Historical Analysis (Unabridged Version)

Executive Summary

Over the last 15 years,

the state of Wisconsin had significantly lower unemployment rates than the nation.
Portage and Marathon Counties had similar unemployment rates; however, Wood County's unemployment rate was significantly higher than Portage and Marathon Counties' unemployment rates.
Portage and Marathon Counties had similar unemployment rates when compared with Wisconsin; however, Wood County had significantly higher unemployment rates than the State of Wisconsin.
Portage, Marathon and Wood Counties' unemployment rates were drastically lower than the national unemployment rate.
Portage and Wood Counties' retailers had similar confidence regarding recent changes in total sales; however; Marathon County retailers were significantly more confident in total sales than retailers in Portage or Wood County.
Portage and Wood Counties' retailers had similar confidence in recent changes in the level of store traffic; however, Marathon County retailers were significantly more confident in the level of store traffic than retailers in Portage or Wood County.
Marathon County retailers were better predictors of future total sales; however, Portage and Wood Counties' retailers were less accurate predicators of future total sales.
Marathon County retailers were better predictors of future store traffic; however, Portage and Wood Counties' retailers were less accurate predicators of future store traffic.
there was an inverse relationship between the central Wisconsin yearly change in the unemployment rate and central Wisconsin's retailer confidence in total sales.
there was an inverse relationship between the national half-year change in the unemployment rate and business confidence in national activity.
there was an inverse relationship between central Wisconsin's retailer confidence in expected store traffic and the central Wisconsin yearly change in the unemployment rate one quarter in the future.
there was an inverse relationship between central Wisconsin's business confidence in expected local activity and the central Wisconsin half-year change in the unemployment rate two quarters in the future.

Introduction

Our main goal is to further enrich the surrounding community's knowledge of the economic conditions in central Wisconsin over the past 15 years, from 1990 to 2004. The purpose of our research was to identify similarities and differences between the central Wisconsin counties' unemployment rates, and to identify possible relationships between retailer confidence indices. Additionally, we compared unemployment rates to business and retailer confidence indices in central Wisconsin to determine whether or not the confidence of businesses and retailers in the economy reflects changes in area unemployment rates.

Data

We used data collected by the Wisconsin Department of Workforce Development (DWD) for the civilian labor force statistics in Portage, Marathon and Wood Counties. These statistics include total workforce, total employed, total unemployed and the unemployment rate for each quarter over the last 15 years (1990-2004). This data is made available on a monthly basis; however, we used the last month of each quarter (i.e. March, June, September, December) for our analyses. Civilian labor force statistics can be obtained from the DWD website at http://worknet.wisconsin.gov/.

The business and retailer confidence survey data is collected by the Central Wisconsin Economic Research Bureau on a quarterly basis. Questions on the survey reflect changes in conditions of the business and retailer sectors in central Wisconsin. Businesses are asked to rate on a scale of 0 to 100 (in increments of 25) the following economic changes in their respective business:

Change in national activity compared the over the past six months
Change in local activity compared with the over the past six months
Expected change in national activity over the next six months
Expected change in local activity over the next six months
Expected change in industry activity over the next six months

A rating of 0, 25, 50, 75 or 100 respectively represents a substantially worse, moderately worse, the same, moderately better or substantially better change in the corresponding economic condition over the specified time period. Retailers are asked to rate in a similar fashion the following economic changes their retail outlet experiences:

Change in the level of total sales compared with the same quarter of the previous year
Change in the amount of store traffic compared with the same quarter of the previous year
Expected change in the level of sales in the next quarter compared with the same time a year ago
Expected change in the amount of store traffic in the next quarter compared with the same time a year ago

Analysis

Our research consists of two primary sets of statistical analysis. First, we examined unemployment rates in a comparative manner. These comparisons include:

-        Portage County vs. Marathon County
-        Portage County vs. Wood County
-        Marathon County vs. Wood County
-        Portage County vs. Central Wisconsin
-        Marathon County vs. Central Wisconsin
-        Wood County vs. Central Wisconsin
-        Portage County vs. State of Wisconsin
-        Marathon County vs. State of Wisconsin
-        Wood County vs. State of Wisconsin
-        Portage County vs. The United States
-        Marathon County vs. The United States
-        Wood County vs. The United States
-        State of Wisconsin vs. The United States

Second, we examined possible correlations regarding the following:

