there were two somewhat important economic reports for October released this week, coincidentally both rescheduled for release at 8:30 AM on Wednesday, retail sales from the commerce department and the consumer price index from the bureau of labor statistics (BLS)….together they’re often used by economists to generate a metric that goes by ‘real retail sales’, which is supposed to be something analogous to the real personal consumption expenditures metric of GDP, and which is said to be an indicator used by the NBER Business Cycle Dating Committee to determine turning points in the economy, ie, the beginning and end points of official recessions…even the Fed generates a graph for real retail sales from these two reports, constructed by a simple deflation the monthly retail sales results with the change in the consumer price index of the same month…the last time we covered these two reports together we suggested that this metric was an inexact and crude measurement that was prone to leading to incorrect conclusions about the direction of the economy, especially in the way it’s viewed by the Fed, NBER (National Bureau of Economic Research) and most economists…today we’re going to show why it’s wrong, and moreover, point to the correct way real retail sales should be calculated…but before we do that, we’ll take a look at the two reports in question…

October Retail Sales Increase 0.4% on Unexpected 1.4% Jump in Vehicle Sales

according to the Advance October Report on Retail and Food Service Sales (pdf) from the Census Bureau, seasonally adjusted retail and food services sales were at $428.1 billion in October, an increase of 0.4 percent (±0.5%)* from September and 3.9 percent (±0.7%) above October 2012…unadjusted retail sales, extrapolated from a small sampling were reported at $421,946 million, an increase over the revised September sales of $402,927… that asterisk points us to a footnote that tells us Census doesnt not yet have sufficient statistical evidence to determine whether seasonally adjusted sales actually rose in October of not, but that they’re 90% certain the September to October change was between a decrease of 0.1% and an increase of 0.9%; with that caveat, we’ll look at these widely followed advance estimates as reported…

October sales were better than expected; the consensus of analysts had forecast no change in total sales, due to an apparent decrease in October auto sales based on reports from manufacturers; however, the small sampling of auto and other vehicle dealers surveyed for this report indicated their actual sales for the month had increased 1.4% over September, adding to October’s sales increase rather than subtracting from it… including parts dealers, October’s seasonally adjusted sales in the vehicle sector were at $81,871, a 1.3% increase over September; that percentage change is shown in the first column in the table below from the report, where we can also see in the second column that the percentage change from October 2012 for auto and other vehicle dealers was 11.9%, with a 10.6% year over year sales increase for the whole automotive category…also note that total retail sales excluding motor vehicle & parts dealers were up 0.2% for the month, also beating expectations

Oct retail sales table

  it’s also clear from looking at that first column above that most kinds of business, except for building and garden supplies and gas stations, saw seasonally adjusted sales rise, led by a 1.6% month over month increase in the sporting goods, hobby, book & music store group, who saw seasonally adjusted sales rise to $7,714 million for the month…clothing stores rebounded to show a 1.4% month over month increase while selling $37,847 million of clothing and shoes, and electronics and appliance stores also saw October sales rise by 1.4% to $8,699 million, while furniture stores logged a one month sales increase of 1.0% to $8,687 million, as did restaurants and bars, whose October sales were at a seasonally adjusted $46,317 million…also showing month over month sales gains were drug stores (health and personal care) at 0.5%, where $24,327 million of products were sold in October, non-store or online retailers, who moved $37,847 million of merchandise, 0.4% more than September, and general merchandise stores, where sales of $55,303 million were up 0.2% for the month..meanwhile, food and beverage store sales were virtually unchanged, with seasonally adjusted sales rising to $54,693 million in October, from $54,682 million in September, gas station sales fell 0.6% to $45,165 million as gasoline prices fell 2.9%, and sales at building material & garden equipment stores fell 1.9% to $25,793 million…also note that miscellaneous retailers, where sales fell 0.1%, account for another $10,609 million sales for the month…

