Some thoughts on purchasing power parity estimates

I plan to make two points in this post:

1. It’s important to use PPP adjusted data.
2. PPP adjusted data is not always reliable.

Yes, there’s a conflict. But real life is messy.

I recently traveled to China, and that got me thinking about PPP estimates of GDP/person. Here is some data from the IMF (for 2023):

US: Nominal = $80,412 PPP adjusted = $80,412 (by definition)

Austria: Nominal $58,013 PPP adjusted = $69,069

Japan: Nominal = $33,950 PPP adjusted = $52,120

Taiwan: Nominal = $32,339 PPP adjusted = $72,485

Chile: Nominal = $17,254 PPP adjusted = $29,935

China: Nominal $12,541 PPP adjusted = $23,309

Recall my recent post where I argued that Japan’s economic performance has been somewhat disappointing. If we used nominal figures, we’d conclude that it’s been an unmitigated disaster. In 1995, Japan’s per capita nominal GDP was more than 50% higher than that of the US ($44,198 vs. $28,658. Now it’s well below half the US level. I’m not saying that doesn’t matter at all (it may explain the decline in Japanese tourism), but surely it’s not an adequate measure of Japan’s overall position relative to the US.

I’ve been to all of these places in recent years, and for the most part the PPP figures seem to be in the right ballpark. But there’s one exception. Taiwan’s PPP adjustment seems way too large.

I’m not trying to dump on Taiwan. Over at Econlog I recently praised Taiwan, which certainly has a highly successful economy. But Taiwan doesn’t seem even close to Austria in living standards. More like Japan (at best.)

Taken at face value, the IMF is claiming that Taiwan has a far lower price level than mainland China. I recently spent three weeks in China, and then stopped in Taiwan for three days on the way home. My initial reaction was that Taiwan seemed far more expensive than the mainland (albeit considerably cheaper than the US.) An hour long taxi ride to the Beijing airport costs about $15, whereas the roughly equal ride from the Taiwan airport to our hotel was about $45. Food was also more expensive in Taipei. My wife bought the same meat filled pancake that cost 80 cents in Beijing for $2 in Taipei.

Those are just anecdotes, but service prices tend to reflect wages, and wages are obviously much higher in Taiwan than in Mainland China. I can’t even imagine which goods are so cheap in Taiwan that they could explain the IMF finding the overall Taiwanese cost of living to be far lower than the mainland. Manufactured goods like cars are also really inexpensive in China. What am I missing?

In contrast, China’s upward adjustment might be a bit too small. China seemed far cheaper than all of the other countries on the list.

So PPP adjustments are far from perfect. Even so, using nominal GDP/person figures leads to nonsense, like the claim that Japan is in the midst of a Great Depression.

Here’s the FT:

After stagnating under Mao Zedong in the 1960s and 70s, China opened to the world in the 1980s — and took off in subsequent decades. Its share of the global economy rose nearly tenfold from below 2 per cent in 1990 to 18.4 per cent in 2021. No nation had ever risen so far, so fast.   

Then the reversal began. In 2022, China’s share of the world economy shrank a bit. This year it will shrink more significantly, to 17 per cent. That two-year drop of 1.4 per cent is the largest since the 1960s. 

These numbers are in “nominal” dollar terms — unadjusted for inflation — the measure that most accurately captures a nation’s relative economic strength.

Define “economic strength”. BTW, that’s a clumsy way to define nominal share of global economy. Instead of saying “unadjusted for inflation”, the FT should have said “at current exchange rates.”

PS. If I had no access to the IMF data and was told the US GDP/person was $80,000, I would have guessed (PPP adjusted) $75,000 for Austria, $60,000 for Japan, $55,000 for Taiwan, $30,000 for Chile and $26,000 for China. So Taiwan is the one where I’d be way off.

PPS. Yes, RGDP doesn’t exactly equate to living standards (which reflect consumption), but Taiwanese investment is 25.9% of GDP, only slightly above the 24.2% ratio in Austria. A slightly higher investment ratio cannot explain the size of the anomaly. Taiwan’s PPP adjustment is too large.



33 Responses to “Some thoughts on purchasing power parity estimates”

  1. Gravatar of Junio Junio
    26. November 2023 at 12:28

    Apologies, I’m going to be deviating from the post, but I’m curious on your opinion.

    Do you think there is potential in applying a supply and demand framework in analyzing some behavior by state-actors in International Relations?

  2. Gravatar of ssumner ssumner
    26. November 2023 at 13:55

    Junio, You’d have to be more specific. S&D usually applies in competitive markets, less so when government actions are involved, as they have monopoly power. So I’d say “it depends”.

  3. Gravatar of Junio Junio
    26. November 2023 at 14:22

    I suppose I’ve been thinking about it as this:

    It would perceive nation-states as producers and consumers of power, in which each state-actor has their own supply and demand for power.

