Journal of Population Research

, Volume 30, Issue 1, pp 19–37

Understanding stalling demographic transition in high-fertility countries: a case study of Guatemala

Authors

    • Department of GeographyUniversity of Utah
  • Stuart H. Sweeney
    • Department of GeographyUniversity of California
Article

DOI: 10.1007/s12546-012-9094-5

Cite this article as:
Grace, K. & Sweeney, S.H. J Pop Research (2013) 30: 19. doi:10.1007/s12546-012-9094-5
  • 172 Views

Abstract

The debate surrounding the use of period, cohort, and tempo-adjusted measures has framed most of the recent studies evaluating the utility of macro-level fertility indicators. Period measures are susceptible to distortions, due to birth timing changes, but there is currently no universally accepted adjustment technique. Recent comparative analyses have offered some insights but only as applied to the low-fertility developed world setting. The utility of different types of measures in the high fertility context is unclear. Furthermore, regional variation in the pace of fertility transition is characteristic of many less developed countries and is rarely incorporated into macro-level analyses. The purpose of this analysis is to evaluate macro-fertility indicators at the regional and national levels in a high-fertility country, Guatemala, using the four most recent survey data sets. The results support the use of macro-level period indicators and adjusted period indicators of fertility in developing country contexts.

Keywords

PeriodCohortFertilityArea analysisGuatemala

Introduction

Most countries in Central and Latin America experienced a notable and dramatic fertility decline during the last quarter of the twentieth century. In general, once the fertility decline is initiated the transition from high fertility to lower (and in some cases, replacement) fertility proceeds without interruption. Recent research, however, has revealed that modern transitioning societies do not always follow a straightforward trajectory of fertility decline; rather, some societies appear to attain a certain level of fertility before fertility levels stop decreasing and fertility decline stalls (Bongaarts 2008). In Central America, Guatemala stood out during the mid-1990s as the only country showing a potential stall in fertility decline after fertility transition had apparently already begun (Bongaarts 2008). In addition to a country-level stall in the total fertility rate (TFR) Guatemala is characterized by highly variable fertility trajectories at the subnational level. Several of Guatemala’s eight political regions have even recorded significant increases to TFRs during the recent past; see Fig. 1, which presents regional fertility over time.
https://static-content.springer.com/image/art%3A10.1007%2Fs12546-012-9094-5/MediaObjects/12546_2012_9094_Fig1_HTML.gif
Fig. 1

Map of Guatemala with bar-charts of TFR: 1987–2002. Bar charts constructed based on author’s calculations using the 2002 RHS data and 1998/1999, 1995, 1987 DHS data. The eight regions used here are the standard for collection and reporting among Guatemalan government agencies. Bar charts are drawn using the same y axis scale and are comparable across all regions/time periods

Certain historical, political and social factors, among them civil strife and an ethnically divided population, are hypothesized to have limited the supply and demand of family planning and are probably at the root of the country’s high fertility and the apparent stall in fertility decline found in the mid-to-late 1990s (Santiso-Galvez and Bertrand 2004). And, while 2002 country-level fertility rates dropped from the 1998/1999 levels, inconsistent historical trends in fertility (several regions experienced an increase in fertility at some point between 1987 and 2002) and a virtual halt to fertility decline between 1995 and 1998/1999 indicate that Guatemala may not be on a direct path toward replacement fertility. Within a macro-level analysis framework the timing and duration of some of these events, particularly the civil war and the Leftist movement, undermine a straightforward period analysis of the country’s fertility patterns. However, because of the severity of the country’s health and population challenges, it is important to act within the current time frame rather than use cohort behaviour to retroactively assess the motivators and inhibitors of family planning and high fertility.

The purpose of this paper is to move beyond conventional representations of developing-country fertility, the TFR being the most commonly used metric to represent aggregate fertility behaviour, by using several different measures of fertility at the country and regional levels. The following section introduces the period and cohort and regional- and country-level fertility indicators used in this analysis.

Analytic approach

Several measures have been constructed to investigate the effectiveness and relative strengths of macro-level period and cohort metrics within the context of stalling fertility and inconsistent regional fertility trajectories. The metrics, their construction and comments on their utility are presented in Table 1. For this analysis each metric was calculated for each of eight political regions and at each of the four most recent time periods. Decomposing the country-level measure into its geographic parts, eight distinct regions in this case, provides additional information that may help to identify a stall or change in fertility. Country-level rates are aggregate measures of regional rates, so regional patterns of decrease or perhaps increase are masked by the single country-level rate. Building on theories that the ethnic divide1 serves as an important determinant of fertility behaviour, dividing the TFR across the broadly identifiable Indigenous and Ladino regions allows the incorporation of more complex population characteristics than those included in the aggregate country-level rate (Bertrand et al. 2001; De Broe et al. 2005; De Broe and Hinde 2006; Santiso-Galvez and Bertrand 2004; Seiber and Bertrand 2002).

