APPENDIX 1

POPULATION FORECAST FOR JEFFERSON COUNTY AND PORT TOWNSEND
FINAL REPORT

Prepared for the City of Port Townsend

Watterson West Group, Inc. 1121 23rd Avenue East Seattle, Washington 98112 (206) 322-0522

December, 1994

LIST OF FIGURES

1.	Jefferson County Population, 1970-1993	4
2.	Population Natural Increase vs. Migration	5
3.	Employment Change vs. Under-65 Population Change	6
4.	Employment/Under-65 Population Ratios	7
5.	Port Townsend/Unincorporated Population	9
6.	Jefferson County Share of State Population	13
7.	County 65+ Share of State 65+ Population	16
8.	County <65 Share of State <65 Population	17
9.	Jefferson County Population Projections	18
10.	Jefferson County Population, 1970-2014	19
11.	Fifty-Year Population Forecasts	21
12.	Housing Unit Population Projection Method	24
13.	New Housing Units Permitted, 1980-93	27
14.	Total Housing Units, 1993 and 2014, Continuation of Trends	29
15.	Total Population, 1993 and 2014, Continuation of Trends	30
16.	Total Housing Units, 1993 and 2014, Trends with Constraints	31
17.	Total Population, 1993 and 2014, Trends with Constraints	31
18.	Port Townsend Population Projections	33

LIST OF TABLES

1.	Population Estimates for County Planning Areas	9
2.	OFM Forecast and Actual Estimates	11
3.	Jefferson County Population Forecast	19
4.	Jefferson County Housing Unit Projection, 1990-2014	25
5.	Planning Area Housing Unit Data, 1990	28
6.	Trends with Constraints Projection Differences from Continuation of Trends	32
7.	Population Forecast for County Planning Aeaea, 1994-2014	34
8.	City of Port Townsend Population Projections	35
9.	Fifty-Year Projections	36

APPENDIX TABLES

A-1.	OFM Population Forecast for Washington, State, December, 1992
A-2.	Housing Unit Permits by Planning Area, 1980-1993
A-3.	Population for Planning Areas in Jefferson County based on Trends with Constraints, 1990-2014

EXECUTIVE SUMMARY

INTRODUCTION

Two major population forecasts for Jefferson County were previously prepared and published to assist in the Growth Management Act (GMA) comprehensive planning for the County and the City of Port Townsend:

• The Washington State Office of Financial Management (OFM) population forecasts for all counties in the State, published in January, 1992-Countywide only; and

• The Jefferson County Planning Department forecast report. Population Change in Jefferson County: The Next 20 Years, published in October, 1993-Countywide and planning areas (including the City).

The OFM forecast was intended to define a minimum amount of future growth over a 20-year period which County and City plans were to accommodate. Unfortunately, population trends in Jefferson County since 1990 have shown the OFM forecast to be so low that it is not particularly credible or useful for planning.

The County's report addressed this problem and developed additional projections for the County level, along with distributions of a medium scenario to the subcounty planning areas, including the City of Port Townsend.

This population forecasting project builds new forecasts for the County and its planning areas using best-practice forecasting methods. The objective has been to apply the best methods to the technical forecast, while making the assumptions underlying the forecast as clear as possible, and opening the forecasting process to as much input and discussion as time available to the project would permit.

The forecasts in this project assume that there are to be three urban growth areas (UGAs) designated for the County-one in the City and two in the unincorporated County. If this policy assumption later changes, then the population forecasts developed here may not be valid as recommended.

 

HISTORICAL TRENDS

• Since 1970, Jefferson County's population has increased at an average rate of 3.5 percent per year, far higher than the rate for Washington State as a whole;

• Almost all of the population growth has been due to net in-migration to the County, rather than from the natural increase (births minus deaths) of the County's population;

• Net in-migration for the over-65 age group has been consistent and steadily increasing over time, while for the under-65 age groups net in-migration was concentrated in the later 1970s and early 1990s.

• For most years since 1970, the growth in the under-65 aged population has been related to changes in the County's job base. Since 1990, the under-65 population has risen fast, but employment growth has been sluggish. One explanation is the emergence of a "new" economic base of entrepreneurs, the semi-retired, pan-time professionals, and telecommuters, along the with "service economy" of retail, construction, and consumer services supported by their incomes.

• The City of Port Townsend's population has grown from 5,241 in 1970 to 7,940 in 1994, for an average annual rate of increase of 1.7 percent. But since-1990, the City has averaged 3.2 percent per year in population growth.

• Much of the County's population growth since 1970 has occurred outside of Port Townsend. The largest growth areas, besides Port Townsend, have been the Quimper Peninsula, the Tri-Area, and the Port Ludlow/Oak Bay area.

 

COUNTY FORECAST

Methods and Assumptions. The projection of the Jefferson County population is based on the County's share of the population in Washington State as a whole, separately for the 65-and-over and the under-65 age groups. This approach builds in the dramatic effects of the aging of the baby boom generation in the long-term future.

Three types of assumptions are necessary for the County forecasts:

• the future change in the share of the State's population for each age group;

• the Statewide forecast-the OFM population forecast for Washington State as a whole prepared in December, 1992 (updated to December, 1993);

• the potential constraints on County population growth in the future-available water supply, infrastructure requirements, and the job base-as well as policies to accommodate future growth.

 

Twenty-Year Forecast. Three alternative projections were developed.

1. The OFM forecast for Jefferson County, which implies a constant 1993 share of the State's population. The OFM forecast is already tracking very low compared with current 1994 population estimates, and thus is called the low projection.

2. The second County projection is based on the average increase in County share of the State's population over the-past 24 years. The average increase in share is applied to both the 65+ and the under-65 age groups. The average increase in share projection produces a total population of 38,337 for Jefferson County in 2014. This becomes the medium 20-year projection.

3. The third projection for the County is based on the peak period increase in County share of the State's population in the past. The peak increase in share is applied to both the 65+ and the under-65 age groups, producing a projected total population of 45,227 for the County in 2014. This is considered the high 20-year projection.

Over time, the County's population growth first diminishes somewhat from the high rate of the early 1990s, but accelerates again around 2010 because of the effects of the rapidly expanding pool of retirement-age people in Washington on historically desirable retirement counties like Jefferson.

The medium projection, based on the historical average rate of increase in shares of the State population for both the 65+ and the under-65 age groups, contains the most reasonable assumption of share increase, and thus is the most likely to occur. The medium projection is therefore recommended as the 20-year forecast for Jefferson County.

 

Fifty-Year Forecast. A 50-year population was developed to be useful in capital facilities, water, and utility planning. This is intended to be consistent with the 20-year future. Therefore, the 50-year forecast takes two alternative directions starting from the most likely medium projection of 38,337 at 2014:

1. One continues the average increase in the County's share of the State's population for the duration of the period, reaching 61,541 at 2044. This is the high alternative;

2. The other adopts a constant 2014 County share of the State's population for the remaining thirty years, for 49,110 at 2044. This is the low alternative.

The high alternative takes population growth to a level which may not be supportable with current ground and surface water resources. The low alternative stays more or less below that level. For a 50-year population growth at the high end, some changes in water supply or demand might be necessary.

 

SUBCOUNTY DISTRIBUTION

Methods and Assumptions. The population distribution to subcounty areas, including the City of Port Townsend, starts from the recommended (medium projection) Jefferson County forecast. The method for distributing the County population forecast to the twelve County planning areas is the Housing Unit Method. Under this method,

future housing units are projected at the Countywide level,

the housing units are then distributed to the planning areas, and

finally they are converted to population within planning areas.

Three types of assumptions and calculations underlie the housing unit approach to population distribution:

• the Countywide housing unit growth, which is based on the County population forecast, but includes assumptions on average household size, vacancy rates, and percent of units for seasonal use;

• the planning area housing unit growth, which is based on the average shares of County building permit activity for each planning area in past years;

• assumptions on the key factors of average household size, vacancy rates, and percent seasonal units for each planning area, which come from estimated 1990 data and County trends.

Two alternative distributions were developed for each subcounty planning area:

1. One is the Continuation of Trends projection, in which the recent trends in the planning area's housing unit growth share of the County are extended throughout the 20-year projection period.

2. The other is the Trends with Constraints, in which the past trends in housing unit growth are modified by considerations of physical and policy constraints on particular planning areas. Where future growth is constrained, that growth is redistributed to planning areas with Urban Growth Areas that have the potential to receive the growth. Constraints on planning areas are severe, moderate, or slight, depending on the seriousness of the potential limitations on future housing unit growth.

For the Trends with Constraints distribution, it is necessary to assume that comprehensive planning under the GMA will be able to address the physical and policy constraints and will provide at least the three designated interim UGAs as alternative locations for future growth.

Twenty-Year Forecast. Projections for the two distributions were developed, along with additional projections for the City of Port Townsend.

• Under the Continuation of Trends projection, most of the population growth in the County over the 20-year period is concentrated in Port Townsend, the Quimper Peninsula, the Tri-Area, and Port Ludlow/Oak Bay-the areas that have received most of the housing unit growth in recent years. Several other planning areas show growth that is high relative to their existing numbers of housing units and population.

• The Trends with Constraints projection concentrates more of the growth in Port Townsend and Port Ludlow/Oak Bay. Constraints reduce particularly the Quimper Peninsula and Marrowstone Island planning areas, as well as the Center/Inland Valleys area. The Tri-Area planning area remains about the same.

The Trends with Constraints projection contains the assumptions and the distribution most likely to occur within the County over the 20-year period. The Trends -with Constraints projection is therefore recommended as the 20-year forecast for planning areas within Jefferson County. It is contingent on the designation of the three UGAs as assumed.

For the City of Port Townsend, two additional projections were developed in order to indicate the total range of possibilities for the 20-year growth of the City.

• The assumptions and calculations of the Continuation of Trends distribution were applied to the low (OFM) Countywide projection, and the Trends with Constraints distribution was applied to the high Countywide projection.

• The indicated range for the City population in 2014 is from 10,229 on the low end to 17,854 on the high end, or from a low growth rate averaging 1.3 percent per year to a high growth rate averaging 4.1 percent per year.

Fifty-Year Forecast. A 50-year population distribution was made for Port Townsend and the adjacent planning areas (Quimper Peninsula, Marrowstone Island, and the Tri-Area).

• The 50-year distribution starts from the most-reasonable Trends with Constraints distribution at 2014, and then takes two alternative directions, based on the high and low 50-year projections at the County level.

• For Port Townsend, the high projection is 22,451 for 2044, and the low projection is 17,395.

The narrowness of the range is due to the tightness of the County population range for 2044 and to the use of the Trends with Constraints approach for both 50-year projections.

 

RECOMMENDATIONS

1. For the Countywide population forecast, the medium projection of 38,337 in the year 2014 is recommended for adoption. This projection series is based on the most reasonable assumption of the change in the County share of the State's population, and thus is the most likely to occur. It is a reasonable, defensible projection to use as a 20-year population forecast.

2. For the subcounty population distribution, the Trends with Constraints projection contains the assumptions and the distribution most likely to occur within the County over the 20-year period. It is based both on past trends in housing development and on considerations of physical and policy constraints that will potentially affect particular planning areas. The Trends with Constraints distribution for the County planning areas to 2014 is recommended for adoption as the 20-year population forecast.

3. For the City of Port Townsend, possible population futures range from a low of 10,229 to a high of 17,854, as of the year 2014. These represent average annual growth rates over the 1994-2014 period of 1.3 percent to 4.1 percent. The Trends with Constraints projection is a population of 13,013 in 2014, or an average of 2.5 percent growth per year. This projection is recommended for adoption as the 20-year population forecast.

4. One important caveat for these recommendations is that if the three assumed UGAs are not formally designated (that is, if Port Townsend is the only UGA) then the recommended population distribution, and perhaps also the total County forecast, is at risk.

I. INTRODUCTION

About this Project

This population forecasting project is intended to build upon and refine previous forecasting work for Jefferson County and its subcounty planning areas. There have been two major population forecasts for Jefferson County which were prepared and published to assist in the Growth Management Act (GMA) comprehensive planning for the County and the City of Port Townsend: (1) the Washington State Office of Financial Management (OFM) population forecasts for all counties in the State, published in January, 1992, and (2) the Jefferson County Planning Department forecast for the County and its planning areas. Population Change in Jefferson County: The Next 20 Years, published in October, 1993.

Under the GMA and the County-Wide Planning Policy adopted by Jefferson County and Port Townsend, the OFM population forecast for the County1 defines a minimum amount of future growth over a 20-year period which the County and City plans are to accommodate. The County and the City then are to consider the OFM forecasts as a low or base projection, and prepare alternate population forecasts that establish a medium and high projection for the 20-year GMA planning horizon.

Unfortunately, even though the OFM population forecast for Jefferson County was to define the minimum growth for planning, it is so low that it is not particularly credible or useful as a guide for County and City comprehensive planning. The official population estimate for Jefferson County in 1994 is almost halfway to the 2012 OFM forecast which took off from the 1990 population estimate. According to the OFM forecast, Jefferson County has only 5,500 additional people to accommodate between 1994 and 2012.

The County's report addressed this problem and developed additional projections for the County level, along with distributions of a medium scenario to the subcounty planning areas, including the City of Port Townsend.

This population forecasting project, as described in this report, builds new forecasts for the County and its planning areas using best-practice forecasting methods.2 The intention is to apply the best methods to the technical forecast, while making the assumptions underlying the forecast as clear as possible, and opening the forecasting process to as much input and discussion as time available to the project would permit.

1 The 1992 OFM population forecasts for GMA planning are countywide only. No subcounty-city or planning area-forecasts were prepared or published by the OFM.

2 See the Washington State Department of Community Development guidebook on population forecasting. Predicting Growth and Change in Your Communities: A Guide to Subcounty Population Forecasting, June, 1993.

The forecasts are a product of their stage in the GMA comprehensive planning process within the County and City. As such, they reflect the inputs and assumptions available at the time. A key underpinning of the forecasts, for example, is that there are to be three urban growth areas (UGAs) designated for the County-one in the City and two in the unincorporated County. If this policy assumption later changes, then the population forecasts as developed here may not be valid.

A Note on Forecasts and Forecasting

Population forecasting plays a central role in comprehensive planning and planning decisions. After all, planning is about the future, and there must be some information about the future in order to plan. The GMA acknowledges this and includes population forecasting as an integral part of the GMA planning process.

Forecasts are statements about the future, but we all know that the future is unknowable. Any number of events will occur which will change the future from what we think we see today. That is why in doing a forecast it is important to state all assumptions about the future as clearly as possible. Assumptions about the future underlie all forecasts. The best forecasts are based on the best or most reasonable assumptions.

The best forecasts also use as much information as is available. This includes not only data on past trends in an area, but also on the future by linking to credible forecasts made on larger geographic areas, such as the state or nation. Through this means, broader demographic and economic futures can be applied to county and subcounty forecasting.

Forecasts come in several varieties. The most common are extrapolations of past trends into the future. The past is known, and a strict use of past trends may minimize the number or breadth of assumptions necessary to the forecast. But the future may or may not resemble the past. If some information is known that may affect the future, then it could be important to bring this to the forecasting. Such information could involve likely physical capacity constraints or impending policy changes that will cause deviations from past trends.

Finally, a fundamental problem with highly visible forecasts, like subcounty population forecasts, is that they are somewhat self-fulfilling prophecies. A most-likely population forecast for subcounty planning areas becomes even more likely after public and private decisions are made using the forecasts. The forecasts themselves help to influence the outcome of future development. This underscores the importance of the population forecasting process, but it also blurs the differences between forecasts based on past trends and any policy considerations.

What is in this Report

This report documents the population forecasting process that was undertaken starting in June, 1994, covering Jefferson County, the City of Port Townsend, and the other planning areas of the County. The next section of the report presents some information on past trends in Jefferson County and the City of Port Townsend in population components, employment, and housing. All of these are relevant to projecting the population into the future.

Then the population futures for the County as a whole are examined and projected, producing a forecast range both for the 20-year and 50-year planning horizons. The projection methods and assumptions are clearly identified.

The Countywide forecasts are then distributed to subcounty planning areas, including the City of Port Townsend. Again, the methods and assumptions are identified and discussed, and two population projections are developed for the 20-year growth management period (note: four are developed for the City).

Finally, there is a section summarizing the forecasting materials and making recommendations for forecast adoption and use.

II. HISTORICAL TRENDS: FACTS ABOUT JEFFERSON COUNTY AND PORT TOWNSEND

Composition of the Population

Figure 1 shows the total population for Jefferson County from 1970 through 1993, broken out by the 65-and-over and under-65 age population components. Two trends stand out from this graph. One is that while the overall County population growth has been substantial-more than doubling over this period-it has come unevenly, with fast growth in the late 1970s and the early 1990s interspersed with much slower periods. Over this period, the County averaged 3.5 percent per year population growth, compared with 1.9 percent for Washington State as a whole.

The second trend is the steady year-by-year increases in the 65-and-over population. The growth of this population has averaged 5.3 percent per year, compared with 3.1 percent per year for the under-65 group. The Statewide growth of the 65+ population averaged 2.7 percent per year over this period. The over-65 component of the County population has risen from 13.5 percent in 1970 to 20.0 percent in 1993. Approximately this same 20 percent share of the county population also existed in Clallam and San Juan counties in 1993.

The composition of the population growth is further revealed in Figure 2. Here the' natural increase (net births over deaths) is compared with net migration by five-year periods since 1970 for the under-65 and 65+ age groups separately. The first conclusion to be drawn from this graph is that the natural increase of the Jefferson County population has contributed very little to overall population growth. By comparison, natural increase has been almost as important as net migration in the population growth of the State over this period. This difference is due to both to a relatively high death rate in the County, compared with the State, and to a relatively low birth rate.3

So, net migration is the source of nearly all of the net increase in the County population since 1970 (over 11,800 of the almost 12,900 overall growth from 1970-93). But Figure 2 also indicates significant differences in who migrated to the County during different time periods. For the 65+ age group, the net in-migration has been consistent and steadily increasing for every five-year period since 1970. For the under-65 ages, however, there are only two big periods for net in-migration-1975-80 and 1990-94-also the years of the great surges in overall population growth in the County.

3 The Jefferson County birth rate for the period 1980-93 was 11.2 per 1,000 residents, compared with 15.9 for Washington State. The County death rate for 1980-93 was 9.3 per 1,000 residents, versus 7.6 for the State.

Employment and Population

Changes in the under-65 year old population in an area are usually associated with economic conditions. Figure 3 charts the annual rates of change in Jefferson County employment levels compared with the rates of change in the under-65 population. Although the patterns are not always clear, there is an apparent relationship between rises and falls in the employment change rate and the under-65 population change rate. When the employment growth becomes low or negative, the-population change follows, with a lag of about a year.

Interestingly, this pattern seems to break down almost completely after 1990. Low employment growth has existed alongside strong growth in the under-65 population. Figure 4 presents another slant on this. The lower line plots the ratio of total County employment to the under-65 County population. The upper line shows the same ratio for the State as a whole. The pattern of the State ratio is rising through the entire period, with pauses or declines through recessions of the early 1980s and early 1990s. The County's ratio is much flatter, though with the same general shape, until 1990, when the ratio moves sharply opposite to the State. Population growth in Jefferson County is again shown to diverge from employment growth, and from the State pattern.

Jefferson County also has a very high proportion of self-employed workers-a ratio of almost 50 percent of the number of wage and salary workers-compared with only 20 percent Statewide.4 The County ranked second in the State in the number of businesses per 1,000 population, with 91.7 compared with 57.7 Statewide, and fifth in the rate of growth of new businesses.5

What all this suggests is that in recent years Jefferson County has begun to break free from the traditional notion of economic base, with not only a continuing steady increase of retirees bringing their own income sources, but also a substantial rise in under-65 population with "footloose" occupations and jobs. This means people moving to the County without jobs but starting their own businesses, people moving their personal businesses into the County, people semi-retired but working part-time either on their own or for a non-County business, and people "telecommuting" as employees to businesses in Seattle or beyond, with computers and occasional actual commute trips.

4 Data for 1991 from the Bureau of Economic Analysis, U.S. Department of Commerce.

5Data for 1991 from the Washington Stale Department of Revenue.

Coupled with this new economic base would be a rise in the "service economy" of retail, construction, and consumer services (along with government) that is supported by the incomes from retirees and the new businesses. These tend to be relatively lower wage jobs, but are still readily filled by workers, many themselves migrating in from outside the County.

There is anecdotal evidence of high turnover rates of population in the County, with many recent high school graduates leaving the County and many other newcomers arriving to take the available service jobs. The 1990 Census reported that almost 30 percent of the Jefferson County population had not lived in the County in 1985, with over half of the new people coming from outside the State of Washington. But the 1980 Census reported over 40 percent of the population had not lived in the County in 1975, with most corning from out of State. Interestingly, in 1990, only 22 percent of the Port Townsend population was from another county in 1985, compared with almost 35 percent of the Hadlock/Irondale population.

Population Distribution

The population growth of Jefferson County since 1970 is presented again in Figure 5, but with the breakdown of Port Townsend and unincorporated County areas (the rest of the County). Much of the growth of the County has occurred outside of Port Townsend. The proportion of the County population living outside of Port Townsend has risen from 51 percent in 1970 to 67 percent in 1993. As shown in the graph, during the rapid growth periods for the County-the late 1970s and the early 1990s-most of the growth was in the unincorporated areas. From 1976 to 1981, over 86 percent of the population growth of the County was in the unincorporated areas. From 1989 to 1993, it was over 81 percent. During slower growth periods, the distribution was more even.

The City of Port Townsend's population has grown from 5,241 in 1970 to 7,940 in 1994, for an average annual rate of increase of 1.7 percent. But since 1990, the City has averaged 3.2 percent per year in population growth.

There are few historical data on the population distribution by other planning areas within the County. There are, however, estimates based on 1990 Census data, along with building permit/housing unit data for previous years, as shown in Table 1.

