Manufacturing Employment in North Carolina, 1995-2000

2232015

Manufacturing Employment in North Carolina 1995-2000

Econometrics (ECN 422-003)

aed1718UNC Wilmington

Executive Summary

Introduction

After analysis of the manufacturing jobs in previous years 1996 through 1999 it is evident that the mean of manufacturing jobs that will exist in 2000 will follow a downward sloping trend Through creating of the variable EmpManf2000 a prediction of jobs that will be filled in the manufacturing sector was predicted using data in previous years County manufacturing employment in North Carolina in year 2000 (EmpManf2000) totaled 799984 summed across all counties Through this variable calculation it is estimated that 8163 manufacturing jobs in North Carolina will be employed in the year 2000 The maximum was noted at 56229 jobs and the minimum was noted at a low 79 jobs the median in 2000 was estimated at 4688 jobs which is indeed less than the mean This difference indicates that the distribution of county manufacturing employment numbers is skewed to the right with more counties having employment numbers less than the mean and fewer counties having employment numbers greater than the mean

Literature Review

Through a close examination of the general employment of North Carolinians within the Manufacturing Industry this analysis aims to further understand employment trends forecast future employment and suggest troubleshooting of possible employment issues The truth is simple manufacturing has been a part of North Carolinarsquos economies for decades However over the past 20 years the job market of Manufacturing has been dwindling It is not uncommon to hear an array of statements similar to ldquoNorth Carolina is no longer a manufacturing staterdquo (The Herald-Sun) Although the importance of the sector there is little doubt of the subtle decline of employment North Carolinarsquos economy is crucially intertwined with manufacturing Manufacturing has been a path to development cause of economic growth and most importantly creates jobs that are held central to the economy

It has been the tactical success of rich nations over the last several hundred years to create a high-quality manufacturing sector in order to develop national wealth and power as Erik Reinert shows in his book ldquoHow Rich Countries Got Richhellipand Why Poor Countries Stay Poor The growth of manufacturing machinery output and technological improvements in that machinery are the main drivers of economic growth No machinery industries no sustained economic growth Most jobs depend on manufacturing -- and reviving the sector could provide tens of millions of new jobs which ultimately played a factor in the eradication of the Great Recession

Along with smart tax policy North Carolinas manufacturing future needs smart workers who are constantly gaining innovative knowledge learning new skills and sharpening attitudes (The Herald-Sun) Sufficient physical infrastructure to support growth predictable state and local operating conditions a fair and balanced legal climate and reliable competitively priced energy are all factors that could adhere to the lack of job in the past decade or so ldquoWithout a robust revival in the manufacturing sector we can kiss our status as a great economic power goodbyerdquo (Belton)

Going back to 1992 the manufacturing activity had reportedly continues to climb (Belton) It was stated in USA Today that ldquoThe NAPM index is based on a monthly survey of purchasing executives at 300 industrial companies A reading above 445 indicates the economy is growing The index has been above that since May A reading above 50 indicates manufacturing activity is increasingrdquo This statistic had been an inkling that the industry would start to grow

In 1998 it was projected that manufacturing would take a new approach with new-product development efforts to boost the sector To combat ever-shrinking product life cycles the Deloitte amp Touche report states many companies are teaming with customers to design new products rather than merely push products into the market(Sheridan) However such hope was quickly dismissed through the progress of the next 8 years

A potential factor regarding the downward sloping relationship of manufacturing jobs could be related to a decline in wages North Carolina was ranked ldquoat or near the bottomrdquo of a variety of statistical measures of economic health including the ranks of state manufacturing wages (Kniesner) in 2001 Even in despite of the boom of the last few decades many Southern states have made little or no progress in raising income levels primarily because their economies are still dominated by low-wage industries Thomas Kniesner a professor at the University of Chapel Hill says wages are low because many ldquoworkers are unskilled and the region has major industries such as textiles that cannot command high prices for their productsrdquo In recent news Government officials have made recent attempts to assist in job creation and promotion of manufacturing Governor Pat McCrory in 2014 is planning to install 228 jobs by 2018 in the manufacturing department of GKN Driveline in Catawba County (Plus Media Solutions) Though there is scarce literary information on the progress and recession within the manufacturing sector there is plenty of empirical data that can be studied In order to fully understand the projected data from the actual results an analysis will be conducted to truly understand the trends the manufacturing faction of North Carolina is undergoing

Data

The following data was obtained from the LINC database system (httplincstatencus) administered by the State of North Carolina This data contains information on all North Carolina counties from the plain coastal and mountain regions The data collected covered information from 2000 taken yearly A deviation in the data occurs with the inclusion of yearly employment in manufacturing jobs from1996 to 2000 There are 101 observations covered in the data Variable definitions and sources (the LINC database includes data from other government agencies and databases) for each variable are listed in Table 1

