Good News! There was no housing crash.

At least according to the US government.

The BLS claims that housing prices are up 2.1% in the last 12 months.  Why does this matter?  For all sorts or reasons, but first let’s try to figure out what really happened.  According to the BLS, housing makes up nearly 40% of the core basket of goods and services.

Category    weight     inflation

Housing     39 %             2.1%

Other         61%              1.4%

Overall      100%             1.7%

Suppose that instead of rising 2.1%, housing costs have actually fallen 2.1% over the past 12 months?  In that case the core rate would be zero.  Which number seems more likely?  For much of the past year house prices have been falling at more than 2% a month.

At this point you may be wondering how the government could be so stupid as to think house prices have risen 2.1% over the past year.  Well they may be stupid, but not THAT stupid.  They are trying to measure something called the “rental equivalent” cost of housing.  Yet by all accounts even monthly rents are falling rapidly.  So why do they show 2.1% inflation of housing prices?  I’m not sure; perhaps someone can answer these two questions:

1.  Do they factor in the effect of deals that include one or two months of free rent?  I don’t think they do.  These are much more common during a downturn, and are disguised price cuts.

2.  Do they survey all apartments, or just newly rented units?  I hope it is the latter, as rents on older leases are not true prices at all, and should definitely not be included in any price index.   They are analogous to monthly payments on a mortgage.

I saw an article (that I can’t cite) which estimated that accounting for disguised rent cuts the actual rental rate is now falling at about a 5% rate.  This means that while BLS data shows 1.7% core inflation, the true rate may well be negative.  Indeed it is not obvious that the actual core inflation rate is any higher than the overall or “headline rate,” which was negative 1.4% over the past 12 months.  Furthermore, it could get even worse, as the other major component in the core rate—the price of services—is closely related to wage rates.  And wage inflation is now slowing sharply.

Why does any of this matter?  In my recent debate with Lee Ohanian he made these assertions:

In contrast, I view deflation as less likely than inflation “” in part because deflation would be so damaging that the Federal Reserve cannot afford to allow it to occur, and would adjust its policy accordingly.

.   .   .

At the end of the day, Federal Reserve policy has a large impact on whether deflation or inflation occurs, and given that the costs of deflation outweigh those of modest inflation, I expect the Fed will adjust policy to prevent significant deflation from occurring. The Fed’s moves have successfully done this over the last nine months, and would continue to do it if needed.

Professor Ohanian and I agreed on many points; we both saw Fed policy as determining the rate of inflation, and we both expected slightly over zero inflation going forward.  The difference is that I thought that was too little inflation, and he thought it was about right.  Thus I favored a more expansionary monetary policy, while he argued the economy did not need any more stimulus.  But after seeing this core inflation data, I’m inclined to think we were both too optimistic.  Contrary to what Ohanian asserted, it looks like the Fed has not successfully prevented deflation over the past nine months.  If Ohanian believes that deflation would be “damaging” to the economy, and if he opposes monetary stimulus on the grounds that we have avoided deflation, then it certainly seems important to figure out whether we have in fact avoided deflation.  Especially given that he has been a consultant to the Minneapolis Fed.

As my long-time readers know by now, I have always favored using NGDP as both an indicator and target of monetary policy.  Indeed I would prefer NGDP even if inflation was measured precisely.   But of course the inflation numbers that we work with, the numbers that eminent professors at Ivy League schools plug into their models, seem pretty much worthless to me.  What exactly is the inflation rate trying to measure?  In a world where goods are always changing, there is no obvious way to measure inflation.  I suppose you could try to insist on some sort of utility-based concept, inflation as the rise in wages necessary to hold utility constant. But that would end up merely measuring wage inflation, as peoples’ expectations rise with every technological advance.

There is no perfect measure of inflation, only measures that are more or less convenient for various purposes.  If I am interested in measuring the “cost of living,” I might want an inflation measure that includes rental equivalent.  If I am interested in the damaging effects of deflation on the economy, I would not care at all about rents; rather I would be interested in the prices of new houses and apartment buildings.  Why?  Because if those prices fall sharply (even if rents are stable) then in a world of sticky wages the firms that build houses and apartments will dramatically reduce output.  Thus deflation will cause a steep recession.  And this will be true regardless of whether rents on existing apartments have risen or fallen.

In the end, I keep coming back to NGDP.  It is something real, not a number dreamed up by a government statistician.  (Well, even NGDP involves a bit of dreaming.)  If the PC industry sells $34 billion in computers, then they sell $34 billion in computers.  The inflation rate for PCs?  Well, let’s see, I think that new Dell is 29% niftier than the old one, what do you think George?

If I wrote a macro book, I’d try to leave inflation and the price level out entirely.  In the end, you might have to include it in the growth chapters, and perhaps exchange rates (for traded goods only), but nowhere else.  The Fisher effect?  I’d just use expected NGDP growth instead of expected inflation.  BTW, this is why China isn’t even close to a liquidity trap despite deflation–check out their NGDP growth numbers.  It is expected NGDP growth not expected inflation, which determines whether you will be in a liquidity trap.

My model of business cycles would use only two variables, NGDP and aggregate hours worked.  After all, you can’t have RGDP if you don’t have inflation.  Unemployment also seems too arbitrary.  What’s the difference between “unemployed” and “discouraged worker” and “involuntarily part time?”  Employment is more definite, but even that glosses over the full time/part time distinction.  So I think total hours worked is best.