Table 1 POPULATION ESTIMATES FOR COUNTY PLANNING AREAS
	County Planning Areas6
	1	2	3	4	5	6	7	8	9	10	11	12
1990 Population	7001	2076	686	3315	891	1075	1111	722	369	1123	1090	947
Growth 1980-90*	934	480	134	597	78	166	430	174	67	164	267	18
Growth 1990-93*	739	436	79	369	86	99	479	83	19	84	107	4
*Based on building permit shares of total unincorporated Cunty permits

6 Planning area names are as follows:

1.	PortTownscnd	7.	Port Ludlow/Oak Bay
2.	Quimpcr Peninsula	8.	Shine/Paradise Bay
3.	Marrowstone Island	9.	CoyIe/Toandos Peninsula
4.	Tri-Area	10.	Quilcenc
5.	Discovery Bay	11.	Brinnon
6.	Center/Inland Valleys	12.	West End

As indicated in Table 1, the most rapid population growth has occurred in Planning Areas 1,2,4, and 7 (Port Townsend, Quimper Peninsula, Tri-Area, and Port Ludlow/Oak Bay). Most rural, non-retirement areas have received relatively little population growth, either before or after 1990.

III. POPULATION FUTURES COUNTYWIDE

OFM County Forecast

As noted in Section I, the county forecasts prepared by the OFM for growth management planning were developed during 1991 and published in January, 1992. They began with a State forecast for each year from 1990 to 2012, and then allocated the State population to counties based on a variety of historical and projected demographic trends. The county forecasts included neither final corrections to the 1990 Census counts nor official estimates for 1991 and 1992. As a result, the Jefferson County forecast series started out low for 1990, and got cumulatively lower than the official population estimates for the 1991-93 period. Table 2 depicts this problem.

Table 2 OFM FORECAST AND ACTUAL ESTIMATES
Year	Actual Estimate	OFM Forecast
1990	20.406	20,146
1991	21.600	20,583
1992	22,500	21,029
1993	23,500	21,484
1994	24,300	21,950
1995		22,425
1996		22,735
1997		23,050
1998		23,368
1999		23,692
2000		24,019
2001		24,448
2002		24,884
2003		25,328
2004		25,780
2005		26,640
2006		26,709
2007		27,186
2008		27,671
2009		28,165
2010		28,669
2011		29,176
2012		29,693

Nonetheless, the GMA and the County-Wide Planning Policies clearly state that the OFM population forecast for Jefferson County will be the minimum growth for which the County and the City will plan. If the OFM population forecast is extended out to 2014 at the same rate of growth as its last years (1.77 percent between 2011 and 2012), then the minimum forecast of population for Jefferson County in 2014 is 30,754. This figure will be carried forward as the OFM forecast.

Projection Method in This Project

In this work, the projection of the County population through the 20-year future is based on the County's share of the population in Washington State as a whole. In particular, we examine the trend of past County shares of the State population, and then make an assumption about how that trend in share will change in the future. This is done not just for the total population, but for the population disaggregated into a 65-and-over age group and an under-65 age group.

There are several reasons why the share of State population method is appropriate for projecting Jefferson County population:

• it relates the County population to the source (through migration) of much of the growth in the County, for both 65+ and under-65 age groups;

• it makes use of the demographic forecasting done at the State level, including detailed data on births, deaths, and migration patterns by age, and a powerful demographic forecasting model;

• it provides a check or benchmark on the County forecasts, so that they will remain in a reasonable range relative to State patterns;

• it gets away from projecting growth rates from some arbitrarily selected period in the past.

Figure 6 shows the historical pattern of Jefferson County's share of the State's population, along with two projections of its share into the future-one is the OFM forecast discussed above, the other is the recommended forecast in the County government's 1993 report.

Note first how the Jefferson County share of the State's population has increased sharply and consistently since 1970. Then look at the projection of that share implied by the OFM forecast. The OFM forecast assumes that the County's share of the State's population will not change at all from 1993 through 2014, after 20 years of increases. There is little that can be said to justify this assumption.

The County's 1993 forecast, on the other hand, continues the recent (1990-93) trend of increase in the County share, but with a rapidly declining rate of increase after about 2000, until it reaches a constant share by the year 2014. This is certainly a credible assumption, though one that needs examination.

In this forecasting work, we look at the County's shares of the State's population for the 65+ age group and the under-65 age group separately, since the patterns of growth in these two groups were shown to be so different, and since we know that Statewide and nationwide the relative numbers of these groups will be changing dramatically out in the future.

We also develop three alternative projections for each group's share of the State population for that group.

• the low projection is the constant 1993 share assumption implied in the OFM population forecast for the County;

• the medium projection takes the average rise (for each age group) from the 1970-93 period, with a flattening of the rise over the forecast period;

• the high projection uses a higher than the average rise from the past, with a flattening of the rise over the forecast period.

Together, these alternative projections form a range that can be considered to bracket the reasonable possibilities for future changes in the County's share of the State's population. When combined with a State population forecast, they lead to a range of population forecasts for the County.

Finally, a 50-year population forecast is developed which is consistent with the 20-year future. The 50-year forecast starts from the most-likely medium projection at 2014 and then takes two alternative directions. One continues the average increase in the County's share of the State population for the duration of the period. This is termed the high alternative. The other adopts a constant 2014 County share of the State population for the remaining thirty years. This is called the law alternative.

Assumptions for County Forecasting

The key assumptions for the projection of the Countywide population include (1) the future change in share for the 65+ and under-65 age groups in Jefferson County, and (2) the State forecast on which the shares are based. In addition, there is the matter of potential constraints on the ability of the County to continue to absorb additional population over long periods into the future.

Future Change in Share. As described above, the alternative projections for the County population contain markedly different assumptions about the share of the State's population the County will have in the future. Which one or ones are the most reasonable? Obviously, the OFM assumption of no future change in share, after more than 20 years of increase, is unreasonable. But what rate of increase is appropriate?

Rapid rates of increase are difficult to maintain over long periods of time. If anything. they are more likely to diminish than to accelerate. Thus, a higher increase in the County share of the State's population in the future than in the past is possible but not probable. Thus, the medium projection track-increase in future shares at the average rate of the past, with some diminishing of the increase over the long term-is the most reasonable assumption. The high projection, with a higher than historic increase in share of the State's population, is useful to establish a range of possibilities, but should not be rated as a probable future.

Applying these share change assumptions separately to the 65+ and the under-65 age groups allows the population in each age group to grow independently, depending on their relative scale in the State in the long-term future. This builds in the demographics of aging and migration in the State population and the effect this may have on Jefferson County.

Future State Population. The other key assumption in these projections is the growth in the State's population in the future, in both the 65+ and the under-65 age groups. Besides the county forecasts prepared for the GMA planning process in 1992, the State OFM develops a long-term demographic forecast for Washington State as a whole, updated annually. These forecasts use a complex demographic model that includes detailed data on births and deaths by age in the population, along with a variety of detailed migration information. Its future projections are determined by the aging of the population and the best available estimates of migration patterns among states. The resulting forecasts are not perfect and will not be exactly on target, but certainly are reasonable and build in a great deal of important information about the future.

The State population forecast that the changes in shares are applied to is the one published by the OFM in December, 1992. This is the most recent annual forecast available in fully published form. A review of the December, 1993, forecast numbers found relatively few changes from 1992.7 Therefore, the OFM State forecast at the end of 1992 was used in this forecasting work.

Constraints in Jefferson County. There is considerable concern on the part of local government staff and elected officials that there are significant constraints to the ultimate amount of population that can be accommodated in Jefferson County. These concerns focus on (1) water supply, (2) infrastructure requirements, and (3) the job base.

The water supply for County population and businesses is composed of Port Townsend's surface supply from the Quilcene River and community and individual wells tapping ground water. It should be acknowledged, however, that there are serious constraints to providing future water service to certain areas of the County. In particular, the Tri-Area's water supply system is currently at or near capacity, and significant capital expenditures will be necessary to serve future growth. Additionally, and irrespective of infrastructure limitations, the ultimate population which can be sustained using both surface and groundwater sources has yet to be determined.

Undoubtedly, serving continued population growth will require substantial investment in water, wastewater, and road infrastructure. Moreover, renovation and improvement of existing systems which are deficient will be costly and difficult to achieve. Although the problems associated with providing expanded infrastructure are a valid concern, they are less a constraint to growth than a financial problem associated with growth. In any event, these are decisions to be made by policy makers during the GMA planning process.

Reference has already been made to the possibly changing relationship in Jefferson County between the traditional job base and population growth. The growth of the County population, especially that of retirees, is less connected to job opportunities in the County than ever before. More and more, the population growth is creating its own economic base. This trend is likely to continue into the future. Any lack of expansion in the traditional basic economy is therefore not going to stop or even retard the potential for population growth in the County.

7 This conclusion was reinforced by comments from OFM staff in recent telephone conversations.

Twenty-Year Projections

The first step in the Countywide population projections is to develop the future year County shares of the State's population for each age group. Figure 7 presents the shares for the 65-and-over age group. The historical pattern has been of almost unbroken, continuous increase at about the same rate. The 65+ age group's share rose from about 0.4 percent to just under 0.8 percent of the State's 65+ population from 1970 to 1993.

For the future, a low projection of this group's share of the 65+ population Statewide would be a constant share at 1993 levels-as implied by the OFM forecast for Jefferson County.8 A medium projection takes the average rate of increase in the 65+ population share and continues that rate of increase into the future, but with some flattening to reflect greater size of the County, higher costs, fewer prime sites (e.g., waterfront, view), and some constraint due to water supply limits. The high projection takes the highest single-year rate of increase in the 65+ population share and continues that rate of increase into the future, regardless than the average for any historical period. It too flattens somewhat to reflect the above considerations.

8 The OFM forecast did not disaggregate or project by age group.

The same procedure is applied to the under-65 age group in Figure 8. The historical pattern for this age group is much more irregular. The low projection of this group's share of the under-65 Statewide population is again a constant share at 1993 levels. The medium projection is again the average rate of increase in the under-65 population share continued into the future, with minimal flattening of the curve because of less sensitivity on the part of this age group to the availability and cost of prime sites than for the 65+ population seeking retirement amenities. The high projection takes the average rate of increase in the under-65 age population share since 1990 and projects that rate of increase indefinitely into the future, with minimal flattening.

When these share projections are multiplied by the State population forecast for each age group and summed together, we get the actual population projections presented in Figure 9. The low projection, based on constant shares of the State population, is the OFM forecast for Jefferson County. As noted earlier in this section, it starts from 1990, rather than 1994. The medium projection, based on the average rate of increase in shares of the State population, reaches 38,337 for the year 2014. The high projection, based on the highest past rates of increase in shares of the State population, reaches 45,227 by 2014.

All three projections bend upward toward the end, despite the flattening of the share curves in Figures 7 and 8. This is due to the demographics contained in the State population forecast. The 65-and-over age group is expected to begin to increase at an explosive rate after about 2010, because that is when the post-war "baby boomers" start turning 65 years old. Actually, in the State forecast, the percent of the population aged 65 and older stays about constant at 11.5 percent from 1994 to 2004, then begins to increase, accelerates fast after 2008, and reaches 14 percent by 2014. The effect of this rapidly expanding pool of retirement-age people on the historically desirable retirement counties will be enormous. The effect of this demand alone will be the major factor in Jefferson County growth at this point.

