(Daniel Acker/Bloomberg News)

Taxation is the cornerstone of all modern states. How states collect taxes shows both the quality of their institutions and societal norms. And for scholars, studying taxation provides a unique perspective not only on fiscal systems, but also why people follow norms and rules.

But in my recent research with Clara Volintiru and Sven Steinmo, we find that not everyone is equally likely to follow the tax rules. Across studies in four countries, we found that women were less likely to cheat — even in countries that had different levels of gender equality.

How we did our research

For our study, we invited more than 2,000 students in the four countries — the United States, Italy, the United Kingdom and Sweden — to take part in an experiment. When subjects arrived at a lab, we asked them to sit down and perform a basic clerical task: Copy a line of text from a piece of paper to a spreadsheet on the computer. For each row copied correctly, they received 10 cents. At the end of the experiment, this was converted into real money.

After each clerical task, students were presented with a tax scenario and were asked to report their income based on their earnings from the clerical task. There were three clerical tasks, and, after each clerical task, subjects were confronted with three unique tax scenarios. Therefore, there was a total of three clerical tasks and nine reporting rounds.

The scenarios ranged broadly. In round one, participants paid taxes for no other reason than for the fact that we asked them to report their taxes. In rounds two and three, all of the tax revenue was put in a general fund and redistributed equally between participants, regardless of what they contributed. In round three, we doubled the general fund. For rounds four through six, we adjusted the tax rates, so that in round four it was 10 percent; in round five, 30 percent; and in round six, 50 percent. Finally, in rounds seven through nine, subjects were given two types of progressive taxation and a charity round in which we donated their tax revenue to a real-world charity.

Students could report however much of their income they wanted. Although we didn’t state this explicitly, students, of course, understood that they’d keep more money if they didn’t report all their earnings, just as is true with actual taxpayers.

There was a catch, however. We told them there was a 5 percent probability of being audited. If we caught them cheating, they would have to pay a fine equal to twice the taxes owed. Of course, with such a small chance of being audited and a rather small fine, from a rational choice perspective, the best choice would always be to report no income. (We’ve used this approach for other experiments, as you can see in a previous Monkey Cage post). We used the students’ choices about how much to report to examine whether some people are more willing to comply with tax law than others.

Women were more honest about their income, in every country

As you can see in the graph below, in every country, women were on average more honest about their income than were men.

Does this hold up beyond students? We think so. We found similar results when we conducted a study with adults in the United Kingdom, which lends weight to the idea that our study may be valid beyond this sample. In the adult sample, we had 90 individuals with an average age of 35, compared with an average age of 24 in our student population. Further, Jim Alm and colleagues do not uncover any significant differences between student and non-student populations.

In our study in the United States, men on average reported 60 percent of their income, while women reported 73 percent, for a gender gap of 13 percent. The gender gap was even larger elsewhere — 19 percentage points in Italy, 22 points in the United Kingdom and 25 points in Sweden.

Curiously, the gap was largest in Sweden, which has among the highest levels of gender equality in the world, and the smallest in the United States, which ranks considerably lower.

 

We had assumed that we would find that men and women would behave more similarly in countries with more gender equality. Our results don’t seem to point to such a pattern.

Egalitarian Swedish institutions might have been expected to foster similar behaviors in men and women, but they haven’t. In addition to our tax compliance experiment, we conducted another experiment with the same students which examined levels of altruism. In this experiment, men and women in Sweden behaved quite similarly, meaning that both men and women were equally likely to give part, or all, of their money to a randomly assigned partner. That wasn’t true in Italy, the United States and the United Kingdom. We believe that the very redistributionist nature of Swedish institutions encourages such altruism.

But apparently men and women see taxes quite differently.

What does the difference mean?

Like many other political scientists and sociologists, we have believed that behavioral change results from institutional change. Our results challenge that assumption. Apparently norms are sticky and have long-lasting effects on behavior. Normative expectations and perceptions of how women ought to behave appear to have real behavioral consequences.

We believe that this is why what political scientist Jessica Preece calls “gendered psyches” continue to have strong effects on behavior — in this case, honesty, that may not respond to recently effected institutional changes.

We hope other scholars will conduct studies that draw on our data to further our understanding of these behavioral differences. The gender gap in tax compliance is complicated and important. Perhaps other scholars can help explain why institutionalized gender equality can be so different from the realities on the ground.

John W. D’Attoma is a postdoctoral research fellow at the European University Institute studying tax compliance, and, most recently, the emergence of trust in the sharing economy. Find him on Twitter @john_dattoma. This article draws on the “Willing to Pay” project funded by the European Research Council.