How do wages (and thus shelter affordability) figure into these models? Is this simply captured in the inflation figure?
What about I recall hearing stories of the 1970s and earlier that it was harder to get a mortgage than today as banks were more risk-averse in lending to individuals.
The model as developed this far doesn't capture wages, but doesn't ignore them either. In short, wages factor in through the input scenario, rather than the model itself!
As I illustrated in the section "How to understand a system", a model can be built up piece by piece, and what I have here is only a piece of a full model. Because of that, it requires certain inputs to model the future -- like the assumption that housing prices stay flat, or that interest rates go to a certain level, or that inflation in other categories drops to 2%.
A more complete economic model would be able to be run forward into the future without any such assumptions, because all the inputs it needs to make those predictions would be drawn from its own outputs. Things that look like independent variables to a partial model, like this one, can end up being dependent variables in a more complete model. I aim to add some components in that direction in part 2 of this series (which is still being written) but I'm not going to develop a full model of the economy.
The shelter costs in my model are derived from inputs like the price of houses, the interest rate, and costs in other categories. That should be _fairly_ accurate independent of wages. But, if wages fell or failed to grow, that would strongly impact all three of those inputs to the shelter model. For example, housing prices might crash deeply as people are forced to sell their houses to pay the bills that their wages don't; interest rates might reverse as the central bank notices the low wages indicate a weakening labour market; inflation in other categories (affecting things like lumber prices and the cost of maintenance) might decrease as well.
On the other hand, rising wages could help support the housing market, create more impetus for faster rate hikes, and so on. All of those inputs, affected by wages, were taken-as-given in the model as it is so far.
I expect banks are about to get a lot more risk-averse in lending. As interest rates go up, so will the stress-test cutoff, and I'm already hearing that some banks are reducing the duration they'll hold a rate for a borrower who qualifies for a mortgage.
I think your missing variable is population growth. The 1950 through 1970 are the peak years of a) the baby boom, and b) post-war immigration. Rental pricing is decided at the margin, and people needed ever-more space through those decades. And the stock of rental changes slowly, with a long delay. https://en.wikipedia.org/wiki/Population_of_Canada#/media/File:Canada_immigration_graph.png
* Wouldn't that predict more rapid rental price gains over those decades? My surprise was how the CPI indicated that the increase in the nominal price of rent was so little, rather than so much.
* The way that my model accommodates confounders like population growth (and it's for sure not perfect) is that it still takes housing prices as an input. An increase in population would result in pressure on all forms of shelter, including both rented and owned housing. My model estimates rent based on the opportunity cost of home ownership, which will grow along with housing costs in general when population pressure drives them up. If a baby boom drove up rental costs only, without affecting the cost of homeownership, it would either drive a lot more people to switch from renting to buying, or incentivize a lot of property investors to buy up homes and rent them out, or both.
Your chart of real rents shows a sustained and basically flat plateau 50-70, and then a decline which matches the period of the boom. I speculate that the 40s would show the matching rise for the beginning of the boom. But your comments on the mix are pointed. My (unjustified) speculation is a mix of: rent control legislation (1945-1955, then 1974 on on Ontario), the differential lag between suburban housing vs purpose built rental, fashion and explicit policy preferences for suburban growth as part of the highway system rollout (sfh over rental) and more. But I’m over my head there.
I think the variable is not directly population growth, but rather "housing stock" as the Reserve Bank of Australia call it, I believe. This is the float in the market of how many properties (shelters) are available to purchase.
This effectively decouples population, as it is simply a measure of supply and demand with the basis that more housing stock there is the lower the demand is to purchase.
A more subtle secondary variable (that could be considered to provide a leading indicator, so a first derivative you might say) is days to sell, or percentage clearances at auction, which is just the percentage of auctions resulting in a sale.
Low stock and high clearance rates (or short days to sell) create FOMO which becomes like a bubble and leads into non linear territory and the desire to buy, and buy now, becomes almost viral. You hear people saying, "I need to buy now, no matter what, because the prices are rising so much faster than I can save for a deposit" . It is in these conditions people can "flip" houses in weeks for profit, but also when people over commmit, because FOMO.
