Recent Posts

Wednesday, August 12, 2015

Inflation And Debt Burdens

Nathan Tankus recently published an article entitled "No, Inflation Doesn’t Erode the Burden of Debts" (h/t Tom Hickey at Mike Norman Economics). Nathan correctly diagnoses that mainstream economists (he cites Paul Krugman) routinely jump between the concepts of "inflation" and "CPI inflation" (Consumer Price Index), and that a rise in the level of the CPI does not help out borrowers. What matters is wage inflation. I have made the same complaint myself. However, the mainstream logic is not exactly incorrect, it is just based on an embedded assumption about the long term. This seems to be done for reasons of politics, or more accurately, attempting to hide politics from economics.

Nathan Tankus writes:
The source of the confusion here is that it is popular do divide key economic variables by abstract measures of the prices across the economy (such as the Consumer Price Index). This makes sense sometimes but in other situations become nonsensical. Debt is one such example. In the United States most people have debts denominated in U.S. Dollars. A useful proxy for the burden of debt is the ratio between an individual’s or sector’s nominal debt to its nominal income.
The burden of this debt falls when they can refinance at a lower nominal interest rate, their nominal income rises or they default. 
In other words, a household is not helped in paying its mortgage if the price it pays for groceries rises. But it will be easier to make mortgage payments if the household's wage increases. (For governments, taxes are levied on nominal activity, and so government revenue will typically rise even if just consumer prices rise. For government finance, what usually matters is nominal GDP growth, which is enhanced by a rising GDP deflator.)

He also predicts the response that I give.
Some will object that Krugman and many others are simply using increases in the price level as a proxy for increasing nominal income. The problem with this is that they never connect their arguments about inflation to income levels. 
Yes, economists who argue that "inflation" reduces the burden of debt are using consumer price inflation as a proxy for "generalised inflation" (which I describe below). This is technically incorrect. But as the chart below shows, there's a fairly strong correlation between CPI inflation and wage inflation, for very good reasons. A divergence between the two has implications for the wage and profit shares of national income.

Chart: Wage And Consumer Price Inflation Trends (

Why do many economists interchange "generalised inflation" and consumer price inflation? My guess is that they have inherited this quirk from central banks. Central bankers do not wish to say that they do not want wages to rise, rather they focus on the politically popular fight against rising consumer prices. Since central banks drive the research agenda in macroeconomics, their political obsessions are picked up by academic economists. (Additionally, many mainstream economists would argue that the two measures cannot diverge, since prices and wages are linked by marginal productivity. This gives a theoretical basis for treating the two concepts as being interchangeable. As some form of post-Keynesian, I question such marginal analysis.)

UPDATE: Nick Edmonds commented:
Any statement in economics that X causes Y carries an implicit ceteris paribus assumption. The question then is what exactly are we assuming is constant. In this case, are we assuming nominal wages or real wages are unchanged? I think most economists would take it as read, when talking about inflation (even CPI inflation) that it is real wages that we are assuming are constant. Tankus obviously reads it the other way.
This is probably the best explanation of why this shorthand is used. I did not think of it when I wrote this article, since I (like Tankus) would not assume that real wages are constant when the level of consumer prices changes. Wages are generally not indexed any more, and so nominal wages no longer follow movements in the CPI.

In summary, Nathan Tankus is correct that mainstream economists are sloppy when they interchange "consumer price inflation" and "generalised inflation." That said, this shorthand is broadly correct in the medium term. Although I am critical of mainstream economics, I see little value in quibbling over a point in which I basically agree with. The rest of the article gives an explanation of why the two notions of inflation are linked.

Wage Inflation Vs. Consumer Price Inflation Vs. Generalised Inflation

In order to explain the concept of "generalised inflation," I will revert to discussing a highly simplified two sector economy -- one with just a business sector and the household sector -- which is further subdivided into workers and firm owners. (If you wish to avoid class conflict overtones, feel free to pretend that the workers and firm owners are the same households.)