also note the third and fourth columns on the table above, which represent the 2nd estimate of retail sales for September…as originally reported, retail sales were down 0.1% to $425.9 billion in September; that estimate has now been revised to a preliminary figure of $426.369 billion, and that works out to a 0.0% change from the seasonally adjusted sales of $426.355 billion in August…the major revision to September sales was in in motor vehicles and parts sales, which were originally reported at $79,773, down 2.2% from August; that’s been revised to show September motor vehicle and parts sales at $80,815 million, down just 1.2% from August….other retail sales groups which saw sales revisions of more than 0.2% included clothing stores, which were originally reported to have seen September sales fall 0.5%, which has now been revised to a 0.9% drop; the sporting goods, hobby, book & music store group, originally reported to have seen just a 0.5% change who’ve now seen their September totals rewritten to indicate a 1.3% sales gain; general merchandise stores, initially reported as up 0.4% which has been revised to just a 0.1% increase; and restaurants and bars, where September sales were reported up 0.9% three weeks ago, which has now been revised to just a 0.2% gain from August…on net, this revision suggests a small upward revision to the durable goods component of PCE in 3rd quarter GDP, partially offset by an even smaller downward revision to non-durable goods…

our FRED bar graph below shows the monthly percentage sales change for each of the 12 major retail sales categories since last October….the monthly data for each of the past 13 months is represented by a grouping of 12 bars, with each type of retail sales represented by its own color code in each group, wherein an increase in sales appears above the ‘0’ line and a decrease below it…from left to right in each group is a dark blue bar representing the percentage change in motor vehicles and parts sales, a red bar indicating the change at general merchandise stores, followed by the percentage change at food and beverage stores in green, the sales change restaurants and bars in mauve, the change at gas stations in orange, the change non-store or online retailers in sky blue, the change at building and garden supply stores in light green, the percentage change at drug stores in mustard, the change in sales at clothing stores in in pink , the change at electronics and appliance stores in purple, the change at furniture stores in yellow, and the percentage change in sales at stores specializing in sporting goods, books or music in pale blue…(click to enlarge)

FRED Graph 

Slowest Inflation Since 2009 as October Consumer Prices Fall 0.1% on Lower Gas Prices

also seeing a delayed release on Wednesday, the Consumer Price Index for October from the Bureau of Labor Statistics was one of the economic reports most profoundly effected by the government shutdown because its input data is normally collected by an army of field agents who visit thousands of retail stores, service establishments, rental units, and doctors’ offices, et al each month to track changes in price in the items appearing in the index…to compensate for the loss of 16 days of such data collection, BLS personnel normally devoted to in-office maintenance work were redirected into field data collection and index production, and as a result BLS managed to collect roughly 75 percent of the amount of prices normally used in constructing the CPI, albeit only during the last two weeks of the month, absent prices from the prior days…understanding that this may make the results less reliable, we’ll go with what’s reported, since considering the methodology there wont be any revision anyhow…

the seasonally adjusted Consumer Price Index for All Urban Consumers (CPI-U) decreased 0.1% in October; entirely due to lower energy prices, as the energy price index fell 1.7% on gasoline prices that were 2.9% lower in October than September.. meanwhile, the food price index rose 0.1%, and the Core CPI, which is the index of all prices less food and energy, also rose 0.1%…the unadjusted CPI, which is based on prices over 1982 to 1984 equal to 100, fell from 234.149 in September to 233.546 in October, and the unadjusted energy index fell from 248.513 to 238.524, while the food index rose from 237.522 to 237.871 and the Core CPI rose from 234,782 to 235.162…the CPI is now up just 0.96%  to 231.317 on a year over year basis, the lowest annual inflation rate since October 2009, while the Core CPI has logged a 1.68% (rounded to 1.7%) annual rate of inflation for all items less food and energy…

other than 2.9% lower prices for gasoline, items in the energy price index which fell in price in October included fuel oil, which was down 0.6%, the index for propane, kerosene and firewood, which slipped 0.4%, and other motor fuels, which also were 0.4% lower; the energy services index was off 0.2%, as a 1.0% decrease in the price of natural gas was partial offset by a 0.1% increase in the cost of electricity…on a year over year basis, the energy price index has fallen 4.8%, as the 9.5% decrease in energy commodities outweighed the 3.3% increase in energy services…