-        Retailer confidence in the level of total sales vs. a comparable change in the unemployment rate in central Wisconsin
-        Retailer confidence in the amount of store traffic vs. a comparable change in the unemployment rate in central Wisconsin
-        Business confidence in the level of national economic activity vs. a comparable change in the national unemployment rate
-        Business confidence in the level of local economic activity vs. a comparable change in the unemployment rate in Central Wisconsin
-        Retailer confidence in the level of expected sales vs. a comparable change in the unemployment rate in central Wisconsin
-        Retailer confidence in the amount of expected store traffic vs. a comparable change in the unemployment rate in central Wisconsin
-        Business confidence in the level of expected national economic activity vs. a comparable change in the national unemployment rate
-        Business confidence in the level of expected local economic activity vs. a comparable change in the unemployment rate in central Wisconsin

Methods

Population Mean Rates of Unemployment

The following formulas were found in the fourth edition of Introduction to Probability and Statistics by J. Susan Milton and Jesse C. Arnold. When testing for significances in differences of population mean rates of unemployment from area to area, we first tested for a difference in population variance. In all tests, the critical region is based on an alpha level of five percent. Based upon the result of the test for differences in variances we used the following test statistics for testing the population mean rate of unemployment:

If population variances were equal then we used the pooled variance given by

in order to use the pooled T test statistic which equals

If population variances were unequal then we used the Smith-Satterthwaite degrees of freedom given by

in order to use the unequal variance test statistic which equals

Paired Data between a Retailer Confidence Index and a Retailer Expected Confidence Index

Data was paired to make a valid comparison between what retailers thought would happen in the future quarter against what actually happened in either total sales or store traffic. In order to analyze this statistically, we used the paired T test statistic given by,

where D-bar is the average difference between the confidence index in the current quarter and the expected confidence index given in the previous quarter, and S_d is the standard deviation of all the differences.

Correlation between a Confidence Index and a Comparable Change in the Unemployment Rate

When testing for significances in linear regressions between two random variables, namely a confidence index and a change in unemployment rate over a similar time frame, we used the following test statistic:

We also double checked these tests, as a demonstration, with the test statistic for the Pearson correlation coefficient given by

In order to maintain a consistent comparison of random variables, alterations to the unemployment rate data were made. The confidence indices for each question have a reference period. On the retailer confidence survey, questions about changes in total sales and store traffic are compared either to the current quarter of the previous year or the next quarter of the previous year. For example, when a retailer gives a response of greater than 50 (less than 50) in the current quarter, that means total sales or store traffic have increased (decreased) compared with the same quarter of the previous year. Thus, a comparable change in the unemployment rate was determined by using the following formula

*100

When businesses are asked about recent changes in economic activity the comparison is over the last six months and future changes in economic activity are compared over the next six months.

Findings

Average Quarterly Unemployment Rates Over the Last 15 Years

Area	Unemployment Rate
Portage County	4.36%
Marathon County	4.31%
Wood County	4.87%
Central Wisconsin	4.47%
Wisconsin	4.34%
United States	5.60%

Table 1 shows the comparison of the average unemployment rate over the last 15 years between the State of Wisconsin and the nation, which are 4.34% and 5.60% respectively. The data indicate that the unemployment rate over the last 15 years was significantly lower for Wisconsin compared with the nation. On average, Wisconsin had a better workforce environment than many of the other states in the union. Specifically, this difference in the average quarterly unemployment rate over the last 15 years may be due to Wisconsin's strong work ethic and public education system compared to the average state.

Table 2 shows each of the central Wisconsin counties compared with a central Wisconsin aggregate (Portage, Marathon and Wood counties combined) unemployment rate. The comparison is between the average quarterly unemployment rate over the last 15 years for Portage, Marathon and Wood counties, which are 4.36%, 4.31%, and 4.87% respectively, and the central Wisconsin average quarterly unemployment rate of 4.47%. Clearly, the result that each county's unemployment rate does not significantly differ from the central Wisconsin average is expected; however, this shows that none of the counties exhibited extreme differences in unemployment rates over the last 15 years. From this, we can conclude that the employment situations in Portage, Marathon and Wood Counties were relatively stable.

Table 3 shows a pair wise comparison of each of the three central Wisconsin counties to one another. The two significant results are the comparisons between Portage and Wood Counties and Marathon and Wood Counties. The data indicate that Wood County's average unemployment rate over the last 15 years of 4.87% was significantly higher than Portage and Marathon Counties' unemployment rates of 4.36% and 4.31% respectively. However, the data also indicate that Portage and Marathon Counties' unemployment rates were statistically equivalent over the last 15 years. A possible explanation of this deals with how the municipalities are structured in Marathon and Portage Counties. In Marathon there is a large city, Wausau, adjacent to two smaller towns, Rothschild and Schofield. Similarly, Portage County has a city, Stevens Point, which is adjacent to two smaller towns, Whiting and Plover. On the contrary, Wood County has two smaller cities, Wisconsin Rapids and Marshfield, which are not adjacent. It seems likely that the geographic relationship of the municipal populations in Marathon and Portage Counties are more conducive to a healthy employment situation. In addition, Wood County's high dependence on paper manufacturing is likely to have played a role in this outcome.