Forecast Conclusion

The medium population projection for Jefferson County is shown in Figure 10, broken down by the two age groups. This projection is based on the average rate of increase in shares of the State population for each age group. This is the most reasonable assumption of share increase, and the most likely to occur. The medium projection is therefore recommended as the 20-year forecast for Jefferson County.

Table 3 summarizes this County population forecast

Table 3 JEFFERSON COUNTY POPULATION FORECAST
Year	Population
1970	10,661
1980	15,965
1990	20,406
1994	24,300
1999	27,860
2004	31,088
2009	34,535
2014	38,337

Fifty-Year Projection

A 50-year population projection for Jefferson County was developed to be consistent with the 20-year future described above. The 50-year projection starts from the most-likely medium County projection-the recommended 20-year forecast-at 2014 and then takes two alternative directions, both based on shares of the State's population.

First, since the OFM population forecast for the State (December, 1992) did not go beyond 2020, it was necessary to extend the forecast to 2044. The OFM State forecast to 2020 exhibited a diminishing rate of growth for the population in the later years of the forecast period. The curve of this growth rate was then extended beyond 2020 to the horizon year of 2044, thus continuing the diminishing rate of growth for the State's population.

Then, two alternative directions for the County's share of the State's population were developed. One direction continued the average increase in the County share of the State's population throughout the extended projection period. This is a generous assumption in the face of a slowing State population growth. This is termed the high alternative. The other alternative adopted a constant 2014 County share of the State's population for the remaining thirty years. While this represents a break with the pre-2014 increases in share, it is a moderate, conservative assumption of the long-term future of the County's population. This is called the law alternative.

These two 50-year alternatives are presented graphically in Figure 11.

These two alternatives indicate a range of reasonable possibilities for the County population over the 50-year period. Long-term future population growth could reasonably occur anywhere within this range. The high alternative takes population growth to a level which may not be supportable with current ground and surface water resources. The low alternative stays more or less below such a level. For a 50-year population growth at the high end, some changes in County water supply or demand might be necessary.

IV. SUBCOUNTY POPULATION DISTRIBUTIONS

Subcounty Areas and Growth Management Planning

For purposes of this forecasting work, Jefferson County is divided into twelve subcounty areas, the current County Planning Areas as designated by the County:

1.	Port Townsend
2.	Quimper Peninsula
3.	Marrowstone Island
4.	Tri-Area
5.	Discovery Bay
6.	Center/Inland Valleys
7.	Port Ludlow/Oak Bay
8.	Shine/Paradise Bay
9.	Coyle/Toandos Peninsula
10.	Quilcene
11.	Brinnon
12.	West End

The City of Port Townsend (Planning Area 1) is the only incorporated city in Jefferson County. All other planning areas are unincorporated areas.

The Growth Management Act, read in concert with the County-Wide Planning Policy, suggests that should be three Interim Urban Growth Areas (UGA) within the County. One is the City of Port Townsend. The others are the Tri-Area UGA and the Port Ludlow UGA. Both of these lie within the planning areas bearing their names (Planning Areas 4 and 7, respectively), but are designated as smaller geographic areas than the entire planning areas. Under the GMA, the UGAs are intended as the focus for future population and employment growth, as well as related public investments in services and infrastructure. The non-UGA areas of the County are mostly rural, but locations within rural areas can be designated as rural town centers, or zones with some urban characteristics. A possible example of such a center might be an existing town like Quilcene.

Under the GMA, past trends in the location of population and employment growth are important to recognize, but planning under the GMA is intended to lead to a future which is somewhat different from the past, in directing growth away from resource lands, rural areas, and environmentally sensitive areas, and toward designated urban areas which can be effectively served with public infrastructure. The population distribution described in this section will follow the GMA by indicating first a distribution based exclusively on past trends and then one that takes account of potential physical constraints and policy directions. The latter distribution assumes that the three interim UGAs will be designated as UGAs for GMA planning, and will thus be available to serve their intended purposes for the broad diversity of future growth in Jefferson County.

Method of Population Distribution

The population distribution to subcounty areas starts from the Jefferson County forecast described in the previous section of this report. Of the three 20-year projections, the medium projection was recommended as the County population forecast. Under this forecast, the County population rises from an estimated 24,300 in 1994 to 38,337 in 2014.

The method for distributing the County population forecast to the twelve County planning areas is the Housing Unit Method. Under this method, future housing units are projected and distributed to planning areas, and then converted to population. A flow diagram of the method as applied to Jefferson County is presented on the next page as Figure 12.

There are several advantages to the Housing Unit Method for subcounty population distributions:

• the subcounty population forecast is tied directly to actual housing units permitted and built within each planning area;

• it may be easier for planners and policy makers to visualize growth in housing units in each planning area than the number of people;

• the method utilizes explicit technical assumptions at each stage of developing forecasts;

• the method is considered more reliable than using population growth rates or other techniques.9

There are potential disadvantages as well. Most important are the facts that the method may be more complex than other techniques, and that more assumptions are required for its application.

In Figure 12, the distribution process starts from the total population for the County as a whole, which was developed as described in Section III. For any future year, this population is taken through average household size to get the number of households, or occupied housing units, for the County. Then using an average vacancy rate produces the number of year-round housing units, and the addition of seasonal units leads to a total count of housing units Countywide. In order to get housing units for each planning area, the proportion of building permits in each planning area is applied to the County total. Then the process reverses, and moves from total housing units to year-round housing units to occupied housing units and finally to total population in each planning area.

Each circle in the Figure 12 diagram indicates an assumption or calculation that must be made for each future year distribution to planning areas. These are discussed in detail below.

9 This is the conclusion of a recent article in a professional planning journal, "Evaluating the Housing Unit Method," Journal of the American Planning Association , Vol. 60, No. 2, Spring, 1994.

Assumptions and Calculations for Distributions

Countywide Housing Unit Growth. The first step in the distribution process is developing a projection of total County housing units for each future year in the 20-year planning period. Starting from the County population forecast, this is accomplished with assumptions and calculations on average household size, vacancy rates, and a factor for seasonal units.

Table 4 presents the numbers for this process.

Table 4 JEFFERSON COUNTY HOUSING UNIT PROJECTION 1990-2014.
Year	Total Population	Average HH Size	Occupied Hsg Units	Vacancy Rate	Yr Round Hsg Units	Percent Seasonal	Total Hsg Units
1990	20406	2.365	8627	0.071	9284	0.157	11014
1991	21600	2.366	9129	0.071	9825	0.156	11645
1992	22500	2.352	9565	0.071	10293	0.156	12190
1993	23500	2.356	9976	0.071	10736	0.155	12703
1994	24288	2.362	10284	0.071	11067	0.154	13084
1995	25149	2.362	10649	0.071	11460	0.153	13538
1996	25918	2.359	10987	0.071	11824	0.153	13956
1997	26572	2.354	11289	0.071	12148	0.152	14328
1998	27222	2.347	11598	0.071	12481	0.151	14708
1999	27860	2.339	11909	0.071	12816	0.151	15090
2000	28501	2.331	12229	0.071	13160	0.150	15483
2001	29143	2.324	12543	0.071	13498	0.149	15861
2002	29785	2.317	12855	0.071	13834	0.148	16237
2003	30433	2.311	13170	0.071	14173	0.147	16615
2004	31088	2.306	13482	0.071	14508	0.146	16989
2005	31750	2.300	13803	0.071	14854	0.145	17373
2006	32429	2.295	14132	0.071	15208	0.144	17766
2007	33099	2.289	14458	0.071	15559	0.143	18156
2008	33817	2.283	14813	0.071	15942	0.142	18580
2009	34535	2.275	15179	0.071	16335	0.141	19017
2010	35251	2.266	15560	0.071	16745	0.140	19471
2011	35961	2.256	15938	0.071	17152	0.139	19916
2012	36758	2.247	16356	0.071	17602	0.138	20410
2013	37555	2.239	16772	0.071	18049	0.136	20900
2014	38337	2.232	17177	0.071	18485	0.135	21375

The first row contains U.S. Census data for 1990. Beyond that, average household size declines from 1990, following the rate of decrease projected in forecasts for other parts of Washington State and for the U.S. as a whole. The decrease in Jefferson County in the future is far slower than in the two decades before 1990. In 1980, for example, the average household size for the County was 2.51, then falling to 2.365 in 1990.

The average vacancy rate for year-round housing units in Jefferson County was 7.1 percent in 1990, according to the Census. The conventional approach in projecting over a long period like 20- years is to assume the same average vacancy rate throughout. Vacancy rates rise and fall in cycles, but over the long term they move rather little.

In 1990, the Census found that 15.7 percent of the total housing units in Jefferson County were used seasonally, for vacation purposes, and were not available for year-round occupancy. This is a relatively high proportion for a county's housing stock. It is important, therefore, to be able to factor out seasonal units when moving from the total housing in place or being built to the population living in housing units.

Although the 1980 Census figures on seasonal units were not entirely comparable with those from 1990, the trend in the seasonal proportion appeared to be slightly downward. For the future projections, this moderate decline was continued. While there are no exact guidelines on trends in seasonal unit proportions, as a county grows relative to its state its land and housing become relatively more costly and therefore less likely to be built or remain for seasonal use only. There will certainly be parts of Jefferson County that will be almost as amenable to seasonal units as now, but in others, such as waterfront locations, this may be less so. Many existing units will be converted to full-time occupancy. New seasonal construction will not be as great. The decrease in the seasonal unit proportion in the County from 15.7 percent in 1990 down to 13.5 percent in 2014 implies a decline in the share of new housing that is for seasonal use from 14.5 percent in 1990 to 8.2 percent in 2014.

Using these assumptions and calculations, the projection of total housing units in the County in 2014 is 21,375.

Planning Area Building Permit Trends. The history of new housing unit building permits by County planning area is displayed in Figure 13. The one exception noted is that, because building permit tabulations were not available for the City of Port Townsend before 1986, the data for Planning Area 1 are the annual changes in total housing units compiled by the Office of Financial Management as part of its annual population estimate program. These should be functionally equivalent to building permits, and perhaps a little more conservative.

Most of the County's building activity during the past 14 years occurred in Planning Areas 1,2,4, and 7, or Port Townsend, Quimper Peninsula, Tri-Area, and Port Ludlow, though some of the lower permit numbers were in planning areas (such as Marrowstone Island) that have small bases, so that the impact was still significant.

These relative shares of housing unit building activity over this 1980-93 period were the basis for distributing the annual change in total County housing units down to each planning area. This is a continuation of past trends in growth of housing units. This is a major, but unavoidable, assumption in the population distributions. A modification in this assumption is introduced for the alternative "constrained" projection.

Planning Area Data and Assumptions. Applying the planning area shares of the County housing unit change is the key driver to obtain total housing unit projections for each planning area in future years. In order to move from total housing units to population, the process is to estimate year-round housing units, then occupied housing units, and finally the population living in the housing units.

