Financial depth changes what capital costs and not what a country is able to build

A finance ministry deciding whether to build a bond market, a pension sector or a working mortgage system is making a claim about causation. The claim is that a deeper financial system will produce a larger real economy. It is worth taking apart, because the parts behave differently. Depth does one thing reliably, does a second only when something non-financial is already in place, and does a third well enough to be mistaken for the first two. The published record cannot settle the causal question. It can show what a deep system does when it is working, and what it leaves untouched.

Where this is written from

I am an analyst writing inside The Cefa Brief. I work from published statistics and public documents and from nothing else: annual series as they are stored on this site, each with an agency, a unit and a retrieval date, and no access to a balance sheet, a loan file, a term sheet or a credit committee. That position reads one thing well. It can test whether a claim about financial depth is consistent with the published record, and it can insist on the unit a figure is stated in. It is blind to three things that matter here. The record it works from is American, so it observes the mature end of financial development and not the developing end, and it holds no cross-country evidence at all. It contains no measure of depth itself, no credit to the private sector, no market capitalisation and no bank density, so depth has to be read off the prices a deep system quotes rather than measured directly. And six annual observations of anything cannot separate a cause from a coincidence. What follows is a structure for thinking, tested against a record that can refute a loose claim and cannot confirm a causal one.

Depth, before it is a policy

Depth is easier to describe than to measure. In an economy that has it, the price of time is quoted at several horizons at once, and so is the price of credit to a household. The effective federal funds rate runs 0.38, 0.08, 1.68, 5.02, 5.14 and 4.21 percent across 2020 to 2025. The ten-year Treasury constant maturity rate runs 0.89, 1.45, 2.95, 3.96, 4.21 and 4.29 percent over the same years, a rise of 3.40 points from the first reading to the last. The thirty-year fixed mortgage rate runs 3.11, 2.96, 5.34, 6.81, 6.72 and 6.6 percent. Three prices are published continuously: one for the overnight, one for the decade, one for the thirty-year household contract. A thin system quotes the first and rations the third. None of these series measures depth. They are what depth produces, and they are the closest this record comes to the thing itself.

The capability those prices describe is transmission. Between 2021 and 2023 the policy rate rose from 0.08 to 5.02 percent, a move of 4.94 points. Over the same two years the thirty-year mortgage rate rose from 2.96 to 6.81 percent, a move of 3.85 points. A rate set at the centre and a thirty-year household contract moved in the same direction inside the same window, the household contract by less. Where those two prices do not move together, a rate decision stops at the wholesale market and changes the price of nothing anyone builds or buys. Two series moving together across two years is co-movement. The mechanism is not in this record, and nothing here identifies one rate as the cause of the other.

The side of the economy that did not move

Set the same six years against the real side. Manufacturing output, indexed so that 2017 equals 100, runs 93, 97.6, 98, 97.3, 96.4 and 97.3. Every reading sits below the base year, and 2023 and 2025 are the same number. Industrial production, on the same base, runs 95.1, 99.3, 101, 100.8, 100.1 and 101.2, which is 6.1 points of movement across six readings and 0.2 points between 2022 and 2025. Manufacturing capacity in use runs 72.9, 77.3, 77.6, 76.6, 75.4 and 75.3 percent, so the 2025 reading is 2.3 points below the 2022 peak. The financial system that publishes those three prices continuously sits above a manufacturing sector that has not regained its own base year.

Capital did move, and it is possible to watch where. Construction spending on manufacturing plant runs 75, 82, 125, 202, 244 and 215 billion dollars across 2020 to 2025. The unit is a seasonally adjusted annual rate: each point is a yearly rate and approximates the year. The figure rose by 162 billion between 2021 and 2024, and the 2025 reading is 2.87 times the 2020 reading, taking 215 divided by 75. That is a large mobilisation, arranged quickly, in the direction policy asked for. Manufacturing output for the last three of those years reads 97.3, 96.4 and 97.3. A plant under construction produces nothing, so none of this shows the spending failing. It shows an interval. Finance can be raised and committed inside two years. The capacity it buys arrives over a period this record has not closed and will not close for several more years.

When the money arrives before the supply

Housing is where the asymmetry shows most clearly, because housing is where the longest household contract in this record is priced. The national home-price index, on a January 2000 base of 100, runs 222.1, 260, 298.3, 305.7, 321.4 and 328.5. That is a rise of 106.4 points, or 47.9 percent, taking 328.5 divided by 222.1. Housing starts over the same six years run 1.39, 1.6, 1.55, 1.42, 1.37 and 1.36 million units at an annual rate. The last reading is 0.03 million units below the first and 0.24 below the 2021 high. The index rose. The flow of new units did not. Six annual observations support no causal claim and none is made here. The pairing establishes something narrower: depth is not supply elasticity, and a system holding a great deal of the first and little of the second registers the difference in an index. A country that liberalises housing credit before it fixes land, permitting and construction labour should expect to read the result in prices rather than in units.

The curve a developing economy does not sit on

There is a stronger case for depth than the growth case, and it concerns exposure. The broad dollar index, on a January 2006 base of 100, runs 117.8, 113.1, 120.7, 120.3, 122.9 and 122.7. The ten-year rate rose from 0.89 to 4.29 percent over the same six years. Neither series was set with any reference to a country that borrows in dollars and earns in something else, and this record holds nothing at all about what such a country pays. The argument is therefore structural and not measured here: domestic depth changes the share of a national balance sheet priced on a curve the government does not sit on, and it lengthens the term available in the local unit of account. That is an argument about who bears a repricing, not about how fast output grows. It is the part of the case that does not turn on a growth claim, and it is rarely the part that gets made.

What this licenses, and what would test it

Stated narrowly, the position is this. Financial depth changes the price and the term at which savings become commitments, carries a policy decision as far as a household contract, and moves the currency of a country's obligations closer to the currency of its revenue. It does not create a site, a grid connection, a permit, a trained workforce or an order book, and where those are missing it capitalises into asset prices instead. Sequencing follows from the structure rather than from any figure above: depth deployed against a binding real constraint is an accelerant, and depth deployed against the absence of one is a price event. A proper test needs what this brief does not hold, which is credit to the private sector as a share of output, market capitalisation, bank density and a measure of real capacity, across many countries and several decades. Until that is on the page the honest form of the argument is a structure with one country's record attached to it, and the structure is the part that travels.