Tuesday, April 1, 2014

In praise of clear thinking over blind modeling

I have been a long advocate of clear thinking over blind obsessions over modeling techniques (see my last post Being forced to think for yourself...priceless).

Friedman vs. Yellen
Consider the difference between the approaches taken by the late Milton Friedman and the current Fed chair Janet Yellen. Friedman used to quip that the Fed could be replaced by a computer. Monetary policy could be set using a simple formula of targeting money supply growth. His concept was based on the equation:

P = Price
Q = Quantity
M = Money supply
V = Monetary velocity
In other words, you could control the rate of growth in the economy by controlling the money supply, since monetary velocity is constant over the long-run.

Good quants know that all models have limitations. As we found out during the Great Recession, velocity plunged despite the best efforts of central bankers to raise the monetary base. Even though money supply grew at what might have been thought as an alarming rate during normal times, growth was anemic or nonexistent as velocity plunged. The moral of this story: All models have limits, or worse, they can be gamed.

Yellen: Multi-factor modeler
By contrast, consider Janet Yellen's latest speech to the National Intragency Community Reinvestment Conference. Instead of relying on any single rule, e.g. Evans Rule or some other variation of the Rule, she used a far more nuanced approach to analyze the economy. Firstly, she considers the number of part time workers who would like a full time job as one measure of the degree of slack in the economy:
For example, the seven million people who are working part time but would like a full-time job. This number is much larger than we would expect at 6.7 percent unemployment, based on past experience, and the existence of such a large pool of "partly unemployed" workers is a sign that labor conditions are worse than indicated by the unemployment rate. Statistics on job turnover also point to considerable slack in the labor market. 
 Yellen then went on to point to the lack of wage pressure as another indication of economic slack:
A second form of evidence for slack is that the decline in unemployment has not helped raise wages for workers as in past recoveries. Workers in a slack market have little leverage to demand raises. Labor compensation has increased an average of only a little more than 2 percent per year since the recession, which is very low by historical standards.
She also referred to the plight of the long-term unemployed worker and noted that they appeared to be not very different from everyone else. Since the long-term unemployed find it very difficult to find a job, they could create real headwinds for the growth outlook:
A third form of evidence related to slack concerns the characteristics of the extraordinarily large share of the unemployed who have been out of work for six months or more. These workers find it exceptionally hard to find steady, regular work, and they appear to be at a severe competitive disadvantage when trying to find a job. The concern is that the long-term unemployed may remain on the sidelines, ultimately dropping out of the workforce. But the data suggest that the long-term unemployed look basically the same as other unemployed people in terms of their occupations, educational attainment, and other characteristics. And, although they find jobs with lower frequency than the short-term jobless do, the rate at which job seekers are finding jobs has only marginally improved for both groups. That is, we have not yet seen clear indications that the short-term unemployed are finding it increasingly easier to find work relative to the long-term unemployed. This fact gives me hope that a significant share of the long-term unemployed will ultimately benefit from a stronger labor market.
Finally, she highlighted how the falling participation rate skewing the unemployment rate, which could render the Evans Rule ineffective:
A final piece of evidence of slack in the labor market has been the behavior of the participation rate–the proportion of working-age adults that hold or are seeking jobs. Participation falls in a slack job market when people who want a job give up trying to find one. When the recession began, 66 percent of the working-age population was part of the labor force. Participation dropped, as it normally does in a recession, but then kept dropping in the recovery. It now stands at 63 percent, the same level as in 1978, when a much smaller share of women were in the workforce. Lower participation could mean that the 6.7 percent unemployment rate is overstating the progress in the labor market.
Past Fed chairs might have used such a venue to deliver an academically oriented speech, but not Janet Yellen. Notwithstanding the perceived dovish tone of the speech, this Fed chair demonstrated that she goes beyond the academic application of rules and models and uses a multi-factor approach in a thinking way to the conduct of monetary policy.

The regulator play whack-a-mole
The problem of model weakness creates a dilemma for regulators in applying standards, such as the EBA's stress tests or Basel standards for banking. On one hand, there is an urge to appear evenhanded by applying a one-size-fits-all approach to regulation. On the other hand, the banking industry is filled with people whose jobs is to get around rules and to game existing regulations.

As an example, I can understand the intent of the Volcker Rule is to minimize the risks taken by investment banks in a TBTF banking structure. However, its implementation has spawned a devil-is-in-the-details nightmare of rules upon rules as regulators play a game of whack-a-mole with bankers. Instead, a simpler approach might be to try and get at the heart of the problem by properly aligning incentives with risk taking. Bring back the partnership investment bank. If they want to lever up the balance to 40 to 1 and take risks, let them! If they fail, it will be the personal net worth of the investment banking partners on the line.