Finally, I am in Australia and my sister in law works for the biggest conveyancer (is licensed to settle the sale transaction and change the title of real estate) in our city and she says the single biggest indicator of whether people are prepared to take out a loan is not the size of the loan, but how confident they are they will have work for the next 3-5 years, sometimes less in a fast rising market. This is because they feel that worst case they can always sell up and not make a loss, as along as they can stay in for 1-3 years and get a good capital increase, despite loss of capital to fees and taxes.
I like your work, I am a control systems engineer (though in practice not usually as theoretical as here) and always wondered why economists did not lean more on control systems theory, especially regards stability and controlability. But also, as I have pointed out above, there are some non-linearaties that can easily creep in once the market conditions are not so steady and humans are involved, then all bets may be off.
Completely agree, but isn’t that the tricky bit? That actual controls available are very different between the players: the central bank can use interest rates to modulate house construction rates (2nd order from the stock, with a 5-10 year lag), cities can vary zoning, fees, and rent control (same). Higher governments might build hosing directly (almost direct control), or change immigration levels ( direct control of next years demand side). All of them are themselves subject to democratic control, but with different sub-grouping and indirection (regular direct elections of municipal politicians, indirect appointments of ban officials, etc). The political economy of it is incoherent, and each player works with the available variables.
> They are not the ones in charge. They are not even the source of money.
That will come as news to them! Central banks around the world take stabilizing inflation as part of their mandate, or central to it. Certainly the Bank of Canada does. They seek to do this through tools like setting upper and lower boundaries on the interbank interest rate and, more recently, purchasing bonds.
> Turns out interest rate hikes run into a fundamental problem with the accounting.
Well, the article you are commenting on is detailing precisely one potential interest/price spiral that the Bank of Canada should be concerned about. Because shelter, like energy, is a fundamental input to almost all goods, I wouldn't be surprised if this results in an increase in costs of other goods as well.
That doesn't mean it's the wrong move. As I've explained, I believe this is a right-half-plane zero, and that means inflation might go up before it goes down.
"They seek to do this through tools like setting upper and lower boundaries on the interbank interest rate and, more recently, purchasing bonds."
Yes. They don't work - for the reasons explained in the paper, and the feedback loops in the article linked. The steering column isn't linked to the wheels. The system assumptions on which they are based are wrong (exogenous money and general equilibrium) and naive (the response of individuals and businesses to rate rises and price changes is not necessarily to borrow less or repay loans).
In academic terms 'they have uncertain distribution characteristics'. The amount of input on that particular zero is likely going to have to be Volcker levels to have any material effect and which will bring on recession and unemployment.
"Because shelter, like energy, is a fundamental input to almost all goods"
Shelter is derived from energy and labour. The roots of the current inflation lie in energy acquisition, and are amplified via shelter as you rightly point out. The current problems is likely due to a shift in the terms of trade relative to energy. Messing around with interest rates is unlikely to change that - particularly when the rest of the world is doing the same.
More likely, as you've discovered, it will push inflation higher than if they hadn't bothered and just let the taxation auto stabilisers drain the money from the system.
> Yes. They don't work... the steering column isn't linked to the wheels.
That's entirely plausible, and I don't take a strong view on it one way or the other. I'm more interested in looking at how the economic system works, from within the framework of how the central bank *assumes* it works.
That's because, ultimately, they're the ones at the wheel (even if it's not connected). When evaluating how best to steer, they'll be using the models they trust. Even if that wheel is disconnected from the steering column, I find myself very interested in anticipating how they're going to try to turn it. After all, my focus is on control theory first, and economics second.
If the economy doesn't react to their feedback the way they expect, that's important too. But my uneducated guess is, that steering wheel is probably connected to the steering column more tightly in Canada than it is in the US.
> is likely going to have to be Volcker levels to have any material effect and which will bring on recession and unemployment
Sometimes, that can be the unfortunate consequence of a borderline-uncontrollable system. It's great when a system can always be smoothly steered to reach any desired operating point by continuous gentle application of small corrective inputs, but in real systems, that luxury doesn't always exist. Especially if you find yourself suddenly so far away from the target, as we have now.
