Tariffs Cause (Price-)Inflation

28 July 2025

Quite a few economists who argue in favor of free international trade none-the-less claim that tariffs do not cause inflation.


A small share of these economist argue quite terribly, by treating one of the original definitions of inflation as if it is the only permissible definition.

Under those original definitions, inflation referred, depending upon just who was using the term, to increases in the stock of money, to such increases effected by debasement or by the issuance of debt instruments, or to increases in the stock of money not matched by general increases in supplies of commodities.

The general public was so used to the idea that increases in the money stock would lead to general increases in prices that inflation came to be treated as if meaning a general increase in prices. An interesting study might be written on how this secondary, operational definition came to be primary; certainly, removing what had been the primary definition from popular consciousness made it easier for policy-makers to ignore the ideas that increases in the money stock would lead to general increases in prices and that general increases in prices were likely to have been caused by increases in the money stock. But, as unfortunate as this change in definition may have been, the fact remains that, not only in popular discourse but in the discourse of economists, the most common use of inflation now is in reference to a general increase in prices.

If the idea of a general increase in prices were itself incoherent, then one might still have a basis for rejecting attempts to use inflation thus. But, as things stand, attempts to argue against the conceptual content of claims about price-inflation by insisting that inflation can only mean some sort of increase in the money stock, even when the term has been used by one's opponent, are abusive.

(Mind you that I advocate avoiding use of the bald term inflation to refer to general increases in prices.)


The larger share of economists tolerate or accept the use of inflation to mean price-inflation, but none-the-less argue that price-inflation is not amongst the ills caused by tariffs.

These economists quite rightly distinguish an increase in the price of one good from a general increase in prices, and likewise increases in the prices of many but not most goods from a general increase in prices; but these economist do not attend properly to the elasticity of the commodities upon which tariffs are placed, and especially to the cross-price elasticities of other commodities. It is as if these economists imagine money to be reallocated with no over-all change in the quantities of commodities purchased.

If the price of some commodity is increased, we must infer that people make one of three choices,

  1. buy the same quantity of the commodity, but buy less of other commodities, or
  2. expend just as much money on the commodity, but buy less of it, or
  3. shift some of their expenditures to buying other resources, buying less or none of the commodity the price of which increased.

Economically rational agents — and we must suppose real-world people to behave in a manner that is approximated by economic rationality — exchange their resources, including money, for the ends most useful to them. Thus,

  1. if they buy less of other commodities, they lose the marginal uses of those other commodities;
  2. if they buy less of the original commodity, they lose the original marginal uses of that commodity; and
  3. if they move some or all of their money to purchases of other commodities, these must be commodities that provided less valuable marginal uses (at the new level of expenditure) than did the original commodity (at the prior level of expenditure).

No matter which the sort of the choice, usefulness is lost as a consequence of the price increase.

To the extent that the commodities are factors of production — and, ultimately, all commodities are factors of production, though some quite neglibily — productivity is reduced. If we may adopt the metaphor in which multipliers have meaning, then we may say that, ceteris paribus, an increase in the cost of any resource has a multiplier of less than 1. If a resource has significant application across many productive processes, or significant application to some productive process an output of which in turn has significant application across many productive processes, then the loss of productivity will have wide-spread significance.

Recall, now, that M · ν = Σ (pi · qi) where M is the quantity of money, ν is the average rate at which the monetary unit is exchanged in trade, pi is the price of the i-th commodity, and qi is the time-rate at which quantities of the i-th commodity are exchanged. A general loss of productivity results in a general decrease in quantities supplied of the various commodities. Unless the money supply or velocity are correspondingly decreased, prices must generally increase. No mechanism for decreasing the money supply is typically entailed in the increase of the price of a commodity. The technologic aspect of moving money will become more costly, which has a slowing effect on the average rate at which the monetary unit is exchanged, but certainly not usually so as to fully offset the general loss of commodities. We should expect a general increase in prices.

Tariffs, then, cause a price-inflation, by way of their effects on production. The inflationary effect of a tariff on some commodities will be so negligible as to be indetectable; the effect of tariffs on other commodities may be profound.

A Prediction of Violence

4 October 2024

I believe that violence will erupt amongst those tasked with counting the votes and with overseeing that counting, for the upcoming General Election. More specifically, I expect that Democrats will attempt forceably to exclude Republicans in these tasks, that Republicans will respond with counter-force, and that the mainstream of the media will present the fighting as caused by the Republicans.

I began predicting such behavior weeks ago, but I have not previously posted the prediction here.