Age specific fertility rates (ASFRs) and TFR

The use of TFR to characterize and classify populations into high, middle and low fertility is widespread. Virtually every governmental and non-governmental agency interested in family planning and women’s health evaluates country, regional, and global fertility using TFR. TFR and its controversial variants (see Bongaarts and Feeney 1998; Sobotka 2003; von Imhoff and Keilman 2000) are used by virtually all population scientists and have been an important tool used to identify the recent trends in Latin American fertility decline (Rosero-Bixby et al. 2008; United Nations 2007). The construction of the TFR first requires the construction of ASFRs (see Table 1). In the numerator a frequency count of the number of births within a period is calculated for each of seven 5-year age groups. The number of person years of exposure, again per age group, serves as the denominator. For this analysis women’s ages and birth experiences are recoded into number codes by calculating the number of months between the event and January 1900 (century month codes).2 Using this type of coding allows a woman to contribute as little as 1 month to one age-specific denominator and allows her to divide her exposure to more than one age group and as many as two (De Broe and Hinde 2006). TFR is then constructed by summing the ASFRs.
Table 1

Period and cohort fertility indicators

Measure (abbv.)

Method of calculation

Type

Remarks

Age-specific fertility rate (ASFR)

\( \,=\,{}_{n}F_{x} = {}_{n}B_{x} /{}_{n}W_{x} \)

nBx is births to women aged x, x + n

nWx is number of women aged x, x + n

Period

Sensitive to timing changes and may therefore show age specific increases in magnitude distorted by timing effects (Ni Bhrolchain 1992; Bongaarts and Feeney 1998)

Total fertility rate (TFR)

\( \,=\,5\sum\limits_{x = 15}^{45} {{}_{5}F_{x} } \)

Period

Synthetic cohort measure; sensitive to timing changes; useful to compare behaviour across regions, time; aggregate measures reveal different patterns from sub-populations (Bongaarts and Feeney 1998)

Adjusted total fertility rate (adjTFR)

\(\,=\, 5\sum\limits_{x = 15}^{50} {\sum\limits_{p = 1}^{15} {\frac{{{}_{5}F_{x,p} }}{{1 - (MAB_{p,t + 1} - MAB_{p,t} )}}} } \)\( {}_{n}F_{x,p} = \frac{{{}_{n}B_{x,p} }}{{{}_{n}W_{x} }}, \) is the age, order fertility rate

\( MAB_{p,t} = \sum\nolimits_{x = 15}^{45} {x\frac{{B_{x,p,t} }}{{B_{p,t} }}} \) is mean age at birth

Period

Synthetic cohort measure with tempo effects removed; sensitive to small changes in mean age at birth; sensitive to small sample sizes (Bongaarts and Feeney 1998; Bongaarts 1999, 2008)

Cohort parity progression ratios (PPR)

\( \,=\,{{\sum\limits_{p = i + 1}^{15} {{}_{5}W_{40,p} } } \mathord{\left/ {\vphantom {{\sum\limits_{p = i + 1}^{15} {{}_{5}W_{40,p} } } {\sum\limits_{p = i}^{15} {{}_{5}W_{40,p} } }}} \right. \kern-\nulldelimiterspace} {\sum\limits_{p = i}^{15} {{}_{5}W_{40,p} } }} \)

Cohort

Occurrence-exposure measure of true transition patterns; not subject to timing fluctuations (Frejka and Calot 2001)

Completed cohort fertility rate (CFR)

\( \,=\,\frac{1}{{W_{{\{ x > 40\} ,p}} }}\sum\limits_{p = 1}^{15} {{}_{p}W_{{\{ x > 40\} ,p}} } \)

Cohort

Measure of true completed fertility; not subject to timing fluctuations (Kim and Schoen 2000; Imhoff 2001)

Tempo and TFR

TFR is sensitive to the timing of births. If, in one period, women were to delay their fertility by 1 year, assuming no change in eventual completed fertility, period ASFRs would be temporarily depressed and would result in a decreased TFR for that period. Likewise, women can advance their fertility schedules, which would result in an inflated TFR. Bongaarts and Feeney (1998) developed a straightforward method of tempo adjustment that incorporates annual change in age at each birth order into the calculation of a tempo-adjusted TFR. As indicated in Table 1, to construct the adjusted TFR the difference between estimates of mean age at birth for each birth order is used to adjust the age-order-specific fertility rate (AOSFR). Without adjustment, the AOSFR can be summed across order then across age to construct the traditional TFR. With adjustment, an increase or decrease in mean age at birth results in a divisor smaller or greater than one, and the adjusted TFR is larger or smaller than the unadjusted TFR.

