Temperature in the Iberian Peninsula: Trend, seasonality, and heterogeneity¹¹1Partial financial support from “la Caixa” Foundation, grant LCF/PR/SR20/52550012-Climate change and economic challenges for the Spanish society, and from the Spanish Government grant PID2022-139614NB-C22 (MINECO/FEDER) is gratefully acknowledged. We are also grateful to Jouni Helske for his help with some of the codes used in this paper and to participants at the 69th International Symposium of Forecasters (July, 2022), Rome-Waseda Time Series Symposium (October, 2022), 16th International Conference on Computational and Financial Econometrics (December, 2022), 6th International Conference on Econometrics and Statistics (August, 2023), and 35th Mexican National Statistical Forum (September, 2023). The usual disclaimer applies.

Climate change can be defined as the variation in the joint probability distribution that describes the state of the atmosphere, oceans, and freshwater, including ice; see Hsiang and Kopp (2018). According to the latest fifth and sixth assessment reports of the International Panel for Climate Change (IPCC, 2014, 2023), one of the most critical aspects of climate change is global warming, described by the evolving distribution of temperature.

The stand of the literature using econometric models to describe the evolution of temperature focuses on average temperature, which is obviously an important characteristic of the distribution.²²2Although we focus on econometric models, there is extensive literature based on deterministic climate models; see, for example, Diebold and Rudenbush (2022b) for a comparison between both types of models in the context of measuring ice volume in the Arctic. However, average temperature alone is not enough to reflect complicated climate variations. In an early paper, Katz and Brown (1992) show that the frequency of extreme weather events is relatively more dependent on any changes in the variability than on the mean of temperature, with this sensitivity being relatively greater the more extreme the event. Consequently, policy makers should not rely on scenarios of future temperature involving only changes in means but should also consider evolving variability of temperature; see, for example, Phella, Gabriel and Martins (2024), who fit quantile regressions with time-varying parameters to generate scenarios for extreme temperatures and their relation with anthropogenic emissions.

Our paper contributes to the important literature on econometric time series modelling of temperatures in two main ways. The first contribution is methodological. Instead of analysing only average temperature, we use the rich joint information contained in the minimum/maximum temperature interval. By doing so, we add to the information of the average temperature, represented by the centre of the interval, the information about the log-range between the maximum and minimum temperature and, consequently, information associated to extreme temperatures.

Furthermore, the proposed model decomposes the centre (average) and the range (variability) of temperature into potentially stochastic trends, seasonal and transitory components. This specification encompasses deterministic trends as a limiting case and, consequently, allows to distinguish between deterministic and stochastic behaviour in the long-run trend of average temperature. It also allows us to model the strong pattern of seasonality associated with temperature and determine whether it changes over time.

In particular, at a given location, we propose using a state space representation of the potentially non-stationary and seasonal system of centre/log-range temperature; see Harvey (1997) about the advantages of using unobserved components models instead of the widely used VAR models based on differencing and cointegration and Pretis and Hendry (2013), who suggest using a state-space approach due to the conflicting results often encountered when unit root and cointegration tests are used in the context of monthly data. The unobserved component models framework allows us to decide about the nature of the trends and seasonal components of the centre and log-range temperature, using the Kalman filter and smoothing (KFS) algorithms to extract the components together with measures of their uncertainty. Structural time series models have been implemented before in the context of average temperatures by, for example, Bloomfield (1992), Woodward and Gray (1993), Visser and Molenaar (1995), Zheng and Bahsher (1999) and Stern and Kaufmann (2000) for annual data, and by Good et al. (2007) and Proietti and Hillebrand (2015), and Hillebrand and Proietti (2017), for monthly data. However, instead of analysing point average temperatures, we propose unobserved component models for interval temperatures. Our methodology is closely related with the smoothing procedure proposed by Maia and Carvalho (2011), who fit two independent models to the mid-points and ranges of the intervals and estimate the smoothing parameters by minimising the interval sum of one-step-ahead forecast errors; see Harvey and Jaeger (1993), who show that the steady-state Kalman filter for the local linear trend model takes the form of Holt’s recursions with suitably defined smoothing constants. However, Maia and Carvalho (2011) do not consider the seasonal component. Furthermore, instead of fitting two independent models for the centre and log-range, we propose modelling them jointly to allow for potential relations between them.

On top of modelling the system of centre and log-range temperature at a given location, we also consider modelling them simultaneously at a large number of locations. To deal with the heterogeneity in the joint evolution of centre and log-range temperature observed at various locations, we propose using a multi-level Dynamic Factor Model (ML-DFM), which is estimated using the KFS algorithms. The ML-DFM allows to represent the joint evolution of a large system of time series of centre/log-range temperature, assuming that some of their trends are common to all locations while others may be common to some subsets of locations. Consequently, ML-DFM can represent some commonality in the evolution of temperatures while allowing, at the same time, for some idiosyncratic movements that explain the heterogeneity often observed when looking at temperatures at different locations. Marotta and Mumtaz (2023) also propose a ML-DFM for annual average temperature. However, our proposal considers not only modelling the average temperature but also its variability at several locations. Furthermore, we propose using data-driven procedures to identify regions with specific regional common factors.