The process requires some base data on each planning area, and some assumptions about the future trends in those data. The base data are presented in Table 5.

Table 5 PLANNING AREA HOUSING UNIT DATA 1990
Planning	Estimated	Average	Occupied	Vacancy	Yr Round	Seasonal	Total
Area	Population	HH Size	Hsg Units	Rate	Hsg Units	Factor	Hsg Units
1	7001	2.32	3027	0.059	3217	0.02	3280
2	2076	2.15	953	0.055	1005	0.19	1240
3	686	2.20	307	0.058	325	0.17	393
4	3315	2.50	1307	0.058	1383	0.06	1467
5	891	2.55	344	0.055	363	0.13	418
6	1075	2.40	441	0.065	471	0.05	498
7	1111	2.10	521	0.110	584	0.18	708
8	722	2.40	296	0.100	328	0.35	504
9	369	2.30	158	0.075	170	0.40	286
10	1123	2.30	481	0.075	519	0.40	870
11	1090	2.30	467	0.075	503	0.40	845
12	947	2.88	.324	0.221	415	0.18	505
County	20406	2.37	8627	0.071	9284	0.16	11014

These data are estimates derived from 1990 Census data. The Census geographic areas did not correspond to County planning areas, so estimates and adjustments were needed to fit the data to the planning areas. These became the base data for each planning area.

The seasonal unit factors varied greatly by planning area. These base year data were then projected to follow the pattern of slight decline described for the County as a whole. Thus, planning areas with the highest base year seasonal factors drop farther during the 20-year projection period than those with low base seasonal factors. The higher the seasonal factor, the higher the proportion of the total housing stock that is for seasonal use.

The average vacancy rates also varied by planning area, but with the exception of Planning Area 12 (West End) the variation was small. As with the County projection, the vacancy rates for each planning area were held constant throughout the projection period.

There were somewhat greater variations by planning area in the average household size for 1990. The lowest average sizes tended to be in areas with significant numbers of retired persons. Non-urban family areas had the highest average household sizes. The 1990 average for each planning area was projected to follow the County pattern in average household size over the 20-year period, which was generally one of moderate decline. Thus, each planning area would follow the same downward trend, but starting from a different level.

Twenty-Year Projections

Continuation of Trends. Applying these planning area data to the County housing unit projections presented above produces the projections of total housing units and total population by planning area which are summarized in Figures 14 and 15, respectively. Since these projections were developed with a straight application of the 14-year past average shares of County building permit activity, these projections represent a continuation of those past trends into the 20-year future, and are termed the "Continuation of Trends" projection.

The Continuation of Trends projection shows the housing unit growth to be concentrated in the four planning areas (1,2,4, and 7) which dominated the recent building permit trends. Because of different seasonal factors and average household sizes, the population projection is a little different, with population growth more concentrated in Port Townsend and the Tri-Area.

Trends with Constraints. But there are good reasons why the past trends in building activity may not continue indefinitely into the future. One is the matter of local water supply, which is already a significant obstacle to new building in Planning Area 3 (Marrowstone Island). Water supply and septic system difficulties have also occurred in Planning Area 2 (Quimper Peninsula) west of Port Townsend.

In addition to these physical constraints, adoption by Jefferson County of a new GMA comprehensive plan will discourage certain types of new development in non-UGA areas of the County, especially those with rural characteristics and a low inventory of platted lots.

In effect, these are factors which can be expected to alter the path of the future trend in planning area housing unit change from that which existed in the past. In order to account for these differences, a new housing unit and population projection was developed that constrained some of the new development in certain planning areas and redistributed that development proportionally to the anticipated UGAs.

This "Trends with Constraints" projection is summarized in Figures 16 and 17.

In the Trends with Constraints projection, there were several levels of constraint applied to individual planning areas. The most severe was made for Planning Area 3 (Marrowstone Island), to reflect the effects of water supply problems. Moderate-to-severe constraints were applied to Planning Area 2 (Quimper) to reflect the negative effects on future development of water supply and septic difficulties, and to Planning Area 6 (Center/Inland) to reflect effects of the GMA comprehensive plan on development in rural areas. Very light constraints were applied to Planning Areas 5 (Discovery Bay), 10 (Quilcene), and 11 (Brinnon), to reflect the more moderate effects of the GMA comprehensive plan on these rural areas, which contain platted lots and/or potential rural town centers.

The Port Townsend planning area, which is currently the County's only interim UGA, and the Port Ludlow/Oak Bay planning area, which will likely be designated a final UGA, received most of the redistribution of constrained development. The Tri-Area, which also may become a final UGA, remained at about the Trends projection level-in effect a constraint-because of physical and policy concerns about potential housing unit development above the Trends growth level in that area. Some other planning areas, such as Planning Areas 6 (Shine/Paradise Bay) and 9 (Coyle/Toandos) gained just slightly from the imposition of constraints on other rural areas.

The differences between the Continuation of Trends and the Trends with Constraints projections are summarized in Table 6 below.

Table 6 TRENDS WITH CONSTRAINTS PROJECTION DIFFERENCES FROM CONTINUATION OF TRENDS
	Housing Units		Population
	Number	Pct of Chg	Number	Pct of Chg
Planinng Area	Difference	1993-2014	Difference	1993-2014
1. Port Townsend	+647	+31.2	+1215	+29.9
2. Quimper Peninsula	-640	-45.8	-1045	-47.8
3. Marrowstone Island	-166	-55.8	-281	-59.3
4. Tri-Area	-46	-3.2	-105	-3.6
5. Discovery Bay	-19	-8.2	-40	-9.4
6. Center/Inland Valleys	-190	-46.7	-391	-50.9
7. Port Ludlow/Oak Bay	+511	+36.9	+772	+37.8
8. Shine/Paradise Bay	+33	+9.5	+47	+9.2
9. Coyle/Toandos Penin.	+6	+4.9	+7	+4.3
10. Quilcene	-57	-14.3	-77	-14.0
11. Brinnon	-95	-17.5	-128	-17.2
12. West End	+15	+56.3	+27	+166.6

Overall, the application of these constraints to the Continuation of Trends projection leads to modest shifts among planning areas. For example, although the shifts would not necessarily occur only between two planning areas, it could be observed that about 1,000 population is caused by the development constraints to shift from Quimper Peninsula into Port Townsend.

Population Range for Port Townsend

Even the redistribution caused by the application of development constraints produces a very narrow range of future population possible for Port Townsend and the other probable UGA planning areas. But this is with a fixed County population forecast level the medium projection that is recommended as the Countywide forecast. Applying the Continuation of Trends and the Constrained projection to the low and high County projections opens a wider range of population possibilities that the City could conceivably face.

A highest projection for the City was developed by applying the Trends with Constraints distribution to the high County population projection. A lowest City projection was formed with the Continuation of Trends distribution and the low County projection, which is the OFM County forecast, adjusted for the discrepancies in the 1990-94 estimate period. These highest and lowest projections for the City produce a rather wide range of population possibilities for the City, as shown in Figure 18.

The 2014 population projections for the City imply a wide range of growth rates for the population from 1994 to 2014. The lowest projection (10,229) is an average of just 1.3 percent per year, while the highest projection (17,854) averages 4.1 percent growth per year over this period. The Trends projection for the City averages 2.0 percent per year, and the Trends with Constraints projection is an average of 2.5 percent per year through the projection period.

Forecast Conclusion

The Trends with Constraints population projection for the planning areas within Jefferson County is summarized in Table 7 below, starting with the 1990 population estimates. This projection takes into account both the historical trends in development of housing units and population change in each planning area relative to the County as a whole, but also the real physical constraints on further development in the future that are facing some of the planning areas, as well as the potential effects of policy changes likely under the GMA planning program.

Table 7 POPULATION FORECAST FOR COUNTY PLANNING AREAS Trends with Constraints Projection, 1994-2014
	County Planning Areas10
Year	1	2	3	4	5	6	7	8	9	10	11	12
1990	7001	2076	686	3315	891	1075	1111	722	369	1123	1090	947
1994	7940	2721	802	4082	1041	1273	1765	830	395	1249	1231	960
1999	9023	3173	891	4782	1142	1425	2304	954	432	1372	1392	969
2004	10122	3446	943	5399	1231	1511	2891	1082	471	1477	1534	981
2009	11469	3619	966	6026	1319	1565	3600	1217	512	1580	1670	992
2014	13013	3736	976	6705	1410	1601	4468	1367	558	1688	1813	1001

The Trends with Constraints projection in Table 7 is based on the medium population projection for Jefferson County, which was recommended as the 20-year forecast for the County as a whole. The Trends with Constraints projection contains the assumptions and the distribution most likely to occur within the County over this period. This projection is therefore recommended as the 20-year forecast for planning areas within Jefferson County.

10 Planning area names are as follows:

1.	Port Townscnd	7.	Port Ludlow/Oak Bay
2.	Quimper Peninsula	8.	Shine/Paradise Bay
3.	Marrowstone Island	9.	Coyle/Toandos Peninsula
4.	Tri-Area	10.	Quilcene
5.	Discovery Bay	11.	Brinnon
6.	Center/Inland Valleys	12.	West End

The four projections developed for the City of Port Townsend are summarized in Table 8 below. In addition to the Continuation of Trends and the Trends with Constraints projections for all planning areas within the County, a Highest projection and a Lowest projection were developed for the City in order to provide a full range of possible futures for the City's population. As with the rest of the planning areas, the Trends with Constraints projection is recommended as the 20-year forecast for the City.

Table 8 CITY OF PORT TOWNSEND POPULATION PROJECTIONS
Year	Lowest	Continuation of Trends	Trends with Constraints	Highest
1970	5,241	5,241	5,241	5,241
1980	6,067	6,067	6,067	6,067
1990	7,001	7,001	7,001	7,001
1994	7,940	7,940	7,940	7,940
1999	8,353	8,924	9,023	10,037
2004	8,849	9,750	10,122	12,279
2009	9,570	10,797	11,469	14,881
2014	10,229	11,798	13,013	17,854

Fifty-Year Projection

A 50-year population distribution to the Port Townsend and the adjacent planning areas was developed to be consistent with the 20-year future described above. Specifically, the 50-year projection starts from the most-reasonable Trends with Constraints distribution- the recommended 20-year forecast-at 2014 and then takes two alternative directions, based on the high and low 50-year projections at the County level.

As summarized in Table 9, these 50-year projections give a relatively narrow range for population in Port Townsend and- the adjacent planning areas (Quimper Peninsula, Marrowstone Island, and the Tri-Area) for 2044. The narrow range is produced by two factors underlying the projection. First is that the 50-year projections for the County as a whole formed a rather tight range, because of diminishing growth at the State level and because of potential constraints on resources within the County. The second factor is the takeofF point of the same recommended County and planning area forecasts at the year 2014, which was assumed in order to keep the 50-year projections consistent with the 20-year forecast. Finally, the Trends with Constraints approach was assumed for both high and low alternatives in recognition of physical and policy constraints and for consistency in the County distributions between 20-year and 50-year projections.