Then we’d need an AS/AD model.  What would it look like without prices and RGDP?  You could put AH (aggregate hours) on the horizontal axis, and actual minus expected NGDP on the vertical axis.  There is no AD curve, and the AS curve crosses the horizontal axis at the point where the aggregate number of hours worked equals the natural rate.  It has the normal positive slope, like any other AS curve.  If NGDP exceeds expectations then AH exceeds the natural rate, and vice versa.

Or you could follow Earl Thompson and put average nominal wage rates on the vertical axis.  In that case all you would have to do to is have monetary policy stabilize next period’s expected wage rate, perhaps with currency redeemable into futures contracts linked to the average wage rate next period.  That would keep employment at its Walrasian equilibrium value, eliminating suboptimal business cycles.  Productivity shocks could still impact RGDP, but remember we don’t have RGDP in our model, only AH.  Hours worked should remain fairly stable, growing with population and changes in demand for leisure.

In 1982 Earl Thompson called it a perfect monetary  system.  I still don’t see anything better out there.



34 Responses to “Good News! There was no housing crash.”

  1. Gravatar of rb rb
    22. July 2009 at 04:47

    Sorry, I am sure this is the wrong place to ask but, have you considered commenting on any of Bernanke’s testimony or his op-ed from yesterday? Thanks

  2. Gravatar of pj pj
    22. July 2009 at 05:09

    The BLS measures neither rent nor housing prices. They use “owner’s equivalent rent” which comes from a telephone survey of homeowners asking them the question, “If you were to rent your home, what monthly rent would you expect to receive?” The answers to this question determine the OER index. The OER rents famously lag turning points in the housing market by several years. OER will turn within a year and will show deflation for a number of years, underestimating inflation when house prices finally turn.

  3. Gravatar of ssumner ssumner
    22. July 2009 at 05:25

    rb, That will be my next post.

  4. Gravatar of B.B. B.B.
    22. July 2009 at 05:48

    Sorry, but no blue ribbon today.

    (1) Conceptually and practically, an index of wages is no easier than an index of consumer prices. The same issues bedevil both: changing quality, changing mix, changing region, hidden discounts. How would you treat bonuses, commissions, stock options, awards, profit sharing? What about fringe benefits, pensions, health insurance?

    (2) I think you are dead wrong on marginal vs. average price. Just because a price is set in a long-term contract does not mean it does not belong in a price index. The theory of price indexes refers to prices consumers use to buy. Whether the prices are floating spot prices or set for a year at a time makes no difference. Of course, if a lot of prices are set by contract, we get Keynesian “sticky prices,” which affect business cycle dynamics, and inertia in the inflation rate. Well, yes.

    (3) I think the BLS does a good job with measuring the CPI. But I agree with you that implicit homeowners rental (which is not the same as apartment rents) is flawed. But what is the alternative? Putting mortgage payments into the CPI means that the nominal mortgage rate should be part of the CPI, which is nonsense. We tried that and it did not work.

    (4) Perhaps the Fed should have a special index to target, and the CPI should not be that index. Fair enough. That does not mean the CPI is a flawed index.

    (5) NGDP is not a pure dollar figure, even though it is expressed in dollars. It is filled with imputations. In 2007, imputations were 14.7% of NGDP. (Table 7.12 in Survey of Current Business, NIPA) Notably, the implicit homeowners’ rentals are a large part of GDP (7.7%). There is also a substantial amount of imputations of the value of services rendered without payment by banks, amounting to a huge $427 billion in 2007. You want to target NGDP, you are targeting imputations also.

    Perhaps the Fed ought to target a GDP Minus Imputations figure.

    (6) I do think that NGDP has a lot of merit as a target, but are you asking for a constant NGDP growth rate? Or could the NGDP target vary with, say, the unemployment rate?

    (7) You can argue that we would be better off if NGDP had increased at +4% in Q4:2008 and Q1:2009. I would agree. But if the Fed can neither control velocity nor forecast it perfectly, what should it do? NGDP growth was not all that low before the disasters of September 2008. The financial panic caused a plunge in velocity. Given the lags in monetary policy, if the Fed had gone to QE by the end of September, how much difference would it have made?

  5. Gravatar of Nick Nick
    22. July 2009 at 06:13

    I’m honestly not terribly familiar with the calculation, but I think part of this is the offset within the OER calculation for energy prices. Utilities costs are subtracted from total rents. If rents stay the same but utilities costs drop, the implied OER actually rises. We’ve seen dropping rents and dropping utilities costs over the past year, so I would guess that’s why you’re ending up with this type of a number.

  6. Gravatar of Bill Woolsey Bill Woolsey
    22. July 2009 at 06:24


    Pretty awful.

    Exactly how does the Fisher effect work with expected nominal income growth? Is not being in a “liquity trap” the only thing we want to know about interest rates?

    What is it that you get when you subtract nominal income growth from a nominal interest rate? What is it that you add nominal income growth too in order to get equilibrium nominal interest rates? Why do borrowers and lenders care about that?

    In my view, real interest rates coordinate spending on consumer goods with the production of consumer goods through time. They coordinate saving and investment.

