Why Not Start From Public Debt/GDP?

Vietnam's public debt/GDP is low; its government bond yields are low too, but that second number must be read with care: the buyers are mostly commercial banks (buying to satisfy liquidity ratios) and the Social Security fund - the yield of a captive-buyer market is a price compressed by mandatory demand, carrying a weaker market signal than it appears to, as a later section will show. All that can be said with confidence is that domestic-currency default risk is not the thing to worry about - and even that does not mean unlimited room to spend. The U.S. can read Debt/GDP its own way because it borrows in the currency the whole world needs to hold; Japan has a very deep domestic bond market with most of its debt held by domestic investors. The mechanism researchers call original sin (economics) puts Vietnam in a different group: like most emerging economies, it cannot borrow internationally in its own currency - the VND bond market, however much deeper than before, is still not treated by foreign investors as a long-term store of value the way USD or JPY are. The main risk therefore takes a different shape: rather than a budget slowly eroded by domestic-currency interest costs, the thing to fear is a sudden devaluation that inflates foreign-currency debt, imports, and other dollar obligations overnight. It is worth naming who actually holds that risk: after a decade of restructuring, government debt has fallen to about 32% of GDP (2024), roughly three quarters of it domestic - the government's own external debt is down to around 7-8% of GDP and shrinking. The "devaluation balloons public debt" channel is therefore much narrower than a decade ago; today's currency mismatch sits mainly in corporate foreign-currency debt and the whole economy's import bill. That changes the address, not the constraint: when the shock arrives, firms and the FX market are where the symptoms break out first, while FX reserves - and ultimately the budget - remain the party that has to treat them.

No foreign example is needed: Vietnam has been through this exact mechanism itself. The post-2008 stimulus package pumped credit and subsidized interest rates across the board; the bill arrived two years later - inflation shot above 18% in 2011, the VND was devalued repeatedly (9.3% in February 2011 alone), and FX reserves fell to the $12-13 billion range, around 1.5 months of imports. Public debt stayed within safe limits the entire time; what broke was the exchange rate and trust in the currency. The 1997 Asian crisis is the larger version of the same lesson: South Korea entered that year with public debt/GDP of only about 13%, with Thailand and Indonesia similarly low - all three still fell, because the blow came from the currency and sudden capital flight, not from the budget. The comparison does have one mismatch worth recording: the 1997 trio let foreign capital move in and out almost freely and carried mountains of short-term private foreign-currency debt - exactly the conditions for a sudden stop; Vietnam keeps relatively tight capital controls, so the same kind of pressure usually takes a slower detour - through expectations, dollar hoarding, and informal channels, as 2022 and 2024 showed. Slower does not mean immune; it means the clock counts in quarters instead of weeks. For an emerging economy, the question worth asking first is still whether the country has enough dollars to withstand a currency shock or a sudden capital outflow; "how much more can the government borrow" comes second.

That external constraint has been there all along. What 2025-2026 added is a more striking development: the internal constraint started speaking up too - through the very people who run the economy's largest funding channel.

"Not Flush With Capital": A Real Statement

The statement is real and was widely quoted in the business press. At the July 18, 2026 conference on "removing bottlenecks - unlocking resources - promoting growth," Nguyen Thanh Tung, chairman of Vietcombank's board, said:

The most common misreading turns this into "banks are about to run out of money to lend." Vietcombank itself runs a loan-to-deposit ratio (LDR) under Circular 22/2019 below 70%, well under the 80% cap he cited. The vault is not empty. The problem sits on a different layer: the size and structure of the entire system's funding base relative to the volume of projects about to land on the economy.

The Arithmetic Behind The Statement

The arithmetic presented at the conference is fairly simple: growth of around 10% a year requires total investment of roughly 40% of GDP, while domestic savings reach only about 36.5% of GDP. That gap has to be filled with foreigners' capital: FDI, international borrowing, foreign investors buying bonds and equities. Remittances do not belong on that list - they are already counted into national income and thicken the 36.5% itself; nor do domestic capital markets, which only redirect existing savings rather than create new ones. The subtraction is a planning number rather than an audited one - 40 and 36.5 come from two different statistical systems; the "three statistical universes" note below goes into detail - but the order of magnitude matches the five-year plan itself: the 2026-2030 targets require roughly 38.5 quadrillion VND of total social investment, with the state covering some 8.5 quadrillion - the remaining ~30 quadrillion must come from financial markets, FDI, firms, and households: spread over five years of GDP, squarely in the ~40%-a-year zone. Meanwhile, the largest domestic funding channel - bank credit - was already carrying outstanding loans equal to 134% of GDP at the end of 2024, and an estimated ~148% of GDP by mid-2026 (a later section shows the arithmetic). The well still has water; the whole village is about to build more houses.

The 40 - 36.5 arithmetic also hides an identity that makes it heavier than it looks. In national accounting, savings minus investment equals the current account: S − I = CA. Investing 40% of GDP on 36.5% savings is therefore not two numbers standing side by side for effect: if the two numbers shared one statistical frame, it would be a commitment to run a current account deficit of about 3.5% of GDP every year - importing real capital, steadily, from the rest of the world. In fair weather, a gap that size is unremarkable for an industrializing economy - disbursed FDI alone runs about $25 billion a year, enough to cover the $17-18 billion it implies. The catch is the starting point: Vietnam's, per the IMF, is a current account surplus of 5.8% of GDP in 2023 and 6.6% in 2024. Run at full throttle, the plan implies a swing of roughly 10 percentage points of GDP in the external balance (exact size depends on the statistical set; the direction - from large surplus toward deficit - does not) - and this is where the piece's two constraints merge into one: the domestic funding gap and the exchange-rate pressure are not two parallel stories, they are two faces of the same number, published every quarter under the name "current account".

The volume gap is only half of the statement; the other half concerns the system's safety ceilings. A billion-dollar project does not just need money - it eats into several technical limits at several banks at once: the capital adequacy ratio (CAR), single-borrower and borrower-group exposure limits tied to the bank's own equity, and the cap on short-term funding used for medium and long-term loans. A bank can have room under one ceiling while pressed against another; to keep disbursing it must either raise capital or wait for the ceiling to be lifted - neither is quick. For the state-owned group - the banks expected to carry the large projects - the nearest wall among these ceilings is equity capital itself: their CARs are among the thinnest in the system, and thickening them has for years depended on petitioning to retain profits instead of paying cash dividends into the budget. And that wall is moving closer: by law, a bank may not lend any single client more than 15% of its equity (25% for a related group of clients) - and the 2024 Law on Credit Institutions is stepping those limits down to 10%/15% by 2029, with the mid-2026 notch at 13%/21%. For billion-dollar projects, the ceiling hit first is usually not a shortage of deposits but a shortage of equity capital.

The solutions proposed at the conference follow that same diagnosis: let the Ministry of Finance issue international bonds; develop the corporate bond market - currently only about 10% of GDP, versus more than 40% in a few regional peers; and fix collateral rules so banks can support that market. All three share one spirit - finding capital outside the banking system - and the rest of this piece walks through each of those exits to see where it leads.

Not Just Vietcombank's Message

The stronger version of this message, delivered earlier, came from the State Bank of Vietnam itself. In June 2025, while still in office, Governor Nguyen Thi Hong (now a Vice Chairwoman of the National Assembly; the governorship passed to Pham Duc An in April 2026) told a National Assembly Q&A session that Vietnam's credit-to-GDP ratio had reached 134% at the end of 2024, among the highest in the world, and warned:

The 134% is an end-2024 marker - and it is aging fast. In 2025 credit grew 19.01%, taking the outstanding stock to 18.58 quadrillion VND - the fastest pace in a decade, nearly double nominal GDP growth (~11.6%); by end-June 2026 the stock had passed 19.97 quadrillion, up 18.1% year on year. In late 2025 the head of the SBV's own monetary policy department was already citing credit/GDP at 146%; add the first half of 2026 and a rough estimate lands around 147-149% of GDP. In other words, between the warning being read out in the National Assembly and this piece going up, the economy's leverage ratio added roughly 13-15 points of GDP - the 134% figures in the rest of this piece should be read as a floor, not a ceiling.

She listed the project pipeline through 2030: roughly 2,000 km of new expressways, the North-South high-speed railway with a preliminary investment of about 1.71 quadrillion VND ($67.3 billion), airports, seaports, Power Development Plan VIII, and COP26 commitments. Her demand was concrete: agencies must identify clearly where the capital will come from and how borrowing and repayment will work, rather than defaulting to bank credit. Even for social housing - a politically high-priority area - her position was consistent: "capital from the banking system is only a supplementary solution, not the decisive policy"; programs needing long-term, low-rate funding should use budget funds entrusted through banks or a dedicated fund, instead of piling the burden onto credit institutions.

What "Relying On Bank Credit" Means

"Relying heavily on bank credit" means that when firms or the government need money to invest, the money comes mostly from bank loans rather than other channels. Picture the economy as one very large family: want to build a house - borrow from a bank; start a company - borrow from a bank; build a factory - borrow from a bank; and through the BOT wave of the 2010s, build an expressway - borrow from a bank as well. When nearly everyone lines up at the same well for water, sooner or later the well feels the strain. People seem to love solving every problem with a single tool, then act surprised when the tool starts to buckle. And this well has a feature few wells have: the bucket sits in the operator's hand. Every year the SBV assigns each bank a credit-growth quota - the "room" - so how much water gets drawn in total, and who draws how much, is an administrative decision, not a price. The roadmap for abolishing the room has been discussed for years; as of mid-2026 the bucket has not left the operator's hand - a detail worth remembering for the end of this piece, when guessing how the funding gap will get filled.

Why Piling It All Onto Banks Is A Problem

This maturity mismatch is not a paper risk - Vietnam has already run a full experiment with it. The transport-BOT wave of 2011-2016 was financed almost entirely by bank credit: funding raised at under-12-month terms lent into toll roads with 15-20 year payback periods, outstanding credit peaking around 105 trillion VND. The ending showed up at the close of the decade: roughly a third of completed projects collected far less toll revenue than their financial plans assumed, banks were left restructuring the debt with bad-debt ratios several times the system average, the SBV warned again and again, and lenders shut the door on new BOT almost in unison. That is why the 2021-2025 expressway wave had to switch to public investment - phase 2 of the North-South expressway moved all 12 subprojects onto the state budget, openly because banks had lost their appetite - not because the old model was refuted in theory, but because it had demonstrated its own limit with real money: a well of short-term funding cannot nourish a 20-year toll asset. And the BOT lesson is not only about maturities - much of the wreckage came from the contracts: tolls held down, plans changed, parallel roads opened, with no clause saying who covers the difference. Until the state-private risk allocation is written down explicitly (what the revised PPP framework is attempting), switching funding sources only changes who holds the same unclaimed risk. Now, with the new pipeline calling PPP projects by name again, the question "what source besides credit?" returns intact - the lesson was sitting at home all along.

Why does Vietnam still depend on banks this much? The familiar answer - the corporate bond market is small (about 10% of GDP), institutional investors are thin, the stock market isn't deep enough - merely restates the symptom. Ask why it stayed small and the picture changes: Vietnam's capital market is no slow-growing child waiting for its growth spurt. It stayed shallow because for decades, every link in the system had a perfectly practical reason not to need it.

Seen from the saver's side: a bank deposit is an implicitly protected investment. In the living memory of Vietnamese depositors, no deposit has ever been wiped out - however weak a bank gets, it is merged or put through a mandatory transfer. A corporate bond is the opposite: when things go wrong, bondholders have almost no effective tool to collect - bankruptcy procedure is rarely used in practice, and collateral resolution drags on for years. Same dong of savings: one comes with someone to shoulder the risk for you, the other leaves you carrying it - the deposit is the smart play under the rules as written, more than any habit. Seen that way, the 2022-2023 shock is hard to call an "accident" that set the bond market back: it was the moment the market discovered that many people buying bonds at bank counters thought they were making a high-interest deposit. A market inflated by misunderstanding deflates once people understand correctly.

Seen from the household side: Vietnamese families have always done their long-term saving in land, hardly ever in a financial product. Land has beaten inflation across multiple cycles, carries virtually no holding tax, and requires trusting no one's financial statements. Add the memory of double-digit inflation in 2008-2011, and holding a piece of paper promising fixed VND for 20-30 years demands a kind of trust that history has not yet had time to build. As long as real estate remains the household's default "pension fund," the long-term portfolio will stay anchored in land rather than long bonds - money spent on land does not vanish from the system (it merely changes hands to the seller), but demand for long paper never forms, and long-term institutions lack the raw material to grow. Seen from the issuer's side: a bank loan built on relationships and collateral is far more comfortable than a bond issue, which requires disclosing information transparently to thousands of strangers. And seen from the policy side: bank credit is the easiest channel to steer - credit-room quotas, policy rates, guidance through state-owned banks - while a capital market prices risk in the open and does not negotiate; a system used to managing capital flows with administrative tools will naturally grow fastest in the channel those tools can reach. When all four sides - savers, households, issuers, policymakers - behave rationally within the existing structure, the 134% did not fall from the sky: it is the sum of millions of rational decisions, piling the risk up in one place.