Addendum (2024:11/16): My prediction here has of course been thoroughly falsified. I believe now that violence did not erupt because it was made pointless for Democrats by Republican monitoring of the collection of ballots and by the margin of Republican victory.

More Refactoring

4 September 2024

The axiom of Generalized Decomposition from Formal Qualitative Probability may be refactored to [image of mathematic formula] This refactoring is mathematically trivial, exploiting two automorphisms, but exhibits the principle more elegantly.

Refactoring

23 August 2024

The axiom of Disjunctive Presumption in Formal Qualitative Probability [image of mathematic formula] may be more simply stated as [image of mathematic formula]

The Significance of Underlying Variance for Social Outcomes

8 August 2024

Measures — quantities to which some arithmetic can be meaningfully applied — can be fitted to some human attributes, even if not to others. When attempting to compare some populations to others, an assumption is made that the properties of the individuals within these populations are subject to quantification of some sort, and that the quantities are commensurable across populations. But usually these assumptions are implicit and unrecognized, and even those who have some awareness that they are dealing with quantities very often don't have a proper grasp of elementary issues.

Very often, people try to understand distinct populations in terms of some notion of averages. If each and every member of every population were exactly average in every regard, then averages would be perfect measures of the populations as such. More generally, if for any two populations the share of that population deviating from average by some specific amount were the same, averages would be sufficient for any comparison of the attributes of populations, except for population sizes. But if the overall variance from average in one population is different from that in another, then thinking in terms of averages can go very, very wrong.

Here are hypothetic distributions for some attribute within two populations, each population having the same number of members:[1] [two lognormal distributions of equal median but of different variance] For both Population A and Population B, the median[2] of the attribute is the same, but Population B has more variance from the arithmetic mean than does Population A. Even though each of these two populations have the same median, more members of the population of greater variance are below some some measure, and more members of that same population are above some measure. [two lognormal distributions of equal median but of different variance] Population B2 [two lognormal distributions of equal mean but of different variance] has the same variance as Population B, but the same arithmetic mean (rather than median) as Population A. Again, even though the center is, by some measure, the same for both populations, more members of the population of greater variance are below some some measure, and more members of that same population are above some measure.

Even if a population has a higher center than Population A, if it has a greater variance then it will dominate the lower range of measures below some measure. [two lognormal distributions of different mean and variance] And even if a population has a lower center than Population A, if it has a greater variance then it will dominate the higher range of measures beyond some measure. [two lognormal distributions of different mean and variance]

If a measurable attribute correlates positively with social success, then ceteris paribus, a population of higher variance is going to dominate both social winners beyond some level and social losers below some level. If a population generally has greater variance amongst its attributes, then — discarding the assumption of ceteris paribus — that population is going to dominate both social winners beyond some level and social losers below some level; even if it has the same median or same mean or even a lower center than another populations.

In fact, though I cannot readily graph the cases in which attributes are only partially ordered and not measurable, the reader should see that the underlying point does not depend upon the measurability of the attributes, but only upon one population having greater propensity for variance than another.

But

  • If one is only looking at the losers and thoughtlessly assuming that their numbers are explained by averages, then one is going inappropriately to infer that the population is generically inferior.
  • If one is only looking at the losers, while thoughtlessly assuming that the averages are the same and that nothing else about the population itself can explain the difference in outcomes, then one is going inappropriately to infer that the population are victims of systemic bias.
  • If one is only looking at the winners and thoughtlessly assuming that their numbers are explained by averages, then one is going inappropriately to infer that the population is generically superior.
  • If one is only looking at the winners, while thoughtlessly assuming that the averages are the same and that nothing else about the population itself can explain the difference in outcomes, then one is going inappropriately to infer that any rival population are victims of systemic bias.

In each case, if we look at the other end of the distribution, the thoughtless conclusion falls apart.

When the last of these errors is made, an attempt may be undertaken to offset illusory bias, by putting an institutional thumb on the scales to shift Population A generally forward, until the the number of social winners at every level is at least the same. But notice what is then really happening as the relative outcomes for most members of the population of greater variance fall increasingly below those of the population of less variance — at previously targetted levels the population of lower variance comes to enjoy greater social success than does the population of greater variance. And notice that the population of greater variance necessarily still dominates above some value, albeït that the value increases as the institutional thumb comes down ever harder in a misguided attempt to match the upper tails of the distribution.

Only actual systemic bias can bring the number of social winners across populations into equality above any given level of social success beyond the center; and, the larger the population, the greater the required bias for such an outcome, and the more that most of the population of greater variance are victimized.

