by Michael Pettis, China Financial Markets
I typically think of four main factors that determine whether or not an economy can function efficiently as a single currency or economic zone. They primarily determine the level of adjustments costs and how these costs are to be shared, so that regions more able to bear the costs absorb a larger share of those costs:
- Labor mobility. If workers can move from one part of the economy to another with low frictional costs (which range from legal restrictions to transportation to language and other social barriers), this reduces the distortions caused by differing growth rates in different parts of the economy. Workers move easily from where they are less valuable to where they are more valuable, reducing downward pressure on wages in the weaker parts of the economy and increasing the value of labor overall.
- Capital mobility. The same is true if capital can move easily from one part of the economy to another. Declining prices and costs in the weaker parts of the economy should attract investment from the stronger parts.
- Fiscal policy. Mechanisms that allow an entity (usually the government) to transfer wealth from rapidly growing sectors to more slowly growing sectors – most obviously income taxes used to fund unemployment benefits – help reduce the disparity between sectors that are growing at very different speeds. This helps stabilize the economy overall.
- Monetary policy. The transmission of monetary policy should be consistent both in timing and effect so that interest rates reflect the needs of the different parts of the economy. This is always hard to do unless each economic entity imposes strict capital controls and has its own discrete monetary policy, and while capital controls are hard enough for a country to impose successfully, they are much harder to impose at a sub-country – e.g. provincial – level.
Although I think China is clearly much more integrated as an optimal currency zone than Europe is today, it is probably less integrated than the US (I will use the US and Europe as the two extreme cases between which China falls). China of course does not have the problems of multiple sovereignty and taxation that Europe does, but there are still important frictional costs among provinces and regions that exceed those among US states and regions and that may make an adjustment to slower growth bumpier than expected.
One of the most obvious such frictions is in labor mobility. Even on paper China is far from being a single, unitary labor market. Some of the constraints on labor mobility are the typical constraints for any large country, for example the costs involved in moving from one place to another very different place – cultural differences, travel expenses, even language barriers – but in China there is an additional important legal constraint: the “hukou” system, or system of residential permits. The always-useful Wikipedia describes it like this:
A hukou is a record in the system of household registration required by law in the People’s Republic of China…A household registration record officially identifies a person as a resident of an area and includes identifying information such as name, parents, spouse, and date of birth.
This matters especially for China’s huge population of migrant workers. Technically in China a worker’s ability to work in any part of the country is constrained by the hukou system.
In 1958, the Chinese government officially promulgated the family register system to control the movement of people between urban and rural areas. Individuals were broadly categorised as a “rural” or “urban” worker. A worker seeking to move from the country to urban areas to take up non-agricultural work would have to apply through the relevant bureaucracies.
The number of workers allowed to make such moves was tightly controlled. Migrant workers would require six passes to work in provinces other than their own. People who worked outside their authorized domain or geographical area would not qualify for grain rations, employer-provided housing, or health care. There were controls over education, employment, marriage and so on.
The effect of the hukou system is to create a multi-tiered caste system for workers in which there are more desirable and less desirable hukous, usually based on differences in urbanization, growth and income:
Reformation of hukou has been controversial in the PRC. It is a system widely regarded as unfair by citizens of the PRC, but there is also fear that its liberalization would lead to massive movement of people into the cities, causing strain to city government services, damage to the rural economies, and increase in social unrest and crime. And yet, there has been recognition that hukou is an impediment to economic development. China’s accession to theWorld Trade Organization has forced it to allow reformation to hukou in order to liberate the movement of labor for the benefit of the economy.
…Chan and Buckingham’s (2008) article, “Is China Abolishing the Hukou System?” argues that previous reforms have not fundamentally changed the hukou system, but have only decentralized the powers of hukou to local governments. The present hukou system remains active and continues to contribute to China’s rural and urban disparity.
Most migrant workers in large cities are technically there illegally because they do not have the appropriate hukou. This means among other things that they have limited access to social services, including health care and schooling for their children. They also have limited legal standing in case of conflicts with their managers and bosses, including conflicts over late pay (in China wages for migrant workers are often paid months in arrears).
While most Chinese recognize that the hukou system is both unfair and detrimental overall to the economy, there is, not surprisingly, opposition to real reform, which would anyway be extremely disruptive without significant changes in the abilities of municipal governments to fund themselves. Eliminating the hukou altogether, for example, would require that workers who are currently non-hukou residents be treated the same way as hukou residents, so that logically city governments would have either to increase their revenues commensurately, or to reduce the services they currently provide to the latter. Almost certainly this would cause social service costs in cities with large migrant worker populations, including health and education, to surge, and so the largest and fastest growing cities strongly oppose real hukou reform.
Ordinary citizens with “upper caste” hukou also oppose significant reform. In China, as in most places, there is prejudice towards migrant workers, who are blamed for crime, low wages, dirty neighborhoods and drug usage. I would imagine that when overall economic growth slows and unemployment rises, tension and conflicts between those with “good” hukou and those without would only deteriorate. I would also assume that if unemployment were to rise, local government officials would naturally take steps to increase discrimination against outsiders.
As of now China effectively has a reasonably high level of labor mobility, especially because most migrant workers have moved from their villages to cities still within their own provinces, and so provincial leaders are responsible for them anyway. For this reason the default assumption among most analysts is that China has a reasonably high degree of labor mobility.
The interesting question however should be how the system would respond to conditions of much lower growth, higher unemployment, and potential labor tension. In my classes at Peking University I teach that government officials, especially regulators, should have the same pessimistic mind-set as bond investors. Instead of thinking about all the ways things can go right, they must think about what their constraints will be when things go wrong.
When we think about how labor mobility in China will affect a slowing economy, in other words, we should not assume that we will have the same level of labor mobility as we have had during the past several years. When there are plenty of jobs available and leaders of fast growing cities and provinces are eager to encourage immigration, the hukou system may work differently than when officials are worried about unemployment in their jurisdictions. We cannot say for certain, especially when the response of Beijing is uncertain, but it is certainly plausible that the hukou system will be used to reduce labor mobility if slowing GDP growth rates are associated with higher unemployment.
In China much if not most “fiscal” activity actually occurs through the banking system, with bank loans playing the role of fiscal expenditures and low or even negative deposit rates playing the role of fiscal revenues. Of course in a country whose financial system is dominated by banks, monetary policy is conducted largely through the banks. This means that capital mobility and the transmission of both fiscal and monetary policy is determined at least in part by the structure and mobility of the banking system.
China’s banking system is divided very broadly into national banks and local and provincial banks, with each side comprising roughly 50% of the total. In the aggregate, local banks, which include city banks and rural cooperatives, are generally believed to be much weaker than the national banks. Their relatively smaller branch networks force them to rely more heavily than their national rivals, with their large retail branch networks, on corporate deposits and purchased funds. In addition their assets are probably in far worse shape than the assets of the large banks, in large part because their lending policies are more likely to be subject to the needs of local officials.
I assume that one of the consequences of an economic slowdown in China will be a surge in potential insolvency among the smaller banks, a large part of which are probably already insolvent but kept liquid by confidence in the banking system. If (when) this happens, I suspect that the response of the financial authorities will be, rather than liquidate bad banks and pay depositors out of government revenues, to merge them into large banks. It would be politically risky to allow banks to default on depositors, and policymakers always prefer to bury expenses, hoping that over time they will resolve themselves.
Bailing out banks rather than closing them down can be economically justified under very specific conditions, formally identified long ago by Walter Bagehot, but the historical precedents suggest that protecting insolvent banks, as the US did in response to the S&L crisis of the 1980s, or Japan in the 1990s, nearly always ultimately raises costs significantly. Nonetheless in my opinion there is a very high probability that over the next ten years, rather than see the closing down of weaker banks, the Chinese financial landscape will evolve towards larger but operationally hobbled banks, much like the “zombie” banks that emerged in Japan in the 1990s.
Small banks tend to be very local in their operations, but I think in China even large national banks operate to an important extent on a provincial basis, with deposits collected in each province going largely to fund loans made within the province. What is more, governance within provincial branches, even those of the largest national banks, tend to be dominated by the needs of local government officials. This means that to some extent a national bank may be more aptly described as a group of provincial banks with headquarters (usually) in Beijing. As an aside, Anne Stevenson-Yang and Andrew Collier have both done very good work in trying to disentangle governance structures within the local banking systems.
Some of the financial reforms described in the Third Plenum are aimed precisely at modifying the ways in which local officials can influence the capital allocation process. Already there is a lot of concern among regulators about the risks associated with the ability of local officials to direct funding into favored projects, and as the economy slows, these risks will become much more obvious. We will see, of course, just how successful – and speedy – governance reform is likely to be in reducing the various kinds of frictional costs, but for now I think it is safe to assume that we cannot count on full, unfettered capital mobility within the banking system, and there is always the chance that unless Beijing forcefully intervenes, local responses to an economic slowdown will result in more, not less, localization of the banking function.
This also implies that the transmission of monetary and fiscal policy is not likely to be consistent across the country. To the extent that fiscal expenditures take place through the banking system, without transfers among provinces it may actually be more difficult for provinces experiencing economic problems to grow deposits fast enough to expand “fiscally”.
Will the costs of rebalancing be spread efficiently?
As growth slows within the Chinese economy overall, and as certain provinces and regions experience especially significant growth declines, I would guess that frictional costs and constraints on both labor and capital will remain, or even rise, and with them the transmission of monetary policy is likely to become more complex. Provinces that have relied heavily on government infrastructure spending – and the real estate boom that seems to accompany heavy government spending – may be especially affected, and these provinces are generally the poorer, Western provinces.
So what are the implications of frictions that constrain the adjustment process? If there were few relevant distortions or frictional costs within China, as China’s economy slowed, the rebalancing process would be difficult in any case, but at least there would be no tendency to distribute the costs of rebalancing unevenly, with heavier costs being assigned to regions less able to bear them. Because constraining labor mobility raises the costs for provinces with higher unemployment, and lowers it for provinces with lower unemployment, low frictional costs implies that labor mobility is not constrained. The same is true for capital mobility.
While the way costs are distributed among the provinces within China matters to individual provinces, of course, it might matter a lot less overall to China if losses for one province are matched with gains for another. To understand what might happen in China it might be helpful to think about the post-crisis experiences of the US and Europe, the two extreme cases of large currency zones in which frictional costs in the former are negligible and in the latter are highly constraining. Comparing how the US adjusted to the 2007-08 crisis with the way Europe did may shed insight on the process.
There are of course many costs within the US economy that keep labor and capital mobility from being frictionless (bus fare, if nothing else), but these are low enough to make the US very clearly an optimal currency zone, and so they spread the costs of the adjustment process more evenly in the US. This is not to say of course that the US adjusted quickly and painlessly from the crisis, but when we consider the other extreme, that of Europe, it is clear that the US adjustment could have been much more painful.
In Europe the adjustment following the crisis has been much more difficult, and this can be explained at least in part by the combination of high debt and high frictional costs (especially because of localized sovereignty) that protected stronger economies at the expense of weaker economies. These frictional costs made the adjustment easier for stronger economies, like Germany, but, I would argue, harder for Europe overall.
As peripheral Europe deleveraged, if Germany had taken steps to increase domestic demand, either German debt or European unemployment would have still risen (or some combination of the two), in the latter case with some, or even most, of the rise in unemployment occurring in Germany, but it would have risen by less than it actually did. Germany was able to take advantage of significant constraints in European labor mobility, along with Berlin’s near-total control of German fiscal policy and its disproportionate influence on European monetary policy, to transfer most of the adjustment costs to peripheral Europe, and this made Europe overall worse off.
The important point for China is not so much the genesis of the crisis but whether the higher adjustment costs for peripheral Europe were matched by equivalently lower adjustment costs for Germany, with European unemployment overall remaining unchanged. I would call this a “zero-sum” adjustment, in which what was worse for one side was equivalently better for another.
If Europe had undergone a zero-sum adjustment, it would be easy to argue (at least for German policymakers) that Europe’s adjustment might have been costly but it was well managed. I would argue, however, that this was not the case. A zero-sum adjustment could only have happened in principle if the rebalancing of internal European demand had occurred many years earlier, before debt levels became too high.
Early rebalancing was always unlikely of course. Like in China in 2007, when then-Premier Wen famously acknowledged China’s own unsustainable internal imbalances, or in the US in the years leading up to the 2007 crisis, when the collapse in the US savings rate had long been evident, European policymakers were unwilling to adopt costly adjustment policies until they were forced to do so, even though the delay sharply raised the cost of the adjustment.
Financial distress costs
It is important to understand the role of debt in forcing up the adjustment costs. High debt levels throughout Europe, even in Germany, made the rebalancing of internal European demand much more painful for Europe overall for at least two reasons.
- Leverage embeds pro-cyclicality into an economy – something that I have discussed many times before – so that the costs of a crisis increase as agents engage in self-reinforcing behavior in response to the crisis itself.
- Financial distress costs are non-linear, so that a transfer of costs is not symmetrical, with benefits for one equaling costs for the other.
To take the first reason, as the crisis forced peripheral Europe to deleverage, with no mitigating rise in German demand, the combination of high debt and significant downward pressure on prices and wages forced Europe into the highly pro-cyclical financial distress process, which I would argue resulted ultimately in much higher unemployment overall than would have otherwise been the case. The combination of debt and asset and wage deflation forced the country into a downward spiral of rising financial distress costs in which the debt burden worsened unemployment and unemployment worsened the debt burden.
There is nothing new about this. History makes it very clear that during periods of unbalanced growth and rising debt, the longer we postpone the adjustment the more expensive it becomes, and I would argue that it is precisely because high debt levels create a self-reinforcing process of financial distress (George Soros calls it “reflexivity”) that this must be true.
But this tells us nothing about how frictional costs and constraints within the Chinese economy might affect the cost of its adjustment. It is the second reason that matters more in the discussion of whether or not we can think of China as an optimal currency zone, and is why I would say that Europe represents the other extreme when we try to understand why a “more optimal” currency zone adjusts less painfully than a “less optimal” currency zone.
Except when debt levels are very low, or so high relative to debt capacity that the debt is practically worthless, the cost of a “unit” of deterioration is greater than the benefit of a “unit” of improvement, and depending on the amount of debt and the size of the unit, the difference between the two can be extremely high.* If distortions and frictional costs among different provinces are extremely high, as they were among different European countries, the cost of an overall rebalancing of the Chinese economy will be disproportionately borne by the weaker provinces.
Apart from issues of fairness, this might matter less to Beijing if it simply meant a distribution of the costs from one part of the country to another. In that case China’s overall economic growth rate would still decline as part of the rebalancing process, but it would be a zero-sum decline, and the fact that the costs were borne unevenly within the country would not affect the overall economy. Anhui’s loss of one dollar – to make up an example in which Anhui suffers more than Zhejiang from the rebalancing process – would be Zhejiang’s gain of one dollar, and although political and social costs for China might change for the worse, overall economic costs for China would not rise.
But, as occurred in Europe, debt changes all of this, and this change is a consequence of the fact that financial distress costs are not linear. If the weaker parts of the Chinese economy are heavily indebted – which is likely to be the case for both the weak and strong provinces and regions – constraints in the way costs are distributed across the economy no longer result in a zero-sum game. Lowering Zhejiang’s cost at Anhui’s expense, in other words, would almost certainly reduce China’s overall wealth.
Financial distress costs do not change in a linear fashion mainly because changes in risk perception, once the market starts to worry about debt levels, causes agents to change their behavior in ways that automatically worsen the debt burden (for those who are interested I have discussed this in several earlier issues of my newsletter as well as in my book, The Volatility Machine). Unless Zhejiang’s debt burden was much worse than that of Anhui, in other words, transferring a dollar of the adjustment cost from Anhui to Zhejiang would reduce China’s overall economic growth if both provinces were heavily indebted. Anhui’s financial distress costs would rise by more than Zhejiang’s would decline.
This is not simply an academic matter. Most analysts were surprised by the severity of the European crisis relative to that of the US, even though in the aggregate Europe’s economy seemed to have entered into the crisis in better shape (some would say much better shape) than that of the US. It had less debt overall, for example, a tiny trade deficit, no subprime mortgages, and a healthier savings rate.
The reason Europe suffered more may largely be because of the frictional costs that made Europe much less optimally a single currency zone. High levels of debt created a kind of negative convexity that ensured that the overall impact of forcing weaker countries to bear a larger share of the costs of adjustment had necessarily to raise overall financial distress costs for Europe, even if weaker and stronger economies had suffered from equally heavy debt burdens, but it was made worse by the fact that the former tended to have more debt.
We need to bear this in mind as we see how China rebalances and how easily rebalancing costs are transmitted through the economy, especially as debt continues to grow while we wait for the adjustment to take place. Is China more like the US or more like Europe? I cannot really say, especially when the answer will depend in part on the timing and sequencing of the reforms Beijing implements.
The pace of both hukou reform and of governance reform in the financial sector will determine in part the answer to this question, just to state the most obvious ways in which the timing and sequencing of reform matters. If China is more like Europe – that is if there are significant frictional costs that make it hard to spread the costs of rebalancing evenly across the overall economy – it is a near certainty that debt will become a much bigger problem and financial distress costs for China’s economy overall will be higher.
There are unfortunately no easy ways to measure the extent and evolution of these kinds of frictional costs, but it is certainly worth keeping an eye on the way the costs of rebalancing are distributed nationally as growth continues to slow. I think there are three clear implications that investors and policymakers may want to consider:
- The distribution of growth in China during the period of rapid growth has been extremely uneven, even when Beijing was committed to rebalancing growth among the poorer and richer provinces and has committed substantial resources to doing so. This may reflect substantial frictions within the Chinese economy and real institutional differences that have resulted in hard-to-change disparities in what I have elsewhere called “social capital”. If that is the case, as the Chinese economy continues to slow, significant growth disparities between regions and provinces are likely to remain.
