(minor updates since original post)
This is an edited excerpt from the work in progress sixth draft of the Candidate Scoring System. I figured it would be useful for people who are interested in capitalism and socialism. It's far from perfect, but I have not found any remotely satisfactory overview anywhere else. This can be a starting point for people who want to debate whether economic systems should be treated as a priority in EA.
Capitalism is an economic system where the means of production - land, machinery, investment capital - are mostly held by private owners. Socialism is an economic system where they are owned by the public in a collective or governmental organization. For a variety of reasons, it could be important to take a stance on the question of which is preferable.
However, it’s not straightforward to judge capitalism and socialism because the question is simply too vague to be a valid basis for direct research. There are many very different ways that these systems can be realized, so modern economic and political science scholarship focuses on narrower and better-defined issues. Acemoglu and Robinson (2015) note, “we do not believe the term capitalism to be a useful one for the purposes of comparative economic or political analysis… both Uzbekistan and modern Switzerland have private ownership of capital, but these societies have little in common in terms of prosperity and inequality because the nature of their economic and political institutions differs so sharply.” Socialism is also very broad. Deep ideological disputes exist among leftists in America, concrete policy proposals are rare and controversial, and there is no clear conception of what socialism would actually look like (New Republic). The Socialism 2019 conference contains almost no discussion of the structure of a socialist economy.
If we want to evaluate the desirability of generic increases or decreases in the probability of socialist change, we cannot select a specific policy proposal that we would prefer. Instead we must survey the various possible forms and components of socialism and produce a vague expectation over the lottery of possibilities.
First, when it comes to the centrally planned economic models of 20th century socialist states, economists overwhelmingly regard them as inferior to capitalist economies. There is disagreement on whether early Soviet industrialization was expedited much by central planning (Krugman 1994, Allen 2005), but the human costs are unequivocal. The entire suite of socialist policies reduced Soviet agricultural productivity by about 50% (Johnson and Brooks 1983). Central planning went awry in the USSR when poor leadership appeared later on (Allen 2001, Allen 2005).
The communist system has been disastrous for Cuba, the US embargo is not the sole cause of their problems (Jales et al 2018, Salazar-Carrillo and Nodarse-León 2015, Ribeiro et al 2013, Ward and Devereux 2012). Cuba was also buoyed by significant Soviet aid and trade subsidies during the Cold War, and more recently is supported by $5 billion annual remittances from expatriates, a very large amount (relative to the size of its economy) which might give it a bonus sufficient to greatly exceed the losses from the embargo. While Cuba’s literacy and healthcare metrics are reportedly very high now unlike most aspects of their society, they were also relatively high before the revolution and have not improved at a stellar rate (Salazar-Carrillo and Nodarse-León 2015). Also, Cuba’s actual healthcare quality is worse than reported (Berdine et al 2018). That being said, Cuba’s hurricane preparedness is quite good (Jacobin).
The fact that these states were generally oppressive autocracies explains their totalitarian crimes but not their economic deficiencies. In addition, there are plausible theoretical explanations for economic failures in centrally planned socialism. One is the famous “calculation problem,” another is the reduction of individual incentives to work harder, another is Shleifer and Vishny (1991)’s argument that central planning creates opportunities and incentives for planners to artificially create shortages in order to collect bribes.
The consensus against these models is shared not just by practically all economists, but also by historians and social scientists, and by policymakers across the world – including the Communist Party of China. Of course the majority of modern socialists also disavow them, so they are unlikely to be repeated. That being said, it’s not straightforward to assume that they won’t be repeated. Many leaders of failed historical socialist movements and nations were avid scholars of Marxist theory, not making ignorant mistakes about its content. Socialist regimes also varied significantly in their ideologies, such as the various stages of Soviet doctrine, Maoism, Juche, and other systems, yet all ran into similar problems. Therefore, historical failures cannot be attributed to a simple and avoidable ideological error on the part of socialist leaders. Moreover, a small number of modern leftists do defend the track record of these regimes. This means there is a non-negligible chance of a repetition.
Still, most socialists nowadays espouse alternative models. Some advocate ParEcon, an economically radical idea which is difficult to properly evaluate. However it is often criticized by leftists and gets little attention these days. Socialists usually support a more straightforward increase in public ownership and control of economic decisions, extending the ideas of modern regulatory and welfare states and workplace democracy while falling short of proper central planning. Probably the most notable proposal for this is Schweickart (2011). We have not seen any rigorous, holistic evaluation of such schemes, and such an evaluation may simply be impossible until they are tried. But we can look specifically at the major components of these visions, which have been studied in isolation.