  both of the major components of the food price index, food at home and food away from home, were up 0.1% in October, as was the overall food index; among categories of food at home, prices for cereals and baked goods fell 0.4% for the month, as bread prices fell 3.0%, flour prices fell 0.9%, and the rice pasta and cornmeal index rose 2.2%, while dairy products were 0.2% lower as a 1.3% drop in cheese prices and a 1.4% drop in ice cream prices offset an 0.3% increase in milk prices…meanwhile, prices for meats, poultry, fish, and eggs rose 0.6% on a 1.0% increase in pork prices, 1.5% higher prices for fish and other seafood, and 1.8% higher egg prices, while fruit and vegetable prices were 0.2% higher as lettuce prices rose 4.0%, citrus fruits rose 1.9%, fresh fruits other than apples, bananas and citrus rose 3.2%, while prices for processed fruits and vegetables fell 1.2%; in the beverage group, which saw prices rise 0.4% in October, roast coffee prices increased 0.7% and carbonated drinks fell 0.1%, while the catch all category of other food at home saw a 0.2% price decline, as sugars and sweets rose 0.4%, snacks rose 0.5%, while both butter and margarine and spices and seasoning prices fell 1.1%…in the food away from home grouping, fast food prices rose 0.2%, full service restaurant prices were unchanged, and food at work and school cost on average 0.8% more than it did in September…

  among core prices that increased in October, the cost of shelter rose 0.1%, as rents for one’s primary residence rose 0.2%, homeowners equivalent rent rose 0.2% while room prices at hotels and motels fell 4.0%; the cost of apparel fell 0.5%, as men’s and boys clothing rose 0.8% as prices for pants and shorts were up 10.2%, women’s and girls’ apparel fell by 0.8% on 2.2% lower prices for girls clothing, while the prices for footwear fell 0.6%…the overall transportation index, which includes fuel costs, showed a 0.7% decline, but prices for transportation commodities less fuel were unchanged, as prices for new cars and trucks fell 0.2% while prices for used cars and trucks rose 0.3% and parts and accessory prices fell 0.1%…meanwhile, prices for transportation services rose 0.7% in October as public transportation costs rose 2.2% on a 3.6% spike in airline fares…the medical care index showed no change for the month as medical commodities were up 0.3% in price on a 0.6% increase in nonprescription drugs prices while costs for medical care services were off 0.1% on 0.2% lower hospital and related services; in addition, the recreation index rose 0.1% as prices for recreation commodities were unchanged overall as prices for video & audio equipment fell 0.4%, offsetting increases of 0.4% in prices for sporting goods and pets and pet supplies, while recreation services prices increased 0.2% on a 0.4% increase in cable and satellite service and a 05% increase in veterinary and other pet services..the last major price index, for education and communication, showed a 0.2% increase in October as education costs were up 0.4% and communication prices were unchanged; within education and communication commodities, which were down 0.4%, textbook prices were up 1.0% and personal computers fell 1.3% in prices, while education and communication services rose 0.3% for the month as college tuitions were up 0.4% and telephone services rose 0.2%…

  our FRED graph below shows the relative price change in each of these major components of the CPI-U since January 2000, with all the indexes reset to 100 as of that date to make for an apples to apples comparison, as two of the composites were restructured in 1997 and the rest were based on 1982-4 prices…in blue, we have the track of the change in the price index for food and beverages, which tracks pretty close to the track of the CPI-U, which is shown in black; in red, we have the change in the price index for housing, which includes rent and equivalents, utilities, repairs and other homeowners costs like insurance and which at 41% of the CPI also tracks close to the CPI… in violet, we have the price index for apparel, which has been the only index to show a net price decline over the decade…the transportation index, in brown, shows the impact of volatile gas prices on the cost of transportation, while the price index for medical care in orange has obviously risen the most over the entire period…in addition, education and communication prices are tracked in dark green, and the track of the recreation price index is shown in light green..

