Search This Blog

Wednesday, September 29, 2004

Time to Dismount Your High Horses

Update: I've thought about this some more and worked with the numbers and have concluded that there are minor errors in my methodology. I don't think those errors will have a substantial effect on the results, but at this point, I'm not sure. I'll post a new entry with an update when I figure it out. The main problem is that I assumed an intercept at (0,0) for Adjusted Federal Expenditures versus GSP per capita which may not be correct. I guess I should get off my high horse too!

Update 2: I definitely need to get off my high horse. The data does show the effects of progressive taxation. Thus the rest of this post below is incorrect and should be ignored.

Andrew Sullivan links to a post on TaxProf Blog titled "Red States Feed at Federal Trough, Blue States Supply the Feed." It shows two tables which follow:

States Receiving Most in Federal Spending Per Dollar of Federal Taxes Paid:

1. D.C. ($6.17)
2. North Dakota ($2.03)
3. New Mexico ($1.89)
4. Mississippi ($1.84)
5. Alaska ($1.82)
6. West Virginia ($1.74)
7. Montana ($1.64)
8. Alabama ($1.61)
9. South Dakota ($1.59)
10. Arkansas ($1.53)

States Receiving Least in Federal Spending Per Dollar of Federal Taxes Paid:

1. New Jersey ($0.62)
2. Connecticut ($0.64)
3. New Hampshire ($0.68)
4. Nevada ($0.73)
5. Illinois ($0.77)
6. Minnesota ($0.77)
7. Colorado ($0.79)
8. Massachusetts ($0.79)
9. California ($0.81)
10. New York ($0.81)
The States in bold letters in the top table tend to vote Republican and the States in bold letters in the bottom table tend to vote for Democrats. The general idea is that the top ten states in terms of federal spending per federal tax dollar are being subsidized by the bottom 10 states.

There are many possible implications, almost all of them negative towards middle America and Republicans. Indeed, Andrew Sullivan writes "A new study reveals just how welfarist much of Middle America is" regarding TaxProf's post. Sullivan's post with this derogatory comment can be found in the entry "Bush's Conservatism".

But before we get carried away about welfare queens in middle America and evil Republicans screwing states that vote Democrat, we should consider if maybe there is a perfectly straightforward explanation for the data. It turns out that there is.

First, let's imagine what a perfectly fair system of federal expenditures and taxation might look like. In such a system, each state would receive identical per capita federal expenditures, including government contracts, entitlements, etc. In other words, if a state had twice as many people as another, it would get twice as much federal spending as the other.

On the tax side, it's a little tricker. In some sense, for perfect "fairness", if a state has twice as many residents, those residents should pay twice as much federal tax. Under this scenario, every man, woman, and child would pay about $5,000 in taxes per year, regardless of income or wealth. However, this would require that the poor (and children) pay a far, far higher percentage of their income and wealth in taxes, in some cases greater than 100%. This is extremely regressive and very few people would consider that to be fair.

A less regressive structure would be a flat tax and/or consumption tax. Then everybody pays the same percent of what they make or consume. In this case, the total federal income tax paid by the residents and businesses within that state would be very close to a constant percentage of the State's total income (or of a closely related measure called Gross State Product). That's still much more regressive than our current tax system. Nonetheless, it seems quite fair for each state to pay an equal percentage of it's Gross State Product.

In the following table, I've shown by State, the Gross State Product Per Capita and Adjusted Federal Expenditures Per Dollar of Taxes. After the table comes my analysis.
State 2001 Gross State Product ($ millions) 2000 Population Gross State Product Per Capita ($) Adjusted Federal Expenditures Per Dollar of Taxes by State
Alabama 121490 4447100 27319 1.61
Alaska 28581 626932 45589 1.82
Arizona 160687 5130632 31319 1.20
Arkansas 67913 2673400 25403 1.53
California 1359265 33871648 40130 0.81
Colorado 173772 4301261 40400 0.79
Connecticut 166165 3405565 48792 0.64
Delaware 40509 783600 51696 0.85
Florida 491488 15982378 30752 1.00
Georgia 299874 8186453 36631 1.01
Hawaii 43710 1211537 36078 1.52
Idaho 36905 1293953 28521 1.34
Illinois 475541 12419293 38291 0.77
Indiana 189919 6080485 31234 0.99
Iowa 90942 2926324 31077 1.22
Kansas 87196 2688418 32434 1.14
Kentucky 120266 4041769 29756 1.46
Louisiana 148697 4468976 33273 1.44
Maine 37449 1274923 29374 1.31
Maryland 195007 5296486 36818 1.20
Massachusetts 287802 6349097 45330 0.79
Michigan 320470 9938444 32245 0.90
Minnesota 188050 4919479 38226 0.77
Mississippi 67125 2844658 23597 1.84
Missouri 181493 5595211 32437 1.32
Montana 22635 902195 25089 1.64
Nebraska 56967 1711263 33289 1.19
Nevada 79220 1998257 39645 0.73
New Hampshire 47183 1235786 38181 0.68
New Jersey 365388 8414350 43424 0.62
New Mexico 55426 1819046 30470 1.89
New York 826488 18976457 43553 0.81
North Carolina 275615 8049313 34241 1.07
North Dakota 19005 642200 29594 2.03
Ohio 373708 11353140 32917 1.02
Oklahoma 93855 3450654 27199 1.47
Oregon 120055 3421399 35089 1.00
Pennsylvania 408373 12281054 33252 1.08
Rhode Island 36939 1048319 35236 1.06
South Carolina 115204 4012012 28715 1.32
South Dakota 24251 754844 32127 1.59
Tennessee 182515 5689283 32080 1.24
Texas 763874 20851820 36633 0.92
Utah 70409 2233169 31529 1.14
Vermont 19149 608827 31452 1.12
Virginia 273070 7078515 38577 1.47
Washington 222950 5894121 37826 0.91
West Virginia 42368 1808344 23429 1.74
Wisconsin 177354 5363675 33066 0.87
Wyoming 20418 493782 41350 1.05

