What if we had a simpler tax code? What if the amount you owe in taxes was computed by determining your overall income, and then taking a percentage of that income, where the percentage itself depended on income? Apart from a few flat-taxers who carry the idea too far and think this percentage should be a fixed number at all incomes, most people believe that income tax should be progressive, taxing higher income levels at a higher percentage. Thanks to the Statistics of Income posted by the IRS, we can look back at the taxes collected, imagine we have been operating under such a tax code, and then determine what that percentage curve would be.
This hypothetical tax code only reflects averages, and would have some people pay less and some more than they do today. Let’s stipulate up front that Adjusted Gross Income isn’t a perfect indicator of actual income, but it’s an indicator nonetheless. It’s a good bet that someone with $100,000 in AGI has more money than someone with $30,000 in AGI.
So what do the historical data tell us? The above graph shows the average income tax percentage paid by married couples at different income levels over a 20-year period ending in 2015, the most recent year for which data has been published at https://www.irs.gov/pub/irs-soi/. Dollar amounts in the graph have been adjusted by the CPI to compensate for inflation.
A major feature of the graph that stands out is the fact that it peaks somewhere around an income level of about a million dollars per year, and then begins to drop off again. Let’s call this “the Romney effect”, where the lower capital gains rate and other tricks allow those with the highest incomes to pay lower percentages of that income in taxes. So if there’s anything that needs fixing in our imaginary simplified tax code, it’s that right there. If we wanted to make the tax code more fair, we could start by leveling off that curve instead of allowing it to go downhill as income climbs.
The next interesting thing to observe about these curves is how they have changed over time. The colour groupings in the graph show periods with similar tax patterns. And we can correlate those historical groupings with actual tax policy changes, as follows:
In dark blue, we have the Clinton era of the late 1990’s. These were the overall highest effective tax collection rates in the time period we’re considering. The available data from the earliest years doesn’t extend as high along the income spectrum, except for the last year of that period, which reveals the same downward curve shown later years. It is probably safe to assume that in the other Clinton years we would see that same trend if we had the data separated as it is in the more recent spreadsheets from the IRS.
The orange curves are a transitional period between the Clinton years and the Bush years. They resemble the Clinton rates, but begin to drop downward at some income levels.
The red curves show the “Bush tax cuts” in full effect, starting in 2003. The effective tax rate percentages are lower at all income levels. That of course translates into vastly more dollars, the higher the income level goes.
The light blue curves show an interesting phenomenon — a two-year drop in tax rates that is most pronounced for incomes under $100,000. This this the tangible effect of the “Obama tax cuts”, the stimulus that reduced taxes for middle-class earners in 2009 and 2010. After 2010, the trend returns to where it was before the stimulus.
And then we come to the yellow curves, the three most recent years that we currently have data for. These reflect the “fiscal cliff” legislation of 2012, which increased the top tax rates. These curves look like the Bush tax code for lower incomes, and like the Clinton tax code for higher incomes. And that seems about right — this is the most steeply progressive tax code we have had in the period from 1996 to 2015. It gives the middle class a break, and collects more of the government’s revenue from those who can spare more. It’s still defective — it still curves downward on the right edge, but that bug could be fixed.
And that’s where we stand today. This is the tax policy we’re leaving behind, the one the Republican congress decided needed fixing. They didn’t simplify it by eliminating loopholes and complexity and applying the above curves across the board for all income. They changed the rules to help out higher incomes the most, borrowing money from our future revenues. And just as every previous major change in tax policy is vividly revealed in the SOI data, we can assume the new tax code will show up too, starting in 2018. Then we’ll really know, with real empirical data, who’s paying what under this new tax code.
If we wanted a simpler tax code that wasn’t redistributive, providing different gains or losses at different income levels, we would expect that the curve wouldn’t move between the old and new policy — it should collect the same amount of tax from each income band, just doing it with a simpler formula. It’s easy enough to compute the above curves. If we wanted a simple tax code, we could start right there, treating all income the same and taxing it at the same rates as our imaginary empirical tax code of 2015.