Table 9 FIFTY-YEAR PROJECTIONS
	Pt Townsend	Planning Areas	Total
Year	1	2,3,4,	1,2,3,4	County
1990	7,001	6,077	13,078	20,406
1994	7,940	7,605	15,545	24,300
2014	13,013	11.417	24.430	38,337
High Projection
2020	15,059	12,474	27,532	43,366
2030	18,240	14,118	32,356	51,187
2040	21,214	15,652	36,866	58,501
2044	22,451	16,291	38,742	61,541
Low Projection
2020	13,995	11,924	25,919	40,751
2030	15,486	12,694	28,180	44,417
2040	16,868	13,408	30,276	47,815
2044	17,395	13,680	31,075	49,111

Thus, the range indicated by these high and low projections is intended to be realistic and consistent with the 20-year population forecast. It does not encompass the widest possible range of population possibilities for Port Townsend and the adjacent planning areas over the entire 50-year period. It is a target for long-range planning for utility needs based on the shorter-range population forecasts for GMA comprehensive planning purposes.

V. SUMMARY AND CONCLUSIONS

County Forecast

The population forecast for Jefferson County as a whole was developed from projections based on the County's shares of the Washington State population for the 65-and-over and the under-65 age groups separately. The assumptions underlying the projections, therefore, were the future year shares of the State's population for each age group and the State forecast itself. In addition, potential constraints on County population growth in the future, such as water supply and infrastructure, were identified and considered.

Three alternative projections were developed. The first was the OFM forecast for Jefferson County, which implied a constant 1993 share of the State's population. The OFM forecast is already tracking very low compared with current 1994 population estimates, and thus is called the low 20-year projection.

The second County projection was based on the average increase in County shares of the State's population over the past 24 years. The average increase in share was applied to both the 65+ and the under-65 age groups.. The State forecast was the December, 1992, OFM forecast, which changed little to the December, 1993, update. This average increase in shares projection produced a total population of 38,337 for Jefferson County in 2014. This became the medium 20-year projection.

The third County projection was based on the peak period increase in County shares of the State's population in the past. The peak increase in shares was applied to both the 65+ and the under-65 age groups, producing a projected total population of 45,227 for the County in 2014. This is considered the high 20-year projection.

All three projections bend upward toward the end of the period, despite a flattening of the increases in County shares of the State's population toward the end of the 20-year period. This is due to the demographics of the aging of the baby boomers contained within the State population forecast. The effect of this rapidly expanding pool of retirement-age people in Washington on the historically desirable retirement counties will be enormous.

The medium projection, which is based on the historical average rate of increase in shares of the State population for both the 65+ and the under-65 age groups, contains the most reasonable assumption of share increase, and thus is the most likely to occur. The medium projection is therefore recommended as the 20-year forecast for Jefferson County.

A 50-year population was developed which is consistent with the 20-year future. The 50-year forecast starts from the most-likely medium projection at 2014 and then takes two alternative directions. One continues the average increase in the County's share of the State's population for the duration of the period. This is termed the high alternative. The other adopts a constant 2014 County share of the State population for the remaining thirty years. This is called the low alternative.

These two alternatives indicate a range of reasonable possibilities for the County population over the 50-year period. Long-term future population growth could reasonably occur anywhere within this range. The high alternative takes population growth to a level which may not be supportable with current ground and surface water resources. The low alternative stays more or less below that level. For a 50-year population growth at the high end, some changes in water supply or demand might be necessary.

Subcounty Distribution

Based on the recommended 20-year County population forecast, distributions of future population to subcounty planning areas were developed. The method for distributing the County population was based on housing unit growth in each planning area. Under this method, future housing units are projected and distributed to planning areas, and then converted to population.

Assumptions and calculations underlying the housing unit approach to population distribution are of three types. First is the countywide housing unit growth, which is based on the population forecast, but includes assumptions on average household size, vacancy rates, and percent of units for seasonal use. Second is the planning area housing unit growth, which was based on the average shares of County building permit activity for each planning area over recent past. Third, there are assumptions on the key factors of average household size, vacancy rates, and percent seasonal units for each planning area, which were based on estimated 1990 data and County trends.

Two alternative distributions were developed. One was the Continuation of Trends projections, in which the recent trends in housing unit growth share of the County were extended throughout the 20-year projection period. The other distribution was the Trends with Constraints, in which the past trends in housing unit growth were modified by considerations of physical and policy constraints on particular planning areas. Where future population growth was constrained, that growth was redistributed to planning areas which will likely include final Urban Growth Areas with the potential to receive the growth. Constraints on planning areas were severe, moderate, or slight, depending on the seriousness of the potential limitations on future housing unit growth.

Under the Continuation of Trends projection, most of the population growth in the County over the 20-year period is concentrated in Port Townsend, the Quimper Peninsula, the Tri-Area, and Port Ludlow/Oak Bay-the areas that have received most of the housing unit growth in recent years. Several other planning areas show growth that is high relative to their existing numbers of housing units and population.

The Trends with Constraints projection concentrates more of the growth in Port Townsend and Port Ludlow/Oak Bay. Constraints reduce particularly the Quimper Peninsula and Marrowstone Island planning areas, as well as the Central/Inland Valleys area. The Tri-Area planning area remains about the same. This projection assumes the existence of three UGAs, one incorporated and two unincorporated, available to receive the diverse types of future population growth in Jefferson County.

The Trends with Constraints projection contains the assumptions and the distribution most likely to occur within the County over the 20-year period. This projection is therefore recommended as the 20-year forecast for planning areas within Jefferson County.

For the City of Port Townsend, two additional projections were developed in order to indicate a total range of possibilities for the 20-year growth of the City. To do this, the assumptions and calculations of the Continuation of Trends distribution were applied to the low (OFM) Countywide projection, and the Trends with Constraints distribution was applied to the high Countywide projection. The indicated range for City population in 2014 was from 10,229 on the low end to 17,854 on the high end, or from a low growth rate averaging 1.3 percent per year to a high growth rate averaging 4.1 percent per year.

A 50-year population distribution was made for Port Townsend and the adjacent planning areas (Quimper Peninsula, Marrowstone Island, and the Tri-Area). In order to be consistent with the 20-year forecast, the 50-year projection starts from the most-reasonable Trends with Constraints distribution at 2014 and then takes two alternative directions, based on the high and low 50-year projections at the County level. These alternatives produce a relatively narrow range for population in Port Townsend and the adjacent planning areas in 2044. For Port Townsend, the high projection is 22,451 for 2044 and the low projection is 17,395. The narrowness of the range is due to the tightness of the County population range for 2044 and to the use of the Trends with Constraints approach for both 50-year projections.

Recommendations

For the Countywide population forecast, the medium projection of 38,337 in the year 2014 is recommended for adoption. This projection series is based on the most reasonable assumption of the change in the County share of the State's population, and thus is the most likely to occur. It is a reasonable, defensible projection to use as a 20-year population forecast.

For the subcounty population distribution, the Trends with Constraints projection contains the assumptions and the distribution most likely to occur within the County over the 20-year period. It is based both on past trends in housing development and on considerations of physical and policy constraints that will potentially affect particular planning areas. This distribution for the County planning areas to 2014 is recommended for adoption as the 20-year population forecast.

For the City of Port Townsend, possible population futures ranged from a low of 10,229 to a high of 17,854, as of the year 2014. These represent average annual growth rates over the 1994-2014 period of 1.3 percent to 4.1 percent. The Trends with Constraints projection is a population of 13,013 in 2014, or an average of 2.5 percent growth per year. This projection is recommended for adoption as the 20-year population forecast.

One important caveat for these recommendations is that if the three assumed UGAs are not formally designated (that is, if Port Townsend is the only UGA) then the recommended population distribution, and perhaps also the total County forecast, is at risk.

APPENDIX A
SUPPLEMENTARY TABLES

Table A-1 OFM POPULATION FORCAST FOR WASHINGTON STATE, December, 1992

	Total	Percent	65 and Over	Percent		Total	Percent	65 and Over	Percent
Year	Population	Change	Population	of Total	Year	Population	Change	Population	of Total
1970	3,413,250	0.7%	327,160	9.6%	1995	5,467,043	2.2%	630,475	11.5%
1971	3,436,300	0.7%	334,560	9.7%	1996	5,560,504	1.7%	640,695	11.5%
1972	3,430,300	-0.2%	341,900	10.0%	1997	5,630,192	1.3%	646,516	11.5%
1973	3,444,300	0.4%	350,480	10.2%	1998	5,698,418	1.2%	652,978	11.5%
1974	3,508,700	1.9%	359,380	10.2%	1999	5,764,701	1.2%	658,898	11.4%
1975	3,597,890	1.7%	368,510	10.3%	2000	5,830,092	1.1%	666,264	11.4%
1976	3,634,891	1.9%	378,550	10.4%	2001	5,895,849	1.1%	673,836	11.4%
1977	3,715,375	2.2%	391,420	10.5%	2002	5,961,647	1.1%	681,459	11.4%
1978	3,836,200	3.3%	404,360	10.5%	2003	6,027,642	1.1%	690,463	11.5%
1979	3,979,000	3.7%	418,340	10.5%	2004	6,093,862	1.1%	700,944	11.5%
1980	4,132,353	3.8%	431,562	10.4%	2005	6,160,339	1.1%	712,746	11.6%
1981	4,229,278	2.3%	445,045	10.5%	2006	6,227,071	1.1%	728,156	11.7%
1982	4,276,549	1.1%	457,651	10.7%	2007	6,293,641	1.1%	742,099	11.8%
1983	4,307,247	0.7%	472,024	11.0%	2008	6,360,109	1.1%	765,807	12.0%
1984	4,354,067	1.1%	484,518	11.1%	2009	6,426,736	1.0%	789,350	12.3%
1985	4,415,785	1.4%	499,349	11.3%	2010	6,493,491	1.0%	812,804	12.5%
1986	4,462,212	1.1%	514,298	11.5%	2011	6,560,636	1.0%	834,896	12.7%
1987	4,527,098	1.5%	529,516	11.7%	2012	6,628,182	1.0%	873,232	13.2%
1988	4,616,886	2.0%	543,773	11.8%	2013	6,696,211	1.0%	911,327	13.6%
1989	4,728,077	2.4%	557,965	11.8%	2014	6,764,498	1.0%	946,706	14.0%
1990	4,866,663	2.9%	571,403	11.7%
1991	5,000,371	2.7%	583,511	11.7%
1992	5,116,671	2.3%	595,461	11.6%
1993	5,240,900	2.4%	607,499	11.6%
1994	5,350,293	2.%	620,034	11.6%	2020	7,170,607		1,193,822	16.6%

Table A-2. HOUSING UNIT PERMITS BY PLANNING AREA, 1980-1993

	Port	Quimper	Marrowstn		Discovery	Center/	Pt. Ludlow/	Shine/	Coyle/
	Townsend	Peninsula	Island	Tri-Area	Bay	Inland	Oak Bay	Paradise	Toandos	Quilcene	Brinnon	West End	County
Year	1	2	3	4	5	6	7	8	9	10	11	12	Total
1980	52	65	19	67	10	14	29	8	13	10	20	1	308
1981	20	30	9	29	6	7	27	9	3	10	27	1	178
1982	35	8	9	35	6	13	17	6	8	12	10	2	161
1983	58	20	8	28	6	12	22	7	1	8	17	0	0187
1984	64	27	9	39	3	8	11	9	3	13	17	3	206
1985	52	28	4	41	5	12	13	13	3	10	28	0	209
1986	129	25	6	36	4	14	15	16	0	012	29	0	286
1987	142	45	7	40	4	15	22	20	8	13	16	2	334
1988	31	41	14	42	4	7	18	14	3	13	16	2	205
1989	88	56	11	72	8	17	135	23	6	17	12	2	447
1990	110	95	15	105	9	19	125	21	3	19	20	1	542
1991	130	115	22	70	6	26	111	16	6	20	31	1	554
1992	117	95	18	83	45	24	99	21	4	20	24	1	551
1993	107	97	13	91	11	29	96	12	4	33	29	0	522
Total	1135	747	164	778	127	217	740	195	65	210	296	16	4690

Source: For Planning Areas 2-12, Jefferson County Permit Center, Building Division, Annual Report, 1993.
For the City of Port Townsend, OFM Population Trends for Washington State, 1980-93.