    Supply of Power: The Production
    This supply of power resources signifies what a nation-state possesses in terms of asserting itself geopolitically.

    Supply Resources:
    -Military Strength
    -Economic Resources
    -Technological Advancements
    -Diplomatic Influence

    Demand of Power: The Consumption
    This demand for power resources signifies the needs, desires, or strategic aims of a nation-state that seems to enhance its influence, security, or economic interests.

    Demand Resources:
    -Security Enhancement
    -Geopolitical Influence
    -Economic Growth and Stability
    -Access to Vital Resources
    -Projection of Ideological Influences

    Nations seek to obtain power, which would be the commodity here. I’m a little unconfident with everything, but I hope it’s somewhat interesting.

  4. Gravatar of Solon of the East Solon of the East
    26. November 2023 at 15:49

    In general, I agree with this assessment. Although I always suspect living standards in the US are overstated, PPP or otherwise.

    Residential rents and housing costs certainly play havoc. An apartment in Osaka rents for $400 that would cost $2500 in Los Angeles.

  5. Gravatar of Matthias Matthias
    26. November 2023 at 16:23

    Solon, by comparison, the US has a lot cheaper produce and groceries in general than Japan.

    Japan has some very high end produce, but the bottom of their market looks rather sad. (Partially because their imports are restricted.)

    Yes, housing costs in what Kevin Erdmann calls America’s closed access cities are absurd. At least absurd compared to where they could be with adequate building.

    On an absolute scale, our housing costs in Singapore aren’t cheaper, but at least we have a better excuse of having a genuinely high density and tiny landmass, and we are building up and up already.

  6. Gravatar of Scott Sumner Scott Sumner
    26. November 2023 at 18:03

    Junio, Instead of S&D, you might be better off with PPFs and indifference curves.

  7. Gravatar of Junio Junio
    26. November 2023 at 18:14

    Scott, thank you for the advice. I’ll be looking into that.

  8. Gravatar of Ricardo Ricardo
    26. November 2023 at 19:34

    Same experience.Taiwan is one of the most expensive countries in East and Southeast Asia.

  9. Gravatar of Ryan Ryan
    27. November 2023 at 09:19

    Can you do Canada?

  10. Gravatar of ssumner ssumner
    27. November 2023 at 09:50

    Ryan, When I visit Canada, it seems only about 10% below the US. So the IMF’s estimate of $59,813 seems too low.

  11. Gravatar of Christian List Christian List
    27. November 2023 at 14:13

    Purchasing power parities (PPP) indicate how many units of a currency have to be paid for a certain amount of goods and services, right? So essentially a high number simply means that you have to pay more, right? Not the opposite. Goods and services are fixed.

    Now divide the PPP by the nominal exchange rate to calculate the price level, and normalize not to 1, but to 100. So that an idiot like me can understand what you’re talking about.

    This data exists for all countries except Taiwan, which is probably because China doesn’t want data on Taiwan to exist, see World Bank etc. They have 170 countries, but Taiwan is not included because China would not accept it.

  12. Gravatar of DocMerlin DocMerlin
    27. November 2023 at 15:26

    The methodology used for housing costs drastically under-price housing, which is a huge share of expenses. They ask people how much to rent the home they own. People then give a price that is far below market rates. (because people who aren’t looking for a home or to rent out a home, give the rate from some years back when they last checked on prices) Instead they should look at actual home sales and use that.

  13. Gravatar of Ray Lopez Ray Lopez
    27. November 2023 at 16:10

    I’ve said it before on this site (and gotten some push-back from our host) and I’ll say it again: nominal GDP is what matters, not “PPP”. PPP uses an artificial dollar that assumes a pizza in Beijing is the same as a pizza in New Jersey because they have the same ingredients. Nothing could be further from the truth (fun fact: Hawaiian pizza, which my Asian partner likes, was invented by a Canadian Greek chef).

    BTW, if you read this far, I posted online that the casualty ratios between Hamas/Palestine and Israel will be proportional to the nominal per capita GDP between the two areas (based on my original research). That’s a ratio of about 14x and it means Palestine’s Gaza strip will suffer about 17k casualties. Right now it’s about 11k so another 6k to go. I have spoke.

  14. Gravatar of Tacticus Tacticus
    27. November 2023 at 20:07

    I think it’s only fair to note that the FT piece you are referencing is an opinion piece, not a journalistic article or an editorial from the board. Worse than that, it’s an opinion piece by Ruchir Sharma… who seems intent on always making a gigantic ass of himself in his articles. I have no idea why they let him continue to write this drivel or why anyone listens to him at all.

