The International Simutrans Forum

 

Author Topic: Interest Rates and Large Game Observations  (Read 12611 times)

0 Members and 1 Guest are viewing this topic.

Offline Sarlock

  • Devotee
  • *
  • Posts: 1340
  • Languages: EN
Interest Rates and Large Game Observations
« on: October 08, 2013, 07:03:20 PM »
I have been playing the 7000x2000 map I provided here and have advanced from 1750 to 1885 thusfar.  I have a bunch of modifications I want to make to the map (as well as adding in some "roughness") and playing it has been the best way to highlight areas to modify.  Passenger networks have been running smoothly and very profitably, almost exclusively using sea transport (first sail, then steam) up until 1850.

While playing, the impact of interest has overshadowed all over revenue streams.

In the early years, interest has little impact on gameplay, except for those times that I drew from my credit line to fuel expansions through capital purchases.  The impact of interest on a negative cash balance is very appropriate and something to heavily consider while planning an aggressive debt-based expansion.  Certainly no changes required here.

The impact of interest on later years, once a significant profit is generated, is significant and explosively exponential.  Over a short period of time, 25 or so years, this impact is less important, but over a period of 135 years thusfar interest quickly passes all over revenue sources and makes everything else insignificant.  I should note that this same behaviour has disproportionally impacted revenues in the online server game as well.



As demonstrated, my passenger lines are very profitable at this stage in 1885.  9.4 million in revenue, 5.4 million in profit (a bottom line performance any business owner would love to have!).  The impact of interest, however, is massive.  It is 4x operating profit and even with an aggressive capital purchasing program at 7.4 million, cash flow is still a positive 18 million.  Without the impact of interest, this would have been a negative cash flow year... signifying the need to generate future profits to pay off the capital purchase.

You can see on the graph the exponential nature of the interest.  The vast majority of the interest earned at this point is made off of interest earned in years past.  It is an exponential system that is running out of control and over the next 50 years of gameplay will completely outstrip the significance of every other number.  Even at this early game stage in 1885 I could completely disregard the profitability of my network and run at a -15 million or more loss and still generate a positive cash flow.

It is difficult to calculate exactly the total amount of interest I have earned over the first 135 years of this game, but I've tried to extrapolate this figure as best I can, based on profit levels over the life of the game.  I currently have 1.4 billion in cash and approximately 400 million of this was from profits and 1.0 billion is from interest.  Over the next 50 years, at 6 million/year profit, I will generate 300 million more in cash (less any capital purchases).  The interest over the next 50 years, however, will be 2.5 billion, bringing me to over 4 billion in cash.  One doesn't need to look further than the online game to know what happens to interest revenues once we get even further in years.  Placed in the hands of an aggressive player, interest revenue could be used as a way to maliciously cannibalize another player's network without any regard to profitable operations.  (or crash the server, ha ha)

My suggestion would be to remove interest as a positive balance revenue source, or at least allow this to be separately set from the debt interest rate (and default to 0).  I am setting my interest rate to 0 for now to correct this.

From a modeling of reality standpoint, interest is certainly a component, though usually as a return on capital investment.  Corporations that have excess cash generally dividend off their surplus funds to shareholders to allow them to invest it elsewhere.  Unless you're Apple.  But even they finally relented and recently issued a dividend.

The notion of inflation has been mentioned before, but I suspect that this may just overcomplicate the economic model.  Setting an inflation rate equal to the interest rate would just counteract the effect of each and return to a neutral 0+/- situation.  Having inflation impact things such as standard maintenance costs may be helpful in early games, however, as standing capital costs are very expensive on the low revenue streams of early transport in pak128.Britain.  But this could probably be addressed by lowering/raising the operating cost of these early stations in graduated steps, or increasing the revenue potential of some early transport (some sea vessels are very profitable, at least).  Inflation is partially baked in to the system through the increasing revenue streams of later, faster transport... but this could be counteracted with balancing measures.

I'll stop here.  I used to be an accountant, I could go on for days...  ;)


Other observations

Industries:

The distribution of factories is fairly nice, though cross connections do not seem to be following the "max_factory_spacing_percentage = 33" limitation in simuconf (using default pak128.Britain-Ex settings).



The bottom cattle farm and dairy are almost 6000 east-west tiles apart across the map and a further 1000 north-south tiles apart.  The closest cattle farms are half a map away.  I have many examples of such a distant cross connection on this map.  It seems like the limitation is not used.

I focused on serving industries that have closer connections instead.


City Growth:

This has been mentioned before and will be addressed through numerous upcoming changes to the city growth system, but city growth sprawls far too much.  I have one megalopolis of 290,000 citizens that sprawls over an area 100 tiles wide by 300 tiles long (approx 500km2) and it's only 1885.

