Germany’s Missing €500 billion

THE PERCEPTIVE Simon Tilford of the Centre for European Reform has noticed a curious feature of Germany’s external accounts: her net international investment position (NIIP) reveals net “foreign assets have risen much less rapidly than the accumulated current account surpluses, leading to … losses: around €580 billion since 1999.”

That is, the value of Germany’s stock of external financial assets minus liabilities measured at market prices is worth substantially less than implied by cumulated flow of current account surpluses. This “loss” of roughly 18 percent of GDP—a decent chunk of change—deserves attention. Why save so much abroad if it is being spilled onto the ground?

To explain this gap, Tilford suggests—contrary to the narrative of an ageing society attaining higher yields abroad—Germany’s external returns fall short of expectations. Given these losses, it would be in her interest to rein in her surplus and invest more at home.

Meanwhile, Martin Sandbu of the Financial Times’ Free Lunch column picks up on this discrepancy, arguing that since surpluses are being “diffused” by “poor … lossmaking … investments,” German savings are not the concern some suggest—if she insists on investing abroad then Germany’s steady loss represents others’ gains.

Against this, Sandbu references in passing another interpretation, due to Matthias Busse and Daniel Gros of the Centre for European Policy Studies in Brussels. They argue current account data to be more reliable; in contrast, measurement errors likely creep into Germany’s net IIP—being the difference between two large and “imperfectly measured” stocks. They therefore propose to focus on the yield on German investments abroad, observing “returns … have remained above most domestic returns” and therefore “German savers have fared quite well over the last decade in their foreign investments.” Thus, net IIP figures can be misleading; surpluses are perfectly rational given relative returns.

What’s going on?

Figure 1 uses the latest external accounts data— consistent with the IMF’s Balance of Payments Manual 6 (BPM6) methodology—for Germany, plotting from end-2003 the cumulative current account balance, financial account, and the change in net IIP. The former two series being cumulative flows, the latter a change in (net) stocks. This illuminates the gap under consideration. Germany’s current account surplus sums to about €2,310bn over the period, the change in net IIP is €1,790; sure enough, the gap is roughly €530bn.

The chart immediately points to one source of discrepancy. Germany’s financial account flows only cumulate to €2,180bn, the gap with the current account being inevitable “errors and omissions” in collection of balance of payments (BOP) data. Indeed, these sum to €130bn—per year perhaps €10bn (say ½ percent of GDP). Either the current account surplus is overstated or financial account flows understated. We cannot say which. But we have captured one-quarter of the missing €530bn already.

But what of the remaining €400bn?

Recall how net IIP data record the change in asset and liability stocks due to flows, recorded in BOP, plus valuation adjustments—due to exchange rate, price, or other adjustments impacting stocks. We can therefore decompose the IIP gap by contrasting the cumulative financial flows with the change in IIP stocks for each investment category—as in the table below.

The accumulation of assets since end-2003, for example, according to BOP was €4,070bn. But the change in the outstanding stock of financial assets over this time was €4,530bn—meaning assets have increased in value by roughly €460bn more than flows. The same is true of liabilities, however—for example, non-resident holders of Bunds, recorded as part of portfolio liabilities, have experienced large capital gains as yields approached zero.

Overall, the positive asset valuation adjustment (+€460bn) is more than offset by that due to liabilities (-€530bn) in net IIP; the total contribution to the IIP shortfall is exactly €66bn. This accounts for an additional 12 percent of the €530bn IIP gap. Together with errors and omissions, we have filled 37 percent of the shortfall.

What next? Only one item in the external accounts remains in “financial derivatives (excluding reserve assets) and employee stock options”—FDs for short.[i]

Since, per BPM6, flows due to FDs are only recorded on a net basis—not gross—it is impossible to deconstruct the stock-flow contributions exactly as in the table above. However, we can decompose Germany’s FD transactions in BOP—a measure of external gains or losses on derivatives positions—in three different ways, by: product, sector, and country, shown in the three figures that follow.

Typically, through time, Germany registers “losses” on FDs—positive flows. The total since 2003 is €305bn. But to think of these as losses is somewhat misleading. If these represent hedges—say on foreign exchange value of foreign denominated securities—then “losses” may simply offset by gains elsewhere in the net IIP, so should not be viewed in isolation.

Nevertheless, since 2010, positive FDs flows mostly reflect forward contracts. However, during GFC large losses on options were recorded. In 2007 and 2008, for example, positive options flows of €110bn (of €163bn in total options over the 14-year period) were recorded, only partially offset by gains in following years.

Employee stock options are immaterial.

In terms of sector, FD flows over the period mainly accrue to monetary financial intermediaries excluding the Bundesbank (€190bn) and other financial intermediaries (€92bn). Direct losses due to non-financial corporates and households are minimal (€22bn).

In 2007 and 2008, when noted losses on options positions were largest, banks (€60bn) and other financial intermediaries (€54bn) were hit roughly equally.