Assuming you agree that fairness would require each State get the same per capita federal expenditure and each State should pay an equal portion of it's Gross State Product in federal taxes, then we might expect some sort of relationship in the data between Gross State Product per Capita and Adjusted Federal Expenditures. In particular, we might expect poor States with low GDP to pay a lower per capita amount in taxes (even though taxed at the same flat rate), yet receive the same federal expenditures per Capita. This would cause the Adjusted Federal Expenditures measure higher for such states. Indeed, this is exactly what happens. The following table shows the correlation between the two last columns in the previous table:

Correlation Between Gross State Product per Capita and Adjusted Federal Expenditures

Correlation Coefficient Correlation Coefficient Squared
All States -0.62 0.38
Excluding Outliers (Alaska, Maryland, Virginia) -0.76 0.58

The square of the correlation coefficient corresponds to the fraction of the variation of one column that is explained by (or correlates with) the other column. The sign of the correlation coefficient (not squared) tells us if it us a positive or negative relationship.

Sure enough, excluding outliers, 58% of the differences of Adjusted Federal Expenditures can be explained by the differences in per capita income between the States. I've excluded Virginia and Maryland because they border D.C., and do get more federal expenditures than would be expected using the Gross State Product model. I've excluded Alaska because it is an extreme outlier and it has low population and Gross State Product. To leave it in would allow it to substantially skew the results even though it represents a minscule amount of federal expentitures and taxation. That same argument could be made to exclude other small population states such as North Dakota and Hawaii, which would also, it turns out, increase the correlation. But I left the rest of the small States in.

Gross State Product (column 2) comes from the Bureau of Economic Analysis, State Population comes from the census bureau, and Adjusted Federal Expenditures comes from The Tax Foundation (this is what TaxProf linked to). Unfortunately, the data for the various columns are for slightly different years. However, all of this data varies slowly so it is unlikely to have materially changed the result.

So if 58% of the differences in Adjusted Federal Expenditures is explained by Gross State Product per Capita, what about the other 42%? Well, The Tax Foundation lists other possibilities and I'll comment on each.
One factor affecting rankings is that federal spending on defense and other procurement dollars are often funneled to the states of powerful members of congress. Also, state governments can grab more federal grant money by manipulating their spending to comply with federal regulations.
Well, maybe so, but they don't happen to mention which states those might be so I have no way of checking. Funny that the list these first when clearly they are a minor effect at best.
Another factor is demography. States with more residents on Social Security, Medicare and other federal entitlements tend to rank high.
Once again, maybe, but if you look at the data, states you might think likely candidates for this effect such as Florida and Arizona, don't seem to be out of line.
Similarly, high spending levels in Virginia, Maryland and the District of Columbia are explained by the predominance of federal employees.
This one is clear and that is why I eliminated Maryland and Virginia from the mix.
Finally, states with higher incomes per capita%G—%@such as Connecticut%G—%@pay higher federal taxes per capita thanks to the income tax's progressive structure, which increases federal taxes per dollar of federal spending received in return.
But as I've shown, even if the taxes were flat instead of progressive, most of the differences are explained anyway. It's not clear to me that the "progressive" portion of the taxation makes much difference at all.

The Tax Foundation doesn't mention randomness. Sheer chance is going to introduce some differences in Adjusted Federal Expenditures. In other words, the chance that every state has identical Federal Income per capita is zero. In fact, chance could possibly explain all of the rest of the differences.

Anyway, for those on their high horses who see vast republican conspiracies to bleed the blue states for the benefit of the welfarist red states, I think it's time to dismount.

No comments: