Hidden flows: Asset prices and eurosystem balance sheets

To suppose large and successive [external] balances to be formed into a debt is to assume an accumulation of debt which is almost equally incredible.

Henry Thornton, Paper Credit, 1805

PDF: Hidden information_FINAL.

Charts: Hidden information_chartpack.

THERE WAS ONCE a time when international macroeconomics was conducted largely through the prism of the central bank’s balance sheet. Absent all but patchy data or guesswork on external current or financial transactions—such as the overall goods and service balance or net lending abroad—changes in central bank reserve assets, typically specie, served as the only real indicator of net flows against the rest of the world—if not, of course, internal drain. Thus, monitoring balance sheets became crucial for investors.

While often neglected today, the central bank balance sheet continues to prove essential for parsing private financial flows and investor sentiment. This was brought into sharp relief when the eurosystem’s TARGET2 clearing system, and network of credits and debits across national central banks (NCBs), became essential for tracking financial flows at the height of the euro area crisis.

More recently, the ECB’s asset purchase program (APP) has provided liquidity across the system that has resulted in widening TARGET2 balances once more—and ever-increasing uneven distribution of eurosystem liquidity. Figure 1.

Germany’s TARGET2 credit, at €913bn in July, has come to mirror the combined debit position of Italy and Spain—at €471bn and €403bn respectively. And at first blush, the TARGET2 debits of Italy and Spain, which have steadily increased since the inception of APP, appear of the same genus—the two giants of the periphery experiencing broadly comparable net private financing flows as expressed through their debit balance.

However, care in interpretation is needed; not all TARGET2 debits are alike. How so?

While most of the public-sector purchase program (PSPP) is undertaken according to the eurosystem capital key, there are other aspects of APP—the ABS, covered-bond, and corporate sector purchase programs—which are also important and implemented by the ECB and select NCBs. Thus, while the total stock of assets held under APP amounts to €2.5trn as of August, only 82 percent of this total related to PSPP purchases; €0.45trn relate to the smaller purchase programs. Figure 2.

Moreover, within PSPP, purchases are split between national government bonds as well as supranational debt—initially supranational purchases were 12 percent 0f total PSPP, decreased to 10 percent in April, 2016. Figure 3.

And purchase of supranationals is determined by “a specialisation approach, [whereby] only a few NCBs buy securities issued by European supranational institutions… independent of the domicile of these international or supranational institutions” (emphasis added.)

These supranational assets include the European Investment Bank (EIB), European Financial Stability Facility (EFSF) and European Stability Mechanism (ESM), for example.

Consider the following. The ECB balance sheet component of the eurosystem is only published annually. Between end-2014 and end-2017, ECB holdings of assets for monetary policy purposes increased €211bn, the counterpart to an increase in intra-eurosystem liabilities (i.e., TARGET2 debit) of €193bn, visible in Figure 1.

However, the system-wide increase in ABS, covered-bond, and corporate sector purchases, as well as purchases of supranationals within PSPP, over the same period was €573bn. In other words, under the totality of APP there are roughly €360bn in assets held for monetary policy purposes—beyond national government bond purchased by capital key under PSPP—which show up on the asset side of NCBs of the eurosystem.

For this reason, interpretation of the evolution of TARGET2 debits becomes nuanced. Imagine a NCB buying a supranational on behalf of the system, not domiciled locally. Then there will be an automatic increase in that NCB’s assets and, mechanically through TARGET2, liabilities. But this in no way reflects net private flows to the economy in question. For example?

Begin with the case of the Banco de España (BdE) which, thanks to the continuation of historical balance sheet reporting beyond euro membership, still provides a geographical breakdown of asset holdings. Table 1 summarizes BdE stock of assets as of end-2014 and end-July 2018.

Total assets of the BdE increased €363bn over the period, as expected during APP. However, domestic asset holdings only increased €251bn—comparable with cumulative purchases, through August, of SPGBs and related local assets of €254bn under PSPP. An additional €67bn assets over this period, however, relates to claims on the rest of the world—i.e., non-euroarea assets. And ECB reporting of total Spanish asset holdings for monetary policy purposes of €328bn as of August 3^rd suggests BdE has engaged in the purchase of non-domestic assets under APP.

Given that most supranational assets eligible under APP are domiciled in the EU but not euroarea, the above data would be consistent with Spain undertaking supranational purchases under the PSPP on behalf of the system, for example.

Regardless, this is also confirmed by the balance of payments (BOP) data.

Figure 4 shows the net financial flows of BdE as reported by in Spanish BOP. Witness, since 2015, the steady accumulation of portfolio investment assets by BdE alongside other investment liabilities, meaning TARGET2 debits.

Figure 5 shows BdE holdings of euro-denominated securities by non-resident geographical breakdown. The imprint of PSPP is apparent in claims on rest of the world (ROW.)

Why does this matter? It means some of the “deterioration” in the Spain’s TARGET2 balance since APP-inception—perhaps €20 billion per year—does not represent net private flows, rather the mechanical accumulation by BdE of foreign assets. And just as this might mislead as to private transactions today, the necessary future financial inflows to offset APP sales later will be correspondingly reduced.