Table A-3. POPULATION FOR PLANNING AREAS IN JEFFERSON COUNTY, based on Trends with Constraints, 1990-2014

	Port	Quimper	Marrowstn		Discovery	Center/	Pt. Ludlow/	Shine/	Coyle
	Townsend	Peninsula	Island	Tri-Area	Bay	Inland	Oak Bay	Paradise	Toandos	Quilcene	Brinnon	West End	County
Year	1	2	3	4	5	6	7	8	9	10	11	12	Total
1990	7001	2076	686	3315	891	1075	1111	722	369	1123	1090	947	20406
1991	7230	2274	720	3608	917	1127	1350	762	376	1158	1127	952	21600
1992	7530	2434	751	3734	926	1172	1495	781	382	1181	1162	953	22500
1993	7740	2594	783	3922	1021	1224	1649	813	389	1212	1197	957	23500
1994	7940	2721	802	4082	1041	1273	1765	830	395	1249	1231	960	24288
1995	8216	2847	821	4244	1061	1319	1884	850	401	1281	1266	959	25149
1996	8436	2948	843	4397	1085	1353	1995	879	410	1308	1302	962	25918
1997	8628	3029	861	4527	1105	1379	2094	904	418	1330	1333	965	26572
1998	8824	3104	877	4656	1124	1403	2198	929	425	1352	1363	967	27222
1999	9023	3173	891	4782	1142	1425	2304	954	432	1372	1392	969	27860
2000	9227	3238	904	4908	1160	1446	2415	979	440	1393	1420	971	28501
2001	9437	3298	916	5032	1178	1465	2529	1005	447	1414	1449	973	29143
2002	9653	3352	926	5155	1196	1482	2646	1031	455	1436	1478	976	29785
2003	9876	3402	936	5279	1214	1498	2767	1057	463	1457	1506	979	30433
2004	10122	3446	943	5399	1231	1511	2891	1082	471	1477	1561	981	31088
2005	10374	3486	949	5520	1248	1523	3021	1108	479	1498	1588	984	31750
2006	10635	3523	955	5644	1266	1535	3156	1135	487	1518	1614	986	32429
2007	10900	3555	959	5765	1283	1545	3295	1161	495	1538	1670	989	33099
2008	11183	3589	963	5895	1301	1556	3445	1189	504	1559	1696	991	33817
2009	11469	3619	966	6026	1319	1565	3600	1217	512	1580	1723	992	34535
2010	11758	3646	968	6155	1336	1573	3760	1245	521	1600	1754	993	35251
2011	12047	3667	970	6282	1353	1579	3922	1273	529	1620	1784	995	35961
2012	12366	3696	973	6425	1372	1588	4100	1305	539	1643	1813	997	36758
2013	12689	3720	976	6567	1391	1596	4283	1336	549	1666	1784	999	37555
2014	13013	3736	976	6705	1410	1601	4468	1367	558	1688	1813	1001	38337

APPENDIX B
SUMMARY OF PROJECTION METHODOLOGY

Several different population forecasts are described in this report. They include 20-year forecasts for Jefferson County as a whole, for planning areas within the County, and for the City of Port Townsend, as well as 50-year forecasts for the County and the Port Townsend area. Different methods were used to develop each of these forecasts. These various methods are detailed in this appendix.

Jefferson County

The projection of the County population through the 20-year future is based on the County's share of the population in Washington State as a whole. In particular, we examine the trend of past County shares of the State population, and then make an assumption about how that trend in share will change in the future. This is done not just for the total population, but for the population disaggregated into a 65-and-over age group and an under-65 age group.

There are several reasons why the share of State population method is appropriate for projecting Jefferson County population:

• it relates the County population to the source (through migration) of much of the growth in the County, for both 65+ and under-65 age groups;

• it makes use of the demographic forecasting done at the State level, including detailed data on births, deaths, and migration patterns by age, and a powerful demographic forecasting model;

• it provides a check or benchmark on the County forecasts, so that they will remain in a reasonable range relative to State patterns;

• it gets away from projecting growth rates from some arbitrarily selected period in the past.

In this project, we look at the County's shares of the State's population for the 65+ age group and the under-65 age group separately, since the patterns of growth in these two groups were shown to be so different, and since we know that Statewide and nationwide the relative numbers of these groups will be changing dramatically out in the future.

We also develop three alternative projections for each group's share of the State population for that group.

• the low projection is the constant 1993 share assumption implied in the OFM population forecast for the County;

• the medium projection takes the average rise (for each age group) from the 1970-93 period, with a flattening of the rise over the forecast period;

• the high projection uses a higher than the average rise from the past, with a flattening of the rise over the forecast period.

Together, these alternative projections form a range that can be considered to bracket the reasonable possibilities for future changes in the County's share of the State's population. When combined with a State population forecast, they lead to a range of population forecasts for the County.

The key assumptions for the projection of the Countywide population include (1) the future change in share for the 65+ and under-65 age groups in Jefferson County, and (2) the State forecast on which the shares are based. In addition, there is the matter of potential constraints on the ability of the County to continue to absorb additional population over long periods into the future.

Future Change in Share. As described above, the alternative projections for the County population contain markedly different assumptions about the share of the State's population the County will have in the future. Which one or ones are the most reasonable? Obviously, the OFM assumption of no future change in share, after more than 20 years of increase, is unreasonable. But what rate of increase is appropriate?

Rapid rates of increase are difficult to maintain over long periods of time. If anything, they are more likely to diminish than to accelerate. Thus, a higher increase in the County share of the State's population in the future than in the past is possible but not probable. Thus, the medium projection track-increase in future shares at the average rate of the past, with some diminishing of the increase over the long term-is the most reasonable assumption. The high projection, with a higher than historic increase in share of the State's population, is useful to establish a range of possibilities, but should not be rated as a probable future.

Applying these share change assumptions separately to the 65+ and the under-65 age groups allows the population in each age group to grow independently, depending on their relative scale in the State in the long-term future. This builds in the demographics of aging and migration in the State population and the effect this may have on Jefferson County.

Future State Population. The other key assumption in these projections is the growth in the State's population in the future, in both the 65+ and the under-65 age groups. Besides the county forecasts prepared for the GMA planning process in 1992, the State OFM develops a long-term demographic forecast for Washington State as a whole, updated annually. These forecasts use a complex demographic model that includes detailed data on births and deaths by age in the population, along with a variety of detailed migration information. Its future projections are determined by the aging of the population and the best available estimates of migration patterns among states. The resulting forecasts are not perfect and will not be target, but certainly are reasonable and build in a great deal of important information about the future.

The State population forecast that the changes in shares are applied to is the one published by the OFM in December, 1992. This is the most recent annual forecast available in fully published form. A review of the December, 1993, forecast numbers found relatively few changes from 1992.1 Therefore, the OFM State forecast at the end of 1992 was used in this forecasting work.

Constraints in Jefferson County. There is considerable concern on the part of local government staff and elected officials that there are significant constraints to the ultimate amount of population that can be accommodated in Jefferson County. These concerns focus on (1) water supply, (2) infrastructure requirements, and (3) the job base.

The water supply for County population and businesses is composed of Port Townsend's surface supply from the Quilcene River and community and individual wells tapping ground water. It should be acknowledged, however, that there are serious constraints to providing future water service to certain areas of the County. In particular, the Tri-Area's water supply system is currently at or near capacity, and significant capital expenditures will be necessary to serve future growth. Additionally, and irrespective of infrastructure limitations, the ultimate population which can be sustained using both surface and groundwater sources has yet to be determined.

Undoubtedly, serving continued population growth will require substantial investment in water, wastewater, and road infrastructure. Moreover, renovation and improvement of existing systems which are deficient will be costly and difficult to achieve. Although the problems associated with providing expanded infrastructure are a valid concern, they are less a constraint to growth than a financial problem associated with growth. In any event, these are decisions to be made by policy makers during the GMA planning process.

Reference has already been made to the possibly changing relationship in Jefferson County between the traditional job base and population growth. The growth of the County population, especially that of retirees, is less connected to job opportunities in the County than ever before. More and more, the population growth is creating its own economic base. This trend is likely to continue into the future. Any lack of expansion in the traditional basic economy is therefore not going to stop or even retard the potential for population growth in the County.

1 This conclusion was reinforced by comments from OFM staff in recent telephone conversations.

Subcounty Population Distribution

The population distribution to subcounty areas starts from the Jefferson County forecast described above. Of the three 20-year projections, the medium projection was recommended as the County population forecast. Under this forecast, the County population rises from an estimated 24,300 in 1994 to 38,337 in 2014:

The method for distributing the County population forecast to the twelve County planning areas is the Housing Unit Method. Under this method, future housing units are projected and distributed to planning areas, and then converted to population. (A flow diagram of the method as applied to Jefferson County is presented as Figure 12 on page 24 in the report).

There are several advantages to the Housing Unit Method for subcounty population distributions:

• the subcounty population forecast is tied directly to actual housing units permitted and built within each planning area;

• it may be easier for planners and policy makers to visualize growth in housing units in each planning area than the number of people;

• the method utilizes explicit technical assumptions at each stage of developing forecasts;

• the method is considered more reliable than using population growth rates or other techniques.2

There are potential disadvantages as well. Most important are the facts that the method may be more complex than other techniques, and that more assumptions are required for its application.

In Figure 12, the distribution process starts from the total population for the County as a whole, which was developed as described above. For any future year, this population is taken through average household size to get the number of households, or occupied housing units, for the County. Then using an average vacancy rate produces the number of year-round housing units, and the addition of seasonal units leads to a total count of housing units Countywide. In order to get housing units for each planning area, the proportion of building permits in each planning area is applied to the County total. Then the process reverses, and moves from total housing units to year-round housing units to occupied housing units and finally to total population in each planning area.

2 This is the conclusion of a recent article in a professional planning journal, "Evaluating the Housing Unit Method," Journal of the American Planning Association, Vol. 60, No. 2, Spring, 1994.