Firstly, we assume the economy is in a steady state. There are 90 workers, and they produce 100 units of goods every day. The daily wage is $1, and the units of goods are sold for $1. The total wage bill is $90, and we assume that workers spend all of their income. This means that they purchase 90 units of goods. The remaining 10 units are purchased by firm owners. The total revenue for the business sector is $100, which means that profits are $10. This $10 dollar profit is paid out to capitalists, which allows them to finance their purchases.

What happens if prices changes (if we assume the economy enters a new steady state)? I look at three generic cases.
  • Consumer Price Inflation.  If the price per unit goes to $1.10 (10% consumer price inflation), the workers will only be able to purchase 81.8 units (roughly) of the 100 units produced. However, if the firm owners step up their purchases to 18.2 units (which costs $20), the full 100 units are sold. The firm will have total profits of $20, which allows it to pay a dividend that allows business owners to finance their purchases. All that has happened is that there has been a shift in the distribution of output towards business owners.
  • Wage inflation. If wages rise by 10% to $1.10, but good prices are fixed at $1, the total wage bill would be $99. Workers would buy 99 units of production, and business owners would only be left with a $1 dividend, allowing them to buy 1 unit. In this case, the income share has shifted towards labour.
  • Generalised Inflation. If both wages and goods prices rise by 10%, all that happens is that all dollar amounts in the original case are multiplied by 1.1, while the distribution of units (which is a real quantity) is unchanged. (In other words, workers buy 90 units for $99, and the capitalists buy 10 units for $11.) Any nominal debts would be effectively shrunk relative to the rising price level.
In other words, the differential between consumer price and wage inflation just tells us about the distribution of incomes within the economy. Since mainstream macro is highly allergic to the concept of income distributions for political reasons, this is obscured in conventional textbook descriptions of the inflation process.

Complications like foreign trade only slightly modifies the picture. For example, rising import prices just means that real incomes for foreign entities (mainly multinational corporations) is increasing at the expense of domestic real incomes.

If we start looking at real-world data instead of simplified models, we need to start adjusting the logic for worker productivity. If the amount of output per worker rises over time, wages would need to rise by that amount in order to keep the profit share of income stable. For example, this would allow the consumer price level to be constant while wages rise.

The rise in the profit share of national income over recent decades demonstrates that wage and consumer prices can diverge. But there are limits to the process.
  • If wages rise too much relative to consumer prices, profits fall, and the corporate managements will do their best to correct the situation by widespread firing of workers. In any event, profit margins are typically not that large as a percentage of total revenue, so there is limited room for this process to run before businesses run into insolvency problems.
  • If consumer prices rise faster than wages, the increased profits are generally not plowed back into spending on goods and services. For example, if dividends (or stock buybacks) are increased, the high income households that are the recipients of the increased nominal income will not spend 100% of the increase. If the rising prices are going to foreign producers, the increased proceeds are typically invested into foreign currency reserves. This creates a revenue shortfall for the business sector; in my simplified model, not all production would be sold. The increase in inventories would eventually lead to a recession. (Broadly speaking, this is what is happening in real world economies in response to the increasing profit share of national income.) 
In summary, it is technically incorrect that consumer price inflation translates exactly into generalised inflation. But at the same time, the two measures are going to be relatively close to each other in the medium term, so people can get away with blurring the two concepts.

See Also:

(c) Brian Romanchuk 2015


  1. Higher (lower) wages do not necessarily result in materially lower (higher) profits. For example, higher wages increase business costs. However, if these higher wages are all spent, they will also increase business revenue by the same amount, leaving profits unchanged. In the real world, the impact on spending of a change in wages is not 1 to 1, of course, but there is enough of a correspondence such that wages changes do not have a major impact on profits.

    1. Agreed. The issue is that the relative pricing between wages and prices determines the breakdown between profits and wages. If workers produce more than they can buy with their wages, the extra production has to be bought somehow. Possibilities include investment (which includes inventories), government purchases, and purchases by the recipients of dividends.