Our second contribution is the empirical analysis of monthly centre and log-range temperatures observed at 68 locations spread throughout the Iberian Peninsula over nearly a century from January 1930 to December 2020. This analysis is particularly interesting because of the severe heat waves experienced in Southern Europe in recent years; see, for example, Kew et al. (2019). After analysing minimum/maximum temperatures separately in four selected locations, namely Barcelona, Coruña, Madrid and Seville, we show that the main characteristics of the centre temperature in each of these four cities can be represented by a smooth stochastic trend. The fact that the slope of the trend changes over time can explain why some authors find upward trends while others conclude about the hiatus in warming (global mean temperature has not risen significantly over the last two decades). Furthermore, we observe some heterogeneity in the trends of center temperatures in these four locations, with different slopes of the trend at the end of the observation period, December 2020. In particular, the slope of the trend is larger in Barcelona and smaller in Seville. We also show that the seasonal component of centre temperature is stochastic. When looking at the log-range, we conclude that it can be represented by a stochastically evolving level with stochastic seasonality. As expected, at the end of the sample period, the log-range is clearly larger in the two locations in the interior of the Iberian Peninsula, Madrid and Seville, and smaller in the two coastal cities, Barcelona and Coruña. Another important conclusion from the empirical analysis of these four cities is that the centre and log-range temperature are not correlated in each location, allowing us to model both characteristics separately.

After fitting separate models to the system of centre/log-range temperature in each city, we consider a multivariate ML-DFM for the simultaneous analysis of such series of temperatures observed at 68 locations in the Iberian Peninsula. First, we show that the commonality in the long-run trend of average temperature is strong, with the regional factors being somehow weaker, although also relevant for understanding the long-run behaviour of average temperature. However, when looking at log-range temperature, we find that global and regional factors are equally important.

The rest of this paper is organized as follows. In Section 2, we briefly survey the literature using econometric models to describe the dynamic evolution of temperatures and the still open debates. Section 3 describes the methodology proposed in this paper to model centre and log-range temperature. In particular, we describe how using state space models and the Kalman filter allows us to estimate trends and seasonal components at a given location. We also describe the ML-DFM and the implementation of KFS to extract the common factors. Section 4 describes the data and the main empirical characteristics of the centre/log-range systems at Barcelona, Coruña, Madrid and Seville. Section 5 fits state-space models to extract trends and seasonal components in each of these four locations of the Iberian Peninsula. Section 6 is devoted to the simultaneous analysis of centre and log-range temperature in a large number of locations spread over the Iberian Peninsula using ML-DFMs. Finally, Section 7 concludes the paper.

When describing global warming, the interest is on the evolution of the distribution of temperature either at a given location or jointly at several locations. However, regardless of whether the analysis is carried out by looking at temperature observed at a particular location or simultaneously at several locations, the stand literature focuses on analysing the evolution of average temperature, i.e. the central tendency of the distribution, based on a large variety of statistical and econometric approaches, periods and frequency of observation, and locations. Many of these studies find an upward trend in average temperature. Increasing trends are found by, for example, Deng and Fu (2019), who compare several methods for extracting cycles from daily temperatures, and Barbosa, Scotto and Alonso (2011), who analyse daily temperatures in several locations in Central Europe. These latter authors use quantile regressions to fit linear trends to the quantiles of average daily temperature. Note that, by doing this, they are analysing the cross-sectional distribution of average temperature but not the distribution of temperature itself. Similarly, Scotto, Barbosa and Alonso (2011) use Extreme Value Theory (EVT) to analyse the extremes of the distribution of daily average temperature in Europe and the spatial distribution of extreme events; see also Wang et al. (2021), who use Generalized EVT to look for fingerprints on temperatures. A similar approach is considered by Gadea Rivas and Gonzalo (2020, 2022), who use regression analysis to estimate quantiles of the full cross-sectional distribution of average temperatures or the distribution of average daily temperatures within a given year, and by Chang et al. (2020), who construct densities of global average temperature in the Southern Hemisphere (measured as anomalies concerning average temperatures over the base period 1961-90) and conclude that there is persistence in the mean, while non-stationarity is less evident in the variance. Kruse-Becher (2023) also considers cross-sections of temperatures to model their averages and ranges with adaptive methods that are robust under structural changes. Finally, Marotta and Mumtaz (2023) analyse the presence of global and regional common factors in annual temperatures observed in 160 countries from 1900 to 2020. They found an upward trend in the global factor. Similarly, Bogalo, Poncela and Senra (2024) also find an upward trend when analysing average monthly temperatures in 12 cities of Southern Europe observed monthly from January 1950 to December 2022. Note that although most evidence is about average temperature having an upward increasing trend, it is essential to point out that several authors describe what is known as the hiatus in warming; see, for example, Schmidt, Shindell and Tsigaridis (2014), Pretis, Mann and Kaufmann (2015), Estrada and Perron (2017), Medhaug et al. (2017) and Miller and Nam (2020).

