China’s labyrinthine shadow banking complex has been a source of concern for years.
It’s always been readily apparent that credit expansion via the country’s network of impossibly convoluted back channels has served to inflate bubbles in all manner of assets by encouraging rampant speculation. It’s also painfully obvious that healthy deleveraging and the purging of misallocated capital has been severely impaired by the existence of end-around credit facilities.
Further, the risk associated with these activities is repackaged and passed around so many times that, to quote a recent SocGen note, it is “nearly impossible to have a clear idea who is responsible for what when things go wrong.”
What we do know is that when these channels get squeezed, we tend to see the rapid unwind of trades they’re used to finance. But again, it’s not always apparent where the unwinds are going to surface because no one knows for sure where all the money has ended up. That’s why you see this kind of rolling boom-bust dynamic that hits everything from equities to real estate to fixed income to commodities, depending on the circumstances.
Starting late last year, Beijing moved in for the proverbial kill by putting upward pressure on interbank rates. For the shadow conduits, this will be a slow death. Because if the PBoC were to make it quick, it would most certainly not be painless. Indeed, given the scope of the problem, a sudden unwind would likely trigger a financial crisis in China which would, in short order, spread to EM more generally and from there, to the global economy. The WMP industry alone has grown to some CNY30 trillion (and CNY60 trillion by some measures):
As we’ve shown on any number of occasions, China’s credit impulse is a leading indicator for all kinds of macro economic variables and more generally, when it comes to private sector credit creation, there simply is no credit impulse outside of the country.
As noted here at the outset, this story has been simmering for years, but occasionally something snaps and it’s thrust back into the spotlight. In 2015, it was the unwind of a half-dozen backdoor margin lending channels and the concurrent equity market meltdown that got everyone’s attention. Late last year, following the December Fed hike, it became apparent that rising interbank rates were creating a mini-crisis in China’s bond market. And then most recently, it appeared that the PBoC’s efforts to squeeze the shadow banking complex might have led directly to metals mayhem.
This time, however, may indeed be different in that China looks set to rein in this problem once and for all. Or, if that’s too strong, then I think it’s entirely fair to say that the effort to tighten via liquidity facility management and interbank rates is going to be a sustained push rather than a transitory event.
That means every analyst who covers China is on a mission to break down what’s going on. Well on Sunday morning, we get the latest from Goldman’s MK Tang who warns that “the credit deceleration in Q1 was more pronounced than reflected by the official data, as the shadowiest of shadow credit (i.e., credit not captured in TSF) slowed considerably amid a tighter monetary and regulatory stance.” His full note is below and you should definitely give it a read.
Strong Q1 growth in China has brought about a jump in industrial profits and some relief on the nonperforming loan (NPL) pressures. But market sentiment has swung to concern in recent weeks in light of risk that policymakers might have been “over-tightening”. While the focus is on the implications for subsequent growth, the starting place from which the tightening has occurred partly underpins the concern:
- Under the hood of Q1 data, a big source of the real growth acceleration was exports. In contrast, fixed investment was fairly soft—by either the official national accounts data or our propriety investment tracker, we estimate that real fixed investment grew roughly 3%yoy in Q1, vs. 5-6%yoy in 2016 (Exhibit 1), although investment price deflator could be noisy. As we discussed previously, China’s PPI reflation does not look likely to drive a significant pickup in real capex and growth. Even so, the lack of strength in real fixed investment so far might have been somewhat a disappointment.
- Further weighing on the fairly soft real fixed investment trend are various policy tightening measures, including the PBOC’s hawkish bias in the interbank funding market (Exhibit 2). While a key policy objective there is to crack down on financial leverage (i.e., borrowing by financial institutions to conduct carry trade), tighter liquidity in the system impinges on the financial conditions for the overall economy, putting further pressure on fixed investment. This would not be necessarily worrying if the policymakers had good foresight and were fully coordinated, so any tightening would be simply intended to lean against the wind. However, this is not always the case, especially if/when each individual policy agency reacts to the political leadership’s high-level macro goals, one of which for this year has been repeatedly highlighted as tackling the financial sector’s underlying fragility. In general, some of the major shifts in the authorities’ policy stance in recent years were not well-calibrated, at least initially (e.g., 2013 interbank rate spikes, 2015 local government fiscal reform), and this adds to the perception that a similar situation could happen again this time.
In this report, we try to link these two developments through the prism of the underlying trend of credit. In particular,
- We provide preliminary update of our “money-implied” credit measure through Q1 this year. We first introduced this measure a year ago to account for shadow banking credit extended to the real economy that may not be captured in the official total social financing (TSF) data. Our estimates suggest that the deceleration in credit has been sharper than indicated by TSF (even after adjusted for municipal bond issuance). The credit slowdown has likely been a reason for the fairly soft real fixed investment growth so far this year.
- Next, we briefly discuss the two broad interest rate categories in China: i.e., benchmark rates vs. market rates, and why moves in the latter matter importantly for credit conditions and growth, even when the former is unchanged.