I would love to see the population figure scaled up based on map scale size.  A city 500km2 in size would likely have several million inhabitants (but only generate the passenger flows at scale).  I believe the intention is to include a "population" setting in the building .dat files which would be excellent.  In order to properly compute population, you need a setting in simuconf for the pakset to indicate what its population scale is (ie: one building is 25m2) so that the game knows to multiply that building tile's population by 25 to derive the population at a 125m2/tile scale.

I tried modifying the cityrules.tab settings for "renovation_percentage" and "minimum_building_density" but they didn't seem to have the desired effect on constraining outward sprawl and concentrating urban density as I expected.  This isn't worth investigating due to the major changes upcoming, but worth noting.

I'd love to contribute to developing the logic of the new city growth system.

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18422
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Interest Rates and Large Game Observations
« Reply #1 on: October 08, 2013, 07:12:54 PM »
Thank you very much for your input! It is much appreciated. Inflation is a necessary thing to add, I think, as without it many real life dynamics cannot adequately be modelled over a long time period (I intend to have lots of different inflation rates for different things to model rising labour costs, etc.). A separate question is whether to allow a separate setting of the credit interest rate (and setting it to zero) until we introduce inflation: that may need to be considered for the next major version.

As to industry spacing, this might be a bug - if you can confirm that this does not work as apparently intended, can you post a separate bug report so that I can look into this?

On city growth, any contribution to disucssion about how to implement it would be most welcome, albeit at present I am currently engaged in trying to make the final parts of the passenger generation system work satisfactorily.

Offline MCollett

  • *
  • Posts: 214
  • Languages: en
Re: Interest Rates and Large Game Observations
« Reply #2 on: October 08, 2013, 07:56:13 PM »
While playing, the impact of interest has overshadowed all over revenue streams.
[...]
Corporations that have excess cash generally dividend off their surplus funds to shareholders to allow them to invest it elsewhere. 
I don't think the interest rate is in principle the problem (although the positive rate should certainly be different from and less than the negative rate), as much as the the lack of dividends.  The cash pile you have built up is huge (especially for the date!), so naturally it earns lots of interest; the issue is the existence of the pile in the first place. 

In Simutrans the player is very much in an operational role: he is a manager or chief executive, concerned with making the company's day-to-day business work efficiently and profitably; unlike in, say, Railroad Tycoon, he is not (also) the company president dealing with the sharemarket, dividends and investors, or with his own personal fortune.  In fact, the Simutrans player could quite well be a state employee managing a government service.  So this leaves open the question of what the dividend rate should be.

Having said all that, your suggestion of setting positive interest to zero is an effective (partial) fix, justifiable on the grounds that any interest earned is being issued as a dividend.  A realistic dividend rate would be higher still.

Quote
I have one megalopolis of 290,000 citizens that sprawls over an area 100 tiles wide by 300 tiles long (approx 500km2) and it's only 1885.
For comparison, in 1891 Inner London had a population of over 4 million in an area of about 300 km2, while Manchester and Liverpool both had populations of about half a million in an area of about 100 km2.  So yes, sprawl is the right word for what you have.

Best wishes,
Matthew

Offline Sarlock

  • Devotee
  • *
  • Posts: 1340
  • Languages: EN
Re: Interest Rates and Large Game Observations
« Reply #3 on: October 08, 2013, 09:07:31 PM »
It's an interesting topic.  Having a large cash reserve is certainly useful for future expansion pushes: laying long sections of track, new vehicles, etc.  Having too large of a cash reserve ends up making profitable operations somewhat unimportant.

I see it the same as you, MCollett: we are the operational department or either a private or public transportation company: the cash reserves are there for us to draw from as required for future growth.  Maintaining a profitable operation is one of our primary mandates.  Just because our company has been profitable in the past does not excuse us from continuing to be a profitable company.  We'd be fired if we started to run massive losses just because our interest income is enough to offset!

Makes me wonder if there is a mechanism we can install that starts to deplete excess cash reserves that the player isn't using (whether it be corporate dividends or the government reaching in and skimming excess cash).  This forces the player to continue to focus (long term) on profitable operations no matter how good performance in the past has been.  It would have to be high enough so as to not impact the player on an aggressive growth campaign or fund several years of bad losses.  A real world company would either issue these reserves as dividends and/or reinvest them in other operational branches or acquisitions.

Offline ӔO

  • Devotees (Inactive)
  • *
  • Posts: 2345
  • Hopefully helpful
  • Languages: en, jp
Re: Interest Rates and Large Game Observations
« Reply #4 on: October 08, 2013, 09:47:45 PM »
If the player can just give cash away to the public player (and only the public player), that would work too.

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18422
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Interest Rates and Large Game Observations
« Reply #5 on: October 08, 2013, 10:08:26 PM »
Ahh, the "dividend rate" as some have described it here aims at the effect that I intend to achieve with the cost of capital feature that I plan here (players start with nothing and have to borrow money by way of bond/long term loan rather than overdraft to invest in infrastructure). The equivalent of a dividend (since simulating an entire stock market is rather beyond the reach of Simutrans) is to represent the cost of long-term capital financing by the means that I there describe.