    On a different note, which of these various countries would you prefer to live in? I’d go for Austria, personally…

  15. Gravatar of Tacticus Tacticus
    27. November 2023 at 20:10

    @ Junio,

    No one has any idea of how to measure ‘power’ in International Relations, so trying to apply a supply and demand framework to it is nonsensical, in my mind.

  16. Gravatar of kangaroo kangaroo
    27. November 2023 at 20:27

    Or you can use the general form:

    1. It’s important to use economic data.
    2. Economic data are not always reliable.


    Scott: if you’re still planning on voting Democrat to keep Trump out of office, I really don’t know what to say. Maybe just run and get yourself a foot long veggie at Subway, the Sandwich Shop Monopoly, before Lina Khan and Elizabeth Warren break it up into 4000 individual companies.

  17. Gravatar of Solon of the East Solon of the East
    27. November 2023 at 23:58


    If you are still reading this, Kevin Erdmann recently substacked to the effect that “America has become Los Angeles.”

    If you want affordable housing in America today, you can live in the Rust Belt or Missouri.

    If you want low-cost housing in Japan, you can live in Sapporo. The ski-resort town known for its scenic qualities.

    I wish I was Japanese.

    BTW, I live in Thailand. Probably the Philippine highlands would have been a better idea. So it goes.

  18. Gravatar of Solon of the East Solon of the East
    27. November 2023 at 23:59

    Side question:

    Is it fair to say that Fed policy is, in the main, effected through more or less lending on real estate by commercial banks?

  19. Gravatar of Viennacapitalist Viennacapitalist
    28. November 2023 at 00:09

    glad you appreciate Austria 🙂

  20. Gravatar of ssumner ssumner
    28. November 2023 at 14:07

    Christian, Not sure what you are trying to say . . .

    Tacticus, I’d prefer to live in Japan if I were a foreigner, Austria if I were a local resident.

    Solon, No.

  21. Gravatar of Christian List Christian List
    28. November 2023 at 15:03

    What I tried to say in my broken English:

    PPP indicates the price one has to pay for a specific basket of goods. Or not?

    Doesn’t that mean that the same basket of goods in US dollars costs 80k in the USA, 72k in Taiwan and 23k in China? So China is considerably cheaper than Taiwan? And Taiwan is a bit cheaper than the US. So pretty much it is in line with your experience?

    The ratio Taiwan to China is pretty close to the prices you mentioned: The factor or multiplier is approximately 3. Aren’t these exactly your price differences, that you mentioned.

    In China you paid 5$ for a service that costs 15$ in Taiwan. Or 80 cents in China for food, and then 2$ in Taiwan. The PPP pretty much exactly. A Taiwanese person could buy about three times as much in China (if they had dollars!) and that’s what you’ve experienced. Where am I going wrong again?

  22. Gravatar of Junio Junio
    28. November 2023 at 17:10

    Tacticus, you have a valid point. The definition of power would vary among between schools of thought, but I have a suspicion that although there are practical differences in how they conceptualize power, the general principles or mechanisms in how they analyze power dynamics between state-actors wouldn’t necessarily deviate from each other, aside from the nuance that those differences might provide. Quantifying power, due to it’s qualitative nature, is the more difficult part, I think.

  23. Gravatar of mbka mbka
    28. November 2023 at 18:47


    let me take this one. Also a nice test for me if Scott strikes me down on my understanding of PPP adjustment, ha!

    PPP is an adjustment _to GDP_. It uses a basket of goods. It is not the price of that basket of goods. It is the adjustment factor for exchange rates and prices of goods. The resulting output “GDP at PPP” is now a new GDP that tells you what that particular GDP income is worth in terms of purchasing power inside that country.

    Assume your rice bowl is $1 in China, $3 in Taiwan and $10 in the US at current exchange rates. Assume that NGDP is $10,000 in China, $30,000 in Taiwan and $100,000 in the US at current exchange rates. Assume your PPP adjustment is based solely on the price of that one rice bowl, which is the basket of goods here. Then your PPP adjusted GDP is $100,000 for all three countries because they can all afford an equal number of rice bowls with their respective GDP. They have the same GDP at purchasing power parity and traditionally the new number is levelled up or down to US numbers = PPP adjustment of 1. The PPP adjustment factor in this example is [(US cost of rice bowl)/(country’s cost of rice bowl)]. If next year the Chinese GDP rises to $20,000 and the rice bowl there still costs $1, all at current exchange rates, then the new PPP adjusted Chinese GDP is now $200,000.

  24. Gravatar of Scott Sumner Scott Sumner
    29. November 2023 at 09:50

    Christian, You asked:

    “Doesn’t that mean that the same basket of goods in US dollars costs 80k in the USA, 72k in Taiwan and 23k in China?”

    No, those are GDP per capita comparisons, not price level comparisons. The IMF data implies that Taiwan has a lower price level than China.