- We then estimate the impact of the past increase in market rates on credit for the rest of the year. Because of lagged transmission, our baseline estimate of credit will likely continue to slow in the coming months. We also quantify this in terms of potential drag on activity growth, based on our previous research on credit impulse to growth.
- We end with our interest rate outlook. We see signs of progress in financial deleveraging in recent weeks. This, along with the headwinds to growth from the past tightening efforts, argue for an incremental less hawkish monetary stance. Indeed, there have been recent signals reflecting policymakers’ awareness of over-tightening risk. And the experience from the last key political transition year (2012) suggests that we should continue to expect nimble policy management to deliver the full-year growth target, even though that does not remove the risk of material intra-year volatility.
Credit slowdown in Q1 ’17: end of era for the shadowiest corner of shadow credit?
Since 2015, the official TSF data has no longer been able to include all the credit extended to the real economy, as we discussed previously. This is because of not only the municipal bond swap (for which it is fairly easy to adjust), but also the additional ways in which shadow banking has been conducted in recent years (which are much harder to account for). To circumvent these data issues, we introduced our own measure of credit last year: we look at the mirror image of credit—i.e., “money”, which is a metric related to households and non-financial corporates’ financial investment. The basic idea is that credit generation is essentially a money creation process, hence an effective gauge of “money” can give a good sense of the pace of credit. And even the shadowiest part of shadow credit—i.e., the credit that is not even reflected in the shadow credit components of TSF—can in principle be captured in our money-implied measure.
From this money-credit linkage perspective, while many investors took comfort from the fact that recent TSF data surprised on the upside, slow M2 growth (which surprised on the downside in March/April) points to caution about interpreting the actual credit conditions in the real economy.
More broadly, as we have done previously, in addition to bank deposits which are part of M2, our money flow measure includes also households and non-financial corporates’ non-deposit financial investments, which have proliferated in recent years. Due to a couple of data limitations, we cannot yet provide a final update of our measure beyond 2016Q2. But here we give preliminary estimates of our measure through Q1 2017, based on certain assumptions. Exhibit 3 details our measure (preliminary estimates for 2016Q3-2017Q1, the darkest-shaded columns).
Our preliminary estimates put credit flow in Q1 ’17 at about RMB 7.7tn (vs. RMB 7.4tn implied by official data—i.e., TSF after adjusted for muni bond swap). As a share of GDP, at 33% it is materially lower than the average in 2016 (36%). And the pace of credit slowdown over the last couple of quarters is also more pronounced based on our estimates than implied by official data. As share of GDP, averaging over 4 quarters, credit flow fell by 8pp of GDP in the past year on our measure, vs. 2pp as implied by official data (Exhibit 4). Alternatively, in terms of year-on-year change in stock, credit decelerated by 5pp in the past year (from 21.5%yoy to 16.5%yoy) on our measure, vs. less than 2pp as implied by official data (from 17%yoy to 15.5%yoy).
The recent realignment in the levels of credit measured by our money-implied approach with the official data is probably not surprising. Just as a loosening in policy stance prompted a jump in the shadow credit that is not captured in TSF two years ago, it seems reasonable that the recent tightening has hit the same kind of credit particularly hard. Besides less accommodative interbank liquidity, stricter prudential rules—in particular the PBOC’s enhanced MPA implemented in Q1—were the likely driver of the slowdown there. And the considerable credit deceleration has probably been a main restraint on real fixed investment growth.
Which interest rates matter?
Before we turn to discuss the impact of monetary tightening on credit conditions, a quick word on China’s fairly segmented interest rate system is probably useful. There are two broad categories of interest rates: 1. Benchmark rates. These are rates set by the PBOC, partly as legacy of the pre-liberalized regime where banks’ loan rates and deposit rates were quite rigidly tied to these rates. Changes in benchmark rates are not “monetary” in nature strictly speaking, as they are more rules-type policy and do not directly entail any changes in the quantity of money. Since 2014/15, the authorities have granted broad rate-setting autonomy to banks. But, in practice, banks still heavily reference benchmark rates in their lending and deposit-taking activities (e.g., rates on corporate and mortgage loans with longer than 1-year tenor are often re-set each year based on the prevailing benchmark rates). Therefore, benchmark rates matter for the average financing cost for banks (for which deposits typically account for the majority of the liabilities). Likewise, they also matter importantly for the average debt servicing cost for bank borrowers (Exhibit 5). Currently, bank loans account for about 60% of all outstanding credit.
2. Market rates. These are rates driven by market activities—the more relevant ones include interbank repo rates (instantaneous short-term funding cost), longer-dated swap rates (reflection of market’s expectations on future interbank rates), and bond yields. These rates do not necessarily move in line with benchmark rates. For banks (and other financial institutions), these rates matter because they determine the marginal financing cost, as any incremental yuan of funding required by banks would likely be sourced from the wholesale market. The PBOC can manage (though not directly dictate) market rates through the quantity and interest rates of its monetary operations (OMOs, MLF, etc.) and RRR changes.
Many believe that benchmark rates are the most important rates for credit conditions and growth in China, with market rates playing limited roles through their indirect impact on financial asset prices. But in practice, market rates are key as well—not only do they determine corporates’ bond issuance cost, they also affect behaviors of financial intermediaries such as banks. In particular, lower market rates encourage financial institutions to borrow more from the market and intermediate the funds to the real economy as well as to secondary financial markets. This is quite clearly manifested in the last couple of years, where the sharp fall in market rates in 2015 gave rise to the subsequent boom in shadow banking credit and the surge in financial leverage. Exhibit 6 gives a visual demonstration of the importance of market rates for credit conditions— market rates tend to feed through to even bank lending rates (after adjusted for benchmark rates), with a lag of roughly 6 months, although the “beta” is not large (at about 30%).
In other words, benchmark rates are just one of the many aspects of PBOC policy stance that matters for credit conditions. We summarize the transmission of various PBOC actions to ultimately the real economy in a stylized flow chart in Exhibit 7.
Further headwinds to credit conditions in the pipeline
The relationship shown in Exhibit 6 above also suggests that the past increase in market rates will probably continue to feed through to credit conditions over the next few months. To more systematically test this, we estimate a VAR on 1-year swap rate and credit (our money-implied credit flow measure as share of GDP), controlling for benchmark rates, instantaneous cost of liquidity (7-day repo rate), and the global financial crisis period. The sample is monthly running from January 2008 through March 2017.
Our result suggests that a 25bp shock to 1-year swap rate would reduce monthly credit flow meaningfully by 3pp of GDP after about 6 months, with the peak impact of 5pp coming after about 12 months (Exhibit 8).
Going forward, assuming that all interest rates remain at the current level through year-end, then our result would suggest that credit may continue to slow, from 33% of GDP in Q1, steadily to 29% of GDP in Q4 ‘17 (Exhibit 9). In terms of growth in credit stock , the result would imply a deceleration from 16.5%yoy in Q1, to 15% by year-end.
Suppose credit flow indeed evolves in line with Exhibit 9, based on our previous estimates on credit impulse to growth, the credit slowdown could imply a reduction of roughly 50bp in annualized sequential real GDP growth through the rest of this year.
A not-so-stable path of stability? But nimble policy management is likely
With the expectation that the authorities would be focused on growth stability ahead of the party congress late this year, our baseline forecast is for a more moderate 10-20bp annualized sequential real GDP growth deceleration in H2 this year.
Growth could indeed slow faster than in our baseline as credit drags (in fact, our China CAI already decelerated by 50bp annualized in the last three months vs. the prior three months, and stood at about 6.5% in April). And it is fair to note that “stability” is a relative concept in terms of scale and time horizon. For instance, in 2012—the last important political transition year, while full-year growth comfortably met the target, intra-year volatility was quite pronounced, with growth slowing rapidly around the middle of the year. The risk of such a soft growth episode and intra-year volatility in the months ahead does not look trivial.
That said, an experience learned from 2012 is that with the party congress in sight, policymakers were prompt in easing when growth weakened, led by a large slowdown in industrial production (Exhibit 10). While growth (and markets) remained under pressure for a period of time as the policy support took time to work its way through to boost the economy, the scale of easing was large enough for a strong rebound to materialize before too long (in fact, the associated large impulse to leverage was partly what prompted the subsequent harsh PBOC tightening in the following year). A similar policy reaction function is likely this year, in our view.
Financial deleveraging in progress, gives room for PBOC to stay easier on the brake
In fact, there are signs that the authorities have incrementally eased back on their hawkish bias, as reflected in the more neutral tone of the PBOC’s Q1 policy report and Premier Li’s recent comment on keeping financial markets stable. Interbank repo rate fixings have also moderated somewhat in recent days. We see reasons why the PBOC will probably remain slightly easier on the brake for the time being, even though we do not expect a strong press on the gas unless growth slows significantly.
In particular, there are early signs that financial leverage has come off more meaningfully in recent weeks, possibly also in response to the ongoing prudential tightening. Some of the metrics that we discussed last November as reflections of elevated financial leverage have moderated. For instance, weighted average maturity of funds’ repo borrowing has continued to rise, and is largely back to the historical norm as of end-April (Exhibit 11). The spread between bond yields and swap rates, a rough proxy for leveraged position in the bond market, has recently widened materially, albeit still quite depressed (Exhibit 12). Also, the spread between R007 (general 7-day repo rate that covers all interbank borrowers) and DR007 (which covers only depository institutions) has narrowed somewhat (Exhibit 13) and seem to have become calmer even as underlying interbank liquidity conditions tighten (Exhibit 14), suggesting that liquidity pressures faced by funds have probably become less severe.
As the process of financial deleveraging advances, the risk-reward of further tightening seems a bit less compelling now (though on the other hand, any major easing could muddle signals on deleveraging commitment and re-ignite moral hazard, hence we believe the hurdle for that is also fairly high). This should imply a larger scope for the authorities to rebalance their focus more toward growth, paying for some insurance against the credit headwinds in the pipeline