In a previous post by Obey, The unbearable weirdness of Greg Mankiw he quoted Mankiw as follows: 

The expression "create or save [4 million jobs]," which has been used regularly by the President and his economic team, is an act of political genius. You can measure how many jobs are created between two points in time. But there is no way to measure how many jobs are saved. Even if things get much, much worse, the President can say that there would have been 4 million fewer jobs without the stimulus. [...] So he gave us a non-measurable metric.

If Dr. Mankiw is bothered by "non-measurable metrics", he should abandon economics and take up some other occupation. Economics is largely based on "non-measurable metrics".

Let's start with "market capitalization", which is the outstanding shares of a corporation times market price. This is "non-measurable" because if you actually attempt to sell that many shares, or even a large percentage, the market price would change. Furthermore, it is well known that there are several differing estimates of the value of a corporation, such as "book value", "enterprise value", "takeover value" and "breakup value". Depending on how they are made by an investment bank, the latter estimates will vary, and in any case they will likely consist of ranges and qualified estimates. Think of "Antiques Roadshow" and the appraiser's advice that the pair of vases will fetch $10,000 to $15,000, maybe $18,000 in a well attended auction, and they aren't really a pair so they might fetch more if sold separately, and in any case insure them for $20,000.

Even more non-measurable is "total stock market wealth". It may be of academic interest to know that if you sum up the market capitalization of all stocks traded in the market you get a large number. But if market cap is non-measurable, stock market wealth is even more so. Alas, due to the herd mentality of investors, the theoretical value of the market as a whole cannot be estimated with much more precision that the value of individual corporations. Prices are correlated, they go up and down together, and the Central Limit Theorem doesn't really apply.

Another non-measurable is "housing wealth", which takes the estimated appraised values of homes and sums them up over the country. This is even more mythical, since houses aren't standard commodities like stocks, and the market data on individual selling prices is even more dubious. There are reasons why homes might be sold at higher or lower than actual market prices.

In general, any computation of "market prices" times "quantity of items" summed over all the items is "non-measurable", unless the aggregate quantity of items is less than a small multiple of the daily market volume. Such aggregate sums are completely theoretical, and they have no more objective existance than unicorns or dragons.

Unfortunately, these non-measurable, market-based numbers aren't just used for theoretical discussions of psychological wealth effects - when "market-based accounting" is used for large banks, they sustitute for other estimates which may be more accurate.

For example, if a bank has a few 100 billion of mortgage-backed securities on its books, the quantity exceeds a few days volume in the market, and the market-based value is therefore theoretical and non-measurable. It is no sounder than the alternative methods of valuation based on historical prices and anticipated cash flows.

Even when individual firms have smaller quantities on their books, the same problem of non-measurability exists when speakng of the collection of firms. You can't take the amount of mortgage-backed securities held by each firm, multiply them by the market prices, sum over the firms, and then make assertions like "regional banks have another $500 billion in mortgage losses to go". The most that can be said is that they might have additional losses, but we can't measure accurately what they will be.