Why Not Shift It All To Government Bonds?

The natural next question: Vietnam does issue 10-30 year government bonds, so why not use that channel instead of bank credit? The textbook answer - public debt rises, the credit rating comes under pressure, crowding out - sounds fine, but it misses the very argument this piece opened with: Vietnam's binding constraint is not the public debt ratio. To answer honestly, ask a more concrete question: if the government issued another few tens of billions of dollars in bonds, who exactly would buy them?

Look at who holds Vietnamese government bonds today and the answer is somewhat ironic: mostly commercial banks and the Social Security fund - the IMF has said plainly that this investor base is too concentrated, lacking private pension funds and large insurers. The low government bond yields noted at the start of this piece carry this structure's fingerprints: rate expectations, inflation, and liquidity still shape the price, but when a large share of demand is mandatory, the yield gets pressed below what a market of voluntary buyers would demand. Which means "shifting from bank credit to government bonds" is largely the same money changing doors: household deposits still sit on bank balance sheets; the banks simply buy bonds instead of making loans. The maturity mismatch does not disappear, but nor does it keep its shape: government bonds can be repoed with the SBV, carry a 0% risk weight, and count toward liquidity reserve ratios - liquidity risk genuinely falls compared with a project loan sitting dead on the book. What stays fully intact is interest-rate risk: holding 10-30 year paper on 6-month deposits means that when rates rise, the mark-to-market loss sits right on the balance sheet. The well is still the same well of domestic savings at 36.5% of GDP: issuing bonds creates no new savings, it only decides who draws first. And when the government draws first and draws big, the private sector - factories, data centers, power plants - queues behind: crowding out here sits closer to arithmetic than to theory. Closer - not equal: the well's bottom is not rigid; higher rates coax out more savings, and banks create money when they lend. But every notch of that elasticity carries a posted price - interest rates, inflation, and ultimately the exchange rate, as the later part of this piece returns to.

And if the buyers are foreign investors - the international bond option the July 18 conference itself proposed - the loop closes exactly where this piece began: borrowing in dollars means trading maturity risk for currency risk, and original sin grants no exemption to the public sector. The government bond channel therefore remains necessary - for the government's own share of the work - but it is no magic wand: it only solves the problem when there are genuine long-term buyers, which circles back to the gap analyzed above - missing pension funds, missing insurers, missing institutions able to hold a 30-year piece of paper. Until that investor class is built, switching channels is just moving debt between pockets of the same coat.

The Path Taking Shape: Lock The Money In, Collect It Directly

While the long-term investor class is not yet born, the rapid-fire moves of 2026 sketch out a different approach, with two prongs. Prong one: keep the money inside the yard. In February 2026, Directive 12/CD-TTg ordered the SBV to urgently find ways to mobilize the foreign currency and gold bars held by households and to set up a national gold exchange. At the same time, gold-bar trading was tightened, with fines up to 400 million VND and license revocation and an invoice required for every transaction; crypto assets are being piloted on domestically operated exchanges. Read the pieces fairly: not every piece is a lock - the revision of Decree 24 under way since 2024 leans toward opening (ending the SJC gold-bar monopoly, licensing raw-gold imports), and legalizing crypto pulls a gray capital channel into the light. But set against the fines, the per-transaction invoices, and the requirement that exchanges operate onshore, the common denominator still emerges: every door through which household savings could leave the VND - gold, dollars, crypto - is being fitted with a meter, to be seen and taxed; whether those doors then get turned toward open or toward locked is a choice still left pending.

Prong two: the state collects the money itself, without waiting for the banks. In the first half of 2026 the tax service collected 1.38 quadrillion VND, up 16.6% year on year - e-commerce alone brought 167.9 trillion, up 44% - after presumptive taxation for household businesses was abolished from January 1, 2026 and tax obligations started being computed straight from e-invoices and bank cash-flow data. In parallel, the State Treasury was tasked with issuing 500 trillion VND of government bonds this year - and to sell them, the average issuance yield had to creep up to 4.09% a year, 0.83 points above 2025.

The circle closes at the most thought-provoking spot: the tax money and the bond proceeds do not travel far - they sit at the Treasury, and the Treasury deposits them with banks. At the end of March 2026, Treasury deposits in the system reached about 626.7 trillion VND, over 99% of it at the four state-owned banks, up sharply from end-2025 - and by June the government had to ask the SBV to study raising the share of Treasury deposits counted as bank funding, handing liquidity back to the very system it had just drained. The state draws water from the village well into its own tank, then lends it back to the village through four state-owned doors - the well holds no more water; there is simply one more person standing in the middle deciding who gets to draw.

In fairness, the Treasury balance has one more cause, more mundane than any deliberate hoarding: the money cannot be spent fast enough. Public investment disbursement is the machinery's chronic illness - in 2021-2024 it ran at roughly 73-83% of the annual capital plan - appraisal, land clearance, land valuation; money already allocated just sits at the Treasury waiting for paperwork. The detail matters because it flips a hidden assumption of the whole mobilization story: for years, the binding constraint of Vietnamese fiscal policy has usually not been the ability to raise money, but the ability to spend it on time. This does not make the loop above harmless - the state still pulls purchasing power in first and returns it to the economy slower than it draws, and that lag is a net drain of liquidity from the private sector - it just means the culprit is not a master plan of accumulation, but a spending pipe clogged on the very segment the state itself manages.

Both prongs can easily backfire, in two different ways. On the capital locks, the China lesson above carries a precondition that tends to get dropped: Beijing could lock the yard because behind it stood a trillion-dollar trade surplus - close the valve and water still pours in on its own. Vietnam sits in the opposite position, needing inflows more than ever, and to foreign investors "easy in, hard out" is the most frightening signal a market can send. As for residents, controls do not erase the need for shelter - they only push it into channels that are darker and more expensive: the onshore-offshore gold gap, crypto through private wallets, mis-invoiced trade. The gap between Vietnamese and world gold prices really has two layers: the base is supply monopoly - Decree 24/2012 reserves the gold-bar brand and imports to the state, which is why the gap narrows whenever imports are loosened; the swings on top are the thermometer of shelter demand - and the tighter supply is locked, the higher the thermometer jumps. "Mobilizing the dollars held by households" runs straight into a policy contradiction that Directive 12 does not mention: since late 2015, the ceiling on USD deposit rates in Vietnam has been 0% a year - the pillar of the de-dollarization strategy the SBV has defended for a decade. To pull dollars in through official channels you have to pay interest, which means reversing that policy with your own hands and inviting dollarization back in; keep the 0% ceiling, and "mobilize" can only mean persuading people to sell their dollars outright for VND - something they could always do, and have mostly chosen not to. As for "mobilizing household gold," Vietnam has tried it: banks were once allowed to take gold deposits and lend in gold, and the mechanism had to be shut down in the early 2010s because of the risks it created; mobilization succeeds only when depositors trust they will get back exactly what they handed over - the same 30-year trust this whole piece keeps finding unbuilt.

On the tax prong, the paradox sits inside the opening arithmetic itself: the target is 10% growth, which needs the private sector spending and investing harder, while every extra dong of tax collected is a dong of purchasing power leaving the private sector ahead of schedule - a subtraction that stays net only for as long as spending lags collection, and slow disbursement, as just noted, is the chronic disease. The tax base's response also had its first draft in mid-2025: when point-of-sale invoicing was tightened, the first reaction of many household businesses was not to pay more but to flounder, shut down, or shrink. Read that signal at its proper size: total collections still rose 16.6%, meaning the shrinking edge has not yet shown up in the total - "the tax base is contracting" is for now a hypothesis to watch, not a conclusion. But that very edge - small household businesses, the informal sector - is exactly where the base-broadening policy is aimed, and it is signaling in the opposite direction. What the two prongs share: they treat the symptom - money not flowing where the state wants it - rather than the cause, which is why people do not hand it over voluntarily. Every extra turn of the screw teaches households one more lesson in keeping wealth beyond the system's reach - wearing down exactly the trust a long-term capital market needs.

Back To The Opening Constraint: Where Do The Real Dollars Come From?

The domestic funding channel is one half of the problem; the other half circles back to the constraint this piece opened with. A large investment cycle drags in imports of machinery, materials, and energy - real dollar demand - while Vietnam's true policy headroom is measured in FX reserves and the exchange rate. If the public sector pumps in more money and dollar inflows from exports, FDI, and remittances don't keep pace, the pressure lands on the exchange rate, and the SBV has to choose between selling reserves and keeping interest rates higher.