As to keeping the cash pile to a manageable level, inflation should do that for us. There is also the possibility of forcing players to build and upgrade headquarters in some specific circumstances and giving them a very high maintenance cost, but the difficult question is what those circumstances should be. One possibility might be to link levels of headquarters to allowing the player access to certain sorts of statistics and data that are currently available to everyone for free, which are not so important with a smaller network, but become increasingly important for a large network. However, features such as that are a long way off.

Might there be something to be said for disabling credit interest entirely until inflation and/or cost of capital is implemented?

Offline Jando

  • *
  • Posts: 454
Re: Interest Rates and Large Game Observations
« Reply #6 on: October 08, 2013, 11:10:36 PM »
I reckon it's hard (as in impossibly hard) to balance.

The Simutrans revenue model is based on distance, thus the larger the map the more revenue potential you have. Once you've unlocked that revenue potential you're swimming in money. And - like in the real world - when you have plenty of spare money then interest rates will make sure you get even more money. :) That's what you're seeing in your finance screen.

I don't think anything can be done. Ideas like inflation or disabling interest at all will hurt people that play smaller maps (smaller maps = less distance = less revenue). Especially for transport where the player doesn't need to pay for the track (like in public roads or see transport) distance is what determines revenue, and as such there seems to be a direct relation between size of map and amount of money the player ends up with.

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18422
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Interest Rates and Large Game Observations
« Reply #7 on: October 08, 2013, 11:13:59 PM »
Why should it be impossibly hard to balance? Transport prices are largely based on distance in the real world, too, but there are countervailing dynamics, many of which I plan to model, as set out in this thread.

Offline MCollett

  • *
  • Posts: 214
  • Languages: en
Re: Interest Rates and Large Game Observations
« Reply #8 on: October 08, 2013, 11:35:15 PM »
Ahh, the "dividend rate" as some have described it here aims at the effect that I intend to achieve with the cost of capital feature that I plan here (players start with nothing and have to borrow money by way of bond/long term loan rather than overdraft to invest in infrastructure).
This would only work in the short term; once the player has made enough profit to pay off the initial bonds (and can fund further expansion from cash on hand) the problem is back.  In the real world, dividends are expected to increase with the value of the company, not be fixed at a level determined by the starting capital. 
Quote
Might there be something to be said for disabling credit interest entirely until inflation and/or cost of capital is implemented?
Yes, at least until then.

Best wishes,
Matthew

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18422
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Interest Rates and Large Game Observations
« Reply #9 on: October 08, 2013, 11:45:44 PM »
Hmm - it is difficult to know what to do about this dividend issue without either (a) simulating the stock market in full (which is not feasible) or; (b) having a system so abstracted that it does not really model reality in a useful way and becomes arbitrary (preventing players from saving to invest later, for example). Also, the joint stock company with paid up share capital is not the only sort of organisation capable of running a transport network, and yet dividends apply only to such companies.

Offline Sarlock

  • Devotee
  • *
  • Posts: 1340
  • Languages: EN
Re: Interest Rates and Large Game Observations
« Reply #10 on: October 09, 2013, 01:48:55 AM »
Great discussion thusfar.  Thank you for the link, James.  I have been trying to slowly catch up on all of the old stuff in these forums from before I joined as an active participant and hadn't gone through that post in detail yet.  I shall take the time to digest its contents fully.

If the player can just give cash away to the public player (and only the public player), that would work too.

Taxation.  Basically an inverse interest model but with a graduated payout based on increasing levels of income.  It's a model that is simple and has merit.

We all love taxes.

Quote
without either (a) simulating the stock market in full (which is not feasible) or; (b) having a system so abstracted that it does not really model reality in a useful way and becomes arbitrary (preventing players from saving to invest later, for example).

We certainly do not want to design a complex economic model to take care of this small issue (we want to devote our precious CPU resources to more important computations).  Right now interest is a small part of a planned system and because it doesn't have the other counterbalances in place, it behaves in the manner that it does.  Disabling certainly seems like the logic step at this point until it has a role to play in a larger system.

Dividends, taxes, payment to crown corporation (government), whatever we want to call it, excess cash is stripped off if the transportation does not require it for regular operations.  These payments are all, in real examples, graduated according to how much excess cash there is -- the more cash, the higher the payout to the corporate masters.  Basically, a parachute on runaway cash reserve growth (spend it or lose it).

Implementing a basic debt model would be nice.  It certainly shouldn't be complicated: a credit limit, similar to what exists now, is calculated and the player can borrow up to that amount at a fixed rate for a fixed period of time.  I would make it as simple as a +/- button to borrow/repay bonds using a set multiple that increases in rounded increments as the player's borrowing capacity changes.  Rather than having each bond a unique entity with its own interest rate and payment term, just have a blended interest rate that goes up or down according to the percentage of bonds the player has against its total borrowing capacity.  This interest rate represents an average rate that the company pays for all of its outstanding debt.  Then, each month, the company pays this interest rate (or, rather, 1/12th of it) plus a certain percentage of repayment of principle, say, 2% (which approximates a 5 year repayment).


In its simplest form, an interest rate is a reward for accumulating a large cash balance.  Until we reach this its effects are fairly minimal and serve more as a 2.5% surcharge on our transportation revenue.

Going to digest that large post of James and put the kids to bed.  More to come later.

Offline sdog

  • Devotee
  • *
  • Posts: 2039
Re: Interest Rates and Large Game Observations
« Reply #11 on: October 09, 2013, 03:36:25 AM »
Hmm - it is difficult to know what to do about this dividend issue without either (a) simulating the stock market in full (which is not feasible) or; (b) having a system so abstracted that it does not really model reality in a useful way and becomes arbitrary (preventing players from saving to invest later, for example). Also, the joint stock company with paid up share capital is not the only sort of organisation capable of running a transport network, and yet dividends apply only to such companies.

Base a dividend on the ratio of net worth and net income. The whole point is to prevent the player from saving up money for later investments. That's where the player has to get money from the capital market again.

This is also not too far away from reality, where the previous investors want to get their share when the company is highly profitable. For further investments, new investors are found (which could be the old ones re-investing their dividend of course. they get the first draft on newly issued stocks).

As soon as the company is in the negatives, dividends cease completely. This has the nice side effect, that it gives negative feedback, helping a trend towards zero.

A few years back there was a sweet spot in the pak-set balance, where it was quite possible to stay profitable, but very difficult to expand. This stagnation caused bancrupcy in the long term, when vehicles became too outdated and declining speed bonus let revenues drop. A perfect situation, as it described very well what happens with stagnating companies in a developing market.

Roughly simulated dividends as a capital cost might make such situations much more likely.

Offline Banksie_82

  • *
  • Posts: 47
Re: Interest Rates and Large Game Observations
« Reply #12 on: October 09, 2013, 05:00:19 AM »
Here’s my thoughts…

How about a combination of tax and dividends, each with subtle differences in their calculation but both being a progressive system.

I believe that most (at least personal) tax systems around the world use a progressive system, but I’m not so sure about for companies. Obviously dividends are usually at the discretion of the company boss, but hear me out anyway.

Income should be taxed based on the amount earned (operating profit?)

Say
Everything less than X per month/year is taxed at 0%
From X to Y at say 15%
From Y to Z at say 30%
Everything above Z at say 45%

Where X, Y and Z, as well as the actual rates could vary with time defined by the pak set. Maybe a little trial and error to get the balance right.

Similarly, dividends (although as far as the player is concerned, just another tax) could be paid in a similar fashion but based on bank balance at the end of the month/year rather than income.

I believe, if balanced correctly, that this system will make it easier to start a small transport company and at the same time reduce the likelihood of runaway profits and bank balances.

Also, you will be able to keep the interest rate as it is, but making it more important for small companies than a bit of a joke for large ones.

Offline MCollett

  • *
  • Posts: 214
  • Languages: en
Re: Interest Rates and Large Game Observations
« Reply #13 on: October 09, 2013, 05:05:51 AM »
How about a combination of tax and dividends, each with subtle differences in their calculation but both being a progressive system.

I believe that most (at least personal) tax systems around the world use a progressive system, but I’m not so sure about for companies. Obviously dividends are usually at the discretion of the company boss, but hear me out anyway.

Company tax is never progressive.  But the general idea of having a tax on profit plus a dividend (or equivalent) payout based on cash balance is reasonable.

Best wishes,
Matthew

Offline Banksie_82

  • *
  • Posts: 47
Re: Interest Rates and Large Game Observations
« Reply #14 on: October 09, 2013, 05:46:00 AM »
 Company tax is never progressive? That’s interesting, I know it’s not in my country but I wasn’t sure if that was the case everywhere.

Is there a particular reason that often you will find progressive tax for people but not companies?

The reason I ask is, my understanding of the reasons for a progressive system for individuals is mainly for social reasons (political ideologies aside). That is, the more wealthy subsidise the poor, creating more even distribution of wealth in society. Thereby providing more social security (in the literal sense of the phrase) and more cohesive society for the benefit of all.

There aren’t the same considerations when dealing with a corporate entity… in real life.

But Simutrans isn’t real life. In a multiplayer game, we want lots of participants, all having a good time and running profitable networks. We don’t want the case in the current online game where a couple of companies are so wealthy they actively try to waste funds while start-ups struggle to turn a profit.

A progressive tax/dividend system will enable the pak developers far more scope to make the game easier to start a company but harder (compared to now) to run and expand an already quite large company.

Offline Sarlock

  • Devotee
  • *
  • Posts: 1340
  • Languages: EN
Re: Interest Rates and Large Game Observations
« Reply #15 on: October 09, 2013, 06:23:51 AM »
Canada has a two level corporate tax structure.  A lower rate for profits up to $500,000/year for Canadian-based corporations and then a general corporate rate that applies after that.  This is done to give a slight advantage to small Canadian-run companies.

The reasons why corporate taxes are not tiered like personal taxes are multi-fold, but the primary one is that you do not want to unfairly penalize a large corporation over a small one that produces the same good(s).  Just because corporation A is larger than corporation B doesn't mean that B should get a competitive advantage because of that.  With individuals, that same competition doesn't exist (unless you're counting who can afford more Porsches).

I think a blend of a taxation and dividends (or whatever you want to call it in a public company when the government skims the excess funds to general revenue-basically an extra tax) would be an efficient way to place a drag on large cash balances while, as you say, allowing a smaller company the freedom to expand without encumbrances.  The rate can increase until such a point is reached where no further cash accumulation can occur beyond a certain level.  I probably wouldn't even tax profits and just cash or perhaps net assets (including cash)... a capital tax.  (so that the player can't buy a ton of vehicles just to hide their extra cash from the tax man)  Set that at a rate higher than the interest rate and you will naturally create a cash limitation.

I certainly do like the idea of even a large company having to dip in to the bond market to fund a large expansion.  Presumably solid profits will quickly pay the bonds off, but this presents a continual risk to the player to always design and maintain a good profitable network.  Large companies can earn large profits... but can also go bankrupt very quickly when things change (look no further than Blackberry).

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18422
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Interest Rates and Large Game Observations
« Reply #16 on: October 09, 2013, 09:21:14 AM »
This is all interesting discussion - thank you! Regarding tax, this is already planned, albeit as part of a wider system in which the public player does not have unlimited finances, and must raise revenue in order to spend it by taxation and borrowing.

As to the dividend model proposed, the ideas are very interesting, but I do worry that explaining to players where their money is evaporating to might be rather challenging; does anyone have any suggestions as to how to make this readily understandable to players with only modest modifications to the GUI?

Offline Sarlock

  • Devotee
  • *
  • Posts: 1340
  • Languages: EN
Re: Interest Rates and Large Game Observations
« Reply #17 on: October 09, 2013, 02:30:34 PM »
Indeed, I've been thinking the same thing: the concept might be somewhat confusing (and frustrating) upon first experience with it, unless it is implemented correctly and intuitively labelled/presented.

Certainly something to think on, as this is an addition that is likely quite some time away.  In the meantime being able to freeze rates (while retaining the debt rate) is probably a good interim measure.

If I had programming knowledge with C++ I'd certainly be happy to help.  I might start to learn (I only have a basic understanding of it) but I fear I'd still be a year or two away from being much help as the learning curve is steep.  Perhaps I may start that journey anyhow.

Offline MCollett

  • *
  • Posts: 214
  • Languages: en
Re: Interest Rates and Large Game Observations
« Reply #18 on: October 09, 2013, 08:37:39 PM »
As to the dividend model proposed, the ideas are very interesting, but I do worry that explaining to players where their money is evaporating to might be rather challenging; does anyone have any suggestions as to how to make this readily understandable to players with only modest modifications to the GUI?

I would have thought a red line in the Finance screen that said 'Company tax' and another one that said 'Dividends' would have been self-explanatory.  Something like 25%-30% tax on profits and dividends of 6% of net worth per annum would be about right.  The dividend rate should certainly be higher than any positive cash interest rate.  If you wanted to put a positive spin on it, the tax could go the the public player, and the dividend payments could help drive economic growth (towns and/or industry).

Best wishes,
Matthew

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18422
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Interest Rates and Large Game Observations
« Reply #19 on: October 09, 2013, 08:50:04 PM »
The tax going to the public player is already planned, although not high on the list of priorities. The real trouble that I see with dividends is that there is no satisfactory way to simulate the great flexibility that a company in reality has in deciding when to pay a dividend and how much to pay, whilst also simulating the practical limitations on that flexibility, the latter of which are especially complex. Without simulating both, the actual in-game dynamic of "dividends" becomes so different from the real life economic dynamic intended to be simulated that it causes more problems than it solves, so far as I can make out.

Offline ӔO

  • Devotees (Inactive)
  • *
  • Posts: 2345
  • Hopefully helpful
  • Languages: en, jp
Re: Interest Rates and Large Game Observations
« Reply #20 on: October 09, 2013, 08:52:53 PM »
shouldn't taxes only be on incomes and not net worth?

