The Trump Administration has said that it wants to help communities address the nation's serious housing affordability challenges, but the President's 2021 budget would do the opposite, slashing housing assistance and community development aid next year by $8.6 billion, or 15.2 percent (not counting the impact of inflation).
Dan Little is kind of a wonky Marxist. He is deep into sociological lingo in this this (also wonky) piece. Nonetheless the techniques of proving, or at least making strong arguments, about causation in social sciences is an important and provocative subject, including in economics.
This diagram provides a map of an extensive set of methods of causal inquiry in the social sciences. The goal here is to show that the many approaches that social scientists have taken to discovering causal relationships have an underlying order, and they can be related to a small number of ontological ideas about social causation. (Here is a higher resolution version of the image; link.)
We begin with the idea that causation involves the production of an outcome by a prior set of conditions mediated by a mechanism. The task of causal inquiry is to discover the events, conditions, and processes that combine to bring about the outcome of interest. Given that causal relationships are often unobservable and complexly intertwined with multiple other causal processes, we need to have methods of inquiry to allow us to use observable evidence and hypothetical theories about causal mechanisms to discover valid causal relationships.
The upper left node of the diagram reviews the basic elements of the ontology of social causation. It gives priority to the idea of causal realism -- the view that social causes are real and inhere in a substrateof social action constituted by social actors and their relations and interactions. This substrate supports the existence of causal mechanisms (and powers) through which causal relations unfold. It is noted that causes are often manifest in a set of necessary and/or sufficient conditions: if X had not occurred, Y would not have occurred. Causes support (and are supported by) counterfactual statements -- our reasoning about what would have occurred in somewhat different circumstances. The important qualification to the simple idea of exceptionless causation is the fact that much causation is probabilisticrather than exceptionless: the cause increases (or decreases) the likelihood of occurrence of its effect. Both exceptionless causation and probabilistic causation supports the basic Humean idea that causal relations are often manifest in observable regularities.
These features of real causal relations give rise to a handful of different methods of inquiry.
First, there is a family of methods of causal inquiry that involve search for underlying causal mechanisms. These include process tracing, individual case studies, paired comparisons, comparative historical sociology, and the application of theories of the middle range.
Second, the ontology of generative causal mechanisms suggests the possibility of simulations as a way of probing the probable workings of a hypothetical mechanism. Agent-based models and computational simulations more generally are formal attempts to identify the dynamics of the mechanisms postulated to bring about specific social outcomes.
Third, the fact that causes produce their effects supports the use of experimental methods. Both exceptionless causation and probabilistic causation supports experimentation; the researcher attempts to discern causation by creating a pair of experimental settings differing only in the presence or absence of the "treatment" (hypothetical causal agent), and observing the outcome.
Fourth, the fact that exceptionless causation produces a set of relationships among events that illustrate the logic of necessary and sufficient conditions permits a family of methods inspired by JS Mills' methods of similarity and difference. If we can identify all potentially relevant causal factors for the occurrence of an outcome and if we can discover a real case illustrating every combination of presence and absence of those factors and the outcome of interest, then we can use truth-functional logic to infer the necessary and/or sufficient conditions that produce the outcome. These results constitute JL Mackie's INUS conditions for the causal system under study (insufficient but non-redundant parts of a condition which is itself unnecessary but sufficient for the occurrence of the effect). Charles Ragin's Boolean methods and fuzzy-set theories of causal analysis and the method of quantitative comparative analysis conform to the same logical structure.
Probabilistic causation cannot be discovered using these Boolean methods, but it is possible to use statistical and probabilistic methods in application to large datasets to discover facilitating and inhibiting conditions and multifactoral and conjunctural causal relations. Statistical analysis can produce evidence of what Wesley Salmon refers to as "causal relevance" (conditional probabilities that are not equal to background population probabilities). This is expressed as: P(O|A&B&C) <> P(O).
Finally, the fact that causal factors can be relied upon to give rise to some kind of statistical associations between factors and outcomes supports the application of methods of inquiry involving regression, correlation analysis, and structural equation modeling.
It is important to emphasize that none of these methods is privileged over all the others, and none permits a purely inductive or empirical study to arrive at valid claims about causation. Instead, we need to have hypotheses about the mechanisms and powers that underlie the causal relationships we identify, and the features of the causal substrate that give these mechanisms their force. In particular, it is sometimes believed that experimental methods, random controlled trials, or purely statistical analysis of large datasets can establish causation without reference to hypothesis and theory. However, none of these claims stands up to scrutiny. There is no "gold standard" of causal inquiry.
This means that causal inquiry requires a plurality of methods of investigation, and it requires that we arrive at theories and hypotheses about the real underlying causal mechanisms and substrate that give rise to ("generate") the outcomes that we observe.
This week, Project Syndicate catches up with Angus Deaton, the 2015 Nobel laureate in economics and a professor emeritus at Princeton University.
Project Syndicate: Before the US midterm elections in 2018, you challenged claims that President Donald Trump had been good for the economy, highlighting, among other things, that strong stock-market performance reflects the redistribution of income from labor to capital, at the expense of the vast majority of Americans. Trump’s defenders would no doubt point to low unemployment and rising wages, even for those at the bottom, in defending the president’s performance. In advance of November’s presidential election, what should American voters understand about the Trump economy?