- To the extent that these growth disparities remain, the distribution of adjustments costs is not likely to be zero-sum. High debt levels will ensure that the cost of China’s adjustment will be higher than expected and that it will be higher as a function of frictional constraints. For this reason it is important to understand whether China, as an integrated currency zone, is closer to the US or to Europe in terms of the “smoothness” in the way costs are distributed. We need especially to watch how slowing growth affects labor mobility and capital mobility.
- Policymakers of course will determine policy based on political considerations, but from a purely economic standpoint reforms aimed at reducing debt and increasing capital and labor mobility should take precedence over other reforms. This analysis does however leave Beijing with a contradiction. We know that moral hazard has been an important reason for investment misallocation in China, and one important step in reducing moral hazard might be to eliminate Beijing’s explicit guarantee of provincial and municipal debt. This would force local governments to borrow only in ways that can be justified economically, or else lose access to funding.
But eliminating Beijing’s guarantee protects the credit of the central government at the expense of local government creditworthiness, and so threatens to make financial distress costs much worse. In that case we might consider that the greater the internal frictions, the more reluctant Beijing should be to withdraw its guarantee, at least until the rebalancing process has been complete and the debt burden resolved.
Convergence is linear, and unlikely
Whereas I argue that the growth disparities among Chinese provinces probably reflect frictional constraints and differences in social capital, and so are likely to have an adverse impact on China’s adjustment, I should point out that not everyone agrees. In fact some analysts argue that it is precisely because there are such great disparities within China that high growth rates are likely to be maintained for the foreseeable future, with at least one analyst arguing in 2013 that China’s growth rate would remain between 7-8% through nearly the end of the decade precisely because of provincial growth disparities.
I admit that I am bewildered by his logic, which seems implicitly to be based on a widely-held but historically suspect theory of economic convergence. It requires implicitly a model of development in which the development process is linear, so that development occurs automatically as knowledge and capital is transferred from richer entities to poorer ones.
My own development model, which owes a lot to economists like Albert Hirschman, assumes that rapid growth is by nature unbalanced and heavily constrained by domestic institutions, so that the development process is specific (i.e. not easily transferred), non-linear, bumpy, and subject to reversals – because rapid growth creates deep imbalances that must be reversed. I also believe that these reversals can be brutal enough to prevent long-term convergence, which is why unlike most economists I am more impressed by successful adjustments than I am by growth miracles. More importantly, I believe development is based less on the transfer of knowledge and capital than on the transformation of domestic institutions (legal, financial, political, etc.).
In fairness, I have to acknowledge that there is no consensus on the topic. There are economists who see things very differently from me who treat convergence as a fairly linear process that, once you have figured out how to do it, can be applied to any variety of economic entities. In that case the fact that some province are poorer than others guarantees that China will grow fairly automatically if for no other reason than that capital will be transferred from the richer to the poorer regions and convergence will do the rest.
If like me you do not believe that development occurs automatically as capital is added, then the idea that income levels of poor provinces will easily converge to the income levels associated with rich provinces becomes not just suspect but in fact almost backwards. Poorer provinces are poorer because their institutional characters make it harder for local businesses and households to exploit capital productively, and pouring money into poorer provinces will simply destroy value more quickly than pouring it into richer provinces. Not only will it not result in faster growth overall, it will actually result in slower growth overall.
We will see what happens, but as China adjusts, I would expect that rather than converge, the income difference between the poorer and the richer provinces will persist and maybe even widen as financial distress costs for these poorer provinces rise faster than they will for the richer provinces. If local governments for provinces that are growing more quickly try to protect themselves by making it harder for migrant workers with the wrong hukou to immigrate, or by making it harder for capital to emigrate, China will slide away from a more optimal to a less optimal currency zone, and overall growth will slow.
One way that Beijing can lower the adjustment costs for China overall, in other words, is to reduce the power of local governments to intervene in labor or capital mobility. Another way is to use central government credibility to protect the creditworthiness of local government debt, even if this leads to moral hazard. China will almost certainly not force a rapid write-down of bad debt. This means that financial distress in China will persist, and over the long term, in exchange for less short-term disruption, the total cost of rebalancing the economy will be much higher.
* For those who think more easily in option terms, this becomes obvious by comparing debt with its optional equivalent, a short put on the “assets” of the borrower. Except when the delta of the option is close to zero (the borrower is so highly creditworthy that changes in the value of its asset have almost no impact on the value of debt), or when it is close to one (the value of assets is so low compared to the value of debt, that a change in the value of the assets is matched almost one-for-one with the change in the value of debt), an increase in creditworthiness has a smaller positive impact on the value of the debt than the negative impact of an equivalent reduction in creditworthiness.