by Michael Pettis, China Financial Markets
As analysts and official entities like the World Bank continue to downgrade their forecasts for medium-term growth in China, I have been asked increasingly often for the reasons I believe that 3-4% average annual growth rates is likely to be the upper limit for China during the adjustment period. In this article I want to explain how I arrived at my numbers.
The analysis is fairly straightforward and those looking for a very complex econometric model are likely to be disappointed, but I have always believed that, unlike physics or cooking, anything in economics that cannot be easily explained to an educated layman with 9th-grade algebra and a little bit of calculus is likely to be useless (and if he knows some probability theory, the most beautiful branch of math in my opinion, he is able to soar).
Before beginning I should make two points. First, for many years I assumed that the “adjustment” period would begin shortly after the beginning of the administration of President Xi Jinping and Premier Li Keqiang, that is, from 2013 or 2014, and would run through the presumed end of their term in 2023.
In fact I may have been overly pessimistic. It now seems to me that China actually began adjusting economically, although in a very limited way, in 2012, when we first started to see growth slow as Beijing became increasingly worried about the astonishing increase in debt. This probably occurred because the big wage increases in 2010-11, which were counterbalanced by the sharp drop in real interest rates during that period, were finally able to take effect in 2012 when real interest rates rose sharply once again. Of course whether the adjustment begins in 2012, 2013, or 2014 probably doesn’t matter much to the analysis, but it is good news, I think, that it may have started earlier than I originally expected.
There is, by the way, nothing especially important about ten years except that it is a round number. If China adjusts much more aggressively than I expect, the adjustment period could easily occur within less than ten years, although this is unlikely to be the case because it will be politically difficult to pull off.
On the other hand, if the adjustment period is much longer than ten years, perhaps because the relationship between investment growth and consumption growth is highly positive, or because political opposition is fiercer than expected, growth rates might be a little higher on average in the first few years but the period of stagnant growth would last longer than ten years and there would be a much higher risk of an economic collapse.
As for the second point, I don’t really think of my numbers as being growth predictions. In fact more generally I do not like to make predictions and frankly have no idea of how to go about doing so, especially if it is likely to involve the complex econometric models.
What I prefer to do, and find more useful at least for my way of thinking, is to try develop an understanding of the overall system under clearly specified (I hope) assumptions, and then work through the logic of the system to see what the various outcomes can be. In other words I am not trying to predict what will happen but simply to list the various scenarios that are consistent with the model and to state explicitly what are the assumptions needed for those scenarios to occur. If the assumptions are plausible, then so is the scenario. If not, then they are not.
As I see it, 3-4% is what it takes for my arithmetic to work within plausible scenarios. It is, in other words, the upper limit of the average growth rate that allows me to work out arithmetically the growth in debt, consumption, investment and GDP needed for the amount of economic adjustment that will rebalance the economy by the minimal acceptable amount, without making some fairly implausible assumptions.
This means that depending on how aggressive Beijing is during the reform process, China’s actual growth might be higher if Beijing engineers a much more aggressive program than I think plausible of transferring resources from the state sector to the household sector, thereby forcing up the household income and household consumption shares of GDP. It will be lower if there are more adverse shocks to trade or the financial sector, or if political opposition to the reforms is fiercer than expected. In other words I really think of 3-4% average annual growth as the plausible upper limit of GDP growth, assuming no massive privatization program.
Let me state my assumptions. As everyone now recognizes, rebalancing in China requires that consumption grow significantly as a share of GDP over the next decade or more. China currently reports household consumption as representing about 35% of GDP, which is an almost surreally low number. By how much would consumption have to rise for meaningful rebalancing to have occurred?
Before answering, how meaningful is this 35% number? A number of analysts have regularly argued that the official data seriously understates both income and consumption in China, and so China’s real household consumption is much higher. This may well be true, but I think there are at least three counterarguments. First, while it is true that some consumption is not included in the official data, at the same time there is quite a lot in there that should not count as consumption, or at least not for the purposes of understanding the rebalancing process.