FRED Graph 

We Find Real Retail Sales Rose .75% in October

now, as we mentioned in opening, an economic metric called “real retail sales” is commonly constructed by using the CPI to adjust retail sales for the same month…the reasoning behind this is that we’d want to know how many units of product were sold, not just the dollar value, and if prices were rising (or falling), the retail sales figures, which do not adjust for price changes, we’d get a skewed result…for instance, if the retail sales report told us the dollar sales of apples were up 5% month over month, the actual change in the number of apples sold couldn’t be determined unless we knew if prices for apples changed or not, and by how much…and the reason we would want to know real retail sales early is because real personal consumption expenditures (adjusted for inflation) constitute 70% of GDP, and although that is mostly spending for services, spending for durable goods like cars and appliances, and non-durable goods, like food, clothing and gasoline sill make up a third of expenditures, or over 23% of GDP…in the common method of computing real retail sales for October, the increase in dollar value retail sales of 0.4% is adjusted for prices that were 0.1% lower, suggesting that real retail sales had increased by 0.5% for the month (or 0.47%, using two decimal places)…but as we’ve seen in reviewing these two reports together, the expenditures they cover don’t really match up; in fact, more than 60% of price changes covered by the CPI are for services, not products sold at retail at all….so, if we use the CPI to adjust retail sales for inflation, there’s a good chance there will be months where the price change in many of the items sold at retail would go in a different direction than the majority of the CPI components…to put it another way, by adjusting retail sales with the CPI, we’d often end up adjusting sales of cars, clothing, and appliances with the price changes of doctor’s visits, apartment rents, airfares, and tuition increases…the only correct way to adjust retail sales with the CPI would be to itemize retail sales and adjust it with the corresponding CPI component price change…we’ll run through an example of how that could be done here using our October data, but as you’ll see it’s a very tedious process better left to a computer program…

the largest component of retail sales is sales at motor vehicle & parts dealers, which at a seasonally adjusted $81,871 million in October accounted for 19.1% of total retail sales; of that, $75,035 million was from vehicle sales, and the rest was sales at parts and tire stores….and as we saw, sales at car dealers increased 1.4% in October…we have two components of the CPI which are suitable to adjust the vehicle sales for inflation; prices for new cars and trucks, which were down 0.2% and prices for used cars and trucks, which were up 0.3%; the weighting of new vehicles in the CPI was 3.133, and the weighting of used vehicles was 1.889, which means that new vehicles were five-eighths of sales, and used vehicles accounted for three-eighths; so we want to adjust 5/8ths of the 1.4% increase up by 0.2%, and 3/8ths of it down by 0.3%, which quite coincidentally gives us a real vehicle sales increase of just over 1.4%, not much change….also buried in that aggregate motor vehicle data are sales of parts and tires, which were up 0.48% in October, while the corresponding prices for motor vehicle parts and equipment were down 0.1%, resulting in a real sales increase of .58%…so in a quick and dirty calculation, we have real retail sales for the motor vehicle & parts dealers component up 1.346% for October…moving on, another large component of retail sales is sales at food and beverage stores, which accounted for 12.8% of retail sales and saw dollar value sales virtually unchanged in October…applying the 0.1% increase in the ‘food at home’ index of the CPI to those sales, we thus find that real retail sales at food and beverage stores fell 0.1% in October…there were also $46,317 million in sales at restaurants and bars in October, or 10.8% of retail sales; those sales increased 1.0% month over month, and we’d deflate them with “food away from home” from the CPI, which indicated a 0.1% MoM price increase; thus real retail sales at bars and restaurants increased 0.9% in October…$45,165 million in sales at gasoline stations were down 0.6% according to the retail sales report, where they accounted for 10.6% of sales; however, the gasoline component of the CPI tells us that gasoline prices were down 2.9%, so real sales at gasoline stations were up around 2.3%…sales at furniture stores, at just over 2% of retail sales, were up 1.0%, which was not broken down into types by the retail sales report, whereas there are prices for a dozen separate items that might be found at a furniture store on the detailed CPI list…so to adjust furniture sales for inflation, we’d use the overall ‘household furnishings and supplies’ component, which includes several non-furniture items but is still mostly furniture with appropriate weightings; prices for that index were down 0.2%, so we’d conclude that real furniture sales were up 1.2% in October…sales of $55,303 at general merchandise stores, which were up 0.2% and account for 12.9% of retail sales, presents another problem, in that these stores sell a wide range of merchandise for which there’s probably over a hundred CPI entries…we’ll adjust those sales with the price change for consumer “commodities less food and energy commodities”, which includes all retail items except food, energy, and which was down 0.1% for October, and thus results in a real sale increase of 0.3% at general merchandise stores…the $8,699 million of sales at electronic and appliance stores were not broken down in October, but in prior months appliances, TV & camera stores accounted for about 3/4ths of those sales and computers and software stores were the rest; together they accounted for just over 2.0% of retail sales and increased 1.4% in nominal terms…in the CPI index, appliances, unchanged in price, are a household item, TVs, down 0.6% and cameras, down 0.1% are entertainment, while personal computers, down 1.3%, are educational; using the CPI weightings of each, we figure a 1.6% real sales increase in TV and appliances sales, and a 2.7% real sales increase in PCs and related equipment, giving us a 1.9% increase in real unit sales for the retail group…computing the other retail businesses in a similar manner, we find that real sales at clothing stores, which account for over 4.9% of retail sales, were up 1.9% in October; while real sales at drug stores, which account for 5.7% of sales, were up 0.4% based on weighted price indexes of personal care products and drugs…real sales at sporting goods, hobby, book and music stores, which account for 1.7% of retail, are up 1.6%, same as the actual increase, when adjusted with the unchanged recreational commodity index, and real sales for non-store retailers, which account for 8.8% of retail and adjusted as general merchandise, were up 0.5% in October…we also find that real sales at building and garden supply stores, which at $25,793 million account for 6.0% of retail sales, when adjusted with prices for tools, hardware, outdoor equipment and supplies, were down 2.0% in October and that miscellaneous store retailers, at 2.5% of the total, saw no change in real retail sales…