Put another way, we can adjust Spain’s TARGET2 debit over the period for the accumulation of foreign assets. And this “shadow” TARGET2 debit would be a better indicator of cumulative private flows since APP.

Next contemplate Italy, the other giant of the euroarea periphery. Banca d’Italia (BdI) does not provide a comparable decomposition of assets by geographical location, so far as I am aware. But we do know the following.

BdI holdings of securities issued by euroarea residents from end-2014 through August increased €306bn. This compares with the ECB reporting on cumulative PSPP purchases of BTPs over the same period of €356bn, although the former represents transactions beyond those only for monetary policy purposes.

Alternative reporting by BdI puts the stock of assets held for monetary policy purposes at end-July of €383bn. Since data reporting only goes back to 2017 on this measure, we cannot make a comparable assessment of the change in asset holdings since 2014. But it suggests BdI has not been engaged in comparable non-resident asset purchases under APP. And this is indeed played out in the BOP data.

Witness Figure 6. BdI has hardly engaged in the purchase of non-resident portfolio assets. Roughly speaking, in contrast to Spain, the entire accumulation of TARGET2 debits in Italy therefore represents net flows of private investors.

Recall Figure 1. Indeed, Italy’s TARGET2 debit drifted below Spain’s for the first time in 2016. But this might be expected given Italy’s larger size and thus purchases under APP.

What Figure 1 hides, however, is relative purchases of external assets under APP—purchases which mechanically drives a TARGET2 debit without reflecting meaningful action on the part of private investors. Thus, APP has in some sense exaggerated Spain’s debit position relative to Italy.

Figure 7 shows Spain’s actual TARGET2 debit as well as one estimate of the adjusted, or “shadow” debit—which adjusts for euro securities held against ROW.

What does this have to do with asset prices? Consider Figure 8 which shows—from July, 2012—Spain’s adjusted TARGET2 position minus Italy’s debit (right scale, € billions) against the Italy bond yield spreads over Spain (left scale, bps).

Net private transactions against non-residents, as reflected in residual liability claim by the respective NCB on the eurosystem, have steadily deteriorated for Italy relative to Spain. From about €150bn larger debit position at the time of Draghi’s “whatever it takes” speech, Spain’s TARGET2 (a smaller economy) has improved relative to Italy such that today Italy is about €120bn worse off.

And these net private flows are reflected in the yield spread of Italy over Spain. Where Italy traded through Spain in 2012, this has now reversed. Arguably, the relative TARGET2 balance is suggestive of the future yield spread.

For amusement, Figure 9 shows the BdE asset and liabilities by residency (ignoring “unclassified” items.) The increasing role of claims on ROW is evident, as is the overall impact of PSPP, reflected in liabilities to the euroarea. But a growing pocket of domestic liabilities—bank deposits—represent a relatively recent trend, suggestive of improved net private flows serving as a buffer for the local economy against future adverse flows.

Is there more to be said on the growing nuance needed when interpreting TARGET2 today? Consider another, more egregious case. It is well known that Greece does not participate in the PSPP—they are precluded from the program and it looks unlikely they will benefit from substantial purchases before the program ends. But this does not mean the Bank of Greece (BoG) is not engaged in other aspects of APP. And they have indeed been accumulating their own small share of assets held for monetary policy purposes across the eurosystem. Figure 10.

 

BoG assets for monetary policy purposes bear the imprint of the earlier Securities Market Program (SMP) from 2010, but these were small and were steadily rolling off. Then, in 2015 Greek holdings of assets grew sharply—from €6bn end-2014 to €64bn end-August.

Since BoG also produces a breakdown of assets by geography, Table 2 gives the decomposition previously shown for Spain.

Since end-2014 total assets held by BoG are hardly changed—up by €3bn. But domestic assets—relating to emergency liquidity assistance and repo operations by the domestic banking system—have fallen €44bn, while foreign assets have increased €47bn. These foreign assets relate, of course, to the implementation by BoG of non-PSPP aspects of the APP.

We can apply a similar adjustment to Greece’s TARGET2 (plus banknote related liabilities) to get a measure of the true “shadow” balance therefore. Figure 11 does this. And in the case of Greece this balance notionally turned positive around the turn of the year. Were it not for APP purchases, net private capital inflows to Greece (and banknotes returned from beneath the mattress) would have been enough to generate a TARGET2 credit by now. It seems likely that over the next year or so if private capital inflows remain robust, Greece will attain an actual credit balance on TARGET2—remarkable for a country that has been constantly berated for being a debtor to the eurosystem.

As a corollary of this, BoG recently attained the largest net foreign asset position since euro inception. Figure 12.

Other countries that witnessed capital inflows and a turnaround from debtor to creditor within the eurosystem—such as Cyprus, Ireland, and Slovakia—also saw a sharp compression in government bond yields. One technical constraint for Greece that prevents such compression relates to the continued limits on domestic banks from buying GGBs which remain largely in non-resident hands.

Still, growing liquidity within Greece will contribute to improved credit creation, growth, ratings upgrades, and eventually a new set of investors to drive down yields—closer to peripheral comparators. If the shadow TARGET2 position is any indicator, Greek assets remain the cheapest in the euroarea by a considerable margin.

END.