    My lesson from the last 2 years is that the reality of multiple interest rates (risk premia and the yield structure) is important. And not just because of a supposed liquidity trap.

    The real wage rate coordinates the demand for consumer goods (now and in the future) with leisure. And, of course, the constellion of real wages impact occupation choice.

    What happens when the population grows? Hours works stay the same? What about structural and technological unemployment? They don’t exist? They don’t change?

    Developing a macroeconomic theory that depending on your understanding of conditions in 2009 is pretty pointless, in my opinion.

    This must be some kind of Chicago thing. I am interested in how it all fits together and how monetary disequilibrium disrupts things.

    In my view, stable growth of nominal income is useful because it provides the best envirornment for microeconomic adjustment. And explaining that without looking at relative prices (real wages and real interest rates) is pointless.

  7. Gravatar of Bill Woolsey Bill Woolsey
    22. July 2009 at 06:43


    It don’t see any point in stabilizing the cost of living. Why ignore the prices of capital goods and government goods?

    Measuring the cost of living might be interesting if you want to estimate what is happening to real standards of living over time.

    But the role of the price level in stabilization policy has to do with monetary disequilibirum. Net upward pressure on prices suggests net excess demands for goods and services, and so, an excess supply of money. And vice versa.

    To me, justifying some way of measuring the price level based upon its ability to track the cost of living and estimate what is happening to standards of living is beside the point. The goal should be, I would think, to try see what is happening to the net excess demands in the economy.

    I think that measures of aggregate nominal expenditure get at this better. GDP (nominal) is one such measure. I prefer final sales of goods and services. But the imputing you describe with GDP would be included there too.

  8. Gravatar of David Pearson David Pearson
    22. July 2009 at 06:50


    In your (hopefully upcoming) post on Bernanke, perhaps you could address why he keeps referring to monetary policy as “highly accommodative”. He repeated this yesterday in his Congressional testimony.

    Do you believe policy is even “somewhat accommodative” given that all the base growth has gone into excess reserves, hours worked are declining, average wages are on the verge of declining, house and other real estate prices are declining, etc? If so, then is the Chairman:

    a) engaging in “mis-direction”?
    b) defining “accommodative” as something other than stimulating AD?
    c) misunderstanding monetary stimulus?

  9. Gravatar of ssumner ssumner
    22. July 2009 at 06:52

    pj, I just read a 30 page article on this, which is on the BLS’s website, and they do more than ask questions to homeowners, they also measure the rent on actual apartments. The housing figure I use includes both owner occupied and rental units. They have rental units rising 2.7% over the past 12 months.


    1. Wages are complicated, but not the wages required for a wage index. You want to stabilize an index of the wages that are sticky, if your goal is to stabilize employment. So you don’t care about things like bonuses. The bigger problem is that many jobs (like mine) don’t pay hourly wages. That’s one reason I am sticking with NGDP targeting.

    2. I am afraid you are wrong here, but it is a common mistake. Indeed I once published an article on the point (with Ross Newman and Davil Gulley.) A sticky price is when a catalogue sets a price and doesn’t change it for an extended period of time. Say a year. What makes it a price is that the consumer can still decide how much to buy at that sticky price. It makes it the marginal cost of buying one new unit of the good. Rentals under leases are different. When you sign a one year lease you are essentially giving the landlord a bond, promising specified nominal payments at specified dates. Unlike with sticky prices, you cannot take advantage of fixed rental rates by consuming more apartments if the overall price level rises (and hence real rental rates fall o nexisting leases) because any new apartment rented will rent at market clearing rates. That is why the BLS should only use new rentals. And it also explains why their data lags reality so badly.
    Here is an analogy. An auto worker gets say $25 an hour, and a professor gets $10,000 a month. But only one of those is sticky wages. If the price level suddenly rises the auto company can take advantage of sticky wages by having the workers work more hours. But the professor’s contract is for a fixed amount of labor over a year. Hence the seemingly “fixed” monthly wages are simply a bond that the university gives the professor, which promises periodic nominal payments. Professors don’t have sticky wages. If everyone worked on one year annual contracts like mine, there would be no wage stickiness in the economy.

    Current rent payments are no more current “prices” then are the current monthly payments on a 30 year mortgage you took out in 1980.

    3. The solution is to try to measure current market rents, which in a weak market often include one or two free months rent. And also only include newly-rented apartments. Then assume house rents are comparable to apartment rents for the same size and location. There will still be an unmeasured gap, but the rates of change should be similar.

    4. I agree.

    5. I agree. (in the post I hinted that NGDP also had some guesswork.)

    6. I’d just go with NGDP, it is dangerous to try targeting unemployment, as the natural rate changes.

    7. I favor targeting the forecast, and I favor “level targeting.” The forecasts for NGDP were far below 5% in my view last August and September, so the Fed would have needed a much more expansionary policy.

    Level targeting helps stabilize expectations. Thus if NGDP started falling before the Fed could do anything, expected NGDP growth would rise, in order to “catch up” to the target path. But the expectation that this would occur will help prevent NGDP from falling in the first place. Those who understand currency bands under a fixed exchange rate regime may be familiar with this concept. It requires credibility, and when the Fed lost credibility early last October I knew we were in big trouble, and the stock market knew it too.

    BTW, Although I strongly disagreed with you on point 2, those were great comments. You are at least partly, and in some cases completely, correct on almost all your points.