For example three of the fastest growing consumption categories year after year are gold and jewelry, household furnishings, and household appliances. I would argue that all of these should really count as investment, and certainly the latter two will drop dramatically as investment, especially in real estate, drops (and the former will probably drop as the financial repression tax is eliminated). As a rule any item of consumption the demand for which is highly correlated to investment should be treated as investment for our purposes, not consumption. So when analysts point out that a lot of consumption is paid for by businesses on behalf of employees, and so does not show up as consumption in the data, they are right. But they are also largely irrelevant. The consumption that we care about is consumption unrelated to investment, because it is this consumption that must rise as investment drops.
Second, if both consumption and income are understated, as they may well be, this does not necessarily mean that the consumption share of GDP is more than 35%. This would only be the case if the ratio of hidden consumption to hidden income is greater than 35%.
If most of the hidden consumption and income belong to the rich or very rich, as is commonly assumed, it may well be that the true ratio is lower, not higher, than 35%. Many analysts are muddled about the differences between absolute consumption levels and the consumption ratio, and so they believe that if they can show that consumption is higher than claimed by the National Bureau of Statistics, the imbalance is less of a problem. It isn’t. What matters is the consumption share of all that is produced, and if both GDP and total consumption are higher than the official numbers, China’s imbalance is not necessarily better. It may even be worse, and it is the imbalance that matters to China’s growth prospects.
Finally we are not talking about small imbalances here. The reported consumption share of GDP is astonishingly low, perhaps the lowest ever recorded, and so the error in the data would have to be enormous for it to matter to the rebalancing debate. Even if it turns out that 20% of Chinese consumption, and none of its income, were hidden and unrecorded in the NBS data, China would still easily have the lowest consumption rate of any major economy in the world.
What is the right level of consumption and investment?
Globally, consumption represents a fairly stable 65% of GDP. Over the past decade this average has encompassed a group of high-consuming countries, such as the United States and peripheral Europe, whose average consumption exceeded 70% of GDP, as well as a group of low-consuming countries, mainly in Asia, whose average consumption, excluding China, ranged from 50% to 58% of GDP.
This distribution of over- and under-consumption should change in the next few years. It is unlikely that the high-consuming countries will be able to maintain their excess levels of consumption for the rest of this decade, and indeed their consumption rates have already come down substantially, with more probably to come. Peripheral Europe is in crisis, and the United States is taking steps to raise its savings rate so as to reduce its current account deficit. Japan, although already a relatively low-consuming country, is also likely to try to increase its savings rate so that its massive debt can be funded by patient domestic savers, rather than by impatient foreigners.
This means that the rest of the low-consuming countries are also unlikely to be able to keep their consumption levels as low as they have in the past. A world with low-consuming countries requires high-consuming countries in order to balance.
If global consumption drops in the high consuming countries, with no corresponding rise in the low-consuming countries, it is unlikely that investment will rise quickly enough to replace it (why invest if no one is going to buy the output?), and so the global economy must respond with enough of a contraction in GDP to maintain consumption at roughly 65%. The great consumption and savings imbalances of the past that led to the current crisis, in other words, have to adjust. This means that if there are no longer large economies consuming 70% of more of their national income, the world is unlikely to be able easily to accommodate large economies consuming just 50–58% of their national incomes.
Let us assume, nonetheless, that the world can accommodate a minimal amount of Chinese rebalancing. Within a decade Chinese household consumption, according to this assumption, will rise to no more than 50% of GDP, as difficult as it will be for the world to accept such low consumption from its second-largest economy. This will mean that China still produces far more than it absorbs – especially if investment growth were to come down sharply, and it would mean that the rest of the world would be forced to absorb excess Chinese production without resorting to trade intervention.
For the sake of completion let us make a second assumption that, because the world is unable to accommodate such a low consumption level for the world’s second largest economy, global pressure forces an even more dramatic change in Chinese household consumption so that it rises to 55% of GDP, instead of 50% as in our first assumption, in ten years. Both of these assumptions can be modeled in a way that combines consumption and GDP growth to arrive at the desired outcomes, and I will ignore the possibility that if the world forces China to raise its consumption rate to 55% in ten years, this will probably happen through negative growth and trade disputes, thus making all my numbers overly optimistic.
So much for our assumptions about the consumption rate – the second set of assumptions involves investment. Currently China has the highest investment share of GDP in the world, with investment comprising 46% of GDP or more, depending on how it is calculated. A 2012 IMF paper that I have cited in earlier issues of this newsletter shows investment as high as 49% of GDP, and it calculates excess investment, i.e. the spread between actual investment levels and the level predicted by an historical model, which began growing around the year 2000, as being 5-10 percentage points.
What is the appropriate level for a country like China? A number of studies have examined other high-investment developing countries during their growth miracle stages, and for most of these countries investment peaked out briefly at 35-40% of GDP (in Malaysia, Thailand and Singapore investment did at one point exceed 40%, but in each case only for a very brief period). In emerging markets investment is typically around 30% of GDP.
How should China compare to these countries? Some analysts argue that China, a very poor country, suffers from a capital stock that is too low, and so the optimal investment level should be much higher. This reasoning, I hope I was able to show in my June blog entry, is based on a fallacy. The optimal amount of investment for any country depends not on how far it is from the capital frontier but rather on its level of social capital, and this implies that very poor countries should optimally have lower levels of capital stock per capita than richer countries.
China’s capital stock per capita, for example, is higher than that of Mexico, and much higher than that of Russia and Brazil, three other large developing countries that are substantially richer than China and whose workers are more productive. When analysts say that China’s capital stock is relatively low, they are completely befuddled. It is low compared to the richest and most productive countries in the world, as it should be, but it is high compared to other developing countries, and even compared to developing countries with much higher levels of productivity
Because investment shares in other developing countries peaked out at 35-40%, some analysts argue that this is the appropriate level of investment for China. There are of course a number of problems with this argument but two stand out especially. First, the countries for which investment peaked out at 35-40% of GDP nearly all had subsequent periods of very difficult adjustment, with burgeoning growth in debt and either sharp economic contraction or many years of very slow growth, during which periods the investment share of GDP dropped substantially.
It is not at all clear, in other words, that for these countries 35-40% was the optimal investment share of GDP. This was probably already too much investment because in many if not most cases it was subsequently followed by many years of low or even negative growth, probably as the economy was forced to grind its way through the debt associated with the excess investment. The optimal level, in other words, was probably much closer to 30%.
Second, even if 35-40% was somehow the optimal level for China all along, investment in China has substantially exceeded this level for many years, so it seems obvious that an appropriate adjustment should mean not that investment drops from 46% of GDP to 35-40%, but rather that is drops to something well below 35% for many years before returning to the “optimal” level. This, I think, is just common sense.
Making the implicit explicit
So I assume that investment must drop as a share of GDP. One way or another, either because Beijing forces changes in the growth model, or because Beijing does nothing and allows debt to build to the point where debt capacity constraints are breached, after which investment collapses automatically and the investment share of GDP will drop substantially.
How long will it take? I am going to assume Beijing has ten years to bring investment levels down to the new “optimal” level just to make my calculations easier, but as in the case of taking ten years for consumption to adjust, I think this is an heroic and frankly implausible assumption. Debt levels are simply too high in China for it to continue this level of investment growth for so many more years.
To repeat the exercise, then, let me make two separate assumptions – that investment will drop to 40% of GDP in ten years and that investment will drop to 35% of GDP in ten years. In either case I will assume that investment is currently 46% of GDP, although it is probably closer to 49%.
It turns out that it is fairly simple arithmetic to work out the implications of each of these assumptions relative to GDP growth. Rather than start with growth assumptions in consumption and investment and use these to determine what the corresponding GDP growth rate is likely to be, I thought it would be more useful if I reverse the process and simply assume a bunch of GDP growth rates ranging from -2% to +10%. These are the different average GDP growth rates possible under different scenarios for the next ten years.
We will assume two sets of adjustments for investment and consumption. The “easier” adjustment scenarios have household consumption growing from 35% of GDP to 50% of GDP, while investment declines from 46% of GDP to 40% of GDP. The “tougher” adjustment scenarios have household consumption growing from 35% of GDP to 55% of GDP, while investment declines from 46% of GDP to 35% of GDP.
The table below lists the consumption and investment growth rates needed for rebalancing to take place at each of the highlighted GDP growth rates.
Table: GDP, consumption, and investment growth in a rebalancing China
|Assumed GDP growth rate||Consumption growth as the consumption share rises from 35% to 50%||Consumption growth as the consumption share rises from 35% to 55%||Growth in investment as the investment share drops from 46% to 40%||Growth in investment as the investment share drops from 46% to 35%|
To read the table, let us start by assuming, as an example, that we believe the average GDP growth rate over the ten-year period will be 6%. For China to do a minimal amount of rebalancing that gets consumption to 50% of GDP and investment to 40% of GDP, we can quickly figure out what the corresponding growth rates of consumption and investment must be. Consumption must grow by 9.9% a year and investment must grow by 4.5% a year to get us there.
Notice the reason why I do it this way rather than the “normal” way most other economists would. Instead of estimating what I expect the growth rates in consumption and investment will be, and then calculating the implicit GDP growth rate from those numbers, I start with an assumed GDP growth rate and then calculate what the implicit growth rates in consumption and investment must be in order for rebalancing to take place. I am not making predictions, in other words. I am simply working out logically what any GDP growth rate must imply in terms of consumption and investment growth rates in order for China to rebalance.
This allows me to make statements like this: If you think that China’s GDP will grow by 7% a year over the next decade, and if you expect a minimal amount of rebalancing, then you are implicitly predicting that consumption will grow by 10-11% a year for ten years and that investment will grow by 4-5.5%. If you believe these two implicit predictions are plausible, then your 7% prediction is also plausible.
Trade is a residual
Notice of course that for the changes to work we are implicitly assuming that the GDP share of the sum of other consumption (government and business) and the current account surplus changes automatically to allow the equation to work. So if consumption rises from 35% of GDP to 50% of GDP, for example, while investment falls from 46% of GDP to 40% of GDP, the other sources of demand (mainly other consumption and the current account) must have reduced their share of GDP from 19% to 10%. This probably means a sharp contraction in the country’s current account surplus and perhaps even a current account deficit.
I want to state again that these numbers are not predictions. They are simply the arithmetically necessary growth rates that are consistent with our assumptions. To return to the interpretation of the table, let us assume again that China does the minimal amount of rebalancing so that in ten years household consumption is 50% of GDP and investment is 40% of GDP, what are the investment and consumption growth rates consistent with, say, 6% GDP growth, and are they plausible?
It turns out that average GDP growth rates of 6% require, as an arithmetical necessity, that household consumption grow by 9.9% a year over the next ten years and that investment grow by 4.5%, after many years of high double digit growth and more recently growth in the low double digits. Is this plausible?
I would argue that positive investment growth rates for another ten years are highly likely to result in our reaching debt capacity constraints well before the end of the decade, so I am skeptical about the investment implications of this scenario. By the way some analysts have mischievously pointed to the very poor construction quality in China to argue that investment growth rates have to stay high just in order to account for higher-than-estimated depreciation costs, and that this suggests that China can grow faster than what we might otherwise assume.
This of course is nonsense. The fact that buildings and infrastructure are poorly constructed means that China is worse off, not better off, and the investment projects will ultimately be required to generate sufficient returns to pay off even more debt than originally estimated. Because it is debt capacity constraints that constrain investment, anything that creates debt without creating additional productivity to service the debt cannot possibly be a solution. Higher-than-expected depreciation increases debt relative to debt-servicing capacity.
I would also argue, more importantly, that if annual investment growth drops to 4.5%, and GDP growth to 6%, it will be very difficult, without significant and politically painful transfers from the state sector to the household sector, for consumption to grow at anywhere close to 9.9% a year for ten years. Consumption growth is, after all, positively correlated with investment growth, especially in the internal provinces upon which a lot of useless investment has been lavished.
In order to get Chinese households to increase their consumption by nearly 10% every year, I would argue that household income would have to grow at that rate, which means that wages, interest rates, and the value of the renminbi should in the aggregate increase rapidly to get consumption to rise fast enough, and of course since it is precisely low wage growth, low interest rates, and an undervalued currency that goose GDP growth, reversing them is not consistent with high GDP growth.
Can consumption grow at close to 10% for ten years while household income grows much more slowly? Yes, of course it can, if the household savings rate declines, but as China’s economy slows and as concerns about debt rise, it seems to me a tad optimistic to assume that the household savings rate will decline sharply. Rising income and rising uncertainty both suggest that we should expect higher, not lower, household savings rates, which in turn imply that household income must grow faster, not slower, than household consumption.
All of this suggest to me that while 6% GDP growth for the next ten years might not be impossible, it is extremely unlikely because it requires what are to me implausible assumptions about the ability to maintain and increase already-high levels of investment without increasing the debt burden unsustainably and about the rise in the growth rate of household income as both GDP and investment growth drop sharply. This is why even 6% annual GDP growth rates, which are still lower than most current growth projections for China, are implausibly high, in my opinion.
What about if you believe that reducing investment is a much more urgent priority than raising consumption? In that case you might argue that China can grow at 6% while the household consumption share of GDP rises to 50% and the investment share of GDP declines to 35%.
In that case you are implicitly assuming that household consumption will grow on average by 9.9% a year for ten years while investment grows by 3.1% a year. Is this possible? Of course it is. Is it plausible? Again, only if you believe that investment growth can drop sharply while the growth in household consumption rises to nearly 10% a year for ten years.
So what is plausible? My working assumption, which I acknowledge is probably still optimistic, is that somehow or the other Beijing can keep household consumption growing at around 7-8% a year, even with a sharp decline in the investment growth rate and with the pressing need to clean up the banking system (and remember that traditionally, in China and elsewhere, cleaning up the banking system always means finding ways of getting the household sector to pay for the losses).
I know many consider assumption this to be a little optimistic, but if Beijing is worried about the social implications of adjustment, this is probably the target it will need to meet, and Beijing can do so even with much slower GDP growth if the leadership implements mechanisms that transfer wealth from the state sector to the household sector. I discuss why this is the right growth rate for to target in more detail in a recent piece published on the Carnegie Endowment website and in an OpEd piece in the Financial Times.
The table above shows that if China is to do the minimal amount of rebalancing, which requires that the world accommodate for another ten years large Chinese trade surpluses, and that debt can continue to grow – quickly but at a lower rate than in the past – for another ten years without pushing China up against its debt capacity constraints, 7-8% growth in household consumption is consistent with roughly 3-4% growth in GDP. It is also consistent with more or less no growth in investment, which would after ten years bring the investment level down to 35% of GDP.
These numbers are, I think, plausible if still a little optimistic. This is something, in other words, that I think Beijing can reasonably pull off – if it is able to manage political opposition from the domestic elite – because they can transfer resources from the state sector to the household sector at a pace necessary to keep the growth rate of household income and household consumption fairly high. However GDP growth rates significantly above 3-4%, I would argue, require assumptions that are unlikely to be met unless Beijing is able radically to transform its attitude to state ownership and the power of the elite, and so embark on a major transfer of assets from the state to the household sector.
This is why I have argued since 2009 that that 3-4% average GDP growth for a decade is likely to be the upper limit once Beijing seriously begins to rebalance the Chinese economy, and if the administration of President Xi and Premier Li is able to pull this off, it would be a huge accomplishment. China would rebalance substantially, the problem of debt would have been managed relatively well, and the income of average Chinese households will have nearly doubled over the decade. The key assumption, of course, is that in the face of a sharp drop in investment, Beijing is nonetheless able to maintain current high levels of consumption growth.
Before closing it is worth pointing out that many analysts have told me that they do not think it is possible for household income growth to exceed GDP growth for many years. But why not? After all state income growth exceeded household income growth for many years, and if Beijing reverses the mechanism that accomplished this – albeit with political difficulty – it can reverse the relative growth rates. More importantly, Japan did just this after 1990, when GDP grew by around 0.5% annually but household income and household consumption grew by between 1% and 2%. The US did this too in the early 1930s when, if I remember correctly, household income and household consumption dropped by a lot less than GDP (around 35%) and investment (around 90%).
But notice these two examples. One occurred under conditions of no growth and the other under conditions of negative growth. Severely unbalanced systems always rebalance in the end, but the process of rebalancing is rarely easy.