Asia-Pacific Investment Report 2012 — Who Pays for Rebalancing

By 2012 the rebalancing debate had a settled diagnosis and no settled politics.

The diagnosis: the region's largest economies had investment shares of output that were extraordinarily high by any historical comparison, and household consumption shares that were correspondingly low. The 2009 stimulus described in the Asia-Pacific Investment Report 2009 had pushed this further.

Why this cannot continue indefinitely is arithmetic rather than ideology. Investment is not a final use of output — it is a means of producing future output. A factory is valuable because it will make goods someone buys. If consumption does not eventually rise to absorb what the investment produces, the investment does not earn a return, and the capital is wasted regardless of how it was recorded when it was spent.

So the investment share has a ceiling set by the consumption that eventually justifies it. The debate was about where that ceiling sat, not whether it existed.

The obstacle was distributional, and this is the report's central point.

The high investment share was not an accident. It was funded by a set of policies that transferred income from households to the enterprises and governments doing the investing:

• Deposit rates were capped below what a market would set, so savers earned less and borrowers paid less.
• Household savings had few alternative destinations, given capital controls and limited domestic investment options — so the low return could not be escaped.
• Land and energy were often supplied below market cost to industrial users.
• And exchange rate management favoured exporters over consumers of imports.

Each policy has an identifiable beneficiary. Rebalancing means ending the transfers, which means the beneficiaries lose — and they are concentrated, organised and influential, while the gainers are hundreds of millions of households with no equivalent representation.

Everyone agreed on the destination. Rebalancing was slow because it was not a policy problem with a technical solution. It was a transfer, and transfers have losers who know who they are.

Why an investment share has a ceiling

The arithmetic is worth developing, because it is frequently mistaken for a value judgment about consumption versus saving.

Output is used in a limited number of ways: consumed by households, consumed by government, invested, or exported net of imports.

Investment is different in kind from the others. Consumption is a final use — the point of the activity. Investment is intermediate. It creates capacity that produces future output, and it is worthwhile only if that future output is wanted.

The chain that must close:

1. Investment creates productive capacity.
2. Capacity produces goods and services.
3. Someone must buy them — domestic consumers, government, or foreigners.
4. The revenue must be sufficient to justify the capital and service the debt that funded it.

If step three fails, the investment does not earn a return. The capacity sits idle or operates at a loss, and the debt raised to build it is a claim on income that the asset is not generating.

Two escape routes and their limits:

Export the output. This works and was the model's original logic. Its limit is other economies' willingness to run the corresponding deficits, which the Asia-Pacific Investment Report 2008 shows became binding after 2008.

Invest more. Building more capacity generates demand today for the output of yesterday's capacity — a construction boom consumes steel from the steel plants built last year. This works for a while and is self-referential: the demand justifying investment is more investment. It postpones the reckoning and enlarges it.

The observable indicator is the capital-output ratio — how much capital is required to produce a unit of output. A rising ratio means each additional unit of investment produces less, which is the empirical form of the ceiling approaching. World Bank and Penn World Table data cover this free for every economy.

Financial repression as a transfer

The mechanism funding the high investment share deserves plain description, because it is usually discussed in technical language that obscures who is paying.

The core policy: a ceiling on the interest rate banks may pay depositors, set below the rate a competitive market would produce and often below inflation.

What follows:

• Households receive a low, sometimes negative, real return on their savings.
• Banks fund themselves cheaply and lend at controlled spreads.
• Borrowers — largely enterprises and government-linked entities — obtain capital below its market cost.
• The difference is a transfer from savers to borrowers, continuous and large.

Why households could not escape it:

• Capital controls prevented investing abroad, so the money could not leave.
• Domestic capital markets were shallow, so alternatives were limited.
• Property was the main alternative, which contributed directly to the property demand described in the Asia-Pacific Investment Report 2010.
• And precautionary saving was necessary given limited social insurance, so households could not simply save less.

The scale is substantial. A gap of several percentage points, applied to household deposits equal to a large fraction of output, transfers a meaningful share of GDP annually from households to borrowers.

