Feb

24

The above plots three asset price bubbles (defined in hindsight as big price run-ups followed by big declines); in Nasdaq, Nikkei 225, and nominal Shiller Home Price Index.

All three were sampled on a monthly basis. Values were transformed by taking the log (to allow % change in value to scale), and further adjusting the log values so all three bubbles peaked at the same nominal value. Timescale is in month numbers, with all three aligned so their respective bubble peaks coincided in time.

The Nasdaq cuts off at the present, rebounding much better than Nikkei did after the same elapsed time after peak. Nasdaq also had a faster/steeper gain than Nikkei, and commensurately faster/steeper decline. Nominal house prices show less noise and a slower move - rather similar to Nikkei - might be expected with lower liquidity.

A story still unfolding.

My friend asked me the other day about how I said lower "noise" might be a result of lower liquidity? How do I figure?

Perhaps incorrectly: How does transaction price behavior change when average transaction time goes from seconds to months? In the case of financial instruments, large bid-ask and long time between transactions could increase volatility. OTOH volatility spikes usually correlate with volume spikes.

The smoothness of Shiller HPI may also be due to price anchoring; loans don't get funded unless the house appraises, and appraisals are mostly based on recent comparable sales.

It is also possible that HPI is more flexionistic and with less valid price discovery than Nasdaq or Nikkei. 


Comments

Name

Email

Website

Speak your mind

1 Comment so far

  1. David Merkel on February 25, 2012 9:49 am

    The phenomenon you speak of is very normal when you talk about debt-financed illiquid investments. Most bonds don’t trade; most true bond private placements never trade.

    But they all have to be priced. Accounting must be done monthly, quarterly, annually. So we set up complex pricing grids to transform pricing data from the few transactions that do happen into prices for everything else.

    There’s a lot of math involved, and assumptions that tend to enforce smooth transition of prices from one state to the next, because anything that does not move slowly will get questioned by those receiving the accounting. Any big move requires a big reason, and when transactions are sparse, big reasons are few, until enough data confirms the transition to the new state.

    It gets worse when the illiquid investments are funded largely by equally illiquid debt. Don’t let the seeming liquidity of GSE-based MBS fool you. Underneath it all are a zillion illiquid residential real estate transactions, which grow more illiquid as market-based loan-to-value ratios approach 90%, and become frozen near 100%. During the bust, appraisal and underwriting standards rise, making new transactions yet more difficult. And, yes, appraisals use these same grids, and adjust slowly to new information.

    So when Case-Shiller or the GSEs come to these data, they impose smoothing — really it is Bayesian statistics at work, where new data slowly modifies prior calculations. Really, it’s all you can do; to do otherwise and allow for rough adjustments would bring complaints.

    But the product of all this is a lot of serial correlation, momentum if you will, to the statistics, and pressure to the downside on home prices for the near future.

Archives

Resources & Links

Search