Adjusting the period TFR for timing effects may reveal more information about period influences on fertility and is especially relevant in spatial and temporal fertility comparisons. Moreover, the Bongaarts-Feeney method serves as a useful and prominent tool in contemporary analysis of developed-country fertility. Its absence in fertility analysis of less developed countries is notable (Bongaarts 1999; Lutz et al. 2003). An examination of the median age at first birth values, available through the Guatemalan DHS reports, can provide preliminary information supporting the need for a tempo adjustment (Bongaarts 1999). Because of an increase of approximately one year per decade (comparing median age for women 25–29 and 30–34) there is an indication of potential timing effect affecting the 1995 TFR (Bongaarts 1999). The differences in median ages in the 1987 and 1998/1999 data are minimal and show no major indication of a tempo effect. However, as the tempo adjustment technique is a mathematical variant of the traditional TFR, the absence of timing changes will result in adjusted rates that do not differ from the unadjusted rates. Adjusting the TFR when no timing changes are present should be no different from the traditional TFR and allows for the risk-free calculation of adjusted TFR for each time period. Furthermore, decomposing this commonly employed measure into fertility level and timing provides additional information using already available data. The tempo adjustment technique developed by Bongaarts and Feeney, slightly adjusted to accommodate DHS data, will thus be applied to the three earliest data sets to develop tempo-free measurements of country- and regional-level TFR (Bongaarts 1999).

It is important to note, however, that using survey data to calculate tempo-adjusted measures of fertility is not as precise as employing census data or birth registry data, the data types commonly used in tempo analysis of developed populations. Survey data are very susceptible to sampling and measurement error, which are very likely to affect the mean age at birth. Moreover, the small sample sizes, fewer than 1,000 women in some of the strata in this analysis, severely reduced the number of higher-order births. To adjust for the reduced sample size of women experiencing higher-order births, tempo adjustment and mean age calculation were only performed for four or fewer births.

Parity progression ratios (PPRs) and cohort fertility rate (CFR)

The use of PPRs to evaluate trends in this region of the world is uncommon. PPRs incorporate both age and parity and are therefore valued as insightful measures of fertility trends (Feeney 1991; Ni Bhrolchain 1992, 2007; Sheps and Menken 1973). PPRs capture the movement of an individual from one parity to the next. This measure differs from the ASFRs or average completed fertility in that rates of movement out of one parity and into the next rely on the exposures only of women who are actually at risk for entering into a higher parity (see Table 1). For example, when calculating the rate of 15–19-year-old women who move into parity two during a particular time period, the count for this age group that has already attained parity one serves as the divisor (as opposed to the entire population aged 15–19). The primary focus will be the rates of movement, particularly comparing the movements among the lower parities to those among the higher parities of women who have already completed their childbearing: those between the ages of 40 and 49. While Latin American women have shown a general tendency toward early and universal motherhood, time- or region-dependent variation in the transition to higher-order births may account for fertility variation (Rosero-Bixby et al. 2008). The CFR, the calculation of average total births among women who have completed their childbearing, will also be calculated for the same populations used to construct the PPRs. Since the CFR is a direct measure of completed family size, the hypothetical cohort construct and period-tempo distortions are irrelevant. Waiting until cohorts have completed their childbearing enables the analysis of factors that may have affected or shaped past fertility behaviour. In this case, women aged 40-49 (in 1987 the data were collected only for women 44 and younger) are considered to have completed their fertility and the average number of lifetime births to these women is calculated. While all of these cohorts began bearing children during the war, they may demonstrate different patterns of behaviour based on their peak years of childbearing and the most intense times of violence or leftist activity, thus potentially capturing region-specific period effects.

Data

The Demographic and Health Survey (DHS) data collected in 1987, 1995, and 1998/1999 by the Guatemalan Instituto Nacional de Estadstica and Measure/DHS+ will be used (MSPAS 1989, 1995, 1999). The DHS is the largest continuing survey in the world and is the primary source of data on population, health and socio-economic indicators for developing nations. DHS data are invaluable for conducting fertility analysis with respect to regional characteristics in this population. Not only do the DHS results provide extensive information regarding individual and family health and cultural norms, the large sample size, more than 12,000 respondents in some cases, can be used to provide a detailed overview of large-scale trends and ultimately enables regional and temporal comparisons of fertility. Moreover, the 1998/1999 survey represents the first large-scale data collection of reproductive health information of inhabitants of the Petén, the northernmost region of Guatemala. Previously, this extremely impoverished population was excluded from surveys and was therefore largely unrepresented in policy decisions. The Reproductive Health Survey (RHS) collected in 2002 with the assistance of the Centers for Disease Control (CDC), will also be used (MSPAS 2002). For the purpose of fertility analysis, the RHS and DHS are essentially identical. They consist of similar sample sizes and weighting schemes, and contain similar reproductive health and fertility information. The sampling designs for all four surveys are virtually identical (see Footnote 2). The surveys were based on a multistage cluster sampling design. The 22 departments of Guatemala (subsets of the eight regions) contain census clusters that were delineated by the 1994 Population Census. Census clusters were then randomly selected from each department and households were randomly selected from within the selected census clusters. The sampling scheme was developed to ensure that each region is proportionally represented in the sample. DHS and CDC calculate weights for each individual to adjust for over or under sampling. Reflecting standard use, weights were used throughout the analysis.