    Nick, You may be right. I am relying on a study that I unfortunately cannot cite, which says that if the BLS factored in things like free months rent the actual change would be about negative 5%. But I don’t expect anyone to take my word. However, if the BLS is looking at rents under contracts signed months earlier (as I think they are), they are certainly understating the fall in free market rents, which is what matters in macro.

    The other issue is that when people worry about “deflation” as something that causes macro instability, it isn’t rents that matter, but the prices of new homes, which are clearly falling fast.

  10. Gravatar of ssumner ssumner
    22. July 2009 at 07:06

    Bill, I can’t answer your question about the Fisher effect until you tell me what inflation is. What is the definition in a world were goods are constantly changing in a qualitative sense? I thought the textbook said something like the income increase you’d need to keep your utility constant. But that makes no sense, becasue in a “keeping up with the Jones” world (which is the one we actually live in) that would mean wage inflation, not price inflation.

    When those statisticians come to the conclusion that all the nifty add-ons to a Dell computer mean its price fell 29%, how does that judgment affect nominal interest rates? Obviously not at all. So how does inflation affect nominal interest rates, if we can’t even define what inflation is?

    The Fisher effect is only useful as a device for explaining how different nominal growth tracks will affect nominal interest rates. My claim is that nominal interest rates will track NGDP growth more closely than the CPI, because the real interest rate is strongly correlated with expected real growth. Do you disagree?

    I also plan to stick with the NGDP targeting rule. But there is a real beauty to the Thompson proposal that I think is under-appreciated. It gets right at the central macro problem–involuntary unemployment. But it has zero chance of being adopted, whereas NGDP targeting has a small chance.

    David, I’ll try to keep that in mind. The huge quantity of comments (which is good) also slows down my new posting rate.

  11. Gravatar of The Ambrosini Critique » Blog Archive » Crap The Ambrosini Critique » Blog Archive » Crap
    22. July 2009 at 08:03

    […] “If everyone worked on one year annual contracts like mine, there would be no wage stickiness in the economy.” — Scott Sumner […]

  12. Gravatar of Lorenzo (from downunder) Lorenzo (from downunder)
    22. July 2009 at 11:38

    The Australian Bureau of Statistics collects its housing price data rather differently from the BLS.

  13. Gravatar of Bill Woolsey Bill Woolsey
    22. July 2009 at 13:47


    The Fisher effect would exist if there were no such thing as the CPI.

    Ex post, each person must determine the purchasing power of money for themselves. Does the money buy more or less? I don’t see why the equal utility approach is a problem from the point of each person making an evaluation.

    Ex ante, lenders expecting inflation lend less and borrowers expecting inflation borrow more. The nominal interest rate rises. The expected real interest rate (which people care about) is the difference between the nominal interest rate and the expected inflation rate.

    This would be true if no one tried to measure the price level.

    Now, ex post, why should I care about the difference between the growth rate of the rest of my nominal income and the nominal interest rate I earned?

    The real interest rate tells me whether I moved consumption from the past to the present. Your measure (the nominal interest rate less the growth of my total nomiminal income) tells me what?

    You can talk oall day about problems with the CPI, but you need to tell me why I should care.

    So, people save less if nominal income is expected to grow faster. I think I can see that people would borrow more if they expected higher nominal income in the future.

    And so nominal interest rates rise. And….What do we get? Equilibrium is obtained where the nominal interest rate minus the expected growth rate of nominal income is equal to… what? Zero? Why is that true necessarily? Or is it something else?

    Suppose the productive capacity of the economy is dropping. The price level is stable. Aggregate demand is dropping with productive capactiy. Nominal income is shrinking. Nominal interest rates must be negative?

    In my opinion, developing interest rate theory based on nothing more than “higher nominal income growth will raise nominal interest rates” leaves a lot out.

    Suppose we lived in a world of unchanging quality? Suppose we lived in a world of constant relative prices? So, the concept of inflation is unambigous. Then your argument against traditional theory is wrong. Real interest rates are unambigious.

    In my view, adding ambiguity doesn’t suddenly mean that real interest rates don’t matter.

    Just because economists can’t test theories doesn’t make them worthless. Like I said, the Fisher effect exists even when no one calculates a price level statistic. And even if no one wrote down the prices so that no economist can later construct a price index and test it, it stil exists.

  14. Gravatar of Bea Bea
    22. July 2009 at 14:19

    Unrelated to this post — did you know the text in your RSS feed has been taken over by a Cialis advertiser/spambot?

  15. Gravatar of » This May Make Robert Shiller and SocSec recipients happy… » This May Make Robert Shiller and SocSec recipients happy…
    22. July 2009 at 14:58

    […] it doesn’t do that much for the rest of us. Via The Ambrosini Critique, Scott Sumner discovers there was no housing crash: The BLS claims that housing prices are up 2.1% in the last 12 months….According to the BLS, […]

  16. Gravatar of q q
    22. July 2009 at 15:58

    here are a couple of bls links:

  17. Gravatar of StatsGuy StatsGuy
    22. July 2009 at 16:44

    To ssummner:

    I believe there is one major component that could contribute to a nominal rise in housing rents being a “real” event: the resetting of ARM mortgages. The reset schedule dipped dramatically in 2008 as subprimes were phased out, but then picked up in late 2008/2009 as various other ARMs (including primes) all reset… Although a lot of these rates were not significantly hurt due to the low rate structure at present, many of these homeowners have so-so credit, and thus pay steeply higher interest payments due to risk (which was discounted in teaser rates or interest-only loans). The total number of loans hitting the reset date was scheduled at 20-30 billion per month (though some may have been refi’d, if their owners have jobs still).