A multi-factor approach to regulation
In other cases, where it is difficult or impossible to get at the heart of the problem, a multi-factor approach of incentive alignment or characteristics measurement may be warranted, much in the spirit of what Yellen used to model the degree of slack in the economy.

For instance, I was talking to a friend about the problems that Vancouver faced with foreign property buyers. Vancouver, like London, which has become the preferred locale for billionaires, has seen an influx of foreign buyers drive up property prices to such an extent that puts housing ownership for many local residents. One American parallel that I am familiar with is the Hamptons - local residents cannot afford to live there as wealthy buyers from New York City have driven prices to stratospheric levels.

In my discussion, one proposal  on the table is to tax non-resident property owners at a higher rate than residents. Notwithstanding the debate on the merit of such a proposal, the implementation of such a scheme is highly problematical as most regulations can be easy to game. After all, there is an army of tax lawyers, accountants and other consultants whose job it is to try and get around these rules.

In such a case, one approach might be to employ a multi-factor approach to the problem in defining the activity that the regulator is trying to incentivize and dis-incentivize. Supposing that the intent of the regulations is to encourage owner-occupancy of single family housing and condominiums. One way to implement such a regulatory regime is to set the default property tax rate at a very high level. If the normal property tax rate is 1% of the property value per annum, set it higher at, say, 5% or 10%. Then create a system which grants owner-occupiers of the property tax credits for engaging in owner-occupier like activities. The taxpayer could accumulate tax credit points based on activities such as:
  • Having one or more children at the local school, whether private or public;
  • Showing utility bills over a certain amount which shows that the house is occupied and not vacant; and
  • Showing that you voted in the last municipal, provincial or federal election.
Accumulate enough points and your property tax bill goes back down to the normal rate for resident. Otherwise, you pay the non-resident rate.

Understand that I am not necessarily endorsing the philosophical approach underlying this tax proposal as the topic is up for debate. The proposal is also incomplete as it ignores the taxation treatment of rental housing stock. Regardless, I engaged in this exercise as an illustration of how to use a multi-factor modeling techniques to achieve a regulatory aim. As well, the factors should be relatively uncorrelated to each other, otherwise you wind up with a confusing mishmash of rules. Too often, regulators approach the problem from a legal framework and spend too much time obsessing over the details of a model or a regulatory regime while glossing over the intent of the regulations. That`s when you get rules piled on top of rules, much like the tax code.

When clear thinking triumphs over blind modeling and rule application, we all win.

Cam Hui is a portfolio manager at Qwest Investment Fund Management Ltd. (“Qwest”). The opinions and any recommendations expressed in the blog are those of the author and do not reflect the opinions and recommendations of Qwest. Qwest reviews Mr. Hui’s blog to ensure it is connected with Mr. Hui’s obligation to deal fairly, honestly and in good faith with the blog’s readers.”

None of the information or opinions expressed in this blog constitutes a solicitation for the purchase or sale of any security or other instrument. Nothing in this blog constitutes investment advice and any recommendations that may be contained herein have not been based upon a consideration of the investment objectives, financial situation or particular needs of any specific recipient. Any purchase or sale activity in any securities or other instrument should be based upon your own analysis and conclusions. Past performance is not indicative of future results. Either Qwest or I may hold or control long or short positions in the securities or instruments mentioned.

1 comment:

Rik said...

Problem here is that 99% of the people have their personal preferences (over-)represented in the models they use if they can basically make them up.
You need safeguards to assure that is not happening even for people with the best intensions. Just look at inflation measures and now unemploymentfigures to see that is necessary. The room here was much smaller and still the most convenient stuff came on top.

Another approach would be start from the basics of the model. It is based on assumptions (either explicitly mentioned or implicit (more as an integral part of the bigger picture).

Here as your example shows unemployment as a measure for the economy seems to have some problems. Size of the workforce for instance.
If you would try to find a logical explanation for that you very likely bump into the issues that are relevant.
This is a different approach from what Yellen does, imho it is a lot more objective (but more difficult to do as well).

Same with part time vs full time. The model is based upon the idea that that relation is more or less stable over the period you are looking at. It clearly isnot now. Modelingwise you probably have to amend your model for it. Eg add on fulltime basis hours effectively being unemployed by part timers compared to the hours they would like to work. Corrected for the percentage historically for instance.

Wages is another point. Imho with another explanation. Wages in the West seem still way too high compared to those inm say China (seen productivity (as in what a worker really can produce not if he is so cheap that you use him iso a machine like in the West, but mainly quality wise). Looks to me that a correction is taking place here. As the labourmarket is volatile now because of the crisis the correction goes faster than in normal times.
Likely seen the competition internationally on the international labourmarket we will see a powershift from labour to capital (and with it a repricing) as well. Bargaining power of the average Western worker simply has reduced considerably. Therefor one would expect downward pressure even in a normal economy. With a lot of loops/side effects (via consumption etc).