  2. I read Tankus differently. It seems to me that Tankus also confuses the popular notion of inflation, which is an increase in the cost of living, with the technical meaning of inflation, which includes the cost of labor. In fact, the title of his article invokes the popular meaning, which is not the intended meaning when people talk about inflation reducing the burden of debt. That merely adds to the general confusion.

    I seldom read Krugman, but my impression was that he appeals to the general technical term. Certainly both Nick Rowe and Bill Mitchell, whom I read more frequently, use inflation in the general sense. Tankus, I am afraid, is still mired in confusion. He even supposes that deflation would be kind to workers. That may be logical, but it shows that his understanding of inflation is faulty.

    We live in a time of increasing income inequality. In addition, after the recent financial crisis we bailed out the banksters. The upper crust of our economy had a quick recovery, but the rest of us have been mired in a depression, and may still be. Bailing out the rest of us may or may not induce inflation, but if it did that would be a good thing, would it not? It would ease the burden of debt.

    1. When he quotes Krugman, Krugman refers to "inflation" being a specific value, which corresponded to CPI inflation (I believe). I doubt that anything I wrote would be news to him; it's just that he was sloppy in that case when writing about "inflation." Other economists use that shorthand.

      "Bailouts" are a tricky concept. The "bailouts" that happened during the crisis could be defended as lender-of-last-resort operations. (One could argue that the bailouts should have been less of a sweetheart deal, but I do not know enough the details to argue either way. I was too busy dealing with the financial market effects of the crisis to care about whether the bailouts were fair.) Lender-of-last-resort operations appear unfair (why do banks get treated specially?), but you need to do them to avoid meltdowns in the banking system.

      Can we extend the same sort of treatment to non-banks? The equivalent of what was done for the banks would be for the government to extend loans to households, which would allow them to service their debt. The first thing to note is that this is effectively a bailout to lenders who would have otherwise suffered credit losses. It then becomes a political problem - who gets access to those loans? People who are not eligible might be unhappy. Actually, such a policy would not add much to final demand, so it would not be particularly inflationary. It is just a movement of financial claims amongst various entities.

      Handing out cash to households would be inflationary. How you divide up the cash is going to cause political problems. However, if the amounts are small, it would not help people who borrowed huge amounts of money to buy a house.

    2. I apologize for getting personal about Tankus. That was not my intention, and it is beneath the dignity of this blog.

    3. Thank you for your informative response. :)

      FWIW: At the time I was in favor of the bank bailouts, although I also think that heads should have rolled. But when Paulson said that it was so that the banks could start lending again, a 15 minute web search suggested that that was bullshit. Historically, after a combination recession and bank panic, it has taken years for the bank lending to recover. I also thought that a good bit of the bank bailouts should have been in the form of bailing out their debtors. Two birds with one stone, as it were. It worked in Europe, but there, too,

      Indeed, bailing out the 97% would raise political questions, but that is true of all bailouts. It is unfortunate, in my view, that the backlash against the bank bailouts told against a general stimulus. I understand the anger, but it was directed against the wrong target.

    4. Oh, what I meant to say was that the indirect bank bailouts in Europe were not large enough to bail out the debtors, as well.

  3. Brian Thanks. Good point. Or if workers spent more than their wages, their extra production could be purchased.

  4. Any statement in economics that X causes Y carries an implicit ceteris paribus assumption. The question then is what exactly are we assuming is constant. In this case, are we assuming nominal wages or real wages are unchanged? I think most economists would take it as read, when talking about inflation (even CPI inflation) that it is real wages that we are assuming are constant. Tankus obviously reads it the other way.

    1. Actually, I would normally treat nominal wages as constant in the short term when discussing the effect of a change in the consumer price level as well. The days when most people had cost-of-living escalators in their contracts are long gone.

