This latter group of studies is interesting, but of limited use for the purpose of this study. First, they do not focus closely enough on CZMA activities. Second, they do not provide a national perspective, but rather, a set of disaggregated context-specific estimates. B-3 Research Design and Results Table B-3 motivates the formal statistical tests that are reported in this section. That table ranks the states participating in the CZM program according to different measures of coastal GNP change from 1978 to 1985, and real spending on CZMA activities (Section 306) from 1982 to 1985. The first column reports the absolute change in coastal GNP in billions of 1982 dollars. The third column is the percentage change in coastal GNP between 1978 and 1985. The last two columns report changes in coastal GNP on an average annual basis. Note that the top five states in terms of growth received an average of $8.8 million in CZM spending, whereas the bottom five states received only $2.8 million, on average. Additionally, only one state ranked in the top five in coastal GNP growth received less in CZM funding than any state ranked in the bottom five. Nonparticipating states have average growth ($4.97 billion), more like the low CZM recipient states ($.12 billion) than like the high CZM recipient states ($42.42 billion); and no nonparticipating state had greater growth than any of the five states with the fastest growth. Table B-3 also shows that both total and annual average growth rates were higher for the top five states than for the bottom five and the non-participants. 16 Tables B-4 and B-5 report results of two types of correlation tests-simple Pearson correlations and Spearman's rank-order correlations. In Table B-4 the CZM spending data from column two of Table B-3 are correlated with the change in coastal GNP data from the first column of Table B-3. In Table B-5 the data in column three of Table B-3 are used instead of the data in column one from that table. In both tables five measures of output change are employed. The first three are the components of coastal GNP (based on payroll data), discussed at length in Section A. The fourth output measure is total coastal GNP, also based on payroll data (that is, the sum of the data used for columns one through three). The fifth measure is a broader definition of coastal GNP which includes all economic activity in coastal counties. This range of definitions is used to ensure the robustness of the results.17 All correlation coefficients are calculated using data from participating states alone. Four of the five Pearson coefficients in Table B-4 are positive and statistically significant at a P value of 0.05 (the exception is coast-linked GNP, which is positive but not significant). The values of the significant coefficients range from 0.5017 to 0.575. Three of the five rank-order correlations reported in Table B-4 are statistically significant, ranging in value from 0.4338 to 0.4901. The results in Table B-5 are uniformly different. There, none of the coefficients is significant at P=0.05.18 The question that arises following an examination of Tables III-4 and III-5 is: why do states with large absolute increases in coastal GNP also have large total CZM spending, while states with large percentage increases in coastal GNP do not (and vice-versa)? The answer lies partly in the law of small base numbers and partly in the formula by which federal funds are allocated. The law of small base numbers implies that small states tend to have high growth rates and large states tend to have slow growth rates, all else equal. The fact that large states appear to experience large total CZM spending would seem to imply that they have an advantage in obtaining funding over their smaller counterparts partly because the formula for CZM allocations uses population. To account for that fact, population should be a control variable in the regression analysis. NOTE: GNP-1 is coastal GNP based on payroll data. GNP-2 is GNP-originating in all activities in coastal counties. Table B-5 Correlations between Percentage Change in Real Output and CZM Expenditures NOTE: GNP-1 is coastal GNP based on payroll data. GNP-2 is GNP-originating in all activities in coastal counties. In addition to running correlations, a series of ordinary least squares (OLS) regressions were estimated, with average annual growth in coastal GNP as the dependent variable. These regressions relate the measure of output to the level of real CZM spending averaged over the years for which data are available, population, and a series of dummy (or indicator) variables. The dummy variables identify the region of the country in which the state is located and whether or not the state was a participant in the program.19 Results from these regressions are generally consistent with the simple correlation analysis reported above. The full regression results are presented in Appendix Tables A7 to A11. The Appendix also contains a table of variable names, a table of descriptive statistics from the model, and an accounting of dummy variables. Three basic conclusions come out of the regression analysis. First, CZM program expenditures are never both statistically significant and negatively related to the growth in coastal GNP. This implies, at least, that CZM spending does not have deleterious effects on state economies. In fact, under many specifications, CZM spending is statistically significant and positively related to coastal output growth. The second general conclusion that can be drawn from the regression analysis is that the control variable for population affects