The fastest transmission channel for that pressure - 2022-2025 demonstrated it live - is not even the import bill, but the USD-VND interest rate differential. Normally, dong deposits pay more than dollar deposits, so anyone holding dollars wants to swap into dong for the interest. But every time the Fed holds dollar rates high while the SBV presses dong rates down to support growth, the math flips: holding dollars now pays the higher interest and gains again if the exchange rate rises. So exporters bring their dollars home but hold off selling them to banks, speculative money turns the same way, and reserves bleed even in years when the current account runs a fat surplus - because the SBV must sell dollars to hold the rate. That is why bank treasury desks read exchange-rate pressure each morning off swap points, not off the goods balance - and that is how the foreign-currency constraint tightens through expectations, before any real flow has changed direction.

This constraint does not wait for a project to import anything. Even a build that uses only domestic materials and labor still consumes an enormous amount of VND, and that VND has to come from somewhere. Raised from real savings - bonds, taxes, deposits - it merely moves purchasing power from one pocket to another. Created fresh, without mobilization, the outcome depends on how much slack the economy has left: with idle capacity, new spending can call up real output; but in an economy already levered near 150% of GDP and chasing double-digit growth, little such slack remains - total purchasing power swells while the supply of goods has not yet caught up, and the excess sooner or later finds its way into imports, gold, and dollars. Foreign-currency demand still shows up, just by a detour - through prices and exchange-rate expectations rather than through machinery invoices. For a currency the world does not hold, "spending in VND" shields very little: whatever is spent beyond real savings must eventually be absorbed somewhere - by inflation, the exchange rate, interest rates, or FX reserves; and as long as the exchange rate is held steady, that road leads back to the same two doors: higher interest rates, or a thinner reserve cushion.

What about earning the dollars through exports - the road China used to bankroll the entire land-leverage game above? At more than $400 billion a year in export turnover, Vietnam can look like it owns a similar machine. But Le Xuan Nghia, former vice chairman of the National Financial Supervisory Commission, broke that number down at a mid-2025 conference on building a self-reliant economy: of the 400-plus billion, roughly $300 billion belongs to FDI firms; Vietnamese firms produce about $100 billion, half of it agriculture - and the value that truly stays with domestic industry is only about $17 billion. He calls it an industry doing "contract work" for the world: "we are exporting on their behalf, not our own." Vietnam's export machine, in other words, generates turnover more than it generates net dollars for the country - much of the value flows back to parent companies abroad as imported components and repatriated profits. The $17 billion figure should be read as a policy-advocacy number more than an official statistic, because it mixes value-added accounting with foreign-currency flows: the wages, electricity, domestic suppliers, and taxes the FDI sector pays are all dollars that stay in Vietnam, and domestic value added in exports on OECD TiVA-style estimates runs around 40-50% of turnover - many times $17 billion. The cleaner yardstick for "net dollars" sits in the balance of payments, and there the official evidence - a primary income account running about $18-20 billion negative each year, mostly FDI profits returning to parent companies - ironically still supports the "exporting on their behalf" argument, just with a number that measures the right thing.

And just as the capital-mobilization story heats up, the goods balance has flipped sign. The last 12 quarters of NSO data show the trade surplus thinning from 2024 onward despite the occasional rebound quarter, and in Q1 2026 flipping into a deficit of about $3.7 billion - a signal worth watching to see whether it is input stockpiling for the projects or genuine foreign-currency demand:

The cushion behind it is also thinner than it used to be. According to World Bank WDI and IMF data, Vietnam's total FX reserves rose from roughly $55.5 billion in 2018 to a peak around $109.4 billion in 2021, then fell to about $86.5 billion in 2022 as the SBV had to intervene during a period of broad dollar strength, recovered to roughly $92 billion in 2023, but slipped back to about $83 billion in 2024. The IMF's 2025 Article IV puts that end-2024 level at roughly 2.5 months of projected imports - below the IMF's ARA benchmark. In other words: the big investment cycle is starting just as the goods balance changes sign and the dollar cushion grows relatively thinner. Foreign capital - FDI, international borrowing, international bonds - therefore becomes the variable that decides whether this cycle passes smoothly, hardly a "supplementary channel" anymore.