Although, if you call it a property tax for asset worth, then I can understand.

Offline jameskuyper

  • *
  • Posts: 78
  • Languages: en
Re: Interest Rates and Large Game Observations
« Reply #21 on: October 09, 2013, 09:08:55 PM »
Company tax is never progressive. ...

See <http://www.irs.gov/pub/irs-pdf/i1120.pdf> (much of the rest of the government is shut down, but parts of the IRS web site are still working). Look at the "Tax Rate Schedule" given on page 17.

Offline MCollett

  • *
  • Posts: 214
  • Languages: en
Re: Interest Rates and Large Game Observations
« Reply #22 on: October 09, 2013, 09:19:37 PM »
The real trouble that I see with dividends is that there is no satisfactory way to simulate the great flexibility that a company in reality has in deciding when to pay a dividend and how much to pay, whilst also simulating the practical limitations on that flexibility, the latter of which are especially complex.

The company owners do have flexibility in determining dividends, and modelling the real limitations on that decision would indeed be complex.  But as I said further up the thread, the player in Simutrans is not really in the role of owner or chairman (or of the relevant politician, if the company is publicly-owned), but rather in that of a manager or chief executive.  The dividend rate is a policy decision that is taken over the player's head, and which he just has to live with.  It would perhaps be reasonable though to give new start-ups a partial dividend holiday: increment the dividend from 0% to the target rate over the first few years of the company's life, and leave it fixed thereafter.

Best wishes,
Matthew

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18422
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Interest Rates and Large Game Observations
« Reply #23 on: October 09, 2013, 09:33:02 PM »
The company owners do have flexibility in determining dividends, and modelling the real limitations on that decision would indeed be complex.  But as I said further up the thread, the player in Simutrans is not really in the role of owner or chairman (or of the relevant politician, if the company is publicly-owned), but rather in that of a manager or chief executive.  The dividend rate is a policy decision that is taken over the player's head, and which he just has to live with.  It would perhaps be reasonable though to give new start-ups a partial dividend holiday: increment the dividend from 0% to the target rate over the first few years of the company's life, and leave it fixed thereafter.

The trouble is that the decisions of higher up are not taken in a vacuum: deciding where and whether to invest in a large infrastructure project would be a joint decision between the senior board officials and the executives represented by the human player, for example. The decisions as to the dividend would in real life reflect the actual performance and plans of the company, and this would have very important economic and fiscal consequences.

A dividend holiday for startups, meanwhile, is likely to be confusing to players: how exactly is this to be explained in the GUI, and precisely what would count as a "startup"?

Offline MCollett

  • *
  • Posts: 214
  • Languages: en
Re: Interest Rates and Large Game Observations
« Reply #24 on: October 09, 2013, 09:36:49 PM »
Look at the "Tax Rate Schedule" given on page 17.

OK, 'Hardly ever': there's a reduced rate for companies whose profits are around or below the level of one person's income.  All the other rates are just minor variations on the theme of 34%-35%.  The two (anti-progressive) bands at a higher rate are designed exactly to compensate for the initial reduced rate bands, so that the net result for a large company is equivalent to a flat tax of 35%. 

Best wishes,
Matthew

Offline Sarlock

  • Devotee
  • *
  • Posts: 1340
  • Languages: EN
Re: Interest Rates and Large Game Observations
« Reply #25 on: October 09, 2013, 09:38:18 PM »
We would not want to issue dividends during years of financial loss (though many large companies do maintain their dividends during periods of what they perceive as temporary losses).  We certainly wouldn't want the system to hurt an unprofitable company that is trying to turn itself around before bankruptcy.

Explaining the logic of this system is certainly important or the player will be confused by the variable nature of the impact to their bottom line.  The system would have to be very simple and carefully itemized in the financial reports.

And since it's a long way from potential implementation, we have lots of time to think about it :)

Offline MCollett

  • *
  • Posts: 214
  • Languages: en
Re: Interest Rates and Large Game Observations
« Reply #26 on: October 09, 2013, 09:59:27 PM »
The trouble is that the decisions of higher up are not taken in a vacuum: deciding where and whether to invest in a large infrastructure project would be a joint decision between the senior board officials and the executives represented by the human player, for example.
To be sure.  But modelling that process would require, at the least, some sort of stockmarket simulation, and you have made it clear that you don't want to go in that direction.  Failing that, an averaged or constant dividend is much closer to reality than none at all.  (And there have been plenty of real companies that have operated not very differently 8) .)

Quote
A dividend holiday for startups, meanwhile, is likely to be confusing to players: how exactly is this to be explained in the GUI, and precisely what would count as a "startup"?
I don't see the problem.  A year or two after starting a single-player game, or joining a multi-player one with a newly-created company, the player gets a ticker-tape message saying something like  "The board has decided to increase the dividend rate from 0% to 2%";  similar messages occur a couple more times, until the target rate is reached.