If you're driving fast and you notice the traffic light turning yellow up ahead at a busy intersection, easing gently off the gas pedal isn't going to be enough to avoid a collision. You have to slam on the brakes, hard, or run the light. If you're also being tailgated by an eighteen-wheeler with a sleepy driver hauling a hundred tons of cargo, then you might not have _any_ trajectory of inputs available to you that avoids an accident. And if it turns out the brake pedal is also mis-wired and accelerates the car, then... well, I hope at least the seatbelts are buckled.
To me, it's not really worth arguing the details of whether it's better to run the light or slam on the brakes. Either way we'll be relying way more on luck than any responsible driver should ever need to. Instead what I hope we can learn is _why_ we somehow trapped ourselves in that situation in the first place, and why it's so important that we avoid letting that happen in the future. Even if it means driving a bit slower, etc.
> More likely, as you've discovered, it will push inflation higher than if they hadn't bothered and just let the taxation auto stabilisers drain the money from the system.
I'm glad you mentioned taxation. That's of course another huge input; within certain limits, the government can pretty freely increase or decrease the money in circulation and in a pretty fine-grained way, more directly than the central bank. I have no doubt that the government's steering wheel is more tightly connected than the bank's.
I think there's a lot the government can do (and I have my own opinions about what they should do), but from a feedback systems perspective, the government is a little less interesting because they don't have an inflation mandate. Inflation is certainly not popular, and the government would rather it be low than high, but their reaction to high inflation won't necessarily be taking action to suppress it. (An understatement). Suppressing inflation usually requires a pretty unpopular policy, which is why they've outsourced that job to a (hopefully) independent party like the central bank.
So I can't realistically think of the government as a feedback controller. They're more of a wild card. I think they have an important role to play, for sure. But it will be easier to predict how the central bank responds to government policy, than how the government responds to central bank policy.
How do wages (and thus shelter affordability) figure into these models? Is this simply captured in the inflation figure?
What about I recall hearing stories of the 1970s and earlier that it was harder to get a mortgage than today as banks were more risk-averse in lending to individuals.
The model as developed this far doesn't capture wages, but doesn't ignore them either. In short, wages factor in through the input scenario, rather than the model itself!
As I illustrated in the section "How to understand a system", a model can be built up piece by piece, and what I have here is only a piece of a full model. Because of that, it requires certain inputs to model the future -- like the assumption that housing prices stay flat, or that interest rates go to a certain level, or that inflation in other categories drops to 2%.
A more complete economic model would be able to be run forward into the future without any such assumptions, because all the inputs it needs to make those predictions would be drawn from its own outputs. Things that look like independent variables to a partial model, like this one, can end up being dependent variables in a more complete model. I aim to add some components in that direction in part 2 of this series (which is still being written) but I'm not going to develop a full model of the economy.
The shelter costs in my model are derived from inputs like the price of houses, the interest rate, and costs in other categories. That should be _fairly_ accurate independent of wages. But, if wages fell or failed to grow, that would strongly impact all three of those inputs to the shelter model. For example, housing prices might crash deeply as people are forced to sell their houses to pay the bills that their wages don't; interest rates might reverse as the central bank notices the low wages indicate a weakening labour market; inflation in other categories (affecting things like lumber prices and the cost of maintenance) might decrease as well.
On the other hand, rising wages could help support the housing market, create more impetus for faster rate hikes, and so on. All of those inputs, affected by wages, were taken-as-given in the model as it is so far.
I expect banks are about to get a lot more risk-averse in lending. As interest rates go up, so will the stress-test cutoff, and I'm already hearing that some banks are reducing the duration they'll hold a rate for a borrower who qualifies for a mortgage.
I think your missing variable is population growth. The 1950 through 1970 are the peak years of a) the baby boom, and b) post-war immigration. Rental pricing is decided at the margin, and people needed ever-more space through those decades. And the stock of rental changes slowly, with a long delay. https://en.wikipedia.org/wiki/Population_of_Canada#/media/File:Canada_immigration_graph.png
Thanks for your comment!
