LEIPLINE Newsletter - November 2009

The Jacksonville Economy in the Third Quarter of 2009


This is the twelfth installment of the LEIPLINE. Our focus each quarter is generally on the four variables for which we collect data and the implications of those data for the Jacksonville MSA overall. Although we provide two articles this quarter, the shift in the real GDP numbers suggests that the recession MAY be waning, implying the need for some commentary beyond just the local data that we collect. We emphasized MAY due to a lack of confidence.


The third quarter national numbers reflected a sizable uptick in most subcomponents of real GDP including personal consumption expenditures (PCE), residential investment, and government purchases of goods and services; even after initial revisions. The increase in residential investment is particularly positive, but non-residential investment was very weak and state and local government spending declined marginally. The PCE increase was at least half due to the major fiscal stimuli associated with "cash for clunkers" and first home-buyer programs, but given the consistent declines in PCE excepting 2009.1 going back into early 2008, the non-stimulus increases, however small, were welcomed. Whether this is the beginning of the end of the recession will dramatically depend on the Christmas season, with weakness in holiday shopping perhaps signaling a "W" pattern, and a strong shopping push meaning that 2010 will reflect recovery.


Our second article this month relates to the statistical relationship between retail sales and unemployment. We developed a forecasting model that suggests that there is a definitive casual relationship from unemployment to retail sales, but not the other way around. The model also forecasted unemployment for next month well above where it was in October, but retail sales stronger in December 2009 than in December 2008.

The Macro Economy

In a climate with so many stimuli; it is difficult to distinguish private sector improvement from government driven spending. The business cycle reflects the aggregate changes in real GDP in total, but it is made up of its component parts. The Obama administration's plan has consistently been to make sure that low private sector spending is replaced by mostly federal government expenditures. Economic theory going back to John Maynard Keynes argues that the multiplier effects associated with enhanced spending are indistinguishable across spending groups (e.g., consumers, businesses, or government). However, public works programs produce infrastructure development that does not necessarily create sustainable private sector growth; and stimuli, particularly through bailouts and niche-targeted programs, beget more demands for bailouts instead of efficiency gains due to the need to maintain operations in an adverse climate.


The financial markets clearly reflect this latter perspective as the bailed out firms' stocks and bonds have prospered despite reduced demands for products and less business-as-usual activities. Many of these corporations are employing stimulus money to maintain revenues while their costs are slashed through layoffs and retrenchment. The combination of these two factors has produced enhanced profitability, but unsustainably so without recovery in sales. Although the pattern of consumers employing their homes as ATMs (as one economist phrased it recently) is a thing of the past, without recovery in consumer confidence and spending, true recovery in employment and economic growth is a long way off.


The inflation news confirms this reality. Excepting June 2009, the inflation rate through October suggests an overall inflation rate for the year of around 2.0%, which is still depressed relative to historical norms. October's numbers were driven almost exclusively by jumps in new and used car prices as "cash for clunkers" disappeared. Until consumers regain their drive to spend, price increases will be unlikely and cost cutting and deep discounts, while not expressly reflected in CPI numbers, will persist. Substantial inflation would almost necessarily have to derive from the massive fiscal and monetary stimuli of the last year, unless as the evidence supports, the greater available money is not being spent or lent.


The national unemployment rate regained its climb in the third quarter virtually universally. Double digit unemployment, which has not been seen in the U.S. since 1982, became widespread. Breaking down the unemployment numbers is both interesting and foreboding. Because the industries that have been affected most (construction, financial services, and real estate) are dominated by male employees, recent articles have referred to current conditions as a "man-cession." Because many of the industries employ more skilled workers, the impact has been greater on college-educated workers than in past recessions. Since recent college graduates have the least experience in the workforce, their prospects are worse than during most historical downturns. For decades, western European countries have generated very large and stubborn quantities of what economists call frictionally unemployed workers (those new to the labor force or in between jobs) - particularly among the young. There is serious concern that such an underclass of skilled workers may arise in the U.S. as a result of this recession, and that unemployment rates may not return to the 4-6% level for a very long time. Regardless, we are not aware of any economists who expect employment to recover very much or quickly in the next year.


Interest rates have remained at extremely low levels, with declines in long term rates, particularly mortgage rates, at levels not seen since the 1950s. Amazingly, from a long term perspective, the treasury can continue to sell intermediate term securities to yield less than 1% interest due to all of the uncertainty of the economies near-term strength. If the economy was regaining its typical growth, given the huge surplus of loan able funds available in the credit markets, interest rates should rise. However, since they are not, it is hard to conclude that the economy can be forecasted to grow at third quarter levels in either the fourth quarter, or early next year

The trade sector regained some degree of growth in the third quarter, but largely reflecting increased demand and supply from China. The Chinese Yuan's value is still too low to reduce the Chinese trade surplus, and its rise would be instrumental in permitting recovery elsewhere in the world. However, trade values for both imports and exports are well below magnitudes from 2007 and before, so the general weakness in world economies persists, despite the more positive news in 2009.3.

What Does It All Mean?