Although the literature focused on modelling average temperature is relevant, there is some agreement that more is needed to reflect the complicated climate variations. Consequently, policy makers should rely on future climate scenarios involving more than only changes in means; see, for example, the recent scenario analysis by Phella, Gabriel and Martins (2024). In this direction, on top of modelling average temperature, several authors consider modelling the evolution of range temperature, computed as the difference between the maximum and minimum temperatures within a given period. These studies often find a downward trend in temperature variability, although the evidence is not as clear as for the positive trend of average temperature. For example, Vose, Easterling and Gleason (2005), Dupuis (2014), Qu, Wan and Hao (2014) and Meng and Taylor (2022) show that increases in minimum temperatures have been more relevant than increases in maximum temperatures in the globe (the first), different regions of the US (the second and third), and four cities in Spain (the last). Xu et al. (2013) analyse minimum and maximum daily temperatures in 825 stations in China observed from 1951 to 2020 and conclude that the diurnal temperature range has significantly decreased at 49% of the stations, with significant increases being identified at 3% of them. Diebold and Rudebusch (2022a) propose modelling the average and range temperature using separate regressions with deterministic trends, seasonal dummies and their interactions. The two univariate models are fitted to average and range temperature observed daily in selected cities in the US from 1960 to 2017. Instead of considering separate regressions, Meng and Taylor (2022) propose two alternative methods to model Interval-Valued Time Series (IVTS) of minimum and maximum temperature. First, they propose modelling the two-time series separately by fitting models with deterministic trends and seasonal components and allowing for interactions between minimum and maximum temperatures. Alternatively, IVTS of minimum and maximum temperature are modelled by a bivariate VARMA-MGARCH model. Using daily data in four Spanish cities observed from 1951 to 2015, their results also suggest a decrease in the diurnal temperature range and an increase in trend. Note that they do not restrict the model to avoid crossing minimum and maximum temperature at a given moment. The authors recognise that, although unlikely, the temperatures may cross.

On top of whether or not temperature variability should be modelled together with the evolution of average temperature, several important debates about modelling the dynamic characteristics of average and range temperature are very active these days. First, there is a growing literature about the nature of the trend in temperature; see, for example, the discussions by Coggin (2012) and Proietti and Hillebrand (2015). Within this literature, one of the most important controversies is whether trends in temperatures (or other climatological variables) should be modelled by assuming that they are deterministic or stochastic. For example, Fatichi et al. (2009) examine trends of daily average temperatures recorded in 26 stations in Tuscany (Italy) and conclude that a deterministic trend can be regarded as the most appropriate model only for a subset of 9 stations. Kaufmann, Kauppi and Stock (2010), Kaufmann et al. (2013) and Chang et al. (2020) are also among those who support the presence of stochastic trends in temperatures. However, Gao and Hawthorne (2006), Gay, Estrada and Sánchez (2009) and Gadea Rivas and Gonzalo (2020, 2022) argue that trend-stationary processes better characterise temperature. Furthermore, Chen, Gao and Vahid (2022) state that deterministic non-linear trends can represent global temperature. Seidel and Lanzante (2004), Gil-Alana (2008), Estrada, Gay and Sánchez (2010), Pretis and Allen (2013), Estrada, Perron and Martínez-López (2013), Estrada and Perron (2017), Friedrich et al. (2020), Kim et al. (2020) and Gadea-Rivas, Gonzalo and Ramos (2023) further propose models with deterministic trends with breaks, while Friedrich, Smeekes and Urbain (2020) and Friedrich et al. (2020) propose smooth non-parametric trends.³³3It is essential to note that it is challenging to establish the nature of trend in a time series with a finite sample and, consequently, there is still a methodological debate about the power of tests for trend-stationarity or difference-stationarity; see, for example, the general discussions by Diebold and Senhadji (1996), Phillips (2005, 2010) and Rao (2010) and the comments by Stern and Kaufmann (2000) and Pretis and Hendry (2013) in the particular context of climate variables. There are also several proposals for modelling temperatures based on long-memory models; see, for example, Baillie and Chung (2002), Gil-Alana (2005), Ventosa-Santaularia, Heres and Martínez-Hernández (2014), Mangat and Reschenhofer (2020) and Vera-Valdés (2021).

Another important debate when addressing the evolution of temperature has to do with its seasonal variation, which is the most prominent source of climate variability, so climate change could also be reflected in it; see Pezzulli, Stephenson and Hannachi (2005) and Proietti and Hillebrand (2015) for its importance. Bogalo, Poncela and Senra (in press) show that seasonality is the main feature of monthly average temperatures observed in 12 southern European cities, capturing 88.3% of the total variability, while the trend accounts for 1% of the total variability. Consequently, there is also an interest in knowing the most appropriate model for the seasonal components of temperature. Several authors have suggested that the seasonal pattern in temperatures varies over time with more intense warming in winters than in summers; see Harvey and Mills (2003), Vogelsang and Franses (2005), Cohen et al. (2012), Proietti and Hillebrand (2015), Hillebrand and Proietti (2017), and Bogalo, Poncela and Senra (2024). Furthermore, Hillebrand and Proietti (2015) also conclude that seasonal warming trends may vary across locations, while Bogalo, Poncela and Senra (in press) conclude that seasonal variations are primarily common.