  25. Gravatar of dlr dlr
    29. November 2023 at 13:43

    Hi Scott.

    Apologies for being off topic for PPP.

    We now have NGDI for Q3. It grew 5.1% QoQ, annualized, a relative increase from the past few quarters. Here is YoY NGDI from Q122 to Q323:

    10.9%, 10.4%, 9.7%, 6.4%, 5.3%, 3.6%, 3.1%

    Here is the average of NGDI and NGDP YoY from Q122 to Q323:

    10.8%, 10.1%, 9.4%, 6.8%, 6.2%, 4.8%, 4.7%

    Here is YoY AHE of all employees, private for the same quarters:

    5.6%, 5.6%, 5.3%, 4.9%, 4.5%, 4.4%, 4.3%

    It’s true, NGDP grew 8.9% QoQ in Q3, far higher than target. But you have warned that when the statistical discrepancy between NGDP and NGDI was high it is better to focus on either NGDI, an average of GDI and GDP, or wages.

    The Fed didn’t tighten policy?

  26. Gravatar of ssumner ssumner
    29. November 2023 at 18:49

    dlr, It depends what you mean by tighten. I bit less expansionary that in 2022? Yes. Tight in an absolute sense? No.

    They became less expansionary, but didn’t become tight.

  27. Gravatar of dlr dlr
    30. November 2023 at 05:13

    yes but when someone is going 85mph in a 55 and slows to 60mph, you haven’t done a helpful job narrating when all your chapters along the way are titled “the driver continues to speed as ever.” the salience of the change in trend has to be a bigger part of the story. the regime may be highly imperfect, the slowing may be incomplete (i would say this is probably true but debatable given uncertainty about which data to use, what expectations are, and short term potential output), but the clear and substantial decline in speed toward target is not well summarized by a series of “the speed is too high” reminders.

    i just think my favorite monpol blog should have more fidelity to what has actually happened here. yes, the fed created an ad hoc, ill-advised, asymmetrical FAIT regime on the fly. and yes, nominal growth exploded under its watch, and it ran a serious risk of chasing that nominal growth with faux tightening that was *actually* loosening. instead they in fact successfully tightened despite not have the anchor of a level targeting or at least symmetrical regime, and they tightened quite a bit. they are closer to target today than they are to the peak nominal growth. this is a big deal, and kind of surprising given the lack of disruption to real output, and i think you have underplayed it. you don’t?

  28. Gravatar of Solon of the East Solon of the East
    30. November 2023 at 08:56

    3% pce core Oct. on year.

    3% inflation is a problem?

  29. Gravatar of ssumner ssumner
    30. November 2023 at 09:23

    dlr, This might help. Look at NGDP relative to the 4% trend line. In 2021 and 2022, it soared well above trend. How about 2023? Is it getting even further above trend, or is it leveling off at 4%? Or is it returning to the trend line as in NGDPLT?

    Given the flawed FAIT regime, I think by 2023 it was too late to return NGDP to the trend line. But it would have been nice to not see it getting further and further above trend.

  30. Gravatar of dlr dlr
    30. November 2023 at 10:23

    this is just another way of saying the fed failed to, or more accurately chose not to, immediately or even quickly reach its nominal target once it finally seemed to realize it should be tightening. true. but *all* the close counterfactuals are important, not just that one. and it did tighten successfully (inexorably even) if not instantly, seemingly avoiding much worse counterfactuals such as nominal growth plateauing, cycling, or even worsening from high levels. and it made quite a bit of progress with, as yet, minimal real consequences.

    i think it’s very fair to ask if the fed just got lucky. it seems incredibly hard to create an asymmetrical non-level regime that hopes to use as hoc discretion to gradually fine tune nominal expansion downward over many quarters or years before starting a new trend line. but it’s also important to notice that in this case it looks like they have thus far managed to exactly that.

  31. Gravatar of Dr Richard Dr Richard
    30. November 2023 at 12:05


    Will you make a post about the 5.2% Real GDP growth in the Q3 and what it means?

  32. Gravatar of ssumner ssumner
    30. November 2023 at 14:45

    dlr, I just have a much more negative view of Fed policy in recent years than you do. I’ll do a post on this soon.

    Dr. Richard, I already did a post on NGDP in Q3, I typically don’t comment much on real GDP.

  33. Gravatar of Junio Junio
    21. December 2023 at 20:23

    I’ve been tinkering with economics and international relations for the past four weeks; time has passed quickly. I don’t have much to show for it truthfully. It’s been more just thinking about how economic principles may apply. I’m not certain where this will lead me. Possibly nowhere, but it’s raised interesting questions for myself nonetheless. I’m not able to discuss about it with anyone. (I likely wouldn’t feel confident enough about it either way.)

Leave a Reply