While the third quarter GDP values are encouraging, caution is warranted relative to concluding that the recession is over. There may very well have been a sizable structural shift in the way American's and western Europeans spend and save. Diminished spending will and has generated reduced demands for employees, which in turn has exacerbated the declines in business investment and inventory spending. John Maynard Keynes proposed a "paradox of thrift," whereby diminished spending that leads to increased savings throughout the economy, leads to reduced growth and less output and employment for the entire economy. It is hard to deny this outcome over the last nearly two years. The third quarter of 2009 revealed some semblance of increased spending, but until this becomes sustainable, and it promotes reemployment and business investment to meet anticipated future demand, the economies of the world will grow very slowly from their diminished levels. Hopefully, the fourth quarter of 2009 will begin this process, but it appears like the recovery will be very slow and quite bumpy.

The Local Data

Since many of the data change for at least one month after we first report them, we have decided to wait until at least the middle of the following quarter to report each quarter's implications. Since the local CPI is the most significant variable that we analyze, we will start with it.

The 2009.3 Jacksonville CPI

Prices in the Jacksonville MSA through September were down for five of the nine quarters. However, October joined June and July with sizable increases. The annualized inflation rate for the year is still below 1%, reflecting the weakness that has now existed for over eighteen months locally. There were some interesting specific price movements during the third quarter. Cash for Clunkers stimulated major price jumps for new cars during the program which were reversed when the program ended. During the same period, used car prices fell as demand for these vehicles fell. In addition, airfares rose dramatically during the quarter as airlines attempted to enhance revenues during the summer travel period. The housing market made some gains in sales one month, accompanied by declining prices; then sales fell off as prices rose. October registered a sizable increase in prices overall, but almost exclusively, like at the national level, from automobile prices increases.


The outlook for the fourth quarter of 2009 relative to inflation is likely to continue small positive inflation rates. If last weeks "Black Friday" is any indication, the retail sector will produce lower prices despite stronger than expected demand. Although the phenomenon has grown dramatically over the last decade, last Friday seemed to generate larger than normal early morning shopping to take advantage of the deep discounts. If consumers spent a greater than average percentage of their holiday shopping dollars last Friday, in a weak economy overall, then the rest of the shopping season will be poor and further discounting will be necessary in an attempt to clear shelves. For several months, consumers have stayed away from big ticket items. They may have been saving up for holiday purchases, but this has left larger than normal inventories of these goods, which will further dampen price increases. We expect the year to reveal inflation associated with the CPI of less than one percent locally and nationally; which is better than the deflation of 2008, but not much.


Unemployment rates in Jacksonville during 2009.3 continued their rise to above the 10% by September, one month ahead of the national experience. Jacksonville's rates have now been ahead of the national numbers for the last three months counting October. LEIP only seasonally adjusts the unemployment rate (we do not calculate it), but doing so provides valuable short term insight into the changes from month to month. The unadjusted rates are currently approaching 11%. However, the seasonally adjusted values are still in the low tens. While the unadjusted rate stayed the same from July to August, the seasonally adjusted rate jumped close to 0.5%. The implication was that when the school year began again in August, contrary to most years, the number of workers unemployed did not fall as much as usually occurs (due to students leaving the work force), and unemployment rose dramatically.


The outlook for the fourth quarter is likely to suggest a further weakening of the employment numbers. Retailers have not hired the same number of seasonal workers as is typical. This combined with normal end of the year layoffs and the influx of December college graduates suggests that 11% plus seasonal unemployment is likely locally by the end of the year. Our second article included this quarter not only implies 11%, but closer to 12%. As addressed above, we remain very concerned that a structural shift in unemployment towards higher norms is in our future, and as an educator and future college graduates, this is a very disconcerting potential.

Leading Indicator (LEI)

The LEIP leading indicator is not forecasting a rosy winter. In February, 2009, the national LEI and the LEIP LEI were virtually identical in value. However, while the national numbers have climbed by over five percentage points, the local indicator was actually lower in October than it was in February. Since the LEIP LEI is a combination of five national indicators and six local ones, the blame for the local weakness has to be placed on the local components. Local stocks have taken a beating so far this year, particularly those in real estate and financial services. Although they recovered somewhat by the October numbers, new claims for unemployment insurance are thousands monthly above the norm for prior years. Building permits now stand at one-quarter of what they were over the 2001-07 time period, and employment advertizing has been lower for six of the last nine months. The only two relatively positive components are consumer confidence and interest rate spreads, but both of these have been just slightly positive for the year. The outlook for sales, revenues, and local growth still remain quite weak for the foreseeable future.

Stock Price Index

While local stocks recovered some during the third quarter of 2009 along with the rally in the DOW, October was another bad month for local stocks. The local headquarters stocks are now 1,800 below the DOW, while the local presence stocks are now 3,600 below the DOW. For the last two quarters we implied that local stocks seem to have been hit harder than the DOW companies due to the substantial presence of financial companies in Jacksonville, and the third quarter numbers reveal that we were correct. However, construction and real estate lags have now also contributed significantly to the weakness. This does not bode well for the speed of the local recovery in employment, real estate (especially commercial), and growth.  