Angus Deaton: The US has undergone a long and very slow recovery from the Great Recession. But that recovery is attributable mostly to the actions of previous administrations, though Trump’s tax cuts also played a role. And while the employment rate has risen (as it always does when the economy is doing well), and unemployment is low, the employment rate remains lower than before the Great Recession, and substantially lower than in 2000.
In fact, the long-run trend of less-educated Americans dropping out of the labor force shows no sign of having abated. The quality of jobs is deteriorating as large firms outsource many of their operations. Meanwhile, the stock market is booming in response to giveaways to corporations.
PS: You’ve highlighted the correlation between US counties with elevated mortality rates for white people – especially owing to what you and Anne Case call “deaths of despair,” caused by suicide, drug overdose, and alcohol abuse – and counties that voted for Trump. You’ve also pointed out that deregulation, which Trump has fervently embraced, can literally kill – the opioid crisis being a case in point. In other words, Trump’s policies are simultaneously deepening and capitalizing on his voters’ misery. The same goes for the gutting of US environmental regulations. What would Trump’s successor, whenever he or she arrives, have to do to break this cycle?
AD: I’m not sure that it is a cycle that feeds into itself, or that is in any way inevitable. It is just bad policy. The opioid crisis should not have happened. (It has not happened in other countries.) US regulators allowed it. Those regulatory failures were not introduced – or ameliorated – by the Trump administration.
Corporate interests are not the only ones that matter. There is nothing stopping a new administration from recognizing that and working for others.
PS: Information and communications technology “need not doom us to a jobless future,” you argued last year; “enlightened policymaking still has a role to play.” What policies – or policy goals – should top the list?
AD: New technologies have been the primary source of prosperity for more than 200 years, but they often cause disruption, sometimes for a long time. A more comprehensive social safety net would be a good start. In my view, one key is to reduce the incredibly high costs of health care in the US, and to break the link between health care and employment. What I do not think would solve the problem – or even help – is a universal basic income.
PS: As someone who argues for economic policies that look beyond growth, what do you think of alternatives to GDP, such as Bhutan’s Gross National Happiness index, the UN’s Human Development Index, or the US-based Genuine Progress Indicator? Are such broader metrics useful, or are they, like US poverty data, too vulnerable to manipulation to underpin real-world change?
AD: Some such broader metrics have proved their usefulness as supplementary information, but they are not alternatives to GDP. In my view, the priority should be to improve GDP itself, excluding many of the things that do not improve human wellbeing (like useless but profitable medical procedures), and ensuring a much better account of distribution – in particular, how personal disposable income is spread across different groups.
That said, the pre-eminent position of GDP is undeserved. In fact, if GDP is interpreted as a measure of how well the economy is serving its people, it is often extremely misleading.
BY THE WAY . . .
PS: When it comes to addressing rent-seeking in the US, you’ve identified policies that would help – such as reframing and enforcing antitrust laws – but noted in 2018 that you are “not very optimistic” about their implementation. Do any of the Democratic presidential contenders give you more hope? Alternatively, are there other less-than-ideal, but still useful policies that seem more feasible?
AD: One key step would be to improve the public’s understanding of the extent of rent-seeking – for example, through investigative journalism of particular cases. News organizations such as The Washington Post and CBS News did a terrific job of exposing the abuses around opioids.
But the bad guys are still extremely powerful. The health-care industry has five lobbyists for every member of Congress, and those who do their bidding rarely seem to suffer electoral consequences. Little wonder so many young people think American democracy is broken.
All of the Democratic contenders give me more hope than Trump does. But change will be difficult for anyone.
PS: Using money as a proxy for utility simplifies things when building formal economic models, but philosophers, as you’ve noted, reject the equation of financial gain and wealth with happiness. Assuming that the economics profession and society alike would benefit if economists were familiar with the relevant philosophical arguments, which works should be required reading?
Another excellent book by philosophers who think hard about economics is Economic Analysis, Moral Philosophy, and Public Policy, by Daniel Hausman, Michael McPherson, and Debra Satz. In it, they work through a number of different philosophical and ethical systems and show how they would apply to a series of ethical questions in economics. I think all economists should read it.
SUBSCRIBE NOW
Subscribe today and get unlimited access to OnPoint, the Big Picture, the PS archive of more than 14,000 commentaries, and our annual magazine, for less than $2 a week.
SUBSCRIBE
PS: The US federal government recently reported that Americans’ life expectancy rose for the first time in four years in 2018. The increase was only a month, but it was driven by the first decline in drug-related deaths in 28 years, including in hard-hit states like Ohio. Has the US turned the corner?
AD: I certainly hope so, especially as far as bringing the opioid epidemic under control is concerned. But there is a long way to go, and the other deaths of despair – suicides and deaths from alcoholic liver disease – continue to rise, especially among those without a college degree. A one-month increase in life expectancy is a terrible outcome, compared with the two-year increase that the US used to achieve every decade, and that other rich countries continue to come close to achieving, albeit somewhat more slowly.
PS: You’ve described the immediate aftermath of winning the Nobel Prize in Economics as “being transported to a fairyland.” What was most surreal, and most rewarding, about that experience?
AD: The Swedes have been giving these prizes for more than a century, and they have perfected the process. The Swedish people – from ordinary working people to the royal family – have embraced the Nobel prizes, so winners become guests of the entire country, standing at the center of a glittering and magical winter festival that lasts a full week.
You get to bring your co-authors, your friends, and your family. My grandchildren will remember it all their lives. It really was like being in one of the fairylands that I read about in books or saw in theaters as a child in Edinburgh.