Finally, among the hazards encountered when modelling climate-related time series, Pretis and Hendry (2013) point out the spatial variation of temperature trends, suggesting that there may be unmodelled heterogeneity. Dupuis (2014) and Scotto, Barbosa and Alonso (2011) also find heterogeneity, analysing minimum and maximum temperatures observed in 12 locations in the South-western US and extremes in average daily temperature in Europe. Recently, Estrada, Kim and Perron (2021) contribute to this debate by showing that the response of high latitudes to increases in radiative forcing is much larger than elsewhere in the world, with warming being more than twice the global average. Other authors find different trends in average temperature depending on the location are Chang et al. (2020), Holt and Teräsvirta (2020) and Gadea Rivas and Gonzalo (2022).⁴⁴4It is also important to remark that, although in this paper we focus on urban stations, it could also be interesting to extend the analysis to non-urban ones; see, for example, Hausfather et al. (2013) for an analysis of the impact of urbanization on temperature trends and the discussion by Estrada and Perron (2021). The heterogeneity in climate change may have important consequences for policy makers; see Kaufmann et al. (2017), who argue that the scepticism about climate change could partially be caused by its spatial heterogeneity, and Zaval et al. (2014) and Binelli, Loveless and Schaffuer (2023), who causally link perceived changes in local temperature to changes in global warming beliefs. Furthermore, Holt and Teräsvirta (2020) find evidence of co-shifting of hemispheric temperatures, which aligns with earlier evidence from Kaufmann and Stern (1997) on the co-movement of hemispheric temperature evolution and the anthropogenic character of climate change. Recently, Gao, Linton and Peng (2024) propose a nonparametric panel model for climate data with seasonal and spatial variation allowing the trends to vary by location and season. Marotta and Mumtaz (2023) specify a DFM with stochastic volatility to model annual temperatures in 160 countries around the world, assuming that there are global and country-specific factors. They conclude that the common factor explains a considerable portion of the average temperature globally.

In this section, we describe the statistical methodologies proposed to represent centre/log-range temperature, first, at a given location and, second, at several locations simultaneously. The approach used for the first case is based on unobserved components time series models and the use of the KFS algorithms to extract the unobserved trends and seasonal components. On the other hand, the methodology designed to study several locations is based on ML-DFM with the factors extracted using KFS.

In this section, we describe the main empirical characteristics of centre and log-range temperature in the Iberian Peninsula.

The results of fitting model (1) separately in the four locations considered, Barcelona, Coruña, Madrid and Seville, are reported in the Online Appendix. The reported estimates show that, in concordance with the descriptive analysis carried out in the previous section, and regardless of the particular location considered, the shocks of the trend, seasonal and irregular components of the centre and log-range are uncorrelated.¹³¹³13All calculations are carried out using the KFAS library developed by Helske (2017) in the R programming environment. Consequently, in this section, we fit separate DF-BSM to centre and log-range at each location.

The estimated parameters of model (1) are reported in Table 2, which also reports the RWD and IRW statistics as well as those for the null of the seasonal variances being zero.¹⁴¹⁴14Note that, based on previous analysis not reported to save space, the seasonal variances of the last five frequencies are assumed to be equal. Some diagnostic statistics can be found in Figures 6 to 9. We observe that, regardless of whether we look at centre or log-range temperature and of the location, the residuals are uncorrelated. Furthermore, as expected, although normality seems to be an appropriate approximation for centre temperature, log-ranges have heavy tails, mainly in the two interior locations.

Several important conclusions about the nature of the trend and seasonal component of the centre (average) and log-range (variability) temperature in each of the four locations considered can be extracted from the estimated variances reported in Table 2. First, we can observe that, regardless of the location, the trend of the centre temperature is represented by an integrated random walk with $\hat{\sigma}^{2}_{\eta}$ being not significantly different from zero, while $\hat{\sigma}^{2}_{\xi}\neq 0$. This implies a stochastic trend with a smooth evolution. This result concurs with the works that conclude that the trend of average temperature is stochastic. Furthermore, according to the integrated random walk model, the slope of the trend can evolve over time.

Figure 10 displays the estimated (smoothed) trends of centre temperature in the four selected locations together with 95% bounds obtained using the MSEs delivered by the Kalman filter. The figure also shows that there is a smoothly increasing trend over the last 90 years. Consider, for example, the centre temperature at Barcelona, which starts around $15^{\circ}C$ in 1930 while the estimated underlying level in December 2020 is $17.298^{\circ}C$. Note that even if we consider the uncertainty in the estimated trend, the increase of 2 degrees in the centre temperature over the last 90 years is clearly significant. We can also observe in Figure 10 that the slope of the trend is changing over the sample period, and, except for Barcelona, it is smaller since 2000. This deceleration is in concordance with the hiatus of average temperature described above. However, note that there is a recent acceleration since the second decade of the XXI Century, which is more pronounced in Barcelona and less in Seville. Table 2 also reports the estimated trends and slopes at the end of the sample period, December 2020, for the centre at each location. Note that, in December 2020, the estimated level of the trend is largest in Seville (18.896) and Barcelona (17.294) followed by Madrid (15.889) and Coruña (14.301). Even more important is to observe that the estimated slopes of the trend are larger in the coastal locations (0.004) and smaller at the interior locations, Madrid (0.003) and Seville (0.002), with these two latter slopes being not significant.