      But your explanation makes quite a bit of sense. Since I did not study economics in school, but picked it up on the job, I do not follow the same embedded assumptions.

    2. Hi, Nick.

      My interpretation of "the price level" without any other qualification is the abstraction of prices, P, in the equation,

      MV = PQ

      in which Q stands for goods and services, including labor. So that Tankus's statement, "Some will object that Krugman and many others are simply using increases in the price level as a proxy for increasing nominal income," is incorrect. The price level is not a proxy for nominal income in the form of wages, it includes wages as the price of labor.

  5. Hyman Minsky has a comprehensive discussion of inflation in his book Stabilizing an Unstable Economy. If I recall correctly he predicts consumer goods price inflation via a shift from production of consumption goods toward more investment goods. I think this shift would also increase profits when applying the Kalecki-Levy profits identity - which Minsky uses to understand the aggregate economy. I think Minsky favors an economy with more consumption spending and less investment spending for increased employment and lower inflation near the full-employment boundary.

  6. This is a very thoughtful response, thanks. The problem with assuming they are using inflation as a proxy for nominal wage increases is a)nominal wages can grow along with productivity b)it is the rise in unit labor costs that (partially) drives the rise in inflation, not the other way around. So either the statement is just wrong or it reduces to a much more obvious statement.

    The other problem is it depends on what policy you implement to cause inflation. the government could easily cause inflation by paying capitalists not to produce output over and above a certain level. this would drive up profit margins without necessarily benefiting worker's wages. Additionally, Krugman makes this point in order to get traction around increasing "inflation expectations". he's never said "and this mechanism works through workers demanding higher nominal wages". When you're incoherent it is easy to come up with incoherent policies that cause a divergence in the data.

    1. Thanks for the response. I agree that it is a frustrating blind spot, but I see it all the time. One of my objectives here is to explain economist logic, and that includes giving the details of the embedded assumptions they use, even if I do not agree with those assumptions.

      I think Nick Edmonds' point that economists have been trained to think about wages as "real wages", without asking whether that holds up empirically, probably explains the assumption. It may have made sense in earlier decades, but not so much now.

    2. Absolutely agree. Following Keynes I think people care much more about their nominal wages and it takes large and persistent price level swings (and institutional power) for them to adjust their nominal wage demands because of the price level. he specifically calls trade unionists "more reasonable economists" because of this point in the general theory.

      I'm doing a lot of work on the antebellum period and even than (a time a lot of economists think had "flexible" prices) the rough data suggests they had persistently falling/rising real wages with the price level.

    3. Just curious. Which bellum?

  7. I agree with both Nathan and Brian. One thing I would ad is that the "inflation erodes debt" claim is more commonly made in regard to public debt, and there it is more unequivocally correct. Inflation increases nominal GDP more or less by definition. The other thing I think Nathan is missing is that it is not wages but income that is relevant for debt repayment capacity, and income of course includes profit income. Nathan is assuming that debt is mostly found lower in the income distribution, where wage income dominates, but I'm not convinced this is true.

    On the larger point, I think Nathan is absolutely right to criticize the common practice, shared by economists and policy/finance types, of "correcting for inflation" by mechanically subtracting/dividing by the CPI.

    1. I mentioned the case of governments, but we still need to be careful about which inflation measure to use. If a country is an oil importer and the price of oil rises, it might not help the government too much. (It would depend on how taxes are levied; the taxes on gasoline are flat taxes, tax revenue would presumably fall due to lower oil consumption.)

    2. More "inflation decreases savings." We can agree on that, no?

  8. Also, have you seen this?

    Very smart criticism of the use of CPI by central banks.


Note: Posts may be moderated, and there may be a considerable delay before they appear.

Although I welcome people who disagree with me, please be civil.

Please note that my spam comment filter appears to dislike long "anonymous" posts. I get no warning about this, and only go through my "spambox" infrequently. The best bet it to keep comments short, and if you think the spam filter struck, let me know with a short comment.