Methodology

Utilization of a testing hypothesis about two means was used with a PROC TEST command Using a one-same t-test we determined whether New Hanover Countyrsquos value of EmpManf2000 was statistically significant from the mean value of EmpManf2000 for all North Carolina counties together

Utilization of an independent samples t-test was used to determine whether average NC manufacturing employment was larger in urban counties or rural counties

Utilization of a dependent samples t-test was used to determine whether a change occurred in NC manufacturing employment change from 1996 to 2000

To determine if the EmpManf has changed from 2000 to 2016 we must calculate the required sample

size required to estimate the mean EmpManf for 2016 To calculate the sample size formula n=( t ∙ sE )

2

is used Use ldquosrdquo (standard deviation from PROC MEANS was used) E= +- 5000 workers Calculations of t are based on the 95 confidence level

Utilization of a cross tabulationcontingency table (for all NC counties together) for each of the following combinations of categorical variables ManfCategory amp GeoRegion and ManfCategory amp UrbpopGTrurpop Additionally a (Pearson) Chi-square Test of the null hypothesis was applied to test if the categorical variables do not affect one another

The predicted value of variance of EmpManf2000 is 70000000 (H0) A Chi-squared test is used to determine if the predicted variance is statistically different from the PROC MEANS table value for variance The null hypothesis used is as follows H0 σ2=70000000 H1 σ2ne 70000000F-test is conducted in SAS to determine if the variation in EmpManf2000 is greater than in urban counties than in rural counties

Utilization of a correlation matrix was used to for the measurement variables The Pearson Correlation coefficient was used to see if the two variables move together in the same direction (positive r) or the two variables move in opposite directions (negative r)

An initial regression was run on the data collected The model reading EmpManf2000 = β0 + β1LandArea + β2SchoolSpendPP + β3PropTaxRate + β4PavedMilesArea + β5PavedMilesAreaSq + β6LNMuniH2OArea + β7LNDocsPer1000 + β8MtnRegion + β9CstRegion + u This regression was run to see which X variables affected the corresponding Y variable

Results

Table 2 presents descriptive statistics for the numerical measurement variables used in the study These results can be viewed in Table Section of this report Calling attention to the EmpManf2000 variable the maximum was 56229 which was on course with the slight descent that started in 1996 As for the minimum calculated at 79 is quite close to the other additional years The sum of the data followed the downward trend as well which in 1999 was 818263 and then in 2000 was 799984 The median as well was on point with the downward sloping relationship The variable EmpManf2000 has a skewness calculation of +256 which is greater than 0 This number describes that the distribution of this variable is skewed to the right with more counties having employment numbers less than the mean and fewer counties having employment numbers greater than the mean The kurtosis calculation appears at 732 which is greater than 3 suggesting the distribution of data has more of a peak than a normal bell-shaped distribution

Additionally the trend in mean of EmpManf from 1996 to 2000 at first seemed to have a downward sloping trend However after looking closer the data spiked in 1998 and then continued to descend With regards to the ldquonrdquo or sample size being equal to less than 100 suggests missing values for that specific variable The variables with an ldquonrdquo value equal to 100 suggest all categories counties in North Carolina were present

Figure 1 simulates a horizontal frequency distribution for each categorical variable (ordinal nominal and character) of the coast mountain and plain regions of North Carolina that can be seen in the Figure section of this report The frequency of the GeoRegion variables and the breakdown of manufacturing jobs are the highest in the plain regions of North Carolina at 60 The coast and mountain areas are relatively low lying at 17 for the coast and 23 for the mountains The second part of Figure 1 explains the Category of Manufacturing jobs and the corresponding frequency The results can also be viewed in Figure 1

To determine whether New Hanover Countyrsquos value of EmpManf2000 was statistically significant from the mean value of EmpManf2000 for all North Carolina counties together a test was run of the corresponding data set The data concluded that at a confidence level of 95 the data is statistically significant

Additionally an independent sample t-test to determine whether the mean value of EmpManf2000 for urban counties (UrbpopGTrurpop = 1) is significantly different from the mean value of EmpManf2000 for rural counties (UrbpopGTrurpop = 0) using a PROC TEST command Conclusively the two values are statistically significant from 0 with a 95 confidence level This states the average NC manufacturing employment was larger in rural counties than urban counties

Utilization of a dependent samples t-test was used to determine whether a change occurred in NC manufacturing employment change from 1996 to 2000 using a PROC TEST command The data revealed that a change in employment did occur from 1996 to 2000 with a confidence level of 95