now that we’ve done this exercise of computing the real retail sales percentage change of each business group covered by the retail sales report, we have to add them together to come up with a final figure for real retail sales, so we’ll express that math in a detailed longhand expression so we all can see what we’re doing…real retail sales for October = 19.1% * 1.346% + 12.8% * -0.1% + 10.8% * .9% + 10.6% * 2.3% + 2% * 1.2% + 12.9% * .3% + 2.0% * 1.9% + 4.9% * 1.9% + 5.7% * .4% + 1.7% * 1.6% + 8.8% * .5% + 6.0% * -2.0% + 2.5% * 0% = + .75%, assuming we’ve done the math right…so our calculation of real retail sales is apparently nearly 0.3% higher than the simplistic method of just deflating total sales with the CPI…one obvious reason for that was the gas station component, which would have been carried as down 0.5% when deflated with CPI-U, but are apparently up 2.3% when deflated with gasoline prices…perhaps some combination of gasoline and tires, batteries and accessories would have been a more appropriate deflator, but we have no data on which to base that percentage…another reason that this workout produces a higher real retail sale result is more valid; that being that prices for items bought at retail (aka commodities in CPI jargon) have been trending down in price while prices for services have generally been up; that’s clear just from reviewing some of what we saw in this month’s CPI core data; it makes no sense to deflate car sales (transport commodity) with the 3.6% spike in airfares (transport service)…but that’s the kind of thing actually being done when the NBER, the Fed and other economists deflate retail sales using the entire CPI…

Job Openings and Hiring at 5 Year Highs

there were also two more employment reports released on Friday, both at 10 AM, as the Bureau of Labor Statistics is quickly catching up on its shutdown delayed backlog…the Job Openings and Labor Turnover Summary for September, which includes estimates of the number and rate of job openings, hires, and separations by industry and by geographic region, uses the phrase “little changed” in describing the month over month change in the first five metrics mentioned in the opening summary, in recognition of the fact that what changes did occur fall within the margin of error in this report, as they only survey 16,400 establishments to create this report, in contrast with the 145,000 businesses and government agencies surveyed for the Current Employment Statistics, aka the establishment survey...what they actually reported is that that the seasonally adjusted number of job openings rose by 69,000, from 3,844,000 in August to 3,913,000 in September, the highest level since before the recession, with 32,000 new openings in retail trade while manufacturing job openings, primarily in the durable goods area, fell by 16,000…the unadjusted number of job openings was a bit higher at 3,950,000, with 531,000 of those in retail, up from the 455,000 retail job openings in August…the seasonally adjusted number of workers hired (or rehired) in September also hit a new 5 year high, at 4,585,000, up 26,000 from the 4,559,000 hired in August…the biggest jump in hiring came in the professional and business services sector, where 1,012,000 were hired, 33,000 more than were hired in August…total separations were up as well, by 21,000 to 4,426,000 in September, as the leisure and hospitality sector saw 51,000 more employees fired, laid off, quit or otherwise discharged than were in August…the difference between those hired and those separated in September was thus 159,000, which is pretty close to the 163,000 payroll jobs gained in September indicated by the most recent establishment survey revision…this JOLTS survey had been showing roughly 20,000 less net new jobs per month than indicated by the Current Employment Statistics throughout this year…