Results

Aggregate population measures, largely applied to developed countries, may be less informative or even misleading in the context of the developing world. The following section presents the results of each of the previously introduced indicators as they apply to Guatemala. While the results reflect the explanatory capability of the metrics in this particular context, they also contribute to a broader understanding of macrolevel fertility indicators in the developing world. In each of the following subsections the national-level time-dependent results are presented, followed by the regionally disaggregated results. Table 2 summarizes the results.
Table 2

Summary of patterns of Guatemalan fertility indicators

 

National-level patterns

Regional-level patterns

Are regional patterns represented by national patterns?

Measure

General

Specific

General

Specific

Period

Age specific fertility rate

Decrease

Over time indicates decreases in magnitudes and general leftward shift

Inconsistent

Over time some regions show decrease in magnitude and leftward shift while other regions show no clear indications of change

No

Total fertility rate

Decrease with mid-1990s stall

Small decline with plateau in middle years followed by quick paced decline during

Inconsistent

Half the regions experience an increase at one point during the period

No

Adjusted total fertility rate

Decrease following early-1990s stall

No real decrease for first two periods followed by steepening decline at third period

Inconsistent

Six regions experience an increase, two show consistent decline

No

Cohort

Parity progression ratios

Decrease

Decrease among higher order births, consistent among lower order births

Inconsistent

Most regions do not reveal decreasing trends in rates of transition to higher parity

No

Completed cohort fertility

Decrease

Overall decline with increase in middle years

Inconsistent

Half the regions have values with increasing magnitudes at one point

No

ASFRs and TFRs

ASFRs sum to the TFR. In lower-TFR regions or time-periods ASFRs, across all ages, should generally be smaller. In a setting that is in transition from high to low fertility the peak of the ASFR will generally move left toward younger ages in conjunction with overall decreased magnitudes, often at each age. Figure 2 presents the ASFR calculated for each region and each time period. The variation in the regions over time is immediately apparent. The metropolitan region, home to the capital city and characterized as the region with the lowest current fertility rate, shows relatively consistent behaviour for the earliest age group. An increase does appear in the ASFR in 1995 and 1998/1999 for the 20-24 age group, but then declines significantly in 2002. The 2002 ASFR points to lower fertility across virtually all age categories with the exception of the higher age categories, which merge across the years. The northern region, as expected, generally has the highest ASFRs across regions with consistently high rates among both older and younger women. This region, however, shows a leftward shift consistent with fertility decrease but no net changes in magnitude. The Petén region demonstrates a leftward shift in combination with a decrease in magnitude, resulting in a lower TFR for the latter time period. Comparing the ASFRs of high and low TFR regions underscores the effect that fertility later in life (presumably to women who have a relatively large number of children) has on the magnitude of the TFR.
https://static-content.springer.com/image/art%3A10.1007%2Fs12546-012-9094-5/MediaObjects/12546_2012_9094_Fig2_HTML.gif
Fig. 2

Regional age-specific fertility rates of Guatemalan women aged 15–49: 1987–2002. Author’s calculations using data from 2002 RHS and 1998/1999, 1995, 1987 DHS. Data from 1987 were collected only for women aged 15–44

The 1995 tempo-adjusted TFR is 5.4, which is 0.3 births larger than the unadjusted TFR. This larger value reflects increases in tempo-adjusted ASFRs for three of the four age categories considered and suggests that an increase in mean age can account for most of the fertility decline from 1987 to 1995 shown by the conventional TFR. While the mean age at first birth does not show any marked decline between 1995 and 1987, the increases in the mean age at second and fourth births are driving much of the differential in the TFR. The tempo-adjusted rates portray high fertility for 1987 and 1995, which declined to the value of 1998/1999, followed by continued decline to the 2002 level. This image of fertility change differs from the image portrayed by the unadjusted rate that shows more of a decline from1987 to 1995, followed by a plateau (or stall) from 1995 to 1998/1999 and then another decline. Applying the adjustment and evaluating order-specific TFRs we see that the increase in the mean age at second birth, probably by increased spacing between the first and second births, may have caused the TFR to appear lower in 1995 and 1998/1999 than the adjusted TFR. Certainly, this holds true for the TFR of order two in both time periods. Therefore, despite the relative consistency in age at first birth over the time periods, significant changes to spacing and age at second birth have affected the TFR which thus may appear to stall in the future if women stop increasing the spacing of their first and second births.

