Why has the share of income going to those at the very top of the U.S. income ladder grown so rapidly over the past 35 years while that same share in France remained relatively flat? Understanding the difference in income trends across countries at the top of the income spectrum has been a challenge for economists and other researchers for a while now. A new National Bureau of Economic Research working paper released yesterday offers one explanation: the difference in payoffs for engaging in entrepreneurship.

Charles I. Jones of Stanford University and Jihee Kim of the Korea Advanced Institute of Science and Technology start their investigation into top incomes by first recognizing that the distribution at the top is a so-called Pareto distribution. This means there’s a reoccurring level of inequality within each top income share. The top 1 percent, for example, have about 40 percent of the income going to the top 10 percent, and the top 0.1 percent have about 40 percent of the income going to the top 1 percent, and so-on.

What can explain a shift in the distribution that concentrates even more income at the top? Jones and Kim quickly dismiss skill-biased technology change, which they find can only explain a shift of the income curve over to the right (a larger difference between the top and bottom) but not the steepening of the curve itself over the past (the rapid increase at the top). What they instead believe explains the sharp rise is a change in the rewards for entrepreneurship, based on a new model they present in the paper.

Let’s start with the quite broad definition of entrepreneurship that Jones and Kim use in their model. A programmer who starts a business in Silicon Valley counts as an entrepreneur, as does a musical artist who writes a hit song and a middle manager at a firm who gets promoted after coming up with a new management process.

The two authors then factor in several developments that can increase the share of income going to the top of the income distribution. One is technological change—such as the invention of the Internet—that can allow for greater business opportunities and thus greater income gains for entrepreneurs. Another is a reduction in red tape. Yet intriguingly, these two factors don’t affect long-term economic growth but do increase top end inequality in their model.

What they do find are two factors that can affect both inequality and growth. An increase in the share of the population that works in research and development boosts economic growth, according to their model, as does a reduction in the ability of already existing firms to block innovation. They find that both of these factors result in more “creative destruction,” reducing the share of income going to those at the top as more entrepreneurs can enter the market.

So what we can take away from the paper? Well, the rise in top end inequality within a country might be the result of positive developments, such as the Internet, or negative developments, such as monopolies or oligopolies crushing new innovations. Factors that increase top end income inequality can be good or bad for economic growth. Just another reminder that looking at the specifics is always important.