Quick Note: PPP Adjustments

All construction cost figures that I sign my name to adjust currencies for purchasing power parity, or PPP. In other words, I convert currencies across countries in PPP terms, not exchange rate terms. This is not how everyone else does this; the World Bank analysis of global high-speed rail costs converts currencies by exchange rate, and, since the yuan is undervalued, concludes Chinese construction costs are below world average, whereas in fact they are above average.

Why PPP?

Because nearly all of the costs of the construction of infrastructure are local. Labor is almost entirely local, and materials are as well, since concrete is made locally rather than imported. Foreign expertise and machinery are internationally traded; in those cases, currency devaluations can lead to cost overruns, but the proportion of the cost that is traded remains low.

India: an indigenization plan from the 2000s was quoted as reducing costs by 10-15%. The rupee’s exchange rate value is lower than its PPP value by a factor of about 3.3; indigenization reducing costs by 10-15% is compatible with around 20% of the total value of the original contract being imported.

Philippines: I spoke with a DOTR planner, who said that 90% of the value of civil works is local, and only 10% is imported, such as foreign expertise and imported material; the planner said that rolling stock is imported, but our construction cost estimates exclude rolling stock when possible.

Why not wages?

Because while the bulk of costs are domestic, they are not labor in developing countries. In Turkey, which is not much poorer than Southern Europe, costs split as 20% labor (US: 55%), 40% permanent materials, 10% construction materials, 30% construction equipment. The 80% non-labor costs are mostly domestically-produced, at local wages, but also at local productivity. If Turkey could produce everything at the same productivity as a richer country, it would just be a richer country. This goes even more so for actually poor countries like India and the Philippines.

The impact of PPP

With PPP adjustment, GDP per capita ceases to be a significant correlate of construction cost per km, except through the tendency of poor countries to build more elevated and fewer subway lines. This was not the original intention of the adjustment, which was to smooth dollar-euro difference, but it’s suggestive that it’s correct and meaningful.


  1. michaelrjames

    I think it is good you have posted this. I find the subject difficult so I hope some others comment on it. I (re)posted Michael Thomas’ comments on your article comparing Arab construction costs because he seemed to speak with some authority (but I have no real idea) and commonsense, and ok it seemed to largely agree with my points on that same article. For convenience I will repost his comments here:

    2019/05/06 – 16:55 Michael Thomas
    I wish to clarify something about purchasing power parity or PPP. The correct use of PPP is to compare the cost of consurmption of a basket of goods from one country to another while scaling the cost to reflect exchange rate differences between those two countries. Some things you can do with PPP is compare the cost of a haircut, or a head of lettuce or a gallon of gas. or a metro ticket or even monthly rents. When you use PPP to compare the cost of something like a skyscraper or a subway project, it is utterly devoid of meaning because those things are not part of anyone’s basket of goods or services. The only way to compare capital expenditure at that scale is just to compare it in strictly nominal terms.

    2019/05/07 – 12:39 Michael Thomas
    Capital expenditures like metro systems are wholly non-tradable. One cannot pick up and sell a local metro system in another town or country. If a good is wholly non-tradable, then the PPP value is the nominal value. Conversely, the nominal value is the PPP value. This is a standard point when calculating PPP.

    Here’s another perspective (perhaps of the World Bank & IMF?:

    The new f[GDP] figures mark a significant downward revision in previous estimates of the prices of goods and services in many developing countries, pushing down their cost of living. The result is that their economies, as adjusted for purchasing power, appear to be much larger than those derived from the previous baseline in 2005. That study, released in 2007, went in the opposite direction, revising prices in countries like China and India upward, and lowering their estimated G.D.P. levels.
    Those dramatic swings from one survey to the next are one reason why Louis Kuijs, the chief China economist for the Royal Bank of Scotland, who for years worked for the World Bank in Beijing, finds himself using the P.P.P. measure less often, instead comparing the size of economies by using exchange rates.
    “Having observed these huge changes in estimates I’ve become a bit wary of these estimates,” Mr. Kuijs said in a telephone interview. “The market can be wrong but at least it’s a pretty objective measurement, and nobody can quibble about whether it was that number or whether it was 10 percent higher.”