FRED Graph

monthly job separations are further divided into those who quit, those who were fired or laid off or otherwise let go, and ‘other’ separations, which includes retirement and employee deaths; in seasonally adjusted data, 2,342,000 workers reportedly quit their jobs in September, down a bit from the 2,364,000 who quit in August, another 1,727,000 were laid off or otherwise discharged by their employer, an increase from August’s 1,676,000 discharges, and other separations numbered 303,000…the unadjusted data shows 2,567,000 worker quits, with nearly a million of those who had worked in retail or food service quitting their jobs, and 1,848,000 layoffs or discharges, with the largest number of those, 431,000, coming from professional and business services…our FRED bar graph above shows the monthly change in each of the major metrics from this series over the past two years…each grouping of five bars represents the changes that occurred in one month, with increases above the ‘0’ line and decreases below it…the blue bars indicate the change in the number of job openings each month, while the orange bars represent the change in the number hired for the month; in addition, the red bars indicate the change in layoffs and discharges, the green bars indicate the change in the number of those who quit each month, and each violet bar shows the monthly change in the total number of separations…you might notice that increases in overall job turnover have slowed somewhat from last year, and in fact, from the years before it, which you can view if you click on the FRED graph for a larger view and change the “Observation Date Range” for “All Bar Series”…

Regional and State Job Creation and Unemployment Rates Show Little Improvement

the other jobs report released by the BLS on Friday was the Regional and State Employment and Unemployment Summary for October, and like the US employment summary usually released on the first Friday of each month, this report is actually two reports, one which reports new payroll job creation as reported by businesses and agencies, and the other on the status of employment or lack of it as reported by households, combined with state data and estimates…so basically what this report does is takes the data from the October jobs report which we covered two weeks ago and breaks both job creation and household employment statistics down by region and by state…so like that report, data collection was delayed by the shutdown, and there was a similar divergence in reporting methods between the two surveys, wherein furloughed government employees were still considered employed by the establishment survey, while they were supposed to be marked as unemployed on temporary layoff in the household survey…

seasonally adjusted state data from establishments showed the states with the largest seasonally adjusted increases in payroll jobs were Florida with 44,600, including 11,300 in professional and business services and 11,000 in leisure and hospitality, California with 39,800, including 12,200 government jobs, and North Carolina with 22,200, including 7,600 in education and health, while Kentucky lost 12,600 jobs and Washington’s payroll employment was reduced by 8,100…the largest percentage job increase occurred in Wyoming, which saw payroll jobs rise by 1.0%, while Delaware, Florida, and Nevada all saw job increases of 0.6%…percentage-wise,  the largest month over month decrease in employment occurred in Kentucky at 0.7%, followed by South Dakota at 0.6% and Washington at 0.3%…the map below, taken from page 23 of the full pdf of this report, shows the year over year percentage change in payroll employment in each state, from greater than 3.1% in white to a decrease in employment in black… 

October YoY percentage change payroll employment by state

from the seasonally adjusted state labor force and unemployment data, we find that Nevada has the highest state unemployment rate at 9.3%, followed by Rhode Island with 9.2% and Michigan with 9.0%, while North Dakota continues to have the lowest jobless rate at 2.7%…only the District of Columbia saw its unemployment rate rise measurably, as it increased from 8.6% to 8.9%, whixh was more than likely mostly shutdown related…meanwhile Missouri saw it’s unemployment rate fall 0.4%, from  6.9% in September to  6.5% in October, as did South Carolina, where the unemployment rate fell from 7.9% to 7.5%…meanwhile, the unemployment rate for North Carolina fell 0.3%, from 8.3% to 8.0% , and five states saw their jobless rate fall by 0.2%: Nebraska, where the rate went from 4.1% to 3.9%, Minnesota, where the rate fell from 5.0% to 4.8%, Maine, where the jubless rate 6.7% was down from 6.9% in September, Delaware, where unemployment fell from 7.0% to 6.8%, and Georgia, where the September jobless rate of 8.3% gave way to the 8.1% rate in October…the map below, also taken from the full pdf report, indicates the unemployment rate in each of the 50 states by darkness of its shading, from those with an unemployment rate below 5.4% in white to those with an unemployment rate between 8.5% and 9.9% with the densest cross hatching..