Each circle in the Figure 12 diagram indicates an assumption or calculation that must be made for each future year distribution to planning areas. These are discussed in detail below.

Countywide Housing Unit Growth. The first step in the distribution process is developing a projection of total County housing units for each future year in the 20-year planning period. Starting from the County population forecast, this is accomplished with assumptions and calculations on average household size, vacancy rates, and a factor for seasonal units.

U.S. Census data for 1990 was the baseline for projections of all County housing data (these are presented in Table 4 on page 25). From 1990, average household size is assumed to decline, following the rate of decrease projected in forecasts for other parts of Washington State and for the U.S. as a whole. The decrease in Jefferson County in the future is far slower than in the two decades before 1990. In 1980, for example the average household size for the County was 2.51, then falling to 2.365 in 1990.

The average vacancy rate for year-round housing units in Jefferson County was 7.1 percent in 1990, according to the Census. The conventional approach in projecting over a long period like 20 years is to assume the same average vacancy rate throughout. Vacancy rates rise and fall in cycles, but over the long term they move rather little.

In 1990, the Census found that 15.7 percent of the total housing units in Jefferson County were used seasonally, for vacation purposes, and were not available for year-round occupancy. This is a relatively high proportion for a county's housing stock. It is important, therefore, to be able to factor out seasonal units when moving from the total housing in place or being built to the population living in housing units.

Although the 1980 Census figures on seasonal units were not entirely comparable with those from 1990, the trend in the seasonal proportion appeared to be slightly downward. For the future projections, this moderate decline was continued. While there are no exact guidelines on trends in seasonal unit proportions, as a county grows relative to its state its land and housing become relatively more costly and therefore less likely to be built or remain for seasonal use only. There will certainly be parts of Jefferson County that will be almost as amenable to seasonal units as now, but in others, such as waterfront locations, this may be less so. Many existing units will be converted to full-time occupancy. New seasonal construction will not be as great. The decrease in the seasonal unit proportion in the County from 15.7 percent in 1990 down to 13.5 percent in 2014 implies a decline in the share of new housing that is for seasonal use from 14.5 percent in 1990 to 8.2 percent in 2014.

Using these assumptions and calculations, the projection of total housing units in the County in 2014 is 21,375.

Planning Area Building Permit Trends. As shown in Figure 13 on page 27, most of the County's building activity during the past 14 years occurred in Planning Areas 1, 2, 4, and 7, or Port Townsend, Quimper Peninsula, Tri-Area, and Port Ludlow, though some of the lower permit numbers were in planning areas (such as Marrowstone Island) that have small bases, so that the impact was still significant.

These relative shares of housing unit building activity over this 1980-93 period were the basis for distributing the annual change in total County housing units down to each planning area. This is a continuation of past trends in growth of housing units. This is a major, but unavoidable, assumption in the population distributions. A modification in this assumption is introduced for the alternative "constrained" projection.

Planning Area Data and Assumptions. Applying the planning area shares of the County housing unit change is the key driver to obtain total housing unit projections for each planning in future years. In order to move from total housing units to population, the process is to estimate year-round housing units, then occupied housing units, and finally the population living in the housing units.

The process requires some base data on each planning area, and some assumptions about the future trends in those data. The base data are presented in Table 5 on page 28. The data are estimates derived from 1990 Census data. The Census geographic areas did not correspond to County planning areas, so estimates and adjustments were needed to fit the data to the planning areas. These became the base data for each planning area.

The seasonal unit factors varied greatly by planning area. These base year data were then projected to follow the pattern of slight decline described for the County as a whole. Thus, planning areas with the highest base year seasonal factors drop farther during the 20-year projection period than those with low base seasonal factors. The higher the seasonal factor, the higher the proportion of the total housing stock that is for seasonal use.

The average vacancy rates also varied by planning area, but with the exception of Planning Area 12 (West End) the variation was small. As with the County projection, the vacancy rates for each planning area were held constant throughout the projection period.

There were somewhat greater variations by planning area in the average household size for 1990. The lowest average sizes tended to be in areas with significant numbers of retired persons. Non-urban family areas had the highest average household sizes. The 1990 average for each planning area was projected to follow the County pattern in average household size over he 20-year period, which was generally one of moderate decline. Thus, each planning area would follow the same downward trend, but starting from a different level.

Applying these planning area data to the County housing unit projections produces the projections of total housing units and total population by planning area. Since these projections are developed with a straight application of the 14-year past average shares of County building permit activity, these projections represent a continuation of those past trends into the 20-year future, and are termed the "Continuation of Trends" projection.

But there are good reasons why the past trends in building activity may not continue indefinitely into the future. One is the matter of local water supply, which is already a significant obstacle to new building in Planning Area 3 (Marrowstone Island). Water supply and septic system difficulties have occurred in Planning Area. 2 (Quimper Peninsula) west of Port Townsend.

In addition to these physical constraints, there will be a GMA comprehensive plan, which, although the plan's policy content is far from established, is likely to discourage certain types of new development in non-UGA areas of the County, especially those with rural characteristics and a low inventory of platted lots.

In effect, these are factors which can be expected to alter the path of the future trend in planning area housing unit change from that which existed in the past. In order to account for these differences, a second housing unit and population projection constrains some of the new development in certain planning areas and redistributed that development proportionally to assumed UGAs, and is termed the "Trends with Constraints" projection.

Population Range for Port Townsend

Even the redistribution caused by the application of development constraints produces a very narrow range of future population possible for the City of Port Townsend (and the other assumed UGA planning areas). But this is with a fixed County population forecast level, the medium projection that is recommended as the Contywide forecast. Applying the Continuation of Trends and the Constrained projection to the low and high County projections opens a wider range of population possibilities that the City could conceivably face.

A highest projection for the City is developed by applying the Trends with Constraints distribution to the high County population projection. A lowest City projection is formed with the Continuation of Trends distribution and the low County projection, which is the OFM County forecast, adjusted for the discrepancies in the 1990-94 estimate period. These highest and lowest projections for the City produce a rather wide range of population possibilities for the City.

Fifty-Year Projections

The 50-year projection for Jefferson County as a whole starts from the most-likely medium County projection-the recommended 20-year forecast-at 2014 and then takes two alternative directions, both based on shares of the State's population.

First, since the OFM population forecast for the State (December, 1992) did not go beyond 2020, it is necessary to extend the forecast to 2044. The OFM State forecast to 2020 exhibits a diminishing rate of growth for the population in the later years of the forecast period. The curve of this growth rate is then extended beyond 2020 to the horizon year of 2044, thus continuing the diminishing rate of growth for the State's population.

Then, two alternative directions for the County's share of the State's population are developed. One direction continues the average increase in the County share of the State's population throughout the extended projection period. This is a generous assumption in the face of a slowing State population growth. This is termed the high alternative. The other alternative adopts a constant 2014 County share of the State's population for the remaining thirty years. While this represents a break with the pre-2014 increases in share, it is a moderate, conservative assumption of the long-term future of the County's population. This is called the low alternative.

A 50-year population distribution to the Port Townsend and adjacent planning areas is developed to be consistent with the 50-year County projection and with the 20-year population distribution to planning areas. Specifically, the 50-year distribution at 2014 and then takes two alternative directions, based on the high and low 50-year projections at the County level.

These 50-year projections give a relatively narrow range for population in Port Townsend and the adjacent planning area (Quimper Peninsula, Marrowstone Island, and the Tri-Area) for 2044. The narrow range is produced by two factors underlying the projection. First is that the 50-year projections for the County as a whole formed a rather tight range, because of diminishing growth at the State level and because of potential constraints on resources within the County. The second factor is the takeoff point of the same recommended County and planning area forecasts at the year 2014, which was assumed in order to keep the 50-year projections consistent with the 20-year forecast. Finally, the Trends with Constraints approach was assumed for both high and low alternatives in recognition of physical and policy constraints and for consistency in the County distributions between 20-year projections.

APPENDIX C
CONSULTANT QUALIFICTIONS

TIME WATERSON is President of the Watterson West Group, Inc., a consulting firm located in Seattle. The consulting group specializes in data collection and analysis, forecasting, modeling, and information systems at the regional and urban scale for economic, demographic, land use, transportation, and housing projects.

Dr. Watterson was associated with the Puget Sound Council of Governments (now the Puget Sound Regional Council) and its technical program from 1979 to 1991, as Regional Economist, then as Forecasting Manager, and from 1986 as Director of Technical Services. In these roles, he worked closely with local and state governments, transit and other public agencies, and the research and data users in the region. Besides personally conducting numerous modeling, forecasting, data collection, and analysis projects, he was responsible for developing several new data and reporting programs, including the Puget Sound Transportation Panel survey, the subcounty current employment data reporting program, and the linked application of land use and travel demand models.

He holds a Ph.D. from the University of Pennsylvania, and an M.R.P. from the University of North Carolina. He is a member of the Urban and Regional Information Systems Association, the American Economic Association, the Regional Science Association, and two committees of the Transportation Research Board-the Committee on Transportation and Land Development and the Committee on Transportation Planning Applications. He is also a member of the Seattle Economists Club, the Seattle-Everett Real Estate Research Committee, and the University of Washington Urban Design and Planning Department's Professionals Council.

Project work in the area of forecasting includes:

• Subcounty Forecasting Guide for Washington State. The Watterson West Group, under contract to the Washington State Department of Community Development, developed a guidebook for conducting subcounty population forecasting under the Growth Management Act of 1990. The guide, published as Predicting Growth and Change in Your Communities: A Guide to Subcounty Population Forecasting, is intended to assist counties and cities in understanding the role of forecasts in comprehensive planning, in organizing an effective forecasting process, in evaluating assumptions, and in actual data preparation and forecasting for their areas.

• PSRC 1992 Subarea Forecasts. The Watterson West Group (under subcontract to Dick Conway & Associates) prepared updated long-range population and employment forecasts for the Puget Sound Regional Council, covering forecast and analysis zones (FAZ) throughout King, Kitsap, Pierce, and Snodhomish Counties. The project contained a substantial process component, including briefings, outreach, and surveys, as well as data base development and model development and application. The baseline forecasts (2000, 2010, 2020), combined with Vision 2020 alternative forecasts, were designed to assist in growth management planning programs.

• Burlington 1993 Employment Forecasts. The Watterson West Group (under subcontract to William Popp and Associates) prepared a set of long-range forecasts of employment by industry sector for the comprehensive planning process of the City of Burlington. The project included estimation of current employment, a 20-year employment forecast, and a projection of employment at full buildout-all for the City as a whole and for each traffic zone within the City. The forecasting utilized a variety of information from county, state, and city sources.

• Seattle City Light Economic and Demographic Modeling and Forecasting. In this contrcted project, Dr. Watterson developed and economic and demographic modeling system and used it to prepare long-term forecasts for use in the Seattle City Light load forecasting models. The modeling system was based on an adaptation of the STEP regional econometric model and of the DRAM/EMPAL activity model for the SCL service area. The forecasts were more internally consistent and more specific both to the SCL service area and the independent variables required for the biennial load forecasting process than previous economic/demographic inputs.