Furthermore, in Table 4, the data demonstrate that Wood County is once again the odd one out. According to the data, Portage and Marathon Counties had similar unemployment rates as Wisconsin whereas Wood County's unemployment rate of 4.87% was significantly higher than the state's unemployment rate of 4.34%. In this case, the noted difference in Wood County's unemployment rate from the State of Wisconsin cannot be attributed solely to the layout of its municipalities. Education and health care services typically account for nearly a quarter of Wood County's employment creating less variety in the labor market. The lack of variety in the labor market may have lead to poor job opportunities for people who were looking for work in other sectors. Also, the reliance on a struggling paper manufacturing industry, along with the less than optimal geographic relationship of its municipal populations, could possibly be creating the atypical unemployment rates. In general, Portage and Marathon Counties had more stable employment situations with respect to Wisconsin than that of Wood County's situation.

Table 5 enforces the notion that, on average, central Wisconsin had better employment conditions than the nation. In all instances, when the average unemployment rate over the last 15 years for each of the central Wisconsin counties were compared with the nation they were significantly lower. This may be due to the same reasons mentioned above for Table 1.

Table 6 begins our examination of business and retailer confidence regarding the economy. We first looked at pair wise comparisons of retailers in each county in central Wisconsin to examine if their confidence in total sales, compared with the same quarter of the previous year, was significantly different. The data indicate retailers in Portage and Wood Counties experienced similar changes in total sales, on average, over the last 15 years. On the other hand, when Portage or Wood County were compared with Marathon County the difference in confidence was significant. It should be noted that Marathon County experiences a much higher level of economic activity than either Portage or Wood County (evidence of this comes from the Wisconsin Department of Revenue county sales tax distribution data)-this could be the cause of the difference in confidence. When data is collected on Marathon County, it is considered to be a metropolitan statistical area. This means that Wisconsin considers this area to be of greater concern than that of Wood or Portage Counties due to its strong economic influence on the surrounding area.

Table 7 is similar to Table 6 except with regard to store traffic. Retailers in Portage and Wood Counties had similar confidence over the last 15 years regarding store traffic compared with the same quarter of the previous year. However, when Portage or Wood County was compared with Marathon County there was a significant difference in confidence levels. Once again, this may be the result of the reason stated above in our findings from Table 6. Furthermore, the fact that Marathon County is a metropolitan statistical area may be contributing to the results in Table 7.

Table 8 shows a comparison of the confidence indices between the level of retailer confidence in total sales and the retailer confidence in expected sales. This examination should demonstrate whether or not the retailers in the central Wisconsin counties were good predictors of the future over the last 15 years. The data indicate that retailers in Portage and Wood Counties were less accurate in terms of forecasting future changes in total sales than those in Marathon County. This means that the survey is more useful in predicting what will occur in the next quarter given the expected confidence in total sales by retailers in Marathon County. The opinions of retailers in Portage and Wood Counties were determined to be less helpful in predicting what total sales will be like in the future quarter.

Table 9 shows a similar comparison to that of Table 8 except with regard to store traffic. Retailers in Marathon County were good predictors of future store traffic. However, retailers in Portage and Wood Counties were not good predictors of future store traffic. These results are identical to the results of Table 8 except that it is with regard to store traffic. In both Table 8 and Table 9, the fact that Marathon retailers were better predictors of the future may be due to greater economic stability. That is to say that growth may have been more constant in Marathon County and more variable (or less predicable) in Portage and Wood Counties. Economic stability in Marathon County may be attributable to the greater amount of economic activity as evidenced by sales tax distributions and a more robust employment situation.

Table 10 changes our focus to linear regressions between changes in unemployment rates and confidence indices. The yearly change in the unemployment rate (from quarter to quarter) was multiplied by 100 to be comparable to the various confidence indices. The first comparison was between the central Wisconsin yearly change in the unemployment rate and retailer confidence in total sales. The linear regression on these two variables was found to be significant. This means that if the central Wisconsin unemployment rate decreases, retailers have a high confidence in total sales. Likewise, if the central Wisconsin unemployment rate increases, retailers have a low confidence in total sales. Support of this comes from the fact that a higher unemployment rate results in a decrease in the average amount of disposable income per consumer. Thus, one would expect that less disposable income directly influences the level of sales in an economy.

