopulation scenario with population Stabilization. The economic mechanisms yield higher long-run per capita living standards under Stabilization than Depopulation, even accounting for the economic harms of climate damages. The temperature reduction benefits of Depopulation are small relative to Stabilization because of population momentum, which ensures that total population sizes in these scenarios remain similar for many decades despite different birth rates. Population sizes diverge significantly only after the majority of cumulative historical emissions are locked-in.

16:30 - Heather Browning, “Determining the line of acceptable welfare”

Under several different plausible theories in population ethics (e.g. sufficientarian, negative utilitarian), there is an asymmetry in value between increasing utility through decreasing suffering, and through increasing pleasure. In these cases, it will thus be important to identify the dividing line at which welfare switches from negative to positive, or unacceptable to acceptable. Particularly within animal welfare, while the aim of most animal welfare science is to assess animal welfare, it is often far less clear what types of recommendation to make based on these assessments. There is a difference between work simply aimed at the comparative goal of improving welfare (i.e. increasing the level of welfare from its previous baseline) and that aimed at the absolute goal of achieving good welfare. The latter involves a normative element, in setting a judgement line regarding what counts as acceptable welfare. In this paper I explore what this means both conceptually and empirically, as well as two of the possible ways of setting this cutoff line – at the ‘zero point’ where negative turns to positive welfare (often taken as the turning point for a life worth living), and as a fixed line set by reference to some target population (e.g. wild members of the species or the welfare levels achieved through current institutional best practice). I conclude by noting that while there may be no single privileged way of setting this cutoff line, it is important to be clear on which line we are choosing in different contexts, and our reasons for doing so.

17:10 - Isabel Juniewicz, "Global Decline in Fertility Rates"

Fertility rates have been declining globally, with an increasing share of the countries in the world experiencing persistently and sometimes substantially below replacement fertility rates. While globally the fertility rate is still currently above replacement, it is projected to decline below replacement within decades, and this decline has negative implications for long term future growth. The first part of this talk will discuss the magnitude, prevalence and correlates of the decline, with a focus on the decline even among many but not all of the highly religious groups that have been recently held up as exceptions to this trend. The second part of this talk will discuss the effects of various government policies, such as baby bonuses, cash transfers, parental leave policies, as well as the difficulty of assessing the effectiveness of such programs on completed cohort fertility.