October unemployment rate by state 

Manipulation of Employment Data is Possible but Not Easy

and since we’ve been talking unemployment rates, there was a story in the NY Post this week wherein a Census Bureau employee claimed to have falsified survey data for the monthly unemployment report during the run-up to to the 2012 election, which resulted in a precipitous decline in the September unemployment rate from 8.1% to 7.8%, and thereby helped Obama win the election…it shouldn’t be much of a surprise that this was picked up by the right wing media, including the likes of Rush Limbaugh, so since the allegation that an employment report was manipulated, and is still being manipulated, is in the news, maybe we should look at how feasible that would be… 

recall that the unemployment rate comes from a survey 60,000 households representing the employment status of 110,000 individuals per month…these surveys are conducted by the Census bureau for the labor department, and the BLS tells us that roughly 2,200 census employees from district offices interview those chosen to participate (based on a regional & demographic mix) first at their home, then three times later by phone before they are dropped out of the survey for a year…so that means the average Census employee is responsible for 50 responses to the survey each month…since there are more employed than unemployed, the opportunity for one employee to switch unemployed to employed is minimal; ie, with the labor force participation rate less than 65%, the average mix of those in the caseload for one census employee would include 29 employed, 3 unemployed, and 18 not in the labor force…thus, with each answer to the survey translating into a little over 22,000 in the final totals (based on a working age population around 245 million), it’s unlikely one rouge employee could reduce the number of unemployed in the final tally by more than 70,000…so with a margin of error of +/- 300,000 in the monthly change in the number unemployed and +/- 0.2% in the employment rate already assume for this survey, it would take at least 5 census employees falsifying every unemployed person they interview to even swing a change larger than the margin of error…

part of the subsequent media discussion after this article revealed that Census offices in the New York and Philadelphia regions were having trouble meeting their quotas of 90% successful interviews, and that in Philadelphia in particular, they were alleged to have filled the gap with fake interviews…we can almost imagine a supervisor under pressure to meet a quota telling employees do whatever you have to, even make up data, to cover his backside, and that a supervisor made such a statement was part of the allegation…but if that’s the case, it didn’t show up in the data; the unemployment rate in the Philadelphia region actually rose in September 2012…what we have to understand is that most census employees are career number crunchers; so we would not be surprised to find that some started work at the Census under Reagan or even earlier…so their politics is probably in a similar proportion to that of the general population…if it was known in the agency that some employees were fudging reports to make Obama look better, as was alleged, those opposed to him could much more easily switch employed to unemployed than vice-versa, simply because there are more employed in each caseload… 

BEA Misses September Inventory Increase; Implies +0.3% Revision to 3rd Quarter GDP

finally, we’ll take a quick look at a bit of old business…if you recall last week, we were trying to piece together a few missing pieces of September data that weren’t available to the BEA when they computed 3rd quarter GDP two weeks ago; this week we got another bit, when the Census released the Manufacturing and Trade Inventories and Sales report for September (pdf)…what we pointed out when the report on wholesale inventories was released last week was that the BEA, in estimating 3rd quarter GDP, added a a technical note (pdf) that indicated one of their assumptions was that wholesale and retail inventories other than motor vehicles would decrease by $7.6 billion in September, so when wholesale inventories already had shown a $2.0 billion increase last week and forecasts were for retail inventories to also increase, it looked like the BEA had seriously underestimated inventory buildup in September….this week all the results are in; total seasonally adjusted manufacturers’ and trade inventories at the end of September were at $1,679.1 billion, up 0.6% (±0.1) from August; manufacture’s inventories were at $634.0 billion, up from $631.3 billion, retailers inventories were at $538.8 billion, up from $534.1 billion, and as reported last week, September wholesale inventories were at $506.3 billion, up from $504.3 billion in August…thus total inventories were up by $9.4 billion, and only $3.3 billion of that was motor vehicles…with both the CPI and wholesale prices showing almost no inflation, there wont be much of a deflator, so it’s a good bet that J.P. Morgan, Barclays, and Macroeconomic Adviser economists are probably right, and that 3rd quarter GDP will be revised up by 0.3% or more as a result…but that only means that goods produced earlier in the year on still sitting on manufacturers, wholesalers, and retailers shelves, which will have to be drawn down before much more new goods can be produced…

(the above is my weekly commentary that accompanied my sunday morning links emailing, which in turn was mostly selected from my weekly blog post on the global glass onion…if you’d be interested in getting my weekly emailing of selected links that accompanies these commentaries, most coming from the aforementioned GGO posts, contact me…)