Unfortunately, since no data are available beyond 2002, we cannot calculate a tempo-adjusted measure of fertility for the 2002 time period. However, we can evaluate the order-specific TFR to develop intuition into the potential for tempo effects on these latest fertility rates. With rates of 0.76 for TFRs of orders one and two suggesting that 76 % of women will have a first or second birth during their lifetimes, we can identify the presence of tempo effects. As Guatemala is a developing country where motherhood is still highly valued, it is very unlikely that 24 % of women will remain childless or that 24 % of women will have single-child families. In fact, it is much more likely that women have started delaying the onset of childbearing, which is affecting the numbers of first and second births. This likely increase in age at first birth presents an optimistic view into the future of childbearing in Guatemala. Once women stop delaying, however, it may appear as though TFR has reached a plateau or has increased (Bongaarts 1999). Policy-makers and public health specialists must be made aware of the potential for an apparent, not actual, delay in the future of fertility decline in Guatemala.

Combining the results from the traditional and adjusted rates allows us to understand fertility behaviour more completely and highlights other aspects of the fertility decline, such as shifts in duration or birth spacing and changes in fertility quantum. Figure 3 presents both the adjusted and unadjusted rates. Regional tempo adjusted TFRs are presented in Table 3. Because of small regional sample sizes, regional rates must be interpreted cautiously and are presented here to illustrate potential applications for tempo-adjusted techniques in a developing-country setting. In most cases, regional TFRs and adjusted regional TFRs show similar fluctuating or decreasing patterns. In the southeast and central regions, however, the conventional TFR indicates a decrease in fertility while the adjusted TFRs show an inconsistent pattern. The metropolitan and southwest regions experience relatively large increases in fertility (as compared to the unadjusted rates and the preceding-years rates) when the traditional measure is adjusted for tempo effects for the 1995 period. These changes may result from declines in higher-order births and may reflect changes in durations between births, suggesting that these regions may ultimately be trend-setters in the reduction of higher-order births. As the national-level rates are weighted composites of the regional rates, the large increases in 1995 regional rates in these populous regions are driving the increase in the adjusted country rate for the same period, despite decreases in the rates of four of the other regions. Therefore, some significant regional-level fertility decreases are masked by the fertility increases of other regions. These regional trends suggest that developing family planning programs with attention to regional characteristics may be beneficial.
https://static-content.springer.com/image/art%3A10.1007%2Fs12546-012-9094-5/MediaObjects/12546_2012_9094_Fig3_HTML.gif
Fig. 3

TFR and adjusted TFR: 1987–1998/1999. Note: Author’s calculations based on data from the 2002 RHS and 1998/1999, 1995 1987 DHS surveys

Table 3

Comparison among Guatemalan regional unadjusted and adjusted TFR, 1987–2002

 

2002

1998/99

1995

 

ENSMI

TFR

95 % CI

adjTFR

ENSMI

TFR

95 % CI

adjTFRa

ENSMI

TFR

95 % CI

adjTFR

Region

Metropolitan

3.2

3.3

3.2–3.4

4.3

4.2

3.3–5.3

5.0

3.9

3.9

3.2–4.6

4.2

North

6.5

6.5

6.1–6.8

5.5

5.5

4.4–6.6

5.8

6.7

6.7

6.1–7.3

5.9

Northeast

4.7

4.3

4.1–4.4

5.4

5.4

3.9–6.9

5.2

5.1

5.1

4.2–6.0

4.6

Southeast

4.4

4.2

4.0–4.5

5.1

5.1

3.9–6.3

4.8

5.7

5.7

4.9–6.2

6.6

Central

4.2

4.2

4.0–4.4

5.0

5.0

4.2–5.7

5.0

5.3

5.3

4.7–5.8

4.8

Southwest

5.0

4.7

4.5–4.9

5.3

5.2

4.6–5.7

4.7

5.5

5.5

5.0–5.9

5.9

Northwest

5.5

5.4

5.1–5.7

6.2

6.2

5.4–7.0

6.7

6.8

6.8

6.2–7.3

6.8

Peten

5.8

5.9

5.4–6.3

6.8

6.9

6.1–7.7

8.8

    

Country

 

4.4

4.4

4.0–4.7

5.0

5.0

4.6–5.5

5.0

5.1

5.1

4.8–5.5

5.4

TFR and adjTFR represent author's own calculations based on data from the 2002 RHS data; 1998/99, 1995 and 1987 DHS data. Values calculated by ENSMI (Encuesta Nacional de Salud Materno Infantil), the Guatemalan reproductive health survey division, are included for comparison. Exact duplication of the ENSMI values was not possible as details of their technique corresponding to 2002 cannot be calculated until data from the subsequent 2007 survey are available.

a Because of an unusually large increase in the mean age of fourth birth in 1998/99 in the Metro. Region this rate is only adjusted through parity three.