Second, we look now at the results for the seasonal component of the centre. Regardless of the location, we observe that the seasonal component is stochastic, with the corresponding variances significantly different from zero. The evolution of the estimated smoothed seasonal component of the centre temperature in each of the four locations is plotted in Figure 11. Our results are in line with those works that support an stochastic evolution of the seasonality of temperatures, even if, as observed in Figure 11, this evolution is very smooth; see also the results by Bogalo, Poncela and Senra (2024), who also conclude that there are changes in the seasonal pattern.

However, the trend of the log-range has a constant slope equal to zero, implying that $\mu_{t}$ is a random walk. Henceforth, the level of log-range is non-stationary implying that minimum and maximum temperatures are not cointegrated. This finding is relevant because it shows that coldest and warmest months over time do not follow the same stochastic pattern in the long run. Figure 12 plots the smoothed estimated level of the log-range temperature, and shows a random evolution without a particular trend. Table 2 reports the estimated value of $\mu_{T}$ and shows that it is larger in Madrid and Seville (2.48 and 2.57, respectively) than at the coastal cities (2.20 at Barcelona and 2.18 at Coruña). The dispersion between minimum and maximum temperatures has increased in recent years.

Finally, regardless of the location, the seasonal component of the dispersion between minimum and maximum temperature is strong and has a stochastic evolution with the corresponding variances being significantly different from zero; see Figure 13.

It is well known that local climate can exacerbate the impact of global warming; see, for example, Estrada and Perron (2021). Consequently, disentangling the local and global drivers of warming can guide local adaptation policies. In this section, we fit a multi-level DFM to the centre and log-range temperature observed at 68 locations in the Iberian Peninsula. As discussed in previous sections, the seasonal component is ubiquitous. Therefore, we extract the seasonal component from the centre and log-range temperature before analysing the common patterns in temperature trends.¹⁵¹⁵15The seasonal component is extracted from each series after fitting a structural model by subtracting the estimated filtered seasonal component.

In this paper, we propose using unobserved component models to represent the dynamic evolution of centre and log-range temperature observed monthly at different locations in the Iberian Peninsula over the last century. By doing so, we contribute to the literature on climate change in two main ways. First, from a methodological point of view, we use the information not only of the centre (average) temperature but also information associated to extreme (minimum and maximum) temperatures. Also, the methodology implemented in this paper allows the incorporation of seasonal patterns and the distinction between deterministic and stochastic components in temperatures. Finally, the methodology can be implemented to represent centre and log-range temperature not only at a given location but also at a large number of locations. Consequently, it allows the simultaneous analysis of the common components and the heterogeneity of climate change and the analysis of the relative importance of both elements. This methodology can obviously be implemented to analyse temperature in other areas of the world.

Second, in the particular application to model temperatures in the Iberian Peninsula, we conclude that, at each particular location, the centre and log-range temperature can be modelled separately. Furthermore, they are characterised by strong seasonal stochastic components, which are similar but different in magnitude at different locations evolving very smoothly over time. We also conclude that the evolution of the centre temperature can be represented by a smooth integrated random walk, which implies changes in the slope of the trend with a recent increase at some particular locations. However, the evolution of the distance between minimum and maximum temperatures is characterised by an evolving level modelled as a random walk.

Finally, when jointly modelling centre and log-range deseasonalised temperature at several locations, we show that the centre temperature are characterised by a strong global component with relatively weaker regional components. However, the distance between the minimum and maximum temperature is characterised by global and regional components with similar weights.

Bai, J. and S. Ng (2005), Tests for skewness, kurtosis, and Normality for time series data, Journal of Business & Economic Statistics, 23(1), 49-60.

Baillie, R.T. and S.K. Chung (2002), Modeling and forecasting from trend-stationary long memory models with applications to climatology, Journal of Forecasting, 18(2), 215-226.

Banbura, M., D. Giannone and L. Reichlin (2011), Nowcasting, in Clements, M.P. and D.F. Hendry (eds.), Oxford Handbook of Economic Forecasting, Oxford University Press.

Barbosa, S.M., M.G. Scotto and A. Alonso (2011), Summarizing changes in air temperature over Central Europe by quantile regression and clustering, Natural Hazards and Earth System Sciences, 11, 3227-3233.

Binelli, C., M. Loveless and B.F. Schaffner (2023), Explaining perceptions of climate change in the US, Political Research Quarterly, 76(1), 365-380.

Bloomfield, P. (1992), Trends in global temperature, Climatic Change, 21, 1-16.

Bogalo, J., P. Poncela and E. Senra (2024), Understanding fluctuations through multivariate circulant singular spectrum analysis, Expert Systems with Applications, 123827.

Busetti, F. and A.C. Harvey (2003), Seasonality tests, Journal of Business & Economic Statistics, 21(3), 420-436.

Campbell, S.D. and F.X. Diebold (2005), Weather forecasting for weather derivatives, Journal of the American Statistical Association, 100(469), 6-16.

Chang, Y., R.K. Kaufmann, C.S. Kim, J.I. Millers, J.Y. Park and S. Park (2020), Evaluating trends in time series of distributions: A spatial fingerprint of human effects on climate, Journal of Econometrics, 214(1), 274-294.