Finally, we can trace the country-counterpart to these FD flows, using the Bundesbank’s bilateral financial account flows data. Historically, Germany’s FD flows were largely recorded against the United Kingdom. For example, the correlation coefficient between total FD flows and that against the UK was more than 0.9 in the decade through 2000, while that recorded against France was less than 0.2. From the millennium—perhaps related to the inception of the euro?—the flow of FDs in Germany’s financial account has pivoted towards France. Indeed, since 2003, €138bn flows have accrued against France and only €116bn against the UK. Moreover, most of this was recorded during GFC, when FD flows against France equalled €94bn—when options losses were at a peak.

The bilateral losses against France are particularly interesting. The growing importance of derivatives trading by German banks in the period prior to GFC has been noted, as well as greater losses of Germany relative to France during GFC (see here). The BOP data appear to suggest that one reason for the high relative German losses is that French banks benefited from their derivatives positions against Germany. In other words, Germany provided the hedge for French banks during GFC, helping mitigate losses at that time. But this, like the financial positions then, is mainly speculation.

In any case, we have now precisely accounted for Germany’s missing €500bn, summarized in the chart below as cumulative flows since 2003. As the chart reiterates, most “losses” were incurred during two stress periods.

First, in 2007/08, during the onset and crescendo of GFC, a cumulative gap of about €200bn developed. A substantial part of this (over €100bn) reflected losses on options positions against France, noted above. But there were other valuation losses during this period on net other investment and net FDI positions.

Second, in 2010/11, during the euro area crisis, valuation on net portfolio investment assets created a further dent. Between early- and end-2011, losses on net portfolio investment asset of about €130bn were recorded. This will reflect both losses on German holdings of peripheral bonds during peak stress, but also the bund’s status as haven pushing down German yields.

            Errors and omissions, meanwhile, fill much of the remaining gap.

It is also noteworthy how, since 2011, the €500bn gap has levelled off. Continued flows relating to FD positions are offset by valuation adjustments elsewhere; net IIP has moved in line with the current account.

To summarize, the German data on external asset and liability stocks and flows is complete. It is not true that errors in recording IIP stocks require instead focus on the current account. While Germans make terrible macroeconomists, they make formidable accountants. There is nothing much wrong with the data. All we lack is interpretation.

But this narrative also unveils flaws in Sandbu’s sanguine interpretation of this gap. In fact, Germany’s losses occur largely during period of financial stress—revealing dangers of unchecked capital flows and external imbalances. Germany’s net IIP losses are not, therefore, a harmless means of dissipating her growing external claims. Rather external losses reflect the destabilizing influence of surplus savings and large gross capital flows across borders—leaving financial crises in their wake.

And this also refutes Tilford’s claim that Germany’s external returns have failed to deliver. Indeed, absent destabilizing financial crises, Germany’s net IIP gap would not have materialized. And it has now settled down. The gap need not grow further—and rate normalization and pull to par of bonds could see it close.

But what about the relative returns on German investments abroad? These are shown in the chart below. As Busse and Gros note, Germany has indeed generated a higher yield on external assets than non-residents have achieved from investment in Germany. But it is noteworthy that the yield gap on portfolio investments has closed in recent years, while other investment returns are more-or-less identical through time. As such, the only advantage Germany currently gains is the yield on direct investment abroad, which has generated about 1.7 percentage points more than investments in Germany on average since 2007. But this is also closing.

But this is not the end of the story. First, low relative returns in Germany could simply reflect the lack of domestic demand there. Any attempt to generate domestic-led growth could change this calculus. Second, in any case, inwards foreign direct investment in Germany experienced substantially larger valuation gains over the past decade (see table above). A measure of total return–that is, yield plus valuation gain or loss–on FDI (not shown) generates very similar returns on inward- as outward-FDI. In other words, the case for Germany achieving a greater return abroad is not clear-cut.

The case for continued net external asset accumulation by Germany is not clear cut.

***

[i] Now…   a technical note: BPM6 introduced FDs as a separate functional category in BOP reporting for the first time (see here). FDs, such as forwards and options, are used for “risk management, hedging, speculation, and arbitrage” with prices linked to some underlying financial instrument. And since employee stock options, while not derivatives, take a similar form—say, the right to purchase equity as remuneration—these are folded into this category. And it turns out that a large part of Germany’s “missing” €500bn can be traced to FDs.

BPM6 manual gives the helpful example of an exporter hedging currency risk with a forward contract (see page 163). This helps intuition. In short: at inception of the hedge, no entries in external accounts are needed—the initial forward position is zero; not true of options. As asset prices change, the value of the forward contract changes, reflected in IIP (asset or liability). Only when the contract is finally settled is the FD transaction recorded in both the BOP as financial account (flow) and net IIP (stock) as a transaction within the period and final valuation adjustment. And the FD transaction equals total net IIP valuation adjustment (but with opposite sign) returning the IIP stock position to zero.

And so, gains or losses due to FDs positions can distort changes in net IIP when compared to the cumulative current and financial account flows. Concretely, in the forward contract example, losses due to hedged exchange rate risk for an exporter imply a $1,200 goods surplus is met with an increase in “currency and deposits” on financial account of only $1,000 alongside a negative $200 FD liability flow (positive $1,200 total) because of losses on the forward contract. Net IIP only increases $1,000—and thanks to FDs, the change in net IIP deviates from the current/financial account flows. Such is the cost of hedging exchange rate risk—with the benefit that local currency value of goods is assured in advance. Technical aside over.