Another very simple way in which the dividend rate could respond to the current state of the company is to cap dividend payments in any year to (after-tax) profits for that year (or possibly for the previous one).

Best wishes,
Matthew

Offline sdog

  • Devotee
  • *
  • Posts: 2039
Re: Interest Rates and Large Game Observations
« Reply #27 on: October 10, 2013, 02:03:25 AM »
There are quite a lot of companies out there where dividends are very constantly paid, without much changes over the decades.

When modelling a complex system, you can not always accurately model every aspect. Some things have to be modelled in a simpler way. Think about a reasonable rule, that averages to roughly the same result as what is often represented in reality. A fixed dividend rate seems to be about right. While it does not address the complexity of the economic simulation behind it, it creates a drain. Ie is the interface to the part of the economic model that is covered.

Just think about how much more complex the processes behind a company tax are. You would have to model chaninging philosophical views over the centuries, and with todays globalisation the interaction of different regions competing with each other. A fixed (perhaps time dependent) tax rate seems to be fine enough though.

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18422
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Interest Rates and Large Game Observations
« Reply #28 on: October 10, 2013, 09:48:27 AM »
I should be very interested in any real life data that can support a constant dividend rate when in profit as being close enough to reality not to cause distortions in the game: it strikes me that I might not know enough about the financial affairs of companies (both now and historically) to know precisely what is correct about this.

Even so, the dividend rate is still in principle part of the cost of capital, and one relevant consideration is that different companies have different rates of issued share capital, some companies buying back their own shares to reduce the amount of dividend that must be paid, others issuing fewer shares to begin with.

Offline MCollett

  • *
  • Posts: 214
  • Languages: en
Re: Interest Rates and Large Game Observations
« Reply #29 on: October 10, 2013, 10:49:53 AM »
one relevant consideration is that different companies have different rates of issued share capital, some companies buying back their own shares to reduce the amount of dividend that must be paid, others issuing fewer shares to begin with.

This is an area where modern and 19th-century ideas diverge somewhat.  The modern way of thinking is in terms of price-to-earnings ratios: the share price reflects the profit, and a 'reasonable' total dividend is the same regardless of how many shares are extant; whether there are a million 10 dollar shares or 10 million 1 dollar shares, the market capitalisation is 10 million dollars.  In the 19th century people cared much more than today about the 'par value' of shares, that is, the initial or nominal price of the shares; dividend rates were seen as high or low not so much in relation to the current share price, as in relation to the par value, so buying back stock would indeed tend to reduce the (total) dividend expectation.  One occasionally encounters references to e.g. the Bridgewater canal making extraordinarily high percentage profits consistently for many years.  This almost certainly means that the profits were a high percentage of the par value, not that the value of the company grew exponentially rapidly over the period; in modern terms, there might have been high percentage profits for a short initial period, but this then translates to increased capital value, and later profits should be considered in relation to this increased value.

To put the same distinction somewhat differently: modern notions of a reasonable dividend are scaled by the market value of the company, which in turn is roughly proportional to (expectations of) profit; 19th-century notions were scaled by the book value.

Best wishes,
Matthew

Offline Sarlock

  • Devotee
  • *
  • Posts: 1340
  • Languages: EN
Re: Interest Rates and Large Game Observations
« Reply #30 on: October 10, 2013, 02:46:51 PM »
Most large corporations pay a pretty consistent dividend rate between about 1% and 4% of market value.  A lot of the "blue chip" large companies have been paying a steady dividend for decades and some over 100 years.

While market value is subject to the whims of the stock market, it does tend to parallel net worth over time.  In fact, you could safely ignore the idea of share capital entirely and base a dividend rate off of the net worth and it would nicely approximate the same effect.

You could then install two simple rules:

-A dividend is only issued with two successive profitable years.
-A dividend is cancelled after two successive loss years.

Since dividends are based off of net worth (Assets + Cash) this acts as a natural brake on runaway cash growth.  If cash levels are reasonable then the dividend payout isn't that large.  If cash levels grow (in the game I cite, cash was 99% of my net worth) then dividend payments grow as well and this cap future cash growth.  This leaves a large company with enough cash to operate and fund capital purchases while still always being conscious of its bottom line profit.

You could pick a payout percentage based on what you figure the limit of cash should be.  Some multiple of assets, as our choice of dividend percentage would directly impact this.

Offline wlindley us

  • Devotee
  • *
  • Posts: 958
    • Hacking for fun and profit since 1977
  • Languages: EN, DE
Re: Interest Rates and Large Game Observations
« Reply #31 on: October 10, 2013, 03:53:34 PM »
If there were a tax rate, a dividend rate, shares of stock, and a resulting goodwill factor, these could work as:
  • game sets tax rate is fixed at a certain percentage (presumably no tax below a certain threshold)
  • user can pick the dividend rate, from zero to any number
  • user can issue or buy back stock shares; the price fluctuates partially randomly,  partially in tune with any economy factors, and also by the user's goodwill factor... this would greatly help funding your initial network
  • user's goodwill factor is determined by the last 10 years' dividends, and by charitable gifts (paying to building landmarks like schools and hospitals, perhaps)
  • purchase price for land could also decrease as user's goodwill factor increases
One potential corporate goal then would be maximum goodwill instead of just making money.