If the two populations are not equal in size, then the foregoing analysis would simply need to entail talk of proportionality. But one might as well speak and write of two populations of the same size, because the real-world application involves two populations that are very close to the same size in most first-world nations. The greater variance of one of those two populations is a consequence of the greater chaos in the formation of one of that population's chromosomes and of a lack of redunancy for another.

Unfortunately, most of the attempts to analyze what has been happening has entailed ham-fisted theorizing about differing averages.


[1] A population with finite membership cannot perfectly conform to a continuous distribution function; but, the larger the population, the less the necessary non-conformance.

[2] The median for each population is the point such that as many members of the population are above it as are below it.

On Distributions of Measurable Human Attributes (A Prologue)

8 July 2024

Often, when talking about the distribution of measurable human attributes, people refer to the bell curve, which is to say to a Gaussian distribution, more commonly known as a normal distribution.

One immediate difficulty is that the Gaussian distribution extends symmetrically without lower limit to measurements with positive probability, whereas the natural measures of most of the attributes that will interest us have lower limits of possibility (typically at or above zero). For example, no one has negative weight or negative height. Simply truncating the lower bound of a Gaussian distribution usually doesn't make a great deal of sense, because few people will even be near the lower bound, rather than a fair number at it or just barely above it.

Instead, the distribution will more typically look something like this:

Mind you that measures can always be transformed, and a measure that has a lower bound of b can be transformed into a measure without lower bound simply by the device of subtracting the bound and then taking a logarithm: measure1(x) = loge[measure0(x) - b] Some set of transformations can surely be used to arrive at one with a distribution that is well approximated by a Gaussian distribution. But, for the most part, I'd rather use natural or familiar measures than manipulate the data to arrive at a Gaussian distribution, especially as one otherwise typically needs to invert the transformations at the end of the analysis, to make sense of things.

In the near future, I plan to post an entry about misreading the consequences of different variances in different human populations. What I have to say could all be expressed in terms of Gaussian distributions, but I don't want to do so, nor did I want that future entry to begin with a discussion such as that here.

It's a Bit Late

16 June 2024

The immoderate political left began speaking and writing of late[-stage] capitalism with the end of the First World War.

The idea has been that the industrialized world has entered the final stage of something called capitalism, with a revolutionary change to some form of socialism just around the corner

…for more than a century now.

(Actually what we've seen is a slow, grinding transformation away from the use of genuine markets to an administrated order, which in more recent years has threatened to become a neo-feudalism. The transformation began well before the First World War, as technocratic thinking began displacing liberalism.)

Certainly Unprofessional

15 June 2024

To-day, I updated some software on a Samsung electronic tablet. The latest version comes with an ability to suggest changes on-the-fly to what I've written. These changes include to wording that Samsung calls Professional, to what they call Casual, or to what they call Social. What they call Social seems to be what they call Casual, with the addition of hashtags. What they call Professional certainly isn't very professional.

The software may already have been patched to address this specific error, but I'm sure that many other errors remain, and they will propagate. For decades now, a great many people, even people who imagine themselves as native speakers of the English language, have been uncritically accepting what software has told them about the language.

I will sometimes use software to check my spelling, but I don't simply make every change that it suggests. I find that, in analyzing my writing, grammar-checkers so often flag constructions that are actually correct and so seldom find real errors that these checkers are not worth running. And a person who relies upon software for tone is falsifying his or her relationships.

A Minor Up-Date

14 June 2024

I still await a first decision from the journal to which on 24 February I submitted my paper on Sraffa. On 27 May, I queried the editor about when I might expect a decision. My query was really to ensure that the paper weren't mislaid; I've more than one unfortunate experience of such a thing. On 29 May, I received a reasonable and polite response that he hoped to have a report from the reviewer within a month.


The mission statement of the journal from which I yanked that paper on 21 February declares

Our goal is to provide a definitive answer within one month of submission.

I yanked the paper, submitted on 23 January, after it had idled for four weeks without being sent to the editor-in-chief. On 28 March they none-the-less reported it as placed in the hands of the editor-in-chief.

When I'd yanked the paper, when I received this notice, and when I received an apology and plea for patience, I sent copies of my replies (declaring and reïterating that the paper were no longer on-offer to them) to that editor-in-chief. None-the-less, on 07 June — more than four months after my submission — the editor sent a rejection on the grounds that the work were primarily exegetic. I think that the behavior of the editor and of the administrative staff betrays a gross disorganization that makes a mockery of the supposèd goal of reaching a decision within a month.

There Is No Pie

26 May 2024

Imagining all of a society's various generation and allocation of goods and services as if the creation and distribution of one big pie is very much analogous to imagining all of the sexual interactions of that society as one enormous orgy.