Plenty Of Water, Too Few Pipes

Put the situation in plain words: the money is there, but in the wrong shape - and nothing suggests the shape is about to change. Domestic savings of 36.5% of GDP live in 6-12 month deposit books and in plots of land, and as the earlier section showed, that is each person's smart play under the current rules; the rules have not changed. An infrastructure project needs the exact opposite: someone willing to hand over money and not ask for it back for 20-30 years. The layer of institutions built for that job - pension funds, life insurers - does exist in Vietnam, but it is far too small for the need, and it is moving backwards. Social Security is the dominant government bond buyer, as an earlier section noted, but it is a state fund that buys almost nothing else - not the investor who will hold a power plant's bond. Life insurers held total assets of about 860 trillion VND at end-2024 (some 7% of GDP) and are an important long-bond buyer, but they just lost a generation of customers to the mis-selling scandals at bank counters - new premium revenue has fallen year after year since 2023. And private pension funds, nearly a decade after getting a legal framework, remain too small to register in any macro arithmetic. Set that scale against a project pipeline in the hundreds of billions of dollars and it is a drop beside a bucket - so every long-term funding need still ends up knocking on the only door open: banks, which live on 6-month deposits.

None of this means every door is shut - some are genuinely opening, right in this window. The stock market has just been upgraded by FTSE Russell to secondary emerging market status (announced October 2025, taking effect in 2026), opening the way for foreign inflows; the international financial center project in Ho Chi Minh City and Da Nang has its framework resolution from the National Assembly; the PPP framework is being revised toward more explicit risk-sharing. But measure them at their true size: foreign money in equities can leave within a session, no holder of 30-year infrastructure paper - and in practice foreigners have been net sellers for months running, cumulatively for years. The upgrade widens the pipe; the domestic long-term investor class - the missing character of this whole piece - does not appear on its own.

One more distinction that blanket statements tend to skip: not every project in the pipeline draws from the same well. The North-South high-speed railway - the biggest number in this piece at $67.3 billion - is, under Resolution 172/2024/QH15, a public investment project funded by the state budget and phased over more than a decade, roughly 1-1.5% of GDP a year, with almost no commercial bank borrowing: its burden falls on the budget, and on dollar demand for imported equipment. The real pressure on the banking well sits elsewhere: the generation and grid projects of Power Development Plan VIII - LNG, wind, transmission, where investors borrow commercially and import equipment and fuel in dollars - and the PPP expressway block, which domestic investors fund mostly with domestic credit. This decomposition does not make the problem lighter; it just delivers each bill to the right address: the budget carries the railway, the banking well carries power and expressways, and FX reserves carry everyone's dollar bill.

The most realistic thing that can be said about the "solution" is that the projects will not wait. Railway, power, and expressway timelines are set by political terms and growth targets; building a long-term investor class is measured in decades. Those two clocks disagree, and when they disagree, the experience of every economy that went first points the same way: the gap gets filled the familiar way. Banks will be "mobilized" to lend - and the mobilization tool has been in hand since the start of this piece: the credit-room bucket, handed out more generously to precisely the banks carrying the projects; safety ceilings will be loosened - raising the short-term-funding cap from 30% to 40% in mid-2026 was the first step, and notably it came before the investment cycle had properly begun; the budget will lean in; whatever remains will be borrowed in foreign currency. Each is a way of borrowing from the future: more maturity risk, more currency risk, another layer of debt on a credit structure already nearing 150% of GDP.

The three beliefs needed before people hand money over for 30 years - low inflation held for decades, a VND that does not devalue deeply, courts that recover something in a default - cannot be decreed into existence, and the third was just demonstrated in reverse by the 2022-2023 bond shock. A generation of savers learned the lesson "don't hold long paper" with their own money; that kind of lesson does not fade within one political term. Put the pieces side by side and the picture is fairly plain: Vietnam is launching a high-growth cycle just as bank liquidity runs tight, while household businesses and SMEs are still reeling from the formalization shock of taxes and e-invoicing, an apparatus wary of legal risk disburses slowly even the money it already has, and a huge public investment pipeline demands a quality of capital allocation above the machine's current execution capacity. The most likely scenario is therefore no mystery: the investment cycle will get funded anyway, risk will quietly accumulate on bank balance sheets and in foreign-currency obligations, and everything will look fine until the property cycle turns or the dollar strengthens globally - two things that never announce themselves in advance. Read to the end, the message of 2025-2026 is an admission: the old model has reached its limit but will have to run for at least one more cycle, because the institutions meant to replace it have not had time to grow. The village will find out how deep the well is at the exact moment everyone needs water - the worst possible time to find out.