Parity progression ratios (PPRs) and cohort fertility rate (CFR)

The parity progression analysis presents a different perspective on fertility in Guatemala. As the rates in Table 4 are based on the oldest cohort from each survey, they capture cohort rather than period fertility trends. The values measure the probability that a woman who already had, for example, two children would proceed to have a third. In 2002 in the metropolitan region the probability of this event is 0.79, whereas in the north region the probability is 0.98. Regionally, declines consistent with the reduction of the observed TFR coincide with declines in the progression to higher-order births, especially births of order four or more in the metropolitan region, which is also characterized by the lowest TFRs. This presents the likely scenario that motherhood and the second birth are virtually universal and that fertility decline is occurring as a result of choices made by older women to reduce or stop their childbearing at a certain parity.
Table 4

Parity progression ratios and fertility rates of women aged 40–49a, 1987–2002

Region

0 → 1

1 → 2

2 → 3

3 → 4

4 → 5

5 → 6

6 → 7

7 → 8

Cohort fertility rate

Metropolitan

2002

0.94

0.95

0.79

0.69

0.72

0.76

0.75

0.81

4.29

1998/99

0.95

0.97

0.91

0.76

0.71

0.79

0.74

0.60

4.96

1995

0.95

0.93

0.78

0.72

0.71

0.65

0.67

0.65

4.01

1987

0.96

0.96

0.93

0.84

0.74

0.77

0.66

0.74

5.28

North

2002

0.95

0.98

0.98

0.92

0.89

0.88

0.84

0.66

6.83

1998/99

0.92

0.97

0.93

0.88

0.94

0.87

0.79

0.78

6.08

1995

0.96

0.98

0.96

0.94

0.96

0.87

0.83

0.90

7.17

1987

1.00

0.97

0.92

0.89

0.84

0.83

0.77

0.76

7.08

Northeast

2002

0.96

0.96

0.94

0.78

0.81

0.80

0.82

0.74

5.43

1998/99

0.99

0.99

0.86

0.83

0.81

0.80

0.80

0.82

5.70

1995

0.92

0.95

0.95

0.87

0.85

0.77

0.87

0.78

5.66

1987

0.93

0.97

0.93

0.90

0.85

0.83

0.75

0.78

5.71

Southeast

2002

0.98

0.99

0.96

0.88

0.86

0.80

0.78

0.75

6.48

1998/99

0.99

0.99

0.93

0.92

0.92

0.93

0.94

0.95

5.59

1995

0.96

0.99

0.97

0.96

0.91

0.86

0.81

0.78

6.87

1987

0.98

0.96

0.92

0.90

0.85

0.82

0.77

0.77

5.86

Central

2002

0.95

0.96

0.86

0.83

0.69

0.86

0.83

0.74

5.04

1998/99

0.93

0.96

0.93

0.88

0.80

0.83

0.75

0.68

5.43

1995

0.95

0.95

0.93

0.88

0.84

0.82

0.81

0.71

5.83

1987

0.98

0.97

0.92

0.87

0.78

0.80

0.72

0.72

6.14

Southwest

2002

0.97

0.95

0.90

0.87

0.85

0.83

0.82

0.68

5.81

1998/99

0.96

0.96

0.95

0.92

0.84

0.85

0.72

0.76

6.18

1995

0.99

0.97

0.94

0.90

0.87

0.86

0.82

0.76

6.51

1987

0.92

0.97

0.92

0.89

0.83

0.83

0.76

0.75

6.06

Northwest

2002

0.99

0.97

0.97

0.91

0.90

0.88

0.83

0.79

7.03

1998/99

0.96

0.94

0.96

0.94

0.91

0.87

0.85

0.82

6.69

1995

0.98

0.97

0.96

0.93

0.90

0.89

0.85

0.77

6.94

1987

0.95

0.97

0.92

0.89

0.82

0.83

0.75

0.74

6.13

Peten

2002

1.00

0.97

0.95

0.90

0.95

0.85

0.86

0.76

6.93

1998/99

0.98

1.00

0.95

0.96

0.93

0.91

0.89

0.88

7.78

NA for 1995

         

NA for 1987

         

Country

2002

0.96

0.96

0.88

0.81

0.81

0.82

0.81

0.75

5.41

1998/99

0.95

0.97

0.92

0.85

0.81

0.83

0.78

0.73

5.69

1995

0.96

0.95

0.89

0.85

0.84

0.81

0.80

0.76

5.61

1987

0.96

0.97

0.93

0.90

0.85

0.82

0.74

0.77

5.87

Author’s calculations based on 2002 RHS data and 1998/1999, 1995, 1987 DHS data. The cohort fertility rate presented here is not directly related to the transition rates but rather a measure of the average number of completed births among women in the specific cohort

