Housing Dynamics Theory       The literature suggests the process of gentrification consists of three key elements that can be analyzed using national census data. One of these elements is a rising occupancy status. This is indicated by the increase in the proportion of owner-occupied houses in the total housing stock. The assumed trend representing gentrification is an indicator of housing to move from rental occupancy to owner-occupancy (Logan, 1982).       In order to represent shifts in owner-occupancy, an index was created by dividing owner-occupancy dwelling counts for one year by the rental occupancy for the same year. A value less than one indicates renter occupancy dominates the study area where as a value greater than one indicated owner-occupancy is the predominant housing type. Data and Method       Census Canada data from the years 1981 to 1996 was used for the small scale and medium scale analysis. Private dwelling counts, owner-occupancy counts, and rental-occupancy counts were used. At the largest scale, the data for sub-areas was compiled by Statistics Canada but broken down into sub-areas by the City of Vancouver. At the smallest scale, the census sub-division data of the Downtown Eastside (DTES) as a whole is compared with Central Vancouver (CVAN) for the years 1981 - 1996. The next stage in the analysis is a comparison of the three census tracts within the DTES. The largest scale analysis is a comparison of six sub-areas within the DTES. Findings Sub-Division Level       At the smallest scale, the census sub-division data, of the Downtown Eastside (DTES) is compared with Central Vancouver (CVAN) for the years 1981 - 1996. Overall, the housing index shows a decrease in DTES of 14% and CVAN by 10%. When examined in detail, both decreases occurred between the years 1981 to 1986. Both census sub-divisions are characterized by dominant rental occupancy (figure 8.1) as a percentage of total dwellings, owner-occupancy decreased for DTES and CVAN by 11% and 3% respectively. Rental occupancy units increased for DTES and CVAN by 10.8% and 3.1%. Census Tract Level       The increase in the number of private dwellings between 1981 and 1996 was consistent for census tracts 57 and 58, with increases of 1715 and 1770, but tract 59.01 showed an increase of 5,850 dwellings (Figure 8.2). This is an increase of approximately 230% over the other two tracts.       Although census tract 57 showed an increase in the number of dwellings, the proportion of owner-occupancy dwellings within this tract has increased 7.8% from 1981 to 1996. The percentage of private dwellings occupied by renters increased by approximately the same amount - 7.7%. The index ratio of homeowners to renters decreases by 12% within this census tract, with values of 0.28, 0.18, 0.24 and 0.16. This index shows this tract is still dominated by private rental units and does not demonstrate the shift towards owner-occupancy dwellings as expected during gentrification. In fact, the number of homeowners has decreased while the number of rental units has increased.       The trends in tract 58 are very similar to trends seen in tract 57. The number of owner-occupancy dwellings within tract 58 decreased 6.4% in contrast with an increase of 5.6% in the number of private rental units. The index ratio for this census tract also indicates a high proportion of renters vs. homeowners, although the proportion is much higher than in census tract 57. The index ratios are 0.08, 0.03, 0.23, 0.01 - indicating almost 100% rental occupancy. According to this index, gentrification is not occurring within this census tract.       Tract 59 experienced the most growth in the number of private dwellings within the time period studied. Although 5,850 new private dwellings came into existence, the balance of owner-occupancy to rental units did not shift significantly. Data analysis detected a 1.4% increase in owner-occupancy dwellings and a 1.3% decrease in rental occupancy. There is no clear indication from these numbers that gentrification is occurring within this census tract. Again, although the index ratio values fluctuate by 10%, they are very low -- 0.10, 0.02, 0.12, and 0.12. It is possible that gentrification is occuring here at a low level.       One event that is quite evident in this census data is the sharp decrease in rental units during 1991 for all the census tracts. This reflects the 'purge' of local residents from their homes to accomodate Expo 1986 tourists (Blomley, 2000). Sub-area Level       During the time period 1986 to 1996, the number of private dwellings increased for five zones within the DTES and decreased in Chinatown. Chinatown lost 65 units within this time period. This can be broken down to the complete loss of the 20 owner-occupancy dwellings in existence in 1986. The housing index for Chinatown is zero for all three years census data was available. In 1996 it was comprised completely of 565 rental occupancy units.       Thorton Park (Railway) is completely dominated by rental-occupancy units, with the number of units increasing from 155 to 230 between 1986 to 1996.       Oppenheimer's housing index has remained constant from 1986 to 1996 at 0.2. Again, this reflects 98% renter-occupancy. It is interesting to note that this zone experienced an increase of 1110 in private dwellings -- this increase occurred only in rental-occupancy units. This increase is not reflected in the housing index because the proportion of renter-occupancy from 1986 to 1996 does not change.       The housing index in Gastown is essentially zero for all three years. The area is 99% renter-occupancy.       Victory Square has a 5.7% decrease in the number of renter-occupancy units while owner-occupancy rates increase to 5.2%. This shift is not sufficient to alter the housing index significantly -- values range from 0.03 to 0.09. Conclusion       Our current knowledge about gentrification as an ongoing urban process is limited because researchers have relied on several idealized sequential stage theories or models of how gentrification is assumed to unfold through time and space (Engels, 1999). This approach has some serious limitations because it does not take into account how the various elements of the process -- economic, social, and political -- are brought together in the processes. It assumes simple cause and effect relationships, such as an inverse relationship between the tenure status of in-moving gentrifiers (owner-occupancy) and the gentrified out-movers (renter-occupancy). In addition, the question of who the homeowners and renters are (local, non-local)and what they intend to do with the property are not considered.       Engels (1999) study in Glebe contradicts the standard view that as gentrification proceeds, renters are displaced by higher-income home owners. He found that "the shift away from private landlordism to owner-occupancy resulted in only an 11.5% decline of private landlordism over a 25 year period. Also, a large private rental sector persisted despite active gentrification processes occurring. Thirdly, Glebe had a consistent history of owner-occupancy with 50% in 1960 to 62.4% in 1984.       Several of these features mentioned by Engels are apparent in the DTES. In Vancouver's DTES, there is no shift towards owner-occupancy at any level, and this would lead one to believe, according to the housing index, that gentrification is not occurring in this area. In fact, all scales examined show a very high percentage of rental-occupancy. In the DTES, we suspect from our fieldwork conducted that the gentrifiers are not owner-occupancy, but are 'upscale' renter-occupancy. This type of gentrification is not detected using a housing index along. In order for the housing index to have any meaning, it needs to be combined with other data that represent who the renters and homeowners are. This type of data consists of land assessment values, occupation, education, rent, and urban amenities. Error       Two assumptions were made in the above analysis. The industrial sub-zone area was not included because it only contained three residential units. In addition, the census tract boundaries extended further west than the DTES political boundaries, but this was not considered a problem because there were no residential units (according to our real estate data) in this area. Any error in the Census Canada data is documented in the appropriate Census Canada document. Rounding errors were noticed in the Census data, but these never exceeded 5 dwelling units and therefore were considered minor.