    So why does this relate to rental charges? Because a large portion is derived from _rental equivalent_, which is essentially by asking people “If someone were to rent your home today, how much do you think it would rent for monthly, unfurnished and without utilities?”

    Typically, people respond to questions like this by answering something related to what they _currently_ pay in their own monthly mortgage payments (if they have a mortgage), so as their monthly rate goes up, their answer also goes up. This is like using stated preference instead of revealed preference (i.e. the market price) – deeply flawed. Homeowners are well known to be living in denial until the market crushingly brings home reality when they are forced to sell.

    In other words, rising “rental” prices may be “real”, in the sense that consumers are being soaked for expensive debt payments. But this does not actually reflect an increase in _willingness to pay_, but rather a long term contractual obligation with a backloaded payment schedule.

    That is _not a good thing_, and if the Fed interprets this as strength in the price level… ugh. I don’t want to think about it.

  18. Gravatar of StatsGuy StatsGuy
    22. July 2009 at 17:11


    “…real interest rates coordinate spending on consumer goods with the production of consumer goods through time. They coordinate saving and investment.”

    That’s the supply side story, which only works if we focus solely on flows and equillibrium, not on storable stocks which contribute to cyclical disequillibrium. The demand side story is that rates also coordinate consumption over time, which matters when goods are storable and/or have high longevity. For instance, housing.

    “The real wage rate coordinates the demand for consumer goods (now and in the future) with leisure.”

    [Don’t forget the past, via debt…]

    Again, this ignores stocks and focuses narrowly on flows. Real wages declined for the median earner for 20 years, but perceived asset appreciation (and an increase in leverage vis-a-vis the debt-to-gdp ratio) sustained demand. The missing link was perceived asset wealth – and that’s why the finance sector blindsided so many economists. Until economists fully integrate asset pricing theory and wealth into money supply/demand and AS/AD, we’ve got some gaps. And good luck on that…

    BB: “I think you are dead wrong on marginal vs. average price. Just because a price is set in a long-term contract does not mean it does not belong in a price index.”

    Again, this means that you use a model that focuses on flows only, and ignores stocks. Why is that important? Because getting a house mortgage is like buying a reverse bond – you are arguing that we should only focus on the FLOW of money to cover the mortgage, but the net-present-value of the mortgage adjusts immediately, and homeowners react (unless they live hand to mouth). Thus, homeowners behave as if they suffer an income loss (or, equivalently, a rise in prices) because they are reacting to future expectations.

    This is one reason why the Fed is so late to the party – not only do they have to wait 3 months for the data, but they always focus on flow-models. The stock market, and consumers, all correctly anticipated the magnitude of the AD shift – the Fed did not.

    So why this occupational bias? I think the answer is obvious – if wealth were to be integrated into the standard model, we’d end up with the result that distributional equity can impact economic productivity. That is anathema to libertarians, so they avoid thinking about it.

  19. Gravatar of This May Make Robert Shiller and SocSec recipients happy… | Bear Market Investments This May Make Robert Shiller and SocSec recipients happy… | Bear Market Investments
    22. July 2009 at 18:00

    […] but it doesn’t do that much for the rest of us. Via The Ambrosini Critique, Scott Sumner discovers there was no housing crash: The BLS claims that housing prices are up 2.1% in the last 12 months….According to the BLS, […]

  20. Gravatar of ssumner ssumner
    23. July 2009 at 05:06

    Bill, Obviously I agree that there is a Fisher effect in the broad sense that faster nominal aggregate growth raises nominal rates. And the CPI is the nominal aggregate traditionally used in the Fisher effect. That part I am fine with. But might other nominal aggregates work better? Let’s accept you constant quality assumption, so our CPI is very accurate. Does my argument break down? I don’t think so.

    Consider a world of stable price levels. Now assume that wage inflation goes from a 0% long run trend to 10% long run trend. (Obviously this implies that real wage inflation is also accelerating.) Now let’s think about what happens in the loanable funds market. I say that potential borrowers are now willing to pay much higher nominal rates. They think “wow, with those pay raises coming along I can buy a big house, and be bailed out of my onerous mortgage by rising monthly paychecks.” Isn’t this exactly the sort of thinking that produces a Fisher effect when there is price inflation? I.e. that “I will be bailed out by inflation?”

    So this thought process shifts the demand for loanable funds to the right, just like higher expected inflation would. Now what about the supply of loanable funds? Here is where I am a bit less sure, but I’ll take a stab at it. Take the sudden jump to 10% wage growth—that suggests nominal and real wages will rise twenty-fold over the next 30 years. So if you are making a loan you’d think “gee, the $10 dollars I loan now requires an hour of my labor, but when the loan is repaid in 30 years it will merely require 3 minutes of my labor. I’d rather consume it today and work an extra three minutes in the future.” If I am right, then this also shifts the supply of loanable funds to the left, just as in the Fisher effect.

    So here is my argument.