Chen, L. J. Gao and F. Vahid (2022), Global temperatures and greenhouse gases: A common features approach, Journal of Econometrics, 230(2), 240-254.

Coggin, T.D. (2012), Using econometric methods to test for trends in the HadCRUT3 global and hemispheric data, International Journal of Climatology, 32(2), 315-320.

Cohen, J.L., J.C. Furtado, M. Barlow, V.A. Alexeev and J.E. Cherry (2012), Asymmetric seasonal temperature trends, Geophysics Research Letters, 39, L04705.

Delle Chiaie, S., L. Ferrara and D. Giannone (2022), Common factors of commodity prices, Journal of Applied Econometrics, 37(3), 461–476.

Deng, Q. and Z. Fu (2019), Comparison of methods for extracting annual cycle with changing amplitude in climate series, Climate Dynamics, 52, 5059-5070.

Diebold, F.X. and G.D. Rudebusch (2022a), On the evolution of U.S. temperature dynamics, in Chudik, A., C. Hsiao and A. Timmermann (eds.), Essays in Honor of Hashem Pesaran: Prediction and Macro Modeling, Advances in Econometrics, 43A, Emerald Publishing Limited.

Diebold, F.X. and G.D. Rudebusch (2022b), Probability assessments of an ice-free Arctic: Comparing statistical and climate model projections, Journal of Econometrics, 231(2), 520-534.

Diebold, F.X. and A.S. Senhadji (1996), The uncertain unit root in real GNP: Comment, American Economic Review, 86(5), 1291-1298.

Doz, C., D. Giannone and L. Reichlin (2011), A two-step estimator of large approximate dynamic factor models based on Kalman filtering, Journal of Econometrics, 164(1), 188-205.

Doz, C., D. Giannone and L. Reichlin (2012), A quasi-maximum likelihood approach for large, approximate dynamic factor models, Review of Economics and Statistics, 94(4), 1014-1024.

Dupuis, D.J. (2012), Modelling waves of extreme temperature: the changing tails of four cities, Journal of the American Statistical Association, 107(497), 24-39.

Dupuis, D.J. (2014), A model for nighttime minimum temperatures, Journal of Climate, 27(19), 7207-7229.

Estrada, F. and P. Perron (2017), Extracting and analyzing the warming trend in global and hemispheric temperatures, Journal of Time Series Analysis, 38, 711-732.

Estrada, F. and P. Perron (2021), Disentangling the trend in the warming of urban areas into global and local factors, Annals of the New York Academy of Sciences, 1504, 230-246.

Estrada, F., C. Gay and A. Sánchez (2010), A reply to “Does temperature contain a stochastic trend? Evaluating conflicting statistical results” by Kaufmann et al., Climatic Change, 101(3), 407-414.

Estrada, F., D. Kim and P. Perron (2021), Spatial variations in the warming trend and the transition to more severe weather in midlatitudes, Scientific Reports, 11(145).

Estrada, F. P. Perron and B. Martínez-López (2013), Statistically derived contributions of diverse human influences to twentieth-century temperature changes, Nature Geoscience, 6(12), 1050-1055.

Fatichi, S., S.M. Barbosa, E. Caporali and M.E. Silva (2009), Deterministic versus stochastic trends: Detection and challenges, Journal of Geophysical Research, 114, D18121.

Friedrich, M., S. Smeekes and J.-P. Urbain (2020), Autoregressive wild bootstrap inference for nonparametric trends, Journal of Econometrics, 214(1), 81-109.

Friedrich, M., E. Beutner, H. Reuvers, S. Smeekes, J.-P. Urbain, W. Bader, B. Franco, B. Lejeune and E. Mahieu (2020), A statistical analysis of time trends in atmospheric ethane, Climatic Change, 162, 105-125.

Gadea Rivas, M.D. and J. Gonzalo (2020), Trends in distributional characteristics: Existence of global warming, Journal of Econometrics, 214, 153-174.

Gadea Rivas, M.D. and J. Gonzalo (2022), A tale of three cities: climate heterogeneity, SERIEs, 13, 475-511.

Gadea-Rivas, M.D., J. Gonzalo and A. Ramos (2023), Trends in temperature data: Micro-foundations of their nature, arXiv:2312.06379.

Gao, J. and K. Hawthorne (2006), Semiparametric estimation and testing of the trend of temperature series, Econometrics Journal, 9(2), 332-355.

Gao, J., O. Linton and B. Peng (2024), A non-parametric panel model for climate data with seasnal and spatial variation, Journal of the Royal Statistical Society. Series A-Statistics in Society, 187(1), 158-177.

Gay, G.C., F. Estrada and A. Sánchez (2009), Global and hemispheric temperatures revisited, Climatic Change, 94(3-4), 333-349.

Gil-Alana, L.A. (2005), Statistical modelling of the temperatures in the Northern hemisphere using fractional integration techniques, Journal of Climate, 18(24), 5357-5369.

Gil-Alana, L.A. (2008), Time-trend estimation with breaks in temperature time series, Climatic Change, 89, 325-337.

González-Rivera, G., Y. Luo and E. Ruiz (2020), Prediction regions for interval-valued time series, Journal of Applied Econometrics, 35(4), 373-390.

