Coincident Index

The Conference Board Coincident Economic Index (CEI) for Japan is a composite average reflecting current economic activity. It averages 4 published series on employment, production, income and sales, and is highly correlated with real GDP without being GDP itself.

It is not a survey, with no respondents, and not a forecast; forecasting is the role of its companion Leading Economic Index (LEI). It is also distinct from the Cabinet Office coincident CI, a separate Japanese index with base 2020 = 100. The 2016 = 100 base is arithmetic, not a boom or bust line, and no single component equals the index.

Each release covers 1 reference month at 10:30 AM ET, about 6 to 8 weeks later; next date is announced 1 release ahead. Monthly updates revise the prior 6 months, with missing values estimated then replaced; annual benchmark recomputes history, resets base and factors, breaks comparison.

LEI and CEI appear with monthly and 6-month changes, levels at 2016 = 100, tables with contributions and 1-month and 6-month diffusion. Read monthly and 6-month direction plus breadth; the system signals peaks and troughs, with no buy or sell rule.

4 series drive the index with volatility-equalizing weights from a 2009 vintage that shift at benchmarks.

• Number of Employed Persons, 53.6%: captures employment in persons, seasonally adjusted; largest and least volatile; rises when hiring tops firings.
• Industrial Production, 11.7%: captures physical output of manufacturing, mining and utilities; most cyclical, most volatile input.
• Wage and Salary Income, 16.7%: captures aggregate labor income in real-time estimates; tracks spendable income.
• Real Retail, Wholesale, and Manufacturing Sales, 18.0%: captures real spending through the distribution chain; quarterly manufacturing leg is linearly interpolated to monthly.

The January 8, 2009 release for November 2008 put the CEI at 100.5 (2004=100), down 0.9%. It had fallen 0.4% in October and 0.1% in September, while the LEI fell 2.6% in November and 10.4% over 6 months. Over 6 months through November it fell 3.3%, a negative 6.4% annual rate, its weakest since 1975. Six-month diffusion stood at 0.0%, with no component advancing.

Yet 2 of 4 components rose, employment and wage income, while production and sales fell and dominated, so 1 part never equals the index. Real GDP, inflation-adjusted output, fell 1.9% at an annual rate in the third quarter, its second straight quarterly fall since 2001.

1. Monthly change: Compute symmetric % change for each of the 4: 200 x (new minus old) divided by (new plus old). Series already in % form would use simple difference, but none of the 4 do. Deflate real legs before computing.
2. Standardize: Multiply each % change by its standardization factor, set inversely to volatility, to get contributions in points. For example, employment up from 63550 thousand to 63910 thousand rises 0.56%, times its factor gives about 0.30 points.
3. Sum: Add the 4 contributions to get total monthly change. In the illustration, 0.30 minus 0.99 plus 0.13 minus 0.35 equals about minus 0.90%.
4. Rebuild levels: Recurse the index from 100 in the first month: new equals old times (200 plus change) divided by (200 minus change). The illustration moves 101.4 down 0.90% to about 100.5.
5. Rebase and maintain: Rebase to average 100 in base year 2016 by scaling preliminary levels; rebasing happens at benchmark, not monthly. Inputs arrive seasonally adjusted, quarterly legs are linearly interpolated to monthly, and missing values are estimated then replaced. When a leg is missing, the other factors rescale to sum to 1.
6. Breadth: Score components 1.0 for a rise above 0.05%, 0.5 for a move within 0.05%, and 0.0 for a fall; average to 100-scale diffusion. Publish for 1-month and 6-month spans from actual changes; factors use 4 decimals, contributions 2, the index 1.