    1. If both wages and prices rise together, the inflation rate will clearly predict the Fisher effect. But equally so will the wage inflation rate.

    2. If the wage and price inflation rate do not rise together, then it is the wage inflation rate that really matters.

    Ergo, the Fisher effect is basically about wage inflation, or nominal income inflation, or NGDP/person inflation. I’m not sure how population growth figures in, so I may have been wrong about total NGDP.

  21. Gravatar of ssumner ssumner
    23. July 2009 at 05:27

    Lorenzo, If I understand it correctly, I think I like the Australian approach better.

    Bea, Thanks, we are working on a new platform that is more bug-proof, and hope to have it soon.

    q, Very helpful. I think that confirms both of my assumptions in the post. The BLS uses old leases, and doesn’t count free months rent (which are disguised price cuts.) Neither matter in the long run, but my point is the CPI is a horrible way of looking at sudden price level changes over the business cycle. NGDP is much better.

    Macro needs a new language. The term ‘deflation’ should refer to falling NGDP, not a falling CPI.

    Statsguy, That is a very interesting point. I think the technique used by the BLS is very flawed, but I hadn’t thought of that angle, which sounds very plausible. By the way, your comment about ARMs makes me realize how foolish those loans were. If people are getting hammered in this environment, just imagine how bad it would be if interest rates were still at normal levels! I’m not opposed to all ARMs, but the government should require at least 20% down on loans made by any Federally-insured banking institution, or any “too big to fail” non-bank bank. Taxpayers should not be on the hook for that sort of insanity.

  22. Gravatar of Bill Woosley Bill Woosley
    23. July 2009 at 06:10


    The Fisher effect isn’t that higher inflation leads to higher interest rates.

    It is the nominal interest rate is equal to the equilibrium real interest rate plus the expected inflation rate.

    None of this is observable except the nominal interest rate.

    It is a theory.

    It is not based upon the CPI. It existed before there was a CPI. You are confusing the efforts of economists to test the theory using some kind of model of real interest rates and inflation expectations and data about lots of things, with the theory.

    Again, showing that more rapid nominal wage growth leads to higher nominal interest rates tells us.. what?

    Think about the neutrality of money. Money is neutral in regards to what? Relative prices. Real interest rates are relate prices. The “fisher effect” is about how inflation doesn’t impact real interest rates.. with complications, of course.

    What does the difference beween the nominal interest rate at the growth rate of wages get you? I guess it is the change in the amount of labor one could afford to hire with your money?

    By the way, the normal way to look at this is that if future income is expected to be higher, then saving decreaes and the real interest rate needed to coordinate current consumption and investment is higher.

    Depending on monetary institutions (and so, what kind of inflation rate is expected) credit markets will generate nominal interest rates consistent with those real interest rates.

    In place of this theory about real interest rates as relative prices that coordinate production and consumption through time, you are giving me a theory that says that higher nominal wage inflation leads to higher nominal interest rates. And calling this the Fisher effect.

    To me, it seems like you are not interested in undersanding the economy, but rather focused on one problem–how raising expectations of nominal income in 2010 will effect nominal interest rates in 2009.

    I have little doubt that under current conditions, higher expecations of nominal income will raise nominal interest rates. The inflation part is the Fisher effect, and the expected increase in real output impacts saving and investment and the natural interest rate. At the same time, all of these things should impact credit markets too.

    But that doesn’t mean that the real interest rate is no longer an important concept. Complain all you want about the CPI. Tell me that economists can’t measure the real interest rate. So?

  23. Gravatar of ssumner ssumner
    24. July 2009 at 05:00

    Bill, You said:

    “The Fisher effect isn’t that higher inflation leads to higher interest rates.

    It is the nominal interest rate is equal to the equilibrium real interest rate plus the expected inflation rate.

    None of this is observable except the nominal interest rate.

    It is a theory.”

    I disagree. What you have just described is a variant of the Fisher equation. I think the Fisher effect assumes that changes in expected inflation result in higher nominal rates, which is equivalent to assuming that the real rate is not affected by changes in the expected inflation rate. Otherwise someone could say “I assume that a 5% increase in expected inflation causes a 5% decrease in the equilibrium real rate, leaving the nominal rate unchanged.” That statement is consistent with your equation, but certainly not consistent with the Fisher effect as it is generally understood.

    Regarding the more important issue, I think economists wrongly assume that “inflation” is the rise of goods prices. We know from the circular flow diagram that the economy is symmetrical; there are factors of production and there is output. At some point economists decided that changes in output prices are “inflation,” but change sin input prices are . . . . well I don’t know what they call changes in input prices.

    I have just as much right to call “inflation” wage inflation as price inflation. Whether input prices or output prices are more relevant for any given model is an issue to be determined, not simply assumed. There is no theory that says price inflation is automatically the right inflation rate to put in the Fisher equation, just as there is no theory that says price inflation is the right inflation rate for the Phillips curve. Indeed I think Phillips had it right, we should be using wage inflation for the Phillips curve.

    Occam’s razor says don’t make the model more complicated than necessary. If we use price inflation for the Fisher effect, then we must separately deal with the impact of a change in wage inflation on the real interest rate. Why not simplify things by just looking at wage inflation? Then we can say that higher expected wage inflation should raise nominal interest rates one for one. What could be simpler?