One aspect of many socialist plans is greater government regulation of the economy. But Hall and Lawson (2014) collected 198 relevant empirical studies published in highly selective social science journals, and we add one more recent one (Jackson 2017). The result is that economic freedom corresponds with good outcomes in 68% of studies and bad outcomes in just 4% of studies. Hall and Lawson find that this result might be weakened by publication bias but find no evidence to indicate that it would be overturned. In a more recent, narrower and simpler literature review, Horpedahl et al (2019) argue that economic freedom generally helps achieve the aims of social justice (which is good for social welfare, ceteris paribus). The think tanks which produce the rankings of economic freedom – mainly the Fraser Institute, but also the Heritage Foundation – are conservative, but highly ranked (see reports here) and the economic freedom rankings are commonly accepted in the academic literature. Now it’s worth noting that other aspects of socialism could temper the downsides of free markets and thereby reduce the necessary level of economic regulations, but it’s not clear whether a socialist government would be inclined to take advantage of this opportunity.
Another aspect of many socialist plans is a larger government with more public spending. It is an open question whether or not government administrative agencies are inefficient compared to private companies. However, heavy public spending does lead to less economic growth (Bergh and Henrekson 2011, Matteo and Summerfield 2017). Of course, public spending can be better targeted for issues like inequality and externalities to outweigh economic costs, so a balance must be struck. However, our issue evaluations have often found cases where increasing government spending would be neutral or harmful. Therefore, sweeping plans for much more government spending seem like a poor idea.
Another aspect of many socialist programs is the replacement of privately owned enterprises (POEs) with state-owned enterprises (SOEs). This is a central aspect of China’s market socialism, though Western socialists frequently reject it with the label “state capitalism.” Two comprehensive literature reviews have shown that SOEs are inferior to POEs (Megginson and Netter 2001, Shirley and Walsh 2001). Also, two recent studies have found that Chinese SOEs are inferior to POEs (Boeing et al 2015, Fang et al 2015). Goldeng et al (2008) found that POEs outperformed SOEs in Norway in the 1990s. However, some of the recent work questions this point of view. Jakob (2017) looks at an international dataset and finds that there is no difference in performance between POEs and SOEs. A metanalysis by Bel et al (2010) found that privatization of local waste and water services has no effect. Omran (2004) suggested that privatization of Egyptian firms in 1994-1998 did not create significant improvements. Demsetz and Villalonga (2001) found that there is no systematic relationship between ownership and performance in US enterprises. But even if SOEs are just as good as POEs on the margin, it would probably be worse to push radical changes to implement many more of them. Meanwhile on the theoretical side, Shleifer and Vishny (1994) argue that an economy of state-run enterprises merely magnifies the flaws of democratic governance (and American governance is indeed flawed, and will remain so for the foreseeable future).
State-owned enterprises receive varying support from leftists depending on the industry. A more popular SOE program would be nationalization of the finance industry, outlined by Schweickart (2011) under the label “social control of investment.” But just as in other industries, public banks are less efficient than private banks (Megginson 2003). Furthermore, public banking in the West would cut investment in foreign economies, as public banks would be politically mandated to support projects which maximize employment for the domestic population. This would allow more severe poverty in the developing world.
Another major plan of many contemporary socialists is the use of worker cooperatives, firms which are owned and managed by the workers. Pérotin (2012, 2015) summarizes research to show that worker cooperatives have broadly positive impacts. As for firms which are merely owned by workers, Kruse (2016) summarized existing research to show that worker ownership is modestly positive for both firm performance and employee welfare, though a study by Monteiro and Straume (2018) found inconclusive and potentially negative impacts on firm efficiency in Portugal. But a significant downside of worker-owned and especially worker-managed firms in the US is that they discourage outsourcing to needier workers in poorer countries. The 20th century’s socialist programs also suggest that mandatory collectivization could have very bad effects, though their problems were arguably caused by state control and mismanagement rather than the mere fact that they were collectives.
Leftists often allege that institutional bias from the capitalist system distorts the views and research of economists and historians, so that socialist programs are better than they seem. But we have seen little evidence of such bias going on, and it is inconsistent with two major academic trends: the academic popularity of Marxist doctrine in economics in the late 19th and early 20th century, and the widespread, almost unquestioning academic acceptance of anti-capitalist views in certain subfields of the humanities today. And we haven’t seen any evidence of censorship or censure of anticapitalist views in economics departments. On the other hand, empirical analysis of socialist regimes is actually likely to be over-optimistic because they have typically been autocracies, which deceive their audiences (Gregory 1990, Kornai 1992, Martinez 2019). American economics textbooks systematically overestimated Soviet growth during the Cold War (Levy and Peart 2011). Cuba’s official statistics are notoriously unreliable (Salazar-Carrillo and Nodarse-León 2015, Berdine et al 2018).