Three consequences that link directly to other findings in this archive:

Capital was cheap, so more of it was used. The capital intensity of growth is partly a response to its subsidised price — a rational reaction to an administered signal rather than a cultural preference for heavy industry.

Household consumption was suppressed, both because income was transferred away and because low returns on savings required saving more to reach any target.

And the search for higher returns drove the products described in the Asia-Pacific Investment Report 2013 — wealth management products existed because the regulated deposit rate was unattractive. The demand was created by the policy.

Financial repression is not a distortion of the growth model. It is the growth model's funding mechanism, and rebalancing means switching it off.

Why the politics are the hard part

The distributional structure explains the pace of change better than any technical account.

Who benefits from the existing arrangement:

• Enterprises with access to cheap capital — capital-intensive, often large, frequently state-linked.
• Local governments funding investment through land and borrowing, per the vehicles described in the Asia-Pacific Investment Report 2009.
• Exporters, from a competitive exchange rate.
• The construction and heavy industry sectors, whose demand is the investment itself.

These groups are concentrated, identifiable, organised, and have direct channels to policy.

Who benefits from rebalancing:

• Households, through higher deposit returns and higher consumption.
• Service sectors, which are labour-intensive and less favoured by cheap capital.
• Smaller private firms, which generally lacked access to the subsidised credit.

These groups are diffuse, numerous, unorganised, and individually gain modestly — a few percentage points on a deposit balance is real but not mobilising.

This is the classic political economy of reform: concentrated losses and diffuse gains produce organised resistance and unorganised support, regardless of the aggregate benefit.

Two additional frictions:

• The losses are immediate and the gains are gradual. Raising deposit rates damages borrowers' finances at once; the consumption response builds over years.
• And the transition has genuine financial stability risk. Enterprises and local entities that borrowed at subsidised rates may not be viable at market rates. Raising rates quickly could force widespread distress, which is a legitimate reason for caution rather than merely an excuse.

The observed path was gradual and partial, which the Asia-Pacific Investment Report 2015 and 2018 track.

The demographic clock

A second constraint was arriving on a schedule known decades in advance, and its predictability is what makes it analytically valuable.

The working-age share of the population had been rising for decades, which contributed to growth through three channels:

• More workers per dependant, so output per person rose even with flat productivity.
• High saving, because people in their working years save most.
• And an abundant labour supply, which kept wages competitive.

All three reverse when the working-age share peaks:

• Dependency ratios rise, so output per person falls unless productivity rises to compensate.
• Saving falls as retirees draw down.
• And labour becomes scarcer, raising wages — good for households and consumption, challenging for a cost-competitive export model.

The timing was known. Demographics are the most forecastable variable in economics — everyone who will be of working age in twenty years has already been born, so the projections are close to arithmetic.

Why a known constraint still binds:

• Nothing acute happens on the peak date. The change is gradual, so there is no forcing event.
• The policy responses are slow. Raising participation, improving productivity, and adjusting retirement ages take decades.
• And the fertility decisions that would alter it are individual, responding to housing costs, childcare and career structure rather than to policy exhortation.

The interaction with rebalancing is direct and helpful. A shrinking workforce raises wages, which raises household income share, which supports consumption. Demographics push in the direction rebalancing requires — which is genuinely fortunate, and slower than the investment share needed.

The Japan Investment Report 2019 covers the economy furthest along this path, and the Asia-Pacific Investment Report 2022 covers the region's later position.

What the middle-income transition actually is

The "middle-income trap" was much discussed in this period and is often described imprecisely. The useful version is about which growth inputs work at which stage.

Catch-up growth has identifiable sources:

• Moving labour from low-productivity agriculture to higher-productivity manufacturing. Large gains, available only while surplus rural labour exists.
• Adopting technology developed elsewhere. Cheap and fast, because the development risk was borne by someone else.
• Capital deepening — giving workers more equipment, which raises output per worker sharply when starting from a low base.

All three exhaust:

• The labour transfer ends when surplus agricultural labour is absorbed.
• Technology adoption ends as the economy approaches the frontier — there is progressively less to copy, and what remains is the hardest.
• Capital deepening exhibits diminishing returns. The tenth machine per worker adds far less than the first.

Beyond that point, growth must come from innovation and productivity improvement, which require completely different inputs: research capability, competitive product markets, capital allocated by return rather than by direction, and institutions supporting new firms displacing incumbents.

The trap is not a wall. It is the point at which the inputs that produced catch-up growth stop working, and the inputs required for frontier growth are institutional rather than physical.

Why the transition is genuinely difficult: the institutions that deliver catch-up growth — directed credit, coordinated industrial policy, protected national champions — are frequently the opposite of the ones that deliver frontier growth, which requires competition, capital reallocation and tolerance of failure. The transition means dismantling arrangements that worked, which is politically harder than building new ones.

The India Venture Capital Report 2017 and Asia-Pacific Investment Report 2019 cover economies at different points on this path.

Where rebalancing becomes visible

If rebalancing is slow and partial, the practical question is how to tell whether it is happening at all. The headline consumption share is the worst available indicator, and better ones exist.

Why the consumption share moves slowly and misleadingly:

• It is a ratio, so it changes only when consumption grows faster than everything else — which can take years to become visible even when the underlying shift is real.
• It is measured with substantial error in economies where informal activity, imputed housing services and household surveys are all imperfect.
• And it can move for irrelevant reasons, such as an investment slowdown that raises the consumption share without any improvement in household welfare.

Better indicators, all of which respond faster:

Services share of employment. Services are labour-intensive and consumption-driven. A rising services employment share means the economy is reallocating people toward meeting domestic demand, which is the physical form of rebalancing.

Real deposit rates. Since financial repression is the funding mechanism, the deposit rate relative to inflation measures directly whether the transfer is being reduced. It is published monthly and needs no interpretation.

Wage share of national income. Rebalancing requires households to receive more of what is produced. The wage share captures this at the source, before any decision about saving or spending.

Social insurance coverage. Since precautionary saving is a rational response to its absence, the coverage and generosity of health, pension and unemployment provision is the mechanism through which the saving rate eventually falls. The ILO publishes this free.

And import composition. A rebalancing economy imports more consumer goods relative to capital goods and commodities. This is observable in customs data monthly, and it is difficult to misreport.

The consumption share is the outcome everyone watches and the last thing to move. The wage share, the real deposit rate and the services employment share all move first, and all are published free.

Applying these to the period gives a mixed but readable picture — real deposit rates improved gradually, the services share rose steadily, and the wage share moved least. Which is consistent with the political economy above: the reforms that were easiest to implement were the ones with the least concentrated opposition.

What an allocator could act on

Watch the capital-output ratio, not the investment share. A rising ratio means each unit of investment produces less, which is the observable form of the ceiling approaching. Penn World Table and World Bank data cover it free.

Identify who pays for a growth model. Capped deposit rates transfer income from households to borrowers continuously, and that transfer is the model's funding mechanism rather than a distortion of it.

Read reform pace from the distributional structure. Concentrated losers and diffuse gainers predict slow, partial change regardless of the aggregate case for it.

Expect demand for higher-yielding products where returns are administered. Suppressed deposit rates create the demand that non-bank products exist to meet, so the policy predicts the product.

Treat demographic turning points as scheduled, not as risks. The working-age peak is close to arithmetic and its direction on wages, saving and consumption is knowable decades ahead.

Assess whether an economy's institutions match its growth stage. Directed credit and protected champions deliver catch-up growth and obstruct frontier growth, so the transition requires dismantling what worked.

What 2012 established

• The investment share has an arithmetic ceiling set by the consumption that eventually justifies the capacity.
• Financial repression was the funding mechanism, transferring income from households to borrowers.
• Rebalancing is distributional, with concentrated losers and diffuse gainers determining its pace.
• The demographic tailwind was scheduled to reverse, pushing helpfully toward rebalancing but slowly.
• The middle-income transition requires different institutions, not more of the ones that produced catch-up growth.