a In 1987 data were collected only through age 44

In the north region, where observed TFRs are the highest and have recently shown signs of increase, the likelihood of progressing to higher-order births generally remains higher than in the other regions and exhibits little indication of a decrease. The plot in Fig. 4 presents the 2002 cohort parity progression ratios with north and metropolitan regional PPRs being the highest and lowest fertility regions of the time. The shading provides the boundaries of the highest and lowest overall rates of transition from the entire sample of regions. Most of the differences in completed cohort fertility can be explained by the different rates of transition between parities three and seven, where the higher-fertility region has significantly higher transition rates. The less than half a child variation in country-level CFRs does not show the variation that the period TFR shows. In fact, it would be difficult to infer from this cohort measure that the country had recently experienced significant political, social and economic change. The majority of the women who contributed to the construction of these indicators began and completed their reproductive lives during the war; changes in completed fertility, if they exist, therefore may not be apparent until the next generation completes childbearing.
https://static-content.springer.com/image/art%3A10.1007%2Fs12546-012-9094-5/MediaObjects/12546_2012_9094_Fig4_HTML.gif
Fig. 4

2002 Cohort parity progression ratios of the highest and lowest fertility regions and corresponding confidence intervals. Author’s calculations based on the 2002 RHS data and 1998/1999, 1995, 1987 DHS data. Shaded area represents the range of 2002 parity transition rates over the eight regions. Transition rates on the vertical axis are as reported in Table 1. For each parity the rate is the proportion of women moving from the specified parity into the next parity

Despite the relatively stable country-level CFR, regional rates show significant intra- and interregional variation. In comparing cohorts born between 1938 and 1947 to those born between 1953 and 1962, the central and metropolitan regions show the largest decrease in completed fertility values. In general, the mean completed family sizes show considerable variation despite the expectation of smoother patterns. Although restricted to only the two most recent time periods (representing women born between 1949–1958 and 1953–1962), the Petén region nonetheless shows the third largest decline in completed fertility. Both the northwest and southeast show increases in completed fertility values.

Discussion

Fertility behaviour is extremely complex in stable societies, and the chaos wrought by war, famines, or economic shocks makes fertility analysis in less developed countries even more difficult. Ultimately, we would like to find patterns in fertility metrics that allow us to document and understand past behaviour while also designing policy interventions toward meeting the family planning needs of women. The problem highlighted in this paper, and building on others such as Ni Bhrolchain (1992) and Sobotka (2003), is that there is no single metric that provides definitive insight into fertility. Instead, the different measures reflect different facets of the underlying behaviour. The split into period and cohort perspectives is, in some ways, an over-rigid dichotomy that masks the diversity, complementarity and importance of underlying metrics.

In general, the results here can be interpreted as a dominance of period effects; in essence, the cohort measures reflect and capture the impact of major period disturbances such as the civil war and delayed contraception. The cohort perspective is based on the idea that women in a birth cohort are homogeneous in choosing their ultimate number of children. If interrupted by period events they revert back to their originally established fertility goals. The period approach, alternatively, provides more room for variation throughout a woman’s reproductive period, nearly 40 years, and builds on the assumption that fertility aspirations are sensitive to a variety of factors that are continually changed and renegotiated.

The period measures used here provide up-to-date information on births with respect to the age composition of the population. TFR is easily calculated from limited survey data and produces a metric that is commonly relied on by specialists and non-specialists to represent the average lifetime fertility of a woman during a particular time period. It provides a measure that reflects contemporary factors, and regional or temporal variability can be explored as dependent on other demographic, economic or political factors. While TFR has known flaws, primarily its sensitivity to changes in timing, the metric can be altered to explore timing changes in childbearing. Combining the adjusted measure with the traditional TFR requires little additional data but provides information that forces the analyst to consider the interplay of exogenous factors on an individual’s reproductive decisions.

TFR and ASFR, the classic measures of period fertility, reveal a general decrease in fertility at the country level with ASFRs decreasing for each age period. The insight provided by the timing adjustment highlighted the impact of the war on fertility behaviour and suggests that Guatemala was generally moving steadily toward lower fertility, but that fertility increased upon the conclusion of the war, leading to an apparent stall in the decline. The impact of the war and the Peace Accords on fertility in Guatemala has not been directly evaluated and the short-term or long-term impacts of extended civil strife or humanitarian crises on fertility and family formation in Latin America have not been extensively explored (Hill 2004; McGinn 2000; Yucesahin and Ozgur 2008). High fertility rates and fertility stagnation are viewed as the response to social and political factors inhibiting fertility regulation, and not as increased demand for children (Santiso-Galvez and Bertrand 2004); however, increases in fertility as families attempt to compensate for losses experienced during the war are also a likely cause.