Good, S., G. Corlett, J. Remedios, E. Noyes and D. Llewellyn-Jones (2007), The global trend in sea surface temperature from 20 years of advanced very high resolution radiometer data, Journal of Climate, 20, 1255-1264.

Harris, I., T.J. Osborn, P. Jones and D. Lister (2020), Version 4 of the CRUTS monthly high-resolution gridded multivariate climate dataset, Scientific Data, 7(109).

Harvey, A.C. (1989), Forecasting, Structural Time Series Models and the Kalman Filter, Cambridge University Press, Cambridge.

Harvey, A.C. (1997), Trends, cycles and autoregression, Economic Journal, 107, 192-201.

Harvey, A.C. (2001), Testing in unobserved components models, Journal of Forecasting, 20(1), 1-19.

Harvey, A.C. and A. Jaeger (1993), Detrending, stylized facts and the business cycle, Journal of Applied Econometrics, 8, 231-247.

Harvey, A.C. and T. Mills (2003), Modelling trends in central England temperatures, Journal of Forecasting, 22, 35-47.

Hausfather, Z., M.J. Menne, C.N. Williams, T. Masters, R. Broberg and D. Jones (2013), Quantifying the effect of urbanization in U.S. historical climatology network temperature records, Journal of Geophysical Research: Atmospheres, 118, 481-494.

Helske, J. (2017), KFAS: Exponential family state space models in R, Journal of Statistical Software, 78(10).

Hillebrand, E. and T. Proietti (2017), Phase changes and seasonal warming in early instrumental temperature records, Journal of Climate, 30(17), 6795-6821.

Hindrayanto, I., J.A.D. Aston, S.J. Koopman and M. Ooms (2013), Modeling trigonometric seasonal components for monthly economic time series, Applied Economics, 45(21), 3024-3034.

Hodrick R.J. and E.C. Prescott (1997), Postwar US business cycles: an empirical investigation, Journal of Money, Credit and Banking, 24,1-16.

Holt, M.T. and T. Teräsvirta (2020), Global hemispheric temperatures and co-shifting: A vector shifting-mean autoregressive analysis, Journal of Econometrics, 214(1), 198-215.

Hsiang, S. and R.E. Koop (2018), An economist’s guide to climate change science, Journal of Economic Perspectives, 32, 3-32.

IPCC (2014), AR5 Synthesis report: Climate Change 2014. Contribution of working groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. [Core Writing Team, R.K. Pachauri and L.A. Meyer (eds.)]. IPCC, Geneva, Switzerland.

IPCC (2023), AR6 Synthesis report: Climate Change 2022. Contribution of working groups I, II and III to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. [Core Writing Team, Aldunce, P. et al.]. IPCC, Geneva, Switzerland.

Katz, R.W. and B.G. Brown (1992), Extreme events in a changing climate: Variability is more important than averages, Climatic Change, 21, 289-302.

Kaufmann, R.K. and D.I. Stern (1997), Evidence from human influence on climate from hemispheric temperature relations, Nature, 388(6637), 39-44.

Kaufmann, R.K., H. Kauppi and J.H. Stock (2010), Does temperature contain a stochastic trend? Evaluating conflicting statistical results, Climatic Change, 101, 395-405.

Kaufmann, R.K., H. Kauppi, M.L. Mann and J.H. Stock (2013), Does temperature contain a stochastic trend: linking statistical results to physical mechanisms, Climatic Change, 118, 729-743.

Kaufmann, R.K., M.L. Mann, S. Gopal, J.A. Liederman, P.D. Howe, F. Pretis, X. Tang and M. Gilmore (2017), Spatial heterogeneity of climate change as an experimental basis for skepticism, Proceedings of the National Academy of Sciences, PNAS, 114(1), 67-71.

Kew, S.F., S.Y. Philip, G. Jan van Oldenborgh, G. van der Schrier, F.E. Otto and R. Vautard (2019), The exceptional summer heat wave in Southern Europe 2017, Bulletin of the American Meteorological Society, 100(1), S49-S53.

Kim, D., T. Oka, F. Estrada and P. Perron (2020), Inference related to common breaks in a multivariate system with joined segmented trends with applications to global and hemispheric temperatures, Journal of Econometrics, 214, 130-152.

Kruse-Becher, R. (2023), Adaptive now- and forecasting of global temperatures under smooth structural changes, mimeo.

Maia, A.L.S. and F.A.T. de Carvalho (2011), Holt’s exponential smoothing and neural network models for forecasting interval-valued time series, International Journal of Forecasting, 27(3), 740-759.

Mangat, M.K. and E. Reschenhofer (2020), Frequency-domain evidence for climate change, Econometrics, 8(3), 28.

Marotta, F. and H. Mumtaz (2023), Vulnerability to climate change: Evidence from a dynamic factor model, Oxford Smith School of Enterprise and the Environment Working Paper no. 23-06.

Megdhaug, L., M.B. Stolpe, E.M. Fisher and R. Knutti (2017), Reconciling controversies about the “global warming hiatus”, Nature, 545(7652), 41-47.

Meng, X. and J.W. Taylor (2022), Comparing probabilistic forecasts of the daily minimum and maximum temperature, International Journal of Forecasting, 38, 267-281.