The Bottom Line, Locally

Overall, 2009.3 was the sixth consecutive weak quarter for Jacksonville and the nation in regards to the data LEIP collects or has access to in a consistent format. Retail sales continued to fall, unemployment rose to values well above optimal levels, and inflation remained muted, indicative of continued weakness. The LEI was somewhat higher in the third quarter, but began to fall again in October. The outlook for Jacksonville continues to be weak for the latter stages of 2009.


The media have concluded at the national level that the recession is over based on the third quarter GDP numbers. The NBER, the group responsible for actually signaling the beginning and end of recessions, has legitimately remained silent. However, whereas Jacksonville developed a reputation as recession-proof in the past due to population and real estate growth, neither of these important drivers is working toward recovery any longer. Jacksonville's recovery will have to be driven by consumers regaining their preferences and confidence to spend and by local businesses gearing up for additional inventory to meet future demand. Neither of these impulses is evident in the current data, consequently, our outlook is not as strong as we would like. Both Florida and Jacksonville need to develop new strategies to promote growth because the old stimuli are unlikely to return for a long time, if at all. 

Forecasting Duval County Retail Sales based on Unemployment: What to Expect for the Rest of 2009

Retail Sales is probably the most revealing indicator of the health of a localized economy. Since consumer spending is typically between 65% and 75% of total GDP, and retail sales measure consumer spending, retails sales tend to drive economic growth. The research in this paper focuses on Duval County, Florida retail sales for the period from January 1998 until February 2009 in order to forecast retail sales for the rest of 2009.


One of the fundamental determinants of consumer spending is employment, since income drives the largest share of consumption (with wealth driving the rest), and income derives for wage or salary employment primarily. Unemployment rates are an effective proxy for the employment sector, and available through the LEIP program, so they were employed in time-series regression analysis. However, unemployment effects on consumption tend to precede the impact on retail sales, so the methodology employed treats unemployment as a lagged variable. To identify a formula that precisely describes the relationship between lagged values of unemployment and retail sales requires the use of a Vector Autoregressive model (VAR). This model captures lags for each variable and reveals the effects of the lags on the dependent variable (retail sales).


Proper time series analysis requires that all variables be stationary. Stationary variables are needed to assure that the model produces reliable forecasts. After employing augmented Dickey-Fuller tests for stationary, the result implied that both variables are stationary.


In addition, testing for Autocorrelation and partial autocorrelation has to be accomplished for each variable separately. After determining the number of significant lags that are suggested to be included in the regression, the statistician must determine the nature of the residuals. These residuals can be obtained from the respective unemployment and sales VAR models and run against Bartlett's white noise test. The Residuals should lie in the middle of the distribution, which they did.

The next step is to identify causality, which mandates a Granger Causality Test. The Granger Causality test assumes that the information relevant to the predication of the respective variables, Unemployment and Retail Sales, is contained solely in the time series data of these variables.


The Hypotheses are as follow:

For Sales:

H0: Unemployment does not granger cause Sales

H1: Unemployment does granger cause Sales

The Granger causality Wald tests are revealed in the table below, where ls is the log of sales and lu is the log of unemployment.



 Equation  Excluded chi2   df Prob > chi2 
ls lu 115.4 22 0.000
ls ALL 115.4 22 0.000
lu Is 37.114 22 0.023
lu ALL 37.114 22 0.023



In this case we can reject the Null Hypothesis for Sales at the 1% level, since the p-value is zero. This means that Unemployment granger causes Retail Sales. Retail Sales do not granger cause Unemployment. By clarifying causality first, the VAR regression can be minimized because the only significant part of the regression is the part where unemployment defines Retail Sales.

The final output can be used to forecast and the following results and graph are shown below:






lsls Sales lslu Unemployment  
Mar-09 21.887972 3204981003 2.2616223 9.60%
Apr-09 21.933771 3355179137 2.2573535 9.56%
May-09 21.884456 3193732077 2.3210864 10.19%
Jun-09 22.141862 4131316672 2.4279194 11.34%
Jul-09 22.086865 3910241603 2.3915117 10.93%
Aug-09 21.858662 3112406319 2.3526528 10.51%
Sep-09 21.976854 3502889373 2.3794212 10.80%
Oct-09 21.880942 3182528998 2.4286709 11.34%
Nov-09 21.837785 3048102187 2.4680964 11.80%
Dec-09 22.120041 4042143672 2.4759156 11.89%



The data have to be converted from logarithms to identify meaningful results (i.e., the second and fourth columns after the dates). The final results show an estimated increase in unemployment to 11.89% in December 2009 and Retail Sales having two peaks, in June 2009 and December 2009, with significant drop offs in sales in between the peaks. Given the seasonality in retail sales, the peak in December is not surprising, but the June peak is less understood. However, the ultimate implications are that unemployment (currently around 9.6% in the Jacksonville MSA) will go over two percentage points higher by the end of the year, and retail sales are likely to fall from June levels until they spike again at the end of the Christmas season, but significantly above the end of 2008 level of $2.91 billion.