    This is why I had some trouble with the adjustment for cost of Morocco’s HSR, ie. it introduces a 3 fold difference which is extreme. The thing is is that it is really only an issue where there is a big difference between nominal and PPP estimates of GDP, and that really only happens for developing countries, with China in between:
    Country:……nom GDP*……PPP-GDP*….nom/PPP
    Indonesia …….$1.16…………..$3.51…….0.33

    FWIW, the argument about local versus imported components must be highly variable. While concrete is always local, by far the most expensive ingredient in it, cement, is often imported. Even rich countries import it these days (in a kind of “exporting carbon pollution” mechanism) and I suspect Morocco might import it because Morocco has been very dependent on imported energy (though it is building lots of solar-CSP) and that would make French imports cheaper (if France uses its cheap nuclear electricity to make it). Likewise the steel for HSR projects is a specialty steel and I doubt places like Morocco or Indonesia or Saudi Arabia/UAE make it.
    I remain undecided, except tending to the conclusion that for comparisons between rich countries it doesn’t matter (as long as no one makes a big deal about 10-20% difference) while it seems to introduce more distortion than realistic adjustment for developing countries perhaps including China. Final point, and this seems a bit deja vu so I expect to be corrected, but doesn’t China pay a premium for their very fast construction?

    • Sassy

      > Some things you can do with PPP is compare the cost of a haircut, or a head of lettuce or a gallon of gas. or a metro ticket or even monthly rents. When you use PPP to compare the cost of something like a skyscraper or a subway project, it is utterly devoid of meaning because those things are not part of anyone’s basket of goods or services. The only way to compare capital expenditure at that scale is just to compare it in strictly nominal terms.

      I find this logic kinda weird. The fact that skyscrapers and subways aren’t in the PPP basket makes it more useful for comparing the cost of skyscrapers and subways, not worse. If PPP was being adjusted for subways, then adjusting by PPP would be adjusting by a measure that is adjusted by the thing you’re adjusting.

      The PPP cost of a subway is how much stuff out of the PPP basket you could have gotten for that money, if you didn’t spend it on a subway. It’s the relative cost of the subway vs other stuff.

      > The market can be wrong but at least it’s a pretty objective measurement, and nobody can quibble about whether it was that number or whether it was 10 percent higher

      I agree there is value to being wrong consistently, however, there is also value to being potentially less wrong.

      • Kamoro

        By his own logic, haircuts (an example he used for PPP) shouldn’t be subject to PPP either. You can’t chop off your head and sell it to somebody else.

        Obviously we’re talking about the service, not the finished product, so it’s a false comparison to talk about picking up and selling a metro system.

        • Sassy

          In his defense, the service of hair cuts can be moved though. You can visit a country, get a hair cut, then take your haircut home with you. You can’t visit a country, get a subway built, then take the subway home with you. However, he’s still wrong that PPP is only relevant for stuff in the PPP basket.

      • ericson2314

        Subways shouldn’t be in there, but ideally you would have some sort of PPP basket just of the goods/services the construction uses: labor, concrete, heavy machinery, etc.

        It is true the more one thinks about this stuff, the harder it gets: it’s not like production of those things could be flawless shifted to other sectors if the construction didn’t happen, either, except for perhaps the imported parts. This is part of the reason I don’t care about costs per-se, but rather because they are useful proxify for shit politics, bad state capacity, delays, etc.

        • michaelrjames

          @ericson2314: “ideally you would have some sort of PPP basket just of the goods/services the construction uses: labor, concrete, heavy machinery, etc.”

          This problem and often inappropriateness of PPP is why alternatives have been created, the most well-known one being the Big Mac Index. I made a half-hearted attempt to add this to the table in my first post but gave up, partly because the data is too incomplete and perhaps more significantly, it can be as misleading as PPP. However in my comments on the original article on Arab costs I did note that for Morocco the BMI was only 16% divergent from the exchange rate, compared to 300% for PPP. But of course that could be for other, inappropriate, irrelevant reasons, and may well be the case for Morocco (one of only three African countries, apparently, with Macdonalds). Most alternatives suffer the problem of using a single item (iPods, etc).