The sample size of counties required to predict the EmpManf next year at the 95 confidence level is 18 (1737) with an allowable error of +- 5000 workers The sample size of counties required to determine if EmpManf has changed from 2000 to 2016 is also 18 (1737)

After running a PROC FREQ command to produce a cross tabulationcontingency table (for all NC counties together) for each of the following combinations of categorical variables ManfCategory and GeoRegion and ManfCategory and UrbpopGTrurpop For each Contingency Table SAS was used to conduct a (Pearson) Chi-square Test of the null hypothesis to see if the categorical variables do not affect one another Looking at the data the Chi-Square value was estimated that geographic region and urban population do not share a relationship

A Chi-squared test is used to determine if the predicted variance is statistically different from the Proc Means table value for variance The null hypothesis used is as follows H0 σ2=70000000 H1 σ2ne 70000000 The proc means table value for variance EmpManf2000 is 113070628 At the 95 CL χ2

critical =12671 χ2test=60051 In this case we reject H0 and accept H1 The predicted variance is statistically

different

F-test is conducted in SAS to determine if the variation in EmpManf2000 is greater than in urban counties than in rural counties The F-test concluded that Ftest 989 gt Fcritical 175 Showing that EmpManf2000 is greater in urban counties

Results corresponding to the PROC CORR command are to be found in Table 4 Of the many observations that can be made based on the results of the correlation analysis the following are to be noted counties with larger populations have more manufacturing jobs per 10000 people The number of manufacturing jobs per 10000 people in a county appears to affect the county unemployment rate Also the number of manufacturing jobs per 10000 people in a county appear to affect the percent in poverty

Table 4 presents the results of the initial regression analysis The regression results indicate that relationship that exists between X and Y and determines if it is linear or non-linear In this model the Y variable or dependent variable was EmpManf2000 With the X variables or independent variables consisting of LandArea SchoolSpendPP PropTaxRate PavedMilesArea PavedMilesAreaSq LNMuniH20Area LNDocsPer1000 MtnRegion and CstRegion The following results were calculated from the initial regression analysis

The F-value (or F-test number) of this regression was placed at 1922 This suggests that the regression as a whole is statistically significant The p-value of the equation was lt0001 which demonstrates the possible error of this equation to the data Clearly the model is well fitted to the data represented The Adjusted R-square value was calculated at 06283 Since this equation has more than 2 X-variables we utilized the Adjusted R-Square to tell us the percent variation in this case ~63 in the Y data that is explained by this regression model

The SER for the regression is 648319396 The SER in this case is quite high This implies that the model has a larger spread which increases the standard of error of the forecast This does not necessarily suggest the model is incompetent It just explains the vast variation in the data values

The parameter estimates explain the estimate values of the ldquobrdquos (small betas) in the regression equation The parameter estimates explain with ever 1 unit increase in the listed x-variable there will be a calculated increase the parameter estimate on the regression result in the corresponding y-variable When a value in the parameter estimates is negative this suggest a negative relationship between that particular x-variable and the corresponding y-variable PavedMilesArea and CstRegion are the only x-variables that have a negative slope The other values can be viewed in Table 5 in the Tables section

T-values are another commonly observed regression result These values are the t-test numbers for the Brsquos (big betas) the actual parameter not estimates in the equation They also correlate to the null and alternative hypotheses The following X variables appear to strongly affect Y LandArea SchoolSpendPP PropTaxRate PavedMilesArea PavedMilesAreaSq and LNMuniH20Area

The estimated model equation readsEmpManf2000 = -20334 + 957630LandArea + 2703592SchoolSpendPP + 320837075PropTaxRate -21657PavedMilesArea + 961560374PavedMilesAreaSq + 270670158LNMuniH2OArea + 29303640LNDocsPer1000 + 273682077MtnRegion ndash 258646822CstRegion + u

Discussion and Conclusion

Cited SourcesBelton Beth Manufacturing Continues To Climb USA Today 1992 MONEY 1B Web 20 Feb 2015

Kneisner Thomas LOW WAGES PERSIST IN NORTH CAROLINA The New York Times 1981 Section D Page 1 Web 17 Feb 2015

Plus Media Solutions Washington NC Gov Mccrory GKN Driveline To Create 228 Jobs In Catawba County 2015 2 Web 21 Feb 2015

SHERIDAN John Companies Emphasize Product Innovation Industry Week (1998) 13 Print

The Herald-Sun NC Still Manufacturing State And Thats A Great Thing 2015 1 Web 22 Feb 2015

