Access to the internet and the rise of social media has overloaded voters with information and exposed them to a proliferation of fake news. Using political budget cycles, or the tendency for politicians to increase the budget in run-up to elections to win more votes, as a proxy for misinformation, Fabio Padovano and Pauline Mille show in new research that voters who score higher on the OECD’s  Programme for International Student Assessment and achieve a higher level of education are better able to hold politicians to account.

Research in public choice and political economy has traditionally held the view that the more informed voters are, the better equipped they are to oversee politicians and prevent chicanery. Supporting evidence for this thesis comes from the empirical literature on conditional political budget cycles (PBCs). Several studies have shown that, in countries where the media are freer and enjoy larger distribution within the population, incumbent governments are less prone to manipulate the budget before elections to “buy” votes (all other things being equal). Scholars generally consider such practices a symptom of a malfunctioning democracy, because PBCs essentially amount to politicians fooling voters. More generous budgets before elections actually imply higher tax liabilities after; voters must be shortsighted and poorly informed not to correctly interpret the consequences of lax fiscal policies before elections and vote for politicians who resort to these manipulations. Hence, the importance of the media’s role in disseminating the information necessary for the public to recognize these practices and deprive politicians who use them of any electoral reward.

Yet, beyond a certain level, more information may become too much information, as well as information of lower quality. The digital revolution, including the diffusion of the internet and uptake of social media platforms, has significantly reduced the costs of supplying new information. The much greater quantity of news which voters receive makes it more difficult for them to assess the news’ relevance and factual accuracy. Secondly, the diffusion of social media has contributed to the spread of fake news and biased information, thus significantly decreasing the average quality of information available to voters.

In the past, the traditional media’s concerns about their reputation encouraged them to find and publicize  reliable sources of information. Nowadays, by contrast, social media platforms allow information from various sources to be shared, retweeted, and distributed in a matter of seconds, contributing to a blurring of the boundary between professional journalists and amateurs. Combined, these two phenomena overload voters with information or outright fake news, creating a classical problem of “signal extraction.” Many political scientists, media commentators, and watchdogs concur that the fact that more and more voters base their electoral choices on “noisy” and biased information poses a threat to the proper functioning of democracies. Examples that motivate such concerns include the 2016 U.S. presidential campaign, when fake information was widely spread, allegedly in favor of Donald Trump. Also in 2016, during the Brexit referendum campaign, the tabloid The Sun published the (unsubstantiated) news that Queen Elizabeth II was secretly in favor of Brexit, despite her duty to remain politically neutral. True or not, the news influenced the final result. Lastly, biased information about phenomena such as immigration and the causes, nature, and appropriate policies towards the Covid-19 pandemic have fueled populism in many European countries.

What are the possible remedies? Several contributions from different social sciences concur that education offers the best solution in solving the signal extraction problem created by over-information and fake news. Education promotes long-term memory, cognitive development, and emotional intelligence. Developing the capacity for long-term memory by teaching consolidated literary, scientific, and cultural knowledge at an early age has been shown to help individuals, even in later stages of their lives, to recognize valuable information and to filter out fake news. Similarly, several studies demonstrate that recognition of fake news heavily depends on someone’s level of “emotional intelligence,” i.e., the ability to understand and regulate one’s emotional reactivity and to relate emotions to context, a capacity which is in turn positively correlated with educational attainment. All this suggests that a) education enables voters to address the signal extraction problem mentioned above; and b) the average educational attainment of the public should act as a constraint on politicians’ tendency to generate PBCs. This is precisely the analysis that we have conducted in a recent study.

Specifically, we empirically examined whether the quality and level of a voting population’s education reduces the PBC. Our sample comprised 46 (self-declared) electoral democracies between 2000 and 2019. Without going into technicalities, the regression model that we estimated is the same used in the conditional PBC literature to explain the magnitude of the cycle, augmented by indicators of the country’s level and quality of education. We have proxied each country’s quality of education by its average results in each round of the Programme for International Student Assessment (PISA), an OECD study intended to evaluate and compare the achievement of 15-year-old students in mathematics, science, and readings/humanities. We measure the level of education by two quantitative indicators: the share of the population with a secondary (high school) degree and a tertiary (college and university) degree.