  24. Gravatar of StatsGuy StatsGuy
    24. July 2009 at 05:48

    Question for ssumner:

    How does wage-rate-as-inflation deal with a situation where there is an exogenous shock to supply (bumper crop, or drought)? What about a long-term technology trend that alters the relative value of skilled vs. unskilled labor and causes skilled labor (aka, human capital) in one industry to become displaced (a sunk cost, and therefore permanently less productive)? Or cheap imports into a country with high wealth stocks (aka, the US) from a country with low wealth stocks (aka, China)?

    Inflation seems to be classically a theory of money and consumption demand, and equalizes with production demand only when consumption and production are in equillibrium (which is almost never).

    More Questions for BW:’

    You write: “You can talk oall day about problems with the CPI, but you need to tell me why I should care.”

    The Fisher effect is a theory (as you note) with unmeasurable variables, and it can’t be proven wrong. Importantly, it does not account for the formation of expectations. This expectation-formation function is drastically different in a regime where the money supply is not visible and inflation is not measured. The example you cite is a world in which consumers don’t see the money supply and only have local visibility on prices. In such a world, the most reasonable inflation trend is based on recent history and other visible dynamics.

    That is not our world. In our world, we _purport_ to have visibility on inflation, and we have decent visibility on money supply. In so far as the inflation metric commonly used is incorrect, however, and people continue to believe in it… the expected inflation can be a consistently biased measure of real inflation.

    If it is consistently biased upwards, then nominal interest rates will be “too high” in the sense that real interest rates (which, circularly, are defined by the Fisher Equation as nominal rates – expected inflation) will be so high that they trigger significant underinvestment.

    That’s why CPI matters – because people believe it to be a measure of actual inflation and therefore it influences expectations of inflation. A bias in CPI of 1% or 2% is a very serious matter indeed, which you very casually dismiss as theoretically uninteresting.

  25. Gravatar of Current Current
    24. July 2009 at 06:32

    I’m mostly with Bill on this one, but I haven’t thought about it that carefully.

    Scott’s criticisms of CPI are correct (and old). The question though in all of these issues is “what do you want the number for”. Using CPI for long term income measurements is different from using it to guide monetary policy which is again somewhat different from using it for monetary theory. Perhaps we should create three different indices for those problems.

    Statsguy, “libertarian” economists certainly do not ignore net present value of assets. Though you are correct that the Fed seem to. Have you read about the Austrian Business Cycle Theory?

    Also, a house is a capital good. The utility had from occupying a house for a period of time is a consumer good. All this depends on definitions though. Keynesians and Monetarists use different definitions. Paul Davidson wrote: “Similarly how many have worked through Chapters 1 and 2 of Friedman’s (1957) Theory of Permanent Income and realized that Friedman defines saving (p. 11) to include the purchase of new durable goods including clothing etc. while for Keynes, saving involves the decision not to purchase durables or nondurables by households?”

  26. Gravatar of Jon Jon
    24. July 2009 at 07:27

    Scott Writes:

    I have just as much right to call “inflation” wage inflation as price inflation. Whether input prices or output prices are more relevant for any given model is an issue to be determined, not simply assumed.

    I agree–as was clear from my discussion of whether real GDP was up or down. You either measure the produce itself or you measure the income. Both should allow you to estimate an inflation rate. However I have few quibbles. First, you use the term ‘input prices’. Labor is an input but I do not think that input prices per-se are the cut-point you mean. Input != Income. Second, many consumer goods are commodified–we can discuss the cost of a lb of flour through the years fairly well. How can we do the same for labor wages or capital yield? I don’t think they are homogeneous enough.

    “Similarly how many have worked through Chapters 1 and 2 of Friedman’s (1957) Theory of Permanent Income and realized that Friedman defines saving (p. 11) to include the purchase of new durable goods including clothing etc. while for Keynes, saving involves the decision not to purchase durables or nondurables by households?”

    This remark deserves to be in bold. I was recently driven up the wall by a CNBC article decrying rising savings rate as disastrous. They then cited the Paradox of Thrift, QED. Apparently it never occurred to them that ‘Thirft’ in Keynes’s way is a rather technical term. It isn’t enough to know about people’s deposit-rate; you need to know whether the banks are turning over deposits into loans. Holding cash at home was much more popular in Keynes’s day than today–and that makes a big difference.

  27. Gravatar of Current Current
    24. July 2009 at 09:01

    FYI it’s from Davidson’s chapter in “Modern Macroeconomics” by Snowden and Vane. I got it from Davidson’s website.

    Keynes himself is much to blame for these things, he was careless in his explanations. He threw around accusations about how destructive saving supposedly is with abandon.

    I’m not sure if holding cash at home was much more popular in Keynes day.

  28. Gravatar of ssumner ssumner
    25. July 2009 at 05:46

    Statsguy, The price shock from a crop failure is a great example of where wage targeting beats price targeting. If you target the price level when food prices soar, then you must depress the prices of manufactured goods. Because wages are sticky, this causes unemployment. The beauty of wage targeting is that if there is a food price shock, the country is fundamentally a bit poorer, so you keep nominal wages on track and let food prices rise. This lowers real wages, and keeps workers employed.