Offline Sarlock

  • Devotee
  • *
  • Posts: 1340
  • Languages: EN
Re: Interest Rates and Large Game Observations
« Reply #32 on: October 10, 2013, 04:51:57 PM »
Installing just a dividend rate would work very successfully.  Can call it "Excess Cash Distribution/Tax" or something.

After running a few models through a spreadsheet, I'd base it off of cash balance only, nothing to do with assets.  5% of cash works nicely to cap cash flow growth.  A more profitable company can reach a higher cash ceiling than one with lower profits, but in all cases there is a maximum potential cash balance.

I used a simple calculation:

Cash Distribution* = (Cash Balance - Assets) * Cash Distribution Rate (/12 if doing monthly)
*Distribution is $0 if the company loses money for 2 continuous years, resumes after 2 profitable years are achieved

The first part ensures that we do not distribute more cash than we have in assets.  This means that we are only distributing cash in excess of our asset total - which means that the player will always have a significant amount of protected cash on hand equal to its capital asset value.  We then distribute out any excess cash at the distribution rate.

Depending on profits, the company can still amass a nice amount of cash.  A company with $10m assets and $10m cash will pay no distribution "tax/dividend".  If it is earning $5m per year (cash flow) then it will start paying 5% of the cash balance over $10m each year.  It will still be able to reach a total cash total of $110m (which is our original $10m plus 1/5%=20x$5m profits = $100m).

Very simple calculation and can be put right on the Interest line on the finance window (rename the label to Interest/Taxes or something).

Now we've put a nice cap on potential cash reserves -- our player will always have to focus on profitable operations over the long term (short term losses can be tolerated if the player has a nice cash reserve built up).


As an aside, most corporations finance capital expansion with debt even if they have significant cash reserves (they may do part cash and part debt).  Shareholders prefer leverage on the balance sheet - if offers a better potential return on investment.

Offline sdog

  • Devotee
  • *
  • Posts: 2039
Re: Interest Rates and Large Game Observations
« Reply #33 on: October 10, 2013, 07:03:50 PM »
Most large corporations pay a pretty consistent dividend rate between about 1% and 4% of market value.  A lot of the "blue chip" large companies have been paying a steady dividend for decades and some over 100 years.

While market value is subject to the whims of the stock market, it does tend to parallel net worth over time.  In fact, you could safely ignore the idea of share capital entirely and base a dividend rate off of the net worth and it would nicely approximate the same effect.

You could then install two simple rules:

-A dividend is only issued with two successive profitable years.
-A dividend is cancelled after two successive loss years.

Since dividends are based off of net worth (Assets + Cash) this acts as a natural brake on runaway cash growth.  If cash levels are reasonable then the dividend payout isn't that large.  If cash levels grow (in the game I cite, cash was 99% of my net worth) then dividend payments grow as well and this cap future cash growth.  This leaves a large company with enough cash to operate and fund capital purchases while still always being conscious of its bottom line profit.

You could pick a payout percentage based on what you figure the limit of cash should be.  Some multiple of assets, as our choice of dividend percentage would directly impact this.
I think it would provide very good game dynamics while being a realistic concept. Afterall, you do not have to reflect everything that's was ever possible on the capital markets. You can pick one way and giv this a kind of deus ex machina: the board of directors is conservative and requires a reliable pattern of dividend payments.

If it works out very well, you could think about refining this. For example a new 19 century company would be venture capital, while a late 20th century company would be something for boring rentiers. Ie delayed very large returns or steady sustainable returns on the other.

Quote
I should be very interested in any real life data that can support a constant dividend rate when in profit as being close enough to reality not to cause distortions in the game: it strikes me that I might not know enough about the financial affairs of companies (both now and historically) to know precisely what is correct about this.
This can be seen indirectly by the fuss markets make when there are slight disruptions. Or on the pride companies in their reliability with this regard.

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18422
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Interest Rates and Large Game Observations
« Reply #34 on: October 10, 2013, 07:27:57 PM »
Thank you very much all for your input on this. Further thought shall have to be given to this at implementation, but this is some very worthwhile insight. I should note, however, that I think that it is important to keep conceptually separate this concept of a dividend with the quite separate concept of tax, which should be based on a percentage of net profit/cash flow, as the two are quite distinct in reality, and the actual tax (but not the dividends) should go to the public player to finance the public player's own operations, whereas the dividend should not.