Each indicator has its pros and cons. The advantage of the PISA scores is that they reflect the different academic standards in various disciplines achieved by students about to receive a similar (secondary education) degree from different countries. The main disadvantage is that these scores apply to students who have not yet obtained the right to vote. Hence, PISA scores should be considered a proxy of the attention that each country devotes to education. On the other hand, the share of the country’s population with at least a college or university degree has the advantage of referring to students with the right to vote, but does not discriminate for differences in the quality of different sample countries’ university systems. Such indicators do exist, the Shanghai rankings being one, but they are biased for the purpose of our analysis, since, for example, American Ivy League universities attract students (and educate voters) from all over the world. The sample size is determined by the availability of the PISA scores: the program started in the year 2000 and had its last comparable round of tests in 2018, before Covid-19 impacted teaching methods.

Several interesting results emerge from the analysis. First, higher PISA scores and larger portions of the population having attained at least a tertiary education appear to be negatively correlated with politicians’ propensity to manipulate the budget before elections. Secondary education never shows any explanatory power when the PISA scores are included in the regression model. This suggests that, for a voter to become able to evaluate and control politicians’ conduct, what matters is not just holding a degree, but the quality of education obtained.

Second, when proxies for education are considered together with those for the diffusion of information, especially through the internet, only education appears to place an effective check on governments’ incentives to generate a PBC. In other words, access to information does not alone equip citizens to better monitor their politicians’ behavior. This confirms that education does enable voters to solve the signal extraction problem caused by over-information and fake news. 

To further investigate that point, we broke the sample countries into two subsamples according to the diffusion of internet and social media. In countries with a diffusion of internet above the average, where the phenomena of information overload and fake news are likely to be more serious, both the quality and level of education appears to play an even more important role, as the larger size of the associated coefficients reveals. In other words, when the quality of information is lower and its quantity larger, education matters more. Investing in education therefore appears the most promising path towards refocusing the political debate onto factual grounds, with the positive consequences for the proper functioning of democratic systems that would be likely to follow.

Third, when the PISA scores are disaggregated by subjects, high scores on humanities appear to have a slightly greater impact in terms of limiting the PBC than science and math. Testing in humanities-oriented subjects tends to evaluate to a greater extent than STEM subjects not only the student’s ability to read longer and more complex texts, but also his/her knowledge about history, which can serve as a valuable discipline for the interpretation of political and social trends. The observed advantage of humanities over science and math is not large (0.0093 in absolute value for humanities, 0.0084 for science and 0.0066 for math, all significant at the 5% level) but it is not negligible in statistical terms. Fourth, to complete our understanding of the interplays between education and politicians’ electoral behaviors, we have compared countries by their levels of democracy, measured by the Polity IV database. This database uses a variety of indicators to distinguish well-functioning democracies (with a score from 6 to 10) from malfunctioning ones (with a lower score; also called “open anocracies”). More democratic countries typically invest in education more and information circulates more freely there. Yet, information could be a double-edged sword: on the one hand, governments in less democratic countries can manipulate and distort information more easily; on the other hand, fake news may spread more easily in more democratic countries. 

Once the two subsamples are split (around a mean Polity IV value of eight for our sample), we find that in more democratic countries, both the quality of education at the high school level and the share of the population with tertiary education act as a constraint on the PBC. Conversely, in less democratic countries, we observe that only the share of the population with a university degree is able to constrain policy distortions. A possible explanation is that in those countries, higher education is more difficult to control than secondary-level education. Yet, in none of these subsamples do the proxies for the diffusion of information appear to affect the PBC or to do it more than the countervailing effect of education.

Winston Churchill used to say that “democracy is the worst form of government, except for all those other forms that have been tried from time to time.” As such, it must be protected. Investing in education, especially in the quality of education that schools and universities provide, appears to be the best remedy against the threats that information overload and fake news pose to the proper functioning of electoral processes and democratic governance. Possibly, this requires assigning a more prominent role to the teaching of humanities, as well as accompanying verbal reasoning and literacy skills from an early age, compared to the current emphasis on STEM. Investing in human capital does not only promote the country’s technological and economic progress, but also its political well-being.

Articles represent the opinions of their writers, not necessarily those of the University of Chicago, the Booth School of Business, or its faculty.