    Long run shifts between skilled and unskilled workers could distort thing a bit, but I think less than you might imagine. These shifts occur gradually over time, but the key macro problems are caused by sudden fall in NGDP, like last winter, which then causes almost all wage groups to suffer pay cuts. Because those cuts occur gradually, however, they also cause unemployment in the short run. I think the sectoral shifts you refer to happen too gradually to affect things much in the short run, and in the long run the market sets the relative wage of skilled and unskilled.

    Current, I agree my complaints on CPI quality are old. On your other point, I actually said something similar in another post/comment. For instance, I have no problem using the CPI for a study of how fast living standards have improved. I agree that wages would be useless for that endeavor. I prefer wage inflation for business cycle research and the Fisher effect.

    Jon, You said:

    “Second, many consumer goods are commodified-we can discuss the cost of a lb of flour through the years fairly well. How can we do the same for labor wages or capital yield? I don’t think they are homogeneous enough.”

    This is a good point, but I think one always must bear in mind the purpose to which a variable is used. If nominal wage stickiness is the problem, but wages are always measured on a per hour basis, then it doesn’t really matter if the quality of labor is changing under a wage target monetary policy. You ware trying to set monetary policy where nominal wages don’t get out of equilibrium. And that will be where wages are stable in the aggregate. And that’s true even if workers are getting smarter or lazier or whatever. The price level then adjust to keep real wages at the appropriate level.

    I’ll have to think about your second point a bit more, I haven’t worked through the implications of the split between Keynes and Friedman on savings. You might be right. As an aside, I agree with those who say the “paradox of thrift” is misnamed, the real problem is when people hoard real cash balances.

    Thanks Current. I agree about Keynes and saving. He was being provocative in the 1930s.

  29. Gravatar of Mark Mark
    29. July 2009 at 05:45

    Your blog posts are incredible…(I say that because I agree with most of them)…just added your feed and will be following you regularly.

    Just a week or so ago, I had done a back of the envelope kind of plot of what CPI might look like if it used the Shiller Home price index and/or the FHFA house price index instead of OER. I couldn’t run down how the weights have changed over time (currently around 23-24% of CPI), so I just assumed a constant weight of 25%. (I also should note that I backed out a 1% real return off home prices, based on some of Shiller’s work on real home price returns over the last 100+yrs. Tried to eliminate the argument that BLS uses to justify OER…that home prices reflect investment and shouldn’t be used…Shiller gives a compelling case that homes overall are mostly inflation hedges and not much of an investment over long periods of time.)

    Prior to yesterday’s Case-Shiller release and the recent FHFA release, deflation was running at a rate between 2 to 6%. Also, Fed was too restrictive based on this rough guesstimate as early as the 4th quarter in 2006, when the rate of change of these adjusted CPI numbers dropped dramatically.

    I cannot help but think that history will end up giving the Fed a decent share of blame for the severity of the crisis… Between this and essentially ignoring the dramatic drops in velocity that began as early as 2007 and continued through 2008 and essentially culminated with the bankruptcy of Lehman. (In my world, sharp drops in t-bill yields are caused by a dramatic shift in the demand for money…)

  30. Gravatar of Current Current
    30. July 2009 at 01:26


    I should probably talk about this a little more. Keynes was not simply attacking the holding of money. He was attacking saving.

    In his view saving put money into the hands of speculators who then used it as he describes in his liquidity-preference account of interest. So, saving makes the economy more unstable.

    He thought that the only long-term solution was for capitalists to be tied to their investments. This is a very common prejudice in Britain. The landed aristocracy are seen as better stewards over their property in the long term than modern capitalist businesses and stock-market “speculators”.

    So, saving by the lower classes is seen as bad since it comes under the control of what Keynes saw as the more rapacious and damaging strain of capitalism.

  31. Gravatar of ssumner ssumner
    30. July 2009 at 04:43

    Mark, Thanks for that very informative comment. I agree with you. My only slight quibble is that I think monetary poolicy was not too weak in late 2006, because I look more at NGDP rather than prices. And the real economy was still very strong at that time because of booming exports etc. But other than that minor point, I think all your observations are excellent.

  32. Gravatar of TheMoneyIllusion » Deflation: Is it coming? Is it here? Has it come and gone? TheMoneyIllusion » Deflation: Is it coming? Is it here? Has it come and gone?
    12. July 2010 at 11:31

    […] year ago I did a post discussing flaws in the way the US government computes price indices.  Official government […]

  33. Gravatar of Deflation: Is It Coming? Is It Here? Has It Come and Gone? Deflation: Is It Coming? Is It Here? Has It Come and Gone?
    12. July 2010 at 23:33

    […] year ago I did a post discussing flaws in the way the US government computes price indices.  Official government […]

  34. Gravatar of TheMoneyIllusion » A few notes on the GDP revisions TheMoneyIllusion » A few notes on the GDP revisions
    2. August 2010 at 07:22

    […] The newly revised GDP accounts paint a very different picture of the recession.  Instead of two quarters of steeply falling RGDP (2008:Q4, and 2009:Q1) there are now three really bad quarters.  The third quarter of 2008 is now estimated to have seen a 4.0% plunge in RGDP.  Nominal growth slowed abruptly from the roughly 5% norm of preceding years, to only 0.4%.  And even that is probably overstated, as the nominal GDP numbers rely on rent imputation values for housing that almost comically overstate inflation during a housing crash (as I’ve discussed in previous posts.) […]

Leave a Reply