The regional disaggregation of TFR was done to identify high- and low-fertility regions and to determine their effect on the apparent national-level stagnated fertility of the late 1990s and the high fertility rate in general. At the regional level the TFRs and ASFRs presented patterns of consistent decline for some regions, while for other regions no clear fertility pattern emerged. The metropolitan region experienced increases in TFR during the time period consistent with the timing of the apparent country-level stall. Adjusting the TFR for timing effects, however, places the peak of the metropolitan fertility in 1995 with a subsequent decrease in 1998/1999, a similar shape to the country’s curve. Up to 1998/1999 no other region had a TFR lower than 5.0. Many of the remaining regions show stagnating rates until 1998/1999; adjusted rates offer little additional information.

Given that the metropolitan region contains nearly 50 % of the entire population, fertility patterns and rates attributed to metropolitan residents dominate the overall country-level rates. Perhaps, outside the metropolitan region, fertility reduction strategies were facilitated by improved supply of contraception/family planning and development programs initiated after the war, or perhaps there are ideational differences underlying motivation for large families. Evaluating either traditional TFR or adjusted TFR, however, clearly indicates that the fertility patterns of the metropolitan region are drastically different from those of the other Guatemalan regions and are perhaps inappropriately masking region-specific fertility adjustment regimes that are diverging from the presumed national fertility transition. In future work a closer analysis of birth spacing and the timing of births may help shed light on regional specific transitions to determine if ideational changes are influencing fertility behaviour or if other factors (such as contraceptive availability and economic development) are bringing about variation over time and across the country.

Many demographers view cohort fertility measures as superior to period fertility measures. Were period measures less effective at representing fertility than cohort measures in this case? The cohort measures highlight overall declines in the transition to higher-order births and reveal noticeable differences in the behaviour of the highest fertility region compared to the lowest-fertility region. The average completed cohort fertility, despite being lauded by pro-cohort methodologists as showing less variability than period measures, reveals a pattern of variation similar to the period fertility measures. Any attempt to explain the variation results in the analysis of period in influences on each of the cohorts, rather than any cohort-specific factors in behaviour. Certainly, the fertility level offered by the average completed cohort fertility is indisputably valid; however, if the highest rates and the lowest rates (both in terms of region and time) were similarly identified by period measures, which also served as measures of the effect of social, political, and economic change, then arguably period measures provide a richer and up-to-date view of demographic patterns. Additionally, the ability to decompose the period TFR to evaluate the effect of changes in timing enables further disaggregation of the standard metric and reveals more about the role of period factors in reproductive health decisions. While varying rates in transitions to higher-order births, as measured by the cohort PPRs, did indicate that decreasing fertility results from lower completed fertility, period PPRs would probably provide similar information relevant to the current context rather than as a retrospective fertility account. These measures will be addressed in future research. Beyond providing a true measure of fertility, the average cohort fertility did not improve understanding of the fertility transition in Guatemala as indicated by the period TFR.

Conclusion

Decomposing the country-level fertility rate to explore timing effects and using period and cohort measures have helped to highlight the effect of period factors on both cohort and period fertility rates. Using techniques beyond the traditional TFR has provided more information about macro-level fertility behaviour and reinforces the importance of using a broad perspective when expressing a country’s demographic trends. Presenting the country-level TFR as the sole measure of fertility behaviour can be misleading. However, in combination with tempo-adjusted measures and geographically disaggregated measures, the period TFR yields more timely information than cohort indicators. Timeliness is particularly valuable in relatively unstable developing-world contexts, and period indicators are probably more immediately informative than cohort indicators. In conclusion, the combination of the tempo-adjusted (at the regional and country level) with the traditional TFR best represents the fertility of the developing world.

The limitations of this research underscore the need for high-quality birth records and related data. Censuses frequently exclude detailed information on births or their timing and are collected relatively infrequently. Surveys are conducted more frequently, but not often enough and they only collect data from a sample of the population. There is hardly any clinic and hospital information which could provide contextual-level information about supply and demand for contraception. Without the collection of quality data at regular intervals, the explanation for Guatemala’s delayed and prolonged fertility transition will never be fully understood. Without accurate information about who is giving birth and where births are occurring, it is virtually impossible to effectively allocate the limited resources. Many countries in Latin America and beyond (sub-Saharan Africa, for example) that are characterized by stalled or slowing fertility declines will benefit from a more nuanced understanding of the fertility transition. Enhancing our knowledge of contemporary transitional behaviour will facilitate more effective policy development and resource allocation. Improved understanding of timing changes in high-fertility contexts and the development and application of metrics that accommodate often-neglected high-fertility regions in the developing world would also be extremely valuable in future reproductive health and demographic analysis.

Footnotes
1

Luis Rosero-Bixby suggested (personal communication) that Guatemala is better viewed as two Guatemalas rather than one unified population.

 
2

The 1987 DHS was designed to collect an equally weighted sample so sampling design differs slightly.

 

Acknowledgments

We thank Dr. James Holland Jones for comments made on an earlier draft of this paper and we thank an anonymous reviewer for comments and suggestions. We also are very grateful for the support of the editor, Dr. Edith Gray.

Copyright information

© Springer Science & Business Media B.V. 2012