          In fact your suggestion, or implication, is that the ICP could calculate different PPPs for different purposes, using a specific basket of goods & services tailored for different sectors like construction, retail etc. Since they collect data on thousands of items it wouldn’t be so difficult for them to do (but extremely laborious for anyone starting from scratch; incidentally they only update PPP calculations rarely, something like once a decade so it can be out of date too).
          Here’s another perspective:

          In general, the purchasing power parity (PPP) theory works miserably when applied to real-world data. In other words, it is rare for the PPP relationship to hold true between any two countries at any particular point in time. In most scientific disciplines, the failure of a theory to be supported by the data means the theory is refuted and should be thrown out or tossed away. However, economists have been reluctant to do that with the PPP theory. In part this is because the logic of the theory seems particularly sound. In part it’s because there are so many “frictions” in the real world, such as tariffs, nontariff barriers, transportation costs, measurement problems, and so on that it would actually be surprising for the theory to work when applied directly to the data. (It is much like expecting an object to follow Newton’s laws of motion while sitting on the ground.)
          In addition, economists have conceived of an alternative way to interpret or apply the PPP theory to overcome the empirical testing problem. The trick is to think of PPP as a “long-run” theory of exchange rate determination rather than a short-run theory. Under such an interpretation, it is no longer necessary for PPP to hold at any point in time. Instead, the PPP exchange rate is thought to represent a target toward which the spot exchange rate is slowly drawn.

          • Henry Miller

            The problem with the Big Mac index is in many countries the Big Mac is marketed as a luxury food item – for the rich. Thus the Morocco 16% vs 300% PPP difference probably tells us more about McDonald’s pricing strategy in Morocco than anything about relative prices. When looking at any one thing you have to account for this, the PPP as a basket of many things helps to balanced that factor out.

          • ericson2314

            I am confused how the big mac index is an alternative of PPP weighting not example of it?

            As to the “Here’s another perspective” yes that sounds right to me. Much of the optimistic theories surrounding floating exchange rates are empirically unfounded. If doing PPP weighting implies this theory of convering exchange rates holding, that is a subtlety I’ve completely missed — I would think in the short term looking at prices rather than exchange rates is certainly in tune with beliefs that the market is “short term wrong”, and compatible with it being “long term wrong” too.

    • Alon Levy

      Yes, and that’s incorrect. The original reason I used PPPs was because I was comparing dollar and euro costs, and it’s dumb to compare these are exchange rates given that there are constant fluctuations. American costs did not get 40% more expensive compared with European costs between 2007, when the euro was approaching $1.50, and 2017, when it was less than $1.10.

    • michaelrjames

      Oops, I meant to include this part from that NY Times article:

      The I.C.P. itself cautions against using its purchasing power estimates to compare developed economies, like the United States, with developing ones, like China, India and Brazil. According to the report, the margin of error when making such comparisons can be as much as plus or minus 15 percent. The P.P.P. measure is much more useful when comparing economies at similar levels of economic development in order to determine whether exchange rates are overvalued or undervalued, Mr. Kuijs said.

      • Nilo

        Is there any country that goes from cheap to expensive or expensive to cheap even if we assume 15% error in either direction?

        I.E. is it messing with the story we are telling?

        • Tom M

          Unlikely given we’re usually talking integer multiples or even orders of magnitude (hello New York city) differences in cost. The potential error is a footnote in this discussion.

          • michaelrjames

            @Tom M: “The potential error is a footnote in this discussion.”

            Agreed. Which is what I meant when I wrote: “for comparisons between rich countries it doesn’t matter (as long as no one makes a big deal about 10-20% difference)”

            Even Switzerland, amongst rich countries, is only a bit out of this range. The Nordics are within the range. OTOH, Singapore is the exact opposite, with a currency apparently undervalued by the same amount as China’s.
            Country:……nom GDP*……PPP-GDP*….nom/PPP
            Singapore…………$0.374 ……………..$0.600……..0.62

    • Paul

      Might be worth adding Switzerland to your table:
      Country:……nom GDP*……PPP-GDP*….nom/PPP

      Switzerland is interesting because the currency is highly overvalued in the exchange rate market (since it’s a safe reserve currency) but they also invest a lot in rail infrastructure.

      • Alon Levy

        Yes, and FWIW, in the Nordic countries, there aren’t real differences in costs between Finland, which is on the euro, and the other three, which are on overvalued-relative-to-PPP national currencies. I didn’t think of it before and I don’t think it’s a very strong line, because the SEK/EUR and DKK/EUR exchange rate ratios are only maybe 15-20% higher than their respective PPP ratios (the NOK/EUR ratio is more like 30% higher), but I think it’s somewhat indicative.

    • James Green

      Purchasing power parity, meaning you adjust prices across countries to account for the relative costs of living rather than just looking at the market exchange rates.

  2. Pingback: Why We Adjust Costs for PPP | Pedestrian Observations

Leave a Reply

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

WordPress.com Logo

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

Facebook photo

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

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.