Table 11 shows the linear regression between the central Wisconsin yearly change in the unemployment rate and retailer confidence in store traffic. The regression was determined not to be significant using the typical standards of statistics. However, if we were to treat this comparison more leniently then we might find the correlation between unemployment rate changes and retailer confidence in store traffic to be significant. If that were the case, then an increase in the central Wisconsin unemployment rate results in lower store traffic. The analysis of this relationship is directly related to what was discussed above about Table 10. A decrease in disposable income results in consumers frequenting retailer outlets less often in order to make purchases.

Table 12 shows the linear regression between the national half-year change in the unemployment rate and business confidence in national activity. The data indicate that this regression is significant by a large margin. This means that the business confidence in recent changes in national economic activity correlates well with similar changes in the national unemployment rate.

Table 13 shows the linear regression between the central Wisconsin half-year change in the unemployment rate and business confidence in local activity. The regression was not significant. This means there is no apparent relationship between the central Wisconsin half-year change in the unemployment rate and business confidence in local activity. A possible explanation of the results in Table 12 and 13 may be due to the fact that it is easier to predict economic activity on a larger scale than a smaller scale. For example, a business that provides services to hundreds or thousands of customers nationally can gauge economic activity much better compared with the few customers they service on a local level. Another factor contributing to this correlation may be due to the stability of national activity in comparison to the variability in local activity.

Table 14 shows the linear regression between central Wisconsin's retailer confidence in expected total sales and the central Wisconsin yearly change in the unemployment rate one quarter in the future. The regression was not significant. This implies that a change in the unemployment rate is not necessarily dependent on how confident retailers are regarding expected total sales. One might expect an inverse relationship between these two variables where an increase in retailer confidence implies a lower unemployment rate; however, the lack of a significant relationship may be due to a lag effect. In other words, increased confidence in any part of the economy may not be immediately noticed through an indicator such as the unemployment rate until some period after that increase took place.

Table 15 shows the linear regression between central Wisconsin's retailer confidence in expected store traffic and the central Wisconsin yearly change in the unemployment rate one quarter in the future. The regression was significant. This implies that a change in the level of retailer confidence regarding expected store traffic corresponds with a change in the unemployment rate. Although correlation does imply that one variable relates to another, it does not imply causality. However, it is possible that if entrepreneurs expect to perform well in the future, more likely than not economic conditions will improve due their increased activity. This is an illustration of self-fulfilling prophecy.

Table 16 shows the linear regression between business confidence in expected national activity and the national half-year change in the unemployment rate two quarters in the future. The regression was not significant. This means there is no apparent relationship between business confidence in expected national activity and the national unemployment rate. This lack of association is most likely due to the relatively small nature of our survey. Although businesses in the local area have a large impact on the central Wisconsin economy, their impact nationally is quite small. Thus, in order to develop a meaningful correlation between business confidence and the national unemployment rate, the scope of our survey would need to encompass businesses throughout the nation. Also, lag times may be contributing to our result in table 15 as discussed in table 14.

Table 17 shows the linear regression between business confidence in expected local activity and the central Wisconsin half-year change in the unemployment rate two quarters in the future. The data indicate that there is a correlation between these two variables. Clearly, it seems reasonable that an increase in business confidence in local activity, taken from a survey of central Wisconsin businesses, would correspond to a lower unemployment rate in central Wisconsin. This reinforces the notion that as businesses in an area become more confident in local activity the area experiences a better employment situation. Lag times may not be contributing to our results as significantly as in the other tables because confidence in local activity may have a more direct relationship with the actual level of economic activity in central Wisconsin than retailer confidence in total sales as examined in table 14 or business confidence in national activity as examined in table 15. In order for retailers to increase the number of workers at their outlets, they must first experience increases in revenue. On the other hand, if a business expects an increase in activity then they must accommodate for such expectations. Therefore, we see a more immediate impact on the unemployment rate and thus lag times are not as significant with regard to confidence in expected local activity.

Sources

Central Wisconsin Economic Research Bureau Business and Retailer Confidence data can be found in part on the web at http://www.uwsp.edu/business/cwerb/, or by calling 715-346-3774 for further information.

Department of Workforce Development Local Area Unemployment Statistics can be located at http://worknet.wisconsin.gov/, by clicking Data Analyst on the left, Data Tables in the submenu, and then Query for LAUS.

Department of Workforce Development Industry Data for Wisconsin counties can be located at http://worknet.wisconsin.gov/, by clicking Data Analyst on the left, Data Tables in the submenu, and then Query for Non-Metro County Industry Employment Estimates.

Wisconsin Department of Revenue County Sales Tax Distributions can be located at http://www.dor.state.wi.us/report/c.html.

TABLE 1