Government Report or Dataset

Tables

Table 1 Variable Definitions

Variable Description Measurement UnitsCntyName North Carolina county names CategoricalPopCens Population of NC at time of Census NumericalGeoRegion Location of county by geographical region CategoricalLandArea County land area Numerical square milesUnempRate The average annual number of unemployed in the

county as a percentage of the average annual civilian labor force in the county

Numerical

MedianFamInc Median Family Income Numerical dollarsMeanFamInc Average income of all families including those with

no income or loss in the prior calendar yearNumerical dollars

PctInPov Percent of persons in poverty in the calendar year prior to the census

Numerical percent

PropTaxRate property tax rate in dollars of tax per $100 of property value

Numerical decimal

PubSchSAT (2007-2008) Sum of all SAT scores for the math critical reading and writing sections divided by the number of scores for school year 2007-2008

Numerical

PM10Area Estimated amount of emissions to the air of particulate matter consisting of particles smaller than 10 micrometers (PM10) for a one-year period as measured in short tons (2000 pounds) from area sources

Numerical

AgeMedian Median age of population NumericalEmpManf1996EmpManf1997 EmpManf1998 EmpManf1999 EmpManf2000

All nonfarm private industry full-time and part-time manufacturing industry employment by year

Numerical

EmpConst2000 All nonfarm private industry full-time and part-time construction industry employment

Numerical

EmpServ2000 All nonfarm private industry full-time and part-time service industry employment

Numerical

EmpFarm2000 All full-time and part-time farm industry employment NumericalOldUnits Number of housing units greater than 30 years old NumericalHospitalBeds Defined as general acute care beds in hospitals NumericalVetPop Veteran Population NumericalAge65More People 65 and older NumericalBlackPop People who define themselves as Black NumericalVotesRepPres Total votes cast for Republican Candidate NumericalVotesDemPres Total votes cast for Democratic Candidate Numerical

Table 2 PROC MEANS Results

Table 2 PROC MEANS Results (Continued)

Variable N Maximum Minimum Sum Mean Median

PopCensLandAreaUnempRateMedianFamIncMeanFamIncPctInPovPropTaxRatePubSchSATPM10AreaAgeMedianEmpManf1996EmpManf1997EmpManf1998EmpManf1999EmpManf2000EmpConst2000EmpServ2000EmpFarm2000HBper10000EMper10000ESper10000Vetsper10000Age65per10000Blackper10000PercentRepubManfjobsGTservjobsUrbpopGTrurpop

100100100100100100100100100100100100

98989898

100100100

98100100100100100

98100

695454009488400000

13000000067149008142700

23900000010300000

1684002010400

467000000584740058201005827500569520056229003964100

18565800298600

579453147310534380030163986236029623375

076942931000000010000000

4149001726400000

1300000030186003898100

7100000004300000

1203005460000000

250000000750000000750000000770000000820000000790000000

21800000001860000000

00

802638813448300795467065898656317714742

18766420603028223

00

8049313004871090

4622000000417754100513086800

143420680285000

1431970034676900

375050863909008546290084306100818263007999840033604200

1320559008497500

2172599348255

12243926102429461421360121530420

563765897340000000330000000

80493134871090000

4622000041775415130868

14342000006802850

143197346769

375050000863909854629860266834962816310342900

13205598497500000

2172589839539035474

122439102429142136215304

056376590346938803300000

47879004629000000

4100000041485005001800

13400000006500000

144650250500

373500000548700535000522150491150468850189000510850

7435000000198597017

9377375731110263

9696287026138809191655

0571486300

Table 3 PROC FREQ Results

Table 4 PROC CORR Procedure Results

Table 5 Regression Results

Figures

Figure 1 PROC CHART Results

GeoRegion Cum Cum

Freq Freq Percent Percent

sbquo

coast sbquo 17 17 1700 1700

sbquo

mountain sbquo 23 40 2300 4000

sbquo

plain sbquo 60 100 6000 10000

sbquo

Šƒƒƒƒˆƒƒƒƒˆƒƒƒƒˆƒƒƒƒˆƒƒƒƒˆƒƒƒƒˆ

10 20 30 40 50 60

Frequency

ManfCategory Cum Cum

Freq Freq Percent Percent

sbquo

VeryLow sbquo 20 20 2000 2000

sbquo

Low sbquo 34 54 3400 5400

sbquo

Medium sbquo 21 75 2100 7500

sbquo

High sbquo 16 91 1600 9100

sbquo

VeryHigh sbquo 9 100 900 10000

sbquo

Šƒƒƒƒˆƒƒƒƒˆƒƒƒƒˆƒƒ

10 20 30

Frequency