Do you want your taxes smooth or spiky?

Cliffs, Discontinuities, Thresholds:

Don Marron illustrates that cliffs make  little sense for soda taxes here or at

Cliffs create problems for the income tax here

For marijuana grows, cliffs’ problems are discussed here or at


There follows a dialog that could be organized to make the point:  Most cliffs are silly.


James Wimberley says:

October 19, 2013 at 8:06 pm

In Britain, excise taxes (true specific or per unit taxes, not sales taxes) on booze follow this logic. Table here. For spirits and strongly fortified wines like liqueurs, it’s a pure tax on alcohol content. For wine, beer and cider, it’s in various bands by alcohol concentration. Presumably this is because it’s much easier to administer. Most wine and beer will fall in two quite wide standard bands (2.8%-7.5% for beer, 5.5%-15% for wine). The exciseman only has to count the bottles, not test them.

A hybrid scheme like this is surely easier to run than a pure THC-based one. That is, for a standard low concentration joint, you tax by weight or packet. For the higher-strength stuff, you tax by THC content, more intrusively. This creates an administrative incentive to market low-strength product.


  • Pat Oglesbysays:October 20, 2013 at 2:15 pmI can’t figure out why bands (they may be called step functions or they may be said to create cliffs) make sense today.The first cliffs I know of – six of them — arose in the first U.S. Federal liquor tax under Hamilton and Washington, the one that sparked the Whiskey Rebellion. 
  • The measuring device mandated by statute was “Dicas’s hydrometer,” which operated by adding weights “only one at a time.”  I don’t follow the technology completely, but a step function seems to make sense as a result in that iterative process:  “then put in the hydrometer . . . which of the weights will sink it to some of the divisions upon the stem.”  We are more accurate today. Maybe bands are easier to administer, but if you can measure something, why not just apply some rate to what you measure?But these bands, step functions, or cliffs persist in the United States as in the United Kingdom. Here are the Federal alcohol tax rates.  We tax liquor by potency, beer by volume, and wine with bands or cliffs. Since the wine rate goes up at 14 percent, we see a good bit of 13.9 percent alcohol content in wine.  Maybe inertia in the law explains these bands.


◦                                  James says:October 20, 2013 at 5:31 pmIt’s not like income where step functions make no sense now, with spreadsheets available to all. Suppose you tax wine and beer by alcohol content, not in bands. Every single vintage of wine – and even a small merchant will carry hundreds – has its own rate; every time a craft brewer brings out a new brew, he has to register it with the excise. No, bands make a lot of sense. Clustering at the top end of the band isn’t a major problem, it improves predictability for the customer. KISS!Reply

▪                                                    Pat Oglesby says:October 21, 2013 at 1:36 amThanks for thinking about this minor matter – we are getting down into the capillaries.I’m all for simplicity. I just think bands (cliffs) are meaningfully simpler now only if the winemaker (or brewer, in the United Kingdom) doesn’t measure alcohol content at all, and just assumes it falls within a particular band for tax purposes. I wonder if that happens much. We see alcohol by volume listed on beer and wine bottles routinely here. Maybe disclosure is mandated.Pre-computer, it certainly used to be meaningfully simpler to lump together (here I rely on imagination rather than knowledge of batch size) a batch of 223 gallons at 12.9 percent alcohol by volume, a batch of 371 gallons at 12.3 percent alcohol by volume, and a batch of 406 gallons at 13.1 percent alcohol by volume – and apply a single rate to the whole 1000 gallons. But these wine bands or cliffs in current U.S. law pre-date the computer age, and seem like vestige of an earlier time.This kind of percentage threshold in the tax law is quite different from, say, the 2 percent floor on itemized deductions here in the United States. If you go over the threshold, each dollar is deductible. There’s no cliff or whipsaw. With the alcohol bands or cliffs, one hundredth of a percent difference in potency (we are in Theoryland here) creates an enormous difference in tax. That makes for silly results, and for game-playing and brinksmanship. That’s just my take. This may be the smallest issue around, but still worth getting to the bottom of — and I’m not sure I have yet.



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: