One of the most frustrating aspects of any discussion about China’s economic prospects is in the confused way with which economists and analysts, who often seem to suffer from what Giuseppe Gabusi calls “GDPism,” employ the concept of GDP. We all think we know what the term is supposed to mean. It is generally assumed that GDP is the total value of all goods and services produced by an economy, so we think of it as a measure of wealth, or as a measure of debt-servicing capacity, and we assume that it is a measure that can be compared across countries.

But it isn’t a measure of wealth or of debt-servicing capacity, and GDP comparisons across countries are often meaningless. GDP measures certain types of economic activity that we agree to define as GDP, but these measures are conditioned by the institutions that mediate economic transactions. Among those institutions, for example, are the accounting conventions used to recognize the value of inventory, or to write down investment, in such a way that reported GDP is artificially changed by these conventions.

GDP Requires a Mark-to-Market Mechanism

To explain why we need to tease out the full consequences of this insight to understand the Chinese economy, I will go through a simple exercise to arrive at conclusions which are, or should be, obvious but which are nonetheless ignored in many discussions about the Chinese economy.

Point 1: GDP does not directly distinguish between activity that increases a country’s wealth and activity that doesn’t.

Consider the case of a country in which there is a corporate entity, which we will call China Corp., that decides to invest $100 in each of two separate projects. One project, Value Bridge, is economically justified—it could be, for example, a bridge that relieves traffic congestion significantly and so sharply lowers manufacturing costs for a neighboring industrial park. The other project, Nowhere Bridge, turns out to have no economic value—a bridge to nowhere, for example, that receives no traffic.

In either case, China Corp. ends up with two $100 investments on the asset side of its balance sheet. The counterbalancing credit shows up either as a $200 reduction in cash or a $200 increase in debt (in this case, let us assume the latter). Net assets are unchanged, and there is no impact on the company’s income statement, although in the future there will be, as interest is paid on the debt and the bridges begin to generate revenues. If the revenues are greater than the debt-servicing costs, these remaining revenues will continue to be added to GDP in the future.

In the current period, Value Bridge and Nowhere Bridge contribute to an increase in economic activity in exactly the same way. They both cause GDP to rise by some amount equal to the $100 value of the investment times some multiplier, which, to make our lives easier, we will assume is one. The amount by which the productive investment boosts the GDP calculation, in other words, is exactly the same as the amount by which the nonproductive investment boosts the GDP calculation.

The point is that when the investment is made, the GDP calculation has no way to distinguish between the productive investment project and the nonproductive investment project, even though the former makes the economy richer by contributing to its productive capacity while the latter does not. Both productive and nonproductive investments boost reported GDP in the same way.

Point 2: The mechanism by which GDP is forced to record real changes in productive capacity is some form of marking-to-market mechanism.

Let us now assume that China Corp. recognizes that Nowhere Bridge is a bridge to nowhere and creates no economic value at all, so it must take a full write-down on the asset. The Nowhere Bridge investment project, which appears on its balance sheet as an asset, is then written down to zero in the form of a $100 credit to the balance sheet. The counterbalancing debit is made in the form of an expense that reduces the net profits of China Corp. by $100.

When we use the income approach to calculating GDP, the writing down of the asset associated with Nowhere Bridge causes a $100 loss to China Corp., and so reduces its retained earnings by $100. This reduces GDP by lowering the value-added component by that amount.

This is how the GDP calculations adjust in a market economy to recognize bad investment decisions. They are written down and the loss is recognized in the income statement. This is the mechanism by which the GDP calculation is forced to reconcile the accounting records with real value creation in the economy.

Notice how this happens: at the time the investment is made, GDP is boosted by the $100 investment in Value Bridge and is also boosted by the same amount by the $100 investment in Nowhere Bridge. When the investment in Nowhere Bridge is recognized to have created no value, the Nowhere Bridge asset is written down on the balance sheet by $100, the amount of the investment, and there is a counterbalancing $100 debit to the income statement that causes a reduction in the net profits of the business sector.

This reduction in business profits reduces the value-added component of the GDP calculation, and the economy’s reported GDP conforms to the underlying economic reality. In a market economy, the value of the investment in Value Bridge increases GDP whereas the value of the investment in Nowhere Bridge does not.

Point 3: In an economy in which there is no mark-to-market mechanism, reported GDP will be higher by the amount of wasted investment.

Let us now assume there are two countries, China A and China B. The two resemble each other in every respect except that in China B, there is no formal mark-to-market mechanism because all loans are guaranteed by the government sector. In each country, there is a company, China Corp., that makes the two investments described above.

In both countries, in other words, there is a $100 investment in Value Bridge and a $100 investment in Nowhere Bridge, and in both cases GDP is increased by the investment amount. The difference is in the accounting treatment of the write-down. In China A, as described above, the investment in Nowhere Bridge is written down to zero, and the accompanying loss reduces China Corp.’s retained earnings, which removes the impact of the investment on China A’s GDP. In China B, however, the write-down does not occur, so the investment in Nowhere Bridge remains on the books at $100, and there is no $100 loss on the income statement that reduces retained earnings by $100.

Remember that in terms of underlying economic reality, including the value of all goods and services produced by the economy, China A and China B are identical, but when it comes to reporting economic activity, there is a difference between the two. China A’s GDP will be lower than China B’s GDP by the amount of the write-down. What’s more, China Corp. in China B will have $100 more in assets than its counterpart in China A, and this will be matched on the other side of the balance sheet by $100 more in retained earnings.

Although the two economies are identical, in other words, China B will have a higher GDP than China A. More generally, in economies that do not recognize investment misallocation, and that do not write down overvalued assets, reported GDP will be higher than it otherwise would be, as will the reported value of total assets.

Point 4: Even if wasted investment is not formally recognized and written down, it will not create a permanent increase in GDP.

This difference, however, is not permanent. Over time, the difference in the recorded value of the two countries’ assets will be amortized to zero, typically because China B will record a higher depreciation expense than China A, so in the future China A’s businesses will record higher net earnings than China B’s, and its assets will depreciate more slowly, until eventually the financial statements of both entities will once again be identical and their GDP levels will be the same. I should quickly note that to make it easier to understand, I am assuming away the impact of financial distress and other balance sheet effects, but in fact these effects will reduce China B’s real and reported earnings even further.

The net result is that during the period in which China A and China B are investing nonproductively on a sufficiently large scale, China B will always show a higher level of GDP than China A, and total assets held by China B’s residents will always have a higher reported value than total assets held by the residents of China A. At some point in the future, however, their relative positions will be reversed, and China A will begin to report higher GDP growth than China B, until eventually the two countries have identical balance sheets and income statements.

How to Read China’s GDP Data

But remember that the two countries are identical. They produce identical amounts of wealth and value. It is only the failure to write down assets that causes an economy to record higher GDP than otherwise. This means that if one believes that China has misallocated investment and has failed to write much of it down, then one has no choice but also to believe that:

  1. Chinese GDP is overstated every year by the net amount of wasted investment made during the year that has not been correctly written down. If, for example, a government entity borrows $100 to invest in a project that over the rest of its life causes only $60 worth of additional goods and services to be produced, China’s reported GDP will be overstated by $40.
  2. At some point in the future, when China begins explicitly or implicitly to recognize and write down bad debt, this overstatement will be reversed and the country’s reported GDP will be understated by the amount of implicit amortization.
  3. Chinese wealth is also overstated by the amount of wasted investment made during the year that has not been correctly written down. In the example above, not only is China’s reported GDP overstated by $40, but Chinese entities collectively claim to have $40 more wealth than they actually do.
  4. When China begins explicitly or implicitly to recognize and write down bad debt, this is simply the obverse of assigning the losses to one economic sector or another, and it will happen one way or another. Put differently, as the bad debt is eventually written down, households, businesses, and/or government entities will discover that they are poorer than they thought by $40. This is why, as I have written many times, the process of deleveraging, which includes writing down bad debt, consists of nothing more than assigning debt-servicing costs to one economic sector or another.
  5. Notice the impact on savings. GDP is equal to consumption plus savings, and if reported GDP overstates the value of economic activity, the reported savings rate also overstates the value of savings by the amount of wasted investment in each period. This also means that the value of the stock of Chinese savings is overstated by the amount of investment and bad debt that must be written down.

Perhaps Growth Has Already “Collapsed”

Think about the following three countries. All of them have had a decade of 10 percent annual GDP growth driven by high investment growth rates and boosted by highly inverted balance sheets, with debt growing from nearly zero to 60 percent of GDP. But now, in all three countries, we have reached the point at which the return on additional investment is negligible. Here is what follows in each country:

  1. In Country A, investment growth drops to zero, driving annual GDP growth down to 2 percent, where it remains for the next five years. Debt also grows by 2 percent to remain steady at 60 percent of GDP.
  2. In Country B, the authorities have a growth target of 6 percent and encourage government entities to keep investment levels high enough to reach the target, even if this means an explosion of debt. For the next five years, GDP grows by 6 percent, while debt soars from 60 percent of GDP to 260 percent of GDP.
  3. Finally, in Country C, the authorities have implemented significant institutional reforms aimed at making the economy more market-driven and these reforms have caused a massive redistribution of wealth in such a way that consumption growth surges. The growth in consumption encourages additional growth in investment so that GDP growth drops to 6 percent and remains there for the next five years, with debt growing at the same pace to remain steady at 60 percent of GDP.

Clearly, China most closely resembles Country B, but which of the other two countries more closely resembles Country B? Many economists are so impressed by GDP figures as the most fundamental description of the underlying economy that they don’t really distinguish between Country B and Country C, and because Country C enjoys healthy growth, they assume that Country B does too. One of the more consistently confused figures has been Stephen Grenville, a former Australian central banker, who recently said:

The China bears have been around for years, continuously predicting the end of China’s stellar growth story . . . So far, so good. China’s unsustainable double-digit growth came to an end in 2008, coinciding with the global financial crisis. Growth was artificially stimulated for a couple of years but only by huge fiscal and financial stimulus. Since then growth has settled to a still-outstanding 6.5%.

If you assume, as Grenville clearly seems to believe, that there is no relationship between the growth in economic activity and the growth in GDP, then he is right to describe China’s current GDP growth as “still-outstanding.” But that seems a pretty astonishing assumption. If GDP growth had not been artificially boosted by credit expansion, then it is hard to understand why Beijing has been trying urgently to get credit growth under control for over five years but has not even been able to prevent it from accelerating.

In fact, I would argue that “the end of China’s stellar growth story” has already occurred, and occurred quite a long time ago. Growth in the Chinese economy has collapsed, but growth in economic activity has not collapsed (let us assume, with Grenville, that somehow the reduction in GDP growth from over 10 percent to 6.5 percent does not represent a slowdown in economic activity). The growth in economic activity has instead been propped up by the acceleration in credit growth and by the failure to write down investments that have created economic activity without having created economic value. In that case, high GDP growth levels simply disguise the seeming collapse of underlying economic growth in a way that has happened many times before—always in the late stages of similar apparent investment-driven growth miracles.

But there is no way to get around the logic of debt: either the debt proceeds went to fund a productive investment, in which case debt-servicing costs are fully covered by the additional productivity generated by that investment, or they were not. If they were not, the debt was created either to fund an expense or to fund an unproductive investment, and in either case the associated debt-servicing cost must be assigned to some other economic entity.

Most of the remaining so-called bull scenarios for China implicitly assume that existing losses on investments that have not been correctly written down will never be recognized. At first, these bull scenarios had the virtue of consistency—the bulls believed that there was no need to recognize the existence of unrecorded losses because these losses in fact did not exist.

Later on, as it became obvious to everyone that there certainly were unreported losses on Chinese balance sheets that emerged from the failure to write down unproductive investment, and that these losses were significant, the bull scenarios began to incorporate implicit assumptions—implicit but never acknowledged by the bulls—that some mechanism or the other would permit the economy never to have to recognize these losses. The most popular form of these mechanisms typically involved some kind of debt monetization or a return to financial repression, but as I have shown many times before, all these do is pass on the losses in hidden ways to the household sector. In that case, the losses are eventually recognized, and in fact recognized in the most damaging way possible to the economy in the long run.

A Digression Into Accounting

I don’t pretend to have any great knowledge of GDP accounting—what I describe above is really quite basic stuff—but for those who are interested, what is seen as the correct way to record a debt transaction in which the proceeds are used to fund an unproductive investment is to record the expenditure as an expense. In that case, net earnings are reduced by the amount of the wasted spending, and net assets are reduced by the amount of the debt. Typically, this is done in two stages:

  1. The economic entity borrows some amount, say $100, and uses the proceeds to make an investment. This will have no impact on its income statement, but both its assets (in the form of investment) and its liabilities will rise by $100, with no impact on its net wealth.
  2. When the $100 investment is recognized to have no value, the economic entity will write down the value of the asset by $100, with no change in debt, of course, and its expenses will rise by $100. In that case, both net earnings and net assets are reduced by $100.

The two steps together are the equivalent of recording an unproductive investment as an expense. But, as I show above, if the economic entity fails to write the unproductive investment down to its true economic value, the second stage is avoided. This changes the way the unproductive investment was recorded without changing in any way the underlying economic entity.

China’s GDP: Different Than Assumed

The reason I have initiated this discussion of GDP is to explain why we systematically misinterpret the information provided by Beijing’s regular GDP data releases. On July 17, 2017, China’s National Bureau of Statistics reported that, at 38.1 trillion renminbi (or $5.7 trillion), China’s GDP for the first half of 2017 at comparable prices was 6.9 percent higher than it had been in the first half of 2016. This came in slightly above consensus expectations of 6.8 percent.

Analysts and investors treated the news of higher-than-expected GDP growth in much the same way they would if it had been about Brazil, a European country, India, Japan, or the United States. The Shanghai-Shenzhen CSI Index, for example, reacted positively, rising 2.3 percent over the next three days. Several economists also saw July’s data release as good news.

But I have long argued that when reported growth comes in above consensus expectations, the implications for China’s economic prospects are not the same as they would be for most other large economies. In the latter cases, higher-than-expected growth might suggest that we should revise upward our longer-term growth expectations because the underlying economy is performing better than expected. In China, however, when reported growth comes in above consensus expectations, it does not imply a stronger economy. The higher demand that drove growth is unlikely to have been a consequence of underlying health—rather, it is far more likely to have been created by a temporary increase in economic activity in response to government decisions to maintain high levels of GDP growth.

This means debt will have grown faster than it otherwise would have. Higher-than-expected GDP growth should therefore suggest that we revise downward our longer-term growth expectations. It simply means that a higher level of unrecorded losses must be written down in the future and, because it implies more debt than otherwise, financial distress costs in the future will also be higher.

We typically use GDP as if it were an expression of the total value of goods and services produced by the economy, but it clearly is no such thing. For those who are interested, Diane Coyle has written a gem of a book about the development of GDP (GDP: A Brief But Affectionate History), which goes a long way toward explaining how GDP can be useful as a measure for the size of an economy and when it isn’t useful.

GDP can be a fairly good proxy for the total value of goods and services produced by the economy, one that allows us to compare two economic entities or to calculate the growth rate of an economic entity. But this is only true under fairly easily defined conditions. One of the key conditions is that the entities being compared must have consistent mark-to-market mechanisms. The reported GDP of the two economic entities cannot be meaningfully compared if in at least one of them there is a significant amount of misallocated investment that does not generate enough productive capacity to justify the cost of the investment. I discussed this further in a May 2014 essay explaining why the PPP adjustment for China—which begins with an assumption that China’s GDP is calculated in a way that is consistent with the U.S. calculation of GDP—is a thoroughly meaningless exercise.