Miller, J.I. and K. Nam (2020), Dating Hiatuses: A statistical model of the recent slowdown in global warming and the next one, Earth System Dynamics, 11, 1123-1132.

Mitchell, T.D. and P.D. Jones (2005), An improved method of constructing a database of monthly climate observations and associated high-resolution grids, International Journal of Climatology, 25(6), 693-712.

Nyblom, J. and A.C. Harvey (2000), Testing against smooth stochastic trends, Journal of Applied Econometrics, 16(3), 415-429.

Palmer, T. and R. Stevens (2019), The scientific challenge of understanding and estimating climate change, Proceedings of the National Academy of Sciences, 116(49), 24390-24395.

Pezzulli, S., D. Stephenson and A. Hannachi (2005), The variability of seasonality, Journal of Climate, 18, 71-88.

Phella, A., V.J. Gabriel and L.F. Martins (2024), Predicting tail risks and the evolution of temperatures, Energy Economics, 131, 107286.

Phillips, P.C.B. (2005), Challenges of trending time series econometrics, Mathematics and Computers in Simulation, 68(5), 401-416.

Phillips, P.C.B. (2010), The mysteries of trend, Macroeconomic Reviews, 9(2), 82-89.

Pretis, F. and M. Allen (2013), Breaks in trends, Nature Geoscience, 6(12), 992-993.

Pretis, F. and D.F. Hendry (2013), Comment on “Polynomial cointegration tests of anthropogenic impact on global warming” by Beenstock et al. (2012)-some hazards in econometric modelling of climate change, Earth System Dynamics, 4, 375-384.

Pretis, F., M.L. Mann and R.K. Kaufmann (2015), Testing competing models of the temperature hiatus: assessing the effects of conditioning variables and temproal uncertainties through sample-wide break detection, Climatic Change, 131, 705-718.

Proietti, T. and E. Hillebrand (2015), Seasonal changes in central England temperatures, Journal of the Royal Statistical Society, Series A (Statistics and Society), 180(3), 769-791.

Qu, M., J. Wan and X. Hao (2014), Analysis of diurnal air temperature range in the continental United States, Weather and Climate Extremes, 4, 86-95.

Rao, B.B. (2010), Deterministic and stochastic trends in time series models: a guide for the applied economist, Applied Economics, 42(17), 2193-2202.

Rodríguez-Caballero, C. and M. Caporin (2019), A multilevel factor approach for the analysis of CDS commonality and risk contribution, Journal of International Financial Markets, Institutions & Money, 63, 101–144.

Schmidt, G.A., D.T. Shindell and K. Tsigaridis (2014), Reconciling warming trends, Nature Geoscience, 7(3), 158-160.

Scotto, M.G., S.M. Barbosa and A.M. Alonso (2011), Extreme value and cluster analysis of European daily temperature series, Journal of Applied Statistics, 38(12), 2793-2804.

Seidel, D.J. and J.R. Lanzante (2004), An assessment of three alternatives to linear trends for characterizing global atmospheric temperature changes, Journal of Geophysical Research: Atmospheres, 109(D14).

Stern, D.I. and R.K. Kaufmann (2000), Detecting a global warming signal in hemispheric temperature series: A structural time series analysis, Climate Change, 47, 411-438.

Ventosa-Santaulalia, D., D.R. Heres and L.C. Martínez-Hernández (2014), Long-memory and the sea-level temperature relationship: A fractional cointegration approach, PloS One, 9(11), e113439.

Vera-Valdés, J.E. (2021), Temperature anomalies. Long-memory and aggregation, Econometrics, 9(1), 9.

Visser, H. and J. Molenaar (1995), Trend estimation and regression analysis in climatological time series: An application of structural time series models and the Kalman filter, Journal of Climate, 8(5), 969-979.

Vogelsang, T.J. and P.H. Franses (2005), Are winters getting warmer?, Environmental Modelling Software, 20, 1449-1455.

Vose, R.S., D.R. Easterling and B. Gleason (2005), Maximum and minimum temperature trends for the globe: An update through 2004, Geophysical Research Letters, 32, L23822.

Wang, Z., Y. Jiang, H. Wan, J. Yan and X. Zhang (2021), Toward optimal fingerprint in detection and attribution of changes in climate extremes, Journal of the American Statistical Association, 116(533), 1-13.

Wijngaard, J.B., A.M.G. Klein Tauk and G.P. Können (2003), Homogeneity of 20th century European daily temperature and precipitations series, International Journal of Climatology, 23, 679-692.

Woodward, W.H. and H.L. Gray (1993), Global warming and the problem of testing for trend in time series data, Journal of Climate, 6, 953-962.

Xu, W., Q. Li, X.L. Wang, S. Yao, L. Cao and Y. Feng (2013), Homogenization of Chinese daily surface air temperatures and analysis of trends in the extreme temperature indices, Journal of Geophysical Research: Atmospheres, 118(17), 9708-9720.

Zaval, L. E.A. Keenan, E.J. Johnson and E.U. Weber (2014), How warm days increase belief in global warming, Nature Climate Change, 4(2), 143-147.

Zheng, X. and R.E. Basher (1999), Structural time series models and trend detection in global and regional temperature series, Journal of Climate, 12, 2347-2358.