We have had this argument before. A disaster develops: a natural disaster like Katrina, a financial and economic disaster like the Great Depression, or most recently, a health disaster like the coronavirus pandemic. Soon people who see goverment as a source of beneficent compassion, not to mention possessor of unmatched resources, begin to badger and crow. “See? Without government, we cannot act effectively. The disaster only gets worse. If you would stop running down our public institutions, they could help all of us a lot better.”
Disasters come and go, but the basic argument does not change. The argument’s logical structure assumes that government intervention benefits people at large, and that without government action, we would all be worse off. Yet the logic cannot possibly produce that conclusion. It only yields that conclusion because one takes the premise for granted, that goverments do good. Consider these paragraphs from a New York Times article that advocates government action in the coronavirus pandemic:
As the global financial system teetered on the brink in the fall of 2008, a Republican president and congressional Democrats, recognizing the severity of the crisis, worked together to pass a bailout package for the financial industry that despite its many flaws and compromises helped to prevent a second Great Depression.
The coronavirus pandemic poses a different kind of challenge, threatening the nation’s health as well as its economic prosperity, but it must be met with the same kind of audacious, coherent and coordinated response.
I can only say: “Give me a break, people.” Can you show me the bailouts prevented “a second Great Depression,” any more than you can show me government’s response did not make things worse? When we have a question of public policy in front of us, we have to weigh costs and benefits, a comparison almost as uncertain after the fact as it is before. Does anyone, for example, want to compare benefits realized by bailout recipients against costs the bank panic imposed on people who lost their homes for good? Do we know how many suicides we might have prevented if more families had been able to keep their homes? Of course not.
If you say, “All proposals for government intervention have this problem. You cannot design public policies in a laboratory,” we respond, “That’s the point. Only a controlled experiment yields conclusions about what would have happened without intervention.” That is why we call them controlled experiments. You cannot draw any conclusions about what would have happened in 2008 if government had not intervened in the financial crisis – or let’s call it by its true name, a bank panic. Outcomes might have been better without intervention, they might have been worse, or they might have been a mix of better and worse outcomes.
Yet when analysts see an intervention they approve, they conclude its effects were, on balance, beneficial. You need not offer any evidence for an article of faith. Your affinity for the intervention, and your general sense that others share your affinity, supports your conclusion. So people who like government interventions spread their rose petals twelve years later, to promote the idea that their pet intervention had great, redemptive outcomes that only government could effect.
A comparative argument of this type has no more support than the alternate view, that we would have been better off if government had not intervened at all. Both sides operate entirely on what-if territory that depends on speculation. People who argue in favor of the actual course taken have no logical grounds for comparison, either with outcomes that would have occured with no intervention, or with different interventions. They have only speculation, and confidence in their premises.
What-if arguments are fun, for people who like intellectual play. They can reveal assumptions, predilections, and premises, as well as weak links in argumentative chains. What-if proposals reveal nothing concrete or persuasive about unrealized outcomes. Nor do they help you evaluate actual outcomes. They give you an evening’s pleasure over a glass of port, and in the right group, test your conversational skills.
When we do try to draw lines of causation between a particular intervention and a specific outcome, we have to rely on data. Data include not only economic evidence, but evidence about intentions, decisions, general motives, personal relationships and loyalties, secondary effects, and so on. Historians, social scientists, and policy analysts – specialists every one – generally agree that variety in types of evidence improves one’s confidence about conclusions.
The same holds when we evaluate outcomes, independent of what caused them. These principles of logic and evidence hold for analysis of military engagements, pandemics and epidemics, health care delivery systems, management of civil conflict, crime prevention, business regulation, environmental effects, and so on. They should make us modest in our arguments when we promote a particular intervention, whether we place our hopes with government action, or elsewhere.
To return to our first case, evidence about lines of causation related to the 2007-2008 bank panic may help us understand causal links in this context, including causes and effects related to government interventions. Evidence about what did occur does not help you reach conclusions about, or evaluate developments that did not occur. We cannot know what would have occurred on another causal path, whatever predilections we might have to speculate. Well-constructed thought experiments help us understand a problem. They do not help us predict, either what will happen, or what would have happened.