the magnitude of the coefficient of CZM spending, and sometimes the significance. However, the policy variable still appears positive and significant in most models.20 The third general conclusion is that the dummy variables do not play a significant role in explaining the economic growth of the coastal zone. The participation dummy is never significant by itself, and it is individually significant in the presence of the regional dummies in only one of the models: for coast-linked activity without AVGPOP. All the dummies are jointly significant only in the case of coast-linked activity. In other words, the null hypothesis that all the dummy coefficients are zero can be rejected only for the case of coast-linked activity. This result suggests that coast-dependent activity, coastal services, and aggregations of the three components of coastal GNP all grow at a similar rate regardless of the region or of participation in the CZM program. It also suggests that the regions are not growing in the same way with respect to coastlinked activity.21 The fact that there is not a significant relationship between CZM spending and coast-linked GNP should not be surprising. Recall that coast-linked activities are those that use products from coast-dependent industries in their production processes, or produce intermediate inputs for coastdependent businesses. Much of the coast-linked activity is located in noncoastal counties. Since CZM spending is concentrated in coastal counties, its stimulative effect cannot be expected to be felt where most of the coast-linked activity takes place. There are input-output relationships between coast-dependent and coast-linked businesses, but those apparently are not strong enough to transmit the effect of CZM spending that is felt by the coast-dependent activities. Table B-6 summarizes the coefficient estimates for the CZM spending variable. The values in the first three rows of the table indicate the degree to which a one dollar increase in CZM spending is associated with increases in coastal payroll. The last two rows show the relationship between changes in CZM spending and alternative measures of coastal GNP. Models 3 and 4 correspond with the regressions using a population control variable. These results indicate that a dollar of CZMA spending from federal government sources is associated with more than an $11 increase in payroll deriving from coast-dependent activity (or, approximately $30 in GNP), and at least a $262 increase in payroll due to coastal services (or some $700 in GNP). At the same time, CZM spending is shown to have little statistical association with coast-linked economic activity.22 For all sectors aggregated together (GNP1), a dollar of CZMA spending from federal sources is related to at least an $822 increase in coastal GNP. Table B-6 Dollar Increase in Coastal Output Associated with a Dollar of CZM Spending NOTE: GNP1 is coastal GNP based on payroll data. GNP2 is GNP-originating in all activities in coastal counties. Values above the dotted line are dollars of payroll, and below the line are dollars of GNP. The values in Table B-6 must be interpreted with some care. They do not indicate the amount of economic activity induced by CZMA spending, for two reasons. First, the regression models developed here are too crude to be used to test causality. The significant statistical relationships that are shown to exist between CZMA spending and economic activity are consistent with the view that coastal protection enhances demand for coast-related goods and services, and hence, the value of the ocean sector. But to conclude with certainty that a one-way relationship exists would require a more completely-specified model and finer data than are available.23 Second, the federal CZM program leverages other spending on coastal protection, from states and local governments. The coefficients estimated in a model that included this other spending would be lower than what has been estimated using federal outlays alone. The failure to account for state-local coastal spending does not invalidate what has been done in this study. To a large degree, state spending is a fixed percentage of federal spending, because states are required to “match” each $0.80 in federal monies with an additional $0.20. Thus, it is straightforward to rescale the coefficient estimates in Appendix Tables A7 to All to account for the match. The bias of our estimates would not increase. The estimates were not rescaled because the principal concern in this report is with the relationship between federal spending and economic outcomes. Local supplements to federal-state CZMA funds are not necessarily in proportion to federal outlays, and therefore, could change the properties of the estimators if included in the analysis. Unfortunately, data on local supplements were not available. Finally, critics might argue that models such as the one used here simply show the relationship between state population and coastal GNP growth, because population is included in the Section 306 allocation formula. That is not a problem for two reasons. First, a reasonable proxy for coastal area population is used in Models 5 and 6.24 Second, states do not necessarily receive the amount of 306 funds to which they are entitled. States do not apply for all the funds that are available, and sometimes turn back unused amounts.25 B-4 AN ALTERNATIVE RESEARCH DESIGN FOR MEASURING THE Rather than relating CZM spending and coastal GNP as we did in the previous section, we could simply ask how much more people were willing to pay to live, work, or recreate on the coast == |