To summarize, government control of the economy seems bad, heavy public spending seems bad, state-owned enterprise seems bad, and worker cooperatives could be good or bad depending on the context of their implementation. All of these conclusions seem relatively robust on their own. However, these programs may have interactions with each other so that a multifaceted socialist program could not be reliably judged merely by taking the sum of its parts. Socialists usually believe that interactions will be positive – in other words, socialist projects could perform much better than what studies indicate if their environment has more socialist aspects in other ways. However, we have seen no good arguments for this view. And negative interactions seem equally possible. The benefits of socialist programs might have diminishing returns as they stack – in other words, perhaps modest reforms would be sufficient to capture the potential benefits of socialist ideas, with additional reform being a pointless pursuit of unnecessary purity. And an excess of socialist programs could concentrate too much power in the government, creating incentives and opportunities for totalitarianism and abuse as we saw in the 20th century. So the question of interactions doesn't give us much reason to take a side, though it does yield another layer of uncertainty which weakens our previous conclusion that most socialist programs would be bad.
Political philosophy does not change the picture. Cohen (2009) argues that socialism would theoretically be more compatible with ideal principles like equality and fairness, but Brennan (2014) demonstrates that Cohen’s argument does not work. Brennan further argues that capitalism is superior in terms of theoretical alignment with ideal principles, but Hall (2014) finds that this part of his argument fails. In any case, such appeals to ideal principles have little relevance for real economic and social outcomes. Similarly, Marx argued that exploitation and alienation are inherent to capitalism (Wolff 2003), but there is no good evidence showing that this hurts aggregate well-being relative to alternatives. In fact, the capitalist organization of alienating and exploitative labor (assuming Marxist definitions of the terms) has led to substantial improvements in population size and quality of life over the history of industrial society. Fuerstein (2015) meanwhile argues that capitalism weakens democracy and further shows that this weakening of democracy has real social costs. However, he does not compare this to possible ways that socialist systems might weaken democracy, and he outlines solutions that are compatible with capitalism.
Socialist change might come in the form of violent revolution, which would have a number of additional domestic and international costs. American leftists have increasingly normalized violent tactics in the last several years (Wikipedia, ROCIC). Even if socialism were desirable, it may not be good enough to outweigh the risks of violence. And even if socialism were desirable enough to outweigh the costs of violence, the event of a failed violent revolution would still be a clearly bad thing. This gives an extra reason to prefer maintenance of the existing economic system and avoid insurrections in the first place.
Despite these issues, there is still substantial value of information in testing a better model of socialism, if a good pathway to running a good test can be identified. No one has attempted the most recent, refined socialist plans. If a nation demonstrated that they can work well, then many other nations could improve their own policies accordingly. If a nation tried and failed, then they could eventually return to capitalism and their experience would inform people in other countries to refrain from pursuing the matter. This benefit is large though it is slightly offset by the risk that a socialist movement or government would reduce the prospects for experimentation with other (i.e. right-wing or radical centrist) economic proposals.
Looking at the general degree of uncertainty in the issue, and especially taking experimental value into account, it does seem possible to sketch out a socialist proposal that would probably be preferable to continuing the current trend of capitalist policies. However, that doesn’t mean that the actual results of a socialist movement would meet this high standard. In fact, there are reasons to be specifically pessimistic about the results of a socialist movement. Trends among many current Western leftists – overconfidence in their point of view, disinterest in policy planning, economic denialism and folk-economic beliefs, historical revisionism and apologia, dehumanization and violence towards political opponents (Wikipedia, ROCIC), censorship and censure of internal dissent, internal sectarianism, explicit opposition to objectivity and rationality, and vindictive attitudes on social justice politics – increase the chances of repeating the failure modes of 20th century socialist programs. These trends might be considered relatively benign in the context of a minority movement in a liberal society with stable institutions, but become more dangerous if they are prevalent in a revolutionary majority which is establishing its own institutions and leadership.
Overall, there is decent but not decisive evidence against socialism. It’s plausible that some forms of socialism, particularly those which leverage worker cooperatives while maintaining competitive markets and a reasonably sized government, could be better than capitalism; this seems worth testing but has not been substantiated well enough to justify activism and reform across the board. In addition, there is no guarantee that generic socialist forces in America would lead to a good socialist system rather than a bad socialist system. Nor is there a guarantee that socialist change would be a peaceful process. So broadly increasing the stability of the capitalist system seems like a good thing.
This finding is too weak and uncertain to be justified as a main cause priority. However, it still has interesting implications for Effective Altruism. Numerous writers have alleged that charity is worse than it seems because it could reinforce capitalism, and argued that EAs have a duty of justice to promote new economic systems that could help the developing world escape poverty. However, the finding here means that this argument must be reversed. Charity is extra good because it reinforces capitalism, and EAs have a duty of justice to reinforce (and refine) capitalism in order to help the developing world. Of course neither of these arguments actually make sense - the number of people whose political attitudes are actually changed by philanthropy is negligible as far as anyone can tell, and the "duty of justice" is a moral falsehood. But as far as academic debates are concerned, a good deal of institutional orthodoxy will have to be revised in the light of this new idea that capitalist systems are usually superior to socialist systems.