* Wouldn't that predict more rapid rental price gains over those decades? My surprise was how the CPI indicated that the increase in the nominal price of rent was so little, rather than so much.
* The way that my model accommodates confounders like population growth (and it's for sure not perfect) is that it still takes housing prices as an input. An increase in population would result in pressure on all forms of shelter, including both rented and owned housing. My model estimates rent based on the opportunity cost of home ownership, which will grow along with housing costs in general when population pressure drives them up. If a baby boom drove up rental costs only, without affecting the cost of homeownership, it would either drive a lot more people to switch from renting to buying, or incentivize a lot of property investors to buy up homes and rent them out, or both.
Your chart of real rents shows a sustained and basically flat plateau 50-70, and then a decline which matches the period of the boom. I speculate that the 40s would show the matching rise for the beginning of the boom. But your comments on the mix are pointed. My (unjustified) speculation is a mix of: rent control legislation (1945-1955, then 1974 on on Ontario), the differential lag between suburban housing vs purpose built rental, fashion and explicit policy preferences for suburban growth as part of the highway system rollout (sfh over rental) and more. But I’m over my head there.
I think the variable is not directly population growth, but rather "housing stock" as the Reserve Bank of Australia call it, I believe. This is the float in the market of how many properties (shelters) are available to purchase.
This effectively decouples population, as it is simply a measure of supply and demand with the basis that more housing stock there is the lower the demand is to purchase.
A more subtle secondary variable (that could be considered to provide a leading indicator, so a first derivative you might say) is days to sell, or percentage clearances at auction, which is just the percentage of auctions resulting in a sale.
Low stock and high clearance rates (or short days to sell) create FOMO which becomes like a bubble and leads into non linear territory and the desire to buy, and buy now, becomes almost viral. You hear people saying, "I need to buy now, no matter what, because the prices are rising so much faster than I can save for a deposit" . It is in these conditions people can "flip" houses in weeks for profit, but also when people over commmit, because FOMO.
Finally, I am in Australia and my sister in law works for the biggest conveyancer (is licensed to settle the sale transaction and change the title of real estate) in our city and she says the single biggest indicator of whether people are prepared to take out a loan is not the size of the loan, but how confident they are they will have work for the next 3-5 years, sometimes less in a fast rising market. This is because they feel that worst case they can always sell up and not make a loss, as along as they can stay in for 1-3 years and get a good capital increase, despite loss of capital to fees and taxes.
I like your work, I am a control systems engineer (though in practice not usually as theoretical as here) and always wondered why economists did not lean more on control systems theory, especially regards stability and controlability. But also, as I have pointed out above, there are some non-linearaties that can easily creep in once the market conditions are not so steady and humans are involved, then all bets may be off.
Completely agree, but isn’t that the tricky bit? That actual controls available are very different between the players: the central bank can use interest rates to modulate house construction rates (2nd order from the stock, with a 5-10 year lag), cities can vary zoning, fees, and rent control (same). Higher governments might build hosing directly (almost direct control), or change immigration levels ( direct control of next years demand side). All of them are themselves subject to democratic control, but with different sub-grouping and indirection (regular direct elections of municipal politicians, indirect appointments of ban officials, etc). The political economy of it is incoherent, and each player works with the available variables.
"So if you were the central bank, how would you control a system like this?" Why the central bank?
They are not the ones in charge. They are not even the source of money.
https://www.ucl.ac.uk/bartlett/public-purpose/publications/2022/may/self-financing-state-institutional-analysis
Turns out interest rate hikes run into a fundamental problem with the accounting.
https://new-wayland.com/blog/interest-price-spiral/
> They are not the ones in charge. They are not even the source of money.
That will come as news to them! Central banks around the world take stabilizing inflation as part of their mandate, or central to it. Certainly the Bank of Canada does. They seek to do this through tools like setting upper and lower boundaries on the interbank interest rate and, more recently, purchasing bonds.
> Turns out interest rate hikes run into a fundamental problem with the accounting.
Well, the article you are commenting on is detailing precisely one potential interest/price spiral that the Bank of Canada should be concerned about. Because shelter, like energy, is a fundamental input to almost all goods, I wouldn't be surprised if this results in an increase in costs of other goods as well.
That doesn't mean it's the wrong move. As I've explained, I believe this is a right-half-plane zero, and that means inflation might go up before it goes down.
"They seek to do this through tools like setting upper and lower boundaries on the interbank interest rate and, more recently, purchasing bonds."
Yes. They don't work - for the reasons explained in the paper, and the feedback loops in the article linked. The steering column isn't linked to the wheels. The system assumptions on which they are based are wrong (exogenous money and general equilibrium) and naive (the response of individuals and businesses to rate rises and price changes is not necessarily to borrow less or repay loans).
In academic terms 'they have uncertain distribution characteristics'. The amount of input on that particular zero is likely going to have to be Volcker levels to have any material effect and which will bring on recession and unemployment.
"Because shelter, like energy, is a fundamental input to almost all goods"
Shelter is derived from energy and labour. The roots of the current inflation lie in energy acquisition, and are amplified via shelter as you rightly point out. The current problems is likely due to a shift in the terms of trade relative to energy. Messing around with interest rates is unlikely to change that - particularly when the rest of the world is doing the same.
More likely, as you've discovered, it will push inflation higher than if they hadn't bothered and just let the taxation auto stabilisers drain the money from the system.
> Yes. They don't work... the steering column isn't linked to the wheels.
That's entirely plausible, and I don't take a strong view on it one way or the other. I'm more interested in looking at how the economic system works, from within the framework of how the central bank *assumes* it works.
That's because, ultimately, they're the ones at the wheel (even if it's not connected). When evaluating how best to steer, they'll be using the models they trust. Even if that wheel is disconnected from the steering column, I find myself very interested in anticipating how they're going to try to turn it. After all, my focus is on control theory first, and economics second.
If the economy doesn't react to their feedback the way they expect, that's important too. But my uneducated guess is, that steering wheel is probably connected to the steering column more tightly in Canada than it is in the US.
> is likely going to have to be Volcker levels to have any material effect and which will bring on recession and unemployment
Sometimes, that can be the unfortunate consequence of a borderline-uncontrollable system. It's great when a system can always be smoothly steered to reach any desired operating point by continuous gentle application of small corrective inputs, but in real systems, that luxury doesn't always exist. Especially if you find yourself suddenly so far away from the target, as we have now.
If you're driving fast and you notice the traffic light turning yellow up ahead at a busy intersection, easing gently off the gas pedal isn't going to be enough to avoid a collision. You have to slam on the brakes, hard, or run the light. If you're also being tailgated by an eighteen-wheeler with a sleepy driver hauling a hundred tons of cargo, then you might not have _any_ trajectory of inputs available to you that avoids an accident. And if it turns out the brake pedal is also mis-wired and accelerates the car, then... well, I hope at least the seatbelts are buckled.
To me, it's not really worth arguing the details of whether it's better to run the light or slam on the brakes. Either way we'll be relying way more on luck than any responsible driver should ever need to. Instead what I hope we can learn is _why_ we somehow trapped ourselves in that situation in the first place, and why it's so important that we avoid letting that happen in the future. Even if it means driving a bit slower, etc.
> More likely, as you've discovered, it will push inflation higher than if they hadn't bothered and just let the taxation auto stabilisers drain the money from the system.
I'm glad you mentioned taxation. That's of course another huge input; within certain limits, the government can pretty freely increase or decrease the money in circulation and in a pretty fine-grained way, more directly than the central bank. I have no doubt that the government's steering wheel is more tightly connected than the bank's.
I think there's a lot the government can do (and I have my own opinions about what they should do), but from a feedback systems perspective, the government is a little less interesting because they don't have an inflation mandate. Inflation is certainly not popular, and the government would rather it be low than high, but their reaction to high inflation won't necessarily be taking action to suppress it. (An understatement). Suppressing inflation usually requires a pretty unpopular policy, which is why they've outsourced that job to a (hopefully) independent party like the central bank.
So I can't realistically think of the government as a feedback controller. They're more of a wild card. I think they have an important role to play, for sure. But it will be easier to predict how the central bank responds to government policy, than how the government responds to central bank policy.